Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Stilbene oxides have been shown to induce various phase I and II drug-metabolizing enzymes (Meijer and DePierre, 1983; Seidegård et al., 1979), ornithine decarboxylase (Oguro et al., 1991), a rate-limiting enzyme of polyamine biosynthesis and metallothionein (Sato et al., 1995). Both TSO and CSO, when administered in vivo to animals, could be converted into the corresponding diols by epoxide hydrolase (Watabe and Akamatsu, 1972, 1974) and/or conjugated with GSH by GSH S-transferase (deSmidt et al., 1987; Gill et al., 1983). The latter enzymatic metabolism of TSO could lead to a prompt depletion of hepatic GSH content (Oguro et al., 1991; Sato et al., 1995).

We have previously shown that many compounds capable of depleting hepatic GSH in rats are potent inducers of hepatic HO as well as ornithine decarboxylase and S-adenosylmethionine decarboxylase (Oguro et al., 1990; Yoshida et al., 1987). We have also demonstrated that both TSO and CSO induce ornithine decarboxylase, S-adenosylmethionine decarboxylase and metallothionein with a concomitant decrease in hepatic GSH content (Sato et al., 1995; Oguro et al., 1991). TSO also induces HO activity in normal Sprague-Dawley rats and the derived mutant hyperbilirubinuria rats (Oguro et al., 1996b). It is now fairly well established that HO is induced by oxidative stress resulting from GSH depletion or by some compounds that generate active intermediates in vivo and in vitro (Applegate et al., 1991; Oguro et al., 1990). However, the question of whether the compound is capable of inducing HO remains because general GSH depletors decrease GSH rapidly and it is difficult to prevent this depletion in vivo.

TSO and CSO are classified as phenobarbital-type P-450 inducers (Seidegård et al., 1979, 1981). Thus, if CSO can induce HO, these stilbene compounds may be useful for the study of the interrelated regulatory mechanisms in heme and heme-related proteins in the liver because it is still unclear how HO, P-450s and heme are regulated simultaneously.

Because there is a stereoselective difference in the metabolism of TSO and CSO by both GSH S-transferase(s) (deSmidt et al., 1987; Gill et al., 1983) and epoxide hydrolase (Gill et al., 1983; Watabe and Akamatsu, 1972), in the present study, we investigated the relation between HO induction and GSH depletion by the stereoisomers of stilbene oxide as well as by stilbene, parent compounds of stilbene oxides. Thus, this study was undertaken to examine the effects of TSO, CSO and their parent compounds on hepatic heme metabolism in relation to their abilities to deplete GSH and induce P-450s.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Chemicals. TSO, CSO, TS and CS were purchased from Aldrich Chemical (Milwaukee, WI). BSO was from Sigma Chemical (St. Louis, MO). PFDA was kindly donated by Dr. T. Ikeda (Sankyo Co., Tokyo, Japan). Deoxycytidine-5-[-³²P]triphosphate (3000 Ci/mmol) was obtained from Japan Isotope Association (Tokyo, Japan). All other reagents used were of the highest grade commercially available.

Animals and treatment. Male Wistar rats (6 weeks old) were obtained from Nippon Seibutu Zairyo Center (Tokyo, Japan). The rats were fed a commercial solid diet and water ad libitum for 1 week. Lighting was maintained on a 12-hr light/dark cycle, and temperature was maintained between 21° and 24°C .

Tissue preparation. Food was withdrawn for 24 hr before the experiment. The livers were perfused in situ with 0.9% NaCl, excised and homogenized with 5 vol of 1.15% KCl. The microsomal fractions were obtained by differential centrifugation, suspended in 0.1 M Na⁺/K⁺-phosphate buffer, pH 7.4, and used for the determinations of P-450 content and HO activity, and for immunoblot analysis.

Enzyme assay. P-450 content was determined from a CO-difference spectrum as described by Omura and Sato (1964). ALAS was determined according to the method of Marver et al. (1966). HO activity was assayed according to the method of Tenhunen et al. (1970), by using the 105,000 × g supernatant fractions from the control rats as the source of biliverdin reductase as described previously (Oguro et al., 1990). Hepatic GSH content was determined according to the method of Ellman (1959) as described by Costa and Murphy (1986). Protein concentration was measured according to the method of Lowry et al. (1951).

