Involvement of Ser-451 and Ser-452 in the catalysis of human gamma-glutamyl transpeptidase.

The serine residue required for catalysis of γ-glutamyl transpeptidase was identified by site-specific mutagenesis of the conserved serine residues on the basis of sequence alignment of the light subunit of human, rat, pig and two bacterial enzymes. Recombinant human γ-glutamyl transpeptidases with replacements of these serine residues by Ala were expressed using a baculovirus-insect cell system. Substitutions of Ala at Ser-385, −413 or −425 yielded almost fully active enzymes. However, substitutions of Ala at Ser-451 or −452 yielded enzymes that were only about 1% as active as the wild-type enzyme. Further, their double mutant is only 0.002% as active as the wild type. Kinetic analysis of transpeptidation using glycylglycine as acceptor indicates that the Vmax values of Ser-451 and −452 mutants are substantially decreased (to about 3% of the wild type); however, their Km values for L-γ-glutamyl-p-nitroanilide as donor were only increased about 5 fold compared to that of the wild type. The double mutation of Ser-451 and −452 further decreased the Vmax value to only about 0.005% of the wild type, while this mutation produced only a minor effect (2-fold increase) on the Km value for the donor. The kinetic values for the hydrolysis reaction of L-γ-glutamyl-p-nitroanilide in the mutants followed similar trends to those for transpeptidation. The rates of inactivation of Ser-451, −452 and their double mutant enzymes by acivicin, a potent inhibitor, were less than 1% that of the wild-type enzyme. The Ki value of the double mutant for L-serine as a competitive inhibitor of the γ-glutamyl group is only 9 fold increased over that of the wild type, whereas the Ki for the serine-borate complex, which acts as an inhibitory transition-state analog, was more than 1,000 times higher than for the wild-type enzyme. These results suggest that both Ser-451 and −452 are located at the position able to interact with the γ-glutamyl group and participate in catalysis, probably as nucleophiles or through stabilization of the transition state.

important role in glutathione metabolism. It catalyzes the transfer reaction of a ␥-glutamyl moiety from glutathione and related compounds to a variety of amino acids and dipeptides. The transfer of the ␥-glutamyl moiety to water leads to hydrolysis (1,2,3). Both the large and the small subunit of ␥-glutamyl transpeptidase are encoded by a common messenger RNA (4,5,6). The enzyme is translated as a single chain precursor, which yields subunits post-translationally by proteolytic processing (7,8,9).
A catalytic nucleophile, such as is found in thiol-or serineclass proteases, is assumed to form a covalent linkage with a ␥-glutamyl group because reactions catalyzed by ␥-glutamyl transpeptidase are thought to proceed via a ␥-glutamyl-enzyme intermediate (10). Although ␥-glutamyl transpeptidase from mammalian species possesses a unique thiol on the light subunit, which has the catalytic domain (11,12), this cysteine residue is not required for catalysis (13). On the other hand, several studies suggest that a serine residue is involved in catalysis (14,15). This is also supported by the recent observation that modification of ␥-glutamyl transpeptidase with Nacetylimidazole led to a stabilized intermediate, and permitted the detection of a ␥-glutamyl enzyme in which the ␥-glutamyl moiety was bound on the light subunit (16). The nature of the linkage between enzyme and ␥-glutamyl group was found to be consistent with an ester. Thus, the nucleophile in the active site was proposed to be a hydroxyl group, probably a serine residue on the light subunit. However, since the stabilized ␥-glutamyl enzyme is hydrolyzed upon denaturation by guanidinium ions, the catalytic residue has not yet been identified. The detailed mechanism of the catalysis of ␥-glutamyl transpeptidase is still unclear because its crystal structure is unknown. Several residues, however, have been identified to be at or near the active site (13,16,17,19,20).
