Inactivation of Bacterial D-Amino Acid Transaminases by the Olefinic Amino Acid D-Vinylglycine

D-Vinylglycine (2-amino-3-butenoate) functions as a transamination substrate and irreversible inactivator of the homogeneous pyridoxal phosphate-dependent D-amino acid transaminases from Bacillus subtilis and Bacillus sphaericus. In the absence of alpha-ketoglutarate as co-substrate, vinyl-glycine causes little if any inactivation of either enzyme; in the presence of excess alpha-ketoglutarate, both enzymes are inactivated with pseudo-first order kinetics. The limiting rate constant for inactivation of the B. sphaericus enzyme is 1.9 min-1, for the B. subilis enzyme it is 0.36 min-1. The number of catalytic events before inactivation is about 450 for the B. sphaericus enzyme and about 800 for the B. subtilis enzyme; that is, about 0.2% inactivation in each catalytic cycle for the former enzyme and 0.15% for the latter. Comparisons are made with the L-aspartate amino-transferase from pig heart which is inactivated completely in one catalytic cycle and the L-alanine aminotransferase which is not inactivated in many cycles. Comparisons are also made between the likely mode of D-transaminase inactivation produced by vinylglycine and the mode of inactivation induced by beta-chloro-D-alanine.


o-Vinylglycine
(Z-amino-3-butenoate) functions as a transamination substrate and irreversible inactivator of the homogeneous pyridoxal phosphate-dependent n-amino acid transaminases from Bacillus subtilis and Bacillus sphaericus. In the absence of cy-ketoglutarate as co-substrate, vinylglycine causes little if any inactivation of either enzyme; in the presence of excess a-ketoglutarate, both enzymes are inactivated with pseudo-first order kinetics. The limiting rate constant for inactivation of the B. sphaericus enzyme is 1.9 mix', for the B. subtilis enzyme it is 0.36 min-'. The number of catalytic events before inactivation is about 450 for the B. sphaericus enzyme and about 800 for the B. subtilis enzyme; that is, about 0.2% inactivation in each catalytic cycle for the former enzyme and 0.15% for the latter. Comparisons are made with the L-aspartate aminotransferase from pig heart which is inactivated completely in one catalytic cycle and the L-alanine aminotransferase which is not inactivated in many cycles. Comparisons are also made between the likely mode of D-transaminase inactivation produced by vinylglycine and the mode of inactivation induced by fl-chloro-n-alanine. D-isomers of amino acids are important constituents of bacterial metabolism.
The D-enantiomers of alanine and glutamate, for example, are found in the peptidoglycan layer of bacterial cell walls, and a variety of D-amino acids are components of peptide antibiotics.
One key enzyme is the bacterial alanine racemase, which functions physiologically to provide D-alanine for cell wall biosynthesis. This enzyme is a target of the antibiotic cycloserine (1) and of the P-halo amino acids fluoroalanine' (2) and chloroalanine (3,4). Another potentially important enzyme in provision of precursors for cell wall biosynthesis is D-amino acid transaminase, which could be a significant source of D-glutamate. This transaminase has also been found to be susceptible to chloroalanine. 2 We report here that the olefinic amino acid D-vinylglycine (2-amino-3-buten- in a total volume of 2 ml at 22". At time zero, 30 /*g of Damino acid transaminase were added. At each interval two O.l-ml aliquots were removed.
One aliquot was added to citrate buffer, pH 2.2, and a portion of this sample was applied to the amino acid analyzer.
At this pH the n-amino acid transaminase is rapidly inactivated; this aliquot measures the production of glutamate in the reaction mixture from vinylglycine and n-ketoglutarate.
To the second O.l-ml aliquot 3 ~1 of n-alanine (0.75 M) were added and incubated at 22" for 5 min. The increase in ["Clglutamate was determined as indicated above.
The difference in the amounts of ['%]glutamate in the two aliquots represents the activity remaining in the reaction mixture. The number of catalytic events before inactivation of the B. sphaericLts enzyme was determined by incubation of the enzyme in 1 ml of 2 rnM m-vinylglycine and 1 mM a-ketoglutarate (1000 dpm/ nmolf in 50 mM potassium pyrophosphate buffer, pH 8.5, at 25". The lOO-~1 aliquots were withdrawn at intervals and injected onto l-ml Dowex 50-H+ (400 mesh) columns.
Each column  is consistent with inactivation proceeding only from a preformed E -I complex.
The slope of each line in Fig. 2  This expectation was realized for the transaminase from B. subtilis (Fig. 3) as well as that from B. sphaericus (Fig. 4).
