Structural and Kinetic Analyses of Arginine Residues in the Active Site of the Acetate Kinase from Methanosarcina thermophila*

Acetate kinase catalyzes transfer of the γ-phosphate of ATP to acetate. The only crystal structure reported for acetate kinase is the homodimeric enzyme from Methanosarcina thermophila containing ADP and sulfate in the active site (Buss, K. A., Cooper, D. C., Ingram-Smith, C., Ferry, J. G., Sanders, D. A., and Hasson, M. S. (2001) J. Bacteriol. 193, 680–686). Here we report two new crystal structure of the M. thermophila enzyme in the presence of substrate and transition state analogs. The enzyme co-crystallized with the ATP analog adenosine 5′-[γ-thio]triphosphate contained AMP adjacent to thiopyrophosphate in the active site cleft of monomer B. The enzyme co-crystallized with ADP, acetate, Al3+, and F- contained a linear array of ADP-AlF3-acetate in the active site cleft of monomer B. Together, the structures clarify the substrate binding sites and support a direct in-line transfer mechanism in which AlF3 mimics the meta-phosphate transition state. Monomers A of both structures contained ADP and sulfate, and the active site clefts were closed less than in monomers B, suggesting that domain movement contributes to catalysis. The finding that His180 was in close proximity to AlF3 is consistent with a role for stabilization of the meta-phosphate that is in agreement with a previous report indicating that this residue is essential for catalysis. Residue Arg241 was also found adjacent to AlF3, consistent with a role for stabilization of the transition state. Kinetic analyses of Arg241 and Arg91 replacement variants indicated that these residues are essential for catalysis and also indicated a role in binding acetate.

Phosphoryl transfer is a key reaction in numerous biological processes, playing roles in signaling mechanisms, energy transfer, and energy storage in both eukaryotic and prokaryotic cells (1). One of the earliest phosphoryl transfers identified was the phosphorylation of acetate by ATP to form acetyl phosphate (AcP) 1 and ADP, described in 1944 by Lippman (2). This re-versible reaction is catalyzed by acetate kinase, which is widely distributed among anaerobic prokaryotes playing a central role in energy-yielding metabolism by synthesizing ATP from acetyl phosphate generated in fermentation pathways. The enzyme also plays an essential role in the fermentation of acetate to methane, which accounts for most of the one billion metric tons of methane produced annually from the decomposition of organic matter by anaerobic microbial consortia (3). In Methanosarcina thermophila, acetate kinase catalyzes the first step in the pathway by activating acetate to acetyl phosphate prior to transfer of the acetyl moiety to CoA catalyzed by phosphotransacetylase (4,5). In later steps of the pathway, the acetyl moiety is further metabolized to methane and carbon dioxide (6).
Although acetate kinase was one of the first enzymes to be investigated mechanistically, details remain elusive; indeed, the first crystal structure was obtained only recently for the M. thermophila enzyme, identifying acetate kinase as a member of the acetate and sugar kinase-Hsp70-actin (ASKHA) structural superfamily and the best candidate for the common ancestor of this family (7). The earliest kinetic studies of the enzyme from Escherichia coli suggested a ping-pong mechanism (8), and evidence for a covalent phosphoryl intermediate supported this mechanism (9, 10); however, it was later shown that the phosphoryl-enzyme complex is not kinetically competent (11). Additionally, the discovery that the E. coli acetate kinase is able to phosphorylate enzyme I of the phosphotransferase system (12) and CheY (13) in vitro indicates the phosphoenzyme functions in sugar transport. Later investigations reported inversion of the stereochemistry about the phosphorous (14) and isotope exchange kinetics inconsistent with the covalent kinase mechanism (15) and supporting a direct in-line phosphoryl transfer. More recently, the acetate kinase from M. thermophila was shown to be inhibited by components of a putative transition state analogue ADP-AlF x -acetate (16) in which the AlF x is proposed to mimic the meta-phosphate in a direct phosphoryl transfer mechanism. No structural evidence for either the covalent or in-line mechanism has been reported previously.
