Angiogenesis is a multistep process involving a diverse array of molecular signals. Ligands for receptor tyrosine kinases (RTKs) have emerged as critical mediators of angiogenesis. Three families of ligands, vascular endothelial cell growth factors (VEGFs), angiopoietins, and ephrins, act via RTKs expressed in endothelial cells. Recent evidence indicates that VEGF cooperates with angiopoietins to regulate vascular remodeling and angiogenesis in both embryogenesis and tumor neovascularization. However, the relationship between VEGF and ephrins remains unclear. Here we show that interaction between EphA RTKs and ephrinA ligands is necessary for induction of maximal neovascularization by VEGF. EphA2 RTK is activated by VEGF through induction of ephrinA1 ligand. A soluble EphA2-Fc receptor inhibits VEGF-, but not basic fibroblast growth factor-induced endothelial cell survival, migration, sprouting, and corneal angiogenesis. As an independent, but complementary approach, EphA2 antisense oligonucleotides inhibited endothelial expression of EphA2 receptor and suppressed ephrinA1- and VEGF-induced cell migration. Taken together, these data indicate an essential role for EphA receptor activation in VEGF-dependent angiogenesis and suggest a potential new target for therapeutic intervention in pathogenic angiogenesis.
Angiogenesis, the formation of new blood vessels from pre-existing vasculature, is a multistep process involving a diverse array of molecular signals. These include factors that stimulate endothelial cell proliferation, migration, and assembly, as well as recruitment of perivascular cells and extracellular matrix remodeling. Three families of receptor tyrosine kinases (RTKs) have emerged as critical mediators of angiogenesis; these are the vascular endothelial growth factor (VEGF), Tie, and Eph RTK families (1, 2). VEGF (VEGF-A/VEGF165) is a potent angiogenic factor in both embryonic development and in adult disease states, such as cancer. VEGF and its RTKs, Flt-1/VEGFR1 and Flk-1/KDR/VEGFR2, are required for the development and remodeling of blood vessels during embryogenesis (3–8). Moreover, VEGF signaling plays a crucial role in pathogenic angiogenesis, including the recruitment and maintenance of tumor vasculature, and blocking VEGF function using soluble receptors, neutralizing antibodies, or pharmacologic inhibitors significantly abrogates tumor angiogenesis and progression (9–14). Recent evidence indicates that VEGF cooperates with other angiogenic factors, such as angiopoietins and their RTK, Tie2, in regulating vascular remodeling and growth in tumors (15–17). However, the relationship between VEGF and ephrins remains unclear.
The Eph family of RTKs and their ligands, originally identified as critical determinants of embryonic patterning and neuronal targeting (18), also regulates vascular development (1, 2, 19). Targeted disruption of ephrinB2, EphB4, or EphB2/EphB3 results in embryonic lethality due to defects in primary capillary network remodeling and subsequent patterning defects in the embryonic vasculature (20–22), suggesting that Eph RTKs and their ligands are critical for vascular development during embryogenesis. The A class ligand, ephrinA1, has also been implicated in angiogenesis. EphrinA1 was originally identified as a tumor necrosis factor α (TNF-α)-inducible gene in human umbilical vein endothelial cells (HUVECs) (23), and is expressed in the developing vasculature during embryogenesis (24). Moreover, ephrinA1 induces endothelial cell migration and capillary assembly in vitro, and angiogenesis in a corneal pocket assay in vivo (25, 26), suggesting a role in neovascularization of adult tissues. Indeed, expression of ephrinA1 and its receptor EphA2 was observed in breast tumors and associated vasculature (27), and blocking EphA receptor activation impaired tumor angiogenesis (28). These studies indicate that Eph signaling is critical for normal blood vessel development as well as pathogenic angiogenesis.
In this study, we provide evidence that Eph RTKs and their ligands are necessary for induction of maximal angiogenesis by VEGF. We show that ephrinA1 is a downstream target gene product induced by VEGF. Blocking EphA receptor activation inhibits VEGF-dependent endothelial cell migration, sprouting, and survival in vitro and angiogenesis in vivo. These effects are specific to VEGF, as blocking EphA class signaling has no impact of fibroblast growth factor (FGF)-mediated angiogenesis. As VEGF is a critical mediator of angiogenesis in pathogenic events such as cancer, our data suggest a potential new target for anti-angiogenic therapy.
