Nrf2 Possesses a Redox-sensitive Nuclear Exporting Signal in the Neh5 Transactivation Domain*

NF-E2-related factor 2 (Nrf2) is the key transcription factor regulating the antioxidant response. Previous studies identified a nuclear localization signal (NLS) in the basic region and a nuclear exporting signal (NES) in the leucine zipper domain of Nrf2. In this study, we characterize a new functional NES (175LLSIPELQCLNI186) in the transactivation (TA) domain of Nrf2. A green fluorescence protein (GFP)-tagged Nrf2 segment (amino acids162-295) called GFP-NESTA exhibited a cytosolic distribution that could be disrupted by L184A mutation or leptomycin B treatment. Chimeric expression of this NESTA with a nuclear protein GAL4DBD could expel GAL4DBD into the cytoplasm. A variety of oxidants, including sulforaphane, tert-butylhydroquinone, and H2O2, could effectively induce nuclear translocation of GFP-NESTA. Mutational studies showed that cysteine 183 may mediate the redox response of NESTA. The discovery of multiple NLS/NES motifs in Nrf2 and the redox sensitivity of NESTA imply Nrf2 may be self-sufficient to sense and transduce oxidative signals into the nucleus, consequently initiating antioxidant gene transcription.

Antioxidant response is an important cytoprotective reaction. When exposed to oxidative stress, mammalian cells can respond with a rapid and coordinated expression of diverse antioxidant genes such as heme oxygenase 1 (HO-1) 3 (1), ␥glutamyl-cysteine synthetase/ligase, NAD(P)H:quinone oxidoreductase 1 (2)(3)(4), and the phase II detoxifying enzymes such as glutathione S-transferase (GST) (2,3) and UDP-glucuronosyltransferase (4). Pivotal to the antioxidant response is a tran-scription factor called Nrf2 (nuclear factor-erythroid 2-related factor 2) (5). Nrf2 knock-out mice display reduced constitutive and inducible expression of HO-1 and GST (6 -9). Nrf2 null mice are also more susceptible to the chemical toxin treatments (10,11). Nrf2 is a member of the basic leucine zipper (bZIP) transcription factor subfamily featuring a Cap 'n Collar motif (5). Like many transcription factors, Nrf2 signaling is regulated by subcellular localization (12). Under homeostatic conditions, Nrf2 molecules are predominantly sequestered in the cytoplasm. When stimulated with oxidative or electrophilic compounds, Nrf2 proteins can quickly translocate to the nucleus and form heterodimers with bZIP proteins called small Maf (musculoaponeurotic fibrosarcoma) proteins (13,14). Nrf2/ Maf heterodimer formation can enhance the specificity and binding of Nrf2 to a cis-acting enhancer called antioxidantresponsive element (ARE) (15) located in the promoter of many antioxidant and phase II detoxification genes (16 -19).
Until now, Nrf2 signaling is generally considered to be regulated by a cysteine-rich protein called Keap1 (Kelch-like ECHassociated protein 1) (20,21). Nrf2 forms a dimer with Keap1 in vitro (22) and probably in vivo as well. An ETGE motif (17) and a DLG motif (23) of the Neh2 (Nrf2 ECH homology 2) domain of Nrf2 have been elucidated to mediate cooperative binding with the Kelch/double glycine repeat domain of Keap1 (19,24). Keap1 is an actin-binding protein (19). Treatments with compounds to dissolve the cytoskeleton can result in Nrf2 nuclear accumulation (25). So the cytosolic sequestering of Nrf2 is attributed to Keap1 retention. Keap1 is also identified as a Cullin 3-dependent adaptor protein for ubiquitin ligase ubiquitinprotein isopeptide ligase (26 -28). Therefore, Nrf2 molecules may not only be sequestered by Keap1 but also subjected to constant degradation. In vitro, the Nrf2/Keap1 dimer can only be formed under reducing conditions (22) and can be disrupted by the treatment of phyto-oxidant sulforaphane or the phenolic compound tert-butylhydroquinone (tBHQ) (22). Some cysteine residues in Keap1, such as Cys-151, Cys-273, and Cys-288, are found to be critical in Nrf2 retention and release (24,29). Based on these observations, it is hypothesized that Keap1 functions as a redox-switch for Nrf2 signaling (22). Whereas reducing conditions favor Keap1 retention of Nrf2, oxidative signals can induce Keap1 to quickly release Nrf2, leading to nuclear translocation of Nrf2 (20).
Although the Keap1 anchoring model seems to successfully explain the repression and activation of Nrf2 signaling in response to the changing redox conditions, some controversial observations are reported recently. It is reported that the cytosolic distribution of Keap1 is maintained by active nuclear export rather than cytoskeleton anchoring (30,31). In certain cell lines, such as hepatoma HepG2 and H4IIEC3 cells, remarkably high amounts of endogenous Nrf2 are found in the cell nucleus at unstimulated conditions (32). Even if the Keap1 anchoring model is still applicable in these cells, the amount or activity of cytosolic Keap1 proteins may not be sufficient to sequester Nrf2. Furthermore, the basal expression of Nrf2driven antioxidant genes such as HO-1, the modifier subunit of glutamyl-cysteine ligase, and Prdx1 (peroxiredoxin I) remained unchanged in the livers of Keap1 knock-out mice (33). Conceptually, the necessity of Keap1 in Nrf2 signaling is based on the assumption that Nrf2 is unable to maintain cytosolic segregation by itself under quiescent conditions. In addition, the redox-switching role of Keap1 assumes that Nrf2 per se lacks redox responsiveness, or the redox responsiveness Nrf2 is functionally irrelevant to its nuclear translocation. To date, only one reactive cysteine is characterized in the basic region of Nrf2 (34). Mutation of this cysteine to a serine only alters its DNA binding affinity but fails to alter the subcellular localization of Nrf2 (34). These data therefore seem to provide further support to those assumptions. However, with the rapid progress in Nrf2 studies, those assumptions may merely reflect our limited knowledge about Nrf2. Recent studies have identified a nuclear export signal (NES) in the ZIP domain of Nrf2 (NESzip) and a bipartite nuclear localization signal (NLS) in the basic region of Nrf2 (bNLS) (35,36). These discoveries raise the question whether Nrf2 possesses other functional NES or NLS motifs. In this study, we identify a new functional NES located in the transactivation (TA) domain of Nrf2. The existence of multiple NES and NLS motifs in Nrf2 enables Nrf2 to maintain cytosolic segregation by itself under quiescent conditions. Furthermore, unlike the NESzip, this NES TA possesses a reactive cysteine residue (Cys-183). An enhanced green fluorescence protein (EGFP)-tagged Nrf2 segment (amino acids 162-295), called EGFP-NES TA , exhibited a dosage-dependent nuclear translocation when treated with sulforaphane. Therefore, Nrf2 can sense the redox signal and translocate to the nucleus. These discoveries suggest Nrf2 may be able to transduce redox signals in a Keap1-independent manner. Keap1, however, may provide an additional regulation of the quantity of Nrf2 both at basal and at inducible conditions.

