Irreversible Inactivation of Protein Kinase C by Glutathione*

The tripeptide glutathione (GSH) is the predominant low molecular weight thiol reductant in mammalian cells. In this report, we show that at concentrations at which GSH is typically present in the intracellular milieu, GSH and the oxidized GSH derivatives GSH disulfide (GSSG) and glutathione sulfonate each irreversibly inactivate up to 100% of the activity of purified Ca2+- and phosphatidylserine (PS)-dependent protein kinase C (PKC) isozymes in a concentration-dependent manner by a novel nonredox mechanism that requires neither glutathiolation of PKC nor the reduction, formation, or isomerization of disulfide bridges within PKC. Our evidence for a nonredox mechanism of PKC inactivation can be summarized as follows. GSSG antagonized the Ca2+- and PS-dependent activity of purified rat brain PKC with the same efficacy (IC50 = 3 mm) whether or not the reductant dithiothreitol was present. Glutathione sulfonate, which is distinguished from GSSG and GSH by its inability to undergo disulfide/thiol exchange reactions, was as effective as GSSG in antagonizing Ca2+- and PS-dependent PKC catalysis. The irreversibility of the inactivation mechanism was indicated by the stability of the inactivated form of PKC to dilution and extensive dialysis. The inactivation mechanism did not involve the nonspecific phenomena of denaturation and aggregation of PKC because it obeyed pseudo-first order kinetics and because the hinge region of PKC-α remained a preferential target of tryptic attack following GSH inactivation. The selectivity of GSH in the inactivation of PKC was also indicated by the lack of effect of the tripeptides Tyr-Gly-Gly and Gly-Ala-Gly on the activity of PKC. Furthermore, GSH antagonism of the Ser/Thr kinase casein kinase 2 was by comparison weak (<25%). Inactivation of PKC-α was not accompanied by covalent modification of the isozyme by GSH or other irreversible binding interactions between PKC-α and the tripeptide, but it was associated with an increase in the susceptibility of PKC-α to trypsinolysis. Treatment of cultured rat fibroblast and human breast cancer cell lines withN-acetylcysteine resulted in a substantial loss of Ca2+- and PS- dependent PKC activity in the cells within 30 min. These results suggest that GSH exerts negative regulation over cellular PKC isozymes that may be lost when oxidative stress depletes the cellular GSH pool.

PKC isozymes contain multiple Cys residues in their catalytic and regulatory domains (14,15), including Cys-rich sequences present in the regulatory domain that are critical to the phorbol ester responsiveness of the isozyme family (2,16). A highly reactive Cys residue of unknown function that is expressed in the catalytic domain of each PKC isozyme subfamily is subject to S-thiolation by the synthetic peptide-substrate analog N-biotinyl-Arg-Arg-Arg-Cys-Leu-Arg-Arg-Leu, i.e. an intermolecular disulfide bridge is formed between the Cys residue and the synthetic peptide, and this modification inactivates the enzyme (17,18). The susceptibility of PKC isozymes to inactivation by an S-thiolating peptide-substrate analog suggests that PKC activity might also be subject to redox regulation by the endogenous molecule glutathione (18). The tripeptide glutathione (GSH), L-␥-glutamyl-L-cysteinyl-glycine, is the predominant low molecular weight thiol reductant in mammalian cells (19). Protein S-glutathiolation, i.e. the formation of a disulfide-linked protein-GSH complex, is a selective protein modification that can be induced in cells by mild oxidative stress (20). GSH disulfide (GSSG) has been shown to oxidatively regulate the function of several purified enzymes, including carbonic anhydrase III, aldose reductase, and HIV-1 protease by S-glutathiolation, and in each case the effects of GSSG can be fully reversed by the reducing agent dithiothreitol (DTT) (21)(22)(23). Mammalian cells typically contain millimolar concentrations of GSH, e.g. 0.5-10 mM, and oxidized GSH gen-erally amounts to less than 1% of the total cellular GSH content (19,24). In this report, we show that at GSH concentrations typically present in the intracellular milieu, GSH and oxidized GSH derivatives irreversibly inactivate purified PKC by a novel nonredox mechanism that does not involve glutathiolation of the enzyme. N-Acetylcysteine, a precursor of cellular GSH (25,26), likewise irreversibly inactivated purified PKC in a nonredox manner. Treatment of cultured cells with N-acetylcysteine induced a substantial decline in the level of cellular PKC activity, providing evidence that the PKC inactivation mechanism observed with N-acetylcysteine and GSH in the purified enzyme system may also be operative in mammalian cells.

MATERIALS AND METHODS
Rat brain PKC was purified to near-homogeneity according to silverstained polyacrylamide gels by a previously described method (27). The histone kinase activity of the purified PKC preparation was stimulated 10 -15-fold by 0.2 mM Ca 2ϩ and 30 g/ml PS but was unaffected by either Ca 2ϩ or PS alone. The purified PKC preparation is a mixture of the isozymes ␣, ␤, ␥, ⑀, and (18). A fully active catalytic domain fragment of PKC was generated from the purified PKC preparation by limited trypsinolysis with a yield of Ͼ50%, as described previously (17,28). Where indicated, the catalytic domain fragment was purified from regulatory domain fragment and residual intact PKC by DEAE ionexchange chromatography using a 0.0 -0.4 M NaCl gradient (17,28). The histone kinase activity of the catalytic domain fragment preparation was stimulated less than 1.5-fold by Ca 2ϩ and PS. Purified, baculovirus-produced recombinant human PKC-␣ was purchased from Pan Vera Corp. (Madison, WI). Immunoblot analysis of PKC-␣ was done as described previously (29) using an ECL detection system (Amersham Pharmacia Biotech) and monoclonal anti-PKC-␣ (Transduction Laboratories, Lexington, KY). Dialysis of PKC-␣ was done as described under "Results," using a 0.5-3 ml Slide-A-Lyzer unit with a molecular mass cut-off of 10,000 Da. (Pierce). Electrospray ionization mass spectrometric analysis of PKC-␣ was done at the University of Texas-Houston Analytical Chemistry Center. Samples of purified recombinant control PKC-␣ and GSH-inactivated PKC-␣ (10 pmol/l) were prepared for analysis by dialysis against 5% acetic acid under bubbling nitrogen for 48 h. Horse skeletal muscle apomyoglobin (16951.5 Da) and bovine serum albumin (66430.3 Da) served as standard proteins and were analyzed in parallel. The accuracy of molecular masses determined from the spectra was within Ϯ20 Da.
