Nrf2 Possesses a Redox-insensitive Nuclear Export Signal Overlapping with the Leucine Zipper Motif*

Basic leucine zipper (bZIP) protein Nrf2 is a key transcription factor mediating the antioxidant response. Under homeostatic conditions Nrf2 is an-chored to cysteine-rich Keap1 and sequestered in the cytoplasm. When challenged with oxidative stress, Keap1 functions as a redox-sensitive switch and re-leases Nrf2. Subsequently, Nrf2 translocates into the cell nucleus and binds to a cis-acting enhancer called the antioxidant response element located in the pro-moters of a battery of cytoprotective genes and ini-tiates their transcription. In this study we identify a canonical nuclear export signal (NES) ( 537 LKKQLST-LYL 546 ) located in the leucine zipper (ZIP) domain of the Nrf2 protein. The enhanced green fluorescent pro-tein-tagged ZIP domain of Nrf2 (amino acids 503–589) exhibited a CRM1-dependent cytosolic distribution that could be abrogated by site-directed mutations or treatment with the nuclear export inhibitor, leptomycin B. Ectotopic expression of the Nrf2-NES could also exclude the GAL4 DNA binding domain into the cytoplasm. This NES overlapped with the ZIP motif in Nrf2, suggesting that the formation of heterodimers between Nrf2 and other bZIP proteins may simultaneously M m M m M examined using a Nikon Eclipse E600 epifluorescent microscope and a Nikon C-SHG1 UV light source purchased from Micron-Optics (Cedar Knolls, NJ). HepG2 cells were cultured on ethanol-sterilized glass coverslips and transfected with 1 (cid:2) g of EGFP-Nrf2zip or its mutants using the Lipofectamine method (Invitrogen) and further cultured in F-12 me- dium for 36 h. Before microscopic examination, some coverslips were counterstained with 0.5 (cid:2) g/ml DAPI in 50% glycerol phosphate-buff- ered saline solution to visualize the position of the cell nuclei. The EGFP signals and DAPI signals were examined using fluorescein iso- thiocyanate and DAPI filters, respectively. The epifluorescent images were digitized using a Nikon DXM1200 camera and Nikon ACT-1 software (version 2). Images were superimposed using SPOT 3.5.2 software (Diagnostic Instrument Inc.). For time-lapse imaging, HepG2 cells transfected with EGFP-Nrf2zip were maintained at 37 °C and 5% CO 2 in 35-mm glass-bottomed dishes (MatTek Corp., Ashland, MA) and subjected to the treatments of LMB or redox compounds. The fluorescent signals were examined using a Zeiss Axiovert 200M inverted epifluorescent microscope. The fluorescent images were captured using a AxioCam MR monochrome camera and assigned pseudo-color using AxioVision 4.1 software.

Although solid progress has been made in the understanding of the Keap1-Nrf2 cytosolic interaction, the mechanism governing nuclear translocation of Nrf2 is largely unknown. Great progress has been made in elucidating the mechanisms underlying nucleocytoplasmic transportation. Active nucleocytoplasmic transport through nuclear pores is mediated by a variety of nuclear importin and exportin proteins (32). In a Ran GTPasedependent fashion, these nuclear importin or exportin proteins recognize the nuclear location signals (NLSs) (33) or nuclear export signals (NESs) (34 -36) on cargo proteins and facilitate their transport. In the case of nuclear export, CRM1 (chromosome region maintenance) has been characterized as a nuclear exportin (37,38). CRM1 can effectively export human immunodeficiency virus Rev protein in a Ran-regulated manner (38). The first leucine-rich NESs were identified in human immunodeficiency virus Rev protein (35) and in an inhibitor of cAMPdependent protein kinase (34). Using a randomization and selection approach to compare a variety of functional NES motifs, a consensus leucine-rich NES motif has been proposed (36). CRM1-mediated nuclear export can be effectively inhibited by the cytotoxin leptomycin B (LMB) (39). LMB can bind covalently to a cysteine residue in CRM1 (39) and interfere with the binding of both Ran and the cargo proteins (37,40). Several functional NES motifs have been identified in some bZIP proteins. One NLS and one redox-sensitive NES have been identified in the CNC/bZIP proteins Bach 1/2 (41) that function as co-repressors in HO-1 transcription (42,43). Cadmium-mediated (43) and heme-regulated NES (43,44) activities have also been identified in Bach 1. In addition, a NES has also been identified in the bZIP protein TCF11 (45).
In the present study we have identified a canonical leucinerich NES overlapping with the leucine zipper motif of Nrf2. This Nrf2-NES binds to CRM1 and is sensitive to LMB. Site-directed mutations targeting the critical leucine residues in the NES abolished its nuclear export behavior. Furthermore, expression of this Nrf2-NES can exclude the GAL4 DNA binding domain from the nucleus, which is otherwise predominantly nuclear. Unlike Bach 1/2, the Nrf2-NES appears to be redox-insensitive. Understanding this NES and the mechanisms controlling nuclear localization of Nrf2 will deepen our understanding of Nrf2 signaling and the overall antioxidant response.

