ETS2 Mediated Tumor Suppressive Function and MET Oncogene Inhibition in Human Non–Small Cell Lung Cancer

This article is featured in Highlights of This Issue, p. 3329

Lung cancer remains the leading cause of cancer-related deaths in the United States and worldwide (1). Non–small cell lung cancer (NSCLC) represents the majority of diagnosed lung cancer cases and is associated with a relatively poor 15% overall 5-year survival rate (2). Understanding the molecular profiles of NSCLC as well as elucidating the roles of oncogenes and tumor suppressors in the development of this malignancy is expected to identify aberrant signaling pathways and molecular targets for therapy.

V-ets erythroblastosis virus E26 oncogene homolog 2 (ETS2) belongs to the ETS family of transcription factors and controls gene expression by binding to numerous genes with GGA(A/T) ETS response elements (ERE) thus impacting a broad spectrum of cellular functions including proliferation, differentiation, migration, transformation, and apoptosis (3–5). During embryonic development, ETS2 was shown to insure proper development of the trophoblast (6) and to regulate endothelial cell survival during embryonic angiogenesis (7). Moreover, ETS2 was reported to be activated in response to extracellular mitogenic signaling mediated by the Ras oncogene (8) in murine fibroblasts, to maintain telomerase gene expression in breast cancer cells (9) and, notably, to exhibit both tumor-promoting and -suppressive effects in different types of carcinomas (10–15). However, ETS2 expression and function in human lung cancer is still unknown.

Our current lung adenocarcinoma expression profiling efforts revealed marked downregulation of ETS2 transcript expression in lung tumors compared with paired normal lung tissues prompting us to examine ETS2 function in lung cancer pathogenesis. In this study, we found that low ETS2 immunohistochemical protein expression was significantly associated with shorter time to recurrence in NSCLC. Moreover, we found that ETS2 negatively regulates cellular growth, migration and invasion, MET oncogene phosphorylation and activation, as well as HGF-mediated signaling. Our findings reveal for the first time a potential tumor-suppressive role for ETS2 in lung cancer pathogenesis that is, in part, mediated by the inhibition of MET oncogenic signaling.

All human tissues were obtained from the MD Anderson Cancer Center Lung Cancer Specialized Program of Research Excellence (SPORE) tissue bank (Houston, TX) and had been classified using the 2004 World Health Organization classification system as described before (16). All specimens were obtained from patients, who underwent surgery at the same institution from 2003 to 2005, under a protocol that was approved by the MD Anderson Cancer Center Institutional review board. Detailed clinical and pathologic information was available for most of these cases and included patients' demographic data, smoking history (never smokers or ever smokers, patients who had smoked at least 100 cigarettes in their lifetime), and pathologic tumor–node–metastasis (TNM) staging.

A collection of 80 lung adenocarcinomas and 30 nontumoral paired tissues were snap-frozen and preserved in liquid nitrogen for total RNA extraction and microarray profiling. For each tissue sample, the percentage of malignant tissue was calculated and the cellular composition of specimens was determined by histologic examination (J. Fujimoto) following hematoxylin–eosin (H&E) staining. All malignant samples retained contained more than 40% tumor cells.

For NSCLC tissue microarray analysis (TMA), we obtained archived formalin-fixed paraffin-embedded (FFPE) samples from surgically resected lung cancer specimens from the lung cancer tissue bank. The tissue microarray analyzed in this study was composed of 201 NSCLC tumor specimens [135 lung adenocarcinomas and 66 squamous cell adenocarcinomas (SCC); Supplementary Table S1]. After histologic examination of NSCLC specimens, the NSCLC TMAs were constructed by obtaining three 1 mm diameter cores from each tumor at 3 different sites (periphery, intermediate, and central tumor sites). The TMAs were prepared with a manual tissue arrayer (Advanced Tissue Arrayer ATA100, Chemicon International) as described previously (16).

Lung cancer cell lines were either originally purchased from the American Type Culture Collection (ATCC; H441, H2291, H3255, H1299, H1693, H522, H1792, H23, and H2009) or were obtained from Dr. Adi Gazdar (University of Texas Southwestern, Dallas, TX) (DFC1024, HCC4006, H2228, H1650, and H1944) and were grown in Dulbecco's modified Eagle medium-F12 low glucose medium supplemented with 10% FBS and maintained in humidified 5% CO₂ incubator. Information on the smoking status of patients from whom the cell lines were isolated was obtained from the ATCC or from Dr. Adi Gazdar. All cell lines used in the study were authenticated by short tandem repeat DNA fingerprinting using the PowerPlex 16 HS System (Promega). For MET inhibition studies, lung cancer cells were treated with the MET tyrosine kinase inhibitor, PHA-665752 (SelleckBio) at a 100 nmol/L final concentration. For treatment with hepatocyte growth factor (HGF) and activation of MET, cells were washed twice with PBS, serum-deprived overnight, and then treated with HGF (EMD Millipore), at a final concentration of 50 ng/mL in serum-free cell culture medium.