The synthetic peptide RKRTLRRL (Ͼ98% pure) was prepared at the Synthetic Antigen Core Facility at the University of Texas M. D. Anderson Cancer Center. The tripeptides ␥-Glu-Gly-Gly, Tyr-Gly-Gly, and Gly-Ala-Gly and the amino acids Gly, Glu, and N-acetyl-Cys were purchased from Bachem Bioscience Inc.  IN). Protein assay solutions and SDS-polyacrylamide gel electrophoresis reagents, including molecular mass markers, were obtained from Bio-Rad. Go6976 was purchased from Calbiochem. MCF7-MDR cells and R6-PKC3 fibroblasts were kindly provided by Dr. Kenneth Cowan (National Institutes of Health) and Dr. I. Bernard Weinstein (Columbia University, NY), respectively, and tissue culture reagents were purchased from Life Technologies, Inc. GSH, glutathione disulfide (GSSG), glutathione sulfonic acid (GSO 3 ), histone IIIS, PS, ATP, DTT, L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated trypsin, phenylmethylsulfonyl fluoride, DEAE-Sepharose, and all other reagents were purchased from Sigma.
Protein Kinase Assays-The Ca 2ϩ -and PS-dependent histone kinase activity of purified PKC was measured as described previously (28). The histone kinase reaction mixture (120 l) contained 20 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , 0.2 mM CaCl 2 , 30 g/ml PS (or none), 6 M [␥-32 P]ATP (5000 -8000 cpm/pmol), 0.67 mg/ml histone III-S, and 5 ng of purified PKC. In indicated experiments, histone was replaced with the synthetic peptide substrate RKRTLRRL (20 M) (28). A 10-min reaction period at 30°C, which yields linear kinetics, was initiated by the addition of [␥-32 P]ATP. The reaction was terminated on phosphocellulose paper, and histone (or peptide) phosphorylation was quantitated as described previously (28). GSX (GSH, GSSG, or GSO 3 ) was added to PKC assay mixtures as described under "Results." In some experiments, PKC or its catalytic domain fragment was preincubated with GSX alone for 5 min at 30°C and then briefly kept on ice prior to its addition to assay mixtures. The preincubation mixtures were diluted 12-30-fold into assay mixtures as specified under "Results." In other experiments, PKC and GSX were added directly to PKC assay mixtures as separate components. All assays were performed in triplicate and expressed as the mean value Ϯ S.D. Purified casein kinase 2 and a casein kinase 2 assay kit were purchased from Upstate Biotechnology (Lake Placid, NY). Casein kinase 2 was assayed by monitoring the phosphorylation of the synthetic peptide substrate RRRDDDSDDD according to the instructions provided by the manufacturer. Each reaction mixture contained 20 ng of casein kinase 2.
Measurement of Ca 2ϩ -and PS-dependent PKC Activity in N-Acetylcysteine-treated Cultured Mammalian Cells-Human breast cancer MCF7-MDR cells and rat R6-PKC3 fibroblasts were chosen for analysis because they express levels of Ca 2ϩ -and PS-dependent PKC activity that can be measured accurately by assays of DEAE-extracted cell lysates. The abundance of Ca 2ϩ -and PS-dependent PKC activity is primarily due to enforced expression of PKC-␤ 1 in the R6-PKC3 cells (30) and to the increase in PKC-␣ expression that occurred in association with the selection of the multidrug resistant line MCF7-MDR by doxorubicin (31). Prior to treatment with N-acetylcysteine, cells were cultured as previously reported, with MCF7-MDR cells in Eagle's minimum essential medium containing 5% heat-inactivated fetal calf serum (32) and R6-PKC3 cells in Dulbecco's modified Eagle medium containing 10% heat-inactivated fetal calf serum and 50 g/ml G418 (30). The culture media also contained nonessential amino acids, vitamins, sodium pyruvate, L-glutamine, and penicillin-streptomycin (30,32). Near-confluent cells cultured in T150 flasks (approximately 5 ϫ 10 7 R6-PKC3 cells and 4 ϫ 10 7 MCF7-MDR cells) were treated with Nacetylcysteine at the indicated concentration in culture medium (without serum) buffered with 80 (R6-PKC3) or 100 (MCF7-MDR) mM Tris-HCl, pH 7.5, for 30 min at 37°C. The treatment conditions employed were determined to have no effect on cell viability, as measured by trypan blue exclusion. At the end of the treatment period, PKC was extracted from the cells by a previously described procedure (33). Briefly, cells were washed with phosphate-buffered saline, harvested from the flasks in ice-cold Buffer A (20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 20 g/ml soybean trypsin inhibitor, 100 g/ml leupeptin, 0.25 mM phenylmethylsulfonyl fluoride) containing 1% Triton X-100, 5 mM chelators (EDTA and EGTA), and 15 mM 2-mercaptoethanol, and stirred for 1 h at 4°C. All subsequent procedures were done at 4°C. Cell lysates were centrifuged for 15 min at 13,800 ϫ g to remove debris and then loaded onto 0.5 ml DEAE-Sepharose columns equilibrated in Buffer A. After washing each column with 3 ml Buffer A, Ca 2ϩ -and PS-dependent PKC was eluted with 2 ml of Buffer A containing 0.2 M NaCl, and the protein concentration of the eluted sample was determined (33). Part of the sample was mixed with 2ϫ SDSpolyacrylamide gel electrophoresis sample buffer for immunoblot analysis, and the remainder was reserved for PKC assays.
