Inhibitor of Nrf2 (INrf2 or Keap1) Protein Degrades Bcl-xL via Phosphoglycerate Mutase 5 and Controls Cellular Apoptosis*

INrf2 (Keap1) is an adaptor protein that facilitates INrf2-Cul3-Rbx1-mediated ubiquitination/degradation of Nrf2, a master regulator of cytoprotective gene expression. Here, we present evidence that members of the phosphoglycerate mutase family 5 (PGAM5) proteins are involved in the INrf2-mediated ubiquitination/degradation of anti-apoptotic factor Bcl-xL. Mass spectrometry and co-immunoprecipitation assays revealed that INrf2, through its DGR domain, interacts with PGAM5, which in turn interacts with anti-apoptotic Bcl-xL protein. INrf2-Cul3-Rbx1 complex facilitates ubiquitination and degradation of both PGAM5 and Bcl-xL. Overexpression of PGAM5 protein increased INrf2-mediated degradation of Bcl-xL, whereas knocking down PGAM5 by siRNA decreased INrf2 degradation of Bcl-xL, resulting in increased stability of Bcl-xL. Mutation of PGMA5-E79A/S80A abolished INrf2/PGAM5/Bcl-xL interaction. Therefore, PGAM5 protein acts as a bridge between INrf2 and Bcl-xL interaction. Further studies showed that overexpression of INrf2 enhanced degradation of PGAM5-Bcl-xL complex, led to etoposide-mediated accumulation of Bax, increased release of cytochrome c from mitochondria, activated caspase-3/7, and enhanced DNA fragmentation and apoptosis. In addition, antioxidant (tert-butylhydroquinone) treatment destabilized the Nrf2-INrf2-PGAM5-Bcl-xL complex, which resulted in release of Nrf2 in cytosol and mitochondria, release of Bcl-xL in mitochondria, increase in Bcl-xL heterodimerization with Bax in mitochondria, and reduced cellular apoptosis. These data provide the first evidence that INrf2 controls Bcl-xL via PGAM5 and controls cellular apoptosis.

INrf2-null mice demonstrated persistent accumulation of Nrf2 in the nucleus and led to postnatal death from malnutrition, resulting from hyperkeratosis in the esophagus and fore stomach (10). Reversed phenotype of INrf2 deficiency by breeding to Nrf2-null mice suggested that a tightly regulated negative feedback might be essential for cell survival (11). The systemic analysis of 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 INrf2 affinity for Nrf2 and elevated expression of cytoprotective genes, which resulted in drug resistance and cell survival in lung cancer cells (12,13). Unrestrained activation of Nrf2 in cells increases a risk of adverse effects, including survival of damaged cells, tumorigenesis, and drug resistance (2). Therefore, it appears that cells contain mechanisms that autoregulate cellular abundance of Nrf2 (14,15). Structural and functional analyses of INrf2 identified an evolutionarily conserved Kelch (DGR) domain, which interacts with several proteins. Although Nrf2 is a well known substrate for INrf2, the DGR domain of INrf2 has been reported also to bind other proteins, including Nrf1, PGAM5, prothymosin-␣, fetal Alz clone 1, and IKK␤ (16 -20). It is noteworthy that binding of a protein with the INrf2DGR region does not always lead to degradation of the protein. Recently, we have shown that prothymosin-␣ interacts with the DGR domain of INrf2, and this interaction is required for nuclear localization of INrf2 (21). Therefore, INrf2 and its interacting partners play several different roles in cell signaling and survival.
Cellular apoptosis is a critical process that is dysregulated in tumorigenesis (22). Bcl-2 family proteins regulate cell death and survival (23,24). The Bcl-2 family includes more than six anti-apoptotic proteins, including Bcl-2 and Bcl-xL, and many pro-apoptotic members such as Bax and Bak (25,26). Bcl-2 and Bcl-xL share regions of sequence similarity as well as a C-terminal hydrophobic region required for membrane localization (27). Bcl-2 and Bcl-xL appear to function in the same apoptotic pathway, and both confer resistance to multiple chemotherapy agents. Overexpression of either protein is usually associated with poor prognosis in many human cancers. However, in some cancer types, multiple anti-apoptotic proteins are expressed (28,23) and have opposite effects on prognosis, indicating that there may be subtle but clinically and biologically relevant functional differences between family members. Experiments in mice with deletion of individual anti-apoptotic genes indicate that the phenotypes are not identical, presumably because of differential tissue expression of the various members (29). The mechanisms of action of Bcl-2 and Bcl-xL are complex, with many postulated interactions with other proteins, and the role of any single interaction in the final phenotype at the cellular level remains unknown.
