Eph proteins have emerged as critical drivers affecting tumor growth and progression in human malignancies. Our The Cancer Genome Atlas (TCGA) data analysis showed that EphB3, a receptor tyrosine kinase, is frequently coamplified with PIK3CA in head and neck squamous cell carcinoma (HNSCC). We therefore hypothesized that EphB3 amplification plays a protumorigenic role in HNSCC and that EphB3 and PIK3CA are cooperating oncogenes that contribute toward its pathogenesis. This hypothesis was not experimentally supported, because EphB3 knockdown failed to alter HNSCC tumor cell growth in vitro or in vivo with an orthotopic model. However, responsiveness of EphB3 knockdown tumors to the PI3K inhibitor, BKM120, was significantly decreased in terms of both tumor growth delay and survival. This is correlated with an increase in prosurvival proteins, S6 and BcL-XL, in the EphB3 shRNA tumors treated with BKM120 compared with controls. We further observed that EphB3 knockdown resulted in increased migration in vitro and increased EMT gene signature in vivo. To explain these results, we examined EphB3 phosphorylation levels in HNSCC at baseline. Although total EphB3 levels were high, we found low phospho-EphB3 levels in HNSCCs. Forced EphB3 phosphorylation with an ephrin-B2–Fc fusion protein resulted in decreased HNSCC migration and cell growth, and enhanced response to BKM120 in vitro. These data collectively indicate that progression of HNSCC selects for low/inhibited EphB3 activity to enhance their survival and migratory abilities and decrease response to PI3K signaling. Therefore, strategies focused on activating EphB3 might be helpful to inhibit tumor growth and enhance sensitivity to PI3K inhibitors in HNSCC. Mol Cancer Ther; 17(9); 2049–59. ©2018 AACR.
Scientific developments over the past few years have enhanced our understanding of dysregulated cell signaling pathways potentiating cancer progression, and Eph receptors have emerged as attractive candidates for therapeutic intervention. The paradoxical nature of Eph activities in different cancer types (1), however, poses a significant challenge, and we are now beginning to understand the dichotomy of Eph receptors and how their dysregulated expression can promote or inhibit tumor progression.
EphB3 is a receptor protein tyrosine kinase that interacts with ephrin-B ligands to trigger downstream signaling and has been reported as a key player in multiple cancers (2–5). For instance, EphB3 has been shown to play a tumor-suppressing function in colorectal cancer (4–7). In non–small cell lung cancer (NSCLC), disparate functions for EphB3 have been described previously (2, 3). A report by Ji and colleagues has described significant upregulation of EphB3 in both patient tumor specimens and cancer cell lines and a correlation was observed between EphB3 levels and patient pathologic features (3). They further documented that high levels of EphB3 increased proliferation and migration both in vitro and in vivo in a kinase-independent manner (3). Furthermore, knockdown of EphB3 resulted in decreased tumorigenesis and metastasis in vivo (3). A study published by the same research group showed that EphB3 receptor is hypophosphorylated in 75% of NSCLC samples, despite its high expression levels (2). Forced activation of EphB3 kinase in EphB3-overexpressing NSCLC cancer cells was shown to inhibit cell migratory capability in vitro as well as in metastatic seeding in vivo (2).
Interrogation of the The Cancer Genome Atlas (TCGA) database led us to identify EphB3 as a new gene target with high copy number amplification in head and neck squamous cell carcinoma (HNSCC). Furthermore, we found that both EphB3 and PI3KCA, present on chromosome 3q (8), are frequently coamplified in HNSCC. We therefore hypothesized that EphB3 amplification plays a protumorigenic role in HNSCC and that EphB3 and PIK3CA are cooperating oncogenes that contribute toward its pathogenesis. We undertook a loss-of-function approach with shRNA knockdown to examine the effects on HNSCC growth, migration, and sensitivity to PI3K inhibitors. Our data surprisingly showed that EphB3 knockdown does not alter tumor growth, but promotes migration, upregulation of epithelial-to-mesenchymal transition (EMT), and decreases responsiveness to PI3K inhibitors. In light of these data, which refute our original hypothesis, we subsequently found that despite high levels of total EphB3, low levels of baseline EphB3 phosphorylation are observed in HNSCC. Furthermore, forced phosphorylation of EphB3 with ephrin-B2–Fc fusion protein decreased HNSCC migration and cell growth, and enhanced responsiveness of these cells to PI3K inhibitor. These novel findings improve our understanding of the role of EphB3 in HNSCC and provide a potential therapeutic strategy for the treatment of this cancer, particularly in the setting of PI3K inhibitors.
