Identification of N-Acetylcysteine as a New Substrate for Rat Liver Microsomal Glutathione Transferase A STUDY OF THIOL LIGANDS*

N-Acetyl-L-cysteine serves as an efficient substrate for the rat liver microsomal glutathione transferase with l-chloro-2,4-dinitrobenzene as second substrate (8.8 t 0.37 pmollmin mg). The activity is actually higher than that obtained with glutathione (2-4 pmollmin mg). In examining the activity of liver subcellular fractions, no activity with N-acetyl-L-Cys could be detected in dia- lyzed or N-ethylmaleimide-treated (in order to remove endogenous glutathione) cytosol. The activity in rat liver microsomes was 0.11 * 0.007 pmollmin mg, which is accounted for by the content of microsomal glutathione transferase. Thus, N-acetyl-L-Cys can be used as a specific substrate for determining the coqjugating activity of microsomal glutathione transferase. N-Acetyl-L-Cys was also shown to function as a substrate for the enzyme when other second substrates than l-chloro-2,4-dinitro-benzene (with varying electrophilicity) are used. The pH dependence of microsomal glutathione

thiols did not yield enzyme-bound complexes. The fact that the enzyme can stabilize Meisenheimer complexes from non-substrate thiol analogues of glutathione offers new possibilities for examining the substrate interactions of glutathione transferases.
Membrane-bound microsomal glutathione transferase (reviewed in Refs. 1 and 2) is characterized by its ability to be activated by sulfhydryl reagents (3) and trypsin (4). The enzyme has a unique amino acid sequence and immunological properties, in comparison to its cytosolic counterparts (5,6). On the functional level, as is the case for other members of the glutathione transferase group of enzymes (see Refs. 7 and 8 for reviews), it is involved in the detoxication of numerous carcinogenic, mutagenic, toxic, and pharmacologically active compounds (9). Polyhalogenated hydrocarbons and (phospho-) lipid hydroperoxides appear to be particularly interesting groups of substrates for the microsomal glutathione transferase (10,11).
In a previous examination of the structural features of glutathione required for activity and activation utilizing various glutathione analogues (12), we had hoped to see some striking differences to cytosolic glutathione transferases (13). However, no specific substrate was found. In the present study we report that N-acetyl-L-cysteine is a unique and efficient substrate for the microsomal glutathione transferase.
Meisenheimer complex stabilization by glutathione transferase has been shown previously to occur with glutathione and 1,3,5-trinitrobenzene (14). In the present study we have investigated the Meisenheimer complex stabilization by microsomal glutathione transferase with 1,3,5-trinitrobenzene and glutathione analogues. Quite remarkably, even analogues that are inactive as substrates for the enzymatic reaction give rise to Meisenheimer complex stabilization by the enzyme. This finding offers new possibilities for studying the substrate-enzyme interactions of glutathione transferases. The data were obtained by non-linear regression with the program Pennzyme (17) except in the case of N-acetyl-L-cysteine and activated enzyme where data from Lineweaver-Burk plots was used. Specific activities are expressed as means * S.D. (n = 3-5) using 10 m~ thiol substrate.  ously provided by Dr. Anton E. P. Adang (University of Leiden, The Netherlands). 1,3,5-Trinitrobenzene was a & from Nobel Chemistry (Karlskoga, Sweden). All other chemicals were of reagent grade and were obtained from common commercial sources.

Methods
Male Sprague-Dawley rat liver subcellular fractions were prepared according to (16). In order to remove endogenous glutathione, cytosol was dialyzed overnight against 0.25 M sucrose or treated with 1 m~ N-ethylmaleimide at 4 "C. In order to activate the glutathione transferase activity of microsomes, these were treated with 1 m~ N-ethylmaleimide at 4 "C prior to assay. Rat liver microsomal glutathione transferase was purified as described (3).
Activation of the purified enzyme with 2-5 m~ N-ethylmaleimide was performed at 4 "C in 10 m~ potassium phosphate, pH 8,1% Triton X-100, 0.1 m~ EDTA, 1 m~ GSH, 20% glycerol, and 0.1 M KCl. When maximal activity was reached (within 5 min.), the reaction was terminated by addition of GSH to give a final concentration of the free thiol of approximately 1 m~.
Glutathione was removed from the purified enzyme by dialysis against 10 m~ potassium phosphate, pH 7,O.l m~ EDTA, 20% glycerol, and 50 m~ KC1 (two daily changes). Removal was checked by testing Meisenheimer complex stabilization without the addition of GSH and took 4S72 h. The glutathione-free enzyme was activated by addition of 0.5 m~ NEM.'  with the glutathione-free enzyme (that had been activated with NEM) Enzyme inactivation with group-selective reagents was performed at mom temperature by the inclusion of diethyl pyrocarbonate (10 m~) , trinitrobenzenesulfonate (2 m~) , pyridoxal 5"phosphate (2 m~) , and phenylglyoxal (5 m~) .
These reagents are selective toward histidine, lysine, lysine, and arginine, respectively. NEM treatment of the enzyme prevented the known sulfhydryl reaction of trinitrobenzenesulfonate. Aliquots were withdrawn for assay at different time points as indicated in Fig. 3

