Keap1 – Nrf2 Interaction Suppresses Cell Motility in Lung Adenocarcinomas by Targeting the S100P Protein

Purpose: Kelch-like ECH-associated protein 1 (Keap1) is an E3 ligase participated in the cellular defense response against oxidative stress through nuclear factor erythroid-2 – related factor 2 (Nrf2). However, the role of Keap1 in regulating cancer motility is still controversial. We investigated the contribution of the Keap1 – Nrf2 axis in the progression of non – small cell lung cancer (NSCLC). Experimental Design: The expression of Keap1 and Nrf2 was examined via immunohistochemistry, real-time PCR, and Western blot analysis in a cohort of NSCLC tissues and cells. A series of in vivo and in vitro assays was performed to elucidate the contribution of the Keap1 – Nrf2 axis in lung cancer mobility and progression. Results: Keap1 expression was decreased in specimens from NSCLC patients with lymph node metastasis compared with patients without metastasis. Higher Keap1 expression levels were correlated with the survival of NSCLC patients. Moreover, manipulation of Keap1 expression affected cell migration/invasion abilities. Depletion of Nrf2 relieved the migration promotion imposed byKeap1suppression. Mechanistic investigations found that S100P was downregulated in bothKeap1-overexpressing and Nrf2-knockdown NSCLC cells. Overexpression of Keap1 and knockdown of Nrf2 both suppressed S100P expression in NSCLC cells. Knockdown of S100P inhibited cell migration in highly invasive NSCLC cells and also relieved the migration promotion imposed by Keap1 suppression in weakly invasive NSCLC cells. Conclusions: Our ﬁ ndings suggest that Keap1 functions as a suppressor of tumor metastasis by targeting the Nrf2/S100P pathway in NSCLC cells. In addition, overexpression of Keap1 may be a novel NSCLC treatment strategy and/or useful biomarker for predicting NSCLC progression. Clin Res; ECH-associated protein 1 (Keap1), were known to regulate redox-balancing genes, which mediate the adaptive response to reactive oxygen species (ROS) and xenobiotics of cancer cells. Here, we demonstrate that the Keap1 – Nrf2 axis also participated in eric0718 regulating cell motility in non – small cell lung cancer (NSCLC) cells. Clinical evalua-tions also show a strong correlation between the Keap1 – Nrf2 axis and the prognosis of NSCLC patients. Of note, we found thatthelymphnodestatusofNSCLCspecimenswaspositively correlated with the Nrf2 expression level and was inversely correlated with the Keap1 expression level. We also identify S100P, a known metastatic gene in NSCLC cells, as the key downstreameffector oftheKeap1 – Nrf2axis.Thus,theKeap1 – Nrf2 – S100P axis has an unexpected oncogenic role in NSCLC metastasis, and thus could serve both as a biomarker and as a therapeutic target.