Acivicin (L-(␣S,5S)-␣-amino-3-chloro-4,5-dihydro-5-isoxazoleacetic acid), a potent inhibitor of ␥-glutamyl transpeptidase, inactivates the enzyme by its covalent attachment in or near the active site (2,21,22). By use of isotopically labeled acivicin, amino acid residues to which the agent bound have been identified. These are Thr-523 in rat enzyme (17), Ser-405 in pig and Ser-406 (equivalent to Ser-405 in rat and pig) in human (20), all of which are on the light subunit. However, the human mutant ␥-glutamyl transpeptidase in which these corresponding residues were replaced by Ala was found to be almost fully active, and the inactivation by acivicin was as rapidly for the mutant as for the wild-type enzyme; thus, these residues are not essential for catalysis (20). These findings suggested that another hydroxyl group reacts with acivicin initially, followed by transfer of acivicin moiety to the aforementioned residues. Such a primarily reactive hydroxyl group is likely to be identical to the serine residue involved in formation of ␥-glutamyl enzyme intermediate because of structural similarity between the ␥-glutamyl moiety and acivicin, and because of the requirement of a nucleophile for both the ␥-glutamyl substrate and acivicin to react with the enzyme.
In the current study, we selected the conserved serine residues as candidates for mutagenesis by using amino acid sequence alignment of the light subunits of human, rat, pig and two bacterial ␥-glutamyl transpeptidases. These candidates were examined by site-specific amino acid substitutions in order to identify the hydroxyl group(s) responsible for inactivation by acivicin and for catalysis.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases and DNA modifying enzymes were obtained from New England Biolabs. Oligonucleotide primers were synthesized by Integrated DNA Technologies, Inc. L-␥-Glutamylp-nitroanilide, glycylglycine and other common reagents were purchased from Sigma.
Preparation of Recombinant Baculoviruses-The purified transfer plasmids and BaculoGold DNA (PharMingen) were subjected to cotransfection to obtain recombinant viruses, as described previously (13,20,25). Transfection was performed with Lipofectin (Life Technologies, Inc.) (26). The generated recombinant viruses were amplified to more than 5 ϫ 10 7 plaque forming units per ml prior to use for expression experiments. General manipulation of recombinant viruses was carried out as described (27).
Enzyme Activity Assay-The standard assay for ␥-glutamyl transpeptidase activity was performed at 37°C using l mM L-␥-glutamyl-p-nitroanilide as a donor substrate and 20 mM glycylglycine as an acceptor in 0.1 M Tris-HCl buffer (pH 8.0) (2). One unit of activity is defined as the quantity of enzyme that releases l mol of p-nitroaniline per min.
Kinetic Analysis-Enzymatic activity for transpeptidation was assayed at 37°C using 0.125 -2 mM L-␥-glutamyl-p-nitroanilide as a donor substrate in 0.1 M Tris-HCl buffer (pH 8.0) with 20 mM glycylglycine as an acceptor in order to obtain the kinetic parameters for the donor substrate. To determine the parameters for the acceptor, a fixed concentration, l mM, of L-␥-glutamyl-p-nitroanilide as a donor was used and the acceptor (glycylglycine) concentration was varied from 2.5 to 20 mM (or from 0.125 to 4 mM for some mutant enzymes). Kinetic parameters for hydrolysis activity were assessed using L-␥-glutamyl-p-nitroanilide (2.86 -320 M) as the substrate in the absence of an acceptor. Since the donor substrate also can serve as an acceptor at the relatively high concentration (autotranspeptidation), and this reaction proceeds faster than hydrolysis, the autotranspeptidation apparently appears as a substrate-activation, which gives a down-curvature in the high concentration range on the double-reciprocal plot (13,36). Therefore, the parameters for hydrolysis were calculated using the data of the range in which a substrate-activation kinetics did not emerge. Release of p-nitroaniline was monitored at 410 nm using a Cary 210 spectrophotometer (Varian). In some studies on the double mutant whose activity was too low to permit continuous monitoring, the release of p-nitroaniline was assessed by end point assay. The reaction rates were determined after subtracting the contribution of the spontaneous hydrolysis of the substrate without enzyme. Kinetic parameters were calculated using nonlinear regression analysis based on the Marquardt algorithm.
