Kinetic and Stereochemical Comparison of Wild-type and Active-Site K145Q Mutant Enzyme of Bacterial D-Amino Acid Transaminase*

D-Amino acid transaminase (EC 2.6.1.21), from Bacillus sp. YM- 1, a thermostable enzyme with pyridoxal 5’-phosphate as coenzyme and a target for the design of novel antimicrobial agents, catalyzes the reversible transfer of an amino group between D-alanine and CP ketoglutarate to form pyruvate and D-glutamate, respectively. To explore the catalytic role of Lys-145, which binds the coenzyme, a site-specific mutant en- zyme, K145Q (in which Lys-145 had been mutated to glutamine) constructed earlier (Futaki, S., Ueno, H., Martinez del Pozo, A., Pospischil, M. A., Manning, J. M., Ringe, D., Stoddard, B., Tanizawa, K., Yoshimura, T., and Soda, K. (1990) J. Biol. Chem. 265, 22306-22312) was compared to the wild-type enzyme for its kinetic parameters. Initial velocity studies and partial reaction isotope exchange experiments showed that the low activity of the mutant enzyme (about 1.5% the activity of the wild-type enzyme with saturating sub- strates) is an intrinsic property, confirming that con-taminating enzymes do not account for the low activity of the K145Q mutant exchanged, only 1-2% of the L-isomer of alanine was found.

ll To whom correspondence should be addressed. netic results on the K145Q mutant enzyme together with the findings on the relative racemization rates and the NMR protein exchange data suggest that an alternate base catalyzes abstraction of the Q proton of substrate in this mutant D-amino acid transaminase. D-Amino acid transaminase from bacteria is a target enzyme for the development of novel antimicrobial agents because it catalyzes the synthesis of D-glutamate and D-alanine, important constituents in the bacterial cell wall (1). Studies on its mechanism of action are important in order to accomplish that goal. Like L-amino acid transaminases (2), it employs pyridoxal 5"phosphate (PLP)' as coenzyme (3,4). Reactions catalyzed by L-amino acid-specific enzymes proceed via a Ping-Pong kinetic mechanism and consist of two halfreactions, each of which is comprised of three major steps. A similar mechanism is proposed for D-amino acid transaminase in Scheme I, which also outlines some mechanistic questions posed in this study. In the first step, referred to as transaldimination, the coenzyme in an internal aldimine structure with the t-NH2 group of Lys-145 in the wild-type enzyme, forms a Schiff s base with an amino acid substrate to form an external aldimine with the concomitant release of the e-amino group of Lys-145. The second step, which is the critical 1,3prototropic shift involving abstraction of a proton from the CY carbon of the amino acid and donation of a proton to the aldehydic carbon of the coenzyme, may proceed via a quinonoid intermediate (5) to yield a ketimine product. The final step for the first half-reaction is the hydrolysis of the ketimine and release of the keto acid. The second half-reaction is the reversal of these steps but with a different a-keto acid and the formation of its corresponding amino acid. In L-aspartate transaminase Lys-258 has the dual function of binding the coenzyme and abstracting the a proton of substrate (2). Whether it performs other functions in this enzyme is not known.
Elucidation of the role(s) of the coenzyme-linked Lys-145 of D-amino acid transaminase has been a focus of our studies on this enzyme. Hence, we have constructed the active-site K145Q mutant enzyme by site-specific mutagenesis (6). With this attenuated mutant enzyme, it was possible to study some intermediates in the reaction pathway in a conventional spectrophotometer, e.g. the formation of the external Schiff base with D-alanine, which occurred over the period of several minutes and its subsequent transformation to the ketimine, which took place over a period of hours. These results suggested that some amino acid side chain other than Lys-145, which was absent in the mutant enzyme, was responsible for these slow transformations. In the present study, we compare the kinetic details of the reaction pathway of the wild-type and mutant enzymes in an effort to understand the basis for these observations.
