Oncogene (2002) 21, 2191 ± 2200 ã 2002 Nature Publishing Group All rights reserved 0950 ± 9232/02 $25.00 www.nature.com/onc

Site-directed mutagenesis of cysteine to serine in the DNA binding region of Nrf2 decreases its capacity to upregulate antioxidant response element-mediated expression and antioxidant induction of NAD(P)H:quinone oxidoreductase1 gene

David Bloom1, Saravanakumar Dhakshinamoorthy1 and Anil K Jaiswal*,1

1Department of Pharmacology, Baylor College of Medicine, One Baylor Plaza, Houston, Texas, TX 77030, USA

NF-E2 related factor 2 (Nrf2) is a CNC/b-zip protein mation (Thelen et al., 1993), ionizing radiation (Breen that regulates antioxidant response element (ARE)- and Murphy, 1995), and cellular respiration (Grisham mediated expression, and antioxidant induction, of and McCord, 1986). These stimuli result in the detoxifying enzyme genes, including NAD(P)H:quinone production of free radicals, including reactive oxygen oxidoreductase1 (NQO1). A comparison of Nrf2 from species (ROS). ROS, and xenobiotic and antioxidant di€erent species, and with other b-zip proteins, revealed generated electrophiles, attack DNA and other cellular the presence of a highly conserved cysteine residue at macromolecules, causing aging, neurodegenerative dis- position 506 in the DNA binding domain of Nrf2. Site- eases, arthritis, arteriosclerosis, and tumor induction directed mutagenesis was used to mutate this cysteine to and promotion (Breen and Murphy, 1995; Grisham serine. Transfection/over expression experiments in hu- and McCord, 1986; Ward, 1994; Rosen et al., 1995). man hepatoblastoma (Hep-G2) cells demonstrated that Cells have developed a defense mechanism to detoxify mutant Nrf2 (mNrf2), containing the C506S mutation, and remove electrophiles and ROS. During times of was signi®cantly less ecient in activating ARE-mediated electrophilic and oxidative stress, a battery of detoxify- gene expression, and induction in response to tert-butyl ing genes is coordinately induced through the anti- hydroquinone (t-BHQ), as copmpared with wild-type oxidant response element (ARE) (Rushmore and Nrf2. N-ethyl malemide (NEM), a sulfhydryl cross- Pickett, 1993; Jaiswal, 1994). These genes include linker, inhibited Nrf2 but not mNrf2C506S-mediated NAD(P)H:quinone oxidoreductases (NQOs), which expression of NQO1. This further implicated the cysteine compete with cytochromes P450 and P450 reductase at position 506 in Nrf2 regulation of ARE-mediated gene and catalyze two-electron reductive metabolism and expression. Nuclear localization experiments revealed detoxi®cation of quinones (Joseph et al., 1994; that C506S mutation did not a€ect the retention of Radjendirane et al., 1997; Talalay et al., 1995; Riley Nrf2 by INrf2/Keap1 in the cytosol, or its release in and Workman, 1992; Ernster, 1987); glutathione response to antioxidants. However, band and supershift S-transferases (GSTs), which conjugate hydrophobic assays showed a signi®cant reduction in the binding of electrophiles and ROS with glutathione (Pickett and mNrf2C506S to the NQO1 gene ARE as compared with Lu, 1989; Tsuchida and Sato, 1992); UDP-glucurono- wild-type Nrf2. Therefore, the C506S mutation in Nrf2 syl transferases (UDP-GT), which catalyze the con- lowered its anity for the ARE, leading to decreased jugation of glucuronic acid with xenobiotics and drugs, expression, and antioxidant induction, of NQO1. leading to their excretion (Tephly and Burchell, 1990); Oncogene (2002) 21, 2191 ± 2200. DOI: 10.1038/sj/ epoxide hydrolase (EH), which inactivates epoxides onc/1205288 (Oesch et al., 1991); g-glutamylcysteine synthetase (g-GCS), which plays a role in glutathione metabolism Keywords: NQO1; ARE; detoxifying genes; Nrf2; (Mulkahy et al., 1997); ferritin-L gene, which plays an cysteine; redox regulation important role in iron storage (Wasserman and Fahl, 1997); and heme oxygenase-1 (HO-1), which catalyses the ®rst and rate-limiting step in heme catabolism Introduction (Choi and Alam, 1996). In response to xenobiotic and antioxidant stress, Cellular oxidative stress is caused by a variety of Nrf2, a basic leucine zipper transcription factor (bZIP), external and internal stimuli, including xenobiotics and binds to the ARE and positively regulates ARE- antioxidants (Dhakshinamoorthy et al., 2000), in¯am- mediated induction of detoxifying genes, including NQO1 (Venugopal and Jaiswal, 1996). Nrf2 is also known to regulate ARE-mediated expression and induction of GST Ya, g-GCS and HO-1 genes *Correspondence: AK Jaiswal; E-mail: [email protected] Received 5 October 2001; revised 3 January 2002; accepted 7 (Venugopal and Jaiswal, 1996; Nguyen et al., 2000; January 2002 Jayepaul and Jaiswal, 2000; Wild et al., 1999; Alam et Redox regulation of Nrf2 D Bloom et al 2192 al., 1999). Under normal conditions, Nrf2 is held in the amount of pcDNA-mNrf2 in the same experiment cytoplasm by a cytosolic inhibitor INrf2/KEAP1 (Itoh resulted in 25% repression of ARE-mediated CAT et al., 1999; Dhakshinamoorthy and Jaiswal, 2001). gene expression. The other two concentrations of the When cells are challenged by oxidative stress, Nrf2 is expression plasmids also showed signi®cant di€erences released from INrf2/Keap1 and translocates to the between the eciency of Nrf2 and mNrf2 to upregulate nucleus, where it activates transcription. The signal, ARE-mediated CAT gene expression. This di€erence in which causes this release and activates Nrf2, is not activity was not due to a di€erence in the amount of clearly understood. Nrf2 has been shown to hetero- expression. HepG2 cells transfected with V5 tagged dimerize with c-Jun and small Maf proteins (MafG and Nrf2, or mNrf2, were lysed with RIPA bu€er, 100 mg MafK). These heterodimeric complexes bind to the of the whole cell extracts were resolved on an SDS ± ARE, and alter ARE-mediated gene expression PAGE. Western blots were performed with the V5 (Venugopal and Jaiswal, 1998; Dhakshinamoorthy antibody and demonstrated equal expression of the two and Jaiswal, 2000; Itoh et al., 1997; Nguyen et al., proteins in transfected Hep-G2 cells (Figure 2b). 