Properties of aspartate racemase, a pyridoxal 5'-phosphate-independent amino acid racemase.

Aspartate racemase from Streptococcus thermophilus contains no pyridoxal 5'-phosphate or other cofactors such as FAD, NAD+, and metal ions. It was affected by neither carbonyl reagents such as hydroxylamine nor sodium borohydride but was strongly inhibited by iodoacetamide and other thiol reagents. Aspartate, cysteate, and cysteine sulfinate were the only substrates. The Km values for L- and D-aspartate were 35 and 8.7 mM, respectively. The enzyme catalyzed the exchange of alpha-hydrogen of the substrate with the solvent hydrogen. Racemization of L-aspartate in 2H2O showed an overshooting in the optical rotation of aspartate before the substrate was fully racemized. This shows that the removal of alpha-hydrogen of the substrate is at least partially rate-determining. When L- or D-aspartate was incubated with aspartate racemase in tritiated water, tritium was incorporated preferentially into the product enantiomer. The results strongly suggest that aspartate racemase contains two hydrogen acceptors.

Racemization is superficially a simple reaction. For example, it is accomplished by the removal of a-hydrogen bound to a chiral carbon of the substrate and subsequent nonspecific return of a hydrogen to the carbon. However, amino acids are racemized only slowly under ordinary conditions: half-lives of aspartic acid and alanine in the racemization at 25 "C are 3,500 and 12,000 years, respectively (Bada, 1985). This is because of the high dissociation energy of the C"-H bond.
Amino acid racemases and epimerases catalyze the racemization and epimerization of amino acids, respectively, and most of them require PLP' as a coenzyme. The racemization proceeds through the formation of aldimine Schiff base between the substrate amino acid and PLP (Snell and Di Mari, 1970). However, several other amino acid racemases require no coenzymes. Proline racemase (EC 5.1.1.4) catalyzes the proline racemization effectively without PLP by action of a pair of cysteinyl residues at the active site of the enzyme (Cardinale and Abeles, 1968;Rudnick and Abeles, 1975). Knowles and his co-workers studied the energetics and mechanism of the enzyme reaction , 1987a, 1987bFisher et al., 1986aFisher et al., , 1986bFisher et al., , 1986c * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 5 Present address: Nagaoka University of Technology, Nagaoka 940-21, Japan.

EXPERIMENTAL PROCEDURES
Purification of Aspartate Racemase-We purified aspartate racemase from E. coli HB101-pAG6-2 (Yohda et al., 1991) to homogeneity by the same method as reported previously (Yohda et al., 1991) except that a step of gel filtration with Cellulofine GCLlOOOm was added after the butyl-Toyopearl hydrophobic chromatography.
Enzyme and Protein Assay-Aspartate racemase was assayed at 30 "C with a mixture (0.2 ml) containing 0.1 M potassium phosphate buffer (pH 8.0), 24 mM 4-aminoantipyrine, 24 mM N-ethyl-N-(2hydroxy-3-sulfopropyl)-m-toluidine (from Dojin), 2 units of horseradish peroxidase (from Toyobo), 28 units of L-glutamate oxidase (from Yamasa Shoyu), 0.2 mM D-aspartate, and aspartate racemase in a final volume of 0.2 ml. L-Aspartate formed was determined with L-glutamate oxidase, and the hydrogen peroxide produced reacted with 4-aminoantipyrine, N-ethyl-N-(2-hydroxy-3-sulfopropyl)-m-toluidine, and peroxidase to form a condensation product which absorbs at 555 nm. The production of L-aspartate from the D-enantiomer was followed in the same manner except that L-glutamate oxidase and Daspartate were replaced by D-aminO acid oxidase (2.2 units) and Laspartate, respectively. Racemization of aspartate, cysteate, cysteine sulfinate, and other amino acids was also followed by measurement of a change in optical rotation at 365 nm with a Perkin-Elmer polarimeter model 241. The reaction mixture containing 0.2 M substrate, 0.1 M Tris-HC1 (pH 8.0), and aspartate racemase was incubated in a polarimeter cell (1 ml) a t 30 "C.
Protein was determined by the method of Bradford (1976), with bovine serum albumin as a standard, or from the absorption coefficient of the enzyme (A;%, a t 280 nm, 5.3), which was calculated according to the method of Kuramitsu et al. (1990). One unit of enzyme was defined as the amount of enzyme that catalyzes the formation of 1 pmol of L-aspartate from the D-enantiomer/min.