A Soluble EphA2-Fc Chimeric Receptor Inhibits VEGF-, but not FGF-induced Angiogenesis
Given that both VEGF and ephrins regulate angiogenesis, and that VEGF is known to cooperate with other angiogenic factors to modulate blood vessel formation, we investigated whether EphA receptor activation is required for VEGF-mediated angiogenesis. To achieve this goal, we have utilized a soluble chimeric protein, EphA2-Fc, as a blocking reagent. This approach was taken because mutant mice deficient for individual Eph family members are either embryonic lethal or do not display overt phenotypes (20–22, 29). The EphA2-Fc soluble receptor fusion proteins consist of extracellular domain of the EphA2 receptor and the Fc portion of the human immunoglobulin IgG1. Because there is promiscuous binding between Eph receptors and ligands within the same subclass, this soluble receptor variant prevents multiple ephrinA class ligand interactions with endogenous receptors, effectively blocking signaling through class A Eph receptors. As shown in Fig. 1A, endothelial cells treated with ephrinA1 in the presence of excess EphA2-Fc showed a dramatic reduction in endogenous EphA2 receptor phosphorylation as compared to cells treated with ephrinA1 alone. However, EphA2-Fc does not affect the phosphorylation of VEGFR2 (Flk-1) (Fig. 1B), demonstrating the specificity of EphA2-Fc to class A ephrins. In the mouse cornea, soluble EphA2-Fc also markedly inhibited the angiogenic response induced by ephrinA1 (Fig. 1C), showing the efficacy of this reagent in blocking A class Eph receptor activation and neovascularization in vivo.
We next investigated whether ephrin/Eph A class signals are required for VEGF-dependent angiogenesis. To do this, we used a well-established assay in which hydron pellets impregnated with either test or control proteins are implanted into mouse corneal pockets (30). Angiogenesis induced by exogenous factors in the normally avascular cornea was then documented and quantified. Consistent with previous observations, ephrinA1, VEGF, and basic fibroblast growth factor (bFGF) each induced corneal neovascularization (Figs. 1C and 2, A and B) (16, 25, 31). To determine if VEGF function is dependent upon signaling by ephrinA class ligands, we implanted hydron pellets impregnated with VEGF or bFGF in the presence or absence of EphA2-Fc. Soluble EphA2-Fc receptor itself does not induce angiogenesis in the cornea. Rather, the addition of EphA2-Fc in the hydron pellet markedly inhibited the angiogenic response induced by VEGF (Fig. 2A). This effect is specific, as EphA2-Fc did not affect angiogenesis induced by bFGF, another potent inducer of angiogenesis (Fig. 2B) (32). These data suggest that EphA receptor activation is specifically required for VEGF-induced angiogenesis in vivo.
Soluble EphA2-Fc Receptors Inhibit VEGF-induced Endothelial Sprouting
Sprouting of new capillaries from pre-existing blood vessels is a hallmark of angiogenesis, and VEGF has been shown to induce sprouting activity in an in vitro capillary sprouting assay (32). To determine whether VEGF-induced sprouting activity is affected by blocking EphA receptor activation, we cultured endothelial cells on collagen-coated beads, and assessed sprouting in fibrin gels in response to ephrinA1-Fc, VEGF, bFGF, or VEGF or bFGF in the presence of EphA2-Fc or control Fc proteins, or medium alone. As shown in Fig. 3, endothelial cells extended from collagen-coated beads in response to ephrinA1, VEGF, bFGF, or VEGF or bFGF plus control Fc proteins. In contrast, EphA2-Fc inhibited sprouting activity induced by VEGF, but not FGF, suggesting that VEGF-induced endothelial cell sprouting requires EphA class receptor function. Taken together, our findings revealed that activation of EphA receptor is required for VEGF-mediated endothelial cell sprouting, an essential step of angiogenic process.
Soluble EphA2-Fc Receptors Inhibit VEGF-Induced Cell Survival but Do Not Affect Cell Proliferation
To determine which steps of VEGF-induced angiogenesis require EphA receptor activation, we performed a series of in vitro angiogenesis assays. First, we tested whether blocking EphA receptor function affects VEGF-induced cell proliferation. Endothelial cells were treated with VEGF or bFGF in the presence or absence of EphA2-Fc, and cell growth was measured by [3H]thymidine incorporation. Consistent with the observation that ephrin signaling does not affect endothelial cell growth, ephrinA1-Fc did not affect cell proliferation, nor did EphA2-Fc impair VEGF- or bFGF-induced cell proliferation significantly (Fig. 4A) (1, 18).