Electrophoresis and immunoblot analysis. The microsomes (10 µg of protein) were solubilized in SDS, and the proteins were separated by polyacrylamide gel electrophoresis (3% stacking gel, 10% separating gel) according to the method of Laemmli (1970). After electrophoresis, the proteins were transferred to nitrocellulose membranes (80 mA, 50 min; BioRad, Hercules, CA). Western blots were performed using a monoclonal antibody specific for CYP2B1/2 (kindly donated by Drs. T. Masuko and T. Hashimoto, Tohoku University, Sendai, Japan) and several polyclonal antibodies specific for CYP2B1/2, 3A2, 2C6, 2C11 and 2E1 as reported previously (Imaoka et al., 1989, 1990; Funae and Imaoka, 1993). The bands were identified by developing a peroxidase reaction with a mixture of 4-chloro-1-naphthol and H₂O₂.

RNA isolation. Total RNA was isolated from pooled rat livers using the acid guanidinium thiocyanate-phenol-chloroform extraction method (Chomczynski and Sacchi, 1987) as described previously (Oguro et al., 1996b). The RNA yield, purity and integrity were determined by absorbance at 260 nm, A_260nm/A_280nm ratio (>1.5) and 1% agarose/1× TAE gel electrophoresis, respectively.

Northern-blot analysis of RNA. The denatured total RNA (20 µg) was fractionated on a 1.0% agarose-formaldehyde gel. RNA samples were loaded with ethidium bromide (1 µg), and loading efficiency was checked by UV after electrophoresis. After electrophoresis, RNA was transferred onto a nylon membrane (NY 13 N, Schleicher & Schuell, Dassell, Germany). The fixed RNA was probed with the following ³²P-labeled cDNAs: HO-1 (0.9 kb, HindIII/EcoRI fragment of pRHO-1 cDNA; Shibahara et al., 1985), c-jun (2.1 kb, EcoRI/ClaI fragment of pBS rcjun-2 cDNA; Riken, Tokyo, Japan; Kitabayashi et al., 1990), CYP2B1 (1.25 kb, Pst1 fragment of rpcP-450pb1 cDNA; Riken), CYP2E1 (1.6 kb, PvuII/EcoRI fragment of hP450IIE cDNA; Japanese Cancer Research Resources Bank, Tokyo, Japan; Umeno et al., 1988) and GAPDH (0.5 kb, Pst1 fragment of GAPDH cDNA; kindly donated by Dr. K. Nose, Department of Microbiology, School of Pharmaceutical Sciences, Showa University, Tokyo, Japan). The filters were baked, prehybridized and hybridized at 42°C to ³²P-labeled cDNA probes as described previously (Oguro et al., 1996b). After hybridization, the membranes were washed with 2× standard saline citrate/0.1% SDS at 42°C for 1 hr and 0.5× standard saline citrate/0.1% SDS at 42°C for 1 hr. Semiquantification of the bands was performed with a bioimaging analyzer (model BAS3000; Fuji Photo Film Co., Tokyo, Japan).

Statistical analysis. The results were analyzed by Student's t test.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Figure 1 shows the time course effects of TSO and CSO on hepatic P-450 and GSH contents, and ALAS and HO activities. Hepatic GSH content decreased rapidly after the administration of either TSO or CSO and reached a nadir (~40% and ~20% of the control level, respectively) at 3 hr. GSH content returned to control levels by 24 hr after treatment with either TSO or CSO. The marked increase in HO activity was observed after administration of either stilbene oxide. HO activity started to increase at 3 hr and reached a peak at 24 hr (9-fold that of the control) after TSO and CSO administration. The return of HO activity to control levels was gradual, and the enzyme activity was ~4- and ~2-fold that of the control at 48 hr after TSO and CSO administration, respectively. Maximum ALAS activity (~3-fold over control) was observed at 12 to 48 hr after TSO injection. The increased ALAS activity by CSO (3-fold over control) was also sustained from 12 hr to 48 hr. The increase in hepatic P-450 content in CSO-treated rats (1.2-fold over control at 48 hr) was less pronounced than that seen in TSO-treated rats (1.7-fold over control at 48 hr).

