Hsp90 Interaction with INrf2(Keap1) Mediates Stress-induced Nrf2 Activation*

INrf2(Keap1) functions as an adapter for Cul3/Rbx1-mediated degradation of Nrf2. In response to stress, Nrf2 is released from INrf2 and translocates inside the nucleus leading to activation of cytoprotective proteins critical in protection against adverse effects including cancer. We demonstrate here a novel role of heat shock protein 90 (Hsp90) in control of the INrf2 and Nrf2 activation. Hsp90 interacted with INrf2 that leds to stabilization of INrf2 during heat shock stress. Domain mapping showed the requirement of INrf2-NTR and the Hsp90-CLD region for interaction of Hsp90 with INrf2. Heat shock and antioxidants induced Hsp90, and casein kinase 2 (CK2) phosphorylated INrf2Thr55. This led to increased Hsp90-INrf2 interaction, dissociation of the Rbx1/Cul3·INrf2·Nrf2 complex, and activation of Nrf2. Inhibitors of CK2 and Hsp90, and mutation of INrf2Thr55 abolished the Hsp90-INrf2 interaction and downstream signaling. INrf2 is released from Hsp90 once the heat shock or antioxidant stress subsidized, thereby allowing INrf2 to interact with Nrf2 and facilitate Nrf2 ubiquitination and degradation. The results together demonstrate a novel role for the stress-induced Hsp90-INrf2 interaction in regulation of Nrf2 activation and induction of cytoprotective proteins.

The Nrf2⅐INrf2 complex serves as a sensor of chemical and radiation-induced oxidative and electrophilic stress (1,2). Nrf2 resides in the cytoplasm where it interacts with INrf2 (inhibitor of Nrf2) or Keap1 (Kelch-like ECH-associated protein 1). INrf2 functions as a substrate adaptor protein for a Cul3/Rbx1-dependent E3 ubiquitin ligase complex to ubiquitinate and degrade Nrf2 thus maintaining steady-state levels of Nrf2 (3)(4)(5). The mechanisms by which Nrf2 are released from INrf2 under stress have been actively investigated. One mechanism is that cysteine thiol groups of INrf2 act as sensors of oxidative stress and are modified by the chemical inducer causing formation of disulfide bonds between cysteines of two INrf2 peptides. This results in a conformational change that renders INrf2 unable to bind to Nrf2 (6,7). On the other hand, we and others have shown that antioxidant-induced protein kinase C␦ (PKC␦) phosphorylates Nrf2Ser40 leading to dissociation of Nrf2 from INrf2, which stabilize Nrf2 and allows it to translocate in the nucleus (2,8,9). More recently, we demonstrated that the two mechanisms work in concert with each other (2). The Nrf2 in the nucleus binds with the antioxidant response element (ARE) 2 of the promoter region of antioxidant genes and increases their expression. Nrf2 up-regulates several genes encoding phase II detoxification enzymes and antioxidant proteins, such as NAD(P)H:quinine oxidoreductase-1 (NQO1), ␥-glutamylcysteine synthetase, heme oxygenase 1, thioredoxin reductase-1, and thioredoxin (10). Transcriptional activation of Nrf2 and its downstream proteins have been shown to be critically important in protection of cells from oxidative stress-and chemical-induced damage of liver and lung tissues (1,11,12).
Cho et al. (13) demonstrated that Nrf2 knock-out mice are more sensitive to hyperoxic injury of the lung. The primary astrocyte of the Nrf2 Ϫ/Ϫ mice is also more susceptible to oxidative stress and inflammation than that of Nrf2 ϩ/ϩ mice (14). Leung et al. (15) showed that deficiency of Nrf2 results in severe oxidative stress. These observations, collectively, imply that Nrf2 is a master regulator of ARE-driven transcriptional activation for antioxidant genes in maintaining the homeostasis of redox status within cells. On the other hand, additional evidence suggested that persistent accumulation of Nrf2 in the nucleus is also harmful. INrf2-null mice demonstrated persistent accumulation of Nrf2 in the nucleus that led to postnatal death by malnutrition resulting from hyperkeratosis in the esophagus and forestomach (16). The reversed phenotype of INrf2 deficiency by breeding to Nrf2-null mice suggested that tightly regulated negative feedback might be essential for cell survival (11). The systemic analysis of the INrf2 genomic locus in human lung cancer patients and cell lines showed that deletion, insertion, and missense mutations in functionally important domains of INrf2 results in reduction of the INrf2 affinity for Nrf2 and elevated expression of cytoprotective genes that resulted in drug resistance and cell survival in lung cancer cells (17,18). Unrestrained activation of Nrf2 in cells increases a risk of adverse effects including survival of damaged cells, tumorigenesis, and drug resistance. Therefore, it appears that cells contain mechanisms that autoregulate cellular abundance of Nrf2 (19). Indeed, these findings suggest that INrf2/Nrf2 signaling plays an important role in cell survival in normal cells, as well as drug resistance in cancer cells (12).