MATERIALS AND METHODS
Cell Culture and Chemicals-Human hepatoma HepG2 cells, cervical squamous cancerous HeLa cells, and HEK293 cells were obtained from ATCC (Manassas, VA). HepG2 cells were maintained as a monolayer using F-12 medium supplemented with 10% fetal bovine serum, 1.7 mg/ml sodium bicarbonate, 0.1 unit/ml insulin, 0.5ϫ minimal essential amino acids, 100 units/ml penicillin, and 100 g/ml streptomycin. HeLa and HEK293 cells were cultured as monolayers using minimum Eagle's medium supplemented with 10% fetal bovine serum, 2.2 mg/ml sodium bicarbonate, 100 units/ml penicillin, and 100 g/ml streptomycin. Leptomycin B was purchased from Calbiochem. 4Ј,6diamidine-2Ј-phenylindole (DAPI) and isopropylthio-␤-D-galactoside were purchased from Sigma. Sulforaphane was purchased from Lkt Laboratory (St. Paul, MN). Diethyl maleate (DEM) was purchased from Aldrich. N-Acetyl-L-cysteine and GSH were purchased from Sigma. Plasmid Constructions-The full-length wild type cDNA of human Nrf2 was kindly provided by Drs. Yuet W. Kan and Jefferson Chan (University of California at San Francisco). The ZIP domain of Nrf2 (amino acids 503-589) was PCR-amplified (Table I) and subcloned into pEGFP-C1 vector (Clontech, Palo Alto, CA) by XhoI/BamHI digestion. The resulting plasmid was designated enhanced green fluorescent protein (EGFP)-Nrf2zip. The GAL4 DNA binding domain (DBD) was amplified (Table I) and subcloned into EGFP-Nrf2zip by PstI/BamHI digestion to form a EGFP-Nrf2-NES-GAL4DBD fusion protein. PCRamplified GAL4DBD (Table I) was also subcloned into pEGFP-C1 as an XhoI/BamHI fragment to form an EGFP-GAL4DBD fusion protein.
Site-directed Mutagenesis-Alanine substituted mutations were performed using the QuikChange XL site-directed mutagenesis kit purchased from Stratagene (La Jolla, CA) according to the manufacturer's instruction with few modifications. Briefly, both sense and antisense mutagenic oligonucleotide primers (Table I) were designed to mutate leucine to alanine. The primers were synthesized and PAGE/high performance liquid chromatography-purified by Intergrated DNA Technologies, Inc. (Coralville, IA). Mutagenesis reactions were performed in 50-l reaction solutions containing 100 ng of template DNA, 125 ng of sense and antisense mutagenic primers, 1ϫ reaction buffer with dNTP supplements, 3 l of QuikSolution, 2.5 units of Pfu Turbo DNA polymerase, and double distilled water. Mutagenesis reactions were performed as follows: denaturation at 95°C for 1 min, 18 cycles of 95°C for 50 s, 60°C for 50 s, and 68°C for 7 min and concluded by a 7-min extension at 68°C. The parental methylated dsDNA plasmids in the reaction mix were digested with DpnI at 37°C for 3 h. Afterward, the reaction products were transformed into XL10-Gold cells (Stratagene). The mutant plasmids were extracted and verified by DNA sequencing.
Transient Transfection and Reporter Gene Activity Assays-HeLa or HEK293 cells were plated in six-well plates at a density of about 4.0 ϫ 10 5 cells/well. Twenty four hours after plating, cells were transfected using Lipofectamine 2000 (Invitrogen) according to manufacturer's instructions. Briefly, 2 g of pHM6-Nrf2 were added into 250 l of Opti-MEM together with 0.5 g of ARE-Luc reporter and 0.5 g of pRSV-␤-galactosidase plasmid, which was included for the normaliza-  A AAA ACT CGA GAA CTG GAA AAT TA GTA GAA CTA G-3Ј  Ϫ  5Ј-A A AAA GGA TCC CTA GTT TTT CTT AAC ATC TGG C-3Ј   Nrf2zip L537A  ϩ  5Ј-T GAC AAA AGC CTT CAC CTA GCG AAA AAA CAA CTC AGC ACC T-3Ј  Ϫ  5Ј-A GGT GCT GAG TTG TTT TTT CGC TAG GTG AAG GCT TTT GTC A-3Ј   Nrf2zip L541A  ϩ  5Ј-CTT CAC CTA CTG AAA AAA CAA GCC AGC ACC TTA TAT CTC GAA GTT-3Ј  Ϫ  5Ј-AAC TTC GAG ATA TAA GGT GCT GGC TTG TTT TTT CAG TAG GTG AAG-3Ј   Nrf2zip L544A  ϩ  5Ј-CTG AAA AAA CAA CTC AGC ACC GCA TAT CTC GAA GTT TTC AGC ATG-3Ј  Ϫ  5Ј-CAT GCT GAA AAC TTC GAG ATA TGC GGT GCT GAG TTG TTT TTT CAG-3Ј   Nrf2zip L546A  ϩ  5Ј-A CAA CTC AGC ACC TTA TAT GCC GAA GTT TTC AGC ATG CTA C-3Ј  Ϫ  5Ј-G TAG CAT GCT GAA AAC TTC GGC ATA TAA GGT GCT GAG TTG T-3Ј   Nrf2zip 4P mutant  ϩ  5Ј-CAC CTA GCG AAA AAA CAA GCC AGC ACC GCA TAT GCC GAA GTT TTC AGC-3Ј  Ϫ  5Ј-GCT GAA AAC TTC GGC ATA TGC GGT GCT GGC TTG TTT TTT CGC TAG GTG-3Ј EGFP-Gal4DBD tion of transfection efficiency as reported previously (46). Lipofectamine was added into another tube of 250 l of Opti-MEM in a 1:2.5 ratio to the amount of plasmid and incubated at room temperature for 5 min. The plasmid solution was then mixed with Lipofectamine solution with vigorous agitation and incubated at room temperature for 30 min. Right before transfection cell culture medium was changed to Opti-MEM. Cells were incubated with transfection solution for 4 h, changed to fresh minimum Eagle's medium, and cultured for an additional 36 h before harvest. Cells were then washed twice with phosphate-buffered saline, scraped, and incubated in 1ϫ reporter lysis buffer (Promega) on ice for 30 min. After centrifugation, a 5-l lysate was mixed with luciferase substrate (Promega), and the ARE-luciferase activity was measured using a Sirius luminomenter (Berthold Detection System). ␤-Galactosidase activity was measured as described before (47). The value of ARE-luciferase activity was normalized by ␤-galactosidase activity.