The assay employed for the Ca ϩ -and PS-dependent PKC activity of the eluted samples (10 g of sample protein/assay) was modified from the assay used with purified PKC (see above, under "Protein Kinase Assays") as follows. The pseudosubstrate-based synthetic peptide substrate 10 M [Ser 25 ]PKC-(19 -31) (34) was employed as the phosphoacceptor substrate, the cofactor Ca 2ϩ was present at 1 mM, and the Ca 2ϩand PS-dependent PKC-selective inhibitor Go6976 (35) was employed at 100 nM to distinguish Ca 2ϩ -and PS-dependent PKC activity from background kinase activity. Ca 2ϩ -and PS-dependent PKC activity was calculated by subtracting the cpm obtained from assay mixtures containing Ca 2ϩ, , PS, and Go6976 from the cpm obtained from assay mixtures containing Ca 2ϩ and PS.

RESULTS
As an initial test of whether PKC is subject to regulation by glutathiolation, we preincubated a purified rat brain PKC preparation with GSSG at the concentrations shown in Fig. 1, diluted each preincubation mixture 30-fold into a PKC assay mixture, and measured the Ca 2ϩ -and PS-dependent histone kinase activity of the enzyme. Preincubation with GSSG inhibited PKC in a concentration-dependent manner, achieving 50% inhibition at about 2.6 mM GSSG and complete inhibition at Ն5.0 mM GSSG (Fig. 1, closed circles). The loss of 100% of the activity indicated that each Ca 2ϩ -and PS-dependent isozymic component of the PKC preparation was subject to inhibition by GSSG. In a control experiment, the preincubation step was omitted, and GSSG was added directly to PKC assay mixtures at the 30-fold diluted final concentration corresponding to each preincubation mixture. Under these conditions, inhibition was negligible ( Fig. 1, open circles). This indicates that the inhibition of PKC achieved by GSSG in Fig. 1 (closed circles) was a result of the preincubation step and was stable when the preincubation mixture was diluted. These results were consistent with an irreversible inactivation mechanism that could involve glutathiolation of PKC.
Glutathiolation of proteins is readily reversed by reducing agents, such as DTT (21,23,36). To test whether GSSGinduced inactivation of Ca 2ϩ -and PS-dependent PKC (Fig. 1, closed circles) involved PKC glutathiolation, DTT was included in the PKC/GSSG preincubation mixtures. The efficacy of GSSG in inactivating Ca 2ϩ -and PS-dependent PKC was virtually unaffected by the presence of either 100 mM DTT ( Fig. 1, open triangles) or 250 mM DTT ( Fig. 1, open squares) in the preincubation mixtures. We also determined that removal of the ␤-mercaptoethanol (5 mM) in the PKC preparation by gel filtration (17,18) prior to preincubation of the enzyme with GSSG had no effect on the concentration dependence of GSSG inactivation of PKC (data not shown). These results ruled out glutathiolation as the mechanism of GSSG inactivation of Ca 2ϩ -and PS-dependent PKC.
We next examined whether GSH and GSO 3 , a GSH analog that cannot participate in thiol/disulfide exchange reactions, could inactivate Ca 2ϩ -and PS-dependent PKC. When PKC was preincubated with GSH and then diluted 30-fold into reaction mixtures, GSH inactivated PKC with roughly half the potency of GSSG, achieving 50% inactivation at about 5.1 mM GSH (Fig.  2, closed triangles). No inhibition was observed if the preincubation step was omitted, and GSH was added directly to reaction mixtures at the corresponding final (30-fold diluted) concentrations (Fig. 2, open triangles), providing evidence for an irreversible inactivation mechanism. The observed potency of GSH in the inactivation of Ca 2ϩ -and PS-dependent PKC was consistent with the equivalence of 2 GSH with DTT-reduced GSSG (Figs. 1 and 2). The capacity of GSH to inactivate PKC was not merely a nonspecific effect of the free sulfhydryl of the tripeptide, because the reducing agent DTT did not inactivate PKC at concentrations as high as 250 mM (Fig. 1). Furthermore, the inactivation potency of GSH could not be attributed to trace amounts of GSSG, because it was unaffected by the presence of 100 mM DTT in PKC/GSH preincubation mixtures (data not shown). We also examined the ability of GSH to inactivate a purified preparation of the Ser/Thr kinase casein kinase 2 (37). In these experiments, casein kinase 2 was preincubated with GSH under the conditions employed in Fig. 2, and the preincubation mixture was diluted 12-fold into casein kinase assay mixtures. Preincubation with 5 and 10 mM GSH, which effected approximately 40 and 90% inactivation of PKC ( Fig. 2), resulted in a loss of only 8 Ϯ 9% and 23 Ϯ 2% of casein kinase 2 activity, respectively (values are averages obtained from three independent experiments done in triplicate). These results indicate that the potent inactivation of PKC achieved by GSH at physiological concentrations ( Fig. 2) cannot be attributed to broad and nonspecific effects of the tripeptide against isolated protein kinases.
