Nuclear Import and Export Signals in Control of Nrf2*

Nrf2 binds to the antioxidant response element and regulates expression and antioxidant induction of a battery of chemopreventive genes. In this study, we have identified nuclear import and export signals of Nrf2 and show that the nuclear import and export of Nrf2 is regulated by antioxidants. We demonstrate that Nrf2 contains a bipartite nuclear localization signal (NLS) and a leucine-rich nuclear export signal, which regulate Nrf2 shuttling in and out of the nucleus. Immunofluorescence and immunoblot analysis revealed that Nrf2 accumulates in the nucleus within 15 min of antioxidant treatment and is exported out of nucleus by 8 h after treatment. Nrf2 mutant lacking the NLS failed to enter the nucleus and displayed diminished expression and induction of the downstream NAD(P)H:quinone oxidoreductase 1 gene. The Nrf2 NLS sequence, when fused to green fluorescence protein, resulted in the nuclear accumulation of green fluorescence protein, indicating that this signal sequence was sufficient to direct nuclear localization of Nrf2. A nuclear export signal (NES) was characterized in the C terminus of Nrf2, the deletion of which caused Nrf2 to accumulate predominantly in the nucleus. The Nrf2 NES was sensitive to leptomycin B and could function as an independent export signal when fused to a heterologous protein. Further studies demonstrate that NES-mediated nuclear export of Nrf2 is required for degradation of Nrf2 in the cytosol. These results led to the conclusion that Nrf2 localization between cytosol and nucleus is controlled by both nuclear import and export of Nrf2, and the overall distribution of Nrf2 is probably the result from a balance between these two processes. Antioxidants change this balance in favor of nuclear accumulation of Nrf2, leading to activation of chemopreventive proteins. Once this is achieved, Nrf2 exits the nucleus for binding to INrf2 and degradation.

ventive genes, and regulate ARE-mediated expression and induction in response to a variety of stimuli including antioxidants (1)(2)(3). Nrf2 is most potent among the three protein factors in regulation of basal and induced expression of antioxidant enzyme genes (1). The studies have provided clear evidence that Nrf2/ARE-mediated coordinated expression and induction is a mechanism of critical importance in cellular protection against oxidative stress and neoplasia (1)(2)(3). Mice lacking the Nrf2 gene exhibited a marked decrease in the expression and induction of antioxidant enzyme genes, including NAD(P)H:quinone oxidoreductase 1 (NQO1) and increased sensitivity to chemically induced neoplasia (4,5).
INrf2 (inhibitor of Nrf2) or KEAP1 (Kelch-like ECH-associated protein 1) retains Nrf2 in the cytoplasm (1,6,7). INrf2 leads to the proteosomal degradation of Nrf2 in the normal cell cytoplasm (8 -13). The exposure to antioxidants leads to dissociation of Nrf2 from INrf2. Nrf2 is stabilized and translocates into the nucleus. Nuclear translocation of Nrf2 leads to coordinated activation of expression of a battery of chemopreventive genes (1). INrf2 appears to be a specific cytosolic inhibitor for Nrf2, because it does not interact with Nrf1 or Nrf3. 2 The signal transduction pathway from antioxidants to the INrf2-Nrf2 complex leading to the release of Nrf2 from INrf2 has been extensively studied but remains largely unknown (1). The cytosolic factor(s) that catalyze antioxidant-induced modifications of Nrf2 and/or INrf2 remain largely uncharacterized. However, this is an active area of significant research, and many studies have been published that have investigated the antioxidant-induced post-translational modifications of Nrf2 and/or INrf2 leading to the release of Nrf2 from INrf2 (1). Several cytosolic kinases that include protein kinase C, mitogen-activated protein kinase, p38, and phosphatidylinositol 3-kinase have been shown to modify Nrf2 and participate in the mechanism of signal transduction from antioxidants and xenobiotics to the ARE (12, 14 -23). Other studies have demonstrated binding of inducers to sulfhydryl groups of cysteines in INrf2 that leads to the loss of the association of Nrf2 with INrf2, leading to the activation of ARE-mediated gene expression (13,24,25). Disruption of INrf2 in mice leads to postnatal death, probably from malnutrition resulting from hyperkeratosis in the esophagus and forestomach, presumably due to accumulation of Nrf2 in the nucleus (26). Interestingly, double knockout mice deficient in both INrf2 and Nrf2 were born normal and survived (27). These studies revealed that increased concentration of Nrf2 in the nucleus leads to adverse effects on cell growth and survival.
The studies as described above clearly establish a major function of Nrf2 in coordinated induction of defensive proteins for cellular protection. Given Nrf2 ability to function as transcription factor, one would predict a tight correlation between Nrf2 availability in the nucleus and activation of defensive proteins. Mechanisms that regulate the levels and activity of Nrf2 in the nucleus have not been fully characterized. The nuclear localization signal that regulates nuclear translocation of Nrf2 is uncharacterized. In addition, the nuclear export of Nrf2 and the nuclear export signal in Nrf2 remain unknown. Furthermore, the antioxidant regulation of import and export signals of Nrf2 also remains unknown.
The purpose of the current study was to examine the nuclear import and export of Nrf2. Toward this end, a bipartite nuclear localization signal (NLS) was identified in the C terminus of Nrf2, which is required for Nrf2 nuclear import and which promoted nuclear import when fused to a heterologous green fluorescence protein (GFP). Deletion of the NLS led to the loss of nuclear localization of Nrf2 and diminished transcriptional activity of Nrf2 on downstream genes, including NQO1 in transfected cells, as compared with wild type Nrf2. A nuclear export signal (NES) was also identified in the C terminus of Nrf2, the deletion of which caused accumulation of Nrf2 in the nucleus. This NES could function as an independent export signal when fused to a heterologous protein and was sensitive to leptomycin B. Antioxidant treatment led to nuclear accumulation of endogenous and transfected wild type Nrf2 but not an NLS mutant of Nrf2 within 15 min. Nrf2 but not an NES mutant exited the nucleus by 8 h after antioxidant treatment. This exiting of Nrf2 was sensitive to leptomycin B. Further studies showed that NES-mediated nuclear export of Nrf2 is required for degradation of Nrf2 in the cytosol. These results led to the conclusion that both nuclear import and export of Nrf2 play a significant role in basal expression and antioxidant induction of a battery of chemopreventive proteins, crucial for cellular growth and survival.