Expression and Purification of CRM1 Protein-CRM1 cDNA (IM-AGE 5267242) was obtained from IMAGE cloning consortium. PCR primers were designed (sense, 5Ј-CGGT GGA TCC ATG CCA GCA ATT ATG ACA ATG TTA G-3Ј; anti-sense, 5Ј-CGG TGG TAC CTT AAT CAC ACA TTT CTT CTG GAA TC-3Ј) to subclone CRM1 into the vector pQE30 (Qiagen) via BamHI/KpnI digestion, which adds a histidine tag of MRGSHHHHHH in the amino end of CRM1. The construct was confirmed by sequencing. Escherichia coli M15 cells (Qiagen) were transformed with pQE30-CRM1 and grew in 100 ml of LB medium containing ampicillin (100 mg/liter) and kanamycin (25 mg/liter) at 37°C overnight and then inoculated into 1 liter of LB supplemented with both antibiotics. The expression of His-CRM1 was induced by the addition of 0.5 mM isopropylthio-␤-D-galactoside. After induction for 4 h at 30°C, the cells were harvested by centrifugation at 4000 ϫ g at 4°C for 30 min. The pellet was resuspended in buffer A (50 mM NaH 2 PO 4 , 0.5 M NaCl, and 10 mM imidazole, pH 8.0) containing 1 mg/ml lysozyme and incubated on ice for 30 min. Cells were disrupted by sonication. The lysed cells were centrifuged at 18,000 ϫ g at 4°C for 15 min. 1 ml of 50% nickel nitrilotriacetic acid slurry was added to 4 ml of supernatant lysate and mixed gently by shaking at 4°C for 60 min. The mixture was then loaded into an empty polypropylene column (Qiagen) and washed twice with buffer B (50 mM NaH 2 PO 4 , 0.5 M NaCl, and 20 mM imidazole, pH 8.0). His-CRM1 protein was eluted with buffer C (50 mM NaH 2 PO 4 , 0.5 M NaCl, and 250 mM imidazole, pH 8.0).
Expression and Purification of GST Fusion Proteins and GST Pulldown Assay-Nrf2zip and Nrf2zip 4-point mutant were subcloned into pGEX-2T vector (Amersham Biosciences) to add a GST tag to their N termini. The pGEX-Nrf2zip/4p plasmids were transformed into E. coli XL-1Blue (Stratagene) and cultured in LB medium supplemented with ampicillin. After overnight induction with 0.5 mM isopropylthio-␤-Dgalactoside at 30°C, the cells were sonicated and incubated in 1% Triton X-100/phosphate-buffered saline solution at ice for 0.5 h. After 15 min of centrifuging at 12,000 rpm, the supernatant cell lysates were incubated with glutathione-Sepharose-4B beads (Amersham Biosciences) at room temperature for 0.5 h and then loaded into the PD-10 column (Amersham Biosciences). After 3 washes with phosphate-buffered saline, the GST fusion proteins were eluted with 10 mM glutathione. For the GST pull-down assay, 2 g of GST or GST-Nrf2zip/4p proteins were mixed with 10 l of glutathione-Sepharose-4B beads. After incubation at 4°C for 0.5 h, each sample was added to 100 ng of purified CRM1 protein and further incubated at 4°C for 2 h. The beads were then washed extensively with incubation buffer (10 mM sodium phosphate, pH 7.3, 150 mM NaCl, 0.1% Triton X-100, and 1 mM 2-mercaptoethanol) and analyzed by immunoblotting.
Immunoprecipitation and Western Blotting-HepG2 cells cultured in 10-cm Petri dishes were harvested 36 h after transient transfection of Nrf2zip or the Nrf2zip4p mutant in cell lysis buffer containing 10 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, 1 mM NaF, 100 M Na 3 VO 4 , 1% sodium deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, and 1% protease inhibitor. The protein concentrations of cell lysates were determined by the Bradford method. For immunoprecipitation, cell lysates containing 500 g of proteins were mixed with 0.1 g of purified CRM1 protein. The solutions were first incubated with 20 l of protein A-conjugated-Sepharose 4B beads (Zymed Laboratories Inc., San Francisco, CA) to clean up endogenous IgG. The solutions were then incubated with 1 g of mouse anti-GFP antibody at 4°C for 1 h and then mixed with 40 l of protein A-Sepharose beads and tumbled overnight at 4°C. The immunoprecipitation solutions were then centrifuged, and the pellets were washed 3 times with the lysis buffer. The protein complexes were then eluted with loading buffer, heated at 80°C for 5 min, and subsequently analyzed by Western blotting. For Western blotting, either the immunoprecipitation products or lysates containing 20 g of proteins for HO-1/␤-actin assay were resolved by 4 -15% linear gradient SDS-polyacrylamide gel (Bio-Rad) electrophoresis and transferred to polyvinylidene difluoride membrane using a semi-dry transfer system (Fisher). The membrane was blocked with 5% nonfat milk in Tris-buffered saline with Tween 20 containing 20 mM Tris-HCl, pH 7.6, 8 mg/ml NaCl, and 0.2% Tween 20 at room temperature for 1 h. The membrane was then probed with polyclonal rabbit anti-CRM1 (1:500), anti-Nrf2 (C-20) (1:500), goat anti-HO-1 (1:500), or anti-␤-actin (1:5000) in 3% nonfat milk, Tris-buffered saline with Tween 20 at 4°C overnight. After washing 3 times with Tris-buffered saline with Tween 20, the membrane was blotted with peroxidase-conjugated anti-rabbit or anti-goat secondary antibody (1:5000 dilution) at room temperature for 1 h. The protein was detected using the ECL mixture from Bio-Rad.