EXPERIMENTAL PROCEDURES
Cell Culture and Chemicals-Human cervical squamous cancerous HeLa cells were obtained from the ATCC (Manassas, VA). HeLa cells were cultured as monolayer using minimum essential 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. Reducing GSH, hydroperoxide (H 2 O 2 ), diethyl maleate (DEM), and propidium iodide (PI) were purchased from Sigma. Sulforaphane was purchased from LKT Laboratories   To examine the two NES-like  motifs individually, two segments of Nrf2, NES TA (amino acids  162-295) and NES INT (amino acids 288 -485), were PCR-amplified and subcloned into pEGFP-C1 vector (Clontech) by XhoI and BamHI digestion. The resulting plasmids were designated EGFP-NES TA and EGFP-NES INT , respectively. For fluorescence resonance energy transfer (FRET) examination, the PCRamplified NES TA segment was inserted into the pECFP-C1 vector (Clontech) by XhoI/BamHI digestion. The resulting plasmid was designated as ECFP-NES TA . As a positive control, N-terminal segment (amino acids 1-174) of Nrf2 was also PCRamplified and subcloned into the pECFP-C1 vector. The resulting plasmid was designated ECFP-Nrf2 NT . To examine the interaction between NES TA and Keap1, full-length Keap1 was PCR-amplified and subcloned into the pEYFP-N1 vector (Clontech). The resulting plasmids were designated EYFP-Keap1. For NES TA ectopic expression, GAL4 DNA binding domain (DBD) was PCR-amplified and subcloned into the ECFP vector or inserted into ECFP-NES TA plasmid by PstI/ BamHI digestion to form an ECFP-GAL4DBD or ECFP-NES TA -GAL4DBD chimera. For mutation studies, the key leucine residue of the NES TA (Leu-184) or the reactive cysteine 183 (Cys-183) was substituted with alanine. The mutagenic primers were synthesized and PAGE purified by Integrated DNA Technologies (Coralville, IA). Mutagenesis reactions were performed using QuikChange XL mutagenesis kit purchased from Stratagene (La Jolla, CA) according to the protocol described previously (35). Mutations were verified by DNA sequencing. The method of siRNA knockdown of Keap1 has been described in detail before (50) with few modifications. The sequence of siRNA primers is 5Ј-GGCCUUUGGCAUCAU-GAACTT-3Ј (sense) and 5Ј-GUUCAUGAUGCCAAAGGC-CTG-3Ј (antisense). The primers were synthesized, and high pressure liquid chromatography-purified by Integrated DNA Technologies. The siRNA primers were annealed to form double strand Keap1 targeting siRNA with TT and TG overhangs and transfected into HeLa cells using the Lipofectamine method.
Transient Transfection and Reporter Gene Activity Assays-Transactivation activity assay has been described in detail before (37). Briefly, HeLa cells were plated in 6-well plates at ϳ4.0 ϫ 10 5 cells/well. Twenty four hours after plating, cells were transfected using the Lipofectamine method according to manufacturer's instructions. For each well, 200 ng of ECFP-NES TA -GAL4DBD or its C183A mutant and 50 ng of GAL4-Luc reporter were added into 125 l of Opti-MEM. Lipofectamine 2000 (Invitrogen) was added into another tube of 125 l of Opti-MEM in a 1:2.5 ratio to the amount of plasmids 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. Cells were incubated with transfection complexes for 3 h, changed to fresh minimum essential medium, and cultured for 16 h. Cells were treated with 5 and 12.5 M sulforaphane for an additional 6 h before harvesting. Cells were washed twice with phosphatebuffered saline (PBS), scraped, and incubated in reporter lysis buffer (Promega) on ice for 30 min. After centrifugation, 10 l of lysate was mixed with luciferase substrate (Promega), and the ARE-luciferase activity was measured using a Sirius luminometer (Berthold Detection System). Protein concentration was measured using the Bradford method. Luciferase activity was normalized by protein concentration.
Cell Fractionation-The protocol to extract nuclear and cytoplasmic proteins has been described before (38) with minor modifications. Briefly, HeLa cells were cultured in 60-mm Petri dishes and transfected with 5 g of plasmids using the Lipofectamine method (Invitrogen). After 24 h, cells were rinsed with ice-cold PBS and harvested with cell lysis buffer A (50 mM Tris-HCl, 10 mM NaCl, 5 mM MgCl 2 , 0.5% Nonidet P-40, pH 8.0). After incubation on ice for 10 min, the samples were centrifuged at 12,000 ϫ g for 15 min. Supernatants (cytosolic extract) were collected. Nuclear pellets were washed twice with cell lysis buffer A and then resuspended in high salt buffer B (20 mM HEPES, 0.5 M NaCl, 1 mM EDTA, 1 mM dithiothreitol, pH 7.9), vortexed, and centrifuged. Supernatants (nuclear extract) were collected. The protein concentration of each sample was measured. To generate homogeneous electrophoretic patterns, cytosolic proteins were diluted in buffer B. 20 g of nuclear proteins and 10 g of cytosolic proteins were loaded for immunoblot analysis.
