Oncogene (1998) 17, 3145 ± 3156 ã 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Nrf2 and Nrf1 in association with Jun proteins regulate antioxidant -mediated expression and coordinated induction of genes encoding detoxifying enzymes

Radjendirane Venugopal and Anil K Jaiswal

Department of Pharmacology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030-3498, USA

Antioxidant response element (ARE)-mediated expres- sequences ¯anking the core sequence have been shown sion and coordinated induction of genes encoding to contribute to the ARE-mediated expression and detoxifying enzymes is one mechanism of critical induction (Prestera et al., 1993; Bergelson et al., 1994; importance to cellular protection against oxidative stress. Xie et al., 1995; Wasserman and Fahl, 1997). In the present report, we demonstrate that nuclear Nuclear factors c-Jun, Jun-B, Jun-D, c- transcription factors Nrf2 and Nrf1 associate with Jun Fos, Fra1, Nrf1, Nrf2, YABP, ARE-BP1 and Ah (c-Jun, Jun-B and Jun-D) proteins to upregulate ARE- have been reported to bind to the AREs from mediated expression and coordinated induction of the various genes (Li and Jaiswal, 1992; Venugopal and detoxifying enzymes in response to antioxidants and Jaiswal, 1996; Friling et al., 1992; Liu and Pickett, xenobiotics. Nrf-Jun association/heterodimerization and 1996; Vasliou et al., 1995; Wasserman and Fahl, 1997). binding to the ARE required unknown cytosolic factor(s). Among these transcription factors, c-Jun, Jun-B, Jun- Nrf2 containing one mutated leucine in its D, c-Fos, Fra1, Nrf1 and Nrf2 bind to the human region was more ecient in upregulation of ARE- NQO1 gene hARE (Li and Jaiswal, 1992; Venugopal mediated , as compared to Nrf1 with and Jaiswal, 1996). Nrf1 and Nrf2 positively and c-Fos two mutated leucines. and Fra 1 negatively regulate the hARE-mediated expression and induction of NQO1 gene in response to Keywords: NF-E2 related factors; Jun; antioxidant xenobiotics and antioxidants (Venugopal and Jaiswal, response element; regulation of expression; detoxifying 1996). Nrf1 and Nrf2 are b-zip (leucine zipper) proteins enzyme genes that do not heterodimerize with each other and require another leucine zipper protein for its activity (Chan et al., 1993; Moi et al., 1994; Venugopal and Jaiswal, 1996). Recently, Nrf27/7 mice lacking the expression Introduction of Nrf2 were generated that showed signi®cantly reduced expression and induction of NQO1 gene Antioxidant response element (ARE) also referred to indicating a physiological (in vivo) role of transcription as electrophile response element (EpRE) was identi®ed factor Nrf2 in the regulation of expression of NQO1 in the promoters of several genes encoding detoxifying gene (Itoh et al., 1997). enzymes (Rushmore and Pickett, 1993; Jaiswal, 1994; In the present report, we demonstrate that nuclear Mulkahy et al., 1997). ARE is known to mediate transcription factors Nrf2 and Nrf1 associate with Jun expression and coordinated induction of the various (c-Jun, Jun-B and Jun-D) proteins to regulate ARE- detoxifying enzyme genes. The detoxifying enzymes mediated expression and coordinated induction of include NAD(P)H:quinone oxidoreductases (NQOs), NAD(P)H:Quinone Oxidoreductase1 (NQO1) and which catalyze two-electron reduction and detoxifica- Glutathione S-transferase Ya (GST Ya) subunit tion of quinones; glutathione S-transferase (GSTs), genes. We also demonstrate that heterodimerization which conjugate hydrophobic electrophiles and reactive and binding of Nrf-Jun proteins require unknown oxygen species (ROS) with glutathione; UDP-glucur- cytosolic factor(s). We further demonstrate that Nrf2 onosyl transferase (UDP-GT), which catalyze the with one mutated leucine in its leucine zipper region conjugation of glucuronic acid with xenobiotics and performs better than Nrf1 containing two mutated drugs; epoxide hydrolase, which inactivates epoxides; leucines. In addition, we show that cellular levels of g-glutamylcysteine synthetase, which is a rate-limiting Jun proteins play an important role in Nrf1 and Nrf2 enzyme in the de novo synthesis of glutathione and so regulated ARE-mediated expression and induction. on (Venugopal et al., 1997). Nucleotide sequence analysis of the various AREs revealed that they contain AP1/AP1 like elements arranged as inverse or Results direct repeats separated by three or eight nucleotides followed by a GC box (Jaiswal, 1994). Mutational The nucleotide sequences of human NQO1 gene analysis of the ARE identi®ed GTGAC***GC as the hARE, mutant NQO1 gene hARE, rat GST Ya gene core of the ARE sequence (Rushmore et al., 1991; Xie ARE, mutant GST Ya gene ARE and human et al., 1995). Additional cis-element and nucleotide collagenase gene TRE are shown in Figure 1. Both NQO1 and GST Ya gene AREs contain two AP1/AP1 like elements followed by `GC' box. The ARE core Correspondence: AK Jaiswal sequence `GTGAC***GC' is highly conserved between Received 8 April 1998; revised 22 June 1998; accepted 22 June 1998 NQO1 and GST Ya genes. The ARE is known to Nrf association with Jun R Venugopal and AK Jaiswal 3146 expression of c-Jun, presumably through c-Jun+c-Fos pathway. c-Jun is known to form stable heterodimer with c-Fos that binds to the TRE (AP1) and increase the expression of target genes (Angel and Karin, 1991). Mutant hARE did contain the 5' AP1 like element in 3'?5' orientation making it look like a TRE element. The results with TRE-tk-CAT indicated that Nrf+Jun proteins that mediate hARE-regulated gene expression, are signi®cantly less e€ective in mediating TRE- regulated CAT gene expression in transfected cells. Figure 1 Nucleotide sequence of NQO1 and GST Ya genes ARE These results were expected because TRE-regulated and collagenase gene TRE. Nucleotide sequences of human gene expression is known to be mediated by Fos+Jun NQO1 gene ARE (hARE), mutant hARE, rat GST Ya gene ARE, mutant GST Ya gene ARE and human collagenase gene and not by Nrf+Jun proteins (Angel and Karin, 1991). TRE are shown. Human NQO1 gene hARE and GST Ya gene In similar experiments as described above, over- ARE contains AP1 (TRE) and AP1 like (TRE like) sequences expression of Jun-B and Jun-D along with Nrf1 and which are boxed. The 3' AP1/AP1 like element was mutated in Nrf2 also upregulated hARE-mediated CAT gene NQO1 and GST Ya gene ARE to construct mutant NQO1 gene hARE and mutant GST Ya gene ARE. X represent mutated expression in transfected Hep-G2 cells (Figure 3). sequences However, overexpression of c-Fos repressed the Nrf1 and Nrf2-mediated upregulation of hARE-mediated CAT gene expression in transfected Hep-G2 cells (Figure 4; compare lanes `Nrf1' with `Nrf1+c-Fos' mediate high basal expression of NQO1 and GST Ya and `Nrf2' with `Nrf2+c-Fos'). In related experiments, genes and their induction in response to xenobiotics overexpression of c-Fos alone also repressed the (b-NF) and antioxidants (t-BHQ) (Rushmore et al., hARE-mediated basal expression of CAT gene 1991; Xie et al., 1995). Mutations in the ARE core (Figure 4; compare c-Fos with LNCX). The repressive sequences are known to abolish basal expression and e€ect of c-Fos was highly reproducible. Replacement of induction in response to xenobiotics and antioxidants hARE-tk-CAT with phNQO1g1.55CAT (1.55 kb of (Rushmore et al., 1991; Xie et al., 1995). human NQO1 gene driving CAT gene) Overexpression of Nrf1 and c- showed identical results as described above for Jun individually in transfected Hep-G2 cells led to less hARE-tk-CAT plasmid (data not shown). than twofold increase in hARE-mediated CAT gene The treatment of Hep-G2 cells overexpressing expression (Figure 2 upper left panel). However, Nrf1+c-Jun and Nrf2+c-Jun with di€erent concentra- overexpression of Nrf1 and c-Jun together caused tions of b-NF showed signi®cant increases in hARE- more than sevenfold increase in the hARE-mediated mediated CAT gene expression (Figure 5). The higher CAT gene expression. Similar results were observed by doses of b-NF (25 and 50 mM) was more or less similar co-transfection and overexpression of Nrf2 and c-Jun or less e€ective in induction of hARE-mediated CAT in Hep-G2 cells (Figure 2 upper left panel). Interest- gene expression as compared to 10 mM b-NF. This ingly, Nrf2 was more ecient than Nrf1 in the presumably may be due to the saturation of system upregulation of hARE-mediated CAT gene expression because of overexpression of Nrf+c-Jun. Similar in transfected Hep-G2 cells. Under similar conditions, results were also obtained by treatment of transfected mutant hARE-mediated CAT gene expression was Hep-G2 cells with tertbutyl hydroquinone (t-BHQ) signi®cantly lower as compared to the hARE-mediated (data not shown). CAT gene expression (Figure 2 upper right panel). The increasing amount of Nrf1 and Nrf2 with a Overexpression of Nrf1, Nrf2 and combinations of constant amount of c-Jun resulted in concentration Nrf1 and Nrf2 with c-Jun had little or no e€ect on dependent increased expression of hARE-mediated mutant hARE-mediated CAT gene expression (Figure CAT gene (Figure 6; top and middle left panels). 2 upper right panel). In related experiments, the c-Jun Increasing the concentration of c-Jun by transfection of expression plasmid (LNCX-c-Jun) was replaced with v- 2.5 mg of LNCX-c-Jun plasmid with a constant Jun and v-JunL3P expression plasmids (LNCX-v-Jun; concentration of Nrf1 and Nrf2 also resulted increased LNCX-v-JunL3P) (Figure 2 lower left panel). v- expression of hARE-mediated CAT gene (Figure 6; top JunL3P is a derivative of v-Jun containing third and middle right panels). However, higher concentra- leucine mutated to proline in its leucine zipper region tions of c-Jun plasmid (5 and 10 mg in case of Nrf1 and (Kataoka et al., 1995). This mutation is known to 10 mg in case of Nrf2) repressed the hARE-mediated severely weaken the heterodimerization of v-Jun with CAT gene expression as compared to 2.5 mg of c-Jun other leucine zipper factors (Kataoka et al., 1995). The plasmid. In other words, higher increases in c-Jun replacement of c-Jun with v-Jun showed similar results concentration is less e€ective than lower increases in c- as with c-Jun. v-Jun combined with Nrf1 and Nrf2 led Jun concentration in the upregulation of hARE- to signi®cant increases in the hARE-mediated CAT mediated CAT gene expression. Interestingly, increas- gene expression in transfected Hep-G2 cells. However, ing the concentration of Nrf2 with a constant replacement of c-Jun or v-Jun with v-JunL3P resulted concentration of Nrf1 resulted in concentration in signi®cant decreases in Nrf+Jun -mediated hARE dependent increase in hARE-mediated CAT gene regulated CAT gene expression (Figure 2; lower left expression (Figure 6; lower panel left side). Under panel). The collagenase gene TRE showed similar similar conditions, increasing the concentration of Nrf1 results as mutant ARE (Figure 2 lower right panel). with a constant concentration of Nrf2 reduced the However, both mutant ARE- and collagenase gene upregulation of Nrf2 regulated hARE-mediated CAT TRE-mediated expression was increased due to over- gene expression (Figure 6; lower panel right side). Nrf association with Jun R Venugopal and AK Jaiswal 3147

