Control of activity through oxidative modification at the conserved residue Cys66 of aryl sulfotransferase IV.

Oxidation at Cys66 of rat liver aryl suflotransferase IV alters the enzyme's catalytic activity, pH optima and substrate specificity. Although this is a cytosolic detoxification enzyme, the pH optimum for the standard assay substrate 4-nitrophenol is at pH 5.5; upon oxidation, the optimum changes to the physiological pH range. The principal effect of the change in pH optimum is activation, which is manifest by an increase in K'cat without any major influence on substrate binding. In contrast, with tyrosine methyl ester as a substrate, the enzyme's optimum activity occurs at pH 8.0; upon oxidation, it ceases to be a substrate at any pH. The presence of Cys66 was essential for activation to occur, thereby providing a putative reason underlying the conserved nature of this cysteine throughout the phenol sulfotransferase family. Mapping of disulfides by mass spectrometry showed the critical event to be the oxidation of Cys66 to form a disulfide with either Cys232 or glutathione, either one is effective. These results point to a mechanism for regulating the activity of a key enzyme in xenobiotic detoxication during cellular oxidative stress.

A sulfotransferase prepared in homogeneous form from the cytosol of rat liver incongruously has at least two very different pH optima depending on the test substrate. The activity toward the phenolic substrates, 2-naphthol and 4-nitrophenol for example, is highest near pH 5.5 but very poor in the physiological range where sulfation of the substrate tyrosine methyl ester is optimum (1,2). The enzyme, designated variously as aryl sulfotransferase IV (1,3,4) and tyrosine-ester sulfotransferase (EC 2.8.2.9) (5, 6), catalyzes the transfer of the sulfuryl group of PAPS 1 to any of a wide range of low molecular weight phenols, hydroxylamines, alcohols, and amines (1,3,4,7). This broad specificity and affinity for lipophilic compounds is characteristic of the enzymes of detoxication, the group of enzymes that prepares xenobiotics for ready excretion (8). The availability of the recombinant liver enzyme from Escherichia coli (6) has allowed the extension (4) of previous studies of mechanism (3) and provided the impetus for an examination of the apparently low activity and inappropriately low pH optimum of an enzyme believed to function in the cytosol.
The sulfotransferase reaction can be measured by two spectrophotometric assays reliant on the absorbance of 4-nitrophenol at pH 7. The first assay is in the physiological direction with PAPS as the sulfate donor and 4-nitrophenol as the acceptor resulting in the products 4-nitrophenyl sulfate and PAP (Equation 1) (2). The second assay is the two-stage transfer of the sulfuryl group from 4-nitrophenyl sulfate to form PAPS in the presence of PAP and the subsequent transfer to 2-naphthol as the acceptor phenol (Equation 2) (2). Under the assay conditions used, activity in the physiological direction was almost 10-fold lower than in the artificial transfer reaction. Upon prolonged storage, enzyme preparations were found to modestly increase their physiological activity at the expense of the transfer activity. These changes were abolished by addition of EDTA or thiols and were induced by disulfides. Significantly, glutathione disulfide, the major form of cellular oxidized disulfide (9,10), was an efficient activator.
Each monomer of the homodimeric sulfotransferase contains five cysteine residues. The first of these from the amino-terminal, Cys 66 , is highly conserved among phenol sulfotransferases (11,12). Although a role for the cysteine residues in these enzymes has been suspected, their function has remained elusive. In N-ethylmaleimide inactivation of a rat liver enzyme, two of the five cysteine residues were found to react with the reagent and were protected from reacting by both PAPS and 4-nitrophenol (13). Replacement of Cys 70 , the first cysteine residue from the amino-terminal end in the human form of phenol sulfotransferase, with serine resulted in the loss of N-ethylmaleimide inactivation, indicating that this cysteine was not essential for catalytic activity (11). Participation of the first cysteine was suggested in the binding of the nucleotide substrate, since it was found that ATP-dialdehyde, a PAP analogue and affinity label, was covalently linked to the rat enzyme at Cys 66 and the adjacent Lys 65 (12).
Here, we describe a sequence of changes in the enzyme's activity profile brought about by oxidation of the cysteine residues of aryl sulfotransferase IV. Replacement of each of the five cysteine residues by site-directed mutagenesis, in combi-nation with mapping of disulfides by mass spectroscopic methods, has provided details of the underlying mechanism.