The possibility that the low activity of a mutant enzyme is due to contamination by the wild-type enzyme was rigorously excluded in order to be certain of the conclusions reached through studies on this attenuated enzyme. Another way to distinguish between the intrinsic catalytic activities of the wild-type and mutant enzyme is the use of steady-state kinetic methods. Thus, we have conducted a detailed comparative investigation of the kinetic mechanism of both wild-type and mutant enzymes. We employ initial velocity studies to determine the kinetic parameters for the substrates in the forward and reverse reactions, and partial reaction isotope exchange techniques on D-alanine/pyruvate and D-glutamatela-ketoglutarate, the two substrate-product pairs, to determine the binding constants for substrates. We have also explored the possible presence of dead-end complexes in the kinetic mechanism of D-amino acid transaminase. Based on these results, a minimal kinetic mechanism for this enzyme has been constructed, and minimal values have been assigned to the individual rate constants.
The ability of a transaminase to catalyze an exchange of the Ca proton of the amino acid substrate has been shown in L-alanine transaminase (7,8). Those results were confirmed and extended by Oshima and Tamiya (9) and by Cooper (10). Julin et al. (11) reported a 99% exchange of the Ca proton of L-aspartate with water catalyzed by Lys-258 of L-aspartate transaminase. A similar exchange might be expected for Damino acid transaminase. As shown in Scheme I, all the steps along the reaction pathway are freely reversible. Thus, the catalytic base that abstracts the proton from D-alanine could exchange it with solvent and reprotonate the quinonoid to reform alanine. If carried out in DzO, the resulting product would be [~u-~H]alanine. Ca proton-solvent exchange studies employing both D-and L-alanine with wild-type as well as K145Q mutant enzymes and the degree of racemization have been studied in order to elucidate the role(s) of Lys-145 in Damino acid transaminase in these reactions. The racemization could result either by a donation of a proton by the same base (12) or by a second base as in serine hydroxymethyltransferase (13). Racemization of amino acids by L-aspartate transaminase has been reported by Kochhar and Christen (14), where protonation from the opposite side is carried out by a water molecule (15). In this communication, we report the results of such proton NMR exchange studies as well as the kinetic constants and other side reactions catalyzed by both the wildtype and the K145Q mutant D-amino acid transaminase in order to understand more fully the role of Lys-145; some of these side reactions have been described for other PLP enzymes (16).
Purification of Wild-type and Mutant D-Amino Acid Transaminase-Wild-type and the active-site mutant enzyme K145Q of Damino acid transaminase were purified by a minor modification of methods described earlier (6,17,18). Since the coenzyme PLP is not covalently linked to the protein in the mutant enzyme as it is in the wild-type enzyme, caution was taken in the heat treatment step of the mutant enzyme purification to ensure that there was no inactivation. Therefore, the 53 "C temperature used for heat treatment in the purification of the wild-type enzyme was lowered to 48 "C for preparation of the K145Q mutant enzyme. In addition, a second DEAE-Sepharose column was employed. The protein was applied to this column (1.5 X 30 cm), previously equilibrated in 50 mM potassium phosphate, pH 7.6,0.2 mM EDTA, 50 p~ PLP, and 0.01% 2-mercaptoethanol, and eluted with a linear gradient of 0.08-0.13 M KC1 (1.5 liters). The final step introduced in the purification procedure of the mutant enzyme involved a Blue Sepharose affinity chromatography column (1 X 35 cm) previously equilibrated with this buffer without KC1 to remove traces of NADH oxidase and D-alanine oxidase, which interfere in the kinetic assays. After application of the protein, the column was washed with 100 ml of the same buffer; NADH oxidase, which is not detected by SDS-PAGE, adheres to this column, and the transaminase does not. Fractions containing enzyme activity were pooled and concentrated in a Amicon centricon to 10-20 mg/ml and stored in 30-60-~1 aliquots at -80 "C. SDS-polyacrylamide gel electrophoresis showed that the purity of the mutant and wild-type enzymes was >99%.
