Evidence for a dual function of EphB4 as tumor promoter and suppressor regulated by the absence or presence of the ephrin-B2 ligand


  • Raphael Rutkowski,

    1. Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
    Search for more papers by this author
    • *

      R.R. and I.M.-W. contributed equally to this work

  • Inga Mertens-Walker,

    Corresponding author
    1. Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
    • Institute of Health and Biomedical Innovation, Queensland University of Technology, Cells and Tissues, 60 Musk Avenue, Kelvin Grove, Queensland 4059, Australia
    Search for more papers by this author
    • *

      R.R. and I.M.-W. contributed equally to this work

  • Jessica E. Lisle,

    1. Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
    Search for more papers by this author
  • Adrian C. Herington,

    1. Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
    Search for more papers by this author
  • Sally-Anne Stephenson

    1. Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave, Kelvin Grove, QLD 4059, Australia
    Search for more papers by this author


Overexpression of the receptor tyrosine kinase EphB4 is common in epithelial cancers and linked to tumor progression by promoting angiogenesis, increasing survival and facilitating invasion and migration. However, other studies have reported loss of EphB4 suggesting a tumor suppressor function in some cancers. These opposing roles may be regulated by (i) the presence of the primary ligand ephrin-B2 that regulates pathways involved in tumor suppression or (ii) the absence of ephrin-B2 that allows EphB4 signaling via ligand-independent pathways that contribute to tumor promotion. To explore this theory, EphB4 was overexpressed in the prostate cancer cell line 22Rv1 and the mammary epithelial cell line MCF-10A. Overexpressed EphB4 localized to lipid-rich regions of the plasma membrane and confirmed to be ligand-responsive as demonstrated by increased phosphorylation of ERK1/2 and internalization. EphB4 overexpressing cells demonstrated enhanced anchorage-independent growth, migration and invasion, all characteristics associated with an aggressive phenotype, and therefore supporting the hypothesis that overexpressed EphB4 facilitates tumor promotion. Importantly, these effects were reversed in the presence of ephrin-B2 which led to a reduction in EphB4 protein levels, demonstrating that ligand-dependent signaling is tumor suppressive. Furthermore, extended ligand stimulation caused a significant decrease in proliferation that correlated with a rise in caspase-3/7 and -8 activities. Together, these results demonstrate that overexpression of EphB4 confers a transformed phenotype in the case of MCF-10A cells and an increased metastatic phenotype in the case of 22Rv1 cancer cells and that both phenotypes can be restrained by stimulation with ephrin-B2, in part by reducing EphB4 levels.

EphB4 is a member of the largest subfamily of receptor tyrosine kinases (RTKs) and is commonly overexpressed in cancer cells from several different tissue origins including prostate,1, 2 breast,3 colon,4 head and neck5 and ovary.6, 7 Various studies exploring the role of EphB4 overexpressed in cancer cells have suggested that EphB4 can promote tumor development by stimulating angiogenesis8, 9 increasing cancer cell survival3, 6, 10 and facilitating invasion and migration.11, 12 Overexpression of EphB4 in the mammary epithelium of MMTV-neuT transgenic mice accelerated the onset of tumor formation and increased metastasis to the lungs.13 In xenograft models of breast and prostate cancer, knockdown of EphB4 significantly inhibits tumor growth indicating that EphB4 directly contributes to tumor progression.2, 3 Similarly, targeting EphB4 using a monoclonal antibody also significantly reduced tumor growth in vivo.14 Collectively, these studies show that in the absence of ligand, EphB4 receptors can contribute to tumor progression.

Paradoxically, EphB4 has also been implicated as a tumor suppressor.15–17 This apparent contradiction may be reconciled, at least in part, if distinct EphB4 signaling pathways are differentially regulated by ligand-dependent and ligand-independent mechanisms and this may depend on the cell context.18, 19 These ligand-independent actions of Eph receptors can be achieved by the following: receptor dimerization and constitutive activation as a result of overexpression,20, 21 phosphorylation in the absence of ligand22 and both kinase-dependent and -independent functions.23, 24 A similar conundrum has been described for the EphA2/ephrin-A1 system. In prostate cancer, kinase-dependent and -independent functions of EphA2 regulated by the absence or presence of ephrin-A1 ligand, influence the metastatic phenotype23 and adenoviral delivery of ephrin-A1 to breast cancer cells reduced their tumorigenic potential.25 Similar results were also obtained in other cell types where overexpression of EphA2 induced chemotactic migration, which was abolished in the presence of ephrin-A1 or shRNA targeting EphA2.26 These results indicate that in contrast to other RTKs such as EGFR, which in response to EGF ligand stimulation usually enhances oncogenic signaling, ligand stimulation of Eph receptors can be tumor suppressive.