EXPERIMENTAL PROCEDURES
Materials-Hydroxyapatite-agarose (SpectraGel/HA) was purchased from Spectrum (Los Angeles, CA); 4-aminobenzenesulfonyl fluoride and dithiothreitol from ICN Biomedicals (Aurora, OH) and bis-tris propane from Calbiochem; sucrose and glycerol were from J. T. Baker Inc., and all other compounds not designated were at the highest grade available from Sigma. Thin layer chromatography plates of DEAE-cellulose (number 13255) and Silica Gel (number 13179) were from Eastman Kodak Co.
Sulfotransferases-Recombinant wild type and mutant sulfotransferases were isolated from E. coli BL21(DE3) bearing the pET3c11 vector (6). The enzyme isolation procedure was the same for all of the sulfotransferase preparations (2) and resulted in greater than 95% homogeneity for each as determined by SDS-polyacrylamide gel electrophoresis (14). Cys 3 Ser mutants were prepared by site-directed mutagenesis using polymerase chain reaction amplification of oligomers of 30 nucleotides with a single base substitution as the mutant primer (6). The expected mutation was confirmed by sequencing the entire mutant cDNA. Of the two forms of the enzyme, only the ␤ form, which is free of bound PAP (2), was used here. Since the enzyme was not expressed by E. coli bearing the plasmid for mutant C232S, a double mutant containing C232S and M228I was used.
Enzyme Assays-The details of the spectrophotometric standard transfer assay and the physiological assay have been presented (2). The standard assay (Equation 2) measures the exchange of the sulfuryl group from 4-nitrophenyl sulfate to 2-naphthol in the presence of 20 M PAP at pH 7.0 and is determined by measurement of the formation of 4-nitrophenol (⑀ 400 ϭ 10.5 ϫ 10 3 at pH 7.0). Note that two reactions are actually involved: the initial generation of PAPS, the sulfate of which, in turn, is transferred to a second phenol (2). In the physiological assay (Equation 1), 4-nitrophenol is conjugated with PAPS as the donor. All reactions are all carried out at 25°C in a Hitachi U3110 spectrophotometer in cuvettes of 1-cm light path. Rates were linear with time and protein concentration when average absorbance changes at 400 nm of less than 0.025/min were followed for 3 min.
Both of these assays are limited to the pH range of 6.5-7.5, because 4-nitrophenol has a pK a of 7.14. This limitation was avoided in the physiological tracer assay by the inclusion of radioactive [ 35 S]PAPS in Reaction 1 along with 1 mM acceptor substrate in a final volume of 100 l containing succinate-NaOH or bis-tris propane-HCl at the required pH values. The reaction was initiated by addition of PAPS (740 m) containing 0.1 Ci [ 35 S]PAPS per vessel and carried out at 37°C for 10 min before termination with 25 l of 2 M acetic acid. After centrifugation, a 3-l aliquot was used for the separation of 35 S-labeled product from [ 35 S]PAPS by thin layer chromatography (4,15). Radiolabeled entities were quantified by phosphorimaging of the TLC plates with a Fuji BAS-1500 Phosphorimager. The reaction was linear with respect to time and concentration over the range 0.02-20 nmol product formed per incubation mixture.
Incubations with GSSG and GSH-Sulfotransferase solutions were diluted to a volume of 1 ml containing 100 mM bis-tris propane-HCl at pH 7.0, 1 mM EDTA, 12.5% (v/v) glycerol, and 125 mM sucrose (buffer A). The preparation was exchanged through a Pharmacia PD-10 Sephadex G25 column equilibrated with buffer A to remove reducing agent present from the storage buffer and diluted to 7.5 M enzyme in the same buffer based on an ⑀ 280 of 6.07 ϫ 10 4 that is assumed to be identical for wild type enzyme and mutants.
Enzyme was incubated with 1 mM GSSG, GSH, or with various ratios of GSSG:GSH at 25°C; higher temperatures led to precipitation of the enzyme. Samples were withdrawn at intervals for measurement of activity in the standard and physiological spectrophotometric assays and for analysis by SDS-polyacrylamide gel electrophoresis (14). In a separate series of trials, PAP and 4-nitrophenol, each at concentrations between 0.2 and 200 M were added during incubation with GSSG.