General Procedures-All steady-state kinetic experiments were conducted most times in triplicate in 50 mM bis-Tris-HC1, pH 7.2, at 25 "C. All buffers and substrate solutions used in the experiments were filtered through 0.2-pm UNIFLO filter units and stored at -20 "C. Solutions of radiolabeled substrates, pyruvate, and u-ketoglutarate were freshly prepared before each experiment. Data analysis was conducted by method of Grubmeyer et al. (19) and as described in Bhatia et al. (20). Data-fitting programs HYPER, COMP, NON-COMP, SEQUEN, and PING PONG are from Cleland (21). Experiments to determine KM values for substrates, and K, values for deadend complexes were conducted on the overall forward and reverse reactions as described by Equation 1 as follows.
D-Alanine + ketoglutarate = pyruvate + D-glutamate (1) The terms forward and reverse reactions employed in this work refer to the overall reactions as described in Equation 1. Partial reaction isotope exchanges were conducted on each substrate-product pair, namely D-alanine/pyruvate and D-glutamate/a-ketoglutarate (Scheme 11). The theory and methodology for partial reaction isotope exchanges are well established and have been described by Cleland (22), Purich and Allison (231,and Boyer (24).
Spectroscopic Enzyme Assay-D-amino acid transaminase was assayed employing a method that couples the pyruvate produced from D-alanine to the NADH-dependent lactate dehydrogenase-catalyzed reaction, as described previously (25). Assays were performed on a Varian Cary 2200 spectrophotometer or on a Zeiss PMQII instrument both with cuvettes thermostatted at 25 "C. For the assay, a 1-ml reaction mixture contained 5 mM a-ketoglutarate, 200 mM D-alanine, 100 PM NADH and 0.4 mg/ml lactate dehydrogenase in 50 mM bis-Tris, pH 7.2, at 25 "C. The reaction was initiated by addition of enzyme. One unit of enzyme activity is defined as the amount of enzyme necessary to convert 1 pmol of NADH to NAD per min at 25 "C, which corresponds to the reduction of an equivalent amount of pyruvate produced from D-alanine, as described previously (25).
Initial Velocity Studies-Previous studies on the mutant D-amino acid transaminase indicated low amounts of activity that could be detected by the standard spectroscopic assay (6). Furthermore, with D-amino acid substrates alone there were complete spectral transformations of the enzyme, i.e. from E-PLP to E-PMP, consistent with intrinsic activity of the mutant enzyme. In the present work we have developed a sensitive assay that employs radiolabeled substrates to provide a convenient method for the determination of kinetic parameters and to permit a study of the reverse reaction catalyzed by Damino acid transaminase.
k3 SCHEME 11. Minima1 kinetic mechanism for wild-type and mutant D-amino acid transaminase.
was maintained at 5 times its KM value. To determine the kinetic parameters for the forward reaction, a 100-pl reaction mixture contained D-alanine, [a-5-"CC]ketoglutarate (200,000 cpm) in 50 mM bis-Tris-HC1, pH 7.2 at 25 "C. D-[l-"CC]Alanine was also employed as the radiolabeled substrate (200,000 cpm) in some experiments. Mutant or wild-type enzyme (0.2-10 pg) was added to initiate the reactions and after appropriate times (3-10 min) the reaction was terminated by the addition of trichloroacetic acid to a final concentration of 10%. The reaction mixture was applied to a AG50W-X2 cation exchange column (Bio-Rad; 0.4 X 4 cm). The keto acid was eluted from the column with water, and the amino acid was separately eluted with 2 N HCI; each eluant (200 p l ) was collected into a scintillation vial. The samples, which were neutralized with 10 N NaOH before addition of 5 ml of Ready Safe (Beckman), were counted for 2 min in a LKB Model 1218 liquid scintillation counter. For the reverse reaction, a 100-pl reaction mixture contained D-glUtamate and [l-14CC]pyruvate (200,000 cpm) in 50 mM bis-Tris-HCI, pH 7.2, at 25 "C.
To investigate possible dead-end complexes in the kinetic mechanism of D-aminO acid transaminase, one substrate was maintained below its Kr value, and the second was varied between 0.5 and 10 times its K, value. For the E-PMP. Ala dead-end complex, a 100-111 reaction mixture contained 0.01-0.70 M D-alanine, [a-B-"C]ketoglutarate (200,000 cprn), and a-ketoglutarate (1 p~ for the wild-type, 100 p~ for the K145Q mutant enzyme). For the E-PMP.Glu experiments, a 100-pl reaction mixture contained 0.5-7 mM D-ghtamate, [l-"C]pyruvate (200,000 cprn), and pyruvate (1 mM with the wildtype enzyme and 50 mM with the K145Q enzyme). For the E-PLP .
