Cytochrome P450 2E1 Responsiveness in the Promoter of Glutamate-Cysteine Ligase Catalytic Subunit. L-glutamate L-cysteine

Cytochrome P450 2E1 Responsiveness in the Promoter of Glutamate-Cysteine Ligase Catalytic Subunit Natalia Nieto, Montserrat Marí, and Arthur I. Cederbaum Previous studies have shown cytochrome P450 2E1 (CYP2E1)-dependent transcriptional up-regulation of glutamate-cysteine ligase (GCL). To identify sequences mediating constitutive and induced expression of the catalytic subunit of GCL (GCLC), a series of deletion mutants from the 5 -flanking region ( 3,802 to 465) were transfected into control (C34) and CYP2E1-overexpressing (E47) HepG2 cells. Increased luciferase expression, both basal (2- to 3-fold) and following exposure to ethanol, arachidonic acid (AA), or AA plus iron, was detected in E47 cells with the full-length but not shorter reporter vectors. Basal induction was blocked by CYP2E1 inhibitors and catalase. Basal and inducible luciferase expression in E47 cells was blunted by the full-length construct mutated in the ARE4 site. Catalase and diallyl sulfide prevented basal and AA-induced messenger RNA (mrna) levels of GCLC and the modulatory subunit of GCL (GCLM). Preincubation with low doses of AA increased glutathione (GSH) levels as well as GCLC and GCLM mrnas, and this protected against H 2 O 2 and menadione toxicity. Primary hepatocytes from pyrazole-injected rats with high levels of CYP2E1 showed an increase in GSH levels as well as GCLC and GCLM mrnas compared with saline controls, and this was prevented by diallyl sulfide. In conclusion, redox-sensitive elements directing constitutive and induced expression of the GCLC in CYP2E1-expressing cells are present in the ARE4 distal portion of the 5 -flanking region, between positions 3,802 and 2,752, perhaps a reflection of metabolic adaptation to CYP2E1-generated oxidative stress. (HEPATOLOGY 2003;37:96-106.) Abbreviations: GSH, glutathione; ATP, adenosine triphosphate; GCL, glutamate-cysteine ligase; GCLC, catalytic subunit of glutamate-cysteine ligase; GCLM, modulatory subunit of glutamate-cysteine ligase; ARE, antioxidant response element; EpRE, electrophile response element; CYP2E1, cytochrome P450 2E1; ROS, reactive oxygen species; AA, arachidonic acid; mrna, messenger RNA. From the Department of Pharmacology and Biological Chemistry, Mount Sinai School of Medicine, New York, New York. Received July 18, 2002; accepted October 3, 2002. Supported by U.S. Public Health Service grants AA03312 and AA06610 from the National Institute on Alcohol Abuse and Alcoholism (to A.I.C.) and the Revson Fellowship and a grant from the Alcohol Beverage Medical Research Foundation (to N.N.). Presented in abstract form at the 2002 annual meeting of the American Association for the Study of Liver Diseases in Boston, MA. Address reprint requests to: Arthur I. Cederbaum, Ph.D., Department of Pharmacology and Biological Chemistry, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1603, New York, NY 10029. E-mail: arthur.cederbaum@mssm.edu; fax: 212-996-7214. Copyright 2003 by the American Association for the Study of Liver Diseases. 0270-9139/03/3701-0017$35.00/0 doi:10.1053/jhep.2003.50003 Glutathione (GSH) is one of the most abundant antioxidants in cells. 1-6 There are several mechanisms by which cells maintain their intracellular GSH content, such as GSH redox cycling, direct uptake, and de novo synthesis. 7,8 Most cells, under normal and oxidative stress conditions, depend on de novo synthesis to maintain their intracellular GSH levels. 5 De novo synthesis of GSH involves 2 adenosine triphosphate (ATP)-requiring enzymatic steps: L-glutamate L-cysteine GCL ATP O -glutamyl-l-cysteine adenosine diphosphate P i (1) GS -glutamyl-l-cysteine L-glycine ATP O GSH adenosine diphosphate P i. (2) The first step is rate limiting and catalyzed by glutamate-cysteine ligase (GCL). GCL is composed of a heavy or catalytic (GCLC; relative molecular mass, 73,000 dalton) subunit and a light or modulatory (GCLM; relative molecular mass, 30,000 dalton) subunit, which are encoded by different genes in both rats and humans. 9-13 Although the heavy subunit is active catalytically, it has a higher Michaelis constant value for glutamate and a lower inhibitory constant value for GSH compared with the 96

HEPATOLOGY, Vol. 37, No. 1, 2003 NIETO, MARÍ, AND CEDERBAUM 97 holoenzyme. 14,15 Thus, the light subunit plays an important regulatory role for the overall function of the enzyme and allows the holoenzyme to be catalytically more efficient and less subject to inhibition by GSH than the heavy subunit alone. GCL is regulated physiologically by feedback competitive inhibition by GSH 15,16 and by the availability of its precursor, L-cysteine. 1,5,7 Overexpression of glutathione synthase failed to increase GSH levels, whereas overexpression of GCL increased GSH levels, consistent with the fact that GCL is the rate-limiting enzyme of GSH synthesis. 17,18 The 5 -flanking region of the human GCLC has been cloned and sequenced. Various putative transcription factors such as AP-1 (c-fos/c-jun)/tre, AP-1 like, nuclear factor B, AP-2, Sp-1, metal response element, and antioxidant response (ARE)/electrophile response (EpRE) elements have been identified. 19-23 In a series of deletion mutants created from the 5 -flanking region ( 3,802 to 465) of the GCLC cloned into a luciferase reporter vector and transfected into HepG2 cells, Mulcahy et al. 20,21 have shown that elements directing constitutive and induced expression of GCLC are present in the distal portion of the 5 -flanking region, between positions 3,802 and 2,752. Sequence analysis showed the presence of several putative consensus response elements in this region, including 2 potential ARE elements (ARE3 and ARE4), separated by 34 base pairs, of which ARE4 seemed the most critical. 20,21 Ethanol induces cytochrome P450 2E1 (CYP2E1), which can metabolize and activate numerous hepatotoxicants in the liver. 24 CYP2E1 is a loosely coupled enzyme that produces reactive oxygen species (ROS) in high amounts. 25-27 Overexpression of CYP2E1 in HepG2 cells induces GCLC and GCLM through a transcriptional mechanism. 