Residues essential for catalysis and stability of the active site of Escherichia coli adenylosuccinate synthetase as revealed by directed mutation and kinetics.

Examined here by directed mutation, circular dichroism spectroscopy, and kinetics are the relationships of five residues, Asp13, Glu14, Lys16, His41, and Arg131, to the catalytic function and structural organization of adenylosuccinate synthetase from Escherichia coli. The D13A mutant has no measurable activity. Mutants E14A and H41N exhibit 1% of the activity of the wild-type enzyme and 2-7-fold increases in the Km of substrates. The mutant K16Q has 34% of the activity of wild-type enzyme and Km values for substrates virtually unchanged from those of the wild-type system. Mutation of Arg131 to leucine caused only a 4-fold increase in the Km for aspartate relative to the wild-type enzyme. The dramatic effects of the D13A, E14A, and H41N mutations on kcat are consistent with the putative roles assigned to Asp13 (catalytic base), His41 (catalytic acid), and Glu14 (structural organization of the active site). The modest effect of the R131L mutation on the binding of aspartate is also in harmony with recent crystallographic investigations, which suggests that Arg131 stabilizes the conformation of the loop that binds the beta-carboxylate of aspartate. The modest effect of the K16Q mutation, however, contrasts with significant changes brought about by the mutation of the corresponding lysines in the P-loop of other GTP- and ATP-binding proteins. Crystallographic structures place Lys16 in a position of direct interaction with the gamma-phosphate of GTP. Furthermore, lysine is present at corresponding positions in all known sequences of adenylosuccinate synthetase. We suggest that along with a modest role in stabilizing the transition state of the phosphotransfer reaction, Lys16 may stabilize the enzyme structurally. In addition, the modest loss of catalytic activity of the K16Q mutant may confer such a selective disadvantage to E. coli that this seemingly innocuous mutation is not tolerated in nature.

Adenylosuccinate synthetase (AMPSase) 1 (see Ref. 1 for review) catalyzes the following reversible reaction in the presence of Mg 2ϩ ions: GTP ϩ IMP ϩ aspartate i GDP ϩ adenylosuccinate ϩ phosphate (P i ). This reaction is the first committed step in the formation of AMP from IMP on the pathway for de novo purine nucleotide biosynthesis and is an integral part of the purine nucleotide cycle in muscle (2). The reaction mechanism of AMPSase centers on 6-phosphoryl-IMP, formed putatively by the nucleophilic attack of the 6-oxyanion of IMP on the ␥-phosphate of GTP. A second nucleophilic substitution reaction by the amino group of aspartate on the C-6 of 6-phosphoryl-IMP yields adenylosuccinate and P i (3). Two Mg 2ϩ ions are involved in the reaction mechanism (4). One Mg 2ϩ is in the active site, associated with the phosphate moiety of the guanine nucleotide and the N-formyl group of hadacidin, an inactive analog of aspartate (5). However, crystallographic investigations have yet to reveal the location of the second Mg 2ϩ .
On the basis of preliminary crystal structures of ligated AMPSase, which are now complete (5,6), several residues are in positions of putative significance to catalytic function, ligand binding, and/or structural organization of the active site. Asp 13 of AMPSase hydrogen bonds with N-1 of IMP and approaches the sixth coordination site of a pentavalently coordinated Mg 2ϩ . Asp 13 is putatively a catalytic base in the abstraction of the proton from N-1 of IMP (5). Glu 14 hydrogen bonds to the backbone amides 10 and 12 of the P-loop (residues 8 -17 of AMPSase) and to NZ of Lys 16 . Glu 14 may stabilize the conformation of the P-loop, provide electrostatic charge balance in the active site, and/or orient NZ of Lys 16 with respect to the ␥-phosphoryl group. Lys 16 of AMPSase corresponds to the essential lysine of the consensus P-loop sequence GXXXXGK (7,8). Lys 16 interacts with the ␤-phosphate of GDP and/or anions (nitrate and phosphate) bound to the ␥-phosphoryl site (5,6). Lys 16 putatively stabilizes the pentavalent transition state of the ␥-phosphoryl group during the phosphotransfer reaction. His 41 interacts with phosphate groups located at the ␤and ␥-phosphoryl binding sites. His 41 is a putative catalytic acid in the phosphotransfer reaction (5). Finally, Arg 131 , initially considered a candidate for binding aspartate (9), may stabilize the closed (ligand-bound) state of the active site by folding over and perhaps hydrogen bonding with the loop that recognizes the ␤-carboxylate of aspartate.
Reported here are mutations of Asp 13 , Glu 14 , Lys 16 , His 41 , and Arg 131 , which further probe the roles of each residue in the function of AMPSase.

