The Role of Glutathione in the Isomerization of Δ5-Androstene- 3,17-dione Catalyzed by Human Glutathione Transferase A1-1*

Human glutathione transferase (GST) A1-1 efficiently catalyzes the isomerization of Δ5-androstene-3,17-dione (AD) into Δ4-androstene-3,17-dione. High activity requires glutathione, but enzymatic catalysis occurs also in the absence of this cofactor. Glutathione alone shows a limited catalytic effect.S-Alkylglutathione derivatives do not promote the reaction, and the pH dependence of the isomerization indicates that the glutathione thiolate serves as a base in the catalytic mechanism. Mutation of the active-site Tyr9 into Phe significantly decreases the steady-state kinetic parameters, alters their pH dependence, and increases the pK a value of the enzyme-bound glutathione thiol. Thus, Tyr9 promotes the reaction via its phenolic hydroxyl group in protonated form. GST A2-2 has a catalytic efficiency with AD 100-fold lower than the homologous GST A1-1. Another Alpha class enzyme, GST A4-4, is 1000-fold less active than GST A1-1. The Y9F mutant of GST A1-1 is more efficient than GST A2-2 and GST A4-4, both having a glutathione cofactor and an active-site Tyr9 residue. The active sites of GST A2-2 and GST A1-1 differ by only four amino acid residues, suggesting that proper orientation of AD in relation to the thiolate of glutathione is crucial for high catalytic efficiency in the isomerization reaction. The GST A1-1-catalyzed steroid isomerization provides a complement to the previously described isomerase activity of 3β-hydroxysteroid dehydrogenase.

The biosynthesis of hormones such as testosterone, progesterone, and corticosteroids includes 3␤-hydroxysteroid oxidation followed by a ⌬ 5 to ⌬ 4 isomerization of the 3-ketosteroid product. In bacteria, such as Pseudomonas species, the two consecutive steps are catalyzed by two distinct enzymes, a 3␤-hydroxysteroid dehydrogenase and a 3-ketosteroid isomerase (1). Mammalian tissues contain 3␤-hydroxysteroid dehydrogenases, which have been found to catalyze also the isomerase reaction (2)(3)(4). Extending the studies of the isomerization (Fig. 1) of ⌬ 5 -androstene-3,17-dione (AD) 1 to mammalian tissues, Benson and Talalay (5) identified a glutathione (GSH)dependent enzyme in rat liver, which was later identified (6) with a major glutathione transferase (GST). The multiple forms of GST are divided into different classes, based on their primary structures. The different classes of soluble mammalian GSTs have been denoted 2 Alpha, Mu, Pi (8), Theta (9), Sigma (10), Kappa (11), Zeta (12), and Omega (13). The proteins are dimers, each subunit containing a largely conserved GSH-binding site (G-site) and a more promiscuous site (H-site) for the second, electrophilic, substrate. In GST A1-1 about 16 of the 25 residues in the cavity of the active center are regarded as H-site residues and the remaining 9 as G-site residues (14). The active-site Tyr 9 residue plays an important role in stabilizing and orienting the thiolate form of GSH in the conjugation reactions catalyzed by GST A1-1 (14,15). A crevice between the two subunits of soluble GSTs may serve as an additional binding pocket distinct from the active site (16,17).
Extensive studies have been made of the role of the GSTs in detoxication reactions between GSH and different electrophiles both in normal cells (18 -20) and in neoplastic cells where they possibly give rise to multiple drug resistance in tumors subjected to chemotherapy (21,22). Structural determinants governing protein folding and stability of GSTs have also been identified (23). The three-dimensional structures (24,25) and the catalytic functions (20) of naturally occurring GSTs have been intensively studied, and their redesign into enzymes with novel properties has been accomplished by means of protein engineering (26,27).
However, the GST-catalyzed isomerization of AD has not been subjected to detailed mechanistic studies despite its theoretical interest and its possible physiological significance. This GST reaction complements the activity of the bifunctional 3␤-hydroxysteroid dehydrogenase/⌬ 534 isomerase (3␤-HSD/ Iso), but its relative efficiency and the role of GSH in catalysis have not previously been evaluated.
