The Binding of Substrates and a Product of the Enzymatic Reaction to Glutathione S-Transferase A*

The binding of substrates and a product to glutathi- one S-transferase A from rat liver was studied by use of equilibrium dialysis and equilibrium partition in a two-phase system. The radioactive substrates glutathi- one and bromosulfophthalein as well as a product of glutathione and 3,4-dichloro-1-nitrobenzene, S-(2- chloro-4-nitrophenyl)glutathione, gave hyperbolic binding isotherms with a stoichiometry of 2 mol per mol of enzyme (i.e. 1 molecule per subunit). Glutathione (and glutathione disulfide) had an equilibrium (disso-ciation) constant for the binding of about 10 PM, whereas bromosulfophthalein and the product had equilibrium constants of about 0.5 PM. All ligands showed the same binding stoichiometry, and competi- tion experiments involving unlabeled ligands indicated that glutathione and the glutathione derivatives were binding to the same site. Low affinity sites appeared to exist in addition

The binding of substrates and a product to glutathione S-transferase A from rat liver was studied by use of equilibrium dialysis and equilibrium partition in a two-phase system. The radioactive substrates glutathione and bromosulfophthalein as well as a product of glutathione and 3,4-dichloro-1-nitrobenzene, S-(2chloro-4-nitrophenyl)glutathione, gave hyperbolic binding isotherms with a stoichiometry of 2 mol per mol of enzyme (i.e. 1 molecule per subunit). Glutathione (and glutathione disulfide) had an equilibrium (dissociation) constant for the binding of about 10 PM, whereas bromosulfophthalein and the product had equilibrium constants of about 0.5 PM. All ligands showed the same binding stoichiometry, and competition experiments involving unlabeled ligands indicated that glutathione and the glutathione derivatives were binding to the same site. Low affinity sites appeared to exist in addition to the specific high affinity sites (one per subunit) for all ligands tested. The binding studies are fully consistent with a steady state random kinetic mechanism for the enzyme.
One of the glutathione S-transferases (l), transferase B, was originally discovered as a binding protein capable of forming tight complexes with various endogenous compounds as well as with xenobiotics (see Ref. 2). It was named ligandin by Litwack et al. (2), and it has been suggested that in the capacity of a binding protein it may serve intracellularly in the transport of compounds having hydrophobic regions (see Ref. 1). Recently, quantitative models of facilitated diffusion were considered for ligandin (3). In contrast to the extensive investigations of GSH S-transferase B (ligandin), the characterization of the other GSH S-transferases in terms of binding properties has been very limited (1). A study of nonsubstrate ligands was made on four GSH S-transferases from rat liver, and it was concluded that all of the transferases possess binding properties similar to those of ligandin (GSH S-transferase B) (4).
In connection with the study of the kinetic properties of the transferases it became essential to have information available about the binding of substrates. Extensive kinetic investigations have been carried out with GSH S-transferase A and the substrates 3,4-dichloro-1-nitrobenzene and GSH (5, 6), but information on binding of the reactants to the enzyme is lacking. In view of the complex rate behavior of the enzyme (5, 6) it was desirable to study the binding of reactants, in order to evaluate the possibility of cooperative binding as a cause of non-Michaelian kinetics. Some preliminary data have previously been reported (6, 7).

MATERIALS AND METHODS
S-(2-Chloro-4-nitrophenyl)glutathione was prepared enzymatically (6), and the ""S-labeled compound was synthesized by the same procedure using [""SIGSH (Schwarz/Mann). The specific radioactivities used in the binding studies were about 0. GSH S-transferase A from rat liver, identical with form II of "GSH S-aryltransferase," was purified and assayed as earlier described (6,8). The protein concentration was determined with a micro-biuret method (9) using bovine serum albumin as a standard. The binding of ligands to the enzyme was studied by two methods: equilibrium partition in a two-phase system and equilibrium dialysis.
