GSK-3� Acts Upstream of Fyn Kinase in Regulation of NuclearExport and Degradation of NF-E2 Related Factor 2*
Received for publication, December 11, 2006, and in revised form, March 20, 2007 Published, JBC Papers in Press, April 2, 2007, DOI 10.1074/jbc.M611336200
Abhinav K. Jain and Anil K. Jaiswal1
From the Department of Pharmacology, Baylor College of Medicine, Houston, Texas 77030
NF-E2-related factor 2 (Nrf2) regulates expression and coor-dinated induction of a battery of chemoprotective genes inresponse to oxidative and electrophilic stress. This leads to pro-tection against oxidative stress and neoplastic diseases. Nuclearimport and export of Nrf2 play a significant role in control ofnuclear levels of Nrf2 and thus the expression of Nrf2 down-stream genes. Tyrosine kinase Fyn phosphorylates tyrosine 568of Nrf2 that leads to the nuclear export of Nrf2. In this study, weinvestigated theupstream factor(s) in regulationof Fyn andFyn-mediated nuclear export of Nrf2. The investigations shed lighton a novel mechanism of Nrf2 regulation in response to oxida-tive stress. We demonstrate that GSK-3� acts upstream of Fynkinase in control of nuclear export of Nrf2. Chemical and shortinterfering RNA-mediated inhibition of GSK-3� led to nuclearaccumulation of Nrf2 and transcriptional activation of the Nrf2downstream gene nqo1. Chemical and short interfering RNAinhibition of GSK-3� and Fyn individually and in combinationrevealed that both kinases follow the same pathway to regulatenuclear export of Nrf2. We further demonstrate that hydrogenperoxide phosphorylates tyrosine 216 of GSK-3�. This leads toactivation of GSK-3�. The activated GSK-3� phosphorylatesFyn at threonine residue(s). Phosphorylated Fyn accumulates inthe nucleus and phosphorylates Nrf2 at tyrosine 568. This leadsto nuclear export, ubiquitination, and degradation of Nrf2.
NF-E2-related factor (Nrf2)2 is a nuclear transcription factorthat binds to antioxidant-response element (ARE) and regu-lates expression and coordinated induction of a battery of che-moprotective genes in response to antioxidants, oxidants, andradiations (1). This induction involves a mechanism essentialfor cellular protection against oxidative and electrophilic stressand neoplastic diseases (1). Nrf2-nullmice are born normal andviable, indicating that Nrf2 is not required for development andgrowth of mice (2). Nrf2-null mice express significantly lowerlevels and no induction of chemoprotective proteins that
include NAD(P)H:quinone oxidoreductase 1 (NQO1), gluta-thione S-transferase Ya subunit (GST-Ya), �-glutamylcysteinylsynthetase, and heme oxygenase (HO-1) (1). These mice dem-onstrate slower wound healing and pulmonary emphysema inresponse to exposure to tobacco smoke (3, 4).A cytosolic inhibitor of Nrf2, INrf2 (inhibitor of Nrf2) or
Keap1, retains Nrf2 in the cytoplasm (5, 6). The INrf2-Nrf2complex serves as cellular sensor of oxidative and electrophilicstress (1). The cellular exposure to oxidants, antioxidants, andradiations antagonizes this interaction and leads to the releaseof Nrf2 from INrf2 (1). Nrf2 translocates into the nucleus andinduces the expression of chemoprotective proteins. Electro-philic adduction of selected cysteines in INrf2 and PKC phos-phorylation of Nrf2 are known to contribute to the release ofNrf2 from INrf2 (1). Disruption of INrf2 in mice leads to post-natal death, probably frommalnutrition resulting fromhyperk-eratosis in the esophagus and forestomach, presumablybecause of excessive nuclear accumulation of Nrf2 (7). There-fore, sustained nuclear accumulation of Nrf2 is lethal to thecells. Indeed, longer retention of Nrf2 in the nucleus leads toapoptotic cell death (8).The abundance of Nrf2 inside the nucleus is tightly regulated
by positive and negative factors that control nuclear import,binding to ARE, export, and degradation of Nrf2 under normaland stress (chemical/radiation) conditions (9–12). The earlyresponse to stress leads to nuclear import of Nrf2 resulting incoordinated activation of chemoprotective genes. The delayedresponse to stress is Fyn-mediated phosphorylation ofNrf2Y568 inside the nucleus (13). Tyrosine 568 phosphoryla-tion leads to nuclear export of Nrf2 (13). However, upstreamevents that control signal transduction fromoxidative and elec-trophilic stress to Fyn that phosphorylates Nrf2Y568 leading tonuclear export of Nrf2 remains unknown. In this study, weinvestigated the upstream factors that regulate Fyn and nuclearexport and degradation of Nrf2.
MATERIALS AND METHODS
Construction of Plasmids—The construction of pGL2B-NQO1-ARE and pcDNA-Nrf2 has been described previously(14). The construction of pcDNA-Nrf2-V5, pcDNA-Nrf2Y568A-V5, and pCMV-FLAG-mINrf2 was also described previously(11). Modified pCMV vectors were used to clone the FLAG-tagged Fyn protein. Mouse Fyn cDNA was amplified from theIMAGE clone obtained from ATCC using the following prim-ers: forward 5�-GCGCTCTAGAGAATTCGTCGAGACCAT-GGGCTGTGTG-3� and reverse 5�-CGCGGATCCGATATCC-AGGTTTTCACCAGGTTGGTA-3�. The PCR-amplified DNAcontainedXbaI andBamHI restriction sites at the 5� and 3� end,
* This work was supported by National Institutes of Health Grants RO1GM47466 and RO1 ES012265. The costs of publication of this article weredefrayed in part by the payment of page charges. This article must there-fore be hereby marked “advertisement” in accordance with 18 U.S.C. Sec-tion 1734 solely to indicate this fact.
1 To whom correspondence should be addressed: Dept. of Pharmacology,Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030. Tel.: 713-798-7691; Fax: 713-798-7369; E-mail: [email protected].