In the present investigation the isomerization of AD catalyzed by the major human liver enzyme GST A1-1 has been studied in detail. This abundant isoenzyme was shown to be significantly more active with AD than the isomerase component of the bifunctional 3␤-HSD/Iso previously described. Marked differences among related Alpha class GSTs in their ability to catalyze the isomerization of AD as well as the significance of an ionized thiol group of GSH acting as a base in the catalytic mechanism were also demonstrated. A2-2, and GST A4-4 were used to produce enzyme in Escherichia coli. The GST A1-1 clones have been described in detail (15,28). The expression of wild-type GST A1-1 was based on the pET21 vector (Novagen, Inc., Madison, WI) into which it was subcloned by Ann Gustafsson. 3 cDNA encoding GST A2-2 (29) was amplified from a human hepatoma cDNA library and engineered into the pGTac⌬Eco vector (30) by Kristina Svensson in our laboratory. 4 These enzymes were expressed and purified using HiTrap SP cation exchange as described earlier for GST A1-1 by Gustafsson and Mannervik (31) with one or two additional steps of cation exchange chromatography. This procedure yielded highly pure and concentrated protein solutions as confirmed by SDSpolyacrylamide gel electrophoresis and spectrophotometric analysis. GST A4-4 was provided by Dr. Ina Hubatsch (32). Extinction coefficients used for enzyme subunit concentration determination: ⑀ ϭ 24,700 M Ϫ1 cm Ϫ1 for GST A1-1, Y9F, and GST A2-2 (33); ⑀ ϭ 15,900 M Ϫ1 cm Ϫ1 for GST A4-4. 5 Assays of Enzyme Activity-The assay system for the results presented in Table I, contained 50 mM Tris-HCl at pH 8.0, 1% (v/v) methanol as solvent for the substrate, 100 M AD and, when used, GSH or GSMe at 1 mM concentration. GST A1-1 subunit concentrations were 3.65 M in the cases with GSMe or no GSH and 7.3 nM in the case with GSH.
Specific activity measurements with CDNB (1 mM) were made in 100 mM sodium phosphate at pH 6.5 with 1 mM GSH and monitored at 340 nm. Specific activity measurements and steady-state kinetic experiments (except for the pH dependence studies) with AD (100 M) were made in 50 mM Tris-HCl at pH 8.0 with 1 mM GSH and monitored at 248 nm.
Parameter values presented in Table II, other than the  The reactions, in which the isomerization of AD in the presence of enzyme and GSH was measured, were initiated by the addition of the enzyme.
Kinetic studies were made on a Shimadzu UV-PC2501 spectrophotometer at 30°C. Regression analyses were made with the computer program package SIMFIT (34), and parameter values reported are given with S.D. values.
Competitive Substrate and Inhibition Experiments-Competition experiments were performed by measuring the conjugation of GSH with the alternative substrate o-CF 3 CDNB at 5 mM GSH with four different concentrations of AD (0, 75, 150, and 300 M) and o-CF 3 CDNB varied from 20 to 400 M in 44 measurements. o-CF 3 CDNB is a CDNB analogue that is monitored the same way as CDNB but which gives improved saturation kinetics (35). Similarly, inhibition of the isomerization reaction by GSHex was measured in a series of reactions where AD was varied from 20 to 300 M in 42 measurements at 5 mM of GSH and different concentrations of GSHex (0, 2, 4, and 8 M). The results of these measurements are presented in Table III.
Determination of the pK a Value of the GSH Thiol-The spectrum of free GSH in solution was obtained at different pH values, and the ionization of the thiol group was monitored by measuring the absorbance of the thiolate at 239 nm. Spectra of GSH bound to GST A1-1 and to Y9F were obtained by subtracting the spectra of free GSH at different pH values and of free enzyme at pH 6 from the spectrum of the sample. Concentration of enzyme was 9 M, and the GSH concentration was 250 M; measurements were done in 100 mM sodium phosphate.
pH Dependence of the Isomerization Reaction Catalyzed by GST A1-1-Kinetic parameters for GST A1-1 and Y9F were determined at different pH values with 2 mM GSH, at least 80% saturating in the crucial part of the pH range examined, and varying AD concentrations. Parameters for GST A1-1 in the absence of GSH were also obtained. Sodium phosphate buffers of 100 mM were used in the pH range 4.7-8. Above pH 8 and up to 9.7, 100 mM ethanolamine HCl was used.