Recently, INrf2 is shown to target anti-apoptotic Bcl-2 protein for degradation and control cellular apoptosis (30). In the present report, we investigated the novel role of INrf2 in the regulation of anti-apoptotic factor Bcl-xL. INrf2, through its DGR domain, interacts with PGAM5 proteins, which interact with Bcl-xL. Interestingly, we show that INrf2-Cul3-Rbx1 complex facilitates both PGAM5 and Bcl-xL ubiquitination and degradation. The data also revealed that PGAM5 proteins control INrf2-mediated degradation of Bcl-xL. Therefore, PGAM5 protein acts as a bridge between INrf2 and Bcl-xL interaction. Further, studies showed that overexpression of INrf2 degrades both PGAM5 and Bcl-xL proteins, which increases/activates cellular pro-apoptotic factors and apoptosis. However, antioxidant (t-BHQ) 2 treatment destabilized Nrf2-INrf2-PGAM5-Bcl-xL complex in mitochondria, leading to the release of Nrf2 and increased Bcl-xL heterodimerization with Bax and reduced cellular apoptosis.
Subcellular Fractionation and Western Blotting-Hepa-1 cells were seeded in 100-mm plates and transfected/treated as displayed in the figures. For making whole cell lysates, the cells were lysed in RIPA buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 0.2 mM EDTA, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride, and 1 mM sodium orthovanadate) supplemented with 1ϫ protease inhibitor mixture (Roche Applied Science). Mitochondrial and cytosolic lysates were prepared by standard procedures. The isolated mitochondria were washed with HEPES buffer and lysed in RIPA buffer. 60 -80 g of proteins were separated on SDS-PAGE and transferred to nitrocellulose membranes. The membranes were blocked with 3% nonfat dry milk and incubated with anti-INrf2 (E-20) (1:1000), anti-Nrf2 (H-300) (1:1000), anti-Bcl-xL (H5) (1:1000), anti-Bax (N20) (1:1000), anti-Bcl-2(N-19) (1:1000), anti-Mcl-1(S-19) (1:1000), and anti-ubiquitin (P4D1) (1:1000) antibodies, all purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-FLAG-HRP, anti-HA-HRP, and anti-␤-actin antibodies were obtained from Sigma and used in 1:10,000 dilutions to probe the Western blots. Anti-V5 antibody and anti-V5-HRP antibody were obtained from Invitrogen, and anti-caspase-3 antibody was purchased from Cell Signaling. To confirm the purity of cytoplasmic and mitochondrial protein fractionation, the membranes were reprobed with cytoplasm-specific, anti-lactate dehydrogenase (Chemicon) and mitochondria-specific anti-cytochrome c or CoxIV antibodies (Cell Signaling). Jurkat cells were treated with 100 ng/ml Killer TRAIL soluble human recombinant protein (Enzo Life Sciences, catalog no. ALX-201073). We generated and purified bacterial PGAM5L-His-tagged protein. The polyclonal antibodies against the PGAM5L form were generated in rabbits (Pacific Immunology) and purified. The membranes were washed three times with TBST, and immunoreactive bands were visualized using a chemiluminescence ECL system (Amersham Biosciences). The intensity of protein bands after Western blotting were quantified by using QuantityOne version 4.6.3 image software (ChemiDoc XRS, Bio-Rad) and normalized against proper loading controls. In related experi-ments, the cells were treated with 50 M t-BHQ or DMSO as a vehicle for different time intervals.
Immunoprecipitation-For immunoprecipitation, 1 mg of whole cell extracts or 300 g of mitochondrial lysates were equilibrated in RIPA buffer and precleaned by Protein AG Plusagarose (Santa Cruz Biotechnology, Inc.), and then extracts were incubated with respective antibodies (1 g) at 4°C overnight. Immune complexes were collected by the addition of Protein AG-agarose and centrifugation. The immune complexes were washed three times with RIPA buffer containing 0.25% Nonidet P-40, and proteins were resolved by 10 -12% reducing SDS-PAGE and transferred to nitrocellulose membrane. The membranes were blocked with 3% nonfat dry milk and incubated with their respective primary and secondary antibodies. Immunoreactive bands were visualized using a chemiluminescence ECL system (Amersham Biosciences). Cell pellets were lysed in RIPA buffer containing 1% SDS. One mg of the lysate (ϳ100 l) was diluted to 10-fold with RIPA buffer. After precleaning, samples were immunoprecipitated with 2 g of antibody or anti-FLAG beads (15 l), as indicated in the figures. Immune complexes were collected by the addition of Protein AG-agarose. Immune complexes were boiled with SDS sample buffer, and denatured samples were resolved by SDS-PAGE followed by immunoblotting with anti-HA and anti-ubiquitin antibodies.
