Mutagenesis and chemical rescue indicate residues involved in beta-aspartyl-AMP formation by Escherichia coli asparagine synthetase B.

Site-directed mutagenesis and kinetic studies have been employed to identify amino acid residues involved in aspartate binding and transition state stabilization during the formation of beta-aspartyl-AMP in the reaction mechanism of Escherichia coli asparagine synthetase B (AS-B). Three conserved amino acids in the segment defined by residues 317-330 appear particularly crucial for enzymatic activity. For example, when Arg-325 is replaced by alanine or lysine, the resulting mutant enzymes possess no detectable asparagine synthetase activity. The catalytic activity of the R325A AS-B mutant can, however, be restored to about 1/6 of that of wild-type AS-B by the addition of guanidinium HCl (GdmHCl). Detailed kinetic analysis of the rescued activity suggests that Arg-325 is involved in stabilization of a pentacovalent intermediate leading to the formation beta-aspartyl-AMP. This rescue experiment is the second example in which the function of a critical arginine residue that has been substituted by mutagenesis is restored by GdmHCl. Mutation of Thr-322 and Thr-323 also produces enzymes with altered kinetic properties, suggesting that these threonines are involved in aspartate binding and/or stabilization of intermediates en route to beta-aspartyl-AMP. These experiments are the first to identify residues outside of the N-terminal glutamine amide transfer domain that have any functional role in asparagine synthesis.

A great deal of interest in asparagine metabolism has resulted from the finding that certain leukemias can be treated by the administration of L-asparaginase (see review, Ref. 1). Experiments suggest that the effectiveness of this protocol is dependent upon decreasing the circulating amount of asparagine (2). Although administration of L-asparaginase is accepted as an essential component of modern therapy, it is fraught with serious side effects and plagued by the appearance of resistant leukemias. An alternative, or adjunct, approach to the use of L-asparaginase might be to lower circulating asparagine by inhibiting asparagine synthetase (AS), 1 the enzyme responsible for its production. Of several hundred compounds that have been evaluated as AS inhibitors, however, none have exhibited sufficient potency and specificity to warrant clinical consideration (3). This failure can be partly explained by the lack of detailed mechanistic information on AS.
Two classes of enzymes catalyzing asparagine synthesis have been described that possess no sequence similarity and may consequently have arisen by convergent evolution. Ammonia-dependent asparagine synthetases in prokaryotes such as Klebsiella aerogenes and Escherichia coli (4 -6) can employ only ammonia as a nitrogen source (Reaction 1). 6 -17141 The second group of asparagine synthetases, on the other hand, is present in both prokaryotes and eukaryotes and employs glutamine as the predominant source of nitrogen in obtaining asparagine from aspartate and ATP (Reaction 2), although ammonia can be employed as an alternative to glutamine (7)(8)(9). In addition, this class of synthetases acts as glutaminases in the absence of aspartate (Reaction 3). E. coli contains two unlinked genes coding for asparagine synthetases (7). Asparagine synthetase A (AS-A), the product of the 990-bp asnA gene for which the complete nucleotide sequence is known, has been isolated and exhibits strictly ammonia-dependent activity (5,6,10). The nucleotide sequence for the 1662-bp asnB gene, encoding asparagine synthetase B (AS-B), has also been cloned and sequenced (11). Based on its primary amino acid sequence, AS-B is a purF (class II) amidotransferase, possessing an N-terminal cysteine residue that is essential for glutamine-dependent activity (12).
Our recent work describing a number of site-specific AS-B mutants has identified specific amino acids in the N-terminal glutamine amide transfer (GAT) domain that are critical to glutamine-dependent nitrogen transfer (13,14). Detailed kinetic analysis of wild-type enzyme and these mutants using alternate substrates and heavy atom isotope effects has also yielded new insights into the mechanistic role of the AS-B GAT domain (15,16). Identification of key residues in the N-terminal region of AS-B was aided by the availability of sequences for the glutamine-utilizing domains of other purF amidotransferases. On the other hand, both BLAST and FASTP analyses indicate that the primary structure of the C-terminal synthetase domain in asparagine synthetases is unique, with the exception of a small region, termed the P-loop-like motif, that is present in a number of ATP-dependent enzymes that release AMP and PP i as reaction products (17). The involvement of this motif in ATP utilization during asparagine synthesis, however, has yet to be validated using site-directed AS mutants.
