A migratory role for EphrinB ligands in avian epicardial mesothelial cells


  • Sonia M. Wengerhoff,

    1. Northwestern University, Feinberg School of Medicine, Department of Pediatrics, Chicago, Illinois
    Search for more papers by this author
  • Amy R. Weiss,

    1. Northwestern University, Feinberg School of Medicine, Department of Pediatrics, Chicago, Illinois
    Current affiliation:
    1. Newton-Wellesley Hospital, 2014 Washington Street, Newton, MA 02462
    Search for more papers by this author
  • Kathryn L. Dwyer,

    1. Northwestern University, Feinberg School of Medicine, Department of Pediatrics, Chicago, Illinois
    Search for more papers by this author
  • Robert W. Dettman

    Corresponding author
    1. Northwestern University, Feinberg School of Medicine, Department of Pediatrics, Chicago, Illinois
    2. Children's Memorial Research Center, Developmental Biology Core, Chicago, Illinois
    • 303 E. Chicago Avenue, Chicago, IL 60611
    Search for more papers by this author


Little is known about the molecules that mediate the attachment of proepicardial cells to the heart. Ephrins are cell surface ligands for Eph tyrosine kinase receptors, molecules known to play a role in cell adhesion and migration. Here, we detected EphrinB ligands in proepicardial and epicardial mesothelial cells (EMCs) using reverse transcriptase-polymerase chain reaction, immunoblotting, immunolocalization, and EphB-Fc binding. Aggregated EphB-Fc fragments clustered ephrinB1 ligands on living EMCs indicating that they are cell surface expressed. In vitro assays demonstrated that ephrinB ligands participate in EMC migration but not cell adhesion. Localization studies in hearts at Hamburger and Hamilton stage 30 and older revealed that ephrinB1 is expressed in the epicardium and subepicardial mesenchyme of the atrioventricular sulcus. EMCs treated with platelet-derived growth factor-BB expressed smooth muscle markers but not ephrinB1. Our study supports an early role for ephrinB ligands for migration of epicardial cells and a later role in perivascular fibroblasts of coronary vessels in the atrioventricular sulcus. Developmental Dynamics 239:598–609, 2010. © 2009 Wiley-Liss, Inc.


The coronary vessels are derived from the mesoderm of the septum transversum. At about the time the early heart loops, a group of cells posterior to the sinoatrial region of the heart coalesce into a villous protrusion called the proepicardial organ (PE; Virágh and Challice,1981; Schulte et al.,2007). The PE grows cranially toward and attaches to the inner curvature of the heart (Manner,1993; Viragh et al.,1993; Nahirney et al.,2003). Outgrowth of the PE appears to be triggered by localized liver-derived factors, which may include fibroblast growth factors (FGFs) and is simultaneously inhibited by coexpression of bone morphogenetic protein-2 (BMP2) and FGF2 at the base of the PE (Kruithof et al.,2006; Ishii et al.,2007). Once attached to the heart tube, PE-derived epicardial mesothelial cells (EMCs) migrate both superficially toward the outflow tract and invasively into the subepicardium and myocardium. The invasive step is generally referred to as epicardial–mesenchymal transformation (EMT) and the cells are called epicardially derived mesenchymal cells (EPDCs). EMT is triggered by several factors including transforming growth factor-beta (TGFβ), FGF, and platelet-derived growth factor (PDGF; Compton et al.,2006; Olivey et al.,2006; Mellgren et al.,2008; Sridurongrit et al.,2008; Tomanek et al.,2008) and inhibited by soluble VCAM-1 (Dokic and Dettman,2006).

After invading, EPDCs participate in the formation of several different components of the adult heart, including coronary vessels, cardiac fibroblasts of the myocardial interstitium, and mesenchymal tissues of the valvuloseptal complex (Perez-Pomares et al.,1997,1998,2002; Dettman et al.,1998; Gittenberger-de Groot et al.,1998). During epicardial formation endothelial and blood cells begin to appear in the subepicardial mesenchyme. Early vessels grow and remodel into first an endothelial plexus through vasculogenesis and subsequently into a mature vessel wall through a process that is likely to involve vascular endothelial growth factor-A (VEGF-A; Tomanek et al.,2006; Nesbitt et al.,2009). Despite an ever growing understanding of the movements and fates of EMCs, precious little is understood about how these events are regulated at the molecular level.

The Eph tyrosine kinases are a family of receptors that play an important role in organogenesis and vascular development along with their cell-surface ligands the ephrins (Cowan and Henkemeyer,2002). In mice, ephrinB2 is primarily expressed in cells of arteries, including endothelial (Wang et al.,1998; Hayashi et al.,2005), smooth muscle, and adventitial cells (Gale et al.,2001; Shin et al.,2001). Genetic studies in mice have demonstrated that ephrinB2 functions in endothelium to facilitate angiogenic remodeling (Wang et al.,1998; Adams et al.,1999; Gerety et al.,1999) and in smooth muscle and adventitial cells for recruitment to the vessel wall (Foo et al.,2006). These studies implicated reciprocal repulsive interactions with the EphB4 receptor, which is expressed predominantly in veins. Other EphB receptors and ephrinB ligands have been implicated in systemic vascular formation of the mouse (Adams et al.,1999). In vitro, ephrinB ligands promote endothelial capillary-like assembly, cell attachment, and angiogenesis of endothelial cells (Daniel et al.,1996; Zhang et al.,2001). Given the importance of Eph/ephrin signaling in vascular development, we postulated that these molecules are likely to play a role in coronary vascular development.

Here, we present evidence that, in avian embryos, ephrinB ligands participate in epicardial and coronary vascular development. We found that the PE expresses transcripts for both ephrinB1 and ephrinB2. Proteins and transcripts are present in the epicardium and then become limited to perivascular fibroblasts of the atrioventricular sulcus. Using in vitro migration and cell adhesion assays, we present evidence that ephrinB ligands are likely to play a role in migrating PE cells but not adhesion. Thus, like other developing vascular systems, the heart uses ephrinB ligands during blood vessel development.


