On the Multiplicity of Rat Liver Glutathione ,%Transferases*

Rat liver glutathione S-transferases have been purified to apparent electrophoretic homogeneity by S- hexylglutathione-linked Sepharose 6B affinity chromatography and CM-cellulose column chromatography. At least 11 transferase activity peaks can be resolved including five Yb size homodimeric isozymes, two Y, size homodimeric isozymes, one Y, homodimeric isozyme, one Y, homodimeric isozyme, and two Y,-Y, heterodimeric isozymes. Distribution of the GSH peroxidase activity among the CM-cellulose column frac- tions suggests the existence of further multiplicity in this isozyme family. Substrate specificity patterns of the Y b subunit isozymes revealed a possibility that each of the five Yb-containing isozymes is composed of a different homodimeric Yb size subunit composition. Our findings on the increasing multiplicity of glutathi- one S-transferase isozymes are consistent with the no-tion that multiple isozymes of overlapping substrate specificities are required to detoxify a multitude of xenobiotics in addition to serving other important physiological functions.

Rat liver glutathione S-transferases have been purified to apparent electrophoretic homogeneity by Shexylglutathione-linked Sepharose 6B affinity chromatography and CM-cellulose column chromatography. At least 11 transferase activity peaks can be resolved including five Yb size homodimeric isozymes, two Y, size homodimeric isozymes, one Y, homodimeric isozyme, one Y, homodimeric isozyme, and two Y,-Y, heterodimeric isozymes. Distribution of the GSH peroxidase activity among the CM-cellulose column fractions suggests the existence of further multiplicity in this isozyme family. Substrate specificity patterns of the Y b subunit isozymes revealed a possibility that each of the five Yb-containing isozymes is composed of a different homodimeric Yb size subunit composition. Our findings on the increasing multiplicity of glutathione S-transferase isozymes are consistent with the notion that multiple isozymes of overlapping substrate specificities are required to detoxify a multitude of xenobiotics in addition to serving other important physiological functions.
The glutathione transferases (GST,' EC 2.5. 1.18) are a family of dimeric proteins that are multifunctional in drug biotransformation and in xenobiotic metabolism (see Ref. 1 for a review). Historically, rat hepatic GSTs, which are abundant cytosolic proteins, have been the most extensively studied. But GSTs from other rat tissues also have been purified and analyzed (1). The expression of rat GSTs has been shown to be tissue-specific (2-4). Two major classes of liver subunits, Y, (Mr = 25,600) and Y, (M, = 28,000), are expressed in kidney GSTs but not in heart, lung, spleen, seminal vesicles, and testis GSTs (5, 6), while the liver Yb class subunits are not expressed in kidney GSTs (2, 3). Improved purification procedures and advanced molecular cloning experiments in this area have enabled us to develop a concept of the multiplicity of GSTs that far exceeds the three-subunit hypothesis (Ya, Yb, and Y,) proposed by Bass et al. (7). Recent results * This research project has been supported by United States Public Health Service Grant ES 02678. 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.
from several laboratories, including ours, have revealed that Y, and Yb each may represent a family of ClOSelY related subunits. We have reported on the Y, subunit sequence microheterogeneity in the two cDNA plasmids, pGTR261 and pGTB38 (5, 8). Sheehan and Mantle (9) have reported the presence of a second Y, dimer GST (GST-F) in addition to the basic Y, dimer GST (GST-L) (10). Mannervik and Jensson (11) have purified three basic Yb-containing GST isozymes that they suggest are products of two different Y b subunits as two homodimers and one heterodimer. Hayes' results (12) on peptide fingerprinting and subunit resolutionreconstitution support the proposal that GST-C may be a heterodimer of GST-D and GST-A. In this communication, we present results on the multiplicity of Y b -and Y,-containing GSTs from rat liver, and we suggest that high multiplicity of GSTs with overlapping substrate specificities may be essential to their multiple roles in xenobiotics metabolism, drug biotransformation, and protection against peroxidative damage.

EXPERIMENTAL PROCEDURES
Chemicak and Enzyme Reagents-Chemicals and substrates for GST and GSH peroxidase assays were as described in previous publications (2, 13).