Immunoprecipitation and Western Blotting-Twenty four hours after transient transfection of ECFP-Nrf2 NT , ECFP-NES TA , and EYFP-Keap1, HeLa cells were harvested 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 (IP), 50 g of cell lysates expressing ECFP-Nrf2 NT or ECFP-NES TA were mixed with 100 g of cell lysate expressing EYFP-Keap1. The mixture was incubated with 20 l of protein A-conjugated Sepharose 4B beads (Zymed Laboratories Inc.) to remove endogenous IgG. Solutions were then incubated with 1 g of mouse anti-GFP antibody at 4°C for 1 h, subsequently mixed with 40 l of protein A-Sepharose beads, and tumbled at 4°C overnight. The IP solutions were then centrifuged. Pellets were washed three times with the lysis buffer. Protein complexes were eluted by loading buffer and heated to 80°C for 5 min and subsequently analyzed via Western blotting. For Western blotting, either the IP products or lysates containing 20 g of protein were resolved by 4 -15% linear gradient SDS-PAGE (Bio-Rad) 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 (TBST) containing 20 mM Tris-HCl, 8 mg/ml NaCl, and 0.2% Tween 20, pH 7.6, at room temperature for 1 h. The membrane was probed with polyclonal rabbit anti-Nrf2 (H-300) (1:750), goat anti-Keap1 (1:750), anti-lamin A (1:250), and anti-␤-actin (1:1000) in 3% nonfat milk TBS-T at 4°C overnight. After washing three times with TBST, the membrane was blotted with peroxidase-conjugated secondary antibody (1:5000 dilution) at room temperature for 1 h. Proteins were visualized using the ECL mixture from Bio-Rad.
Epifluorescent Microscopy-Expression and subcellular distribution of EGFP-NES TA/INT or their mutants in response to treatments of LMB or oxidative compounds were examined using a Nikon Eclipse E600 epifluorescent microscope and a Nikon C-SHG1 UV light source purchased from Micron-Optics (Cedar Knolls, NJ). HeLa cells were cultured on ethanolsterilized glass coverslips and transfected with 1 g of EGFP-NES TA or its mutants by using the Lipofectamine method (Invitrogen) and cultured in minimum essential medium for 24 h. Prior to microscopic examination, some coverslips were briefly fixed in methanol for 2 min, rinsed three times in PBS, and then counterstained with PI (5 g/ml PI in PBS solution supplemented with 100 g/ml RNase A) to visualize the positions of cell nuclei. The EGFP signals were examined using a fluorescein isothiocyanate filter, and the PI signals were examined with a Texas Red filter. The epifluorescent images were digitized using a Nikon DXM1200 camera and Nikon ACT-1 software (version 2). Images were superimposed using Adobe Photoshop CS version 8.0. For the kinetics study, the cell numbers exhibiting whole cell or cytosolic distribution pattern in response to sulforaphane treatments were counted. The ratio of whole cell distribution was calculated and plotted against the durations of sulforaphane treatment using a curve fitting software (Origin 7.0) based on the Michaelis-Menten equation.
Confocal Microscopy and FRET Assay-To examine ectopic expression, HeLa cells were cultured in glass bottom dishes (MatTek, Ashland, MA) and transfected with plasmids expressing ECFP-GAL4DBD, ECFP-NES TA -GAL4DBD, or its L184A mutant. For FRET assay, HeLa cells were transfected with plasmids expressing ECFP-Nrf2 NT , ECFP-NES TA , and EYFP-Keap1. Twenty four hours after transfection, cells were examined using a Zeiss LSM510 laser scanning confocal microscope (Zeiss, Thornwood, NY) with a ϫ63 water-immersion objective. We used a sensitized emission method for the FRET assay (39,40). Three filter sets were used to detect the donor (ECFP), acceptor (EYFP), and FRET signals. The FRET signal is corrected for spectral bleed through and contamination of donor and acceptor fluorescence according to Youvan's models (40) as indicated in Equation 1, The abbreviations uses are as follows: Fc is FRET concentration; bg is background intensity; cf is correction factor; fret is FRET signal; don is donor signal; and acc is acceptor signal.
The FRET concentration was normalized to donor and acceptor concentrations according to Equation 2, For data acquisition, the donor (ECFP) channel was excited with an argon laser line at 457 nm, and the emission was detected using a bandpass filter of 475-525 nm. The acceptor (EYFP) channel was excited at 543 nm and its emission was detected at 545-600 nm. The FRET channel was excited at 457 nm, and the emission was detected at 545-600 nm. For data analysis, we used the LSM510 SP2 software (version 3.2) to subtract donor and acceptor bleed through and normalize against acceptor (EYFP) and donor (ECFP) intensity.

RESULTS
Dissecting Individual NES-like Motif of Nrf2-Cytoplasmonuclear transportations are mediated by the importin and exportin proteins. They recognize a specific nuclear localization signal (NLS) or nuclear export signal (NES) on cargo proteins (41). Consensus leucine-rich NES motifs can be formulated as ⌽ 4 (X) 2-3 ⌽ 3 (X) 2-3 ⌽ 2 X⌽ 1 , where ⌽ represents hydrophobic amino acids, and X represents any amino acid (42). To be a valid NES, ⌽ 1 and ⌽ 2 must be hydrophobic amino acids, and the positions of ⌽ 3 and ⌽ 4 may wobble (42). In Nrf2, besides the documented NESzip (35), there are several NES-like motifs. In this study, we examined one NES-like motif ( 175 LLSIPELQCLNI 186 ) located in the Neh5 transactivation domain (TA) and one NES-like motif ( 296 LLNGPIDVS-DLSL 308 ) located in the intervening domain (INT), with the ⌽-corresponding residues underlined. Both NES-like motifs have legitimate hydrophobic residues at the ⌽ 1-3 positions. The ⌽ 4 residue for NES TA is isoleucine 178. The ⌽ 4 residue of NES INT is a glycine 299 residue or a proline 300 residue; none are long chain hydrophobic amino acids. Close to the Gly-299 residue, there are tandem leucines (residues 296 -297) and an isoleucine residue (301). To examine whether these two NESlike motifs are functional, we made two enhanced green fluorescence protein (EGFP)-tagged chimeras, EGFP-NES TA containing the TA segment of Nrf2 (amino acids 162-295) and EGFP-NES INT containing the INT segment of Nrf2 (amino acids 288 -485) (Fig. 1). These two constructs were so designed that they contained no other consensus NES, NLS, or NES/NLS-like motifs. In addition, the elucidated Keap1binding motifs were also carefully excluded.
