Involvement of Notch1 signaling in malignant progression of A549 cells subjected to prolonged cadmium exposure

Cadmium exposure is known to increase lung cancer risk, but the underlying molecular mechanisms in cadmium-stimulated progression of malignancy are unclear. Here, we examined the effects of prolonged cadmium exposure on the malignant progression of A549 human lung adenocarcinoma cells and the roles of Notch1, hypoxia-inducible factor 1α (HIF-1α), and insulin-like growth factor 1 receptor (IGF-1R)/Akt/extracellular signal-regulated kinase (ERK)/p70 S6 kinase 1 (S6K1) signaling pathways. Exposing A549 cells to 10 or 20 μm cadmium chloride (CdCl2) for 9–15 weeks induced a high proliferative potential, the epithelial-mesenchymal transition (EMT), stress fiber formation, high cell motility, and resistance to antitumor drugs. Of note, the CdCl2 exposure increased the levels of the Notch1 intracellular domain and of the downstream Notch1 target genes Snail and Slug. Strikingly, siRNA-mediated Notch1 silencing partially suppressed the CdCl2-induced EMT, stress fiber formation, high cell motility, and antitumor drug resistance. In addition, we found that prolonged CdCl2 exposure induced reduction of E-cadherin in BEAS-2B human bronchial epithelial cells and antitumor drug resistance in H1975 human tumor-derived non-small-cell lung cancer cells depending on Notch1 signaling. Moreover, Notch1, HIF-1α, and IGF-1R/Akt/ERK/S6K1 activated each other to induce EMT in the CdCl2-exposed A549 cells. These results suggest that Notch1, along with HIF-1α and IGF-1R/Akt/ERK/S6K1 signaling pathways, promotes malignant progression stimulated by prolonged cadmium exposure in this lung adenocarcinoma model.

exposure on Notch1 signaling in lung epithelial cells and its possible involvement in cadmium-induced malignant progression have not yet been clarified.
Therefore, we examined whether prolonged cadmium chloride (CdCl 2 ) exposure induces malignant progression in A549 human lung adenocarcinoma cells through the activation of Notch1 signaling. Our results show that cadmium exposure induces Notch1-dependent EMT, stress fiber formation, high cell motility, and antitumor drug resistance in A549 cells. We also found that hypoxia-inducible factor 1␣ (HIF-1␣), a stressresponsive transcription factor, enhanced the transcriptional activity of Notch1 and that insulin-like growth factor 1 receptor (IGF-1R)/Akt/extracellular-signal regulated kinase (ERK)/p70 S6 kinase 1 (S6K1) cascade could interplay with Notch1 signaling and HIF-1␣ in response to cadmium exposure.