Figure 2 E€ect of overexpression of Nrf1 and Nrf2 along with c-Jun, v-Jun and v-JunL3P on hARE-, mutant hARE- and TRE- mediated CAT gene in Hep-G2 cells. The Hep-G2 cells were co-transfected with ®ve micrograms of reporter plasmid hARE-tk- CAT, mutant hARE-tk-CAT or TRE-tk-CAT and 5 mg of expression plasmids LNCX (vector alone), LNCX-Nrf1, LNCX-Nrf2, LNCX-c-Jun, LNCX-v-Jun, LNCX-v-JunL3P individually and in the combinations as shown. LNCX-Nrf1R and LNCX-Nrf2R contained Nrf1 and Nrf2 in reverse orientation. These plasmids upon transfection in Hep-G2 cells did not increase the concentration of Nrf1 and Nrf2. On the other hand, LNCX-Nrf1C and LNCX-Nrf2C contained Nrf1 and Nrf2 cDNA in correct orientation. These plasmids upon transfection in Hep-G2 cells resulted in overexpression of Nrf1 and Nrf2 as described in Materials and methods. Five micrograms of RSV-b-galactosidase plasmid was included in each case as internal control of transfection eciency. The total amount of DNA for transfection in each case was normalized to 20 mg with LNCX (vector alone plasmid). Forty-eight hours after transfection, the cells were analysed for b-galactosidasae and CAT activities. The acetylated [14C]chloramphenicol in the upper spots on TLC plates were cut out, counted and used to calculate CAT activity. The various results are presented as mean+s.e. of ®ve independent transfection experiments

Similar to the hARE-mediated CAT gene expres- c-Jun suggests that 5' AP1 like sequence in GST ARE sion, overexpression of Nrf1 and Nrf2 together with is not as speci®c as the 5' AP1 in the hARE. Jun proteins upregulated the GST Ya ARE-mediated The alignment of DNA binding regions of Nrf1 and CAT gene expression in transfected Hep-G2 cells Nrf2 with Jun (c-Jun, Jun-B and Jun-D), Fos (c-Fos) (Figure 7; data shown only for c-Jun). Nrf2 was more and Fra (Fra1) revealed presence of highly conserved ecient than Nrf1 in upregulating the GST Ya ARE- redox labile cysteine residue in `Basic region' and mediated CAT gene expression. Replacement of GST several conserved leucines in the leucine zipper regions ARE with mutant GST ARE resulted in the loss of (Figure 8). Interestingly, two leucines (#4 and 5) in the expression and e€ect of overexpression of Nrf+Jun in leucine zipper region of Nrf1 and one leucine (#4) in transfected Hep-G2 cells. It may be noteworthy that the leucine zipper region of Nrf2 were found mutated response of GST mutant construct to c-Jun was as compared to Jun, Fos and Fra proteins. di€erent than mutant hARE (compare Figure 7 with Immunoprecipitation experiments with c-Jun anti- Figure 2). Mutant hARE and not mutant GST ARE bodies showed co-precipitation of 65 kDa Nrf1 and was responsive to overexpression of c-Jun. The virtual 66 kDa Nrf2 with 39 kDa c-Jun protein (Figure 9a). absence of a response of the GST mutant construct to The presence of Nrf1 and Nrf2 proteins in the complex Nrf association with Jun R Venugopal and AK Jaiswal 3148