2000; Wild et al., 1999). To determine if mNrf2 would respond to antiox- bZIP proteins, including c-Jun and c-Fos, contain a idants, both wild-type Nrf2 and mNrf2 were transiently highly conserved cysteine in their DNA binding transfected into HepG2 cells. The cells were then domains (Abate et al., 1990). This cysteine residue treated with 100 mM or 200 mM t-BHQ for 16 h. mNrf2 has been shown to be a site of redox regulation in activity was induced in response to 100 mM t-BHQ c-Jun, a€ecting its binding to the AP1 site (Abate et treatment, but again the response was attenuated as al., 1990). Interestingly, all species of Nrf2 also compared to wild-type Nrf2 (Figure 3). 200 mM t-BHQ contains this cysteine residue in their DNA binding treatment caused an inhibition of Nrf2 activity. mNrf2, region (Venugopal and Jaiswal, 1998). In addition, it however, was still induced by 200 mM t-BHQ treat- has been reported that ARE-mediated expression and ment. induction of the NQO1 gene is sensitive to sulfhydryl N-ethyl malemide (NEM) was used to determine if modifying agents (Li and Jaiswal, 1994). The present C506, in Nrf2, is a target for oxidation-reduction studies were performed to determine the role of this (redox) regulation. Hep-G2 cells cotransfected with the cysteine residue in Nrf2 regulation of ARE-mediated reporter plasmid, NQO1 ARE-tk-CAT, expression gene expression, and induction in response to anti- plasmid pcDNA, or pcDNA-Nrf2 were treated with oxidants and xenobiotics. 0.25 mM or 0.5 mM NEM. These cells showed a In this report, we show that site-directed mutagenesis signi®cant repression of ARE-mediated CAT expres- of cysteine to serine in the DNA binding domain of sion (Figure 4). However, in a similar experiment, the Nrf2 leads to a signi®cant decrease in the capacity of mNrf2 regulation of ARE-mediated CAT expression Nrf2 to upregulate ARE-mediated NQO1 gene expres- was more or less una€ected after treatment with NEM. sion, and induction in response to t-BHQ. The C506S GFP tagged proteins were used to show the mutation did not a€ect Nrf2 interaction with INrf2/ interaction of Nrf2/mNrf2 with INrf2 (Figure 5). The Keap1. This mutation also had no e€ect on the transfection of QT6 cells, with EGFP-Nrf2 or EGFP- heterodimerization of Nrf2 with MafG, or the binding mNrf2 led to the expression of the respective proteins of the Nrf2/MafG complex to the NQO1 gene ARE. in both the cytosolic and nuclear compartments However, binding of the mNrf2/c-Jun complex to the (Figure 5, left panels). The treatment of transfected ARE was signi®cantly reduced as compared to QT6 cells with DMSO had no e€ect on the localization Nrf2-c-Jun. Therefore, the cysteine at position 506 of of Nrf2 or mNrf2 (data not shown). The coexpression Nrf2 is required for interaction with speci®c hetero- of INrf2 with EGFP-Nrf2 or EGFP-mNrf2 retained dimeric partner proteins and/or binding of these the respective proteins in the cytosol (Figure 5, middle heterodimeric complexes to the ARE. panels). The treatment of these cells with t-BHQ caused the release of Nrf2, or mNrf2, which then translocated to the nucleus (Figure 5, right panels). Results Nrf2, mNrf2, MafG-V5 and c-Jun were in vitro translated; the translation was con®rmed by SDS ± The cysteine residue at position 506 of Nrf2 was PAGE (Figure 6b) and Western analysis (data not successfully converted to a serine residue using site shown). These proteins were used in band and super directed mutagenesis. The molecule created, shift assays. Nrf2-c-Jun heterodimers, MafG-V5 homo- Nrf2C506S, is referred to as mNrf2 (Figure 1). Over dimers and Nrf2/mNrf2-MafG-V5 heterodimers all expression of either wild-type Nrf2 or mNrf2 in HepG2 bound to the NQO1 gene ARE (Figures 6 and 7). cells lead to a dose dependent increase in ARE activity Nrf2-c-Jun binding to the ARE required pretreatment (Figure 2a). However, the activity of mNrf2 was with cytosolic extract, as reported earlier (Venugopal substantially lower than that of the wild-type Nrf2 at and Jaiswal, 1998). The endogenous proteins in the all the three concentration of the plasmid used for cytosolic extract created a slight background. This was transfection. The transfection of Hep-G2 cells with taken into account when analysing the results. 0.15 mg of pcDNA-Nrf2 plasmid led to more than Densitometric analysis of the Nrf2-c-Jun and mNrf2- 100% increase in ARE-mediated CAT gene expression. c-Jun bands revealed nearly a threefold decrease in However, the replacement of pcDNA-Nrf2 with similar binding of mNrf2-c-Jun to the ARE as compared to