Expression of ephrinb1 and EphrinB2 Ligands in Proepicardium and Epicardium

While three mammalian ephrinB ligands have been identified, there are only two ephrinB orthologs in the chick genome database: ephrinB1 and ephrinB2 (Hubbard et al.,2005). To determine if ephrinB transcripts are expressed in the PE, we performed reverse transcriptase-polymerase chain reaction (RT-PCR) from total RNA isolated from Hamburger and Hamilton stage (HH) 16 chick PEs (Fig. 1A). We observed ephrinB1 and ephrinB2 amplification products from HH16 PE and cultured HH24 epicardial total RNA (not shown) indicating that these transcripts are expressed in the PE and epicardium. To determine if superficial epicardium expressed ephrinB ligands, we isolated total protein from HH24 epicardial cultures and analyzed it using immunoblotting (Fig. 1B). Using an antiserum that recognizes a C-terminal epitope common to all ephrinB isoforms (pan anti-ephrinB), we observed two bands of sizes consistent with ephrinB1 (50 kDa) and ephrinB2 (48 kDa). This antiserum recognized a band of the same mobility in mouse lung total protein, used as a positive control for ephrinB1. Using an antiserum that recognizes an internal epitope in the extracellular domain of human ephrinB1, we observed a band at 50 kDa, consistent with the predicted molecular weight for chicken ephrinB1 and consistent with the mobility of the band in the mouse lung extract. These observations supported the hypothesis that the developing chicken epicardium expresses ephrinB ligands.

Figure 1.

Expression of ephrinB ligands in avian epicardium. A: Reverse transcriptase-polymerase chain reaction (RT-PCR; 35 cycles) of ephrinB transcripts in total RNA isolated from dissected Hamburger and Hamilton stage (HH) 16 PEs. B: Immunoblots of total protein isolated from chick epicardial mesothelial cells (EMCs) cultured from HH24 hearts probed with pan anti-ephrinB (left) and anti-human ephrinB1 (right). Total protein from mouse lung extract (MLE) was used as a positive control for ephrinB1. C: Confocal microscopy of cultured chick EMCs stained with anti-ephrinB1. Dotted line indicates the leading edge of the migrating monolayer. Arrows indicate general direction of migration. D,E: Confocal microscopy of cultured chick EMCs stained with pan anti-ephrinB at the leading edge of the monolayer (D) and several cell lengths behind (E). Arrowheads indicate areas where intercellular junctions are visible in (D). F: Confocal image of an EMC monolayer incubated with goat anti-rabbit Alexa 488 as a control. Inset shows DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride) staining of the same cells. G: Confocal image of an EMC monolayer stained with rabbit anti-human ZO1. H–J: Confocal images of 0.32-μm optical sections through an EMC nucleus stained with DAPI (H), Wilm's Tumor factor 1 (I) and pan anti-ephrinB (J). Scale bars = 50 μm in C, 20 μm in D–G, 10 μm in H–J.

Localization of EphrinB Ligands in Epicardium

We stained epicardial monolayers migrating from explanted HH24 chick hearts with anti-human ephrinB1 and observed that the antiserum reacted with superficial epicardial cells. Here, we observed cytoplasmic staining. Interestingly, the intensity of staining was greater in cells at the leading edge of the monolayer and less in cells nearer to the explanted heart (Fig. 1C). We stained similar monolayers with pan anti-ephrinB (Fig. 1D). In this case, we observed cytoplasmic staining as well as staining in intercellular junctions and nuclei (Fig. 1D, arrowheads). Junctional staining with pan anti-ephrinB was more evident in areas of the monolayer behind the leading edge (Fig. 1E) where we have previously observed that junctional complexes such as adherens junctions and tight junctions are better organized (Dokic and Dettman,2006). Nuclear staining was present in this area of the monolayer (Fig. 1E). We performed three experiments to validate nuclear reaction by the rabbit pan anti-ephrinB antiserum. Nuclei did not react with goat anti-rabbit immunoglobulin G (IgG) in the absence of pan anti-ephrinB (Fig. 1F). Tight junctions but not nuclei were labeled when monolayers were stained with an antiserum to the tight junction protein zona occludens-1 (ZO1; Fig. 1G). This also independently demonstrated that the secondary antibody was not responsible for labeling the nuclei in our experiments with pan anti-ephrin. Finally, using confocal microscopy, we optically sectioned nuclei of monolayers stained with pan anti-ephrin and WT1, a mesothelial marker found in the nucleus. Here within 0.32-μm slices, we observed that nuclei were labeled with DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; Fig. 1H), and reacted with anti-WT1 (Fig. 1I) and pan anti-ephrin (Fig. 1J).

To confirm the presence of ephrinB ligands in vivo, we stained coronal sections of hearts with pan anti-ephrinB (Fig. 2A) and anti-ephrinB1 (Fig. 2B–E). Figure 2A shows a section of a HH30 heart stained with pan anti-ephrinB and anti-integrin α5 to label the vascular endothelium. Here, the epicardium and the perivascular mesenchyme reacted with this antiserum. Unlike what we observed in cultured cells, there did not appear to be nuclear staining in EPDCs at this stage with pan anti-ephrinB (Fig. 2A). Figure 2B shows a section of a HH26 heart stained with anti-ephrinB1. Here, the epicardium and interstitial cells reacted with the antiserum. Figure 2C shows a control experiment in which a section from the same heart was stained with anti-cytokeratin to label the epicardium. In these sections, labeled cells are not visible in the myocardial interstitium. At higher magnifications, cells labeled with anti-ephrinB1 appear in sites consistent with the endocardium (Fig. 2D) and within the myocardium (Fig. 2E). In all cases, this antiserum labels the cytoplasm of cells.

Figure 2.