28 This induction is associated with an increase in GSH levels and is blocked by antioxidants. The effects of ethanol or pro-oxidants believed to be important in alcoholic liver disease, such as polyunsaturated fatty acids (arachidonic acid [AA]) and iron 29 on GCL induction, have not been determined. In the present study, deletion analysis was used to identify the sequences that modulate the constitutive and oxidant-induced expression of the human GCLC gene 5 flanking region, and we have evaluated a possible link between the pro-oxidant/antioxidant status of CYP2E1- expressing cells, GCL messenger RNA (mrna) expression, and GSH levels on exposure to ethanol, AA, and iron. Materials and Methods Cell Culture, Standard Protocols, and Chemicals. E43 and E47 cells are HepG2 cells that constitutively express human CYP2E1 after transfection with pci-neo- CYP2E1 and selection. 30 C34 cells are HepG2 cells transfected with pci-neo and do not express CYP2E1. 3A4 cells are HepG2 cells that express human CYP3A4 for comparison with results obtained with human CYP2E1 (E47 or E43 cells). Most of the chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). Primary hepatocytes from saline- or pyrazole-treated rats (200 mg/kg body wt/day for 2 days, followed by an overnight fast) were isolated from male Sprague-Dawley rats (200 g) (Charles River Laboratories, Wilmington, MA) as described by Wu and Cederbaum. 31 The protocol for this study was approved by the Mount Sinai Institutional Animal Care and Use Committee, and rats received humane treatment. Pyrazole was used to elevate the level of CYP2E1 in hepatocytes, whose content and activity was measured by Western blot and by the rate of oxidation of p-nitrophenol to p-nitrocatechol according to previous protocols. 32 Anti-GCLC and anti-gclm antibodies were kindly provided by Dr. T. J. Kavanaugh (University of Washington). 33,34 Northern blot analysis for GCL mrna and the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide viability assay were performed as described. 28,32 The adenovirus-lacz and the adenovirus containing the complementary DNA for catalase were from Dr. Jingxiang Bai of our laboratory. 35 Western blot analysis for GCLC and GCLM protein and for CYP2E1 was performed as previously described. 28,32-34 Transfection Experiments. Reporter DNA constructs containing upstream sequences of the GCLC promoter linked to the firefly luciferase gene (Luc) were kindly provided by Dr. T. Mulcahy (University of Wisconsin). 21 GCLC sequences span from 3,802 to 465 ( 3,802/GCLC 5 -Luc), from 2,752 to 465 ( 2,752/GCLC 5 -Luc), from 1,286 to 465 ( 1,286/GCLC 5 -Luc), and from 814 to 465 ( 814/GCLC 5 -Luc). For 3,802mA4/GCLC 5 -Luc, a single base mutation (T 3 G) was introduced into the ARE4 of 3,802/GCLC 5 -Luc by site-directed mutagenesis. This mutation disrupts both the ARE4/EpRE and the internal AP-1 site. These 5 -deletion mutants and the luciferase reporter vector pgl3-basic were cotransfected along with the renilla luciferase reporter vector (prl-null) as a control for transfection efficiency. Transfections were performed using Fugene 6 (Boehringer, Indianapolis, IN) as DNA carrier plus 100 nmol/l of each of the deletion constructs and 20 nmol/l of prl-null vector. pbluescript II KS- was added to complete the amount of transfected plasmid DNA to 1 g. Luciferase activity was determined using the dual-luciferase reporter assay system (Promega, Madison, WI). All 3 vectors, pgl3-basic, prl-null, and pbluescript II KS-, were purchased from Promega.

98 NIETO, MARÍ, AND CEDERBAUM HEPATOLOGY, January 2003 Fig. 1. (A) Mutational analysis of the GCLC 5 -flanking sequence. A 4.2-kilobase HindIII fragment ( 3,802 to 465) from the 5 -flanking region of the GCLC gene was cloned into the HindIII site of the luciferase reporter vector pgl3-basic to create a recombinant plasmid 3,802/GCLC 5 -Luc. A series of progressively smaller transgenes were created by digesting 3,802/GCLC 5 -Luc with the restriction enzyme indicated on the restriction map at the bottom of the figure. For 3,802mA4/GCLC 5 -Luc, a single base mutation (T 3 G) was introduced into ARE4/AP-1 element/site of 3,802/GCLC 5 -Luc by site-directed mutagenesis. These constructs were prepared by Dr. Jerry Gipp and provided by Dr. Timothy Mulcahy (University of Wisconsin). (B) Activities of the GCLC reporter transgenes in different cell lines. Firefly luciferase activity associated with transfection of the different GCLC promoter/luciferase transgenes in C34, 3A4, E43, and E47 cells is shown. The results are corrected by transfection efficiency using the luciferase activity from a cotransfected renilla luciferase reporter vector (prl-null) and by protein content and are normalized to the corrected firefly luciferase activity detected in each cell line transfected with the luciferase reporter vector pgl3-basic in which the recombinant expression vectors were created. Results are expressed as average numbers of 4 determinations SE. *P.05, **P.01, ***P.001 compared with C34 cells transfected with each construct. GSH Assay. GSH was determined by the method of Tietze. 36 Statistics. Results are expressed as mean SE, with the number of experiments indicated in the figure legends. Statistical evaluation was performed using Student s t test. Results Identification of the CYP2E1-Responsive Element in the Promoter of GCLC. Previous studies have shown that GCLC mrna levels were elevated in E47 cells compared with C34 or CYP3A4 HepG2 cells. 28 Transfection studies with the series of deletion constructs of the GCLC promoter shown in Fig. 1A were performed to attempt to map the CYP2E1-responsive element. Two days after transfection, cells were lysed and luciferase activities were measured (Fig. 1B). There was a 3- to 4-fold increase in the activity of the 3,802/GCLC 5 -Luc expression vector in the E47 cells (P.001) compared with the activity in C34 cells, which do not express CYP2E1. A 2-fold increase was found in E43 cells, which express a lower level of CYP2E1 compared with E47 cells. Importantly, HepG2 cells expressing a different P450, CYP3A4, did not show an increase in activity of the 3,802/GCLC 5 -Luc expression vector. The increase found with the E47 and E43 cells was blunted by transfection with the mutated construct ( 3,802m4A/GCLC 5 -Luc). Of interest is the fact that the 3,802m4A/GCLC 5 -Luc harbors a base substitution that affects both the ARE4/EpRE and the embedded AP-1 binding site, a redox-sensitive transcription factor. This region of the GCLC promoter seems to be critical for the constitutive and induced expression of the human GCLC gene 21 and is now shown to be important as a CYP2E1-responsive element. Ethanol, AA, and AA Plus Iron Induce Expression of GCLC and GCLM mrnas and Affect GSH Levels. CYP2E1 generates high amounts of ROS. 25,26 Ethanol elevates CYP2E1 levels, and ethanol metabolism by CYP2E1 generates the 1-hydroxyethyl radical among other toxic species. AA, as a component of cell membranes, is a target for auto-oxidation and is susceptible to lipid peroxidation. In animal models, polyunsaturated fatty acids enhance the toxic potential of ethanol. 