Access to the crystal structure (7) and production of the M. thermophila acetate kinase in E. coli (17) have allowed experimental approaches not previously employed to investigate the catalytic mechanism of this enzyme. The structure of the homodimeric acetate kinase co-crystallized with ATP (the ATP-AK structure) reveals ADP in a cleft with contacts that are conserved in the nucleotide binding sites of other ASKHA family members, which identifies the active site of the M. thermophila acetate kinase. The active site contains Arg 91 and Arg 241 , a result consistent with roles for these residues in substrate binding, catalysis, or both. It was hypothesized that Arg 91 binds acetyl phosphate and Arg 241 binds acetate based on a postulated binding site identified in the crystal structure (7). The low specific activity reported for Arg 91 and Arg 241 replacement variants relative to the wild type is consistent with a role for both arginines in stabilizing the pentacoordinate transition state for the postulated direct in-line mechanism (16,18); however, the low activity of the variants precluded a determination of the steady state kinetic parameters. The ATP-AK structure contains only ADP, with the ␤-phosphate repelled by a sulfate ion and pointing away from the arginines; thus, the catalytically competent orientation of the ␥-phosphate of ATP relative to Arg 91 and Arg 241 is unknown. Here we present two novel M. thermophila acetate kinase structures obtained by co-crystallization with either the ATP analog ATP␥S (the ATP␥S-AK structure) or components of the putative transition state analog ADP-AlF X -acetate (the AlF 3 -AK structure) that, along with kinetic analyses utilizing an improved assay, allow us to further examine the roles for Arg 91 and Arg 241 . The results also provide the first structural evidence for the proposed acetate binding site and a direct in-line phosphoryl transfer mechanism.

EXPERIMENTAL PROCEDURES
Materials-Chemicals were purchased from Sigma, VWR Scientific Products, or Fisher. The pH values of ATP, ADP, and ATP␥S stock solutions were adjusted to 7.0 with sodium hydroxide, and concentrations were determined utilizing the extension coefficient (⑀ 259 ϭ 15.4 ϫ 10 3 M Ϫ1 cm Ϫ1 ). ATP and ADP stock solutions were prepared to be equimolar with magnesium chloride. Acetyl phosphate concentrations were determined by assay with hexokinase/glucose 6-phosphate dehydrogenase/acetate kinase. The pH of the acetate stock solution was adjusted to 7.0 with potassium hydroxide. Crystallization materials were obtained from VWR Scientific Products or Hampton Research.
Heterologous Production and Purification of Acetate Kinase-Plasmids for the R91A, R91L, R91K, R241A, R241L, and R241K variant acetate kinases previously generated were utilized for this study (18). The wild-type and variant acetate kinases were overproduced in E. coli BL21(DE3) (F-dcm ompT hsdS (rB-mB-) gal (DE3)) and purified as described previously (16, 18 -20). Protein purity was examined by SDS-PAGE (21), and protein concentrations were determined by the Bradford method (22), using Bio-Rad dye and bovine serum albumin as the standard. The yields and dimeric state of the variants were similar to those of the wild type enzyme (data not shown).
Enzymatic Assays-The hydroxamate assay adaptation of the Lipmann and Rose methods (2, 23, 24) detects acetyl phosphate formation and was previously used to determine the kinetic parameters in the forward (ADP/acetyl phosphate synthesis) direction. In this study, the kinetic parameters were determined utilizing an enzyme-linked assay system with pyruvate kinase and lactate dehydrogenase as previously described by Allen et al. (25) and utilized by Aceti with slight modifications (24 where E is concentration of enzyme in g/ml, and V max is the maximum ⌬A 340 /min determined from the Michaelis-Menten equation. When determining kinetic parameters in the reverse direction (ATP/ acetate synthesis), the previously described enzyme-linked assay was used (24). Assay components were 100 mM Tris (pH 7.4), 0.2 mM dithiothreitol, 10 mM MgCl 2 , 4.4 mM glucose, 1 mM NADP, 10 units of hexokinase (yeast), and 10 units of glucose-6-phosphatase (yeast). The ADP concentration was held in excess at 5 mM when K m(AcP) was determined, and the AcP concentration was held in excess at 10 mM when K m(ADP) was determined. Enzyme concentrations varied from 1 to 50 g/ml, depending upon enzyme activity to yield a linear rate over the duration of the assay. Kinetic constants were determined using nonlinear regression to fit data using the program Kaleidagraph (Synergy Software, Reading, PA).