EphrinB ligands are expressed in epicardium in vivo. A: Confocal microscopy of a Hamburger and Hamilton stage (HH) 30 chick heart section (coronal) stained with pan anti-ephrinB antiserum (green) and anti-α5 integrin to label vascular endothelium (red). Stars indicate lumens of coronary vessels. B: Confocal microscopy of a HH26 chick heart section (coronal) stained with pan anti-ephrinB1 (green), mAb MF20 (red) to delineate the myocardium, and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; blue) to visualize nuclei. C: View of a similar section from the same heart stained with pan anti-cytokeratin (green), mAb MF20 (red), and DAPI. D: Higher magnification of a HH26 heart section similar to (B) where the antiserum has labeled cells (green) directly facing the lumen of the heart within the trabeculae (lum). E: Confocal image of the myocardial interstitium of a HH28 chick heart section (coronal) stained with anti-ephrinB1. Here, other channels were removed to highlight the reaction of the antiserum with the cytoplasm of interspersed isolated cells. Arrowheads point to three such cells. F: Confocal microscopy of coronal sections through a HH28 chick heart incubated with mouse EphB3-FC fragments and visualized with goat anti-human Alexa 488. G: View of the atrioventricular (AV) sulcus (avs) showing binding of EphB3-Fc to the epicardium (epi) and epicardially derived mesenchymal cells (EPDCs). Cells are counterstained with propidium iodide (red). H: Control section incubated only with secondary antibody. Scale bars = 20 μm in A, 50 μm in B,C,F,G, 25 μm in D, 10 μm in E.

To confirm ephrinB ligand expression in these sites we incubated sections with EphB3-FC fragments and then visualized this interaction using goat anti-human IgG coupled to Alexa 488. In this experiment, we observed that the EphB3-FC protein interacted with superficial epicardium and EPDCs (Fig. 2F,G). Sections incubated with the human IgG control did not react with epicardium (Fig. 2H). Unlike the anti-ephrinB1 antiserum, EphB3-Fc did not react strongly with cells within myocardium or endocardium.

Cell Surface Expression of EphrinB1

The antiserum to human ephrinB1 reacted with cultured epicardial cells; however, its expression was diffuse in the cytoplasm. Thus, we did not know if the protein was surface expressed. To test if ephrinB proteins are expressed on the surface of chick EMCs, we treated serum-starved chick EMCs with aggregated EphB1-FC and EphB3-Fc fragments. Cells were treated for 45 min, fixed, and probed with anti-ephrinB1 (Fig. 3). Clustering was absent in cells that were treated with aggregated human IgG (Fig. 3A,D). Clustering of proteins was observed when cells were treated with aggregated EphB1-FC fragments (Fig. 3B,E). This confirmed that reaction of anti-ephrinB1 with chick EMCs is specific to a protein that can be clustered by aggregated EphB-Fc fusion proteins and indicated that at least some ephrinB1 ligands are presented on the surface of cultured chick EMCs.

Figure 3.

EphrinB1 is cell surface expressed. A–F: Confocal microscopy of cultured chick epicardial mesothelial cells (EMCs) stained with anti ephrinB1 (green), Texas-Red-X phalloidin, and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). A,D: Cells were incubated with aggregated human immunoglobulin G (IgG). B,E: Cells were incubated with aggregated EphB1-Fc fusion proteins. C,F: Cells were incubated with nonaggregated goat anti-human IgG. A–F: Regions within white boxes in (A–C) are magnified in (D–F). B,E: Clustering of ephrinB1 is visible as speckles in (B) and (E, arrowheads). Speckles are absent in the controls (A,C,D,F). Scale bars = 10 μm in A–C, 5 μm in D–F.

EphB Receptors Are Expressed in the Developing Myocardium

EphrinB ligands could mediate distinct adhesive and migratory effects on EMCs, including attraction or repulsion (Holland et al.,1996; Santiago and Erickson,2002). EMCs require integrins for migration on the myocardial surface, and in their absence, migrations does not occur (Yang et al.,1995; Sengbusch et al.,2002). For this reason, it is not likely that ephrinB ligands are sufficient to mediate EMC superficial migration. It is, however, likely that they could modify integrin-mediated migration (Huynh-Do et al.,1999,2002; Deroanne et al.,2003; Meyer et al.,2005). We, therefore, reasoned that before attachment of the PE, the myocardium would express an EphB receptor to interact with EMC expressed ephrinB ligands.

To determine if EphB receptors are expressed in the heart before PE attachment, we performed RT-PCR on total RNA isolated from HH16 hearts that were dissected free from PEs. We observed that of the four chick EphB receptors, three were amplified: EphB1, EphB2, and EphB3 (Fig. 4A). The presence of two bands in the EphB3 reaction suggested the presence of an alternatively spliced transcript, and this was confirmed by sequencing the smaller amplification product (data not shown). Transcripts for EphB5 were not amplified. While this experiment does not define which EphB receptors are expressed by the myocardium, it does demonstrate that EphB transcripts are present in the HH16 heart tube. We next performed in situ hybridization analysis of HH16 and HH17 embryos using antisense probes to chicken EphB1, EphB2, and EphB3 (Fig. 4B). Here, we observed that the antisense probe to EphB3 had the strongest reaction with the heart tube. Thus, before the attachment of the PE, EphB1, EphB2, and EphB3 transcripts are present in the heart tube, with EphB3 appearing to be the highest expressed.

Figure 4.

EphB3 expression in the developing chicken heart. A: Reverse transcriptase-polymerase chain reaction (RTPCR; 35 cycles) of EphB receptor transcripts from total RNA isolated from E16 heart tubes. B: Whole-mount in situ hybridization of EphB antisense (top) and sense probes (bottom) to Hamburger and Hamilton stage (HH) 16 and 17 chick embryos. C: Immunoblotting of total protein isolated from (left to right) HH19 trunk, HH19 forelimb, HH16 ventricle, HH19 ventricle, and HH25 ventricle. The immunoblot was probed with rat anti-mouse EphB3 as indicated at the right. The mobility of the 100-kDa marker is indicated on the left.