To determine the effect of EphA signaling in VEGF-induced cell survival, endothelial cells were growth factor/serum starved to induce apoptosis. At the time of growth factor and serum withdrawal, endothelial cells were treated with VEGF, bFGF, or VEGF or FGF in the presence of EphA2-Fc or control Fc proteins. Compared to cells treated with VEGF and VEGF plus Fc control, cells treated with VEGF plus EphA2-Fc showed a significant reduction in cell survival, as determined by TUNEL assay (Fig. 4B) and trypan blue exclusion (data not shown). However, bFGF-induced cell survival was not affected by EphA2-Fc soluble receptor treatment. Taken together, these data suggest that EphA class signaling is not required for the mitogenic activity of VEGF, but is required for VEGF-mediated survival of endothelial cells.
EphA2 Antisense Oligonucleotides or Soluble EphA2-Fc Receptors Inhibit VEGF-Induced Endothelial Cell Migration
To test the effect of EphA2-Fc on VEGF-mediated endothelial cell migration, we used both “wound closure” and modified Boyden chamber assays. As shown in Fig. 5A, the rate of endothelial migration to close a circular wound in a confluent endothelial cell monolayer was increased in serum-free medium supplemented with ephrinA1-Fc, VEGF, bFGF, or VEGF or bFGF plus Fc control protein. In contrast, EphA2-Fc significantly reduced the migration of endothelial cells into the “wounded” area in response to VEGF, but not bFGF. Consistent with “wound closure” data, excess EphA2-Fc also significantly inhibited migration of endothelial cells in response to VEGF in a modified Boyden chamber assay (Fig. 5B).
As an independent, but complementary, approach to determine the role of EphA receptor activation in VEGF-induced angiogenesis, we utilized a published antisense oligonucleotide to specifically inhibit EphA2 receptor expression (34). Endothelial cells transfected with antisense oligonucleotides show approximately a 3-fold reduction in endothelial endogenous EphA2 receptor expression, whereas EphA2 levels in cells transfected with control inverted antisense oligonucleotides are not significantly affected (Fig. 5C). Inhibition of EphA2 receptor expression results in a significant reduction in ephrinA1- and VEGF-induced migration in the wound closure assay (Fig. 5D) and in the modified Boyden chamber assay (Fig. 5E), indicating that EphA2 receptor is required for VEGF-induced endothelial cell migration.
VEGF Induces Endothelial Expression of EphrinA1 Ligand and Phosphorylation of EphA2 Receptor
Given that EphA receptor activation is required for several steps in VEGF-dependent angiogenesis, we investigated whether ephrinA1, the ligand for EphA2 receptor, is a downstream target gene induced by VEGF signaling. EphrinA1 was previously shown to be induced by TNF-α, and to mediate TNF-α-induced corneal angiogenesis (25). As shown in Fig. 6, like TNF-α, VEGF induced ephrinA1 expression in both HMEC-1 and HUVEC endothelial cells, as judged by Northern blot (Fig. 6A) and Western blot analyses (Fig. 6B).
Because ephrinA1 is angiogenic, this result suggests that the induction of ephrinA1 and subsequent activation of the EphA receptors could be partially responsible for the angiogenic effects of VEGF. To address this possibility directly, we asked whether VEGF could induce EphA2 phosphorylation. As shown in Fig. 6C, exposure of endothelial cells to VEGF resulted in phosphorylation of EphA2 receptor. This activation of receptor was apparently due to the induction of ephrinA1 because treatment of cells with VEGF in the presence of soluble EphA2-Fc receptor resulted in inhibition of endogenous EphA2 receptor phosphorylation (Fig. 6C). Addition of a control Fc protein had no effect on EphA2 phosphorylation or ephrinA1 expression (data not shown). Thus, the level of expression of ephrinA1 correlated directly with the extent of phosphorylation of endogenous EphA2 receptor, whereas the absolute amount of EphA2 remained unchanged. Taken together, these data demonstrate that VEGF induces ephrinA1 expression in endothelial cells, suggesting a juxtacrine mechanism for activation of EphA receptor signaling.