Determination of KЈ m and K d -The apparent K m for PAP and 4-nitrophenyl sulfate were determined by the transfer assay procedure and, for PAPS and 4-nitrophenol, by the physiological assay. Sulfotransferase was incubated for 1 h with 1 mM GSSG at 25°C or with 1 mM DTT before addition of substrates. Data were analyzed with the En-zymeKinetics (Trinity Software) program.
K d was estimated at 25°C with a Hitachi F2000 spectrofluorimeter by following changes in fluorescence due to the binding of ANS to the enzyme (16). Sulfotransferase, 750 nM, was incubated at 25°C in a quartz cuvette of 1-cm cross-section with 10 M ANS in 100 mM bis-tris propane-HCl at pH 7.0. Fluorescence was measured at 465 nm after excitation at 380 nm and observed as a decrease for the addition of all ligands with the exception of an increase in the case of PAP binding. For the determination, reduced enzyme was prepared by measurement in 1 mM DTT; oxidized enzyme was prepared by incubating a 7.5 M protein solution for 1 h with 1 mM GSSG. Where necessary, corrections were made for dilution; absorbance was sufficiently low that correction for inner filter effects was not required.
The binding of ligands to multiple sites, which is the case here, has been described by Knott based on the Scatchard equation, and using the PCMLAB program (17). Fluorescence titration curves of sulfotransferase by ANS may be analyzed by a simple extension of this technique. The dye is assumed to bind to a number of independent sites, on the protein, each with a characteristic association constant and a change in intensity of fluorescence. Experimental data were fit to yield the values of the association constants and intensity changes that are presented in Table I. In all cases, good fits were obtained assuming the presence of only two binding sites. When the change in fluorescence was a linear function of added dye concentration, the binding was assumed to be stoichiometric, i.e. with an association constant Ͼ 15,000 M Ϫ1 .
Titration of Protein Thiol-The number of available free cysteine thiol groups of the sulfotransferase were estimated after exchange through a PD-10 column into buffer A. Enzyme at a concentration of 7.5 M was incubated with 0.3 mM DTNB at 25°C. Formation of the 2-nitro-5-thiobenzoate ion was followed at 412 nm (⑀ ϭ 14,150 M Ϫ1 cm Ϫ1 ) (18).
Preparation of Samples for Mass Spectrometry-Sulfotransferase, 7.5 M in buffer A, was incubated with 1 mM GSSG. The reaction was stopped at 1, 4, and 24 h by exchange through a PD-10 column into 10 mM HCl. Under conditions of proteolysis by trypsin, however, the reaction with GSSG was stopped for this purpose by alkylating remaining thiol groups by incubation with 45 mM 4-vinylpyridine at 25°C for 1 h in the dark; this was adequate to assure elimination of migration of disulfide bonds. After dialysis against 50 mM ammonium bicarbonate at pH 8, the protein was cleaved by incubation at 37°C for 2 h with 2% (w/w) trypsin. Control samples were treated in the same manner. All samples were lyophilized before being dissolved in 0.05% trifluoroacetic acid.
Mass Spectrometry-Electrospray ionization mass spectrometry was performed on a Fisons Instruments VG Quattro II triple quadrapole mass spectrometer (Manchester, UK) fitted with a Megaflow electrospray source. The Quattro II was controlled by a digital DECpcLPx 466d2 computer running Fisons MassLynx 2.1 software. The LC interface to the mass spectrometer consisted of Shimadzu LC10AD solvent delivery modules.
Flow injection analysis of the complete protein was conducted by injecting samples into a 40 l/min flow of 50% acetonitrile, 50% water, which carried the sample plug into the electrospray source. ESI-MS instrument settings were as follows: source temperature ϭ 100°C; N 2 drying gas ϭ 500 liters/h; nebulizing gas ϭ 20 liters/h; probe voltage ϭ 4 kV; cone voltage ϭ 33 V; probe position: on axis. Acquisition was carried out from m/z 200 -2000 Da over 4.5 s in a CONTINUUM scanning mode. The calibration file was created for analysis of protein using a solution of myoglobin. Deconvolution of ions was accomplished using the "Maximum Entropy Program" (MaxEnt, Fisons, Manchester, UK).