Potent, irreversible inactivation of Ca 2ϩ -and PS-dependent PKC was also achieved by preincubation of the enzyme with GSO 3 (Fig. 2, closed circles). We next tested whether preincubation with GSH and the oxidized derivatives would inactivate PKC in assays employing the synthetic peptide-substrate RKRTLRRL (28) in lieu of histone. GSH, GSSG, and GSO 3 were each equally effective in inactivating the histone and synthetic peptide phosphorylation reactions of PKC (data not shown). The ability of GSO 3 to inactivate PKC, taken together with the DTT insensitivity of GSH-and GSSG-mediated PKC inactivation, definitively demonstrates that the inactivation mechanism does not require reduction, formation, or isomerization of disulfide bridges within PKC.
To determine whether the catalytic domain of PKC was sufficient for GSH inactivation of the enzyme, we generated a Ca 2ϩ -and PS-independent catalytic domain fragment of PKC by limited trypsinolysis of the purified rat brain PKC preparation (17, 28) (see "Materials and Methods"). The catalytic domain fragment was preincubated with GSH and oxidized GSH derivatives as described under "Materials and Methods" and then diluted 12-fold into assay mixtures. Fig. 3 shows that GSH, GSSG, and GSO 3 each effected concentration-dependent inactivation of the Ca 2ϩ -and PS-independent histone kinase activity of the catalytic domain fragment. No inactivation was observed when the preincubation step was omitted and GSH, FIG. 1. Irreversible inactivation of PKC by GSSG. Purified rat brain PKC was preincubated with GSSG at the concentration shown for 5 min at 30°C, briefly placed on ice, and then diluted 30-fold into histone kinase assay mixtures. The Ca 2ϩ -and PS-dependent histone kinase activity of PKC was measured; it represents the total activity of PKC-␣, PKC-␤, and PKC-␥ (18). Shown is the % inactivation of PKC achieved by preincubation with GSSG alone (q) and in the presence of 100 mM DTT (‚) and 250 mM DTT (Ⅺ). Also shown is the % inhibition of PKC achieved by omitting the preincubation step and directly adding GSSG and PKC to assay mixtures as separate components to achieve the 30-fold diluted final concentration (E), i.e. the GSSG concentration in the assay mixtures was 30-fold less than the GSSG concentration indicated on the x axis. 100% activity was 3.77 Ϯ 0.32 pmol of 32 P transferred per min. For other experimental details, see "Materials and Methods." GSSG, or GSO 3 was diluted 12-fold into reaction mixtures as a component separate from the catalytic domain fragment of PKC. Catalytic domain fragment that was purified from residual intact PKC and the regulatory domain fragment by DEAE chromatography (28) was likewise subject to inactivation by GSH, GSSG, and GSO 3 (data not shown). Although GSH, GSSG, and GSO 3 were each about 1.5-fold more potent in the inactivation of the catalytic domain fragment compared with inactivation of Ca 2ϩ -and PS-dependent PKC, the relative potencies of GSH, GSSG, and GSO 3 against the catalytic domain fragment of PKC and intact PKC were similar, with GSH being about half as potent as GSSG and GSO 3 (Figs. 1-3). These results indicate that the mechanism of GSH inactivation of PKC is catalytic domain-directed.
In the above experiments, PKC or its catalytic domain fragment was preincubated with GSH or oxidized GSH derivatives at inactivating concentrations in the absence of PKC substrates and cofactors. To test whether PKC substrates and cofactors might protect PKC from inactivation by GSX (GSH, GSSG, or GSO 3 ), we next added inactivating concentrations of GSX directly to the PKC assay mixtures, which contained 0.2 mM CaCl 2 , 30 g/ml PS, 10 mM MgCl 2 , 6 M [␥-32 P]ATP, 0.67 mg/ml histone III-S, and 5 ng of purified PKC. Under these conditions, the IC 50 values of GSSG, GSH and GSO 3 against Ca 2ϩ -and PS-dependent PKC activity estimated by graphical analysis were 2.2, 4.2, and 1.7 mM (Fig. 4). These IC 50 values are actually somewhat smaller than those obtained when PKC was preincubated in the absence of cofactors and substrates with GSSG (IC 50 ϭ 2.6 mM), GSH (IC 50 ϭ 5.1 mM), and GSO 3 (IC 50 ϭ 2.4 mM) prior to its addition to assay mixtures (Figs. 1 and 2). Thus, the presence of PKC substrates and cofactors in PKC assays at concentrations sufficient to optimally support Ca 2ϩand PS-dependent PKC activity does not afford protection against GSH inactivation of PKC.
