Role of oxidants and antioxidants in the induction of AP-1, NF-kappaB, and glutathione S-transferase gene expression.

Transcription factors AP-1 and NF-κB have been implicated in the inducible expression of a variety of genes in response to oxidative stress. Recently, based on the observation that butylated hydroxyanisole (BHA) and pyrrolidine dithiocarbamate (PDTC) induce AP-1 binding activity and AP-1-dependent gene expression and assuming that these compounds exert an antioxidant effect, it was claimed that AP-1 is an antioxidant-responsive factor. To determine whether AP-1 can be responsive to both oxidant and antioxidant, we examined the nature of BHA and PDTC inducing activity. Using EPR spectroscopy to detect semiquinone radicals, we demonstrate the autoxidation of BHA metabolite tert-butylhydroquinone (TBHQ) to tert-butylquinone. The kinetics of TBHQ-mediated generation of ·OH radicals were monitored in intact hepatoma HepG2 cells by EPR spin trapping technique. Exogenous catalase inhibited the rate and amount of ·OH radical formation and the induction of AP-1-mediated glutathione S-transferase (GST) Ya gene expression by BHA and TBHQ, thus indicating the intermediate formation of H2O2 in the metabolism of these chemicals. Furthermore, we show that the induction of AP-1 and NF-κB activities and GST Ya gene expression by BHA and TBHQ is due to a pro-oxidant activity, since this induction was inhibited by thiol compounds N-acetyl cysteine and GSH. Similarly, induction of AP-1 and GST Ya gene expression by PDTC was inhibited by N-acetyl cysteine and GSH. The present findings do not support the notion that the induction of AP-1 by BHA, TBHQ, or PDTC is an antioxidant response and demonstrate that both AP-1 and NF-κB activities are induced by oxygen radicals.

Transcription factors AP-1 and NF-B have been implicated in the inducible expression of a variety of genes in response to oxidative stress. Recently, based on the observation that butylated hydroxyanisole (BHA) and pyrrolidine dithiocarbamate (PDTC) induce AP-1 binding activity and AP-1-dependent gene expression and assuming that these compounds exert an antioxidant effect, it was claimed that AP-1 is an antioxidant-responsive factor. To determine whether AP-1 can be responsive to both oxidant and antioxidant, we examined the nature of BHA and PDTC inducing activity. Using EPR spectroscopy to detect semiquinone radicals, we demonstrate the autoxidation of BHA metabolite tert-butylhydroquinone (TBHQ) to tert-butylquinone.

The kinetics of TBHQ-mediated generation of ⅐ OH radicals were monitored in intact hepatoma HepG2 cells by EPR spin trapping technique. Exogenous catalase inhibited the rate and amount of ⅐ OH radical formation and the induction of AP-1-mediated glutathione S-transferase (GST) Ya gene expression by BHA and TBHQ, thus indicating the intermediate formation of H 2 O 2 in the metabolism of these chemicals. Furthermore, we show that the induction of AP-1 and NF-B activities and GST Ya gene expression by BHA and TBHQ is due to a pro-oxidant activity, since this induction was inhibited by thiol compounds N-acetyl cysteine and GSH. Similarly, induction of AP-1 and GST Ya gene expression by PDTC was inhibited by N-acetyl cysteine and GSH. The present findings do not support the notion that the induction of AP-1 by BHA, TBHQ, or PDTC is an antioxidant response and demonstrate that both AP-1 and NF-B activities are induced by oxygen radicals.
Reactive oxygen species, such as superoxide anion O 2 . , H 2 O 2 , hydroxyl radical ⅐ OH, organic peroxides, and radicals, are generated endogenously by all aerobic cells as byproducts of a number of metabolic reactions (1). Oxidative stress, which is an excess production of reactive oxygen species, can damage cells by lipid peroxidation and alteration of protein and nucleic acid structure. To prevent oxidative damage and allow survival in an oxygen environment, mammalian cells have developed an elaborate antioxidant defense system that includes nonenzymatic antioxidants (e.g. glutathione and vitamins C and E) as well as enzymatic activities such as superoxide dismutases, glutathione peroxidase, glutathione reductase, catalase, and other hemoprotein peroxidases (2). In addition, drug-metabolizing enzymes such as glutathione S-transferases, glucuronosyl transferases, and NAD(P)H:quinone reductase, by removing compounds capable of generating reactive oxygen species, decrease the level of oxidative stress and are also part of the antioxidant defense (2). Recent findings indicate that oxidative stress conditions enhance the expression of genes encoding antioxidant enzyme activities such as glutathione S-transferase (3,4), ␥-glutamyl cysteine synthetase (5), and heme oxygenase (6). Reactive oxygen species were also shown to be responsible for the inducible expression of genes involved in inflammatory and immune responses (7)(8)(9).