Prolonged cadmium exposure induces EMT, stress fiber formation, and high cell motility in A549 cells
First, we examined the proliferative potential of A549 cells following exposure to varying doses (1-20 M) of CdCl 2 for 1-14 weeks. Cells exposed to 5, 10, or 20 M CdCl 2 for 1 week showed lower proliferation compared with cells without CdCl 2 exposure (supplemental Fig. S1a). This lower proliferation became less marked after 8 weeks of exposure (supplemental Fig. S1a). Conversely, cells exposed to 10 or 20 M CdCl 2 for 10 (supplemental Fig. S1a) and 14 weeks (Fig. 1A) showed higher proliferation than cells without CdCl 2 exposure. Although A549 cells without CdCl 2 exposure maintained a typical epithelial morphology and were organized in compact islets, cells A, cells incubated with 0, 1, 5, 10, or 20 M CdCl 2 for 14 weeks were seeded and allowed to proliferate for 5 days. Total cell numbers were counted each day using a cell counter. Each value is the ratio of the cell numbers at Day 0 and reflects the mean Ϯ S.D. (error bars) of three experiments with duplicate assays in each experiment. B, cells were incubated with 0, 5, 10, or 20 M CdCl 2 for 10 weeks, and their phase-contrast micrographs were taken. Scale bar, 100 m. C, cells were incubated with 0, 1, 5, 10, or 20 M CdCl 2 (Cd) for 10 weeks. Cell lysates were subjected to Western blotting using the indicated antibodies. Immunoblots shown are representative of at least three independent experiments. D, cells were incubated with 0 or 20 M CdCl 2 for 10 weeks and stained with phalloidin-FITC for F-actin and DAPI for DNA. Scale bar, 20 m. E, cells were incubated with 0, 10, or 20 M CdCl 2 for 10 weeks and allowed to migrate for 20 h after scratching. Results of quantification of the area covered by cells are also shown (E). Each value (mean Ϯ S.D. of three experiments) was expressed as a percentage of the area covered. Scale bar, 250 m. **, p Ͻ 0.01, significant difference between the samples. Malignant progression in cadmium-exposed A549 cells exposed to 10 or 20 M CdCl 2 for 1 week changed in morphology to a rounded shape (supplemental Fig. S1b). In contrast, cells exposed to 10 or 20 M CdCl 2 for 10 weeks showed an EMT-like spindle shape without definite cellular damage (Fig.  1B). This EMT-like change was also observed in cells exposed to 5 M CdCl 2 for 32 weeks (supplemental Fig. S1c). Furthermore, expression of E-cadherin decreased consistently, and that of N-cadherin and vimentin increased in cells exposed to 1-20 M CdCl 2 for 10 weeks in a dose-dependent manner (Fig.  1C). EMT-inducing stressors such as transforming growth factor ␤ (TGF-␤) have been reported to induce the formation of stress fibers (23). Exposure to 20 M CdCl 2 for 10 weeks also produced many stress fibers in A549 cells (Fig. 1D). Woundhealing assays showed that exposure to 10 or 20 M CdCl 2 for 10 weeks increased cell motility (Fig. 1E). In addition, the mRNA level of matrix metalloproteinase 2, a type VI collagenase that facilitates tumor invasion (24), was elevated in prolonged CdCl 2 -exposed cells (supplemental Fig. S2). Taken together, these results indicate that CdCl 2 exposure for more than 8 weeks induced EMT, stress fiber formation, and high cell motility in A549 cells. For the subsequent experiments, we used A549 cells that were incubated with 20 M CdCl 2 for 9 -15 weeks because cadmium-induced EMT and high cell motility were observed in these cells with prolonged CdCl 2 exposure (data not shown).

Notch1 is involved in prolonged cadmium exposure-induced antitumor drug resistance in A549 cells
We determined the viability of prolonged CdCl 2 -exposed A549 cells treated with cisplatin, gemcitabine, and etoposide, antitumor drugs commonly used in lung cancer chemotherapy, using trypan blue exclusion assays. Prolonged CdCl 2 exposure resulted in lower cell death induced by treatment with cisplatin ( Fig. 3A), gemcitabine (Fig. 3B), and etoposide (data not shown), indicating the development of higher antitumor drug resistance. Deletion of Notch1 expression with siRNAs (Notch1-1 and Notch1-2) partially but significantly increased cell death induced by cisplatin (Fig. 3C) and gemcitabine (Fig.  3D) in prolonged CdCl 2 -exposed cells. In addition, knockdown of Snail increased cell death induced by cisplatin (supplemental Fig. S4a) and gemcitabine (supplemental Fig. S4b) in these cells. These findings indicate that prolonged CdCl 2 exposure induced antitumor drug resistance through Notch1 signaling in A549 cells.