Heat shock protein 90 (Hsp90) is a molecular chaperone and is one of the most abundant proteins expressed in cells (20). Hsp90 is a member of the heat shock protein family up-regulated in response to stress. In unstressed cells, Hsp90 plays a number of important roles, which include assisting in folding, intracellular transport, maintenance, and degradation of proteins, as well as facilitating cell signaling (21). Hsp90 is known to associate with the non-native structures of many proteins that have led to the proposal that Hsp90 is involved in protein folding in general. Furthermore, Hsp90 has been shown to suppress the aggregation of a wide range of "client" or "substrate" proteins and hence acts as a general protective chaperone (22,23). However, Hsp90 is somewhat more selective than other chaperones. For example, many cancer cells overexpress a number of proteins involved in cell survival including PI3K and AKT. Inhibition of these two proteins triggers apoptosis. Hsp90 stabilizes the PI3K and AKT proteins. Hence inhibition of Hsp90 appears to induce apoptosis through inhibition of the PI3K/ AKT signaling pathway (24). Another important role of Hsp90 in cancer is stabilization of mutant proteins such as v-Src, the fusion oncogene Bcr/Abl, and p53 that appear during cell transformation. It appears that Hsp90 can act as a "protector" of less stable proteins produced by DNA mutations (25).
In the present study, we demonstrate a novel role of Hsp90 in regulation

EXPERIMENTAL PROCEDURES
Plasmid Construction-INrf2 and mutants were previously described (26). The INrf2T55A mutant was generated using the Gene Tailor site-directed mutagenesis kit (Invitrogen). The various domain deletion mutants of the Hsp90 protein were cloned into pcDNA3.1/V5-His/Topo vector by TA cloning. The primer sequences used for construction of the various Hsp90 mutant plasmids and FLAG-INrf2T55A are shown in supplemental Table S1. The construction of pcmx-FLAG-INrf2, pcDNA-INrf2-V5, and PGL2b-NQO1-ARE have been described previously (26). All plasmids were confirmed by DNA sequencing.

Hsp90-INrf2(Keap1) Interaction and Nrf2 Activation
(catalog number C7363), was dissolved in DMSO and Hepa-1 cells were treated with 10 -60 nM concentrations. The Hsp90specific inhibitor geldanamycin was obtained from LC Laboratories and Hepa-1 cells were treated with 2 M for different time intervals. Cells were harvested and regulation of specific proteins were analyzed. Generation of stable Flp-In T-REx HEK293 cells expressing tetracycline-inducible INrf2 is described previously (26).
Immunoprecipitation (IP) and Phosphorylation Analysis-Hepa-1 cells treated or transfected with the indicated plasmids for 36 h or control 293 cells or FLAG-INrf2-293 cells were treated with tetracycline (0.5 g/ml) for 24 h. The cells were lysed in RIPA buffer containing 1 mM PMSF, 1 mM sodium vanadate, a serine/threonine phosphatase inhibitor mixture (Sigma), and protease inhibitors (Roche Applied Science). For immunoprecipitation, 1 mg of whole cell lysates were precleaned by protein AG Plus-agarose (Santa Cruz Biotechnology) and extracts were incubated with the respective antibodies (1 g) overnight at 4°C. Immune complexes were collected, analyzed by immunoblotting, probed with anti-INrf2 antibody or anti-phosphothreonine antibody (Stressgen), and visualized using a ECL chemiluminescence (Amersham Biosciences).
Immunofluorescence-Hepa-1 cells were grown in Lab-Tek II chamber slides and separately transfected with INrf2-GFP, INrf2-NTR-2XGFP, or INrf2⌬NTR-GFP for 32 h, fixed in 2% formaldehyde, and permeabilized by treatment with 0.25% Triton X-100. Cells were washed twice with PBS and incubated with 1:1000 dilution of mouse Hsp90 primary antibodies in 2% BSA for 12 h at 4°C. The cells were then washed twice with PBS and incubated with Alexa Fluor 594-conjugated anti-mouse secondary antibodies (Invitrogen). Similarly, Hepa-1 cells were also transfected with Hsp90-V5 and deletion plasmids tagged with V5, fixed, permeabilized, and incubated with anti-V5-FITC antibody and goat anti-INrf2 antibodies followed by Alexa Fluor 594-conjugated anti-goat secondary antibodies. After immunostaining, the cells were observed under a Nikon fluorescence microscope and photographed.