In eukaryotic cells enhanced production of reactive oxygen species and oxidative stress conditions are induced by a variety of stimuli, which include ionizing radiation, exposure to drugs and xenobiotics, or binding of cytokines to cell surface receptors (10). Current evidence indicates that the different stimuli use reactive oxygen species as signaling messengers to activate transcription factors and induce gene expression. Recently it was shown that the activation of transcription factor NF-B by different stimuli, such as cytokines TNF 1 and interleukin-1, phorbol ester, cycloheximide, lipopolysaccharide, and in some cell types H 2 O 2 , is inhibited by antioxidant compound NAC and metal chelators (7,8,11,12). Based on these observations it was concluded that the activation of NF-B, which occurs by a posttranslational mechanism involving dissociation of inhibitory protein IB (13), is controlled by reactive oxygen species and the intracellular redox state. Furthermore, using cell lines stably overexpressing H 2 O 2 -degrading enzyme, catalase, or cytoplasmic copper/zinc-dependent superoxide dismutase, which enhances H 2 O 2 production from superoxide, Schmidt et al. (14) show that among reactive oxygen species H 2 O 2 acts as a messenger for TNF-and okadaic acid-induced activation of NF-B.
Transcription factor AP-1 is an ubiquitous regulatory protein complex that interacts with AP-1 binding sites of target genes to regulate transcription in response to environmental stimuli (15). The AP-1 factor is composed of protein products of members of fos and jun proto-oncogene families, which form homodimeric (Jun-Jun) or heterodimeric (Fos-Jun) complexes. The expression of fos and jun genes was shown to be induced by a variety of extracellular stimulatory agents (e.g. serum, growth factors, phorbol esters, calcium ionophore, neurotrans-* This work was supported in part by the Israel Cancer Research Fund and the Leo and Julia Forchheimer Center for Molecular Genetics. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § Supported in part by a Gertrude and Benjamin R. Harris Scholarship. ʈ To whom correspondence should be addressed. Tel: 972-89-343639; Fax: 972-89-344118. 1 The abbreviations used are: TNF, tumor necrosis factor; BHA, 3(2)tert-butyl-4-hydroxyanisole; BHT, butylated hydroxytoluene; DDTC, diethyldithiocarbamate; DMPO, 5,5Ј-dimethyl-1-pyroline N-oxide; GST, glutathione S-transferase; NAC, N-acetyl cysteine; PDTC, pyrrolidine dithiocarbamate; Q c b , 2-phenyl-4-(butylamino)naphthol[2,3-h]-quinoline-7,12-dione; Q n , 2-phenyl-5-nitronaphthol[2,3-g]-indole-6,11-dione; TBHQ, tert-butylhydroquinone; TBQ, tert-butylquinone; TEMPO, 2,2,6,6-tetramethylpiperidine-1-oxyl-4-one. mitters, and ionizing radiation) that promote cell proliferation and transformation and neuronal excitation (15)(16)(17)(18). Recent studies on the induction of xenobiotic-metabolizing enzyme glutathione S-transferase Ya gene expression by a variety of structurally unrelated chemical agents have indicated the involvement of AP-1 in the cellular response to chemical stress. It was shown that the induction of a mouse glutathione S-transferase (GST) Ya gene expression, is mediated by a regulatory element, EpRE, composed of two adjacent AP-1-like binding sites that binds and is transactivated by the AP-1 complex (19,20). Regulatory elements with similar structure were observed to mediate the inducible expression of several other genes encoding drug-metabolizing enzymes (21).
Exposure of hepatoma cells to chemical inducers of GST Ya gene expression such as ␤-naphthoflavone, 3-methylcholanthrene, tert-butylhydroquinone, trans-4-phenyl-3-butene-2one, phorbol 12-myristate 13-acetate, dioxin, phenobarbital, hydrogen peroxide, arsenite, arsenate, and heavy metals was found to induce an increase in AP-1 binding activity (21)(22)(23)(24). The finding that chemical agents of diverse structure all activate GST Ya gene expression through the induction of AP-1 transcription factor that interacts with the EpRE enhancer, led to the assumption that the various chemicals may produce a common transduction signal responsible for AP-1 induction. It was observed that chemical inducers of GST Ya gene expression can generate by metabolism reactive oxygen species or can modify thiol compounds, both of which may cause depletion of reduced GSH and a change of intracellular redox equilibrium toward a more oxidizing environment (25). Recent studies support the hypothesis that the cellular redox state plays an important role in the induction of AP-1 and AP-1-mediated activation of GST Ya gene expression and indicate that this induction is associated with an increase in intracellular oxidant level (3,23). The role of reactive oxygen species in the induction of AP-1 activity and GST Ya gene expression and their effect on intracellular GSH levels were recently studied by exposing hepatoma cells to quinones with different capacities to generate oxygen radicals (4). The findings indicate a correlation between quinone-mediated production of ⅐ OH radicals, a decrease in GSH levels, and the induction of AP-1 binding activity and GST Ya gene expression. In view of the regulatory role played by the AP-1 transcription factor in the inducible expression of GST Ya and other drug-metabolizing enzymes by oxidants and thiol reagents (3,21), the induction of AP-1 appears to be part of a general mechanism of regulation of gene expression by oxidative stress.