2 The abbreviations used are: Nrf2, NF-E2-related factor; ARE, antioxidant-re-sponse element; NQO1, NAD(P)H:quinone oxidoreductase1; Me2SO, di-methyl sulfoxide; MG132, proteasome inhibitor; Ub, ubiquitin; PP2, spe-cific inhibitor of Src kinases including Fyn kinase; RT, reverse transcription;siRNA, short interfering RNA; HRP, horseradish peroxidase; HA, hemagglu-tinin; PKC, protein kinase C; LDH, lactate dehydrogenase.
respectively. The amplified DNA was digested with XbaI andBamHI and subcloned into the FLAG vector digested with sim-ilar enzymes. The resultant plasmid was designated as pCMV-FLAG-mFyn. The pCMV-HA-Ub was received as a generousgift fromDr, ShigekiMiyamoto (Department of Pharmacology,University of Wisconsin Medical School).Cell Culture, Co-transfection of Expression Plasmids, and
Luciferase Reporter Assay—Human hepatoma (HepG2) cellswere grown inmonolayer cultures in 6-well plates in minimumessential medium-� supplemented with 10% fetal bovineserum. Transient transfections were done in cells grown to�50% confluence using the Effectene transfection reagent(Qiagen,Valencia, CA). Cellswere co-transfectedwith 0.2�g ofreporter construct (human NQO1-ARE-Luc) and 10 times lessquantities of firefly Renilla luciferase encoded by plasmid pRL-TK. Renilla luciferase was used as the internal control in eachtransfection. To analyze the effect of GSK-3� inhibitors onNQO1-ARE activity, the transfected cells were treated for 12 hwith the indicated inhibitor (lithium chloride or TDZD-8 orPP2) in the concentrations as indicated in the figures. Lithiumchloride was purchased from Sigma, and TDZD-8 and PP2were purchased from Calbiochem and were the highest purity
available. After the treatment for thespecified times, the cells were washedwith 1� phosphate-buffered salineand lysed in 1� Passive lysis bufferfrom the Dual-Luciferase� reporterassay system kit (Promega, Madison,WI).The luciferase activitywasmeas-ured using the procedures describedpreviously (15).siRNA Transfection—Mouse GSK-
3� siRNA, Fyn siRNA, and LaminA/C siRNA (control) were purchasedfrom Dharmacon and were trans-fected using Lipofectamine transfec-tion reagent (Invitrogen) followingthe manufacturer’s protocol. Forluciferase assay, mouse Hepa-1 cellswere co-transfected with 0.2 �g ofreporter construct (human NQO1-ARE-Luc) and 10 times less quan-tities of firefly Renilla luciferaseencoded by plasmid pRL-TK asdescribed above along with theindicated amounts of GSK-3� orFyn siRNA. The luciferase analysiswas done as described above. ForWestern analysis, Hepa-1 cells in100-mm plates were transfectedwith siRNA in different doses, andsubcellular fractionation and im-munoblotting were done as de-scribed below.Subcellular Fractionation and
Western Blotting—HepG2/Hepa1cells, seeded in 100-mm plates andtreated/transfected as displayed in
the figures, were washed twice with ice-cold phosphate-buff-ered saline, scraped in phosphate-buffered saline using a rubberpoliceman, and centrifuged at 500 rpm for 5 min. Biochemicalfractionation of the cells was done using the nuclear extract kit(ActiveMotif, Carlsbad, CA) following themanufacturer’s pro-tocol. For making total cell lysate, the cells were lysed in RIPAbuffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40,0.5% deoxycholic acid, 0.1% SDS, 1 mM phenylmethylsulfonylfluoride, and 1 mM sodium vanadate supplemented with prote-ase inhibitor mixture (Roche Applied Science), and phosphataseinhibitor mixtures I and II (Sigma)). The protein concentrationwas determined using the protein assay reagent (Bio-Rad). 100�g of total cell lysate or cytosolic or nuclear fractions wereresolved on a 10% SDS-polyacrylamide gel, Western-blotted,and probed with anti-Nrf2 antibody, anti-Fyn antibody (bothfrom Santa Cruz Biotechnology), anti-GSK-3� antibody (CellSignaling), anti-Tyr(P)-GSK-3 antibody (BIOSOURCE), anti-FLAG-HRP antibody (Sigma), or anti-NQO1 antibody (11). Toconfirm the purity of subcellular fractionations, the extractswere Western-blotted with cytoplasm-specific anti-lactatedehydrogenase (LDH) antibody (Chemicon International,Temecula, CA) and nuclear specific anti-Lamin B antibody
FIGURE 1. siRNA-mediated inhibition of GSK-3� leads to nuclear accumulation and stabilization of Nrf2.A–C, siRNA-mediated inhibition of GSK-3�. A and B, Western analysis. Hepa-1 cells were transfected withGSK-3� siRNA in concentrations as indicated. The transfected cells were harvested, subjected to subcellularfractionation to prepare cytosolic and nuclear extracts (Nuc Ext), or lysed to obtain total cell lysate. The cytosolicand nuclear extracts (A) and total cell lysate (B) were analyzed by Western blotting and probing with anti-GSK-3� and anti-Nrf2 antibody. The blots were also probed with anti-Lamin B (nuclear specific), anti-LDH(cytosol-specific), and �-actin (equal loading) antibodies. GSK-3� and Nrf2 levels were normalized to �-actinlevels by using QuantityOne Image software, and the fold amount of protein is plotted versus the amount ofGSK-3� siRNA. The densitometry results are presented as � S.E. of three independent experiments, and arepresentative blot is shown. C, ARE-luciferase assay. Hepa-1 cells co-transfected with NQO1-ARE luciferasereporter, firefly Renilla luciferase, and increasing concentrations of either control siRNA (left panel) or GSK-3�siRNA (right panel) were harvested and analyzed for luciferase activity. D, Western analysis. Hepa1 cells trans-fected with GSK-3� siRNA were lysed in RIPA buffer as in B and analyzed for NQO1 and GSK-3� protein levels byimmunoblotting. E, RT-PCR. In a similar experiment, total RNA was prepared from Hepa1 cells transfected as inD, and RT-PCR was done using the primers as indicated under “Materials and Methods.”