Enzyme Assays-GSH peroxidase activity was measured spectrophotometrically by Paglia and Valentine's procedure (14) as modified by Reddy et al. (13), with cumene hydroperoxide as the substrate. The GST activities were determined with l-chloro-2,4-dinitrobenzene (CDNB) and other compounds according to published procedures (15).
Purification of GSTs from Rat Liver Cytosol-Livers (200 g) from male Wistar rats (-500 g, body weight) were used in this enzyme preparation. The procedures are those used to purify GSTs from sheep liver and other rat tissues as published earlier (2, 13), except that two linear gradients, 0-50 mM KC1 and 50-200 mM KC1, in 10 mM Napi, pH 6.0, were used in succession to elute the various isozymes from the CM52 column. The peak fractions of each GST activity peak against CDNB were analyzed by NaDodS04-polyacrylamide gel electrophoresis as described (16).

RESULTS AND DISCUSSION
Multiplicity of Y b Subunit GST Isozymes-Ten GST activity peaks are resolvable by the CM-cellulose chromatography with the 0-50 mM KC1 buffer gradient. They are represented by fractions 15, 32, 44, 76, 89, 97, 118, 124, 146, and 162 in Fig. 1 and designated as peaks I-X. An additional isozyme activity eluted by 50-200 mM KC1 buffer gradient, fraction 210, has been reported as the Ye-subunit GST isozyme of rat liver (17). From the NaDodS04-polyacrylamide gel electrophoresis pattern in Fig. 1, fractions 15, 32, 44, 89, and 124 represent isozymes containing Yb size subunit(s). Peak 111 (fraction 44) can be further resolved into two homodimeric isozymes by DEAE-cellulose chromatography: an isozyme of y b size subunitb) and an isozyme of Y. size subunit(s).' Peak I isozyme (fraction 15), which we referred to as the anionic GST isozyme(s), is a mixture of microheterogeneous GST isozymes because of its asymmetric GSH peroxidase activity peak. The leading fractions have GSH peroxidase activities, but the trailing fractions of the peak contain little or no GSH peroxidase activities against cumene hydroperoxide. The asymmetric pattern remains after subsequent DE52 column chromatography (18). Two distinct GST peaks with only one C. C. Reddy and C.-P. D. Tu, unpublished results.  containing GSH peroxidase activity, however, were obtained by high performance liquid chromatography on a sulphopropyl cation exchange column.3 The other three peaks represented by fractions 32 (peak 11), 89 (peak V), and 124 (peak VIII) are also isozymes of Y h subunits. Although the specific activity of fraction 32 against CDNB is intermediate between that of fraction 89 and 124, fractions 89 and 124 are not directly comparable with those of the three "A,C type" of GSTs reported by Mannervik and Jensson (11). According to these authors, GST-A2 has the highest specific activity toward CDNB and DCNB.
We would suggest an alternative explanation that each of the isozymes in Peaks 11,111, V, and VI11 may be a homodimer of different Y b subunits with similar structural and immunological properties. The peak I (fraction 15) GST is composed of a mixture of isozymes with similar chromatographic properties but very different GSH peroxidase activities. This peak may contain GST-D activities and fractions N1 to N3 as described by Hayes (12). No direct comparisons can be made, however, since Hayes did not report any GSH peroxidase activities for any of his purified GSTs. The proposed Y h multiplicity in rat liver GST isozymes can be supported by results from different laboratories. For example, reconstitution of urea or guanidine HCl denatured GST-C (presum-ably a YblYbz heterodimer) did not result in a 1:2:1 (Yb~Yb~:Yb~Ybz:YbzYbz) ratio of CDNB conjugation activity after CM-cellulose chromatography. Tryptic peptides fingerprinting experiments revealed more peptides for the putative heterodimeric isozyme than the combination of peptides from the two corresponding putative homodimeric isozymes (12,19). Two-dimensional gel electrophoretic separation of GST subunits revealed at least four spots of Yb size mobility. Their levels of expression are perturbed during chemical hepatocarcinogenesis (20). These results are consistent with our observed Yb subunit multiplicity.
