Antioxidant-induced INrf2 (Keap1) tyrosine 85 phosphorylation controls the nuclear export and degradation of the INrf2–Cul3–Rbx1 complex to allow normal Nrf2 activation and repression

Oxidative stress is the pathogenic outcome resulting from an imbalanced ratio between production of reactive oxygen species (ROS) and the cellular antioxidant capacity (Landriscina et al., 2009). Oxidative stress is induced by a vast range of factors, including xenobiotics, drugs, heavy metals and ionizing radiation (Kaspar et al., 2009). High concentrations of ROS are hazardous for living organisms because they damage all cellular components, including DNA, lipids and proteins (Ames, 1983; Brazilai and Yamamoto, 2004; Chakravarti and Chakravarti, 2007), leading to many pathological conditions, including cancer, neurodegenerative diseases, arthritis, atherosclerosis and inflammation (Grisham and McCord, 1986; Ames et al., 1995; Ward, 1994; Breen and Murphy, 1995; Rosen et al., 1995). It is apparent that uncontrolled levels of ROS can cause a wide range of disorders. Therefore, it is evident that living organisms have evolved mechanisms that are responsible for protection against oxidative stress.

The transcription factor nuclear factor erythroid 2-related factor 2 (Nrf2, also known as HEBP1) is required to combat increases in oxidative stress. Nrf2 is a member of the family of leucine zipper or ‘cap'n'collar’-containing nuclear factor proteins (Kaspar et al., 2009; Dhakshinamoorthy and Jaiswal, 2001; Itoh et al., 1999). Nrf2 forms a heterodimer with a small Maf protein and binds to the antioxidant response element (ARE) regulating expression and induction of many genes encoding antioxidant and cytoprotective proteins, including NAD(P)H:quinone oxidoreductases (NQO1 and NQO2), glutathione S-transferase Ya subunit and heme oxygenase-1 (Kang et al., 2004). In the absence of cellular stress, Nrf2 is tethered within the cytoplasm by an inhibitory partner, inhibitor of Nrf2 (INrf2), also known as ‘Kelch-like erythroid cell-derived protein with CNC homology (ECH)-associated protein 1’ (Keap1) (Itoh et al., 1999), which interacts with the actin cytoskeleton (Kang et al., 2004). This interaction is between a single Nrf2 protein and INrf2 dimer (Tong et al., 2006). INrf2 serves as a substrate linker protein for interactions with the cullin 3 (Cul3)-based E3 ubiquitin ligase to regulate the stability of Nrf2 (Cullinan et al., 2004). Covalent conjugation of proteins by ubiquitin usually involves three enzymatic activities for activating (E1), conjugating (E2) and ligating (E3) ubiquitin to a substrate (Furukawa and Xiong, 2005). In this case, Nrf2 serves as the substrate, whereas Cul3 serves as a scaffold protein that forms the E3 ligase complex with ring box protein 1 (Rbx1) that recruits a cognate E2 enzyme (Kobayashi and Yamamoto, 2006). INrf2, through its N-terminal BTB/POZ domain, binds to Cul3 (Geyer et al., 2003) and, through its C-terminal Kelch domain binds to the substrate Nrf2, leading to the ubiquitylation and proteasomal degradation of Nrf2 through the 26S proteasome (Stewart et al., 2003; Nguyen et al., 2003). Exposure to a number of stressors and inducing agents leads to dissociation of Nrf2 from INrf2, thereby rescuing Nrf2 from proteasomal degradation and permitting entry of Nrf2 into the nucleus.

Recent studies have shown that persistent accumulation of Nrf2 in the nucleus is harmful. Nrf2 regulates the expression of several multidrug resistance-associated protein (MRP) efflux transporters in responses to oxidative stress (Maher et al., 2007) that could lead to chemotherapeutic drug resistance. INrf2-null mice demonstrated persistent accumulation of Nrf2 in the nucleus that led to postnatal death from malnutrition resulting from hyperkeratosis in the esophagus and fore-stomach (Wakabayashi et al., 2003). 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 a reduction of INrf2 affinity for Nrf2 and elevated expression of cytoprotective genes (Padmanabhan et al., 2006; Singh et al., 2006). Taken together, uncontrolled activation of Nrf2 in cells increases the risk of adverse effects, including tumorigenesis. By contrast, stress-induced activation of the Nrf2 pathway in normal cells is tightly regulated and confers cytoprotection against oxidative or electrophilic stress and carcinogens. Therefore, it is evident that cells possess multiple mechanisms that regulate the cellular abundance of Nrf2.