While raising interesting evolutionary questions, the observation that sequence similarities between AS-A and AS-B do not exist (11) complicates the identification of critical catalytic residues in the C-terminal AS-B synthetase domain. Multiple sequence alignments for several glutamine-dependent asparagine synthetases show 152 conserved residues and 92 positions in which conservative replacements are present. Of the seven regions possessing four, or more, consecutive conserved residues, two of these are in the GAT domain and appear to mediate only nitrogen transfer and glutamine hydrolysis. In this paper, we present the results of random, and site-specific, mutagenesis experiments on two of the other five regions defined by residues 317-330 and 484 -500, respectively ( Fig. 1). While the latter region does not appear to possess a functional role in catalysis and/or substrate binding, we report evidence supporting the hypothesis that Arg-325, Thr-322, and Thr-323 mediate ␤-aspartyl-AMP formation, a key intermediate in asparagine biosynthesis (5,18).

MATERIALS AND METHODS
Restriction and modifying enzymes were purchased from Promega (Madison, WI), Life Technologies, Inc., or New England Biolabs (Beverly, MA). Deoxyadenosine 5Ј-[␣-35 S]thiotriphosphate triethylammonium salt (S p isomer, 1000 Ci/mmol) was purchased from Amersham Corp. All other reagents were the highest possible quality. Oligonucleotide primers were synthesized on an Applied Biosystems 380B DNA synthesizer by the DNA Synthesis Core Facility of the Interdisciplinary Center for Biotechnology Research at the University of Florida. Polymerase chain reactions (PCR) were performed on an Ericomp (San Diego, CA) thermocycler using the GeneAmp DNA Amplification Reagent Kit with AmpliTaq from Perkin-Elmer. Thirty-five cycles consisting of denaturing at 94°C for 1 min, annealing at 54°C for 1 min during megaprimer reactions or 52°C for 1 min during megaproduct reactions, and extension at 72°C for 1 min were followed by a 10-min completion cycle at 72°C. Megaprimers were purified by polyethylene glycol precipitation (0.6 volumes of 20% polyethylene glycol 8000 in 2.5 M NaCl were added to the PCR reaction, incubated at 37°C for 10 min, centrifuged at 10,000 rpm for 10 min, and washed with 2 volumes of 80% EtOH) or agarose gel electrophoresis. After gel electrophoresis, the PCR product was extracted using a Gene CleanII Kit from Bio 101, Inc. (Vista, CA). Double-stranded DNA sequencing was performed using the U. S. Biochemical Corp. Sequenase 2.0 Sequencing kit. Preparation of the template for sequencing was performed by the alkaline lysis method.
Sequence Alignments-The amino acid sequences of known glutaminedependent asparagine synthetases were aligned using a simplification of the progressive alignment method of Feng and Doolittle (21), as implemented in the program PILEUP in the GCG Sequence Analysis Software Package (22).
Construction of Mutants-All directed random mutants were constructed using the PCR megaprimer strategy (23) using template pETB, and oligonucleotide primer pairs SS123 and SS82 or SS124 and SS182 (Table I). Primer SS123 and SS124 contained degenerate oligonucleotide sequences for the construction of the mutants and a silent change inserting a restriction site, which was used for screening. The megaproduct was constructed using the megaprimer and SS65, which was 5Ј of the coding region of the gene. The final product was cut with the appropriate restriction enzymes, gel-purified, and ligated back into pETB. The identity of the mutated insert was verified by sequencing.
Site-directed mutants with single amino acid changes were obtained as follows. A mutant in which Thr-322 was replaced by serine (T322S), produced in the directed random mutagenesis experiments, was used as the template to construct a cloning cassette. This template was chosen because a unique SalI site, 3Ј of the mutagenic area, was created during the original mutagenesis. Megaprimer strategy was again utilized with primers SS196 and SS148 (Table I) creating a unique KpnI site 5Ј of the mutagenic area. The megaproduct reaction utilized the megaprimer and SS224 (Table I). This 534-bp product was digested with PvuII and SalI and cloned into template T322S, thus creating a new template pETB-KS, containing a 126-bp cassette between the KpnI and SalI sites. pETB-KS was then used as the template to construct single random mutants of Thr-323 and Arg-325. SS148, containing the KpnI site and the corrected sequence for Thr-322, along with SS197 (Thr-323) or SS195 (Arg-325) were used to generate a 126-bp product which was cut by SalI and KpnI and cloned into pETB-KS digested with the same enzymes.
Additional site-specific mutations were generated utilizing pETB-KS as a template and SS224 and SS351-SS363 and SS370 (Table I) to  produce T322V, T322S, T322A, T323V, T323S, T323A, R325A, R325K, V321A, D320A, Y319A, T318A, E317A, and E317Q. The 534-bp product was digested with SalI and KpnI and cloned into pETB-KS digested with the same enzymes. The identity of mutated inserts generated by PCR and the adjoining cloning sites was confirmed by sequencing before protein expression.