Materials and Methods
TCGA data analysis
Whole-genome sequences from all cancer-type cohorts (27 cohorts total) of TCGA (http://cancergenome.nih.gov) were accessed via cBioPortal (http://cbioportal.org) and queried for any genomic alterations in EPHB3. Cohorts containing significant amplification of EPHB3 including head and neck provisional (n = 530), lung squamous cell carcinoma provisional (n = 530), and cervical cancer provisional (n = 309) were requeried for alterations in PIK3CA gene commonly found altered in head and neck cancers. EPHB3 is mentioned in RSEM (RNA-seq by expectation maximization) units.
For survival analysis, the HNSCC RNAseq dataset was downloaded from TCGA and patients with oral cavity tumors (n = 314) were selected. For the purpose of this study, classifications of alveolar ridge, buccal mucosa, lip, oral tongue, and floor of mouth were reclassified to oral cavity. Patients were sorted by EPHB3 gene expression and divided into quartiles. Patients in the upper quartile were classified as high EPHB3 and patients in the lower quartile were classified as low EPHB3. Overall survival and disease-free survival were calculated by Kaplan–Meier method using log-rank tests for comparisons. Univariate Cox proportional model was used to calculate the HR. Two-sided P values were reported for all survival analyses.
Cell lines and reagents
The human HNSCC cell lines CAL27 and Fadu were obtained from the ATCC. MSK921 and Detroit 562 cell lines were obtained from Dr. X.J. Wang's laboratory (University of Colorado, Anschutz Medical Campus, Aurora, CO). MSK921 cells were maintained in RPMI1640 medium (Gibco). CAL27, Fadu, UM-SCC25, UM-SCC1, and Detroit 562 cells were maintained in DMEM (Gibco). All these cell lines were authenticated by short tandem repeat testing. Murine B4B8 and LY2 squamous cell carcinoma cells were obtained from the laboratory of Dr. Nadarajah Vigneswaran (UTHealth, Houston, TX). Both these cell lines were maintained in DMEM/F12 (Gibco). MOC1 and MOC2 cell lines were obtained from the laboratory of Dr. Young J. Kim (Johns Hopkins University, Baltimore, MD), whereas MEER and MOE cell lines were obtained from Dr. John Lee (Sanford Health, Sioux Falls, SD). All cell lines were grown in the presence of 10% FBS and primocin at 37°C, 5% CO2 and checked for Mycoplasma contamination. The class I PI3K inhibitor, BKM120, was purchased from Selleck Biochemicals and prepared as 100 mmol/L stocks in DMSO with aliquots stored at −20°C.
Generation of shRNA knockdown clones
Lentivirus-encapsidated shRNA vectors (Sigma, pLKO.1) were purchased from the University of Colorado Cancer Center Functional Genomics Facility (Anschutz Medical Campus, Aurora, CO). EphB3 knockdown was achieved by transducing cells with shRNA against human EphB3 (Sigma-Aldrich catalog no. TRCN0000006428) or mouse EphB3 (Sigma-Aldrich catalog no. TRCN0000023351). Cells were transduced with control shRNA (Sigma-Aldrich catalog no. SHC016) in parallel. Transduced cells were selected with 1 μmol/L or 6 μmol/L (CAL27 and Ly2, respectively) puromycin and following selection, cells were maintained in 1 μmol/L dose. For Cal27 cell lines, we were able to generate four viable clones with EphB3 knockdown. For Ly2 cell line, A and B were the viable clones we were successfully able to generate and expand for experiments. We used the clones that showed approximately 60% knockdown compared with control shRNA group.
CAL27 cells (control shRNA and EphB3 knockdown clones) were seeded at a density of 1.5 × 105 cells per well in 6-well plates and maintained in complete growth medium containing 10% FBS and primocin. Approximately 72 hours postplating, cells were fixed with 10% TCA (LabChem) and viability was measured by sulforhodamine B (SRB) assay (ACROS Organics). For Trypan blue assay, cells were collected by trypsinization. To each 10-μL aliquot of cells, 10 μL of Trypan blue dye was added. Cell counting was performed using an automated T-20 cell counter (Bio-Rad). The experiment was replicated twice and each condition had 3 to 4 replicates.
Stimulation with clustered Fc protein
Control ChromoPure mouse IgG-Fc fragment (Jackson ImmunoResearch Laboratories) or rmEphrinB2-Fc chimera (R&D Systems) was preclustered with anti-hIgG1 Fc antibody (R&D Systems) at a 1:3 ratio for 30 minutes at 4°C with shaking. Preclustering ensures rapid and effective receptor activation once ligands come into contact range. Preclustered Fc protein was added to cell culture media at a final concentration of 3 μg/mL. Cells were stimulated with preclustered ligand for 15 minutes at 37°C. Following stimulation, cells were transferred to ice and washed two times with 1× cold PBS. Cells were detached by scraping and transferred to a microfuge tube. Cells were pelleted, supernatant removed, and pellet was lysed in 1× cold Cell Lysis Buffer (Cell Signaling Technology) containing 1× Halt Protease Inhibitor Cocktail (Thermo Scientific) and Phosphatase Inhibitor Cocktail 2 (Sigma). Protein concentration was determined by BCA assay (Thermo Scientific).