Transient Transfection/siRNA Interference Assay-Hepa-1 cells were plated in 100-mm plates at a density of 1 ϫ 10 5 cells/ plate 24 h prior to transfection. In the related experiments, the cells were transfected with 1 g of the indicated plasmids using Effectene transfection reagent (Qiagen) according to the manufacturer's instructions. After 36 h of transfection, cells were harvested and immunoblotted. INrf2 siRNA, PGAM5L siRNA, and Bcl-xL siRNA were used to inhibit INrf2, PGAM5L, and Bcl-xL proteins, respectively. Control GAPDH siRNA and INrf2 siRNA were purchased from Dharmacon. PGAM5L siRNA and Bcl-xL siRNA were obtained from Ambion. In most cases, Hepa-1 cells were transfected with 25, 50, and 75 nM INrf2, PGAM5L, and Bcl-xL siRNA separately using Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer's instructions. Thirty-two hours after transfection, cells were harvested, and INrf2, PGAM5L, Bcl-xL, and actin proteins were analyzed by Western blotting. In related experiments, Jurkat cells were transfected with 25 and 50 nM INrf2 siRNA. siRNA was mixed with Lipofectamine RNAiMAX reagent in Opti-MEM medium (1 ml), and the mixture was incubated at room temperature for 15 min. Then the reaction mixture was coated onto 100-mm tissue culture plates for 10 min. Exponentially grown Jurkat cell suspensions in RPMI medium without antibiotics (4 ml; 10 6 cells) were added into the plates, and cells were incubated at 37°C for 24 h followed by treatment with TRAIL protein (100 ng/ml) for 30 h. Cells were harvested, lysed, and immunoblotted.
Immunofluorescence-Hepa-1 cells were grown in Lab-Tek II chamber slides. Cells were fixed in 2% formaldehyde and permeabilized by the treatment of 0.25% Triton X-100. Cells were washed twice with PBS and incubated with a 1:1000 dilution of sheep cytochrome c antibody along with goat INrf2, rabbit PGAM5, and mouse Bcl-xL antibody separately at 4°C for 12 h. Then cells were washed twice with PBS and incubated with anti-sheep FITC-conjugated second antibody or Alexa Fluor-594-conjugated anti-goat, anti-rabbit, and anti-mouse second antibodies (Invitrogen). After immunostaining, cells were washed twice with PBS, stained with Vectashield containing nuclear DAPI stain, and mounted. Cells were observed under a Nikon fluorescence microscope and photographed.
Real-time PCR-Hepa-1 cells were transfected with increasing concentrations of FLAG-INrf2, or endogenous INrf2 expression was knocked down by siRNA (25-75 nM) for 30 h, and cells were harvested. The total RNA was isolated using the RNeasy minikit (Qiagen). 250 ng of total RNA was subjected to reverse transcription using a high capacity cDNA reverse transcription kit (Applied Biosystems). After synthesis of cDNA at 37°C for 60 min, the PCR was performed using the 7500 real-time PCR system as per the manufacturer's instructions. Bcl-xL (ID: Mm00437783_m1) and control Gusb (ID: Mm00446953_m1) TaqMan gene expression assay probe primers were used for PCR and quantitative Bcl-xL gene expression. The data were analyzed and plotted.
DNA Fragmentation Assay-Hepa-1 cells or control 293 cells or INrf2-293 cells were plated at a density of 2000 cells/well in 96-well plates. After 20 h, Hepa-1 cells were transfected with pcDNA or INrf2-V5 construct for 12 h. Similarly, control 293 cells and INrf2-293 cells were treated with 0.5 g/ml tetracycline for 12 h, and all cells were exposed to varying concentrations of etoposide for 72 h. A photometric enzyme immunoassay was performed for the quantitative in vitro determination of cytoplasmic histone-associated DNA fragments (mono-and oligonucleosomes) after etoposide exposure to cells using the Cell Death Detection ELISA kit (Roche Applied Science) and as per the manufacturer's instructions. Each combination of cell line and drug concentration was set up in eight replicate wells, and the experiment was repeated three times. Each data point represents a mean Ϯ S.D. and is normalized to the value of the corresponding control cells.
TUNEL Assay-For the TUNEL assay, Hepa-1 cells were transfected with the indicated plasmids treated with etoposide or etoposide plus t-BHQ (50 M) for 48 h. The Dead End Fluorometric TUNEL assay kit (Promega) was used as per the manufacturer's protocol. TUNEL-positive cells were counted from three independent experiments and plotted.
MTT Cell Survival Assay-Hepa-1, Hek293, HepG2, and Jurkat cells were plated at a density of 5000 cells/well in 24-well plates and allowed to recover for 12 h. Then cells were transfected with the indicated constructs or control siRNA or INrf2 siRNA for 20 h, and cells were treated with etoposide or etoposide plus t-BHQ for 48 h as indicated in the figures. Jurkat cells were also transfected with control siRNA or INrf2 siRNA and treated with TRAIL (100 ng/ml) for 30 h. Cells were incubated with fresh MTT solution (200 l/well; stock 5 mg/ml in PBS) at 37°C for 2 h, and absorbance at 570 nm was measured. Each combination of cell line and drug concentration was set up in eight replicate wells, and the experiment was repeated three times. Each data point represents a mean Ϯ S.D. and is normalized to the value of the corresponding control cells.
Statistical Analyses-The data real-time PCR and immunoblotting band intensities were analyzed using a two-tailed Student's t test. Data are expressed as mean Ϯ S.D. of three independent experiments. The error bars indicate 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).  showed siRNA inhibition of INrf2 and marginal increases in Bcl-xL mRNA (Fig. 1C, bottom). The results together suggested that INrf2 mediated degradation of PGAM5 and Bcl-xL.