Determination of Inhibition Constants for ATP␥S and Hydroxylamine-Inhibition constants for hydroxylamine and ATP␥S were determined by a 5 ϫ 5 matrix of conditions that systematically varies inhibitor and substrate concentrations (hydroxylamine versus ATP or acetate; ATP␥S versus ATP). Assays contained 60 mM HEPES, pH 7.0, 5 mM MgCl 2 , 16.7 units of pyruvate kinase, 36 units of lactate dehydrogenase, 3 mM phosphoenolpyruvate, 0.2 mM NADH, and 1 g/ml wildtype acetate kinase, with 200 mM acetate when ATP varied and 5 mM ATP when acetate varied. ATP concentrations were varied between 20 M and 1 mM, whereas acetate concentrations were varied from 0.2 to 10 mM. ATP␥S concentrations ranged from 0 to 300 M. Hydroxylamine concentrations ranged from 0 to 1 M. Kinetic parameters for inhibition were determined by linear regressing using the MINITAB program and a value for ␣ of 2.0 (26).
Guanidine Rescue of Activity-The ability of the hydrogen-donating guanidine to rescue the activity of the arginine variants was determined by including guanidine hydrochloride (GdnHCl) in the assay solution. Wild-type acetate kinase was assayed in the presence of increasing concentrations of GdnHCl to determine maximum concentration permissible in the assay conditions before enzyme activity is affected, and 200 mM GdnHCl was determined to be the maximum concentration tolerated (data not shown). Kinetic constants for wild-type, R91A, and R241A acetate kinases were determined utilizing the forward reaction assay solution (described under "Enzymatic Assays") in the presence of 200 mM guanidine hydrochloride. Substrate concentrations, enzyme concentrations, and assay times are as previously described.
Crystallization and Data Collection-The hanging drop method was used to co-crystallize acetate kinase with ATP␥S as previously described for co-crystallization of the enzyme with ATP (7,27). Wild-type acetate kinase (0.5 mg/ml) was incubated with 1 mM ATP␥S, 1.5 mM MgCl 2 , 315 mM (NH 4 ) 2 SO 4 , and 25 mM Tris (pH 7.4) in a drop that was equilibrated against a reservoir of 1.7 mM (NH 4 ) 2 SO 4 for 2 h at room temperature. The drop was then equilibrated against a reservoir of 0.8 mM (NH 4 ) 2 SO 4 at 37°C, with small crystals first appearing overnight and reaching maximum size at 14 days. Crystals are stable for at least 3 months. A single ATP␥S-AK crystal was transferred to a saturated glucose solution as a cryoprotectant and flash frozen in a liquid-N 2 stream. Data were collected at 100 K on the F2 beam line at the Cornell High Energy Synchrotron Source (Cornell University, Ithaca, NY), and image files were processed with DENZO/SCALE-PACK (28).
The hanging drop method was also used to co-crystallize acetate kinase in the presence of acetate, ADP, and AlF X . Prior to use in crystallization trials, 100 mM AlCl 3 and 50 mM NaF were pre-equilibrated overnight. Wild-type acetate kinase (0.5 mg/ml) was incubated with 1 mM ADP, 1.5 mM LiCl, 0.1 mM AlCl 3 plus 0.5 mM NaF, 10 mM acetate, 315 mM (NH 4 ) 2 SO 4 , and 25 mM Tris (pH 7.4) in a drop that was equilibrated against a reservoir of 1.7 mM (NH 4 ) 2 SO 4 for 2 h at room temperature. The drop was then transferred to a reservoir of 0.8 mM (NH 4 ) 2 SO 4 . Crystallization was allowed to proceed as described above, although the crystals reached a smaller macroscopic size. The crystals were transferred to a saturated glucose solution as a cryoprotectant and frozen in liquid N 2 . Data were collected at 100 K at Argonne National Synchrotron (Argonne, IL), and image files were processed with DENZO/SCALEPACK (28).