Recent studies, however, based on the assumption that the inducible effect of 2(3)tert-butyl-4 hydroxyanisole (BHA) and pyrrolidine dithiocarbamate (PDTC) on the AP-1 binding activity and AP-1-mediated reporter gene expression is due to their antioxidant activities, have concluded that AP-1 is an antioxidant-responsive factor (26,27). Thus, AP-1 activity would be induced by antagonistic signals, by both oxidative stress and antioxidants. The answer to this apparently paradoxical situation may be found, in our opinion, in a reconsideration of the chemical and biological features of BHA, its metabolite tertbutylhydroquinone (TBHQ), and PDTC. Phenolic compounds such as BHA and BHT, termed phenolic antioxidants due to their chain-breaking action during autoxidation of lipids, are utilized for food preservation and suppression of lipid peroxidation in biological materials. However, these phenolic antioxidants were found to exhibit also tumor-promoting activity in rodents, an activity that seems to be related to the formation of oxidized metabolites (28). Indeed, BHA is oxidatively demethylated in mammalian tissues to TBHQ, which by autoxidation to tert-butylquinone (TBQ) may produce reactive oxygen spe-cies by redox cycling (29). Thus, BHA may have the potential to act both as oxidant and antioxidant. This raises the question of which of the two activities is involved in the induction of AP-1-mediated gene expression.
In the present study we investigated whether the induction of AP-1 activity and GST Ya gene expression by BHA and its metabolite TBHQ is due to generation of reactive oxygen species or to an antioxidant activity. The formation of semiquinone by auto-oxidation of TBHQ both in vitro and in intact cells was detected by EPR spectroscopy. The generation of ⅐ OH radicals by TBHQ metabolism was monitored by DMPO-⅐ OH spin trapping and EPR spectra measurements. In this report we present evidence that the induction of AP-1 and NF-B activities and GST Ya gene expression by phenolic antioxidants BHA and TBHQ is due to oxidation and quinone-mediated generation of oxygen radicals.
Cell Cultures and Plasmids-Human HepG2 and rat H4II (a differentiated cell line that still expresses GST Ya gene) hepatoma cells were grown in F12 and Dulbecco's modified Eagle's medium (1:1) with 10% fetal calf serum (32). HepG2 and H4II cells have been previously found to yield similar results concerning AP-1 binding of nuclear extracts, transient expression of GST Ya gene constructs, intracellular measurements of GSH levels, and quinone-mediated generation of ⅐ OH radicals (3,4,21). The EpRE Ya-cat plasmid construct, containing the EpRE 41-base pair enhancer of GST Ya gene ligated into the Ϫ187 site of its promoter driving the expression of CAT gene (32), was transfected for transient expression into HepG2 cells as described (3).
RNA Extraction and RNA Blot Analysis-Total cellular RNA was prepared by the guanidium thiocyanate extraction method (34) and was fractionated by electrophoresis on 1% agarose-formaldehyde gels followed by transfer onto nitrocellulose filters. The RNA blot was hybridized with 32 P-labeled probes for GST Ya cDNA and rRNA.
Measurement of Cellular Sulfhydryl Groups-Total -SH groups were measured in intact HepG2 cells by the EPR spectroscopy method of Weiner (35) as described previously by Pinkus et al. (4) and in cell extracts by optical methods, as described by Bergelson et al. (3), using the Ellman reagent (36) and by the Griffith (37) procedure .