To test if EphB3 proteins were present in the developing heart, we isolated protein from dissected HH16 ventricles. We chose hearts with visibly unattached PEs and removed both the presumptive outflow tract and atria (sinus venosus). We also isolated total protein from HH19 and HH25 ventricles. For positive controls, we isolated total protein from dissected HH19 trunks and forelimbs. This was done by removing the head cranially, and the rest of the embryo caudally, leaving the heart and forelimbs. The heart, trunk, and forelimbs were then segregated for protein isolation. The trunk region includes portions of the neural tube, neural crest, and somites that have been shown by in situ hybridization to express EphB1, EphB2, and EphB3 (Krull et al.,1997; Santiago and Erickson,2002). The forelimb was also demonstrated to be a site of EphB3 expression by in situ hybridization (Santiago and Erickson,2002). Proteins were analyzed by immunoblotting (Fig. 4C). Using a rat monoclonal antibody raised against the N-terminus of mouse EphB3, we were able to detect a prominent band of approximately 100 kDa (Fig. 4C). This band was present in all the samples tested. Together with the results from RT-PCR and in situ hybridization, we conclude that of the EphB receptors tested, EphB3 is present in the developing chick heart during the time that the PE attaches and EMCs migrate superficially to form the epicardium.

Eph Receptors Alter Integrin-Based EMC Migration but Not Adhesion

To test if EphB receptors alter epicardial cell migration in vitro, we explanted HH24 chick hearts to dishes coated with mammalian EphB-FC fragments. We predicted that repulsive effects would be manifested as either inhibition or reduction of EMC migration from explanted hearts. Alternately, attractive effects would be manifested as an increase in EMC migration. Cells were allowed to migrate onto the plates from the hearts for 18 hours. Hearts were removed from dishes and the remaining epicardial monolayers were fixed. The area of each monolayer was determined using light microscopy. EphB-Fc fragments are coupled to the Fc portion of the human IgG heavy chain. We, therefore, used dishes coated with bovine serum albumin (BSA) and human IgG as controls for the Eph-Fc coated dishes. We also used dishes coated with human serum fibronectin (FN) as a positive control. Here, we observed that EMCs migrated from explanted hearts to form monolayers that covered 3.69 mm2 for FN and 1.82 mm2 for IgG (Fig. 5A). As compared with coating with BSA and IgG, EphB1-Fc and EphB2-Fc increased the average size of monolayers to 2.58 mm2 and 2.14 mm2, respectively. EphB3-Fc (1.79 mm2) did not increase or decrease migration. Only the increased migration on Eph-B1-FC was statistically significant. This effect was not as robust as FN, a known mediator of EMC migration. From this experiment, we conclude that surfaces coated with mammalian EphB receptors can augment epicardial migration relative to BSA and IgG. Because we did not observe a repulsive effect with EphB-Fc coating, this experiment supports the hypothesis that epicardial expressed ephrinB ligands are involved in attractive migration of chick EMCs on the heart surface.

Figure 5.

EphrinB ligands play a role in epicardial cell migration. A: Bar graph showing the average size of areas of epicardial monolayers migrated from Hamburger and Hamilton stage (HH) 24 chick hearts on surfaces coated with fibronectin (FN; 5 μg/ml) or EphB-Fc fragments (5 μg/ml). The height of each bar reflects the mean area of the monolayers in square millimeters. The five-point star indicates a P value of 0.004 for FN compared with bovine serum albumin (BSA; 1% w/v). The dagger indicates a P value of 0.02 for EphB1-Fc compared with BSA coating. The line above the other bars indicates that none of these values are significant for these variables compared with BSA coating. B: Bar graph showing relative adhesion of chick EMCs to the indicated proteins coated at a concentration of 5 μg/ml. The values on the y-axis are optical density units read at 540 nm for crystal violet staining. Increased values reflect more adhered cells. The five-point star indicates a P value of 0.01 for the comparison between FN coating and BSA coating. The line above the other bars indicates that none of these values are significant for these variables compared with FN coating. C: Bar graph showing the average size of areas of epicardial monolayers migrated from HH24 chick hearts on surfaces coated with FN or FN and EphB1-Fc fragments. The height of each bar reflects the mean area of the monolayers in square millimeters. The five-point star indicates a P value of 0.0008 for FN compared with BSA. The dagger indicates a P value of 0.04 for FN (1 μg/ml) + EphB1-Fc (5 μg/ml) compared with FN (1 μg/ml). The eight-point star indicates a P value of 0.05 for FN (1 μg/ml) + EphB1-Fc (5 μg/ml) compared with FN (1 μg/ml) + EphB1-Fc (5 μg/ml) + GRGDSP (100 μM). In all graphs, the error bars represent standard error of the mean. The statistical test was a Student's t-test. White numbers in each bar indicate the sample size for that experiment.

EphrinB ligands could also alter integrin-based adhesion of EMCs (Huynh-Do et al.,1999). To test this, we performed cell adhesion assays with FN and rat EphB1-FC. We chose EphB1-FC, because it had the greatest effect on migration. We observed that EphB1-FC was not sufficient to increase EMC adhesion to FN (Fig. 5B). FN increased cell adhesion by approximately 20%, but neither IgG nor EphB1-Fc significantly increased this adhesion to FN. Thus, ephrinB ligands do not alter integrin-based adhesion in EMCs.