37 We therefore considered the possibility that both ethanol and AA, as a representative polyunsaturated fatty acid, could increase GCL activity through a transcriptional mechanism and elevate GSH levels to help the cell cope with the undergoing oxidative stress. Iron was also chosen because it is known to potentiate ethanol and AA toxicity both in vivo 38 and in vitro. 39,40 Evidence for oxidative stress in

HEPATOLOGY, Vol. 37, No. 1, 2003 NIETO, MARÍ, AND CEDERBAUM 99 Fig. 2. (A) Northern blot analysis of GCLC and GCLM mrnas following treatment with ethanol, AA, and AA plus iron. A total of 10 g ofrna isolated from C34 and E47 cells treated with 100 mmol/l ethanol or 30 mol/l AA for 24 hours or 10 mol/l AA for 23 hours plus 10 mol/l ferric nitrilotriacetate for 1 hour were size fractionated in an 0.9% agarose/2.2 mol/l formaldehyde gel, pressure blotted onto a nylon membrane, cross linked, and hybridized with labeled probes for GCLC purchased from American Type Culture Collection (Manassas, VA) and GCLM generated by polymerase chain reaction amplification of a 512 base pair spanning from 411 to 923 base pairs. Subsequently, the membrane was hybridized to a glyceraldehyde-3-phosphate dehydrogenase loading control probe. Results are expressed as relative GCLC and GCLM mrna expression, average values (n 3) SE, with the untreated C34 cells labeled as C34 control assigned a value of 1. *P.05, ***P.001 for E47 versus C34 cells for each treatment. Comparisons were also made for the effect of the treatment versus each nontreated cell line ( P.05, P.005, P.001). (B) Western blot analysis of GCLC and GCLM proteins after treatment with ethanol, AA, and AA plus iron. Twenty-five micrograms of protein of C34 and E47 cells treated as described in A were resolved in an 8% or 10% polyacrylamide gel electrophoresis for GCLC and GCLM, respectively, transferred onto polyvinylidene difluoride membranes, and blotted with antibodies for GCLC and GCLM kindly provided by Dr. T. J. Kavanagh (University of Washington). (C) GSH levels. A parallel experiment was performed to measure intracellular GSH levels at 24 hours of culture in C34 and E47 cells using the GSH recycling method of Tietze. 34 Results are expressed as nanomole GSH per milligram of protein and are average values of 4 determinations SE. (**P.01, ***P.001 E47 cells compared with C34 per treatment). Comparisons were also made for the effect of the treatment versus each nontreated cell line ( P.05, P.005, P.001). E47 cells generated mainly via metabolism of CYP2E1 under ethanol, AA, and AA plus iron treatments has been reported. 31,39 mrna levels for GCLC and GCLM were about 2-fold higher in E47 cells compared with C34 cells in the absence of any treatment (Fig. 2A, control E47 and C34 blots). Northern blot analysis showed an increase in GCLC mrna levels of about 40% to 50% with ethanol and AA plus iron and almost 100% with AA in E47 cells compared with nontreated E47 cells; GCLM seemed less responsive, showing increases with AA and AA/iron of about 50% and 30%, respectively (Fig. 2A). GCLC mrna levels in C34 cells were not significantly altered by these treatments. Western blot analysis showed a parallel 2-fold increase in both basal GCLC and GCLM protein levels in E47 cells compared with C34 cells, in agreement with the elevated mrna levels. Incubation in the presence of ethanol or AA further increased both GCLC and GCLM protein expression in both cell lines (2-fold in

100 NIETO, MARÍ, AND CEDERBAUM HEPATOLOGY, January 2003 Fig. 3. (A) Ethanol, (B) AA, and (C) AA plus iron further increase the activity of the 3,802/GCLC 5 -Luc reporter construct in E47 cells. Twenty-four hours after transfection, C34 and E47 cells were treated with 100 mmol/l ethanol or 30 mol/l AA and incubated for an additional 24 hours before cell harvest and lysis. For the combined treatment, cells were incubated with 10 mol/l AA for 23 hours and with 2.5 mol/l ferric nitrilotriacetate for 1 hour. Luciferase activity for each group was corrected by transfection efficiency and for protein content and normalized to that of the pgl3 basic vector in each cell line. Results are expressed as average numbers of 4 determinations SE. *P.05, **P.01, ***P.001 for each treatment (ethanol, AA, or AA plus iron) versus no treatment for each cell line. C34 cells and 3-fold in E47 cells) (Fig. 2B). Interestingly, the effects of ethanol and AA on GCLC and GCLM protein levels were more dramatic than on the respective mrna levels. Whether this reflects effects on translation of the mrna or on stability of the corresponding protein remains to be evaluated. To determine whether these changes in GCL subunits mrna and protein reflected increased GSH synthesis under oxidative stress conditions, total GSH levels were determined in cell lysates (Fig. 2C). In the absence of any treatment, there was a 2-fold increase in the basal levels of GSH between both cell lines (P.001) in agreement with the elevated mrna and protein levels of GCL. Treatment with either ethanol or AA further increased GSH levels in both cell lines, although levels remained higher in the E47 cells. These increases in GSH levels approximate the increases in GCL protein by ethanol and AA. Incubation with iron plus AA depleted GSH levels in both cell lines by about 50%, despite the increase in GCLC mrna and protein levels (Fig. 2A-C). This likely reflects the increase in cellular toxicity and consumption of GSH as an antioxidant under these conditions of oxidative stress. 