Total RNA was isolated from cells using the RNeasy kit from Qiagen according to the manufacturer's instructions. Lung adenocarcinoma and normal lung tissue samples were homogenized using Omni plastic disposable probes and an Omni (TH-115) homogenizer (Omni International) for 1 minute on dry ice after which total RNA was isolated using Trizol reagent according to the manufacturer's instructions. Total RNA was quantified using the Nanodrop 1000 spectrophotometer (Thermo Scientific). RNA quality was assessed based on RNA integrity numbers generated by the Agilent Bioanalyzer 2000 (Agilent) according to the manufacturer's instructions.

Total RNA was isolated from lung adenocarcinomas (n = 80) and adjacent normal lung tissues (n = 30). RNA was also isolated from lung cancer cells transfected with scrambled siRNA and ETS2-specific siRNA (3 replicates each, n = 6). RNA samples were analyzed by microarray expression profiling using the Affymetrix Human Gene 1.0 ST platform (Affymetrix) according to the manufacturer's instructions. Briefly, in vitro transcription and cDNA target preparation were conducted using the WT expression kit (Life Technologies). A total of 2.5 μg of fragmented and labeled cDNA was generated using the Affymetrix GeneChip WT Terminal Labeling and Controls Kit and hybridized onto Human Gene 1.0 ST arrays according to the manufacturer's instructions (Affymetrix). Arrays were washed, stained, and processed using Affymetrix GeneChip Fluidics Station 450 systems after which they were imaged using Affymetrix GeneChip Scanner 3000 7G for subsequent generation of raw data (*CEL files). Microarray data were submitted to the Gene Expression Omnibus and were MIAME compliant. Lung tumors and normal lung tissue were deposited under series GSE43458 (samples GSM1062765-GSM1062874) and array data of transfected cells were submitted under series GSE43459 (samples GSM1062875-GSM1062880). Raw data were normalized using Robust Multichip Array and log2 transformed using BRB-ArrayTools v 4.3.0 developed by Dr. Richard Simon and the BRB-ArrayTools Development Team (Biometric Research Branch, Rockville, MD; ref. 17). Genes significantly differentially expressed between lung adenocarcinomas and normal lung tissues were selected based on a false discovery rate (FDR) threshold of less than 0.001 and a 1.5-fold change between the 2 groups (n = 2,665 transcripts). Genes differentially expressed between H441 lung cancer cells transfected with ETS2-specific siRNA compared with cells transfected with scrambled siRNA were selected on the basis of a P < 0.01 (n = 1,816 transcripts; Supplementary Table S2). Functional pathway analysis was conducted using the commercially available software Ingenuity Pathways Analysis (IPA) according to the manufacturer's instructions.

siRNAs against ETS2 and MET as well as scrambled (control) siRNA were synthesized by a proprietary design as SMARTpool siRNA (Dharmacon, Thermoscientific). Knockdown of ETS2 or MET expression was conducted using Lipofectamine RNAiMAX (Life Technologies) according to the manufacturer's instructions. ETS2 overexpression was achieved using an ETS2 cDNA clone (Origene) inserted into a Cytomegalovirus plasmid pCMV-XL5 vector. Cells were transfected with the expression vectors using Lipofectamine 2000 (Life Technologies) according to the manufacturer's instructions. For HGF treatment experiments, transfections were conducted one day before overnight incubation with serum-free cell culture medium.

Transwell cell migration was quantified by seeding cells (n = 8 × 10⁴) in serum-free medium onto the top layer of 24-well BD BioCoat 8.0 μm PET membrane inserts (BD Biosciences). Cells (n = 8 × 10⁴) in serum-free medium were seeded onto the top layer of 24-well BD BioCoat Matrigel invasion chambers (BD Biosciences) for quantitative assessment of cell invasion. After 48 hours, migrating or invading cells were washed with PBS, fixed with 10% formalin, and stained with 0.5% crystal violet then counted using bright-field microscopy. For HGF treatment experiments, a concentration of 50 ng/mL of HGF in serum-free medium was supplied to the top layer. Qualitative assessment of cell migration was conducted by the wound-healing assay in which cells were scratched with a 200 μL pipette tip after which the wound was monitored for closure. All conditions were conducted in 3 replicates.