Figure 3 E€ect of overexpression of Nrf1 and Nrf2 together with Jun-B and Jun-D on hARE- and mutant hARE-mediated CAT gene expression. Five micrograms of reporter plasmid hARE-tk-CAT or mutant hARE-tk-CAT were mixed with 5 mg of expression plasmids LNCX (vector alone), LNCX-Nrf1, LNCX-Nrf2, LNCX-Jun-B and LNCX-Jun-D individually and in di€erent combinations. LNCX (vector alone) plasmid was used to normalize the amount of expression plasmid to 10 mg in each transfection. Five micrograms of RSV-b-galactosidase plasmid was used in each case as control of transfection eciency. Forty-eight hours after the transfection, the cells were analysed for b-galactosidase and CAT activities. The various results are presented as mean+s.e. of three independent transfection experiments

immunoprecipitated with c-Jun antibody was con- extract resulted in cytosolic extract concentration ®rmed by SDS PAGE, Western blotting and probing dependent binding of Nrf2+c-Jun to the hARE with antibodies against Nrf1 and Nrf2 (Figure 9b). We (Figure 10b). In the same experiment cytosolic extract also detected small amounts of 110 kDa protein in by itself did not show any binding to the hARE both Nrf1 and Nrf2 cases that cross-reacted with (Figure 10b). The mobility of the hARE-Nrf2+c-Jun antibodies against Nrf1 and Nrf2. 110 kDa and complex in polyacrylamide gel was similar to that 66 kDa Nrf1 proteins are the product of the same observed with hARE-Hepa-1 nuclear protein complex gene (Chan et al., 1993). The peptide sequences of (Figure 10b). In related experiments c-Jun was replaced 110 kDa Nrf1 protein is same as 66 kDa Nrf1 protein with v-Jun and v-JunL3P (Figure 10c). v-JunL3P is a except that 110 kDa protein is longer at its N-terminus derivative of v-Jun containing third leucine mutated to region due to use of a distal 5' ATG (methionine). proline in its leucine zipper region (Kataoka et al., Similar observations have also been reported for Nrf2 1995). This mutation is known to signi®cantly reduce (Moi et al., 1994). It may be noteworthy that Nrf2 and the capacity of v-Jun to form heterodimer with other Nrf1 bands were signi®cantly higher intensity as leucine zipper proteins (Kataoka et al., 1995). Band compared to c-Jun band. This could be due to higher shift experiments with hARE and in vitro translated turnover/better labeling of Nrf1 and Nrf2 as compared Nrf2+v-Jun showed similar results as Nrf2+c-Jun. to c-Jun. In a related experiment, c-Jun co-precipitated Nrf2+v-Jun required preincubation with cytosolic with Nrf2 antibody (data not shown). Overexposure of extract to bind to the hARE (Figure 10c). Replace- immunoblot in one of the experiment of Figure 9a ment of v-Jun with in vitro translated v-JunL3P revealed presence of several proteins in addition to resulted in the loss of binding of Nrf2 to the hARE Nrf1 and Nrf2 that associate with c-Jun (Figure 9c). (Figure 10c). Band shift assays with GST Ya gene The c-Jun antibodies also immunoprecipitated Nrf1 ARE and in vitro translated Nrf2+c-Jun demonstrated and Nrf2 from untransfected Hep-G2 cells (data not similar results as NQO1 gene hARE (Figure 10d). shown) indicating the presence of Nrf+Jun complexes Replacement of Nrf2 with Nrf1 in experiments shown in normal (untransfected) cells. in Figure 10 resulted in similar observations as Nrf2 In vitro translated c-Fos and c-Jun proteins bind to with the exception that higher concentrations of Nrf1 the hARE in band shift assays presumably due to was required to observe binding of Nrf1+c-Jun to the presence of AP1/AP1 like elements within the hARE hARE (data not shown). In a related experiment the (Figure 10a). Under similar conditions, Nrf2+c-jun incubation of in vitro translated Nrf2+c-Jun with Nrf2 and Nrf1+c-Jun failed to bind to the hARE (Figure and c-Jun antibodies resulted in super shifting of 10a ± data shown only for Nrf2). Preincubation of cytosolic extract-dependent binding of Nrf+Jun in Nrf2+c-Jun with varying amounts of Hepa-1 nuclear band shift assays with hARE (Figure 11). The super extract did not result in binding of Nrf2+c-Jun to the shifted bands were observed with both Nrf2 and c-Jun hARE (Figure 10b). However, preincubation of antibodies indicating the presence of these proteins in Nrf2+c-Jun with di€erent concentrations of cytosolic the cytosolic extract dependent hARE+(Nrf+Jun) Nrf association with Jun R Venugopal and AK Jaiswal 3149