( a d ) .
Enzyme Assays-Activities were assayed in 0.1 M potassium phosphate, pH 6.5, containing 0.1% Triton X-100 at 30 "C and varying glutathione analogue concentrations with 0.5 m~ l-chloro-2,4-dinitrobenzene. All experiments involved duplicate or triplicate determinations and were repeated at least once. kcat and K,,, values were derived by non-linear regression with the program Pennzyme (17). In the case of N-acetyl-L-cysteine and activated enzyme, these values are from Lineweaver-Burk plots. Activities with 4-substituted 1-chloro-2-nitrobenzenes and 1,2-dichloro-4-nitrobenzene (0.5-1 m~) were assayed in 0.1 M potassium phosphate, pH 6.5, containing 0.1% Triton X-100 at 30 "C with 10 m~ N-acetyl-L-cysteine at the wavelengths described in Ref. 18. The non-enzymatic rates were measured and subtracted from the observed enzymatic activities. In no case were they more than 10% of the enzyme-catalyzed rates (except for t-Cys: 550%).
The pH dependence of k,,JK,,, was determined in the following buff-

Substrates of Microsomal Glutathione Thansferase 73
em: 0.1 M citratdphosphate (pH 5.0-6.0), 0.1 M potassium phosphate (pH 6.5-8.0), 0.1 M Tris-HC1 (pH 8.5-9.0). All buffers contained 0.1% 'Riton X-100. kcllJKm was determined directly under first order conditions with one substrate et low concentration and the other at a saturating concentration. Preincubation at the different pH levels was employed to ascertain enzyme stability under the assay conditions. Except for activated enzyme at pH 5 (where activity decreases by 50% in 2 min), no decrease was observed during the normal assay period (30 8). Curves were fitted and pK, values determined where possible with the program UltraFit (Biosoft, Cambridge, United Kingdom). The formation constants and extinction coefficients for the Meisenheimer complexes were obtained by the spectrophotometric titration of the enzyme-GSH or enzyme-glutathione analogue complex (5-10 n m GSH or analogue was added to 40-60 p enzyme) with stepwise additions of 1,3,5-trinitrobenzene (0400 m) in 0.1 M potassium phosphate, pH 7, 1% Triton X-100,0.05 m EDTA, 10% glycerol, and 25 x m KC1 at room temperature as described (14). The background complex formation in the absence of enzyme (usually negligible) was measured with all analogues used and subtracted from the observed values in the presence of enzyme.
The apparent differences in binding energies of analogues and GSH were calculated according to Ref. 19.
Protein was determined by the method of Peterson (20) with bovine serum albumin as standard.

RESULTS AND DISCUSSION
The activities and kinetic parameters of rat liver microsomal glutathione transferase with various truncated glutathione derivatives and CDNB are shown in Table I