Epifluorescent Microscopy-The expression and subcellular distribution of EGFP-tagged Nrf2zip and Nrf2zip mutants were examined using a Nikon Eclipse E600 epifluorescent microscope and a Nikon C-SHG1 UV light source purchased from Micron-Optics (Cedar Knolls, NJ). HepG2 cells were cultured on ethanol-sterilized glass coverslips and transfected with 1 g of EGFP-Nrf2zip or its mutants using the Lipofectamine method (Invitrogen) and further cultured in F-12 medium for 36 h. Before microscopic examination, some coverslips were counterstained with 0.5 g/ml DAPI in 50% glycerol phosphate-buffered saline solution to visualize the position of the cell nuclei. The EGFP signals and DAPI signals were examined using fluorescein isothiocyanate and DAPI filters, respectively. The epifluorescent images were digitized using a Nikon DXM1200 camera and Nikon ACT-1 software (version 2). Images were superimposed using SPOT 3.5.2 software (Diagnostic Instrument Inc.). For time-lapse imaging, HepG2 cells transfected with EGFP-Nrf2zip were maintained at 37°C and 5% CO 2 in 35-mm glass-bottomed dishes (MatTek Corp., Ashland, MA) and subjected to the treatments of LMB or redox compounds. The fluorescent signals were examined using a Zeiss Axiovert 200M inverted epifluorescent microscope. The fluorescent images were captured using a AxioCam MR monochrome camera and assigned pseudo-color using AxioVision 4.1 software.

Structure of Nrf2 and Design of Fusion Proteins-The
Nrf2 protein can be divided into two parts (Fig. 1). The C terminus contains the CNC, basic region, and ZIP domains. The N terminus contains the acidic transactivation domain and a Keap1 binding Neh2 domain (11,21). There are three putative NLSs (33) in human Nrf2. In the N terminus a region from amino acids 26 -37 that is enriched in basic amino acids may function as a NLS. In the C terminus there is a canonical bipartite NLS localized in the basic region (amino acids 486 -502) featuring 2 tandems of 3-4 basic amino acids separated by 10 amino acids (33). Our preliminary data found that this putative bipartite NLS indeed functioned strongly as an NLS. 2 In addition, there is another putative NLS ( 580 KSKK 583 ) located at the end of the C terminus. In searching for a NES we found embedded in the leucine zipper domain a leucine-rich motif ( 537 L*KKQLSTL*YL 546 ) that conforms to the canonical NES motif of ⌽XXX⌽XX⌽X⌽ (34)(35)(36), where ⌽ stands for long chain hydrophobic amino acid residues, and X can be any amino acid. Interestingly, two key residues of this 2 W. Li, M. R. Jain, and A. N. T. Kong, unpublished data. . To examine whether the putative NES plays a functional role in Nrf2 translocation, we designed and constructed a probe protein containing this NES. Because the position of this NES is very close to the bipartite NLS (amino acids 486 -502), we dissected them from each other by constructing a fusion protein called EGFP-Nrf2zip. EGFP-Nrf2zip consists of an EGFP tag and a segment of the C terminus of human Nrf2 (amino acids 503-589), which begins right after the bipartite NLS in the basic region and preserves the entire ZIP domain (Fig. 1).
Nrf2 Possesses a Functional NES Located in the ZIP Domain-In the majority of HepG2 cells, EGFP-Nrf2zip proteins exhibited a predominantly cytosolic distribution pattern (Fig.  2, A-C) in which green fluorescence was confined to the cytoplasm, with cell nuclei devoid of fluorescence ( Fig. 2A). The positions of cell nuclei were confirmed by DAPI staining (Fig.  2B, arrows) superimposed with the green fluorescent image (Fig. 2C). In contrast, when EGFP alone was expressed, green fluorescence was observed throughout the cell (data not shown). We found that EGFP-Nrf2zip proteins were confined in cytoplasm in more than 85% of the transfected HepG2 cells (Table II). In 13.9% of the cells an evenly distribution pattern was observed (Table II). There were barely any cells showing nuclear distribution. Similar results were also found when EGFP-Nrf2zip was expressed in HeLa and HEK293 cells, although the cytosolic distribution of EGFP-Nrf2zip was more pronounced in HeLa and HEK293 cells (data not shown).