We also tested for protective effects of PKC substrates and cofactors against GSX inactivation of Ca 2ϩ -and PS-dependent PKC by including substrates/cofactors in the PKC-GSX preincubation mixtures prior to dilution into histone kinase assay mixtures. Inclusion of the activating cofactors 1 mM CaCl 2 and 30 g/ml PS in PKC/GSSG preincubation mixtures did not afford protection against GSSG inactivation of Ca 2ϩ -and PSdependent PKC but instead reduced the IC 50 of GSSG estimated by graphical analysis to 1 mM (Fig. 5). Similarly, the presence of 1 mM CaCl 2 and 30 g/ml PS in PKC/GSH preincubation mixtures reduced the IC 50 of GSH to about 3.8 mM (data not shown). Histone also failed to afford substantial protection against GSH inactivation of PKC. The percentage of inactivation of Ca 2ϩ -and PS-dependent PKC achieved by preincubation with 10 mM GSH in the absence of histone (100 Ϯ 1%) was negligibly affected by including histone III-S in the preincubation mixtures at concentrations of 0.5 mg/ml (100 Ϯ 1% inactivation), 1.0 mg/ml (93 Ϯ 4% inactivation), and 1.5 mg/ml (89 Ϯ 5% inactivation). As seen in Table III, a comparison of the inactivation of recombinant PKC-␣ achieved by 10 mM GSH when 3 mM MgATP was present (87 Ϯ 11% inactivation) or absent (97 Ϯ 4% inactivation) in PKC-␣/GSH preincu-FIG. 2. Irreversible inactivation of PKC by GSH and GSO 3 . The % inactivation of purified rat brain PKC by preincubation with GSH and GSO 3 was measured. The Ca 2ϩ -and PS-dependent histone kinase activity of PKC was assayed subsequent to preincubation of the enzyme with GSH (OE) and GSO 3 (q) at the concentrations shown, or the preincubation step was omitted, and GSH (‚) and GSO 3 (E) were added to assay mixtures independently of PKC. In every case, the GSH/GSO 3 concentration in the assay mixture was 30-fold less than the concentration indicated on the abscissa. 100% activity was 2.87 Ϯ 0.14 pmol of 32 P transferred per min. For other experimental details, see the legend to Fig. 1 and "Materials and Methods."

FIG. 3. Irreversible inactivation of the catalytic domain of PKC by glutathione.
A catalytic domain fragment of PKC generated by limited trypsinolysis of the purified rat brain PKC preparation was preincubated with GSH (E), GSSG (Ⅺ), or GSO 3 (‚) for 5 min at 30°C, briefly placed on ice, and then diluted 12-fold into histone kinase assay mixtures. The % inactivation of the Ca 2ϩ -and PS-independent histone kinase activity of the catalytic domain fragment achieved by preincubation with GSH, GSSG, and GSO 3 is shown. 100% activity was 1.52 Ϯ 0.2 pmol of 32 P transferred per min. For other details, see "Materials and Methods." bation mixtures (which were extensively dialyzed to remove excess nucleotide prior to dilution into histone kinase assay mixtures) indicates that MgATP affords little protection against GSH inactivation of PKC.
Having determined that preincubation of PKC with glutathiones (GSH, GSSG, and GSO 3 ) and the direct addition of the glutathiones to PKC assay mixtures at physiological GSH concentrations (5-10 mM) both potently antagonize the enzyme (Figs. 1, 2, and 4), we next asked the question of whether PKC antagonism by glutathiones reflected selective interactions between PKC and the tripeptides or whether, on the other hand, any tripeptide at a concentration of 10 mM could nonspecifically antagonize PKC activity under the same conditions. Of the three tripeptides that we analyzed, ␥-Glu-Gly-Gly bore the most structural resemblance to GSH, and it effected Ͼ90% antagonism of PKC activity at a concentration of 10 mM, whether the peptide was preincubated with PKC or directly added to PKC assays (Table I). In contrast, the tripeptides Tyr-Gly-Gly and Gly-Ala-Gly were related to GSH only by virtue of a C-terminal Gly residue. These peptides had negligible effects on PKC activity (Ͻ10% antagonism), whether preincubated with the enzyme or directly added to PKC assay mixtures at a concentration of 10 mM (Table I). These results clearly show that GSH antagonism of PKC cannot be ascribed to nonspecific effects of millimolar concentrations of a small peptide on the purified enzyme. We also analyzed the ability of amino acid constituents of GSH to antagonize PKC. At a con-centration of 10 mM, Glu and Gly had little or no effect on PKC activity, whereas N-acetyl-Cys resembled GSH in that it potently inactivated PKC in a DTT-insensitive manner (Table I). ␥-Glu-Gly-Gly 93 Ϯ 8 100 Ϯ 1 Tyr-Gly-Gly 9 Ϯ 8 0 Ϯ 7 Gly-Ala-Gly 4 Ϯ 5 9 Ϯ 9 Glu 19 Ϯ 18 3 Ϯ 4 N-Acetyl-Cys 100 Ϯ 2 9 9 Ϯ 1 N-Acetyl-Cys ϩ 100 mM DTT 100 Ϯ 1 N D c Gly 0 Ϯ 7 0 Ϯ 5 a Purified rat brain PKC was preincubated with the test reagent (10 mM) for 5 min at 30°C, briefly placed on ice, and then diluted 30-fold into histone kinase assay mixtures.
b Test reagents were added directly to histone kinase assay mixtures to a final concentration of 10 mM. The Ca 2ϩ -and PS-dependent histone kinase activity of PKC was assayed as described under "Materials and Methods." c ND, not determined.

FIG. 4. PKC inhibition by GSX added directly to assay mixtures.
The % inhibition of the Ca 2ϩ -and PS-dependent histone kinase activity of purified rat brain PKC by the presence of GSH (OE), GSSG (f), and GSO 3 (q) in assay mixtures at the concentrations shown was measured. 100% activity was 9.51 Ϯ 0.78 pmol of 32 P transferred per min. For other experimental details, see the legend to Fig. 1 and "Materials and Methods."

FIG. 5. Irreversible inactivation of Ca 2؉ and phosphatidylserine (CaPS)activated PKC by GSSG.
Shown is the % inactivation of purified rat brain PKC achieved by preincubation of the enzyme with GSSG at the concentrations indicated in the presence of the activating cofactors 1 mM CaCl 2 and 30 g/ml PS. Preincubation mixtures were prepared and diluted into histone kinase assay mixtures as described in the legend to The Ͼ90% inactivation of PKC by 10 mM ␥-Glu-Gly-Gly and 10 mM N-acetyl-Cys (Ϯ 100 mM DTT) provides evidence that distinct structural features of GSH may contribute to its ability to antagonize PKC.