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
JUNE 1, 2007 • VOLUME 282 • NUMBER 22 JOURNAL OF BIOLOGICAL CHEMISTRY 16503
(Santa Cruz Biotechnology). The levels of protein on aWesternblot were quantitated by using QuantityOne Image software(ChemiDoc XRS, Bio-Rad) and normalized against properloading controls.RT-PCR—RNA was prepared from Hepa-1 cells transfected
with GSK-3� siRNA or treated with LiCl using “RNeasy minikit” (Qiagen). 100 ng of RNA was used for amplification in anRT-PCRusing the “SuperscriptOne-stepRT-PCRkit” (Invitro-gen) following the manufacturer’s protocol. The primers usedwere as follows: Nrf2 forward 5�-GAGATGAGCTTAGGGC-3�, Nrf2 reverse 5�-GTTTTTCTTTGTATCTGGCTTCTTG-3�; NQO-1 forward 5�- ATGGCGGCGAGAAGAGCCCTG-3�,NQO1 reverse 5�-TTATTTTCTAGCTTTGATCTGGTT-GTC-3�; glyceraldehyde-3-phosphate dehydrogenase forward5�-ACCACAGTCCATGCCATCAC-3�, glyceraldehyde-3-phos-phate dehydrogenase reverse 5�-TCCACCACCCTGTTGCT-GTA-3�. The amplified products were analyzed by resolving ona 2% agarose gel.Phosphorylation Analysis—HepG2 cells were seeded in
100-mm plates and transfected with 0.5 �g of pCMV-FLAG-mFyn plasmid. 24 h after transfection, the cells were pretreatedwith LiCl/PP2 for 8 h followed by treatment with hydrogenperoxide (H2O2) � LiCl/PP2 for 1 and 4 h. At the end of treat-ment, the cells were harvested, and biochemical fractionationwas done following the procedures described above. To analyzetyrosine phosphorylation of GSK-3�, 100 �g of total cell lysate
was resolved on 10% SDS-PAGEand immunoblotted with phos-phospecific (Tyr(P))-GSK-3 anti-body (BIOSOURCE). The samemembrane was reprobed with anti-GSK-3� and anti-�-actin antibod-ies. To analyze the Fyn phosphoryl-ation, 1 mg of nuclear extract wasused to immunoprecipitate eitherwith anti-Thr(P) antibody (Cell Sig-naling) or anti-FLAGM2 beads(Sigma). Briefly, nuclear extractsupplemented with 1 mM phenyl-methylsulfonyl fluoride, 1 mMsodiumvanadate, Ser/Thr phospha-tase inhibitor mixture (Sigma), andprotease inhibitor mixture (RocheApplied Science) was incubatedwith 2.5 �g of antibody overnight at4 °C with shaking. 40 �l of washedprotein A beads (Santa Cruz Bio-technology) were added and incu-bated for 1 h at 4 °C with shaking.The slurrywas centrifuged at 10,000rpm for 30 s, and the supernatantwas discarded. The beads werewashed twice with RIPA buffer. 25�l of SDS sample dye was added tothe beads and boiled, and immu-noprecipitates were resolved on10% SDS-polyacrylamide gel fol-lowed by immunoblotting with
anti-Ser(P) or anti-FLAG-HRP antibody.In Vitro Kinase Assay—Purified GSK-3�, phospho-GSK-3
peptide, and Fyn kinase were obtained from Upstate Biotech-nology, Inc. His-Nrf2 was purified from bacterial lysates usingHis-Trap columns (GE Healthcare). 100 ng of GSK-3 peptide,Fyn kinase, or His-Nrf2 were incubated with active GSK-3� inthe kinase assay buffer (20mMTris, pH 7.5, 10mMMgCl2, 5mM
dithiothreitol, 200 �M ATP, and 1.5 �Ci of [�-32P]ATP) for 30min at 30 °C. Phospho-GSK-3 peptide (3 kDa) was used as apositive control for in vitro GSK-3� kinase reaction. Reactionwas stopped by adding 2� SDS-gel loading dye followed byboiling for 5 min. The samples were then resolved on a 12%SDS-polyacrylamide gel followed by autoradiography.Ubiquitination ofNrf2—HepG2 cellswere seeded in 100-mm
plates and co-transfected either with pcDNA-Nrf2-V5 (1.0�g),pCMV-HA-Ub (0.5 �g), or pCMV-FLAG-mINrf2 (0.25 �g) indifferent combinations as indicated in the figure. The trans-fected cells were treated with either Me2SO or 50 mM
GSK-3� inhibitor LiCl or 20 �M proteasomal inhibitorMG132 for 5 h. The cells were lysed in RIPA buffer, and 100�g of protein were analyzed by SDS-PAGE, Western blot-ting, and probing with anti-V5, anti-FLAG, and �-actin anti-bodies. In a similar experiment, 1 mg of protein was immu-noprecipitated with anti-V5 antibody and immunoblottedwith anti-HA antibody.
FIGURE 2. Inhibition of GSK-3� by chemical inhibitors leads to nuclear accumulation and stabilization ofNrf2. A and B, Western analysis. HepG2 cells were treated with LiCl or TDZD-8 for 2 h in concentrations asindicated. The cells were harvested, subjected to subcellular fractionation to prepare cytosolic and nuclearextracts (Nuc Ext), or lysed to prepare total cell lysate. The cytosolic and nuclear extracts (A and B, left panels) andtotal cell lysate (A and B, middle panels) were analyzed by Western blotting and probing with anti-Nrf2 anti-body. The blots were also probed with anti-Lamin B (nuclear specific), anti-LDH (cytosol-specific) and anti-actin(equal loading) antibodies. A and B, right panels, ARE-luciferase assay. HepG2 cells were co-transfected withNQO1-ARE luciferase reporter and firefly Renilla luciferase as described under “Materials and Methods.” Thirtysix h after transfection, the cells were treated with either TDZD-8 (10, 25 or 50 �M in top panel) or lithiumchloride (10, 25, or 50 mM, bottom panel) for 8 h. Cells were harvested, lysed, and analyzed for the luciferaseactivity. The results are presented as � S.E. of three independent experiments, and each experiment was donein triplicate. C, Western analysis. Hepa1 cells were treated with GSK-3� inhibitor; LiCl were lysed in RIPA bufferas in B and analyzed for NQO1 protein levels by immunoblotting. D, RT-PCR. In a similar experiment, total RNAwas prepared from Hepa1 cells treated as in C, and RT-PCR was done as described in Fig. 1.