To further examine the possibility that the observed cytosolic distribution was mediated by the NES, we performed site-directed mutagenesis of the NES. We found that a singlepoint mutation, leucine 544 to alanine, could convert the cytosolic distribution into a whole cell distribution pattern ( Fig. 2D; Table II). Similar results were also found with L537A, L541A, and L546A mutants (data not shown). The percentages of cells with a whole cell distribution pattern were similar among these four single point mutants (Nrf2zip-1P) (data not shown) and in strong contrast to the results with EGFP-Nrf2zip. Two point mutations of L537A and L546A of Nrf2 also exhibited an evenly distributed pattern ( Fig. 2E; Table II). Furthermore, we constructed a mutant ablating all four of these leucines and observed the same whole-cell distribution pattern ( Fig. 2F; Table II). Thus, it appears that single point mutations in the Nrf2-NES are sufficient to disable the nuclear-exporting function. No further increase in disruption was observed with increasing multiple-point mutants relative to the mutant carrying a single substitution.
One might have predicted that disruption of the NES by mutation would lead to a nuclear condensation pattern rather than a whole cell distribution pattern because the construct has a putative NLS (residues 580 KSKK 583 ) to promote nuclear localization. Import in the absence of an NES for export might lead to accumulation in the nucleus. There are two possible explanations. One is that 580 KSKK 583 is either not a functional NLS or is a weak NLS. However, the 580 KSKK 583 motif, where the lysine residue (K) can be substituted with another basic amino acid arginine (R), is conserved in all Nrf2 molecules cloned from different species (see Fig. 7), suggesting functional importance. Alternatively, Nrf2zip may possess additional unidentified NESs or other unknown inhibitory elements to tightly control the nuclear localization of Nrf2zip. Further experiments are needed to solve this puzzle.
The Nuclear Exporting Activity of Nrf2-NES Is CRM1-dependent-Next, we examined whether the function of Nrf2-NES is mediated by binding to the nuclear exporting protein, CRM1 (40). Our in vitro GST pull-down assay showed clearly that GST-Nrf2zip bound to CRM1 (Fig. 3A). In contrast neither GST nor the GST-Nrf2zip-4p mutant bound to CRM1 (Fig. 3A). Our in vitro CRM1 pull-down results also showed that only Nrf2zip bound to CRM1 (data not shown). In agreement with our GST assay result, our immunoprecipitation assay also showed that when 100 ng of CRM1 protein were incubated with 500 g of the lysate of HepG2 cells transfected with EGFP-Nrf2zip or EGFP-Nrf2zip-4p in the presence of mouse anti-GFP antibody and protein A conjugated-Sepharose beads, only the co-precipitation of CRM1 and EGFP-Nrf2zip was detected (Fig. 3B). In contrast, no binding was detected between CRM1 and the Nrf2zip-4p mutant (Fig. 3B). To further examine the CRM1-dependent nuclear export activity of Nrf2-NES, we treated HepG2 cells expressing EGFP-Nrf2zip with LMB, an inhibitor of CRM1 (39). Time-lapse imaging showed that when treated with 10 nM LMB, the fluorescence of EGFP-Nrf2zip migrated into the nucleus within 10 -20 min (Fig. 3C). A similar time course of disruption of nuclear export of transcription factor Bach 2 by 10 ng/ml LMB was reported previously (41). Percentage assay also confirmed that after treatments with LMB, the distribution pattern of EGFP-Nrf2zip was converted into a whole cell distribution pattern (Table II). In contrast, when HepG2 cells expressing EGFP were treated with LMB, no nuclear translocation/condensation was observed (data not  shown). Therefore, these data suggest that the observed cytosolic distribution pattern of EGFP-Nrf2zip was very likely maintained by the active expulsion mediated by the NES in a CRM1-dependent manner. Ectotopic Expression of Nrf2-NES Can Expel GAL4DBD into the Cytoplasm-To further examine whether the Nrf2-NES alone is sufficient to exert nuclear exporting function, we expressed the Nrf2-NES (amino acids 503-567) ectotopically in a well known nuclear protein, the DNA binding domain of yeast protein GAL4 (Gal4DBD) (48). We constructed an EGFPtagged fusion protein by fusing the Nrf2-NES to the N terminus of GAL4DBD (Fig. 1). When expressed alone, EGFP-GAL4DBD was found in the nucleus in more than 70% of the cells (Fig. 4A; Table II). In contrast, more than 60% of HepG2 cells expressing EGFP-Nrf2-NES-GAL4DBD exhibited a cytosolic distribution pattern ( Fig. 4B; Table II). Site-directed mutation of the leucine residue corresponding to Leu-544 of Nrf2-NES in EGFP-Nrf2-NES-GAL4DBD was sufficient to abolish the cytosolic distribution pattern but failed to restore the nu-clear distribution pattern (Fig. 4C; Table II). LMB treatments could also convert the cytosolic distribution of EGFP-Nrf2-NES-GAL4DBD into an almost whole cell distribution pattern, with nearly 90% of the cells (Table II) exhibiting an evenly distribution pattern (Fig. 4D, arrow) and only about 9% of the cells showing nuclear accumulation of green fluorescence (Fig.  4D, arrowhead).
These ectotopic expression data clearly showed that Nrf2-NES could expel a heterologous nuclear protein into the cytoplasm. Considering the fact that GAL4DBD possesses an innate NLS (48), our data suggest that the Nrf2-NES appeared to be a stronger driving force than the NLS of GAL4DBD. Failure to restore the nuclear condensation pattern of GAL4DBD when the Leu-544 residue of Nrf2-NES is mutated is puzzling. There may be some unidentified NES or other inhibitory element(s) that deters the complete nuclear import of Nrf2-NES(L544A)-GAL4DBD. Further studies are needed to address this issue.