Nrf2 Possesses a Functional NES Located in the TA Domain-When expressed in HeLa cells, EGFP-NES TA exhibited a predominantly cytosolic distribution ( Fig. 2A). The green fluorescence of EGFP-NES TA was found confined in the cytoplasm, with cell nuclei devoid of fluorescence ( Fig. 2A). The positions of cell nuclei were also confirmed by PI counter-staining ( Fig. 2A) superimposed with green fluorescent image ( Fig. 2A). Cell percentage assay showed that EGFP-NES TA proteins were cytoplasmic in 83.5% of the transfected cells (Table 1). In the remaining cells (16.5%), an evenly distributed pattern was observed (Table 1). There was virtually no cell showing nuclear distribution. In contrast to NES TA , NES INT did not appear to be a functional NES. In HeLa cells, EGFP-NES INT is localized to both the cytoplasm and the nucleus ( Fig. 2A). Therefore, despite the similarity of NES INT to the canonical NES, NES INT probably is a pseudo-NES. To determine whether the cytosolic distribution of EGFP-NES TA is mediated by the NES TA motif, we performed site-directed mutagenesis of NES TA . Substitution of leucine 184 with alanine (L184A) was sufficient to convert the cytosolic distribution of wild type NES TA into a whole cell distribution pattern (Fig. 2B). All HeLa cells expressing EGFP-NES TA L184A exhibited a whole cell distribution pattern (Table 1). Furthermore, we examined whether the function of NES TA is mediated by binding with nuclear exporting protein, CRM1 (chromosome maintenance region 1) (43). One hour of treatment with LMB, a specific CRM1 inhibitor (44), could convert the distribution of EGFP-NES TA into a whole cell pattern ( Fig.  2B; Table 1). These data are in agreement with a previous study that LMB inhibited NESzip cytosolic localization (35). Combined, these data suggest that the observed cytosolic distribution pattern of EGFP-NES TA is likely maintained by active export in a CRM1-dependent manner.
Chimeric Expression of NES TA Can Expel GAL4DBD into the Cytoplasm-To examine whether the NES TA alone is sufficient to cause nuclear export, we fused the NES TA to the DNA binding domain of the yeast GAL4 protein (GAL4DBD), a well known nuclear protein (45). We constructed an ECFP-tagged chimerical protein by inserting the NES TA segment to the N terminus of GAL4DBD (Fig. 1). When expressed alone, ECFP-GAL4DBD was nuclear in 65.1% of the transfected cells (Fig.  2C, arrowheads; Table 1). In contrast, more than 80% of cells expressing ECFP-NES TA -GAL4DBD chimera exhibited a cytosolic distribution pattern ( Fig. 2C; Table 1). A point mutation in the leucine residue (Leu-184) in the NES TA abolished the cytosolic distribution and restored the nuclear distribution (Fig. 2C, arrowheads; Table 1). In addition, LMB treatments could also convert the cytosolic distribution of NES TA -GAL4DBD into a nuclear distribution pattern (data not shown). Because the GAL4DBD possesses an innate NLS (45), these data suggest that the NES TA is stronger than the NLS GAL4DBD , and NES TA could expel a heterologous nuclear protein GAL4DBD into the cytoplasm. The observation that L184A mutation and LMB perturbation of NES TA -mediated exporting could restore the nuclear distribution of GAL4DBD provides further evidence suggesting the disabled NES TA might give way to the NLS GAL4DBD activity.
The NES TA Is Redox-sensitive-Because Nrf2 plays the pivotal role in transducing oxidative signals, we next asked whether NES TA -mediated nuclear exporting activity can be modulated by redox compounds. Based on the chemical structures, oxidative compounds, also known as phase II inducers, can be divided into 10 classes (46). In this study, we examined the redox reactivity of EGFP-NES TA to the treatments of H 2 O 2 (hydroperoxide), DEM (Michael acceptor), tBHQ (quinone), and sulforaphane (isothiocyanate). When EGFP-NES TA -expressing cells were treated with solvent dimethyl sulfoxide (Me 2 SO) as a negative control for 3 h, an uninterrupted cytosolic distribution was observed ( Fig. 3A; Table 2). In contrast, 1 h of treatment with 1 mM H 2 O 2 or 1 mM DEM could almost completely convert the cytosolic distribution of EGFP-NES TA to a whole cell pattern ( Fig. 3A; Table 2). A similar disruptive effect on nuclear exporting activity of the redox-sensitive NES of the bZIP protein Bach2 was reported using similar dosages (47). A pronounced translocation effect was also observed when EGFP-NES TA -expressing cells were treated with tBHQ (200 M) and sulforaphane (50 M) for 3 h ( Fig. 3A; Table 2). In contrast, the cytosolic distribution of EGFP-NESzip was not disrupted when treated with 1 mM DEM and 50 M sulforaphane (35). Common to the treatments of all four oxidants, virtually no nuclear condensation of EGFP-NES TA was observed, probably reflecting the fact that EGFP-NES TA lacks any NLS motif. So the maximal inducing effect elicited by oxidants was a 100% whole cell distribution. We also examined the effect of reducing compounds on EGFP-NES TA . After 3 h of treatment of 1 mM reducing GSH, a sulforaphane-conjugating agent, the cytosolic distribution of EGFP-NES TA was slightly increased ( Fig. 3A; Table 2), probably reflecting inhibition of basal activation of EGFP-NES TA . Co-treatment of 1 mM GSH could completely abolish the mobilizing effect of sulforaphane ( Fig. 3A; Table 2).
Because there is a cysteine residue (Cys-183) embedded in the NES TA motif, we mutated the Cys-183 residue to examine whether this cysteine residue is a reactive cysteine mediating the redox reactivity of the NES TA motif. The C183A mutant significantly attenuated the translocation effect exerted by H 2 O 2 , DEM, tBHQ, and sulforaphane as compared with wild type EGFP-NES TA . A considerable percentage of EGFP-NES TA   (Table  2), especially when treated with tBHQ and sulforaphane for 3 h (Fig.  3A, arrowhead). Although tBHQ and sulforaphane could mobilize wild type EGFP-NES TA into the cell nucleus in more than 70 and 80% cells, respectively, only ϳ20 and 50% of the cells expressing the C183A mutant exhibited whole cell distribution after 3 h of tBHQ and sulforaphane treatments ( To confirm that oxidants could induce EGFP-NES TA translocation, we did a cell fractionation assay. We found EGFP-NES TA proteins were mainly in the cytosolic extract after treating with Me 2 SO for 3 h (Fig.  3B). In comparison, 3 h of treatment with 50 M sulforaphane increased EGFP-NES TA nuclear localization (Fig. 4B).