Previous studies have shown that INrf2 exits out of the nucleus by means of possession of a nuclear export sequence (NES) or chromosome region maintenance-1 (Crm1)-dependent nuclear export mechanism in response to oxidative stress (Velichkova and Hasson, 2005; Sun et al., 2007). Recently, we have shown that prothymosin-α mediates the nuclear import of the INrf2–Cul3–Rbx1 complex to degrade nuclear Nrf2 in the late response (post-induction of Nrf2) to oxidative stress (Niture and Jaiswal, 2009). We have also shown that Nrf2 controls the cellular abundance of Cul3 and Rbx1, leading to increases in the ubiquitin ligases in the late response to oxidative stress (Kaspar and Jaiswal, 2010a). Therefore, in this report, we investigated the nuclear export of INrf2 in the early response (pre-induction of Nrf2) to oxidative stress. We also investigated whether the nuclear export of INrf2 also led to the nuclear export of Cul3 and Rbx1. We propose that INrf2 will export out of the nucleus prior to Nrf2 stabilization and localization in the nucleus, and the mechanism of export is attributable to tyrosine phosphorylation. In this study, we also examined whether Cul3 and Rbx1 export out of the nucleus by using INrf2 and studied the effects a mutation of INrf2 can have on downstream defensive genes.

To investigate the pre-induction response of INrf2 in response to antioxidative stress, the subcellular localization of INrf2 was followed by immunoblotting. Western blot analysis shows that endogenous and overexpressed INrf2 exported out of the nucleus within 0.5 hours of treatment (Fig. 1A,B, also see quantitative densitometry values below the blots). Interestingly, Cul3 and Rbx1 also exported out of the nucleus within 0.5 hours, and this followed the same kinetics as those of INrf2. To study the means by which INrf2 exports out of the nucleus, preliminary experiments were performed using inhibitors. Genistein, a tyrosine kinase inhibitor, given concurrently with t-BHQ prevented the transient decrease in the levels of nuclear INrf2 protein (compare Fig. 1C with Fig. 1A). These results suggest that the nuclear export might be dependent upon tyrosine phosphorylation. Leptomycin B (LMB), a specific inhibitor of proteins containing nuclear export signals (Fukuda et al., 1997), was also given simultaneously with t-BHQ (Fig. 1D). LMB blocked the antioxidant-induced nuclear export of endogenous INrf2. Again, Cul3 and Rbx1 followed the same actions as INrf2 and did not export out of the nucleus. These results demonstrate that INrf2 and Cul3–Rbx1 might be interacting with Crm1 and that the nuclear export of Cul3 and Rbx1 is blocked by LMB.

Analysis of the mouse INrf2 amino acid sequence identified four different putative tyrosine phosphorylation sites. Site-directed mutagenesis mutations were performed on the four aforementioned tyrosine residues. Subcellular localization followed by immunoblotting was performed to investigate whether any one of the tyrosine residues was implicated in the nuclear export of INrf2. INrf2 mutants Y141A, Y208A and Y255A all showed an antioxidant-mediated nuclear export at 0.5 hours when exposed to t-BHQ (Fig. 2B–D). In addition, Y141A and Y255A were unstable when the experiments were carried out, and so MG132, a proteasome inhibitor, was used to stabilize these two proteins. Interestingly, INrf2 mutant Y85A was the only mutant to show nuclear accumulation in the presence of t-BHQ (Fig. 2A). As INrf2Y85A showed nuclear accumulation, we probed for Cul3 and Rbx1 to see their localization in the same experiment. Surprisingly the nuclear export of Cul3 and Rbx1 were also stymied in the presence of INrf2Y85A and t-BHQ. These results convey that phosphorylation at tyrosine residue 85 was required for the nuclear export of INrf2 and Cul3–Rbx1.

Analysis of the mouse Cul3 and Rbx1 amino acid sequence revealed three putative tyrosine phosphorylation sites in Cul3 and one putative tyrosine phosphorylation site in Rbx1. Site-directed mutagenesis mutations were performed on the tyrosine residues mentioned previously. Subcellular localization followed by immunoblotting was performed to investigate whether any one of the tyrosine residues was implicated in the nuclear export of Cul3 and Rbx1. Surprisingly, Cul3 mutants Y74A, Y432A, Y764A and Rbx1 mutant Y106A all showed an antioxidant-mediated nuclear export at 0.5 hours when exposed to t-BHQ (Fig. 3A–F), suggesting that tyrosine phosphorylation of Cul3 and Rbx1 is not required for their nuclear export. However, in the cases of Cul3Y764A and Rbx1Y106A, these proteins seemed to behave differently compared with the other mutant proteins by accumulating slightly in the cytosol after nuclear export at 0.5 hours.