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
16504 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 22 • JUNE 1, 2007
After a series of experiments, we found that siRNA againstGSK-3� inhibited GSK-3� and led to nuclear accumulation ofNrf2 as analyzed by immunoblotting and probing with anti-GSK3� and anti-Nrf2 antibodies (Fig. 1A). The blots were rep-robed with anti-LDH (cytosol-specific) and anti-Lamin B(nuclear specific) antibodies to confirm the purity of cytosolicand nuclear fractions. GSK-3� is a serine/threonine kinase thatregulates vital processes of cell division and apoptosis (16).Interestingly, siRNA-mediated inhibition of GSK-3� also led tostabilization of Nrf2 (Fig. 1B) and induction of the Nrf2 down-stream gene nqo1 ARE-luciferase expression (Fig. 1C, rightpanel). The nuclear accumulation and stabilization of Nrf2 wasGSK-3� siRNA concentration-dependent. Nuclear accumula-tion ofNrf2 afterGSK-3� siRNA transfection resulted in higherNQO1 protein levels (Fig. 1D, left panel). This increase inNQO1 protein was because of Nrf2-mediated increased NQO1transcription as analyzed by RT-PCR analysis (Fig. 1E). Theincreased nuclear accumulation and stability of Nrf2 was notbecause of increased transcription of Nrf2 as Nrf2 transcriptlevels did not change after GSK-3� siRNA transfection (Fig.1E). In a related set of experiments, the treatment of cells withchemical based GSK-3� inhibitors, lithium chloride (LiCl) and
TDZD-8, showed similar results asobserved with GSK-3� siRNA(Fig. 2). Both inhibitors enhancednuclear accumulation and stabiliza-tion of Nrf2 in total cell lysates andincreased ARE-luciferase activity(Fig. 2, A and B). Lithium chloride-mediated inhibition of GSK-3�enzyme activity also led to increasedNQO1 protein (Fig. 2C) and geneexpression without affecting Nrf2transcription (Fig. 2D). Theseresults suggested that a decrease inGSK-3� protein or inhibition ofGSK-3� activity led to nuclear accu-mulation and stabilization of Nrf2without affecting its transcription.The nuclear accumulation of Nrf2in response to GSK-3� inhibitors isbecause of blocking of nuclearexport of Nrf2 as reported previ-ously (17).Earlier, we have shown that Fyn
kinase-mediated phosphorylationof Nrf2Y568 is essential for thenuclear export of Nrf2 (13). There-fore, the treatment of human hepa-toblastoma (HepG2) cells with theFyn inhibitor PP2 led to nuclearaccumulation and stabilization ofNrf2 due to blocking of nuclearexport of Nrf2 (13). The inhibitionof GSK-3� also led to nuclear accu-mulation and stabilization of Nrf2due to the inhibition of nuclear
export of Nrf2 (Fig. 1). This raised questions if GSK-3� isupstream to Fyn because Fyn directly phosphorylatedNrf2Y568 or GSK-3� and Fyn regulated nuclear export of Nrf2by two independent mechanisms? We performed experimentsto determine whether GSK-3� inhibitor LiCl and Fyn inhibitorPP2 act independent of each other or in concert leading tonuclear accumulation of Nrf2. HepG2 cells were treated withdifferent doses of either LiCl or PP2 or both in combination;cytosol and nuclear fractions were prepared and analyzed bySDS-PAGE, immunoblotting, and probing with Nrf2 antibody(Fig. 3A). The blot was stripped and reprobed with anti-LDH(cytosolic marker) and anti-Lamin B (nuclear marker) antibod-ies. The cytosolic fractions had undetectable amounts of Nrf2,therefore not shown. The results demonstrated that Nrf2 accu-mulated in the nucleus in a PP2 dose-dependent manner (Fig.3A, lanes 1–3). Low and high dose LiCl treatment also led todose-dependent nuclear accumulation of Nrf2 (Fig. 3A, com-pare lanes 1, 4, and 7). However, when LiCl was included with alow dose of PP2, Nrf2 followed PP2 dose-dependent nuclearaccumulation (Fig. 3A, lanes 4–6). Finally, a combination ofhigher doses of PP2 (1 �M) with different doses of LiCl couldnot result in increased nuclear accumulation of Nrf2 as com-paredwith PP2 alone (Fig. 3A, lanes 7–9). These results indicate
FIGURE 3. GSK-3� and Fyn regulate Nrf2 via same pathway. A, Western analysis. PP2 and LiCl led to nuclearaccumulation of endogenous Nrf2. HepG2 cells were treated with PP2 (0.5 and 1 �M) or LiCl (10 and 50 mM) ora combination of both for 2 h. Cells were harvested and fractionated, and nuclear fractions were immuno-blotted with anti-Nrf2 and anti-Lamin B antibody. Nrf2 levels were normalized to Lamin B levels as describedearlier, and the fold amount of Nrf2 in the nucleus in untreated or treated cells is shown (lower panel).B, ARE-luciferase assay. HepG2 cells were co-transfected with NQO1-ARE luciferase reporter and firefly Renillaluciferase as described under “Materials and Methods.” Thirty six h after transfection, the cells were treated witheither PP2 or lithium chloride as in A for 8 h. Cells were then harvested and lysed to analyze the luciferaseactivity. The results are presented as � S.E. of three independent experiments, and each experiment was donein triplicate. C, ARE-luciferase assay. Hepa-1 cells were co-transfected as in B along with GSK-3� or Fyn siRNA incombination and doses as indicated, lysed, and analyzed for luciferase activity. D, Western analysis. Hepa-1cells were transfected with siRNA in combinations as indicated. 48 h later, cells were lysed to obtain total celllysate. 50 �g of lysate was probed with antibodies against Nr2, GSK-3�, Fyn, and �-actin. Protein levels werenormalized to �-actin levels as described earlier, and the fold amount of protein is plotted.
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
JUNE 1, 2007 • VOLUME 282 • NUMBER 22 JOURNAL OF BIOLOGICAL CHEMISTRY 16505
that inclusion of LiCl with PP2 did not increase further thenuclear accumulation of Nrf2 over that observed with PP2alone. In a related experiment, HepG2 cells were transfectedwith the NQO1 gene ARE-luciferase. The transfected cellswere treated with different doses of LiCl or PP2 alone or incombinations and analyzed for luciferase activity (Fig. 3B).The results showed similar increase in luciferase activitywith PP2 alone or in combination with LiCl. In other words,the LiCl had no effect on PP2 induction of ARE-luciferasegene expression. To further prove these observations, weused siRNA against both Fyn and GSK-3� alone or in com-binations. In accordance with our earlier published results(13), ARE-luciferase activity increased after siRNA-medi-ated inhibition of Fyn in a dose-dependent manner (Fig. 3C).Similar results were obtained with GSK-3� siRNA. However,when both the siRNAs were used in combination, the ARE-luciferase activity followed the pattern as observed by inhi-bition of Fyn alone indicating that both these kinases affectNrf2 by a similar pathway (Fig. 3C). Also, inhibition of boththe kinases together did not show a further increase in Nrf2protein stabilization as seen by Fyn siRNA alone (Fig. 3D).These results suggested that inhibition of GSK-3� and Fynhad a concerted and not an additive effect on nuclear accu-mulation of Nrf2. In other words, GSK-3� is upstream to Fynin the control of nuclear export of Nrf2.