Screening of Directed Random AS-B Mutants-Several preliminary assays were performed to determine which AS-B mutants warranted further investigation. First, the solubility of the mutant protein was evaluated by SDS-polyacrylamide gel electrophoresis. Second, the soluble cellular fractions containing crude overexpressed AS-B or AS-B mutant enzymes were assayed by measuring the conversion of aspartate to asparagine as monitored by high performance liquid chromatography amino acid analysis on an Applied Biosystems 420A derivatizer and 130A separation system. Standard assay conditions were as follows: 100 mM NH 4 OAc or 10 mM glutamine, 10 mM ATP, 10 mM Laspartate, 17 mM Mg(OAc) 2 , and 50 mM Tris-HCl, pH 7.5. Reactions were initiated with 10 l of crude extract and incubated at 37°C for 15 min before being terminated by the addition of trichloroacetic acid, to a 4% final concentration, and filtered through a 0.2-m syringe filter. All reactions were performed in duplicate. Procedures for the purification and expression of pETB and mutant enzymes have been described elsewhere (13). Protein concentrations were determined with an assay kit supplied by Bio-Rad using ␥-globulin to construct a standard curve.
Kinetic Characterization of Wild-type AS-B and AS-B Mutants-Apparent affinity constants (K m (app) ) for AS-B substrates were determined by incubating purified wild-type or mutant AS-B in reaction mixtures (total volume 160 l) in which all but one of the substrates were saturating, i.e. at approximately 10 times their K m (app) value, unless otherwise noted. The highest aspartate concentration used in these experiments was 100 mM, so AS-B mutants for which the apparent aspartate K m (app) was greater than 10 mM were not saturated by this substrate. All assays contained 100 mM Tris-HCl, pH 8.0, 8 mM MgCl 2 , and the appropriate purified enzyme (3-15 g). A 10-fold variation in substrate, centered about K m (app) , was used to determine the effect of substrate concentration for both wild-type and mutant enzymes. The initial velocity of each reaction was determined spectropho-FIG. 1. Sequence alignment of asparagine synthetases showing the residue conservation in the regions investigated in this study. This alignment was obtained using the program PILEUP, a module in the GCG Sequence Analysis Software Package (22), which uses a simplification of the progressive alignment method of Feng and Doolittle (21). A gap weight of 3.0 was used to obtain these results. Sequence numbers are given for the mature proteins in which cysteine is the N-terminal residue. tometrically by following the production of pyrophosphate during asparagine synthesis (Sigma, Technical Bulletin No. BI-100). Each assay was run two to four times, and the averages are presented. Rate and concentration data were fit to the Michaelis-Menten equation to give K m (app) and V max using the software program Graph Pad Prism (Graph Pad, San Diego, CA).
Glutaminase Assay-The glutaminase activity of the AS-B mutants was assayed by measuring the formation of glutamate using the reaction of glutamate dehydrogenase in the presence of NAD ϩ (24). Reaction mixtures (100 l total volume) contained 100 mM Tris-HCl, pH 8, and 8.0 mM MgCl 2 with varying concentrations of glutamine (0.5-10 mM). Reactions were initiated by the addition of purified wild-type AS-B or mutant enzyme (0.83 g) and were incubated for 18 min before being terminated by the addition of 20 l of 1 N AcOH. The reaction mixture was then added to 380 l of the coupling reagent (300 mM glycine, 250 mM hydrazine, pH 9, 1 mM ADP, 1.6 mM NAD ϩ , and 2.2 units of glutamate dehydrogenase) and incubated for 10 min at room temperature. The solution absorbance was measured at 340 nm, the amount of glutamate present being determined from a standard curve.
Chemical Rescue of Synthetase Activity in the R325A and R325K AS-B Mutants-Chemical rescue experiments were performed by incubating purified R325A (10.8 g), R325K (13.5 g), or wild type AS-B (3.9 g) with 50 mM aspartate, 10 mM ATP, 10 mM glutamine, 15 mM MgCl 2 , 100 mM Tris-HCl, pH 8, and varying concentrations of guanidinium HCl (GdmHCl), methylamine, urea, thiourea, tetramethylguanidine, or methylguanidine (0.5-50 mM). Asparagine synthesis was quantified by measuring the amount of pyrophosphate released in the enzyme-catalyzed reaction. Independent high performance liquid chromatography experiments were employed to ensure that pyrophosphate and asparagine were formed in a 1:1 ratio under these reaction conditions. All assays were performed in triplicate.