Oncogene Redox regulation of Nrf2 D Bloom et al 2193

Figure 1 Alignment of Nrf2 with other CNC and b-zip proteins, and site-directed mutagenesis of Nrf2 to generate mutant mNrf2C506S. (a) Alignment of DNA binding (Basic) and heterodimerization (Leucine Zipper) regions of Nrf2 with similar regions of c-Jun, Nrf1, Nrf3 and small MafG proteins. The conserved cysteine and leucines are highlighted. (b) Alignment of DNA binding (Basic) region of Nrf2 from di€erent species. The conserved cysteine is highlighted. (c) Protein domains of Nrf2 and mutant Nrf2. The site of cysteine to serine mutation is shown. CNC, Cap`n'Collar wild-type Nrf2-c-Jun, P5.025 (Figure 6a,b). Equimo- lower proportion of Nrf2 and mNrf2 proteins, lar concentrations of Nrf2 and mNrf2 were used in the compared to AP1-binding proteins. It may be band shift assays as demonstrated by autoradiography noteworthy that the NQO1 gene ARE contains a of the in vitro translated proteins (Figure 6c). The perfect AP1 (Jun+Fos) binding site (Jaiswal, 1994). results also demonstrated that both Nrf2 and mNrf2 Therefore, we believe that the di€erence in binding were in vitro transcribed and translated with equal between Nrf2 and mNrf2 in the in vivo band shift was eciencies (Figure 6c). masked due to binding of AP1, and related proteins, to The band shift experiments with NQO1 gene ARE the ARE. and in vitro translated Nrf2, mNrf2 and MafG-V5 revealed binding of MafG-V5-MafG-V5 homodimers and Nrf2-MafG-V5 and mNrf2-MafG-V5 heterodimers Discussion to the ARE (Figure 7). The super shift assays with V5 antibody super shifted the MafG-MafG homo- and Nrf2, a cap`n'collar containing b-zip protein, has a Nrf2-MafG hetrodimeric complexes with the NQO1 clearly established role in the ARE-mediated expres- gene ARE (Figure 7). However, the analysis of the sion, and induction, of NQO1 and other detoxifying shifted and super shifted bands did not show a enzyme genes (Dhakshinamoorthy et al., 2000). Nrf2 di€erence between binding of Nrf2-MafG-V5 and mediated activation of detoxifying enzymes is consid- mNrf2-MafG-V5. ered essential for cellular protection against oxidative The band shift experiments were also performed with stress and other free radical damage (Dhakshina- nuclear extracts from Hep-G2 cells over expressing moorthy et al., 2000). Normally, Nrf2 is retained in Nrf2 or mNrf2 (data not shown). No di€erence was the cytosol by an inhibitor protein, INrf2/Keap1 (Itoh detected. This presumably was because of relatively et al., 1999; Dhakshinamoorthy and Jaiswal, 2001).

