Nuclear factor erythroid 2-related factor 2 enhances carcinogenesis by suppressing apoptosis and promoting autophagy in nickel-transformed cells

Nickel-containing compounds are widely used in industry. Nickel is a known human carcinogen that primarily affects the lungs. Proposed mechanisms of nickel-induced carcinogenesis include disruption of cellular iron homeostasis, generation of reactive oxygen species (ROS), and induction of hypoxia signaling. However, the precise molecular mechanisms of nickel-induced malignant transformation and tumor development remain unclear. This study shows that the transcription factor Nrf2 is highly expressed in lung tumor tissue and in nickel-transformed human lung bronchial epithelial BEAS-2B cells (NiT cells). Additionally, constitutively high levels of Nrf2 play a critical role in apoptosis resistance in NiT cells. Basal ROS levels were extremely low in NiT cells and were correlated with elevated expression levels of both antioxidant enzymes (e.g. catalase and superoxide dismutases) and antiapoptotic proteins (e.g. Bcl-2 and Bcl-xL). These processes are tightly controlled by Nrf2. Autophagy inhibition, induced pharmacologically or genetically, enhanced Ni2+-induced apoptosis, indicating that the induction of autophagy is the cause of apoptosis resistance in NiT cells. Using similar approaches, we show that in NiT cells the inhibition of apoptosis decreases autophagy. We have shown that Stat3, which is up-regulated by Nrf2, controls autophagy induction in NiT cells. Colony formation and tumor growth were significantly attenuated by knockdown of Nrf2 or Bcl-2. Taken together, this study demonstrates that in NiT cells constitutively high Nrf2 expression inhibits apoptosis by up-regulating antioxidant enzymes and antiapoptotic proteins to increase autophagy via Stat3 signaling. These findings indicate that the Nrf2-mediated suppression of apoptosis and promotion of autophagy contribute to nickel-induced transformation and tumorigenesis.

Nickel is a ubiquitous environmental transition metal that is widely used in industrial and medical processes, including electroplating and the manufacture of steel, batteries, and electronic devices (1). Epidemiological studies indicate that chronic occupational exposure to nickel compounds increases the incidence rates of human lung and nasal cancers (2,3). As a result, nickel compounds were classified as human carcinogens by the International Agency for Research on Cancer in 1990 (4). Disruption of cellular iron homeostasis by interfering with iron-dependent enzymes (5) and generation of reactive oxygen species (ROS) 4 contribute to nickel-induced carcinogenesis (6). Induction of the hypoxia-signaling pathway also represents a key mechanism for nickel-initiated carcinogenesis (7,8). Despite many studies, the precise molecular mechanisms by which nickel induces carcinogenesis remain to be elucidated.
Apoptosis, a form of programmed cell death, serves as a natural barrier to carcinogenesis (9). Morphological and biochemical hallmarks of apoptosis include cell shrinkage, membrane blebbing, nuclear condensation, and nuclear DNA fragmentation (10). The apoptotic process is controlled by pro-apoptotic and antiapoptotic members of the Bcl-2 family (9). Nrf2 up-regulates the antiapoptotic proteins Bcl-2 and Bcl-xL by inducing transcription of their respective genes (11,12). However, cancer cells have a variety of strategies to escape apoptosis, such as loss of function of the tumor suppressor protein p53 and an increase in expression of antiapoptotic proteins (13,14).
Autophagy is a cellular degradation pathway that is essential for cell survival under some conditions but can induce cell death under others (15,16). In this way, autophagy can have beneficial or negative effects on human diseases such as cancer, liver disease, and neurodegeneration (17). By recycling nutrients, autophagy promotes tumor growth, metabolism, and survival (18). Autophagy is evidenced by the early appearance of large autophagic vacuoles in the cytoplasm (19 -23). Beclin 1 and microtubule-associated protein 1 light chain 3 (LC3) are two critical components in autophagy. Beclin 1 is the mammalian orthologue of yeast Atg6/Vps 30 and is involved in the regulation of autophagy (24 -28). The other key protein, LC3, is the mammalian homologue of yeast Atg8 and localizes to autophagosomal membranes after post-translational modifications. LC3 exists in two molecular forms as follows: LC3-I (18 kDa) is the cytosolic form, and LC3-II (16 kDa) is incorporated into autophagosome membranes (29,30). The level of LC3-II directly correlates with the number of autophagosomes (30).
Nrf2 regulates antioxidant proteins to neutralize ROS, thereby regulating the cellular redox balance (31,32). Under normal conditions, Nrf2 binds to Keap1 (Kelch-like ECH-associated protein 1) in the cytosol, which promotes the proteasomal degradation of Nrf2. Alternatively, under oxidative stress conditions, Nrf2 is freed from Keap1-mediated repression and is translocated to the nucleus, where it binds to antioxidantresponse elements (AREs) in the promoter regions of genes encoding antioxidant proteins and detoxification enzymes to initiate their transcriptions (33). Constitutively high overexpression of Nrf2 protects cancer cells against oxidative stress and chemotherapeutic agents (34). Constitutive activation of Nrf2 is evident in several human cancer cell lines as well as in tumors (35)(36)(37)(38). Furthermore, high Nrf2 levels in cancer cells correlate with chemoresistance (34, 39 -42). In this study, we explore the role of Nrf2 in nickel-induced carcinogenesis. Our results demonstrate that nickel-transformed (NiT) cells have properties of low ROS levels and apoptosis resistance by upregulating antioxidant enzymes and antiapoptotic proteins. The sensitivity of NiT cells to autophagy induction enhances their resistance to apoptosis. We have shown that the high expression of Stat3 in NiT cells is responsible for the increased autophagy. We have also shown that antioxidant enzymes, antiapoptotic proteins, and Stat3 are tightly regulated by Nrf2 in the transformed cells. High expression of Nrf2 was confirmed in human lung tumor tissues, and its role in malignant transformation was investigated by conducting in vitro and in vivo experiments. Taken together, our results reveal a tumor cell survival mechanism involving the down-regulation of apoptosis and up-regulation of autophagy. Furthermore, they show that Nrf2 is a key regulator of intracellular ROS levels, apoptosis resistance, autophagy sensitivity, and therefore of cell survival and carcinogenesis in nickel-transformed cells.