Immunoprecipitation followed by immunoblotting was used to investigate tyrosine phosphorylation of endogenous INrf2, INrf2–V5 and INrf2Y85A–V5. HepG2 cells were treated with t-BHQ, and immunoprecipitation was performed with antibodies against INrf2 using only the nuclear fraction of each sample. Western blots were probed with anti-phosphotyrosine antibodies and then re-probed with anti-INrf2 antibodies. Antibodies against INrf2 immunoprecipitated phosphotyrosine proteins at 0.5 hours, suggesting that more tyrosine phosphorylation was likely to be taking place (Fig. 4A, top panels). The reverse immunoprecipitation confirmed an increase in interaction between endogenous INrf2 and phosphotyrosine at 0.5 hours (Fig. 4A, lower panels).

In an analogous setting to that of the endogenous INrf2 experiment, HepG2 cells were transfected with vector encoding INrf2–V5 and treated with t-BHQ. As expected, immunoprecipitation with anti-FLAG antibodies produced a result similar to that of the endogenous INrf2 experiment. Anti-V5 immunoprecipitated phosphotyrosine at the 0.5 hour treatment (Fig. 4B, top panels). In the reverse experiment, immunoprecipitation with phosphotyrosine also immunoprecipitated an increase in V5-tagged INrf2 protein at the 0.5 hour treatment (Fig. 4B, bottom panel).

We then determined whether the INrf2Y85A mutant could be tyrosine phosphorylated in response to tBHQ. Anti-V5 antibodies did not immunoprecipitate phosphotyrosine protein in HepG2 cells transfected with INrf2Y85A at any time-point (Fig. 4C, top panels). In the reverse experiment, phosphotyrosine antibodies did not immunoprecipitate V5-tagged INrf2 protein at any time-points (Fig. 4C, bottom panel). Taken together, these results show that endogenous INrf2 and INrf2–V5 show tyrosine phosphorylation at 0.5 hours compared with the INrf2Y85A mutant, suggesting that the INrf2Y85 residue is likely to be phosphorylated.

Even though site-directed mutagenesis of tyrosine residues in Cul3 and Rbx1 resulted in nuclear export after antioxidant treatment, the fact that Cul3Y764A and Rbx1Y106A accumulated in the cytosol following nuclear export (Fig. 3D,F) needed to be investigated further. We were interested to see whether Cul3 or Rbx1 were actually phosphorylated after antioxidant treatment versus the aforementioned mutant proteins. Immunoprecipitation followed by immunoblotting was used to investigate tyrosine phosphorylation of endogenous Cul3–Rbx1, and overexpressed Cul3–Rbx1. HepG2 cells were treated with t-BHQ, and immunoprecipitation was performed with antibodies against Cul3 and Rbx1 using only the nuclear fraction of each sample. Western blots were probed with anti-phosphotyrosine antibodies and then re-probed with antibodies to Cul3 and Rbx1. Surprisingly, antibodies to Cul3 and Rbx1 immunoprecipitated phosphotyrosine proteins at 0.5 hours, suggesting that phosphorylation was taking place (Fig. 5A,B, top panels). The reverse immunoprecipitation confirmed an increase in interaction between endogenous Cul3–Rbx1 and phosphotyrosine at 0.5 hours (Fig. 5A,B lower panels).

In an analogous setting to the endogenous Cul3 and Rbx1 experiments, HepG2 cells were transfected with Cul3–V5 and Rbx1–V5 and treated with t-BHQ. As expected, immunoprecipitation with anti-V5 antibodies produced a result similar to that of the endogenous Cul3 and Rbx1 experiments. Anti-V5 immunoprecipitated phosphotyrosine at the 0.5 hour treatment (Fig 5C,D). In the reverse experiment, immunoprecipitation with phosphotyrosine also immunoprecipitated an increase in V5-tagged Cul3 and Rbx1 protein at the 0.5 hour treatment (Fig. 5C,D, lower left panels).

We then determined whether Cul3 and Rbx1 tyrosine mutants, which showed cytosolic accumulation after nuclear export (Fig. 3E,F), would be able to interact with any tyrosine phosphorylation. Antibodies to V5 did not immunoprecipitate phosphotyrosine protein in HepG2 cells transfected with vector encoding Cul3Y764A or Rbx1Y106A at any time-point (Fig. 5C,D, top right panels). In the reverse experiments, phosphotyrosine antibodies did not immunoprecipitate V5-tagged Cul3 or Rbx1 protein at any time-points (Fig. 5C,D, lower right panels). Taken together, these results show that endogenous Cul3–Rbx1 and overexpressed Cul3 and Rbx1 show tyrosine phosphorylation at 0.5 hours, unlike the mutants, suggesting that the Cul3Y764A and Rbx1Y106A residues are likely to be phosphorylated.

