The formation of the coronary vasculature system involves vasculogenesis and angiogenesis and is dependent on a transient structure called the proepicardium (PE) (reviewed in Wada et al.,2003a,b). The process of coronary vessel morphogenesis is initiated when the PE attaches to the heart at the junction between the embryonic atrium and ventricle at Hamburger and Hamilton (HH) stage 16 (1951) in avians and embryonic day 10.5 in rats (Viragh et al.,1993; Nesbitt et al.,2006). The PE is a cluster of cells that attach to the myocardium and migrate superficially on the heart to form the epicardium (Manner,1992). Some epicardial cells undergo epithelial to mesenchymal transformation (EMT), to invade the subepicardial space and myocardium (Munoz-Chapuli et al., 1996; Perez-Pomares et al.,1997; Dettman et al.,1998; Gittenberger-de Groot,1998; Manner,1999). Mesenchymal cells derived from the epicardium contribute to the coronary vasculature as vascular endothelial cells, vascular smooth muscle cells, and perivascular and interstitial fibroblasts (Mikawa and Fischman,1992; Mikawa and Gourdie,1996; Dettman et al.,1998; Gittenberger-de Groot,1998; Kirby,2002; Reese et al.,2002; Wada et al.,2003; Olivey et al.,2004; Wessels and Perez-Pomares,2004; Tomanek,2005).
Coronary vasculature development arises by a series of events in which vasculogenic endothelial cells form a plexus of tube-like structures (Sabin,1917; Poole and Coffin,1989; Kattan et al., 2004). Angiogenesis occurs as the existing capillary network increases in size via branching and connects to the aortic root forming the coronary arteries (Bogers et al.,1989; Poelmann et al.,1993; Tomanek,1996,2005; Waldo et al.,1990). This process continues as newly formed arteries recruit other PE-derived cells such as vascular smooth muscle cells and fibroblasts to form the vascular network (Tomanek,2005).
Although a wealth of knowledge exists concerning the anatomical development of the coronary vasculature system, much remains to be determined about the molecular mechanisms that regulate this process. Previous reports have suggested that myocardial development and coronary vascularization are interdependent processes because of bidirectional signaling occurring between epicardially-derived cells (EPDCs) and cardiac myocytes (reviewed in Bhattacharya et al.,2006). To date, studies have shown that when the interaction of GATA-4 and FOG-2 transcription factors is perturbed in cells of the myocardium, the epicardium fails to undergo EMT and coronary vascular development is disrupted (Svensson et al.,2000; Crispino et al.,2001). Thus, it appears that cardiac myocytes are critical to the generation of signals that stimulate epicardial EMT and coronary development. Conditioned media obtained from cardiac myocytes has also been shown to enhance epicardial EMT, which suggests that these cells are able to regulate this process by secreted factors (Morabito et al.,2001; Chen et al.,2002).
Several reports suggest that when PE-derived cells come in contact with the myocardium, these cells become exposed to myocardial-derived growth factors that induce the process of coronary vessel morphogenesis (Guadix et al.,2006; Pennisi et al.,2003). In addition, paracrine factors secreted by the myocardium may stimulate the development, growth, and maintenance of the coronary vasculature system (Giordano et al.,2001). Several growth factors including VEGF, FGF, PDGF, and TGF-β have been identified as potential regulators of coronary vessel morphogenesis (Beck and D'Amore,1997; Tomanek et al.,1998,1999; Tomanek and Zheng,2002; Tomanek,2005; Lavine et al.,2006). VEGF-A, in particular, has been extensively recognized as an important growth factor for myocardial vascularization (Tomanek and Zheng,2002; Tomanek,2005). VEGF-A is a single gene deriving three splice products, VEGF120, VEGF165 (4) in murines, and VEGF188. Recent reports indicate that each isoform may have a specific role in vascular patterning during development (Mitchell et al.,2006; Lundkvist et al.,2007).