Introduction
Lung cancer is the leading cause of cancer-related deaths worldwide (1). Non-small cell lung cancer (NSCLC) comprise 85% of lung cancers, and adenocarcinomas (ADC) and squamous cell carcinoma (SCC) are the most common histotypes (2). Metastasis is a significant problem and causes more mortalities than do primary tumors. Distant metastasis is the predominant cause of death in early-stage NSCLC (3).
Nuclear factor erythroid-2-related factor 2 (Nrf2) is a transcription factor that was reported to be highly expressed in cancer cells, especially NSCLC cells and was suggested to be associated with cancer cell behaviors, including regulation of cancer growth, survival, angiogenesis, and chemoresistance (4)(5)(6). The Kelch-like ECH-associated protein 1 (Keap1) gene is a negative regulator of a cell's adaptive response to reactive oxygen species (ROS) and xenobiotics mediated by Nrf2, and contains the BTB, IVR, and DGR domains (7,8). These domains contribute to the dual function of Keap1. Keap1 specifically interacts with the adaptor protein, Cullin 3 (Cul3), and forms an E3 ubiquitin ligase that renders substrate Nrf2 susceptible to rapid degradation (9). Keap1 also functions as a sensor for electrophilic/oxidative stresses. In the absence of stressful stimuli, Nrf2 is rapidly degraded through the proteasome pathway via ubiquitination by the Keap1-Cul3 E3 ligase (9). Upon exposure to oxidative stress, Keap1 is modified, and Nrf2 is released and translocated to nuclei where it specifically recognizes an enhancer sequence known as the antioxidant response element (ARE) resulting in activation of redox-balancing genes [e.g., heme-oxygenase (HO)-1], phase II detoxifying genes [e.g., NADP(P)H quinone oxidoreductase-1 (NQO1)], and drug transporters [e.g., multidrug resistant proteins (MRP); ref . 5]. HO-1, NQO1, and MRPs were also reported to be overexpressed in a variety of solid tumors, including NSCLC (10)(11)(12)(13) and play a role in regulating cancer growth (14), angiogenesis (15), and chemoprevention (16). Although the roles of Keap1/Nrf2 and their wellknown downstream genes in some cancer behaviors have been widely documented, the role of Keap1/Nrf2 and their underlying signaling in the regulation of cell motility have not been extensively studied. S100P is a member of the S100 family of small calcium-binding proteins that were reported to be overexpressed in many different cancers, including NSCLC, and its expression is associated with poor clinical outcomes (17,18). Previous DNA microarray data showed that among four histopathological subtypes of lung cancer (ADC, SCC, large-cell carcinoma, and small-cell carcinoma), S100P was identified as a candidate marker for ADC tumors (19). In terms of regulating tumor metastasis, S100P was demonstrated to increase the invasive ability of breast cancer (20) and to be overexpressed in metastatic tissues of NSCLC (21). Furthermore, the level of S100P in metastasizing lung tumors was correlated with poor patient survival (21). In this study, we explored whether the Keap1/Nrf2 axis plays a role in regulating cell motility and the possible role of S100P in the Keap1/Nrf2 axis-mediated cell motility.
Our results showed that Keap1 is a good prognostic factor the overexpression of which is associated with increased survival and decreased lymph node metastasis of lung cancer patients. We demonstrate that Keap1 can act as a suppressor of tumor migration and/or metastasis by degrading Nrf2, and S100P is an important downstream target of the Keap1/Nrf2 axis-mediated cell motility.

Patients and specimens
Lung specimens were obtained from a total of 238 consecutive patients who underwent surgical resection at National Taiwan University Hospital and Kaohsiung Veterans General Hospital. None of the patients had received preoperative adjuvant chemotherapy or radiation therapy. Approval for the study was obtained from both hospitals' ethics committees. Paraffinembedded surgical specimens from 238 patients were collected for immunohistochemical (IHC) staining for Keap1. Nrf2 staining were performed on 169 specimens of the same cohort with Keap1 staining.

Immunohistochemistry
Paraffin-embedded tissues were incubated in 3% hydrogen peroxide to block endogenous peroxidase activity. The primary antibodies were applied to slides at a dilution of 1:100 (diluted in 3% BSA) and incubated at 4 C overnight. After three washes in PBS, samples were treated with a donkey anti-goat immunoglobulin G (IgG) biotin-labeled secondary antibody (Vector Laboratories, Burlingame, CA, USA) at a dilution of 1:300 (diluted in 0.05% PBS-Tween 20) for 1 hour at room temperature. Bound antibodies were detected with an ABC Kit (Vector Laboratories). Slides were stained with diaminobenzidine, washed, counterstained with hematoxylin. Immunostaining was classified into one of two groups according to both the intensity and extent: low expression was defined as no staining present (a staining intensity score of 0) or positive staining detected in 20% of cells (a staining intensity score of 1), whereas high expression was defined as positive immunostaining present in 20% to 50% of cells (a staining intensity score of 2) or >50% of the cells (a staining intensity score of 3).

Cell lines
Lung adenocarcinoma cells were grown in RPMI-1640 medium with 10% fetal bovine serum (FBS) and 2 mmol/L l-glutamine (all from Life Technologies) at 37 C in a humidified atmosphere of 5% CO 2 /95% air. The CL1-5 subline of a lung ADC cell line was selected from parental CL1-0 cultures with a Matrigel-coated polycarbonate membrane (Collaborative Biomedical; BD Biosciences) in a Transwell invasion chamber as described previously (22). The PC9 and PC14 cell lines were obtained from IBL. Lung ADC cell lines (A549, H928, and H441), breast adenocarcinoma cell line (MCF7), and colorectal carcinoma cell line (HCT116) were obtained from American Type Culture Collection. All cell lines had been authenticated by STR analysis (Mission Biotech) as of December 2014, and all cells had been routinely tested and were negative for mycoplasma by 4 0 ,6-diamino-2-phenylindole (DAPI) staining.
RNA isolation, reverse-transcription polymerase chain reaction (RT-PCR), and microarray Total RNA was isolated using TRizol (Invitrogen) according to the manufacturer's instructions. Reverse-transcription of RNA isolated from cells was performed in a final reaction volume of 20 mL containing 5 mg of total RNA in Moloney murine leukemia virus (MMLV) reverse-transcriptase buffer (Promega), and 200 U of MMLV reverse transcriptase (Promega). The reaction mixture was incubated at 37 C for 2 hours, and the reaction was terminated by heating at 70 C for 10 min. One microliter of the reaction mixture was then amplified by a PCR with specific primers. RNA for the microarray analysis was also isolated from cells with TRizol (Invitrogen), purified on RNeasy columns (Qiagen), and checked for integrity by Agilent testing. Complementary (c)DNA was generated and hybridized to Human Genome U133 Plus 2.0 Arrays (Affymetrix) according to the manufacturer's instructions. The microarray datasets were deposited in Gene Expression Omnibus as GSE66473.