Peptides obtained from tryptic cleavage were separated by reverse phase HPLC using an Ultrasphere C18 column (2.0 mm ϫ 25 cm; Beckman, Fullerton, CA). Elution of the peptides was accomplished with solvents A (0.05% trifluoroacetic acid in water) and B (0.05% (v/v) trifluoroacetic acid in 90% acetonitrile and water); the initial concentration of solvent B was increased linearly from 5 to 60% in solvent A during 80 min at a flow rate of 200 l/min. The column eluate was monitored at 215 nm using a Shimadzu SPD-10AV module detector.
ESI-MS instrument settings were as follows: source temperature ϭ 150°C; N 2 drying gas ϭ 500 liters/h; nebulizing gas ϭ 20 liters/h; probe voltage ϭ 3.6 kV; cone voltage ϭ 35-50 V; probe position: off axis. Acquisition was carried out from m/z 200 -2000 Da over 4.5 s in the CENTROID scanning mode, with unit resolution up to at least 1500 Da based on calibration and resolution optimization using polyethylene glycol at M r 300, 600, and 1000.
The identity of GSH-conjugated peptides was confirmed by LC/MS and tandem MS/MS, using multiple reaction monitoring to specifically look for the neutral ion loss of 129 Da from the parent ion under collision-induced dissociation conditions (19).

Effect of GSH and GSSG-
The aryl sulfotransferase catalyzed the transfer of sulfate from 4-nitrophenyl sulfate to 2-naphthol in the transfer assay system (Equation 2) with a specific activity of approximately 600 nmol/mg/min. In comparison, the activity toward 4-nitrophenol with PAPS as the sulfate donor in the physiological assay system (Equation 1) was 10-fold lower at about 60 nmol/mg/min. Modest increases in the physiological activity could be achieved by incubation of enzyme in buffer A at 25°C for several hours. These increases were eliminated by inclusion of 1 mM concentrations of EDTA, DTT, or GSH, suggesting involvement of an oxidative process. When the enzyme was incubated with 1 mM GSSG, however, a large rise occurred in the physiological activity within 30 -60 min, achieving a maximum of 600 mol/mg/min; thereafter, activity very slowly and steadily declined (Fig. 1A). In contrast, transfer activity decreased upon oxidation with GSSG: the fall was initially rapid, but then became parallel to the slow decline phase of the physiological reaction (Fig. 1A). Incubation of 7.5 M enzyme with 5 and 10 M GSSG achieved specific activities in the physiological reaction of 350 and 460 mol/mg/min, respectively. With these low concentrations of oxidant, maximum activity was sustained but took longer to attain. Using 20 M GSSG, a ratio slightly in excess of 2 mol/enzyme dimer, the decline phase in the activity profile was regained and, at concentrations higher than 50 M GSSG, all activity curves showed essentially similar phases of rise and decline (Fig. 1B).
Altering the redox ratio of GSSG to GSH at a constant 1 mM total glutathione concentration also changed the timing and maximum activities achieved, but only at 1:3 GSSG:GSH (250 m GSSG) and above was there a later decline in activity (Fig.  1C). Addition of 0.2 M bovine liver protein disulfide-isomerase (EC 5.3.4.1) (Sigma) to 7 M sulfotransferase increased the rate of formation of the activated enzyme (not shown). After 30 min of incubation, and using a 1:99 ratio of GSSG to GSH, a 4-fold increase in activity was observed due to inclusion of the isomerase; after 60 min of incubation, the increase was 9-fold.
Inclusion of PAP in the enzyme/GSSG incubation mixture blocked activation. With 1 mol of PAP/dimer or greater, there was no activation at 1 h in the presence of 1 mM GSSG. It was not possible to determine the effect of PAPS on GSSG activation, since PAPS is rapidly hydrolyzed by the enzyme at physiological pH (3). Excess PAP, 200 M, added at any time after exposure to GSSG, arrested all further activity changes and, in the presence of a redox mixture of 1:1 GSSG:GSH, led to slow but complete reversal of the activity changes that had occurred: physiological activity declined and the transfer activity was regained (data not shown).