We performed immunocytochemistry studies to examine the colocalization and interaction of INrf2, Cul3 and Rbx1 proteins in HepG2 and Hepa-1 cells (Fig. 6A,B). The data demonstrated that INrf2, Cul3 and Rbx1 proteins were localized in both cytosolic as well as nuclear compartments of the HepG2 and Hepa-1 cells (Fig. 6A,B, upper three panels). The studies also suggested that INrf2 colocalized with Cul3 and Rbx1 proteins and that the INrf2–Cul3–Rbx1 complex exists in cytosolic and nuclear compartments (Fig. 6A,B, lower two panels). Next, to test whether INrf2 is required for Cul3 and Rbx1 nuclear export, we used siRNA to knockdown INrf2 and looked at the effect it would have on Cul3 and Rbx1 nuclear export. After transfecting INrf2 siRNA, antioxidant treatment failed to induce the nuclear export of Cul3 and Rbx1 (Fig. 6C, left and right panels). It has been previously reported that INrf2 possesses a functional NES2 (nuclear export signal 2) domain between amino acids 301 and 310 (LVQIFQELTL) that interacts with nuclear export protein Crm1, leading to the nuclear export of INrf2 in the late response to oxidative stress (Velichkova and Hasson, 2005; Sun et al., 2007). The NES2 within INrf2 is shown in Fig. 6D. To test our previous experiments where LMB blocked nuclear export of INrf2, Cul3 and Rbx1 in the early response to antioxidants, we sought to investigate whether mutation of leucine residues in the NES2 (L308A and L310A) in INrf2 would result in the stymie of Cul3 and Rbx1 nuclear export. As expected, when the nuclear export of INrf2 was blocked by mutation of the NES2, nuclear export of Cul3 and Rbx1 was also blocked (Fig. 6E). Next, we examined whether INrf2–V5 and INrf2Y85A–V5 could interact with Crm1, a nuclear exporter protein. Anti-V5 antibodies were able to immunoprecipitate FLAG-tagged Crm1 at 0.5 and 1 hours (Fig. 6F, upper panels). In the reverse experiment, anti-FLAG antibodies immunoprecipitated INrf2–V5 at 0.5 and 1 hour (Fig. 6F, upper panels). In a similar experimental setting, cells were co-transfected with INrf2Y85A–V5 and FLAG–Crm1. Anti-V5 antibodies were unable to immunoprecipitate FLAG–Crm1 (Fig. 6F, lower panels), and anti-FLAG antibodies were unable to immunoprecipitate INrf2Y85A–V5 (Fig. 6F, lower panels). Furthermore, to confirm that INrf2Y85A can in fact bind to Cul3–Rbx1 and Nrf2, immunoprecipitations were performed. With INrf2–V5 acting as a positive control (Fig. 7A,C), INrf2Y85A–V5 was able to interact with HA–Cul3 and Myc–Rbx1 in both the forward and reverse immunoprecipitations (Fig. 7B). INrf2Y85A also interacted with endogenous Nrf2 in both the forward and reverse immunoprecipitations (Fig. 7D). Together these results show that INrf2–V5, and not INrf2Y85A–V5, was able to interact with Crm1, suggesting that nuclear export of INrf2, Cul3 and Rbx1 is dependent upon a NES interaction with Crm1 in the early response to antioxidants.

To elucidate the fate of INrf2 after nuclear export has taken place, experiments were performed using the proteasome inhibitor MG132 followed by t-BHQ. After nuclear export of endogenous INrf2 at 0.5 hours, accumulation of INrf2 in the cytosol appeared (Fig. 8A). Cul3 and Rbx1 also displayed kinetics similar to those of INrf2. In a similar experiment, INrf2–V5 was transfected and HepG2 cells were pre-treated with proteasome inhibitor MG132 followed by t-BHQ. INrf2–V5 behaved analogously to endogenous INrf2, revealing accumulation in the cytosol at 0.5 hours (Fig. 8B). Again, Cul3 and Rbx1 also showed accumulation in the cytosol after nuclear export in response to antioxidants. Next, INrf2Y85A was transfected into HepG2 cells and pre-treated with MG132, followed by t-BHQ treatment. As expected, INrf2Y85A failed to transport out of the nucleus and clearly did not demonstrate any cytosolic accumulation (Fig. 8C). As expected, Cul3 and Rbx1 also did not accumulate in the cytosol after MG132 and antioxidant treatment. The effect of MG132 seemed to mediate cytosolic accumulation of INrf2, Cul3 and Rbx1 after nuclear export, suggesting that all three proteins are degraded after nuclear export. To confirm that INrf2, Cul3 and Rbx1 are all ubiquitylated after nuclear export, a ubiquitylation assay was performed. HepG2 cells were transfected with INrf2–V5, Cul3–V5 or Myc–Rbx1 and HA–ubiquitin plasmids. Cells were then immunoprecipitated with antibodies against V5 or Myc and immunoblotted with antibodies to HA. Cells treated with DMSO showed little ubiquitylated proteins in the cytosol, and undetectable levels in the nucleus (Fig. 8D–F). INrf2, Cul3 and Rbx1 ubiquitylation was highest in the cytosol after 30 minutes of t-BHQ treatment following nuclear export. Ubiquitylation decreased noticeably at 4 hours after antioxidant treatment. Taken together, these results demonstrate that INrf2–Cul3–Rbx1 is ubiquitylated and degraded following nuclear export.