Oncogene Redox regulation of Nrf2 D Bloom et al 2194

Figure 2 Nrf2 C506S mutation reduces ARE-mediated NQO1 gene expression. (a) Human hepatoblastoma (Hep-G2) cells were transfected with 0.5 mg of reporter plasmid, NQO1 gene ARE-tk-CAT, alone and in combination with various concentrations of expression plasmids pcDNA (vector alone), pcDNA-Nrf2, or pcDNA-mNrf2. Forty-eight hours after the transfection, the cells were harvested and analysed for CAT activity. The NQO1 gene ARE-mediated CAT activities in Hep-G2 cells over expressing Nrf2 or mNrf2 were compared with CAT activities in Hep-G2 cells expressing endogenous levels of Nrf2. The data are presented as fold- induction in CAT activity due to over expression of di€erent levels of Nrf2 or mNrf2. The data are mean+s.d. of ®ve independent transfection experiments. (b) Hep-G2 cells were transfected with pcDNA-Nrf2-V5 or pcDNA-mNrf2-V5. Forty-eight hours after transfection the cells were harvested and whole cell extracts were made using RIPA bu€er. 100 mg of the extracts were resolved on a 10% PAGE, and Western blots were performed with V5 antibodies and b-actin antibodies

The treatment of cells with antioxidants and xenobio- cysteine residue in c-Jun has been shown to be a site of tics causes Nrf2 to be released from INrf2/Keap1. Nrf2 redox regulation. Mutation of this site led to a then translocates to the nucleus and activates ARE- signi®cant alteration in the expression of c-Jun target mediated expression of NQO1 and other detoxifying genes (Abate et al., 1990). A mutant Nrf2 molecule, enzyme genes. containing a cysteine to serine mutation at position 506 It may be noteworthy that Nrf2 does not bind to the in its DNA binding domain, was generated by site- ARE by itself, but must heterodimerize with another directed mutagenesis. The C506S change in Nrf2 b-zip protein before binding to the ARE. c-Jun has signi®cantly reduced the ability of Nrf2 to activate been shown to heterodimerize with Nrf2 in the ARE-mediated gene expression, and induction in presence of an unknown cytosolic factor(s) and response to antioxidants. In addition, mNrf2 regulation upregulate ARE activity (Venugopal and Jaiswal, of ARE-mediated gene expression demonstrated resis- 1998). Small Maf proteins, including MafG, also form tance to NEM inhibition, as compared to wild-type heterodimers with Nrf2 and bind to the ARE (Itoh et Nrf2. These results demonstrate the importance of al., 1997; Nguyen et al., 2000; Wild et al., 1999; cysteine 506 in redox regulation of Nrf2. This is also Dhakshinamoorthy and Jaiswal, 2000). However, the supported by previous observations that ARE- in vivo role of Nrf2-MafG heterodimers is controver- mediated gene expression is sensitive to sulfhydryl sial; some groups suggest that the heterodimers modifying agents, such as glutathione, and related activate, while others suggest that they repress ARE- molecules (Li and Jaiswal, 1994; Bergelson et al., 1994; mediated gene expression (Nguyen et al., 2000; Wild et Shertzer et al., 1995). al., 1999; Dhakshinamoorthy and Jaiswal, 2000; Itoh et Further experiments demonstrated that the reduced al., 1997; Gong et al., 2001). Clearly, the role of b-zip eciency of mNrf2 to activate ARE-mediated gene proteins that heterodimerize with Nrf2 and participate expression was not due to alterations in its interaction in Nrf2 signaling is not completely understood. with INrf2/Keap1, but due to its reduced binding with An analysis of the DNA binding domain of Nrf2 the ARE. We demonstrated that, like wild-type Nrf2, revealed the presence of a cysteine residue that was mNrf2 interacted with, and was retained in the cytosol conserved across the species, and in other members of by, INrf2/Keap1. Furthermore, mNrf2, like Nrf2, was the Nrf family, including Nrf1 and Nrf3. A similar released from INrf2 in response to antioxidants, and