NiT cells are resistant to cell death, including apoptosis
To generate the nickel-transformed cell line, NiT, we continuously exposed BEAS-2B cells to Ni 2ϩ (50 M) for 4 months. A soft agar assay revealed that this exposure malignantly transformed the cells (Fig. 1A). The NiT cells had a higher proliferative potential than the parental non-transformed BEAS-2B cells, during both short culture periods at high cell density (Fig.   1B) and long culture periods at low cell density (Fig. 1C). Ni 2ϩ induced less cell death in NiT cells than in BEAS-2B cells (Fig. 1, D and E). An apoptosis assay showed that ϳ60% of the BEAS-2B cells were apoptotic following Ni 2ϩ exposure. In contrast, ϳ40% of the NiT cells were apoptotic following the same Ni 2ϩ exposure (Fig. 1, F and G). Furthermore, the cleavage of PARP and caspases 3 and 7, as well as the decrease in the levels of pro-caspase 8, were much more pronounced in the parental BEAS-2B cells compared with NiT cells. These results suggested that NiT cells resist cell death, including apoptosis.

NiT cells are sensitive to autophagy induction
Ni 2ϩ treatment dramatically increased the conversion of LC3-I to LC3-II in NiT cells in a dose-and time-dependent manner, whereas this conversion was not as extensive in the parental BEAS-2B cells (Fig. 2, A and B). Treatment with bafilomycin A1, an inhibitor of autophagosome and lysosome fusion, led to increased LC3-II levels in both BEAS-2B and NiT cells. LC3-II levels were further increased when the cells were treated with a combination of Ni 2ϩ and bafilomycin A1, suggesting that Ni 2ϩ increased autophagic flux, rather than the obstruction to the fusion of autophagosomes with lysosomes ( Fig. 2C). In contrast, when the cells were treated with Ni 2ϩ in the presence of wortmannin, an inhibitor of autophagosome initiation, the Ni 2ϩ -induced up-regulation of LC3-II was attenuated ( Fig. 2C). Following Ni 2ϩ treatment, GFP-LC3 puncta formation increased dramatically in NiT cells but not in BEAS-2B cells (Fig. 2, D and E). To extend the study of autophagic flux in these cells, we used a tandem fluorescenttagged LC3 system. When the transfected cells were exposed to Ni 2ϩ , both yellow (mCherry ϩ /GFP ϩ ) (autophagosome) and red (mCherry ϩ /GFP Ϫ ) (autolysosome) puncta were increased in NiT cells, whereas only yellow low-intensity puncta were increased in normal cells (Fig. 2, F and G). These results indicate that, following Ni 2ϩ exposure, autophagosomes fuse with lysosomes to generate autolysosomes in NiT cells but not in parental BEAS-2B cells.

Autophagy plays opposite roles in normal and NiT cells
The combination treatment of nickel with the autophagy inhibitors sortmannin or 3-methyladenine (3-MA) produced a greater reduction in cell viability and enhanced apoptosis when compared with the nickel-only treatment in NiT cells, whereas cell viability was enhanced and apoptosis was reduced in parental BEAS-2B cells (Fig. 3, A and B). Note that single agent treatments with either wortmannin or 3-methyladenine did not alter cell viability or apoptosis (Fig. 3, A and B, and supplemental Fig. 1). These findings indicate that Ni 2ϩ -induced autophagy in NiT cells is involved in cell survival, whereas autophagy promotes cell death in the parental BEAS-2B cells., Ni 2ϩ -induced cell death was significantly enhanced in autophagy-defective beclin 1-deficient NiT cells when compared with NiT cells transfected with the control shRNA (Fig. 3C). Knockdown of beclin 1 by beclin 1 shRNA transfection did not alter cell viability or apoptosis in either the BEAS-2B or NiT cells (supplemental Fig. 2). Interestingly, inhibition of autophagy in NiT cells by both genetic (beclin 1 inhibition) and pharmacological (bafilo-mycin A1) approaches increased Ni 2ϩ -induced apoptosis over that observed in NiT cells transfected with the control shRNA or in untreated NiT cells, respectively (Fig. 3, D and E). However, these phenomena were not observed in the normal BEAS-2B cells (Fig. 3, D and E). Inhibiting the apoptosis signaling cascade, both pharmacologically (using the caspase inhibitor, Z-VAD; Fig. 3G) or genetically (by overexpressing Bcl-2; Fig. 3H), led to an enhancement of autophagy in Ni 2ϩ -exposed NiT cells. The inhibition of apoptosis by Z-VAD was confirmed by annexin V/PI staining (Fig. 3F), and the effects of Bcl-2 over-expression on cell viability were verified by MTT assay (supplemental Fig. 3). These results suggest that in Ni 2ϩ -exposed NiT cells autophagy inhibition enhances apoptosis and that inhibition of apoptosis enhances autophagy.