Next, we determined the effect ofLiCl inhibition of GSK-3� onhydrogen peroxide-induced nuclearimport and export of Nrf2 and Fyn(Fig. 4). The untransfected andNrf2-V5- or FLAG-Fyn-transfectedHepG2 cells were treated withhydrogen peroxide in the absenceand presence of LiCl and were sub-cellularly fractionated. The distri-bution of Nrf2 and Fyn in nuclearand cytosolic fractions was analyzedby immunoblotting. Hydrogen per-oxide treatment led to import ofNrf2 in the nucleus at 1 h and exportof Nrf2 at 4 h of treatment (Fig. 4A).The nuclear accumulation ofNrf2 isan early response of the Nrf2-INrf2complex to hydrogen peroxide andwas expected as reported earlier (11,13). Hydrogen peroxide treatmentat 1 h released Nrf2 from INrf2.Nrf2 translocated to the nucleusresulting in increased nuclear Nrf2.Nrf2 was exported out of thenucleus at 4 h after hydrogen perox-ide treatment as a result of delayedresponse to oxidative stress (13).Inclusion of LiCl, an inhibitor ofGSK-3�, blocked the nuclear lossof Nrf2 at 4 h of hydrogen per-oxide treatment. Interestingly, Fynresponse to hydrogen peroxide was
opposite that of Nrf2. Endogenous Fyn protein levels werereduced in the nucleus at 1 h but increased at 4 h after hydrogenperoxide treatment of HepG2 cells (Fig. 4A). The LiCl-medi-ated inhibition ofGSK-3� reduced the nuclear gain of Fyn at 4 hafter hydrogen peroxide treatment (Fig. 4A). The exogenouslyexpressed Nrf2-V5 protein showed similar localization pat-terns as observed with endogenous Nrf2 (compare Fig. 4B andFig. 2B, Nuc. Ext panels). Nrf2-V5 accumulated in the nucleusin a LiCl dose-dependentmanner (Fig. 4B, left panel). However,the nuclear export-deficient mutant Nrf2Y568A-V5 showedaccumulation in the nucleus even in the absence of LiCl. This isbecause of loss of tyrosine 568 phosphorylation and the absenceof nuclear export of mutant protein (13). Treatment with LiClhad no effect on nuclear levels of Nrf2Y568A-V5 (Fig. 4B, rightpanel). Hydrogen peroxide treatment in the absence and pres-ence of LiCl also showed similar results for exogenouslyexpressed FLAG-Fyn as observed for endogenous Fyn (com-pare Fig. 4,CwithA, nuclear panels). FLAG-Fyn depleted fromthe nucleus at 1 h post hydrogen peroxide treatment which wasfollowed by accumulation in the nucleus at 4 h of hydrogenperoxide exposure (Fig. 4C). The hydrogen peroxide mediatednuclear accumulation of FLAG-Fyn at 4 h was blocked in pres-ence of LiCl (Fig. 4C, Nuc. Ext panel, compare lanes 1–3 and 4and 5). These results revealed that the treatment ofHepG2 cellswith hydrogen peroxide leads to nuclear loss of Fyn and nuclear
FIGURE 4. LiCl blocked hydrogen peroxide-induced nuclear import of Fyn and hence nuclear export ofNrf2. A, Western analysis of localization of endogenous Nrf2 and Fyn in cells treated with hydrogen peroxideand LiCl. HepG2 cells were untreated or pretreated with 50 mM lithium chloride and then treated with 4 mM
hydrogen peroxide (H2O2) for 1 or 4 h without lithium chloride (�LiCl) or with lithium chloride (�LiCl). After thetreatments, the cells were harvested and nuclear fractions prepared by standard procedures and analyzed byimmunoblotting with anti-Nrf2, anti-Fyn, and nuclear specific anti-Lamin B antibody. Nrf2 and Fyn levels werenormalized to Lamin B levels, and the fold induction in the amount of Nrf2 or Fyn in the nucleus is shown (lowerpanel). B and C, Western analysis of localization of transfected Nrf2 or Fyn. HepG2 cells were transfected withNrf2-V5 or nuclear export-deficient mutant Nrf2Y568A-V5 or FLAG-Fyn. Twenty four h after transfection, thecells were treated with 50 mM LiCl for 30 min or 1 h (B) or as in A (C). Cells were harvested, and cytosol andnuclear extracts were prepared and immunoblotted with anti-V5 (B), anti-FLAG (C), and anti-Lamin B andanti-LDH antibodies as controls. The protein levels were quantitated by densitometry, normalized to propercontrols, and plotted as fold amount of protein versus treatment (lower panels in B and right panel in C).
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
16506 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 22 • JUNE 1, 2007
import of Nrf2 as an early response for activation of genesencoding chemoprotective proteins. Delayed/late response tohydrogen peroxide is to accumulate Fyn in the nucleus to phos-phorylate Nrf2Y568 and activate the nuclear export of Nrf2.The results also revealed that inhibition of GSK-3� with LiClled to the loss of Fyn accumulation in the nucleus and activatednuclear accumulation of Nrf2. Taken together, these resultssuggested that GSK-3� is upstream to Fyn as accumulation ofFyn in the nucleus was dependent on GSK-3� activity. Thehydrogen peroxide-induced nuclear accumulation of Fyn at 4 hafter treatment is likely because of nuclear translocation of Fynafter phosphorylation by GSK-3�.