Cells were seeded in 3 replicate wells for each experimental condition at a density of 3 × 10⁴ cells per well in 12-well plates and then transfected the following day with siRNAs or overexpression vectors. Seventy-two hours after transfection, cells were washed with 1× PBS, trypsinized, mixed with 0.4% Trypan blue solution (Sigma Aldrich) and then counted using the Reichert Bright-Line Hemacytometer (Hausser Scientific) by Trypan blue exclusion principle.

For anchorage-dependent colony formation on plastic, cells were seeded at a density of 150 cells per well in 6-well plates. Plates were monitored for colony development for 2–3 weeks under a phase-contrast microscope after which cells were washed with 1× PBS, fixed with 10% formalin, and stained with 0.5% crystal violet before counting. All experiments were carried out in triplicates and Student t test was used to test for statistical significance. Cells were assessed for anchorage-independent colony formation by growth on soft agar. Cells were seeded (1 × 10⁴/well) in 2 mL culture medium containing of 0.3% (w/v) of granulated agar (Difco, BD Biosciences). Cell-agar suspensions were plated on a 2 mL bottom layer composed of 0.6% (w/v) of agar. Both layers of soft agar were supplemented with 20% FBS. All conditions were conducted in triplicates and colonies were scored 2 to 3 weeks after cell seeding.

Cell monolayers were washed twice with PBS, harvested, and lysed with ice-cold radioimmunoprecipitation assay buffer (Sigma-Aldrich) after which protein lysates (20 micrograms) were subjected to SDS- PAGE and Western blotting. The antibodies used for immunoblotting included those raised against ERK1/2, phosphorylated-ERK1/2 (THR202/TYR204), MET, phosphorylated-MET (Y1234/Y1235; Cell Signaling Technology), ETS2 (Santa Cruz Biotechnology), and β-actin (loading control; Sigma-Aldrich). Antibody binding was detected by enhanced chemiluminescence (Amersham Biosciences Corp.). Band intensities were quantified relative to intensity of β-actin using ImageJ software.

Levels of phosphorylated MET (Y1234/Y1235) were quantified 48 hours after knockdown of ETS2 expression using the phospho-MET(Y1234/Y1235) PathScan sandwich ELISA (Cell signaling Technology) according to the manufacturer's instructions.

A total of 1 μg of RNA was reverse-transcribed using the High Capacity RNA-to-cDNA kit (Life Technologies) according to the manufacturer's instructions and diluted in nuclease-free water. Quantitative real-time PCR (qRT-PCR) was conducted using predesigned TaqMan expression assays for ETS2 (Hs00232009_m1), β-ACTIN (Hs99999903_m1), MET (Hs01565584_m1), RAB3B (Hs01001137_m1), SERPINB5 (Hs00985285_m1), ZEB1 (Hs00232783_m1), SNAI2 (Hs00950344_m1), and SPDEF (Hs01026050_m1; Life Technologies) on an ABI 7300 Real-Time PCR System (Applied Biosystems). Relative quantification was calculated using the 2^−ΔΔCT relative quantification method.

Immunohistochemical (IHC) analysis was conducted on 4 μm FFPE histologic sections using purified rabbit polyclonal primary antibodies raised against ETS2 (1:1000 dilution, clone RB24588, AP9846c, Abgent) and phospho-MET (Y1234/Y1235) (1:250 dilution, catalog number AF2480; R&D systems). Antigen retrieval was carried out using the Dako Target Retrieval System at a pH of 6 (Dako). Intrinsic peroxidase activity was blocked by 3% methanol and hydrogen peroxide for 15 minutes and serum-free protein block from Dako was used for 30 minutes for blocking nonspecific antibody binding. Slides were then incubated with antibodies against ETS2 and phospho-MET (Y1234/Y1235) at room temperature for 65 and 90 minutes, respectively. After 3 washes in TBS, slides were then incubated for 30 minutes with Dako Envision+ Dual Link at room temperature. Following 3 additional washes, slides were incubated with Dako chromogen substrate for 5 minutes and were counterstained with hematoxylin for another 5 minutes. FFPE pellets from lung cancer cell lines displaying positive ETS2 and phospho-MET (Y1234/Y1235) expression were used as positive controls for antibody optimization, whereas samples and whole-section tissue specimens processed similarly, except for the omission of the primary antibodies, were used as negative controls. Immunostaining intensity and reactivity were examined by an experienced pathologist (J. Fujimoto) using a light microscope under a × 20 magnification objective. ETS2 expression was quantified using a 4-value intensity score (0, none; 1, weak; 2, moderate; and 3, strong) and the percentage (0%–100%) of the extent of reactivity. A final expression score was obtained by multiplying the intensity and reactivity extension values (range, 0–300; ref. 16). For analysis of phosphorylated MET expression, sections with membranous immunoreactivity were considered positive for phospho-MET expression as previously described (18, 19) and phospho-MET staining was quantified using a score range of 0 to 3+ (20).

