Active-site Arg → Lys Substitutions Alter Reaction and Substrate Specificity of Aspartate Aminotransferase*

Arg386 and Arg292of aspartate aminotransferase bind the α and the distal carboxylate group, respectively, of dicarboxylic substrates. Their substitution with lysine residues markedly decreased aminotransferase activity. Thek cat values with l-aspartate and 2-oxoglutarate as substrates under steady-state conditions at 25 °C were 0.5, 2.0, and 0.03 s−1 for the R292K, R386K, and R292K/R386K mutations, respectively, k cat of the wild-type enzyme being 220 s−1. Longer dicarboxylic substrates did not compensate for the shorter side chain of the lysine residues. Consistent with the different roles of Arg292 and Arg386 in substrate binding, the effects of their substitution on the activity toward long chain monocarboxylic (norleucine/2-oxocaproic acid) and aromatic substrates diverged. Whereas the R292K mutation did not impair the aminotransferase activity toward these substrates, the effect of the R386K substitution was similar to that on the activity toward dicarboxylic substrates. All three mutant enzymes catalyzed as side reactions the β-decarboxylation of l-aspartate and the racemization of amino acids at faster rates than the wild-type enzyme. The changes in reaction specificity were most pronounced in aspartate aminotransferase R292K, which decarboxylated l-aspartate tol-alanine 15 times faster (k cat = 0.002 s−1) than the wild-type enzyme. The rates of racemization of l-aspartate, l-glutamate, andl-alanine were 3, 5, and 2 times, respectively, faster than with the wild-type enzyme. Thus, Arg → Lys substitutions in the active site of aspartate aminotransferase decrease aminotransferase activity but increase other pyridoxal 5′-phosphate-dependent catalytic activities. Apparently, the reaction specificity of pyridoxal 5′-phosphate-dependent enzymes is not only achieved by accelerating the specific reaction but also by preventing potential side reactions of the coenzyme substrate adduct.

The pyridoxal 5Ј-phosphate (PLP) 1 -dependent enzymes that catalyze transformations of amino acids (for a recent review, see Ref. 1) constitute a few families of evolutionarily related enzymes (2). The member enzymes of such a family use the same protein scaffold to catalyze quite diverse reactions. Apparently, subtle structural differences underlie their catalytic specificity.
Aspartate aminotransferase (AspAT) is probably the most extensively studied PLP-containing enzyme. It catalyzes the reversible transamination of the dicarboxylic L-amino acids, aspartate and glutamate, and the corresponding 2-oxo acids, oxalacetate and 2-oxoglutarate. During the catalytic cycle, the cofactor shuttles between the PLP and the pyridoxamine 5Јphosphate (PMP) forms. High resolution x-ray crystallographic analyses (3)(4)(5)(6) in conjunction with site-directed mutagenesis studies (7)(8)(9)(10)(11)(12)(13)(14) have elucidated the role of several active-site residues. The specificity for dicarboxylic amino acids appears to be based mainly on two active-site arginine residues (Fig. 1). Arg 386 of the small domain binds the ␣-carboxylate group of the substrate, and Arg 292 of the large domain of the adjacent subunit interacts with the distal carboxylate group. The spatial orientation of these key residues is determined by steric constraints and polar interactions. The van der Waals contacts of the guanidino nitrogens of Arg 386 with the side chain carbonyl of Asn 194 and the aromatic ring of Phe 360 effectively delimit the conformational space available to Arg 386 . The guanidino nitrogens of Arg 292 are within hydrogen bonding distance from the carboxylate group of Asp 15 , the side chain amide of Asn 142 , and the hydroxy group of Ser 296 of the adjacent subunit. The side chain of Arg 292 is thus maintained in an extended configuration, which favors interaction with the distal carboxylate group of the incoming substrate (15,12). Arg 386 is invariant in all known aminotransferase sequences. Arg 292 is conserved in most AspAT sequences, and other aminotransferases have variable residues at position 292 (16).
In an attempt to explore the mechanisms responsible for the reaction specificity of AspAT, we re-engineered the substratebinding site of the enzyme by substituting the substrate-binding Arg 292 and Arg 386 with lysine residues. This conservative substitution was expected not to abolish the catalytic apparatus of the enzyme, but to alter the electron repartition and certain bond angles in the coenzyme-substrate adduct, both important determinants of catalytic specificity.