Structure Solution and Refinement-Since the unit cell dimensions of both the ATP␥S-AK and AlF 3 -AK structures were nearly identical to the previously solved ATP-AK structure, molecular replacement was used to determine the structures of the complexes. The program AMORE (CCP4 suite, version 1.1) (29) was used to perform molecular replacement using as a search model the ATP-AK structure deposited in the Protein Data Bank data base (identifier 1G99) (7). Refinement was performed with CNS_solve (version 4.1) (30), and molecular models were built and visualized with O (31). Ideal parameter and topology files for AMP, ADP, SO 4 2Ϫ , NH 3 , AlF 3 , and acetate for CNS_solve were obtained from HIC-Up (32). Parameter and topology files for pyrophosphate were modified to include sulfur in place of one oxygen of the ␥-phosphate for thiopyrophosphate (TPP; Protein Data Bank residue name PIS) for use in CNS_solve. Designations for chains A and B in both structures were assigned so that chain A contains the same heteroatoms as in the ATP-AK structure. Model suitability was determined with PROCHECK (33), and overall molecular refinement statistics are presented in Table IV. The rotation matrix and root mean square values between either the ATP␥S-AK or AlF 3 -AK structure and the ATP-AK structure were calculated with LSQ_MAN (27). Coordinate and structure files have been deposited at the RCSB Data bank with the PDB identifiers 1TUU for the acetate kinase-ATP␥S complex and 1TUY for the acetate kinase-ADP-AlF 3 -acetate complex.

Kinetic Parameters of the Wild-type and Variant Acetate
Kinases-Prior to assessment of the kinetic parameters of the variant acetate kinases, the accuracy of the hydroxamate assay was tested utilizing the enzyme-linked assay described under "Experimental Procedures." Hydroxylamine was found to inhibit wild-type acetate kinase in a nonlinear and noncompetitive fashion versus either acetate or ATP (Fig. 1), as described by Equation 2, where v represents the reaction velocity, V max is the maximal reaction velocity, K a is the  (18,19,24) utilizing the hydroxamate assay. Furthermore, kinetic parameters for the wild-type enzyme in the direction of ATP synthesis have not been determined. Thus, kinetic constants for the wild-type and arginine replacement variants (Tables I and II) were determined in both directions utilizing the enzyme-linked assays.
Although the k cat determined for the wild-type acetate kinase approximated the values (1050 -1596 s Ϫ1 ) reported using the hydroxamate assay, the K m(ATP) and K m(acetate) values determined with the enzyme-linked assay in the direction of ADP synthesis (Table I) were at least 12-and 7-fold lower than those previously reported (18,19,24). When assayed in the direction of ATP synthesis (Table II), the wild-type k cat approximated the value determined in the direction of ADP synthesis ( Table I).
The K m(ADP) approximated the K m(ATP) ; however, the K m(AcP) was nearly 6-fold less than the K m(acetate) .
It was reported previously that all of the variants shown in Tables I and II purified according to the wild-type are dimeric, and the CD spectra of the R91A and R241A variants are nearly identical to wild type, indicating no gross conformational changes in the variants relative to wild type (18). All of the Arg 91 and Arg 241 variants showed large decreases in k cat relative to wild type when assayed in the direction of ADP synthesis, ranging from 250-fold for R91K to 8200-fold for R91A (Table I). The K m (ATP) values determined for all of the Arg 91 variants changed little relative to the wild type, with the largest effect being a 5-fold decrease for the R91A variant; however, the K m(acetate) values increased 93-, 156-, and 26-fold for the R91A, R91L, and R91K variants, respectively (Table I). Only a modest increase in K m(ATP) compared with wild-type was determined for the R241A variant, arguing against an important role in binding ATP. In contrast, large increases were observed in K m(ATP) for the R241L (213fold) and R241K (143-fold) variants. The Arg 241 variants also displayed substantial increases for K m(acetate) : 263-fold for R241A, 100-fold for R241L, and 29-fold for R241K. Notably, the increases in K m(acetate) were severalfold less when Arg 91 or Arg 241 was replaced with a Lys as opposed to the other residues tested.