ANOVA and Student t test were used to test for statistical significance among different groups in the in vitro experiments. Statistical analysis of the IHC data was first summarized using standard descriptive statistics and frequency tabulations. Associations of ETS2 IHC protein expression alone or in combination phospho-MET (Y1234/Y1235) with patient outcome (time to recurrence) was estimated using the Kaplan–Meier method and compared among groups by log-rank statistical tests. Multivariate Cox proportional hazard models were applied to assess the effects of ETS2 and/or phospho-MET IHC expression on time to recurrence and adjusted for tumor stage. All computations were carried out in SAS 9.3 and S-plus 8.2.

We conducted gene expression profiling of lung adenocarcinomas (N = 80) and matched normal lung (N = 30) and identified, using an FDR threshold of 0.1% and a 1.5-fold change cutoff, 2,665 transcripts that were significantly differentially expressed between lung adenocarcinomas and normal lung. Further analysis from our current expression profiling efforts revealed significantly decreased expression of the ETS2 transcription factor transcript in lung adenocarcinomas compared with normal lung tissues (P < 10⁻⁷; Fig. 1A). In silico analysis of a publicly available microarray dataset (21) corroborated our findings and similarly showed a significant decrease in ETS2 expression in NSCLC tumors compared with adjacent normal lung. qRT-PCR analysis confirmed the ETS2 decreased expression (P = 0.003; Fig. 1B). Moreover, ETS2 expression was highly correlated between gene microarray and qRT-PCR in a panel of normal lung and lung adenocarcinoma tissues (R = 0.72, P < 0.001; Fig. 1C). Analysis of ETS2 expression in our gene expression microarray dataset in the context of various clinicopathologic features revealed that ETS2 was significantly lower in smoker compared with never-smoker adenocarcinomas (P < 0.001; Supplementary Fig. S1A) which was confirmed in a subset of adenocarcinoma tumors by qRT-PCR (P < 0.001; Supplementary Fig. S1B). Similarly, ETS2 expression, assessed by qRT-PCR, was lower in established NSCLC cell lines previously derived from smokers compared with those isolated from never-smoker patients (P = 0.05; Supplementary Fig. S2). These findings show that ETS2 mRNA expression is reduced in lung adenocarcinomas relative to normal lung, in particular, in smoker lung tumors.

ETS2 is a canonical transcription factor that affects a broad spectrum of genes by binding to ERE sequences in the genome thereby affecting various molecular and cellular functions (3, 5). Our findings on reduced ETS2 transcript expression in lung tumors prompted us to examine the IHC expression of ETS2 protein in association with patient outcome that has not been characterized before. We analyzed ETS2 expression in a NSCLC (n = 201) TMA (135 adenocarcinomas and 66 SCCs) derived from patients who did not receive neoadjuvant or adjuvant treatment. Representative photomicrographs of low (left) and high (right) ETS2 IHC expression are depicted in Supplementary Fig. S3. Association of ETS2 IHC expression with patient clinical outcome was conducted following dichotomization of patients into groups with relatively higher or lower ETS2 expression than the IHC score of 55, which was determined by the martingale residual. Low ETS2 expression predicted shorter time to recurrence in all-stages (Fig. 2A) and stage-I (Fig. 2B) NSCLC (both P = 0.004 of the log-rank test). Moreover, relatively low ETS2 expression predicted shorter time to recurrence in all-stages (P = 0.03; Fig. 2C) and stage-I (Fig. 2D; P = 0.046) lung adenocarcinoma but not in lung SCC. Importantly, multivariate Cox proportional hazard models showed that, after adjusting for stage, relatively lower ETS2 IHC protein expression was an independent predictor of shorter time to recurrence in NSCLC (P = 0.009, HR = 1.89, 95% confidence interval 1.17–3.06) and in lung adenocarcinoma (P = 0.03, HR = 1.86, 95% confidence interval 1.05–3.31). Our findings on the association of low ETS2 immunohistochemical expression with shorter time to recurrence in NSCLC points to a potential tumor suppressor function for ETS2 in human lung cancer.