View larger version (32K): [in this window] [in a new window]   Fig. 1.   Time course effects of TSO and CSO administration on hepatic P-450 and GSH contents, and ALAS and HO activities. Rats were injected with TSO (2 mmol/kg i.p.) or CSO (2 mmol/kg i.p.) and livers were removed at the times indicated. Control values throughout this experiment were as followed: GSH, 3.84 ± 0.16 µmol/g of liver; HO, 1.89 ± 0.24 nmol/mg of protein/hr; ALAS, 33.8 ± 3.00 nmol/g of liver; P-450, 0.773 ± 0.026 nmol/mg of protein. Symbols represent mean ± S.E. for three rats. *Significantly different from controls at P < .05.

Northern-blot analyses of HO-1, c-jun, CYP2B1/2 and CYP2E1 mRNAs (fig. 2A) and their semiquantification (fig. 2B) revealed that HO-1 mRNA increased significantly after TSO treatment and reached a maximum level (~30-fold) at the 12-hr time point. CSO induced HO-1 mRNA more strongly and quickly than TSO. HO-1 mRNA reached a maximum (~120-fold) at 4 hr after CSO treatment. Both TSO and CSO treatment also produced an increase in c-jun gene expression in parallel with the change in HO-1 mRNA. Treatment with TSO and CSO increased c-jun mRNA to 3- and 16-fold that of control at 6 hr, respectively. There was a continuous increase in liver CYP2B1/2 mRNA from 4 to 24 hr in TSO-treated rats. In the CSO-treated rats, the increase in CYP2B1/2 was maximum (22-fold increase) at 12 hr. In contrast, both TSO and CSO reduced the amount of CYP2E1 mRNA in a time-dependent manner. At 24 hr after TSO or CSO administration, CYP2E1 mRNA was the lowest, ~10% and ~20% of the controls, respectively (figs. 2, A and B).

View larger version (39K): [in this window] [in a new window]   Fig. 2.   Time course study for changes of HO-1, c-jun, CYP2B1/2 and CYP2E1 mRNAs in rat liver by Northern blot analysis. A, Rats were treated as described in the legend to figure 1, and total liver RNA was analyzed by Northern blotting as described in Materials and Methods. Hybridization was done with 32P-labeled cDNAs for HO-1, c-jun, CYP2B1, CYP2E1 and GAPDH. The vehicle alone did not produce an effect on mRNA levels. B, Changes in HO-1, c-jun, CYP2B1/2 and CYP2E1 mRNA levels in the liver treated with TSO or CSO. Each blot in A was calculated by BAS3000, and the extent of the blot for HO-1, c-jun, CYP2B1/2 and CYP2E1 was normalized with one for GAPDH.

Figure 3 shows immunoblot analyses of P-450 isozymes from microsomes of rat livers treated with TSO or CSO at different doses (0.5, 1 and 2 mmol/kg). CYP2B1/2 increased in a dose-dependent manner after TSO treatment, although total P-450 content did not increase appreciably. CSO also induced CYP2B1/2; however, the magnitude was less than that produced by the same dosage of TSO. Other phenobarbital-inducible P-450 isozymes, CYP2C6 and 3A2, were also increased by either TSO or CSO at the doses used in this study. Because polyclonal antibodies against CYP2C11 used in this experiment cross react with CYP2C6, as reported by Imaoka et al. (1990), there were two bands on a membrane. The top band, which represents CYP2C11, a male-specific isozyme, showed slight decrease in accordance with a dose increase in TSO. TSO treatment resulted in a decrease of CYP2E1 in a dose-dependent manner. There was almost no detectable CYP2E1 in liver microsomes 24 hr after TSO (1 and 2 mmol/kg) treatment. CSO (2 mmol/kg) also reduced CYP2E1 and 2C11 apoproteins at 36 hr (data not shown).

View larger version (34K): [in this window] [in a new window]   Fig. 3.   Immunoblot analysis of microsomes obtained from TSO- and CSO-treated rat livers. Rats were injected intraperitoneally with TSO or CSO (0.5, 1 and 2 mmol/kg), and tissues were obtained 24 hr after the administration. Ten micrograms of microsomal protein were loaded onto SDS-polyacrylamide gels, and immunoblot analyses were performed using polyclonal antibodies specific for CYP2B1/2, 3A2, 2C6, 2C11 and 2E1 as described in Materials and Methods. C, Control group.