For the enzyme from the former source, Fig. 3  -It is possible that the inactivation of the enzymes could occur by attack on enzyme nucleophiles not at the active site by the 2-keto-3-butenoate released into solution. To test this possibility, incubations ofB. sphaericus enzyme, m-vinylglycine, and cY-ketoglutarate were performed in 5 mM dithiothreitol, which has been found to protect L-hydroxyacid oxidase from inactivation by nL-vinylglycolate (2-hydroxy-3-butenoate) (11). Although this concentration of dithiothreitol for a IO-min incubation had no effect on enzyme stability or the rate of transamination of n-alanine and cy-ketoglutarate, the rate of inactivation increased considerably. In experiments similar to those of Fig. 4, 25% fewer turnovers were observed from incubations containing dithiothreitol than those of controls. Overnight dialysis at 0" against 1 mM dithiothreitol did not have a similar effect. The possibility that high concentrations of thiol cause a different enzyme nucleophile to be attacked will be examined during labeling studies.

Effects
of Vinylglycine on L-Aspartate Aminotransferase and L-Alanine Aminotransferase -The first report of an enzymatic inactivation by vinylglycine was Rando's recent report of its effect on the L-aspartate aminotransferase from pig heart (6). We have been able to confirm that inactivation: 43% activity loss in a 15min preincubation at 25" with 25 mM vinylglycine; the loss of activity in 15 min at 1 mM vinylglytine was essentially negligible. But, in marked contrast to the bacterial n-amino acid transaminases reported here, there is no stimulation by a-keto acids on the rate of inactivation by vinylglycine either at pH 7.5 or at pH 8.5. This lack of stimulation by keto acids suggest that vinylglycine inactivates pig heart L-aspartate aminotransferase without significant turnover. Direct testing of this idea with nL-vinylglycine, CXket.o['*CJglutarate, and 2 mg ofcommercial L-aspartate aminotransferase for 30 min revealed no more than 400 cpm of [14C]glutamate over a 200 to 250 cpm background, corresponding at most to 0.1 to 0.2 nmol of vinylglycine oxidized/nmol of enzyme.
We have previously reported that the other major transaminase in pig heart muscle, r.-alanine aminotransferase, is not detectably inactivated by m-vinylglycine (7). We have confirmed this insensitivity even in the presence of excess a-keto acids. One question that arises is whether L-alanine aminotransferase is inert because it cannot oxidize vinylglycine to the putative inactivator, 2-imino-3-butenoate.
When vinylglytine turnover was monitored by experiments analogous to those of Fig. 4 A number of different possible pathways for the reaction of vinylglycine with the n-amino acid transaminases are delineated in Scheme 1. Two of these mechanisms lead to enzyme inactivation by covalent alkylation (pathways 2 and 4) while the other two (pathways 1 and 3) lead to catalytic turnover and cY-keto acid production: 2-keto-3-butenoate by pathway 1 and 2ketobutyrate by pathway 3. It is also clear from Scheme 1 that these four routes are grouped pairwise from two different product complexes. If the ES complex is converted by route A to the normal transaminase imine complex, then the conjugated imine can either be released or undergo conjugate addition by an enzyme nucleophile at carbon 4. On the other hand, if the ES complex, after abstraction of the a-H as a proton to form the stabilized cY-carbanion, can undergo a 1,3-prototropic shift, the product in route B will be an enzyme-bound eneamine. Hydrolysis of the eneamine, indicated in pathway 3, would yield the 2-ketobutyrate and pyridoxal-P form of the transaminase. This is in contrast to the keto acid production in pathway 1 where the enzyme would be left in the two-electron reduced pyridoxamine P-form. If the product. eneamine complex is captured instead by an enzyme nucleophile, inactivation would again ensue. But pathway 4 is distinct from pathway 2 in that pathway 4 would predict covalent bond formation at carbon 3 of the inactivator. The situation is further complicated by the possibility that the bound eneamine and bound imine might well be in equilibrium, as is shown in Scheme 1.
Mechanism B3 has been proposed to explain the production of 2-ketobutyrate from L-vinylglycine by the following pyridoxal-P-dependent enzymes: sheep liver threonine deaminase (121, Escherichia coli tryptophan synthetase (13), rat liverycystathionase,Y and Salmonella typhimurium cystathionine ysynthetase.4 However, we do not see any catalytic production of keto acid in the absence of added cr-ketoglutarate, the required co-substrate for transamination.
This result rules out possibility B3 as a significant path to product formation.