We then sought to test the role of ETS2 in NSCLC pathogenesis. H441 lung cancer cells transfected with siRNA specific to ETS2 exhibited significantly reduced ETS2 transcript compared with cells transfected with control scrambled siRNA (P < 0.05) (Fig. 3A). Expression of ETS2 protein was also reduced by 0.6-fold, albeit less than that observed for the transcript, following RNA interference (Fig. 3A). Moreover, ETS2 knockdown significantly increased cell growth (1.55-fold) evidenced by Trypan blue exclusion (Fig. 3B, right) count as well as augmented cell migration (Fig. 3B, left) and invasion (Fig. 3B, middle) by 4.4-fold and 6.2-fold, respectively (all P < 0.05). Conversely, ETS2 overexpression in H1299 lung cancer cells, with a relatively low basal expression of the gene (Fig. 3C), significantly decreased cell migration (Fig. 3D, left) and invasion (Fig. 3D, middle) by 2.6-fold and proliferation by 1.5-fold (Fig. 3D, right; all P < 0.05). Moreover, overexpression of ETS2 in H1299 cells significantly suppressed anchorage-dependent (1.8-fold) and –independent (2.1-fold) colony formation evidenced by growth on plastic and soft agar, respectively (P < 0.05; Supplementary Figs. S4A and S4B). Importantly, siRNA-mediated knockdown of ETS2 expression nearly completely rescued cells from cell growth inhibition mediated by overexpression of the gene suggesting that the observed cell growth inhibition was ETS2-mediated (Supplementary Fig. S4C). It is worthwhile to mention that the fold changes in cellular migration and invasion following knockdown or overexpression of ETS2 were greater than the noted differences in cell growth. These findings suggest a tumor suppressor function for the ETS2 gene in vitro in human lung cancer.

To gain insights into the mechanisms of ETS2 tumor suppressor function, we sought to compare and contrast the transcriptome of cells transfected with scrambled siRNA and ETS2-specific siRNA. Gene expression profiling, using the Affymetrix Human Gene 1.0 ST platform, identified 1,816 transcripts that were significantly differentially expressed, based on a P < 0.01 threshold, in H441 lung cancer cells with ETS2 knockdown compared with their control counterparts (Fig. 4A; Supplementary Table S2). Moreover, functional analysis of the genes using IPA, revealed that ETS2 knockdown modulated key pathways typically activated in cancer, in particular, HGF, integrin, tissue factor, semaphoring, and mitogen-activated protein kinase (MAPK) signaling (all P < 0.001; Fig. 4B). Notably, the pathway analysis revealed that HGF signaling was the top modulated canonical pathway following ETS2 knockdown (P < 10–5; Fig. 4B) and that an HGF-mediated gene interaction network was among the top significant gene networks following topological arrangement of the differentially expressed genes by IPA (Supplementary Fig. S5). In addition, we observed significant induction of genes involved in epithelial mesenchymal transition (EMT; ref. 22; e.g., zinc finger E-box binding homeobox 1/ZEB1 and snail homolog 2/SNAI2), GTPases (e.g., RAB3B), known ETS2 targets (ref. 23; e.g., serpin peptidase inhibitor clade B member 5/SERPINB5), and members of the ETS family (ref. 5; e.g., SAM pointed domain containing ets transcription factor/SPDEF) all of which were confirmed by qRT-PCR analysis following selection by a combination of statistical criteria and pathways analysis (Fig. 4C). Our global transcriptome analysis further points to a tumor suppressor function by ETS2 in human lung cancer cells.