HIF-1␣ transcriptional activity is not required for the activation of Notch1 signaling in prolonged cadmiumexposed A549 cells
It has been reported that HIF-1␣ increases the expression of Jagged2 and anterior pharynx-defective 1 (APH-1), a component of the ␥-secretase complex, through the binding to their promoters (29,30), resulting in the activation of Notch1 signaling. To investigate whether the transcriptional activity of HIF-1␣ is required for Notch1 activation in prolonged CdCl 2exposed cells, we depleted the expression of arylhydrocarbon receptor nuclear translocator (ARNT), an HIF-1␣ binding part-ner for DNA binding (28). Transfection with siRNAs targeted against the human ARNT gene (ARNT-1 and ARNT-2) markedly suppressed ARNT expression (Fig. 5A) and almost completely abolished the cadmium-induced increase in mRNA levels of vascular endothelial growth factor A (VEGF-A), an HIF-1␣ target gene (28) (Fig. 5B). However, there was no sig-nificant difference in Notch1-ICD and HIF-1␣ expression between prolonged CdCl 2 -exposed cells in the presence and absence of ARNT (Fig. 5A). Although prolonged CdCl 2 exposure increased the levels of Jagged2 mRNA (supplemental Fig.  S3c) and its protein (Fig. 5C), depletion of ARNT did not change cadmium-induced up-regulation of Jagged2 expression ( Fig.   Figure 2. Notch1 is involved in prolonged CdCl 2 -induced EMT, stress fiber formation, and high cell motility in A549 cells. A, cells were incubated with 0, 1, 5, 10, or 20 M CdCl 2 (Cd) for 10 weeks. Cell lysates were subjected to Western blotting using the indicated antibodies. B, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA, Notch1 siRNA-1, or Notch1 siRNA-2. Cell lysates were subjected to Western blotting using the indicated antibodies. C, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were treated with 0.1% DMSO or 50 M DAPT for 16 h. Cell lysates were subjected to Western blotting using the indicated antibodies. D, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA, Notch1 siRNA-1, Notch1 siRNA-2, Snail siRNA-1, or Snail siRNA-2. Cell lysates were subjected to Western blotting using the indicated antibodies. Immunoblots shown are representative of at least three independent experiments. e and f, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA, Notch1 siRNA-1, or Notch1 siRNA-2. Transfected cells were stained with phalloidin-FITC for F-actin and DAPI for DNA (E) and used for wound healing assays (F). Scale bar, 20 m (e), 250 m (F). Results of quantification of the area covered by cells are also shown (F). Each value (mean Ϯ S.D. (error bars) of three experiments) was expressed as a percentage of the area covered. **, p Ͻ 0.01, significant difference between the samples. 5C). Treatment with 1 nM echinomycin, an HIF-1␣ transcriptional inhibitor, did not affect the levels of Notch1-ICD and HIF-1␣ expression in prolonged CdCl 2 -exposed cells (supplemental Fig. S5), whereas treatment with 0.3 nM echinomycin fully suppressed HIF-1␣ transcriptional activity (31). Furthermore, there was no significant difference in APH-1a mRNA levels between prolonged CdCl 2 -exposed cells in the presence and absence of HIF-1␣ (supplemental Fig. S6). These findings indicate that HIF-1␣ transcriptional activity is not required for the activation of Notch1 signaling in prolonged cadmium-exposed A549 cells.

Notch1 is involved in prolonged cadmium exposure-induced cellular responses in BEAS-2B and H1975 cells
We examined whether Notch1 signaling regulates prolonged cadmium-induced malignant progression in lung-related can- . HIF-1␣ regulates Notch1 activity in prolonged CdCl 2 -exposed A549 cells. A, cells were incubated with 0, 1, 5, 10, or 20 M CdCl 2 (Cd) for 10 weeks. Cell lysates were subjected to Western blotting using the indicated antibodies. B, total RNAs from prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were subjected to quantitative RT-PCR to determine HIF-1␣ mRNA levels. Data were normalized to ␤-actin expression and reflect the mean Ϯ S.D. (error bars) of three experiments. C, cell lysates from prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were subjected to Western blotting using the indicated antibodies. D, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA, HIF-1␣ siRNA-1, or HIF-1␣ siRNA-2. Cell lysates were subjected to Western blotting using the indicated antibodies. E, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA or HIF-1␣ siRNA-1. Cell lysates were subjected to Western blotting using the indicated antibodies. Results of densitometric analysis are also shown. Immunoblots shown are representative of at least three independent experiments. Data reflect the mean Ϯ S.D. of three experiments. *, p Ͻ 0.05; **, p Ͻ 0.01, significant difference between the samples. cer cells other than A549 cells, including BEAS-2B human bronchial epithelial cells and H1975 human tumor-derived non-small-cell lung cancer cells. Consistent with the previous report (9), we found that prolonged CdCl 2 -exposed BEAS-2B cells showed cadmium resistance depending on Nrf2 expression (supplemental Fig. S8, a-c). In addition, Notch1 knockdown partially suppressed cadmium-induced alterations of E-cadherin and Snail levels and stress fiber formation in these resistant BEAS-2B cells (Fig. 7, A and B). Furthermore, we found that prolonged CdCl 2 -exposed H1975 cells obtained the cisplatin resistance via activation of Notch1 signaling (Fig. 7, C and D).