Frozen tumors were pulverized in liquid nitrogen using a mortar and pestle followed by resuspension in 1× Cell Lysis Buffer containing protease and phosphatase inhibitors. Lysate was diluted to 2 μg/μL in lysis buffer and 250 μL of resulting lysate was incubated with 1:100 dilution of anti-EphB3 antibody (Sigma) overnight at 4°C with end-over-end rotation. Magnetic protein G beads (Bio-Rad) were washed three times with 1× PBS prior to adding lysate–antibody complexes. Beads were incubated with lysate–antibody complexes for 1 hour at 4°C with end-over-end rotation. Beads were washed three times with cold 1:1 lysis buffer:PBS, removing supernatant after each wash. Protein was eluted from beads by adding 1× sample loading buffer (Invitrogen) containing 1× sample reducing agent (Invitrogen) and boiling beads at 95°C for 15 minutes. A total of 200 μg equivalent starting material was resolved on a 10% SDS-PAGE gel and transferred to Immuno-Blot PVDF membrane (Bio-Rad). Membranes were probed for total phosphotyrosine (Cell Signaling Technology) prior to stripping (Restore PLUS Western Blot Stripping Buffer, Thermo Scientific) and reprobed for total EphB3 protein (Abcam). The immunoprecipitation assay was repeated twice.
Boyden chamber migration assay
CAL27 cells or LY2 were serum starved overnight prior to seeding (2.5 × 105 cells/well) in serum-free media into the top chamber of a 24-well plate insert with 8.0 μmol/L pores (BioCoat Control Inserts, BD Biosciences). Bottom chamber was filled with appropriate media containing 10% FBS and preclustered control-Fc or ephrinB2-Fc (R&D Systems). Cells were incubated in a 37°C cell culture incubator and allowed to migrate for 48 hours (CAL27) or 24 hours (LY2). To prepare membranes for imaging, unmigrated cells in the top chamber were removed according to manufacturer's instructions and migrated cells were stained in 0.1% crystal violet (10% methanol), washed in dH2O, and allowed to dry overnight before membrane was cut and mounted on a microscope slide with immersion oil. An image of the membrane was taken at 4× magnification and number of migrated cells were counted using ImageJ software (NIH, Bethesda, MD). Experiment was repeated 2 to 3 times. Each condition had three replicates.
Generation of CAL27 BKM120–resistant cell line
CAL27 cells were continually cultured in increasing doses of BKM120 (1.5, 2.0, and 2.5 μmol/L) until the cells were able to survive and proliferate in 2.5 μmol/L BKM120. Parental and BKM120-resistant CAL27 cells were plated in drug-free media overnight. The following day the cells were treated with DMSO or 2.5 μmol/L BKM120. After 72-hour incubation, relative cell abundance was assessed by SRB assay.
Cell growth assay
CAL27 cells were seeded at a density of 2,000 cells/well in a 96-well plate and stimulated with preclustered Control-Fc or ephrin-B2-Fc as described previously. Cells were stimulated again at 48 hours, and 72-hour postplating. Changes in cell growth between stimulated and nonstimulated groups were analyzed by MTT assay by reading the absorbance at 595 nm on a microplate reader. Data were plotted as % cell growth normalized to 24-hour time point from multiple replicates per experimental condition.
HNSCC cells transfected with control or EphB3 shRNA and tumor tissues treated with vehicle or BKM120 inhibitor treatment were harvested. Cells or tumor tissues were homogenized in RIPA lysis buffer (Millipore), containing protease inhibitor cocktail (Thermo Fisher Scientific Inc.) and phosphatase inhibitor (Sigma) on ice for 30 minutes. The homogenate was centrifuged at 4°C at 13,000 rpm for 20 minutes, and lysates were collected. Protein concentration was determined using the BCA Protein Assay Kit (Thermo Fisher Scientific Inc.). Lysates (20–30 μg) were loaded onto 10%–12% SDS-PAGE gels. Electrophoresis, blocking, probing, and detection of proteins were conducted as described previously (9). Membranes were probed overnight at 4°C with respective antibodies. All primary antibodies against p-AKT (S473), AKT, p-S6 (S235/S236), S6, and Bcl-XL were obtained from Cell Signaling Technology. Anti-EphB3 antibody was obtained from Sigma. Horseradish peroxidase (HRP)-conjugated secondary antibodies were obtained from Sigma. The experiment was replicated two times.