Translational Relevance
Nuclear factor erythroid-2-related factor 2 (Nrf2) and its E3 ligase, Kelch-like ECH-associated protein 1 (Keap1), were known to regulate redox-balancing genes, which mediate the adaptive response to reactive oxygen species (ROS) and xenobiotics of cancer cells. Here, we demonstrate that the Keap1-Nrf2 axis also participated in eric0718 regulating cell motility in non-small cell lung cancer (NSCLC) cells. Clinical evaluations also show a strong correlation between the Keap1-Nrf2 axis and the prognosis of NSCLC patients. Of note, we found that the lymph node status of NSCLC specimens was positively correlated with the Nrf2 expression level and was inversely correlated with the Keap1 expression level. We also identify S100P, a known metastatic gene in NSCLC cells, as the key downstream effector of the Keap1-Nrf2 axis. Thus, the Keap1-Nrf2-S100P axis has an unexpected oncogenic role in NSCLC metastasis, and thus could serve both as a biomarker and as a therapeutic target.

Lentiviral production and infection
Short hairpin (sh)RNAs were purchased from the National RNAi core Facility at Academic Sinica (Taipei, Taiwan). The target sequence of Keap1 shRNA 1 was 5 0 -GCGAATGATCA-CAGCAATGAA-3 0 ; that of Keap1 shRNA 2 was 5 0 -CGGGAGTA-CATCTACATGCAT-3 0 ; that of Nrf2 shRNA 1 was 5 0 -GCTCCTACTGTGATGTGAAAT-3 0 ; that of Nrf2 shRNA 2 was 5 0 -CCGGCATTTCACTAAACACAA-3 0 ; that of S100P shRNA was 5 0 -CTTCAGTGAGTTCATCGTGTT-3 0 ; and that of luciferase shRNA was 5 0 -GCGGTTGCCAAGAGGTTCCAT-3 0 . The lentiviral vector and its packaging vectors were transfected into 293T packaging cells by calcium phosphate transfection. Briefly, 10 6 293T cells were transfected with 10 mg shRNA expressing plasmid together with 10 mg of pCMVDR8.91 (the packaging vector) and 1 mg of pMD.G (the envelope vector). After 5 hours of incubation, transfection medium was replaced with fresh culture medium. Forty-eight hours later, lentivirus-containing medium was collected from transfection and spun down at 1500 rpm for 5 min to pelletize the cell debris, the supernatant was filtered through a 0.45-mm filter, and target cells were infected with fresh lentivirus-containing medium (supplemented with 8 mg/mL polybrene) for 24 hours.

Two-chamber migration and invasion assay
Invasion and migration assays were performed using Transwell inserts for a 24-well plate that contained 8-mm pores (Millipore). Filters coated with Matrigel (BD Labware) were used for the invasion assay. The numbers of cells that migrated and invaded were normalized to the growth rate for each cell line.

Wound-healing assay
The wound-healing assay was performed by creating an artificial "wound" using a 200-mL pipette tip on confluent cell monolayers in 6-well culture plates with serum-containing medium. Photographs were taken at the start and end of the experiment (18 hours for CL1-5 cells and 28 h for CL1-0 cells).

Immunofluorescence microscopy
Cells were grown on coverslips and fixed in 4% paraformaldehyde, permeabilized, and stained with Alexa Fluor 488 Phalloidin (Life Technologies). Slides were examined and photographed using a Zeiss Axiophot fluorescence microscope. Nuclei were counterstained with DAPI.
Mice and the experimental metastasis assay NOD.CB17-Prkdc SCID mice were maintained in a germ-free environment and had ad libitum access to food and water. An aliquot of a suspension (10 7 cells in 0.2 mL of PBS) was then subcutaneously injected into 6-week-old SCID mice. Mice were sacrificed after 6 weeks. Blood was taken to examine circulating tumor cells by seeding in 6-well plates and selected for G418resistant colonies. All animal work was performed according to protocols approved by the Institutional Animal Care and Use Committee of the College of Medicine, National Taiwan University.

