Factors contributing to the inhibition of aspartate aminotransferase by dicarboxylic acids.

At pH 8.0 aspartate aminotransferase (L-aspartate:2-oxoglutarate aminotransferase, EC 2.6.1.1) reacts with the modified substrate, erythro-beta-hydroxy-L-aspartate, to form a mixture of enzyme-substrate complexes absorbing at 492 nm. A variety of dicarboxylic acids were studied spectrophotometrically as competitive inhibitors of this reaction. All of the inhibitory dicarboxylic acids form a complex with the enzyme, absorbing at 362 nm. In addition, some of the dicarboxylic acids form a protonated complex absorbing at about 435 nm. This complex, which is the conjugate acid of that absorbing at 362 nm, is formed only by those dicarboxylic acids which can assume a configuration in which the two carboxyl groups are positioned as in maleic acid. Bulky substituents, such as aromatic rings or even methyl groups, prevent the formation of the protonated complex, presumably because of steric restrictions at the active site. Substitution of the central carbon atom of glutaric acid by heteroatoms of increasing charge density results in a progressive decrease in inhibitory effectiveness, at pH 8, primarily due to a loss of this pH-dependent stabilization of the enzyme-dicarboxylic acid complex. Acids with an aromatic ring are among the most potent dicarboxylic acid inhibitors of this enzyme in spite of the fact that they do not undergo the pH-dependent stabilization of their enzyme complexes. From these observations it was concluded that the affinity of aspartate aminotransferase for dicarboxylic acids is determined as much by the mechanism of binding as by the solvation and steric effects.

. In contrast, the phosphopyridoxal enzyme, aspartate aminotransferase (L-aspartate:2-oxyoglutarate aminotransferase, EC 2.6.1.1) provides an ideal system for the direct study of ligand affinities since the interactions with substrates and inhibitors cause marked changes in the visible absorption spectrum of the prosthetic group (3,9). These spectral changes may be used to determine the dissociation constant from the active site with a high degree of precision (10). However, the unique advantage of direct spectrophotometric studies is that they frequently reveal that the ligand-macromolecule interaction produces more than one complex (3,11,12). Such studies have provided information regarding the mechanism of formation and the interrelationships between such multiple complexes (3,9,13 resnectivelv. From this scheme it is obvious that the observed in a l-ml cuvette at 25". Erythro-P-hydroxy-L-aspartate concentraaffinity of this enzyme for a dicarboxylic acid depends not only upon the ligand dissociation constant, K, = (E)(A)I(EA), but also upon the protolytic dissociation constant, K, = (EA)(H+)/ @HA), for the initial enzyme.dicarboxylic acid complex. In fact, it can be shown that the observed inhibition constant, K,, for a dicarboxylic acid as a competitive inhibitor of aspartate aminotransferase varies with the pH between two limiting values, K, and K,, according to Equation 1.
Previous work showed that pimelic acid and fumaric acid are relatively weak inhibitors of aspartate aminotransferase even at low pH values. On the other hand, the high degree of inhibition by glutaric acid and maleic acid at low pH values has been demonstrated to be due to a low value of the dissociation constant K, for the protolytic step (3,14).
In effect, these observations indicate that a dicarboxylic acid may inhibit by alternative mechanisms, that is, by forming two different inhibitory complexes which are the enzyme 'dicarboxylic acid complexes, EA and EHA, described in Scheme I.
Values of K,, K,, and A,,, used to initiate the nonlinear analysis were provided by a series of linear least squares analyses which are identical with graphical methods described elsewhere (12) and illustrated in Fig. 1. Between 20 and 30 data points were used to determine the three constants for each dicarboxylic acid. All error determinations are in accord with the treatment presented by Cleland (18) This ratio is a sensitive measure of the affinity of aspartate aminotransferase for a particular ligand relative to its affinity for erythro-&hydroxy-L-aspartate (19). A low value of the affinity ratio thus signifies a high affinity of the enzyme for that ligand. K, values which appear in Table I (4) where (H+), the proton concentration, is fixed by the Tris base to Tris-acetate ratio. Initial values of K,, K,, and D, for the nonlinear computer analysis were obtained from linear least squares analyses such as shown graphically in Fig. 2, which have been described earlier (31. Again, the data were fairly well conditioned with respect to the experimental ligand concentrations and pH ranges.

Inhibition of Aspartate Aminotransferase by Dicarboxylic
Acids-The results of the inhibition analysis according to Scheme II are presented in Table I. It is important for the validity of this simple analysis that the value of A,,, was found to be essentially the same regardless of the dicarboxylic acid investigated.