In vivo studies
Female athymic nude mice (6-week old, n = 13–15) were purchased from Envigo. All mice were handled in accordance with the ethics guidelines and conditions set and overseen by the University of Colorado, Anschutz Medical Campus Animal Care and Use Committee. All protocols for animal studies were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Colorado, Anschutz Medical Campus. CAL27 shRNA control and EphB3 shRNA–transduced cells were injected into the right buccal cavity at 1 × 106 cells/mouse in 1:1 serum-free media:Matrigel ratio. BKM120 was administered at a dose of 30 mg/kg daily by oral gavage for a total duration of 2 weeks (5 days/week) in 1:9 1-methyl-2-pyrrolidone:PEG300 for a total of 10 doses of drug. A mix of 1-methyl-2-pyrrolidone:PEG300 solution (1:9 ratio) was used as a vehicle. Tumor growth was measured twice a week using a digital caliper and tumor volume was calculated using the formula: [(smaller diameter)2 × (longer diameter)]/2. Tumors harvested between days 25 and 45 postimplantation were used for Western blotting and IHC analysis. Survival analysis was performed using Kaplan–Meier survival curves. For in vivo studies, survival was calculated on the basis of animal death dictated by tumor volume guidelines (maximum limit: 2,000 mm3) approved in our animal protocol and loss of 20% of total body weight.
IHC analysis was performed using an established protocol (10) on tumors harvested from mice implanted with either CAL27 Control shRNA or CAL27 EphB3 knockdown clones treated with Vehicle or BKM120 using anti-PCNA antibody (1:100 dilution, BD Biosciences), E-cadherin and vimentin (1:100 dilution, Cell Signaling Technology). This was followed by incubation with biotinylated goat anti-mouse IgG or biotinylated goat anti-rabbit IgG secondary antibody (1:400 dilution, Vector Laboratories). Images were captured using a 20× objective on a Nikon microscope. At least 4 to 5 images were captured per tumor section. The quantification was performed by counting the number of positive cells or positive area per high power field using ImageJ software.
All the experiments were performed in duplicate or triplicate and repeated two to three times. Quantitative analyses were performed using Student t test or one-way ANOVA by GraphPad Prism software. A P value of <0.05 was considered significant.
EphB3 is amplified in head and neck cancer
Alterations in EPHB3 across cancer types in the TCGA database were accessed via cBioportal and analyzed. We found that EPHB3 is amplified in multiple human carcinomas, including HNSCC (Fig. 1A). We also performed analysis of microarray gene expression in 30 head and neck cancer cell lines from the Cancer Cell Line Encyclopedia. We found, similar to patients, a wide range of EPHB3 expression (Supplementary Fig. S1). To confirm these data, we examined EphB3 protein levels in multiple human (Fig. 1B) and murine head and neck cancer cell lines (Fig. 1C). Total EphB3 protein levels were found to be variable in multiple human and murine HNSCC, compared with normal respective negative control tissue of adult cerebellum (11) and NIH 3T3 cells (Fig. 1B and C). We also observed some variation in EphB3 levels in human HNSCC cell lines when compared with normal oral keratinocyte (NOK) cell line (Supplementary Fig. S2A and S2B).
Knockdown of EphB3 does not affect cell viability in vitro but increases cell migration
To elucidate the role of EphB3 receptor in HNSCCs, we knocked down the levels of the receptor in both human and murine head and neck cancer cell lines using EphB3-targeting shRNA. Compared with the control transfectants, CAL27 and Ly2 cells transfected with EphB3-specific shRNA showed approximately 55%–80% reduction in the EphB3 levels (Fig. 2A and B). Our data demonstrated that reduced EphB3 expression in CAL27 cells (Fig. 2C) or Ly2 cells (Supplementary Fig. S3A) did not significantly alter tumor cell viability when compared with the control shRNA group. These data were substantiated by SRB assay (Fig. 2D). EphB3 receptor has previously been reported to play a role in tumor cell migration in NSCLC (2). Therefore, we investigated whether shRNA-mediated knockdown of EphB3 would affect cell migration in HNSCC cells using a Boyden chamber assay. A significant increase in migration was evident in EphB3-knockdown CAL27 cells compared with the cells that were transfected with control shRNA (Fig. 2E). In Ly2 cells, cells were EphB3 was knocked down also showed a trend toward increased migration compared with the control counterpart (Supplementary Fig. S3B).