Detection of Free Radical Intermediates-HepG2 cells grown to log phase were harvested by scraping, washed twice in phosphate-buffered saline (pH 7.2), and suspended in the same buffer at a density of 1 ϫ 10 5 to 2.5 ϫ 10 6 cells/ml. ⅐ OH radical formation was followed by EPR of spin adduct DMPO-⅐ OH (38). A typical 200-l incubation mixture for trapping ⅐ OH radicals contained 2.5 ϫ 10 5 cells, 100 mM DMPO, and 2 M to 1 mM TBHQ dissolved in Me 2 SO. For DMPO-⅐ OH radical measurements the Me 2 SO final concentrations were not higher than 0.005% and were obtained by serial dilutions with phosphate-buffered saline of 1 M TBHQ solution in 50% Me 2 SO. The DMPO in phosphate-buffered saline (pH 7.2) was purified before use by passing it twice through a syringe containing activated charcoal. The EPR spectra of DMPO-⅐ OH, semiquinone, and TEMPO radicals were recorded in a Brucker electron spin resonance ER200D-SRC spectrometer in a quartz capillary of 150 l. The standard instrumental parameters were as follows: microwave frequency, 9.7 GHz; incident microwave power, 10 mW; center of the field, 3480; scan range, 100 G; field modulation, 1 G; receiver gain, 6.3 ϫ 10 5 ; and time constant, 640 ms. The EPR spectrum of DMPO-⅐ OH consisted of a quartet (1:2:2:1) with hyperfine splitting constants of a N ϭa H ϭ 14.9 G. The amplitude of the second peak in the quartet of the EPR spectrum of DMPO-⅐ OH (Fig. 1b) was used for further calculations. The EPR spectrum of TEMPO consisted of a triplet (1:1:1) with hyperfine splitting constants of a N ϭ 14.9 G. For computer simulation of EPR spectra the Public EPR Software Distribution developed by D. R. Duling (NIEHS, National Institutes of Health) was used.

⅐ OH Radical Formation in tert-Butylhydroquinone-Treated
Cells-In order to quantitate the TBHQ-mediated formation of ⅐ OH radical in cultured cells, we have used the spin trapping of ⅐ OH by DMPO and measured the EPR spectrum of the resulting DMPO-⅐ OH spin adduct. Fig. 1b shows that incubation of intact HepG2 cells with 2 M TBHQ in the presence of 100 mM DMPO for 15 min before the EPR spectrum was measured, resulted in the formation of an EPR spectrum consisting of a quartet (1:2:2:1) with hyperfine splitting of 14.9 G, which is characteristic for the DMPO-⅐ OH spin adduct (39). An EPR spectrum consisting of six components was observed to be equally produced during incubation of 10 mM TBHQ in phosphate-buffered saline in the presence or absence of HepG2 cells (Fig. 1, c and d). Similar sextet spectra were registered during incubation of TBHQ in phosphate-buffered saline by Schilderman et al. (40). However, using a flat cell and varying the conditions of spectra registration of 10 mM TBHQ solution and the scan range to 25 G we obtained a well resolved EPR spectrum consisting of eight lines of equal intensity (Fig. 1e). This spectrum corresponds to the known EPR spectrum of the semiquinone of TBHQ (41). Computer simulation of the experimental spectrum (Fig. 1f) yielded the hyperfine splitting constants a 2 ϭ 0.08 G, a 3 ϭ 1.65 G, a 5 ϭ 2.13 G and a 6 ϭ 2.82 G, which are close to those estimated for the semiquinone of TBHQ in aqueous solution by Asworth (42). It should be noted that the EPR sextet spectra observed in Fig. 1 (c and d), which were ascribed by others to the semiquinone of TBHQ (40) are actually the result of inadequate experimental conditions for ESR spectra registration. The kinetics of ⅐ OH formation in TBHQ-treated cells were studied by measuring the amplitude of the EPR signal of DMPO-⅐ OH spin adduct (Fig. 1b)  EPR spectra for semiquinone were obtained by incubation of 10 mM TBHQ in phosphate-buffered saline in the presence (c) or absence (d and e) of 2.5 ϫ 10 5 HepG2 cells. The instrumental parameters in a, b, c, and d were as described under "Experimental Procedures," and for semiquinone radical (e) they were as follows: microwave frequency, 9.77 GHz; incident microwave power, 2 mW; field modulation, 320 mG; receiver gain, 5 ϫ 10 5 ; time constant, 320 ms; center of field 3490 G; scan range, 25 G; scan time 500 s. f, computer simulation of TBHQ semiquinone spectrum.
for each TBHQ concentration used there is an increase in the amount of DMPO-⅐ OH production as a function of time of incubation. However, both the kinetics of ⅐ OH radical formation and the maximum amplitude of the spin adduct EPR signal are found to decrease with the increase in TBHQ concentration being almost abolished at 1 mM. BHA, under conditions similar to those described for TBHQ in Fig. 2, did not produce any DMPO-⅐ OH spin adduct signal (data not shown). We should observe that in order to prevent Me 2 SO interference and formation of DMPO-⅐ CH 3 radical spin adduct, all EPR measurements of the DMPO-⅐ OH signal were carried out in the presence of Me 2 SO concentrations up to 0.005%.