GSSG and the Cysteine Mutants-Five mutant enzymes, one for each of the cysteine residues of the protein, were examined for the effects of oxidation with GSSG. Three of them, C232S,M228I, C283S, and C289S, responded with changes in the activity profile identical to those shown for the wild type enzyme upon incubation with GSSG. Mutant C82S attained between 60 and 80% of the maximal specific activity. Mutant C66S was unique in lacking the activation response; instead, C66S exhibited only a gradual time-dependant decline in both physiological and transfer activities (Fig. 1A). As was observed with the wild type enzyme, 200 M PAP blocked the activity changes seen for of all the cysteine mutants in GSSG-containing buffer and caused reversal of the effects in redox buffer containing 1:1 GSSG:GSH (data not shown).
Changes in pH Optima-The tracer physiological assay was used to determine pH activity profiles toward 4-nitrophenol, tyrosine methyl ester, and 1-hexanol as representative acceptor substrates. Assays were performed either in the presence of 1 mM DTT or after treatment of the enzyme with 1 mM GSSG for 1 h at 25°C. When in the presence of the reducing agent DTT, optimum activity for 4-nitrophenol was observed at pH 5.2, whereas activity in the physiological pH region was minimal. From Fig. 2A it can be seen that treatment with 1 mM GSSG leads to a 17-fold increase in optimum activity and a shift in pH optimum to between 6.5 and 7.5 ( Fig. 2A). Each of the cysteine mutant enzymes displayed such a change in activity with the exception of the mutant C66S which, although minimally activated by oxidation, remained active only in the acid pH range (Fig. 2B). Tyrosine methyl ester, an eponymous substrate for the enzyme, and 1-hexanol are dissimilar to 4-nitrophenol in that there is extensive and optimal activity in the physiological pH range (Fig. 2, C and E) under reducing conditions; both compounds cease to be substrates upon oxidation. Mutant C66S displayed pH optima and sulfation rates toward tyrosine methyl ester and 1-hexanol similar to those of the wild type enzyme but differed in that 1-h treatment with GSSG resulted in only a 50% reduction and not abolition of activity (Fig. 2, D and F). The association of enzyme and PAP could be analyzed with the PCMLAB program, revealing the best fit to be attained by a two binding site model for each of the ligands. Stoichiometric binding was found to occur at the first site for PAP, 2-naphthol and 2-naphthyl sulfate regardless of oxidation state of the enzyme. At the second site, however, oxidation of the protein with 1 mM GSSG for 1 h elicited a 30-fold increase in K d for PAP, but only small decreases in K d for 2-naphthol and 2-naphthyl sulfate (Table I). In each instance, the total fluorescence change was greater using the completely reduced enzyme than that treated with GSSG.
Determination of KЈ m and KЈ cat -The spectrophotometric physiological and transfer assays were used to compare apparent K m and K cat for substrates with enzyme in the presence of DTT or after incubation with GSSG for 1 h. Upon oxidation there were changes in the KЈ m and KЈ cat for the wild type and mutant enzymes with the sole exception of the mutant C66S for which both of these parameters remained constant. For the physiological assay system, partial oxidation caused KЈ m increases for 4-nitrophenol and PAPS. Despite the higher KЈ m , KЈ cat was 6 -15-fold greater upon GSSG oxidation (Table II). Similar changes in the apparent KЈ m for PAP and 4-naphthyl sulfate were observed using the transfer assay system, whereas KЈ cat values, in contrast, decreased 2-3-fold in all but mutant C82S for which KЈ cat doubled (data not presented).
Enzyme Thiols-The change in the number of titratable enzyme thiol groups, estimated by reaction with DTNB, is recorded in Table III as a function of time of exposure to GSSG.
Prior to oxidation, 4.0 of the five known cysteine groups per subunit were titratable with DTNB for the wild type enzyme, and between 3.5 and 3.7 of the four thiols were found for the mutants in which a cysteine was replaced by serine. After incubation of the enzymes with 1 mM GSSG, there was a decrease in the number of free thiol groups titratable. After 1 h of incubation, wild type and mutants C82S, C283S, and C289S lost almost two of the initial free thiol groups, but mutants C66S and C232S,M228I only lost about one. After prolonged incubation with GSSG (24 h), virtually all of the initial free thiol groups of the sulfotransferases were unavailable.