Inhibition of ␥-Glutamyl Transpeptidases by Serine and Serine-Borate Complex-Inhibition of enzyme activities by serine and serineborate complex were examined under the conditions described above with a variable concentration of L-␥-glutamyl-p-nitroanilide in the presence of 20 mM glycylglycine. In the L-serine alone experiments, serine concentration was varied between 4 and 16 mM; higher concentrations of L-serine (10 -40 mM) were used for some mutant enzymes. In the serine-borate experiments, L-serine (20 -80 M) was added to the assay mixture containing 10 mM sodium borate, which was saturating concentration for the wild-type enzyme (15). The concentration of L-serine in serine-borate complex inhibition was increased to 40 mM for some relatively resistant mutants.
Inactivation of ␥-Glutamyl Transpeptidases by Acivicin-Enzymes were incubated at 25°C with 0.25 -4 mM acivicin in 0.1 M Tris-HCl buffer (pH 8.0). At several intervals, the activity remaining was determined in a standard assay. Inactivation rates were much slower for some mutants and acivicin concentrations up to 20 mM were used. Since plots of acivicin concentrations versus inactivation rates did not show saturation in the concentration range studied, the inactivation kinetics were treated as a bimolecular reaction.
Protein Determination-Protein was determined by the bicinchoninic acid method using bovine serum albumin as a standard (30).

Amino Acid Sequence Alignment of the Light Subunits-
Amino acid sequences of the light subunit of human (6, 31), rat (5, 6), pig (32), E. coli (33) and Pseudomonas (34) ␥-glutamyl transpeptidases were aligned with the Clustal method (35). The alignment of these five species is shown in Fig. 1. The serine residues at positions 385, 451, and 452 of the human enzyme were ideally conserved. In addition, the residues at positions 413 and 425 in human were found in three other species at the corresponding positions; in Pseudomonas these serine residues are replaced semi-conservatively by threonine. Because of the common structure as a hydroxyl group, these two serine residues were also included as candidates to be investigated.
Specific Activity of the Purified Mutant Enzymes-The mutant enzymes were successfully expressed in insect cells infected with their recombinant viruses, and were produced in sufficient quantity to purify. All purified enzymes exhibited only two bands of M r 44,000 and 24,000, corresponding to a large and small subunit on the sodium dodecyl sulfate-polyacrylamide gel electrophoresis followed by silver-staining, as shown in our previous study (25). Only replacement of Ser-451 or -452 led to great reduction of activity, to about l% of that of the wild-type enzyme, while the other mutations did not significantly decrease activity (Table I). The results show that only Ser-451 and -452, among residues examined, play an important role in enzyme activity, and that others are not required for catalysis. Thus, the double mutation of Ser-451 and -452 (S451A/S452A) was constructed. The double mutant enzyme is dramatically less active; its activity is only 0.002% of that of the wild type (under the standard assay condition).
Kinetic Properties of Mutant Enzymes-In order to characterize the effects of the mutations of Ser-451 and -452, their kinetic parameters were compared with those of the wild-type enzyme. The kinetic parameters for L-␥-glutamyl-p-nitroanilide as a donor in transpeptidation were obtained in the presence of 20 mM of glycylglycine as an acceptor. Alternatively, parameters for the acceptor were determined with l mM L-␥glutamyl-p-nitroanilide as the donor. Mutations of Ser-385, -413, or -425 had little effect on the kinetic parameters for transpeptidation or hydrolysis (Table II and III).
The K m values of Ser-451 and -452 mutant ␥-glutamyl transpeptidases for the donor substrate were only about 4 and 2-times higher than that of the wild-type enzyme, respectively. However, their V max values were 40 times lower (Table II). Furthermore, their double mutation resulted in significant decrease of V max value to about 0.005% of that of the wild type, but the K m for the donor increased only modestly. Similarly, large decreases in V max values, without significant alteration of K m , were also observed in hydrolysis of L-␥-glutamyl-p-nitroanilide by these serine-substituted mutants (Table II).
The decreased activity in the Ser-451 and -452 mutants is not likely to be due to an inability to use glycylglycine as an acceptor substrate because the presence of 20 mM glycylglycine substantially enhanced the V max values of both of these mutants (Table II). Thus, the decrease in activity (transpeptidation and hydrolysis) appears to be due to hindrance in chemical step involving the donor substrate, L-␥-glutamyl-p-nitroanilide. Since the substrate K m values are not significantly increased in the mutants, the amino acid substitutions at Ser-451 and -452 do not seem to affect steady-state binding of the substrates, which suggests that these mutations do not cause a gross conformational alteration. These results show that both Ser-451 and -452 are important in catalysis.
