JNK signaling mediates EPHA2-dependent tumor cell proliferation, motility, and cancer stem cell-like properties in non-small cell lung cancer

Cell lines, plasmids, and reagents

Human NSCLC cell lines were provided by the Specialized Programs of Research Excellence (SPORE) in lung cancer at Vanderbilt-Ingram Cancer Center. The cell lines were validated with DNA finger printing through the Vanderbilt Technology for Advanced Genomics Core facility in the fall of 2013. They were cultured in RPMI-1640 medium with 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin. Human EPHA2 cDNA was obtained from Open Biosystems (Huntsville, AL) and subcloned into pCLXSN retroviral vector containing Neomycin gene for G418 selection. Human c-JUN cDNA and constitutively activated JNK1 and JNK2 were obtained from Addgene (Cambridge, MA) and subcloned into pCLXSN retroviral vector. Hairpin shRNAs targeting human EPHA2 were purchased from Open Biosystems. JNK inhibitor SP600125 was purchased from Cell Signaling (Denvers, MA). Human Phospho-kinase antibody array and Lung cancer tissue microarray were purchased from R&D System (Minneapolis, MN) and US Biomax (Rockville, MD), respectively.

Lentiviral shRNA knockdown and retroviral overexpression experiments

shRNA construct in the pLKO.1 lentiviral vector containing the following EPHA2 targeting sequence was used: 5′-CGGACAGACATATGGGATATT-3′. Vector control (pLKO.1) or EPHA2 shRNA lentiviral particles were produced by co-transfection of HEK 293T cells with targeting plasmids and packaging vectors, psPAX2 and pMD.2G, using lipofectamine 2000 (Invitrogen, Life Technologies). Viral supernatants were collected by centrifugation and were used to infect NSCLC cells for 24 hours. Cells were changed to new growth medium for another 24 hours, followed by puromycin selection (2 μg/ml) (Sigma-Aldrich, ST. Louis, MO) for 3–5 days.

Retroviruses carrying vector (pCLXSN), pCLXSN-EPHA2, pCLXSN-JNK-CA, or pCLXSN-c-JUN were produced by co-transfection of HEK293T cells with overexpression plasmids and packaging vector, pCLAmpho. Viral supernatants were used to infect NSCLC cells, followed by selection of 800 μg/ml G418 (Sigma-Aldrich) for 10 days.

Cell growth Assays

Cell growth was measured by MTT, colony formation, and BrdU assays. For MTT assay, 2.5×10³ cells were plated into each well of 96-well plate in 100μl of complete growth medium. JNK inhibitor was added on the second day after cell attachment. Cell viability was measured by incubating cells with 20μl of 5 μg/ml Tetrazolium salt 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT, Sigma-Aldrich) and quantified by reading absorbance at 590 nm using Microplate reader (Bio-Tek, Winooski, VT). For colony formation assay, 200 or 400 cells in complete growth medium were plated into each well of a 12-well plate. Cells were growing for 10–14 days, and the medium was changed every three days. At the end of the experiment, cell colonies were stained with crystal violet (Sigma-Aldrich) and the foci were photographed. For BrdU incorporation assay, 2×10⁴ cells/well in complete growth medium were plated onto matrigel coated 2-well LabTekII chamber slide. Cells were starved for 20 hours, followed by 10 μg/ml BrdU labeling in the presence of 0.5% FBS for 16 hours. BrdU detection was performed using BrdU staining kit (Invitrogen, Life Technologies). BrdU positive cells were enumerated in four random fields, at 40× magnification, per chamber and proliferation index was calculated as the percentage of BrdU+ nuclei/total nuclei.