Electrophoresis of Sulfotransferase-When wild type and cysteine mutant enzymes were reduced with DTT and subjected to electrophoresis in SDS under conditions free of reducing agent, each of the enzymes migrated as a single band of 34 kDa. Following overnight oxidation with GSSG, the pattern of electrophoretic migration changed to reveal two major bands of equal density. For wild type and for mutants C66S and C82S, the new bands were at 35 and 31.5 kDa with a minor one at 32 kDa; for cysteine mutants C283S and C289S the new major bands were at 35 and 32 kDa with minor bands at 31.5 and 34 kDa (Fig. 3). These major new bands were barely visible at 1 h, becoming apparent only after prolonged oxidation. Mutant C232S,M228I was exceptional as it showed no changes in electrophoretic pattern upon oxidation (Fig. 3). No significant species of greater M r was observed at any time period for any enzyme species.
Mass Spectral Analyses of Glutathione Conjugates-In an attempt to ascribe the activation phenomenon of sulfotransferase to specific cysteine residues, the enzymes were subjected to electrospray ionization mass spectral analysis following timed oxidation by GSSG. A multiply charged envelope of ions was obtained which was deconvoluted using the "MaxEnt" program to yield a molecular mass for the wild type protein reacted with GSSG that was 305 Da higher than the original mass. This clearly established the covalent addition of one glutathione to each enzyme subunit over a 24-h time period (Fig. 4). Formation of this new species was slow but progressive with the ion intensity being relatively low at 1 h, the time at which activation of the enzyme was optimum. Additional low intensity ions at approximately 34,510 Da (addition of 2 GSH) and 34,815 Da (3 GSH) were observed after 24 h incubation with GSSG (Fig. 4).
The sulfotransferase mutants C66S and C232S,M228I underwent the same mass spectral analysis after oxidation by GSSG and yielded identical profiles. The ions obtained again established the addition of a single glutathione to each of these proteins, but with these mutants the reactions occurred rapidly and were near completion after 1 h (Fig. 4). Upon further incubation, ions indicating the addition of a second glutathione to the proteins were detected as shown for the mutant

Control of Sulfotransferase Activity by S-Glutathiolation
C232S,M228I (Fig. 4). The reaction of mutants C283S and C289S with GSSG yielded ions at 1 h that were similar in mass and intensity to those obtained with the wild type enzyme, indicating slow and incomplete addition of a single glutathione; both mutants, however, acquired a second glutathione after 24 h of incubation (not shown). Mutant C82S was the exception because it yielded an ion representing the addition of two glutathione molecules (ϩ610 Da) without passing through a stage at which only a single addition was evident; the intensity of this new ion increased slowly over the 24-h incubation period. When fully oxidized the protein was equally divided between two peaks representing species of the original molecular weight and the diglutathiolated monomer.
Mapping of Disulfides-The possibility of formation of intraprotein disulfides was examined by tryptic digestion and peptide mapping of the GSSG-treated protein. The reaction of GSSG with the enzyme was terminated by the addition of the alkylating agent 4-vinylpyridine to prevent further disulfide interchange. Following alkylation and tryptic cleavage, peptides were separated by liquid chromatography and analyzed by mass spectrometry. Trypsin was chosen for proteolysis because it generated separate peptides for each of the five cysteine residues. The cysteine-bearing tryptic peptides and their masses, based on the known sequence (6), are listed in Table IV (Fig. 5C), that occurred within 1 h, the time of maximal enzyme activity. The total ion count for this peak remained almost constant in intensity thereafter. The appearance of the peak at 39.8 min, represented by ions at m/z 1624.77 (1ϩ), 812.56 (2ϩ), and 542.22 (3ϩ), representing disulfide linked peptides 32 and 34 incorporating Cys 283 and Cys 289 (Fig. 5A), was minor at 1 h, but increased in intensity with time. This peak represents peptides 32, 33, and 34, i.e. the disulfide bond formation prevents the usual tryptic cleavage between the three peptides; this conclusion is supported by the reciprocally diminished peak for peptide 32 incorporating Cys 283 with 4-vinylpyridine.