Oncogene Redox regulation of Nrf2 D Bloom et al 2195

Figure 3 Nrf2C506S mutation diminishes t-BHQ induction of ARE-mediated NQO1 gene expression. Hep-G2 cells were cotransfected with 0.5 mg of reporter plasmid, NQO1 gene ARE-tk-CAT, and 0.15 mg of pcDNA (expression vector alone), pcDNA- Nrf2, or pcDNA-mNrf2. Thirty-six hours after the transfection, the cells were treated with 100 mM or 200 mM t-BHQ for 16 h. The cells were harvested at the end of the treatment and analysed for CAT activity. The results are presented as fold-induction in CAT activity in response to t-BHQ. The data are mean+s.d. of ®ve independent transfection experiments translocated to the nucleus. However, mNrf2-c-Jun activation of ARE-mediated gene expression. This heterodimers bound about threefold less eciently to pathway has to be in addition to the phosphorylation the ARE than wild-type Nrf2-c-Jun, causing the decrease of Nrf2 as reported earlier (Huang et al., 2000). It is in ARE activation. A di€erence in binding between expected that a redox factor(s), like Ref-1, plays a role mNrf2-MafG and wild-type Nrf2-MafG, with the ARE, in the regulation of cysteine 506 (Yao et al., 1994). was not observed. This may be due to structural and However, the role of such factors in the redox functional di€erences between c-Jun and MafG. regulation of Nrf2 remains a topic for further study. In conclusion, the cysteine at position 506 in the DNA binding region of Nrf2 is required for ARE- mediated activation of detoxifying genes, and their induction in response to antioxidants. This cysteine Materials and methods appears to be redox regulated, and is required for ecient binding of Nrf2 to the ARE. The exposure of Plasmid construction/mutagenesis cells to chemicals leads to the generation of electro- The previously constructed LNCX-Nrf2 plasmid (Venugopal philic and oxidative signals (Huang et al., 2000). It is and Jaiswal, 1996) containing the mouse Nrf2 cDNA was possible that these signals lead to the modi®cation of mutated using the TransformerTM Site-Directed Mutagenesis the cysteine in the DNA binding region of Nrf2. This Kit (2nd version) from Clontech. The cysteine at position 506, results in increased binding of Nrf2 to the ARE, and in the DNA binding domain of Nrf2, was mutated to a serine,

Oncogene Redox regulation of Nrf2 D Bloom et al 2196 using the protocol provided by Clontech. Brie¯y described, two Oligos were synthesized. The ®rst, AJ203, was designed to replace the appropriate bases to change the cysteine (TGT) to serine (AGC). The second, AJ204, was designed to remove a unique restriction site, Nru1, in the LNCX plasmid. The LNCX-Nrf2 plasmid was denatured by boiling for 3 min in the presence of AJ203 and AJ204. The mixture was allowed to cool and the new (mutant) strand was synthesized using T4 DNA polymerase on T4 DNA Ligase. Digesting with Nru1 linearized the non-mutated DNA. The mutated plasmids were then chemically transformed into BMH 71-18 mutS (repair- de®cient) E. coli. The cells were grown overnight in 5 ml LB medium containing ampicillin. 1.5 ml of the overnight culture was used to prepare a mini-prep by the boiling-lysis method described in the Clontech manual. Again the plasmid DNA was digested with Nru1 to linearize any unmutated plasmids, and transformed into mutS E coli. The bacteria were grown on LB agar containing ampicillin. Twelve colonies were analysed by restriction digest, and nine appeared to have been successfully mutated. The mutation was con®rmed by sequencing. The wild-type Nrf2 and mutated Nrf2 cDNA's (Nrf2C506S) were then subcloned separately into the pcDNA3.1/V5-HIS TOPO Figure 4 NEM inhibits Nrf2, but not mNrf2, regulation of vector, via PCR and TA cloning to generate pcDNA-Nrf2 and ARE-mediated NQO1 gene expression. Hep-G2 cells were pcDNA-mNrf2 plasmids respectively. These constructs pro- cotransfected with pcDNA (vector alone), pcDNA-Nrf2, or duce proteins with the V5 epitope. Clones were also generated pcDNA-mNrf2. The transfected cells were treated with either that do not have the V5 tag. The V5 epitope contains 14 amino 0.25 mM or 0.5 mM NEM and analysed for CAT activity. The acids in the sequence: Gly-Lys-Pro-Ile-Pro-Asn-Pro-Leu-Leu- data are shown as mean+s.d. of ®ve independent transfection Gly-Leu-Asp-Ser-Thr. The antibodies against V5 are commer- experiments cially available from Invitrogen, California. The construction

Figure 5 Nrf2C506S mutation does not e€ect the interaction of Nrf2 with INrf2/Keap1. Quail ®broblasts (QT6) cells were transfected with EGFP-Nrf2 or EGFP-mNrf2 alone, or in combination with pcDNA-INrf2 plasmid. The transfected cells treated with DMSO or 100 mM t-BHQ for 16 h and analysed for EGFP localization by procedures as described in Materials and methods