High expression of Nrf2 plays a critical role in the survival of NiT cells
Nrf2 regulates intracellular ROS levels in response to oxidative stimuli and toxic substances (43). We investigated whether Nrf2 is involved in apoptosis resistance in NiT cells. NiT cells Non-transformed BEAS-2B cells were exposed to Ni 2ϩ for 4 months. Transformed cells (NiT) were selected, and their transformation was confirmed by soft agar assay (A). BEAS-2B or NiT cells (0.1 ϫ 10 4 ) were seeded into a 96-well plate, and their proliferation rates were analyzed at the indicated times using MTT (B). BEAS-2B and NiT cells (0.2 ϫ 10 6 ) were seeded into 10-cm culture dishes and incubated for 7 days. Each day, the total number of cells in each dish was obtained using a cell counter (C). BEAS-2B and NiT cells were exposed to Ni 2ϩ (0 -2 mM) for 24 h, and cell morphology was visualized microscopically (D). Following treatment of BEAS-2B and NiT cells as described in C, cell viability was assessed using MTT (E), and apoptosis was assessed using annexin V/PI staining followed by flow cytometry (F). Graphic representation of the early apoptotic cells (annexin V ϩ /PI Ϫ ) is presented (G). BEAS-2B and NiT cells were exposed to increasing concentrations (0 -2 mM) of Ni 2ϩ for 24 h, and protein levels of PARP subunits, caspase 8, and c-caspase 3/7 were assessed by Western blot analysis using the indicated antibodies; GAPDH was used as a loading control (H). Photomicrographs presented are representative images for each experimental design. Data presented graphically are the mean Ϯ S.E. of triplicate samples or of three independent experiments, with significant differences indicated as follows: *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001, identified by ANOVA and Scheffe's test.

Nrf2 enhances carcinogenesis in nickel-transformed cells
exhibit a constitutively higher level of Nrf2 than that in nontransformed cells (Fig. 4A). When BEAS-2B and NiT cells were exposed to Ni 2ϩ (0 -2 mM), Nrf2 expression increased in both cell lines (Fig. 4B). The expression levels of Nrf2 were much higher in nickel-exposed NiT cells than those in BEAS-2B cells (Fig. 4B). In a time course experiment, Ni 2ϩ exposure began to increase Nrf2 expression in BEAS-2B cells within 1 h. The increase was sustained for 24 h (Fig. 4C, upper panel). In NiT cells, the basal Nrf2 expression levels were much higher, and Ni 2ϩ exposure led to a sustained increase in Nrf2 over the 24-h period (Fig. 4C, lower panel). More importantly, the Nrf2 target protein NQO1 was dramatically up-regulated up to the 24-h time point, and a second Nrf2 target, HO-1, was up-regulated after 1 h of Ni 2ϩ exposure in NiT cells (Fig. 4C, lower panel). To extend studies on the role of Nrf2 in apoptosis resistance, we used siRNA to silence Nrf2 expression. This approach attenuated basal and Ni 2ϩ -induced expression of Nrf2 in both cell lines (Fig. 4D). Inhibition of Nrf2 expression also increased Ni 2ϩ -induced apoptosis in both cell lines (Fig. 4E). Additionally, silencing Nrf2 in NiT cells attenuated the level of LC3-II as well as the number of cells containing GFP-LC3 puncta (Fig. 4, F, upper panel, and G) but increased the cleavage of PARP and caspases 3 and 7 (Fig. 4F, lower panel). These results suggest that Nrf2 plays an important role in the apoptosis resistance and autophagy sensitivity of NiT cells.