Transient Transfection and Luciferase Assay-Hepa-1 cells were transfected with 1 g of the indicated plasmids using Effectene transfection reagent (Qiagen). 36 h after transfection, the cells were harvested and protein expression was examined by immunoblotting. For luciferase reporter assay, Hepa-1 cells were co-transfected with 0.1 g of human NQO1 promoter ARE-Luc reporter plasmid and 10 times less quantities of firefly Renilla luciferase encoded by plasmid pRL-TK. After 24 h of transfection, the cells were either treated with DMSO or tBHQ at 37°C or incubated for different times at 42°C for heat shock as indicated in figures. Cells were lysed and analyzed for luciferase activity by previously described procedures (27).
siRNA Interference Assay-Hsp90 siRNA was used to inhibit Hsp90 protein by a procedure described previously (27). Hsp90 siRNA, and control siRNA were purchased from Dharmacon. In related experiments, Hepa-1 cells were transfected with 25 to 100 nM Hsp90 siRNA or control siRNA using Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer's instructions. Thirty-two h after transfection, cells were harvested and Hsp90, INrf2, Nrf2, and actin protein levels were analyzed by immunoblotting.
Mass Spectrometry Analysis of INrf2-Hsp90 Interaction and Identification Phosphorylation Sites of INrf2-Control 293 cells or FLAG-INrf2-293 cells were treated with tetracycline (0.5 g/ml) for 24 h to induce FLAG-INrf2 protein in FLAG-INrf2-293 cells. 10 mg of cell lysates were immunoprecipitated with anti-FLAG antibodies, the immune complexes were separated by SDS-PAGE, and gels were stained with Coomassie Brilliant Blue. Gel slices containing bands indicated by arrows in figure were reduced, alkylated, and digested with trypsin. Tryptic peptides were desalted and subjected to LC-MS/MS analysis by the UMB Proteomics Core using a Voyager DEPro with a 20-Hz 337-nm nitrogen laser (Applied Biosystems, Foster City, CA). The Mascot software package was used to match the mass of the peptides with the predicted tryptic peptides generated from the translated human genome and Hsp90 as an interacting protein of INrf2 was identified from INrf2-293 cells.
For INrf2 phosphorylation studies, INrf2-293 cells were treated with tetracycline for 24 h followed by treatment with DMSO or tBHQ (4 h). Ten mg of whole cell lysates were immunoprecipitated with anti-FLAG antibodies, the immune complexes were separated by SDS-PAGE, and the gels were stained with Coomassie Brilliant Blue. INrf2 bands were excised, cut into ϳ1 ϫ 1-mm pieces, digested with trypsin, subjected to reverse phase liquid chromatography, and analyzed. All MS analyses were performed using an LCQ Deca (Thermo Scientific, Waltham, MA) mass spectrometer equipped with a nanospray ionization source. Peptides were introduced into the mass spectrometer via a 75-m inner diameter/15-m tip  inner diameter C18-packed PicoFrit column (New Objective, Woburn, MA). The spray voltage was 2.0 kV and the heated capillary temperature was 200°C. MS/MS data were acquired using a top 3 data-dependent acquisition method with dynamic exclusion enabled. MS/MS spectra were searched against a mouse protein data base (from NCBI; 88,212 sequences) using Bioworks with the SEQUEST algorithm. Details of the mass spectrometry experiments are available upon request.
Statistical Analyses-Statistical analysis of the data from luciferase assays were performed using the one-way analysis of variance test and SPSS-16 software. Error bars indicate mean Ϯ S.E. of triplicate samples and comparisons were made using the two-tailed Student's test for repeated measures. Differences between means were accepted as statistically significant at the 95% level (p Ͻ 0.04).