DGR Domain of INrf2 Is Required for Interaction and Ubiquitination/Degradation of PGAM5-Bcl-xL-INrf2DGR
domain is known to interact with Nrf2, leading to ubiquitination and degradation of Nrf2 (21). We generated two additional stable FRT-Hek-293 cells that upon exposure to tetracycline express FLAG-INrf2DGR and FLAG-INrf2⌬DGR (Fig. 4A) for use in examining the requirement of the DGR domain for interaction with PGAM5-Bcl-xL. Control 293, INrf2-293, INrf2DGR-293, and INrf2⌬DGR-293 cells were treated with tetracycline for 24 h. Ten mg of cell lysates were immunoprecipitated using anti-FLAG antibodies, the immune complexes were separated by SDS-PAGE, and gels were stained with Coomassie Brilliant Blue (Fig. 4A). FLAG-INrf2, FLAG-INrf2DGR, and FLAG-INrf2⌬DGR proteins were strongly expressed in these cells as denoted by asterisks above the bands (Fig. 4A). Interestingly, PGAM5 bands were pulled down along with  4A). This suggested that the DGR domain is required for interaction with PGAM5-Bcl-xL. This was also supported by immunoprecipitation and immunoblotting experiments (Fig. 4, B and C). INrf2-293, INrf2DGR-293, and INrf2⌬DGR-293 cells were transfected with PGAM5-V5 constructs and treated with tetracycline for 12 h, and 1 mg of cell lysates were immunoprecipitated with anti-FLAG or anti-V5 antibodies and immunoblotted with the same antibodies (Fig. 4B). Immunoprecipitation of INrf2 and INrf2DGR pulls down both PGAM5L-V5 and Bcl-xL proteins (Fig. 4B, left). In addition, immunoprecipitation of PGAM5-V5 pulls down FLAG-INrf2 and FLAG-INrf2DGR (Fig. 4B, right). However, in the same experiments, INrf2⌬DGR failed to pull down either PGAM5L-V5 or Bcl-xL protein, and PGAM5-V5 failed to pull down INrf2⌬DGR (Fig. 4B, left and  right).
Next, we analyzed the ubiquitination and degradation of PGAM5L-V5 and Bcl-xL proteins in INrf2-, INrf2DGR-, and INrf2⌬DGR-expressing cells after transfection with PGAM5-V5 and HA-ubiquitin constructs (Fig. 4C). Overexpression of INrf2 in INrf2-293 cells by tetracycline degraded both PGAM5L-V5 and Bcl-xL protein and significantly increased PGAM5 and Bcl-xL ubiquitination (Fig. 4C). Interestingly, overexpression of INrf2DGR domain and INrf2⌬DGR protein both failed to degrade PGAM5L-V5 or Bcl-xL protein (Fig. 4C, left) and also failed to ubiquitinate PGAM5L and Bcl-xL proteins (Fig. 4C, right panels). These data clearly demonstrated that the DGR domain of INrf2 was required for binding with the PGAM5L-Bcl-xL complex. However, the DGR domain of INrf2 was not sufficient for ubiquitination and degradation of PGAM5L and Bcl-xL protein because Cul3-Rbx1 ubiquitin E3 ligase complex binds with the BTB domain at the N terminus of INrf2 and mediates ubiquitination and degradation. Therefore, both the BTB and DGR domains of INrf2 are required for ubiquitination and degradation of the PGAM5-Bcl-xL complex.