The Nrf2-NES Appears to Be Redox-insensitive-Because Nrf2 plays a pivotal role in the antioxidant response, we examined whether the nuclear exporting activity mediated by Nrf2-NES would respond to redox signals. Previous studies have shown that phyto-oxidant sulforaphane (SUL), a natural isothiocyanate, and DEM could induce ARE-luciferase activity mediated by Nrf2 (31,49). The maximal induction elicited by 12.5 M SUL could be effectively inhibited by the pretreatments with 5 mM reducing compounds NAC or conjugating agent GSH (31). Our time-lapse imaging data showed that when HepG2 cells expressing EGFP-Nrf2zip were treated with 12.5 M SUL for 1 h, the cytosolic distribution pattern of EGFP-Nrf2zip was unchanged (Fig. 5A-D). Similar results were also found in treatments with 100 M DEM (data not shown). A cell percentage assay also showed that virtually no difference could be detected between EGFP-Nrf2zip-expressing cells with or without the treatments of SUL and DEM for 2 h (Table III). In addition, treatment with reducing compounds of 5 mM GSH for 1 h also failed to disrupt the cytosolic distribution pattern of EGFP-Nrf2zip in HepG2 cells (Fig. 5, E-H ;  Table III). Similarly, treatment with 5 mM NAC for 1 h did not alter the cytosolic distribution of Nrf2zip (data not shown). In contrast, treatment with 10 nM LMB for 20 min did disrupt the cytosolic distribution of EGFP-Nrf2zip (Fig. 3B, Table II). These data suggest that the Nrf2-NES is redox-insensitive. Further experiments such as challenging EGFP-Nrf2zipexpressing cells with redox agents with different chemical structures will reveal whether this Nrf2-NES is truly redox-insensitive.

Leptomycin B Changes the Subcellular Distribution Pattern of Wild Type Nrf2 and Enhances Nrf2
Signaling-Finally we examined whether the Nrf2-NES also plays a functional role for the wild type Nrf2. When EGFP-tagged full-length Nrf2 was expressed in HepG2 cells, a heterologous distribution pattern was observed. Cell percentage assay showed that the majority of the cells (81.1%) showed whole cell distribution (Fig. 6A, Table II), very few cells (1.6%) showed a nuclear condensation pattern (Table II), and 17.4% cells showed the cytosolic distribution ( Fig. 6B; Table II). Because there are three putative NLSs in Nrf2 and three more NES-like motifs in Nrf2 2 in addition to this NES located in the ZIP domain, the observed heterogeneous distribution pattern of EGFP-Nrf2 may reflect the dynamic balance among multivalent nuclear localization signals and nuclear export signals. When EGFP-Nrf2-expressing cells were treated with 10 nM LMB for 2 h to block CRM-1-dependent nuclear export, a predominantly nuclear condensation pattern was observed ( Fig. 6C; Table II), demonstrating that Nrf2 definitely possesses at least one functional NES. Whether the NES of the Zip domain is important in the wild type Nrf2 remains to be further investigated. Further studies are needed to dissect whether the other NES-like motifs contribute in the nuclear export function of Nrf2.
Consistent with the observation that treatments with LMB could mobilize EGFP-Nrf2 into the nucleus, the treatments of LMB also had a functional impact on Nrf2-mediated antioxidant response. In a reporter gene activity assay, 12-h treatments of 25 nM LMB significantly induced Nrf2-mediated ARE luciferase activity (Fig. 6D). Furthermore, 12-h treatments with 25 nM LMB also enhanced the expression level of HO-1, an Nrf2-regulated antioxidant protein (Fig. 6E). It is possible that other unidentified NESs present in Nrf2 may be affected by LMB and account for these biological functions in the wild type protein. At this time it would not be possible for us to test the biological relevance of this newly discovered NES in the ab-sence of the transactivation domain of Nrf2, which is located at the N terminus (Fig. 1). Nevertheless, our current study presented the first clue as to the importance of the NES located in the leucine zipper domain with its involvement in the nuclear localization of Nrf2 and, therefore, will have an impact on the biological transcriptional function in the nucleus. DISCUSSION In the present study we identified for the first time a canonical leucine-rich NES in Nrf2. This Nrf2-NES mediated pronounced CRM-1-dependent nuclear export activities, which were abolished by LMB treatments and site-directed mutations. When expressed ectotopically, this Nrf2-NES could convert the nuclear distribution pattern of Gal4DBD into a cytosolic distribution pattern. These data show that the Nrf2-NES may play a functional role in determining the subcellular localization of Nrf2. It is also quite unusual that the position of the NES overlaps with the leucine zipper motif of Nrf2. Furthermore, this Nrf2-NES appears to be redox-insensitive. Taken together these discoveries help expand our understanding of the mechanisms underlying the transcriptional activation of Nrf2.
Up until now, the mechanistic studies of Nrf2 activation were mainly focused on Keap1 (for review, see Ref. 50). Virtually nothing is known about Nrf2 nuclear translocation after it is released from Keap1. In our current study the discovery of a functional NES in Nrf2 shows that Nrf2 translocation is not a passive or an automatic process. In fact, even though Nrf2 is released from Keap1 in the cytoplasm, our results suggest that there may be active nuclear export of Nrf2 to retain it in the cytoplasm. This NES may need to be masked to achieve nuclear localization. In other words, Keap1 may not be the only arresting force to sequester Nrf2 in the cytoplasm.