Effect of Catalase on ⅐ OH Radical Formation in tert-Butylhydroquinone-and Adriamycin-treated Cells-To study the involvement of H 2 O 2 in ⅐ OH formation by TBHQ and adriamycin, a quinone producer of ⅐ OH radical previously studied by us (4), we have measured the effect of catalase on the kinetics of DMPO-⅐ OH spin adduct formation by these chemicals. Fig. 3 shows that the addition of catalase to an incubation mixture of intact HepG2 cells containing 2 M TBHQ or 100 M adriamy-cin and 100 mM DMPO-⅐ OH reduces both the kinetics and the maximal amplitude of the DMPO-⅐ OH signal. The effect seems to depend on catalase concentration. Moreover, the addition of catalase (0.1 or 1.0 mg/ml) after the DMPO-⅐ OH production reached steady-state levels is found to cause an immediate drop in DMPO-⅐ OH concentrations followed by a continued decrease with kinetics dependent upon catalase concentrations (Fig. 3).
Antioxidant Effects of tert-Butylhydroquinone-Assuming that the decrease in ⅐ OH radical production with the increase in TBHQ concentration observed in Fig. 2 may be due to an antioxidant effect of this chemical, we have studied the effect of TBHQ addition to an active ⅐ OH radical producer such as adriamycin. Fig. 4A indicates that the addition of 1 mM TBHQ to HepG2 cells incubated with 100 M adriamycin and 100 mM DMPO at the point where DMPO-⅐ OH spin adduct formation has reached steady state levels causes an immediate drop in DMPO-⅐ OH level. An inhibition of DMPO-⅐ OH formation is observed in cells preincubated for 30 min with 1 mM TBHQ before the addition of adriamycin. Suspecting that TBHQ may inhibit DMPO-⅐ OH spin adduct signal by reducing the DMPO-⅐ OH spin adduct to diamagnetic hydroxylamine, we have studied the effect of TBHQ on a stable nitroxyl radical such as TEMPO. Aliquots of 100 M TEMPO were incubated with increasing concentrations of TBHQ from 1 M to 70 mM, and the TEMPO radical signal was measured from the amplitude of the middle component of the triplet in EPR spectrum. The results presented in Fig. 4B show a decrease in TEMPO radical signal as a function of TBHQ concentration, which indicates a direct reduction of the nitroxyl radical and production of EPR silent hydroxylamine.
Effect of tert-Butylhydroquinone on Intracellular -SH Level-The intracellular -SH levels were measured in intact HepG2 cells exposed to TBHQ by the EPR spectroscopy method of Weiner (35) and in cell extracts by optical methods using Ellman reagent as described previously (3) or the Griffith (37) procedure. It was observed that 3-h exposure of cells to 100 M TBHQ caused a decrease in total -SH levels of about 18% as measured by the different methods (Table I).
Induction of AP-1 and NF-B Binding Activities by tert-Butylhydroquinone, BHA, and PDTC-TBHQ was previously shown to be an inducer of AP-1 binding activity (3,25). To determine whether this induction is the result of its antioxidant properties or is due to its oxidation to semiquinone/quinone and production of ⅐ OH radicals by redox cycling, HepG2 cells were exposed for 3 h to 30 M TBHQ in the absence of, or after a 60-min preincubation with, antioxidant NAC or GSH. Nuclear extracts were prepared and analyzed by electrophoretic mobility shift assay for AP-1 binding activity using an AP-1 binding site oligonucleotide probe. Fig. 5A shows that, compared with nuclear extracts from uninduced cells, exposure to TBHQ at concentrations where it generates ⅐ OH radicals (Fig. 2) resulted in an increase in AP-1 binding. Exposure of cells to NAC or GSH for 60 min before the addition of TBHQ was observed to inhibit the TBHQ induction of AP-1 binding activity (Fig. 5A). Using an NF-B binding site oligonucleotide  probe we observe also an induction of NF-B binding activity in nuclear extracts of cells exposed to 30 M TBHQ. This induction is inhibited by NAC or GSH (Fig. 5A). Similarly, both AP-1 and NF-B binding activities were induced in the nuclear extracts of cells exposed for 3 h to 30 M BHA, and their induction was prevented by GSH (Fig. 5B). PDTC was claimed to act as an inducer of AP-1 binding activity and AP-1-regulated gene expression and as an inhibitor of activation of NF-B binding activity by virtue of its antioxidant properties (26,27). To test whether PDTC acts as an antioxidant, HepG2 cells were exposed for 3 h to 100 M PDTC, nuclear extracts were prepared, and AP-1 and NF-B binding activities were measured by electrophoretic mobility shift using the respective oligonucleotide probes. Fig. 5C shows that PDTC induced the AP-1 binding activity and that this induction was inhibited by pre-exposure of cells to antioxidant NAC. GSH was also observed to inhibit PDTC induction of AP-1 activity (data not shown). In contrast, PDTC was not found to induce the NF-B binding activity (Fig.  5C).