Site-directed Mutagenesis and Purification of Wild-type and Mutant
AspATs-Oligonucleotide-directed mutagenesis of the wild-type aspC gene of Escherichia coli inserted into the BS M13 vector (17) was performed with the mutagenesis kit of Bio-Rad (18) and the oligonucleotides GC CAC ATT TAC TTT ACC AGA AGC and GA GTA GTT AGC TTT AAT CGC CGC T for the R386K and the R292K mutation, respectively. The mutations were confirmed by determination of the nucleotide sequences. The mutated DNAs were expressed in the Asp-AT-deficient E. coli strain TY103 (19) with the expression vector pK-DHE19 (20).
Wild-type and mutant enzymes were purified as described previously * This work was supported in part by an Italian National Research Council Bilateral Project, the Swiss National Science Foundation Grant 31-45940.95, and the COST ACTION-D7 Program of the European Cooperation in the Field of Scientific and Technical Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Measurement of Aminotransferase Activity-The activity of the purified enzymes toward various amino and oxo acid substrates was determined under single turnover conditions by monitoring the changes in the UV/VIS absorption spectrum of the enzyme-bound cofactor. The reactions were conducted at 25°C in 50 mM 4-methylmorpholine, pH 7.5, containing 9 M (subunit concentration) AspAT and the substrate. The half-reaction from amino acid to oxo acid was followed by measuring the decrease in absorbance at 360 nm and the increase in absorbance at 330 nm due to the conversion of enzyme-bound PLP to PMP; the reverse half-reaction was followed in an analogous manner. The PMP form of the enzymes was prepared by incubation of the PLP form with 1 mM PMP and 5 mM cysteine sulfinate for 30 min at 25°C in the dark followed by Sephadex G-25 chromatography. The reactions were followed with a Beckman 7400 DU spectrophotometer. With rapidly reacting substrates, a stopped-flow apparatus (a pbp-Spectra Kinetic Monochromator 05-109 from Applied Photophysics) with a cuvette of 1-cm path length and a dead time of 2 ms was used. In all cases, the reaction progress curves fitted to single exponential equations with the pseudo-first order rate constant k obs . The values for k cat and K m were obtained from the k obs values at varying substrate concentrations by Lineweaver-Burk analysis using GRAFIT software (Erithacus Software). The calculated values of k cat /K m were in good agreement with the values measured at low substrate concentrations ([S 0 ] Ͻ Ͻ K m ), where k obs /[S 0 ] gives directly k cat /K m .
Steady-state aspartate aminotransferase activity was measured in a coupled assay with malate dehydrogenase in 50 mM 4-methylmorpholine, pH 7.5, at 25°C in the presence of 20 mM 2-oxoglutarate plus 20 mM or 150 mM L-aspartate for wild-type enzyme or mutant AspATs, respectively.

Determination of Dissociation Equilibrium Constants for Competitive Inhibitors and of pKЈ a of Internal Aldimine by Spectral Titration-
Spectral titrations of the PLP form of the wild-type and mutant enzymes were performed at 25°C in 20 mM sodium phosphate. The dissociation equilibrium constants of the wild-type and mutant enzymes for competitive inhibitors were determined at pH 7.5 (9 M subunit concentration) by monitoring the absorbance of enzyme-bound PLP at 430 nm as a function of inhibitor concentration (22). To determine the pKЈ a value of the internal aldimine, 3 ml of enzyme solution were titrated by the repeated addition of 3-5 l of 2 M acetic acid. The ensuing pH values were measured, and spectra were recorded from pH 7.5 to 5.0. In the titration of the wild-type and the three mutant enzymes, the absorption maximum shifted from 360 to 430 nm. The spectral change reflects the protonation of the internal aldimine. The values of K d for competitive inhibitors and pKЈ a of the internal aldimine were obtained by fitting the measured values of absorbance at 430 nm to theoretical dissociation curves using GRAFIT software.