RESULTS
GSH Dependence of the Enzyme-catalyzed Isomerization-In the absence of GSH and enzyme a slow but measurable conversion of 100 M AD into its isomer was observed (Table I). GSH (1 mM) afforded a 2-fold increased reaction rate at pH 8.0. The presence of GST A1-1 under conditions similar to those used for specific activity measurements, but in the absence of GSH, increased the rate of the isomerization about 16 times at 1 M enzyme concentration ( Table I). The presence of both GST A1-1 and GSH further increased the reaction rate 240 times so that the overall rate enhancement was about 3800-fold. Replacement of GSH by 1 mM GSMe gave an enzymatic activity lower than that in the absence of GSH, demonstrating that GSMe inhibits the GST A1-1-catalyzed reaction. The nonenzymatic isomerization rate was the same in the presence and absence of GSMe.
Optimization of Conditions for Specific Activity Determinations-Modifications of the original assay system defining the specific activity (5) increased the specific activity of GST A1-1 from ϳ10 to 40 mol min Ϫ1 mg Ϫ1 and brought the assay closer to physiological conditions. With the new conditions (50 mM Tris-HCl, pH 8.0, 1.0 mM GSH, 100 M AD, and 1%, v/v, methanol) the blank reaction, in the absence of enzyme, was low in comparison with the catalyzed reaction, and data were reproducible.
Steady-state Kinetics of Alpha Class GSTs with ⌬ 5 -Androstene-3,17-dione-Steady-state kinetic experiments were made in which both GSH and AD concentrations were varied independently. In addition, specific activities with both CDNB and AD were determined (Table II). The data were in agreement with a random sequential two-substrate mechanism for all GSTs investigated (cf. Fig. 2). The kinetic parameters were obtained by fitting Equation 1 to the experimental data.
The k cat value for GST A1-1 with AD, 29 s Ϫ1 , is about 50 times, 110 times, and 1200 times higher than the values for Y9F, GST A2-2, and GST A4-4, respectively. The  same order of magnitude except that for GST A2-2, which is 4 -10 times higher. The values for GST A1-1 and GST A4-4 are the same within the experimental error. The reaction rate enhancement afforded by GST A1-1, defined as the ratio of k cat /K m AD to the rate constant of the uncatalyzed reaction was calculated as (1.3 Ϯ 0.1)⅐10 10 M Ϫ1 .
Competing Alternative Substrate and Dead-end Inhibition-AD inhibited the GST A1-1-catalyzed GSH conjugation of the alternative substrate o-CF 3 CDNB in a linear competitive manner (Fig. 3). Nonlinear regression analysis, fitting Equation 2 to the experimental data, gave a K i AD value of 81 Ϯ 6 M (Table III).
The K m o-CF3CDNB was determined to 50 Ϯ 3 M, similar to the previously reported value (35).
Inhibition by GSHex of the GST A1-1-catalyzed isomerization of AD showed linear competition with respect to both AD and GSH. At a constant GSH concentration (5 mM) and varying AD concentrations, a K i GSHex value of 5.8 Ϯ 1.0 M was determined.
Ability of the Enzyme to Lower the GSH Thiol pK a upon Binding-Spectrophotometric titration of the GSH thiolate gave different pK a values for free GSH, GSH bound to GST A1-1, and GSH bound to Y9F (Fig. 4). Upon binding of GSH to wild-type GST A1-1, the thiol pK a of 9.2 is lowered by 2.5 pH units to 6.7, in agreement with the previous estimate of pK a 6.7 (37). The mutant Y9F only decreased the thiol pK a by 2.0 pH units to 7.2.