When assayed in the direction of ATP synthesis (Table II), large decreases in k cat relative to wild-type were observed for all of the variants that were similar in magnitude to the decreases in k cat in the direction of ADP synthesis (Table I). The minor deviations in K m(AcP) for all of the variants relative to wild type argue against a role for these residues in binding acetyl phosphate. Although the 2-fold increase in K m(ADP) observed for the R241A variant relative to wild-type was also minor, moderately larger increases were observed for the R241L (13-fold) and R241K (6-fold) variants.
Guanidine Hydrochloride Rescue of Variants-GdnHCl is reported to rescue the k cat of arginine replacement variants of several enzymes for which arginine is essential (35)(36)(37)(38)(39)(40); thus, rescue of the R91A and R241A variants of the M. thermophila acetate kinase was investigated using the enzyme-linked assay to further address the role of these residues. The k cat of the wild-type acetate kinase was reduced to approximately one-  (Tables I and  II). However, a 10-fold increase in K m(acetate) was observed in the presence of GdnHCl, for which the most straightforward explanation is that GdnHCl occupies space near the acetate binding pocket. Analysis of the R91A variant in the presence of GdnHCl showed a 250-fold decrease in k cat and only modest changes in K m(ATP) and K m(acetate) as compared with the wildtype parameters in the presence of GdnHCl (Table III). However, a comparison of R91A in the presence of GdnHCl (Table  III) revealed a 15-fold increase in k cat , a 4-fold decrease in K m(ATP) , and a 3-fold decrease in K m(acetate) relative to the parameters obtained for this variant in the absence of GdnHCl (Table I).
Analysis of the R241A variant in the presence of GdnHCl revealed an 807-fold decrease in k cat , a 4-fold increase in K m(ATP) , and no significant change in K m(acetate) compared with parameters for the wild-type enzyme in the presence of GdnHCl (Table III). Comparison of the R241A variant kinetic parameters in the presence and absence of GdnHCl showed no appreciable differences in the k cat and K m(ATP) , whereas the K m(acetate) decreased 24-fold in the presence of GdnHCl.
Inhibition by ATP␥S-The first acetate kinase structure (ATP-AK) was obtained by co-crystallization of the M. thermophila enzyme with ATP; however, only ADP was identified in the active site with the ␤-phosphate repelled by a sulfate ion precluding the catalytically competent location of the ␥-phosphate of ATP relative to Arg 91 and Arg 241 (7). Since it is likely that ATP hydrolysis occurred during crystallization, the nonhydrolyzable ATP analogue ATP␥S was co-crystallized with the enzyme in anticipation of generating a more catalytically relevant complex. In order to better interpret the ATP␥S-AK structure, the influence of ATP␥S on acetate kinase activity was evaluated. No acetate kinase activity was detected with the coupled assay system when ATP␥S replaced ATP, a result consistent with no hydrolysis of the ␥-thiophosphate. Inhibition of the wild-type enzyme by ATP␥S was investigated to indicate whether the analog bound to the catalytic ATP binding site. Inhibition was determined in assays with five concentrations of ATP, each versus five concentrations of ATP␥S (a 5 ϫ 5 matrix) and holding acetate ( Fig. 2A) at a 100-fold excess relative to K m (Fig. 2B). Data were fit utilizing the equation for competitive inhibition (Equation 3),    putative transition state analog, ADP-AlF X -acetate, had the same C2 space group and similar unit cell dimensions as previously reported for the enzyme co-crystallized with ATP (16) ( Table IV). The ATP␥S-AK structure was solved by molecular replacement starting with the backbone coordinates reported for the published ATP-AK structure (7), and rigid body refinement was performed utilizing each monomer as the rigid body. The AlF 3 -AK structure was solved by molecular replacement as described above; however, rigid body refinement utilized each of the two domains within each monomer as the rigid body. Refinement of both the ATP␥S-AK and AlF 3 -AK structures resulted in models with a similar C ␣ backbone trace to each other and to the previously reported ATP-AK structure (Fig. 3). The overall structure of each acetate kinase homodimer resembles a bird with its wings spread. The "body" of the bird contains the dimer interface and is formed by the C-terminal domains of each monomer. The "wings" are formed by the N-terminal domains, and the active site of each monomer is located in the cleft between the two domains (Fig. 3). As in the ATP-AK structure, the wings of each monomer in both the ATP␥S-AK and AlF 3 -AK structures were closed onto the body to different extents. As reported for the ATP-AK structure, electron density in the active site of one of the monomers (designated monomer A) for both the ATP␥S-AK and AlF 3 -AK structures was fit to ADP and SO 4 2Ϫ , and the domains were closed less than in monomer B (Fig. 4) (Fig. 4). Monomer B of both the ATP␥S-AK and AlF 3 -AK structures contained electron densities not reported for the ATP-AK[B] structure that will be discussed separately for each new structure.