39 Ethanol, AA, and AA Plus Iron Further Increase the Activity of the 3,802/GCLC 5 -Luc Reporter Construct in E47 Cells. Twenty-four hours after transfection with the different deletion constructs of the GCLC gene, C34 and E47 cells were treated with 100 mmol/l ethanol or 30 mol/l AA and incubated for an additional 24 hours to identify the elements in the GCLC promoter responsive to ethanol, AA, and AA plus iron. For the combined AA plus iron treatment, cells were incubated with a low nontoxic dose of AA (10 mol/l) for 23 hours and with 2.5 mol/l ferric nitrilotriacetate for 1 hour. Treatment with ethanol caused a small 25% increase in luciferase activity in E47 cells transfected with the 3,802/GCLC 5 -Luc construct (P.05) (comparable to the 40% increase in GCLC mrna), whereas treatments with AA and AA plus iron induced luciferase activity about 3-fold (P.001) and 4-fold (P.001), respectively, in E47 cells transfected with the 3,802/ GCLC 5 -Luc construct (Fig. 3A-C). Transfection with the 3,802m4A/GCLC 5 -Luc transgene harboring a mutation in the putative ARE4/EpRE containing an internal AP-1 binding site blunted the increases in luciferase activity in E47 cells under the 3 treatments. No significant changes by treatment with ethanol, AA, or AA plus iron were observed with the other reporter constructs (Fig. 3). These data suggest that the 3,802 to 2,752 region of the GCLC gene is critical for ethanol, AA, and AA plus iron responsiveness. C34 cells showed a similar but considerably weaker response to these additions (Fig. 3A-C). Effect of Antioxidants on Constitutive and Induced GCLC and GCLM mrna Levels. We evaluated if pretreatment with antioxidants such as catalase (2,000 U/mL) or an adenovirus containing the catalase complementary DNA, or vitamin E (50 mol/l), and diallyl

HEPATOLOGY, Vol. 37, No. 1, 2003 NIETO, MARÍ, AND CEDERBAUM 101 Fig. 4. Effect of catalase, vitamin E, or diallyl sulfide treatment on steady-state GCLC and GCLM mrna levels and on the induction by ethanol, AA, or ethanol plus AA. Cells were incubated in the presence or absence of 2,000 U/mL catalase, 50 mol/l vitamin E, or 5 mmol/l diallyl sulfide for 6 hours, after which followed treatment with 100 mmol/l ethanol, 30 mol/l AA, or the combination of ethanol plus AA. For the cells treated with ethanol alone, fresh medium with ethanol was replaced on a daily basis up to 5 days, whereas the cells treated with either AA or AA plus ethanol were collected at 48 hours. Total RNA was isolated using the TRIzol reagent and analyzed by Northern blot for GCLC, GCLM, and glyceraldehyde- 3-phosphate dehydrogenase. Arbitrary densitometric units are indicated under the blots. sulfide, a CYP2E1 inhibitor, could prevent not only the basal increase in GCLC and GCLM mrna levels in E47 cells but also the induction observed by ethanol, AA, and the combination of both. Preincubation with catalase, vitamin E, and diallyl sulfide suppressed the increase in both GCL mrnas produced by overexpression of CYP2E1 in the absence of any treatment (Fig. 4, E47 control blots). Catalase and vitamin E also decreased basal expression of GCL mrnas in C34 cells (Fig. 4, C34 control blots). Moreover, catalase, vitamin E, and diallyl sulfide blunted the increase in both GCLC and GCLM mrnas produced by AA and the combination of ethanol plus AA in E47 cells (Fig. 4). These data suggest that CYP2E1 metabolism in the presence or absence of an added pro-oxidant substrate (i.e., ethanol or AA) with the subsequent generation of ROS is responsible at least in part for the up-regulation of the GCL mrnas, GCL proteins (Fig. 2B), and GSH synthesis. This up-regulation may help the E47 cells to adapt to a state of oxidative stress as a result of the expression of CYP2E1. Catalase and Diallyl Sulfide Suppress Luciferase Activity of the 3,802/GCLC 5 -Luc Reporter Transgene. To extend the mrna results to the reporter constructs, 5 10 3 cells were seeded; cells were either infected with an adenovirus containing the complementary DNA for catalase or with control adenovirus (LacZ) or treated with 5 mmol/l diallyl sulfide for 6 hours. In untreated E47 cells, infection with the adenovirus containing catalase lowered the 3- to 4-fold increase in luciferase activity of the 3,802/GCLC 5 -Luc transgene (***P.001), whereas minimal changes were observed with any of the other constructs in E47 cells or with any construct in the C34 cells (Fig. 5A). Diallyl sulfide also prevented the increase in luciferase activity in the E47 cells transfected with the 3,802/GCLC 5 -Luc transgene, whereas no major changes were observed with the other constructs in E47 cells (Fig. 5B). To further validate the role of CYP2E1, E47 cells were pretreated with other CYP2E1 inhibitors such as 4-methylpyrazole or phenylisothiocyanate before transfection with the 3,802/ GCLC 5 -Luc construct. Luciferase activity was repressed by these CYP2E1 inhibitors compared with nontreated E47 cells (labeled as E47 control; Fig. 6). Moreover, E47 cells transfected with CYP2E1 24 hours before transfection (with the reporter constructs) to further elevate CYP2E1 levels showed about a 2-fold further increase in luciferase activity with the 3,802/GCLC 5 -Luc reporter construct compared with empty plasmid transfected E47 cells, and this increase was blocked about 80% by transfection with the 3,802m4A/GCLC 5 -Luc. In contrast, transfection studies with an antisense CYP2E1, which decreased CYP2E1 levels, blunted luciferase activity in E47 cells cotransfected with the 3,802/GCLC 5 -Luc reporter transgene compared with transfection with the empty vector (Fig. 6). Thus, there seemed to be some association between CYP2E1 levels in E47 cells and expression of the 3,802/GCLC 5 -Luc reporter construct. The CYP2E1 inhibitors and antisense CYP2E1 had no effect on the 3,802/GCLC 5 -Luc activity in C34 cells as anticipated in view of the absence of CYP2E1. Importantly, transfection with CYP2E1 strikingly elevated the 3,802/GCLC 5 -Luc activity in the C34 cells, validating that CYP2E1 and not cell or clonal selection is responsible for elevated activities found in the E47 and E43 cells. Control experiments indicated that the CYP2E1 inhibitors or complementary DNA constructs had no effect on the expression of the 814, 1,286, and 2,752 constructs (Fig. 6, inset). Preincubation With Low Concentrations of AA Increases GSH Levels and Protects Against Toxicity by Pro-oxidants. Because AA elevated GCL mrna levels and expression of the 3,802/GCLC 5 -Luc reporter construct, the functional significance of this elevation was determined. Preincubation of C34 and E47 cells with a