We then investigated the effect of mutation of INrf2Y85 on Nrf2 downstream gene activity. The rationale was that, if INrf2 is unable to export out of the nucleus, it might be possible that it could degrade Nrf2 in the nucleus or transport Nrf2 out of the nucleus, interfering with cytoprotective gene expression. To examine the differences between the negative regulation of INrf2 and mutant INrf2Y85A on Nrf2, ubiquitylation assays were performed (Fig. 9A). HepG2 cells were transfected with INrf2–V5 or mutant INrf2Y85A–V5 along with FLAG–Nrf2 and Ub-HA, pretreated with MG132 and then treated with either DMSO or t-BHQ and analyzed for Nrf2 ubiquitylation and activation.

A comparison of ubiquitylated Nrf2 in the nuclear and cytosolic compartments of DMSO- and t-BHQ-treated INrf2- and INrf2Y85A-transfected cells revealed the following: INrf2 and mutant INrf2Y85A both induced ubiquitylation of Nrf2 in DMSO-treated cells that was noticeably decreased at 0.5 hour following treatment with t-BHQ (Fig. 9A top panels, compare lane 3 with 2, and lane 6 with 5), leading to stabilization and nuclear accumulation of Nrf2 (Fig. 9A, bottom panel). This was expected as t-BHQ antagonizes Nrf2–INrf2 interaction. Interestingly, much higher levels of ubiquitylated Nrf2 were observed in the cytosolic compartment of cells transfected with INrf2 and treated with t-BHQ for 4 hours, in comparison with the nuclear compartment (Fig. 9A, top panels, compare lane 5 between cytosol and nucleus). By contrast, the cells transfected with mutant INrf2Y85A and treated with t-BHQ for 4 hours showed noticeably higher ubiquitylated Nrf2 in the nucleus than in cytosolic compartment (Fig. 9A, top panels, compare lane 7 between cytosol and nucleus). In related experiments, the treatment of INrf2- and INrf2Y85A-transfected cells with t-BHQ for prolonged times led to alterations in the levels of Nrf2 and the Nrf2 downstream gene product NQO1 (Fig. 9B). Nrf2 levels declined in INrf2-transfected cells but increased in mutant INrf2Y85A-transfected cells (Fig. 9B). The NQO1 levels were also higher in mutant INrf2Y85A-transfected than INrf2-transfected cells (Fig. 9B). In similar experiments, an NQO1 gene ARE–luciferase construct demonstrated noticeably higher luciferase activity in mutant INrf2Y85A-transfected cells in comparison with INrf2-transfected cells (Fig. 9C).

INrf2, Cul3 and Rbx1 act as negative regulators of Nrf2 (Kaspar et al., 2009). The regulation of Nrf2, especially its abundance in the nucleus, is important for controlling expression of cytoprotective genes in response to oxidative stress (Kang et al., 2004). As persistent increases in cytoprotective gene expression threaten cell survival (Strachan et al., 2005), Nrf2 is exported out of the nucleus and degraded. The nuclear export and degradation of Nrf2 is activated after INrf2–Cul3–Rbx1 imports into the nucleus and degrades Nrf2, or INrf2 shuttles Nrf2 out of the nucleus, allowing degradation in the cytosol (Velichkova et al., 2005; Sun et al., 2007). INrf2-mediated degradation of nuclear Nrf2 and shuttling of Nrf2 out of the nucleus are delayed or post-induction responses to oxidative stress (Velichkova et al., 2005; Sun et al., 2007). In the current study, we investigated the early or pre-induction response of the INrf2–Cul3–Rbx1 complex to antioxidative stress.

We have demonstrated that INrf2, Cul3 and Rbx1 were exported out of the nucleus soon after exposure to the antioxidant t-BHQ. The antioxidant-induced nuclear export of INrf2–Cul3–Rbx1 appears to be an integral part of Nrf2-mediated activation of cytoprotective genes. Treatment with LMB, an inhibitor of Crm1-mediated nuclear export, stymied nuclear export of INrf2–Cul3–Rbx1, resulting in nuclear accumulation of the protein. The mechanism of INrf2–Cul3–Rbx1 nuclear export was dependent on phosphorylation of tyrosine 85 on INrf2 after stimulation of antioxidant stress. Once the INrf2 was phosphorylated at tyrosine 85 in response to antioxidant, the nuclear export signal in INrf2 interacted with Crm1, which allowed for nuclear export and subsequent degradation of the INrf2–Cul3–Rbx1 complex. Interestingly, mutation of the nuclear export signal in INrf2 not only occludes nuclear export of INrf2, but also Cul3 and Rbx1. siRNA-mediated knockdown of INrf2 also blocked nuclear export of Cul3 and Rbx1, suggesting that Cul3 and Rbx1, both of which lack nuclear export signals, need the nuclear export signal of INrf2 in order to be exported. In this case, INrf2 serves not only as an escort for Nrf2, which had previously been shown (Sun et al., 2007), but also an escort for Cul3 and Rbx1. The kinase responsible for phosphorylation of the complex has not been identified. We believe that the tyrosine kinase has to be stress responsive and is rapidly activated in response to antioxidants.