High Bcl-2 and Bcl-xL expression levels contribute to the resistance of NiT cells to cell death
Members of the Bcl-2 family of proteins are well-known regulators of apoptosis. To investigate whether two antiapoptotic Bcl-2 proteins, Bcl-2 and Bcl-xL, are involved in the resistance of NiT cells to cell death, we analyzed the Bcl-2 and Bcl-xL expression levels in BEAS-2B and NiT cells. We found that NiT cells have higher basal levels of Bcl-2 and Bcl-xL than the nontransformed parental cells (Fig. 5A). In dose and time course experiments, treatment with Ni 2ϩ reduced Bcl-2 and Bcl-xL expression to much lower levels in BEAS-2B cells than in NiT cells (Fig. 5, B and C). To determine the effect of disrupting Bcl-2 and Bcl-xL with pro-apoptotic proteins in the two cell lines, we treated them with the Bcl-2 family inhibitor ABT-263. As expected, this treatment effectively enhanced PARP cleavage and Bim expression in both cell lines, following Ni 2ϩ exposure (Fig. 5D, upper panel). Treatment with ABT-263 also led to a greater reduction in cell viability following Ni 2 exposure in both cell lines (Fig. 5D, lower panel). These cell viability results are consistent with those obtained using the genetic knockdown approaches of Bcl-2 and Bcl-xL (Fig. 5E). These results showed important functional roles for Bcl-2 and Bcl-xL in cell resistance to death. It has been reported that Nrf2 upregulates the expressions of Bcl-2 and Bcl-xL (44,45). To investigate whether Nrf2 up-regulates Bcl-2 and Bcl-xL expression in NiT cells, we transfected the cells with Nrf2- Figure 2. NiT cells are highly sensitive to autophagy induction following Ni 2؉ exposure. Cells were exposed to various concentrations of Ni 2ϩ (0 -2 mM) for 24 h, and the levels of LC3-I and LC3-II were assessed by Western blot analysis (A). BEAS-2B and NiT cells were exposed to Ni 2ϩ (2 mM) for various time periods (0 -24 h), and the levels of LC3-I and LC3-II were assessed by Western blot analysis (B). To analyze autophagic flux, the cells were pre-incubated with bafilomycin A1 (100 nM) or wortmannin (100 nM) for 1 h prior to treatment with Ni 2ϩ (2 mM). After a 24-h incubation, the levels of LC3-I and LC3-II were assessed by Western blot analysis (C). BEAS-2B and NiT cells were transfected with the GFP-LC3 plasmid, treated with Ni 2ϩ (2 mM), and visualized by fluorescence microscopy (D), and the number of cells containing GFP-LC3 puncta was counted (E). BEAS-2B and NiT cells were transfected with the mCherry-EGFP-LC3 plasmid and treated with Ni 2ϩ (2 mM). The yellow puncta (mCherry ϩ /GFP ϩ ; autophagosome) and red puncta (mCherry ϩ /GFP Ϫ ; autolysosome) were visualized using a fluorescence microscope (F), and the number of puncta/cell was quantified (G). Photomicrographs are representative images of each experiment. Data presented graphically are the mean Ϯ S.E. of triplicate samples or of three independent experiments, with significant differences from the vehicle control indicated as *, p Ͻ 0.05, and ***, p Ͻ 0.001, as identified by ANOVA and Scheffe's test. specific siRNA. As anticipated, the siRNA-mediated knockdown of Nrf2 reduced expression levels of Bcl-2 and Bcl-xL (Fig. 5F). Recently, our group found one consensus ARE and six putative AREs in the 3.6-kb Bcl-xL promoter and two putative AREs in the 8-kb Bcl-2 promoter (12). We used ChIP analysis to investigate whether Nrf2 up-regulates the transcription of Bcl-2 and/or Bcl-xL in BEAS-2B and NiT cells by binding to these sequences. Our analysis revealed that Nrf2 binding to the ARE-containing regions of the Bcl-2 (R1, Ϫ278 to ϳϪ2769) and Bcl-xL (F1, Ϫ2992 to Ϫ2984)  (5 mM). Thereafter, cell viability (A) and apoptosis (B) were assessed by MTT assay and annexin V/PI-staining, respectively. NiT cells stably transfected with BECN1 shRNA were treated with Ni 2ϩ (0 -2 mM) for 24 h, and cell viability (C) and the levels of apoptosis-associated proteins (D) were evaluated by MTT assay and Western blot analysis, respectively. BEAS-2B and NiT cells were preincubated with bafilomycin A1 (100 nM) 1 h prior to treatment with Ni 2ϩ (2 mM). After a 24-h incubation, the level of PARP was assessed by Western blot analysis (E). BEAS-2B and NiT cells were incubated with Ni 2ϩ (2 mM) for 24 h in the presence or absence of Z-VAD (20 M) and assessed for apoptosis (F) or protein levels of PARP subunits, caspase 3/7 and LC3 I/II, by Western blot analysis (G). Bcl-2-overexpressing BEAS-2B and NiT cells were incubated with Ni 2ϩ (2 mM) for 24 h, and the expression levels of Bcl-2 and LC3 were determined by Western blot analysis (H). Data presented graphically are the mean Ϯ S.E. of triplicate samples or of three independent experiments, with significant differences from controls indicated as *, p Ͻ 0.05; **, p Ͻ 0.01, and ***, p Ͻ 0.001, identified by ANOVA and Scheffe's test.

Nrf2 enhances carcinogenesis in nickel-transformed cells
promoters was higher in NiT cells than in BEAS-2B cells (Fig. 5G). The ChIP analysis coupled with quantitative realtime PCR also revealed that the extent of Nrf2 binding to the Bcl-2 ARE R1 and Bcl-xL ARE F1 was greater in NiT cells compared with BEAS-2B cells, where Nrf2 binding was minimal (Fig. 5H). These results suggest that the high expression levels of antiapoptotic proteins in NiT cells at least partially result from Nrf2-mediated up-regulation of transcription in NiT cells.

Nrf2 regulates intracellular ROS levels in NiT cells
Intracellular ROS levels serve as an important determinant of cell fate upon exposure to internal or external stimuli (46). We used several methods to measure the intracellular ROS levels of non-transformed BEAS-2B and NiT cells to determine whether the levels of these species differ between the two cell lines. First, we recorded the ESR spectra of BEAS-2B and NiT cells using DMPO for free radical trapping. We detected a 1:2:2:1 quartet ESR signal for the BEAS-2B cells. In contrast, the ESR signal for the NiT cells was extremely weak (supplemental Fig. 4A). The ESR signal was increased in Ni 2ϩ -exposed BEAS-2B cells but not in NiT cells (Fig. 6A). Next, we stained the cells with CM-H 2 DCFDA and analyzed fluorescence intensity using flow cytometry, fluorescence microscopy, or fluorescence spectroscopy (supplemental Fig. 4, B-D). In all of these experiments, the fluorescence intensity was significantly lower in NiT cells than in non-transformed BEAS-2B cells, indicating that NiT cells contain lower levels of ROS. Importantly, these fluorescence intensities were significantly increased only in Ni 2ϩ -exposed BEAS-2B cells, whereas the fluorescence signal was only slightly increased in NiT cells (Fig. 6, A-D). These results suggested that NiT cells have a higher ROS scavenging potential than normal BEAS-2B cells. We also used Western blot analysis to evaluate whether higher levels of antioxidant enzymes could account for the lower ROS levels detected in NiT cells compared with BEAS-2B cells. The major antioxidant enzymes catalase and superoxide dismutases (SODs) 1 and 2 were in fact expressed to a greater extent in NiT cells than in BEAS-2B cells (Fig. 6E). Furthermore, when we exposed the cells to Ni 2ϩ , the expression levels of catalase, SOD1, and SOD2 decreased in a dose-dependent manner in both BEAS-2B and NiT cells; however, NiT cells retained higher levels of these enzymes than BEAS-2B cells did (Fig. 6F). Moreover, pharmacological inhibition of ROS scavengers in NiT cells, by the addition of catalase inhibitor 3AT, the SOD1 inhibitor LCS-1, or the SOD2 inhibitor 2ME, led to a greater reduction in viability following Ni 2ϩ exposure (Fig.  6G). Collectively, these results suggest that the high expressions of antioxidant enzymes in NiT cells allows them to maintain low ROS levels and contributes to their resistance to apoptosis following Ni 2ϩ exposure. . BEAS-2B and NiT cells were exposed to various concentrations of Ni 2ϩ (0 -2 mM) for 24 h (B), or to Ni 2ϩ (2 mM) for various lengths of time (0 -24 h) (C), and the levels of Nrf2, NQO1, HO-1 were assessed by Western blot analysis. To decrease Nrf2 levels, BEAS-2B and NiT cells were transfected with Nrf2-specific siRNA. After overnight transfection, cells were exposed to Ni 2ϩ (2 mM) for an additional 24 h. Nrf2 expression levels were evaluated by Western blot analysis (D), and an apoptosis assay was performed after staining with annexinV/PI (E). In addition, the levels of LC3, PARP, and caspases 3 and 7 were assessed by Western blot analysis (F), and the number of cells containing GFP-LC3 puncta was quantified (G). Data presented graphically are the mean Ϯ S.E. of three independent experiments, with significant differences from controls indicated as *, p Ͻ 0.05; **, p Ͻ 0.01, and ***, p Ͻ 0.001, identified by ANOVA and Scheffe's test). GAPDH was used as loading control in the Western blot analyses.