Vinylglycine is a transaminase substrate when a-ketoglutarate is present as measured by the conversion of radioactive cY-ketoglutarate to radioactive n-glutamate. This result suggests pathway Al is responsible for vinylglycine turnover and implies that the product is 2-keto-3-butenoate.
If path Al is responsible for catalytic turnover, then we believe it more likely that inactivation occurs by way of pathway A2 rather than B4. However, despite the fact that we have no evidence that the n-specific transaminases can isomerize vinylglycine to an eneamine product, it is possible that branch B could occur occasionally, and could be responsible for inactivation. This issue must be left in doubt until the identification of the enzyme-inactivator linkage is elucidated, indicating which carbon is attached to the enzyme residue.
The production of 2-keto-3-butenoate during catalytic turn-Inactivation of D-Amino Acid Transaminases by Vinylglycine 1575 over leaves the enzyme molecules in the pyridoxamine-P form, unable to undergo further reaction with another molecule of vinylglycine, or any other n-amino acid, in the absence of added keto acid. These enzyme molecules are thus protected from inactivation that would proceed during many catalytic cycles, thus the insensitivity of the n-transaminase to vinylglycine in the absence of keto acid. In the presence of (Yketoglutarate, the pyridoxamine-P enzyme molecules are converted back to pyridoxal-P forms, again competent to react with vinylglycine. Further inactivation ensues, at the rate of 0.22% inactivation in any given catalytic cycle for Bacillus sphaericus enzyme or 0.15% inactivation per catalytic event with the Bacillus subtilis enzyme.
These low partitioning ratios between inactivation and normal catalytic transamination contrast with the susceptibility of pig heart L-aspartate aminotransferase to L-vinylglycine. Bando did not report whether detectable turnover accompanied inactivation of that enzyme (61, but we have now tested this point and found none. The L-aspartate transaminase must be nearly 100% inactivated in any cycle where it acts upon vinylglycine. This idea is in accord with the ability of vinylglytine to inactivate the transaminase in the absence of added aketo acid. Any degree of normal transamination would have left that percentage of enzyme in the "protected" pyridoxamine-P form of the enzyme and refractory to complete inactivation until keto acid is added to regenerate the pyridoxal-P form. No such protection is evident in Rando's experiments or ours with this enzyme. The basis for the different partitioning ratios are obscure. It could be a kinetic difference between rates of hydrolysis of bound imine product uersus rates of nucleophilic attack by the susceptible residues of the two transaminases.
Similarly unclear at the molecular level is why L-alanine transaminase utilizes vinylglycine as a transamination substrate but undergoes no detectable inactivation." Finally, the contrast between the behavior of o-vinylglycine and p-chloro-n-alanine with the bacterial n-specific aminotransferases should be noted. We have noted that @hloroalanine undergoes HCl elimination but not normal transamination.2 The a-keto acid-independent HCl elimination pathway is reminiscent of putative pathway B3 of Scheme 1 whereas transamination would have been through Al. Thus the transaminases apparently will generate the product eneamine with the three carbon halo substrate and not the four carbon olefinic substrate. (Chloroalanine inactivation than is most economically envisaged as proceeding by a path analogous to B4 rather than A2, since no chloropyruvate is detectable.) The routing of the enzyme essentially exclusively through path A a In this connection, we have noted (7) that propargyl glycine (2amino-4-pentynoate) inactivates the L-alanine aminotransferase but not the L-aspartate aminotransferase. In preliminary experiments, we have found that o-propargyl glycine irreversibly inactivates the Bacillus subtilis transaminase, suggesting it may show susceptibilities intermediate to the two mammalian enzymes. or B of Scheme 1 then could depend on the difference in electronegativity of the chloro and vinyl /3 substituents in chloroalanine and vinylglycine. Or, it may reflect the inability of the enzyme to supply a proton at carbon 4 of vinylglycine necessary to capture a rearranging allylic carbanion (Scheme 2). If a protonated base is not available to provide general acid catalysis, there may be no driving force for the allylic rearrangement to the eneamine (2 of Scheme 21. Instead, the (Ycarbanion (1 of Scheme 21 may simply undergo the usual 1,3azallylic prototropic isomerization that is normal transamination pathway (to 3 of Scheme 21. In contrast, the chloride ion is a leaving group in itself and can be eliminated from /3-chloroalanine without need for protonation from a BH+ group. Following this line of reasoning, one could see how inactivation from P-chloroalanine processing could be due to alkylation of an eneamine intermediate while inactivation during vinylglycine processing could be from Michael attack on the bound imine product.