The Eph family of RTKs is one of three major families of RTKs, which also include the Tie2 and VEGF receptor families, that regulate blood vessel formation. Recent evidence indicates that VEGF cooperates with angiopoietins, the ligands for Tie2 receptors, in regulating angiogenesis, vascular remodeling, and growth in tumors (15, 17). Our data demonstrate that ephrins also act in concert with VEGF in promoting angiogenesis. Several lines of evidence support this conclusion. First, VEGF induces endothelial expression of ephrinA1 and phosphorylation of endogenous EphA2 receptor. Second, VEGF-induced endothelial cell migration, survival, and sprouting are inhibited by a soluble EphA2-Fc receptor in vitro. Third, blocking EphA2 receptor expression by antisense oligonucleotides also inhibits VEGF-induced endothelial cell migration, implying the role of EphA2 receptor in VEGF-induced angiogenesis. Fourth, blocking EphA receptor activation inhibits VEGF-, but not FGF-induced angiogenesis in corneal assays. Taken together, these results provide evidence for a specific link between the ephrinA1/EphA2 and the VEGF pathways in formation of new blood vessels.
How does inhibition of EphA RTK activation lead to suppression of VEGF-induced angiogenesis? One hypothesis is that EphA signaling may be a part of VEGF signaling cascade to regulate angiogenesis. In support of this hypothesis, we showed that VEGF induced endothelial ephrinA1 expression and phosphorylation of EphA2 receptor (Fig. 6). This activation of receptor was apparently due to the induction of ephrinA1 because treatment cells with VEGF in the presence of soluble EphA2-Fc receptor resulted in inhibition of endogenous EphA2 phosphorylation. Thus, these data support a hypothesis that VEGF induces ephrinA1 expression in endothelial cells and subsequent activation of EphA receptor signaling to promote angiogenesis through a juxtacrine mechanism.
A second hypothesis is that VEGF-induced signaling and ephrinA1-induced signaling are two separate pathways, and activation of EphA RTK signaling positively regulates VEGF signaling pathways to promote angiogenesis. Thus, soluble EphA2-Fc receptor could affect VEGF signaling at several steps. First, EphA2-Fc could affect VEGF receptor phosphorylation, which subsequently affects several VEGF-induced cellular responses. However, as shown in Fig. 1B, we found no effects of EphA2-Fc on VEGFR-2 phosphorylation in response to VEGF stimulation. Second, VEGF is known to inhibit endothelial cell apoptosis by activation of Akt via a PI3 kinase-dependent pathway (35). Blockade of EphA receptor binding to ephrinA ligand could thus affect VEGF-induced PI3 kinase/Akt activation, leading to increased apoptosis. Third, activation of FAK or p38MAP kinase is shown to be required for VEGF-induced cell migration (36). Inhibition of EphA receptor activation either by soluble EphA2-Fc or antisense oligonucleotide treatment could affect FAK/p38 MAP kinase activity, resulting in inhibition of cell migration. Further experiments are required to determine whether blocking EphA receptor activation affects VEGF-induced activation of PI3 kinase/Akt and/or FAK/p38 MAP kinase pathways.
Although there is a bidirectional signaling between both class A and class B Eph receptors and ligands (37–40), it remains unclear whether soluble EphA2-Fc could signal to ephrinA1 ligand on the surface of endothelial cells. Based on our results, EphA2-Fc neither elicit endothelial cell responses above the controls in vitro, nor induce corneal angiogenesis in vivo. However, in principle, soluble EphA2-Fc could inhibit VEGF-induced angiogenesis through ephrinA1-mediated reverse signaling. If this is the case, then reduction of EphA receptor expression is not expected to affect VEGF-induced responses greatly. Because the EphA2 antisense oligonucleotides and soluble EphA2-Fc receptor have similar effects on VEGF-induced cell migration (Fig. 5), the effect of EphA2-Fc is probably due to the inhibition of ephrinA-EphA ligand receptor interaction.