Statistical analysis
Data are presented as the mean AE standard deviation (SD). Statistical analysis was performed using Statistical Package for Social Science software, version 16 (SPSS). Student t test was used to compare data between two groups. Statistical analyses of clinicopathological data were performed by x 2 and Fisher's exact test. Survival curves were obtained using the Kaplan-Meier method. All statistical tests included a two-way analysis of variance (ANOVA). P values of <0.05 were considered statistically significant.

Results
Expression of Keap1 was inversely correlated with the stage and lymph node metastasis and was positively correlated with the survival of lung cancer patients To elucidate the clinical relevance of Keap1 in cancer patients, we analyzed a cohort of 238 lung cancer specimens using an IHC analysis with a Keap1-specific antibody. Keap1 expression was detected in different stages of lung ADCs ( Fig. 1A and B) and normal lung epithelium (Fig. 1C, left). Keap1 was found to predominantly be expressed in the cytosol, and rarely in nuclei. Negative staining used an IgG antibody as a control as shown in right panel of Fig. 1C. Keap1 expression was inversely correlated with the clinical stage and lymph node metastasis (Supplementary Table S1). Notably, early stage (I and II) tumors without lymph node metastasis had higher levels of Keap1 expression compared with late-stage (III and IV) tumors ( Fig. 1A and B). Keap1 expression was significantly correlated with both the overall survival and disease-free survival of lung cancer patients (Fig.  1D). Taken together, the above clinical data indicate that downregulation of Keap1 is a critical event in tumor progression.

Keap1 expression suppressed the migratory/invasive abilities of lung cancer cells
Our clinical findings suggest that Keap1 may play a role in the invasiveness and metastatic ability of tumor cells. We next evaluated the function of Keap1 in cell migration and invasion, which are the most fundamental steps of tumor metastasis. To elucidate a link between Keap1 expression and tumor cell invasiveness, we used a set of lung ADC cell lines (CL1-0, CL1-3, and CL1-5) that were designed to exhibit progressive invasiveness abilities as previously described (22). A Western blot analysis showed that Keap1 protein levels were significantly  attenuated in the highly invasive CL1-5 and CL1-3 lung cancer cells compared with the poorly invasive parental CL1-0 cells ( Fig.  2A, left). In addition to these three cell lines, we further investigated the correlation between Keap1 expression and tumor cell invasiveness in other lung ADC cell lines (PC14, H441, H928, PC9, and A549). We observed that expression levels of Keap1 in these cell lines were also inversely correlated with their migratory/invasive abilities ( Fig. 2A, right). Next, to determine whether Keap1 modulates tumor cell migration and invasion, we overexpressed Keap1 in CL1-5 and A549 cells. The migratory/ invasive abilities of both Keap1 transiently transfected CL1-5 and A549 cells were significantly suppressed (Fig. 2B). Similar results were obtained from Keap1-stably transfected CL1-5 cells (Supplementary Fig. S1). In comparison, knockdown of Keap1 by two Keap1-specific shRNAs significantly increased the cell migratory/invasive abilities of poorly invasive CL1-0 cells (Fig. 2C,  bottom), and the knockdown efficiency of Keap1-specific shRNAs was detected by a Western blot analysis (Fig. 2C, top). In addition, manipulation of Keap1 expression (overexpression or knockdown) had no significant effect on the cell growth rate during 5 days of cell culture (Supplementary Figs. S2A and S2B). Taken together, these results indicate that Keap1 may play important roles in cell motility. In addition to lung cancer cells, a similar effect of Keap1 on regulating cell motility was observed in colon cancer (HCT116) and breast cancer (MCF-7) cells ( Supplementary Fig. S3). Under phase microscopy, we found that highly invasive CL1-5 cells overexpressing Keap1 turned into a round shape from a spindle-like shape (Fig. 2D, left), implying the Keap1 might affect the arrangement of the cytoskeleton. Factin is the main component of the cytoskeletal system and is continuously polymerized and depolymerized in motile cells and is essential to cell motility (24). To further assess whether Keap1 expression suppresses cell motility via arrangements of the actin cytoskeleton, CL1-5 cells transfected vector control or Keap1 and CL1-0 cells transfected luciferase shRNA or Keap1 shRNA were stained with an FITC-conjugated anti-phalloidin antibody and characterized using immunofluorescence microscopy. CL1-5 control cells displayed well-formed F-actin-containing microfilament bundles within the cytoplasm, whereas CL1-5/Keap1 cells contained few microfilament bundles (Fig. 2D, right). In contrast, CL1-0 control cells displayed few microfilament bundles, which were induced to increase by Keap1 shRNA (Fig. 2D, right). These findings indicated that Keap1 expression regulates rearrangements of F-actin-containing microfilament bundles, suggesting that F-actin-containing microfilament bundle rearrangements may be involved in Keap1-mediated cell invasiveness.