Pyr dead-end complex experiments, a 100-pl reaction mixture contained 5-50 mM pyruvate, [l-'*C]pyruvate (200,000 cpm), and Dglutamate (100 pM with the wild-type enzyme). The spectroscopic assay method was employed for the determination of the K, value for the E-PLP . Kg complex in wild-type D-amino acid transaminase.  , and D-glutamate in 50 mM bis-Tris-HCI, pH 7.2, were used. For both exchange reactions, substrates were eluted from the column and analyzed as described above for the initial velocity experiments. Racemization of Amino Acids-A 0.5-1111 reaction mixture containing either D-or L-alanine and wild-type or K145Q mutant D-aminO acid transaminase (0.65 mg) in 50 mM bis-Tris-HC1, pH 7.2, was incubated at 25 "C. Samples (100 pl) were applied to Centricon 10 and centrifuged at 5000 X g for 20 min. Aliquots of the filtrate were removed from the lower chamber, derivatized with Marfey's reagent (26), and analyzed by HPLC to determine the amount of each enantiomer, as described by Martinez del Pozo et al. (27).
NMR Investigation of fH]Alanine Solvent Exchange-Reaction mixtures (1 ml) containing 0.11 mg of wild-type or 1.2 mg of K145Q mutant D-amino acid transaminase were incubated with 100 mM Dor L-alanine in 100 mM phosphate buffer in D,O at 25 "C, pH = 7.6. NMR spectra were collected at 25 "C on a Nicolet 300-MHz instrument. A 5-ps pulse (30" pulse angle) with 1-s delay was employed, and spectra were collected (64 transients) over a range of 550-1450 Hz with 8192 data points. Data were exponentially apodized (line broadening = 0.4 Hz). To quantitate the conversion or exchange of the Cor proton of D-or L-alanine, the areas under the peaks were integrated. Chemical shifts, which were assigned by comparison with standard CHCI, (7.3 ppm), of 1.54 and 3.84 ppm represent the methyl protons and the Ca proton, respectively.

RESULTS
Initial Velocity Studies-The enzyme preparations used in the kinetic experiments show no detectable D-alanine oxidase (<0.01%) or NADH oxidase activity (<0.01%) and contained 1 molecule of PLP per enzyme subunit (6). Whereas the maximal rate for the overall forward reaction ( VFOR) in wildtype D-amino acid transaminase was 258 units/mg (Table I), the rate of the reverse reaction ( VREV) was 30-fold slower (8.6 units/mg). This result was rather unusual and suggested a possible slow off-rate for one of the products of the reverse reaction (D-alanine or a-ketoglutarate). The KM value for Dalanine was found to be high (48 mM, Table I); large KM values for amino acid substrates for transaminases are not unusual, e.g. Hooper and Segal (28) determined a KM of 34 mM for L-alanine with L-alanine transaminase from rat liver and the same enzyme from pig heart has a K@ of 28 mM for L-alanine (29) and that from human liver has a K F of 20.8 mM (30). The KM for D-glutamate with D-amino acid transaminase was 1.2 mM indicating the large binding contribution of the second carboxyl group. Based on a subunit molecular weight of 32,000 (3,4), second-order rate constants of 2.7 X lo3 M" s -~ (kCat/K@) and 3.6 X lo3 M" s" (kc.,/@") can be calculated. Though these values are much slower than the limit set by diffusion, they are in the same range as that for other transaminases, such as L-aspartate transaminase ( kcat/ For the K145Q mutant enzyme, VFOR was 3.6 units/mg and VREV was 45-fold lower (0.08 units/mg). Thus, for the K145Q mutant enzyme, the activity ( VFOR) was about 70-fold lower than that for the wild-type protein, and the KM values for Dalanine and D-glUtamate were 525 and 8 mM, signifying about a 10-fold increase compared to the wild-type enzyme ( Table   I). The KM values for the keto acids, a-ketoglutarate and pyruvate, were 65 and 8.4 mM, about a 100-fold increase over the wild-type enzyme. The 1.5% activity of the K145Q mutant enzyme compared with the wild-type enzyme is higher than the value reported previously because the current studies were performed at saturating concentrations of substrates.