EXPERIMENTAL PROCEDURES
Plasmid DNAs-The construction of pGL2-NQO1-ARE and pcDNA-Nrf2 was previously described (28). The pcDNA-Nrf2 was used as a template to construct several deletion mutants of Nrf2 as shown in Fig. 1. The forward 5Ј-GCAGGACATGGATTTGATTGACATCC-3Ј and reverse 5Ј-GTTTTTCTTTGTATCTGGCTTCTTG-3Ј primers were used to amplify the Nrf2 coding region without a stop codon. The PCRamplified product was TA-cloned in pcDNA3.1/V5-His TOPO to generate plasmid pcDNA-Nrf2-V5 using an expression kit (Invitrogen). pcDNA-Nrf2⌬NLS-V5 was constructed by PCR-amplifying the fragments just before and after the NLS region using the primers forward (5Ј-GCAGGACATGGATTTGATTGACATCC-3Ј) and reverse (5Ј-TATA-TCTCGGATCAATGCGAGCTGAGC-3Ј) for the N terminus fragment and forward (5Ј-GAGAACATTGTCGAGCTGGAGCAAGAC-3Ј) and reverse (5Ј-GTTTTTCTTTGTATCTGGCTTCTTG-3Ј) for the C terminus fragment, respectively. The amplified fragments were ligated and TAcloned into the pcDNA3.1/V5-His vector following the protocol as suggested by the manufacturer. pcDNA-Nrf2⌬NES-V5 was constructed by PCR amplifying the fragments just before and after the NES region using the primers forward (5Ј-GCAGGACATGGATTTGATTGACATC-C-3Ј) and reverse (5Ј-TAGATGGAGGTTTCTGTCGTTTTCTCC-3Ј) for the N terminus fragment and forward (5Ј-GAAGTCTTCAGCATGTTA-CGTGATGAG-3Ј) and reverse (5Ј-GTTTTTCTTTGTATCTGGCTTCTT-G-3Ј) for the C terminus fragment, respectively. The amplified fragments were ligated and TA-cloned into the pcDNA3.1/V5-His vector as described above. For constructing the ⌬Neh2 mutants of pcDNA-Nrf2-V5 and pcDNA-Nrf2⌬NLS-V5, these plasmids were used as template for PCR amplification to amplify respective fragments using forward (5Ј-ACCATGGCCCAGCATATCCAGACAGAC-3Ј) and reverse (5Ј-TATATCTCGGATCAATGCGAGCTGAGC-3Ј) primers followed by cloning into pcDNA 3.1/V5-His TOPO vector. All of the plasmids were confirmed by sequencing and expressed a V5 epitope tag at their C terminus.
GFP, an NLS Fusion Protein-A DNA fragment corresponding to amino acids 494 -511 of the nuclear localization signal (NLS) of Nrf2 was made by PCR. The primers used for PCR were forward (5Ј-TACG-CAGGAGAGGTAAGAATAAAGTCGC-3Ј) and reverse (5Ј-CTCCAGCT-TCCTTTTCCTACAGTTCTG-3Ј), respectively. The resulting PCR product was cloned into pcDNA3.1 TOPO vector (Invitrogen). The plasmid was then digested with KpnI and XbaI, and the digested fragment was subcloned into EGFP-C1 vector (Clontech). The construct was named GFP-NLS, having GFP at the N terminus of the NLS.

Two Yellow Fluorescent Protein (2YFP), an NES Fusion Protein-
The p2YFP vector was a gift from Dr. Yanping Zhang (M.D. Anderson Cancer Center, Houston, TX). A DNA fragment corresponding to amino acids 545-554 of the nuclear export signal (NES) of Nrf2 was made using the oligonucleotides, which were then annealed. The resulting fragment had NheI and AgeI overhangs and was cloned into the corresponding sites in p2YFP vector. The construct was confirmed by sequencing and named p2YFP-NES.
Cell Culture, Co-transfection of Expression Plasmids, and Luciferase Reporter Assay-Human hepatoma (HepG2) cells were grown in monolayer cultures in 6-well plates in minimum essential medium-␣ supplemented with 10% fetal bovine serum, both from Invitrogen. Transient transfections were done in cells grown to ϳ50% confluence using the Effectene Transfection reagent (Qiagen, Valencia, CA) following the manufacturer's protocol. Cells were co-transfected with 0.2 g of reporter construct (human NQO1-ARE-Luc) mixed with pcDNA and different deletion constructs of Nrf2 (pcDNA-Nrf2-V5, pcDNA-⌬Neh2Nrf2-V5, pcDNA-Nrf2⌬NLS-V5, and pcDNA-⌬Neh2⌬NLS-Nrf2-V5) in the quantities mentioned in the figures. The plasmid pRL-TK encoding Renilla luciferase was used as the internal control in each transfection. 36 h after transfection, the cells were induced with 50 M t-butylhydroquinone (t-BHQ) dissolved in Me 2 SO for 16 h. One set of transfected cells were treated with Me 2 SO for the same period of time and was used as vehicle control. For studying the effect of leptomycin B on ARE activity for different time points, 36 h after transfection with reporter plasmid, cells were pretreated for 8 h with 20 ng/ml leptomycin B from Sigma. After 8 h of pretreatment, cells were then treated with either vehicle control Me 2 SO or 50 M t-BHQ with or without leptomycin B for the indicated time periods. After treatment for the specified time, the cells were washed with 1ϫ phosphate-buffered saline and lysed in 1ϫ passive lysis buffer from the Dual-Luciferase Reporter Assay System Kit (Promega, Madison, WI). The luciferase activity was measured using the procedures described previously (28). Each set of transfection experiments was repeated three times. 100 g of the lysates from transfected cells were resolved in 10% SDS-PAGE and Western blotted. Western blots were probed with anti-Nrf2 antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), anti-V5-horseradish peroxidase antibody (Invitrogen), and anti-␤actin antibody (Sigma) to show the overexpression of these proteins.