Keap1 regulates tumorigenecity and invasive ability in an animal model
We next examined the in vivo effects of Keap1 expression on tumor growth and metastasis. Control CL1-0 cells (CL1-0/shLuc) subcutaneously injected into NOD-SCID mice formed smaller tumors than those in mice injected with CL1-0/shKeap1 cells after 6 weeks (Fig. 3A). There were no significant differences in body weights between the CL1-0/shLuc-injected and CL1-0/shKeap1injected groups (Fig. 3B, left), and the ability of these tumors to retain depleted Keap1 was examined (Fig. 3B, right). Seven weeks after inoculation with CL1-0 cells, no metastases were macroscopically visible in the lungs of these mice; perhaps longer monitoring times would be needed to observe such secondary tumors. However, the mice had to be sacrificed after 7 weeks due to large subcutaneous tumors according to guidelines of the Animal Ethics Committee. To circumvent this problem, we collected peripheral blood at the endpoint of the study and cultured it in the presence of G418 for 2 weeks to select transfected cells. Blood from five control cell-injected mice formed no colonies compared with that from four of five CL1-0/shKeap1-injected mice which formed around 40 to 190 colonies (Fig. 3C). We also examined Keap1 expression in colonies derived from these four CL1-0/shKeap1-injected mice to see if the colonies retained the depleted Keap1. Compared with CL1-0/shLuc cells, all of the colonies expressed lower levels of Keap1 protein (Fig. 3D). Our data indicate that Keap1 expression suppresses the growth and extravasative abilities of cancer cells.

Nrf2 is a direct and functional target for Keap1-mediated migratory ability
Nrf2 is negatively regulated by Keap1 and controls expressions of several antioxidant genes that regulate cellular responses to oxidative stress (5). Recently, Nrf2 was also reported to regulate the growth and chemoresistance of lung cancer cells (4,25). To determine the role of Nrf2 in Keap1-mediated suppression of cell motility, we first examined the correlation between Keap1 and Nrf2 and confirmed the normal functioning of the Keap1-Nrf2 axis in our in vitro system. As shown in Fig. 4A, expression levels of Keap1 and Nref2 were inverted in CL1-0 and CL1-5 cells. Compared with control cells, up-and downregulation of Nrf2 were respectively observed in Keap1-depleted CL1-0 cells and Keap1overexpressed CL1-5 cells (Fig. 4A). After treatment of CL1-0 cells with H 2 O 2 for 24 hours, expressions of Keap1 and Nrf2 were respectively down-and upregulated in concentration-dependent manners ( Supplementary Fig. S4A). The Nrf2 downstream antioxidant genes, Gpx, HO-1, and MRP-3, were upregulated in Keap1-depleted CL1-0 and downregulated in Nrf2-depleted  Fig.  S4B). We next determined whether Nrf2 modulates tumor cell motility. From the two Nrf2-specific shRNAs in transiently transfected highly invasive CL1-5 and A549 cells, the migratory abilities of both cells were significantly suppressed (Fig. 4B, left and middle panels), but proliferation was not affected during the experimental period (Supplementary Fig. S5A). In comparison, overexpression of Nrf2 significantly increased the cell migratory ability of poorly invasive CL1-0 cells (Fig. 4B, right). The BTB domain is an important functional domain in Keap1 and is responsible for mediating Nrf2 degradation (26). To further determine the importance of the Keap1-Nrf2 axis in regulating cell motility, the Keap1 BTB-deleted mutant (Keap1-DBTB) was used. CL1-5 cells transfected with wild-type Keap1 resulted in lower Nrf2 protein expression (Fig. 4C, left), Nrf2 transcriptional activity (ARE-Nrf2 luciferase reporter assay; Fig. 4C, right), and cell migratory ability (Fig. 4C, left) compared with the controls, while in CL1-5 cells transfected with Keap1-DBTB, the suppressive effects described above were all significantly reversed (Fig. 4C). To further confirm whether Nrf2 can reverse Keap1-mediated phenotypes, transfection and expression of Nrf2-specific shRNA significantly reduced levels of the Nrf2 protein with a concomitant decrease in the migratory ability of CL1-0 cells that had previously been transfected with Keap1 shRNA compared with the control