Interestingly, upon nuclear export of INrf2, Cul3 and Rbx1, subsequent accumulation in the cytosol was absent. When cells were treated with proteasome inhibitors, INrf2–Cul3–Rbx1 accumulation could be seen in the cytosol, and INrf2 ubiquitylation could also be seen in the cytosol, supporting the conclusion that antioxidants induce proteasome-dependent degradation following nuclear export. These data are in agreement with another study that has shown that the ubiquitin–proteasome machinery is involved in the degradation of Cul3, resulting in accumulation of Cul3 (Kim et al., 2010). Tyrosine mutation on residue Y764 in Cul3 and Y106A in Rbx1 also induced cytosolic accumulation following antioxidant-induced nuclear export. The mutant proteins Cul3Y764A and Rbx1Y106A, unlike the wild-type proteins, could not immunoprecipitate any tyrosine phosphorylation. Hence, it appears that tyrosine phosphorylation might be playing a key role in regulating the degradation of the proteins. The transcriptional regulation of Cul3–Rbx1 has been shown to be mediated through Nrf2 and the antioxidant response element ARE (Kaspar and Jaiswal, 2010). Induction of Nrf2 leads to increases in Cul3 and Rbx1 protein synthesis (Kaspar and Jaiswal, 2010). Therefore, it is possible that tyrosine phosphorylation could be marking these proteins for degradation to prevent an excess of ubiquitin ligases within the cell.

Previous findings have shown that INrf2 is degraded through Cul3-dependent ubiquitin ligases, but the degradation of INrf2 might not be accomplished by the 26S proteasome (Zhang et al., 2005). Our results suggest that INrf2 might be degraded through the 26S proteasome because INrf2 was ubiquitylated in the early response to antioxidant treatment and accumulation of the protein was seen in the cytosol following treatment with proteasome inhibitors. A recent study has shown that the degradation of INrf2 might be controlled through sequestosome 1 (Copple et al., 2010). Sequestosome-1, also referred to as ubiquitin-binding protein p62, is a scaffold protein that has been implicated in both proteasomal and liposomal degradation cascades (Korolchuk et al., 2009). Therefore it appears that multiple mechanisms exist that can regulate the degradation of INrf2.

Investigations also determined the physiological significance of nuclear export of INrf2–Cul3–Rbx1 complex during the pre-induction response to antioxidant. The results suggested that pre-induction nuclear export of the INrf2–Cul3–Rbx1 complex in response to antioxidant assures normal activation and repression of Nrf2 and downstream cytoprotective gene expression. Interestingly, nuclear export mutant INrf2Y85A accumulated in the nucleus but had no effect on Nrf2 activation, presumably owing to the antagonizing effect of the antioxidant on Nrf2 and INrf2. Intriguingly, the nuclear accumulation of export-deficient INrf2Y85A interfered with the post-induction response to rapidly bring down the activated Nrf2 to its basal level. Antioxidant-induced nuclear accumulation of INrf2Y85A led to increased nuclear levels of Nrf2 and higher activation of cytoprotective gene expression. The mechanism of nuclear accumulation of Nrf2 in the presence of mutant INrf2Y85 is not clear. However, our data suggest that nuclear accumulation of INrf2Y85A results in decreased nuclear export and degradation of Nrf2, leading to nuclear accumulation of Nrf2 and higher activation of Nrf2 downstream cytoprotective gene expression. In addition to this, nuclear export and degradation of INrf2–Cul3–Rbx1 might also regulate nuclear levels of this complex during induction and post-induction responses. After nuclear export, the complex is degraded, and Nrf2 controls the induction of the genes encoding INrf2, Cul3 and Rbx1, and so it seems plausible that the export is also to control the protein levels. Excessive INrf2 or Cul3–Rbx1 ubiquitin ligases might be harmful and cause destabilization of the Nrf2 signaling pathway.