INrf2 Physically Interacts with PGAM5 but Not Bcl-xL-Several experiments were performed to examine if INrf2 directly interacts only with PGAM5 or with both PGAM5 and Bcl-xL. To investigate, we knocked down PGAM5L protein by siRNA in Hepa-1 cells, and the levels of PGAM5, Bcl-xL, INrf2, and actin were analyzed by Western blotting (Fig. 5A). Transient transfection of PGAM5 siRNA (50 to 75 nM) decreased PGAM5 protein by 60 -80%. However, Bcl-xL protein levels significantly increased (2-2.5-fold) upon PGAM5 knockdown (Fig.  5A, left). Using the same cell lysates, we analyzed INrf2-PGAM5 and INrf2-Bcl-xL interaction by immunoprecipitation and immunoblotting (Fig. 5A, right). Transfection of PGAM5 siRNA decreased the interaction between INrf2 and PGAM5, as expected, whereas it also decreased INrf2/Bcl-xL interaction to the same magnitude (Fig. 5A, right), suggesting that PGAM5L is required for INrf2/Bcl-xL interaction. PGAM5 at the N terminus is known to contain a motif, NXESGE, that is similar to the Nrf2 motif DEETGE (17). Both of these motifs are binding sites for other proteins. To test whether the 77 NXESGE 82 motif in PGAM5L is involved in the binding to INrf2, a mutant PGAM5L protein was generated in which two alanine substitutions were introduced in place of Glu-79 and Ser-80. V5-tagged plasmids of the wild-type PGAM5L and PGAM5L-E79A/S80A mutant proteins were transfected in Hepa-1 cells along with HA-INrf2 and FLAG-Bcl-xL constructs for 30 h, and cell lysates were immunoblotted (Fig. 5B, top). Immunoblotting data alone indicate that in mutant PGAM5Ltransfected cells, the levels of mutant PGAM5 and Bcl-xL protein were ϳ1.5-fold more than in wild type PGAM5L-transfected cells, suggesting that the NXESGE motif of PGAM5L is required for the binding to INrf2. To support this observation, we performed the forward and reverse immunoprecipitation and immunoblotting experiments. The mutant PGAM5L-E79A/S80A protein did not bind with INrf2 (Fig. 5B, bottom  left), whereas Bcl-xL interaction was the same with wild type and mutant PGAM5. FLAG-Bcl-xL interaction with INrf2 was observed in wild type PGAM5-V5-tranfected cells. However, FLAG-Bcl-xL and HA-INrf2 interaction was abolished in mutant PGAM5L-E79A/S80A-V5-transfected cells (Fig. 5B,  bottom right). In addition, we further investigated whether Bcl-xL has any role in INrf2-PGAM5L interaction. For this, we knocked down Bcl-xL protein by siRNA, and interactions between INrf2-PGAM5L and PGAM5L-Bcl-xL were analyzed (Fig. 5C, left and right). Silencing of Bcl-xL protein by siRNA decreased Bcl-xL protein, whereas PGAM5L and INrf2 levels remained the same (Fig. 5C, left). Importantly, immunoprecipitation/immunoblotting data clearly indicate that knockdown of Bcl-xL protein has no effect on INrf2 and PGAM5L interactions (Fig. 5C, right panels). These results together suggested that INrf2 physically interacts with PGAM5 but not Bcl-xL. In addition, INrf2 interaction with Bcl-xL required PGAM5.
PGAM5 Is Required for Mitochondrial Localization of Nrf2-INrf2-PGAM5-Bcl-xL Complex-PGAM5L contains mitochondrial localization signal between amino acids 9 and 29 and is shown to localize in the mitochondria (31). Therefore, we performed immunohistochemistry analysis to investigate the co-localization of Nrf2-INrf2-PGAM5-Bcl-xL complex to the mitochondria. Imunocytochemistry analysis clearly showed the co-localization of INrf2, PGAM5, and Bcl-xL proteins with mitochondrial cytochrome c protein (Fig. 6A). Interestingly, immunohistochemistry analysis of Nrf2 in the same experiment demonstrated that Nrf2 also co-localized with INrf2-PGAM5-Bcl-xL complex in the mitochondria (Fig. 6A). These results suggested that Nrf2-INrf2-PGAM5-Bcl-xL complex localizes to the mitochondria. This is also supported by a single report earlier (31). Next we examined the role of PGAM5 on the localization of Nrf2-INrf2-PGAM5-Bcl-xL complex in the  (Fig. 6B). Hepa-1 cells were transfected with control and PGAM5 siRNA, and cytoplasmic and mitochondrial fractions were isolated and immunoblotted for PGAM5, Bcl-xL, INrf2, and Nrf2. The immunoblot was also probed with anti-lactate dehydrogenase (cytosolic marker) and anti-CoxIV (mitochondrial marker). Results revealed that PGAM5 knockdown by siRNA significantly reduced the mitochondrial localization of Nrf2-INrf2-PGAM5-Bcl-xL (Fig. 6B). This suggested that PGAM5 proteins are involved in the trafficking of Nrf2-INrf2-PGAM5-Bcl-xL complex to the mitochondria.
Antioxidant Treatment Led to the Release of Bcl-xL-INrf2 contains reactive cysteines (Cys-151, Cys-272, and Cys-282) that in response to chemicals/radiation are oxidized, leading to destabilization of the dimeric structure of INrf2, degradation of INrf2, and release of Nrf2 (1,2). In the present report, we studied the effect of antioxidant t-BHQ on the INrf2-PGAM5L-Bcl-xL interaction. Hepa-1 cells transfected with PGAM5-V5 were exposed to DMSO (vehicle control) or t-BHQ (50 M) for different time periods (2-8 h), and cell lysates were immunoblotted for INrf2, PGAM5L-V5, and Bcl-xL (Fig. 7A). DMSO showed virtually no significant effect on endogenous levels of INrf2, PGAM5L-V5, and Bcl-xL. However, cells upon treatment with t-BHQ (between 2 and 4 h) showed decreased levels of INrf2 and PGAM5-V5 and increased levels of Bcl-xL (Fig.  7A). Forward and reverse immunoprecipitation followed by immunoblotting in similar experiments also analyzed INrf2/ PGAM5/Bcl-xL interactions (Fig. 7B). Forward IP results revealed t-BHQ exposure time-dependent loss in interaction of INrf2 with PGAM5 and Bcl-xL (Fig. 7B, right top three panels). Reverse IP in similar experiments also showed loss of interaction of PGAM5 with INrf2 and Bcl-xL (Fig. 7B, right bottom  three panels). The various results indicated that t-BHQ destabilized INrf2-PGAM5-Bcl-xL complex, leading to degradation of PGAM5 and stabilization of Bcl-xL.