EphB3 and PIK3CA are frequently coamplified in squamous cell carcinomas including HNSCC
Given that PI3K activity has been shown to modulate Eph/ephrin proteins (12, 13), we sought to establish if there is any correlation between EphB3 and PI3KCA as both genes being located on chromosome 3. Genomic alterations in the TCGA database were analyzed for head and neck cancer (n = 530), lung squamous cell carcinoma (n = 530), and cervical squamous cell carcinoma and endocervical adenocarcinoma (n = 309; Fig. 3A). It was noted that approximately 18% to 43% patients have coamplification of EPHB3 and PIK3CA, a catalytic subunit of PI3K (Fig. 3A; Table 1). We also analyzed the levels of PI3 Kinase p110α subunit in our panel of human HNSCC cells and observed variable expression (Supplementary Fig. S2A and S2C).
|Cancer type .||Percentage of all patients with coamplified gene alteration .||P .||Log OR .|
|Cervical SCC and adenocarcinoma||18.77||<0.001||<3|
|Cancer type .||Percentage of all patients with coamplified gene alteration .||P .||Log OR .|
|Cervical SCC and adenocarcinoma||18.77||<0.001||<3|
Abbreviation: SCC, squamous cell carcinoma.
EphB3 downregulation decreases sensitivity to PI3K inhibitor in vitro in head and neck cancer cells
To determine the role of EphB3 in PI3K signaling inhibition, we generated PI3K inhibitor–resistant CAL27 cell line. CAL27 cells were grown in the presence of increasing concentrations of BKM120, a pan-class I PI3K inhibitor (14), until they had acquired resistance to 2.5 μmol/L BKM120. SRB proliferation assay demonstrates that CAL27-BKM120–resistant cells proliferate more slowly than parental CAL27 cells, but are less sensitive to BKM120 treatment than parental cells, which are acutely sensitive to BKM120 (Fig. 3B). DMSO at the same concentration as BKM120 was used as a negative control. Furthermore, we exposed CAL27 parental and BKM120-resistant cells to increasing doses of BKM120. Western blot analysis data demonstrate that exposure of parental CAL27 to PI3K inhibitor for 24 hours decreased the level of EphB3, although the BKM120-resistant cells maintained constitutively low levels of EphB3 (Fig. 3C). Parental CAL27 cells treated with BKM120 had reduced levels of p-AKT and p-S6, downstream effectors of PI3K signaling pathway (Fig. 3C). Levels of p-AKT and p-S6 were increased in CAL27 BKM120–resistant cell line following short-term withdrawal of the BKM120 (Fig. 3C). The p-AKT and p-S6 levels were subsequently reduced after exposure to the PI3K inhibitor for 24 hours (Fig. 3C). The densitometric analysis for this panel is presented in Table 2. In addition, we analyzed these markers in CAL27 cell line stably expressing control or EphB3-specific shRNA treated with vehicle or BKM120. The data show a minor increase in the levels of p-AKT and p-S6 in the CAL27 EphB3 knockdown clones compared with the control shRNA cells in the absence of BKM120 inhibitor. The levels of p-S6 protein are slightly increased in the EphB3 knockdown cells compared with the control samples in the presence of high dose of BKM120 (Fig. 3D). The densitometric analysis is presented in Table 3. These data demonstrate an association between decreased sensitivity to PI3K inhibition and decreased EphB3 levels.
|CAL27 .||BKM120 (μmol/L) .||EphB3/β-actin .||p-AKT/AKT .||AKT/β-actin .||p-S6/β-actin .|
|CAL27 .||BKM120 (μmol/L) .||EphB3/β-actin .||p-AKT/AKT .||AKT/β-actin .||p-S6/β-actin .|
|CAL27 .||BKM120 (μmol/L) .||p-AKT/AKT .||AKT/β-actin .||p-S6/β-actin .||p-S6/β-actin .|
|CAL27 .||BKM120 (μmol/L) .||p-AKT/AKT .||AKT/β-actin .||p-S6/β-actin .||p-S6/β-actin .|
EphB3 knockdown does not significantly alter tumor growth and survival but decreases sensitivity to PI3K inhibitor in an orthotopic model of head and neck cancer
On the basis of the in vitro findings and TCGA data, we next evaluated the effect of EphB3 knockdown on tumor growth in vivo using CAL27 cell line stably expressing control shRNA or EphB3 shRNA. Because the complex interplay between Eph receptors and ligands is highly context specific, it is the best modeled in vivo. To closely mimic the tumor microenvironment, tumor cells were implanted orthotopically in the buccal region of nude mice. Orthotopic tumors best recapitulate characteristics of an aggressive tumor growth model to study genetic and molecular changes associated with tumor progression and development (15). They also serve as useful tools to study the efficacy of therapeutic agents (15). In our experimental model, we observed that knockdown of EphB3 did not significantly affect CAL27 tumor growth compared with the control group (Fig. 4A). Because of the aggressive tumor growth pattern, most of the mice in the EphB3 knockdown and control group subjected to vehicle treatment were sacrificed after day 20 postvehicle treatment.