RESULTS
Hsp90 Interacts with INrf2-We successfully generated stable FLAG-INrf2-HEK293 cells that expressed FLAG-INrf2 protein upon exposure of cells to tetracycline (26). The cell lysates from untreated and tetracycline-treated control 293 and FLAG-INrf2-293 cells were immunoprecipitated with anti-FLAG antibody to identify INrf2 interacting proteins. The immune complexes were separated by SDS-PAGE and stained with Coomassie Brilliant Blue (Fig. 1A,  left panel). Several protein bands that specifically interacted with INrf2 were identified by LC-MS/MS (Fig. 1A, left panel, lane 4). Band A (90 kDa) was tentatively identified as Hsp90 (NCBI accession number NM_001017963.2) with a Protein-Prophet probability score 0.8 (7 unique paptides, 14% sequence coverage). Band B (69 kDa) was identified as INrf2(Keap1) (NCBI accession number NM_001110306) with a ProteinProphet probability score 1 (34 unique paptides, 37% sequence coverage) in two independent experiments. The position and sequences of the Hsp90-and INrf2derived peptides identified by LC-MS/MS mass spectrometry are shown (Fig. 1A, right panel). We performed reverse IP and immunoblotting assays to obtain further support for the unique interaction of Hsp90-INrf2 (Fig. 1, B and C). INrf2 antibody immunoprecipitated endogenous INrf2 from mouse Hepa-1 cells and Hsp90 protein was pulled down with it (Fig. 1B, upper  two panels). Similarly, Hsp90 antibody immunoprecipitated Hsp90 and INrf2 (Fig. 1B, lower two panels). In related experiments, anti-FLAG antibody immunoprecipitated FLAG-INrf2 and associated Hsp90 protein from 293-INrf2 cells but not from control 293 cells (Fig. 1C, lower two panels). In addition, in reverse IP experiments, Hsp90 antibody immunoprecipitated Hsp90 and FLAG-INrf2 from 293-INrf2 but not 293 cells (Fig. 1C, upper two panels). Control IgG in all these experiments did not pull down Hsp90 or the INrf2 protein. Furthermore, the interaction of FLAG-INrf2 and Hsp90-His-tagged protein was analyzed by pull-down experiments. Hepa-1 cell lysates co-transfected with pcDNA3.1Hsp90-His and FLAG-INrf2 plasmids were mixed with protein AG-agarose beads or Ni-NTA beads and Histagged Hsp90 protein was pulled down. The beads were boiled with SDS dye and immunoblotted with anti-FLAG antibody and anti-Hsp90 antibody (Fig. 1D). The data demonstrated that Ni-NTA beads and not protein AG-agarose beads were pulled down by Hsp90-His-tagged protein, and Hsp90 protein pulled down by FLAG-INrf2 protein (Fig. 1D, upper and lower panels). These results together confirmed that INrf2 interacts with Hsp90.
INrf2 through Its NTR Domain Interacts with the CLD Domain of Hsp90-Next we mapped the interacting domains of INrf2 and Hsp90 proteins. INrf2 possesses 5 discrete domains designated as NTR (N-terminal region; 1-60 amino acids),  NOVEMBER 19, 2010 • VOLUME 285 • NUMBER 47 BTB (broad complex, tamtrac and bric-a-bric; 61-178 amino acids), IVR (linker region; 179 -314 amino acids), DGR/Kelch (di-glycine repeats; 315-598 amino acid), and CTR (C-terminal region; 599 -624 amino acids) ( Fig. 2A). Full-length and domain deletion mutants of INrf2 were successfully cloned in 1ϫ GFP vector. Similarly, individual domains of INrf2 were also cloned in 2ϫ GFP vector. Hepa-1 cells were transfected with GFP-tagged INrf2 and INrf2 mutants, and lysates were probed with anti-GFP and Hsp90 antibodies (Fig. 2, B and C). Immunoblot analysis demonstrated expression of the correct size INrf2 of the mutant and Hsp90 bands (Fig. 2, B and C, left  panels). The forward and reverse immunoprecipitation results demonstrated that full-length INrf2 and all domain deletion mutants except INrf2⌬NTR interacted with the Hsp90 protein (Fig. 2B, right panels). The immunoprecipitation results also showed that only full-length INrf2 and the NTR domain of INrf2 interacted with Hsp90 (Fig. 2C, right panels). These results suggested that the NTR domain is required for the INrf2-Hsp90 interaction. Immunocytochemistry analysis supported the immunoprecipitation results (supplemental Fig. S1). Full-length INrf2 and not INrf2⌬NTR showed interaction. These results together with the immunoprecipitation results confirmed that the INrf2-NTR domain interacts with Hsp90.