Discussion
A549 cells are a human lung adenocarcinoma cell line with properties of type II alveolar epithelial cells (42). We found that exposure to CdCl 2 for more than 8 weeks enhanced the proliferative ability of A549 cells. The transcription factor Nrf2 has been reported to be one of the key factors that induce a high proliferative ability in cadmium-transformed BEAS-2B cells (9), and its downstream target, heme oxygenase-1, is involved in the suppression of cadmium toxicity in kidney and pulmonary cells (43,44). Consistent with these findings, knockdown of Nrf2 (supplemental Fig. S9, a-c) and heme oxygenase-1 (supplemental Fig. S9, d-f) reduced the proliferative potential in prolonged CdCl 2 -exposed A549 cells. To clarify the regulatory mechanisms of cadmium-induced malignant progression, we used A549 cells that were incubated with 20 M CdCl 2 for 9 -15 weeks as the in vitro model of prolonged cadmium-exposed lung epithelial cells.
We found that prolonged CdCl 2 exposure induced EMT, stress fiber formation, high cell motility, and antitumor drug resistance in A549 cells. In concordance with the findings that chronic cadmium exposure induces EMT-like characteristics in HPL-1D human peripheral lung epithelial cells (45), cadmium-induced malignant progression was also clearly observed in our model using A549 cells. Furthermore, consistent with our previous findings in HK-2 human renal proximal epithelial cells treated with CdCl 2 (22), an increase in the levels of Notch1-ICD and its downstream targets, Snail and Slug, was found in prolonged CdCl 2 -exposed A549 cells. Notch1 knockdown partially suppressed prolonged CdCl 2 -induced EMT, stress fiber formation, high cell motility, and antitumor drug resistance. In addition, we also found that prolonged CdCl 2 exposure induced reduction of E-cadherin in BEAS-2B cells and antitumor drug resistance in H1975 cells depending on Notch1 signaling. These findings demonstrate for the first time, to our knowledge, that Notch1 signaling is involved in cadmium-induced malignant progression in lung cancer cells, including A549 cells. Because these findings remained in the prolonged CdCl 2 -exposed A549 cells after removal of CdCl 2 from culture medium for 10 weeks, cadmium-induced malignant progression via the Notch1 pathway may be maintained (supplemental Fig. S10).