EXPERIMENTAL PROCEDURES
Materials-Escherichia coli strain XL1-Blue came from Stratagene, a site-directed mutagenesis kit from Amersham Corp., and restriction enzymes from Promega. The chemicals used in this study were obtained from Sigma, and pur A Ϫ strain H1238 was a gift from Dr. B. Bachman (Genetic Center, Yale University).
Site-directed Mutagenesis-Recombinant DNA manipulation was performed using standard procedures (10). The mutagenic primers 2 in this study are 5Ј-TTTACCTTCGGCACCCCATT-3Ј (Asp-13 3 Ala), 5Ј-* This research was supported in part by National Institutes of Health Research Grant NS 10546 and National Science Foundation Grants MCB-9218763 and MCB-9316244. This is Iowa Agriculture and Home Economics Experiment Station (Project 3191) Journal Paper J-17206. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Preparation and Kinetics of Wild-type and Mutant AMPSase-Mutant forms of AMPSase were purified by the procedure described previously (12), except for H41N. The H41N mutant has low solubility in 20 mM potassium P i , pH 7. As a consequence, the eluent from the phenyl-Sepharose column was concentrated in 100 mM potassium P i , pH 7, and then purified by DEAE-HPLC using 100 mM potassium P i , pH 7, with a linear salt gradient from 0 to 1 M NaCl. Protein purity was monitored by SDS-polyacrylamide gel electrophoresis according to Laemmli (13). The concentration of purified protein was determined using the extinction coefficient for wild-type AMPSase at 280 nm (⑀ 280 ϭ 67.85 mM Ϫ1 cm Ϫ1 ) where the concentration refers to monomers. AMPSase activity was determined as described earlier (14). 3-20 g/ml enzyme was used in assays, depending on the activity of each mutant. A GBC model 918 UV/Visible spectrophotometer equipped with a Peltier-Effect temperature controller to maintain the temperature at 25°C was used to monitor absorbance changes at 290 nm.
Stability of the K16Q Mutant AMPSase Activity-Two samples of the K16Q mutant (0.54 mg/ml) in 40 mM Hepes buffer, pH 7.7, were incubated at either 25 or 4°C for 2 h. At different times, 5-l aliquots were removed and added to 1 ml of assay solution containing 150 M GTP, 200 M IMP, 5 mM aspartate, and 2 mM MgCl 2 in Hepes, pH 7.7, and the activity was measured at 25°C. The absorbance change at 290 nm was recorded. A parallel experiment was carried out with the wild-type enzyme for the purpose of comparison with the mutant.
In another study, urea (0.5 M) was used to evaluate enzyme stability involving the wild-type and K16Q mutant AMPSases. The enzymes were incubated with 0.5 M urea at 25°C for different periods of time (0, 1, 2, 4, 8, and 16 min) and then added to the assay solution. Initial rates were measured at 280 nm and 25°C in solutions containing 0.5 M urea,  Circular Dichroism Spectroscopy-Circular dichroism spectra were acquired at room temperature on a Jasco spectropolarimeter, model J-710. Samples (50 -300 g/ml in 10 mM potassium P i , pH 7.0) were placed in a 1-mm cuvette, and data points were obtained from 200 to 260 nm in 0.5-nm increments. Spectra were normalized for a direct comparison.

RESULTS AND DISCUSSION
Growth of Transformed E. coli AMPSase-All transformants of the pur A Ϫ cell line (H1238) grew in the LB medium. D13A and E14A transformants, however, grew at a slow rate, comparable to that of the original pur A Ϫ cell line, which must draw its entire supply of adenine from the LB medium.
Purification of Mutant AMPSases-D13A, E14A, K16Q, and R131L mutants migrate comparably on phenyl-Sepharose and DEAE-HPLC columns to wild-type AMPSase. However, during concentration (Amicon concentrator) of the eluent from the phenyl-Sepharose column, the H41N mutant precipitated. The precipitated protein could be redissolved upon dilution or by increasing the concentration of KP i at pH 7 from 20 to 500 mM. Samples, precipitated and then redissolved, showed no loss of activity, ruling out the possibility of irreversible denaturation. All mutant proteins were more than 95% pure, as judged by SDS-polyacrylamide gel electrophoresis.
Circular Dichroism Spectroscopic Study of the Mutants-The CD spectra of D13A, E14A, K16Q, and R131L mutants are almost identical to that of wild-type AMPSase (data not shown). For these mutants, then, no global conformational change occurs as a consequence of mutation. However, the H41N mutant differs significantly from the wild-type protein in its CD spectrum (Fig. 1), indicating a large structural perturbation. The altered structure of the H41N mutant may be responsible for its low solubility in low salt buffer. Based upon x-ray diffraction studies of AMPSase (7), His 41 hydrogen bonds to Asp 21 , an interaction that may stabilize the loop 42-53 in the absence of ligands. Differences in the CD spectra of wild-type and the H41N mutant may stem from conformational differences in this loop structure in the absence of ligands.