Apoptosis assay

Tumor cells were serum starved for 48 hours and apoptosis measured by Annexin V-FLUOS Staining Kit (Roche) per manufacturer’s instruction. Briefly, cells were gently trypsinized and washed once with serum containing medium. Cell suspensions were incubated with Annexin-V-Fluorescein and Propidium idodide to detect phosphoserine on the outer leaflet of apoptotic cell membranes and to differentiate from necrotic cells, respectively. Annexin-V Fluorescein labeled cells were detected by FACS analysis. For tumor xenografts, apoptosis was measured by TUNEL assay on tumor sections, as described previously (12).

Transwell migration assay

Tumor cell migration was assessed by a modified Boyden chamber assay using 8mm pore size Transwell (Costar). The transwell inserts were pre-treated with 1% BSA in Opti-MEMI medium for 30 minutes. Tumor cells (1×10⁴ of H1975, 2 ×10⁴ of A549, or 3 ×10⁴ of H1650 for 24hr migration, and 2×10⁴ of H1975, 4 ×10⁴ of A549 for 6hr migration) were plated on top of the transwell. Growth medium was added to the lower chamber and cells were allowed to migrate for 6 or 24 hours. Cells that passed through the transwell membrane and stayed on the bottom of membrane were fixed with 4% paraformaldehyde and stained with crystal violet, photographed and counted.

Screen of phospho-kinase array

Human phospho-kinase antibody arrays were purchased from R&D system (Catalog # ARY003). Capture antibodies were pre-spotted in duplicate on nitrocellulose membranes by the manufacturer. Membranes were incubated with blocking buffer for 1 hour at room temperature. Cell lysates derived from control and EPHA2 knockdown H1975 cells (500 μg protein) were mixed with a cocktail of biotinylated detection antibodies. The sample/antibody mixture was then incubated with the array overnight at 4°C, followed by three washes in washing buffer. Streptavidin-Horseradish Peroxidase and chemiluminescent detection reagents were subsequently added and chemiluminescence was detected in the same manner as a Western blot. Densitometry was performed on scanned blots using NIH Image J software version 1.45 to quantify the average pixel density/spotted area after normalization to the pixel density of the background controls. Data are representation of two independent experiments.

Tumor Sphere assay

Cells (0.5×10⁴/well for A549, 1×10⁴/well for H1975) were plated into 6-well plates with ultra-low attachment surface in serum-free medium DMEM/F12 (Gibco, Life Technologies) supplemented with commercial hormone mix B27 (Invitrogen, Life Technologies), EGF (20 ng/ml, R&D Systems, Minneapolis, MN), bFGF (20 ng/ml, R&D Systems), and heparin (2 μg/ml, Stemcell Technologies, Vancouver, BC, Canada). Floating sphere cultures were kept for 7–9 days. EGF and bFGF were replenished every 3 days. Plates were scanned and numbers of tumor sphere per well were enumerated automatically using The GelCountTM system (Oxford Optronix, United Kingdom).

Antibodies, western blot analysis and immunofluorescence

Antibodies against the following proteins were used: EPHA2 (1:1000, D7, Millipore, Temecula, CA); EPHA2 (1:1000, c-20), ACTIN (1:1000, I-19) (Santa Cruz Biotechnology, Santa Cruz, CA); pThr183/Tyr185 SAPK/JNK, SAPK/JNK, pSer63-c-JUN, c-JUN, pSer473 AKT, AKT, pThr202/Tyr204 ERK, ERK, pSer235/236 pS6, S6, NOTCH3, SOX2 (1:1000, Cell Signaling Technology, Danvers, MA); ALDH (1:5000, BD Transduction Laboratories, San Diego, CA). Horseradish peroxidase (HRP)-conjugated secondary antibodies were purchased from Promega (Madison, WI). Western blot analyses were performed by standard methods using indicated antibodies. Densitometry was performed on scanned blots using NIH Image J software version 1.45 to quantify the average pixel density/band area after normalization to the pixel density of the corresponding loading controls. Data are representation of 3–4 independent experiments.