. Alignments of the C ␣ backbone of monomers A from the ATP␥S-AK and AlF 3 -AK structures (ATP␥S-AK[A] and AlF 3 -AK[A]) with monomer
In the ATP␥S-AK[B] structure, electron density in the active site was fit to AMP (Fig. 5) for which contacts (not shown) were the same as reported for the nucleotide base, ribose ring, and ␣-phosphate of the ADP that was reported in the ATP-AK[B] structure. These results, combined with the inhibition results, indicate that ATP␥S bound analogously to the catalytically competent binding of ATP. Additional electron density in the active site of the ATP␥S-AK[B] structure was fit to NH 3 that probably originated from the crystallization solution. The NH 3 was located at the mouth of the acetate binding pocket predicted from the ATP-AK structure (7) and supported by kinetic analyses of replacement variants. 2 The binding pocket is composed of Val 93 , Leu 122 , and Pro 232 and is proposed to accept the methyl group of acetate, thereby positioning the carboxyl group at the mouth of the pocket adjacent to Arg 241 in the approximate position of NH 3 (Fig. 5). Although kinetic analyses of replacement variants suggested a role for Arg 91 in binding acetate, this residue is positioned ϳ7 Å away from NH 3 (Fig. 5). Additional electron density was observed in the active site of ATP␥S-AK[B] that could not be fit to any component of the crystallization conditions. The presence of AMP in this active site suggested the possibility of hydrolysis of the ester bond between the ␣and ␤-phosphates of ATP␥S, resulting in AMP and thiopyrophosphate (TPP, H 2 PO 3 -O-HPO 2 S Ϫ ). For description, the phosphates of TPP are named ␤Ј and ␥Ј, reflecting their original positions in the ATP␥S. Since TPP had not been described previously, the structural topologies and parameters for pyrophosphate were manually modified to simulate TPP and used to fit the postulated TPP to the unidentified electron density in monomer B (Fig. 5), with a subsequent decrease in R free . The model showed a distance of 3.68 Å from the ␤Јphosphate of TPP to the proximal oxygen of the AMP phosphoryl group. These results suggest in situ hydrolysis of the ester bond between the ␣and ␤-phosphates of ATP␥S. The ␤Јand ␥Ј-phosphates of TPP were positioned in line with the ␣-phosphate of AMP and NH 3 (Fig. 5). Notably, the electron density for the ␥Ј-phosphate of TPP showed an interaction with the ⑀ amine group of Arg 241 and proximity to His 180 (Fig. 5). The position of the ␥Ј-phosphate of TPP was similar to the position of sulfate in the ATP␥S-AK[A] structure (Fig. 4A) and published ATP-AK structure (not shown). Furthermore, the ␤Јphosphate of TPP was positioned distinct from the ␤-phosphate of ADP determined for the ATP␥S-AK[A] structure (Fig. 4A) and published ATP-AK structure (not shown).