The single glutathione addition to the wild type enzyme took place at Cys 82 as confirmed by a peak of mass 2156.5 at 40.4 min with a mass 305 Da higher than that predicted for unmodified peptide 9 (Fig. 5B). This formation of the glutathione conjugate was observed as a gradual time-dependent transition from the ions representing peptide 9 incorporating Cys 82 conjugated to 4-vinylpyridine m/z 979.3 (2ϩ) and 653. 3 (3ϩ) to the ions at m/z 1079.3 (2ϩ) and 719.8 (3ϩ) (Fig. 5B). The presence of the glutathione conjugate was verified by LC/MS and tandem MS/MS as the characteristic neutral loss ion of 129 Da under CID conditions. Disulfide mapping of the peptides derived from mutant C82S indicates that a disulfide bond formed between Cys 283 and Cys 289 as it did with wild type enzyme. The formation of the disulfide between Cys 66 and Cys 232 did not occur on all of the protein, however. This incomplete formation of the latter disulfide bond may explain the presence of the two peaks observed with mass spectral analysis of the intact C82S protein, representing nonconjugated and diglutathiolated species.
Both mutants, C66S and C232S,M228I, neither of which is able to form the disulfide bond of the wild type enzyme between Cys 66 and Cys 232 , underwent complete addition of a single glutathione within 1 h of incubation with GSSG. A second glutathione was slowly added to Cys 82 of these mutants upon prolonged incubation with GSSG in a manner similar to that with the wild type enzyme. A mass increase of 305 to peptide 27 confirmed the S-glutathiolation of Cys 232 in the mutant enzyme C66S. Glutathione addition to peptide 6, containing Cys 66 , in the mutant C232S,M228I required indirect detection. The 4-vinylpyridine conjugate and the S-glutathiolated form of peptide 6 were not resolved by HPLC and eluted too close to the solvent front; this was confirmed using a custom synthesized peptide of sequence CGR conjugated with 4-vinylpyridine and, separately, with glutathione. The observation of a neutral ion loss of 129 Da under CID conditions from the peptides in the peak eluting close to the solvent front using LC/MS and tandem MS/MS confirmed that mutant C232S,M228I contained S-glutathiolated Cys 66 after 1-h incubation with GSSG. This conjugate was not observed at zero time for the mutant enzyme or at any time in the wild type preparations. DISCUSSION The characteristics of tyrosine-ester sulfotransferase include several curious elements. For one, the PAP-dependent transfer of sulfate from 4-nitrophenyl sulfate to 2-naphthol (Equation 1) was catalyzed at a rate 10 times greater than the reaction in the physiological direction, i.e. the sulfation of 4-nitrophenol using PAPS as the sulfate source (Equation 2). The reason for this difference emerged when the two activities were found to change upon moderate oxidation of the enzyme. Incubation of enzyme with GSSG led to both rapid loss of the transfer activity and the time-dependent activation of the sulfation of 4-nitrophenol. Further oxidation produced complete loss of activity. There was a direct relationship between the extent and rate of the activation and the concentration of GSSG, but, the subsequent loss of activity was observed only if GSSG exceeded 2 mol/enzyme dimer. In redox buffers containing both oxidized and reduced glutathione, activation was dependent on the re-dox ratio and not the absolute concentration of GSSG. Below a 1:4 ratio of GSSG:GSH, high activity was sustained, suggesting that oxidation and reduction of the enzyme were in a stable equilibrium under these conditions. The equilibrium was disrupted by intervention with PAP, which binds extraordinarily tightly to the enzyme (2), but only under reducing conditions. PAP halts, but does not reverse, the effects of oxidation with GSSG alone. In a redox system, however, PAP bound to the reduced form of the enzyme and prevented reoxidation, thereby reversing the physiological activation and restoring the transfer activity. The binding of PAP (and PAPS) is believed to be at a site close to Cys 66 (12), consistent with the capability of PAP to block oxidation.