This indicates, primarily, that none of the multiple enzyme-substrate intermediates (XES in Scheme II) has an appreciable affinity for dicarboxylic acid, because the distribution of the intermediary complexes was not dependent upon the type or concentration of dicarboxylic acid present. Furthermore, the phosphopyridoxamine form of this enzyme, which is produced by the reaction of the phosphopyrid.oxal form with amino acids, binds dicarboxylic acids (9). Formation of the phosphopyridoxamine enzyme with its complexes would also reduce A,,,. However, the reaction of phosphopyridoxal aspartate aminotransferase with erythro-&hydroxy-r.-aspartate produces very little of the phosphopyridoxamine form (9, 10) and with the low concentrations of dicarboxylic acid used in this study, there was negligible formation of the phosphopyridoxamine enzyme .dicarboxylic acid complexes. Consequently, the dicarboxylic acids were acting as simple competitive inhibitors of the amino acid in that their effect was empirically described by the familiar equation: where K, is the observed substrate dissociation constant in the presence of a concentration (A) of an inhibitor, and K, and K, have their meanings defined in Scheme II.  I  I  I  I  I  I  I  I  I  I  -125 -   in Table  I shows a reasonable agreement. Table  III, which presents a summary of data obtained in both this and previous studies (3,9,14), shows those dicar-   high enough to give a yellow color at pH 8. The choice of pH 8 for Table III was thus arbitrary but is satisfactory to distinguish dicarboxylic acids, capable of forming bidentate ligands, from monovalent anions and from those dicarboxylic acids capable of acting only, as though they were monovalent anions, to form monodentate complexes. Some of the acids which do not form an acidic complex at pH 8 may do so at a lower pH but it is not easy to identify them unequivocally.
The large number of dicarboxylic acid inhibitors of aspartate aminotransferase that are presented in this study allows the analysis of the factors that contribute to the affinity of this enzyme for a particular ligand. These contributing factors can be discussed as belonging to one of the following broad categories: steric effects, mechanistic effects, and solvation effects. However, it must be noted that none of these categories is wholly independent of the other two. The data presented in Table I indicate that the aromatic ring dicarboxylic acids are the most potent inhibitors in this group. However, as revealed in Table III, none of these acids are able to form an acidic EHA complex with the enzyme at pH 8. In contrast, the straight chain dicarboxylic acids, glutaric and thiodiglycolic acids, are nearly as potent inhibitors, but these acids are able to form the acidic complex. These results suggest that the high affinity of the enzyme depends upon different factors for each of these two classes of dicarboxylic acid inhibitors.
In the case of the aromatic ring acids, the high affinity is probably produced by the hydrophobic effect. On the other hand, the high affinity of the enzyme for the straight chain acids is most likely due to the formation of this additional yellow complex.
The contribution of the yellow acidic complex to the binding affinity can be appreciated by consideration of Scheme I and of  3 (left). Space-filling models of four dicarboxylic acids which are able to form a protonated complex with aspartate aminotransferase at pH 8.0. Model A, maleic acid; B, glutaric acid; C, cis-1,Z cyclobutanedicarboxylic acid; and D, meso-tartaric acid. The models are arranged so that the adjacent carboxyl groups are eclipsed and are in a conformation analogous to that of maleic acid. FIG. 4 (right). Space-filling models of dicarboxylic acids that are unable to form a protonated complex with aspartate aminotransferase at pH 8.0. These models are to be compared with those in Fig. 3. Model A, methylmaleic acid; B, @-methylglutaric acid; C, isophthalic acid; and D, dl-tartaric acid.
Equations 1 and 6. These equations predict that the formation of a protonated enzyme.dicarboxylic acid complex (EHA) is responsible for a large amount of the binding energy for those acids that are able to form this complex with the enzyme. However, the data presented here further indicate that dicarboxylic acids must fulfill rather strict steric requirements in order to inhibit in this manner. The first requirement is that these dicarboxylic acids have as an allowable conformation one in which the carboxyl groups are close together. Maleic acid is a model for this particular conformation. A second requirement is that the dicarboxylic acid must be fairly compact. Apparently, steric restrictions at the enzymic site prevent methylmaleic and P-methyl and &I-dimethyl glutaric acids from forming the acidic EHA complex (14).
This second steric restriction is, however, dependent upon the polarity of the substituent groups in a series of related dicarboxylic acids. This is shown in Table III, for various derivatives of succinic acid, which itself is able to form an acidic EHA complex at pH 8.0 (14). As shown, the presence of a methyl group on succinic acid prevents the formation of a protonated complex at pH 8; however, the addition of one polar group (e.g. hydroxyl=malic acid, either L or D) still permits tight binding because of formation of the protonated complex. Furthermore, two vicinal polar groups permit formation of the acidic complex but these groups must be in erythro conformation as can be appreciated by comparing meso-tartaric acid and meso-diaminosuccinic acid with their corresponding dl isomers in Table III. This must mean that the meso isomers bind in the maleic acid conformation without either hydroxyl group interacting unfavorably with the enzyme. On the other hand, the same conformation of the dl isomers is either too strained or, alternatively it does not result in a favorable interaction of a hydroxyl group with a group on the enzyme. It is interesting to note, in this regard, that the erythro isomer of /3-hydroxyaspartate, which has the same stereochemistry of its polar groups as meso-tartaric and meso-diaminosuccinic, forms a unique covalent complex with the enzyme which was not observed with the threo isomer (14), nor, to any significant extent, with any other amino acid.