Oncogene Redox regulation of Nrf2 D Bloom et al 2197

Figure 6 Nrf2C506S-c-Jun heterodimers bind less eciently to the NQO1 gene ARE than wild-type Nrf2-c-Jun heterodimers. (a) 100 000 c.p.m. of the 32P-labeled NQO1 gene ARE was incubated with in vitro translated Nrf2 or mNrf2 alone or in combination with in vitro translated c-Jun, and band shift analysis performed by procedures as described in Materials and methods. CE alone lane also contained rabbit reticulocyte lysate to normalize with other lanes. (b) The density of the bands was determined using Kodak imaging software. The graph represents data from three independent experiments+s.d. A one tailed student t-test was performed. (c)5ml of the in vitro translated Nrf2 and mNrf2 proteins were resolved on a SDS ± PAGE, treated with Amplify solution to enhance the 35S signal, dried and exposed to X-ray ®lm. The two proteins were translated with equal eciency. CE, cytosolic extract; *non-speci®c reticulocyte band

of the hARE-tk-CAT plasmid has been previously described tagged MafG (MafG-V5) protein upon transfection in (Xie et al., 1995). mammalian cells. Primer Sequences for Site Directed Mutagenesis of Nrf2: AJ203: 5' GCC GCC CAG AAC AGC AGG AAA AGG AAG 3' Cell culture and cotransfection of reporter and expression AJ204: 5' GAT CCC CTC GAG AGT TGG TTC plasmids Mouse wild-type Nrf2 and mNrf2 were subcloned separately at Kpn I/Apa I sites of pEGFP vector to produce Human Hepatoblastoma (Hep-G2) and Quail Fibroblast pEGFP-Nrf2 and pEGFP-mNrf2 respectively. These plas- (QT6) cells were grown in monolayer cultures in six well mids generate EGFP tagged Nrf2 and EGFP tagged INrf2 in dishes, as described (Li and Jaiswal, 1992; Dhakshina- transfected cells. The rat INrf2 cDNA encoding the full- moorthy and Jaiswal, 2001). QT6 cells are known to express length INrf2 protein and the chicken MafG cDNA were lesser amount of INrf2 that retains Nrf2 in the cytosol subcloned in pcDNA vector to generate pcDNA-INrf2 and (Dhakshinamoorthy and Jaiswal, 2001). Therefore, a portion pcDNA-MafG plasmids by described procedures (Dhakshi- of the Nrf2 is always in the nucleus. 0.5 mg of reporter namoorthy and Jaiswal, 2001). The MafG cDNA was also plasmid, hARE-tk-CAT, was mixed with di€erent concentra- subcloned in-frame with the V5 epitope of the pcDNA tions of expression plasmid pcDNA3.1 vector alone or expression plasmid as previously described (Dhakshina- pcDNA-Nrf2, pcDNA-mNrf2, and cotransfected in Hep-G2 moorthy and Jaiswal, 2000). These plasmids encode the V5 cells using the CalPhosTM Mammalian Transfection Kit from

Oncogene Redox regulation of Nrf2 D Bloom et al 2198

Figure 7 Nrf2C506S-c-Jun and Nrf2-c-Jun heterodimers bind with similar eciency to the NQO1 gene ARE. The NQO1 gene ARE was 32P-labeled. 100 000 c.p.m. of labeled ARE was incubated with in vitro translated Nrf2 or mNrf2 alone or in combination with MafG-V5 and band shift analysis performed by procedures as described in Materials and methods. In related experiments, the band shift mixture was incubated with V5 antibodies and band shift experiment was performed. SB, Super shifted band, *non- speci®c reticulocyte band

Clontech by procedures supplied by the manufacturer. 0.5 mg in QT6 cells. This is because QT6 cells are de®cient in INrf2 of Rous sarcoma virus (RSV)-b-galactosidase (b-gal) plasmid and a suitable model to study Nrf2 nuclear localization was added to each transfection to normalize transfection (Dhakshinamoorthy and Jaiswal, 2001). The QT6 cells were eciency. The transfected cells were treated with various grown on Lab-Tek II Chamber Slides (Nalge Nunc concentrations of tert-butylhydroquinone (t-BHQ) and International, Naperville, IL, USA). Half a microgram of N-ethylmalamide (NEM) at 36 h post-transfection. The cells plasmids pEGFP-Nrf2 and pEGFP-mNrf2 were transfected were harvested 48 h after transfection in the reporter lysis either alone or in combination with pcDNA-INrf2 using the bu€er supplied with the CAT enzyme assay system from E€ectene Transfection Reagent Kit by procedures as Promega. The b-galactosidase activity was determined using described in manufacturer's protocol. After 36 h of transfec- the assay system from Promega. The appropriate amounts of tion, the cells were washed twice with 16PBS. Then the cell extract, calculated based on b-galactosidase activity, were chamber was removed from the slide and a cover slip was then used to perform the CAT Assay using Promega's liquid placed over it. The slide was observed under the ZEISS scintillation method. ¯uorescent microscope by excitation at 488 nm, and the The transfection of Hep-G2 cells with pcDNA-Nrf2 and images were captured by described procedures (Chan et al., pcDNA-mNrf2 resulted similar expression of the respective 1999). The pictures were taken using the Kodak select black proteins as determined by Western analysis (Figure 2B). and white ®lm on a Nikon UFX camera. For induction Whole cell extracts were prepared using RIPA bu€er, and studies, the cells were treated with 50 mM t-BHQ dissolved in 100 mg of the extracts were run on SDS ± PAGE. Western DMSO for 16 h. The cells treated with the vehicle (DMSO) blots were then performed with the V5 antibody, and b-actin alone were included as control in all induction experiments antibody as a control. NEM and t-BHQ were obtained from Sigma, St. Louis, MO, USA. b-actin antibodies were In vitro transcription/translation purchased from CalBiochem. The in vitro transcription/translation of the plasmids encoding Nrf2, mNrf2 (Nrf2C506S), c-Jun and MafG were Nuclear localization performed using the TNT coupled rabbit reticulocyte lysate The nuclear localization studies of the EGFP-tagged Nrf2 systems (Promega) by procedures as suggested in the and EGFP-tagged mNrf2 protein molecules were performed manufacturer's protocol. Redivue L-[35S] Methionine was