Assay for Newly Generated Activities toward Amino Acids-Mutant AspATs and wild-type enzyme (0.9 mM subunit concentration) were incubated in 50 mM 4-methylmorpholine, pH 7.5, at 25°C with different amino acids and the cognate oxo acids as substrates. Samples of 20 l were withdrawn at different times and immediately frozen in liquid nitrogen. For quantitative analysis of the reaction products, the samples were deproteinized with 1.12 M perchloric acid, derivatized with 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide (from Pierce; Ref. 23), and loaded onto a reverse-phase high pressure liquid chromatography column (Aquapore RP-300; 250 ϫ 4.6 mm). The separated derivatives of amino acid substrates and products were photometrically detected at 340 nm (24). In the assay for serine dehydratation, samples of the reaction mixture were deproteinized and directly analyzed for pyruvate with lactate dehydrogenase and NADH.

RESULTS
Spectroscopic Properties of Mutant AspATs-Virtually identical absorption and CD spectra in the visible region for wildtype and mutant enzymes, in both their PLP and PMP forms (not shown), indicate that the mutations leave the active-site geometry essentially undisturbed. Spectrophotometric pH titration gave a pKЈ a value of 6.4 for the internal aldimine of all three mutant enzymes and of 6.3 for the wild-type enzyme. Apparently, the positive electrostatic potential due to Arg 292 and Arg 386 that is assumed to account for the low pKЈ a of the internal aldimine (4) is maintained in the Arg 3 Lys mutant AspATs.
Changes in Reaction Specificity-Replacement of either Arg 386 or Arg 292 with a lysine residue resulted in a marked decrease in aminotransferase activity toward dicarboxylic substrates. The k cat values of the half-reaction from L-aspartate to oxalacetate were decreased by 2 orders of magnitude in the single mutant enzymes and by 4 orders of magnitude in the double mutant enzyme (Table I). Due to a general increase in K m values by 1-2 orders of magnitude, an even larger decrease in the catalytic efficiency k cat /K m was observed for all mutant enzymes. The kЈ cat values of the overall steady-state transamination reactions catalyzed by the mutant enzymes were decreased commensurately with the decrease in rate of the halfreactions (Table II).
The pH rate profiles for AspAT R386K and AspAT R292K/ R386K did not significantly differ from that of wild-type enzyme (Fig. 2). In contrast, the pH optimum of AspAT R292K was considerably narrower than that of the wild-type enzyme. The decrease in activity at higher pH might be due to the deprotonation of the newly introduced Lys 292 . The affinity of AspAT R292K for aspartate decreased only little with pH (K m of Asp ϭ 14 mM at pH 7.5, 18 mM at pH 9.0, and 36 mM at pH 10.0), indicating that k cat rather than K m is affected at higher pH.
All three mutant enzymes were analyzed for newly generated catalytic activities (Table II). AspAT R292K racemizes L-aspartate, L-glutamate, and L-alanine 3, 5, and 2 times faster, respectively, than the wild-type enzyme. The same mutant enzyme catalyzes the ␤-decarboxylation of L-aspartate to Lalanine with a kЈ cat of 0.002 s Ϫ1 , i.e. 15 times faster than the wild-type enzyme. The ratio between the steady-state rates of ␤-decarboxylation and transamination is 0.004 and 5.9 ϫ 10 Ϫ7 for AspAT R292K and the wild-type enzyme, respectively. Both AspAT R386K and R292K/R386K exhibited a three times higher ␤-decarboxylase activity and a 2-3-fold higher alanine racemase activity than the wild-type enzyme. All three mutant enzymes showed about the same serine dehydratase activity as the wild-type enzyme.
Changes in Substrate Specificity-Aliphatic dicarboxylic and monocarboxylic amino and oxo acids of various length as well as aromatic substrates were tested for transamination (Tables  I and III). As previously shown for AspAT R386K (7) and now for AspAT R292K and AspAT R292K/R386K, the decrease in both k cat and k cat /K m compared with the wild-type enzyme was more pronounced with C 5 dicarboxylic substrates (glutamate and 2-oxoglutarate) than with C 4 dicarboxylic substrates (as- partate and oxalacetate). The k cat /K m value of AspAT R386K but not of AspAT R292K toward C 6 dicarboxylic substrates (2-aminoadipate and 2-oxoadipate) was significantly lower than that toward C 5 dicarboxylic substrates. The activity of all mutant enzymes toward alanine and pyruvate was decreased, while k cat /K m of AspAT R292K toward norleucine and 2-oxocaproic acid was even higher than that of the wild-type enzyme.