Research.
on August 13, 2017. © 2015 American Association for Cancer clincancerres.aacrjournals.org Downloaded from (Fig. 4D). Taken together, these data indicate that the ability of Keap1 to inhibit cell motility is mainly attributable to its capacity to downregulate Nrf2.

Keap1 expression was inversely correlated with Nrf2 in patients with lung cancer
We next investigated whether Nrf2 expression was inversely correlated with Keap1 levels in human lung cancer patients. Representative examples with different Nrf2 scores are shown in Supplementary Fig. S6. IHC analysis of lung cancer specimens revealed an inverse correlation between Nrf2 and Keap1 expressions (tested by Kendall's t, correlation coefficient value of À0.110, P < 0.05; Supplementary Table S2) Representative IHC staining of Keap1 and Nrf2 on serial sections revealed inverse staining patterns in lung ADC tissues (Fig. 5A). Relationships between levels of Keap1 and Nrf2 expressions and the survival rate of patients with lung cancer are shown in Fig. 5B. Patients with tumors exhibiting a high expression level of Keap1 and a low expression level of Nrf2 had significantly higher overall survival time and disease-free survival time than patients with tumors exhibiting a low expression level of Keap1 and a high expression level of Nrf2 (Fig. 5B). The correlation of high Keap1 expression with low Nrf2 expression in human lung cancer patients is consistent with our finding that knockdown of Keap1 can upregulate Nrf2 in lung cancer cells.

S100P acts as a downstream effector in Keap1/Nrf2 signaling regulated cell mobility in lung cancer cells
To further investigate the downstream effector genes of the Keap1/Nrf2 axis that mediate cell motility, we respectively analyzed differentially expressed genes in CL1-5/Keap1 and CL1-5/ shNrf2 cells using Affymetrix HGU133Plus2.0 arrays. From the microarray RNA expression profiles, the tumor metastatic-related gene, S100P (18), was identified as one of the significantly decreased genes in both CL1-5/Keap1 and CL1-5/shNrf2 cells (Fig. 6A). An RT-PCR analysis was used to further validate expression levels of S100P in CL1-5/Keap1 and CL1-5/shNrf2 cells (Fig.  6B, left). Moreover, protein expression patterns of S100P in CL1-0 and CL1-5 cells were similar to that of Nrf2 and contrary to that of Keap1 (Fig. 6B, right). Next, we determined whether S100P affected tumor cell motility. We found that silencing S100P by shRNA delivery to CL1-5 cells drastically reduced the migratory ability of cells as evidenced in a Transwell migration assay (Fig.  6C, left) and wound-healing assay (Fig. 6C, right). The knockdown efficiencies of S100P-specific shRNA on S100P mRNA and protein expressions were respectively detected by an RT-PCR and Western blot analysis (Fig. 6C, left). To further search for functional linkages between S100P-and Keap/Nrf2-mediated cell motility, the level of S100P in CL1-0/shKeap cells was depleted by transfection with S100P-specific shRNA (Fig. 6D, left). As anticipated, transfection with S100P shRNA also effectively reduced the migratory ability of CL1-0/shKeap1 cells as evidence from Transwell migration assay (Fig. 6D, left) and wound-healing assay (Fig. 6D, right). Taken together, the results obtained here show that S100P is involved in lung cancer cell motility and acts as a downstream effector of Keap1/Nrf2 signaling.