102 NIETO, MARÍ, AND CEDERBAUM HEPATOLOGY, January 2003 Fig. 5. (A) Activities of the GCLC reporter transgenes following infection with an adenovirus expressing catalase. Six hours after infection with a catalase-expressing adenovirus, C34 and E47 cells were transfected with the different GCLC promoter/luciferase constructs for 48 hours before cell harvest and lysis. Luciferase activity for each group was corrected by protein content and by transfection efficiency and normalized to the corrected luciferase activity of the pgl3-basic vector per cell line. Similar results were obtained in cells preincubated for 1 hour with 2,000 U/mL catalase (data not shown). Results are expressed as average numbers of 4 determinations SE. ***P.001 for catalase versus no catalase treatment. (B) Diallyl sulfide, a CYP2E1 inhibitor, suppresses luciferase activity of the 3,802/GCLC 5 -Luc reporter transgene. After overnight incubation with 5 mmol/l diallyl sulfide, C34 and E47 cells were transfected with the GCLC promoter/luciferase constructs and the luciferase activity was measured after 48 hours. Luciferase activity for each group was corrected by protein content and by transfection efficiency and normalized to the corrected luciferase activity of the pgl3-basic vector in each cell line. Results are expressed as average numbers of 4 determinations SE. ***P.001 for diallyl sulfide versus no diallyl sulfide treatment. low, nontoxic dose of AA (10 mol/l) for 24 hours increased GSH levels almost 2-fold in E47 cells (P.001) and 35% in C34 cells (Fig. 7B). We considered whether this pretreatment served as a protective tool against a more powerful pro-oxidant stress such as addition of H 2 O 2 and menadione. 41 After overnight incubation with 10 mol/l AA, cells were treated with either 60 mol/l H 2 O 2 or 15 mol/l menadione for 24 hours. C34 cells Fig. 6. The increase in luciferase activity in transfection studies with the full-length promoter of the GCLC is dependent on CYP2E1. C34 and E47 cells were either transfected with the empty vector pci-neo (labeled as control), pci-cyp2e1, or pci-as-cyp2e1 or treated with 2 mmol/l 4-methylpyrazole or 10 mol/l phenylisothiocyanate (PICT) 2 hours before transfection with the different GCLC reporter constructs. Luciferase activities were measured 48 hours after transfection as described in the legend to Fig. 1. Results are expressed as average values of 4 determinations SE. *P.05, ***P.001 compared with the basal response for each construct in the E47 cells (labeled as control in the graph). The luciferase activity for the 3,802/ GCLC 5 -Luc and the 3,802mA4/ GCLC 5 -Luc are shown in the bar graph, whereas the luciferase activity for the rest of the constructs is shown in the table.

HEPATOLOGY, Vol. 37, No. 1, 2003 NIETO, MARÍ, AND CEDERBAUM 103 Fig. 7. Preincubation with low doses of AA increases GSH levels and protects against cytotoxicity by pro-oxidants. C34 and E47 cells were preincubated with 10 mol/l AA for 24 hours followed by treatment with either 60 mol/l H 2 O 2 or 15 mol/l menadione for 24 hours. Experiments were performed to analyze for (A) cell viability (3-[4,5- dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide), (B) GSH levels, and (C) GCLC and GCLM mrna expression. Results in (A) are expressed as percentage of the C34 or E47 nontreated control (taken as 100% viability) and in (B) as nanomole of GSH per milligram of protein; in both cases, they are average values of 4 determinations per time point SE. **P.01, ***P.001 compared with the nontreated control for each cell line. Results in C are expressed as arbitrary densitometric units of duplicate experiments and are indicated under the blots. preincubated with buffer overnight showed 25% and 70% loss of viability when treated with H 2 O 2 and menadione, respectively (Fig. 7A, bars 3 and 4 compared with C34 control). The E47 cells were much more resistant to the toxic actions of H 2 O 2 and menadione (Fig. 7A), which may reflect the fact that the E47 cells have higher basal catalase activity and GSH levels and therefore have greater detoxifying capacity. 41 Pretreatment overnight with AA was able to increase cell viability in both C34 and E47 cells treated with H 2 O 2 or menadione (Fig. 7A). GSH levels were higher when H 2 O 2 or menadione were incubated in the AA-pretreated C34 and E47 cells compared with buffer-pretreated cells (Fig. 7B, 35% higher GSH for C34 cells and 100% higher GSH for E47 cells), suggesting a protective role by the initial increase in GSH. AA increased GCLC mrna levels in both C34 and E47 cells, H 2 O 2 had no effect, and menadione increased both GCLC and GCLM mrnas 2-fold in both cell lines (Fig. 7C). Although the combination of AA plus H 2 O 2 did not change the levels of GCLM in both cell lines, GCLC remained up-regulated, probably explaining the elevation in GSH levels under these conditions (Fig. 7C). The fact that menadione alone did not elevate GSH levels despite induction of both subunits of GCL may reflect increased consumption of GSH due to menadione-induced generation of ROS or increased export. Hepatocytes From Pyrazole-Treated Rats Show Increased GSH Concentration and GCLC and GCLM mrnas. To verify the results obtained in C34 and E47 cells with primary hepatocytes, rats were injected with saline or pyrazole and primary hepatocytes were obtained. Some hepatocytes were incubated overnight in the presence of 5 mmol/l diallyl sulfide to inhibit CYP2E1. Pyrazole induces CYP2E1 protein expression about 3-fold in hepatocytes. 31 Northern blot analysis showed a 2-fold increase in GCLC and GCLM mrnas in hepatocytes from pyrazole-treated rats (Fig. 8A), and this was associated with a 2-fold increase in GSH levels (Fig. 8B, P.001). When primary hepatocytes from pyrazole-injected rats were incubated in the presence of diallyl sulfide, GCLC and GCLM mrnas decreased to basal levels whereas no changes were observed in saline-treated hepatocytes. Similar results were observed for GSH levels (30 nmol/mg of protein for pyrazole-treated hepatocytes vs. 12.5 nmol/mg of protein for pyrazole-treated hepatocytes incubated with diallyl sulfide; P.001). A Western blot is shown in Fig. 8C comparing the expression of CYP2E1 in saline-treated hepatocytes, pyrazole-treated hepatocytes, and E47 cells, as are rates of oxidation of p-nitrophenol, a reflection of CYP2E1 catalytic activity. Discussion The present study extends previous results 28 that GCL mrna levels are elevated in E47 cells to show that GCLC and GCLM mrna expression further increase when the cells are challenged with substrates for CYP2E1 or prooxidants, which further elevate oxidative stress, such as ethanol, AA, and AA plus iron. Moreover, the increase in GCL mrna levels is accompanied by equal or even