We also did the reporter gene activity assay. Because the ECFP-NES TA -GAL4DBD chimera contained both the Neh5 transactivation domain (48) and the DNA binding domain of GAL4, we co-

Keap1-independent Nrf2 Signaling
expressed the ECFP-NES TA -GAL4DBD chimera with the GAL4-luciferase reporter and measured the luciferase induction. At basal conditions, expression of wild type ECFP-NES TA -GAL4DBD elicited high constitutive luciferase activity that could be significantly down-regulated by treatment with 1 mM GSH (data not shown). The luciferase activities induced by wild type ECFP-NES TA -GAL4DBD were significantly higher than the C183A mutant (Fig. 3C), suggesting the C183A mutant may be less responsive at the basal conditions. When we treated cells expressing wild type ECFP-NES TA -GAL4DBD with sulforaphane (5 or 12.5 M) for 6 h, there was pronounced luciferase induction in a dosagedependent manner (Fig. 3C). The folds of induction (1.85 Ϯ 0.10) mediated by the ECFP-NES TA -GAL4DBD chimera in response to 12.5 M sulforaphane treatments appeared a little bit lower than the induction folds (ϳ2.3) mediated by another Nrf2-GALDBD chimera N2 (amino acids 113-251) (37). The difference may be attributed to the fact that N2 contains both the Neh4 and Neh5 transactivation domains, whereas the NES TA only contains the Neh5 domain (48). In comparison with the wild type, only modest luciferase induc-tion was observed for C183A mutant in response to sulforaphane treatment (Fig. 3C), consistent with the hypothesis that Cys-183 is a redox sensor. NES TA Nuclear Translocation Kinetics-After observing NES TA was a redox-sensitive NES, we measured the kinetics of EGFP-NES TA nuclear translocation triggered by oxidants. Because the translocation effect exerted by sulforaphane was close to the physiologically achievable concentrations (49), we chose to study the translocation kinetics elicited by sulforaphane. As mentioned previously, the EGFP-NES TA construct lacks any NLS motif, and the maximal induction effect on EGFP-NES TA was a conversion from a cytosolic distribution to a whole cell distribution pattern. We employed the ratio of cells exhibiting whole cell distribution of EGFP-NES TA as an index of nuclear translocation. We treated EGFP-NES TA -expressing HeLa cells with 5, 12.5, 25, and 50 M sulforaphane at 37°C with 5% CO 2 supplement for various durations. We determined the ratio of cells showing uniform EGFP-NES TA localization patterns in response to sulforaphane treatments and plotted their values against the duration of treatments (Fig. 4A). The kinetics of sulforaphane-induced EGFP-NES TA translocation could be described by two parameters. One is the maximal accumulation (A max ) of EGFP-NES TA in the nucleus. The other parameter is the half-time to achieve maximal translocation (t1 ⁄2 ). The A max values delineated the magnitude of EGFP-NES TA translocation, whereas the t1 ⁄2 indicated the speed of translocation.
The EGFP-NES TA showed a dosage-dependent nuclear translocation in response to sulforaphane treatments (Fig. 4A). In the four tested concentrations, the A max values of EGFP-NES TA appeared to be positively correlated with the sulforaphane concentration ( Fig. 4B; Table 3), i.e. the stronger the oxidative signal, the more EGFP-NES TA will translocate into the nucleus. The t1 ⁄ 2 values are inversely correlated with sulforaphane concentrations ( Fig. 4B; Table 3), suggesting that the stronger the oxidative signal, the faster the Nrf2 influx.
Consistent with the cysteine-mediated redox sensitivity, the C183A mutation significantly slowed down nuclear translocation (Fig. 4A). The t1 ⁄ 2 value of the C183A mutant was almost two times longer than the wild type EGFP-NES TA when treated with 50 M sulforaphane (Table 3). Unlike the t1 ⁄ 2 value, the A max value of the C183A mutant was virtually unchanged compared with wild type EGFP-NES TA in response to 50 M sulforaphane treatment ( Table 3). The functional significance of this differential response of t1 ⁄ 2 and A max of the C183A mutant is unknown.
To exclude the possibility that the observed sulforaphaneinduced translocation effect is mediated by the EGFP tag, we also treated the EGFP-expressing cells with 50 M sulforaphane for 3 h. EGFP protein was nonresponsive to sulforaphane treatment (data not shown). These data suggest EGFP-NES TA translocation is specifically mediated by the NES TA . We also employed the CRM1-specific inhibitor LMB as a positive control. Treatment with 10 nM LMB elicited a rapid and complete translocation, with the A max ϭ 1.0 and the t1 ⁄2 ϭ 2.6 min (Table 3). These results appear much faster than the translocation kinetics elicited by sulforaphane  SEPTEMBER 15, 2006 • VOLUME 281 • NUMBER 37 treatments (Fig. 4A). The measured t1 ⁄2 of LMB treatment is also faster than the formerly reported time constant, measured at room temperature (35,47). The difference may due to the temperature (37°C) used in our experiments.