In this report, we sought to specifically address the role that cardiac myocytes and VEGF164 plays in coronary vessel morphogenesis. This investigation was facilitated by use of two novel models of coronary development. One is an in vitro model system in which PE cells are cultured on a collagen scaffold. In this model, PE cells undergo EMT and develop capillary networks of endothelial tubes (tubulogenesis) when cultured with cardiac myocytes (Nesbitt et al.,2007). The other model system uses an organ culture approach in which embryonic hearts are used to assay epicardial EMT (Dokic and Dettman, 2007). The in vitro collagen cell scaffold can be used as an artificial extracellular matrix by incorporating VEGF molecules into the scaffold. PE cells cultured on these VEGF164-treated scaffolds had increased numbers of endothelial cells and higher endothelial cell proliferation rates in comparison to untreated scaffolds and PE cells co-cultured with cardiac myocytes. However, there was no evidence that the VEGF164-stimulated endothelial cells underwent morphogenesis into endothelial tubes (tubulogenesis). This is opposed to PE cells that were cocultured with cardiac myocytes in which extensive tubulogenesis was observed. Interestingly, inhibiting the VEGF receptor, Flk-1 (VEGFR-2), reduced tubulogenesis in the PE cardiac myocyte cocultures. Organ culture studies found that VEGF164 stimulated epicardial EMT to the same degree as TGFβ1; however, co-stimulation with VEGF164 and TGFβ1 did not increase overall epicardial EMT. These results indicate that myocardially derived VEGF164 drives proliferation and EMT of PE-derived cells, but is not sufficient to induce the morphogenesis of endothelial cells into endothelial tubes. However, the VEGF pathway does appear to regulate coronary tubulogenesis via the flk-1 receptor.
VEGF164 Increases PE-Derived Endothelial Cell Density
Previous studies have described the culturing of PE-derived endothelial cells on a 3-D collagen scaffold (Nesbitt et al.,2007). This 3-D culturing system recapitulated several key steps in avian coronary vasculogenesis including epicardial mesothelium formation, EMT, and de novo endothelial vessel development. Here we tested if cardiac myocytes could influence PE cell development on the 3-D collagen scaffold. HH stage 16 quail PE explants were cultured on scaffolds alone (Fig. 1, top row) or cocultured with rat cardiac myocytes (Fig 1, middle row). Scaffolds were then fixed and stained with DAPI and mAb QH1 to detect endothelial cells. After 2 days, we observed that endothelial cells were less abundant and did not become elongated, as was observed in the PE/CM coculture (Fig. 1, middle row). This demonstrated that cardiac myocytes were able to alter the behavior of PE cells such that they differentiated into networks sooner than on collagen scaffolds alone. When cultures were allowed to incubate for 7 days, we made two observations. First, on the collagen scaffolds, QH1+ cells primarily remained as individual rounded, squamous cells [although as in Nesbitt et al. (2007) some networks were observed]. Second, in PE/CM cocultures, most QH1+ cells were elongated and in networks. We concluded from this experiment that cardiac myocytes were able to induce PE-derived QH1+ cells to change shape and organize into networks.
We next tested the hypothesis that VEGF164 produced by myocytes could cause these changes within the context of the 3-D collagen scaffold. We found by immunoblotting that cultured rat cardiac myocytes synthesize VEGF164 (data not shown). One of the advantages of using the scaffold is that growth factors can be crosslinked within the collagen fibers of the scaffold such that they are slowly released over time (see Supplemental Data). In this experiment, quail PEs were cultured on collagen scaffolds to which recombinant mouse VEGF164 was crosslinked. Here we observed that there were more endothelial cells on the VEGF164-treated scaffolds in comparison to the untreated scaffolds or the PE/CM cocultures (Fig. 1, bottom row and 2). At 7 days of culture, the difference in cell numbers was even more pronounced. However, while we observed elongated cells on VEGF-treated scaffolds at day 2, most of the QH1+ cells at day 7 remained squamous in shape, in contrast to PE/CM cocultures, which contained collections of elongated vessel-like endothelial cells (Fig. 1A, inset in middle row). These were rarely observed in either the day-7 PE-only cultures or on VEGF scaffolds. Moreover, in VEGF164 crosslinked scaffolds, many QH1+ cells retained a squamous, epithelial-like appearance with the characteristic cobblestone morphology (Fig. 1, inset B) in contrast to day-7 PE/CM cocultures where cells were uniformly elongated (Fig. 1, inset A). We concluded from this experiment that VEGF164 plays a role in the genesis of endothelial cells from the PE, but that this role is distinct from that conferred by cardiac myocytes.