The above results raised an interesting question regardingthe mechanism of GSK-3� regulation of Fyn? Additionalexperiments were performed to address this question. Phos-phorylation of GSK-3�Y216 is known to activate GSK-3�(16).We performed experiments to test if hydrogen peroxideactivated GSK-3�? HepG2 cells were treated with hydrogenperoxide and analyzed by immunoblotting with anti-phos-pho-GSK-3�-pY216 antibody (Fig. 5A, left panel). Hydrogenperoxide induced GSK-3�Y216 phosphorylation within 1 hof the treatment. The analysis showed that hydrogen perox-ide also phosphorylated GSK3� at tyrosine 279 (Fig. 5A, leftpanel). The anti-phospho-GSK3�-pY216 antibody is knownto cross-react with phospho-GSK3�-pY279 (16). LiCl andPP2 failed to inhibit hydrogen peroxide-induced tyrosine
phosphorylation of GSK-3� andGSK-3� (Fig. 5A, left, panel).However, AG18 and genistein, twogeneral tyrosine kinase inhibitors,significantly reduced phosphoryl-ation of GSK-3�-pY216 (Fig. 5A,right panel, compare lanes 2, 4,and 6). The inhibition of GSK-3�activation in the presence of gen-eral tyrosine kinase inhibitors alsorevealed that Fyn did not phos-phorylate GSK-3� in response tohydrogen peroxide because PP2,an inhibitor of Fyn, failed to blockhydrogen peroxide-induced GSK-3�Y216 phosphorylation. Theresults did suggest that unknowntyrosine kinase(s) phosphorylateGSK-3�-Y216 in response tohydrogen peroxide that led to acti-vation of GSK-3�. Further experi-ments were performed to confirmthat GSK-3� is upstream to Fynand regulated Fyn in response tohydrogen peroxide. We tested thehypothesis that hydrogen perox-ide activated GSK-3� which thenphosphorylated Fyn resulting inits nuclear localization. Inside thenucleus, Fyn phosphorylatesNrf2Y568 and results in nuclearexport of Nrf2. HepG2 cells were
transfected with FLAG-Fyn, treated with hydrogen peroxidein the absence and presence of GSK3� inhibitor LiCl or Fyninhibitor PP2. Nuclear extract was prepared and used foranalyzing the phosphorylation status of Fyn by immunopre-cipitation with anti-Thr(P) antibody followed by immuno-blotting with anti-FLAG antibody (Fig. 5B, top left panel). Ina similar experiment, the nuclear extract was analyzed byimmunoblotting and probing with anti-FLAG antibody (Fig.5B, right panel). The results revealed that hydrogen peroxideindeed induced time-dependent phosphorylation of Fyn atthreonine residue(s) leading to nuclear accumulation of Fyn(Fig. 5, left and right panels). We also checked the phospho-rylation of Fyn at serine residues by immunoprecipitatingwith anti-FLAG and immunoblotting with anti-Ser(P) anti-body (Fig. 5B, bottom left panel). The results indicate thatFyn is not phosphorylated at serine residues in response tohydrogen peroxide. Untreated cells also showed a lowamount of phosphorylated Fyn in the nucleus. This is pre-sumably because of endogenous cellular stresses that mighthave led to Fyn phosphorylation and nuclear accumulation.The results also revealed that almost all of the Fyn translo-cated in the nucleus is phosphorylated. The treatment withLiCl blocked hydrogen peroxide-induced phosphorylationand nuclear accumulation of FLAG-Fyn, which was in agree-ment with GSK-3� being upstream to Fyn. However, PP2had no effect on phosphorylation status or nuclear accumu-
FIGURE 5. GSK-3� is upstream of Fyn kinase in regulation of nuclear export of Nrf2. A, hydrogen peroxide-induced phosphorylation and activation of GSK-3�. HepG2 cells were pretreated with LiCl or PP2 (in somecases) and treated with 4 mM H2O2 for 1 and 4 h � 50 mM lithium chloride (�LiCl) or 1 �M PP2 (�PP2) (left panel).In a similar experiment, the cells were treated with H2O2 in the presence or absence of AG18 or genistein for 1 h(right panel). In both the cases, cells were lysed and total cell lysate was analyzed by Western blotting (WB) andprobing with anti-(Tyr(P))-GSK3, anti-GSK-3�, and anti-�-actin antibodies. B and C, phosphorylation and local-ization of FLAG-Fyn. HepG2 cells in 100-mm plates were transfected with 1.0 �g of FLAG-Fyn. 24 h aftertransfection cells were treated as in A, left panel. Cells were harvested, lysed, or subjected to subcellular frac-tionation to prepare nuclear extracts (Nuc Ext). For analyzing phosphorylation of FLAG-Fyn, 1 mg of nuclearextract was used to immunoprecipitate (IP) with either anti-Thr(P) or anti-FLAG antibody, and the immunecomplexes were immunoblotted with anti-FLAG-HRP or anti-Ser(P) antibodies, respectively. 50 �g of nuclearextract (B, right panel) or total cell lysate (C) were analyzed by Western blotting and probing with anti-FLAG-HRP and anti-LaminB or anti-�-actin antibodies.
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
JUNE 1, 2007 • VOLUME 282 • NUMBER 22 JOURNAL OF BIOLOGICAL CHEMISTRY 16507
lation of Fyn (Fig. 5B, compare �LiCl with �PP2). In arelated experiment, immunoblot analysis of total cell lysaterevealed similar expression of FLAG-Fyn in untreated andLiCl-treated cells (Fig. 5C). Increased nuclear FLAG-Fyn in�PP2 panels is not because of more nuclear accumulationbut because of more overall FLAG-Fyn expression. The totalcell lysates showed more FLAG-Fyn expression in the �PP2panels (Fig. 5C). This indicated that the alteration in phos-phorylation and nuclear localization of Fyn is not because ofalterations in the expression of FLAG-Fyn plasmid in HepG2cells. The combined results revealed that hydrogen peroxideinduced GSK-3�Y216 phosphorylation leading to activationof GSK-3�. The activated GSK-3� phosphorylated Fyn atthreonine residue(s). Thr-phosphorylated Fyn accumulatedin the nucleus and phosphorylated Nrf2Y568 leading tonuclear export of Nrf2 (Figs. 5 and 4A). To further analyzeGSK-3�-mediated Fyn and Nrf2 phosphorylation, weemployed in vitro kinase assays using purified Nrf2, GSK-3�,and Fyn proteins (Fig. 6). In the initial experiment, GSK-3�was incubated with a 3-kDa GSK3 peptide, and in vitrokinase reaction was performed to test the activity of GSK-3�.GSK-3� successfully phosphorylated GSK peptide in vitro ina kinase dose-dependent manner (Fig. 6A). Bacterially puri-fied Nrf2 was then incubated with active GSK-3� in the pres-ence or absence of LiCl (Fig. 6B, lanes 3 and 4). The Nrf2
phosphorylation by GSK-3� wasnot detected (Fig. 6B, lanes 3 and4, upper panel). This was despitethe presence of both Nrf2 andGSK-3� proteins in the test reac-tion (Fig. 6B, middle and lowerpanels, lanes 3 and 4). In the sameexperiment, phosphorylated Nrf2was detected when purified Nrf2was incubated with active Fynkinase (Fig. 6B, lane 5). It is note-worthy that in in vitro kinase reac-tions GSK-3� and Fyn are auto-phosphorylated (Fig. 6B, upperpanel). In related experiments,GSK-3� and Fyn kinases wereincubated individually and to-gether in an assay reaction to testthe GSK-3� phosphorylation ofFyn. Both GSK-3� and Fyn wereautophosphorylated. The amountof Fyn autophosphorylation was insignificantly high proportions sothat addition of GSK-3� could notgive any conclusive results.We also performed experiments
to investigate the role of nuclearexport of Nrf2, i.e. to determinethe fate of Nrf2 exported from thenucleus. We tested the hypothesisthat Nrf2 exported out of thenucleus binds to INrf2, ubiquiti-nates, and degrades in the cytosol.