We also interrogated whether EphB3 knockdown in head and neck cancer cells alters in vivo sensitivity toward PI3K inhibition. We found that EphB3 knockdown in CAL27 tumors that were administered with BKM120 had significantly enhanced tumor growth by approximately1.5-fold compared with the control BKM120-treated xenografts at day 42 posttreatment with BKM120 (Fig. 4B).
In addition, we investigated whether EphB3 knockdown in CAL27 tumors had an effect on overall survival in vivo. Kaplan–Meier survival curves show that mice implanted with CAL27 knockdown tumors had a median survival of 25 days compared with mice implanted with CAL27 control tumors where the median survival was approximately 38 days (Fig. 4C). Treatment of CAL27 control tumors with BKM120 also provided survival benefit compared with other groups with median survival of approximately 70 days (Fig. 4C). However, EphB3 knockdown tumors treated with BKM120 reduced the median survival to 50 days (Fig. 4C).
To determine proof-of-target, we performed an immunoprecipitation assay to detect the phosphorylation status of EphB3 in the control vehicle versus EphB3 knockdown tumors treated with vehicle or BKM120. We observed decreased levels of p-EphB3 in CAL27 EphB3 knockdown tumor xenografts (Fig. 4D). Further comparisons between the BKM120-treated groups show reduction in the p-EphB3 levels following EphB3 receptor downregulation compared with the control BKM120 treatment (Fig. 4D). We also analyzed the levels of the prosurvival proteins, p-S6, S6, p-AKT, AKT, and Bcl-XL, in the tumors harvested from control and EphB3 knockdown groups treated with vehicle or BKM120 treatment. Our data show a decrease in the prosurvival protein levels of p-S6, S6, AKT, and Bcl-XL in control tumors treated with BKM120 (Fig. 4E). An increase in the levels of p-S6, p-AKT were observed in EphB3 shRNA knockdown tumors following BKM120 treatment compared with similarly treated control transfectants (Fig. 4E).
EphB3 knockdown decreases sensitivity toward BKM120 treatment by affecting markers of tumor cell proliferation and migration in vivo
Given that our in vitro data showed an increase in HNSCC migration following EphB3 knockdown, we sought to determine the mechanisms underlying changes in tumor cell growth observed in vivo particularly following BKM120 inhibitor treatment. IHC analysis on tumor sections was performed on tumors derived from EphB3 knockdown and control tumors with vehicle or BKM120 treatment. Our data show that knockdown of EphB3 did not affect tumor cell proliferation to a significant extent compared with the control vehicle as evident by PCNA staining (Fig. 4F). The CAL27 tumors treated with BKM120 showed reduced PCNA staining compared with the control vehicle–treated group (Fig. 4F). However, when tumors with EphB3 knockdown were treated with BKM120, a significant increase in the levels of PCNA was observed compared with control BKM120 and EphB3 KD vehicle (Fig. 4F), suggesting that EphB3 knockdown decreases response to BKM120 inhibitor treatment and results in increased tumor cell proliferation.
Because we observed an increase in migration of EphB3 knockdown cells in vitro, we also analyzed the levels of E-cadherin (an epithelial cell marker) and vimentin (a mesenchymal marker), proteins reported to play a role in tumor cell migration (16, 17). These two proteins with opposing roles have long been considered as key components of epithelial–mesenchymal transition (EMT) in various cancers (16, 17). The cells that undergo the process of EMT are characterized by morphologic transformation, alteration in the levels of EMT markers that in turn affects cell motility (16, 17). Downregulation of E-cadherin function or expression and gain of function or elevated levels of vimentin have been reported to play a role in cancer progression and metastasis (18, 19). We observed significant upregulation of E-cadherin in control tumors treated with BKM120 compared with the nontreated counterparts (Fig. 4G). However, when EphB3-deficient tumors treated with BKM120 inhibitor were analyzed, we observed significant reduction in E-cadherin expression compared with the control tumors that were subjected to the same treatment (Fig. 4G). Vimentin is an EMT marker that plays a critical role in tumor cell migration and metastasis. Consistent with an EMT phenotype, IHC analysis showed that EphB3 downregulation significantly enhanced the levels of vimentin compared with the control tumors treated with vehicle (Fig. 4H). However, when the control tumors were treated with PI3K inhibitor, there was a significant reduction in vimentin expression compared with control vehicle group (Fig. 4H).