The results shown in Table II demonstrate that purified recombinant human PKC-␣ (38) is subject to inactivation by GSH and oxidized GSH derivatives. PKC-␣ was inactivated by preincubation with GSSG, GSSG ϩ DTT, GSH, and GSO 3 , and in each case, the inactivated form of PKC-␣ was stable following a 12-fold dilution (Table II), providing evidence for an irreversible inactivation mechanism. Comparable inactivation of PKC-␣ was achieved by GSH and the oxidized GSH derivatives whether histone III-S (Table II) or the synthetic peptide RKRTLRRL (data not shown) was employed as the phosphoacceptor substrate. Kinetic analysis of the inactivation of PKC-␣ by 5 mM GSH indicated that the inactivation mechanism was time-dependent and obeyed pseudo-first order kinetics (Fig. 6). The pseudo-first order rate constant k obs , which is the slope of the linear plot ( Fig. 6) (39), was determined to be 0.43 Ϯ 0.02 min Ϫ1 by averaging the values obtained from two kinetic analyses.
As a more rigorous test of the irreversibility of GSX (GSH, GSSG, or GSO 3 )-mediated inactivation of PKC-␣, we determined the stability of the GSX-inactivated form of PKC-␣ to dialysis. We compared the postdialysis recovery of PKC-␣ activity and PKC-␣ protein from samples of GSX-inactivated PKC-␣ and mock-inactivated (control) PKC-␣ subjected to dialysis for 8 -10 h against Tris-HCl buffer, pH 7.5, in the presence of 10 or 25 mM DTT. Based on Western analysis with a PKC-␣ mAb, the postdialysis recovery of PKC-␣ protein from GSX-inactivated PKC-␣ and control PKC-␣ was virtually identical, and the electrophoretic mobility of the isozyme was unchanged whether PKC-␣ was inactivated by GSSG, GSH, or GSO 3 (Fig. 7). Comparison of the PKC-␣ activity recovered postdialysis from GSX-inactivated PKC-␣ versus control PKC-␣ indicated that each of the GSX-inactivated PKC-␣ species (GSSG-, GSH-, or GSO 3 -inactivated PKC-␣) remained fully inactivated following dialysis (Table III). These results firmly establish the irreversibility of the inactivation of PKC-␣ by GSH and oxidized GSH derivatives.
Having established that GSH inactivation of PKC-␣ was stable following extensive dialysis and that inactivated PKC-␣ did not contain covalently attached or tightly bound GSH, we next addressed the question of whether the inactivation was associated with a conformational change in PKC-␣. The susceptibility of PKC isozymes to limited trypsinolysis serves as an indicator of conformational changes that involve alterations in the exposure of the hinge region (8,41,42). Fig. 8 shows a comparison of the sensitivity of control PKC-␣ (A) and GSHinactivated PKC-␣ (B) to limited trypsinolysis. GSH-inactivated and control PKC-␣ (15 g each) were prepared and then dialyzed for removal of GSH from the inactivated PKC-␣ sample, as described in the legend to Fig. 7. Following dialysis, the GSH-inactivated PKC-␣ sample was determined to be fully inactivated (percentage of inactivation ϭ 98 Ϯ 2%). The dialyzed samples were subjected to limited trypsinolysis under identical conditions. Trypsinolysis was measured by Western analysis of the samples (18) with a monoclonal antibody that recognizes PKC-␣ and its catalytic domain fragment (Upstate Biotechnology, Inc.). Fig. 8A shows that in the control PKC-␣ sample, the 82-kDa intact PKC-␣ species (lane 1) persisted following incubation with 1ϫ (lane 2), 2ϫ (lane 3), and 4ϫ (lane 4) units of trypsin/ml. In contrast, Fig. 8B shows that under  Fig.  8 were reproduced in two additional experiments (data not shown). The fact that limited trypsinolysis generated a 50-kDa catalytic domain fragment (18) from both control (Fig. 8A) and GSH-inactivated (Fig. 8B) PKC-␣ indicates that the hinge region remains a major target of trypsinolysis of PKC-␣ following GSH inactivation. The results also indicate that GSH-inactivated PKC-␣ is more sensitive than control PKC-␣ to trypsinolysis, providing evidence that GSH inactivation of PKC-␣ is accompanied by either a stable conformational change in the isozyme or a destabilization of the isozyme structure that persists even after removal of GSH by prolonged dialysis. N-Acetylcysteine is a precursor of cellular GSH that is readily taken up by mammalian cells and incorporated into GSH (25,26). Because GSH and N-acetylcysteine inactivated purified Ca 2ϩ -and PS-dependent PKC isozymes similarly, treatment of cells with N-acetylcysteine afforded a convenient approach to assess whether GSH/N-acetylcysteine-mediated PKC inactivation could also occur in mammalian cells. Fig. 9 shows that when cultured rat R6-PKC3 fibroblasts were treated with 40 mM N-acetylcysteine for 30 min at 37°C, the level of extractable Ca 2ϩ -and PS-dependent PKC activity in the cells was reduced by 53 Ϯ 3%. Similarly, the level of Ca 2ϩand PS-dependent PKC activity in cultured human breast cancer MCF7-MDR cells declined 64 Ϯ 3% in response to a 30 min treatment with 50 mM N-acetylcysteine at 37°C (Fig. 9). We determined by Western analysis that the N-acetylcysteine treatment of R6-PKC3 and MCF7-MDR cells shown in Fig. 9 did not change the level of expression of the Ca 2ϩ -and PS-dependent PKC isozymes (␣, ␤, and ␥) in the cells (experimental error, Յ10%) (data not shown). We also measured the intracellular level of GSH/N-acetylcysteine in each cell line prior to and at the end of the treatment period indicated in Fig. 9, by an   FIG. 6. Kinetics of inactivation. The kinetics of GSH inactivation of PKC-␣ were analyzed by preincubating PKC-␣ with 5 mM GSH at 30°C for the time periods indicated. Preincubation mixtures with and without GSH were analyzed in parallel. Preincubated mixtures were immediately diluted 12-fold into histone kinase reaction mixtures on ice, and PKC-␣ activity was measured as described under "Materials and Methods." Control values (E) show the PKC-␣ activity remaining after preincubation in the absence of GSH for the indicated time period expressed as a percentage of the activity observed prior to preincubation (time ϭ 0 min). q shows the time-dependent loss of PKC-␣ activity induced by GSH; each closed circle denotes the PKC-␣ activity remaining after preincubation with GSH for the indicated time period expressed as a percentage of the control value observed at the same time period (linear correlation coefficient ϭ 0.9945).