The primary physiological ligand of the EphB4 receptor is ephrin-B2,27 which is also membrane-bound and located on neighboring cells. Eph/ephrin interactions lead to bidirectional signaling as both proteins are capable of mediating signals into their respective cell.18 Stimulation of EphB4 forward signaling using soluble ephrin-B2-Fc chimera has been shown to cause cell death of tumor cells both in vitro and in vivo28, 29 which, in breast cancer cells, is mediated via the tumor suppressive Abl/Crk pathway.28 In addition, reverse signaling induced by using soluble EphB4 has also been shown to inhibit tumor growth and angiogenesis.9, 30

To explore the relationship between ligand-dependent and -independent EphB4 activity and to confirm that EphB4 is a tumor promoter and ephrin-B2-induced forward signaling mediates tumor suppressive effects, we generated two models of exogenous EphB4 overexpression. We found that overexpression of functional EphB4 in 22Rv1 (22Rv1-B4) prostate cancer cells and MCF-10A (MCF10A-B4) nontransformed epithelial mammary cells increased anchorage-independent growth, cell migration and invasion, all characteristics indicative of an increased metastatic and transformed phenotype. Extended stimulation with ephrin-B2 reversed the tumorigenic effects induced by EphB4 which supports the hypothesis that the activation of forward signaling of EphB4 by ephrin-B2 diminishes the EphB4-induced tumor promoting effects.

Material and Methods

Cell culture

All cell lines were purchased from the American Type Culture Collection (Manassas, VA). 22Rv1 prostate cancer cells were cultured in RPMI 1640 (Invitrogen, Mulgrave, VIC, Australia) supplemented with 10% FBS and penicillin/streptomycin. MCF-10A cells were cultured in DMEM/F12 (Invitrogen) supplemented with 10% FBS, 20 ng/ml EGF, 100 ng/ml cholera toxin, 0.5 μg/ml hydrocortisone, 10 μg/ml insulin (all Sigma, St. Louis, MO) and penicillin/streptomycin. MDA-MB-468, MDA-MB-231, T47D and MCF-7 breast cancer cells were cultured in DMEM supplemented with 10% FBS and penicillin/streptomycin.


GAPDH antibody was obtained from Abcam (Cambridge, MA), antiphospho-ERK1/2 and antitotal ERK1/2 were from Cell Signaling (Danvers, MA), mouse recombinant chimeric ephrin-B2/Fc was obtained from R&D Systems (Minneapolis, MN), mouse-anti-EphB4 was from Invitrogen (Carlsbad, CA), antiphosphotyrosine 4G10 Platinum mouse monoclonal antibody was from Millipore (Billerica, MA) and β-actin antibody was from Sigma.

Generation of stable cell lines

For the production of polyclonal derivative cell lines that stably overexpressed EphB4, an expression vector containing the full length coding sequence of human EphB4 (Genbank accession number NM_004444) was produced using the pIRES-neo2 vector (Clontech, Mountainview, CA). Fourteen micrograms of Nru1-linearized pIRES-B4 construct or empty vector was combined with 15 μl lipofectamine 2000 (Invitrogen) and transfected into 22Rv1 and MCF-10A cells grown to 80% confluence. Transfected cells were cultured for 24 hr before stably transfected cells were selected using a preoptimized dose of 900 μg/ml G418 (Invitrogen) for 2 weeks.

Western blotting

Cells were lysed with ice-cold RIPA buffer (50 mM Tris pH 7.4, 1% Triton X-100, 0.5% sodium deoxycholate, 150 mM NaCl, 2 mM EDTA, 1 mM sodium orthovanadate, 1 mM NaF, 1 mM PMSF and 10 μg/ml aprotinin) supplemented with protease inhibitors (complete mini-EDTA-free tablets; Roche, Castle Hill, NSW, Australia). Protein lysates were mixed by rotation at 4°C for 1 hr, then insoluble proteins removed by centrifugation at 14,000 rpm. Soluble protein concentrations were determined using the BCA Protein Assay Kit (Pierce, Thermo, Rockford, IL). Protein samples (20–50 μg) were separated on SDS-PAGE gels (4% stacking and 12% separating) for 2.5 hr then transferred overnight at 30 V onto BioTrace™ NT nitrocellulose membrane (Pall, Pensacola, FL) using SDS-transfer buffer (25 mM Tris, 190 mM glycine, 0.1% SDS, 20% methanol and pH 8.5). The membrane was blocked for 1 hr using Western-blocking reagent (Roche) before protein was detected with specific antibodies followed by incubation with a peroxidase-conjugated goat anti-mouse secondary antibody (Pierce) and the Amersham™ ECL Plus Chemiluminescence kit (GE Healthcare, Rydalmere, NSW) following the manufacturer's recommendation, with 5 sec to 5 min exposure to SuperRX X-ray film (Fuji Film Corporation, Japan).