Preliminary Functional Evaluation of C-terminal Segments
317-330 and 484 -500 -An initial evaluation of the functional importance of amino acid residues located in regions 317-330 and 484 -500 was carried out using "directed random mutagenesis." Oligonucleotides SS123 and SS124 (Table I), having a calculated average of two mismatches per oligonucleotide, were used to create the two sets of mutations. Twelve independent clones containing mutations in region 317-330,and 17 independent clones for region 484 -500 were evaluated (Table II). Each of these clones was characterized by sequence determination of the inserts and measurement of glutaminase and asparagine synthetase activities in extracts of the expressed protein. Although random mutagenesis of region 484 -500 gave eight double mutants, two triple mutants, and seven single mutants, representing a variety of conservative replacements (T489S, S492T, and F485L) and changes in local charge and potential structural modifications (P486L:T489K, R484D: E500K, T489K:E500K, and R484C:P486A:E500K), all of the associated proteins retained their ability to catalyze asparagine synthesis (Table II). We therefore conclude that residues 484 -500 are not involved in direct catalysis and/or substrate binding.
In contrast, of the six mutant enzymes associated with single mutations in region 317-330, three were unable to synthesize asparagine, and of the six mutants containing multiple alterations in their sequence, none had detectable AS activity ( Table  II). All of these mutants were soluble and retained their ability to catalyze glutamine hydrolysis (data not shown), suggesting that gross conformational changes were not responsible for the concurrent loss of synthetase activity. Three AS-B single mutants retaining synthetase activity (Y319F, T322S, and T328S) were then purified according to standard procedures (13), and kinetic constants for their glutaminase (Table III) and synthe-TABLE I Oligonucleotides used in construction of site-directed mutants of AS-B The codes for oligonucleotides synthesis are as follows: a, 91% A, 3% T, 3% G, 3% C; c, 91% C, 3% T, 3% G, 3% A; g, 91% G, 3% T, 3% C, 3% A; *, 25% A, 25% T, 25% G, 25% C. 5Ј gas acT TaT gaT gTg acc acT aTT cgc gcG TCG ACa ccg aTg TAT TTA 3Ј

Random 2 SS124
5Ј cgc TTC ccg Tac aac acg cca acc agT aaa gAA GCT Tac cTg TaT CGG GAG 3Ј  Mutants with changes in the region defined by residues 317-330 Mutants with changes in the region defined by residues 484-500 (Tables IV and V) activities were determined. The synthetase activity of site-specific mutants was fully characterized only if the mutant AS-B exhibited essentially unaltered kinetic parameters for glutaminase activity, in the presence and absence of ATP, relative to wild-type enzyme. The observation of unchanged ATP-dependent stimulation of glutaminase activity provided additional evidence that the ATP-binding site was likely unchanged by the mutation. In the case of all three AS-B mutants, the kinetic parameters for glutamine hydrolysis were comparable to those of the wild-type enzyme. On the other hand, although three of these site-specific AS-B mutant enzymes had similar specificity constants, measured as k cat /K m , to those of wild-type AS-B for all substrates, a substantial decrease in k cat and K m (app) for ATP was observed in the glutamine synthetase activity of the T322S mutant. Kinetic Characterization of Thr-322 and Thr-323 AS-B Mutants-The T322A and T323A AS-B mutants exhibited diminished synthetase activity but retained their ability to function as glutaminases. A series of additional mutants was therefore constructed and expressed in which Thr-323 was systematically replaced by isoleucine, leucine, serine, or valine. Kinetic characterization of the resulting AS-B mutants showed that, in all cases, K m (app) for aspartate was increased in both glutamine-and ammonia-dependent asparagine synthesis (Tables  IV and V). Structurally conservative replacements were associated with 4 -12-fold increases, whereas K m (app) for aspartate increased more dramatically for the T323A, T323L, and T323I AS-B mutants (20 -50-fold). Similar trends were observed for the turnover numbers of the Thr-323 AS-B mutants, structural conservation decreasing k cat 2-fold, whereas side chain modification to alanine caused a 20-fold reduction in this parameter, irrespective of the nitrogen source. To investigate whether increases in aspartate K m were due primarily to impaired aspartate binding, we compared the ability of cysteinesulfinic acid (CSA) to inhibit the wild-type and mutant enzymes. CSA, an aspartate analog (25), is a competitive inhibitor with respect to aspartic acid in AS-B-catalyzed asparagine synthesis, having a K I value of 1.45 mM, but is not competitive with respect to any of the other substrates (26). Since it was unlikely that the substrate binding order for the AS-B mutants differed from wild-type AS-B, we obtained relative K I values for CSA by plotting the reciprocal of initial velocity versus CSA concentra-tion, ensuring that substrate concentrations were maintained at a constant ratio (2-fold) to the appropriate K m (app) , making the resulting K I values directly comparable. The wild-type AS-B K I (CSA)/K m(app) ratio of about 8 was used as a guideline in determining whether aspartate and CSA binding were similar in the mutants. Surprisingly, the inhibition constant for CSA could not be determined for any Thr-323 mutant enzymes other than T323S (Table VI) because for these proteins the CSA K I exceeded 100 mM, the maximum concentration used in our assays.