FIG. 7. Postdialysis immunoblot analysis of GSX-inactivated PKC-␣.
Recombinant PKC-␣ (10 g) was inactivated by incubation with 10 mM GSH, 5 mM GSSG, or 10 mM GSO 3 for 5 min at 30°C in 20 mM Tris-HCl, pH 7.5, 10% glycerol; control PKC-␣ was mock-inactivated by incubation as described above but in the absence of GSX. PKC-␣ incubation mixtures (1 ml) were dialyzed for 8 -10 h at 4°C against 500 ml of 20 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 10% glycerol, 0.25 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, plus either 10 or 25 mM DTT as shown. Western analysis of dialyzed GSX-inactivated PKC-␣ samples (ϩ) and dialyzed control PKC-␣ samples (Ϫ) is shown. The samples employed in the Western analysis shown here were also employed in the analysis of catalytic activity shown in Table III. Sample volumes were first equalized by adjusting for slight discrepancies (Ͻ5%) with the addition of buffer. For Western analysis, a 20-l aliquot of each adjusted sample was prepared in SDS-polyacrylamide gel electrophoresis sample buffer (1:1) and then loaded onto the gel. Anti-PKC-␣ mAb (0.05 g/ml) was employed as the primary Ab, horseradish peroxidase-linked sheep anti-mouse Ig (1:300 dilution) was the secondary Ab, and immunoreactive bands were detected by enhanced chemiluminescence. The region of the blots spanning 45-116 kDa is shown for each GSX ϩ/Ϫ paired sample; PKC-␣ is indicated by an arrow at 82 kDa.

TABLE III
The stability of the GSSG/GSH/GSO 3 -inactivated form of PKC-␣ under dialysis GSX-inactivated PKC-␣ and mock-inactivated control PKC-␣ were dialyzed for 8 -10 h under identical conditions against Tris-HCl buffer containing either 10 or 25 mM DTT. For dialysis buffer components and other experimental details, see the legend to Fig. 7. The postdialysis % inactivation was calculated by expressing the activity recovered from the GSX-inactivated PKC-␣ sample after dialysis as a percentage of the activity recovered postdialysis from mock-inactivated PKC-␣. The recovery of PKC-␣ protein was determined to be equivalent for GSX-inactivated and mock-inactivated PKC-␣ by Western analysis (see Fig. 7 established method that subjects deproteinized cell lysates to a 5,5Ј-dithiobis-(2-nitrobenzoic acid) colorimetric assay (43). In R6-PKC3 fibroblasts, prior to treatment, the GSH level was 4.03 Ϯ 0.23 nmol/10 6 cells, and subsequent to treatment, the GSH/N-acetylcysteine level was 10.60 Ϯ 0.33 nmol/10 6 cells. In MCF7-MDR cells, the initial GSH level was 4.24 Ϯ 0.37 nmol/ 10 6 cells, and the GSH/N-acetylcysteine level measured at the end of the treatment period was 12.24 Ϯ 0.04 nmol/10 6 cells. These results provide evidence that a 3-fold increase in the intracellular level of GSH may be sufficient to induce a marked inactivation of Ca 2ϩ -and PS-dependent PKC isozymes in mammalian cells. DISCUSSION The oxidative modification of select proteins by oxidized GSH, i.e. protein S-glutathiolation, has been shown to serve as a reversible mechanism of regulation of several enzymes, including carbonic anhydrase III and aldose reductase (21,22). Liu and Hannun (24) recently reported that at physiological concentrations, GSH reversibly inhibits isolated neutral mag-nesium-dependent sphingomyelinase by a nonredox mechanism but does not antagonize acidic sphingomyelinase. We report here that Ca 2ϩ -and PS-dependent PKC activity is also subject to antagonism by GSH by a nonredox mechanism, but in the case of PKC, the antagonism is irreversible. Incubation with GSH at physiological concentrations fully inactivated PKC, whereas it achieved less than 25% loss of casein kinase 2 activity, indicating that potent inactivation by GSH is not a general property of purified Ser/Thr protein kinases. Our evidence for a non-oxidative/reductive irreversible mechanism of PKC inactivation by GSH at physiological concentrations can be summarized as follows. GSSG antagonized the Ca 2ϩ -and PS-dependent activity of purified rat brain PKC with the same efficacy (IC 50 ϭ 3 mM) whether or not the reductant DTT (Ն100 mM) was present, and 2 equivalents of GSH antagonized PKC to the same extent as 1 equivalent of GSSG in the presence of DTT. GSO 3 , which is distinguished from GSSG and GSH by its inability to undergo disulfide/thiol exchange reactions, was as effective as GSSG in antagonizing Ca 2ϩ -and PS-dependent rat brain PKC catalysis. Purified recombinant PKC-␣ activity was likewise subject to antagonism by GSH, GSSG, and GSO 3 . An irreversible mechanism of PKC inactivation by GSX (GSH, GSSG, or GSO 3 ) was demonstrated by the stability of the inactivated form of PKC to dilution into PKC assay mixtures and to extensive dialysis.