Cells were serum starved for 24 hr and then stimulated for 15 min with 2 μg/ml soluble ligand ephrin-B2/Fc clustered with AffiniPure goat anti-human IgG, Fcγ (Jackson ImmunoResearch, West Grove, PA). Such clustered recombinant ephrin-B2/Fc proteins have been shown to be the active form of the ligand.31 As negative controls, cells were treated with the clustering antibody (Fc) alone or left untreated. Protein was then extracted from cell lines using RIPA buffer and subjected to immunoprecipitation using an EphB4-specific antibody (Invitrogen) and Protein G Sepharose 4 Fast Flow beads (GE Healthcare). Immunoprecipitated proteins were denatured by heating at 95°C for 5 min before being separated on 12% SDS-PAGE gels which were then immunoblotted to detect both EphB4 and phosphorylated EphB4.

Isolation of membrane rafts

Lipid rafts were isolated using OptiPrep (Sigma) as published elsewhere.32 Briefly, membranes were prepared using base buffer (20 mM Tris-HCl, pH 7.8 and 250 mM sucrose) supplemented with 1 mM CaCl2, 1 mM MgCl2 and protease inhibitors (Roche). This was mixed with 50% Optiprep and underlaid below a discontinuous gradient of 0–20% OptiPrep in base buffer. After centrifugation at 20,300 rpm for 90 min at 4°C, 14 sequential fractions were collected from the top of the gradient and proteins precipitated using trichloroacetic acid. Pellets were resuspended in 8 M urea/50 mM Tris buffer, denatured and analyzed by Western blotting. An anti-flotillin antibody was used as a marker for lipid rafts (BD Biosciences, San Jose, CA).33


Cell lines were seeded onto sterile 13-mm glass cover slips in 24-well tissue culture plates at 1 × 105 cells per well, allowed to attach for 24 hr then fixed using 4% paraformaldehyde before being permeabilized using 0.5% Triton X-100/PBS. Cells were blocked in 10% goat serum for 1 hr at room temperature before incubation with 1 μg/ml anti-human EphB4 overnight at 4°C. Primary antibody was detected using an Alexa Fluor 488-conjugated secondary antibody (Invitrogen) and F-actin stained with phalloidin conjugated to tetramethyl rhodamine isothiocyanate (TRITC) (Sigma). Cover slips were mounted on glass slides using Prolong Gold Antifade with DAPI (Invitrogen) and cells were visualized using a Leica SP5 spectral scanning confocal microscope.

Proliferation and apoptosis assays

Cells were seeded into 96-well plates at a density of 1 × 103 cells per well in triplicate samples. After 24 hr of attachment cells were treated with 2 μg/ml soluble clustered ephrin-B2/Fc or Fc alone and proliferation was examined using CellTiter Aqueous One solution (Promega, Sydney, Australia) at 3, 5, 7 and 11 days. For apoptosis assays using the Caspase-Glo™-3/7 and -8 kits (Promega), 1 × 103 cells were seeded in 96-well plates and triplicate samples treated with 2 or 10 μg/ml ephrin-B2/Fc or Fc alone for 3 days before caspase activity was detected according to the manufacturer's instructions.

Soft agar assay

Anchorage-independent growth was analyzed using a two-layer soft agar system. A bottom layer of 0.6% soft agar (Sigma) in RPMI growth medium was poured and allowed to set for 30 min before 2 × 104 cells per well in 0.3% soft agar was layered on top. Medium containing ephrin-B2/Fc or Fc alone was replenished every 3 days. After 21 days of incubation, colonies were examined using a Leica AF6000 wide field microscope and the size of colonies was measured in triplicate wells.

Scratch migration assay

Cells were seeded into 24-well plates at >90% confluency and grown until they formed a monolayer. The monolayer was scratched using a sterile 200-μl pipette tip and the cells washed carefully with PBS to remove cell debris and detached cells. Fresh complete medium containing 5 μg/ml ephrin-B2/Fc or Fc alone was added and images were taken at 0 and 18 hr (MCF-10A) or 24 hr (22Rv1) using a Leica AF6000 wide field microscope at 10× magnification and the wound areas were measured using Tscratch.34

Invasion assay

The assay was performed using BD BioCoat Growth factor reduced Matrigel invasion chambers (BD Biosciences) or the Xcelligence system RTCA DP (Roche) with a collagen I (Sigma-Aldrich) coated 16-well CIM plate. For the latter, 2.5 × 105 22Rv1-VO or 22Rv1-B4 cells were seeded into the upper chamber of the CIM plate in 0.1% FCS-containing medium in the presence of 5 μg/ml soluble clustered ephrin-B2/Fc. Medium containing 10% FCS was used as chemoattractant. Invasion was monitored during the following 9 hr by measuring changes in resistance at the cell-electrode interphase. The experiments were performed in triplicates. For the Matrigel invasion assay, 5 × 104 MCF10A-VO or MCF10A-B4 cells were incubated for 22 hr and cells that had not invaded were removed from the upper chamber using a cotton swab. The membranes were fixed in ice-cold methanol and stained with DAPI. Invading cells on the underside of the filter were counted in five random fields at 200× magnification using a Nikon Eclipse epifluorescent microscope with a Coolsnap camera and the mean number of cells was calculated relative to untreated control cells.