We next tested if EphB1-Fc could alter EMC migration on surfaces coated with FN. We coated charged tissue culture plates with FN at two concentrations: 1 μg/ml and 5 μg/ml. We also coated plates with FN (at the same concentrations above) mixed with EphB1-Fc fragments at a concentration of 5 μg/ml. Plates were blocked in BSA (1% w/v), and HH24 hearts were explanted onto these surfaces and migration was quantified after 18 hr. Here, we found that the addition of EphB1-Fc to the coating mixture increased EMC migration by a modest but significant amount (Fig. 5C). At 1 μg/ml, FN was able to increase migration as compared with BSA coating in our assay, but not significantly. The effect of FN was significant at 5 μg/ml, increasing from 1.59 mm2 to 3.46 mm2. When EphB1-Fc was coated on surfaces along with FN, it increased migration at both concentrations of FN tested. Migration increased from 1.88 mm2 on 1 μg/ml FN to 2.32 mm2 on 1 μg/ml FN with EphB1-Fc, an increase of approximately 23%. Migration increased from 3.46 mm2 on 5 μg/ml FN to 4.03 mm2 on 5 μg/ml FN with EphB1-Fc, an increase of approximately 16%; however, this increase was not statistically significant.

We have shown that chick EMCs express several integrin receptors, including those that bind the RGD core of the cell-binding domain of FN (Pae et al.,2008). Others have demonstrated that α4β1 integrin, which binds to the V25 (CS-1) domain of FN, is expressed by EMCs and that this receptor is essential for PE cell attachment and migration on the surface of the myocardium (Stepp et al.,1994; Yang et al.,1995; Sengbusch et al.,2002). To determine which class of integrins is affected by FN, EphB1-Fc co-coating we used function-blocking peptides. Two peptides were used to test interactions with the cell binding domain: GRGDSP and its control peptide GRADSP. Additionally, two peptides were used to test interactions with the V25 (CS-1) domain of FN: EILDVPST and its control peptide EILEVPST. Because increased migration was statistically significant only when plates were co-coated with 1 μg/ml FN and 5 μg/ml EphB1-Fc (Fig. 5C), we chose these concentrations for our peptide studies. Thus, surfaces were coated with 1 μg/ml FN and 5 μg/ml EphB1-Fc, and peptides were added to the medium at a concentration of 100 μM during the time cells were migrating from hearts. Here, we observed that GRGDSP significantly reduced migration from 2.32 mm2 to 1.94 mm2, a decrease of approximately 16%. EILDVPST did not diminish migration but slightly increased it to 2.43 mm2; however, this increase was not statistically significant. Of the control peptides, GRADSP increased migration and EILEVPST reduced migration, but these changes were not statistically significant. These observations are consistent with the hypothesis that the interaction of EMCs with the cell-binding domain of FN by means of the RGD binding class of integrins is altered by the presence of EphB1-Fc on the substrate during migration.

Adventitial Fibroblasts Express EphrinB Ligands

Based on our observations of ephrinB ligand expression in situ, we predicted that ephrinB ligands persist in EPDCs during coronary vascular development. To investigate this, we performed in situ hybridization on chick hearts using antisense probes to ephrinB1 and ephrinB2. For ephrinB1, we observed that from HH24 to HH36 staining with the antisense probe was in the superficial epicardium and subepicardial mesenchyme (data not shown). This confirmed our previous observations with pan anti-ephrin and anti-ephrinB1. For the ephrinB2 antisense probe, we did not observe appreciable staining in the epicardium after HH24 (data not shown). From this it appears that, after HH24, ephrinB1 becomes the predominant ephrinB ligand in chicken epicardium and its derivatives. Of interest, after HH30 reaction with the ephrinB1 antisense probe appeared most intense in the region of the atrioventricular (AV) sulcus (Fig. 6A). We cut transverse sections of HH36 hearts previously stained by the ephrinB1 antisense probe and observed that the cells which reacted with the probe were located around vessels but not in the vascular media. We stained these sections with anti-smooth muscle α-actin and observed that the purple stain from the in situ hybridization was outside of the red stained region containing smooth muscle (Fig. 6B,C). Anti-ephrinB1 staining of HH40 (4 days older) chick heart sections indicated that some adventitial fibroblasts continued to express ephrinB1 (Fig. 6D).

Figure 6.

EphrinB1 ligand expression in perivascular mesenchyme in the atrioventricular (AV) sulcus. A: Whole-mount in situ hybridization of an ephrinB1 antisense probe to a Hamburger and Hamilton stage (HH) 36 chick heart. The line indicates the transverse plane of section shown in panels (B) and (C). B,C: Transverse sections of a heart similar to the one in (A) that show purple staining in the perivascular mesenchyme. Red staining is anti-smooth muscle α-actin. Stars indicate lumens of vessels. D: Confocal image of a transverse section of a HH40 heart stained with anti-ephrinB1 (green) and anti-smooth muscle α-actin (red). Arrows point to green cells in the vascular adventitia. E,F: Confocal image of an untreated epicardial mesothelial cell (EMC) monolayer stained with anti-ephrinB1 (green), anti-smooth muscle α-actin (red), and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; blue). G,H: Confocal image of a platelet-derived growth factor-BB (PDGF-BB) treated EMC monolayer stained with anti-ephrinB1 (green), anti-smooth muscle α-actin (red), and DAPI (blue). I,J: Confocal image of a PDGF-BB treated EMC monolayer stained with anti-ephrinB1 (green), anti-caldesmon (red), and DAPI (blue). The sensitivity of the photomultiplier tube was 40% less for the red channel in (G) and (I) than in (E). F, H, and J were imaged using identical photomultiplier tube settings and only show the ephrinB1 reaction.

These observations suggested that as EPDCs differentiate into smooth muscle they reduce their expression of ephrinB1. To test this, we used primary EMC cultures from HH24 chick hearts. These cells shift from a mesothelial to a fibroblast-like phenotype and begin to express smooth muscle cytoskeletal genes when treated with platelet-derived growth factor BB (PDGF BB) for 3 days (Landerholm et al.,1999; Lu et al.,2001). We explanted HH24 chick hearts onto collagen coated coverslips and allowed cells to migrate to the surface of the glass. Coverslips were treated with PDGF BB (10 ng/ml) for 3 days. EMCs on coverslips were probed for ephrinB1, smooth muscle α-actin, or caldesmon (Fig. 6E–J). We observed that, before differentiation, EMCs expressed ephrinB1 and smooth muscle α-actin (Fig. 6E,F). After stimulation with PDGF-BB, EMCs became fibroblast-like in appearance and they expressed higher levels of smooth muscle α-actin but not ephrinB1 (Fig. 6G,H). We repeated this experiment using antibodies to caldesmon and made a similar observation (Fig. 6I,J). Thus, as EMCs differentiate into smooth muscle in vitro, they down-regulate ephrinB1 expression. This experiment was consistent with our in vivo observations and suggests that the PDGF-BB stimulated EMCs differentiate into smooth muscle and not into other lineages derived from EMCs like cardiac fibroblasts and pericytes. Thus, down-regulation of ephrinB1 expression could be used as a marker to identify factors that stimulate cardiac fibroblast and pericyte differentiation in vitro.