Returning to the consideration of the enzyme's high affinity for the aromatic dicarboxylic acids, which are sterically prohibited from forming a stabilized protonated EHA complex, it seems reasonable to conclude that these ligands derive much of their binding energy from solvation effects or, in other words, from "hydrophobic interactions" (23). This hypothesis is further supported by consideration of the data in Table I that show the effect of replacing a central carbon atom by heteroatoms of increasing charge density (4). For the series of aromatic dicarboxylic acids, the order of the affinity for the enzyme is: thiophenedicarboxylic acid > isophthalic acid > furandicarboxylic acid > dipicolinic acid. A corresponding series of the straight chain glutaric acid analogues has the order of affinity: thiodiglycolic acid = glutaric acid > oxydiglycolic acid > iminodiacetic acid. Thus the replacement of the central atom by sulfur, carbon, oxygen, and nitrogen results in a progressive decrease in the enzyme's affinity for members of these two series.
In a recent paper, Rogers (4) examined the same series of glutaric acid analogues as inhibitors of glutamate dehydrogenase and found a good correlation between the sigma charge density at the central atom and the extent of inhibition of the enzyme. His data are compared in Table IV with the data presented in this paper. Rogers pointed out that the charge density is an index of the interaction between atoms of an organic compound and water. He thus suggested that the hydration of the dicarboxylic acid is an important factor in the binding reaction. The data presented in this paper also indicate that in the case of aspartate aminotransferase, the more hydrated ligands, iminodiacetic acid and oxydiglycolic acid, have a different mechanism of binding in that they are not able to form an acidic complex (Table III). A probable explanation of this observation is that water molecules introduce a steric restriction by hydrating the oxygen and nitrogen atoms and thereby prevent formation of the protonated complex. That the water of solvation can act as a structural component to affect binding affinities and specificities has in fact been demonstrated in antibody-hapten reactions (24).
It is relevant at this point to mention other studies of phosphopyridoxal enzymes that have examined dicarboxylic acid inhibitors. Haarhoff (25) did a detailed computer kinetic analysis of the inhibition by dicarboxylic acids of the reaction catalyzed by aspartate aminotransferase.
In general, his study showed trends which have been further delineated in the present study. Previous studies of aspartate aminotransferase also have shown that the extent of dicarboxylic acid binding increases with decreasing pH (3,26) even with the phthalic acids (27). Another enzyme, kynurenine transaminase, was shown to be inhibited most strongly and specifically at low pH values, and it was found that adipic acid had the optimum chain length. From these observations, Mason (6) proposed that there must be two cationic sites about 11 A apart, only one of which binds the dicarboxylic acid at high pH. Finally, a study of glutamate decarboxylase found that glutaric, adipic, and pimelic acids were powerful inhibitors, especially at low pH. Fonda (5) therefore concluded that the positive binding sites must be about 6 A apart since neither maleic acid nor fumaric acid reacted. She also pointed out that the positive binding sites on aspartate aminotransferase must be less than 6 A apart since maleic acid is much more effective than fumaric acid as an inhibitor of this enzyme (14).
Despite these many studies, the nature of the protonated aspartate aminotransferase dicarboxylic acid complex has not been satisfactorily established. It has been suggested (3) that one of the carboxyl groups binds as a carboxylate anion, possibly to the same active site histidine residue that is reputed to bind monoanions (26), and that the second carboxyl group binds as an undissociated acid, probably to, or close to, the nitrogen of the pyridoxal phosphate-lysine Schiff base (28), as shown schematically in Fig. 5. Such a representation is consistent with the data presented in this paper. It is implied with this drawing that the dicarboxylic acid ligand undergoes a conformational change with protonation of the EA complex to form EHA, and, as the data in this paper indicate, only certain dicarboxylic acids are sterically able to undergo this conformational change at the active site to form the protonated EHA complex.
In conclusion, when it is observed, as in many studies with dicarboxylic acid enzyme-inhibitors, that the relative potencies vary with the pH, it is plainly an invalid oversimplification to use inhibition data obtained at one pH to attempt to map the active site. This paper shows that with aspartate aminotransferase, this pH variation can be accounted for by assigning two dissociation constants to each dicarboxylic acid inhibi-