Data are presented as mean ± SEM unless stated otherwise. Statistical differences were assessed using either Student's t-test or one-way ANOVA followed by Fisher's least significant difference post hoc test with significance set at p < 0.05. The number of replicates and/or repeat experiments is given in the appropriate figure legends.


Characterization of exogenous EphB4

To ensure exogenous EphB4 is appropriately expressed in the stably selected polyclonal populations, 22Rv1 EphB4 overexpressing (22Rv1-B4), MCF-10A EphB4 overexpressing (MCF10A-B4) and empty vector (22Rv1-VO, MCF10A-VO) cells were subjected to Western immunoblotting. The polyclonal populations of 22Rv1-B4 and MCF10A-B4 cells strongly expressed EphB4 whereas vector only (VO) cells hardly displayed immunoreactivity for EphB4 (Fig. 1a). We have previously published that EphB4 is overexpressed in prostate cancer cell lines.1 To investigate endogenous protein levels of the ligand ephrin-B2, several prostate cancer cells were analyzed for ephrin-B2 expression (Supporting Information Fig. S1). 22Rv1 and LnCaP cells showed no expression of ephrin-B2. To ensure that overexpressed EphB4 in MCF10A cells mirrors endogenous overexpression as seen in breast cancer cells, several breast cancer cell lines were subjected to Western immunoblotting for EphB4. MDA-MB-468 and T47D cells highly expressed EphB4 whereas MCF-7 cells showed slightly lower levels and MDA-MB-231 cells displayed the least expression (Fig. 1b). These results are in agreement with previously published data.28 Immunofluorescence was then used to determine the subcellular localization of the overexpressed EphB4. Specific staining was associated with the plasma membrane of 22Rv1-B4 cells and little immunoreactivity was visible in the wild type (WT) 22Rv1-WT or 22Rv1-VO cell lines or the controls (Fig. 1c). MCF10A-B4 cells exhibited membrane staining of EphB4 as well as punctuate staining in the cytoplasm and increased staining at sites of cell–cell contact. MCF-10A WT cells showed no immunoreactivity for EphB4 (Fig. 1c).

Figure 1.

Characterization of exogenous overexpression of EphB4. (a) Protein expression of EphB4 in overexpressing cell models. 22Rv1 prostate cancer cells and MCF-10A mammary epithelial cells were transfected with pIRES-Neo2 containing EphB4 (B4) or the empty vector (VO) and analyzed using immunoblotting for EphB4. There was a marked increase in EphB4 expression after stable transfection of both cell lines. (b) Western blot of total cell lysates of four breast cancer cell lines probed for EphB4 protein expression and reprobed using anti-actin to confirm equal loading. (c) Localization of EphB4 in 22Rv1- and MCF10A-B4 overexpressing cells. Immunofluorescent staining of EphB4 (green) reveals localization of EphB4 on the surface of 22Rv1-B4 cells with little immunoreactivity seen in WT and 22Rv1-VO-transfected cells. Background controls shown are: mIgG, mouse IgG control; 2° only, secondary antibody only; none, no antibodies. MCF10A-B4 cells show EphB4 staining at the membrane and in the cytoplasm and at sites of cell–cell contact. All cell lines were costained for F-actin using phalloidin (red) to visualize the cytoskeleton and DAPI to visualize nuclei (blue). (d) Lipid-raft preparations and subsequent immunoblotting showed that EphB4 was present in high lipid-containing membrane fractions 1 and 2. Flotillin was used as a lipid raft marker.

Many transmembrane proteins such as RTKs are enriched in lipid rafts/caveolae, which have been implicated as signaling hubs and also appear to be deregulated in cancer.35 To determine whether overexpressed EphB4 is localized to these specialized structures within the plasma membrane, lipid rafts were prepared from 22Rv1-B4 cells and analyzed by Western blotting analysis. Immunoreactivity to the EphB4 protein was mostly seen in fractions 1 and 2, which corresponded to the fractions that contained high-lipid concentration (Fig. 1d). Significantly less EphB4 immunoreactivity was observed in fractions 3 and 4. The protein flotillin is recognized as an independent marker of lipid rafts.33 Flotillin appeared primarily in fractions 1–5 as expected, with less reactivity seen in fractions 6 and 7. These results indicate that overexpressed EphB4 is localized to high lipid-containing lipid rafts.

To determine whether the overexpressed EphB4 was functional and responding to its ligand ephrin-B2, the 22Rv1-B4 cells were stimulated with clustered ephrin-B2/Fc (eB2/Fc), clustering antibody alone (Fc only) or left untreated for 15 min, before phosphorylated proteins were immunoprecipitated from total protein lysates using the 4G10 platinum antibody and phosphorylated EphB4 identified by Western analysis using an EphB4-specific antibody (Fig. 2a, top panel). A duplicate sample was simultaneously immunoprecipitated using an EphB4 (IP: EphB4) antibody followed by immunoblotting for EphB4 to confirm equal protein input (Fig. 2a, lower panel). As expected, significant phosphorylated EphB4 was detected in the cells treated for 15 min with the clustered soluble ephrin-B2/Fc, with little phosphorylation detected in the cells treated with clustering antibody (Fc) alone and no phosphorylation in untreated samples.