Metal Analysis-Metals were determined by the method of standard additions with a Shimadzu AA-670G atomic absorption spectrophotometer equipped with a graphite furnace atomizer. Absorption was monitored at an appropriate wavelength with deuterium arc background correction. All solutions used for metal analysis were passed through a Chelex (Bio-Rad) column (1 X 10 cm) to remove free meta1 ions and were stored in polyethylene bottles.
Modification of Thiol Group-The enzyme (0.1 mg, 8 nmol) was incubated with 0.1-2 mM NTCB in a final volume of 0.2 ml at room temperature. The modification was followed by measurement of the absorbance at 412 nm. The amount of TNB formed from NTCB and aspartate racemase was estimated with the absorption coefficient of 13,600 M cm". An aliquot of the above reaction mixture was removed at appropriate time intervals and added to the assay mixture to make a final concentration of NTCB lower than 0.2 p~.
'H-'H Exchange-The reaction mixture containing 0.1 M L-aspartate, 0.1 M Tris-HC1 (pH 8.0), and 0.05 unit of aspartate racemase in 'HZ0 was incubated at 30 "C. The m-' H exchange of the substrate with the solvent 'H was followed with a Varian XL200 NMR spectrometer. Proton incorporation into the a-position of [a-'Hlaspartate was examined in a similar manner by incubation in 'HzO.
Tritium Incorporation-Tritium incorporation into the racemization product was examined in a mixture (1 ml) containing 0.5 M Dor L-aspartate adjusted at pH 8.0,0.8 unit of aspartate racemase, and 5 mCi of tritiated water. After incubation at 30 "C for 30 or 60 min, the reaction was stopped by boiling for 10 min, followed by centrifugation to remove denatured protein. The supernatant solution was adjusted to pH 3.0 by the addition of concentrated acetic acid and was applied to an Amberlite IR-12OB cation exchange column (bed volume, 2 ml). The column was washed with 50 ml of water, and then the aspartate was eluted with 20 ml of 2 M NH,OH. Radioactivity of the aspartate was measured with a Packard Tri-Carb 300C liquid scintillation system. Tritium incorporated into both isomers was examined with aspartate aminotransferase, which specifically acts on the L-isomer to exchange the a-hydrogen with solvent hydrogen. The solution containing aspartate was adjusted to pH 8.0 with NH40H and incubated with 200 units of aspartate aminotransferase at 37 "C for 12 h. The reaction was stopped by keeping the mixture in boiling water for 10 min followed by centrifugation to remove denatured protein. The mixture was then adjusted to pH 3.0 with acetic acid and applied to an Amberlite IR-120B column (bed volume, 4 ml). The column was washed with 20 ml of deionized water, and the aspartate was eluted with 20 ml of 2 M NHdOH. The washing fraction, which contains tritiated water derived from the L-isomer, and the eluate with NH40H containing the tritium-labeled D-isomer were subjected to radioactivity measurement. The fraction containing aspartate was again treated with aspartate aminotransferase as described above to release thoroughly tritium bound to aspartate.

RESULTS
Cofactor and Prosthetic Group Requirement-Electronic spectrum of the enzyme showed no characteristic absorption except for a peak at 277 nm. Addition of PLP, pyridoxamine 5'-phosphate, FAD, ATP, NAD+, NADP+, NADH, and NADPH at a final concentration of 1 mM did not affect the enzyme activity. This indicates that the enzyme requires none of the cofactors. The enzyme retained its original activity after incubation with 1 mM hydroxylamine HCl or sodium borohydride in 0.1 M Tris-HC1 (pH 8.0) at 30 "C for 30 min. This excludes the possibility that the enzyme contains a carbonyl group participating in catalysis such as pyruvoyl residue.
Metal Cation Requirement-The enzyme was not inactivated by incubation with chelating reagents such as EDTA and a&-dipyridyl at a final concentration of 10 mM. Atomic absorption spectra of the enzyme showed the occurrence of about 0.05 g atom of zinc and 0.001-0.005 g atoms of calcium, vanadium, manganese, iron, cobalt, nickel, copper, selenium, and molybdenum per mol of subunit: the enzyme contains no significant amounts of the metals. Moreover, the addition of Ca2+, V2+, M P , Mn2+, Fe3+, Co2+, Ni", Cu2+, and Zn2+ at a final concentration of 0.1 mM did not increase the enzyme activity. Thus, the enzyme requires none of these metal ions.