NQO1 Protein and Activity-HepG2 cells were grown in 100-mm tissue culture plates and were transfected with 2.0 g of pcDNA or pcDNA-Nrf2 wild type or deletion mutant plasmids using Effectene transfection reagent following the procedures described above. 36 h after transfection, the cells were treated with 50 M t-BHQ for 16 h, and cells were also treated with Me 2 SO as vehicle control. The cells were then harvested, cytosolic extracts were prepared, and NQO1 activity was assayed following the procedures previously described (29). 100 g of cytosolic extracts from the transfected and t-BHQ-treated HepG2 cells were resolved on 12% SDS-PAGE, Western blotted, and probed with antibodies against NQO1 developed in our laboratory (29) and ␤-actin (Sigma).
Immunofluorescence-Mouse hepatoma (Hepa-1) cells were grown in Lab-Tek II chamber slides in Dulbecco's modified essential medium supplemented with 10% fetal bovine serum, both from Invitrogen. Cells were transfected with the desired plasmid using the procedures described above. For studying the localization of endogenous Nrf2, the cells were treated with 50 M t-BHQ for 15 min and for 1, 2, 4, 6, and 8 h. The transfected cells were treated with t-BHQ in the presence or absence of 20 ng/ml leptomycin B (Sigma) for 8 h. Cells were then fixed in formalin (Polysciences, Inc., Warrington, PA) and permeabilized with cold acetone (Fisher). The antibody used for immunostaining the V5tagged protein was anti-V5-FITC (Invitrogen), and for visualizing the endogenous protein, the cells were probed with Nrf2 antibody, (Santa Cruz Biotechnology). The fluorescein isothiocynate (FITC)-conjugated anti-rabbit antibody (Chemicon International, Temecula, CA) was used as secondary antibody by procedures previously described (29). To visualize the nuclei, the cells were stained with Hoechst stain (Bio-Rad). The fluorescent images were captured using appropriate filters in a Nikon eclipse TE 2000-U fluorescent microscope fitted with a Photometrics CoolSnap CF camera, and images were enhanced using Adobe Photo-Deluxe software. In some experiments, the localization of Nrf2 and mutant Nrf2 was also determined by SDS-PAGE, Western blotting, and probing with Nrf2 antibody as described (12).
Subcellular Fractionation and Western Analysis-Hepa-1 cells were grown in 100-mm tissue culture plates, and after reaching 50% confluence, these cells were transfected with 2.0 g of pcDNA-Nrf2-V5, pcDNA-Nrf2⌬NLS-V5, or pcDNA-Nrf2⌬NES-V5 plasmids using the Effectene transfection reagent as described above. Twenty-four hours after transfection, the cells were treated with Me 2 So, t-BHQ, or leptomycin B as indicated in the figures. At the end of treatment, cells were washed twice with ice-cold phosphate-buffered saline, scraped in phosphate-buffered saline using a rubber policeman and centrifuged at 500 rpm for 5 min. Biochemical fractionation of the cells was done using the nuclear extract kit (Active Motif, Carlsbad, CA) following the manufacturer's protocol. Briefly, the cell pellet was resuspended in 1ϫ hypotonic buffer (cytoplasmic buffer) supplemented with complete protease inhibitor mixture (Roche Applied Science), incubated for 15 min at 4°C, vortexed in the presence of detergents, and centrifuged briefly. The supernatant (cytoplasmic fraction) was collected into a prechilled microcentrifuge tube; remaining is the nuclear pellet. The nuclear pellet was washed twice with the cytoplasmic buffer followed by resuspending in the lysis buffer supplemented with 1 mM dithiothreitol and protease inhibitors. The suspension was incubated on a rocking platform at 4°C for 30 min. The suspension was vortexed briefly and centrifuged for 10 min at 14,000 ϫ g at 4°C. The supernatant (nuclear fraction) was collected. The protein concentration was determined using the protein assay reagent (Bio-Rad). 100 g of the cell fractions were resolved on a 10% SDS-PAGE, Western blotted, and probed with anti-V5-horseradish peroxidase antibody (Invitrogen) and ␤-actin antibody (Sigma). To confirm the purity of subcellular fractionation, the extracts were immunoblotted with cytoplasm specific anti-lactate dehydrogenase antibody (Chemicon International, Temecula, CA) and nucleus-specific antilamin B antibody (Santa Cruz Biotechnology).
In Vitro Interaction of Nrf2/Mutant Nrf2 with INrf2-The in vitro transcription/translation of the plasmids encoding Nrf2-V5, Nrf2⌬NES-V5, and INrf2 was performed using the TNT-coupled rabbit reticulocyte lysate system (Promega, Madison, WI) by procedures as suggested in the manufacturer's protocol. Redivue L-[ 35 S]methionine (Amersham Biosciences) was substituted for methionine in the reactions. After the coupled transcription/translation, the proteins were checked for their correct size on 10% SDS-PAGE and Western analysis. Five microliters of the translated proteins were resolved on a 10% SDS-PAGE and analyzed. All of the in vitro transcribed/translated proteins gave bands of the expected size. For the binding assay, 5 l each of in vitro translated protein (Nrf2V5 ϩ INrf2 or Nrf2⌬NES ϩ INrf2) were mixed in protein binding buffer (20 mM Tris, pH 7.5, 50 mM NaCl, 10% glycerol, 0.5% Nonidet P-40, 20 mM sodium vanadate) supplemented with protease inhibitors) and incubated at 37°C for 30 min. The mixture was immunoprecipitated with 2.5 g of either mouse IgG or anti-V5 antibody. Immune complexes were washed twice with protein-binding buffer, boiled with 1ϫ SDS sample dye, and resolved on 10% SDS-PAGE. The gel was treated with Amplify solution (NAMP 100; Amersham Biosciences) to enhance the 35 S signal, dried, and autoradiographed.