104 NIETO, MARÍ, AND CEDERBAUM HEPATOLOGY, January 2003 Fig. 8. Induction of GCLC and GCLM mrna expression and GSH levels in primary hepatocytes from pyrazole-treated rats. Rats were injected twice intraperitoneally with either saline solution or pyrazole (200 mg/kg body wt); 24 hours after the last injection, the livers were perfused with bacterial collagenase to isolate primary hepatocytes. Some hepatocytes were seeded in the presence of 5 mmol/l diallyl sulfide (DAS) to inhibit CYP2E1 activity. Twenty-four hours after plating, total RNA was isolated for Northern blot analysis for (A) GCLC and GCLM mrna expression and samples from different plates were also collected to determine (B) the concentration of GSH. Results are expressed as arbitrary densitometric units and are indicated under the Northern blot or as nanomole per milligram of protein for GSH (***P.001, hepatocytes from pyrazoleinjected rats compared with hepatocytes from saline-injected rats). (C) Expression of CYP2E1 protein in the E47 cells compared with that of the saline- and pyrazole-treated livers. Results are expressed as arbitrary densitometric units under the Western blot. The activity of CYP2E1 was measured by the p-nitrophenol oxidation assay (PNP) and is expressed as picomole per minute per milligram of microsomal protein. greater increases in GCL protein levels. Priming of the liver for ethanol-induced injury by nutrients such as polyunsaturated fat and iron plays a key role in alcoholic liver disease. 29,37,42,43 The changes observed were most pronounced in E47 cells treated with AA and AA plus iron in both the heavy and light subunits of GCL mrnas and the corresponding proteins. Ethanol and AA increased GSH levels in E47 cells, consistent with the induction of the GCL mrnas and proteins. However, the coincubation with AA and iron decreased GSH in both cell lines below the basal levels; this effect may reflect GSH consumption under conditions of a higher pro-oxidant stress. Transient transfection experiments with chimeric constructs driven by different sequences of the GCLC gene indicate that the 3,802 to 2,752 region is essential for CYP2E1-mediated high basal luciferase activity and for ethanol, AA, and AA plus iron responsiveness in E47 cells. Interestingly, this region contains 2 redox-sensitive elements, ARE3 and ARE4, of which ARE4 contains an embedded AP-1 binding site, a transcription factor known to respond to oxidative stress. 21 In the CYP2E1 system, the ARE4 element seemed to be critical because transfection of the 3,802mA4/GCLC 5 -Luc transgene containing a single base mutation (T 3 G) introduced into the ARE4 binding site blunted both the constitutive and inducible luciferase activity. This induction seemed to be CYP2E1 selective because transfection of the fulllength reporter construct into a HepG2 cell line expressing a different P450 isoform, CYP3A4, did not show any increase in luciferase activity. The increase in ARE4 responsiveness was also observed with a different clone of CYP2E1-expressing cells (E43 cells); these cells with a lower CYP2E1 content showed less responsiveness than E47 cells. The increase in GCLC and GCLM mrnas in E47 cells compared with C34 cells is likely due to CYP2E1- dependent generation of ROS. The further increase produced by ethanol, AA, or AA plus iron is also likely due to an increase in oxidative stress. Treatment with catalase or vitamin E prevented the increase in both mrnas in nontreated E47 cells and in E47 cells treated with ethanol, AA, or AA plus iron. The fact that the increase in GCL mrnas in nontreated or treated E47 cells is indeed due to CYP2E1 is indicated by the prevention of such increases by the CYP2E1 inhibitor diallyl sulfide. To determine whether the basal induction of GCLC transcription was indeed mediated by ROS and dependent on CYP2E1 expression, we studied the responsiveness to transfection with the different chimeric constructs in both cell lines in the presence or absence of either catalase or diallyl sulfide. In E47 cells infected with adenovirus containing catalase, the induction of luciferase

HEPATOLOGY, Vol. 37, No. 1, 2003 NIETO, MARÍ, AND CEDERBAUM 105 activity by transfection with the 3,802/GCLC 5 -Luc recombinant expression vector was blunted when compared with cells infected with the adenovirus containing LacZ. Diallyl sulfide also prevented this increase in luciferase activity, as did several other CYP2E1 inhibitors and antisense CYP2E1. These results suggest that CYP2E1- mediated ROS production acts on regulatory sequences mediating basal and induced luciferase activity in transfection experiments with the full-length construct and that these sequences seem to be localized between nucleotides 3,802 and 2,752 of the GCLC gene. The blunting effect of the 3,802m4A/GCLC construct indicates the critical role of the ARE4 in the GCLC promoter. Besides the GCL genes, AREs have been identified in the 5 -flanking region of genes encoding phase II enzymes such as GSH S-transferase, reduced nicotinamide adenine dinucleotide phosphate quinone oxidoreductase 1, and other antioxidant stress proteins such as heme oxygenase 1 and L-ferritin. 19,44,45 These stressassociated proteins are widely recognized to provide protection against the toxicities associated with organic electrophiles and ROS. 46 Multiple transcriptional regulators, including a primary Nrf2-Maf heterodimer, a p160 family coactivator (ARE binding protein 1), and a tertiary coactivator (CBP/p300), nucleate to form a complex at the ARE that regulates transcription of the GCLC gene. 47 All 3 of these transcriptional regulators have been identified in nuclear extracts from HepG2 cells. 