pH Dependence of the Isomerization Reaction Catalyzed by GSTA1-1-The pH dependence of the GST A1-1-catalyzed isomerization of AD, at saturating levels of GSH, shows two pK a values for k cat , at pH 6.1 and 9.5, and two for k cat /K m AD , at 6.1 and 8.7 (Fig. 5). The reaction rate of wild-type GST A1-1 in the absence of GSH is ϳ100 times 6 lower at pH 8.0 and shows only one pK a value at pH 7.8 for k cat and one at pH 8.0 for k cat /K m AD . The corresponding activity for Y9F with GSH is about 35 times lower than the wild-type activity and also displays only single pK a values, for k cat at pH 8.2, and for k cat /K m AD at pH 8.4, respectively, slightly higher than the values in the absence of GSH. At pH 8, in the absence of GSH, and with as much as 12 M Y9F, the reaction rate was lower than the nonenzymatic rate observed with AD in the absence of 6 Reaction rate data derived from Fig. 5A.   enzyme, indicating partial inhibition of the isomerization by binding of AD to the protein.

DISCUSSION
The ⌬ 5 -3-ketosteroid AD spontaneously rearranges to the ⌬ 4 -isomer product at a slow, but indeed, measurable rate. The presence of 1 mM GSH markedly raises this rate, while the same concentration of GSMe gives no detectable effect as compared with the rate observed with AD alone. Since the sulfur of GSMe is blocked by a methyl group, this suggests that the sulfhydryl group of GSH is important for the catalytic effect.
GST A1-1, in the absence of GSH, also increases the rate of the reaction as compared with either the presence of AD alone or the presence of AD together with GSH (Table I). The largest reaction rate enhancement, ϳ4000-fold, is obtained when AD, GSH, and GST A1-1 are all present (Table I). It should be noted that in tissues the enzyme concentration may be 100-fold higher (100 M) with a corresponding higher rate enhancement. However, when GSMe is substituting for GSH, the activity of GST A1-1 decreases to a level below the rate for GST A1-1 and AD alone. It is possible that GSMe binds to the enzyme with its methyl group entering the binding site for AD, thereby blocking entry of the steroid. An alternative explanation would be that the ⌬ 5 -3-ketosteroid transformation is attenuated indirectly by the interaction between enzyme and GSMe. The lack of catalytic effect of GSMe also demonstrates that GSH does not promote the reaction simply through a ligand-induced conformational change of the enzyme into a catalytically competent state.
The simultaneous action of GST A1-1 on AD and o-CF 3 CDNB, an alternative substrate more reactive than its analogue (35), CDNB, results in alternative substrate competition (Fig. 3) with an apparent K i AD value of 81 M (Table III) similar to the K m AD value of 58 M (Table II). The conjugation reaction is measured at pH 6.5 and the isomerization reaction at pH 8.0, but since K m AD values of the data presented in the k cat /K m AD curve (Fig. 5B) show no marked change with pH, these results suggest that the similar K i AD and K m AD values reflect the same interaction with the protein and that the alternative substrates AD and o-CF 3 CDNB compete for the same binding site.
GSHex is a competitive inhibitor of conjugation reactions catalyzed by GST A1-1. Thus, GSHex competes with GSH for the G-site; the hexyl group of the inhibitor binds into the H-site and obstructs the steroid from binding. This, together with the results from the alternative substrate experiments, indicates that the isomerization of AD takes place in the active center of GST A1-1 and that AD occupies the H-site. This rules out the possibility that catalysis takes place in the cleft between the two protein subunits (16,17). The conclusion is supported by results with the homologous rat GST A1-1 showing that AD does not protect against affinity labeling with steroid derivatives that target the nonsubstrate steroid binding site (38,39).