Because TS and CS are metabolized to TSO and CSO, respectively, by P-450, we also examined the effects of TS and CS on GSH content and HO-1 and CYP2B1/2 mRNAs (fig. 4). CS depleted hepatic GSH (25% of the control) in a similar manner to CSO. On the other hand, TS (2 mmol/kg) failed to produce a pronounced effect on GSH content. Also, TS did not induce either HO-1 or CYP2B1/2 mRNA, whereas CS dramatically induced HO-1 mRNA (30-fold increment) and CYP2B1/2 mRNA (10-fold increment).

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Effects of TS and CS on hepatic GSH content and HO-1 and CYP2B1/2 mRNAs. Rats were treated with TS or CS (2 mmol/kg), and the liver was collected 4 hr after the injection. a, Total RNA was electrophoresed on a 1% agarose-formaldehyde gel. Hybridization was done with 32P-labeled cDNA for HO-1 or CYP2B1 and GAPDH. C, Control group. b, The extent of each blot was calculated by BAS3000 and normalized with the extent for GAPDH. c, GSH content was determined as described in Materials and Methods. Symbols represent mean ± S.E. for three rats. *Significantly different from controls at P < .05.

To study the relationship between GSH depletion and HO induction by TSO and CSO, the combined effect of BSO, an inhibitor of GSH biosynthesis, was studied (fig. 5). Because the combined treatment of rats with higher dose of stilbene oxide and BSO was highly toxic to animals, we used the reduced dose of the compound in this experiment. Pretreatment of rats with BSO (4 mmol/kg) before TSO or CSO (0.5 mmol/kg) injection resulted in a more marked depletion of GSH content (50% and 40% of the control, respectively) compared with the use of TSO or CSO alone (80% of the control). HO-1 mRNA was elevated markedly by the combined administration of BSO and TSO or CSO, and a ~3- or ~6-fold increase in the signal was observed compared with TSO or CSO alone, respectively. BSO pretreatment also slightly enhanced the expression of c-jun mRNA by TSO or CSO to a ~10- or ~15-fold increase, respectively. CYP2B1/2 gene expression by TSO was prevented by BSO, and that by CSO was decreased by BSO. The change of the enzyme activity was parallel to that of the gene expression. Namely, TSO or CSO produced a slight increase in HO activity (1.5-fold), and this effect was augmented by pretreatment with BSO (4- and 5-fold of each compound alone; data not shown).

View larger version (49K): [in this window] [in a new window]   Fig. 5.   Effect of pretreatment with BSO on TSO- and CSO-mediated changes of GSH and HO-1, c-jun and CYP2B1/2 mRNAs. Rats were pretreated with BSO (4 mmol/kg) 4 hr before the administration of TSO or CSO (0.5 mmol/kg). Livers were obtained 4 hr after the last injection. a, Total RNA was isolated from the liver and subjected to electrophoresis, blotting and hybridization as described in Materials and Methods. C, Control group. b, The extent of each blot was calculated by BAS3000 and normalized with the extent for GAPDH. c, GSH content was determined as described in Materials and Methods. Symbols represent mean ± S.E. for three rats. #Significantly different from controls at P < .05. *Significantly different from TSO or CSO alone at P < .05.

The effects of pretreatment of rats with BSO on the induction of CYP2B1/2 by TSO and CSO are shown in figure 6. Both TSO and CSO induced CYP2B1/2 apoprotein at 12 hr; however, pretreatment with BSO inhibited CYP2B1/2 induction by either TSO or CSO.

View larger version (44K): [in this window] [in a new window]   Fig. 6.   Effect of BSO pretreatment on the induction of CYP2B1/2 proteins by TSO and CSO. Rats were treated as described in the legend for figure 5, and microsomal protein in the rat liver at 12 hr after the last treatment was fractionated on SDS-polyacrylamide gels. Immunoblot analyses using a monoclonal antibody specific for CYP2B1/2 were investigated as described in Materials and Methods. C, Control group.