Replacement of Thr-322 by alanine, serine, valine, or tyrosine gave AS-B mutants for which k cat was decreased between 5-and 40-fold in either the glutamine-or ammonia-dependent synthetase reaction. K m (app) for aspartate, however, was similar to that of the wild-type enzyme for these mutants (Tables IV  and V) with the exception of T322Y for which it was increased to approximately 40 mM. While CSA binding was affected in an analogous fashion to that observed for aspartate for T322A and T322S, its K I appeared to exceed 100 mM for the T322V AS-B mutant (Table VI).
K m (app) for ATP decreased approximately 3-fold for almost all of the Thr-322 mutant enzymes. We therefore determined the dissociation constant, K d , of ATP with both the T322A and T322S AS-B mutants by measuring the ability of this substrate to protect the mutant enzymes from inactivation by 5Ј-fluorosulfonylbenzoyladenosine (FSBA) 2 (27). After ensuring that the FSBA K I was similar for both of the mutants, the ATP dissociation constants with wild-type AS-B, T322A, and T322V were found to be 18 M, 2 M, and 3 mM, respectively, suggesting that the decrease in K m (app) for ATP underestimated the apparent increased binding affinity of this substrate for the mutant enzymes. We next eliminated the possibility that ATP hydrolysis was becoming uncoupled from synthetase activity in the T322A and T322V AS-B mutants, i.e. that the rate of pyrophosphate formation measured by our standard coupled assay (13) was not equal to the rate of asparagine synthesis. Direct high performance liquid chromatography analysis of product concentrations showed that, in reactions catalyzed by T322A and T322V, the amounts of asparagine and pyrophosphate produced were equal and that the kinetic constants determined using this assay were unchanged from our experiments using the standard assay, within experimental error (data not shown). Kinetic Characterization and Chemical Rescue of Arg-325 AS-B Mutants-Using random mutagenesis methods, Arg-325 was changed to lysine, histidine, isoleucine, leucine, threonine, and glutamine to yield a series of AS-B mutants that exhibited no detectable synthetase activity, but glutaminase activity was unaltered. The catalytic role of this residue was then probed using chemical rescue experiments involving the R325A, and the conservatively altered R325K AS-B mutants, both of which possessed unaltered glutaminase activity (in the presence and absence of ATP) when compared with wild-type AS-B (Table  III). In the presence of saturating substrates, methylamine, urea, thiourea, GdmHCl, methylguanidine, or tetramethylguanidine were added to wild-type AS-B, R325A, and R325K at varying concentrations (0.5-50 mM). The addition of 50 mM GdmHCl to the R325A AS-B mutant restored synthetase activity (150 nmol/min/mg), to a level of about 15% that of wild-type AS-B. Chemical rescue saturated at approximately 50 mM GdmHCl, whether the aspartate concentration was 50 or 200 mM (Fig. 2), with no further activation at concentrations up to 150 mM GdmHCl (data not shown). Plots of initial velocity versus either glutamine or ATP concentration in 50 mM GdmHCl showed normal Michaelis-Menten behavior, and the K m (app) values for ATP and glutamine were similar to those of wildtype AS-B. In contrast, aspartate did not show saturation kinetics, the plot of initial velocity versus aspartate concentration remaining linear up to concentrations of 200 mM regardless of the concentration of GdmHCl in the reaction mixture (5-50 mM) (Fig. 3). Addition of methylguanidine also restored the enzymatic activity of the R325A AS-B mutant, albeit at a much lower rate. To establish whether restoration of R325A activity was a specific effect, we examined the behavior of the AS-B mutant R49A, in which Arg-49 is replaced by alanine. Arg-49 is known to be involved in glutamine recognition and binding, and mutation of this GAT-domain residue severely diminishes glutamine-dependent synthetase activity. 3 No stimulation of the R49A AS-B synthetase activity was detected at any concentration of GdmHCl (data not shown). Kinetic constants were 3 S. K. Boehlein, unpublished observations.  then determined for the glutamine-dependent synthetase reaction catalyzed by R325A in the presence of 50 mM GdmHCl (Table VII). No concentration of any of these exogenous reagents affected glutamine-dependent synthetase activity in either wild-type AS-B or the R325K AS-B mutant.