At saturating levels of GSH, the pH dependence profile for the GST A1-1-catalyzed isomerization of AD has a bell shape that suggests two ionizable groups to be involved in the catalysis. Since AD does not contain any protolyzable groups, the ionizations have to be ascribed the binary GSH-enzyme complex. The pK a value corresponding to the acidic limb of this curve is 6.1 for both k cat and k cat /K m AD (Fig. 5). Because the parameter k cat describes the pH dependence for the enzymesubstrate complex, while the parameter k cat /K m describes the condition where enzyme and substrate are free, the similar pK a values indicate that the binding of substrate does not affect the group responsible for the lower pK a . The thiol group of GSH bound to the wild-type enzyme has a pK a of 6.7 (Fig. 4), somewhat higher than the kinetic pK a of 6.1. Despite this apparent discrepancy the lower kinetic pK a value is ascribed to GSH, because complex mechanisms, including pH-dependent changes of rate-determining steps, may give apparent ionization constants of active-site residues (40). With the exception of Tyr 9 , which has a pK a value of 8.1 (41), there is no other ionizable group in the active site of GST A1-1. Thus, the data suggest that GSH is acting as a base in the isomerization reaction. For the basic limb of the pH profiles, a difference between the pK a values of the parameters k cat and k cat /K m AD was noted (9.5 and 8.7, respectively, Fig. 5). The higher value for k cat indicates that the second of the ionizable groups becomes more basic upon binding of AD, suggesting a more hydrophobic environment.
Without GSH present in the reaction of GST A1-1 and AD, there is only one pK a value at ϳ8 observed in the pH dependence profile. There is a small difference of 0.2 pH units between the estimated pK a values of the k cat and k cat /K m AD curves (Fig.  5), but the values are not significantly different considering the experimental variance. Tyr 9 of GST A1-1 has been assigned a pK a value of 8.1 in the absence of GSH and a pK a value of 9.2 in the presence of GSH (41). These figures correlate well with the pK a values observed in this work and lead to the hypothesis that the side chain of Tyr 9 is the second of the ionizable groups of GST A1-1 catalyzing the isomerization reaction. In the presence of GSH, serving as a base, Tyr 9 promotes catalysis in its protonated form. However, in the absence of GSH, Tyr 9 seems to adopt the role of the base in the catalytic mechanism. A structure of GST A1-1 with AD bound is not available, but the sulfur of GSH and the phenolic oxygen of Tyr 9 are adjacent in the known GST A1-1 structures (14). Thus, both of the proposed functional groups may interact with the reactive region of the AD molecule, allowing for alternative roles of Tyr 9 in catalysis.
A comparison of GST A1-1 and the active-site tyrosine mu-  4. GSH thiolate formation as a function of pH in the absence of enzyme, in the presence of GST A1-1, and in the presence of the active site mutant Y9F. Free GSH (E), GSH bound to wild-type GST A1-1 (q), and GSH bound to Y9F (छ) are shown. Nonlinear regression analyses of the three data sets gave pK a values of 9.17 Ϯ 0.04, 6.7 Ϯ 0.1, and 7.2 Ϯ 0.1, respectively, for the sulfhydryl group of GSH. tant Y9F first demonstrates that the k cat value for the isomerization of AD is decreased 48 times in Y9F (Table II). The K m GSH value is lowered 11 times, whereas the K m AD value is lowered only two times in Y9F. The k cat /K m GSH and k cat /K m AD values for wild-type GST A1-1 are 4 and 21 times higher, respectively, than the corresponding parameters of Y9F. Nonproductive binding of a substrate is expected to decrease both k cat and K m by the same factor, leaving k cat /K m unchanged (42). Thus, the marked effects of the Y9F mutation on k cat and K m GSH suggest that GSH to a large extent is bound in a catalytically nonproductive mode in the mutant. In wild-type GST A1-1 the proposed hydrogen bond between the sulfur of GSH and the phenolic hydroxyl group of Tyr 9 (14) could help orienting the orbitals of the thiolate for optimal catalytic efficiency. The loss of this steering effect in Y9F may account for a major portion of the decreased activity. However, it is clear that also other factors are involved.
The enzyme-bound GSH thiol group has a pK a value that is somewhat higher for Y9F (7.2) than the corresponding value (6.7) for wild-type GST A1-1. This is in agreement with the finding that the hydroxyl group of the active-site tyrosine in a rat Mu class GST contributes to the stabilization of the thiolate in the enzyme-GSH complex (43). However, the shift of the pK a value for GSH does not fully account for the altered pH profiles of the kinetic parameters k cat and k cat /K m AD with pK a values for Y9F 8.2 and 8.4, respectively (Fig. 5). This discrepancy suggests a pH-dependent change in rate-determining step or that at least one additional functional group is partaking in catalysis.