In Vivo Interaction of Nrf2/Mutant Nrf2 with INrf2-Hepa-1 cells were grown in 100-mm tissue culture plates and co-transfected with 2.0 g of either pcDNA-Nrf2-V5 or pcDNA-Nrf2⌬NES-V5 plasmid and 0.5 g of FLAG-INrf2 as described above. Twenty-four hours after transfection, the cells were harvested and lysed in radioimmune precipitation buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, and 20 mM sodium vanadate supplemented with protease inhibitors). 0.5 mg of cell lysates were used for immunoprecipitation with mouse IgG or anti-V5 antibody (Invitrogen). The extract was incubated with 2.5 g of antibody overnight at 4°C with shaking. Forty microliters of washed Protein A beads (Santa Cruz Biotechnology) were added and incubated for 1 h at 4°C with shaking. Immune complexes were washed twice with radioimmune precipitation buffer, boiled with 1ϫ SDS sample dye, and resolved on 10% SDS-PAGE followed by immunoblotting with anti-V5-horseradish peroxidase and anti-FLAG-horseradish peroxidase (Sigma) antibodies.
Degradation of Nrf2 and Mutant Nrf2-Hepa-1 cells were grown in 100-mm tissue culture plates and transfected with 2.0 g of either pcDNA-Nrf2-V5 or pcDNA-Nrf2⌬NES-V5 plasmid. Twenty four hours after transfection, the cells were pretreated with either Me 2 SO or MG132 (20 M) for 8 h and with 20 ng/ml leptomycin B (LMB) wherever indicated. Cells were washed twice with medium and treated with 30 g/ml cycloheximide for different time points (0.5, 1, 2, or 3 h). One set of the cells was left treated with MG132 alone. After completion of the treatments for the indicated time points, the cells were washed twice with ice-cold 1ϫ phosphate-buffered saline. The cells were harvested in phosphate-buffered saline, whole cell lysate was prepared in radioimmune precipitation buffer, and biochemical fractionation of cytosol and nuclei was done as described above. 100 g of the whole cell lysate or cytosolic fraction and 50 g of nuclear fractions were resolved on a 10% SDS-PAGE, Western blotted, and probed with anti-V5, anti-lactate dehydrogenase, anti-lamin B, and anti-␤-actin antibodies.

RESULTS
We generated several V5-tagged deletion mutants of Nrf2 to identify and characterize nuclear import and export signals of Nrf2 and show that the nuclear import and export of Nrf2 is regulated by antioxidants. We further demonstrate that NLSmediated nuclear import is required for activation of AREmediated gene expression. We also demonstrate that NESmediated nuclear export is required for rapid degradation of Nrf2 in the cytosol. The structures of wild type Nrf2 and Nrf2 deletion mutants used in the present study are shown in Fig. 1. Prosite search of Nrf2 amino acid sequence tentatively identified a single copy each of the NLS and NES region at the C terminus of the protein (Fig. 1). The NLS was found located between amino acids 494 and 511, and the NES was between amino acids 545 and 554.
Hep-G2 cells were co-transfected with reporter plasmid NQO1 gene ARE-Luc and different concentrations of expression plasmids encoding Nrf2 or Nrf2-V5 and Nrf2⌬NLS-V5 as shown in Fig. 2  the overexpression of Nrf2, Nrf2-V5, and Nrf2⌬NLS-V5 in transfected cells (Fig. 2, A-C). The transfected cells were also analyzed for ARE-mediated luciferase gene expression (Fig. 2, D-F). The results showed plasmid concentration-dependent overexpression of Nrf2, Nrf2-V5, and Nrf2⌬NLS-V5 in transfected cells (Fig. 2, A-C). The results also showed a plasmid concentration-dependent increase in ARE-mediated luciferase gene expression and induction in response to t-BHQ in cells overexpressing Nrf2 and Nrf2-V5 (Fig. 2, D and E). These results demonstrated that increases in expression of Nrf2 and Nrf2-V5 both up-regulated ARE-mediated gene expression and induction in response to antioxidant (Fig. 2, D and E). The results also revealed that the addition of the V5 tag to the C terminus had virtually no effect on Nrf2 regulation of AREmediated gene expression (Fig. 2, compare D and E). Interestingly, the deletion of the NLS region from Nrf2 failed to increase basal and/or antioxidant-induced expression of AREmediated gene expression with an increase in overexpression of Nrf2⌬NLS-V5 in transfected cells (Fig. 2, compare F with D and E). However, the transfection with 0.5 g of Nrf2⌬NLS-V5 showed insignificant but persistent increase in the basal and antioxidant-induced ARE activity (Fig. 2F, last panel). This increase in ARE activity is presumably a result of the binding of Nrf2⌬NLS-V5 with INrf2 and titration of endogenous Nrf2. Nrf2⌬NLS-V5 has the INrf2-binding Neh2 domain and is expected to bind to INrf2. As a result, some of the endogenous Nrf2 is titrated out and translocates to the nucleus, leading to ARE activation. To explore this possibility, we generated Nrf2 and Nrf2⌬NLS-V5 mutants lacking the Neh2 domain that binds with INrf2. These were designated as Nrf2⌬Neh2-V5 and Nrf2⌬Neh2⌬NLS-V5 (Fig. 1). These plasmids were transfected in Hep-G2 cells to determine the role of Neh2 and NLS domains in the