Discussion
Tumor metastasis is the main cause of cancer-related deaths. Identifying genes that regulate cell migration and metastasis is crucial to understanding this process. In lung cancer, Nrf2 activation in malignant cells is associated with tumor progression and chemoresistance (4,27,28). High levels of nuclear Nrf2 facilitate cancer cell growth and cell survival as a result of the transactivation of cytoprotective genes (4,27,28). In NSCLC, overexpression of nuclear Nrf2 is principally attributable to genetic and epigenetic alterations and the loss of function of its repressor, Keap1 (6,29). Despite these recent findings, characteristics of NSCLC tumors with Nrf2 activation and loss of Keap1 expression and the role of the Keap1-Nrf2 axis in regulating cell motility are not fully understood. The results presented in our investigation indicate that suppression of Nrf2 via shRNA or Keap1 overexpression in CL1-5 and A549 cells were accompanied by acquisition of cell actin cytoskeletal changes and suppression of S100P-mediated motility.
Although knockdown of Nrf2 or overexpression of Keap1 in CL1-5 and A549 cells did not affect cell growth in a short culture period, we found that the migratory ability of these cells could significantly be suppressed under the same conditions. To the present, the role of the Keap1-Nrf2 axis in regulating tumor cell motility is still controversial. For example, a similar result was found by Pan and colleagues who also indicated that knockdown of Nrf2 in glioma cells attenuated the migratory and invasive abilities of cells (30). In addition, we further investigated the role of Keap1 in other cancer types and found that the overexpression of Keap1 in MCF-7 breast cancer cells and HCT116 colon cancer cells also significantly suppressed cell motility. In contrast, Rachakonda and colleagues indicated that suppression of Nrf2 via overexpression of Keap1 in HepG2 liver cancer cells or direct knockdown of Nrf2 in A549 or SW480 colon cancer cells was accompanied by acquisition of enhanced cell motility (31). Discrepancies in cell motility observed by Rachakonda and colleagues and in this work are not currently understood. Rachakonda and colleagues's study only utilized cell culture models, but in our study, we not only demonstrated the role of the Keap1-Nrf2 axis in a cell culture system, but also found that knockdown of Keap1 promoted the metastatic ability in a CL1-0 xenograft model. Otherwise, we found that the lymph node status of Taiwanese lung tumor specimens was positively correlated with the Nrf2 expression level (Supplementary Table S3) and was inversely correlated with the Keap1 expression level (Supplementary Table  S1). Collectively, our results based on multiple lines of evidence support Keap1 being a suppressor of tumor metastasis, and Nrf2 being one of its important targets in regulating this phenomenon.
(Continued.) The GAPDH control demonstrated that equal amounts of RNA were used in the assays performed. B (right), Western blot analysis of endogenous Keap1, Nrf2, and S100P protein levels in CL1-0 and CL1-5 cells. C (left), S100P mRNA and protein levels were respectively detected by RT-PCR and Western blot analyses in CL1-5/scrambled shRNA and CL1-5/sh100P cells (left). Migratory abilities of CL1-5/scrambled shRNA and CL1-5/sh100P cells were evaluated by Transwell migration assays (right) and wound-healing assays (C, right). Multiples of differences are presented as the mean of triplicate experiments compared with control cells. Ã , P < 0.05, compared with control cells. D, migration of CL1-0 cells infected with a lentivirus carrying scrambled shRNA or shRNAs for either Keap1 or Keap1þS100P. Migratory abilities of cells carrying scrambled shRNA, shKeap1, or shKeap1þshS100P were evaluated by Transwell migration assays (D, left) and wound-healing assays (D, right). The knockdown efficiency was shown by a Western blot analysis using specific antibodies (D, left). Ã , P < 0.05, compared with control cells. #, P < 0.05, compared with Keap1 shRNA-infected CL1-0 cells.
In lung cancer, we first discovered that S100P is an important downstream target of the Keap1-Nrf2 axis in regulating cell motility. S100P expression was described in many different cancers, including pancreas, breast, colon, prostate, and lung cancers, and its expression was associated with drug resistance, metastasis, and poor clinical outcomes (17,18,20,(32)(33)(34). In the extracellular space, S100P interacts with the receptor for advanced-glycation end products (RAGE) to promote tumor development (33). Intracellular S100P interacts with the cytoskeletal multidomain protein, ezrin, which is a membrane-cytoskeleton organizer that mediates the rearrangement and function of F-actin and further regulates cancer cell motility (35,36). From these findings, S100P is recognized as a promoter of tumor cell motility. Herein, we found that the expression phenotype of S100P in NSCLC cells was similar to Nrf2, but inversely correlated with Keap1. Both the overexpression of Keap1 and knockdown of Nrf2 attenuated the expression of S100P in CL1-5 cells, whereas downregulation of the expression of Keap1 increased the expression