Induction of AP-1 and NF-B Activities by
Quinones-In a previous study we showed that the induction of AP-1 binding activity by adriamycin and two synthetic quinones, Q c b and Q n , with different capacities to generate oxygen radicals, correlates with their oxygen radical production (4). Thus, adriamycin and Q c b , which can chelate Fe(III) ions and are more effective hydroxyl radical producers, both in vitro (30) and in vivo (4), were also stronger inducers of AP-1 binding activity than Q n . The effect of the different quinones on the induction of NF-B binding activity was presently studied in nuclear extracts of HepG2 cells exposed for 3 h to 2 M concentrations of adriamycin, Q c b or Q n . Fig. 6A indicates that the induction of NF-B binding activity by these quinones correlates with their capacity to produce ⅐ OH radicals, adriamycin and Q c b being stronger inducers that Q n . The formation of H 2 O 2 as an intermediate in the production of ⅐ OH radicals by adriamycin is indicated by catalase inhibition of AP-1 and NF-B induction. Antioxidant NAC is found to inhibit induction by adriamycin of both AP-1 and NF-B binding activities. The specificity of the induced NF-B binding activity was determined by competition experiments with a 100-fold molar concentration of unlabeled oligonucleotides containing NF-B or AP-1 binding sites, respectively (Fig. 6B).
Antioxidants Inhibit Induction of GST Ya Gene Expression by Adriamycin, PDTC, and tert-Butylhydroquinone-We have previously described the induction of GST Ya gene expression by TBHQ (25), adriamycin, and two other quinones with different capacities to produce ⅐ OH radicals and have shown that the induction of GST Ya mRNA by these quinones correlates with their oxygen radical production (4). To further study the effect of oxidants and antioxidants on GST Ya gene expression, hepatoma H4II cells that express this gene were exposed for 3 h to 2 M adriamycin, 100 M TBHQ, or 100 M PDTC in the absence or presence of NAC or GSH. Fig. 7 shows that the induction of GST Ya mRNA by adriamycin, TBHQ, or PDTC is considerably inhibited by the presence of the antioxidants.
NAC, GSH, and Catalase Inhibit Induction of EpRE Ya-cat by PDTC, tert-Butylhydroquinone, and BHA-To further question the nature of the PDTC activity in the induction of gene expression we studied the effect of PDTC on the expression of EpRE Ya-cat construct containing the inducible enhancer of GST Ya gene. HepG2 cells were transfected with EpRE Ya-cat plasmid and treated with PDTC directly or after a 60-min exposure to NAC or GSH. The induction of CAT activity by 10 or 100 M PDTC, which reached 10-and 30-fold respectively, was inhibited by both NAC and GSH (Fig. 8A). Furthermore, the activation of EpRE Ya-cat by 100 M TBHQ is found to be enhanced by concomitant addition of 10 or 100 M PDTC, thus showing that the two chemicals have similar effects. Since catalase is found to inhibit EpRE Ya-cat induction by TBHQ (Fig. 8A), it is evident that the activation of gene expression by Previously, it was shown that NAC and GSH inhibit EpRE Ya-cat induction by TBHQ (3). Similarly, induction of EpRE Ya-cat by BHA is found to be inhibited by GSH and catalase but not by PDTC (Fig. 8B). DISCUSSION In the present study we have addressed the question whether AP-1 can be induced by both oxidant and antioxidant conditions as recently claimed (26,27). We have examined the chemical properties of AP-1-inducing compounds, BHA and its metabolite TBHQ, and questioned the nature of their action on the activation of gene expression. In this study, using EPR spectroscopy, we have observed that TBHQ in buffered solution or in HepG2 cellular suspensions gives rise to paramagnetic molecular species (Fig. 1, c, d, and e). We ascribed these species to a semiquinone form of TBHQ on the basis of similarity of the experimental EPR spectra with those of TBHQ semiquinones described in the literature (41,42). It can be assumed that in solution, the autoxidation of TBHQ to semiquinone (TBQ . ) takes place according to the reaction, REACTION 1 which occurs at a noticeable rate when hydroquinones are in the anionic form (43,44). In fact we observed that the concentration of TBQ . increased dramatically with the pH and oxygen pressure (data not shown). The electron transfer from TBQ . into molecular oxygen follows the reaction,

REACTION 2