To quantify these observations, the density of PE-derived endothelial cells was determined for these different treatments using confocal fluorescence microscopy (Fig. 2). At 2 days of culture, endothelial cell density was significantly higher (P < 0.5) in PE+VEGF cultures as compared to both the PE-only or PE/CM cultures (Fig. 2). This difference in endothelial density was also observed at the day-7 time point. Interestingly, the amount of QH1-dependent fluorescence was not significantly higher in PE/CM cocultures despite the distinct cell shape changes we observed. This suggested that there may be a difference in cell proliferation of PE-derived endothelial cells between the PE + VEGF cultures and the PE/CM cultures.
Endothelial Cell Proliferation on the Tube Substrate
To determine whether the increase in endothelial density was due to proliferation, we developed an endothelial-specific 5′-bromo-2′ deoxyuridine (BrdU) assay. Here we treated scaffold-based PE cultures with BrdU and then stained with the isolectin GS-IB4, which specifically binds to vascular endothelial cells. In this assay, BrdU-positive nuclei, surrounded by isolectin staining, were interpreted to be replicating endothelial cells. PE cells cultured on substrates containing VEGF164 had a significantly higher (approximately 6-fold) endothelial BrdU labeling index than both PE-only cultures and PE/CM cocultures (Fig. 3B). This observation was consistent with our observations in Figure 2 and the hypothesis that VEGF164 induced PE-derived endothelial cell proliferation.
Inhibition of the Flk-1 Receptor Decreases Endothelial Tube Morphogenesis
Though our data indicate that VEGF stimulates PE-derived endothelial cell proliferation, distinct differences between the PE/CM and PE/VEGF cultures were observed with regards to tubulogenesis. Numerous endothelial tubes were found in the PE/CM cultures, but were not observed in the PE+VEGF or PE-only cultures (Fig. 1). During the morphogenesis of endothelial tubes (tubulogenesis), endothelial cells change morphology. Starting as squamous cells with round nuclei, they differentiate into spindle-shaped cells with oval nuclei (Fig. 1). Tubulogenesis, therefore, requires that endothelial cells undergo additional differentiation events as they become incorporated into vessels. To determine the extent of tubulogenesis that occurs in our system, we investigated the long-term effects of PE/CM coculture. Extensive arrays of endothelial tubes were observed in day-14, -21, and -28 cocultures (Fig. 4). However, the number of tubes peaked at day 14. Typical examples of day-14 and -28 cultures are shown in Figure 4.
To test if cardiac myocytes secreted VEGF plays a role in coronary tubulogenesis, PE/CM cocultures were treated with SU1498, a specific inhibitor of the VEGF receptor flk-1. At day 14 of culture, PE/CM cocultures showed on average over 25 endothelial tubes per explant (Fig. 5). In PE/CM cocultures treated with SU1498, the number of endothelial tubes was significantly less, averaging approximately 4 tubes/explant (Fig 4). Thus, inhibition of Flk-1 in cocultures reduced endothelial tube formation by about fivefold. Moreover, we observed very little endothelial tube formation in collagen scaffolds containing VEGF164, which indicates that endothelial tube formation requires VEGF and another signal supplied by cardiac myocytes that requires Flk-1 signaling.