Figure 2.

Overexpressed EphB4 responds to ephrin-B2. (a) EphB4 is phosphorylated after addition of ephrin-B2. 22Rv1-B4 cells were treated with 2 μg/ml soluble clustered ephrin-B2/Fc (eB2/Fc), clustering antibody (Fc only) or left untreated for 15 min followed by immunoprecipitation using the 4G10 platinum phosphotyrosine antibody (IP: pTYR) followed by immunoblotting for EphB4 (IB: EphB4). A duplicate sample was also immunoprecipitated using an EphB4 (IP: EphB4) antibody followed by immunoblotting for EphB4 to assess equal protein input. (b) EphB4 is internalized in response to its ligand. Immunofluorescent staining of EphB4 after treatment of 22RV1-B4 cells with either Fc alone (i) or soluble clustered ephrin-B2/Fc (ii) for 15 min shows internalization of EphB4 after ligand addition [dotted structures in (ii) as opposed to plasma membrane staining in (i)]. (c) ERK1/2 is activated after addition of ligand. Treatment of 22RV1-B4 cells (B4) or VO control cells with clustered ephrin-B2/Fc (5–60 min) led to activation (i.e., phosphorylation) of ERK1/2 (p-ERK1/2). Equal protein loading was assessed by total (t-ERK1/2). (d) EphB4 protein degrades after 24 hr, 72 hr and 11 days of ligand stimulation. 22Rv1 cells were stimulated with 2 μg/ml soluble clustered ephrin-B2/Fc (eB2) or Fc alone and EphB4 protein expression was assessed using immunoblotting. GAPDH was used as a loading control. In both 22Rv1-VO and-B4 treated cells, EphB4 is markedly decreased.

Immunofluorescence was used to examine the localization of EphB4 in 22Rv1-B4 cells after stimulation with ephrin-B2. EphB4 protein was internalized 15 min after addition of soluble ephrin-B2/Fc and appeared in punctuate foci. Residual membrane staining in treated cells was also observed to become more localized, perhaps into preinternalization structures (Fig. 2b). Addition of ligand for 5–60 min to both VO and EphB4-overexpressing 22Rv1 cells resulted in activation of extracellular-signal-regulated kinase1/2 (ERK1/2) with phosphorylation levels in the EphB4 overexpressing cells markedly higher than in VO cells (Fig. 2c). The lower level, transient activation of ERK1/2 in VO cells can possibly be attributed to the low endogenous levels of EphB4 in VO cells (Fig. 1a) or the presence of other ephrin-B2 responsive receptors in these cells (Supporting Information Fig. S2).36 Extended stimulation of 22Rv1-B4 cells for 24–72 hr and 11 days resulted in a decrease in EphB4 protein (Fig. 2d) confirming that degradation is a consequence of internalization after ligand stimulation, a common response of Eph receptors to their ephrin ligands.37, 38 These results show that exogenously expressed EphB4 is functional in lipid rich domains of the cell membrane and responds to ligand stimulation with phosphorylation and degradation as is reported for endogenous Eph receptors.37

EphB4 enhances cell migration and invasion and ephrin-B2 reverses this effect

To determine the effects of overexpression of EphB4 on the migratory and invasive ability of MCF10A and 22Rv1 cells, in vitro migration and invasion assays were carried out. The 22Rv1-B4 cells showed a significant increase in migration as determined by a scratch wound monolayer assay (Fig. 3a). MCF10A-B4 cells displayed a nonsignificant slight increase in migration, but the addition of ephrin-B2 to the EphB4 overexpressing cells significantly reduced cell migration whereas VO cells were unaffected (Fig. 3b). To analyze the effects of EphB4 on cell invasion, 22Rv1-VO and -B4 cells were subjected to an invasion assay using the Xcelligence system (Roche). CIM plates were coated with collagen I (ColI) and invasion was monitored in real time (Fig. 3c). EphB4 overexpressing cells showed significantly increased invasion (p < 0.001), which was abrogated in the presence of ligand (p < 0.05). VO cells also responded with a significant decrease of invasion in the presence of ephrin-B2 (p < 0.05), reflecting the low levels of EphB4 in the parental V0 cells (Fig. 1a). These results demonstrate that EphB4 enhances invasion through ColI and ephrin-B2 reverses this effect. Furthermore, overexpression of EphB4 in noninvasive MCF-10A cells39 significantly increased the number of cells which invaded through Matrigel by 2.5-fold in a Transwell chamber assay compared to VO cells (Fig. 3d). These results show that EphB4 expression is sufficient to convert a nontransformed cell line into an invasive and migratory phenotype. Addition of ephrin-B2/Fc to both the top and the bottom of the Transwell chamber led to a slight, but nonsignificant decline in invasion of MCF10A-B4 cells. To visualize the effect of ephrin-B2 on cell shape and spreading, 22Rv1-VO or -B4 cells were allowed to attach on ColI coated glass slides for 1 hr followed by immunofluorescence analysis of the cytoskeleton by staining for F-actin (Fig. 3e). In the presence of Fc only, 22Rv1-VO cells show less spreading than 22Rv1-B4 overexpressing cells, which display their typical fibroblast-like morphology. In the presence of ligand, 22Rv1-VO cells appear similar to Fc-treated control cells, but EphB4 overexpressing cells are extremely rounded and have not spread. This indicates that activation of EphB4 by its ligand reduces cell spreading. There was no difference in cell spreading in MCF10A cells (data not shown).