INrf2 is known to function as ubiquitin-protein isopeptideligase and in association with Cul3 leads to ubiquitinationand degradation of Nrf2 (18, 19). Ubiquitination analysis ofNrf2 was performed in HepG2 cells transfected withNrf2-V5 or the nuclear export-deficient mutant Nrf2Y568A-V5, HA-Ub, and FLAG-INrf2 in combinations as shown inFig. 7. The transfected cells were treated with vehicle con-trol, Me2SO or GSK-3� inhibitor, LiCl or proteasomal inhib-itor, MG132. The cells were lysed and immunoblotted witheither anti-V5 to detect Nrf2-V5 or Nrf2Y568A-V5, anti-FLAG antibody to detect FLAG-INrf2, or anti-�-actin anti-body to demonstrate equal loading. The lysates were immu-noprecipitated with anti-V5 antibody and immunoblottedwith anti-HA antibody to detect ubiquitination of Nrf2-V5and Nrf2Y568A-V5. The results are shown in Fig. 7. Theco-transfection of Nrf2-V5 with HA-Ub showed ubiquitina-tion and degradation of Nrf2-V5 (Fig. 7, left panel). Inclusionof INrf2 in the transfection increased ubiquitination anddegradation of Nrf2-V5 (Fig. 7, left panel, Me2SO lanes). Thepretreatment of transfected cells with LiCl significantlyreduced ubiquitination and degradation of Nrf2-V5 suggest-ing the role of the nuclear export in degradation of Nrf2. Thetreatment with the proteasome inhibitor MG132 increasedubiquitination without degradation of Nrf2-V5. Interest-ingly, the nuclear export-deficient mutant Nrf2Y568A-V5
FIGURE 6. In vitro kinase assay. A, 100 ng of GSK-3 peptide was incubated with 50, 100, or 200 ng of activeGSK-3� in an in vitro kinase reaction at 30 °C for 30 min. Reaction was stopped by boiling the samples in SDS-gelloading dye, resolved on 12% SDS-polyacrylamide gel, dried, and autoradiographed. Phosphorylated GSK-3peptide and free [�-32P]ATP are indicated. B, bacterially purified Nrf2, active GSK-3�, and active Fyn wereincubated in and in vitro kinase reaction in combinations as displayed. The reactions were stopped as describedin A, and the gels were autoradiographed (top panel). A duplicate of the test reactions was transferred onto thenitrocellulose membrane and stained with Ponceau S stain to show the presence of specified proteins (middlepanel). The membrane was probed with anti-Nrf2 antibody to show the presence of equal amounts of purifiedNrf2 in each reaction (lower panel).
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
16508 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 22 • JUNE 1, 2007
showed very little ubiquitination in the absence or presenceof GSK3� inhibitor LiCl (Fig. 7B, right panel). In otherwords, LiCl-mediated inhibition of GSK3� activity had noeffect on ubiquitination of Nrf2Y568A-V5. However, thenuclear export-deficient mutant Nrf2Y568A proteindegraded efficiently in the presence of INrf2. The treatmentof cells with the proteasome inhibitor MG132 showed ubiq-uitination of Nrf2Y568A without degradation. The com-bined results suggested that the Fyn-phosphorylated Nrf2protein is exported out of nucleus to bind to INrf2, ubiquiti-nate, and degrade. The nuclear export-deficient mutantNrf2Y568 protein binds to INrf2, ubiquitinates, anddegrades inside the nucleus. INrf2 is predominantly shown
to be present in the cytosol (5, 6).However, recent studies haveshown that some of the INrf2 is alsopresent in the nucleus (20, 21).
DISCUSSION
The regulation of Nrf2, espe-cially its abundance in the nucleus,is extremely important for con-trolling the normal and inducedexpression of a battery of chemo-protective genes (1). Exposure ofcells to chemical stress leads to therelease of Nrf2 from its cytosolicinhibitor INrf2 as an early cellularresponse to chemical stress (Fig.8). The release of Nrf2 from INrf2is mediated by PKC and/or cys-teine modification of INrf2 (1). Abipartite nuclear localization sig-nal directs Nrf2 to the nucleus(11). Nrf2 after translocation inthe nucleus heterodimerizes withknown (Jun and small Maf) andunknown nuclear factors, binds toARE, and activates gene transcrip-tion. The increase in expression ofchemoprotective genes neutralizesthe chemical stress. Because persis-tent increase in chemoprotectivegenes expression threatens cell sur-vival, Nrf2 is exported out of thenucleus and degraded. The nuclearexport of Nrf2 is delayed/lateresponse of cells to oxidative/elec-trophilic stress. A leucine-richnuclear export signal at the C termi-nus of Nrf2 has been characterized(11).However, the nuclear export sig-nal in Nrf2 is activated only after Fynis accumulated inside thenucleus thatphosphorylates tyrosine 568 of Nrf2(13). The phosphorylated Nrf2Y568binds to Crm1 and is exported out ofthe nucleus (13).
The studies in the present report demonstrate that GSK-3�is upstream to Fyn in regulation of nuclear export ofNrf2. Phos-phorylation status ofGSK-3� regulates its activity (16).GSK-3�phosphorylated at a serine 9 residue via PKC or other similarenzymes is inactive. Activation of GSK-3� is mediated by phos-phorylation at tyrosine 216 residue and/or de-phosphorylationof serine 9 (16). The hydrogen peroxide in our studies inducedtyrosine 216 phosphorylation of GSK-3� resulting in its activa-tion. The activated GSK-3� phosphorylated Fyn at threonineresidue(s) leading to nuclear localization of Fyn. Fyn phospho-rylates tyrosine 568 of Nrf2 (13). Phosphorylated Nrf2Y568 isexported out of the nucleus, ubiquitinated, and degraded.Recently, GSK3� was shown to phosphorylate Nrf2 at
FIGURE 7. Ubiquitination and degradation of Nrf2. HepG2 cells were co-transfected either withpcDNA-Nrf2-V5/pcDNA-Nrf2Y568A-V5, pCMV-HA-Ub, or pCMV-FLAG-mINrf2 in combinations as shown. Thetransfected cells were treated with either Me2SO or GSK3� inhibitor LiCl or proteasomal inhibitor MG132 for5 h. The cells were lysed in RIPA buffer, and 100 �g of protein was analyzed by SDS-PAGE and immunoblottedwith anti-V5, anti-FLAG, and �-actin antibodies. 500 �g of protein was immunoprecipitated (IP) with anti-V5antibody and immunoblotted (WB) with anti-HA antibody to detect ubiquitinated Nrf2 or Nrf2Y568A-V5.