of S100P in CL1-0 cells. In addition, knockdown of S100P directly suppressed the migratory ability of CL1-5 cells and also reversed the Keap1-mediated increase in the migratory ability of CL1-0 cells. The results suggest that S100P is essential for regulating cell migration by the Keap1-Nrf2 axis. We further asked how the Keap1/Nrf2/S100P signal pathway mediates lung cancer cell motility. In our study, fewer F-actin-containing microfilament bundles and lower ezrin expression were observed in CL1-5/ Keap1 cells compared with control cells. We hypothesized that the Keap1-Nrf2 axis-mediated cell motility can be attributed to regulation of the interaction between S100P and ezrin, and this issue should be further addressed in the future. S100P gene expression is controlled by a promoter region containing multiple regulatory elements, and the luciferase activity produced by different promoter constructs shows that SMAD, STAT, CREB, and SP/KLF binding sites are critical cis-elements that are required for S100P expression in cancer cells (37). Although ARE sites were not found in the S100P promoter, the CREBbinding protein (CBP) was shown to interact with Nrf2 and enhance Nrf2-dependent reporter gene activities (38). CBP can also be recruited by CREB to regulate multiple genes such as S100P in regulating cancer pathogenesis (39,40). It was therefore speculated that Nrf2 may interact with the CREB/CBP complex and thus regulate S100P expression, but this hypothesis needs to be further investigated. Moreover, Nrf2 was also reported to induce upregulation of the STAT3 activator, interleukin (IL)-6 (41), and at this moment, we cannot rule out the possibility that Nrf2 may regulate S100P expression through the IL6/STAT pathway.
Although we have demonstrated that Keap1 possesses the anti-extravasative property on cancer cells in a subcutaneously growing CL1-0 tumor model and further elucidated its downstream effectors, Nrf2 and S100P, play critical roles in regulating cell motility of cancer cells. At present, several reports have indicated that the subcutaneous microenvironment for human visceral tumors is different from their original milieu and this difference may alter the metastastic ability or drug response of cancer cells (42)(43)(44). In contrast to subcutaneous injection, orthotopic metastatic mouse models such as injecting tumor cell suspensions into the orthotopic mouse sites or surgical orthotopic implantation (SOI) of histologically intact cancer fragments in the mouse have been recognized as the better way to promote the cancer metastatic rates and sites in the trans-planted mice reflect the clinical pattern (45,46). According to these reports, an orthotopic xenograft model or SOI model may be a better in vivo model for evaluating the effects of Keap1, Nrf2 and S100P on lung tumor metastasis. In this study, we tried to orthotopically inject mice (n ¼ 4) with CL1-0 cells expressing scramble shRNA, shKeap1, and combinations of shKeap1 and shS100P, respectively. Keap1 knockdown showed increasing trend of luciferase counts on orthotopic tumors and metastastic sites. S100P knockdown, however, showed the trend to abate lung metastasis in Keap1 shRNA-expressing CL1-0 cells in orthotopic left to right lung metastasis assay ( Supplementary Fig. S7). More animal numbers are worth to be done for this orthotopic xenograft model in our future work.
In our study, we also observed significant morphological and cytoskeletal changes in Keap1-manipulated lung cancer cells (Fig.  2D). The assembly and disassembly of the actin cytoskeleton are tightly controlled by three Ras-related GTP-binding proteins, Rho, Rac, and Cdc42 (47). Herein, we also found that overexpression of Keap1 in CL1-5 cells suppressed the expression and activity of RhoA, but not Rac or Cdc42 (Supplementary Fig. S8A). We further found that Keap1 did not affect mRNA levels of RhoA (Supplementary Fig. S8B), but promoted RhoA degradation (Supplementary Fig. S8C). Although further study is required, these results suggest that Keap1 might mediate the degradation of Nrf2 and also RhoA to regulate cell motility in lung cancer cells. In addition, we also evaluated if Keap1 can interact with RhoA directly using the coimmunoprecipitation assay. We found that the Keap1 protein was not able to bind RhoA protein directly (Supplementary Fig. S8D) and the underlying mechanisms involved in the Keap1-mediated RhoA degradation should be further investigated in our future work.
In conclusion, we have shown that Keap1 is endowed with ubiquitin ligase activity to concomitantly degrade Nrf2 and repress the downstream effector, S100P, thereby changing the cytoskeleton and suppressing cell motility. In addition to Nrf2, we first identified RhoA as another downstream target that may be involved in Keap1-mediated cell motility. Moreover, we also showed that Keap1 is a good prognostic factor the overexpression of which is associated with increased survival and decreased lymph node metastasis. Our findings promote a better understanding of the mechanisms of metastasis and may lead to the development of effective therapies of lung cancer.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.