The Their slower rates of formation of the enzyme species (␥-glutamyl enzyme) necessary for the binding of the second substrate (glycylglycine) would cause a relatively low concentration of the ␥-glutamyl enzyme species, leading to lower K m values for the second substrate. These considerations are consistent with impairment of step from the enzyme-first substrate complex to the ␥-glutamyl enzyme in a ping-pong mechanism proposed for the enzyme (10,36).

Inactivation of the Mutant Enzymes by
Acivicin-A nucleophilic residue on the enzyme is thought to be required for formation of the covalent adduct with acivicin ( Fig. 2-B) that leads to inactivation of the enzyme. In order to evaluate which of the serine residues was necessary for interaction with acivicin, inactivation rates of the mutant enzymes were compared with that of the wild-type enzyme. Table IV shows inactivation rates of the mutants as compared to the rates for the wild-type enzyme. Replacements of Ser-385, -413 and -425 had no effect on inactivation rates by acivicin. On the other hand, the inactivation rate was greatly decreased (Ͻ1%) by the sole substitution of either Ser -451 or -452. The second-order rate constants of inactivation of these mutants correlated with their catalytic efficiency for the donor (k cat /K mdonor ϳ V max / K mdonor ), which gives an apparent second-order rate constant in the reaction of the enzyme with the substrate. Virtually, no detectable inactivation was observed for the double mutant. These results show that Ser-451 and -452 are necessary for acivicin inactivation.
Inhibition of the Mutant Enzymes by L-Serine and Serine-Borate Complex-L-Serine is a competitive inhibitor of the L-␥-glutamyl portion of the donor substrate (10), and it is based on the identical structure of L-serine at ␣and ␤-carbon atoms as those of ␥-glutamyl moiety (Fig. 2-C). Because of the structural similarity between the inhibitor and substrate, L-serine may be used to probe the environment of the active site. In addition, L-serine forms a complex with borate by a reversible covalent bond (15). This complex is an inhibitory transition state analog, where the borate portion mimics the tetrahedral form of the ␥-carbonyl carbon. The inhibition of the enzyme by the serine-borate complex involves interaction with a hydroxyl group of the enzyme (Fig. 2-D) (15). Therefore, the hydroxyl  The inhibition constants of the wild-type and mutant enzymes by L-serine or serine-borate complex are given in Table  V. All enzymes examined were inhibited in a competitive manner. The mutant enzymes in which Ser-385, 413 or 425 were replaced by Ala exhibited the same K i values for L-serine and for serine-borate complex as for the wild-type enzyme. In the presence of saturating borate (10 mM), the K i value of the wild type for L-serine decreased to 0.4% of that of L-serine alone. This decrease in K i is attributed to the formation of a reversible linkage between the borate portion of the serine-borate complex and the hydroxyl group(s) of the enzyme. As expected from their K m values for L-␥-glutamyl-p-nitroanilide in the kinetic study, K i values of Ser-451 and -452 mutants for L-serine in the absence of borate were only 5.5 and 3.8 fold higher than the wild type, respectively.
Nevertheless, the usual decrease in K i by addition of borate to the Ser-451, -452 and the double mutants was not as large as for the wild-type enzyme. Thus, substitution at Ser-451 and at Ser-452 greatly abolished (about 300 and 8 fold, respectively) inhibition by the serine-borate complex. Even at the concentration of L-serine higher than that of the borate, these inhibitions depended on the serine concentration. The borate appears to react with the complex of the enzyme and the serine because the borate needs the hydroxyl groups disposed appropriately to form the linkages. Therefore, Ser-451 and -452 residues appear to be necessary to form the linkage between the enzyme and borate. Moreover, inhibition of the double mutant appears to depend only on the serine portion of the serine-borate complex because the K i values of the double mutant for serine-borate complex is much greater (Ͼ1,000 fold) than that of the wild type, while the K i value of the double mutant for L-serine (without borate) is similar to the K i value of the mutant for the serine-borate complex. This is in contrast to the wild-type and other active mutant enzymes. Additionally, the K i of the double mutant for L-serine alone was relatively higher than expected from its K m values for the ␥-glutamyl substrate, compared with the other mutants. This may be due to the lack of a hydrogen    bond between the hydroxyl groups of L-serine as an inhibitor and the two serine residues, Ser-451 and -452, of the enzyme. These results suggest that both Ser-451 and -452 are accessible to the ␥-carbonyl carbon of the ␥-glutamyl substrate.