Hsp90-INrf2(Keap1) Interaction and Nrf2 Activation
Hsp90 protein consists of 4 unique domains that include the N-terminal domain (NTD), charged linker domain (CLD), middle domain (MD), and the C-terminal domain (CTD) (Fig. 3A). We successfully generated N and C terminus domain deletion mutants of Hsp90 to investigate which domain of Hsp90 is required for interaction with INrf2 (Fig. 3A). The various Hsp90 domain deletion mutants were co-transfected with FLAG-INrf2 in Hepa-1 cells. The various plasmids bearing Hsp90 mutants produced the expected size of proteins (Fig. 3B, left panel). Forward and reverse immunoprecipitations and immunoblot analysis was performed to analyze the interaction of INrf2 with full-length and deletion mutants of Hsp90 (Fig. 3B, right  panel). The results demonstrate that the Hsp90 protein deficient in the CLD failed to interact with FLAG-INrf2 and also with endogenous INrf2 (Fig. 3B, right panel, upper five blots). In addition, immunoprecipitation of endogenous INrf2 pulled down wild type Hsp90-V5, ⌬NTD-Hsp90-V5, ⌬CTD-Hsp90-V5, and ⌬CTD⌬MD-Hsp90-V5, but not ⌬NTD⌬CLD-Hsp90 or NTD-Hsp-90 (Fig. 3B, right panel, lower two blots). Furthermore, immunocytochemistry analysis supported the immunoprecipitation results (supplemental Fig. S2). Full-length Hsp90-V5, and not ⌬NTD⌬CLD-Hsp90-V5, showed co-localization with endogenous INrf2. The above results together concluded that the INrf2 NTR region interacts with the CLD of Hsp90.

Casein Kinase 2 (CK2)-mediated Phosphorylation of INrf2Thr55 Is Required for INrf2-Hsp90
Interaction-FLAGtagged INrf2 overexpressing HEK293 cells were treated with tetracycline for 24 h and then treated with DMSO or antioxidant (tBHQ) for an additional 4 h. FLAG-INrf2 protein was purified by anti-FLAG beads, the proteins were separated by SDS-PAGE and stained with Coomassie Blue. INrf2 bands were sliced and subjected to mass spectrometry for phosphorylation analysis. Mass spectrometry analysis identified that threonine 55 (Thr-55) of INrf2 was phosphorylated in cells treated with tBHQ (supplemental Fig. S3). To further confirm threonine 55 phosphorylation of INrf2 we used the site-directed mutagenesis approach to mutate INrf2T55 to alanine (INrf2T55A). Hepa-1 cells co-transfected with FLAG-tagged wild type INrf2 or mutant Flag-INrf2T55A with Hsp90-V5 plasmids and lysates were immunoblotted (Fig. 4A, left panel). All proteins were expressed in comparable amounts in Hepa-1 cells (Fig. 4A,  left panel). Next, using the same cell lysate, we analyzed threo-

Hsp90-INrf2(Keap1) Interaction and Nrf2 Activation
nine phosphorylation of wild type and mutant INrf2T55A proteins by IP and immunoblotting. Forward and reverse IP and immunoblot analysis showed a significant decrease in threonine phosphorylation of the INrf2T55A mutant protein, as compared with the wild type INrf2 protein (Fig. 4A, right  panel). Furthermore, using the same cell lysates we analyzed the interaction of wild type INrf2 or INrf2T55A with transfected Hsp90-V5 protein (Fig. 4B). Forward and reverse IP and immunoblot analysis showed that FLAG-INrf2 interacted with Hsp90-V5. However, interaction between FLAG-INrf2T55A and Hsp90-V5 was completely abolished (Fig. 4B). This suggested that phosphorylation of threonine 55 of INrf2 was required for the INrf2-Hsp90 interaction.
PROSITE search revealed that INrf2Thr55 is a putative site of phosphorylation by CK2 and is conserved across species (Fig. 5A). We used a CK2-specific inhibitor to investigate the role of CK2 in INrf2 phosphorylation and INrf2-Hsp90 interaction. Hepa-1 cells treated with DMSO or different doses of the CK2 inhibitor for 18 h and CK2, INrf2, and Hsp90 protein levels were analyzed by immunoblotting (Fig. 5B, left panels). The CK2 and Hsp90 protein levels did not change in cells treated with the CK2 inhibitor (Fig. 5B, left panels). However, INrf2 showed CK2 inhibitor concentration-dependent decreases (Fig. 5B, left panel). Interestingly, forward, reverse IP, and immunoblotting results demonstrated significant loss of INrf2Thr phosphorylation and INrf2-Hsp90 interaction in cells treated with the CK2 specific inhibitor (Fig. 5B, middle and right panels). Experiments were also performed to test whether CK2-mediated phosphorylation of INrf2Thr55 and Hsp90 interaction with INrf2 protect INrf2 from ubiquitination and proteasomal degradation. Hepa-1 cells were pretreated with proteasome inhibitor MG132 (2 h) followed by treatment with different concentrations of the CK2 specific inhibitor and lysates were analyzed by immunoblotting (Fig.  5C). The data demonstrated that pre-treatment with MG132 significantly stabilized the cellular INrf2 protein level in CK2 inhibitortreated cells. MG132 also caused a minor increase in Hsp90 and CK2 levels in CK2 inhibitor-treated cells (Fig. 5C). The results collectively suggested that CK2-mediated INrf2Thr55 phosphorylation is required for INrf2-Hsp90 interaction and INrf2 stabilization.