Nrf2 enhances carcinogenesis in nickel-transformed cells
To investigate the role of Nrf2 in maintaining low intracellular ROS levels in NiT cells, we transfected these cells with Nrf2specific siRNA. Suppression of Nrf2 expression in NiT cells resulted in elevated intracellular ROS levels, as determined by fluorescence microscopy (Fig. 6H), fluorescence spectroscopy ( Fig. 6I), and ESR (Fig. 6, J and K). Nrf2 knockdown also reduced the expression levels of catalase, SOD1, and SOD2 (Fig. 6L) as well as the activities of catalase and SOD (Fig. 6, M and N). These results suggest that Nrf2 up-regulates the levels of antioxidant enzymes in NiT cells.

Nrf2 enhances carcinogenesis in nickel-transformed cells Nrf2 induces autophagy by activating transcription of Stat3 in NiT cells
To investigate signaling pathways of Nrf2 on autophagy induction, we analyzed Stat3 levels. Compared with non-transformed BEAS-2B cells, total and phosphorylated Stat3 levels were dramatically higher in NiT cells than in parental BEAS-2B cells (Fig. 7A). Treating NiT cells with stattic, a pharmacological Stat3 inhibitor, or genetic knockdown Stat3 with siRNA blocked the Ni 2ϩ -mediated increase in LC3-II levels as well as the number of cells containing GFP-LC3 puncta (Fig. 7, B and  C). To further investigate the role of Stat3 in autophagic activity, we used plasmid DNA to overexpress Stat3 in BEAS-2B cells (Fig. 7D). Overexpressing Stat3 led to significantly higher LC3-II levels and to a greater number of cells containing GFP-LC3 puncta, following Ni 2ϩ exposure (Fig. 7, E and F). These results indicate that Stat3 plays an important role in inducing autophagy in response to Ni 2ϩ exposure. We further investigated whether Stat3 levels are regulated by Nrf2 in BEAS-2B cells. When we used siRNA to suppress Nrf2 expression in NiT cells, the levels of Stat3 and phosphorylated Stat3 were sharply reduced (Fig. 7G). To determine whether Nrf2 transcriptionally activates Stat3 in our model, we performed ChIP analysis. Nucleotide sequence analysis of the 3-kb Stat3 promoter revealed the presence of five consensus AREs as follows: one in the forward strand and four in the reverse strand (Fig. 7H). ChIP analysis demonstrated that, in NiT cells, Nrf2 binds to ARE F1 (Ϫ41 to Ϫ32) and ARE R1 (Ϫ1731 to Ϫ1720) in the Stat3 promoter (Fig. 7I). Interestingly, the degree of Nrf2 binding to these AREs was much greater in NiT cells than in BEAS-2B cells (Fig. 7I). Quantitative real-time PCR results further supported this conclusion (Fig. 7J). These results indicate that Nrf2 directly activates Stat3 transcription and induces autophagy in NiT cells.

Nrf2 plays a critical role in Ni 2؉ -induced malignant transformation and malignant tumorigenesis
We next used a genetic knockdown approach to investigate whether Nrf2 is involved in Ni 2ϩ -induced carcinogenesis. Soft agar and clonal assays revealed that, following prolonged Ni 2ϩ exposure, suppressing Nrf2 expression in BEAS-2B cells whether by siRNA or shRNA transfection significantly impaired colony growth (Fig. 8, A-C). An shRNA-mediated knockdown of Bcl-2 also impeded colony growth in cells following prolonged Ni 2ϩ exposure (Fig. 8D). When we injected the NiT cells that had been exposed to Ni 2ϩ for 4 months into mice, Ni 2ϩ -induced tumor volume was dramatically increased than unexposed parental control cells (Fig. 8, E and F), whereas body weight remained similar to unexposed parental control cells (Fig. 8G). Interestingly, BEAS-2B cells that had been transfected with Nrf2 shRNA prior to the prolonged Ni 2ϩ exposure produced smaller tumors than those that had been transfected with control shRNA prior to this exposure (Fig. 8H). To investigate whether Nrf2 may play a role in lung tumorigenesis in humans, we examined the expression levels of Nrf2 in tumor tissues and adjacent normal tissues obtained from lung cancer patients (Fig. 8I). The expression level of Nrf2 was dramatically higher in lung tumor tissues when compared with adjacent normal tissues from the same patients (Fig. 8I). The mitotic index of tumor tissues was confirmed by Ki67 and hematoxylin and eosin (H&E) staining (Fig. 8I). Overall, our in vitro, in vivo, and clinical data suggest that the Nrf2 plays a pivotal role in the development of lung adenocarcinoma.