Hepatic GSH depletion evoked by TSO and CSO is due to their metabolism by GSH S-transferase. Therefore, GSH S-transferase was inhibited by PFDA, which is known to decrease GSH S-transferase proteins and mRNAs (Schramm et al., 1989), to reduce a limited GSH conjugation of stilbene oxides (fig. 7). Although TSO and CSO are known to induce GSH S-transferase, PFDA pretreatment produced 30% and 50% decreases in GSH S-transferase activity compared with each compound alone (data not shown). The GSH reduction produced by CSO was prevented by PFDA pretreatment. PFDA also inhibited HO-1 mRNA induction by CSO to 30% of the induced level, and it enhanced CSO-mediated CYP2B1/2 mRNA induction to ~9-fold more than CSO alone. The increases in HO activity by TSO and CSO (3.5- and 4.5-fold increase, respectively) were also inhibited significantly by pretreatment with PFDA (data not shown).

View larger version (37K): [in this window] [in a new window]   Fig. 7.   Effects of pretreatment with PFDA on CSO-mediated HO-1 and CYP2B1/2 mRNA changes. Rats were pretreated with PFDA (40 mg/kg) 3 weeks before the injection of CSO (1 mmol/kg), and livers were obtained 4 hr after the last injection. a, Hybridization was done with 32P-labeled cDNA for HO-1, CYP2B1 and GAPDH. C, Control group. b, The extent of blot for HO-1 and CYP2B1/2 were normalized with that for GAPDH. c, GSH was determined as described in Materials and Methods. Symbols represent mean ± S.E. for three rats. #Significantly different from controls at P < .05. *Significantly different from CSO alone at P < .05.

PFDA is a peroxisome proliferator; it is also known to induce CYP4A1 (Funae and Imaoka, 1993; Hardwick et al., 1987). To confirm whether the reduced HO activity by the pretreatment with PFDA could have an effect on TSO- or CSO-mediated CYP2B1/2 induction, immunoblot analysis of the isoform was examined (fig. 8). There was no visible band of CYP2B1/2 in rat livers treated with PFDA alone. TSO (1 mmol/kg) induced more CYP2B1/2 than CSO (1 mmol/kg). Pretreatment with PFDA enhanced the magnitude of the CYP2B1/2 band by TSO or CSO compared with the corresponding compound alone.

View larger version (47K): [in this window] [in a new window]   Fig. 8.   Effect of PFDA on TSO- and CSO-mediated CYP2B1/2 induction. Rats were treated as described in the legend for figure 7, and microsomal protein in the rat liver at 12 hr after the last administration was fractionated on SDS-polyacrylamide gels. Immunoblot analyses using a monoclonal antibody specific for CYP2B1/2 were investigated as described in Materials and Methods. C, Control group.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The results of the present study revealed that TSO, CSO and CS, a parent compound of CSO, are potent inducers of hepatic HO and that this HO induction is related to the ability of these stilbene compounds to decrease GSH content. HO induction evoked by TSO, CSO and CS is regulated at the transcriptional level, like other HO inducers (Applegate et al., 1991; Tacchini et al., 1995). CSO induced HO-1 mRNA more efficiently than TSO. This difference could be due to the difference in their abilities to deplete hepatic GSH. It has been reported that the metabolism of TSO and CSO by both epoxide hydrolase and GSH S-transferase precedes stereoselectivity (Gill et al., 1983; Watabe and Akamatsu, 1972), with CSO being a better substrate for both GSH S-transferase and microsomal epoxide hydrolase (Gill et al., 1983). Thus, CSO may consume GSH more rapidly and cause stronger oxidative conditions than TSO. This hypothesis is supported by the result of the effect of BSO pretreatment on HO-1 gene expression of stilbene oxides. Because BSO did not cause HO-1 gene expression in this experimental conditions, a rapid and drastic GSH depletion could be a trigger that leads to the HO-1 gene expression.

TSO and CSO increased c-jun mRNA in parallel with HO-1 mRNA induction but not c-fos mRNA (data not shown). The oxidative stress has been indicated to activate AP-1 binding activity to a cis-regulatory element of some genes (Bergelson et al., 1994; Friling et al., 1992; Pinkus et al., 1995). The rat HO-1 gene has also shown to include two AP-1-like sequences in the 5-flanking regulatory region (Müller et al., 1987), and HO-1 gene expression by different oxidative stress conditions has been revealed to be associated with AP-1 binding activation (Camhi et al., 1995; Lee et al., 1996; Oguro et al., 1996a). In addition, BSO pretreatment markedly increased c-jun and HO-1 mRNA induction by CSO more than that by TSO. All of these findings suggest that a drastic GSH depletion could induce AP-1 binding activity, resulting in the stimulation of the HO-1 transcription.