We determined that although wild-type D-amino acid trans- indicate that the enzyme-substrate interactions for binding are the same, and that the major difference is in kcat. Thus, the decreased ability of L-alanine to act as substrate is not due to inferior binding compared to D-alanine but rather that its a proton is apparently not close enough to Lys-145. The K145Q mutant enzyme accepts L-alanine = 570 mM, kc.JPkA'" = 1.6 M-' s-') as well as D-alanine as substrates with nearly equal efficiency, i.e. similar kcat/KM values. The kcat value is the same for L-alanine and D-alanine.
Partial Reaction Isotope Exchanges-The ability of an enzyme to catalyze the exchange of radiolabel between each substrate-product pair in the absence of the other substrateproduct pair is an important criteria to establish a Ping-Pong kinetic mechanism (32,33). Partial reaction isotope exchange experiments were performed with both substrate-product pairs D-alanine/pyruvate and D-ghtamate/a-ketoglutarate in an Attenuated Transaminase (Scheme 11). The partial reaction isotope exchange experiments allow for the establishment of Ping-Pong type kinetics for the enzyme and the determination of the dissociation constants for the enzyme-substrate complexes in wild-type and K145Q D-amino acid transaminase. The KI values obtained by this method represent true dissociation constants (32, 34). These exchanges, which were readily detected, confirm a Ping-Pong type kinetic mechanism for wild-type and mutant D-amino acid transaminase. Neither exchange occurred in the absence of the enzyme nor when the enzyme was denatured by boiling in 1% SDS. The rates for the exchanges are summarized in Table I. For the wild-type and mutant enzyme, the rate of the D-alanine/pyruvate exchange was similar to the forward reaction rate (VFOR), whereas the D-glutamate/cY-ketoglutarate exchange rate was 30to 50-fold slower. This result is consistent with Equation 2 (35) (2) where VFoR and VREV represent the maximal rates of the forward and reverse reactions and Vexl and Vexa represent the two exchange rates. Thus, the observation of a considerably slow reverse reaction rate and D-glutarate/a-ketoglutarate exchange rate in both the wild-type and mutant enzymes is not unusual. These results indicate that one of the steps that comprises the a-ketoglutarate/D-glutamate partial reaction is the slow, rate-determining step. The dissociation constants for substrates with wild-type D-amino acid transaminase and mutant are summarized in Table I. In the K145Q mutant enzyme the dissociation constants for the amino acids D-alanine and D-glutamate are about 11-fold higher than for the wild-type enzyme. The KI value for a-ketoglutarate is 153-fold higher and that for pyruvate 95-fold higher than the wild-type enzyme. These results are indicative of a difference in the active-site geometry of wild-type and K145Q mutant D-amino acid transaminase. Although the attenuated K145Q mutant enzyme is significantly less efficient than the wildtype enzyme, it remains a competent enzyme that proceeds via Ping-Pong kinetic mechanism in the same manner as its wild-type counterpart. Dead-end Complexes-The formation of dead-end complexes between the enzyme and one substrate is characteristic of enzymes that proceed via Ping-Pong kinetics. For example, in L-aspartate transaminase, the E-PLP. a-ketoglutarate and E-PMP.aspartate dead-end complexes are formed (36). Initial velocity experiments have been employed to investigate the possible formation of dead-end complexes in the kinetic mechanism for wild-type and mutant D-amino acid transaminases. In these experiments one substrate is maintained at a low concentration (<< KI), while the concentration of the second is varied from 0.5 to 10 times KI. A decrease in the maximal overall reaction rate at high concentration of the varied substrate would suggest the formation of an enzymesubstrate dead-end complex (32). For wild-type enzyme a t low concentrations of D-alanine (in relation to its K I ) and high concentration of Kg (relation to its KI), there was inhibition in the maximal reaction rate (VFOR), indicating the formation of a possible E-PLP .Kg dead-end complex. The KZ value for that complex was determined by the method of Segel (32) to be 3.4 mM. Under our experimental conditions, the formation of E-PLP.Pyr (KI > 50 mM), E-PMP.Ala (KI > 0.7 M ) , or E-PMP. Glu ( K I > 7 mM) dead-end complexes was not detected since there was no inhibition of the maximal reaction rate by high concentrations (relative to KI) of substrate form-ing the complex. In the case of the mutant enzyme no deadend complexes were detected, although limits of the binding constants of such complexes can be assigned, i.e. 20 mM, 700, 660, and 7 mM for the E-PLP. Kg, E-PMP. Ala, E-PLP. Pyr, and E-PMP. Glu complexes, respectively (see "Materials and Methods" for details).