Figure 5. HIF-1␣ transcriptional activity is not required for activation of Notch1 in prolonged CdCl 2 -exposed A549 cells.
A, prolonged 20 M CdCl 2exposed A549 cells (Cd) were transfected with control siRNA, ARNT siRNA-1, or ARNT siRNA-2. Cell lysates were subjected to Western blotting using the indicated antibodies. B and C, prolonged 20 M CdCl 2 -exposed A549 cells (Cd) were transfected with control siRNA or ARNT siRNA-1. Total RNAs were subjected to quantitative RT-PCR to determine VEGF-A mRNA levels. Data were normalized to ␤-actin expression and reflect the mean Ϯ S.D. (error bars) of three experiments. *, p Ͻ 0.05; **, p Ͻ 0.01, significant difference between the samples (B). Cell lysates were subjected to Western blotting using the indicated antibodies (C). Immunoblots shown are representative of at least three independent experiments.
It has been reported that Notch1 is one of the important components that induce EMT under hypoxia (16), and it is also correlated with antitumor drug resistance in T-cell acute lymphoblastic leukemia cells (16), breast cancer cells (17), and A549 lung cancer cells (18). However, knockdown of Notch1 failed to fully suppress up-regulation of N-cadherin and vimentin in prolonged CdCl 2 -exposed cells. It has been reported that TGF-␤ signaling induces EMT (23), and its expression increases in response to cadmium exposure (45,46). When expression of both TGF-␤1 and TGF-␤2 was reduced, cadmium-induced up-regulation of N-cadherin and vimentin was found to be only partially suppressed (supplemental Fig. S11).
These findings suggest that other pathways might also function in parallel with Notch1 and TGF-␤ signaling for the induction of EMT in prolonged CdCl 2 -exposed A549 cells. We found that knockdown of Notch1 did not suppress cadmium-induced upregulation of Nrf2 expression. Conversely, Nrf2 knockdown suppressed the reduction of E-cadherin and elevation of Snail levels without affecting the Notch1-ICD level in prolonged CdCl 2 -exposed A549 cells (supplemental Fig. S12). These findings suggest that Nrf2 may be another signaling pathway that regulates E-cadherin expression in response to the prolonged cadmium exposure.
HIF-1␣ has been reported to activate Notch1 signaling in either a transcriptional activity-dependent or -independent Malignant progression in cadmium-exposed A549 cells manner (28 -32). In response to prolonged CdCl 2 exposure, HIF-1␣ protein but not its transcripts was accumulated in A549 cells, likely due to a perturbance of proteasomal degradation of HIF-1␣. Silencing of HIF-1␣ expression suppressed the cadmium-induced increase in Notch1-ICD and reduction of E-cadherin expression. In contrast, inhibition of HIF-1␣ transcriptional activity with ARNT knockdown or echinomycin treatment failed to suppress Notch1 activation. Although expression of Jagged2, APH-1, and Notch1 has been reported to be up-regulated by HIF-1␣ (28 -30), Jagged2 expression was not affected by the depletion of ARNT. Furthermore, we found that expression of APH-1 and Notch1-NTM was not affected by the depletion of HIF-1␣. These findings suggest that HIF-1␣ induces the activation of Notch1 signaling without depending on its transcriptional activity in prolonged CdCl 2 -exposed A549 cells. In contrast, it has been reported that a transcriptionally inactive mutant of HIF-1␣ can stabilize Notch1-ICD in COS-7 cells (32). HIF-1␣ can also activate Notch1 signaling through a direct interaction with the ␥-secretase complex (31). Further studies are needed to examine these transcriptional activity-independent mechanisms in A549 cells exposed to cadmium.
Conversely, the experiments using Notch1 siRNAs showed that Notch1 signaling induces the accumulation of HIF-1␣ protein in prolonged CdCl 2 -exposed A549 cells. In addition, activation of the Akt/ERK/S6K1 pathway, which can accelerate the translation of HIF-1␣ (35,36), was found to be regulated by Notch1 signaling as has been reported previously (30,(37)(38)(39)(40). Notch1 activation increases the expression of IGF-1R and IGF-1 and results in the activation of the downstream Akt pathway in lung adenocarcinoma cells during hypoxia (37). However, depletion of Notch1 had little to no effect on the expression of IGF-1R, IGF-2, and PTEN but did suppress the phosphorylation of IGF-1R and Akt in prolonged CdCl 2 -exposed A549 cells. The IGF-1R kinase inhibitor PPP at a concentration (200 nM) without detectable cytotoxicity partially suppressed the expression of HIF-1␣ and the phosphorylation of S6K1. The Akt/ERK pathway has also been reported to induce HIF-1␣ expression in human bronchial epithelial cells exposed to cadmium (8). We also confirmed that the ERK/Akt/S6K1 pathway was involved in HIF-1␣ accumulation in prolonged CdCl 2 -exposed A549 cells. These findings suggest that the activation of the IGF-1R/Akt/ERK/S6K1 pathway is involved in the accumulation of HIF-1␣ in prolonged CdCl 2 -exposed A549 cells. Oxidative stressors such as cigarette smoke extract, H 2 O 2 , and ozone induce a ligand-independent aberrant activation of a receptor-type tyrosine kinase (47)(48)(49). Although the possible involvement of IGF-1 cannot be excluded, a Notch1-mediated mechanism independent of ligand binding remains to be clarified. In addition, cadmium exposure-induced activation of other receptor-type tyrosine kinases, including the epidermal growth factor receptor (22,50), might contribute to the subsequent activation of the Akt/ERK pathway. In summary, the present study shows that prolonged CdCl 2 exposure activates Notch1 signaling and induces malignant progression in A549 lung adenocarcinoma cells. In addition, the HIF-1␣ and IGF-1R/Akt/ERK/S6K1 signaling pathways reciprocally activate Notch1 signaling to induce EMT in prolonged CdCl 2 -exposed A549 cells. Further experiments on Notch1 signaling after exposure to lower concentrations of cadmium, including animal models, will provide clues to understanding the malignant progression of lung cancer caused by cadmium exposure or cadmium from cigarette smoking. FlexiTube siRNA, SI02664053; siRNA-2, Hs_ HIF1A_10 Flexi-Tube siRNA, SI04249308), and ARNT (siRNA-1, Hs_ARNT_5 FlexiTube siRNA, SI03020913; siRNA-2, Hs_ ARNT_2 Flexi-Tube siRNA, SI00304220) and non-target siRNA (AllStars Negative Control siRNA) were purchased from Qiagen (Hilden, Germany). See supplemental "Materials and Methods" for more information.

Cell culture and treatments
A549 cells were obtained from the Health Science Research Resources Bank (Japan Health Sciences Foundation, Osaka, Japan) and grown in Earle's minimum essential medium (MEM) with non-essential amino acids supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Gibco, Invitrogen) in a humidified atmosphere of 5% CO 2 , 95% air at 37°C. BEAS-2B cells were obtained from DS Pharma Biomedical Corp. (Osaka, Japan) and grown in Dulbecco's modified Eagle's medium (DMEM) with non-essential amino acids supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere of 5% CO 2 , 95% air at 37°C. H1975 cells were kindly provided from Dr. Kenji Tanabe and grown in RPMI 1640 medium, HEPES with non-essential amino acids supplemented with 1 mM sodium pyruvate, 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin in a humidified atmosphere of 5% CO 2 , 95% air at 37°C. A549 cells were exposed continuously to CdCl 2 (0, 1, 5, 10, and 20 M) for 1-15 weeks. For each experiment, exponentially growing A549 cells were seeded at 3 ϫ 10 5 or 5 ϫ 10 5 cells/well in 6-well culture plates and cultured for 1 day before the experiments. DAPT and PPP were dissolved in dimethyl sulfoxide (DMSO). Cells were incubated with DMSO (0.1%) or each compound for 16 h. BEAS-2B and H1975 cells were exposed continuously to 10 and 30 M CdCl 2 for 10 weeks, respectively. For each experiment, exponentially growing cells were seeded at 3 ϫ 10 5 (for BEAS-2B cells) or 4 ϫ 10 5 cells/well in 6-well culture plates (for H1975 cells).