Hsp90 Stabilizes INrf2-Next we performed experiments to investigate the functional significance of the INrf2-Hsp90 interaction. Hepa-1 cells transfected with Hsp90-V5 showed plasmid concentration-dependent overexpression of Hsp90 (Fig. 6A). Interestingly, the cells also showed a Hsp90-V5-dependent increase in INrf2 (Fig.  6A, left panel). The increase in INrf2 was parallel to the increase in Hsp90. This suggested that Hsp90 stabilized INrf2. Forward and reverse IP and immunoblot analysis revealed that a parallel increase in Hsp90 and INrf2 resulted in enhanced interaction between Hsp90-V5 and INrf2 (Fig. 6A, right panel). Similarly, siRNA inhibition of Hsp90 led to a concentration-dependent decrease in INrf2 (Fig. 6B). In related experiments, treatment of Hepa-1 cells with the Hsp90 inhibitor geldanamycin also led to a significant decrease in INrf2 (Fig. 6C, left panel). IP and immunoblot analysis demonstrated that a decrease in Hsp90 activity in geldanamycin-treated cells also led to a significant loss of INrf2-Hsp90 interaction (Fig. 6C, right panel). These results together suggested that alterations and inactivation in Hsp90 cause similar alterations in INrf2-Hsp90 interaction and INrf2 stability. The results also reveal that Hsp90 interaction with INrf2 leads to stabilization of INrf2.
Heat Shock and Antioxidant Induce Hsp90/CK2, INrf2Thr Phosphorylation, and Hsp90-INrf2 Interaction-We analyzed the role of Hsp90-INrf2 interaction in heat shock and antioxidant stress. Hepa-1 cells were incubated at 37 and 42°C for different time intervals and analyzed for Hsp90, INrf2, and Hsp90-INrf2 interaction (Fig. 7, A-C). Immunoblot analysis revealed that heat shock at 42°C led to a time-dependent increase in Hsp90, INrf2, and CK2 as compared with cells incu-

Hsp90-INrf2(Keap1) Interaction and Nrf2 Activation
localization of Nrf2, resulting in increased expression of Nrf2 downstream genes (2). To test whether heat shock also contributes Nrf2 activation and nuclear stabilization, we exposed Hepa-1 cells to 37 or 42°C for various time periods and cellular localization of Nrf2 was analyzed (Fig. 9A). The data demonstrated that heat shock increased a time-dependent nuclear localization of Nrf2 in Hepa-1 cells (Fig. 9A). Increasing nuclear localization of Nrf2 by heat shock also increased Nrf2 downstream NQO1 promoter luciferase activity similar to antioxidant (tBHQ) treatments (Fig. 9B, left and right panels). In addition, immunoblotting analysis clearly showed that exposure of cells to heat shock (7 h) also increased Nrf2 and NQO1 protein levels as well as Hsp90 and INrf2 levels similar to the tBHQ treatment (Fig. 9C, left and right panels). Collectively, the data presented in Figs. 8 and 9 suggest that heat shock or tBHQ treatment stabilized the INrf2⅐Hsp90 complex, destabilized the Rbx1/Cul3⅐INrf2⅐Nrf2 complex, and increased stabilization and nuclear localization of Nrf2, resulting in increased expression of the Nrf2 downstream gene NQO1.