In summary, we have investigated the early pre-induction response of INrf2, Cul3 and Rbx1 after treatment with antioxidants. We demonstrate that INrf2 exports out of the nucleus with Cul3 and Rbx1 as a complex. The nuclear export of INrf2 is controlled by tyrosine phosphorylation of Y85 and a nuclear export signal interaction with Crm1, leading to INrf2 degradation after nuclear export. The nuclear export of Cul3 and Rbx1 relies on the NES in INrf2 for their own nuclear export and subsequent degradation. Once in the nucleus, Nrf2 activates cytoprotective genes. After completing activation of defensive genes, INrf2–Cul3–Rbx1 is imported back into the nucleus to degrade Nrf2 (Sun et al., 2007) or transport Nrf2 out of the nucleus (Velichkova et al., 2005), causing Nrf2 export and degradation. The possible function of the pre-induction nuclear export of the INrf2–Cul3–Rbx1 complex is to participate in normal activation or repression of Nrf2 and to regulate the protein levels within the cells during periods of stress.

Human hepatoblastoma (HepG2) and mouse hepatoma (Hepa-1) cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA). HepG2 cells were grown in minimum essential α-medium, and Hepa-1 was grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin (40 units ml⁻¹) and streptomycin (40 μg ml⁻¹). The cells were grown in a monolayer in an incubator at 37°C in 95% air and 5% CO₂.

The construction of pcMV-HA-Cul3, pcMV-Rbx1-Myc, HA-Ub, pcMX-FLAG-Crm1, pcDNA-INrf2-V5, pcDNA-INrf141A-V5 and pCDNA-INrf2Y208A-V5 has been described previously (Niture and Jaiswal, 2009; Jain and Jaiswal, 2006; Jain and Jaiswal, 2007; Jain and Jaiswal, 2008). Cul3 cDNA was amplified using primers: forward 5′-AACGCCATGTCGAATCTGAGCAAAGG and reverse 5′-TGCTACATATGTGTATACTTTGCG. The PCR-amplified DNA was subcloned into pcDNA 3.1/V5-HisTOPO vector by TA cloning (Invitrogen, Carlsbad, CA). The resultant plasmid was designated as Cul3–V5. Three tyrosine residues (Y74, Y432 and Y764) present in Cul3–V5 were mutated to alanine using a site-directed mutagenesis kit (Invitrogen). Mutant Y74A was generated by PCR using the mutant forward primer 5′-AAACATGGAGAAAAGCTCGCCACTGGACTAAGA and reverse primer 5′-GAGCTTTTCTCCATGTTTATGCAAAACCAT. Mutant Y432A was generated by PCR using the mutant forward primer 5′-GATGTATTTGAACGTTATGCTAAACAACACCTG and reverse primer 5′-ATAACGTTCAAATACATCTTTTTCTTGCAT. Mutant Y764A was generated by PCR using the mutant forward primer 5′-CCTGAGGATCGCAAAGTAGCCACATATGTAGCA and reverse primer 5′-TACTTTGCGATCCTCAGGTGTTCGTGCCAA. Rbx1 cDNA was amplified using primers: forward 5′-AACGCCATGGCGGCGGCGATGGATGT and reverse 5′-ATGCCCATACTTCTGGAACTC. The PCR-amplified DNA was subcloned into pcDNA 3.1/V5-HisTOPO vector by TA cloning (Invitrogen). The resultant plasmid was designated as Rbx1–V5. One tyrosine residue (Y106) present in Rbx1–V5 was mutated to alanine using a site-directed mutagenesis kit (Invitrogen). Mutant Y106A was generated by PCR using the mutant forward primer 5′-GAGTGGGAGTTCCAGAAGGCTGGGCATAAGGGC and reverse primer 5′-CTTCTGGAACTCCCACTCTCTGTTGTCCAA. Two tyrosine residues (Y85 and Y255) present in INrf2–V5 were mutated to alanine using a site-directed mutagenesis kit (Invitrogen). Mutant INrf2Y85A was generated by PCR using the mutant forward primer 5′-ACGTGACCCTGCAGGTCAAAGCAGAGGACATCC and reverse primer 5′-TTTGACCTGCAGGGTCACGTCACAGAGTTG. Mutant Y255A was generated by PCR using the mutant forward primer 5′-CGTGCATCGACTGGGTCAAAGCAGACTGCCCGC and reverse primer, 5′-TTTGACCCAGTCGATGCACGCGTGGAACAC. Two leucine residues (L308 and L310) present in functional NES2 domain (³⁰¹LVQIFQELTL³¹⁰) of INrf2–V5 were mutated to alanine using a site-directed mutagenesis kit (Invitrogen). For this we used: forward primer 5′-CAGATATTCCAGGAGGCCACGGCGCACAAGCCCA and Reverse Primer 5′-CTCCTGGAATATCTGCACCAGGTAGTC. The mutant plasmid was designated as INrf2ΔNES2–V5. All plasmids were confirmed by DNA sequencing.

In vitro transcription and translation of the plasmids encoding tyrosine mutations were performed using the TNT-coupled rabbit reticulocyte lysate system (Promega, Madison, WI), as described previously (Niture and Jaiswal, 2009). All of the in vitro transcribed and translated proteins gave bands of the expected molecular masses.