Next we investigated the effect of t-BHQ on Nrf2, INrf2, PGAM5, Bcl-xL, and Bax interactions in cytosol and mitochon-dria (Fig. 7C). Hepa-1 cells were treated with DMSO or t-BHQ for 2 h, and cytoplasmic and mitochondrial fractions were isolated. Immunoprecipitation followed by immunoblotting analyzed the various interactions. Immunoprecipitation of PGAM5 pulled down Bcl-xL, INrf2, and Nrf2 in cytosolic and mitochondrial lysates from DMSO-treated Hepa-1 cells (Fig.  7C, left). The treatment with t-BHQ led to reduced interaction of PGAM5 with INrf2 and release of Nrf2 in cytosol as well as in mitochondria. t-BHQ treatment also led to release of Bcl-xL in mitochondria but not in cytosol (Fig. 7C, left). Similar results were obtained in the case of INrf2 immunoprecipitation (Fig.  7C, second panel from the left). Bcl-xL immunoprecipitation demonstrated that Bcl-xL/PGAM5 and Bcl-xL/INrf2 interaction was decreased significantly in the mitochondrial but not in the cytosolic compartment upon treatment with t-BHQ (Fig.  7C, third panel from the left). Bcl-xl protein did not show a strong interaction with Nrf2 in both compartments of the cells. In addition, Nrf2 immunoprecipitation data clearly showed that Nrf2 interacts with INrf2 and PGAM5 in DMSO-treated cells in cytoplasm and mitochondria; however, the interaction was decreased more than 90% when cells were treated with t-BHQ and again no interaction with Bcl-xL (Fig. 7C, right). Interestingly, immunoprecipitation of Bcl-xL and Bax data clearly showed that the mitochondrial interaction of Bcl-xL and Bax was increased after t-BHQ treatment as compared with DMSO treatment (Fig. 7D) and no change in cytoplasmic Bcl-xL and Bax interaction. Collectively, the results demonstrated that antioxidant t-BHQ led to release of Nrf2 in the cytosol and Nrf2 and Bcl-xL in the mitochondria. The release of Bcl-xL in mitochondria led to increased interaction with Bax that is expected to contribute to altered apoptosis.
Overexpression of INrf2 Led to Degradation of PGAM5-Bcl-xL and Enhancement of Etoposide-induced Cytochrome c Release, Up-regulation of Pro-apoptotic Bax, and Increase in Activated Caspase-3/7-Our data suggested that INrf2 mediated degradation of PGAM5-Bcl-xL protein (Fig. 1). This indi- cated that INrf2, through regulation of anti-apoptotic Bcl-xL protein, might influence apoptotic cell death/survival. Therefore, we examined the role of INrf2-mediated degradation of PGAM5 and Bcl-xL in cellular apoptosis. INrf2-293 cells were transfected with PGAM5-V5 and treated with tetracycline, and Hepa-1 cells transfected with INrf2 were treated with two different concentrations of etoposide and analyzed for PGAM5-Bcl-xL degradation, cytochrome C release, caspase-3/7 activity, and cleaved caspase-3 (Fig. 8). INrf2-293 and Hepa-1 cells overexpressing INrf2 showed degradation of PGAM5-Bcl-xL and an increase in Bax (Fig. 8A). Etoposide treatment marginally increased alterations in the various molecules. INrf2 overexpression followed by etoposide treatment in Hepa-1 cells also induced cytochrome c release from mitochondria to cytosol by 1.8-fold (Fig. 8B, compare lanes 2, 3, and 4; data shown only for Hepa-1), increased 1.5-2-fold caspase-3/7 activity (Fig. 8C), . One mg of the same cell lysates from transfected cells was immunoprecipitated with anti-FLAG or anti-V5 antibody, and the immune complexes were immunoblotted with anti-Bcl-xL, anti-FLAG or anti-V5 antibodies (right panels). C, t-BHQ causes Nrf2 release in the cytosol and Nrf2 and Bcl-xL release in the mitochondria. Hepa-1 cells were treated with DMSO or t-BHQ for 2 h, and 1 mg of cytosolic and 300 g of mitochondrial lysates were immunoprecipitated with anti-PGAM5L, anti-INrf2, anti-Bcl-xL, and anti-Nrf2 antibodies and immunoblotted with the indicted antibodies (top panels). The band intensities of the respective panels are shown (bottom panels). C, cytosolic; M, mitochondrial. D, t-BHQ treatment increased heterodimerization of Bcl-xL and Bax protein on mitochondria. Hepa-1 cells were treated with DMSO or t-BHQ for 2 h, and 1 mg of cytosolic and 300 g of mitochondrial lysates were immunoprecipitated with anti-Bcl-xL or anti-Bax antibodies and immunoblotted with indicted antibodies. All experiments were repeated two times, and one representative set of data is presented. and cleaved caspase-3 (Fig. 8D), as compared with etoposidetreated cells expressing endogenous levels of INrf2. These data suggested that INrf2 overexpression degrades PGAM5 and Bcl-xL proteins, promotes etoposide-mediated increases in cellular Bax level, releases cytochrome from mitochondria, and activates caspase-3/7. In similar experiments, we also examined the role of INrf2 in the regulation of anti-apoptotic factors Bcl-2 and Mcl-1. Increasing INrf2-V5 levels in Hepa-1 cells by transient transfection degraded not only Bcl-xL as described above but also Bcl-2 and Mcl-1 (supplemental Fig. S1A). Moreover, a dose-dependent siRNA-mediated INrf2 knockdown stabilized anti-apoptotic proteins Bcl-xL, Bcl-2, and Mcl-1 (supplemental Fig. S1B). These results together indicate that INrf2 down-regulates these anti-apoptotic factors in addition to Bcl-xL to control apoptosis. It is noteworthy that we have recently reported INrf2 degradation of anti-apoptotic factor Bcl-2 (30).  (Fig. 9A). The results demonstrated that overexpression of INrf2 followed by etoposide treatments enhanced DNA fragmentation at least 1.4 -2.1-fold in both cell lines compared with control cells (Fig. 9A, top and bottom).