Recent studies from our laboratory and others suggest that EphA receptor activation plays a critical role in tumor angiogenesis. Elevated expressions of ephrinA1 and EphA2 were observed in various tumors and associated vasculature including colon, breast, and pancreatic carcinoma (27). Furthermore, in a separate manuscript, we reported that EphA receptor activation is required for tumor neovascularization in two tumor models (28). These data suggest that ephrinA1 could function to induce tumor angiogenesis through both juxtacrine and paracrine mechanisms. As VEGF has been shown to be a key regulator in tumor angiogenesis, EphA signaling may be an important mediator of VEGF-dependent tumor angiogenesis. Thus, it is possible that high levels of VEGF expression in tumor cells could induce ephrinA1 expression in adjacent endothelial cells. Engagement of ephrinA1 to EphA2 receptor may then activate juxtacrine signaling to regulate changes in endothelial cell survival and migration to promote angiogenesis. Alternatively, tumor cells may also recruit new blood vessels directly through paracrine signaling between ephrinA1 expressed on tumor cells and EphA2 expressed on surrounding endothelial cells.
In summary, we provide the first evidence that class A Eph RTKs play a critical role in VEGF-dependent angiogenesis. Thus, EphA class receptors and ephrinA ligands may provide potential novel targets for therapeutic intervention in diseases associated with pathogenic angiogenesis.
Materials and Methods
The EphA2-Fc soluble receptor cDNA construct was provided by Regeneron Inc. (Tarrytown, NY) and subcloned into episomal expression vector pCEP4 (Invitrogen, San Diego, CA). pCEP4/EphrinA1-Fc expression vector was provided by Dr. A. Pandy (University of Michigan, Ann Arbor, MI). Recombinant EphA2-Fc and ephrinA1-Fc proteins were either purified from culture supernatant of stable 293T clones expressing these factors using protein A-Sepharose column, or purchased from R&D Systems, Inc. (Minneapolis, MN). The rabbit polyclonal antibody against the nonconserved spacer region of ephrinA1 was provided by Immunex Inc. (Seattle, WA). Anti-EphA2 monoclonal antibody clone D7 and anti-phosphotyrosine monoclonal antibody 4G10 were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). A monoclonal anti-VEGFR-2 antibody (1A8) was kindly provided by Dr. R. Brekken (The Hope Heart Institute, Seattle, WA). Primary endothelial cells HUVEC and HMEC were purchased from Clonetics (Walkersville, MD).
Immunoprecipitation and Western Blot Analysis
EA926 endothelial cells were serum starved overnight and stimulated for 20 min with 0.5 μg/ml ephrinA1-Fc in the presence or absence of 0.5–5 μg/ml EphA2-Fc in serum-free media. Serum-starved HUVECs were stimulated for 15 min with 20 ng/ml VEGF in the presence or absence of 5 μg/ml EphA2-Fc. Cells were lysed in a buffer containing: 1× PBS, 1% Ipegal CA-630 (Sigma, St. Louis, MO), 0.5% sodium deoxycholate, 0.1% SDS with 100 μm of protease and phosphatase inhibitors: aprotinin, phenylmethylsulfonyl fluoride, sodium orthovanadate, and leupeptin. Cell lysates were cleared by sonication and centrifugation. Endogenous EphA2 receptors were immunoprecipitated using either an ephrinA1-Fc protein or an anti-EphA2 monoclonal antibody clone D7 (Upstate Biotechnology, Inc.). Immunoprecipitated proteins were fractionated on 8% SDS-PAGE, transferred to nitrocellulose membrane, and blotted with anti-phosphotyrosine antibodies 4G10, according to manufacturer's instructions. The blots were stripped and reprobed with anti-EphA2 monoclonal antibodies as loading control.
Mouse Corneal Angiogenesis Assay
Mouse corneal angiogenesis assays were performed as described previously (30). Briefly, hydron pellets containing sucralfate with either vehicle alone (PBS or IgG), bFGF, VEGF-A, ephrinA1-Fc, or EphA2-Fc were prepared. Pellets were surgically implanted into corneal micropockets created 1 mm to the lateral corneal limbus of C57/BL6 mice. At day 5 postimplantation, corneas were photographed at an incipient angle of 35–50° from the polar axis in the meridian containing the pellet, using a Zeiss split lamp. The fraction of the total corneal image that was vascularized (VA), and the ratio of pixels marking neovascular capillaries both within the vascularized region (RVD) and within the total corneal image (TVD) were calculated using Bioquant software (Vanderbilt University, Nashville, TN). Statistical analysis was performed by two-tailed, paired Student's t test.