We have observed that, in the developing chick heart, one site of ephrinB ligand expression is in epicardium, EPDCs, and perivascular fibroblasts of the AV sulcus. This expression is at least partially on the cell surface and is likely to play a role in integrin mediated epicardial superficial migration but not adhesion. EphB receptors are expressed in the HH16 heart tube, so it is possible that an interaction between a PE expressed ephrinB ligand and a myocardial expressed EphB receptor plays a role in the initial interface between the two cell types.

EphrinB1 has been implicated in the formation of skeletal structures, the retina, and in neural crest. A common theme for ephrinB1 in these tissues is in the guidance or sorting of cells along migratory pathways. For example, during early neural crest migration, ephrinB1 expressed in the somite can act repulsively on migrating neural crest cells expressing Eph receptors (Krull et al.,1997; Wang and Anderson,1997). In the case of migrating EMCs, the ephrinB ligands are expressed in the migrating cells. Our evidence would suggest that ephrinB ligands play a supportive role in EMC migration. This may involve sensing the myocardium using a similar mechanism through which retinal progenitor cells sense their environment as they migrate into the eye field (Moore et al.,2004; Lee et al.,2006).

Evidence is also accumulating that ephrinB ligands are likely to be important in establishing or maintaining tight junctions. Xenopus, mouse, and chick ephrinB1 show absolute conservation at their C-termini. Lee and colleagues have recently demonstrated that three tyrosine residues in this domain of ephrinB1 play a critical role in regulating the planar cell polarity pathway by interacting with Par6 and Disheveled (Dsh) (Lee et al.,2006,2008). Phosphorylation of these residues can be regulated by either ligation of ephrinB ligands to Eph receptors or by FGF signaling (Lee et al.,2008). FGF signaling can promote phosphorylation of ephrinB1 at Tyr 310. This reduces the interaction of ephrinB1 with par6 and promotes the establishment of tight junctions. FGF signaling can also result in tyrosine phosphorylation of ephrinB1 at residues 324 and 325. This reduces the interaction between ephrinB1 and Dsh, and this alters the ability of ephrinB1 to mediate cell migration (Lee et al.,2009). TGFβ receptors play an important role in regulating epicardial EMT (Olivey et al.,2006; Compton et al.,2007; Sridurongrit et al.,2008) and interact with Par6 to modulate epithelial cell polarity in mammary epithelial cells (Ozdamar et al.,2005). These studies, together with ours, point to a potential role for ephrinB1 in the complex regulation of epicardial migration, cell polarity, and EMT as regulated by the FGF and TGFβ signaling pathways.

Our observation of junctional ephrinB ligand localization with pan anti-ephrinB antiserum supports this hypothesis. There are two possible interpretations of our observation. By virtue of its C-terminal epitope, pan anti-ephrinB recognizes both ephrinB1 and ephrinB2. Therefore, because junctional staining was not observed with the anti-ephrinB1 antiserum, which recognizes the extracellular portion of the molecule, it represents the localization of ephrinB2 in EMCs. Alternatively, because the epitope for the pan anti-ephrin is the C-terminal cytoplasmic domain, junctional staining could represent staining of either ephrinB ligand but that this region of ephrinB1 is not accessible to this antiserum when it is localized in junctional complexes. Based on our experiments, it is not possible to conclude which ligand is present in the junctions. However, staining with the pan anti-ephrinB antiserum does suggest a role for ephrinB ligands in epicardial cell junctions.

In cultured HH24 EMCs, we also observed nuclear staining with pan anti-ephrinB. This localization was not observed in subepicardial mesenchyme of HH30 hearts. A possible nuclear localization for ephrinB ligands has not yet been documented but is consistent with other epithelial examples. Class B ephrins have been demonstrated to be PDZ domain binding proteins. Specifically, the conserved YYKV sequence in the cytoplasmic C-terminal domain has been implicated as a PDZ domain binding site (Lin et al.,1999). PDZ domain proteins can mediate the interaction of proteins and their subcellular localization (Nourry et al.,2003). Of interest, similar to what we observed with pan anti-ephrinB, the zona occludins-2 protein (ZO2) is localized either in tight junctions or the nucleus, and the nucleus is thought to serve as a reservoir for newly synthesized ZO2 (Chamorro et al.,2009). Nuclear import of ZO2 is increased by mechanical injury or chemical stress and during the G1 phase of the cell cycle. Thus, its localization is carefully regulated and there is typically more ZO2 in the nucleus of proliferating cells (Tapia et al.,2009). It is possible that, if ephrinB ligands bind to PDZ domain proteins such as ZO2, their localization pattern could change depending on several factors including cell proliferation (for example in cultured cells), migration and epithelial versus mesenchymal phenotype.

Of the EphB transcripts detected in the HH15–17 heart, EphB3 was expressed at the highest level in the myocardium by RT-PCR and in situ hybridization. EphB3 transcripts have been observed in the cardiogenic mesoderm with continued expression in the primary myocardium during fusion and looping of the heart tube (Baker et al.,2001; Baker and Antin,2003). Additionally, Santiago and Erickson (2002) observed EphB3 transcripts in HH23 hearts by in situ hybridization. In their study, none of the other Eph receptor probes (including EphB1 and EphB2) reacted significantly with the heart at this stage. These observations, along with ours, support a continued expression of EphB3 in the primary myocardial lineage. Thus, the presence of EphB3 in the myocardium before the attachment of the PE suggests a possible role in mediating a molecular interaction between of ephrinB1/ephrinB2 expressing PE cells and EphB3 expressing myocardial cells.