D13A Mutant-The D13A mutant shows no activity using the conventional assay, even with 1 mg/ml protein. Considering the sensitivity of this technique (approximately 1 ϫ 10 Ϫ4 A change/min at 290 nm), the activity must be less than 0.001% wild-type AMPSase. However, GTP and IMP quench the intrinsic fluorescence of the mutant, indicating that substrates bind to the D13A mutant. Crystal structures have revealed a hydrogen bond between the side chain of Asp 13 and N-1 of the IMP (5). Asp 13 , then, may abstract the proton from N-1 to generate the 6-oxyanion of IMP, the putative nucleophile in the attack on the ␥-phosphorus atom of GTP (Fig. 2). The complete loss of activity due to the D13A mutation is entirely consistent with an essential catalytic role for Asp 13 .
E14A Mutant-The E14A mutant exhibits greatly reduced activity. The k cat of E14A is too low to be measured with confidence at pH 7.7. Thus, assays were performed at pH 7.0, where the k cat of wild-type AMPSase increases by 40% and the K m values decrease (Table I). At pH 7.0 the E14A mutant had a k cat of 0.022/s (approximately 1% that of the wild-type enzyme at pH 7.0), and the K m values for substrates are 3-6-fold higher than those of the wild-type enzyme ( Table I). Given that Glu 14 makes hydrogen bonds that stabilize the P-loop in E. coli AMPSase (7), the dramatic fall-off in k cat of the E14A mutant may be due to a conformational perturbation on Asp 13 , which, as noted above, is a putative catalytic base. Alternatively, Glu 14 may be essential to the electrostatic charge balance in the active site. In crystal structures, Glu 14 makes a salt link with Lys 16 (see below).
K16Q Mutant-The NZ atom of Lys 16 probably interacts with the ␤and/or ␥-phosphate groups of GTP (5, 6). The consensus P-loop lysine is putatively essential for stabilization of a pentavalent phosphoryl group in the transition state (8) as is well documented in p21 ras (15) and adenylate kinase (16,17). However, for AMPSase, the kinetic parameters of the K16Q mutant (Table I) are similar to those of the wild-type protein.
The corresponding mutant of p21 ras (K16N) drastically reduces the affinity of nucleotides (15), and the corresponding mutant of E. coli adenylate kinase (K13Q) significantly lowers catalytic activity with a modest effect on substrate affinity (16). The possibility that the K16Q mutant of AMPSase may have reverted to the wild-type protein was eliminated by confirming the sequence of the mutant plasmid in the transformed H1238 cell line. Also we detected no endogenous AMPSase in the H1238 E. coli cell line by Western blot analysis (data not shown). Thus, apparently NE2 of glutamine can substitute for NZ of Lys 16 in maintaining hydrogen bonds. Furthermore, for AMPSase, the positive charge of Lys 16 may not be essential for the stabilization of the transition state. In fact, Lys 16 hydrogen bonds to Glu 14 in ligated complexes of AMPSase, resulting in a charge-balanced ion pair. In order for NE2 of the Gln 16 mutant to take up the position of NZ of Lys 16 , OE1 of Gln 16 must hydrogen bond to Glu 14 (Fig. 3). Thus, the observed Lys 16 -Glu 14 ion pair in the wild-type enzyme may be replaced by a neutrally charged Gln 16 -Glu 14 pair. The net electrostatic charge of the mutated and wild-type active site, then, may be the same. Hence we observe little influence on kinetic parameters. Neither adenylate kinase nor p21 ras have a P-loop residue equivalent to Glu 14 .
If the mutation of Lys 16 has only a modest impact on catalysis, why then is position 16 always a lysine in all known sequences of AMPSase? The explanation may rest with the stability of the mutant. At room temperature, the activity of the K16Q mutant is unchanged for 90 min, but at 4°C the activity decreases significantly relative to that of the wild-type protein (Fig. 4). It is known that hydrophobic interactions are weakened at lower temperatures (18). In addition, the stability of AMPSase dimers decreases with temperature (19). It was also observed that when the mutant and wild-type enzymes were exposed to 0.5 M urea for varied periods of time and then assayed for activity, the K16Q enzyme was significantly less stable than its wild-type counterpart (data not shown). Taken together, these results suggest that the K16Q mutant is less stable than the wild-type protein, provided that one accepts enzyme activity as a criterion of stability. If minor structural alterations in the P-loop lessen the stability of AMPSase, as revealed by subunit complementation experiments (19), glutamine may not be permitted at position 16 in E. coli AMPSase due to selective pressures of evolution. However, if one argues that the conditions described in Fig. 4 and with 0.5 M urea may not be applicable in vivo, it becomes difficult to explain the evolutionary preference for lysine over glutamine at position 16.