Figure 3.

EphB4 overexpression enhances migration and invasion and ephrin-B2 suppresses this effect. (a) Confluent monolayers of 22Rv1-B4 or -VO cells were wounded with a pipette tip and live cell movement was monitored using a time-lapse camera. After 24 hr, the wound area was quantitated and expressed as percent migration. Data are represented as ±SEM of triplicate wells. **p < 0.01 using a student's t-test. Representative photomicrographs are shown at 0 and 24 hr after wounding. (b) Confluent MCF10A-B4 or -VO cells were wounded and photomicrographs taken at 0 and 18 hr (representative images are shown at 18 hr when the experiment was concluded). The wound area was quantitated using Tscratch and data are expressed as percent migration compared to Fc-treated VO (100%) cells. Data represent ± SEM of one experiment carried out in triplicate. (c) Invasion of 22Rv1 cells through collagen I was monitored in real-time using the Xcelligence system (Roche). The cell index indicates increased impedence at the electrodes which correlates with increased number of cells that have invaded through the collagen layer. ***VO + Fc vs. B4 + Fc, p < 0.001; *VO + Fc vs. VO + eB2, p = 0.018; #B4 + Fc vs. B4 + eB2, p =0.022. Data are shown ± SD from one experiment carried out in triplicate. (d) Transwell Matrigel invasion assays were used to determine MCF10A-B4 or -VO invasion in the presence of either Fc or 2 μg/ml soluble ephrin-B2/Fc (eB2). Filters were fixed and stained with DAPI and photomicrographs of nuclei in five independent fields on the underside of the filter were taken and counted. Data are presented as percentage invasion relative to VO Fc-treated cells. Data represent ± SEM of two independent experiments carried out in triplicate. (e) Ephrin-B2 alters spreading of 22Rv1-B4 cells on ColI. 22Rv1-VO or-B4 cells were treated with 5 μg/ml clustered ephrin-B2 and allowed to adhere to ColI coated glass slides for 1 hr. Cells were fixed and stained with phalloidin-TRITC to visualize the cytoskeleton. 22Rv1-VO cells show typical fibroblast-like morphology which is enhanced in 22Rv1-B4 cells. Treatment with ephrin-B2 induces some cell rounding in VO cells and is dramatically increased in EphB4 overexpressing cells. Scale bar 50 nm.

EphB4 promotes anchorage-independent growth, but ephrin-B2 opposes the effect

To further determine whether overexpression of EphB4 could induce a transformed phenotype in the nontransformed MCF10A cell line, cells were grown in soft agar and colony formation was observed. As reported for this cell line, MCF10A-VO cells showed little ability to grow under anchorage-independent conditions and only a few small colonies were visible in the untreated or Fc only treated cells (Fig. 4a). MCF10A-B4 untreated or Fc-treated cells showed a dramatic increase in colony size indicating that overexpression of EphB4 induces neoplastic changes in this cell line independent of ligand (Fig. 4a). When MCF10A-VO cells were treated with ephrin-B2/Fc there was a small but significant increase in colony size whereas MCF10A-B4 cells showed a significant decrease in colony size in the presence of ligand. This suggests that ephrin-B2-induced forward signaling in EphB4 overexpressing cells reverses the oncogenic-like effects of EphB4, but ephrin-B2 stimulation in nontransformed, low EphB4 expressing cells has a moderate positive effect. To investigate if EphB4 overexpression also increases metastatic potential in malignant cells, we analyzed 22Rv1-VO and -B4 cells in the soft agar assay. EphB4 overexpressing cells displayed a significant increase in colony size whereas treatment with ephrin-B2 completely reversed this effect (Fig. 4b) demonstrating that ephrin-B2 causes tumor suppression in EphB4 overexpressing cells.

Figure 4.

EphB4 overexpression enhances colony formation and ephrin-B2 reverses this effect. (a) MCF10A-B4 cells formed significantly larger colonies in soft agar when compared to VO cells. Treatment with ephrin-B2 significantly inhibited the EphB4-induced colony growth. In VO cells, treatment with ephrin-B2 significantly increased colony size compared to untreated or Fc-treated VO cells. (b) 22Rv1-B4 cells formed larger colonies than the VO cells when grown in soft agar for 21 days. Treatment over this time with soluble, clustered ephrin-B2 significantly reduced EphB4-induced colony growth. Photomicrographs show representative colonies. Data are presented as ± SEM of one experiment in triplicate. *p < 0.05, **p < 0.01, ***p < 0.001. B4, 22Rv1 or MCF-10A cells overexpressing EphB4; VO, 22Rv1 or MCF-10A cells transfected with the empty vector; eB2, soluble clustered ephrin-B2; Fc, Fc antibody alone.