Nrf2 regulation of ARE-mediated luciferase and endogenous NQO1 gene expression and induction in response to t-BHQ. Western analysis with V5 antibody revealed overexpression of Nrf2 and Nrf2 deletions in transfected cells (Fig.  3A). Overexpression of wild type Nrf2 showed a significant increase in basal and t-BHQ-induced expression of ARE-mediated luciferase gene expression and endogenous NQO1 activity and protein (compare lanes 1 and 2 in Fig. 3, B-D). Deletion of the Neh2 INrf2-binding domain from Nrf2 also led to an Nrf2regulated increase in basal and t-BHQ-induced expression of transfected luciferase and endogenous NQO1 genes (compare lanes 1 and 3 in Fig. 3, B-D). However, the deletion of the NLS domain from Nrf2 led to significant loss in basal and induced expression as compared with wild type Nrf2 (compare lanes 2 and 4 in Fig. 3, B-D). Interestingly, the deletion of both Neh2 and NLS domains from Nrf2 resulted in the complete loss of capacity of Nrf2 to activate ARE-mediated luciferase and endogenous NQO1 gene expression and induction in response to t-BHQ (compare lanes 1 and 5 in Fig. 3 The Nrf2 NLS (amino acids 494 -511) was aligned with previously characterized NLS from nucleoplasmin, p73, and p53 (Fig. 4A). Results revealed that Nrf2 NLS was highly homologous to NLS from nucleoplasmin, p73, and p53. Nrf2 NLS showed conservation of basic amino acid clusters and was bipartite as observed with nucleoplasmin, p73, and p53 NLS. Hepa-1 cells were transfected with Nrf2-V5 or Nrf2⌬NLS-V5 to analyze the subcellular localization of Nrf2 with or without the NLS. The transfected Hepa-1 cells were biochemically fractionated into nuclear and cytoplasmic fractions. The two fractions were then immunoblotted with anti-V5 antibody. Western analysis revealed that in Me 2 SO-treated cells, wild type Nrf2-V5 was present both in cytoplasm and nucleus (Fig. 4B), whereas the Nrf2⌬NLS-V5 was only present in the cytoplasm (Fig. 4B). The Nrf2⌬NLS-V5 was absent in nuclear fraction. 4B), whereas Nrf2⌬NLS failed to localize to the nucleus even after t-BHQ treatment (Fig. 4B). The same blots were reprobed with specific cytosolic and nuclear markers to show the purity of fractionation. Immunofluorescence assays with FITC-tagged V5 antibodies showed similar results as observed in Western analysis (Fig. 4C). Immunofluorescence was detected in cytosolic and nuclear compartments of Me 2 SO-treated Hepa-1 cells transfected with Nrf2-V5 but only in the cytoplasmic compartment in Hepa-1 cells transfected with NLS-deficient Nrf2⌬NLS-V5 mutant. The localization of Nrf2⌬NLS-V5 in the nuclear compartment was absent. The treatment of transfected Hepa-1 cells with t-BHQ led to nuclear localization of Nrf2-V5 but not Nrf2⌬NLS-V5. The Nrf2 NLS domain was synthesized and cloned in frame with GFP to generate plasmid pGFP-NLS. Immunohistochemical analysis of Hepa-1 cells transfected with pGFP alone showed localization of GFP in cytosol and nucleus. In contrast, similar analysis of Hepa-1 cells transfected with pGFP-NLS demonstrated the presence of most of the GFP-NLS in the nucleus. The GFP-NLS was almost absent in cytosol (Fig. 4D).
Immunofluorescence analysis of endogenous Nrf2 protein in Hepa-1 cells with Nrf2 antibody showed localization of Nrf2 in the cytosol as well as in the nucleus (Fig. 5A, Me 2 SO panel). The treatment of Hepa-1 cells with t-BHQ led to nuclear localization of Nrf2 within 15 min of t-BHQ treatment. Interestingly, Nrf2 started exiting nucleus between 1 and 4 h after treatment and achieved normal localization status at 8 h after treatment. This revealed that Nrf2 might contain an NES in addition to the NLS. Indeed, the analysis of the Nrf2 amino acid sequence demonstrated the presence of a leucine-rich NES in Nrf2 similar to what has been previously characterized in several proteins, including IkB␣, TFIIIA, hDM2, p53, p73, and PKI-␣ (Fig. 5B).
We used nuclear export inhibitor leptomycin B to determine the role of NES in nuclear export and localization of endogenous Nrf2. Hepa-1 cells were treated with Me 2 SO and t-BHQ in the absence or presence of leptomycin B and probed with anti-Nrf2 followed by FITC-tagged secondary antibody to determine localization of endogenous Nrf2 protein in the absence and presence of leptomycin B. Endogenous Nrf2 was found localized both in cytoplasm and nucleus in the absence of leptomycin B (Fig. 5C, ϪLMB). However, the treatment of cells with leptomycin B led to nuclear accumulation of Nrf2 in Me 2 SO-treated cells (Fig. 5C, ϩLMB). The treatment of Hepa-1 cells with t-BHQ for 8 h showed localization of Nrf2 both in cytoplasm and nucleus (Fig. 5D, ϪLMB). However, in the presence of leptomycin B, Nrf2 was predominantly localized to the nucleus (Fig. 5D, ϩLMB). These results clearly demonstrated that leptomycin B blocked nuclear export of endogenous Nrf2 in Me 2 SO-treated as well as t-BHQ-treated Hepa-1 cells. The results also suggested that Nrf2 is actively exported out from the nucleus in a leptomycin B-sensitive manner.