The NES motif we identified in Nrf2 conforms to the canonical leucine-rich NES (34 -36). Canonical leucine-rich NES feature a 10-amino acid motif formulated as ⌽ 1 XXX⌽ 2 XX⌽ 3 X⌽ 4 , where 4 positions of ⌽ are required to be hydrophobic amino acids residues such as leucine, isoleucine, valine, methionine, and phenylalanine, whereas X can be any amino acid (34 -36). Among these four ⌽ residues, ⌽ 3 and ⌽ 4 are critical and must be hydrophobic amino acids to exert nuclear export function (36). In the present study our functional assays provided several lines of evidence that mutation of each of these four ⌽ residues alone or in combination was sufficient to disable the NES function in Nrf2zip.
One salient feature of this Nrf2-NES is its overlapping position with the leucine zipper domain (Fig. 7). Under physiological conditions, for the formation of homo-or heterodimer, the leucine zipper domain of each bZIP monomer assumes a conformation of parallel coiled coil (51) that consisted of 4 -6 heptads formulated as (abcdefg) 4 -6 . The position a and d need to be hydrophobic residues. In the process of dimerization, the a and d residues in one monomer interact with the complementary d and a residue in the opposite monomer, respectively (52). The interaction forms a hydrophobic core essential for dimer stability (53). In the Nrf2-NES motif, ⌽ 1 (Leu-537) and ⌽ 3 (Leu-544) are located at the d position in the fifth and sixth heptad of ZIP domain, respectively. The ⌽ 2 (L541) is located at the a position in the sixth heptad. In other words, this Nrf2-NES occupies three key positions in the ZIP motif. The overlapping of Nrf2-NES and ZIP motifs implies that when Nrf2 forms heterodimers via its leucine zipper with other bZIP proteins, such as its obligatory binding partner small MafG/K proteins, the NES motif may be simultaneously masked. Indeed, we have found that MafG/K could facilitate the nuclear translocation of EGFP-Nrf2zip. 3 Therefore, it appears that the co-localization of NES with ZIP motif may enable the dimerization to play a dual functional role in Nrf2 signaling. Nrf2 heterodimerization via leucine zipper with MafG/K may not only enhance the DNA binding specificity of Nrf2 (14,15) but may also effectively recruit Nrf2 into the nucleus by simultaneously masking the NES activity. Future studies will test this hypothesis.
Because the Nrf2-NES may play a significant role in Nrf2 signaling, we further proceeded to ask whether this mechanism is also employed by other Nrf proteins, i.e. Nrf1 and Nrf3. When the amino acid sequences of the C-terminal ends of Nrf1, Nrf2, and Nrf3 from various species are aligned, it is very interesting to note that the NES motif observed in human Nrf2 is highly conserved in all known Nrf2 sequences with the exception of zebrafish (Fig. 7). The high cross-species conservation of the NES motif underlines the functional importance of this NES motif in Nrf2 activation. In contrast, the Nrf2-NES motif does not appear to be conserved in Nrf1 and Nrf3 (Fig. 7). A careful search in the ZIP domain of Nrf1 and Nrf3 failed to identify any leucine-rich NES motif (Fig. 7). Therefore, this leucine-rich NES appears only conserved for Nrf2 molecules. Nrf1 and Nrf3 have some similar amino acids to the Nrf2 motif possessing hydrophobic residues in position ⌽ 1 , ⌽ 2 , and ⌽ 3 . However, in position ⌽ 4 , Nrf1 and Nrf3 have polar residues (glutamine in Nrf1 and glutamine or histidine in Nrf3) (Fig. 7). ⌽ 4 is critical for NES function (36), which is consistent with our results showing that alanine substitution of Leu-546 was sufficient to abolish the nuclear export activity of human Nrf2-NES. Thus, it seems likely that the polar residues at position 4 of Nrf2-NES may be sufficient to abrogate the NES activity in Nrf1 and Nrf3. This observation of differences among Nrf molecules may challenge the conventional notion of functional redundancy among the Nrf molecules. In gene knock-out experiments, ablation of the Nrf2 molecule did not result in any severe phenotypic changes (54), and it was attributed to redundancy conferred by other Nrf molecules. However, the apparently different NES signals among Nrf molecules as described above suggests that functional redundancy may be more complicated than originally conceived, and further investigation is needed.
Although the NES is an important element in determining the subcellular localization of Nrf2, complete ablation of this NES by a four-point mutation failed to result in the accumulation of Nrf2zip in the nucleus. A similar result was observed in ectotopically expressed Nrf2-NES. Point mutations of Nrf2-NES could not restore the nuclear distribution pattern of GAL4DBD even though GAL4DBD has an NLS (48). In addition, LMB treatment also did not accumulate Nrf2zip in the FIG. 7. Sequence alignment of the C-terminal ends of Nrf1, -2, and -3 including the ZIP domain. The NES as identified in human Nrf2 is also conserved in mouse, rat, chick, and frog. The positions of key residues of NES are highlighted in black boxes. The Nrf2-NES is not conserved in Nrf1 and Nrf3, with one key leucine residue changed in Nrf1 and Nrf3. Furthermore, there is no NES-like motif in the ZIP domain of Nrf1 and Nrf3. The leucine residues for leucine zipper are designated with asterisks. A conserved putative NLS motif found in Nrf2 in different species is also highlighted in a transparent box.