Subcellular fractionation and western blotting have been described previously (Kaspar and Jaiswal, 2010b). Antibodies used in this study were as follows: anti-INrf2 (1:1000) purchased from Santa Cruz Biotechnology (Santa Cruz, CA), anti-Cul3 (1:1000), anti-Rbx1 (1:1000) purchased from Cell Signaling (Danvers, MA), anti-V5 HRP (1:5000), anti-FLAG HRP purchased from Invitrogen, anti-phosphotyrosine (1:1000) and anti-actin (1:5000) purchased from Sigma-Aldrich (St Louis, MO). For immunoprecipitations, anti-Roc1 from Biosource (Carlsbad, CA) was used. To confirm the purity of nuclear–cytoplasmic fractionation, the membranes were re-probed with cytoplasm-specific anti-lactate dehydrogenase (LDH) (Chemicon, Billerica, MA) and nuclear specific, anti-lamin B antibodies (Santa Cruz). In related experiments, the cells were treated with 100 μM tert-butylhydroquinone (t-BHQ) (Sigma), 20 ng ml⁻¹ leptomycin B (LMB), 2 μM MG132, 50 μm genistein (Calbiochem, La Jolla, CA) or DMSO as a vehicle for different time intervals.

Immunoprecipitations were performed based on methods described previously (Kaspar and Jaiswal, 2010b).

Hepa-1 cells were plated in 100 mm plates at a density of 1×10⁵ cells per plate 24 hours prior to transfection. In the related experiments, the cells were transfected with 1 μg of the indicated plasmids using Effectene transfection reagent (Qiagen, Valencia, CA) according to the manufacturer's protocol.

Control and INrf2 siRNA purchased from Dharmacon were used to inhibit INrf2 proteins, by a procedure that has been described previously (Wang and Jaiswal, 2006). Hepa-1 cells were transfected at a 50 nM concentration of control or INrf2 siRNA using Lipofectamine RNAiMAX reagent (Invitrogen) according to the manufacturer's instructions. 32 hours after transfection, cells were harvested and the localization of INrf2, Cul3 and Rbx1 was analyzed by western blotting by probing the membranes with antibodies against INrf2, Cul3 or RBX1.

Hep-G2 cells were transfected with vectors encoding INrf2–V5, Cul3–V5 or Myc–Rbx1 along with Ub-HA for 24 hours and treated with DMSO or t-BHQ (100 μM) for 0.5 and 4 hours. Nuclear and cytosolic fractions were prepared by the Active Motif kit. The lysates were immunoprecipitated with anti-V5 or anti-Myc antibodies followed by western analysis with anti-HA antibody to determine ubiquitylation of INrf2, Cul3 and Rbx1. The effect of INrf2–V5 or INrf2Y85A mutant and antioxidant (t-BHQ) treatments on FLAG–Nrf2 ubiquitylation was also analyzed in HepG2 cells. HepG2 cells were co-transfected with vectors encoding Flag-Nrf2, wild-type INrf2–V5 or mutant INrf2Y85A–V5 and HA–ubiquitin constructs for 24 hours. Cells were pretreated with MG132 (5 μM) for 2 hours and post-treated with DMSO or t-BHQ (100 μM) for the indicated time-periods. Cells were harvested and nuclear and cytosolic extracts prepared by the Active Motif Kit. 500 μg of nuclear and cytosolic extracts were immunoprecipitated with anti-FLAG antibody and western blotted for HA–HRP antibody.

HepG2 or 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 1:500 dilution of anti-goat INrf2, anti-rabbit Cul3 and anti-rabbit Rbx1 antibody in 2% BSA separately, or in combination as indicated, at 4°C for 12 hours. Then cells were washed twice with PBS and incubated with an Alexa-Fluor-594-conjugated anti-goat antibody or FITC-conjugated anti-rabbit secondary antibody (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.

HepG2 cells were grown in monolayer cultures in 12-well plates. After 12 hours, cells were co-transfected with INrf2–V5 or mutant INrf2Y85A–V5 (100 ng per well) along with 100 ng of NQO1 promoter ARE-Luc reporter construct and 10 ng of firefly Renilla luciferase plasmid pRL-TK. Renilla luciferase was used as the internal control in each transfection. After 24 hours of transfection, cells were treated with DMSO or t-BHQ (100 μM for 4 to 16 hours), as indicated in the figures. The cells were washed with saline and lysed in 1X passive lysis buffer from the Dual-Luciferase reporter assay system kit (Promega, Madison, WI). NQO1 promoter luciferase activity was measured and plotted as described by the manufacturer's instructions. The data shown are the mean ± s.d. of three independent transfection experiments.

The data were analyzed using a two-tailed Student's t test. Data are expressed the means ± s.d. of three independent experiments.

We thank our colleagues for helpful discussion.