Discussion
Industrial or environmental exposure to nickel compounds is associated with a higher incidence of human lung cancer (1-3). Nickel has weak mutagenic effects (7,47) and could transform cells by acting on a number of target molecules, including irondependent enzymes, ROS, and proteins involved in hypoxia (5,48). The precise mechanisms underlying its carcinogenic activity are not fully understood.
Nrf2 is a basic leucine zipper transcription factor that serves as a master regulator of cellular redox homeostasis (49). Nrf2 binds to antioxidant-response elements in gene promoters and transcriptionally activates antioxidant enzymes as well as several apoptosis regulatory proteins (31,44,45). Constitutive activation of Nrf2 contributes to malignant transformation, and high expression of Nrf2 has been observed in a variety of tumor cells (38,50). Previously, our group reported that Nrf2 is highly expressed in metal-transformed cells and that this may play a critical role in metal-induced carcinogenesis (11,12). Our immunohistochemical results presented in this study show that Nrf2 expression was markedly higher in lung tumor tissues from patients with a 40-year smoking history when compared with that of the adjacent normal tissues (Fig. 8I). In addition, although BEAS-2B human lung bronchial epithelial cells that had been chronically exposed to Ni 2ϩ produced tumors in mice, BEAS-2B cells in which Nrf2 expression had been suppressed prior to the Ni 2ϩ exposure produced smaller tumors (Fig. 8H). Silencing Nrf2 expression also impaired colony formation by BEAS-2B cells that had been chronically exposed to Ni 2ϩ (Fig. 8, A-C). Collectively, these results support the path-   BEAS-2B cells were exposed to Ni 2ϩ (50 M) for 4 months. The cells were harvested each month, and a soft agar assay was performed (A). NiT cells were transfected with siRNA or shRNA specific for Nrf2, and a clonal assay (B) or soft agar assay (C) was performed, respectively. A soft agar assay was also performed for NiT cells following transfection with shRNA specific for Bcl-2 (D). BEAS-2B cells that had been exposed to Ni 2ϩ (50 M) for 4 months were injected subcutaneously (1 ϫ 10 6 cells/site) into 6-week-old male athymic nude mice. After 2 months, each tumor was dissected, and the tumor volume (E and F) and body weight (G) were measured. After transfection with Nrf2-specific shRNA, BEAS-2B cells were exposed to Ni 2ϩ (50 M) for 4 months. The cells were then injected subcutaneously (1 ϫ 10 6 cells/site) into 6-week-old male athymic nude mice. After 2 months, each tumor was dissected, and the tumor volume was measured (H). Normal and tumor tissues were obtained from lung adenocarcinoma patients (stage IA or IIA) with a smoking history of more than 40 years. After fixation with 4% paraformaldehyde, Nrf2 expression (upper) and proliferation (middle) were examined by immunohistochemical staining. The photomicrographs are representative images of nine samples (I). Graphic data are the mean Ϯ S.E. of three independent experiments, with significant differences indicated as *, p Ͻ 0.05; **, p Ͻ 0.01; and ***, p Ͻ 0.001 identified by ANOVA and Scheffe's test.