Keap1-independent Nrf2 Signaling
Redox-sensitive Translocation of EGFP-NES TA Is Keap1independent-The EGFP-NES TA construct did not contain any known Keap1-binding motif. However, to exclude the possibility that the observed redox-reactive NES TA translocation is mediated by an unknown Keap1-binding motif, we performed FRET assay (39,40). The FRET technique detects direct interaction between molecules. A fluorophore 1 (donor) is tagged on molecule A, and fluorophore 2 (acceptor) is tagged on molecule B, the putative binding partner of A. The emission spectrum of the donor overlaps with the excitation spectrum of the acceptor. If molecule A and B physically bind, the complex will bring the donor and the acceptor in close proximity (i.e. nanometer range); the induced FRET signal can be detected when donor alone is stimulated. The FRET technique can discern molecules that co-localize in the same subcellular compartment but have no direct binding. In this study, we used an ECFP as donor and an EYFP as acceptor. We added an ECFP tag on the NES TA segment (Fig. 1). As a positive control, we also added ECFP tag to an N-terminal segment of Nrf2 (Nrf2 NT , amino acids 1-174), which contains both the Keap1-binding ETGE motif (17) and the DLG motif (23). When expressed alone, EYFP-Keap1, ECFP-NES TA , and ECFP-Nrf2 NT all exhibited cytosolic distributions (data not shown). Co-expression of ECFP-NES TA with EYFP-Keap1 did not induce a FRET signal (Fig. 5, A and B). In contrast, co-expressing ECFP-Nrf2 NT with EYFP-Keap1 produced intense FRET signals (Fig. 5, A and B). Furthermore, immunoprecipitation assays also confirmed that EYFP-Keap1 bound to the ECFP-Nrf2 NT but not to the ECFP-NES TA (Fig. 5C). Thus, the redox-reactive EGFP-NES TA translocation appeared very likely to be a Keap1-independent event.
Multivalent NES/NLS Motifs Determined the Subcellular Distribution of Nrf2-The discovery of a new NES TA motif, in combination with the previously characterized NESzip and bNLS, suggests Nrf2 possesses multiple NES and NLS motifs. , and FRET were detected in three different channels. When 0.5 g/ml ECFP-Nrf2 NT was co-expressed with 0.5 g/ml EYFP-Keap1, intense FRET signals were detected. In contrast, virtually no FRET signals could be detected when ECFP-NES TA was co-expressed with EYFP-Keap1, indicating there is no interaction between NES TA and Keap1. The fluorescence intensity was indicated by the color-coded bar. Scale bar indicates 10 m. B, calculated FRET values after correction of spectral bleed through and contamination of donor and acceptor fluorescence. C, ECFP-Nrf2 NT could be immunoprecipitated with EYFP-Keap1. In contrast, ECFP-NES TA failed to be co-precipitated with EYFP-Keap1. Cell lysates (50 g) expressing ECFP-Nrf2 NT or ECFP-NES TA were mixed with 100 g of lysate-expressing EYFP-Keap1. The protein complex was precipitated by mouse anti-GFP antibody and detected by anti-Nrf2 and Keap1 antibody, respectively.

TABLE 3 Kinetics of SFN-elicited NES TA nuclear translocation
Kinetics study of sulforaphane (SFN)-induced nuclear translocation of EGFP-NES TA or its C183A mutant. CRM1-specific inhibitor LMB was used as a positive control. A max is the maximal accumulation of influx of EGFP-NES TA or its mutant in the nucleus. The t1 ⁄ 2 is the time constant of half-maximal influx. R 2 is the regression coefficient. The values of A max and t1 ⁄ 2 were calculated from more than 600 cells from three independent experiments.

Keap1-independent Nrf2 Signaling
For a cargo protein containing multiple NES/NLS motifs, the tendency of its cytoplasmonuclear transportation is determined by the summation of the activity of each individual NES and NLS. Previous studies show that the driving force of the NESzip is weaker than the bNLS (35,36). Indeed, an Nrf2 segment containing both the bNLS and the NESzip motif ( Fig. 1, EGFP-bNLS-NESzip) exhibited a mainly nuclear distribution ( Fig. 6A; Table 1). A cell fractionation assay also confirmed that nuclear immunoreactivity of EGFP-bNLS-NESzip was stronger (Fig. 6B). In comparison, the NES TA motif appeared to be a stronger NES. An Nrf2 segment containing both the NES TA and the bNLS motif (Fig. 1, EGFP-NES TA -bNLS) showed a predominantly whole cell distribution. More than 75% of cells showed whole cell distribution (Fig. 6A), and only 11.2% cells showed nuclear distribution (Fig. 6A, arrowheads), and 11.9% cells showed cytosolic distribution (Table 1). This distribution pattern suggests the NES TA motif may have equal driving force as the bNLS motif. In full-length Nrf2, the combined nuclear exporting activities of NES TA and NESzip appeared to be able to counterbalance the nuclear importing activities mediated by bNLS with or without other unidentified NLS. We observed a heterologous distribution pattern for EGFP-Nrf2. Nearly 60% cells showed whole cell distribution (Fig. 6A, arrows). Nearly 30% cells showed nuclear distribu-tion (Fig. 6A, arrowheads), and 12% cells showed cytosolic distribution (Fig. 6A). There was a higher percentage of EGFP-Nrf2-expressing cells showing nuclear distribution in comparison with EGFP-NES TA -bNLSexpressing cells ( Table 1). The NES TA -bNLS segment lacked the 162 amino acids in the N terminus of Nrf2. Whether this area contains any unidentified NLS is unknown. In addition, this Nterminal region also appeared to have strong redox reactivity 4 ; therefore, it may be more sensitive to basal oxidative stress. These observations may explainwhythereappearstobeahigher percentage of nuclear distribution in EGFP-Nrf2-than in EGFP-NES TA -bNLS-expressing cells.
In previous studies (35,36) and in this study, a point mutation of a single key leucine residue, i.e. Leu-544 in the NESzip motif and Leu-184 in the NES TA motif, was sufficient to nullify their nuclear exporting activity. Mutation of either the Leu-184 residue or the Leu-544 residue in full-length Nrf2 was sufficient to convert the heterologous distribution of EGFP-Nrf2 into a predominantly nuclear distribution pattern (Fig. 6, A and B; Table 1). In other words, both NES TA and NESzip motif are indispensable to counterbalance the nuclear importing activity mediated by the bNLS motif alone or with other unidentified NLS motifs.