Analysis of Kinetic Mechanism-Based on the steady state kinetic parameters, minimal values for the rates of individual steps of the D-amino acid transaminase-catalyzed reaction were calculated (Table 11). From VFOR, a kcat of 129 s-' (based on a molecular weight of 32,000 per subunit) was calculated for the wild-type enzyme. The corresponding kcat for the K145Q enzyme was 1.8 s-', about 70-fold lower than that for the wild-type enzyme. For the wild-type enzyme, taking kcat/ KAla --2.7 X lo3 M-' s" as the minimal on-rate for D-alanine ( kl, see Scheme 11) and the Kz value ( ko,$k,,) obtained from the D-alanine/pyruvate isotope exchange (Table I), the offrate ( L 1 ) for D-alanine from E-PLP was calculated to be 199 s" (Table 11). The kcat/K@ for D-alanine for the K145Q mutant enzyme was calculated to be 3.4 M" s-', which is about 800-fold lower than that for the wild-type. This rather large effect on kcat/KM may reflect the inability of the mutant enzyme to stabilize the transition state. From the on-rate of D-alanine (kl, see Table 11) and KI of 780 mM, an off-rate of 2.7 s" was calculated. Based on a molecular weight of 32,000, D-alanine/pyruvate isotope exchange rates of 356 (wild-type) and 3.8 (K145Q) units/mg yield rate constants of 178 and 1.9 s-'. Thus, these rate constants can be assigned to the central complex interconversion (k2 and L z ) , and to the off-rate of pyruvate (k3) from E-PMP (Scheme 11). Since in both wildtype and K145Q mutant enzyme the VALapyr exchange rate is faster than the kcat value, neither central complex interconversion (k2 and L z ) nor product off-rates ( k l and k3) are rate-limiting in either enzyme. As described earlier, the onrate for pyruvate (kT3) can be calculated from kCat/flx to be 2.4 x lo4 and 3.0 M" s-l for the wild-type and mutant enzymes, respectively (Table 11). From the kcat/K2 values of 1.6 X lo5 M" s-l (wild-type) and 28 M" sK1 and K F values of 15 @M and 2.3 mM for wild-type and mutant enzyme, off-rates (k-J of 2.4 and 0.006 s" were calculated. For both enzymes, this off-rate is rather slow compared to kcat. However, they reflect the slow reverse reaction rates and slow a-ketoglutarate/D-glutamate isotope exchange rate in both enzymes. Thus, the reverse reaction in both enzymes is limited by a slow offrate of Kg (Scheme 11). The forward reaction may at least be partially limited by an off-rate for D-glutamate. Since net catalysis proceeds with kcat values of 129 s" (wild-type) and 18 s-' (K145Q), these values represent minimal rate constants

TABLE I1
Summary of minimal rate constants in the reaction mechanism of wild-type and mutant D-amino acid transaminases The designations for the rate constants are shown in Scheme 11.

Kinetic constant
Wild-type Reduced Stereoselectivity in an Attenuated Transaminase for kg, k+, and kc (Glu off-rate). The on-rate for D-glutamate ( k -6 ) was calculated from I@'" determined from the D-glutamate/a-ketoglutarate partial reaction isotope exchange experiments. Thus, while the reaction rates are slower and substrate binding constants are higher in the K145Q mutant enzyme, the kinetic mechanism is similar to the wild-type enzyme, i.e. Ping-Pong type with a slow reverse reaction rate resulting from a slow Kg off-rate.