The rate of reaction (2) and the equilibrium concentration of TBQ . and O 2 . depend on the redox potential of TBQ/TBQ . and O 2 /O 2 . known as E1 ⁄2 ϭ Ϫ520 mV (in dimethylformamide) and E1 ⁄2 ϭ Ϫ155 mV (in water), respectively (45,46). Although E1 ⁄2 becomes more positive with an increase in the polarity of solvent, this does not seem be a marked effect for p-benzoquinone (47), and one can assume an E1 ⁄2 for TBQ/TBQ . in water that would still allow it to participate in O 2 . production according to Reaction 2. ⅐ OH radicals will be further generated from O 2 . by Haber-Weiss and Fenton reactions as described by Weiner (48). We have used the spin trapping EPR technique to monitor the generation of ⅐ OH radicals by TBHQ in intact cells. In the presence of spin trapping agent DMPO the TBHQ-induced ⅐ OH radical production was detected by EPR spectra of DMPO-⅐ OH spin adducts (Fig. 1b). The study of DMPO-⅐ OH formation in cells exposed to TBHQ shows a decrease in both kinetics and maximum amplitude of the spin adduct EPR signal with increase in TBHQ concentration (Fig. 2). This may indicate an antioxidant effect since TBHQ, at concentrations above 1 mM, caused an inhibition of DMPO-⅐ OH formation by adriamycin and a decrease in TEMPO nitroxyl radical signal (Fig. 4). BHA, under conditions similar to those where TBHQ is active in generating oxygen radicals, did not produce any DMPO-⅐ OH spin adduct signal (data not shown). This finding may be rationalized by postulating that (i) TBHQ, a metabolite of BHA, is not produced under the experimental conditions described in Fig. 2, which do not allow BHA demethylation, and (ii) BHA itself does not cause radical formation. This interpretation is consistent with the conclusions of Kahl et al. (29) that no oxygen activating properties can be ascribed to BHA itself. Working with intact HepG2 cells we show that the redox cycling of TBHQ/TBQ leading to the generation of ⅐ OH radicals, which is probably accelerated in cells by the availability of enzymes that support this process, occurs with intermediate formation of H 2 O 2 . Thus, the addition of catalase to the culture medium inhibited the ⅐ OH generation in a catalase concentration-dependent fashion (Fig. 3). This effect is surprising since, due to its molecular mass, catalase is not expected to enter the cells. It may, however, be explained by the difusibility of H 2 O 2 through biological membranes, which would allow catalase, without entering the cell, to lower intracellular H 2 O 2 levels by decomposition of extracellular H 2 O 2 and enhance the efflux gradient across plasma membrane. The addition of exogenous catalase to HepG2 cells is presently found to inhibit also the induction of AP-1 and NF-B binding activities by adriamycin (Fig. 6) as well as the induction of EpRE Ya-cat expression by TBHQ or BHA (Fig. 8).
GSH, the most abundant intracellular thiol compound in mammalian cells, plays a major role in maintaining the intracellular redox potential by regulating the levels of reactive oxygen species via the GSH peroxidase/GSSG reductase system. The redox metabolism of quinones generally results in a depletion of reduced GSH, which is due to enhanced detoxification of hydrogen peroxide by GSH peroxidase and oxidation of GSH to GSSG (49). Our present data indicate that a short term exposure (3 h) of HepG2 cells to TBHQ causes a modest (ϳ18%) but reproducible decrease in intracellular -SH levels and support the pro-oxidant notion for the activity of this chemical (Table I). Further evidence that TBHQ and BHA function as oxidants to induce AP-1 binding activity is provided by the inhibitory effect of thiol compounds NAC and GSH on this induction (Fig. 5). In previous studies we have shown that depletion of intracellular GSH levels, by the specific inhibition of ␥-glutamyl cysteine synthetase with L-buthionine-S,R-sulfoximine (50) or by direct oxidation to GSSG by diamide, causes an increase in AP-1 binding activity in the absence of chemical inducer as well as an enhanced induction of AP-1 binding activity by TBHQ. In addition, the depletion of intracellular GSH levels was found to enhance the activation of AP-1-mediated EpRE Ya-cat expression by a variety of chemical agents (3). The inducible increase in the AP-1 binding activity by chemicals was shown to involve the induction of fos and jun gene expression with accumulation of increased levels of the respective mRNAs (21,22,24) and a de novo synthesis of the AP-1 protein components (21,24).
In contrast to AP-1, the activation of transcription factor NF-B in response to pro-oxidant conditions produced by various stimuli (7,9,12) occurs by a post-translational mechanism, which involves the dissociation of the inhibitory protein IB. There are, however, similarities between the induction of AP-1 and NF-B transcription factor activation by reactive oxygen species. Working with adriamycin and two different quinones, Q c b and Q n , with different capacities to generate ⅐ OH radicals, we previously showed that the induction of AP-1 binding activity by these quinones correlates with their oxygen radical production (4). Similarly, in the present study we find that the induction of NF-B binding activity by these quinones is dependent on their capacity to generate ⅐ OH radicals, the two Fe(III) chelating quinones and potent ⅐ OH radical producers, adriamycin and Q c b , being stronger NF-B inducers than Q n (Fig. 6). We may therefore conclude that ⅐ OH radicals probably constitute the reactive oxygen species responsible for the induction of both AP-1 and NF-B transcription factors.