nucleus. This incomplete nuclear import behavior suggested the possibility that other unidentified NESs or inhibitory elements, which are irresponsive to LMB treatments, may exist in Nrf2zip that deter further nuclear import. One possibility may be constitutive phosphorylation of Nrf2zip. There is a consensus mitogen-activated protein kinase site ( 559 PYSP 562 ) in EGFP-Nrf2zip as well as in EGFP-Nrf2-NES-GAL4DBD constructs. In our laboratory we found that a GST-tagged C-terminal segment of Nrf2 was constitutively phosphorylated. 4 Because all of the NLS motifs described to date are tandems of positively charged amino acids, it was proposed that negatively charged phosphorylated groups can deter nuclear import. This hypothesis was proven for the transcription factor NFAT1. NFAT1 is constitutively phosphorylated under basal conditions, and dephosphorylation can indeed facilitate its translocation to the nucleus (55). Further studies will be needed to determine whether phosphorylation of Nrf2 has a role in nuclear translocation. For example, treatment with specific mitogen-activated protein kinase inhibitors or co-expression of phosphatases with EGFP-Nrf2zip NES mutants might result in the accumulation of Nrf2zip in the nucleus, supporting this model. Alternatively, substitution of the phosphate acceptor serine 561 with a neutral amino acid may help to solve this puzzle.
In our present study the subcellular distribution pattern of EGFP-Nrf2zip appeared not to be affected by treatment with oxidants (SUL and DEM) or reducing compounds (NAC and GSH). However, more experiments using other structurally different redox compounds would be needed to conclude that this Nrf2-NES is truly redox-insensitive. Furthermore, other putative NLS and NES in Nrf2 may also need to be examined for redox sensitivity. However, our current results indicating that the Nrf2-NES located in the ZIP domain is redox-insensitive are in striking contrast to the previous finding that the conditional nuclear export of CNC/bZIP protein Bach 2 was mediated by a redox-sensitive NES (41). Igarashi and co-workers (41) illustrated that the redox sensitivity of the NES in Bach 2 was attributed to two key cysteine residues. The fact that there is no cysteine residue in our newly identified Nrf2-NES or the whole segment of Nrf2zip may explain the difference between Nrf2-NES and the Bach 2-NES. If the NES in the ZIP domain as well as other putative NES motifs in Nrf2 are truly redox-insensitive, then Nrf2 may exemplify a different redox-responsive activation mechanism.
Nrf2 signaling, when triggered by oxidants, is a multistep activation process. It is well established that the initial step, the release of Nrf2 from Keap1 retention, could be a redoxsensitive step, at least in the in vitro setting (29,30). If the subsequent steps are redox-insensitive, it could underline the vital importance of the initial step as the rate-limiting step for Nrf2 activation. This may also explain in part why the ablation of the Keap1 gene, which leads to loss of control of Nrf2 activation, is lethal in the mice (56). So far our knowledge on the redox sensitivity of Nrf2 is still quite limited. Recently, a redox-sensitive cysteine was identified in the basic region right in the middle of the bipartite NLS and DNA binding domain (57). Mutation of this cysteine residue did not alter the Keap1 retention and nuclear translocation of Nrf2. However, the mutation attenuated the DNA binding capability of Nrf2 (57). The dissection and analysis of redox sensitivity of individual functional elements of Nrf2 in the future may help to draw a more complete picture of Nrf2 activation triggered by oxidative stress.
The identification of a functional NES in Nrf2 will not only help us to understand the activation of Nrf2, it may also help us decipher the mechanism of deactivation of Nrf2. Although all the efforts up to now have focused on the activation of Nrf2, the effective termination of Nrf2 signaling in the nucleus may also be equally important. Hyperactivity of Nrf2 and a long-lasting antioxidant response may be equally harmful as to the failure to respond to oxidant attacks. Indeed, Nrf2 has a very rapid turnover rate. The measured half-life of Nrf2 is as short as 15 min (58,59). Accumulating evidence shows that after fulfilling its transactivation function, Nrf2 is destined for proteasomal degradation in the cytoplasm (58,59), although some weak degradation activity may also exist within the nucleus (60). Therefore, Nrf2 signaling can be turned on and off rapidly to match rapid changes of the redox status of the cells. Our present study has identified Nrf2-NES as a potential candidate for the deactivation signal of Nrf2. In the process of nuclear localization, the Nrf2-NES could be masked, probably by heterodimer formation with MafG/K via the ZIP domain. After Nrf2 fulfills its transactivation function in the nucleus, the mechanism in exposing the Nrf2-NES, leading to its nuclear export, remains to be elucidated. It will certainly be an interesting topic for future study.
In summary, the present study identified a functional NES in Nrf2 localized in the leucine zipper dimerization domain. It binds to CRM1 and is sensitive to LMB treatments. It could also exclude nuclear protein Gal4DBD into cytoplasm. Furthermore, we also found that this Nrf2-NES appeared to be redoxinsensitive. Because the nucleocytoplasm translocation of transcription factors is the consequence of a dynamic equilibrium of multivalent NLS and NES, the characterization of the NES in Nrf2 is the first step to delineate the complex mechanisms underlying the nuclear import and export of Nrf2, which has a critical impact on the transcription regulation of ARE-mediated cellular cytoprotective genes.