Exchange of fHJAlanine with Solvent D20-Whereas the kinetic experiments described above were conducted under steady state conditions with both substrates present, the C a proton exchange studies were performed with D-alanine alone employing relatively long incubation times (in hours) and higher enzyme concentrations. The exchange of the Ca proton of L-and D-alanine with solvent D20 catalyzed by wild-type and mutant K145Q enzyme was monitored by NMR by observing the loss of the Ca quadruplet at 3.84 ppm. As shown in Fig. 1, the wild-type enzyme catalyzes a relatively fast exchange of the Ca proton of D-alanine with solvent ( k = 0.46 h-'). The wild-type enzyme also catalyzes a very slow exchange of the Ca proton of L-alanine ( k = 0.034 h-l). With a 10-fold increase in the concentration of the mutant enzyme, there was a slow exchange of the Ca proton of L-alanine ( k = 0.045 h-') and D-alanine ( k = 0.034 h-l) (Fig. 2 ) . These constants were calculated taking into account the difference in the concentrations of wild-type and K145Q enzymes. Thus, the Ca proton exchange rates are similar for both L-and Dalanine with the mutant enzyme and close to the value for the exchange rate of the a proton of L-alanine by the wildtype enzyme. This signal, which appears as a doublet at 1.54 ppm, becomes a singlet upon exchange of the Ca proton with D20. In control experiments, there was no decrease in this signal indicating no exchange of fi protons. Furthermore, no exchange of the Ca proton was found in the absence of enzyme.
Racemization of Alanine-Donation of a proton to the opposite side of the Ca carbon of the quinonoid complex (Scheme I) would lead to an amino acid of opposite stereochemistry, i.e. racemization. Compared to the Ca protonsolvent exchange, racemization catalyzed by wild-type and mutant D-amino acid transaminase was found to be lower than the proton exchange. Thus, the rates of racemization of  L-and D-alanine catalyzed by the wild-type enzyme were nearly equivalent and approximately 50-fold slower than the corresponding Ca proton exchange rates, i.e. over a period of 10 h during which 50% of Ca proton of D-alanine had exchanged, only 1-2% of the L-isomer of alanine was found. The K145Q mutant enzyme also catalyzed the racemization of L-and D-alanine at rates 50-times slower than the ca proton exchange rates. Thus, after a 14-h period when 50% of the Ca protons of D-alanine had been exchanged with solvent, the amount of racemization of L-alanine formed was also about 1%. To eliminate the possibility that the racemization was due to bacterial growth, experiments were conducted under sterile conditions after sterile filtration in a UNIFLO 0.2-Fm microfiltration unit in a biological safety cabinet. After completion of the experiment, reaction samples were plated on LB agar media (37). No bacterial growth was observed on the plates when incubated at 37 "C for 2 days.

DISCUSSION
The results in this study demonstrate that although the K145Q mutant D-amino acid transaminase has low catalytic efficiency, i.e. it is an attenuated enzyme whose kinetic profile proceeds via a Ping-Pong mechanism like the wild-type enzyme (Scheme 11). The mutation of Lys-145 leads to an enzyme in which there are changes in active-site geometry as indicated by the increases in binding constants and a loss of stereochemical integrity. Unusual for both wild-type and mutant enzymes were the findings that the reverse reaction rate ( V R E V ) was 30-fold slower than the forward reaction rate ( V F O R ) . However, detailed partial reaction isotope exchange studies revealed a slow a-ketoglutarate/D-glutamate exchange rate and a fast D-alaninelpyruvate exchange rate. The slow reverse reaction rate and Kg/Glu isotope exchange rate were attributed to a slow off-rate for a-ketoglutarate as reflected in its KI (15 pM). Similar studies on the K145Q mutant enzyme reveal that it has a slow Kg off-rate in the reverse reaction.