For immunofluorescence studies, 2×10⁴ cells were plated onto matrigel coated 2-well LabTekII chamber slide. Cells were fixed next day with 4% paraformaldehyde, permeabilized, and blocked with 0.3% Triton X-100, 3% BSA in PBS, followed by incubation with anti-p-c-JUN (1:100, Cell Signaling Technology) overnight at 4°C. The primary antibody was detected by incubation with the Alexa-conjugated secondary antibody (1:500, Invitrogen, Life Technologies) for 1 hour at room temperature. The slides were mounted with ProLong Antifade reagent with Dapi (Invitrogen, Life Technologies) and photographed.

Immunohistochemistry

Lung cancer tissue microarray (TMA) (US Biomax, Rockville, MD) or xenograft tumor sections from A549 and H1975 cells were stained by immunohistochemistry as described previously (13). Briefly, rehydrated paraffin sections were subjected to antigen retrieval in citrated buffer (2 mM citric acid, 10 mM sodium citrated, pH 6.0) using the PickCell Laboratories 2100 Retriever. Endogenous peroxidases were blocked by 3% H₂O₂ for 30 minutes. Sections were incubated with blocking solution and stained with primary antibodies (anti-EPHA2 1:25, anti-p-c-JUN 1:100, anti-ALDH 1:100) overnight at 4°C. Sections were subsequently washed and stained with biotinylated secondary antibodies, followed by avidin-peroxidase reagent and DAB substrate. After hematoxylin counterstain, sections were mounted and photographed on an Olympus BX60 microscope using a digital camera and NIH Scion Image software.

For quantification of lung cancer TMAs, relative expression was scored using a continuous scale as follows: 0 = 0–10% positive tumor epithelium, 1 = 10–25% positive tumor epithelium, 2 = 25–50% positive tumor epithelium, and 3 = >50% positive tumor epithelium/core. TMA cores were scored blind by three independent individuals, the average of which was reported here. Differential expression between tissue samples and correlation between EPHA2 and ALDH expression were quantified and statistically analyzed using the Chi square analysis.

In vivo tumor studies

8-week-old nude mice were purchased from Harlan-Sprague-Dawley and used for vivo tumor evaluation. The animals were housed under pathogen-free conditions, and experiments were performed in accordance with the Association and Accreditation of Laboratory Animal Care International guidelines and with approval by the Vanderbilt University Institutional Animal Care and Use Committee. To evaluate the role of EPHA2 on tumor growth, 2.5×10⁶ EPHA2 knockdown cells were injected subcutaneously into the flank of the nude mice, and the same numbers of vector control cells were injected contralaterally. Tumor progression was monitored by palpation and tumor size measured by a digital caliper. At the end of the 4 weeks, tumor volume was calculated using the following formula: volume=length×width²×0.52.

Flow Cytometry Analysis

The cell profile of ALDH was determined by flow cytometry analysis using the Aldefluor assay kit (Stem Cell Technologies) per manufacturer’s instruction. Briefly, cells were incubated in Aldefluor assay buffer containing efflux inhibitor and the ALDH substrate. Cells that could catalyze substrate to its fluorescent product were considered ALDH+. Sorting gates for FACS were drawn relative to cell baseline fluorescence, which was determined by the addition of the ALDH-specific inhibitor DEAB during the incubation and DEAB-treated samples as negative controls. Nonviable cells were identified by 7-aminoactinomycin D inclusion. Flow cytometry analyses were performed on FACScan or FACSCalibur flow cytometer (BD Biosciences, Oxford, UK) and the data were analyzed using FlowJo software (Tree Star, Ashland, OR).

Statistical analysis

For comparisons between two groups, Student t test was used, and for analysis with multiple comparisons, analyses of variance (ANOVA) or Kruskal-Wallis tests were used, using PRISM software (GraphPad Software, La Jolla, CA). Chi-square analysis was used for categorical outcomes. All tests of statistical significance were two-sided, and the exact tests used for each experiment were listed in the text and in the figure legend. P values less than 0.05 were considered to be statistically significant.