The equilibrium partition method (10) made use of a system formed Binding of Reactants to GSH S-Transferuse A of GSH by addition of 50 ~1 of 10 mM N-ethylmaleimide to the scintillation vial before mixing with Aquasol was necessary to obtain stable counts of radioactivity.
When bromosulfophthalein was used 50 aI of 0.2 M NaH2P04 was added to each scintillation vial to bleach the color of the dye by lowering the pH value.

Binding of a Single Ligand-The
binding of the product of the enzymatic reaction, S-(2-chloro-4-nitrophenyl)glutathione to GSH S-transferase A was studied by two methods. Fig. 1 shows a Scatchard plot of the results of the equilibrium partition method. By this method as well as by equilibrium dialysis the graphs were linear, indicating hyperbolic  Scatchard plot showing a maximal binding of 2 molecules of GSH per enzyme molecule (Fig. 2). Most of the binding studies were carried out at pH 7.3, because this pH value was optimal for the location of enzyme to one of the two phases in the equilibrium partition method. It was checked by use of equilibrium dialysis that the product and GSH gave the same stoichiometry and hyperbolic binding curve at pH 8.0 (Fig. 2), which was the pH value used in most of the previous kinetic experiments. The inhibitor glutathione disultide (8) gave similar results (data not shown). The second substrate 3,4-dichloro-1-nitrobenzene was not available in radioactive form and could accordingly not be studied by the same methodology. However, an alternative substrate bromosulfophthalein was studied by the equilibrium dialysis method and was also found to give a hyperbolic saturation curve and a binding stoichiometry of 2. A summary of the binding parameters is given in Table I 3 shows that addition of a fixed concentration of unlabeled GSH inhibited the binding of labeled S-(2-chloro-4-nitrophenyl)glutathione.
The effect was competitive, indicating that the two ligands bind to the same site. Other ligands which were also shown to decrease the binding of the product were S-hexyl-and S-octylglutathione and bromosulfophthalein. All effects appeared to be competitive.
Competition experiments involving labeled GSH and unlabeled GSH derivatives gave similar results. The second substrate 3,4-dichlorol-nitrobenzene in concentrations up to 1 mM (near the limit of solubility) had no measurable effect on the binding of product. (The effect on GSH binding cannot be studied owing to the enzyme-catalyzed reaction between the two substrates.) The binding curves of the radioactive ligands were hyperbolic in the concentration range considered, whereas the effect of the unlabeled competitors appeared to be nonhyperbolic.
Figs. 4 and 5 show the results of displacement of a fixed concentration of the radioactive product using variable concentrations of unlabeled GSH and bromosulfophthalein, respectively. However, it should be noted that the concentrations used of the competitors are considerably higher than those of the same compounds in labeled form used in the binding studies (cfi Fig. 2). In fact, the concentration range covered with radioactive GSH in Fig. 2 corresponds to the four points closest to the Y axis in The rneasurements were made at 22°C by equilibrium dialysis using 24 pM total concentration of ""S-labeled product acd 21 pM enzyme. The buffer, 50 mM sodium phosphate (PH 7.3), contained 1 mM dithioerythritol. Curvature is expected for a competition experiment involving two ligands, which each display a noncooperative hyperbolic binding isotherm, because the concentration of free labeled ligand increases with increasing concentration of competitor. The points in the graph represent total concentrations of competitor (glutathione); correction for the binding of glutathione to the enzyme does not affect the curve shape. The measurements were made at 22'C by equilibrium dialysis using 24 pM total concentration of '"S-labeled product, 21 PM enzyme, and 50 mM sodium phosphate buffer (pH 7.3). lated from directly measured binding of labeled competitor (I$ Fig. 2). On the other hand, it was found that the lines in the Scatchard plots curve to the right near the V axis (cf Fig.  1) when