Extended ephrin-B2 stimulation of EphB4 causes a significant reduction in cell growth and increases apoptosis

To determine the effect of extended ligand stimulation on cell growth, 22Rv1-B4 and-VO cells grown in 2D culture were stimulated with 2 μg/ml clustered ephrin-B2/Fc for 11 days. At several time-points growth was assessed using the MTT assay. There was little effect of EphB4 overexpression on cell proliferation in standard anchorage-dependent cell culture conditions (Fig. 5a). However, a significant decline in cell growth was noted by day 9 in the ephrin-B2-treated 22Rv1-B4 cells when compared to untreated 22Rv1-B4 cells (Fig. 5a, p < 0.001). Untreated or ligand-treated VO cells showed no significant difference in cell growth.

Figure 5.

Ephrin-B2 inhibits cell growth and causes apoptosis. (a) Stimulation of 22Rv1-B4 cells with 2 μg/ml soluble clustered ephrin-B2/Fc causes a significant decrease in cell growth compared to unstimulated 22Rv1-B4 cells after 9 days of treatment (***p < 0.001). There was no change of cell growth of VO cells in the presence of ephrin-B2. Data are normalized to day 1 of untreated VO cells and displayed as ± SD of two independent experiments carried out in at least triplicate. (b) Stimulation of MCF10A-B4 cells with 2 μg/ml soluble clustered ephrin-B2/Fc causes a significant decrease in cell growth compared to unstimulated MCF10A-B4 cells after 7 days of treatment (***p < 0.001). There was no change of cell growth of VO cells in the presence of ephrin-B2. Data are normalized to day 1 of untreated VO cells and displayed as ± SD of one experiment carried out in triplicate. (c,d,e) Stimulation of 22Rv1-B4 (c,d) or MCF10A-B4 (e) cells with 10 μg/ml soluble ephrin-B2/Fc caused a significant increase in apoptosis after 3 days compared to Fc-treated VO cells as determined by enhanced caspase-3/7 (c,e) and caspase-8 (d) activities. Fc-treated 22Rv1-EphB4 overexpressing cells showed a significant decrease in apoptosis as determined by a decline in both caspase-3/7 and -8 activities (c,d). Addition of ligand had no significant (ns) effect on the caspase-3/7 activity in 22Rv1-VO or MCF-10A-VO cells. Data are normalized to Fc-treated VO cells and displayed as ± SEM of at least one experiment carried out in triplicate. *p < 0.05, **p < 0.01, ***p < 0.001. B4, 22Rv1 or MCF10A cells overexpressing EphB4; VO, 22Rv1 or MCF10A cells transfected with the empty vector; eB2, soluble clustered ephrin-B2; Fc, Fc antibody alone.

In MCF10A-B4 cells, addition of ephrin-B2 led to a significant decrease in cell growth, first noted at day 7 compared to untreated cells (Fig. 5b). MCF10A-VO cells either treated or untreated showed no difference in cell growth. Together, these results indicate that ephrin-B2 stimulation of EphB4 overexpressing cells inhibits cell growth whereas EphB4 overexpression maintains proliferative signals.

To determine whether apoptosis was responsible for the decreased cell growth in response to ligand, cells were treated using clustered ephrin-B2-Fc and the caspase-3/7 and -8 activities were measured. 22Rv1-B4 cells treated with control Fc alone displayed a significant reduction in apoptosis compared to VO cells (caspase-3/7 activity, p < 0.05; caspase-8 activity, p < 0.01) (Figs. 5c and 5d), confirming previous reports that EphB4 is a survival factor in cancer cells.3 Stimulation of EphB4 overexpressing cells with soluble ligand significantly increased caspase 3/7 and 8 activity in the EphB4 overexpressing 22Rv1 cells compared to Fc only treated cells (Figs. 5c and 5d). There was no significant difference in apoptosis in 22Rv1-VO cells treated with ligand. In MCF10A-B4 cells, caspase 3/7 activity was also significantly increased when ephrin-B2 ligand was applied (Fig. 5e). These results demonstrate that EphB4 overexpression protects cells from apoptosis but receptor activation by ephrin-B2 ligand and subsequent forward signaling induces apoptosis.