Our expression profiling analysis showed that HGF signaling was the top significant modulated pathway following siRNA-mediated ETS2 knockdown in lung cancer cells. As HGF is known to function as ligand for the MET receptor oncogene (24, 25), we sought to determine the effects of ETS2 knockdown on the latter tyrosine kinase receptor. Western blotting analysis showed that siRNA-mediated knockdown of ETS2 expression increased phosphorylated levels of MET by 1.2-fold (Fig. 5A). Enzyme-linked immunosorbent assay for phosphorylated MET levels similarly revealed an increase in MET phosphorylation at Y1234/Y1235 following ETS2 knockdown (Supplementary Fig. S6). Conversely, ETS2 overexpression suppressed MET phosphorylation which was rescued by cotransfection of ETS2-targeting siRNA (Supplementary Fig. S7). Importantly, cotransfection of H441 lung cancer cells with siRNA targeting MET (Fig. 5B) abrogated cell migration (Fig. 5C, top), invasion (Fig. 5C, middle) and, to a less pronounced effect, cell growth (Fig. 5C bottom; P < 0.05) mediated by knockdown of ETS2 expression alone (Fig. 5B and C). Moreover, we examined the expression of genes we had identified to be induced by ETS2 knockdown (Fig. 4C) in cells transfected with siRNAs targeting both ETS2 and MET. Cotransfection of cells with MET-specific siRNA significantly attenuated the induction of SNAI2, RAB3B, SPDEF, and SERPINB5 mediated by knockdown of ETS2 alone (all P < 0.05) as well as reduced, albeit not significant, expression of the ZEB1 gene (Fig. 5D). These data show that ETS2 suppresses key features of the lung cell malignant phenotype, at least in part, through inhibition of the MET oncogene.

As we found that ETS2 suppresses cellular migration and invasion through inhibition of MET phosphorylation and activation, we sought to examine the role of ETS2 in HGF-mediated signaling. Following an overnight serum starvation, H441 lung cancer cells were treated with HGF (50 ng/mL final concentration) for one hour. HGF treatment increased ETS2 protein expression by 1.3-fold in H441 cells (Fig. 6A). The increase in ETS2 protein was also noted in additional lung cancer cell lines (Supplementary Fig. S8). Moreover, and as shown in previous studies (26, 27), HGF treatment also increased phosphorylation of both MET and extracellular-regulated kinase (ERK1/2) MAPK (Fig. 6A) in H441 lung cancer cells which was alleviated at later treatment time points (Supplementary Fig. S8A) as previously reported for earlier HGF treatment and ligand stimulation studies (28, 29). Furthermore, expression of ETS2 protein and of phospho-ERK1/2 and phospho-MET was attenuated by cotreatment of cells with the MET tyrosine kinase inhibitor PHA-665752 (100 nmol/L final concentration; Fig. 6A and Supplementary Fig. S8B).

We then sought to assess the relevance of ETS2 to HGF signaling in lung cancer cells. RNA interference-mediated knockdown of ETS2 expression significantly augmented HGF-induced cell migration (1.9-fold) and invasion (1.7-fold; P < 0.05; Fig. 6B, right). These effects were concomitant with increased MET phosphorylation in HGF-treated cells transfected with ETS2-specific siRNA (1.8-fold with respect to control cells) which was higher than that observed in treated cells transfected with control siRNA (1.5-fold with respect to control cells; Fig. 6B, left). Furthermore, we also found that, similar to knockdown of ETS2 alone (P < 0.05), HGF treatment also significantly increased the expression of the SNAI2, RAB3B, SPDEF, ZEB1, and SERPINB5 genes (all P < 0.05; Fig. 6C). It is worthwhile to mention that HGF treatment also significantly increased ETS2 mRNA levels (P < 0.05; Fig. 6C). Importantly, the expression of the aforementioned genes was significantly higher in HGF-treated cells transfected with ETS2-specific siRNA compared with similarly treated cells that were transfected with control siRNA (P < 0.05; Fig. 6C). The observed additive effect of the combination of HGF treatment and ETS2 knockdown on MET phosphorylation was also confirmed in an additional lung cancer cell line (H1944; Supplementary Figs. S9A and S9B). While HGF treatment alone increased phospho-MET protein, as evidenced by Western blotting, by 61-fold, phosphorylated MET protein was increased by 98-fold following HGF treatment in cells with knockdown of ETS2 (Supplementary Fig. S9B). This effect was concomitant with increased cell growth (1.6-fold), rather than invasion, compared with HGF-treated cells with control siRNA (Supplementary Fig. S9C). Moreover and conversely, overexpression of ETS2 in H1299 lung cancer cells decreased HGF-induced phosphorylated MET levels and decreased HGF-induced cell migration evidenced by the wound-healing assay (Supplementary Fig. S9D). Our findings suggest that ETS2 may impact dissimilar phenotypic and antimalignant effects downstream of HGF in different lung cancer cell lines. Interestingly, differential antimalignant phenotypic effects by the MET inhibitor, PHA-665752, in different cancer cell lines were also previously noted in the original report by Christensen and colleagues (26). Our findings suggest that ETS2 inhibits HGF-mediated oncogenic signaling in a negative-feedback manner.