Cell proliferation was measured by [3H]thymidine incorporation. Briefly, HUVECs were grown to 80% confluency in 48-well plates (Nalgene NUNC International, Rochester, NY) and starved for 16 h in growth factor-deprived media. Cells were either incubated in medium alone, or stimulated for 24 h with 5–10 μg/ml ephrinA1-Fc, 20 ng/ml VEGF, or 20 ng/ml bFGF in the presence of 10 μg/ml EphA2-Fc or Fc control protein, in growth factor-deprived media. After 24 h, cells were treated with 1 μCi [3H]dTh for an additional 24 h, harvested by trypsinization and transferred to 3 mm Whatmann filter paper. [3H]dTh incorporation per well was determined using a beta scintillation counter.
HUVECs were cultured to near confluency on silanized slides (Nalge NUNC, Rochester, NY) and starved for 16 h in growth factor-deprived media. Cells were subsequently incubated with 5–10 μg/ml ephrinA1-Fc, 20 ng/ml VEGF, or 20 ng/ml bFGF in the presence of 10 μg/ml EphA2-Fc or Fc control protein, in growth factor-deprived media for 48 h. Cells were then fixed, subjected to TUNEL assay using an ApopTag Red in situ apoptosis detection kit (Intergen Co., Purchase, NY), and counterstained with DAPI (0.05 μg/ml in PBS, Sigma-Aldrich, St. Louis, MO). Apoptotic cells were quantified by numerating rhodamine-stained nuclei in six random 20× fields per sample using Scion Image analysis software. Two samples per experiment were analyzed. Experiments were repeated three times, and results were pooled. Percentage of apoptosis was calculated as the number of rhodamine-stained nuclei/total number of DAPI-stained nuclei. Statistical analysis was performed using the two-tailed, paired Student's t test.
Endothelial cell migration was determined by two independent assays: a modified Boyden chamber assay (9) and a “wound closure” assay (41). For “wound closure” assays, replicate circular “wounds” were generated in confluent human dermal microvascular endothelial cell (HMEC-1, passages 10–12) monolayers using a silicon-tipped drill press. Serum-free media were then supplemented with the indicated factors at the time of wounding. Residual fractional “wound” areas were measured at 0, 2, 4, 8, and 10 h using Bioquant software (Vanderbilt University). Mean fractional residual areas of three wounds were calculated, reflecting migration rates. Experiments were repeated three times and results were pooled. Statistical analysis was performed by two-tailed, paired Student's t test. Briefly, for modified Boyden chamber, polycarbonate filter wells (Transwell, Costar, 8 μm pore, VWR Scientific, West Chester, PA) were coated with 50 μg/ml fibronectin (Sigma-Aldrich), 0.1% gelatin in PBS for 30 min at room temperature, followed by equilibration into DMEM/0.1% BSA at 37°C for 1 h. Growth factor-deprived HUVECs (Clonetics, passages 4–6, 1 × 105) were plated in the upper chamber of the filter well and allowed to migrate toward the undersides of the filters in the bottom chamber containing serum-free media supplemented with 5–10 μg/ml ephrinA1-Fc, VEGF (20 ng/ml), bFGF (20 ng/ml), or VEGF or bFGF in the presence of EphA2-Fc (5 or 10 μg/ml) or control Fc protein. After 5 h, cells adhering to the top of the transwell were removed with a cotton swab, and the cells that had migrated to the underside of the filter were fixed and stained. For quantification, 10 high-power fields were counted on each filter and triplicate filters were counted per experiment. Experiments were repeated twice and data were pooled. Statistical analysis was performed by two-tailed, paired Student's t test.
Endothelial Cell Sprouting Assay
Briefly, HMEC-1 (passages 9–11) were grown to confluency on collagen-coated Cytodex 3 beads (Amersham-Pharmacia, Piscataway, NJ) for 5–7 days using EBM-supplemented media (Clonetics). The beads were plated in a gel matrix containing 5.46 mg/ml fibrinogen (Sigma-Aldrich), 2 units/ml thrombin (Sigma-Aldrich), DMEM/2% fetal bovine serum and the following test and control proteins: 5–10 μg/ml ephrinA1-Fc, 20 ng/ml VEGF, VEGF plus 10 μg/ml Fc, and VEGF plus 10 μg/ml EphA2-Fc. Serum-free media supplemented with test and control proteins were added to the gel matrix every 2 days. Photographs were taken at days 3 and 4 using a Nikon 35-mm camera and Ectachrome 100 film. The number of endothelial cell sprouts exceeding the diameter of the bead was determined for every 40 beads counted per experiment, as described (33). Experiments were repeated three times, and results were pooled and analyzed for statistical significance using two-tailed, paired Student's t test.