DISCUSSION
The studies reported here indicate that Ser-451 or -452, or both, are essential for catalysis of ␥-glutamyl transpeptidase, and that neither of these participate in the binding of the ␥-glutamyl substrate. The results also suggest that these serine residues are positioned in the active site to allow interaction with the ␥-carbonyl carbon of the ␥-glutamyl moiety of the substrate. In addition to their spatial arrangement, the serine residues are nucleophilic so as to give rise to the nucleophilic substitution at the imide moiety of acivicin in the covalentadduct formation. It also suggests that these residues would be primarily responsible for interaction with acivicin prior to the transfer to Ser-406 as proposed in previous studies (20). This seems analogous to the intramolecular transfer of the acyl group as found in fatty acid synthase (42). These findings are consistent with the suggestion that both Ser-451 and -452 are catalytic nucleophiles involved in formation of ␥-glutamyl enzyme intermediate. However, it is also possible that these residues may stabilize the transition state in a manner similar to that of the oxyanion hole found in serine proteases (37).
The rate of hydrolysis of L-␥-glutamyl-p-nitroanilide even by the double mutant, although virtually inactive, is still 2,000 times faster than without enzyme; the k cat of the double mutant was 1.7 ϫ 10 Ϫ4 s Ϫ1 , while non-enzymatic hydrolysis was 9.5 ϫ 10 Ϫ8 s Ϫ1 . This enhancement of the reaction rate may be achieved by a reaction pathway that is different from the normal mechanism. In such a pathway, a catalytic nucleophile may not be necessary. This possibility is similar to that for a subtilisin (a serine protease) mutant, where the catalytic serine was replaced by alanine. The mutant enzyme facilitated the hydrolysis of the acyl-p-nitroanilide substrate 3,000 times faster than in the absence of enzyme (38). This alternative mechanism of subtilisin for hydrolysis of the substrate without the catalytic serine is based on a detailed structure of the active site.
The motif, Pro-Leu-Ser 451 -Ser 452 -Met, is one of the most conservative regions in the complete sequences of ␥-glutamyl transpeptidase from a variety of species. Further, the Ser-452 residue in the human enzyme is the only serine residue conserved among all ␥-glutamyl transpeptidases for which the primary sequences are known and the two related enzymes, human ␥-glutamyl transpeptidase-related enzyme (39) and Pseudomonas cephalosporin acylase (40). The later two are distinct from ␥-glutamyl transpeptidase but have significant sequence homology. These comparisons suggest that these serine residues are of critical importance to enzyme function. Our data obtained using site-directed mutagenesis are in accord with this consideration in terms of sequence homology. Re-placement of Ser-452 by threonine resulted in almost the same catalytic properties as substitution by alanine (data not shown). These results show that the hydroxyl residue at position 452 is restricted to only a serine for enzyme function.
In general, the common sequence motif, Gly-Xaa-Ser-Xaa-Gly (Ala), is conserved at the region including the catalytic serine among serine-class hydrolases (41). Nevertheless, this motif was not found after alignment of the whole sequences of a family of ␥-glutamyl transpeptidases. If the serine residues identified here actually act as catalytic nucleophiles in the mechanism of ␥-glutamyl transpeptidase, the enzyme may utilize either nucleophile to attack the carbonyl moiety of the substrate, and might be distinct from ordinary serine-class hydrolases.
Our current studies identify two serine residues required for catalysis, Ser-451 and -452 in human ␥-glutamyl transpeptidase. These are tentatively proposed as catalytic nucleophiles or alternatively the stabilizing residues of the transition state (Fig. 3). Further structural studies are still needed to ascertain the functions of these serine residues in catalysis. Our findings, however, provide valuable information on the active site chemistry of ␥-glutamyl transpeptidase and its related enzymes.