Substrate Specificity-L-Cysteine sulfinate and L-cysteate were found to be good substrates of the enzyme; they are racemized at a rate of 51 and 88%, respectively, relative to that of L-aspartate. However, the presence of the acidic group at the @-carbon of the substrate is essential; asparagine, cysteine, and alanine were not the substrates. The carbon chain length of the substrate is also important; both isomers of glutamate were inert. L-Proline and L-hydroxyproline were not racemized. The K,,, values for L-and D-aspartate were 35 and 8.7 mM, respectively. The VmaX values of the racemization of L-and D-aspartate were 8.0 and 1.9 mM/ml/min, respectively. When these values were used, the calculated Keg for aspartate racemization was 1.05, in good agreement with the theoretical value (1.0) for the chemical symmetric reaction.
Proline racemase is inhibited by proline at high concentrations because of oversaturation (Cardinale and Abeles, 1968). However, aspartate racemase is not influenced by high concentrations of aspartate up to 500 mM.
Thiol groups of proteins are converted quantitatively to Scyano derivatives specifically with NTCB, and the modification can be easily monitored by determination of TNB. Therefore, NTCB was used to determine the number of reactive cysteine residues of the enzyme and the rate of modification. Aspartate racemase is composed of two identical subunits as described previously (Okada et al., 1991), and six thiol groups were modified per dimer with a 50 x molar excess of NTCB the value is consistent with the total number of cysteine residues of the enzyme which was deduced from the DNA sequence of the structural gene (Okada et al., 1991). These indicate that the enzyme contains no disulfide bonds. When TNB release was monitored upon incubation of the enzyme (2.4 mM) with 100 mM NTCB at 25 "C and pH 8.0, a fast TNB release of 2.0-2.6 mol/mol of enzyme (tlh < 0.2 min) was followed by a slow one (tlh = 13.8 min) corresponding to 3.4-4.0 mol/mol. When a &fold molar excess of NTCB was used for the thiol alkylation, 1 cysteine residue/dimer was modified with a concomitant loss of over 95% activity (Fig.  1). However, upon incubation of the NTCB-inactivated enzyme with an excess amount of thiol (e.g. 10 mM dithiothreitol), the enzyme was reactivated almost fully; the inactivation by NTCB is reversible (Fig. 2). The enzyme was protected from inactivation with NTCB by the addition of 10 mM DLaspartate (Fig. 2).
Isotope Incorporation and Time Course of the Reaction in 2H20-We examined by 'H NMR whether deuterium is in- corporated into the a-position of aspartate during incubation with aspartate racemase in deuterium oxide. Only the (Yproton signal of aspartate disappeared completely after 12 h. The first-order rate of a-proton loss from L-aspartate in 'HZO, which was determined to be about 0.021 min" by 'H NMR, was close to that of change in optical rotation of L-aspartate (about 0.026 min"). When a ,~,~-t r i d e u~r i o a s p a~t e was incubated with the enzyme in normal water, the a-proton signal of aspartate appeared. The observed 'H-*H exchange at the a-position of the substrate indicates that the C"-H bond is cleaved h e~r o l~i c a l l y . T h e racemization probably proceeds through an a-carbanion intermediate after removal of the ahydrogen from the substrate, Removal and return of the ahydrogen possibly proceed concertedly.
The optical rotation of L-aspartate was followed during incubation with aspartate racemase in deuterium oxide. The rotation was initially negative and increased with time to reach a positive extremum beyond zero, then again approached zero on prolonged incubation (Fig. 3A). However, such an overshoot of the optical rotation was not observed when the reaction was carried out in water with either of the enantiomers (Fig. 3B). This is most probably because of a deuterium isotope effect at the a-position as Cardinale and Abeles (1968) and Cleland (1977) have shown for the proline racemase reaction.