Alanine Scanning Mutagenesis of Other Conserved Residues in AS-B Residues 317-330 -
To investigate the roles of other conserved amino acids in the region defined by residues 317-330, single-site AS-B mutants were expressed in which these residues were systematically replaced by alanine. After mutagenesis, sequencing, and enzyme purification, kinetic constants for the resulting AS-B mutants were determined (Tables  III-V). Replacement of Glu-317 by alanine resulted in a large change in the kinetic parameters associated with the glutaminase activity of this AS-B mutant. K m (app) for glutamine, in the absence of ATP, was increased almost 5-fold, and ATP stimulation of hydrolysis activity was lost in the E317A AS-B mu-tant. In the case of the T318A AS-B mutant, the kinetic parameters were consistent with altered aspartate binding. Thus, a 10 -12-fold increase in K m (app) for aspartate was seen when either glutamine or ammonia were used as the nitrogen source (see Tables IV and V). The k cat was decreased approximately 3-fold, and the specificity constant k cat /K m (app) was decreased approximately 30 -40-fold. The kinetic constants obtained when varying glutamine, ATP, or ammonia are very similar to that of wild-type AS-B. For T318A, K m (app) for aspartate (6.6 mM) and the apparent CSA K I (90 mM) gave a K I (CSA)/K m(app) (Asp) of 13.6.
Replacement of Tyr-319 by either alanine or phenylalanine caused only minor alterations in the kinetic parameters for both the glutamine-dependent and ammonia-dependent AS reactions (Tables IV and V). The K I (CSA)/K m (app) (Asp) ratio was about 8, indicating that both aspartate and CSA bind similarly to these mutants and also that aspartate binding was not altered by the mutation.
Substitution of alanine for Asp-320 gave the D320A AS-B mutant for which k cat was decreased from 4-to 6-fold in both glutamine-and ammonia-dependent asparagine synthesis. The associated specificity constant (k cat /K m (app) ) was lowered 14 -22-fold since K m (app) for aspartate was increased 4-fold. In contrast to the modest increase in this kinetic parameter, the K I for CSA became greater than 150 mM (Table VI).
Mutation of Val-321 to alanine caused a 4-fold increase in K m (app) for aspartate in the synthetase activity using any nitrogen source. In addition, there was a 4-fold increase in the K I for CSA, consistent with the hypothesis that the change in K m (app) of aspartate reflects perturbed aspartate binding (Table VI). DISCUSSION Although the detailed mechanism of nitrogen transfer from glutamine remains to be defined for class II amidotransferases, especially in the light of recent kinetic isotope effect measurements (15), ATP-dependent aspartate activation has been dem-   was observed for the E317A, E317Q, T322V,  T322Y, T323A, T323I, T323L, and T323V AS-B mutants at concentrations up to 100  onstrated to proceed via the formation of ␤-aspartyl-AMP for a number of asparagine synthetases (Fig. 4) (5, 18). Thus, the ␤-carboxylate anion of aspartate reacts with the ␣-phosphorus of ATP generating a pentacovalent intermediate (I). Release of pyrophosphate then gives the desired intermediate that can undergo attack by nucleophilic nitrogen to give asparagine and AMP (Fig. 4). In forming (I), however, a number of chemical problems must be overcome by the synthetase domain of the enzyme. First, the resonance stabilized carboxylate, which is ordinarily a poor nucleophile, must be activated. Second, the ␣-phosphorus must be made more electrophilic, and the transition state leading to the formation of the pentacovalent intermediate must be stabilized.
No previous studies have clearly identified the amino acids involved in mediating aspartate binding and/or activation in asparagine synthesis. In an effort to identify catalytically important residues in the synthetase domain of AS-B, two regions that were highly conserved in all known asparagine synthetases were chosen for alteration using random mutagenesis methods (Fig. 1). Of these two regions, numerous single and double mutants in residues 484 -500 showed insignificant effects on the specific activity of the enzyme, indicating that this region was not directly involved in either substrate binding or catalysis. In sharp contrast, preliminary mutagenesis experiments revealed that even conservative changes to several residues in region 317-330 gave mutants possessing no detectable synthetase activity without affecting glutaminase activity or the effect of ATP on the glutaminase activity.
The specific functional roles of conserved amino acids in this region were therefore investigated using a combination of alanine scanning and site-directed mutagenesis. Before being subject to detailed kinetic analysis, each AS-B mutant had to meet several criteria demonstrating its overall structural integrity. First, mutant enzymes had to be soluble. Furthermore, in the absence of a crystal structure for AS-B, we utilized the unique characteristics of the two separable reactions catalyzed by all asparagine synthetases; the synthetase activity of site-specific mutants was fully characterized only if the mutant AS-B exhibited essentially unaltered kinetic parameters for glutaminase activity, in the presence and absence of ATP, relative to wild-type enzyme. If these criteria were met, then we assumed that changes in activity due to site-specific replacements arose from local effects. We note that, of all the AS-B mutants constructed in this study, only E317A and E317Q exhibited diminished glutaminase activity, suggesting that mutation of this residue had caused large structural changes in the enzyme. It is also possible that this residue may function in inter-domain communication, linking the synthetase and the GAT domain through hydrogen bonding or salt-bridge interactions. Consequently, its mutation resulted in perturbation of both the glutaminase and synthetase activities. Validation of this hypothesis, however, awaits detailed structural characterization of AS-B.