Figure 5 E€ect of b-naphtho¯avone (b-NF) on Nrf1+c-Jun and Nrf2+c-Jun regulation of hARE-mediated expression of CAT gene. Hep-G2 cells were co-transfected with reporter plasmid hARE-tk-CAT and expression plasmids LNCX (vector alone) or LNCX-Nrf1+LNCX-c-Jun or LNCX-Nrf2+c-Jun in separate experiments. Five micrograms of RSV-b-galactosidase plasmid was used in each case as control of transfection eciency. Thirty- six hours after the transfection, the cells were treated with DMSO (control) or di€erent concentrations of b-NF. Forty-eight hours Figure 4 E€ect of overexpression of Nrf1 and Nrf2 along with c- after the transfection, the cells were analysed for b-galactosidase Fos on hARE-mediated CAT gene expression. Five micrograms and CAT activities. CAT activities are expressed as mean+s.e. of of reporter plasmid hARE-tk-CAT were co-transfected in Hep-G2 three independent transfection experiments cells along with ®ve micrograms of LNCX(vector), LNCX-Nrf1, LNCX-Nrf2, LNCX-c-Fos individually and in combinations as shown. LNCX (vector alone) DNA was used to normalize the amount of expression plasmid to 10 mg in each transfection). Five micrograms of RSV-b-galactosidase plasmid was used in each peroxidases, and oxidases), which activate carcino- case as control of transfection eciency. Forty-eight hours after gens, and phase II enzymes (quinone oxidoreductases, the transfection, the cells were analysed for b-galactosidase and CAT activities. CAT activities are expressed as mean+s.e. of glutathione S-transferases, UDP-glucuronosyl trans- three independent transfection experiments ferases, sufotransferases, epoxide hydrolases etc), which detoxify them (Talalay et al., 1995). ARE sequences have been characterized in the promoter regions of several genes including NQO1, GST Ya, complex. It may be noteworthy that lowest band oxygenase (HO) and g-glutamylcysteine synthe- observed in band and super shift assays in Figures 10 tase (g-GCS) (Rushmore and Pickett, 1993; Jaiswal, and 11 is due to non-speci®c adsorption of protein(s) 1994; Nebert, 1994; Prestera et al., 1995; Mulkahy et from rabbit reticulocyte lysate. Therefore, lowest non- al., 1997). ARE-mediated expression and coordinated speci®c band was always observed in all cases that used induction of detoxifying enzymes shift the balance in in vitro translated Nrf+Jun proteins for band shift favor of phase II enzymes and provide the required assays. This non-speci®c band has also been observed protection to the cells against electrophilic and by other investigators in band shift assays (Nakabeppu oxidative stress due to exposure to chemicals, drugs et al., 1988). and xenobiotics. Therefore, the nuclear proteins that A model is presented in Figure 12 to demonstrate bind to the ARE and activate detoxifying enzyme the role of Nrf, Jun and Fos proteins in the regulation genes are part of a signal cascade that plays an of ARE-mediated expression and induction of detox- important role in chemoprevention. Nrf1, Nrf2, c-Jun, ifying enzyme genes. Jun-B, Jun-D, c-Fos, Fra1 and Fra2 proteins are known to bind to the NQO1 gene hARE (Li and Jaiswal, 1992, 1994; Xie et al., 1995; Venugopal and Discussion Jaiswal, 1996). Earlier, we have shown that over- expression of transcription factors Nrf1 and Nrf2 The detoxifying enzymes play an important role in positively and c-Fos and Fra1 negatively regulate the determining the risk of cancer in human and hARE-mediated expression of NQO1 gene (Venugopal laboratory animals (Talalay et al., 1995). In other and Jaiswal, 1996). Nrf1 and Nrf2 proteins were also words, development of neoplasia may be regulated by a found to regulate hARE-mediated induction of NQO1 balance between phase I enzymes (cytochromes P450 gene expression in response to b-NF and t-BHQ and P450 reductase, hydroxylases, lipoxygenases, (Venugopal and Jaiswal, 1996). In addition, we Nrf association with Jun R Venugopal and AK Jaiswal 3150 increase in hARE- and GST Ya ARE-mediated gene expression was signi®cantly higher than the additive e€ect of individual proteins; (ii) Overexpression of v- Jun in combination with Nrf1 and Nrf2 also led to signi®cant increases in hARE-mediated CAT gene expression. However, replacement of v-Jun with v- JunL3P resulted in signi®cantly reduced expression of hARE-tk-CAT in transfected Hep-G2 cells. This was due to signi®cantly reduced capacity of v-JunL3P to heterodimerize with other leucine zipper proteins (Nrf1 and Nrf2 in the present study) due to mutation of third leucine to proline in its leucine zipper region; (iii) Nrf1 and Nrf2 co-precipitated with c-Jun in immunopreci- pitation experiments with c-Jun antibodies; (iv) The in vitro translated Nrf+c-Jun/v-Jun proteins bind to the NQO1 gene hARE and GST Ya gene ARE; and (v) v- JunL3P, a mutant form of v-Jun containing lethal mutation in its leucine zipper region failed to heterodimerize/associate with Nrf2 as evident from the loss of binding of in vitro translated Nrf2/v-JunL3P combined to the NQO1 gene hARE. Interestingly, binding of in vitro translated Nrf2+c- Jun and Nrf1+c-Jun to the NQO1 gene hARE and GST Ya gene ARE required preincubation of these proteins with Hepa-1 cytosolic extracts. This indicated that Hepa-1 cytosolic extract contains unknown factor(s) that presumably modify Nrf and/or c-Jun for heterodimerization/binding to the ARE. It is possible that the cytosolic factor(s) may be kinase(s)/ phosphatase(s) or redox protein(s) that are required for modi®cation of Nrf/c-Jun leading to their binding to the ARE. It is noteworthy that Nrf and c-Jun both contain phosphorylation sites and cysteines (potential sites for redox regulation). While a vast amount of knowledge is available on phosphorylation/redox regulation of c-Jun protein, the similar information Figure 6 E€ect of varying concentrations of Nrf1, Nrf2 and c- for Nrf proteins are lacking. Jun proteins on hARE-mediated expression of CAT gene. Five We also found that c-Jun concentration within the micrograms of reporter plasmid hARE-tk-CAT were co- transfected with expression plasmids LNCX-Nrf1, LNCX-Nrf2 cells was important for Nrf1 and Nrf2 regulated and c-Jun in di€erent concentrations and combinations as shown. expression of the CAT gene. The smaller increases in Five micrograms of RSV-b-gal plasmid were used in each case as the concentration of c-Jun resulted in the upregulation internal control of transfection eciency. Forty-eight hours after of hARE-mediated CAT gene expression. However, the transfection, the cells were analysed for b-galactosidase and CAT activities. CAT activities are expressed as mean+s.e. of higher increases in the c-Jun concentration inhibited/ three independent transfection experiments repressed the amount of upregulation of Nrf1 and Nrf2 regulated hARE-mediated CAT gene expression. In addition, overexpression of c-Fos also repressed the hARE-mediated CAT gene expression. These results reported that overexpression of c-Jun led to only indicated that higher concentrations of c-Jun and marginal increases in the hARE-mediated CAT gene overexpression of c-Fos may lead to the formation of expression (Venugopal and Jaiswal, 1996). sucient amounts of c-Jun+c-Fos complex which Nrf1 and Nrf2 are members of leucine zipper (b-zip) interferes with the binding of positive regulatory proteins (Chan et al., 1993; Moi et al., 1994). They do factors Nrf1+c-Jun and Nrf2+c-Jun complexes at not heterodimerize with each other (Chan et al., 1993; the ARE binding site. This may also explain why we Moi et al., 1994; Venugopal and Jaiswal, 1996). did not observe a signi®cant e€ect of overexpression of Therefore, it was expected that Nrf1 and Nrf2 c-Jun on hARE-mediated CAT gene expression in heterodimerize with another b-zip protein to bind to transfected Hep-G2 cells reported by us previously the ARE and activate gene expression. In the present (Venugopal and Jaiswal, 1996). We think that in those study, we have shown that Nrf1 and Nrf2 associate/ experiments, we used a concentration of LNCX-c-Jun heterodimerize with c-Jun to upregulate hARE- plasmid which balanced the positive regulation by mediated expression and b-NF induction of the Nrf1+c-Jun with negative regulation by c-Jun+c-Fos. NQO1 gene. The support for this theory came from These observations are consistent with the role of c-Jun several experimental observations: (i) Overexpression in positive and negative regulation of urokinase gene of Nrf1 and Nrf2 along with c-Jun in transfected Hep- expression (Cesare et al., 1995). The positive and G2 cells resulted in signi®cant increases in NQO1 gene negative regulation of urokinase gene expression hARE and GST Ya gene ARE-mediated CAT gene depends on association of c-Jun with other b-zip expression and induction in response to b-NF. The proteins. c-Jun heterodimerization with ATF-2 results Nrf association with Jun R Venugopal and AK Jaiswal 3151