The possibility that the observed activity in the mutant enzyme was due to contamination by wild-type or endogenous enzymes has received consideration from several perspectives, and the results are completely consistent with the initial observation (6). Since the binding constants obtained for the K145Q mutant D-amino acid transaminase in this communi-in an Attenuated Transaminase cation are strikingly different from those of the wild-type enzyme, the activity observed in the kinetic experiments with the K145Q mutant enzyme must be due to its inherent activity. Furthermore, the possibility that any putative contaminating enzymes would catalyze the partial isotope exchanges, which fit Equation 2, is remote.
The attenuated enzymatic activity of the K145Q enzyme and its ability to catalyze the exchange of the Ca proton of Land D-alanine with solvent indicate that another amino acid side chain but not Lys-145, which is absent in the mutant enzyme, catalyzes the proton abstraction (1,3 prototropic shift) from the amino acid substrate in the mutant enzyme. The low enzymatic activity of the mutant enzyme could arise from an improper anchoring of the coenzyme due to the absence of the internal aldimine with Lys-145, similar to the Y70F mutant of Escherichia coli L-aspartate aminotransferase where a conservative mutation of Tyr-70, which interacts with the coenzyme, to phenylalanine leads to an 85% loss in kcat (38)(39)(40). The mutation of Lys-145 to glutamine results in about a 10-fold increase in the KM of the amino acids and a 100-fold increase in the KM for the a-keto acids. Similar increases in the KI values were also observed. These observations suggest that Lys-145 may also be important in ground state substrate binding. More significantly, the kCat/KM ratio, which reflects transition-state binding, decreases lo4and 105-fold for the amino acids and keto acids, respectively, in the K145Q mutant enzyme. Thus, the low activity in the K145Q mutant enzyme may reflect its inability to stabilize effectively a transition state in the K145Q mutant enzyme reaction mechanism.
Our results indicate that it is likely that Lys-145 is the basic residue that abstracts the Ca proton (1,3 prototropic shift) in wild-type enzyme with optimum efficiency. However, the activity of the K145Q mutant enzyme must be due to another residue, which has been designated BASE I1 in Scheme I. This view is supported by several results presented in this work, including the rates of proton exchange with solvent for both D-and L-alanine. Thus, L-or D-alanine may bind to the enzyme in a similar way as indicated by similar KM values and the 1,3 prototropic shift step with L-alanine as substrate could be carried out by BASE I1 at a slow rate (Scheme I). In the absence of Lys-145 in the K145Q mutant enzyme, the substrate (or the coenzyme-substrate complex) has greater flexibility in movement and a second base (BASE 11) may catalyze the 1,3 prototropic shift from either L-or Dalanine and could abstract the Ca proton from either L-or Dalanine. A single base mechanism involving a quinonoid complex that changes its orientation has been proposed for alanine racemase from Bacillus subtilis and is referred to as the swinging door mechanism (12).
The proton exchange results are also consistent with the second base responsible for abstraction of the a proton of Lalanine for the wild-type enzyme and the a proton of D-or Lalanine for the mutant enzyme (Scheme I). With the wildtype enzyme, the external aldimine between coenzyme and Dalanine permits the most efficient proton abstraction by Lys-145, thus the relatively high exchange rate. Binding of Lalanine allows for an abstraction of the proton by BASE I1 and thus the low exchange rate. In the mutant K145Q enzyme, the external aldimine between coenzyme and alanine cannot interact with the absent Lys-145. Hence, BASE I1 abstracts the proton from L-as well as D-alanine at the same slow exchange rate for both isomers. Therefore, Lys-145 must play an important role in conferring stereochemical integrity to Damino acid transaminase. However, it is not known at present if BASE I1 functions in the wild-type enzyme.
Although our studies strongly suggest that a second catalytic base is present in the active-site of D-amino acid transaminase, its nature remains to be confirmed. It could conceivably be a water molecule. We have recently shown on the basis of chemical modification studies that Lys-267 of Damino acid transaminase is labeled by D-serine (18). We are currently investigating whether "BASE 11" is this side chain. Having reliable kinetic constants for this enzyme, we can now employ more sophisticated kinetic methods (36) to further investigate the chemistry of D-amino acid transaminase.