A second enigmatic element of this sulfotransferase involves the diverse range of pH optima for different substrates. Sulfation of tyrosine methyl ester was optimum at about pH 8, but that for 4-nitrophenol was at about pH 5.5. The low pH optimum of this cytosolic enzyme for substrates such as 4-nitrophenol and 2-naphthol (1) was patently inconsistent with the presumption that the enzyme has a role in the detoxication process. Resolution to this paradox was made possible by the FIG. 4. ESI mass spectra of intact wild type and mutant C232S,M228I proteins, treated with GSSG for increasing periods of time. The ions obtained were deconvoluted using the "Maximum Entropy Program" to yield peaks, indicating the average molecular mass of each protein species. observation that partial oxidation of the enzyme leads to a dramatic change from an acid pH optimum to one in the physiological pH range for these substrates and results in a large increase in the rate of sulfation. Nevertheless, two other substrates, 1-hexanol and tyrosine methyl ester, both of which had activity for the neutral pH range in the reduced enzyme, essentially lost activity at all pH values upon oxidation. We have observed, therefore, changes in substrate specificity as a function of the redox state of the enzyme's environment. The steps underlying the activation process can be understood only in part by considering the information obtained with the cysteine mutants. There are three strong candidates for the mechanism of activation: S-glutathiolation of a critical cysteine residue, formation of a disulfide between two cysteine residues within a subunit, and disulfide bond formation between the two subunits of the homodimer. The last of these can be dismissed because SDS-electrophoresis provided no evidence for the formation of intersubunit disulfides: higher molecular weight species, greater than one subunit, were not formed. The slightly smaller M r species, which were observed on electrophoresis after treatment with GSSG, are probably due to conformational changes resulting from disulfide formation among the five SH groups of each subunit. Data from thiol titration with DTNB following incubation of the enzymes with GSSG indicated only that mutants of Cys 66 and Cys 232 were different in their behavior (Table III) but did not distinguish between the other two candidates.
An answer to the mechanism of GSSG activation required mass spectrometric analysis and disulfide mapping. ESI-MS analysis of the intact protein disclosed that a single glutathione had added to the wild type enzyme, but only on prolonged incubation, far after maximum activation had been achieved.
Although S-glutathiolation of a protein at a conserved cysteine is known to regulate the activity of several enzymes, including human immunodeficiency virus type 1 protease (20), glutathione transferase (21), and the phosphatase activity of carbonic anhydrase III (22), the timing of conjugate formation for the wild type sulfotransferase did not coincide with activation. Disulfide mapping of the tryptic peptides derived after incubation with GSSG confirmed that the event occurring for the wild type enzyme at the time of maximum activation consists of the formation of a disulfide bond between Cys 66 and Cys 232 . The S-glutathiolation of Cys 82 , as well as disulfide bond formation between Cys 283 and Cys 289 , occur subsequently during the period that leads to inactivation. These later events are probably without normal physiological function, since it is unlikely that the enzyme will encounter such high ratios of GSSG to GSH for the protracted periods of time required.
Mutant enzymes C283S and C289S displayed the same profile of changes in activity as the wild type. The only difference observed among these enzymes was the addition of a second glutathione after 24-h incubation with GSSG, presumably because the disulfide between Cys 283 and Cys 289 does not exist. Mutant C82S differed in its lower maximal physiological activity than the wild type; mass spectral data imply that only half of the enzyme forms a disulfide between Cys 66 and Cys 232 , whereas the other portion incorporates 2 mol of glutathione.
The critical element in the activation appears to be centered on the oxidation of Cys 66 . Simply removing the charge due to the thiolate anion at Cys 66 , as in the substitution of cysteine by serine in mutant C66S, does not lead to permanent activation; in fact, that mutant can not be activated. One mode of activation is achieved by formation of a disulfide bond between Cys 66 and Cys 232 . The creation of that disulfide bond is not an abso- lute requirement for activation, however, since a mutant in which Cys 232 is replaced by serine also can be activated; the latter occurs by S-glutathiolation as documented by the finding of a disulfide with glutathione at Cys 66 . In contrast, the Sglutathiolation at Cys 232 is ineffective in activating mutant enzyme C66S. The importance of Cys 66 for activation provides a rational for the conserved nature of Cys 66 throughout the phenol sulfotransferase family (11,12).
The demonstration that the activity of a major detoxication enzyme is critically altered by the redox potential of its environment has potential significance in times of cellular oxidative stress. The activation that has been described for the sulfotransferase occurs within the expected cellular ratios of GSSG:GSH experienced under oxidative stress (23) and is indicated here as being accelerated by cellular enzymes catalyzing protein disulfide exchange reactions (24,25). Furthermore, it has been postulated that under conditions of oxidative stress, even without significant rises in the concentration of cellular GSSG, there will be direct, nonenzymatic coupling of GSH to cysteine thiols through oxygen-radical catalysis (26 -28). The survival advantages conferred on the cell by this means of regulating aryl sulfotransferase IV remain to be investigated as does the mechanism determining the switch in substrate specificity.