Figure 7 E€ect of overexpression of Nrf1, Nrf2 and c-Jun on GST Ya gene ARE-mediated CAT gene expression. Five micrograms of GST Ya ARE-tk-CAT plasmid were co-transfected with expression plasmids LNCX (vector alone), LNCX-Nrf1, LNCX-Nrf2 and LNCX-c-Jun individually and in combinations as shown. Five micrograms of RSV-b-gal plasmid were used as internal control. Forty-eight hours after the transfection, the cells were analysed for b-galactosidase and CAT activities. CAT activities are expressed as mean+s.e. of four independent transfection experiments

Figure 8 Alignment of DNA binding domains of seven leucine zipper (b-zip) genes. Basic and leucine zipper regions of DNA binding domains of c-Fos, Fra1, c-Jun, Jun-B, Jun-D, Nrf1 and Nrf2 are aligned. Conserved cysteine (C) region in basic domain and leucines in leucine zipper regions of the various genes are circled. It may be noteworthy that Nrf1 and Nrf2 contains mutated leucines in their leucine zipper regions as compared to Jun, Fos and Fra proteins. Nrf1 contains two mutated leucines and Nrf2 contains one mutated leucine

in positive regulation of urokinase gene expression. Nrf1 and Nrf2 in association with Jun proteins also However, c-Jun heterodimerization with c-Fos leads to regulated ARE mediated GST gene expression. This negative regulation of urokinase gene expression. was expected because NQO1 and GST Ya genes are The various experiments also indicated that Nrf2 known to be coordinately upregulated by ARE was more ecient than Nrf1 in upregulation of hARE- (Rushmore and Pickett, 1993; Jaiswal, 1994). Re- mediated CAT gene expression. This di€erence in the cently, a 74.3 kDa protein designated as YABP has capacities of Nrf1 and Nrf2 may be attributed to the been shown to bind to GST Ya gene ARE (Liu and number of mutated leucines in their leucine zipper Pickett, 1996). At present, it is not clear if YABP is regions. The leucine zipper region of b-zip proteins are related to 65 kDa Nrf1 and 68 kDa Nrf2 proteins or a important for homo- or heterodimerization of b-zip new class of protein(s). More recently, hMAF, a small proteins for stabilization of the complex and ecient human transcription factor is shown to homodimerize binding at respective cis-element (Angel and Karin, and heterodimerize with Nrf1 and Nrf2 (Marini et al., 1991). Nrf1 contains two mutated leucines as compared 1997). Although hMAF-hMAF homodimers and to one mutated leucine in Nrf2. Therefore, Nrf1-c-Jun hMAF/Nrf1 and hMAF/Nrf2 heterodimers bind to complex will be expected to be less stable than Nrf2-c- the similar NF-E2 binding site, they exert functionally Jun complex leading to more ecient performance of opposite e€ects. The hMAF heterodimerization with Nrf2 as compared to Nrf1. Nrf1 and Nrf2 result in activation of b-globin gene Other Jun proteins (Jun-B and Jun-D) were found to expression. However, hMAF homodimerization leads be more or less equally ecient as c-Jun in association to repression of b-globin gene expression. Because with Nrf1 and Nrf2 leading to the upregulation of hMAF heterodimerizes with Nrf1 and Nrf2, the role of hARE-mediated CAT gene expression. Because of hMAF, if any, in the regulation of Nrf1+c-Jun and similar reasons as observed with c-Jun, Nrf2 combina- Nrf2+c-Jun regulated hARE-mediated expression of tions with Jun-B and Jun-D were more e€ective than detoxifying enzyme genes remains to be determined. Nrf1 with Jun-B and Jun-D respectively. Recently, Nrf2-MafK heterodimer was shown to bind Nrf association with Jun R Venugopal and AK Jaiswal 3152

Figure 9 Immunoprecipitation of Nrf1 and Nrf2 with c-Jun. (a) The Hep-G2 cells were transiently transfected with LNCX-Nrf1C, LNCX-Nrf2C and LNCX-c-Jun to overexpress the respective proteins. Thirty-six hours after the transfection, the Hep-G2 cells were starved for methionine followed by in vivo labeling of proteins with [35S]methionine. The nuclei were isolated, lysed and used to immunoprecipitate with preimmune serum and antibodies against c-Jun. The immunoprecipitated proteins were run on denaturing polyacrylamide gel. The proteins were ®xed and [35S]methionine signal ampli®ed by treatment as described in Materials and methods. The gel was dried and autoradiographed. (b). In a related experiment, the Hep-G2 cells without labeling with [35S]methionine were used for immunoprecipitation experiment as described above. The immunoprecipitated proteins were run on denaturing polyacrylamide gel, Western blotted and probed with antibodies against Nrf1 and Nrf2 and developed with ECL reagents. (c). The immunoblot from one of the experiments as described in (a) was overexposed for three days to detect other proteins that associate with c-Jun

to GST Ya subunit gene ARE leading to the signals (Schrech and Baeuerle, 1994). Two other suggestion that it plays a role in the ARE-mediated eukaryotic transcription factors, induction of GST Ya gene expression (Itoh et al., (SRF) and hypoxia response factor HIF-1 have also 1997). The role of mafK in Nrf1/Jun and Nrf2/Jun been suggested to be altered by reactive oxygen (Meyer regulation of ARE-mediated expression of NQO1 gene et al., 1993). Therefore, it is possible that ROS remains to be determined. generated during metabolic activation of xenobiotics The various steps in the mechanism of signal and antioxidants may lead to the increased expression transduction from xenobiotics and antioxidants to the and/or modi®cation of Nrf and Jun proteins resulting Nrf1, Nrf2, Jun and Fos proteins that bind to the ARE in increased binding of these proteins to the ARE and/ and regulate ARE-mediated basal expression and or increased transcriptional activity of these proteins coordinated induction of the various detoxifying that bind to the ARE resulting in coordinated enzyme genes remain unknown. Xenobiotics and induction of the various detoxifying enzyme genes. antioxidants undergo metabolism to generate electro- Recently, it has been reported that the quinone- philes and reactive oxygen species (ROS) (Halliwell, mediated generation of hydroxyl radicals induces the 1989; DeLong et al., 1987; Figure 12). ROS has been ARE mediated expression of the mouse glutathione S- implicated in the induction of a number of transcrip- transferase gene (Pinkus et al., 1996; Favreau and tion factors in bacteria, yeast and mammalian cells Pickett, 1993). This correlation remains unknown in (Pahl and Baeuerle, 1994). These include OxyR and the case of AREs from other genes including NQO1

SoxRS, which activate more than a dozen genes in gene hARE. However, hydrogen peroxide (H2O2) has bacteria (Zitomer and Lowry, 1992), yAP-1 and yAP-2 been shown to induce the ARE mediated expression of in the yeast, which regulate the expression of oxidative rat GST Ya, rat NQO1 and human NQO1 genes (Li stress response genes superoxide dismutase, glutathione and Jaiswal, 1994; Cesare et al., 1995; Rushmore et al., reductase and glucose 6-phosphate dehydrogenase 1991) and may be an important intermediate in the (Zitomer and Lowry, 1992; Kataoka et al., 1995) and induction of these genes by the antioxidants and Mac-1, which activates catalase gene expression xenobiotics. The role of electrophiles, if any, in the (Jungmann et al., 1993) and NF-kB and AP1 in ARE mediated expression and induction of detoxifying higher eukaryotic cells which regulate the expression of enzyme genes remains unknown. The present studies several genes in immune response and antagonistic also indicate that signal transduction mechanism for Nrf association with Jun R Venugopal and AK Jaiswal 3153