After extensive kinetic analysis of a large number of sitedirected AS-B mutants, it was clear that several mutations resulted in considerable increases in K m (app) for aspartate in both glutamine-and ammonia-dependent asparagine synthesis. To verify that these increases in the Michaelis constant reflected, at least in part, a decreased ability of the enzyme to bind aspartate, we examined whether CSA binding to each AS-B mutant was similarly reduced. In previous studies, we have shown that CSA is a competitive inhibitor with respect to only aspartate and no other AS-B substrate (26). With the one notable exception of the D320A AS-B mutant, we observed a good correlation between increases in K m (app) for aspartate and the loss of the ability of CSA to inhibit asparagine synthesis (Table VI). This behavior supports the idea that mutations raising K m (app) for aspartate reflect an involvement in aspartate recognition and binding by the cognate residue in the wild-type enzyme. In experiments employing conformationally constrained amino acids, we have demonstrated that in the bound conformation of aspartate, all of the ionizable groups are located on one face of the substrate (26). It is therefore possible that the side chains of Thr-318 and Val-321 are positioned to interact with the hydrophobic face of the bound aspartate. In the case of the D320A AS-B mutant, we observed that while K m (app) for aspartate was only moderately increased, the apparent K I for CSA was greater than 150 mM (Table VI), giving a K I (CSA)/K m(app) (Asp) ratio of approximately 61. Hence, the D320A AS-B mutant appears, in contrast to wild-type AS-B, to be able to distinguish structural differences in the carboxylate and sulfonate groups of aspartate and CSA, respectively.
Having tentatively identified residues participating primarily in aspartate binding (Thr-318 and Val-321), we next sought those that had a catalytic function in the mechanism of ␤-aspartyl-AMP formation. The observations that site-specific mutations of Thr-322 mostly affected the turnover number of the synthetase activity, while changes to Thr-323 caused both a decrease in k cat and an increase in K m (app) for aspartate, were therefore noteworthy. Our results suggest that Thr-323 is most likely involved in aspartate binding, given that changes in both kinetic parameters for aspartate appeared to reflect the size and shape of the amino acid replacing threonine in any given AS-B mutant, whereas Thr-322 is important in catalysis. Hence, no substitution for Thr-322 significantly affected K m (app) for aspartate with the exception of tyrosine. The alterations to both kinetic parameters by the latter, however, were most likely the result of structural perturbations due to the size of the tyrosine side chain. Furthermore, mutation of Thr-322 also caused an increased affinity for ATP in the T322A, T322S, and T322V AS-B mutants relative to wild-type enzyme, as reflected by the K m (app) values for ATP (Tables IV and V). That K d for ATP was decreased in these mutants was confirmed by determining the ability of ATP to protect against FSBA inactivation of these three AS-B mutants (see "Results"). Energy released on ATP binding to wild-type AS-B may therefore result in a conformational change facilitating either aspartate binding or the formation of ␤-aspartyl-AMP. Although our data suggest that some structural conservation of the Thr-322 side chain is required for aspartate binding, the K m (app) for aspar- tate was only increased 10-fold for the T322V AS-B mutant. We determined that the ATPase and synthetase activities of the mutant enzymes were not uncoupled since the amounts of pyrophosphate and asparagine formed were equal in independent assays.