VEFG164 Stimulates Transformation of HH26 Epicardial Cells
Recombinant human VEGF was reported to stimulate epicardial EMT in collagen gel assays (Morabito et al.,2001). However, in our experiments in which VEGF164 was crosslinked to the scaffold, we observed that most QH1+ cells were squamous, suggesting that they did not undergo EMT. We therefore tested if VEGF164 could stimulate EMT in chick hearts. Using an organ culture model of epicardial transformation (Dokic and Dettman,2006), the effect of VEGF164 on mesenchyme formation was assayed (Fig. 6). A LacZ-expressing replication-incompetent adenovirus was used to label epicardial cells on the surface of HH26 chick hearts. Previously, all three members of the TGF-β family were shown to stimulate transformation of epicardial cells in HH26 hearts and collagen gels (Dokic and Dettman,2006). At this stage, epicardial cells (Epi, Fig, 6A) transform into mesenchymal cells and move into the subepicardial space and myocardium becoming mesenchymal cells. Here we observed that VEGF164 stimulated epicardial EMT to a similar degree as TGFβ1 as compared to the untreated (UTD, Fig. 6B). Interestingly, no significant difference in EMT was observed between the separate treatments of TGFβ1, VEGF164, or costimulation (Fig. 6B). Thus, while VEGF164 does not stimulate EMT on the collagen scaffold, it stimulates EMT in cultured chick hearts at a rate similar to TGFβ1. This confirms the earlier report of Morabito et al. (2001) and indicates that VEGF164 can synergize with a heart-derived factor to drive epicardial EMT.
In this report, two unique in vitro systems have been used to investigate the cellular and molecular mechanisms of early coronary vascular development. In previous work, our group has demonstrated the ability to grow spontaneously contracting embryonic cardiac myocytes (CM) in a 3-D culture for long periods of time (Evans et al.,2003). Here, we have used this model to investigate the interactions between CM and PE cells during coronary vasculogenesis.
There has been some controversy about the origin of the endothelial cells of the coronary vasculature. As reported recently, there appears to be cellular heterogeneity within the PE. Cells on the outside of the PE are epithelial and express a discrete set of attachment molecules including α4 integrin (Pae et al.,2008). Cells within the core of the PE buds appear to be mesothelial and express adhesion molecules also found in mature vascular endothelium such as α5 integrin. Experiments in this study were carried out using PE cells that had yet to contact the myocardium. Also, PE dissections were carried out so as to include as little liver tissue as possible. However, since the PE forms as a cellular projection that is connected to the liver, there is always the possibility that some liver cells may be present in PE explants. Regardless, numerous endothelial cells (QH1-positive) were reliably cultured from these explants.
Culturing PE explants in the presence of VEGF164 stimulated endothelial expansion. The observation that VEGF164 treatment stimulated endothelial proliferation in PE cells indicates that this VEGF A isoform has the capacity to drive the proliferation of committed endothelial cells as opposed to differentiation of uncommitted PE cells. If VEGF164 induced endothelial differentiation, then we would have expected to see endothelial expansion without an increase in BrdU incorporation.
Our initial hypothesis was that the presence of CM would stimulate endothelial expansion. Though an upward trend was observed at the day-7 time point of the PE/CM cocultures, no significant difference was observed between it and the PE-only cultures. However, there was a difference in the morphology of a subset of endothelial cells of the PE/CM cocultures and the PE-only and VEGF164-treated PE cultures. Tube-like arrangements of spindle-shaped endothelial cells were abundant in the PE/CM co-cultures, but not so in the PE-only and VEGF164-treated PE cultures. This indicates that the presence of CM provides an additional signal that stimulates coronary tubulogenesis. Inhibition of the VEGF-A receptor Flk-1 (VEGFR-2) decreased CM-induced tube formation. Thus, it is possible that other VEFG-A isoforms (VEGF121 or VEGF189) could be regulating this next step in coronary morphogenesis.
As has been shown by Tomanek and colleagues (2006), other signaling molecules are critical in the development of the coronaries. Of particular relevance to our observations is another VEGF family member, VEGF-B. Inhibiting antibodies against this VEGF were found to inhibit coronary artery development. Since VEGF-A and -B bind the Flt-1 receptor (VEGF-R1), we posit that cooperation between Flk-1 and Flt-1 could be necessary for coronary vascular development. Alternatively, the role of VEGF164 may be to simply stimulate endothelial proliferation creating the adequate numbers of endothelial cells necessary for VEGF-B to have its affect.