Northern Blot and Western Blot Analyses
HMEC-1 (42) or HUVECs (Clonetics) were grown to near confluency and starved in endothelial cell basal medium (Clonetics) with 2% fetal bovine serum overnight. Cells were preincubated with 10 μg/ml cycloheximide for 30 min and stimulated with 20 ng/ml VEGF for 0, 2, 5, and 16 h. RNA from stimulated cells were isolated using Trizol reagent (Life Technologies, Inc., Rockville, MD). Ten micrograms of total RNA were resolved on a 1% agarose gel containing 2.2 m formaldehyde, transferred to a nylon membrane, and probed with an ephrinA1 cDNA fragment. Membranes were washed once with 2× SSC at room temperature, followed by two stringent washes (1× SSC at 65°C for 10 min and 0.5× SSC at 65°C for 10 min). The RNA blots were stripped and reprobed with a GAPDH cDNA probe as loading control.
HMEC-1 and HUVECs were maintained in 10-cm dishes (unless otherwise indicated) in endothelial cell basal medium with growth factors (Clonetics). HMEC-1 cells were used to passage 13, and HUVECs to passage 5. Endothelial cells were cultured to 80% confluency, starved for 16 h in medium deficient in growth factors and stimulated with 20 ng/ml VEGF for 0, 2, 5, and 16 h. Cells were lysed in SDS sample buffer, and cell extracts were cleared by sonication and centrifugation. Fifty micrograms of protein from cell lysates were fractionated on 12% SDS-PAGE, and transferred to a nitrocellulose membrane (ECL+, Amersham Pharmacia Biotech, Piscataway, NJ). Membranes were blocked at room temperature for 1 h in Tris-buffered saline containing 0.1% Tween 20 and 5% powdered dry milk, and then incubated with ephrinA1-specific rabbit polyclonal antibodies (P2, specifically against the spacer region of ephrinA1, Immunex) for 1 h. Immunoreactive proteins were detected with anti-rabbit secondary antibodies (Amersham Pharmacia) conjugated to horseradish peroxidase, using enhanced chemiluminescence (Amersham Pharmacia).
Oligonucleotide Transfection of HUVEC Cells
To inhibit the expression of EphA2, HUVECs were transiently transfected with phosphorothioate-modified antisense oligonucleotides (5′-CCAGCAGTACCACTTCCTT GCCCTGCGCCG-3′), or control inverted antisense oligonucleotides (5′-GCCGCGTCCCGTTCCTTCACCATGACGACC-3′) as described (33). Briefly, HUVECs (passages 5–6) were grown at 50% confluency, and transfected with 100 nm of oligonucleotide using oligofectamine reagent (Life Technologies, Inc.) for 4 h at 37°C. Forty-eight hours after transfection, cells were analyzed for effects on VEGF-induced migration by “wound closure” and Boyden chamber assays, as described above. Transfected cells were also analyzed for expression of EphA2 by Western blot analysis using anti-EphA2 monoclonal antibody (D7, Upstate Biotechnology, Inc.).
We thank Erin Thompson for excellent technical assistance, and B. L. M. Hogan, L. M. Matrisian, R. DuBois, M. Boothby, J. W. Thomas, G. Miller, and R. S. Muraoka for helpful discussions and comments on the manuscript. Special thanks to A. Pandey for providing pCEP-4/ephrinA1-Fc construct, R. Brekken for generous gift of anti-VEGFR-2 antibodies, N. Boudreau for providing detailed protocol of endothelial sprouting assay, and A. Pozzi for helping with Scion image analysis.
NIH Grants HD36400 and DK47078; JDF grant I-2001-519; DOD grant BC010265; American Heart Association Grant 97300889N; ACS Institutional Research Grant IN-25-38 (to J. Chen); Vascular Biology Training Grant T32-HL-07751-06 and American Heart Association Fellowship 0120147B (to D. Brantley); Cancer training Grant T-32 CA09592 (to N. Cheng); and a core facilities Grant 2P30CA68485 to the Vanderbilt-Ingram Cancer Center.