EphB3 is present in an inactive state and forced phosphorylation with ephrin-B2–Fc stimulation decreases migration, cell growth, and enhances responsiveness to BKM120 in vitro
Our data suggest that HNSCC selects for low or inhibited EphB3 activity to enhance their survival and migratory abilities and decrease response to PI3K signaling. To reconcile the unexpected data, which refutes the original hypothesis that EphB3 amplification by HNSCC is indicative of protumorigenic role and rather supportive of a hypothesis that EphB3 exerts a suppressor role, we assessed the activation of EphB3 by examining its phosphorylation status at baseline. We found that although the basal level of EphB3 was detectable in both CAL27 and Ly2 cells, the basal phospho-EphB3 levels were low (Fig. 5A and B). Stimulation with ephrin-B2 Fc increased the levels of phosphorylated form of EphB3 (Fig. 5A and B). In addition, forced phosphorylation with ephrin-B2-Fc stimulation significantly decreased CAL27 HNSCC migration in a Boyden chamber assay (Fig. 5C) and cell growth by MTT assay (Fig. 5D) Furthermore, we observed a significant reduction in cell growth when ephrin-B2–stimulated CAL27 cells were treated with BKM120 compared with the control-Fc–treated group (Fig. 5E).
To further understand why HNSCC cell lines amplify EphB3 RNA levels although simultaneously inactivating it, we examined the levels of its ligand ephrin-B2. Ephrin-B2 is membrane bound that signals via reverse signaling and this has been shown to play a critical role in promoting protumorigenic roles in different cancer subtypes (20, 21). Furthermore, we have established it to be a negative prognostic factor in HNSCC (22). Evaluation of Western blot analysis shows variable levels of ephrin-B2 protein in human and mouse HNSCC cell lines (Fig. 5F and G). Analysis of the phosphorylation status of EphB3 in CAL27 and Ly2 cells transfected with control shRNA and EphB3-specific shRNA by immunoprecipitation analysis shows reduced levels of p-EphB3 present in both CAL27 (Fig. 5H) and Ly2 cells (Fig. 5I) transfected with EphB3-specific shRNA as compared with the control shRNA clones.
High EphB3 expression correlates with better overall and disease-free survival in patients with HNSCC
Given that our findings support a model where activation of EphB3 acts in a suppressive role in HNSCC, we interrogated survival outcomes in the TCGA RNAseq dataset. Patients with oral cavity tumors (n = 314) were sorted by EphB3 gene expression and divided into quartiles (Fig. 6A). Patients in the upper quartile (Q1) were classified as high EphB3 cohort and those with lower quartile (Q4) were classified as low EphB3 cohort (Fig. 6A). Survival analysis revealed a trend that patients with high EphB3 expression have better overall survival (median overall survival ∼ 49 months) compared with the patients with low EphB3 expression (median overall survival ∼ 28 months; Fig. 6B). Patients with high EphB3 expression also showed significantly higher disease-free survival rates with median survival approximately 71 months as compared with the patients with low EphB3 expression with median survival approximately 28 months (Fig. 6C).
The family of Eph receptor tyrosine kinases has garnered attention of the scientific community due to their role in both physiologic and pathologic processes (23–25). Dysregulated expression of both Eph receptors and their cognate ligands has been reported to contribute toward tumor progression, and metastasis (23). Aberrant expression of EphB3, a member of the Eph receptor gene family, has been reported in different cancers (2–4). In colorectal cancer, a tumor suppressor function has been described for EphB3 where its loss has been shown to result in invasive cancer in the Apc (min/+) mice (5) and acts as an inhibitor of tumor cell metastasis by regulating intercellular adhesive and repulsive interactions (5). In NSCLC, EphB3 has been found to be amplified, but its silencing suppressed tumor growth and metastasis (3). Yet, in a different manuscript done by the same group a year later, it was reported that inhibition of EphB3 promoted NSCLC migration and metastasis (2). This perplexing dichotomy was attributed to differences in kinase activation status of the EphB3 receptor, whereby it acts as a tumor promoter via “kinase independent” properties and as a tumor suppressor via “kinase dependent” properties.