the concentration of the radioactive ligand was increased. These findings reveal low affinity binding in excess of the stoichiometry of 2 molecules of ligand per enzyme molecule. DISCUSSION The results reported in the present paper show that GSH, GSSG, bromosulfophthalein (an electrophilic substrate), as well as a product, S-(2-chloro-4-nitrophenyl)glutathione, bind noncooperatively to GSH S-transferase A with a stoichiometry of 2 molecules per enzyme molecule (1 molecule per subunit) when their concentrations are kept low. At high ligand concentrations a much weaker nonspecific binding appears to occur. Similar results were obtained by use of equilibrium dialysis in a study of the binding to ligandin (GSH Stransferase B) of ligands such as some steroids, bromosulfophthalein, and the conjugate N-methyl-4-aminoazobenzeneglutathione (13). However, a notable difference between the results of the two studies appears in the stoichiometry, which for GSH S-transferase B was reported as 1 molecule of ligand per enzyme molecule (13), whereas our data indicate that 2 molecules are bound per GSH S-transferase A molecule. An unequivocal explanation for the difference cannot be given, but the finding that the two subunits may be dissimilar in GSH S-transferase B (14, 15) could mean that in this protein only one subunit binds the ligands tested. The subunits of GSH S-transferase A, on the other hand, appear to be identical in their binding properties.
The mutual competition of the ligands (Figs. 3 to 5) indicate that they bind to the same or to overlapping sites on the GSH S-transferase A molecule.
Binding of Reactants to GSH S-Transferase A The finding that bromosulfophthalein displaces S-(2chloro-4-nitrophenyl)glutathione from the enzyme (Fig. 5) indicates that bromosulfophthalein binds to (at least a part of) the same site as the GSH derivative. The effect is not completely reciprocal, because S-(2-chloro-4-nitrophenyl)glutathione could only partly inhibit the binding of bromosulfophthalein (data not shown). Therefore, it appears probable that bromosulfophthalein can bind to different positions on the enzyme, some of which are inaccessible to the glutathione derivative. It has been suggested to explain inhibition experiments (7, 16) that the active center of the enzyme consists of a binding site for the peptide part of GSH and GSH derivatives (G site) as well as a hydrophobic site (H site) for the binding of the second hydrophobic substrate. The latter site would also bind hydrophobic S-substituents of GSH derivatives. The H site must be able to bind the variety of large and small substrate molecules, and the lack of effect of 3,4-dichloro-lnitrobenzene on the binding of product could mean that both the S-substituent of the product and the relatively small substrate molecule could be accommodated simultaneously. The evidence for the binding of the 2-chloro-4nitrophenyl group of the product to the enzyme comes from the about 20-fold higher strength of binding of S-(2-chloro-4-nitrophenyl)glutathione in comparison with GSH or GSSG (see Table I). Another explanation of the lack of effect of 3,4dichloro-1-nitrobenzene might be that this compound has a very low affinity for the enzyme. The assumption of more than one binding site for bromosulfophthalein per subunit is consistent with the complex curve shape in Fig. 5. Results indicating more binding sites for bromosulfophthalein than for other ligands have previously been reported for GSH Stransferase B (13). The binding of GSH and GSSG occurred with very similar dissociation constants (Table I). This finding lends support to the conclusion that they bind to the same site as previously indicated by their competition in kinetic studies (7,16). The similarity in binding strength also indicates that GSSG is bound only with one tripeptide moiety to the enzyme. The value for the dissociation constant of GSH (about 10 PM) implies that more than 99% of the two high affinity sites for GSH are saturated in the cytosol of liver cells, because the concentration of GSH is about 10 mu and the concentration of GSH S-transferase A is about 20 flM (1% of the soluble proteins (1)) (cfi Ref. 17).