VEGF164 was also found to stimulate EMT in organ culture experiments. Interestingly, this stimulation could not be enhanced by co-stimulation with TGFβ. Some studies have demonstrated that individual TGFβ isoforms can repress endothelial tube formation in the avian heart (Holifield et al.,2004). Other studies have supported a role for TGFβ in epicardial EMT. We observed here that VEGF164, in particular, could stimulate epicardial EMT, but only in the heart. This supported the earlier work of Morabito et al. (2001), which demonstrated a role for rhVEGF in epicardial EMT. Neither, TGFβ (Holifield et al.,2004) nor VEGF164 (this study) could drive endothelial tubulogenesis. Thus, both TGFβ and VEGF164 have the capacity to induce epicardial EMT and neither drives tubulogenesis. A critical difference here was our observation that VEGF164 stimulated endothelial cell proliferation. While it is possible that overlapping signaling pathways govern endothelial differentiation, growth, tube formation, and EMT, it is likely that these pathways differ or must synergize. Our present results support the idea that molecular control points of coronary vascular development are separable experimentally using the collagen gel scaffold.
The idea that the TGFβ and VEGF signaling pathways are linked somehow is especially provocative. Perhaps, TGFβs are primarily involved in stimulating epicardial EMT but as is suggested by Holifield et al. (2004), TGFβs simultaneously repress VEGF-mediated events such as tubulogenesis. This would allow cells to differentiate along a non-endothelial path such as into smooth muscle precursors or cardiac fibroblasts. Conversely, as we observed that VEGF164 stimulated epicardial EMT at an equal rate to that of TGFβ1, VEGF164 may stimulate epicardial EMT in other areas where TGFβ is less abundant or repressed. These cells may then be influenced in a greater way by pathways that favor endothelial cells. Together, these observations point to subtleties in the molecular regulation of epicardial EMT, endothelial proliferation, and differentiation that will require future study.
In an attempt to synthesize the data presented here with previous studies, we suggest the following pathway of coronary development. The PE contains cells with differing potentials and fates. Epithelial cells of the PE attach to the myocardium via an α4β1/WT1-meditated process and become definitive epicardium. Mesothelial cells in the core of PE villi are committed to the endothelial line. The interaction of the PE cells with the myocardium would then stimulate FGF signaling that, in turn, induces the sonic hedgehog signal that is required for VEGF A, B, C, and Ang 2 expression (Lavine et al.,2006). Expression of the VEGF164 would then drive EMT and endothelial proliferation and subsequent VEGF-A and VEGF-B expression and TGFβ inhibition would then promote tubulogenesis of the coronary vasculature. These events would then be followed by recruitment of smooth muscle cells and further pruning and stabilization that occurs as blood flow is established after the coronaries become patent with the aorta.
Endothelial Cell Density Assay
The endothelial cell density assay was used to determine the number of endothelial cells on tube cultures at 48 hr and 7 days. This assay was conducted by quantifying the area (in pixels) occupied by QH1 staining (endothelial cell marker) and dividing this number by the area occupied by DAPI pixels within microscopic fields. The process was automated by determining threshold values for endothelial cells using a positive control (stage-14 quail blood islands/CAM assay). Values were determined using Image Pro Plus version 22.214.171.124 (Media Cybernetics, Inc). Pixel values above threshold were counted as positive. At least three biological replicates were assayed per treatment group per time point. Endothelial cell density assays were statistically analyzed by quantifying a minimum of 15 fields. These data were analyzed using the SigmaStat 3.0 Suite of programs. Significance (P < 0.05) was determined for the endothelial cell density assay using a two-way analysis of variance. Pairwise multiple comparison procedures (Holm-Sidak method) revealed that the PE/VEGF vs. PE-only and PE/VEGF vs. PE/CM comparisons were significantly different at both time points.