Figure 10 Band shift assays with ARE and in vitro translated Nrf and Jun proteins. hARE and GST Ya ARE were 32P-labeled with [a-32P]dCTP and klenow. 40 000 to 50 000 c.p.m. (52 ng) of labeled DNA was mixed with in vitro translated proteins or Hepa-1 nuclear or cytosolic extracts and band shift assays performed by procedures as described in Materials and methods. Only shifted bands are shown. The speci®c binding of proteins are indicated by arrows. The lower band is due to non-speci®c interaction of rabbit reticulocyte lysate with the ARE and is observed when in vitro translated proteins are used for binding assays. The non- speci®c lower band was not observed with Hepa-1 nuclear or cytosolic extracts. NE, Hepa-1 nuclear extract; CE, Hepa-1 cytosolic extract. (a). 32P-labeled hARE was mixed with 0.5 mgofin vitro translated c-Fos+c-Jun, Nrf2+c-Jun and b-galactosidase (control) proteins preincubated at 378C for 15 min for heterodimerization and band shift assay performed. (b). In vitro translated Nrf2+c- Jun were preincubated with di€erent concentrations of Hepa-1 nuclear (NE) or cytosolic (CE) extracts at 378C for 15 min for modi®cations/heterodimerization and used in band shift assays with hARE. hARE was also incubated with Hepa-1 nuclear (NE) and cytosolic extracts (CE). (c) hARE was incubated with 0.5 mg in vitro translated Nrf2+v-Jun or Nrf2+v-JunL3P previously preincubated with 20 mg of Hepa-1 cytosolic extract at 378C for 15 min for modi®cation and heterodimerization. In addition, hARE was also incubated with 2 mg of Hepa-1 nuclear extract (NE) or with 0.5 mg of Nrf2+c-Jun preincubated with 20 mg cytosolic extract. (d) Band shift assay with GST Ya gene ARE. Hepa-1 cytosolic extract (15 0006g, 1 h supernatant)

Nrf+c-Jun regulated ARE-mediated expression of However, the identi®cation of unknown cytosolic detoxifying enzyme genes involves unknown cytosolic factor(s) as kinase(s)/phosphatase(s) could not be factor(s). However, the nature of cytosolic factor(s) ruled out. remains unknown. The identi®cation of unknown In conclusion, we have shown that nuclear cytosolic factor(s) in the stimulation of Nrf+Jun transcription factors Nrf2 and Nrf1 associate with binding to the hARE is an important step to our Jun (c-Jun, Jun-B and Jun-D) proteins and upregulate understanding of the mechanism of signal transduction ARE-mediated basal expression of detoxifying enzyme from endogenous substances, xenobiotics and antiox- genes and coordinated induction in response to idants to the Nrf+Jun proteins (Figure 12). The xenobiotics and antioxidants. Nrf-c-Jun association/ unknown cytosolic factor(s) may be a kinase/phospha- binding to the ARE requires unknown cytosolic tase or redox protein(s) that catalyzes the necessary factor(s). Nrf2 containing one mutation in its leucine modi®cation of Nrf and/or Jun proteins to stimulate zipper region was more ecient in upregulation of their heterodimerization and/or binding to the ARE hARE-mediated NQO1 and GST Ya ARE -mediated leading to activation of detoxifying enzyme genes GST genes expression, as compared to Nrf1 that expression. The previous studies demonstrating in- contains two mutated leucines in its leucine zipper creased binding of nuclear factors to the hARE and region. We have also shown that c-Fos+c-Jun comlex increased expression of hARE-mediated CAT gene when present in suciently high amounts, interferes expression in resonse to dithiothretol anf b-mercap- with the binding of positive regulatory complex toethanol (Li and Jaiswal, 1994), may suggest that Nrf+Jun at the ARE binding site resulting in unknown cytosolic factor(s) may be redox protein(s). repression of hARE-mediated CAT gene expression. Nrf association with Jun R Venugopal and AK Jaiswal 3154

Figure 12 A hypothetical model showing upregulation of ARE- mediated gene expression. Endogenous substances, xenobiotics and antioxidants are metabolized by cytochromes P450 and related enzymes to generate electrophiles and reactive oxygen species (ROS). ROS and/or electrophiles may activate unknown cytosolic factor(s) which modify Nrf1, Nrf2 and/or Jun proteins. Nrf1 and Nrf2 proteins associate with c-Jun, Jun-B and Jun-D proteins to upregulate ARE-mediated gene expression. The preliminary experiments also indicate that higher concentrations of Jun proteins result in increased amount of Jun+Fos complex which interferes with the binding of Nrf+Jun complex at the ARE binding site and represses the ARE-mediated gene expression and induction (not shown)

plasmid pGST Ya ARE-tk-CAT. Purine to pyrimidine (and vice versa) changes were incorporated to construct mutant GST Ya gene ARE-tk-CAT plasmid. Both mutant NQO1 gene hARE and mutant GST Ya gene ARE contained mutations in the ARE core sequences resulting in the loss of ARE-mediated gene expression and induction (Rush- more et al., 1991; Xie et al., 1995). The construction of expression plasmids LNCX-c-Jun, LNCX-Jun-B, LNCX- Jun-D, LNCX-c-Fos, LNCX-Fra1, LNCX-Nrf1 and LNCX-Nrf2 have been previously described (Venugopal and Jaiswal, 1996).