ND"
2.59 f 0.22 The decrease in catalytic efficiency for N-acetyl-L-Cys upon NEM-treatment of the enzyme agrees well with previous findings that the free amino function in the Y-L-G~u group of GSH is important for activation (12). In keeping with this prediction, activation did occur with Y-L-G~u-L-CYS.
The pH rate behavior of microsomal glutathione transferase with GSH, Y-L-GIu-L-CYS, and N-acetyl-L-Cys was investigated in terms of kc,JKm for both CDNB and the thiols. The k,,,/K,,, for the thiols at saturating CDNB concentration should reflect the pH behavior of the enzyme-CDNB complex and the free thiol substrate. It is evident (Fig. 1) that there were very small effects of pH on kcaJKmmcthiol,. "he conclusions that can be drawn from this experiment are: 1) no rate-determining ionizations of the free enzyme appear to occur between pH 5.5 and 8; 2) the deprotonation of the sulthydryl has little influence on kcat/K,,,, implying that the enzyme has no preference for the substrate in its thiolate anion form (except perhaps in the case of N-acetyl-L-cysteine). The k,,,/K, for CDNB at saturating thiol should reflect the pH dependence of the enzyme-thiol complex and in that way reveal any effect of the enzyme in lowering the p K a of these thiols (e.g. if no rate-determining titrating group is present on the enzyme itself, which appears to be the case). As can be seen in Fig. 2, the pKa of GSH was lowered to approximately 6, whereas the pK, values of y-~-Glu-t-Cys and N-acetyl-L-Cys were 8.5-9 and 27.5, respectively. Thus, the pKa for GSH in complex with the enzyme appears to be lowered from that of the parent compound (pKa = 8. indicates that the thiolate anion of this compound can substitute efficiently for glutathione and that the binding energy available in the glycine part of glutathione is utilized by the enzyme to lower the pK, of the thiol. Although it is technically difficult to measure rates at higher pH levels, the enzyme does not appear to be able to lower the pK, values of ~-L -G~u -L -C~S and N-acetyl-L-Cys to a great extent (if at all) and the pK, values might thus resemble those of the parent compounds. The activated and unactivated enzyme are affected similarly by pH, except in the case of k,,,/Km for CDNB with glutathione as the saturating substrate. In this case there is a sharper decline of the activated enzyme at low pH, actually becoming lower than the value for the unactivated enzyme. This effect can only be partly attributed to a lower stability of the activated enzyme at low pH. It is interesting to note that the differences between glutathione and Y-L-G~u-L-CYS resemble those that were obtained with cytosolic glutathione transferase 3-3 wild type and the in vitro mutant tyrosine to phenylalanine where the enzyme loses the ability to lower the pKa of glutathione (22). In our case the substrate change brings about the same effect. Cytosolic glutathione transferases purified from rat liver are active with Y-L-GIu-L-CYS, but not with N-acetyl-L-Cys (23). Rat liver cytosol and microsomes were used in order to test whether N-acetyl-L-Cys is a selective substrate for microsomal glutathione transferase. As can be seen in Table 11, no N-acetyl-L-Cys conjugating activity could be detected in cytosol that had been freed from endogenous GSH by dialysis or NEM treatment. Rat liver microsomes, on the other hand, displayed a high activity (Table II), which was fully accounted for by their content of microsomal glutathione transferase. Upon NEM treatment of microsomes, this activity was still significant, but decreased to a much lower level than expected from the activity decrease observed with the purified NEM-treated enzyme (approximately 10%). This could reflect the presence of an endogenous inhibitor (that has been suggested to be present in microsomes; Ref. 24) that inhibits N-acetyl-L-Cys conjugation by the NEM-treated enzyme by as much as 90%. Although it is known that the activated enzyme is more sensitive to inhibition (251, further studies regarding the nature of a possible inhibitor and inhibition characteristics of the purified microsomal glutathione transferase with N-acetyl-L-Cys as substrate are required.

Specific activities of rat liver microsomal glutathione transferase with 10 m M N-acetyl-L-cysteine and different second substrates
In order to investigate whether other substrates than CDNB with lower reactivity would also function as electrophilic acceptors for N-acetyl-L-Cys a range of structurally related compounds were employed (Table 111). All substrates tested were active together with N-acetyl-L-Cys, but, as reactivity of the analogues decreased, the rates were much lower than with GSH. The pattern of activation upon treatment with NEM is, in essence, the reverse of that obtained with GSH (26).
Our findings show that N-acetyl-L-Cys can be used as a selective substrate to study microsomal glutathione transferase and add to the list of unique features of this enzyme. This substrate can be of particular use when subcellular fractions  are studied, since significant levels of cytosolic glutathione transferases (as compared to those expected from simple contamination) are known to be present in microsomes (27) and mitochondria (28).