Overexpression of INrf2 Increased and Treatment with Antioxidant t-BHQ Decreased
We also examined the role of PGAM5, a bridge protein between INrf2 and Bcl-xL, in Hepa-1 cell survival. Overexpression of PGAM5-V5 protein and etoposide treatments decreased cell survival by 20% compared with pcDNAtransfected cells (Fig. 9B, left). In contrast, PGAM5 knockdown by siRNA followed by etoposide treatments increased cell survival by 15-20% as compared with control siRNA-transfected cells (Fig. 9B, right), again suggesting that PGAM5 proteins are involved in Bcl-xL regulation and cell survival.
Next we determined the effect of antioxidant t-BHQ on etoposide-mediated apoptosis. This was analyzed by a TUNEL assay (Fig. 9C). Hepa-1 cells overexpressing INrf2 or Nrf2 were treated with etoposide and t-BHQ as shown in Fig. 9C, and the TUNEL-positive cells were observed under a microscope, counted, and plotted (Fig. 9C)  9C). DNA fragmentation/TUNEL data were further supported by cell survival assays (Fig. 9D). Overexpression of FLAG-INrf2 in Hepa-1 cells significantly reduced (ϳ18%) cell survival upon treatment with etoposide, as compared with etoposide-treated control cells expressing endogenous levels of INrf2. Nrf2 overexpression or t-BHQ treatment showed increased cell survival (20 -30%) compared with cells treated with etoposide alone (Fig. 9D).
siRNA-mediated Knockdown of INrf2 Increased Etoposideand TRAIL-mediated Cell Survival-We investigated the effect of siRNA-mediated inhibition of INrf2 on Bcl-xL stabilization and etoposide-mediated cell survival in three different cancer cell lines. Hepa-1, Hek-293, and HepG2 cells were transfected with different concentrations of either control siRNA or INrf2 siRNA for 24 h and treated with etoposide for 36 h, and cell survival analysis was performed by an MTT assay (Fig. 10A). A dose-dependent knockdown of INrf2 by siRNA in all three cell lines followed by etoposide treatment showed 20 -40% increased cell survival compared with control siRNAtransfected cells (Fig 10A, top, middle, and bottom). The reasons for increasing cell survival after knockdown of INrf2 were further examined in Heap-1 cells (Fig. 10B). INrf2 knockdown and etoposide treatment stabilized cellular Bcl-xL and also decreased INrf2-mediated Bcl-xL ubiquitination compared with control siRNA-transfected cells (Fig. 10B, left and right). These results together suggested that knockdown of INrf2 stabilized Bcl-xL and increased cell survival. Interestingly, we also used cell killer TRAIL protein, which is known to activate the extrinsic apoptotic pathway in cells by the activation of caspase-8. For this we used Jurkat (T cell Lymphoma) cells. Cells were transfected with control siRNA or INrf2 siRNA and treated with TRAIL protein as indicated (Fig. 10C, bottom), and cell survival was measured. Treatment of cells with TRAIL decreased cell survival by 40% compared with untreated cells. However, knockdown of INrf2 followed by TRAIL treatment increased cell survival by 20 -25% compared with control siRNA-transfected and TRAIL-treated cells (p Ͻ 0.01) (Fig.  10C, bottom). In addition, we also confirmed INrf2 knockdown and Bcl-xL stabilization in Jurkat cells by immunoblotting of the same cell lysates with INrf2, Bcl-xL, and actin antibodies (Fig. 10C, top). These data indicate that INrf2 knockdown led to increased level of Bcl-xL that contributed to resistance against extrinsic apoptotic pathway in Jurkat cells.