Oncogene Redox regulation of Nrf2 D Bloom et al 2199 substituted for methionine in the reactions. After the coupled was incubated with nuclear extracts or in vitro translated transcription/translation, the proteins were resolved using proteins. Band and super shift assays were then performed by SDS ± PAGE. Brie¯y, 5 ml of the translated proteins were previously described procedures (Venugopal and Jaiswal, resolved on a 10% gel, treated with Amplify solution (NAMP 1998; Dhakshinamoorthy and Jaiswal, 2000). Ten micro- 100, Amersham Pharmacia) to enhance the 35S signal, dried grams of the nuclear extract or equimolar concentrations of and exposed to X-ray ®lm. in vitro translated proteins were used in the gel shift and super shift experiments. 1.5 ml of anti-V5 antibody, 3 mlof Nrf2 antibody or pre-immune serum, were used in the super Preparation of nuclear and cytosolic extract shift assays. Densitometry anylysis was done using Kodak 1D The nuclear extract from human hepatoblastoma (Hep-G2) image analysis software version 3.5. cells over expressing Nrf2 and mNrf2 were prepared by procedures as described (Venugopal and Jaiswal, 1998). To prepare cytosolic extract, 106100 mm dishes of Hepal cells were grown to 80% con¯uency. The cells were washed with Abbreviations ice cold PBS without calcium or magnesium, then each plate NQO1, First isoform of cytosolic NAD(P)H:quinone was scraped into 10 ml PBS. The cells were pelletted and oxidoreductase1; ARE, antioxidant response element; t- resuspended in 10 volumes of isotonic bu€er (0.15 M NaCl, BHQ, tert-butyl hydroquinone; CAT, chloramphenicol acetyl transferase. 1.5 mM MgCl2,10mM Tris-HCl pH 7.4, NP-40 0.5% (v/v), 1mM DTT, and 1 mM PMSF. DTT and PMSF were added just before use). The cells were dounced 10 times using a loose pestle, and spun at 3500 r.p.m. for 10 min to remove cell debris. The supernatant was spun again for 1 h at 14 000 r.p.m., and then dialyzed against the DNA binding bu€er (20 mM HEPES pH 7.9, 50 mM KCl, 1 mM EDTA, 1mM DTT, 1 mM PMSF, and 10% glycerol. DTT and PMSF were added just before use).

Acknowledgments Band and super shift assays We are grateful to Drs Je€erson Y Chan and Yuet W Kan, The NQO1 gene ARE (Nucleotide Sequence 5'-CAG TCA both from University of California, San Francisco for CAG TGA CTC AGC AGA ATC T-3') was end labeled with providing us the cDNA encoding Nrf2. This investigation gP32 ATP and T4 polynucleotide kinase. The labeled ARE was supported by NIH grant GM47466.