Thiol Substrates of Microsomal Glutathione Dansferase
Microsomal glutathione transferase, like other glutathione transferases (14), can stabilize a dead-end Meisenheimer complex between GSH and 1,3,5-trinitrobenzene (29). Titration of the enzyme-GSH complex with 1,3,5-trinitrobenzene yields the formation constant and extinction coefficient of the Meisenheimer complex (14). The formation constant obtained with glutathione (about 15 m"l) is large compared to that reported in the absence of enzyme (0.028 m -l ) (30) and is within the range of those reported for the cytosolic isoenzymes 3-3 and 4-4 (51 and 0.7 m"l, respectively) (14). In general, a correlation between the magnitude of the formation constant and the catalytic efficiency seems to exist (14). Meisenheimer complexes between glutathione analogues and 1,3,5-trinitrobenzene are also stabilized by the microsomal glutathione transferase (Table IV) and, quite remarkably, even with analogues that are not substrates for catalysis. Formation constants for good substrates are in general above a threshold value of 8 --', with the exception of glutaryl-L-Cys-Gly (3 m"'), which displays a formation constant similar to those of poor substrates and nonsubstrates. The absorbances of the Meisenheimer complexes obtained in the absence of enzyme are very low and have been subtracted. Differences between the activated and unactivated enzyme are, in general, small, indicating that the conformational change that accompanies activation (4) does not increase the complementarity of the enzyme to the dead-end complex and, to the degree that this restlnbles the transition state, does not appear to increase catalytic efficiency by increasing transition state complementarity. The extinction coefficients (Table  IV) (Table IV). L-Cysteinylglycine, N-acetyl+-Cys, cysteamine, and mercaptoethanol did not result in any detectable enzymatic contribution to complex stabilization, showing that the enzyme will not bind any Meisenheimer complex formed in solution to its hydrophobic binding site. The fact that non-substrate glutathione analogues can be utilized to probe the active site of glutathione transferases provides an additional tool for in-depth analysis of future in vitro mutagenesis experiments. The structural changes in glutathione and its analogues result in changes in k,,JK,,, and Kf, which can be used to calculate the corresponding changes in binding energies to the enzyme (8,19). It is of particular interest to compare the k,,/K,,, and Kf for the same pair of substrates. The a-amino group in the Y-L-G~u residue in GSH is, for instance, responsible for a binding energy of 3.2 kJ/mol (based upon comparing the k,,JK,,, values of glutaryl-L-Cys-Gly and GSH (12) with the unactivated enzyme), whereas comparison of the formation constants yields a value of 3.6 kJ/mol. The agreement of these values implies that the binding energy of the amino group is utilized equally effectively in complex stabilization and catalysis and contributes to catalysis in this manner. On the other hand, comparison of k c , & , values for yL-Glu-L-Cys and GSH indicates that the glycine moiety contributes 3 kJ/mol in binding energy, whereas removal of the glycine actually yields a tighter Meisenheimer complex (2.8 kJ/mol increase in binding energy). Clearly, changes in binding energies determined by kinetic and deadend complex binding studies are not comparable for gross structural changes, but appear to be quite accurate for quantification of more subtle substratdsubstrate analogue enzyme interactions. The value of being able to examine non-substrate ligands can be exemplified by comparing the Kf values for 4-Abu-~-Cys-Gly (which is not a substrate) and GSH, from which it can be estimated that the free carboxyl group of the Y-L-G~u residue stabilizes the complex to the unactivated enzyme by 4.4 kJ/mol.
Another valuable approach utilizing analogues of glutathione lies in the possibility of determining specific enzyme-substrate interactions, demonstrated here by the experiments involving group selective reagents. In Fig. 3d it can be seen that phenylglyoxal almost totally inactivated the microsomal glutathione transferase when glutathione or Y-L-G~u-L-CYS are used as substrates. On the other hand, the activity with N-acetyl+ Cys decreased by only 50% aRer 30 min of phenylglyoxal treatment. These observations indicate a possible function for an arginine residue in substrate binding through interaction with the yL-Glu residue of glutathione. Furthermore, since it has been observed (31) that phenylglyoxal can also react with lysine, it is interesting to note that lysine reagents did not yield the same result, indicating that both arginine and lysine residues are involved in substrate binding. Histidine and lysine reagents showed no difference in their inhibition characteristics with any of the substrates tested. The introduction of bulky amino acid-modifying reagents into the active site of an enzyme can, of course, be a much more drastic change than substitution of active site amino acids by in vitro mutagenesis. Therefore,