The R386K substitution decreased the catalytic efficiency toward aromatic amino acids by 1-4 orders of magnitude. In contrast, the k cat /K m values of AspAT R292K for aromatic amino acids were only slightly lower than that of the wild-type enzyme and thus higher by 1-2 orders of magnitude than that toward L-aspartate.
Changes in the Binding of Dicarboxylic Inhibitors-The affinity of the mutant enzymes for dicarboxylic substrate analogs of varying length was compared with that of the wild-type enzyme (Table IV). Dicarboxylic acids bind noncovalently to the enzyme and increase the pKЈ a of the internal aldimine, thereby inducing, at pH 7.5, the conversion of the unprotonated species ( max 360 nm) to the protonated species ( max 430 nm), thus allowing spectrophotometric determination of their dissociation constants (see "Experimental Procedures"). All aliphatic inhibitors (C 4 to C 6 ) with K d values in the millimolar range in the case of the wild-type enzyme, were bound by all mutant enzymes with a K d value that was higher by 1 order of magnitude and corresponded to the K d values of the wild-type enzyme for the too short (C 3 ) and the too long (C 7 ) inhibitors. All enzyme forms showed K d values Ͼ100 mM for monocarboxylic inhibitors (butyrate and n-valerianate). Apparently, the hydrocarbon chain contributes only insignificantly to the binding of the inhibitors. o-Phthalate behaves like C 4 to C 6 aliphatic inhibitors. m-Phthalate, however, is the only inhibitor that is bound more tightly by the mutant enzymes than by the wildtype enzyme. m-Phthalate with its fixed conformation may be assumed to interact more strongly with the more flexible lysine residue than with the arginine residue, the side chain of which is fixed by multiple interactions of its guanidinium moiety (see Introduction). and aromatic amino acids Values were determined under single turnover conditions in 50 mM 4-methylmorpholine, pH 7.5, at 25°C and 9 M subunit concentration by Lineweaver-Burk analysis of the rate of the decrease in A 360 (see "Experimental Procedures"). The concentration ranges of the tested amino acids were 10 -200 mM for dicarboxylic amino acids, 2.5-150 mM for L-alanine, 5-90 mM for L-norleucine, 1-30 mM for L-phenylalanine, and 1-5 mM for L-tyrosine.

DISCUSSION
Arg 3 Lys substitutions are generally considered conservative alterations in protein structure. In AspAT R292K, AspAT R386K, and the double-mutant enzyme, absorption and CD spectra as well as determination of the pK a Ј value of the internal aldimine indeed did not reveal changes in the active-site geometry. Nevertheless, both single mutations reduced the catalytic efficiency (k cat /K m ) for transamination of C 4 and C 5 dicarboxylic substrates by at least 3 orders of magnitude (Table  I). Similar results with AspAT R386K have been reported previously (7). 2 A decrease by 6 orders of magnitude was brought about by the double mutation. Substitution of Arg 292 with aspartate (26), valine, and leucine (27) has been found to have the same effect. Replacement of Arg 386 with tyrosine, phenylalanine (8), or alanine (14) reduced the catalytic efficiency by 4 orders of magnitude. Thus, irrespective of whether Arg 292 or Arg 386 is substituted and irrespective of the nature of the new side chain, the catalytic efficiency is decreased by at least 3 orders of magnitude.
On the basis of chemical modification studies, it was suggested early on that anionic substrates of enzymes are bound by arginine residues (28). In the case of AspAT, determination of the crystal structure of enzyme-substrate analog complexes (3)(4)(5)(6) and of enzymic reaction intermediates (29) has confirmed this prediction. The preference for arginine in AspAT and many other enzymes may be explained by the peculiar strength of the guanidinium-carboxylate interaction due to the resonance-sta-bilized ion pair underlying the two hydrogen bonds that can be formed (30,31). Moreover, arginine is a poor proton donor because of resonance stabilization and hence would probably not function as a general acid catalyst. Evolutionary selection of arginine thus minimizes nonspecific hydrolysis of substrates (28). The greater number of possible hydrogen bonds not only with carboxylate groups of the substrates but also with other polar active-site residues endows the guanidinium-carboxylate interaction with a more strictly defined geometry than could be achieved with lysine. The crucial role of arginine-carboxylate interactions, both for substrate binding and efficient catalysis, is borne out by our results.