~~i~i~~ I~o~r u t i o n -W e determined the amount of tritium incorporated into the substrate and product enantiomers during incubation with the enzyme in tritiated water. The enzyme acts on the product enantiomer accumulated in the reaction mixture, and the tritium labeling is probably based on the primary, secondary, and their derived reactions. Therefore, the reaction was stopped at its early stage when only a part (from 2.5 to 9.4%) of the substrate enantiomer was converted to the antipode. We analyzed the tritium distribution in the L-isomer with aspartate aminotransferase, which acts specifically on the L-isomer to exchange the a-hydrogen with hydrogen of solvent water. Thus, we determined the amount of tritium incorporated into both isomers by comparison of radioactivity of aspartate before and after the treatment with aspartate aminotransferase. The amount of tritium in L-aspartate was determined specifically by measurement of the radioactivity of water containing tritium liberated from the L-isomer with aspartate aminotransferase. As shown in Table I, tritium was incorporated almost specifically into the product enantiomer regardless of isomerism of the substrate used.

DISCUSSION
Two types of mechanisms for amino acid racemization have been proposed one-base and two-base mechanisms. In the one-base mechanism, an a-hydrogen of substrate amino acid is abstracted by a single acceptor site of a racemase. The intermediate is an a-carbanion derived from the substrate and is kept at the active site until a proton is translocated from one face to the other of the substrate. The one-base mechanism is typical of PLP-dependent amino acid racemases (Snell and Di Mari, 1970). In the two-base mechanism, an ahydrogen of amino acid is abstracted on one face as a proton on the other face is incorporated concertedly. Proline racemase (Cardinale and Abeles, 1968;Rudnick and Abeles, 1975) and ~aminop~melate epimerase (Wiseman and Nichols, 1984) reactions are proposed to proceed by the two-base mechanism.
A mechanism of tritium incorporation into substrate and product enantiomers in tritiated water with a single base is shown in Fig. 4. When L-aspartate was incubated with aspar-

TABLE I Tritium incorporation into L-and i?-a.spurtate in tritiated water
Aspartate was racemized in tritiated water as shown under "Experimental Procedures." Tritium incorporated into L-aspartate was examined with aspartate aminotransferase. FIG. 4. One-base mechanism for racemization of aspartate.
Im is the @-carbanion intermediate derived from aspartate by abstraction of the a-proton.
tate racemase in tritiated water, tritium was incorporated preferentially into D-aspartate. This indicates that the rate of formation of tritiated D-aspartate (kp) is much faster than that of the antipode (ks). D-Aspartate is racemized also through the common intermediates ( i e . E-BH.Im and E-B3H. Im). Therefore, when D-aspartate is used as a substrate, tritium should be incorporated preferentially into the D-isomer. However, this was not the case. Thus, the one-base mechanism is excluded. Fig. 5 shows a two-base mechanism for tritium incorporation into substrate and product enantiomers. ~-[a-~H]Aspartate is produced from L-aspartate through the route a-b-c or a-b-d-e, whereas ~-[a-~H]aspartate is through a-b-d-f-e'. The removal of a-hydrogen from the substrate ( i e . steps b, b', and f) is probably rate-determining as evidenced by the reaction in deuterium oxide. Therefore, the rate of the reaction through a-b-c or a-b-d-e should be much faster than that though a-bd-f-e'. In the D to L direction, the rate of route a'-b'-c' or a'b'-d'-e' is faster than that of a'-b'-d'-f-e in the same manner. Thus, the scheme is compatible with the result that tritium is incorporated preferentially into the product enantiomers irrespective of the substrate enantiomers used.
In proline racemase (Cardinale and Abeles, 1968;Rudnick and Abeles, 1975) and diaminopimelate epimerase (Wiseman and Nichols, 1984), a thiol group of cysteine residue serves as a base for the proton transfer, and aspartate racemase also contains an essential thiol group. We have found by sitedirected mutagenesis that Cys-(but no C Y S '~ and CYS''~) is essential for the enzyme activity as described elsewhere.' Cysa is most probably the residue that is subject to selective modification by NTCB. The alkylation of 1 cysteine residue/ dimer with NTCB resulted in a loss of 95% of the activity; the enzyme shows a half-of-the-sites-reactivity as well as proline racemase (Rudnick and Abeles, 1975) and hydroxyproline epimerase (Ramaswamy, 1984), both of which consist of two identical subunits. It is likely that aspartate racemase has a composite active site formed at the interface of two identical subunits in the same manner as proposed for proline racemase (Rudnick and Abeles, 1975).