The conservation of lysine at positions equivalent to 16 in all known sequences of AMPSase may stem alternatively from survival disadvantages associated with a 3-fold reduction in k cat . The wild-type activity of AMPSase in E. coli may barely meet the requirement of the cell for adenine nucleotides. Thus, even a 3-fold reduction in activity of the K16Q mutant relative to the wild-type enzyme could be catastrophic. In an attempt to evaluate the amount of AMPSase in E. coli (TG 1 cells), a quantitative Western blot experiment was employed with antibody against E. coli AMPSase. The signal from a crude extract of TG 1 cells was compared with that of purified AMP-Sase; the amount of AMPSase was approximately 1.5 mg of AMPSase/g of wet TG 1 cells (data not shown). Assuming that the weight percents of total DNA and RNA in E. coli are 1 and 6% (20), respectively, with a quarter of this nucleotide pool in the cell associated with adenine and that the free adenine nucleotide concentration (ATP, ADP, and AMP) is approximately 5 mM, it is possible to calculate the total adenine content of the bacterial cell. With a k cat value of 1.4/s for AMPSase (Table I), the calculated minimum time for a doubling of an E. coli population is approximately 21 min, close to the observed time for population doubling under optimal conditions. Therefore, an E. coli cell containing K16Q AMPSase as the endogenous enzyme should grow at a significantly slower rate than an E. coli cell containing wild-type AMPSase. Such a mutation in a critical enzyme with an extremely low turnover number may not be acceptable for cell survival in a competitive environment.
Based upon the amount of AMPSase in E. coli and its low turnover number, it is tempting to suggest that the AMPSase reaction is the rate-limiting factor in the generation time of the bacterium. Although our calculations seem to support this hypothesis, it is possible that an as yet unrecognized activator of the synthetase may negate this suggestion.
H41N Mutant-The H41N mutant has broadly altered kinetic constants relative to wild-type AMPSase, including a k cat that is approximately 0.01/s (1% of the wild-type enzyme) and K m substrate values increased by 2-6-fold relative to those of the wild-type enzyme ( Table I). The uniform increase in K m values for all substrates implies that the mutation perturbs the entire active site. A hydrogen bond between His 41 and the ␤and/or ␥-phosphoryl groups of GTP is one of three enzymeligand interactions putatively responsible for the 9-Å conformational change in the loop 42-53 (5). The loop 42-53 folds over the guanine nucleotide; the 6-fold increase in K m for GTP may stem from a weakened interaction between Asn 41 and the guanine nucleotide relative to observed His 41 -guanine nucleotide interactions in the wild-type system. Increases in K m for IMP and aspartate to some extent may stem from the decrease in GTP affinity. Synergism in the binding of IMP and GTP is suggested by studies of Wang et al. (21). Furthermore, Mg 2ϩ binds to guanine nucleotides and putatively to the ␣-carboxylate of aspartate (5). Thus, the observed 6-fold increase in the K m for aspartate may be due entirely to the 6-fold increase in the K m for GTP and, presumably, bound Mg 2ϩ . The altered CD spectrum of the H41N mutant may not be relevant to its kinetic properties; the CD spectrum measures the conformation of the unligated mutant, whereas the kinetics probe the ligated mutant. The severely depressed k cat of the H41N mutant is in harmony with crystallographic studies that implicate His 41 as a catalytic acid in the phosphotransfer step. 3 R131L Mutant-The guanidinium group of Arg 131 is close to the loop that binds to the ␤-carboxylate group of aspartate (Fig.  5), and its removal affects the affinity for aspartate (K m , Asp increases 4-fold) without a significant influence on other parameters. Mutation of Arg 303 or Arg 305 to leucine increases K m for aspartate by 100-fold (21); these residues putatively bind directly to the ␤and ␣-carboxylates, respectively, of aspartate. Thus, the 4-fold increase in K m of the R131L mutant is consistent with long range, electrostatic interactions between Arg 131 and aspartate, and the stabilization of the aspartate-bound conformation of the loop 298 -303.
Summary-Mutations at positions 13, 14, 41, and 131 are consistent, as noted above, with their putative roles in catalysis and conformational changes in AMPSase from E. coli. The high activity of K16Q was not anticipated on the basis of recent crystallographic structures of AMPSase, which clearly show Lys 16 making hydrogen bonds with a nitrate anion in the ␥-phosphoryl site (5) and with a P i anion in the ␥-phosphoryl site and the ␤-phosphate group of bound GDP (6). On the basis of crystallographic structures, the role played by Lys 16 appears to be as significant as Arg 305 , His 41 , or Asp 13 , where mutations cause at least a 99% reduction in k cat . Thus, the positive charge on Lys 16 does not play a significant role in stabilizing the transition state. Further mutations at position 16 should reveal whether glutamine is the next best alternative to lysine or whether other substitutions at position 16 result in mutants with significant catalytic capacity.