References

Abate C, Patel L, Rauscher III FJ and Curran T. (1990). ItohK,WakabayashiN,KatohY,ishiiT,IgarashiK,Engel Science, 249, 1157 ± 1161. JD and Yamamoto M. (1999). Genes Dev., 13, 76 ± 86. Alam J, Stewart D, Touchard C, Boinapally S, Choi MK Jaiswal AK. (1994). Biochem. Pharm., 48, 439 ± 444. and Cook JL. (1999). J. Biol. Chem., 274, 26071 ± Jayepaul J and Jaiswal AK. (2000). Biochem. Pharm., 59, 26078. 1433 ± 1439. Bergelson S, Pinkus R and Daniel V. (1994). Cancer Res., 54, Joseph P, Xie T, Xu Y and Jaiswal AK. (1994). Oncol. Res., 36 ± 40. 6, 525 ± 532. Breen AP and Murphy JA. (1995). Free Rad. Biol. Med., 18, Li Y and Jaiswal AK. (1992). J. Biol. Chem., 267, 15097 ± 1033 ± 1077. 15104. Chan PK, Bloom DA and Hoang TT. (1999). Biochem. Li Y and Jaiswal AK. (1994). Eur. J. Biochem., 226, 31 ± 39. Biophys. Res. Commun., 264, 305 ± 309. Mulkahy RT, Wartman MA, Bailey HH and Gipp JJ. (1997). Choi AM and Alam J. (1996). Am. J. Respir. Cell. Mol. Biol., J. Biol. Chem., 272, 7445 ± 7454. 15, 9±19. Nguyen T, Huang HC and Pickett CB. (2000). J. Biol. Chem., Dhakshinamoorthy S and Jaiswal AK. (2001). Oncogene, 20, 275, 15466 ± 15473. 3906 ± 3917. Oesch F, Gath I, Igarashi T, Glatt HR and Thomas H. Dhakshinamoorthy S and Jaiswal AK. (2000). J. Biol. (1991). Molecular aspects of monooxygenases and bioacti- Chem., 275, 40134 ± 40141. vation of toxic compounds. Arinc E, Schenkman JB and Dhakshinamoorthy S, Long II DJ and Jaiswal AK. (2000). Hodgson E (eds). New York: Plenum Press, pp. 447 ± 461. Current Topics in Cellular Regulation, 36, 201 ± 216. Pickett CB and Lu AYH. (1989). Ann. Rev. Biochem., 58, Ernster L. (1987). Chem. Scripta, 27A, 1 ± 13. 743 ± 764. Gong P, Hu B, Stewart D, Ellerbe M, Figueroa YG, Blank V, Radjendirane V, Joseph P and Jaiswal AK. (1997). Oxidative Beckman BS and Alam J. (2001). J. Biol. Chem., 276, Stress and Signal Transduction. Cadenas E and Forman 27018 ± 27025. HJ (eds). New York: Chapman and Hall, pp. 441 ± 475. Grisham MB and McCord JM. (1986). Physiology of Oxygen Riley RJ and Workman P. (1992). Biochem. Pharm., 43, Radicals, pp. 1 ± 18, Waverly Press, Baltimore. 1657 ± 1669. Huang HC, Nguyen T and Pickett CB. (2000). Proc. Natl. Rosen GM, Pou S, Ramos CL, Cohen MS and Britigan BE. Acad. Sci. USA, 98, 379 ± 379. (1995). FASEB J., 9, 200 ± 209. Itoh K, Chiba T, Takahashi S, Ishii T, Igarashi K, Katoh K, Rushmore TH and Pickett CB. (1993). J. Biol. Chem., 268, Oyake T, Hayashi N, Satoh K, Hatayama I, Yamamoto M 11475 ± 11478. and Nabeshima Y. (1997). Biochem. Biophys. Res. Shertzer HG, Vasiliou V, Liu RM, Tabor MW and Nebert Commun., 236, 313 ± 322. DW. (1995). Chem. Res. Toxicol., 8, 431 ± 436.

Oncogene Redox regulation of Nrf2 D Bloom et al 2200 Talalay P, Fahey JW, Holtzclaw WD, Prestera T and Zhang Ward JF. (1994). Int. J. Rad. Biol., 66, 427 ± 432. Y. (1995). Toxicol. Lett., 82 ± 83, 173 ± 179. Wasserman W and Fahl WE. (1997). Proc. Natl. Acad. Sci. Tephly T and Burchell B. (1990). Trends in Pharm., 11, 276 ± USA, 94, 5361 ± 5366. 279. Wild AC, Moinova HR and Mulcahy RT. (1999). J. Biol. Thelen M, Dewald B and Baggiolini M. (1993). Physiol. Rev., Chem., 274, 33627 ± 33636. 73, 797 ± 821. Xie T, Belinsky M, Xu Y and Jaiswal AK. (1995). J. Biol. Tsuchida S and Sato K. (1992). Critical Reviews in Chem., 270, 6894 ± 6900. Biochemistry and Molecular Biology, 27, 337 ± 384. Yao KS, Xanthoudakis S, Curran T, O'Dwyer PJ. (1994). Venugopal R and Jaiswal AK. (1998). Oncogene, 17, 3145 ± Mol. Cell. Biol., 14, 5997 ± 6003. 3156. Venugopal R and Jaiswal AK. (1996). Proc. Natl. Acad. Sci. USA, 93, 14960 ± 14965.

Oncogene