Nrf2 enhances carcinogenesis in nickel-transformed cells
ological significance of Nrf2 signaling in nickel-induced carcinogenesis. To understand underlying molecular mechanisms by which Nrf2 contributes to nickel-induced carcinogenesis, we generated NiT cells by continuously exposing BEAS-2B cells to 50 M Ni 2ϩ for 4 months (51). As anticipated, Nrf2 expression was increased in NiT cells when compared with the parental BEAS-2B cells (Fig. 4A). Furthermore, in NiT cells, the expression of Nrf2 was accompanied by resistance to apoptosis (Fig.  4E). NiT cells also had a higher proliferative potential than their parental cell line (Fig. 1). In exploring a possible link between high Nrf2 expression and apoptosis resistance, we found that, compared with BEAS-2B cells, NiT cells have lower ROS levels (Fig. 6, A-D, and supplemental Fig. 4) and higher levels of the antioxidant enzymes catalase, SOD1, and SOD2 (Fig. 6E). The high expression of antioxidant enzymes is also involved in the resistance of NiT cells to apoptosis (Fig. 6G). Because siRNA-mediated knockdown of Nrf2 in NiT cells enhanced the generation of ROS and attenuated the expression and activities of catalase, SOD1, and SOD2, we conclude that Nrf2 tightly regulates antioxidant enzymes and contributes to maintaining low levels of ROS in these cells (Fig. 6, H-N). Furthermore, compared with BEAS-2B cells, NiT cells express high levels of the antiapoptotic proteins Bcl-2 and Bcl-xL, contributing to the resistance of NiT cells to apoptosis (Fig. 5, A-E). ChIP analysis revealed that Nrf2 binding to the ARE-containing regions of the Bcl-2 or Bcl-xL promoter was greater in NiT cells than in BEAS-2B cells (Fig. 5, G and H). Moreover, siRNA-mediated knockdown of Nrf2 expression in NiT cells reduced the expressions of Bcl-2 and Bcl-xL (Fig. 5F). The up-regulation of antiapoptotic proteins by Nrf2 might play a major role in the apoptosis resistance of NiT cells. Our findings suggest that, by enhancing the levels of antiapoptotic proteins, Nrf2 acts as a survival factor in the Ni 2ϩ -transformed cells.
Autophagy is a degradation pathway that is essential for survival during starvation, hypoxia, immune responses, and chemotherapy exposure (17). Autophagy helps cell survival by inhibiting caspase activities, and, in turn, inhibition of caspase activities induces autophagy (52). We found that Ni 2ϩ increased the levels of LC3-II in NiT cells in a dose-and timedependent manner and to a greater extent than in BEAS-2B cells (Fig. 2, A and B), indicating that NiT cells are sensitive to autophagy induction following Ni 2ϩ exposure (Fig. 2). This phenomenon helps NiT cells to resist apoptosis (Fig. 3). We further confirmed the sensitivity of NiT cells to autophagy induction using several assays of autophagic flux (Fig. 2, C-G). In addition, Ni 2ϩ -induced cell viability was attenuated, and apoptosis was enhanced, by inhibiting autophagy using both pharmacological and genetic approaches in NiT cells (Fig. 3, A and  E). Notably, inhibiting apoptosis by Z-VAD, or by overexpressing Bcl-2, enhanced autophagic flux in the NiT cells (Fig. 3,  F-H). It appears that, in Ni 2ϩ -transformed cells, increased autophagy has a beneficial effect for cell survival (15).
Members of the signal transducer and activator of transcription (Stat) protein family are important inducers of cytokines and growth factors in some cancers (53). Among the seven Stat proteins, Stat3 is constitutively activated most frequently in several cancers (54) and is implicated in various steps of tumor development such as proliferation, survival, invasion, and angiogenesis (55,56). Recent reports describe a relationship between Nrf2 and Stat3. For example, the Nrf2-mediated upregulation of Stat3 has cardioprotective effects in a streptozotocin-induced diabetes model (57). Stat3 and Nrf2 promote the proliferation of neural stem cells (58) and inhibit virally induced inflammatory responses (59). Stat3-dependent autophagy induction correlates with the grade of human glioma and plays a central role in malignant glioma progression (60). We found five consensus AREs in the Stat3 promoter (Fig. 7H). Among the ARE-containing regions of the Stat3 promoter, we were able to detect the binding of Nrf2 to the ARE F1 (Ϫ41 to Ϫ32) and R1 (Ϫ1731 to Ϫ1720) (Fig. 7I). The extent of Nrf2 binding of these regions was higher in NiT cells than in non-malignant BEAS-2B cells (Fig. 7, H and I). Furthermore, siRNA-mediated knockdown of Nrf2 in NiT cells completely depleted the levels of phosphorylated Stat3 and sharply attenuated the total Stat3 levels (Fig. 7G). Notably, these results indicate that the high basal level of Stat3 in NiT cells directly results from Nrf2-mediated transcriptional activation. Presumably, the Nrf2-induced up-regulation of Stat3 contributes to the autophagy sensitivity and apoptosis resistance found in NiT cells. In BEAS-2B cells, overexpression of Stat3 via plasmid DNA transfection increased LC3-II levels as well as the number of cells containing GFP-LC3 puncta (Fig. 7, D-F), whereas inhibition of Stat3 with stattic or siRNA completely abolished the Ni 2ϩ -induced increase in LC3-II levels and in the number of cells containing GFP-LC3 puncta in NiT cells (Fig. 7, A-F). As shown in Fig. 3, we confirmed that autophagy contributes to cell survival in NiT cells in our system. The cell survival function of Stat3-dependent autophagy is consistent with other researcher's findings. For example, Stat3-dependent induction of autophagy was found to play an important role in taxol resistance in human colorectal cancer cells (61), and inhibition of Stat3-dependent autophagy enhances capsaicin-induced apoptosis in human hepatocellular carcinoma cells (62). These results suggest that autophagy and apoptosis have a negative feedback loop in cancer cells and that the regulations of these mechanisms are important for chemotherapeutic strategies.
Overall, our study demonstrates a role for Nrf2 in regulating apoptosis and autophagy, which is mediated by its ability to induce expression of antioxidant enzymes (e.g. catalase, SOD1, and SOD2), antiapoptotic proteins (e.g. Bcl-2 and Bcl-xL), and Stat3, in nickel-induced carcinogenesis (Fig. 9). Constitutive overexpression of Nrf2 enhances the up-regulation of antioxidant enzymes and antiapoptotic proteins in the transformed cells. These phenomena allow Ni 2ϩ -transformed cells to maintain low intracellular ROS levels and resist apoptosis. In addition, the Nrf2-induced up-regulation of Stat3 in NiT cells potentiates autophagy. Activation of autophagy inhibits apoptosis signaling in NiT cells. The suppression of apoptosis and promotion of autophagy contribute to cell survival and carcinogenesis of the Ni 2ϩ -transformed cells. Collectively, our findings provide new insight into the molecular mechanisms underlying nickel-induced autophagy and apoptosis and point toward a possible effective therapeutic strategy for metal-induced carcinogenesis.

Cell culture and treatment
The human lung bronchial epithelial cell line BEAS-2B was from the American Type Culture Collection (CRL-9609). NiT cells were generated as described previously. (63). The transformation and tumorigenicity of the NiT cells were confirmed by soft agar and xenograft assays, respectively. Ni 2ϩ -transformed BEAS-2B (NiT) cells and their parental non-transformed BEAS-2B cells were maintained in DMEM supplemented with 10% heat-inactivated FBS and 1% penicillin/streptomycin (05140122, Thermo Fisher Scientific, Inc).

FITC-Annexin V/PI staining
The apoptosis assay was performed as described previously (64). The scatter parameters of the cells (20,000 cells per experiment) were analyzed with a FACSCalibur system (BD Biosciences) using annexin V (556420, BD Biosciences) and PI (P4170, Sigma). The early apoptotic population in the lower right quadrant (low PI and high FITC signals) and the late apoptotic population in the upper right quadrant (high PI and high FITC signals) were considered as apoptotic cells.