We generated a mutant Nrf2⌬NES-V5 deficient in NES by internal deletion of NES and used this mutant to further confirm the role of NES in nuclear export of Nrf2 in transfected cells. The Hepa-1 cells were transfected with pcDNA-Nrf2-V5 or pcDNA-Nrf2⌬NES-V5, treated with either Me 2 SO or t-BHQ for 8 h, and probed with FITC-tagged V5 antibodies. In related experiments, the transfected cells were subcellularly fractionated, and nuclear and cytosolic fractions were prepared by standard procedures. The nuclear and cytosolic proteins were separated on SDS-PAGE, Western blotted, and probed with V5 antibodies. The purity of nuclear and cytosolic fractions was tested by probing Western blots with anti-lamin B and antilactate dehydrogenase antibodies, respectively. The results of immunofluorescence are shown in Fig. 6, A and B, and results of Western analysis are shown in Fig. 6, C and D. The results showed immunofluorescence both in the cytosol and the nucleus of Me 2 SO-treated Hepa-1 cells expressing wild type Nrf2-V5 (Fig. 6A, Me 2 SO-LMB). This showed that Nrf2-V5 localized both in cytosolic and nuclear fractions. The results with Me 2 SO-treated Hepa-1 cells expressing mutant Nrf2⌬NES-V5 protein showed localization of mutant Nrf2 protein deficient in NES predominantly in the nucleus (Fig. 6B, Me 2 SO-LMB). The treatment of transfected cells with t-BHQ for 8 h showed similar localization of wild type Nrf2-V5 and mutant Nrf2⌬NES-V5 proteins as observed with Me 2 SOtreated cells. The results with Nrf2-V5 were similar as those observed with endogenous Nrf2 protein in Fig. 5A and indicated that t-BHQ-induced nuclear import of Nrf2-V5 was followed by nuclear export of Nrf2-V5 (Fig. 6A, t-BHQ-LMB). Interestingly, the Hepa-1 cells transfected with pcDNA-Nrf2⌬NES-V5 deficient in NES showed a prominent nuclear fluorescence in cells treated with t-BHQ for 8 h, indicating the absence of nuclear export of Nrf2⌬NES (Fig. 6B, t-BHQ-LMB). The pretreatment of transfected cells with leptomycin B blocked the nuclear export of Nrf2 in Me 2 SO-and t-BHQtreated cells as evident from the absence of immunofluorescence in cytosol (Fig. 6A, Me 2 SO ϩ LMB and t-BHQ ϩ LMB). However, no effect of leptomycin B on nuclear localization of Nrf2⌬NES mutant was observed in Me 2 SO-and t-BHQtreated cells (Fig. 6B, Me 2 SO ϩ LMB and t-BHQ ϩ LMB). Western analysis supported immunofluorescence studies (Fig. 6, C and D).
We performed in vitro and in vivo experiments and evaluated Nrf2 and Nrf2⌬NES mutant interaction with INrf2 to confirm that the predominant nuclear localization of Nrf2⌬NES is because of the deficiency of export signal and not because of the loss of interaction with INrf2 (Fig. 7). The plasmids encoding Nrf2-V5, Nrf2⌬NES-V5, and INrf2 were in vitro transcribed and translated. The translated proteins ran at required size on a 10% gel (Fig. 7A, lanes 1-3). The in vitro translated proteins were mixed together in equal amounts and then immunoprecipitated using the anti-V5 antibody and autoradiographed for 35 S signal. The results demonstrated that Nrf2⌬NES interacted with INrf2 similar as wild type Nrf2 (Fig. 7A, lanes 5 and  7). Nrf2⌬NES interaction with INrf2 was also determined in in vivo experiments as shown in Fig. 7B. Hepa-1 cells were co- transfected with V5-tagged wild type Nrf2 and Nrf2⌬NES along with FLAG-tagged INrf2. The cell lysates were immunoprecipitated with IgG or anti-V5 antibody (Fig. 7B). The results clearly demonstrate that INrf2 co-immunoprecipitated with both Nrf2 and Nrf2⌬NES (Fig. 7B, lanes 3 and 5). Therefore, both in vitro and in vivo assays showed that Nrf2⌬NES bound to INrf2 the same as wild type Nrf2 and that predominant localization of Nrf2⌬NES in the nucleus was not due to loss of interaction with INrf2 but was due to the loss of NES from Nrf2⌬NES mutant protein.
The NES domain of Nrf2 was synthesized and cloned in frame with two copies of YFP to generate plasmid p2YFP-NES. Immunofluorescence analysis of Hepa-1 cells transfected with p2YFP vector alone showed localization of YFP in the cytosol and nucleus, mostly in the nucleus, with no effect of leptomycin B treatment on localization of p2YFP (Fig. 8A). Interestingly, p2YFP-NES in a similar assay was found localized predominantly in cytosol (Fig. 8B, left panel). Nuclear localization of YFP-NES was not detected. Interestingly, the pretreatment of cells with leptomycin B blocked the nuclear export of YFP-NES, and YFP-NES was only detected in the nucleus (Fig. 8B,  right panel).
We also analyzed the effect of leptomycin B on expression and t-BHQ induction of the NQO1 ARE-mediated luciferase gene to demonstrate the role of NES in Nrf2 regulation of basal expression and antioxidant induction of downstream genes. The transfection of the NQO1 gene ARE-Luc in HepG2 cells expressed ARE-mediated luciferase activity that was induced in response to t-BHQ (Fig. 9, compare Me 2 SO with t-BHQ). The t-BHQ-induced expression was highest at 8 h after treatment and plateaued thereafter at 16, 24, and 36 h after t-BHQ treatment. In other words, the induction was the same between 8 and 36 h of t-BHQ treatment. This finding is in agreement with our Nrf2 localization data, where Nrf2 attains a normal localization pattern 8 h after t-BHQ induction. Interestingly, the treatment of cells with leptomycin B significantly increased basal expression and t-BHQ induction of ARE-mediated gene expression, especially at 24 and 36 h after Me 2 SO and t-BHQ treatment (Fig. 9, compare Me 2 SO ϩ LMB with t-BHQ ϩ LMB).