One curiosity made apparent by our experiments was that, in our migration experiment, chick EMCs responded the most to rat EphB1-Fc chimeras. This was not consistent with our observation that EphB3 transcripts were the most abundant of the four EphB receptors we tested by RT-PCR and in situ hybridization. We concluded that there must be subtle functional differences between the mammalian Eph-Fc chimeras we used in our assay and avian EphB receptors. So we compared the amino acid sequences of chick EphB3 with rat and mouse EphB1, EphB2, and EphB3 using clustalW2 (Larkin et al.,2007). We found that chick EphB3 was most closely related to rat EphB1. Thus, by phylogeny we predict that, of the mammalian EphB-Fc proteins, rat EphB1-Fc would have the greatest biologic effect on chick EMCs effectively mimicking avian EphB3. It also follows that our experiment demonstrates that, when comparing orthologous gene products, there may be important functional differences between avian and mammalian Eph family members that affect their use in avian experiments.

While several studies have demonstrated in vitro differentiation of EMCs into smooth muscle, none have reported differentiation into perivascular fibroblasts. This is likely because little is known about this differentiation pathway or markers that would define it. Our findings support the idea that ephrinB1 could be used as a marker for fibroblast differentiation, because it is expressed in perivascular fibroblasts in HH36 and older hearts and because it is reduced in PDGF-BB differentiated EMCs in vitro.



Babcock B-300 Chicken (Gallus gallus) hatch eggs were obtained from Phil's Fresh Eggs (Forreston, IL) and incubated in a humidified incubator at 38°C. Embryos were staged according to method of Hamburger and Hamilton (1951).


Recombinant rat EphB1-FC, mouse EphB2-FC, mouse EphB3-FC, mouse ephrinB1-FC, and mouse ephrinB2-FC were obtained from R&D Systems. Texas Red-X phalloidin, 4′, 6 diamidino-2-phenylindole, dihydrochloride (DAPI), propidium iodide, and phalloidin coupled to Texas Red-X were from Invitrogen, Molecular Probes. IgG from human serum was obtained from Sigma. Peptides were purchased from American Peptide and suspended in sterile water. To competitively inhibit binding of RGD binding integrins to the RGD core domain of ECM proteins, we used the GRGDSP peptide. The peptide GRADSP was used as the control. To competitively inhibit binding of α4β1 to the V25 (CS-1) domain of FN, we used EILDVPST. The peptide EILEVPST was used as the control. All peptides were used at a final concentration of 100 μM. Mouse lung extract (Santa Cruz Biotechnology, SC-2300) was used as a positive control for ephrinB1 in immunoblots.


The following antibodies were used in this study. Rabbit anti-human ephrinB1 (sc-1011), and rabbit pan anti-human ephrinB (sc-910) were obtained from Santa Cruz Biotechnology. Goat anti-human IgG, goat anti-mouse IgG and goat anti-rabbit IgG coupled to the fluorophores indicated in the text were obtained from Invitrogen, Molecular Probes. Sheep anti-digoxigenin (DIG) coupled to alkaline phosphatase was obtained from Roche Diagnostics. Mouse anti-α5 integrin (U1α) was a gift from Dr. Ruth Chiquet-Ehrismann. Mouse anti-caldesmon (CALD-5) and anti-smooth muscle α-actin (1A4) were obtained from Sigma. Mouse monoclonal anti-myosin heavy chain (mAb MF20), was obtained from Developmental Studies Hybridoma Bank. Rat anti-mouse EphB3 was obtained from R&D systems. Rabbit anti-cytokeratin (catalog no. Z0622) and mouse anti-human WT1 (clone 6F-H2) were obtained from Dako. Goat anti-human IgG FC fragment specific was obtained from Jackson Immunochemicals. Vector Red staining of paraffin sections was done using a goat anti-mouse Vectastain ABC-AP kit (Vector Laboratories).

Tissue Culture

Primary cultures of chick epicardial cells (EMCs) were grown from HH24 hearts as in Dokic and Dettman (2006) and maintained in M199 medium with varying amounts of (as indicated in the text) fetal bovine serum (Hyclone) in a humidified 5% CO2 tissue culture incubator at 37°C. In some cases, hearts were explanted onto 22-mm glass coverslips coated with rat-tail Collagen 1 (BD Biosciences).


Cells or embryonic tissues were solubilized in ice-cold RIPA buffer (50 mmol/L Tris-HCl pH 8.0, 350 mmol/L NaCl, 1% w/v NP-40, 0.5% w/v deoxycholate, 0.1% w/v sodium dodecyl chloride) containing protease inhibitor cocktail (Sigma, catalog no. P8340) for 5 min, scraped or pulverized with a disposable pestle and spun at 4°C at 10,000 rpm for 10 min. Protein in cleared supernatants was quantified using the Bradford method (Thermo Scientific) and 10 μg of total cellular protein was run on 10% acrylamide sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels (Bio-Rad). Proteins were electroblotted onto Hybond-ECL nitrocellulose membranes (Amersham), which were blocked in 2.5% (w/v) nonfat dry milk in Tris buffered saline with 0.1% Tween-20 (Fisher). Membranes were probed with antibodies in blocking solution at 4°C, overnight. Horseradish peroxidase conjugated secondary antibodies (Cell Signaling Technologies) in blocking solution were incubated with washed membranes for 1 hr at room temperature. Bands were detected using SuperSignal West Femto substrate (Thermo Scientific) on X-ray film.