Phosphorylation of JNK and c-JUN is regulated by EPHA2 in lung cancer cells

Previous studies showed that knockdown of EPHA2 receptor in cultured lung cancer cells inhibited proliferation (11). Conversely, expression of EPHA2 in immortalized BEAS-2B bronchial epithelial cells increased cell survival and invasion (10). Consistent with these observations, shRNA-mediated stable knockdown of EPHA2 in non-small cell lung cancer (NSCLC) lines inhibited tumor cell proliferation and motility (Figure 1 and Supplementary Figure 1). To identify signaling pathways downstream of EPHA2 receptor that mediate tumor-promoting effects in cancer cells, cell lysates derived from vector-control or EPHA2 shRNA knockdown H1975 cells were used to screen two human phospho-kinase antibody arrays (human phospho-kinase antibody array and human phospho-RTK antibody array). As shown in Figure 2A, phosphorylation at Ser63 residue of c-JUN is decreased in EPHA2 knockdown cells, relative to vector control cells, in two independent experiments, suggesting that EPHA2 may regulate the JNK/c-JUN pathway in lung cancer cells.

To determine if depletion of EphA2 affects JNK/c-JUN signaling in response to serum stimulation, H1975 or A549 cells were serum starved overnight and stimulated with 10% FBS following a time course, and phosphorylated levels of JNK and c-JUN were determined by western blot analysis. We found that p-JNK/JNK and p-c-JUN/c-JUN levels were consistently decreased in EPHA2 knockdown cells in four independent experiments (Figure 2B). Interestingly, we also found that total c-JUN levels appear to be decreased in EPHA2 deficient cells. Next, we sought to investigate whether phosphorylation of c-JUN and JNK are altered in other EPHA2-depleted NSCLC cells. Consistent with observations in H1975 and A549 cells, knockdown of EphA2 resulted in decrease of p-c-JUN and pJNK levels in 4 and 5 out of 10 NSCLC lines tested, respectively (Supplementary Figure 2C).

Once phosphorylated, c-JUN enters the nucleus and regulates transcription (14, 15). Accordingly, we assessed nuclear c-JUN expression by immunofluorescence. The number of cells expressing nuclear c-JUN is markedly decreased in EPHA2 knockdown cells relative to vector control cells (Figure 2C), indicating that EPHA2 regulates c-JUN activity. Collectively, pJNK and p-c-JUN levels were reproducibly and significantly decreased in approximately 50% of EPHA2 knockdown cell lines tested, whereas phosphorylation levels of AKT and ERK were not changed in most cell lines (Supplementary Figure 2).

JNK signaling mediates EPHA2-dependent tumor cell proliferation and motility

JNK was originally shown to mediate cell apoptosis in response to a variety of stress signals (14, 15). However, emerging evidence suggests that JNKs play a critical role in cell growth and survival in tumor cells (14–17). Indeed, JNK is activated in NSCLC and promotes neoplastic transformation in human bronchial epithelial cells (18). JNK2-α appears to be the major JNK isoform that plays a critical role in lung cancer (19). Therefore, the JNK/c-JUN pathway may, at least in part, mediate EPHA2-dependent regulation of tumor cell behavior. To investigate whether JNK/c-JUN pathway is necessary and sufficient for cell proliferation and motility, we used two complementary approaches: a JNK kinase inhibitor, SP600125, and two constitutively activated JNK mutants (JNK1-CA and JNK2-CA).