Preparation of whole cell lysates
After incubation, cells were washed with phosphate-buffered saline and lysed with sodium dodecyl sulfate-polyacrylamide gel Laemmli sample buffer. Cell lysates were collected, sonicated, and boiled for 5 min. Protein concentrations were determined using the RC DC Protein Assay (Bio-Rad).

Western blotting
Equal amounts of protein (20 g) were subjected to sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis and transferred onto a nitrocellulose membrane (Hybond-ECL, Amersham Biosciences). The membrane was blocked with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween 20 for 1 h at room temperature. The membrane was then incubated overnight at 4°C with the primary antibody, and protein was detected with a Phototope-HRP Western blot detection kit (Cell Signaling Technology, Inc.). The bands on the developed film were quantified with ImageJ 1.42 (National Institutes of Health, Bethesda, MD). The density of each band was normalized to that of GAPDH.

RNA isolation and reverse transcription-PCR
Total RNA was isolated with RNeasy (Qiagen), and firststrand cDNA was synthesized using the ReverTra Ace qPCR RT kit (Toyobo, Osaka, Japan). Quantitative reverse transcription-PCR (RT-PCR) analysis was performed using the THUN-DERBIRD SYBR qPCR Mix (Toyobo) in the StepOne TM Real-Time PCR system (Life Technologies Japan Ltd.). The expression level was normalized to that of glyceraldehyde-3phosphate dehydrogenase or ␤-actin. PCR primers are listed in supplemental Table S1.

Proliferation assay
4 ϫ 10 4 cells were seeded on 24-well culture plates for 5 days. Each day, the total cell number was counted using a TC10 TM automated cell counter (Bio-Rad).

Fluorescence microscopy
For staining with phalloidin, cells were grown on chamber slides, then fixed in 4% paraformaldehyde at room temperature, permeabilized in 0.5% Triton X-100, and incubated with phalloidin-FITC at 1:1000. Fluorescence was examined by a confocal laser-scanning microscope (Carl Zeiss, Oberkochen, Germany). Notch1, SNAI1, HIF1A, ARNT, NFE2L2,  HMOX1, TGFB1, and TGFB2 by siRNA   Transfection of siRNAs against human Notch1, SNAI1,  HIF1A, ARNT, NFE2L2, HMOX1, TGFB1, and TGFB2 and non-target siRNA into cells was performed using Lipofectamine 2000 (Invitrogen) (for A549 and H1975 cells) or Lipofectamine RNAiMAX (Invitrogen) (for BEAS-2B cells) according to the manufacturer's instructions with some adjustments. The siRNAs were dissolved in nuclease-free water and diluted to 0.2 M with 250 l of Opti-MEM (Invitrogen). 5 l of Lipofectamine 2000 was also diluted 50-fold with Opti-MEM. Equal volumes of these two solutions were mixed (500 l total) and immediately added to 2 ml of culture medium at the time of cell plating. After incubation for 24 h, cells were washed with medium and used for experiments.

Wound-healing assay
Cells were grown in MEM under appropriate conditions until a monolayer was formed. A wound was generated by creating a scratch using a sterile 1-ml pipette tip. Images were acquired immediately after the scratch at 0 h and again after 20or 24-h incubation at 37°C.

Trypan blue exclusion assay
Culture medium was aspirated and reserved. After trypsinization, cells were suspended in MEM, and the culture medium was returned. The mixture was centrifuged to pellet the cells. Cellular suspension and 0.4% trypan blue in Hanks' balanced salt solution were mixed, and the number of viable cells was counted using a TC10 automated cell counter.

Statistical analysis
Results are expressed as the mean Ϯ S.D. Statistical significance was determined by Student's t test. A value of p Ͻ 0.05 was considered to be statistically significant.