Aggregation of EphB-Fc Fragments

EphB1-FC or EphB3-Fc proteins were aggregated using a 10-fold excess of goat anti-human FC for 2 hr at room temperature. As a control, human IgG was aggregated with goat anti-human FC for the same time. Aggregates (5 μg/ml, final concentration) were added for 45 min to serum-starved chick EMCs grown on collagen coated glass coverslips in a 5% CO2 tissue culture incubator at 37°C. A second control was to incubate cells for 45 min with goat anti-human FC to determine if this antibody adhered to cells in a punctate manner. Coverslips were washed three times with phosphate buffered saline (PBS) before fixation in 4% formaldehyde for 10 min. Coverslips were then washed three times in PBS to remove formaldehyde and blocked in PBS containing 0.1% (w/v) bovine serum albumin (Sigma) and 0.1% (v/v) Triton X-100 (Fisher). Cells were probed with anti-ephrinB1, washed, and then probed with goat anti-rabbit IgG. Cells were counterstained with DAPI and phalloidin. Cells were then imaged using confocal microscopy.

Mesothelial Cell Migration Assay

Cell migration assays were performed as was described in Pae et al. (2008) with the following changes. The 30-ml culture dishes were coated overnight with human plasma fibronectin (BD Biosciences) at concentration of 1 or 5 μg/ml (in PBS) and/or EphB-FC fusion proteins at 5 μg/ml (in PBS) at 4°C, washed three times with PBS and blocked for 1 hr in bovine serum albumin (Sigma, 1% v/v) at 37°C in a 5% CO2 tissue culture incubator. Coated and blocked plates were washed three times with serum free M199 medium. A total of 500 μl of serum free M199 medium was added after the third wash, and two to four hearts from HH24 embryos were placed on the surface of each plate and left at room temperature for 30 min to allow hearts to adhere to the surface. Because hearts were approximately the same size, the starting area was approximately equal between each assay. Dishes were placed in a 5% CO2 tissue culture incubator at 37°C for 18 hr. Hearts were removed from the surface of the plates by gentle washing with the remaining medium and cells were washed twice with sterile PBS and fixed in 4% formaldehyde. Plates were stained with hematoxylin as in Pae et al. (2008) and the total area of each monolayer was determined using Metamorph 4.5 (Molecular Devices). In experiments with bioactive peptides, peptides were recently purchased, aliquoted, and stored at −80°C. They were thawed and diluted in serum free medium immediately before hearts were explanted onto dishes. Peptides or other proteins were never reused after thawing.

Cell Adhesion Assay

Cell adhesion assays were performed as was described in Pae et al. (2008) with the following changes. Primary cultures of chick EMCs were generated from 20 to 30 explanted HH24 hearts in serum free M199 and antibiotics. Hearts were removed after 1 day and cells were cultured for 2 days in M199 medium containing fetal bovine serum (FBS; 1% v/v). Ninety-six well tissue culture plates were coated as above, blocked in heat-inactivated bovine serum albumin (10 μg/ml) for 30 min, and then washed with 300 μL of serum free M199. Cells were digested briefly in trypsin (0.25% w/v, 2.21 mM ethylenediaminetetraacetic acid), mechanically disrupted by pipetting and complete disruption of cells was monitored on a phase contrast microscope. Cells were pelleted by centrifugation (5,000 rpm), resuspended in 320 μl of M199 supplemented with FBS (10% v/v), and placed on ice. Cells were counted on a hemocytometer and diluted to a concentration of 350 cells/μl. An equal number of cells (35,000) were added to individual wells in 100 μl serum free medium, the plate was briefly swirled and placed in a 5% CO2 incubator for 45 min at 37°C. Wells were washed three times with PBS followed by aspiration and then fixed in formaldehyde (4% v/v) for 10 min at room temperature. Bound cells were stained with crystal violet (0.1% w/v in absolute ethanol) for 10 min. Wells were washed with water until no residual dye remained and plates were allowed to dry. Dye was solubilized in 100 μL sodium dodecyl sulfate (2% w/v in water) and absorbance was read at 540 nm on a Lab Systems MCC/340 multiscan plate reader. Data are expressed as average optical density of the solution at 540 nm. P values were calculated using a Student's t-test.

In Situ Hybridization

Hybridization with DIG labeled RNA probes was performed following an in situ hybridization protocol adapted from Nieto (Nieto et al.,1996). An alkaline phosphatase conjugated anti-DIG antibody and an NBT/BCIP (nitroblue tetrazolium choride/5-bromo-4-chloro-3-indolyl phosphate) color reaction was used for detection of probe. Photographs of the embryos and hearts were taken using a Nikon Coolpix 5700 digital camera.

Riboprobe Synthesis

EphB1, EphB2, and EphB3 cDNAs were obtained from Dr. Parker Antin. EphrinB1 was amplified from PE mRNA cloned into pCRII (Invitrogen) and sequenced to confirm identity. The following cDNAs were used to generate ribroprobes: EphB1, accession number Z19110, nucleotide range 934–2876; EphB2, accession number M62325, nucleotide range 871–1439; EphB3, accession number Z19061, nucleotide range 1674–3582; ephrinB1, accession number U72394, nucleotide range 801–1145; ephrinB2, accession number AF180729, nucleotide range 1–1002.


Total RNA from dissected PEs or cultured embryonic chick EMCs was isolated using the RNAqueous for RT-PCR kit (Ambion). RT-PCR was done using the Access for RT-PCR kit (Promega). The primers used for RT-PCR are listed in Table 1.

Table 1. Primers Used in This Study
Chick ephrinB1 forward5′AGGAGGCAGACAACACCGTGA3′
Chick ephrinB1 reverse5′TAGGGGATGATGATGTCGCT 3′


The authors thank Dr. Parker Antin for the gift of the cDNAs used in this study. We thank Drs. Hans-Georg Simon, Danijela Dokic, and Maria Luz V. Dizon for critical reading of the manuscript. The Pediatric Cardiopulmonary Disease Laboratory is generously supported by funds raised by the Women's Board of Children's Memorial Hospital, Chicago, IL.