The k cat /K m GSH and the k cat /K m AD values of GST A2-2 are both lower than the corresponding values of GST A1-1 and Y9F. The k cat /K m GSH value is 36 times and the k cat /K m AD value is 480 times lower than for GST A1-1. By this measure GST A2-2 differs from Y9F about as much as Y9F is differing from GST A1-1. This finding contrasts with the notion that Tyr 9 , present in all the naturally occurring Alpha class isoenzymes, should play an important role in the isomerization of AD, in analogy with its significance in conjugation reactions. For example, GST A2-2 has a reasonably high specific activity with CDNB, while Y9F does not (Table II). Again the results indicate that there are amino acid residues in addition to Tyr 9 that are important for the isomerization reaction.
GST A2-2 differs from GST A1-1 by only 11 amino acids out of a total of 221 in a protein subunit. Four of these are situated in the active site and all of them in the H-site. The four mutations in GST A2-2 required to mimic the active site of GST A1-1 are S10F, I12A, F111V, and S216A. The two serine residues in GST A2-2, Ser 10 and Ser 216 , provide a less hydrophobic environment and may afford hydrogen bond interactions between substrate and enzyme. Both GST A1-1 and GST A2-2 have a phenylalanine in the active site but in different positions, Phe 10 and Phe 111 , respectively. In position 111 the residue may be too bulky, whereas in position 10 phenylalanine may actually contribute to productive binding of AD as seen for the Pseudomonas 3-ketosteroid isomerase (44). Furthermore, an isoleucine residue (Ile 12 ) next to the active site tyrosine in GST A2-2 may, for example, sterically hinder the large AD molecule, whereas the same position in GST A1-1 is occupied by the much less bulky alanine. The other seven differences between GST A1-1 and A2-2 involve residues that probably do not interact with the bound AD, even though they might still have an indirect influence on catalysis.
The reaction rate of GST A4-4 with AD is 3 orders of magnitude lower than that of GST A1-1. The structural differences between GST A1-1 and GST A4-4 are extensive, and their amino acid sequence similarity is only 53% overall. Thus, it is noteworthy that there is a measurable activity with AD despite the significant differences in primary structures of GST A4-4 and the other isoenzymes investigated in this work. The K m GSH and K m AD values of GST A4-4 are similar to the corresponding values of GST A1-1, and the major difference in catalytic efficiency between GST A4-4 and GST A1-1 is a reflection of the k cat values that differ by 3 orders of magnitude.
The bacterial 3-ketosteroid isomerase identified in Pseudomonas (1) is one of the most efficient enzymes known. Its catalytic mechanism has been studied in great detail. A key residue of the active site is Asp 38 , the carboxylate that shuttles a proton from C4 to C6 of the substrate AD in the isomerization reaction (45). A tyrosine residue, Tyr 14 , assisted by Asp 99 promotes the isomerization by hydrogen bonding and polarization of the 3-keto group of AD (46,47). Little is known about the catalytic mechanism of the bifunctional mammalian 3␤-HSD/ Iso, but it is noteworthy that mutational studies have shown two neighboring tyrosine residues, Tyr 253 and Tyr 254 , to be important for the isomerase activity (48).
The present study demonstrates the importance of two functional groups for the isomerization of AD catalyzed by GST A1-1. The most prominent role is played by the sulfhydryl group of GSH bound as a cofactor in the active site. It acts in its basic thiolate form and presumably relays a proton from C4 to C6 in AD, thus playing a similar role as Asp 38 in the bacterial isomerase. The second functional group is the phenolic hy- droxyl of Tyr 9 , which is active in its protonated form. This group may promote catalysis not only by stabilizing the GSH thiolate but also by polarizing the 3-keto group of AD. In the absence of GSH Tyr 9 takes over as a base instead of the thiolate of GSH, as indicated by the pH activity profile (Fig. 5) and the lack of activity of Y9F in the absence of GSH.