The results of the binding experiments obtained in the region of higher reactant concentrations indicate that for all of the ligands studied more than two molecules can be bound per enzyme molecule. The measurements were normally restricted to the range of low concentrations, but in Fig. 1 this feature is evident. In some cases it was attempted to resolve the binding curve into two hyperbolas. However, no clearcut stoichiometry could be determined for the binding in excess of 2 molecules of ligand per molecule, and it, therefore, appears as if the additional binding is unspecific. Also in the case of GSH S-transferase B have "primary" and "secondary" binding sites been found, even if the stoichiometry of binding was different, as noted above (13). The competition experiments show that GSH and bromosulfophthalein displace the product of the enzymatic reaction S-(2-chloro-4-nitrophenyl)glutathione (Figs. 3 to 5) from the enzyme. The interaction of the ligands is of the generalized competitive type (18), indicating that they bind to the same site(s) (see Fig. 3). When the effect of variable concentrations of competitor was studied at a constant level of labeled S-(2-chloro-4-nitrophenyl)glutathione, the level chosen was such that 1.6 to 2.0 molecules of labeled ligand were bound per molecule of enzyme, which corresponds to the hyperbolic part of the binding isotherm. In the experiment when unlabeled GSH was used as competitor the data are plotted in a V versus V X [GSH] diagram (Fig. 4). Competition experiments with GSH at pH 9.4 (about half a unit above the isoelectric point of the enzyme) gave the same type of curve as at pH 7.3 (Fig. 4), indicating that the negatively charged GSH is not bound unspecifically at lower pH values owing to a positive net charge of the enzyme molecule. The binding of bromosulfophthalein was more complex (Fig.  5). Also in the case of bromosulfophthalein, the part of the curve corresponding to concentrations below those required for the saturation of the fast two binding sites of the enzyme molecule could be adequately approximated by a straight line. The possibility that the deviations from hyperbolic binding observed were due to an increase of the ionic strength, caused by increasing the ligand concentration, was ruled out by additional binding experiments carried out in the presence of 0.2 M NaCl. The interpretation of the competitive experiments involving GSH (G) and labeled S-(2-chloro-4nitrophenyl)glutathione (P) (Figs. 3 and 4) can be based on the binding scheme presented in Fig. 6. The scheme shows a dimer, which has two noncooperative binding sites, i.e. the microscopic rate constants for the binding of a ligand to a subunit are always the same irrespective of the occupancy of the other subunit. If only one ligand is present three enzyme forms vanish, and the binding equation becomes a simple hyperbola as found experimentally ( Figs. 1 and 2). If two ligands are present simultaneously the binding equation would appear to be a 212 function in the labeled ligand (P in Fig. 6)  A. G and P denote glutathione and S-(2-chloro-ri-nitrophenybglutathione, respectively. The binding equation degenerates to a first-degree expression after cancellation of the common factor of numerator and denominator. factor, and the expression degenerates to contain only firstdegree terms if no cooperativity exists. The binding equation of Fig. 6 was fitted by nonlinear regression methods to the data sets of Figs. 3 and 4 both by using the dissociation constants determined from binding curves with a single ligand (K1 = 0.5 pM, Kz = 7 pM, see Table I) and by using the dissociation constants as parameters in the regression. Both approaches gave good fits. The data were also analyzed under the assumption that the equilibrium constant for the binding of glutathione to the EP complex (&) was different from that for binding of glutathione to a subunit of the unliganded protein (&). Thus, three parameters were used in the curve fitting, but the inclusion of an additional parameter did not improve the fits sufficiently to allow rejection of the simpler model. The third constant (KS) was somewhat less than K2, which would have implied heterotropic cooperativity between the two ligands provided that the difference were significant. However, in view of the lack of cooperative effects of either ligand when studied separately, and considering the experiment,al error, such a heterotropic interaction cannot be supported by existing evidence. Thus, just as the binding studies involving a single ligand, the competition experiments fully support the conclusion that the two binding sites are independent.
In summary, the results show that the binding characteristics of glutathione S-transferase A are fully consistent with the steady state random kinetic mechanism which has previously been founded on initial rate and inhibition data (6, 16).