We followed the stability of Nrf2 and Nrf2⌬NES proteins in transfected cells to determine the functional mechanism of the role of nuclear export of Nrf2 in degradation of Nrf2. The degradation pattern of Nrf2⌬NES was compared with wild type Nrf2 in whole cell lysates and cytosolic and nuclear extracts. Hepa-1 cells were transfected with Nrf2V5 or Nrf2⌬NES-V5. The cells were pretreated with MG132 for 8 h to initially inhibit the proteosomal degradation of Nrf2, leading to accumulation of Nrf2 for degradation studies. The cells were then treated with cycloheximide for different time points to block new protein synthesis and let the accumulated Nrf2 degrade in the cell via proteasomal degradation. This design is frequently used to study the stability/degradation of proteins (11). Whole cell lysate, cytosolic, and nuclear extracts prepared from these cells were immunoblotted with anti-V5 antibody. In whole cell lysate, wild type Nrf2 stabilized after treatment with MG132 and disappeared within 2 h of cycloheximide treatment (Fig 10A, top left panel). On the contrary, Nrf2⌬NES degradation was significantly slower as compared with wild type Nrf2. This was evident from the observation that mutant Nrf2⌬NES was visible in significant amounts even 3 h after cycloheximide treatment (Fig. 10A, top right panel). It is also noteworthy that Nrf2⌬NES-V5 in whole cell lysate showed increased protein content due to nuclear accumulation and stability as compared with wild type Nrf2-V5 that is exported out and degraded (compare the top panels of whole cell lysate in Fig. 10A). In the cytosol, both wild type Nrf2-V5 and mutant Nrf2⌬NES-V5 degraded rapidly (Fig. 10A, cytosolic extract panels). However, the amount of Nrf2-V5 in the MG132 lane was higher than Nrf2⌬NES-V5 (compare lanes 2 in the cytosolic extract panels in Fig. 10A). This was in contrast to lower Nrf2-V5 as compared with Nrf2⌬NES-V5 in MG132-treated whole cell extract (compare lane 2 in whole cell lysate panels in Fig. 10A). The higher amount of Nrf2-V5 in MG132-treated cytosolic extract despite a lower amount in whole cell extract was due to stabilization of newly synthesized and nuclear exported Nrf2-V5. On the other hand, the lower content of Nrf2⌬NES-V5 in MG132-treated cytosolic extract despite the higher amount in whole cell extract was due to the inability of Nrf2⌬NES to be exported out of the nucleus. In the nucleus, the wild type Nrf2 degraded at a faster rate when compared with the mutant Nrf2⌬NES (Fig.  10A, compare the left and right panels under nuclear extracts). In addition, the amount of Nrf2⌬NES in nuclear extract was higher than Nrf2-V5 because of the absence of nuclear export of Nrf2⌬NES-V5 (compare nuclear extract panels in Fig. 10A). All of the results above combined indicated that without the export signal, Nrf2 could not exit the nucleus, and hence the degradation of Nrf2⌬NES was significantly delayed both in the nucleus and whole cell lysate when compared with the wild type Nrf2. In a similar experiment, the cells were pretreated with leptomycin B before treatment with cycloheximide. The results indicated that in the presence of leptomycin B, wild type Nrf2 was stabilized, because the protein was visible even 3 h after cycloheximide treatment. (Fig. 10B, left panel,

DISCUSSION
Nrf2-mediated expression and coordinated induction of a battery of defensive genes including detoxifying enzymes is a mechanism of critical importance in protection against chemically induced oxidative stress and neoplasia (1). Therefore, the signals/mechanisms that regulate nuclear availability of Nrf2 are extremely important for the regulation of expression and induction of defensive genes. In the current study, we identified and characterized a bipartite NLS, which is required for Nrf2 nuclear localization and can promote the nuclear localization of a normally cytoplasmic Nrf2 protein. In addition, we demonstrate, for the first time, that Nrf2 undergoes active nuclear export and that this export is leptomycin B-sensitive and mediated by an NES located in the Nrf2 C terminus.
Nrf2 is normally retained in the cytoplasm by its inhibitor INrf2 (6,7). Antioxidants antagonize this interaction leading to the release of Nrf2 from INrf2. Nrf2 translocates in the nucleus. This leads to increased ARE-mediated gene expression. A search of the Nrf2 amino acid sequence identified an NLS domain in the C terminus of the Nrf2 protein between amino acids 494 and 511 (amino acid sequence RRRGKQKVAAN-QCRKRK). Nrf2 NLS sequence aligned perfectly with well characterized NLS sequences from nucleoplasmin, p73, and p53. The NLS was required for nuclear translocation of Nrf2. This was clearly evident from several observations. Nrf2 mutant lacking the NLS failed to enter the nucleus and displayed diminished expression and induction of downstream genes including NQO1. The small amount of expression and induction observed in Fig. 2 was the same as in vector (pcDNA)transfected control and was due to endogenous Nrf2. In similar experiments, the wild type Nrf2 significantly increased basal and induced expression of the downstream NQO1 gene. The Nrf2 NLS sequence fused to GFP resulted in the nuclear accumulation of GFP, indicating that this signal sequence was sufficient to direct nuclear localization of Nrf2. It appears that Nrf2 contains a single NLS, since no other NLS was found by sequence analysis, and deletion of NLS led to complete loss of entry of mutant Nrf2 in the nucleus.