Removal of Heat Shock or Antioxidant (tBHQ) Leads to Down-regulation of Hsp90 and Restoration of Normal Interaction with INrf2-Hepa-1 cells were incubated either at 37 or 42°C for 7 h. Two sets of cells exposed to heat shock (42°C for 7 h) were returned for incubation at 37°C for another 4 and 8 h and cell lysates were immunoblotted with Hsp90, INrf2, Nrf2, GCLC, and NQO1 antibodies (Fig. 10A, left panel). The results demonstrated that heat shock exposure led to an increase in Hsp90 and stabilization of INrf2, and an increase in Nrf2 and Nrf2 downstream GCLC and NQO1 gene expression. These results are consistent with the results from earlier in our study. Interestingly, removal of heat shock by incubation at 37°C for 8 h led to restoration of Hsp90, INrf2, and Nrf2 to normal levels and increased the GCLC and NQO1 proteins to basal and constitutive levels. However, Nrf2 downstream genes continued to express at higher levels at 8 h of transfer of cells from 42 to 37°C (Fig. 10A). The same Hepa-1 cell lysates were also subjected to IP and immunoblot analysis to investigate INrf2Thr phosphorylation and interaction of INrf2 with Hsp90 and Nrf2 (Fig. 10A, right  panel). The results showed that removal of heat shock (8 h) led to a decrease in INrf2Thr phosphorylation, INrf2-Hsp90 interaction, and increased interaction of INrf2 with Nrf2 (Fig. 10A, right panel). The results also show that the INrf2-Nrf2 interaction increased during heat shock release leads to an increase in Nrf2 ubiquitination and degradation (Fig. 10, A and B). Similar results were obtained when cells were exposed for 8 h to tBHQ and released after 4 -8 h (Fig. 10,  C and D). The data clearly demonstrate that removal of heat shock or tBHQ treatments decreased Hsp90 and INrf2Thr phosphorylation leading to decreased Hsp90-INrf2 interaction, and increased in INrf2-Nrf2 interactions that leads to restoration of normal/basal levels of the various proteins.

DISCUSSION
INrf2 is known to function as an adaptor protein that binds to Cul3/Rbx1 and Nrf2 (10). This leads to degradation of Nrf2. Under basal conditions, the Rbx1/Cul3⅐INrf2 complex is constantly degrading Nrf2. In response to chemical and radiation stress, Nrf2 is released and stabilized. Nrf2 then translocates in the nucleus leading to activation of more than 200 defensive genes critical for cellular protection. One mechanism of Nrf2 release from INrf2 involves chemical modification of INrf2Cys151 followed by PKC␦ phosphorylation of Nrf2S40 (2). In the present report, we present evidence for a novel mechanism involving Hsp90 and CK2 that also contributes to the activation of Nrf2.
Mass spectra and IP and immunoblot studies showed that INrf2 interacts with Hsp90. Domain deletion studies revealed that the NTR region of INrf2 interacts with the charge-linked domain (CLD) region of Hsp90. The deletion of NTR from INrf2 or the deletion of CLD from Hsp90 abrogated the INrf2-Hsp90 interaction. This is the first report on INrf2 interaction with Hsp90. This is also the first report of the role of the NTR region of INrf2 in Nrf2/ARE-mediated gene expression. The FIGURE 9. Heat shock and antioxidant both induce Nrf2 activation and increased ARE-luciferase gene expression. A, heat shock increases nuclear localization of Nrf2. Hepa-1 cells were incubated at 37°C for 7 h or 42°C for 1, 3, and 7 h. Cells were harvested and cytosolic and nuclear extracts were prepared as described under "Experimental Procedures." 80 g of cell lysates were immunoblotted with anti-Hsp90, anti-Nrf2, anti-INrf2, anti-lactate dehydrogenase (LDH), and anti-lamin-B antibodies. B, NQO1 ARE-luciferase activity assay. Hepa-1 cells were transfected with hNQO1-ARE luciferase promoter for 24 h and one set of cells were treated with DMSO or tBHQ for 7 h and another set of cells incubated at 42°C for various time periods as indicated. Heatshocked cells were further incubated at 37°C for an additional 12 h, cells were harvested, and NQO1 promoter luciferase activity was measured as described under "Experimental Procedures." C, immunoblot analysis. Hepa-1 cells were treated with DMSO or tBHQ for 8 h or incubated at 37 or 42°C for 7 h. Eighty g of cell lysates were immunoblotted with the indicated antibodies.
role of the DGR region of INrf2 that is known to bind to Nrf2 in Nrf2/ARE-mediated gene expression is well established (10). The crystal structure reveals that Hsp90 consists of three highly conserved domains, the N-terminal ATP-binding domain (25 kDa), a middle domain (35 kDa), and a C-terminal dimerization domain (12 kDa) (28 -30). Hsp90 exists as a homodimer (31). The N-terminal domain contains a specific ATP binding pocket, which has been well characterized (28,31). The middle domain is highly charged and its major role is to distinguish various types of client proteins and adjust the molecular chaperone for proper substrate activation (32). The C-terminal domain strengthens the weak association between the two N-terminal domains of the Hsp90 dimer. A second ATP-binding site is located in the C terminus, which does not exhibit ATPase activity (33). The molecular chaperone, Hsp90, facilitates the maturation and/or activation of over 100 client pro-teins involved in signal transduction and transcriptional regulation (34 interaction all returned to basal levels. Therefore, it is clear that the Hsp90/CK2-INrf2 mechanism is switched "on" in response to stress and is switched "off" when stress is subsides. Although we have investigated the mechanism of switch on in the current study, the mechanism of switch off remains obscure. The mechanism of the stress-induced increase in Hsp90 and CK2 also remains unknown. It is possible that an increase in Hsp90 leads to interaction with CK2 leading to stabilization of CK2. However, this has to be determined by experiments. It is noteworthy that the Hsp90 protein is known to interact with CK2 and protect CK2 from self-aggregation resulting in increased CK2 kinase activity (22).