Loss of function of the NES TA motif could naturally occur in the cell. Because the NES TA motif is redox-sensitive, the increased oxidative stress may disable the NES TA motif. However, the redox reactivity of NES TA , as observed in segment studies, may not prove functional in the full-length Nrf2 due to the difference of protein folding and solvent accessibility of oxidants. We further examined whether oxidants could mobilize wild type Nrf2 into the nucleus. At a low concentration of 50 M, a 10-min treatment of tBHQ or H 2 O 2 caused pronounced nuclear translocation of EYFP-Nrf2 ( Fig. 7A; Table 4). Similar results were reported in Hepa-1 cells that treatments with 50 M tBHQ can elicit GFP-Nrf2 nuclear translocation within 15 min (36). In MDA-231 cells, 16 h of treatment with 25 M tBHQ can elevate nuclear immunoreactivity of Nrf2 (24). When endogenous Keap1 proteins were knocked down (supplemental Fig. 1) by siRNA (50), the tBHQ-induced nuclear translocation of EYFP-Nrf2 appeared to be uninterrupted (Table 4), suggesting that tBHQ-induced Nrf2 nuclear translocation is mainly 4 W. Li, unpublished data.  SEPTEMBER 15, 2006 • VOLUME 281 • NUMBER 37 independent of Keap1. In comparison to the wild type Nrf2, the C183A mutant exhibited an attenuated translocating response. Under basal conditions, nuclear localized Nrf2 (Fig. 7A, arrowheads) decreased remarkably from 24.3% in wild type Nrf2 to 7.8% in the C183A mutant (Table 4). In parallel, significantly more C183A mutant-expressing cells (27.3%) exhibited the cytosolic localization (Fig. 7A, arrows) than cells expressing wild type Nrf2 (10.9%) ( Table 4). Attenuated ARE-luciferase induction was also observed in the C183A mutation (Fig. 7B). So the redox sensitivity to basal oxidative stress appears to be reduced in the C183A mutant. Furthermore, the C183A mutant is less responsive to oxidant treatments. A 10-min treatment of 50 M tBHQ could induce nuclear translocation in 45.5% of cells expressing wild type Nrf2, and only 19.9% of C183A-expressing cells showed nuclear translocation (Table 4). Unlike wild type Nrf2, the C183A mutant appears not to be responsive to H 2 O 2 treatment (Table  4). Our transactivation assay also showed that although tBHQ and H 2 O 2 could significantly enhance ARE-luciferase activities mediated by wild type Nrf2, virtually no induction was observed in the C183A mutant (Fig. 7B).

Keap1-independent Nrf2 Signaling
In summary, these data clearly show that under normal physiological conditions, the redox reactivity of the NES TA motif enables Nrf2 to detect oxidative signals and transmit them to the nucleus.

DISCUSSION
In this study, we identified a novel redox-sensitive NES TA motif located in the transactivation domain of Nrf2. The discovery of NES TA may draw a quite different picture of Nrf2 signaling.
Previous studies identified a NESzip and a bNLS motif in Nrf2 (35,36). Because the driving force of the bipartite bNLS is stronger than the NESzip (35,36), if Nrf2 only possesses a bNLS and a NESzip, it has a stronger tendency driving toward the nucleus and indeed would need Keap1 to anchor it in the cytoplasm during the unstimulated state. The existence of an extra NES TA motif with a strong driving force, however, may reverse the cytoplasmonuclear transport tendency of Nrf2. Our present study shows that the NESzip and the NES TA in combination can effectively counterbalance the activity of bNLS and provide sufficient nuclear export driving force to hold Nrf2 in the cytoplasm (Fig. 6). If this is true, Keap1 may be dispensable to maintain cytosolic sequestering of Nrf2.
Unlike the NESzip motif, this newly identified NES TA motif appeared to be redox-sensitive. Treatments with diverse oxidative compounds elicited EGFP-NES TA nuclear translocation ( Fig.  3; Table 2). There is a Cys-183 residue embedded in this NES TA motif, bearing resemblance to the reported redox-sensitive NES motifs of the yeast AP-1-like transcription factor (YAP-1) (51) and mammalian bZIP protein Bach2 (47). Our mutation analyses showed that NES TA redox sensitivity may be mediated by the Cys-183 residue. C183A mutation could remarkably slow down translocation kinetics (Fig. 4, A and B; Table 3) and attenuate the luciferase reporter gene activity (Fig. 3C). It is possible that direct sulfhydryl modification of the Cys-183 res- idue inhibits the access and binding of nuclear exportin CRM1 to the NES TA motif and consequently results in nuclear accumulation of EGFP-NES TA . Similar results have been reported in YAP-1 protein (51). Alternatively, intramolecular disulfide bond formation may also disable the NES activities (52). In our EGFP-NES TA construct, there is a highly conserved Cys-226 residue (Fig. 9). Whether the Cys-226 residue functions as a disulfide bond partner of Cys-183 requires further study. To examine these possibilities, it is necessary to detect the presence and type (53) of sulfhydryl adducts of NES TA using the liquid chromatography and tandem mass spectrometry method, which has been successfully used in characterizing the reactive cysteine residues of Keap1 in response to the treatments of tBHQ (54) and sulforaphane (55).
Unlike the CRM1-specific inhibitor LMB, which can covalently modify exportin CRM1 (44) and exert a nearly allor-none effect on NES TA nuclear translocation, sulforaphane could elicit NES TA nuclear translocation in a dosage-dependent way (Fig. 4B). Presumably due to high concentrations of endogenous glutathione, sulforaphane may only partially and reversibly disable the NES TA . The dosage-dependent response of NES TA to sulforaphane suggests that Nrf2 cannot only sense oxidative signals but also precisely transmit the "intensity" of oxidative signals to the nucleus and up-regulate gene transcription accordingly, particularly at low stress conditions.
Based on these observations, we propose a new Nrf2 signaling model (Fig. 8). The Nrf2 molecule possesses multivalent NES/NLS motifs, and their relative driving forces are represented by the size and direction of the arrows (see Fig. 8). Under the unstimulated conditions (Fig. 8A), the combined nuclear exporting forces of NES TA and NESzip could counteract the nuclear importing force of the bNLS. As a result, Nrf2 exhibits a predominantly whole cell distribution (Fig. 6A). Although the majority of Nrf2 molecules remain in the cytoplasm, the residual nuclear Nrf2 may account for the basal or constitutive Nrf2 activities. The observation of a small percentage of cells exhibiting nuclear and cytosolic distribution of Nrf2 (Fig. 6A) may reflect the hyper-and hypo-oxidative condition of individual cells, respectively. When challenged with oxidative stress (Fig.  8B), the redox-sensitive NES TA is disabled, but the redox-insensitive NESzip remains functional (35), and the bNLS motif may remain functionally uninterrupted (34). Because the driving force of the NESzip is weaker than the bipartite bNLS (Fig.  6), the nuclear importing force mediated by the bNLS prevails and triggers Nrf2 nuclear translocation (Fig. 8B).