Measurement of cellular ROS levels
The electron spin resonance (ESR) assay was performed using a Bruker EMX spectrometer (Bruker Instruments, Billerica, MA) and a flat cell assembly, as described previously (65). NiT cells and normal BEAS-2B cells (1 ϫ 10 6 ) were cultured overnight, harvested, and mixed with DMPO (50 mM; catalog no. D5766, Sigma). The Acquisit program was used for data acquisition and analysis (Bruker Instruments). For fluorescence Figure 9. Proposed model of Ni 2؉ -induced carcinogenesis. Non-transformed cells express low levels of Nrf2. Exposure of these cells to Ni 2ϩ leads to weak apoptotic or autophagic signaling. However, the high expression levels of Nrf2 in NiT cells leads to up-regulation of antiapoptotic proteins (e.g. Bcl-2 and Bcl-xL) and antioxidant enzymes (e.g. catalase and SODs), causing them to be less sensitive to apoptotic signaling. Nrf2 also activates Stat3, which leads to increased autophagy. Both the suppression of apoptosis and the promotion of autophagy increase cell survival, apoptosis resistance, and oncogenicity. Negative regulation between apoptosis and autophagy in response to Ni 2ϩ stimuli enhances the carcinogenic potential and survival of transformed cells.

Nrf2 enhances carcinogenesis in nickel-transformed cells
microscopy image analysis, the cells (2 ϫ 10 4 ) were seeded onto glass coverslips placed in the wells of a 24-well plate and incubated overnight. Attached cells were then exposed to CM-H 2 DCFDA (5 M; catalog no. C6827, Thermo Fisher Scientific) for 30 min. After washing with PBS, the coverslips were mounted onto microscope slides, and the cells were observed with a fluorescence microscope (Carl Zeiss, Germany). To determine the fluorescence intensity of the DCFDA signal, the cells (10,000 cells/well) were seeded into a 96-well culture plate and incubated overnight. Cells were treated with CM-H 2 DCFDA (5 M) for 30 min. After two washes with PBS, 2Ј,7Ј-dichlorofluorescein diacetate fluorescence was measured using a Spectramax GEMINIXPS Fluorescence Microplate Reader (Molecular Devices, Sunnyvale, CA). In addition, cells (0.5 ϫ 10 6 cells/well) were seeded into 60-mm culture dishes and incubated overnight. These cells were also exposed to CM-H 2 DCFDA (5 M) for 30 min and processed for flow cytometric analysis (FACSCalibur system, BD Biosciences).

Anchorage-independent colony growth assays
The soft agar assay was performed as described previously (67). Briefly, 0.5% agar (30391023, Thermo Fisher Scientific) (3 ml) in DMEM supplemented with 10% FBS was spread onto each well of a 6-well culture plate. A suspension (1 ml) containing NiT cells or BEAS-2B cells (1 ϫ 10 4 ) was mixed with 0.5% agar/DMEM (2 ml) and plated on top of the 0.5% agar layer. The plates were incubated at 37°C in 5% CO 2 for 2 months, and colonies larger than 50 m in diameter were counted using a light microscope (Thermo Fisher Scientific).

Chromatin immunoprecipitation (ChIP) assay
The ChIP assay was performed using a kit (26156, Thermo Fisher Scientific). Briefly, 90% confluent NiT cells or non-transformed BEAS-2B cells were used for the assay. DNA and proteins were cross-linked by incubating the cells with 1% formaldehyde for 10 min at room temperature. Excess formaldehyde was quenched with glycine for 5 min. The cells were lysed, and nuclei were digested with micrococcal nuclease provided in the kit. The sheared chromatin was diluted (1:10) and immunoprecipitated with 2 g of an anti-Nrf2 or a control IgG antibody. DNA-protein complexes were eluted from the protein A/Gagarose beads using a spin column, and cross-links were reversed by incubating with NaCl at 65°C. The relative levels of Nrf2 bound to the ARE-containing regions of the Bcl-2, Bcl-xL, and Stat3 promoters were analyzed using the MyiQ TM Single-Color Real-Time PCR Detection System (Bio-Rad) with SYBR Green PCR master mix (4472903, Thermo Fisher Scientific). PCR-mediated amplification was performed using a Mastercycler thermal cycler (Eppendorf, Foster City, CA).

Immunohistochemical staining
Human lung adenocarcinoma tissues (stage IA or IIA) derived from patients with a 40-year smoking history were provided by the Biospecimen and Tissue Procurement Shared Resource Facility of the University of Kentucky Markey Cancer Center. Tumor tissues were fixed with 4% paraformaldehyde at room temperature for 24 h, embedded in paraffin, and sectioned (5 m thickness). The tissue sections were deparaffinized, rehydrated, and processed for immunohistochemical staining using a kit (PK6100, Vector Laboratories). Briefly, the sections were incubated in 3% H 2 O 2 in distilled water to block endogenous peroxidase activity. After antigen retrieval, the sections were blocked with normal serum for 20 min and then incubated with primary antibodies for 1 h. Sections incubated with nonspecific mouse or rabbit serum IgGs served as negative controls. After washing with PBS, the sections were incubated with biotinylated secondary antibodies for 30 min. The sections were then washed twice with PBS, incubated with ABC reagent for 30 min, and developed in 3,3Ј-diaminobenzidine solution until the desired staining intensity was achieved.

Statistical analysis
All data are expressed as means Ϯ S.E. For multiple comparisons, one-way analysis of variance (ANOVA) was performed using IBM SPSS Statistics 21 software (IBM Corp., Armonk, NY). Values of p Ͻ 0.05 were considered statistically significant.
Author contributions-Y.-O. S. designed the study and wrote the paper. P. P. performed the ChIP, soft agar, and xenograft assay shown in Figs. 5, 7, and 8. S. P. D. and Z. Z. generated transformed cells and performed autophagy assay shown in Figs. 1 and 2. X. S. conceived and coordinated the study and revised the paper.