It is worthwhile to mention that we also analyzed phosphorylated MET IHC expression (Supplementary Fig. S10) along with that of ETS2 in association with patient time to recurrence. Interestingly, patients with NSCLC (P = 0.01) or adenocarcinoma (P = 0.02) with relatively lower ETS2 expression and higher phosphorylated MET membrane immunoreactivity exhibited significantly shortest time to recurrence compared with other patient subgroups, in particular those with higher ETS2 and higher phosphorylated MET expression (Supplementary Fig. S11A–S11D). These data further suggest that ETS2 suppression of MET oncogene signaling is operative in human lung cancer.

Our expression profiling efforts revealed significant downregulation of the ETS2 transcription factor in lung adenocarcinomas compared with normal lung tissues. We questioned the relevance of ETS2 downregulation to NSCLC biology, given the important cellular and molecular roles this transcription factor plays (3, 5), and found that low ETS2 protein was predictive of shorter time to recurrence and the gene was a negative regulator of lung cancer cell growth, migration, and invasion and of HGF/MET oncogenic signaling. It is worthwhile to note that we had found significant reduced expression of ETS2 in smoker compared with never-smoker lung adenocarcinomas (Supplementary Fig. S1) and, similarly, NSCLC cell lines originally derived from smokers exhibited reduced ETS2 expression compared with lung cancer cell lines isolated from never-smokers. It is worthwhile to mention that in silico analysis of publicly available expression datasets also revealed reduced ETS2 transcript expression in smoker adenocarcinomas as well as marked downregulation of expression of the gene in airways of phenotypically healthy smokers compared with disease-free nonsmokers (data not shown). In addition and notably, ETS2 mRNA expression, part of a genomic signature of phosphoinositide 3-kinase (PI3K) pathway activation, was decreased in the normal bronchial airways of smokers with lung cancer compared with the airways of disease-free smokers (30). It is plausible to suggest that cigarette smoking reduces expression of the ETS2 gene in the course of NSCLC pathogenesis.

Microarray profiling coupled with functional pathways analysis revealed that ETS2 expression is linked to various canonical cancer-associated pathways in lung cancer cells, in particular the HGF pathway. Moreover, qRT-PCR analysis confirmed the array findings and showed upregulation of the ZEB1, SNAI2, RAB3B, SERPINB5, and SPDEF genes following siRNA-mediated knockdown of ETS2 expression concomitant with elevated cell migration and invasion. ZEB1 and SNAI2 (also known as SLUG) are transcription factors with well-established roles in promotion of EMT in various physiologic and pathologic conditions (22). RAB3B is a relatively understudied RAS family member and GTPase that was found to be an important positive regulator of epithelial cell polarity (31) and of the stemness of adult human mesenchymal stem cells (32). SERPINB5, a serpin peptidase inhibitor, was shown to be a target repressed by murine Ets2 (23) and was shown to be upregulated in human lung tumors compared with normal lung tissues (33) and hypomethylated and elevated in lung SCCs (34). SPDEF (also known as PDEF), a member of the ETS-domain protein family (5), was shown to be upregulated in mammary tumors (35) and to positively control cancer cell migration and invasion (36). Notably, SPDEF was found to promote airway epithelial cell (e.g., goblet cell) hyperplasia, inflammation, and mucus hypersecretion (37). It is also noteworthy, that HGF signaling, known for its promigratory and -invasive effects (27), was the top pathway significantly modulated following ETS2 knockdown in vitro. In addition, ETS2 knockdown significantly augmented HGF-mediated induction of all 5 genes. It is reasonable to suggest, given the increased expression of the aforementioned genes following ETS2 knockdown, along with their given previous reported promalignant roles, that ETS2 functions as a lung tumor suppressor in lung cancer cells, at least in part, by regulation of the migration- and invasion-promoting transcriptome.

In this study, we found that ETS2 suppressed HGF-mediated MET phosphorylation and cell migration and invasion in a negative feedback manner. Our findings are in accordance with previous reports showing ETS2 negative feedback regulation of other signaling networks, namely FGF signaling (38) and phorbol ester-induced MAPK signaling (39) by upregulation of MAPK phosphatases (38, 39). Moreover, a previous study by Cao and colleagues showed modulation of ETS2 expression and intracellular localization downstream of HGF/MET pathway activation in mesothelioma cancer cell lines, albeit without functional assessment of the relevance of increased ETS2 expression (40). Because we also showed that ETS2 basally suppressed MET phosphorylation and activation, it is plausible to suggest that ETS2 functions both upstream and downstream of MET oncogene to inhibit cell migration and invasion. In an attempt to further probe the relevance of the identified MET/ETS2 signaling to human lung cancer, we studied the association of both ETS2 and phosphorylated MET IHC protein expression with NSCLC patient outcome. The combination of both decreased ETS2 and increased phosphorylated MET IHC expression identified a subgroup of patients with further reduced time to recurrence compared with patients with high ETS2 and high phosphorylated MET protein or when compared with analysis of ETS2 expression alone. Further studies to confirm the combinatorial effect of low ETS2 protein and high phosphorylated MET IHC expression on reduced time to recurrence in lung cancer or other malignancies are warranted. It is worthwhile to note that MET function was shown to be deregulated in different types of solid tumors and hematologic malignancies by increased transcript expression or copy number gain or amplification (27, 41, 42). It is intriguing to suggest that ETS2 may represent an additional endogenous level of pathway control of MET activation in NSCLC.