We report here that overexpression of EphB4 in the nontransformed mammary epithelial line, MCF-10A, leads to conversion to a malignant phenotype. This is the first report to show that overexpression of EphB4 leads to transformation in MCF-10A cells in a similar manner to that described for ErbB2, H-Ras and EphA2.40, 41, 42 Furthermore, we also show that in 22Rv1 prostate cancer cells, introduction of EphB4 increases metastatic behavior. These ligand-independent effects of EphB4 can be opposed by treatment of EphB4-overexpressing cells with the ephrin-B2 ligand, which leads to decreased anchorage-independent growth, migration, invasion and proliferation, and increased apoptotic activity. Several studies have shown that removal of EphB4 by siRNA or antisense oligonucleotides is also tumor suppressive.2, 3, 6, 43 Additionally, targeting EphB4 with specific monoclonal antibodies that induce degradation of the receptor appears to be a promising avenue as demonstrated in a recent study in ovarian cancer.14 Combined with our results demonstrating that addition of ephrin-B2 to EphB4 overexpressing cells results in receptor degradation and subsequent tumor suppressive effects, it is conceivable that removal of EphB4 by different means is a beneficial therapeutic strategy.

It has been shown that the tumor suppressive function of EphB4 induced by ephrin-B2 involves the Abl/Crk pathway in breast cancer cells.28 Treatment of nonmalignant MCF-10A cells with a soluble EphB4/Fc to inhibit the interaction of EphB4 with ephrin-B2 and thus prevent ligand-dependent (i.e., tumor suppressive) forward signaling disrupted cell junctions and promoted colony formation in soft agar.28 In contrast, expression of a soluble form of EphB4 in A375 melanoma cells led to reduced proliferation in vitro and corresponded with a reduction in microvessel density in vivo; and soluble EphB4 administered to murine xenograft models reduced tumor growth,9, 30 presumably via manipulation of the ephrin-B2 reverse signaling pathway. These opposing effects highlight the complexity of the system and indicate cell context dependence.

Contradictory reports of increased EphB4 expression in some cancers, yet decreased expression in others, highlight the duality of EphB4 action depending on the cell context and the availability of ephrin-B2 ligand.3, 15, 44–46 A growing body of data now supports the hypothesis of a dual function for EphB4 as both a tumor promoter and suppressor.2, 6, 44, 45 This duality of EphB4 function may be regulated by the presence or absence of its sole, physiologically relevant ligand ephrin-B2 via ligand-mediated and ligand-independent signaling pathways, respectively,17 but few studies have formally tested this hypothesis with specific focus on the EphB4/ephrin-B2 interaction. To address this critical question we developed in vitro cell line models that stably overexpress EphB4 in the prostate cancer cell line 22Rv1, which produces minimal ephrin-B2 protein, and the nontransformed mammary epithelial cell line MCF-10A, which expresses significant amounts of ephrin-B2.28 Endogenous expression of ligand does not seem to be able to overcome the oncogenic effect of EphB4, as demonstrated by the results in MCF-10A cells. EphB4/ephrin-B2 interaction is considered to take place in trans, i.e., through interaction between opposing cell types47 and our results would support this notion.

We have shown that overexpressed EphB4 protein was localized to the plasma membrane and enriched in lipid rafts and that it is rapidly phosphorylated and internalized after the addition of exogenous soluble ephrin-B2. Localization of the overexpressed EphB4 in lipid rafts could facilitate the lateral dimerization/oligomerization of EphB4 and therefore potentially enhance both ligand-dependent and ligand-independent signaling.48 Ligand-dependent signaling normally requires the interaction of EphB4 with a cell that expresses ephrin-B2 and triggers both forward signaling through EphB4 and reverse signaling through ephrin-B2.49 Such interaction is important for maintaining cell–cell contacts. It is well known that cancer cells do not form stable cell–cell contacts, so it is likely that in those cancer cells where EphB4 and ephrin-B2 might be coexpressed, effective interaction between ligand and receptor may not occur.

In the absence of exogenous ligand, the overexpressing 22Rv1-B4 and MCF10A-B4 cells showed enhanced formation of colonies in anchorage-independent conditions and increased motility when compared to VO cells, thus clearly demonstrating ligand-independent functions characteristic of tumor promotion and metastasis. In contrast, overexpression of EphB4 in A375 melanoma cells did not affect anchorage-independent growth, but increased migration and adhesion to HUVECs demonstrating that EphB4's ligand-independent functions do not result in the same cell biological consequences in all cell types.50

Little increase in proliferation in EphB4 overexpressing cells grown in monolayer culture was observed, which suggests that overexpressed EphB4 does not have a major role in proliferation. In stark contrast, ligand stimulation of the 22Rv1-B4 and MCF10A-B4 cells for extended periods resulted in significant reduction in cell growth. This result confirms an earlier report where ephrin-B2 reduced cell growth in MDA-MB-435 breast cancer cells, but no mechanistic explanation was given, and it was not clear whether this effect was solely mediated by EphB4 ligand-dependent signaling.8 Our results indicate that the reduction in proliferation is potentially due to increased apoptosis in the ephrin-B2-treated cells as demonstrated by an increase in caspase-3/7 and -8 activities. In conclusion, our data supports the notion that ligand-dependent signaling through EphB4 expressed on cancer cells is tumor suppressive.


We thank Mr. Mark Adams for help with the preparation of lipid rafts and Dr. Leonore de Boer for assistance with the confocal microscopy.