Electron density in the active site of the AlF 3 -AK[B] structure was fit to ADP, AlF 3 , and acetate (Fig. 6) in an arrangement consistent with a transition state analog of the direct in-line phosphoryl transfer mechanism in which the AlF 3 has been proposed to mimic the meta-phosphate (16). The ribose ring and ␣-phosphate of ADP were positioned in the active site similarly to AMP in the ATP␥S-AK[B] structure (Fig. 5), and the ␤-phosphate of ADP and AlF 3 were positioned in the active site similarly to the ␤Јand ␥Ј-phosphates of TPP in the ATP␥S-AK[B] structure. Acetate was found adjacent to AlF 3 and contained in the active site hydrophobic pocket formed by Val 93 , Leu 122 , and Pro 232 placing the carboxyl group of acetate within hydrogen bonding distance of Arg 241 (Fig. 6). This binding site for acetate was previously postulated (7) based on analogy to the substrate binding sites of other ASKHA family members and recently supported by kinetic analyses of site-specific variants. 2 The trigonal planar electron density of aluminum fluoride in the structure indicates that it formed with the stoichiometry AlF 3 as opposed to AlF 4 , consistent with the pH dependence of aluminum fluoride formation (42) and with the stoichiometry suggested by inhibition of the M. thermophila acetate kinase by the transition state analog (16). The plane of AlF 3 is oriented parallel to the ␤-phosphoryl group of ADP and acetate (Fig. 6) and therefore does not represent a true transition state, since a direct in-line transfer mechanism (S N 2 reaction) requires the plane of the AlF 3 to be perpendicular to both ADP and acetate. Although no contacts were evident, the AlF 3 was adjacent to Arg 241 and His 180 , two candidates for stabilizing the transition state (18,20). The results presented here provide the first structural evidence supporting the previously proposed direct in-line phosphoryl transfer mechanism (14 -16).
Several additional nuances of the AlF 3 -AK structure merit further description. Although the ribose of ADP in the AlF 3  structure. The backbone and angles for Gly 331 , a conserved feature of the ASKHA superfamily (43), are maintained in the crystal structure even when contacts to the substrates have changed. Additional electron density, not observed in any other acetate kinase crystal structure, was found within hydrogen bonding distance of acetate in the active site of AlF 3 -AK[B] and was fit to a water molecule (Wat2 in Fig. 6). The function of this water is unknown at present.

DISCUSSION
Kinetic Analysis of Site-specific Replacement Variants-The only kinetic parameters previously reported for the acetate kinase from M. thermophila were determined in the direction of ADP synthesis with the hydroxamate assay where hydroxylamine is a component of the assay mixture (18 -20, 34). Using the enzyme-linked assay in the direction of ADP synthesis, it was shown that hydroxylamine inhibits activity and could influence the kinetic parameters; therefore, the more sensitive and accurate enzyme-linked assay was used to measure the activity of the variants. The k cat values obtained for all of the Arg 91 and Arg 241 variants in both reaction directions establish that these residues are essential for catalysis and support the previously hypothesized role for these active site residues in stabilization of a meta-phosphate transition state in a direct in-line mechanism for phosphoryl transfer from ATP to acetate (16,18). It was previously postulated that Arg 241 also interacts with the carboxyl group of acetate and that Arg 91 interacts with the phosphoryl group of acetyl phosphate based on features of the ATP-AK crystal structure identifying putative binding sites for these substrates (7). However, the inconsequential changes in K m(AcP) relative to the wild-type enzyme in all of the Arg 91 and Arg 241 variants argue against a role for either residue in binding the phosphoryl group of acetyl phosphate. In contrast, substantial increases in K m (acetate) com- pared with wild-type were observed for all of the Arg 91 and Arg 241 variants, supporting the previously proposed role for Arg 241 in binding acetate and suggesting the same role for Arg 91 . The increases in K m(acetate) for the R91K and R241K variants were severalfold less than for the variants in which the arginines were replaced with Ala or Leu, a result consistent with the requirement for a positive charge in positions 91 and 241 to interact with the carboxyl group of acetate.
Interpretation of the K m(ATP) and K m(ADP) values obtained for the Arg 241 variants is not straightforward. The marginal increases relative to wild type for the R241A variant indicate a minor involvement for Arg 241 in binding ATP or ADP; however, large increases in both parameters were observed for the R241L and R241K variants. One possible explanation for these results is that, compared with Ala, the larger side chains of Leu and Lys sterically hinder ATP binding consistent with relatively lower K m(ADP) versus K m(ATP) values determined for the R241L and R241K variants.
The K m(acetate) values for the R91A and R241A variants were found to be significantly lower in the presence of GdnHCl, a result further supporting a role for the guanidino groups of the arginines interacting with the carboxyl group of acetate. Both the lower K m(acetate) and substantially higher k cat of the R91K variant compared with the other Arg 91 variants indicate that the positive charge of the guanidino moiety of Arg 91 enhances the catalytic efficiency. Indeed, the presence of GdnHCl decreased the K m(acetate) of the R91A variant while increasing the k cat . Thus, in addition to the proposed role in stabilization of the transition state, another potential role for Arg 91 may be to orient the carboxyl group of acetate for nucleophilic attack on the ␥-phosphate of ATP.