Cell culture and cotransfection of reporter and expression Figure 11 Super shift with hARE and in vitro translated c-Jun plasmids and Nrf2 proteins. In vitro translated Nrf2+c-Jun were preincubated with 4 ml of pre-immune serum or Nrf2 antibody Human hepatoblastoma (Hep-G2) cells were grown in or c-Jun antibody for 90 min at 48C followed by incubation with monolayer cultures as described (Li and Jaiswal, 1992). 20 mg of Hepa-1 cytosolic (CE) extracts at 378C for 15 min for Five micrograms of reporter plasmid hARE-tk-CAT, modi®cations/heterodimerization and used in band shift assays mutant hARE-tk-CAT, GST Ya ARE-tk-CAT, mutant with hARE by procedures as described in Materials and methods. GST Ya ARE-tk-CAT or TRE-tk-CAT was mixed with 10 SB, Shifted band; SSB, Super shifted band. The lowest band in micrograms of expression plasmid LNCX vector alone or each lane is non-speci®c adsorption of proteins from rabbit LNCX-c-Jun, LNCX-Jun-B, LNCX-Jun-D, LNCX-v-Jun; reticulocyte lysate used to in vitro translate the Nrf2 and c-Jun LNCX-v-JunL3P; LNCX-c-Fos, LNCX-Fra1, LNCX-Nrf1 proteins and LNCX-Nrf2 and cotransfected in Hep-G2 cells by procedures as described (Venugopal and Jaiswal, 1996). Five micrograms of Rous sarcoma virus (RSV)-b-galacto- sidase (b-gal) plasmid was included in each transfection to Materials and methods normalize transfection eciency. The transfected Hep-G2 cells were harvested 48 h after transfection, homogenized by sonication and analysed for the levels of b-gal and CAT Plasmid construction activity. In several experiments, the transfected cells were Nucleotide sequences of human NQO1 gene ARE (hARE), treated with several di€erent concentrations of b-naphtho- mutant hARE, rat GST Ya gene ARE, mutant GST Ya ¯avone (b-NF) and tert-butylhydroquinone (t-BHQ). In gene ARE and human collagenase gene TRE are shown in related experiments, the concentration of an expression Figure 1. The construction of hARE-, mutant hARE- and plasmid was changed with a constant concentration of a TRE-thymidine kinase (tk)-CAT plasmids have been second expression plasmid. described (Xie et al., 1995). Both strands of the GST Ya The proteins encoded by expression plasmids in the gene ARE (region between 7722 to 7689 of rat GST Ya transfected Hep-G2 cells were characterized and measured gene (Rushmore et al., 1991) with BamHI ends were by Western analysis and probing with speci®c antibodies as synthesized, annealed, kinased, and cloned at the BamHI described previously (Venugopal and Jaiswal, 1996). The site of the plasmid pBLCAT2 to generate recombinant densitometric analysis of Western blots indicated increase of Nrf association with Jun R Venugopal and AK Jaiswal 3155 two- to threefold in the expression of transcription factor(s) vector. v-JunL3P is a mutant form of v-Jun (Kataoka et in transfected cells as compared with the vector-transfected al., 1995). The third leucine in the leucine zipper region of controls. The speci®c fold inductions upon transfection with v-Junwasmutatedtoprolinetogeneratev-JunL3P.This 5 mg of expression plasmids were as follows: LNCX-Nrf1R, mutation is known to signi®cantly reduce the ability of v- 1.0-fold; LNCX-Nrf1C, 2.1-fold; LNCX-Nrf2R, 1.0-fold; Jun to heterodimerize with other leucine zipper proteins LNCX-Nrf2C, 2.8-fold; LNCX-c-Jun, 3.1-fold; LNCX-Jun- resulting in the loss of DNA binding (Kataoka et al., D, 2.4-fold; LNCX-Jun-B, 2.2-fold; LNCX-v-Jun, 3.6-fold; 1995). Plasmid DNAs carrying various cDNAs (Nrf2, Nrf1 LNCX-v-JunL3P, 3.3-fold. Increase in the amount of and c-Jun) were digested with appropriate restriction expression plasmids from 5 ± 10 mg of DNA led to further enzymes and treated with proteinaseK following phenol/ increases in the fold expression of respective transcription chloroform extraction and ethanol precipitation. To factors. The fold increases with 10 mg expression plasmids synthesize 5' capped RNA, linearized DNA templates were two times higher than observed with 5 mg expression were transcribed in vitro using a commercial mRNA plasmids. Therefore, the amount of transcription factor(s) capping kit from Promega. RNA was puri®ed according produced from their respective cDNA(s) correlated with the to the manufacturer's manual. One microgram of RNA amount of expression plasmid(s) used in transfections (data was translated in vitro using a rabbit reticulocyte lysate not shown). LNCX-Nrf1R and LNCX-Nrf2R contained the (50 ml of total reaction volume) with or without 50 mCi of respective cDNA in reverse orientation and failed to [35S]methionine (*100 Ci/mmol)] as directed by the overexpress the proteins in transfected Hep-G2 cells. supplier (Promega Biotech). In vitro translated Nrf2, Nrf1, c-Jun, v-Jun and v-JunL3P were stored at 7808C. 35S-methionine-labeled translation products were analysed Antibodies and immunoprecipitation on 12.5% SDS-polyacrylamide gel by procedures as Anti-c-Jun (against peptide MLTQQLQTF) of murine c- described (Laemmli, 1970). A 14C-labeled protein mixture Jun, Anti-c-Fos (against peptide YGKVEQLSPEEEEKR- wasusedformolecularweightstandards.Afterelectro- RIRRERNKMAAA of mouse c-Fos were obtained from phoresis, the gel was ®xed with 10% acetic acid/30% Dr Kevin Ryder (NIH, Bethesda). These peptides are methanol, treated with NAMP 100 from Amersham to speci®c for c-Jun and c-Fos and are selected from regions amplify the 35S signal. The gel was dried and autoradio- distinct from DNA binding domain or transcriptional graphed. In addition, the in vitro translated products were activation domain. Antibodies raised against these pep- con®rmed by SDS ± PAGE and Western analysis. tides speci®cally detect the presence of corresponding proteins and do not interfere in DNA binding or DNA binding activation of transcription. These antibodies have been used to detect c-Jun and c-Fos in super shift assays The nuclear extract from Hepa-1 cells was prepared (Nakabeppu et al., 1988; Li and Jaiswal, 1992). Nrf1 and according to the procedure of Kadonaga (1991). In vitro Nrf2 antibodies were purchased from Santa Cruz translated Nrf2, Nrf1, c-Jun, v-Jun and v-JunL3P were Biotechnology, Inc.). These are speci®cally designed to mixed in various combinations in equimolar ratios and detect Nrf1 and Nrf2 proteins in super shift assays. The incubated at 378Cfor15mininabsenceandpresenceof Nrf1 and Nrf2 antibodies have been previously used to Hepa-1nuclearorcytosolic(15000g, 1 h supernatant) detect Nrf1 and Nrf2 in super shift assays (Venugopal and extracts. DNA binding of in vitro translated proteins to the Jaiswal, 1996). NQO1 gene hARE and GST Ya gene ARE were performed Hep-G2 cells were grown as monolayers in 100 mm petri by band shift assays by procedures as described below. dishes and co-transfected with LNCX-Nrf2C, LNCX-Nrf1C 40 000 ± 50 000 c.p.m. (52ng)ofthe 32P-labeled hARE with LNCX-c-Jun to overexpress the respective proteins. The was mixed with 4 mg poly (dI-dC).poly (dI.dC) and 0.2 ± cells from ®ve transfected petri dishes were washed with PBS, 0.5 mgofin vitro translated and preincubated (378Cfor scraped and nuclei isolated by procedure as described 15 min for heterodimerization) of Nrf2, Nrf1, c-Jun, v-Jun (Dignam et al., 1983; Kadonaga, 1991). The nuclei were and v-JunL3P individually and in various combinations in lysed in 1 ml of RIPA bu€er (50 mM Tris pH 7.4, 150 mM the presence of 25 mM HEPES (K+), pH 7.8, 12.5 mM