Wild-type AspAT shows an inverse relationship between k cat /K m values and the side chain length of dicarboxylic amino acids (Table I), as has also been shown in another study (32). The side chain of lysine is shorter than that of arginine, the C ␣ -N distance being 5.64 and 6.50 Å, respectively. However, the Arg 3 Lys substitution cannot be compensated by longer dicarboxylic substrates. Neither of the substituted arginine residues directly participates in the covalency changes catalyzed by AspAT. Apparently, the loss in enzymic activity is due to modes of binding of the substrates that do not allow the catalytic apparatus to become fully effective. Indeed, all Arg 3 Lys mutant AspATs show a decrease in catalytic competence with increasing length of the dicarboxylic substrates, although the binding of dicarboxylic reversible inhibitors is independent of their length (Table IV). Arginine residues that are responsible for the formation of catalytically competent enzyme-substrate complexes and cannot be replaced by lysine without substantial loss in catalytic activity have also been found in enzymes other than AspAT (33)(34)(35)(36)(37).
Consistent with the different roles of Arg 292 and Arg 386 in substrate binding, the effects of their substitution on the activ-    (Tables I and III). The introduction of a lysine residue at position 292, which is situated on the re face of the coenzyme-substrate adduct (Fig.  1), increases the rate of racemization of L-aspartate, L-glutamate, and L-alanine 2-3-fold (Table II). A somewhat greater effect was observed when Trp 140 on the re face of PLP was replaced by histidine (Table V; Ref. 13). Reprotonation of the coenzyme-substrate adduct at C ␣ from the re instead of the si side is a rare event in wild-type AspAT due to the absence of polar residues at the re face and to the almost total exclusion of water in the closed conformation of the substrate-liganded enzyme (24,38). Conceivably, the R292K substitution interferes with the substrate-induced closure of the active site.
The R292K mutation increases not only the racemase activity but also enhances the L-aspartate ␤-decarboxylase activity 15-fold in comparison with the wild-type enzyme (Table II). For efficient ␤-decarboxylation, a proton donor has to operate in close proximity to the position of the leaving carboxylic group (25). The accelerated ␤-decarboxylation observed with AspAT R292K may be explained, at least in part, by the presence of Lys 292 , which is a much better proton donor than arginine and might, assisted by an intervening water molecule, protonate the carbanion intermediate 5 (Scheme 1) produced by ␤-decarboxylation. The reaction pathways of transamination, ␤-decarboxylation, and racemization diverge only after deprotonation at C ␣ of the external aldimine 2 has formed the quinonoid intermediate 3. In transamination, reprotonation of this intermediate at C 4Ј produces the ketimine intermediate 4. In the wild-type enzyme, despite its much higher transaminase activity, both removal of CO 2 from C ␤ and reprotonation at C ␣ from the re side to give the aldimine intermediate 6 are slower than in the mutant enzymes.
In conclusion, neither of the two substrate-binding arginine residues of E. coli AspAT can be replaced by lysine without a marked decrease in the affinity for dicarboxylic substrates and an even larger decrease in the catalytic activity toward these substrates. The exact positioning of the substrate moiety, the additional hydrogen bonds of the guanidinium group to other active-site residues, and the specific electrostatic potential due to the guanidinium groups are all factors that conceivably contribute to efficient catalysis. The present data and earlier results (Table V) provide unequivocal experimental evidence that the reaction specificity of PLP-dependent enzymes is not only achieved by accelerating the specific reaction but also by preventing the occurrence of potential side reactions. Most changes in the delicately poised catalytic machinery of AspAT have indeed resulted not only in a decrease in rate of the specific reaction, i.e. transamination, but also in an acceleration of the side reactions, i.e. ␤-decarboxylation and racemization.