Treatment of either A549 or H1975 cells with JNK inhibitor SP600125 inhibited p-JNK and p–c-JUN levels (Supplementary Figure 3A) and reduced cell viability, colony formation and tumor cell motility in a dose-dependent manner (Figure 3A & 3B, Supplementary Figure 3B), suggesting that the JNK/c-JUN pathway is required in lung cancer cells. To investigate whether the JNK/c-JUN pathway mediates EPHA2 receptor signaling, we utilized previously characterized constitutively activated MKK7-JNK1 and MKK7-JNK2 fusion mutants [here after termed JNK-CA, (20)]. Re-expression of wild-type EPHA2 receptor in EPHA2-knockdown cells rescued p-JNK and p-c-JUN levels, colony formation, and cell motility (Figure 3C–E), demonstrating that phenotypes in EPHA2 knockdown cells are not due to off-target effects. Expression of JNK1-CA, JNK2-CA, or c-JUN rescued phenotypes in the EPHA2 knockdown cells to the comparable level as wild-type EPHA2 (Figure 3C–E), indicating that activated JNK1/2 or overexpression of c-JUN is sufficient to restore function in these cells. Collectively, these data suggest that JNK/c-JUN pathway mediates EPHA2-dependent tumor cell proliferation and motility.

Loss of EPHA2 inhibits JNK/c-JUN signaling and suppresses tumor formation and tumor growth in vivo

To directly determine the in vivo role of EPHA2 in lung cancer, EPHA2 knockdown A549 or H1975 cells were injected subcutaneously into the dorsal flank of a nude mouse, and control cells carrying the empty vector were injected contralaterally in the same mouse. Loss of EPHA2 in knockdown tumors significantly inhibited tumor volume in the xenograft animal models (mean A549 tumor volume: vector control 419 ± 236 mm³ versus EPHA2 shRNA 225 ± 90 mm³; mean H1975 tumor volume: vector control 623 ± 310 mm³ versus EPHA2 shRNA 324 ± 166 mm³, p<0.05, paired t test) (Figure 4A). Tumors were harvested at the end of the studies and tumor sections were subjected to Ki67 immunohistochemistry and TUNEL assay for assessing proliferation and apoptosis, respectively. Consistent with in vitro data, we found that tumor cell proliferation was decreased in EPHA2 knockdown tumors as compared to vector control tumors, whereas apoptosis levels were not significantly changed between the two groups (Figure 4B & 4C). Analysis of tumor lysates revealed that EPHA2 was stably knocked down in tumors after grown in vivo for 30 days. Both phosphorylation and total levels of c-JUN were decreased in EphA2-deficient H1975 tumors, whereas there was a predominant reduction in total JNK and c-JUN levels in both EPHA2-deficient H1975 and A549 tumors (Figure 4D). Expression of EPHA2 and p-c-JUN was further analyzed in adjacent tumor sections by immunohistochemistry. As shown in Figure 4E, EPHA2 receptor and p-c-JUN were co-expressed in the control tumors, whereas nuclear staining of p-c-JUN is markedly decreased in EPHA2 knockdown tumors. Taken together, these results show that EPHA2 promotes tumor growth of NSCLC, such that depletion of EPHA2 limited progression of this aggressive NSCLC xenografts model.

To further investigate whether loss of EPHA2 affects lung cancer formation in vivo, we knocked down EPHA2 in A549 and H1975 cells that express luciferase. EPHA2 depletion in these knockdown cells was confirmed by FACS and western blot analysis (Supplementary Figure 4). Serial diluted EPHA2 knockdown tumor cells (2.5×10⁵, 2.5×10⁴, 2.5×10³) were injected subcutaneously into nude mice, and same numbers of control cells carrying the empty vector were injected contralaterally in the same mouse. Tumor formation was monitored by bioluminescence weekly. Vector control A549 and H1975 cells exhibited a greater frequency of tumor formation, particularly at low numbers of injected cells, compared with EPHA2 knockdown counterparts (Supplementary Figure 4). Collectively, our tumor xenograft data suggest that EPHA2 receptor regulates both tumor formation and tumor growth in vivo.