To our knowledge, this is the first study that investigates EphB3 expression and function in HNSCC. We found that EphB3 is overexpressed in the TCGA HNSCC dataset and is coamplified with PI3KCA. Given that the high levels of expression and the association between Eph receptor tumor–promoting properties and the AKT pathway (2, 12), we hypothesized that EphB3 amplification plays a protumorigenic role in HNSCC and those EphB3 and PIK3CA are cooperating oncogenes that contribute toward its pathogenesis. Our results unexpectedly showed that the EphB3 shRNA knockdown approach had no significant effect on cancer cell viability in vitro as determined by Trypan blue staining and SRB assay, but it increased HNSCC migration, increased EMT markers, and reduced sensitivity to PI3K inhibitors. Furthermore, we found that high EphB3 RNA expression levels correlate with improved disease-free survival outcomes. This led us to investigate the activation or phosphorylation status of the protein at baseline in this context. Our data show that EphB3 is hypophosphorylated at baseline and forced phosphorylation with ephrin-B2 stimulation decreases migration, cell growth, and enhances response to PI3K inhibitor. These results are consistent with the published literature showing that EphB3 receptor activation decreases invasion, migration, and metastasis in NSCLC (2), colorectal cancer (5), and hepatoblast migration (26). Knockdown of EphB3 also reduced sensitivity to BKM120 via an increase in AKT and p-S6 proteins. The altered effects on tumor growth and sensitivity to PI3K inhibitor observed in our tumor model were associated with changes in the expression level of proteins known to play a role in the cell proliferation, survival, and cell migration. PI3K–AKT–mTOR pathway alterations are implicated in almost a third of HNSCC, with PIK3CA commonly mutated and multiple PI3K–AKT–mTOR pathway aberrations associated with disease progression (27, 28). In the recent article published by Swick and colleagues, combined inhibition of PI3K/AKT/mTORC and RAS/RAF/MAPK1 signaling resulted in significant tumor growth inhibition (28). However, the clinical success of agents targeting these pathways has shown limited success. Therefore, it justifies the need to look for alternate agents that can be exploited to maximize the therapeutic response. Our data suggest that EphB3 knockdown reduces sensitivity to PI3K inhibitors. Therefore, a novel generation of compounds that can enhance the activated levels of EphB3 in tumors might show some promise in combination with PI3K inhibitors.
To explain the biology of why a tumor suppressor gene would be amplified, we examined the status of its cognate ligand, ephrin-B2. The biology of EphB–ephrinB signaling is highly complex because both the receptor and ligand can signal upon cell-to-cell contact (23). Forward signaling from the EphB receptor requires the ectodomain of ephrin-B2 for activation, whereas reverse signaling from ephrin-B2 ligand requires the ectodomain of an EphB receptor to signal. Signaling mediated by either a receptor or a ligand or both (bidirectional) can be activated during cell-to-cell interaction and leads to an effect. Activation of ephrin-B2 signaling has been shown to trigger tumorigenesis, a process mediated in part by a proangiogenic effect (20, 21, 29). Our published analysis showed that ephrin-B2 acts as a prognostic indicator of overall survival and response to both chemotherapy and radiotherapy in patients with high-risk HNSCCs (22). We also showed that inhibition of ephrin-B2 signaling results in tumor growth retardation in patient-derived xenograft models of high-risk HNSCC (22). In this study, we show elevated levels of ephrin-B2 in various HNSCCs cell lines. This suggests that cancer cells selects for low or inhibited tumor suppressor activity while at the same time using its ectodomain to stimulate the oncogenic effect of a tumor promoter. However, given the redundancy of EphB–receptor expression in HNSCC, with EphB4 being another ephrin-B2 binding receptor that is ubiquitously expressed, the effect of targeted deletion of EphB3 on ephrin-B2 activation and cell proliferation can potentially be compensated.
In conclusion, our study provides the first evidence establishing the role of EphB3 in head and neck cancers and modulated response to PI3K signaling inhibition. Future studies are warranted to understand the underlying mechanism and interaction between EphB3 receptor and PIK3CA signaling component in our model system that in turn will guide the development of novel therapeutic agents applicable for patients with head and neck cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: S. Bhatia, A. Griego, S.D. Karam
Development of methodology: S. Bhatia, J. Sharma, S.D. Karam
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Bhatia, A. Griego, S. Lennon, C. Rohmer, N. Uyanga, B. Van Court, C. Young, S.D. Karam
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Bhatia, A. Griego, S. Lennon, A. Oweida, C. Rohmer, C. Young, S.D. Karam
Writing, review, and/or revision of the manuscript: S. Bhatia, A. Griego, S. Lennon, A. Oweida, J. Sharma, D. Raben, C. Young, L. Heasley, S.D. Karam
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Griego, J. Sharma, S.D. Karam
Study supervision: S.D. Karam
This work was supported by the funding awarded to S.D. Karam by the NIH under Ruth L. Kirschstein National Research Service Award (T32CA17468), Paul Calabresi Career Development Award for Clinical Oncology (K12) CA086913, and the RSNA grant (RSD1713).
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.