Multiplicity of Y, Family Subunits-The results in Fig. 1 reveal two isozyme peaks of Y, homodimers (fractions 146 and 162) and at least two Y.-Y, heterodimers (fractions 97 and 118). The Y, size subunits in the two Y.-Y, heterodimeric isozymes may not be of identical electrophoretic mobilities because fraction 110, which is at the common region of peaks VI and VII, has three bands after electrophoresis. Substrate specificity patterns of fractions 76, 97, 118, 146, and 162 in Table I did not reveal additivity of single subunit activities as assumed by Mannervik and Jensson, although the majority of the heterodimeric isozyme activities are intermediates of the Y, and Y, homodimeric isozymes. It is possible that the Y, and Y, subunits in the heterodimeric and homodimeric isozymes may not be identical.
The two Y, homodimeric isozymes have different substrate specificity patterns. For example, peak IX (fraction 146) has higher activities than Peak X (fraction 162) against CDNB, ethacrynic acid, and t-butyl hydroperoxide, but lower activities than Peak X against DCNB, 1,2-epoxy-3-(p-nitrophenoxy)propane, 4-nitropyridine N-oxide, A5-androstene-3,17dione, cumene hydroperoxide, and linoleic acid hydroperoxide. They are equally active against p-nitrobenzyl chloride andp-nitrophenyl acetate. Our resolution of two homodimeric Y, isozymes and two heterodimeric Y.-Y, isozymes may also explain the CDNB activity patterns in chromatofocusing separation observed by Mannervik and Jensson (11). They did not show the corresponding protein profile probably due to peptides used for elution. But, their peak I11 (GST-Bz), designated as Y,Y, homodimer, is unusually broad and contains Y, size subunit as shown in their NaDodS04-gel electrophoresis pattern. We suggest that the broadness of their peak I11 is consistent with the existence of multiple isozymes and that their peak I11 may contain a second Y.Y, heterodimeric isozyme and a second homodimeric Y, isozyme in addition to the assigned GST-Bz (or GST-AA according to Jakoby (21)).
The substrate specificity patterns of the three "B,L type" isozyme showed more deviation from the assumption of subunit activity than the corresponding A,C type isozymes. This is again consistent with our observed isozyme multiplicity. Our CM-cellulose chromatographic pattern of liver GST isozymes may be compared with Jakoby et al.'s (21) scheme of six isozymes, E, D, C, B, A, and AA, according to their order of elution off the CM-cellulose column. Recently, Meyer et al. (22) reported that GST-E (another Yb-containing isozyme) did not bind to the S-hexyl-GSH-linked Sepharose-GB column. Therefore, we should not have GST-E in Fig. 1 Further Multiplicity of Rat Liver GSTs-Since the GSTs are more than 95% pure after the S-hexyl-GSH affinity column chromatography as judged by their NaDodS04-gel electrophoresis patterns, subsequent resolution on a CMcellulose column should be considered a more efficient resolution procedure than chromatofocusing. In addition to the 11 major activity peaks, several fractions in Fig. 1 suggest the existence of minor isozymes or isozymes not completely resolved by the current condition. Two such examples can be identified around fraction 65 and between fractions 100 and 110. The former contains both GST and GSH peroxidase activities, but the latter fractions contain only GST activity against CDNB. Some other isozymes could be "hidden" be- neath the many partially resolved peaks between fractions 76 and 140 in Fig. 1.
The protein purification and substrate specificity patterns presented here, together with results from many other laboratories, suggest a much higher multiplicity of liver GST isozymes than previously conceived. We can classify liver GST subunits into at least four classes, Y,, Yb, Y,, and Y, based upon the electrophoretic mobilities observed by Na-DodS04-polyacrylamide gel electrophoresis. The cDNA cloning results have probably revealed a level of microheterogeneity not resolvable by conventional protein purification techniques (5, 8). Considering the non-liver GST subunits that are tissue-specific, the multiplicity of GST isozyme subunits may well approach that of other drug-metabolizing enzymes. Each organism has to cope with many environmental xenobiotics and has to evolve adequate defense mechanisms. Being a conjugation isozyme family, GSTs have been classified as one of the phase I1 detoxification enzymes. Each of the GST isozymes has a limited substrate specificity pattern against many synthetic chemicals such as CDNB, DCNB, etc. Therefore, it is essential to have evolved a large number of different subunits with partially overlapping substrate specificity patterns for efficient metabolism of xenobiotics.