NaCl, 0.5% Triton X-100, 11.6 mM deoxycholic acid and MgCl2,1mM DTT, 20% glycerol (v/v), 0.1% Nonidet 0.1% SDS). One half of the nuclear extract was mixed with P-40 and 0.1 M KCl and incubated at room temperature 40 ml of preimmune serum and another half with similar for 20 min. The samples were electrophoresed at 25 mA, amount of c-Jun antiserum and incubated with shaking at gel-dried under vacuum, and autoradiographed. In related 48C for 1 h. The immunoprecipitate was adsorbed on 25 mlof experiments, the concentration of DTT was varied from protein A sepharose in eppendorf tubes for 2 h at 48C. The 0±10mM in band shift assays to determine its e€ect on protein A sepharose beads were collected by centrifugation binding of Nrf-Jun proteins to the hARE. In other and washed with wash bu€er I (100 mM Tris, pH 8.7, experiments, NQO1 gene hARE was replaced with GST 500 mM lithium chloride and 140 mM of b-mercaptoetha- Ya gene ARE. In related experiments, the in vitro nol) followed by wash bu€er II (100 mM Tris, pH 8.7, 2 M translated c-Jun+Nrf2 proteins were pre-incubated with KCl and 140 mM b-mercaptoethanol) and wash bu€er III speci®c Nrf2 and c-Jun antibodies before band shift assays (1% SDS in RIPA bu€er). The beads were suspended in with hARE. 20 ml of sample bu€er (50 mM Tris pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromophenol blue and 10% glycerol) and loaded on a 12% polyacrylamide gel. The proteins were electrophoresed and the gel was incubated in methanol : acetic acid : water (4 : 5 : 1) to ®x the proteins. The gel was also Abbreviations treated with amplify (NAMP 100) from Amersham to The abbreviations used are: ARE, antioxidant response amplify the 35S signal. The gel was dried and autoradio- element; hARE, human NQO1 gene ARE; GST Ya ARE, graphed. Similar experiment was also performed with GST Ya subunit gene ARE; TRE; TPA response element; untransfected Hep-G2 cells. NQO1, AND(P)H:Quinone Oxidoreductase1; GST Ya, Glutathione S-transferase Ya gene; Nrf1 and Nrf2, NF- E2 related nuclear transcription factor 1 and 2; c-Jun, Jun- In vitro transcription and translation B, Jun-D, c-Fos and Fra1, Leucine zipper containing b-zip The coding sequences of Nrf2, Nrf1, c-Jun, v-Jun and v- family of nuclear transcription factors; b-NF, b-naphtho- JunL3P were separately subcloned in pBluescriptSK+ ¯avone; t-BHQ, tert-butyl hydroquinone; bp, base pairs. Nrf association with Jun R Venugopal and AK Jaiswal 3156 Acknowledgements the cDNAs encoding Nrf1 and Nrf2 and to Dr Kevin We thank our colleagues for helpful discussion. We also Ryder (National Institute of Health, Bethesda) for the gift thank Drs Hagop Youssou®an (Baylor College of Medi- of cDNAs for Jun, Fos and Fra proteins. We are also cine) for critically reading the manuscript. We are grateful grateful to Dr Masakatsu Nishizawa for the gift of cDNA to Drs Je€erson Y Chan and Yuet W Kan, both from encoding v-Jun and v-JunL3P. This investigation was University of California, San Francisco for providing us supported by NIH grant GM47466.

References

Angel P and Karin M. (1991). Biochem. Biophys. Acta, 1072, MoiP,ChanK,AsunisI,CaoAandKanYW.(1994).Proc. 129 ± 157. Natl. Acad. Sci., 91, 9926 ± 9930. Bergelson S, Pinkus R and Daniel V. (1994). Oncogene, 9, Mulkahy RT, Wartman MA, Bailey HH and Gipp JJ. (1997). 565 ± 571. J. Biol. Chem., 272, 7445 ± 7454. Cesare DD, Vallone D, Caracciolo A, Saaaone-Corsi P, Nakabeppu Y, Ryder K and Nathans D. (1988). Cell, 55, Nerlov C and Verde P. (1995). Oncogene, 11, 365 ± 376. 907 ± 915. Chan JY, Han X and Kan YW. (1993). Proc. Natl. Acad. Sci. Nebert DW. (1994). Biochem. Pharm., 47, 25 ± 37. USA, 90, 11371 ± 11375. Pahl HL and Baeuerle PA. (1994). BioEssays, 16, 497 ± 502. De Long MJ, Santamaria AB and Talalay P. (1987). Pinkus R, Weiner LM and Daniel V. (1996). J. Biol. Chem., Carcinogenesis, 8, 1549 ± 1553. 271, 13422 ± 13429. Dignam JD, Lebovitz RM and Roeder RG. (1983). Nucleic Prestera T, Holtzclaw WD, Zhang Y and Talalay P. (1993). Acids Res., 11, 1475 ± 1489. Proc. Natl. Acad. Sci. USA, 90, 2965 ± 2969. Favreau LV and Pickett CB. (1993). J. Biol. Chem., 268, PresteraT,TalalayP,AlamJ,AhnYI,LeePJandChoiAM. 19875 ± 19881. (1995). Mol. Medicine, 1, 827 ± 837. Friling R, Bergelson S and Daniel V. (1992). Proc. Natl. Rushmore TH, Morton MR and Pickett CB. (1991). J. Biol. Acad. Sci. USA, 89, 668 ± 672. Chem., 266, 11632 ± 11639. Halliwell B and Gutteridge JMC. (1989). Free Radicals in Rushmore TH and Pickett CB. (1993). J. Biol. Chem., 268, Biology and Medicine. Clarendon Press: Oxford. 11475 ± 11478. Itoh K, Chiba T, Takahashi S, Ishii T, Igarashi K, Katoh K, Schreck R and Baeuerle PA. (1994). Methods in Enzymol., Oyake T, Hayashi N, Satoh K, Hatayama I, Yamamoto M 234, 151 ± 163. and Nabeshima Y. (1997). Biochem. Biophys. Res. Talalay P, Fahey JW, Holtzclaw WD, Prestera T and Zhang Commun., 236, 313 ± 322. Y. (1995). Toxicology Letts., 82 ± 83, 173 ± 179. Jaiswal AK. (1994). Biochem. Pharmacol., 48, 439 ± 444. Vasliou V, Puga A, Chang C, Tabor MW and Nebert DW. Jungmann J, Reins HA, Lee J, Romeo A, Hassett R, Kosman (1995). Biochem. Pharm., 12, 2057 ± 2068. D and Jentsch S. (1993). EMBO J., 12, 5051 ± 5056. Venugopal R and Jaiswal AK. (1996). Proc. Natl. Acad. Sci., Kadonaga JT. (1991). Methods Enzymol., 208, 11 ± 24. 93, 14960 ± 14965. KataokaK,IgarashiK,ItohK,FujiwaraKT,YamamotoM Venugopal R, Joseph P and Jaiswal AK. (1997). Genes and Nishizawa M. (1995). Mol. Cell. Biol., 15, 2180 ± 2190. expression of DT-diaphorase in cancer cells. In: Forman Laemmli UK. (1970). Nature, 227, 680 ± 685. HJ and Cadenas E. (eds.). Oxidative stress and signal Li Y and Jaiswal AK. (1992). J. Biol. Chem., 267, 15097 ± transduction. Chapman and Hall: New York, pp. 441 ± 15104. 475. Li Y and Jaiswal AK. (1994). Eur. J. Biochem., 226, 31 ± 39. Wasserman W and Fahl WE. (1997). 94, 5361 ± 5366. Liu S and Pickett CB. (1996). Biochemistry, 35, 11517 ± Xie T, Belinsky M, Xu Y and Jaiswal AK. (1995). J. Biol. 11521. Chem., 270, 6894 ± 6900. MariniMG,ChanK,CasulaL,KanYW,CaoAandMoiP. Zitomer RS and Lowry CY. (1992). Microbiolial Rev., 56, 1± (1997). J. Biol. Chem., 272, 16490 ± 16497. 11. Meyer M, Schreck R and Baeuerle P. (1993). EMBO J., 12, 2005 ± 2015.