In view of the present findings, which show that TBHQ is a producer of ⅐ OH radicals and that the induction of AP-1 activity and AP-1-dependent gene expression by this chemical is due to its pro-oxidant properties, we have reconsidered the effect of BHA and its cellular metabolite TBHQ on NF-B activation. Recently Israel et al. (51) and Schulze-Osthoff et al. (9) have observed that BHA inhibits the phorbol 12-myristate 13-acetate-or TNF-induced activation of NF-B binding activity and attributed this inhibition to an antioxidant effect of BHA. We find, however, that a direct treatment of HepG2 cells with TBHQ or BHA, at the relatively low concentration of 30 M, causes an induction of NF-B binding activity (Fig. 5). Since this induction is inhibited by NAC and GSH, it is evident that TBHQ and BHA act as pro-oxidants to induce NF-B. It should be noted that pyrogalol, a triphenol antioxidant, was reported to generate reactive oxygen and induce NF-B binding activity (52). The inhibitory effects of BHA on phorbol 12-myristate 13-acetate or TNF induction of NF-B binding activity and NF-B-mediated human immunodeficiency virus enhancer activation described by Israel et al. (51) and Schulze-Osthoff et al. (9) may still be explained by the large amounts of BHA required (200 -400 M), which at these concentrations may have an antioxidant effect in the lymphoblastoid T (J. Jhan) and monocytic (U937) cell lines studied.
Because the pyrrolidine derivative of dithiocarbamate, PDTC, was shown to suppress activation of NF-B in response to diverse stimuli involving reactive oxygen production, being more potent than thiol compound NAC, its mode of action was attributed to an antioxidant thiol function (26). However, this conclusion may not be correct in view of the fact that dithiocarbamates may induce a variety of biological effects in mammalian cells that are not consistent with this hypothesis. Thus, dithiocarbamates PDTC, DDTC, and disulfiram (which may be reduced by GSH to DDTC), due to their metal-chelating properties, inhibit (copper/zinc) superoxide dismutase activity and were also shown to potentiate oxygen toxicity in animal tissues and cause a decrease in glutathione peroxidase activity and thiol levels (53)(54)(55). It was demonstrated that dithiocarbamates, like other xenobiotics bearing a thiol function, may cause oxidation of GSH by a nonradical mechanism (56). The thiol functions of these compounds are oxidized by microsomal flavin-containing monooxygenases to reactive intermediates sulfenic acids, which by a nonenzymatic reaction oxidize GSH to GSSG, regenerating the parent xenobiotic (57). In this respect dithiocarbamates seem to act catalytically, relatively small amounts leading to the oxidation of several hundred molar equivalents of GSH (56). In fact, exposure of thymocytes to PDTC was found to cause a fast increase in intracellular GSSG level (58). This, and the observation that nonthiol metal chelators, such as phenanthroline and desferal, also inhibit NF-B activation (12) suggests that the inhibitory effects of dithiocarbamates on NF-B activation may be due to their metal chelator properties rather than to an antioxidant effect of their thiol function.
The hypothesis regarding an antioxidant function for PDTC was also invoked to explain the induction of AP-1 activity by this chemical (26,27). In this report we show, however, that the PDTC induction of AP-1 binding activity is completely inhibited by NAC (Fig. 5C). Furthermore, we show that PDTC induction of EpRE Ya-cat expression is inhibited by NAC and GSH and is stimulated by the pro-oxidant activity of TBHQ (Fig. 8). In addition, the induction of endogenous GST Ya gene expression in H4II cells by PDTC, TBHQ, or adriamycin was also inhibited by NAC or GSH (Fig. 7). These findings clearly indicate that the activation of AP-1 transcription factor and GST Ya gene expression by PDTC is not due to an antioxidant effect of this chemical.
In conclusion, the present study presents evidence that phenolic antioxidant BHA and its active metabolite TBHQ have dual capacity to act both as antioxidants and as producers of reactive oxygen. Here we show, for the first time, that the induction of AP-1 and NF-B activities and GST Ya gene expression by BHA and TBHQ is due to the oxidative stress conditions generated by these chemicals. This conclusion is supported by the following findings: (i) exposure of cells to TBHQ causes generation of ⅐ OH radicals and a decrease in intracellular GSH level, (ii) induction of AP-1 and NF-B by BHA and TBHQ is inhibited by antioxidants NAC or GSH and by exogenous catalase, (iii) induction of endogenous GST Ya gene in hepatoma cells or a transfected EpRE-Ya cat construct by BHA or TBHQ is inhibited by NAC, GSH, and exogenous catalase. In view of these findings the regulatory element responsible for the inducible expression of GST Ya gene by TBHQ is actually an oxidative stress response element, and the term "antioxidant response element" (ARE) currently used by a number of laboratories (59,60) to define this element is misleading. In the present study we have also shown that induction of AP-1 and GST Ya gene expression by PDTC is due to an oxidative effect, since it is inhibited by antioxidants NAC and GSH but not by BHA or TBHQ.