This model may not only account for the repression and activation of Nrf2 signaling but may also account for the inactivation of Nrf2 signaling. Accumulating evidence shows that Nrf2 activation can consequently elevate the expression and enzymatic activities of ␥-glutamyl-cysteine synthetase/GST as well as the GSH level in cells (56,57). Considering the reversible nature of sulfhydryl modification by sulforaphane, the elevated GSH levels may favor the restoration of NES TA activity and trigger Nrf2 nuclear export. Further studies are needed to examine this possibility.
Unlike the previous Keap1 model, this new model hypothesizes a Keap1-independent Nrf2 signaling. Neither FRET nor the IP assay could detect NES TA /Keap1 interaction. Inhibiting endogenous Keap1 activity with siRNA also failed to change tBHQ-induced nuclear translocation of Nrf2 (Table 4). Definitive proof   (58). The importance of Keap1 may vary in a spatial-and temporal-dependent fashion. To some Nrf2-driven structural proteins, such as keratin K6 and loricrin of epidermal tissue (59), rigid Keap1 sequestering of Nrf2 appears to be vital during development. The Keap1 null mice were found to die of hyperkeratosis of the esophagus and forestomach before weaning (59). In contrast, the liver-targeted Keap1 knock out appeared to incur no severe phenotypic deficit. Although the activities of some phase II genes such as GST subunits were indeed up-regulated (33), the expression of some other antioxidant and phase II genes such as HO-1, modifier subunit of glutamyl-cysteine ligase, and peroxiredoxin I were unaffected (33). The importance of Keap1 regulation in Nrf2 signaling may depend on its actual expression level. Unfortunately, so far it is unknown whether there is tissue-or cell type-specific difference in the expression ratio of Nrf2 over Keap1. This will require further study. If the new NES TA plays an important function of redox switch for Nrf2 signaling, we asked whether this NES TA motif is conserved across species. When the amino acid sequences of all cloned Nrf2 molecules were aligned with the key ⌽ positions underlined, this NES TA motif is found conserved only in human, dog, bovine, chicken, and Zebrafish (Fig. 9). To our surprise, this NES TA motif appeared not to be conserved in the p45 protein, the rodent, and frog Nrf2 orthologs. In the ⌽ 1 position, a hydrophilic threonine was found in the p45 protein and rodent Nrf2. A polar asparagine residue was found in the frog Nrf2 (Fig. 9, light gray box). According to the formulation of consensus NES motifs (42), the presence of threonine or asparagine may be sufficient to disable the NES activity. Re-examining the human NES TA sequence, we found the NES TA motif may also orient in a reversed direction, as the key ⌽ positions are illustrated by italicized fonts and underlined by dotted lines (Fig. 9). In the reversed orientation, the NES TA motif appears to be conserved in all cloned Nrf2 molecules, including p45, rodent, and frog Nrf2 molecules (Fig. 9). Because the ⌽ 1 residues of functional NES motifs are usually leucine or isoleucine (42), one concern is that in p45 and rodent Nrf2, the ⌽ 1 residue is a phenylalanine (Fig. 9). Occasionally, atypical hydrophobic residues are found at the ⌽ 1 position of functional NES, such as a methionine was found at the ⌽ 1 position of RXR␣ (60). Future expressive assays are needed to examine whether these NES TA -like motifs are functional.
The Nrf2 molecule shares high homology and probably similar origin with Nrf1, including the N terminus (61). In contrast, the Nrf2 molecule shares poor homology with Nrf3 in the N terminus (62). It is very interesting to find that the NES TA motif is also strictly conserved in all cloned Nrf1 molecules, including rodent Nrf1 (Fig. 9).
With regards to the redox sensitivity, the residues corresponding to the reactive Cys-183 in human Nrf2 are conserved in all cloned Nrf2 molecules (Fig. 9), underscoring its potential importance for Nrf2 signaling. In contrast, this cysteine residue is absent in Nrf1 molecules (Fig. 9). The Cys-226-corresponding residue is highly conserved in Nrf1 and Nrf2 molecules ( Fig.  9), with the exception of frog Nrf2 and Zebrafish Nrf1 (unknown). Overall, the conservation difference in the Cys-183 and Cys-226 positions implies that the redox-reactive profiles may be different between Nrf1 and Nrf2 molecules.
In summary, this study identifies a novel NES motif in Nrf2 and shows Nrf2 per se is redox-sensitive. Nrf2 consists of a constitutively active bNLS-NESzip in tandem and a conditional NES TA motif. The NES TA motif functions as a redox-sensitive switch that can be turned on/off by oxidative signals and determine the subcellular localization of Nrf2. Nrf2 appears to be self-sufficient to transduce redox signals, whereas Keap1 may provide additional regulation of Nrf2 signaling. These discoveries may profoundly modify our mechanistic understanding of Nrf2 signaling. FIGURE 9. Sequence alignments of the NES TA motif. Positions of key residues (⌽) in the consensus NES motif, as formulated by ⌽ 4 (X) 2-3 ⌽ 3 (X) 2-3 ⌽ 2 X⌽ 1 , are denoted by the underlined bars. The newly identified human NES TA motif appears to be conserved (transparent boxes) in dog, bovine, chick, and Zebrafish Nrf2. However, the NES TA motif is not conserved (shaded box) in p45 molecule and mouse, rat, and frog Nrf2. In comparison, this NES TA motif is strictly conserved in all cloned Nrf1 molecules. However, the NES TA motif may be conserved in a reversed orientation as designated by dotted underlines and italicized ⌽ positions in all cloned Nrf2, including p45 and rodent Nrf2. The reactive cysteine residue corresponding to the Cys-183 in human Nrf2 is conserved in all cloned Nrf2 molecules (black box) but not in Nrf1 molecules. In comparison, the residue corresponding to Cys-226 in Nrf2 is conserved in nearly all cloned Nrf2 and Nrf1 molecules with the exception of frog Nrf2 and Zebrafish Nrf1 (unknown). In comparison, the NES-like motif of NES INT does not match the NES formula in both orientations.