FIGURE 8. Model depicting the role of GSK-3� in regulating nuclear export of Nrf2 via Fynphosphorylation.
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
JUNE 1, 2007 • VOLUME 282 • NUMBER 22 JOURNAL OF BIOLOGICAL CHEMISTRY 16509
unknown residues with implications in nuclear export of Nrf2(17). However, our in vitro kinase assay data did not show Nrf2phosphorylation by GSK-3�. Also, our data reveal that GSK3�is upstream to Fyn, which phosphorylates Tyr-568 of Nrf2 thatregulated nuclear export of Nrf2. The reasons and significanceof this difference remain unknown.It is noteworthy that PKC is also known to inactivate
GSK-3� (16). The decrease in nuclear Fyn at 1 h after hydro-gen peroxide treatment in Fig. 4A is presumably due to PKC-mediated inactivation of GSK-3�; however, this remains tobe determined. Therefore, the inactivation of GSK-3� andactivation of Nrf2 via serine phosphorylation are both regu-lated by PKC. Together these mechanisms might work inharmonization leading to a synergistic action to earlyresponse to oxidative stress resulting in nuclear accumula-tion of Nrf2 both by activating import and blocking theexport of Nrf2. The signaling events between hydrogen per-oxide and tyrosine 216 phosphorylation of GSK-3� remainunknown. In addition, the tyrosine kinase(s) that phospho-rylate GSK-3�Y216 also remains unknown. The signalingevents between hydrogen peroxide and phosphorylation ofGSK-3� might involve phosphatidylinositol 3-kinase, AKT,PP2A, and PKC that could inactivate or activate GSK-3�during early and delayed/later events (22).In conclusion, we have investigated the Fyn upstream sig-
naling in the regulation of nuclear export of Nrf2 and theregulation of expression of chemoprotective proteins inresponse to oxidative/electrophilic stress. We demonstratethat GSK-3� is upstream to Fyn. The chemical stressinduced phosphorylation of GSK-3�Y216 and activatedGSK-3�. The activated GSK-3� phosphorylated Fyn. Thephosphorylation of Fyn leads to nuclear accumulation of Fynand phosphorylation of Nrf2 resulting in nuclear export,ubiquitination, and degradation of Nrf2. This study dissectsthe mechanism involved in Nrf2 regulation via Fyn kinase asa result of delayed response to oxidative stress.
REFERENCES1. Jaiswal, A. K. (2004) Free Radic. Biol. Med. 36, 1199–12072. Chan, K., Lu, R., Chang, J. C., and Kan, Y. W. (1996) Proc. Natl. Acad. Sci.
U. S. A. 93, 13943–139483. Braun, S., Hanselmann, C., Gassmann, M. G., auf dem Keller, U., Born-
Berclaz, C., Chan, K., Kan, Y.W., andWerner, S. (2002)Mol. Cell. Biol. 22,5492–5505
4. Rangasamy, T., Cho, C. Y., Thimmulappa, R. K., Zhen, L., Srisuma, S. S.,Kensler, T.W., Yamamoto, M., Petrache, I., Tuder, R. M., and Biswal, S. S.(2004) J. Clin. Investig. 114, 1248–1259
5. Itoh, K., Wakabayashi, N., Katoh, Y., Ishii, T., Igarashi, K., Engel, J. D., andYamamoto, M. (1999) Genes Dev. 13, 76–86
6. Dhakshinamoorthy, S., and Jaiswal, A. K. (2001)Oncogene 20, 3906–39177. Wakabayashi, N., Itoh, K., Wakabayashi, J., Motohashi, H., Noda, S., Ta-
kahashi, S., Imakado, S., Kotsuji, T., Otsuka, F., Roop, D. R., Harada, T.,Engel, J. D., and Yamamoto, M. (2003) Nat. Genet. 35, 238–245
8. Strachan, G. D., Ostrow, L. A., and Jordan-Sciutto, K. L. (2005) Biochem.Biophys. Res. Commun. 336, 490–495
9. Bloom, D. A., and Jaiswal, A. K. (2003) J. Biol. Chem. 278, 44675–4468210. Huang, H. C., Nguyen, T., and Pickett, C. B. (2002) J. Biol. Chem. 277,
42769–4277411. Jain, A. K., Bloom, D. A., and Jaiswal, A. K. (2005) J. Biol. Chem. 280,
29158–2916812. Kannan, S., and Jaiswal, A. K. (2006) Cancer Res. 66, 8421–842913. Jain, A. K., and Jaiswal, A. K. (2006) J. Biol. Chem. 281, 12132–1214214. Dhakshinamoorthy, S., and Jaiswal, A. K. (2000) J. Biol. Chem. 275,
40134–4014115. Dhakshinamoorthy, S., Jain, A. K., Bloom, D. A., and Jaiswal, A. K. (2005)
J. Biol. Chem. 280, 16891–1690016. Grimes, C. A., and Jope, R. S. (2001) Prog. Neurobiol. 65, 391–42617. Salazar, M., Rojo, A. I., Velasco, D., de Sagarra, R. M., and Cuadrado, A.
(2006) J. Biol. Chem. 281, 14841–1485118. Cullinan, S. B., Gordan, J. D., Jin, J., Harper, J. W., and Diehl, J. A. (2004)
Mol. Cell. Biol. 24, 8477–848619. Kobayashi, A., Kang, M. I., Okawa, H., Ohtsuji, M., Zenke, Y., Chiba, T.,
Igarashi, K., and Yamamoto, M. (2004)Mol. Cell. Biol. 24, 7130–713920. Karapetian, R. N., Evstafieva, A. G., Abaeva, I. S., Chichkova, N. V.,
Filonov, G. S., Rubtsov, Y. P., Sukhacheva, E. A., Melnikov, S. V., Schnei-der, U., Wanker, E. E., and Vartapetian, A. B. (2005) Mol. Cell. Biol. 25,1089–1099
21. Velichkova, M., and Hasson, T. (2005)Mol. Cell. Biol. 25, 4501–451322. Hennessy, B. T., Smith, D. L., Ram, P. T., Lu, Y., andMills, G. B. (2005)Nat.
Rev. Drug Discov. 4, 988–1004
GSK-3� Upstream of Fyn in Regulation of Nuclear Export of Nrf2
16510 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 282 • NUMBER 22 • JUNE 1, 2007