The model shown in supplemental Fig. S5 depicts a two (early and delayed) phase mechanism of Nrf2 activation in response to antioxidant stress. The early phase of Nrf2 activation involves modification of INrf2Cys151 and PKC␦-mediated phosphorylation of Nrf2S40 that occurs within 1-2 h after exposure to antioxidants (2). The delayed phase spans 2 to 7 or Hepa-1 cells were pre-treated with the proteasome inhibitor MG132 for 4 h and cells were further incubated at various temperatures as shown in A. One mg of cell lysates were immunoprecipitated with control IgG or anti-Nrf2 antibody and immunoblotted with anti-ubiquitin antibody. C, removal of antioxidant (tBHQ) stress. Hepa-1 cells were treated with DMSO or tBHQ for 8 h and further incubated without tBHQ for 4 and 8 h for release. Cells were harvested and 60 g of cell lysates were immunoblotted with the indicated antibodies (left input panel). For analyzing the interaction of Hsp90-INrf2-Nrf2, 1 mg of the same cell lysates were immunoprecipitated with anti-INrf2 antibody and immunoblotted with anti-Hsp90, anti-phosphothreonine, anti-Nrf2, and anti-INrf2 antibodies. D, Nrf2 ubiquitination analysis. Hepa-1 cells were pre-treated with the proteasome inhibitor MG132 for 4 h and the cells were further treated with DMSO or tBHQ or release as shown in C. One mg of cell lysates were immunoprecipitated with control IgG or anti-Nrf2 antibody and immunoblotted with anti-ubiquitin antibody. All experiments were repeated three times and one representative set of data are shown.

Hsp90-INrf2(Keap1) Interaction and Nrf2 Activation
8 h and involves induction of Hsp90 and CK2, CK2 phosphorylation of INrf2Thr55, Hsp90 interaction with phospho-INrf2Thr55 to sequester INrf2, destabilization of Rbx1/ Cul3⅐INrf2⅐Nrf2, uninhibited nuclear localization of Nrf2, and induction of downstream gene expression. In other words, the Hsp90/CK2 mechanism prolongs antioxidant activation of Nrf2. Interestingly, heat shock stress has only a delayed phase of Nrf2 activation presumably due to the absence of chemical modification reactions, a characteristic of the early phase of activation (supplemental Fig. S5).
The present studies demonstrated Hsp90 regulation of INrf2/Nrf2 signaling. This raised an interesting question regarding INrf2 regulation of Hsp90 in normal and stressed conditions because both interacted with each other. The Hsp90 chaperone protein is a remarkably versatile protein involved in stress response and in normal homeostatic control mechanisms (35,36). It interacts with protein kinases, transcription factors, and other proteins, and either facilitates their stabilization as observed for INrf2 or directs them for proteasomal degradation. By this means, Hsp90 displays a multifaceted ability to influence signal transduction, chromatin remodeling, and epigenetic regulation, development and morphological evolution with implications in cell growth, apoptosis, transformation, and cancer. INrf2 titration of Hsp90 might or might not influence these processes and remains to be investigated.
In conclusion, we demonstrated that heat shock and antioxidant treatment induced Hsp90 and CK2. The activated CK2 phosphorylated INrf2Thr55. This led to an increased Hsp90-INrf2 interaction, dissociation of the Rbx1/Cul3⅐INrf2⅐Nrf2 complex, release of Nrf2, and activation of Nrf2 downstream gene expression. Hsp90, INrf2, Hsp90-INrf2 interaction, CK2, INrf2Thr phosphorylation, Nrf2, and Nrf2 downstream gene expression all returned to basal levels once the heat shock or antioxidant stress subsided. INrf2 interacts with Nrf2 and facilitates Nrf2 ubiquitination and degradation. The results together demonstrate a novel role of stress-induced Hsp90-INrf2 interaction in regulation of Nrf2 activation and induction of chemoprotective proteins.