Scaffold-Cultured PE Endothelial VEGF164 Assay
The effect of the VEGF164 isoform on coronary vascular development has been reported previously by Yue and Tomanek (2001). These experiments used a cardiac explants approach that is similar to the organ cultures used below. These experiments served as a starting point for the VEGF164 dose/response experiments. Collagen scaffolds were soaked in a range of concentrations (1, 10, 20, 50 ng/mL) of recombinant mouse VEGF164 (R&D Systems) for 24 hr. The scaffolds were then UV cross-linked at 1,600 × 102 μjoules/cm2 using a Spectroline UV cross-linker. This increases the cross-linking of the collagen fibers and traps the VEGF164 in the scaffold. The amount of UV cross-linking of the collagen scaffold does affect growth factor release from the scaffold (see Supplemental data for details). Endothelial cell density plateaued at 10 ng/ml, and thus this concentration was used in the studies reported here. This is similar to previous reports using the quail coronary development model (Yue and Tomanek,2001). Isolated quail PE cells were placed on the scaffold and allowed to attach for 12–16 hr, after which fresh media was added to the cultures. All scaffolds were cultured in 10% fetal bovine serum in Dulbecco's modified Eagle's medium supplemented with antibiotics.
Endothelial Cell DNA Synthesis Assay
The BrdU assay was utilized to determine the effect of VEGF164 on the proliferation of PE-derived endothelial cells. PE cells alone, PE cells cocultured with cardiac myocytes, and PE cells cultured on the tubular scaffold containing VEGF164 were cultured for 48 hr. After 48 hr of culture, these samples were labeled with BrdU for 2 hr according to instructions supplied by the manufacturer (Roche Applied Sciences, Inc., 2006). The BrdU labeling index for proliferating endothelial cells was defined by the ratio of the number of BrdU-positive/endothelial-positive nuclei divided by the total number of endothelial nuclei within microscopic fields. DAPI was used to stain cell nuclei. Directly conjugated isolectin GS-IB4 (Invitrogen) was used to stain endothelial cells. A minimum of 200 endothelial cells from at least 10 fields was counted per time point for each experiment. Significance differences between experimental and control groups were determined using the SigmaStat 3.0 Suite of programs. Significance (P < 0.05) was determined for the BrdU labeling index between two time points using a Kruskal-Wallis one-way analysis of variance based on ranks.
Endothelial Tube Formation Assay
The endothelial tube formation assay was utilized to determine the number of vessels present on the tubular scaffold. The endothelial tube formation assay was conducted by quantifying the number of endothelial tubes for a minimum of 15 fields per time point using Image Pro Plus version 126.96.36.199. An endothelial tube is defined as a continuous chain of cells with oval nuclei that are positive for endothelial markers. The standard error was determined by using the SigmaStat 3.0 Suite of Programs. Significance (P < 0.05) was determined for the endothelial tube assay using an unpaired t-test.
VEGF Inhibition (Tyrphostin) Assay
In sum, 10 μg/mL of SU 1498 (tyrphostin, Sigma-Aldrich), an inhibitor of VEGFR-2 (flk-1) receptor tyrosine kinase, was added to PE/CM cocultures 12–16 hr after PE implantation. SU 1498 was added to the culture media every 24 hr for a period of 7 days for the endothelial cell density assay and 14 days for the endothelial tube formation assay.
Organ Culture EMT Assay
HH26 chick hearts were cultured in serum-free M199 medium containing AdlacZGFP adenovirus as in Dokic and Dettman (2006). Medium was supplemented with 10 ng/mL VEGF164 (R&D Systems) or 1 ng/mL TGFβ1 (Preprotech). After 48 hr, hearts were fixed in 4% formaldehyde (v/v) for 30 min and incubated in lacZ stain buffer, 0.2M potassium-ferrocyanide, 0.2M potassium-ferricyanide, 10 mM MgCl2, 1 mg/ml BluoGal (Fisher) in PBS at 37°C for 1 hr. Hearts were then embedded in paraffin and 10-μm sections were collected such that each heart was completely sectioned using a Reichert-Jung 2030 microtome. Every third section was then assayed for the presence of blue cells. Cells were scored as either epicardial or mesenchymal (subepicardial, or myocardial). The percentage of mesenchymal cells was determined for each heart. A minimum of 5 hearts was assayed for each group. Significant differences between groups were determined with a z-test (P < 0.05).