The primary goal of the present study was to clarify the role of GSH in the double-bond isomerization reaction, which is quite distinct from the previously investigated GST-catalyzed conjugation reactions leading to detoxication and excretion of reactive electrophiles. However, several lines of evidence suggest that the steroid isomerization effected by GST A1-1 may be physiologically important in mammalian tissues. Pregnenolone is an obligatory intermediate in the biosynthesis of steroid hormones, and in the available alternative metabolic pathways its 3␤-hydroxyl group will eventually be oxidized to a ketone, and its ⌬ 5 double bond will be converted to the resonancestabilized ⌬ 4 isomer. For example, a reaction sequence leading to testosterone involves the 3␤-oxidation of dehydroepiandrosterone to ⌬ 5 -androstenedione followed by the isomerization of the product into ⌬ 4 -androstenedione (2). In mammalian tissues the oxidation is catalyzed by several pyridine-nucleotide-dependent 3␤-hydroxysteroid dehydrogenases. Of the six homologous dehydrogenases identified so far, four NAD ϩ -dependent dehydrogenases are believed to be involved in the biosynthesis of steroid hormones (49). One of them is the principal form in gonads and adrenal glands. Other isoenzymes are expressed in organs such as liver and kidney. Although progress has been made in the study of the differential expression of the multiple forms in adult and fetal tissues, many issues concerning the catalytic properties and the physiological roles of the distinct isoenzymes remain unresolved. In Pseudomonas the 3␤-hydroxysteroid dehydrogenase is accompanied by a highly efficient ⌬ 5 -3-ketosteroid isomerase (1), and the two enzymes can be separated by chromatography. A similar efficient isomerase has not been identified in mammalian tissues, and at least some of the mammalian dehydrogenases have been shown to have intrinsic steroid isomerase activity (4,49) and are therefore referred to as bifunctional 3␤-hydroxysteroid dehydrogenase/⌬ 534 isomerases. The dehydrogenase reaction can be studied separately from the isomerization and vice versa, and it has been suggested that the two distinct catalytic activities reside in distinct active sites (48,51). However, unless there is efficient channeling of the product of the oxidation to the site of the isomerization reaction, there may be substantial leakage of the ⌬ 5 -intermediate to the surrounding medium. The higher K m values for the dehydrogenase expressed in liver and kidney as compared with that found in gonads and adrenal glands (52) suggest that the need for a supplementary steroid isomerase may be particularly high in the former organs.
The present study demonstrates that human GST A1-1 is highly efficient in catalyzing the isomerization of AD. This enzyme is one of the most abundant proteins in liver, kidney, and testis, in which it may exceed 2% of the total soluble protein (53,54). The catalytic efficiency of GST A1-1, 50⅐10 4 s Ϫ1 M Ϫ1 , determined in the present investigation, is obtained at saturating GSH levels (millimolar concentrations), which are achieved in most tissues under normal physiological conditions. For comparison, kinetic parameters are available for 3␤-HSD/Iso from human placenta (48) with a calculated specificity constant (k cat /K m ) of 1.6⅐10 4 s Ϫ1 M Ϫ1 with AD as the substrate. This value is 30-fold lower than that of GST A1-1. In view of the exceedingly high intracellular concentration of GST A1-1, its isomerase activity with AD may exceed by several orders of magnitude that of the 3␤-hydroxysteroid dehydrogenase, particularly in tissues such as liver and kidney. Recent studies 7 have shown that GST A1-1 is almost as active with ⌬ 5 -pregnen-3,20-dione, an intermediate on the pathway to progesterone.
In conclusion, the GST-catalyzed isomerization of AD provides a complement to the enzyme reactions previously associated with 3␤-hydroxysteroid dehydrogenases in mammalian tissues. The steroid isomerization is one of the most efficient of the known GST-catalyzed reactions involving an endogenous substrate. Other examples include 4-hydroxynonenal (32,55) and ortho-quinones derived from catecholamines (50,56), compounds that originate by oxidative processes. The common denominator for these reactions is that only one of the multiple homologous forms of GST has the characteristic high activity with the endogenous substrate, suggesting that the activity has evolved specifically for the catalyzed reaction.