It is important to note that ETS2 has been shown to exhibit both oncogenic and tumor suppressor roles in various malignancies. For example, ETS2 was shown to have mitogenic activity downstream of Ras in murine fibroblasts (8), mediate transformation and tumorigenicity of prostate and breast epithelial cells (13, 14), and to be upregulated in breast, thyroid, prostate, and pancreatic tumors as well as in hematologic malignancies such as leukemias (4, 43–45). Moreover, loss of murine Ets2 expression in tumor-associated macrophages was shown to decrease breast cancer metastasis in mice (23). On the other hand, various studies have revealed tumor suppressor functions and effects mediated by the ETS2 gene. Foos and colleagues showed that overexpression of full-length Ets2, with intact transactivator function, reversed Ras-induced NIH3T3 cell transformation (46) and, in a separate study, Hever and colleagues showed that Ets2 was not required for ErbB2-mediated cell transformation in vitro (47). In addition, mice modeling Down's syndrome and colon cancer exhibited reduced intestinal tumor incidence which was attributed to increased gene dosage of Ets2, located in the 21q22.2 chromosomal locus, in Down's syndrome with trisomy of chromosome 21 (15). In this context and similarly, mice with Ets2-deficient intestinal cells were shown to develop more colon tumors in response to treatment with carcinogens (12). Additional potential mechanisms of ETS2-mediated tumor suppression have been attributed to suppression of miR-196b–induced invasion in gastric cancer cells (11) and promotion of cell-cycle arrest by negative feedback inhibition of MAPK signaling following treatment with the phorbol ester PMA (39). Interestingly, overexpression in lung cancer cells of the frequently epigenetically inactivated lung tumor suppressor RASSF1A was shown to increase ETS2 transcript expression concomitant with arrest of cell-cycle progression (48). Furthermore, ETS2 expression was shown to be downregulated in human lung invasive adenocarcinomas compared with noninvasive bronchioalveolar carcinomas (49). The opposing tumor-repressive and -supportive functions attributable to ETS2 in different tumors may be, in part, related to the dissimilar roles ETS2 may exhibit in tumors of different cell lineages.

In conclusion, we showed, in this study, reduced expression of ETS2 transcript expression in human lung tumors relative to normal lung tissues, association of low ETS2 protein expression with shorter time to recurrence in patients with NSCLC, and inhibition of human lung cancer cell invasion, migration, and growth by ETS2, at least in part, through suppression of MET oncogene phosphorylation and HGF signaling. Our findings provide collective testament to a potential tumor suppressor role of the ETS2 gene in human NSCLC pathogenesis, in part, through inhibition of MET oncogene activation.

No potential conflicts of interest were disclosed.

Conception and design: M. Kabbout, I.I. Wistuba, H. Kadara

Development of methodology: M. Kabbout, L.M. Solis Soto, I.I. Wistuba, H. Kadara

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Garcia, D. Woods, G. Mendoza, C. Behrens

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Kabbout, J. Fujimoto, D. Liu, A.A. Momin, B.P. James, J.J. Lee, I.I. Wistuba, H. Kadara

Writing, review, and/or revision of the manuscript: M. Kabbout, J. Fujimoto, L.M. Solis Soto, I.I. Wistuba, H. Kadara

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Kabbout, D. Woods, C.-W. Chow, I.I. Wistuba, H. Kadara

Study supervision: I.I. Wistuba, H. Kadara

This work was supported in part by a Lung Cancer Research Foundation grant (to H. Kadara), Department of Defense (DoD; W81XWH-04-1-0142 to I.I. Wistuba), and by UT Lung SPORE P50CA70907 (to I.I. Wistuba) and Cancer Center Support Grant CA-16672 from the National Cancer Institute.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.