Our mutagenesis experiments also showed that Arg-325 was absolutely critical for both glutamine-and ammonia-dependent asparagine formation. The catalytic role of this residue was therefore probed by investigating chemical rescue of the R325A and the conservatively altered R325K, AS-B mutants. Both of these mutants met our criteria for structural integrity. Interpretation of these experiments was aided by the growing literature on the rescue of catalytic activity by addition of exogenous compounds to inactive or impaired enzyme mutants (28 -34). For example, exogenous amines rescue an aspartate aminotransferase mutant in which a critical lysine residue is replaced by alanine (28). The only direct precedent for our studies came from experiments in which the activity of carboxypeptidase A mutants lacking Arg-127 was rescued by the addition of several guanidine derivatives (29). In these experiments, the addition of the rescuing agent restored k cat , without affecting K m (app) , an observation consistent with the hypothesis that the guanidine side chain of Arg-127 stabilizes the rate-limiting transition state for peptide hydrolysis (35). For the R325A AS-B mutant, 50 mM GdmHCl restored 15% of the synthetase activity of the wild-type enzyme. Rescue of activity was specific in that methylamine or other guanidine and urea derivatives failed to restore activity to any significant level. Although the saturation behavior of the rescue agent may arise from an alternative, less efficient pathway for catalysis (32), it is likely that higher concentrations of GdmHCl unfold AS-B, limiting the amount of activity that can be restored. The observation that the K m (app) values for both glutamine and ATP in the rescued AS-B mutant are unaltered from their values for wild-type AS-B is consistent with the idea that the rescued enzyme employs an identical catalytic mechanism. Furthermore, aspartate saturation could not be reached experimentally in the rescue of R325A. It is unlikely that aspartate binds to R325A as a complex with GdmHCl, since the relevant dissociation constant is approximately 2 M. Our results, therefore, suggest that the exogenous GdmHCl replaces the side chain functionality of Arg-325 by binding into an appropriate pocket in the active site. Precedence for this idea is found in the crystal structure of the D189S trypsin mutant, which can be rescued by exogenous acetate, in which acetate was observed bound to the active site of the mutant in a manner similar to that of the corresponding aspartic acid residue present in wild-type trypsin (36). To ensure that restoration of activity by exogenous GdmHCl was a specific effect, we examined the ability of this reagent to rescue the R49A AS-B mutant. Arg-49 is a conserved residue in the GAT domain and is likely involved in interacting with the carboxylate group of glutamine. 3 No restoration of activity by GdmHCl was observed for the R49A AS-B mutant.
Arg-325 may play two functional roles in AS-B, being involved either in binding aspartate in a catalytically competent manner or in stabilization of the transition state leading to a pentacovalent intermediate (I). Many enzymes employ the guanidino group for substrate binding (37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48)(49), and if this were the function of Arg-325, then our failure to observe saturation of the rescued enzyme by aspartate might result from an insufficient amount of properly oriented guanidinium ion in the active site at concentrations that do not denature the protein. On the other hand, there is ample precedence in the literature for participation of critical arginine residues in stabilizing transition states involving a nucleotide (50 -59). For example, synthesis of aminoacylated tRNAs by aminoacyl tRNA synthetases proceeds via an ␣-aminoacyl-AMP intermediate, which is chemically similar to ␤-aspartyl-AMP. The availability of x-ray crystal data for several class I and class II tRNA synthetases has also established functional roles for numerous residues, including conserved arginines.
Active site residues in methionyl tRNA synthetase (MetRS) that mediate formation of ␣-methionyl-AMP are similar to those identified in the experiments reported here. For example, Arg-322 in MetRS, which is conserved in one subgroup of 6 class I tRNA synthetases, appears critical in the activation of the methionine ␣-carboxylate. When this residue was changed to a glutamine, the resulting R322Q MetRS mutant exhibited a 60,000-fold decrease in k cat /K m (app) for ATP-PP i exchange (53). In addition, a 25-fold increase in the K m (app) for methionine was observed with no change in the K m (app) values for either ATP or tRNA-Met. These results for the R322Q MetRS mutant were consistent with the hypothesis that the mutation reduced the stability of the methionyl-AMP intermediate by elimination of important contacts with methionine in the transition state. In another series of experiments on MetRS, Tyr-258 was replaced by alanine (60). Mutation of this residue led to a 2,000fold decrease in k cat for the ATP-PP i exchange reaction but had no significant effect on the K m (app) values for ATP or methionine. The rate of pyrophosphorolysis for the Y258A MetRS mutant was also decreased upon the addition of PP i without affecting K m (app) for PP i . This observation was interpreted to suggest that Tyr-258 binds the ␣-phosphate of ATP in the transition state, enhancing the rate of methionyl-AMP formation.
The effects of these changes in the active site of MetRS are strikingly similar to those observed for the AS-B mutations studied in our experiments. Thus, K m (app) values for both glutamine and ATP in asparagine synthesis by the rescued R325A mutant were similar to those for wild-type AS-B, and satura- tion behavior for aspartate, the substrate requiring activation, was not observed under our experimental conditions. Arg-325 therefore appears to play a role in AS-B similar to that of the active site arginine in MetRS. Furthermore, the kinetic behavior of the Thr-322 AS-B mutants paralleled those observed for changes to Tyr-358 in MetRS in that K m (app) values for all substrates were unchanged compared with wild-type AS-B, whereas k cat was significantly decreased in both glutamineand ammonia-dependent asparagine synthesis. Hence, Thr-322 might also interact with the ␣-phosphate of ATP in the transition state leading to pentacovalent intermediate (I) and, subsequently, ␤-aspartyl-AMP. Given the similarities in the kinetic behavior of the mutants of these two enzymes, it will be of interest to determine if there are structural relationships between asparagine synthetases and tRNA synthetases.