CS elevated HO-1 mRNA and enzyme activity (data not shown), whereas TS did not. Because CS, but not TS, depleted GSH as did CSO, the compound may be oxidized efficiently to CSO by P-450, as has been shown in vitro by Watabe and Akamatsu (1974). Therefore, the differences in the effect between TS and CS on HO and CYP2B1/2 could be ascribed to their differential metabolic fates in the liver, although there are no available data concerning the metabolic fate of both compounds in vivo at this time. The repeated treatment of rats with TS induced P-450 (Seidegård et al., 1981); however, TS had no effect on CYP2B1/2 mRNA in the present study. The results again suggest the slow metabolic fate of TS. In addition, the results suggest that TS and CS themselves do not have the ability to induce either HO-1 and CYP2B1/2.

As suggested by Dwarki et al. (1987), the decreased heme pool due to the induction of HO causes a decreased P-450 mRNA transcription as well as newly synthesized apoprotein degradation. However, this study showed that the transcription of the CYP2B1/2 gene seemed to be affected when the transcription of the HO-1 gene changed in accordance with the results of pretreatment with either BSO or PFDA (i.e., the CYP2B1/2 gene expression was stimulated before the heme pool decreased markedly by newly synthesized HO). Although we did not examine a run-on assay for the CYP2B1/2 or HO-1 gene or stability of these two mRNAs, these two genes have been shown to be regulated at the transcriptional level (Gonzalez, 1988; Tacchini et al., 1995). Therefore, our data suggest that there are some factors or pathways that control both CYP2B1/2 and HO-1 gene transcriptions under an oxidative stress produced by TSO and CSO. A regulation of CYP2B1/2 gene expression is not still clear, although there have been many reports concerning a cis-acting DNA element, including a negative regulated element, and nuclear protein factors (Hoffman et al., 1992; Nirodi et al., 1996; Ram et al., 1995; Shaw et al., 1996; Upadhya et al., 1992). Therefore, our results might be useful in the search for another regulator that is involved in either down-regulation of CYP2B1/2 or up-regulation of HO-1.

CSO produced similar effects on heme metabolism, although the maximum ALAS activity produced by the compound tended to be smaller than TSO. The ability of CSO to induce CYP2B1/2 was rather less extensive than TSO, as seen similarly in total P-450 induction after repeated treatment with these compounds (Seidegård et al., 1981). Taken together, it might be possible to point out that CSO disappears from the liver faster than TSO, thus leading to the pronounced increase in heme degradation, so as to retard an available heme pool for P-450 synthesis and/or to restrict access to the machinery of this P-450 synthesis. In this respect, however, further study will be required.

TSO decreased CYP2E1 apoprotein, which is in accordance with the report of Thomas et al. (1987). In addition, this study revealed that both TSO and CSO reduced CYP2E1 mRNA in a time-dependent manner, suggesting that both compounds inhibit CYP2E1 apoprotein synthesis and stimulate its degradation. Stilbene oxides also decreased CYP2C11 apoprotein. It is suggested that the oxidation of P-450 heme by HO is due to a differentiated degree of lability of P-450 (Kutty et al., 1988; Trakshel et al., 1986). Because either CYP2E1 or 2C11 apoproteins started to decrease at 12 hr after TSO injection (data not shown), newly synthesized HO degraded these isozymes, which may be more labile than other isozymes. On the other hand, CYP2E1 and 2C11 have been shown to be down-regulated by cytokines and endotoxin (Morgan, 1993; Morgan et al., 1994). An inflammatory condition that might be produced in the GSH-depleted condition evoked by stilbene oxides could influence a subsequent down-regulation of CYP2E1 gene expression. However, we do not know why the effect of CSO on either CYP2E1 or 2C11 appeared to be slower than that of TSO.

In conclusion, this study has revealed that stilbene oxides are potent HO-1 inducers as well as P-450 inducers in rat livers. This HO-1 induction by stilbene oxides could be related to their abilities to deplete hepatic GSH.