Analysis of Acetate Kinase Crystal Structures-The published ATP-AK structure co-crystallized with ATP contained only ADP in the active site cleft adjacent to sulfate that is proposed to displaced the ␤-phosphate and the inferred ␥-phosphate of ATP from the catalytically competent position (7). The ATP␥S-AK[A] and AlF 3 -AK[A] structures reported here also contained only ADP and sulfate positioned in the active site similarly to the ATP-AK structure; however, monomer B of both new structures was void of sulfate in the active site and revealed new structural information advancing an understanding of substrate binding and catalysis (16,18). The nucleotide base and ribose of AMP in the ATP␥S-AK[B] structure were positioned in the active site similarly to ADP in the published ATP-AK structure, which, combined with results showing inhibition of acetate kinase activity by ATP␥S, indicates that ATP␥S bound to monomer B of the ATP␥S-AK structure analogously to ATP. The close proximity of the AMP phosphate to the ␤Ј-phosphate of TPP suggests that ATP␥S had bound to the active site followed by in situ hydrolysis of the ester bond between the ␣and ␤-phosphates, yielding AMP and TPP. Although the TPP may have shifted slightly away from AMP after hydrolysis, the results suggest that the ␥Ј-phosphate of TPP approximates the active site position of the ␥-phosphate of ATP poised for catalysis.
The ␤-phosphate of ADP in the AlF 3 -AK[B] structure and the ␤Ј-phosphate of TPP in the ATP␥S-AK[B] structure were positioned similarly in the active site although differently from the ␤-phosphate of ADP in the published ATP-AK[A] structure and for monomer A of both the ATP␥S-AK and AlF 3 -AK structures that contain sulfate. In the ATP-AK[A] structure, the sulfate is proposed to displace the ␤-phosphate and the inferred ␥-phosphate of ATP from the catalytically competent position. Furthermore, the ␥Ј-phosphate of TPP in the active site was in the approximate position of the sulfate in the ATP-AK structure, consistent with the previous proposal that sulfate occupies the position of the ␥-phosphate of ATP (7). Although the ATP␥S-AK[B] structure indicates contact of the ␥Ј-phosphate of TPP with Arg 241 , kinetic analysis of the R241A variant suggests that this residue is not involved in binding ATP; instead, the kinetic results suggest that Arg 241 binds acetate.
The AlF 3 -AK[B] structure contained acetate in the active site, the first reported for any acetate kinase structure, positioned in a hydrophobic pocket as previously postulated (7) and supported by recent kinetic analyses of site-specific replacement variants of the enzyme. 2 The active site also contained AlF 3 , shown previously to mimic the planar phosphoryl group derived from the ␥-phosphate of NTPs in the catalytic transition state for a variety of kinases (16, 49 -51). The proximity to the carboxyl group of acetate and the ␤-phosphate of ATP in the AlF 3 -AK[B] structure suggests that AlF 3 also mimics the metaphosphate transition state in a direct in-line phosphoryl transfer mechanism for the acetate kinase from M. thermophila. These results provide the first structural evidence supporting the previously proposed direct in-line mechanism for acetate kinase (16,18). The AlF 3 was positioned in the active site of the AlF 3 -AK[B] structure approximately the same as the ␥Ј-phosphate of TPP and the inferred ␥-phosphate of ATP.
The direct in-line mechanism predicts a requirement for residues to stabilize the trigonal bipyramidal phosphate transition state by coordination with the three equatorial oxygen atoms. Residues Arg 241 and Arg 91 are candidates, based on the kinetic analyses of variants presented here. A role for His 180 in stabilizing the transition state has been proposed based on FIG. 8. Postulated mechanism of acetate kinase from M. thermophila for the forward (AcP-producing) reaction direction. A, ATP and acetate substrate interactions with acetate kinase; B, interactions for the proposed direct in-line mechanism transition state; C, interactions of the products ADP and AcP with acetate kinase. The arrows indicate direction of electron movement described under "Discussion."