The above results together suggested that overexpression of INrf2 degraded the PGAM5-Bcl-xL complex, activated proapoptotic factors, and promoted etoposide-mediated cellular apoptosis. In contrast, the knockdown of INrf2, resulting in increased Nrf2, the overexpression of Nrf2, or t-BHQ treatment, leading to stabilization of Nrf2, all promoted cell survival. Therefore, INrf2 and Nrf2 play opposite roles in the regulation of cellular apoptosis.

DISCUSSION
to stabilize Bcl-2 that forms complex with Bax and contributes to decreased apoptosis and increased cell survival (30). Therefore, it is expected that antioxidant-induced stabilization of anti-apoptotic proteins Bcl-xL and Bcl-2 leads to reduced apoptotic cell death and increased cell survival.
PGAM5 exists in two isoforms that are identical in the N terminal 239 amino acids (17). The longer form (PGAM5L) contains 289 amino acids, and the shorter form (PGAM5-S) contains 255 amino acids. The 16 C-terminal amino acids in PGAM5S are not similar to those of the PGAM5L isoform. The N terminus of the PGAM5 protein contains a conserved NXESGE motif (amino acid 77-82), similar to Nrf2, that binds to the DGR region of INrf2, whereas the C-terminal PGAM domain (amino acids 125-156) binds anti-apoptotic factor Bcl-xL. Interestingly, both isoforms of human PGAM5 contain a large PGAM domain, which begins at amino acid 98 and extends to the C-terminal end (17). In addition to this, PGAM5 proteins also possess an N-terminal mitochondrial localization signal (amino acids 9 -29), which is involved in the mitochondrial localization of PGAM5 and its binding partners to the mitochondria. The present studies used the PGAM5L form. PGAML5-E79A/S80A failed to bind with INrf2 and ubiquitinate/degrade Bcl-xL. This indicated that PGAM5, through the 77 NXESGE 82 domain, binds to INrf2. Because this domain is present in both isoforms, we believe that PGAM5L and PGAM5S both function as a bridge between INrf2 and Bcl-xL.
A hypothetical model demonstrating the role of INrf2 control of Bcl-xL and apoptosis is depicted in Fig. 11. The INrf2 homodimer bound to Nrf2 on one monomer and PGAM5-Bcl-xL on the other monomer exists in the cytosol and mitochondria. Under physiological conditions, INrf2 homodimers promote a Cul3-Rbx1-mediated degradation of Nrf2 and PGAM5-Bcl-xL, thereby contributing to the maintenance of a normal level of Bcl-xL and apoptosis. Oxidative/electrophilic stress leads to the release of Nrf2 and PGAM5-Bcl-xL complex from INrf2 dimers in the cytosol. Nrf2 translocates to the nucleus, leading to activation of cytoprotective gene expression (2). PGAM5-Bcl-xL is directed to mitochondria, where PGAM5 is degraded to release Bcl-xL. In addition, oxidative/ electrophilic stress also leads to release of Nrf2 and Bcl-xL in the mitochondria, resulting in activation of unknown mitochondrial gene expression and an increase in Bcl-xL-Bax dimers. The Nrf2-mediated increased expression of nuclear and mitochondrial cytoprotective gene expression and induction of Bcl-xL-Bax dimers leads to reduced apoptosis and increased cell survival. Further studies are required to investigate Nrf2-regulated genes in mitochondria. Oxidative/electrophilic stress is also known to stabilize Bcl-2 in the cytosol that forms dimers with Bax and contributes to reduced apoptosis and increased cell survival (not shown in Fig. 10) (30). It is noteworthy that the Nrf2-INrf2-PGAM5-Bcl-xL complex in the cytosol might be in addition to the previously characterized Nrf2-INrf2 complex (2) and remains to be further studied.
The stabilization of Bcl-xL (current study) and Bcl-2 (30) from INrf2 and prevention of apoptosis are presumably an important mechanism to save cells from dying in acute stress due to exposure to antioxidants, xenobiotics, drugs, and radiation. Once the exposure effect subsides, the levels of Bcl-xL, Bcl-2, and cytoprotective proteins are brought back to normal, and a normal level of apoptotic cell death is restored. Recent studies have reported increased stabilization/accumulation of Nrf2 due to mutations in INrf2, resulting in loss of function in lung and breast tumors (12,13,33,20). Lung cancer cell line A549 contains INrf2G333C mutant protein that has lost its capacity to bind/degrade Nrf2, leading to accumulation of Nrf2 in the nucleus (13). It has been suggested that higher levels of Nrf2 in A549 cells might have contributed to the survival of these cells in lung cancer. Similarly, our studies demonstrated that stress mediated loss of INrf2-PGAM5 interaction stabilized cellular Bcl-xL that resulted in decreased cellular apoptosis. In conclusion, we demonstrate that both INrf2 and PGAM5 protein contribute to the regulation of Bcl-xL protein and apoptosis.