47 The role, if any, of these regulators in the induction of the GCLC gene in nontreated and pro-oxidant treated E47 cells remains to be determined. Overexpression of CYP2E1 in HepG2 cells caused increased basal levels of GSH as well as GSH transferase and catalase, 28,41 rendering the cells less susceptible to oxidative damage. When C34 or E47 cells were pretreated with a nontoxic dose of AA (10 mol/l) before addition of H 2 O 2 or menadione, cell viability was restored in conjunction with increased levels of GSH. These results suggest that AA via induction of GCL followed by an increase in GSH levels could protect the cells against cytotoxicity exerted by more powerful pro-oxidants. Indeed, in E47 cells, AA itself increased GSH levels by 2-fold and the combined treatment with H 2 O 2 or menadione maintained GSH levels without any cellular toxicity. C34 cells were more prone to oxidant insult, due at least in part to a lower basal concentration of GSH; the combined treatment of AA plus either H 2 O 2 or menadione elevated GSH levels enough to protect the cells when compared with C34 cells treated with H 2 O 2 or menadione alone. This cytoprotective mechanism, especially of AA and menadione, seemed to operate via increasing GCLC and GCLM mrna levels. In the protected E47 cells (and pyrazole hepatocytes), the toxicity eventually induced by ethanol, AA, or iron must overcome the initial protection afforded by the elevated levels of GSH and antioxidant enzymes. Indeed, depletion of GSH markedly elevates the toxicity of these agents and even results in toxicity in the absence of ethanol, AA, or iron, 30,39,41 validating that up-regulation of GSH and antioxidant enzymes is a metabolic adaptation in response to CYP2E1-derived oxidative stress. As an approximation to in vivo models, and to validate the experiments in C34 and E47 cells with primary hepatocytes, rats were either injected with saline (control) or pyrazole (to elevate CYP2E1 content) and primary hepatocytes were isolated. There was a 2-fold increase in the mrnas of both subunits of the GCL and in GSH levels in the pyrazole-treated hepatocytes, which was prevented by diallyl sulfide. This up-regulation of GSH synthesis may, by analogy to experiments with the E47 cells, reflect an adaptation to CYP2E1-dependent oxidative stress. Indeed, inhibition of GSH synthesis by treatment with L-buthionine sulfoximine dramatically increases the sensitivity of pyrazole-treated hepatocytes and E47 cells to oxidative stress. Importantly, long-term ethanol feeding has been shown to increase GCLC mrna levels. 48 Whether this is due to the elevated content of CYP2E1 after long-term ethanol feeding remains to be determined. In summary, overexpression of CYP2E1 in either HepG2 cells or primary hepatocytes induces transcriptional activation of the GCL genes, which increase GCL proteins and GSH levels to protect the cell against CYP2E1-mediated oxidative stress. This transcriptional up-regulation seems to operate through a redox-sensitive element (ARE4) localized 3.1 kilobases upstream of the transcription start site in the GCLC gene. The transcriptional up-regulation is suppressed by antioxidants and CYP2E1 inhibitors. Acknowledgment: The authors thank Dr. Defeng Wu (Mount Sinai School of Medicine) for providing the primary hepatocytes, Drs. Timothy Mulcahy and Jerry J. Gipp (University of Wisconsin Medical School) for the h -GCLC constructs, and Dr. Terrance J. Kavanagh (University of Washington) for the anti-gclc and anti- GCLM antibodies. References 1. DeLeve L, Kaplowitz N. Glutathione metabolism and its role in hepatotoxicity. Pharmacol Ther 1991;52:287-305. 2. Hutter DE, Till BG, Greene JJ. Redox state changes in density-dependent regulation of proliferation. Exp Cell Res 1997;232:435-438. 3. Lu SC. Up-regulation of hepatic glutathione synthesis. Semin Liver Dis 1998;18:331-343. 4. Lu SC. Regulation of hepatic glutathione synthesis: current concepts and controversies. FASEB J 1999;13:1169-1183.

106 NIETO, MARÍ, AND CEDERBAUM HEPATOLOGY, January 2003 5. Meister A. Glutathione. In: Arias IM, Jakoby WB, Popper H, Schachter D, Shafritz DA, eds. The Liver: Biology and Pathobiology. 2nd ed. New York: Raven, 1998:401-417. 6. Suthanthiran M, Anderson ME, Sharma VK, Meister A. Glutathione regulates activation-dependent DNA synthesis in highly purified normal human T lymphocytes stimulated via the CD2 and CD3 antigens. Proc Natl Acad Sci USA1990;87:3343-3347. 7. Fernandez-Checa J, Lu SC, Ookhtens M, DeLeve L, Runnegar M, Yoshida H, Saiki H, et al. The regulation of hepatic glutathione. In: Tavoloni N, Berk PD, eds. Hepatic Anion Transport and Bile Secretion: Physiology and Pathophysiology. New York: Marcel Dekker, 1992:363-395. 8. Ookhtens M, Kaplowitz N. Role of the liver in interorgan homeostasis of glutathione and cysteine. Semin Liver Dis 1998;18:313-329. 9. Gipp JJ, Bailey HH, Mulcahy RT. Cloning and sequence of the cdna for the light subunit of human liver -glutamylcysteine synthetase and relative mrna levels for heavy and light subunits in human normal tissues. Biochem Biophys Res Commun 1995;206:584-589. 10. Gipp JJ, Chang C, Mulcahy RT. Cloning and nucleotide sequence of a full-length cdna for human liver -glutamylcysteine synthetase. Biochem Biophys Res Commun 1992;185:29-35. 11. Huang C, Anderson ME, Meister A. Amino acid sequence and function of the light subunit of rat kidney -glutamylcysteine synthetase. J Biol Chem 1993;268:20578-20583. 12. Yan N, Meister A. Amino acid sequence of rat kidney -glutamylcysteine synthetase. J Biol Chem 1990;265:1588-1593. 13. Seelig GF, Simondsen RP, Meister A. Reversible dissociation of -glutamylcysteine synthetase into two subunits. J Biol Chem 1984;259:9345-9347. 14. Huang CS, Moore WR, Meister A. On the active thiol site of -glutamylcysteine synthetase: relationship to catalysis, inhibition, and regulation. Proc Natl Acad Sci USA1988;85:2464-2468. 15. Huang C, Chang L, Anderson ME, Meister A. Catalytic and regulatory properties of the heavy subunit of rat kidney -glutamylcysteine synthetase. J Biol Chem 1993;268:19675-19680. 