The fact that GSH inactivation of PKC-␣ obeyed pseudo-first order kinetics indicates that the observed inactivation cannot be attributed to GSH-induced random unfolding and/or aggregation of PKC-␣, because random protein unfolding (denaturation) and aggregation do not obey first-order kinetics (44). In addition, the production of a 50-kDa catalytic domain fragment (18) from fully inactivated PKC-␣ by limited trypsinolysis corroborates the kinetic evidence against an inactivation mechanism involving random unfolding (denaturation) of PKC, because it shows that the hinge region of PKC-␣ continues to serve as a preferential target of trypsinolysis even after irreversible inactivation of the isozyme by GSH. An independent line of evidence that GSH inactivation of PKC involves specific interactions between the enzyme and the tripeptide is provided by the observation that PKC is also potently inactivated by the closely related tripeptide ␥-Glu-Gly-Gly but not by the amino acids Glu and Gly or the distantly related tripeptides Tyr-Gly-Gly and Gly-Ala-Gly. GSH and N-acetyl-Cys both induced DTT-insensitive PKC inactivation. The potent inactivation of PKC by ␥-Glu-Gly-Gly and N-acetyl-Cys suggests that at least two structural features of GSH represented by ␥-Glu-X-Gly and X-Cys-X contribute to its ability to inactivate PKC.
We determined that GSH-inactivated PKC-␣ did not contain covalently bound GSH by two experimental approaches, mass spectrometric analysis of the inactivated isozyme and quantitation of radiolabel irreversibly bound to [ 3 H]GSH-inactivated PKC-␣. The latter approach also demonstrated that the inactivated form of PKC-␣ does not contain tightly associated, reversibly bound GSH. In other words, the GSH-inactivated state of PKC-␣ persists in the absence of bound GSH. Proteolytic sensitivity of PKC can be used as a probe of the conformation of the enzyme (8,41,42). Our observation that the GSH-inactivated form of PKC-␣ is much more sensitive than control PKC-␣ to the proteolytic action of trypsin provides evidence that the inactivation mechanism involves either the induction of a stable conformational change in the isozyme that exposes the hinge region or the conversion of the isozyme structure to a destabilized state with increased conformational flexibility at the hinge region. Physical studies of the solution structures of native versus GSH-inactivated PKC-␣ will be required to distinguish between these two possibilities. Comparable inactivation of intact PKC and a proteolytically derived catalytic domain fragment of the enzyme by GSH, GSSG, and GSO 3 indicated that the inactivation mechanism was catalytic domain-directed. The protection against GSX inactivation of PKC offered by the substrates MgATP and histone was negligible, arguing against an active site-directed mechanism. In addition, the activating cofactors Ca 2ϩ and PS failed to protect PKC against GSX-mediated inactivation, providing evidence that both resting and activated conformations of PKC are susceptible to GSX inactivation. Although the close homology among the catalytic domains of PKC isozymes (15) would suggest that all PKC isozymes may be subject to GSH inactivation, the observations reported here are confined to GSH inactivation of isozymes in the Ca 2ϩ -and PS-dependent PKC subfamily, and the question of whether isozymes in the Ca 2ϩindependent PKC subfamilies are subject to GSX inactivation remains to be addressed.
Because N-acetylcysteine is a GSH precursor that is readily taken up by cells, N-acetylcysteine treatment is commonly used as a means of increasing the intracellular GSH pool (25,26). N-Acetylcysteine itself also functions effectively in cells as a free radical scavenger and reducing agent (25,26). In this report, we show that GSH and N-acetylcysteine irreversibly inactivate purified PKC isozymes by a nonredox mechanism with similar efficacy, and we report that brief treatment of cultured rat fibroblasts and human cancer cells with N-acetylcysteine results in a sharp decline in the level of Ca 2ϩ -and PS-dependent PKC activity in the cells, according to assays of extracts prepared from the cells. These results provide evidence that the mechanism of PKC inactivation by GSH/N-acetylcysteine in a purified enzyme system may also be operative in mammalian cells. Physiological stimuli, e.g. transforming growth factor ␤ 1 , tumor necrosis factor, and growth factor withdrawal, have been reported to increase the GSH level in mammalian cells by about 1.5-3.5-fold (45)(46)(47). We observed a marked inactivation of Ca 2ϩ -and PS-dependent PKC isozymes when cells were exposed to N-acetylcysteine under conditions that induced a GSH ϩ N-acetylcysteine level that exceeded the original GSH level by about 3-fold. Thus, our results suggest that the induction of increased GSH levels in mammalian cells by physiological stimuli may in some cases be associated with GSH-induced inactivation of PKC isozymes. Just how GSH-mediated PKC inactivation may regulate the enzymatic activity in cells is currently under investigation.
The importance of PKC as a molecular target in phorbol ester-mediated tumor promotion is well documented (1), but it is not yet clear whether PKC also plays an important role in oxidant-mediated tumor promotion (48). Based on the evidence reported here that GSH can antagonize PKC, we hypothesize that depletion of the intracellular GSH pool and the consequential loss of a negative regulatory mechanism over PKC isozymes may contribute to the tumor-promoting action of oxidants in nontransformed cells.