EPHA2 silencing suppresses ALDH positive stem-like cell-enriched populations, tumor sphere formation, and tumorigenicity in vivo

Recent evidence suggests that human lung cancers harbor cancer stem-like cells (CSCs) or tumor initiating/ propagating cells (TICs/TPCs) that facilitate tumor cell heterogeneity, metastases, and therapeutic resistance (21, 22). Because CSCs/TICs/TPCs are intimately linked with tumor initiation and knockdown of EPHA2 affects tumor formation in vivo, we tested if EPHA2-deficiency affects CSC/TIC/TPC populations in lung cancer cells.

ALDH has recently been identified as a marker for lung cancer cells with stem cell-like properties, and ALDH+ cells were shown to be highly tumorigenic compared to ALDH-counterparts (23, 24). Accordingly we used the Aldefluor assay to quantify the percentage of ALDH+ cells in vector control and EPHA2 knockdown lung cancer cell lines. Cells were incubated in Aldefluor assay buffer containing efflux inhibitor and the ALDH substrate. Cells that could catalyze substrate to its fluorescent product were considered ALDH positive. Sorting gates for FACS were drawn relative to cell baseline fluorescence, which was determined by the addition of the ALDH-specific inhibitor DEAB during the incubation and DEAB-treated samples serve as negative controls. As shown in Figure 5A, a subpopulation of ALDH+ cells was detected by flow cytometry in control A549 and H1975 cells, and the ALDH+ subpopulations significantly decreased in corresponding EPHA2 knockdown cells. Depletion of EphA2 also reduced ALDH+ populations in H23 and H1650 cells (Supplementary Figure 5).

To relate the findings to human lung cancer, we analyzed the expression of EPHA2 and ALDH on adjacent sections of human lung cancer tissue microarrays. Consistent with previous studies, we observed significantly elevated EPHA2 protein expression in human lung tumor samples relative to normal, adjacent control lung tissue (5/8 EPHA2+ normal lung tissue samples versus 25/30 EPHA2+ lung cancer tissue samples; p<0.05, Chi Square analysis). Furthermore, we observed a positive correlation between EPHA2 expression and ALDH expression in a significant number of human lung cancer samples (21/30, 70%) relative to samples that were EPHA2 positive but ALDH negative (4/30, 13%) (p<0.05, Chi Square analysis)(Figure 5B). Together, these data demonstrate the relevance of our studies in human cancer.

In addition to expressing stem cell markers, CSCs/TICs/TPCs are capable of propagation of isolated tumor cells as spheroids in suspension, and exhibit a high capacity for tumorigenic growth in mouse xenografts. We found that knockdown of EPHA2 significantly suppressed sphere-forming ability and in vitro self-renewal capacity of A549 and H1975 cells (Figure 5C). To test the role of EPHA2 in tumorigenic capacity of cancer stem-like cells, ALDH positive populations were sorted from A549 cells by FACS (Supplementary Figure 6) and subsequently transduced with either control virus or lentiviral shRNA against EPHA2 (Figure 6A). Serial dilutions of control or EPHA2 knockdown ALDH-positive cells (10⁴, 10³, 10² cells) were then injected into nude mice and tumor development was monitored weekly by bioluminescence imaging (Figure 6B). Compared to unsorted populations, control ALDH+ cells develop tumors much sooner (7 weeks in unsorted cells versus 3 weeks in ALDH+ cells), suggesting that CSCs/TICs/ TPCs are enriched in ALDH+ populations. Furthermore, while 100% of mice injected with 1000 ALDH+ vector control cells formed tumors at week 4, none of the mice injected with 1000 ALDH+ EPHA2 knockdown cells developed tumor. When 100 ALDH+ cells were injected, 80% of mice carrying vector control cells formed tumors at week 4, compared to 0% of mice carrying EPHA2 knockdown cells (Table 1). Collectively, our data on stem cell marker analysis, spheroid formation assay, and in vivo tumorigenicity suggest a role for EPHA2 in regulating cancer stem-like populations.