Immunofluorescence studies on endogenous Nrf2 proteins in Hepa-1 cells treated with antioxidants demonstrated nuclear localization of Nrf2 within 15 min of antioxidant treatment. Immunofluorescence analysis also demonstrated nuclear export of Nrf2 to the cytoplasm that might have started as early as 1 h after antioxidant treatment and was clearly visible at 8 h after antioxidant treatment. One can argue that newly synthesized Nrf2 might have contributed to the cytoplasmic appearance of Nrf2 in Hepa-1 cells 8 h after antioxidant treatment. This is possible; however, our data indicate that the contribution of new synthesis has to be minimal, because pretreatment of cells with leptomycin B failed to demonstrate the cytosolic presence of Nrf2. Therefore, a majority of the cytosolic presence of Nrf2 in Hepa-1 cells 8 h after treatment with antioxidant was due to nuclear export of Nrf2. The immunofluorescence of Nrf2-V5-transfected Hepa-1 cells with V5 antibody also supported the above conclusions on nuclear export of Nrf2. Amino acid sequence analysis of Nrf2 identified a leucine-rich NES consensus sequence at the C terminus of the Nrf2 protein between amino acids 545 and 554 (amino acid sequence LKRRLSTLYL). The NES consensus sequence was determined from previously reported nuclear export signals from several proteins, including IkB-␣, TFIIIA, hDM2, p53, p73, and PKI-␣ proteins (30). The studies with mutant Nrf2 lacking NES confirmed that Nrf2 NES is functional. Nrf2 was more prominent in the nucleus when its C-terminal NES was deleted by mutation or when nuclear export was inhibited by leptomycin B treatment. The predominant localization of NES-deficient mutant Nrf2⌬NES in the nucleus was not due to lack of its interaction with INrf2, because Nrf2⌬NES interacted with INrf2 the same as wild type Nrf2. Our studies also demonstrated that the Nrf2 C-terminal NES can function as an au- pcDNA-Nrf2-V5, pcDNA-Nrf2⌬NES-V5, and pcDNA-INrf2 plasmids were in vitro transcribed/translated using the TNT coupled reticulocyte lysate system by procedures described under "Experimental Procedures." 5 l of the translated proteins were loaded in the input lane. Equal amounts of proteins were mixed with the binding buffer as indicated and incubated at 37°C for 30 min, and the mixture was immunoprecipitated with either mouse IgG or anti-V5 antibody by shaking at 4°C overnight. The input and immunoprecipitates were resolved on a 10% SDS-PAGE, treated with Amplify solution to enhance the 35 S signal, dried, and autoradiographed. In a similar experiment, the samples were transferred to the nitrocellulose membrane and immunoblotted with anti-V5 antibody. Upper panel, 35 S autoradiograph; lower panel, V5-Western. B, in vivo interaction of Nrf2 and Nrf2⌬NES with INrf2. Hepa-1 cells were seeded in 100-mm plates and co-transfected with plasmids encoding wild type Nrf2-V5 or Nrf2⌬NES-V5 along with FLAG-INrf2 in a 4:1 ratio. 24 h after transfection, the cells were harvested and lysed, and 500 g of lysate was used to immunoprecipitate (IP) with IgG or anti-V5 antibody as described under "Experimental Procedures." The input (one-fifth) and immunoprecipitates were resolved on a 10% SDS-PAGE and immunoblotted (WB) with anti-V5 antibody. The same blot was stripped and reprobed with anti-FLAG antibody. tonomous nuclear export signal when fused to a heterologous protein. These results clearly demonstrated that the NES of Nrf2 is functional and that the Nrf2 protein is subject to active nuclear export that follows the Crm1-mediated export and is sensitive to leptomycin B.
The identification of NES in Nrf2 raised questions regarding the functional mechanism of nuclear export of Nrf2. Our data demonstrated that mutant Nrf2⌬NES protein degraded at a significantly slower rate than wild type Nrf2. This was clearly evident by rapid degradation of wild type Nrf2 but not Nrf2⌬NES. In addition, the treatment of cells with leptomycin B that prevents nuclear export of wild type Nrf2 also prevented degradation of Nrf2 in a similar manner as observed with mutant Nrf2⌬NES protein deficient in nuclear export. The leptomycin B, on the other hand, had no effect on degradation of mutant Nrf2⌬NES because of the absence of nuclear export of NES mutant Nrf2. Therefore, our data clearly suggested that NES is required for Nrf2 export to cytoplasm and nuclear export of Nrf2 is required for degradation of Nrf2.
Given that Nrf2 can function as a transcription factor, we considered that nuclear import and export might affect Nrf2 transactivation function. Not surprisingly, a nuclear importdeficient form of Nrf2 was less able to activate gene expression in transfected cells as also described above. However, Nrf2 protein deficient in NES was more prominent in the nucleus and correspondingly increased expression of ARE-mediated gene expression.
The studies also demonstrate that nuclear export of Nrf2 is functional even in the absence of antioxidant treatment. This was clearly evident from increased nuclear localization of Nrf2 and ARE-mediated gene expression in Hepa-1 cells treated with leptomycin B alone. Wild-type Nrf2 is expressed at low levels in most normal cells and, at least in some cell types, is localized in the cytoplasm. In response to oxidative stress and various other stresses, Nrf2 levels increase, and the Nrf2 pro-tein accumulates in the nucleus. The stress-induced, nuclear accumulation of Nrf2 is likely to result from both diminished nuclear export and continued nuclear import. These results suggested that nuclear and cytosolic distribution of Nrf2 is a balance of nuclear import and export of Nrf2. The studies also raise interesting questions regarding mechanisms that regulate nuclear import and export of Nrf2. It is believed that oxidative stress-mediated sulfhydryl modifications of INrf2 and/or phosphorylation of Nrf2 regulates release and nuclear transport of Nrf2 (1). It is expected that nuclear export of Nrf2 is also regulated by unknown modifications of Nrf2 and remains to be determined. Unknown modifications might include phosphorylation of Nrf2 as observed with other proteins including p53 and p73 (31). Interestingly, a recent report showed nuclear import and export of Keap1 (INrf2) (32). The impact of nuclear shuttling of INrf2 on localization and fate of Nrf2 remains unknown and is an exciting area of investigation.
In conclusion, our results indicate that subcellular localization of Nrf2 is controlled by both nuclear import and export signals and suggest that the overall distribution of Nrf2 is likely to result from the balance between these two processes. Proper control of nuclear import and export is likely to be an important regulatory determinant of Nrf2, since Nrf2 availability in the nucleus has significant impact on cell survival and growth.