16. Misra I, Griffith OW. Expression and purification of human -glutamylcysteine synthetase. Protein Exp Purif 1998;13:268-276. 17. Grant CM, MacIver FH, Dawes IW. Glutathione synthetase is dispensable for growth under both normal and oxidative stress conditions in the yeast Saccharomyces cerevisiae due to an accumulation of the dipeptide -glutamylcysteine. Mol Biol Cell 1997;8:1699-1707. 18. Huang CS, He W, Meister A, Anderson ME. Amino acid sequence of rat kidney glutathione synthetase. Proc Natl Acad Sci U S A 1995;92:1232-1236. 19. Friling RS, Bensimon A, Tichauer Y, Daniel V. Xenobiotic-inducible expression of murine glutathione S-transferase Ya subunit gene is controlled by an electrophile responsive element. Proc Natl Acad Sci U S A 1991;87: 6258-6262. 20. Mulcahy RT, Gipp JJ. Identification of a putative antioxidant response element in the 5 -flanking region of the human gamma-glutamylcysteine synthetase heavy subunit gene. Biochem Biophys Res Commun 1995;209: 227-233. 21. Mulcahy RT, Wartman MA, Bailey HH, Gipp JJ. Constitutive and betanaphthoflavone-induced expression of the human gamma-glutamylcysteine synthetase heavy subunit gene is regulated by a distal antioxidant response element/tre sequence. J Biol Chem 1997;272:7445-7454. 22. Rushmore TH, Morton MR, Pickett CB. The antioxidant responsive element. Activation by oxidative stress and identification of the DNA consensus sequence required for functional activity. J Biol Chem 1991;266: 11632-11639. 23. Yao KS, Godwin AK, Johnson WW, Ozols RF, O Dwyer PJ, Hamilton TC. Evidence for altered regulation of gamma-glutamylcysteine synthetase gene expression among cisplatin-sensitive and cisplatin-resistant human ovarian cancer cell lines. Cancer Res 1995;55:4367-4374. 24. Koop DR. Oxidative and reductive metabolism by cytochrome P450 2E1. FASEB J 1992;6:724-730. 25. Ekstrom G, Ingelman-Sundberg M. Rat liver microsomal NADPH-supported oxidase activity and lipid peroxidation dependent on ethanol-inducible cytochrome P-450 (P-450IIE1). Biochem Pharmacol 1989;38: 1313-1319. 26. Gorsky LD, Koop DR, Coon MJ. On the stoichiometry of the oxidase and monooxygenase reactions catalyzed by liver microsomal cytochrome P-450. Products of oxygen reduction. J Biol Chem 1984;259:6812-6817. 27. Vaz AD, Roberts ES, Coon MJ. Reductive beta-scission of the hydroperoxides of fatty acids and xenobiotics: role of alcohol-inducible cytochrome P-450. Proc Natl Acad Sci USA1990;87:5499-5503. 28. Mari M, Cederbaum AI. CYP2E1 overexpression in HepG2 cells induces glutathione synthesis by transcriptional activation of -glutamylcysteine synthetase. J Biol Chem 2000;275:15563-15571. 29. Tsukamoto H, Lu SC. Current concepts in the pathogenesis of alcoholic liver disease. FASEB J 2001;15:1335-1349. 30. Chen Q, Cederbaum AI. Cytotoxicity and apoptosis produced by cytochrome P450 2E1 in HepG2 cells. Mol Pharmacol 1998;53:638-648. 31. Wu D, Cederbaum AI. Induction of liver cytochrome P4502E1 by pyrazole and 4-methylpyrazole in neonatal rats. J Pharmacol Exp Ther 1993; 264:1468-1473. 32. Nieto N, Friedman SL, Greenwel P, Cederbaum AI. CYP2E1-mediated oxidative stress induces collagen type I expression in rat hepatic stellate cells. HEPATOLOGY 1999;30:987-996. 33. Franklin CC, Krejsa CM, Pierce RH, White CC, Fausto N, Kavanagh TJ. Caspase-3- dependent cleavage of the glutamate-l-cysteine ligase catalytic subunit during apoptotic cell death. Am J Pathol 2002;160:1887-1894. 34. Thompson SA, White CC, Krejsa CM, Diaz D, Woods JS, Eaton DL, Kavanagh TJ. Induction of glutamate-cysteine ligase ( -glutamylcysteine synthetase) in the brains of adult female mice subchronically exposed to methylmercury. Toxicol Lett 1999;110:1-9. 35. Bai J, Cederbaum AI. Adenovirus-mediated expression of catalase in the cytosolic or mitochondrial compartment protects against cytochrome P450 2E1-dependent toxicity in HepG2 cells. J Biol Chem 2001;274: 26217-26224. 36. Tietze F. Enzymatic method for quantitative determination of nanogram amounts of total and oxidized glutathione: applications to mammalian blood and other tissues. Anal Biochem 1969;27:502-522. 37. Tsukamoto H, Matsuoka M, French S. Experimental models of hepatic fibrosis: a review. Semin Liver Dis 1990;10:56-65. 38. Niemela O. Distribution of ethanol-induced protein adducts in vivo: relationship to tissue injury. Free Radic Biol Med 2001;31:1533-1538. 39. Caro A, Cederbaum AI. Synergistic toxicity of iron and arachidonic acid in HepG2 cells overexpressing CYP2E1. Mol Pharmacol 2001;60:742-752. 40. Sakurai K, Cederbaum AI. Oxidative stress and cytotoxicity induced by ferric nitrilotriacetate in HepG2 cells that express cytochrome P450 2E1. Mol Pharmacol 1998;54:1024-1035. 41. Mari M, Cederbaum AI. Induction of catalase, alpha, and microsomal glutathione S-transferase in CYP2E1 overexpressing HepG2 cells and protection against short-term oxidative stress. HEPATOLOGY 2001;33:652-661. 42. De La M Hall P, Lieber CS, DeCarli LM, French SW, Lindros KO, Jarvelainen H, Bode C, et al. Models of alcoholic liver disease in rodents: a critical evaluation. Alcohol Clin Exp Res 2001;25:254S-261S. 43. French SW, Morimoto M, Reitz RC, Koop D, Klopfenstein B, Estes K, Clot P, et al. Lipid peroxidation, CYP2E1 and arachidonic acid metabolism in alcoholic liver disease in rats. J Nutr 1997;127:907S-911S. 44. Favreau LV, Pickett CB. Transcriptional regulation of the rat NAD(P)H: quinone reductase gene. Identification of regulatory elements controlling basal level expression and inducible expression by planar aromatic compounds and phenolic antioxidants. J Biol Chem 1991;266: 4556-4561. 45. Tsuji Y, Ayaki H, Whitman SP, Morrow CS, Torti SV, Torti FM. Coordinate transcriptional and translational regulation of ferritin in response to oxidative stress. Mol Cell Biol 2000;20:5818-5827. 46. Kensler TW. Chemoprevention by inducers of carcinogen detoxification enzymes. Environ Health Perspect 1997;105:965-970. 47. Zhu M, Fahl WE. Functional characterization of transcription regulators that interact with the electrophile response element. Biochem Biophys Res Commun 2001;289:212-219. 48. Lu SC, Huang ZZ, Yang JM, Tsukamoto H. Effect of ethanol and high fat feeding on hepatic -GCS subunit expression in the rat. HEPATOLOGY 1999;30:209-214.