Using neural crest tracings in mouse embryos (Wnt1-Cre-lacZ), we studied the patterning of cardiac neural crest cells during development. Participation of neural crest cells in the formation of the vascular media could not be excluded, although epicardium-derived cells have hitherto been considered responsible for formation of the coronary arterial smooth muscle cells. The endothelial cells of the coronary network derive mostly from the endothelium of the sinus venosus–liver region by vasculogenesis and angiogenesis. However, an epicardium-derived cell origin of some endothelial cells cannot be ruled out. The coronary vasculature is closely related to the differentiating Purkinje network, but isolated epicardium-derived cells are also associated with Purkinje cells. After ablating the pro-epicardial organ in quail embryos, we found severe malformations in the myocardial architecture, leading to the hypothesis that epicardium-derived cells give instructive signals to the myocardium for proper differentiation of the compact and the trabeculated compartments. (Tex Heart Inst J 2002;29:255–61)

The heart is a conglomerate of many differentiated cell types that assemble from precursors in a specified period of development. In a primitive streak-stage embryo, precursors of the cardiomyocytes and endocardial cells can be encountered in the bilateral plate mesoderm in the form of cardiogenic fields. After midline fusion, the fused heart fields differentiate into a myocardial tube, lined on the inner side by the endocardial endothelium. ¹ Only after looping of the single heart tube do other cell populations join the heart tube. These cell populations derive from the pro-epicardial organ ² or PEO that is part of the serous lining of the body wall, close to the liver and sinus venosus. The PEO adheres to the dorsal surface of the heart tube and migrates over its outer surface until completion. ³ The epicardium-derived cells (EPDC) invade the muscular wall and the endocardial cushions and differentiate into several cell types, e.g., interstitial fibroblasts, coronary smooth muscle cells, and adventitial fibroblasts. Their role in the development of the cardiac conduction system will also be discussed. A further population of cells that join the heart is derived from the primitive nervous system, the neural plate. These neural crest cells migrate in an early stage of development through (for example) the branchial arches and target both the arterial and the venous poles of the heart.

Scope of the EPDC. The PEO starts as a multivesicular outcropping (Fig. 1) of the mesothelial body wall. As it covers the highly vascularized sinus venosus–liver area, it attracts vascular sprouts that soon invade the hyaluronan-rich inner space of the vesicles. This vesicular space will transform into the subepicardial space upon migration of the epicardium over the initially bare myocardial outer surface of the heart tube. The subepicardial space will be used by endothelial branches, by neural crest-derived cells, and by EPDC. It is important to note that the early epicardial cells have migration capacities. In a chicken embryo, the migration takes place between Hamburger and Hamilton ⁴ (HH) stages 16 and 25. Near the end of this period, the epicardial lining starts to show epithelial–mesenchymal transformation (EMT), giving rise to a secondary mesenchymal population, found initially in the subepicardial space, and later also in the myocardium. ⁵ It is important to note that these EPDC are highly invasive.

Fig. 1 The pro-epicardial organ (arrow) in a stage 16 quail embryo, which has been stained with the HNK1 antibody and cleared for transparency. This antibody will also stain the atrioventricular and outflow tract cushions and the developing nervous system. (more ...)

During several waves of invasion, EPDC target various locations in the heart, such as the subendocardium, the vascular endothelial tubes, and the cardiomyocytes. The EPDC can be traced by marking, by immunohistochemistry, by in situ hybridization, or by heterospecific transplantation (chicken-quail chimeras). Furthermore, by transgenesis and knock-out strategies, “EPDC-specific” genes can be mutated, which results in specified phenotypes. These can lead to coronary abnormalities and bleeding and therefore might be embryo-lethal.

White Leghorn chicken and Japanese quail embryos were staged according to the criteria of Hamburger and Hamilton. ⁴ Normally developed embryos and chicken-quail chimeras have both been studied; in the latter instance, a quail PEO was transplanted into a chicken pericardial cavity. Furthermore, we have inhibited outgrowth of the chicken PEO, sometimes combining this with replacement by a quail PEO. ⁶ Immunohistochemistry has been used to determine the differentiation of cell populations and to trace the location of quail cells in a chimeric environment.

Chimerization. After the eggs had been incubated for about 3 days, the embryos had developed to HH15–18. The egg was opened and the vitelline membrane locally removed. Subsequently, the pericardial cavity of the embryo was approached at HH15 through the naturally existing body wall hiatus, or at HH16–18 through a slit in the amniotic and pericardial membranes, by means of watchmaker's forceps. A piece of PEO, together with an adjacent piece of liver tissue, was harvested from a quail embryo of similar incubation time and developmental stage and introduced into the chicken pericardial cavity as described earlier. ⁷ Care was taken to insert the additional PEO close to the region of the host PEO. In this way, both chick and quail spread in consort, resulting in a mosaic epicardial lining of the myocardium.

Ablation. We obtained inhibition of outgrowth of the epicardium by ablation, as described by Männer, ⁸ which resulted in complete or partial absence of outgrowth of the epicardium from the venous pole. ⁶ This was confirmed by cytokeratin staining. In a separate set of experiments using HH15 chicken embryos, we implanted a quail PEO (HH15–17) immediately after the inhibition (again by ablation) of outgrowth of the host PEO. ⁶ This procedure resulted in a delay in epicardial outgrowth of approximately 1 day. In all embryos, quail epicardium covered the greater part of the heart. ⁶ Embryos were serially sectioned and subjected to standard immunohistochemical procedures.

Reporter Mouse. The Wnt1 gene is specifically expressed in neural crest cells. On the basis of this finding, a lacZ-containing construct ⁹ has been made that produces blue neural crest cells in a transgenic mouse after the above-described enzyme treatment. We restudied the material published by Jiang and colleagues, ⁹ in order to investigate the cardiac neural crest population migrating to the aortic root.

Immunohistochemistry. For detection of quail cells, an anti-quail nuclear antibody, QCPN, was used. The QCPN antibody, in combination with the quail-specific anti-endothelium QH1 antibody, ¹⁰ enables the observer to differentiate between quail epicardial cells and endothelium. To follow the differentiation of the epicardial cells, we also used the 1E12 smooth muscle antibody, ¹¹ the procollagen type 1 antibody M38, ¹² and the anti-actin muscle marker HHF35. ¹³ The QCPN, QH1, M38, and HHF35 antibodies were obtained from the Hybridoma Bank (Developmental Studies Hybridoma Bank; University of Iowa; Iowa City, Iowa). To follow the development of the conduction system, we used the EAP300 antibody. Detailed information on the lab protocols can be found in Gittenberger-de Groot and colleagues. ⁵

To describe the function of EPDC properly, we have to investigate a number of steps in the developmental pathway. These include 1) migration as a contiguous layer of cells over the myocardial surface; 2) epithelial–mesenchymal transformation; 3) invasion of the cardiac tube; 4) local differentiation into several cell types (smooth muscle cells and fibroblasts); and 5) putative interaction with cardiomyocytes, which may differentiate into Purkinje cells. A schematic representation is shown in Figure 2.

Fig. 2 Schematic representation of the migration and differentiation steps of cells derived from the pro-epicardial organ (the EPDC give rise to smooth muscle cells and fibroblasts) and from the liver/sinus venosus microvasculature (which gives rise to (more ...)

Malfunctioning of PEO and EPDC in 1 of these stages may result in congenital cardiac anomalies, as we have shown by ablation (in part) of the pro-epicardial organ (see below). We also have to realize that the importance of the EPDC is not restricted to prenatal life. It has been estimated that less than 50% of the cardiac wall consists of cardiomyocytes.* Considering the small, nominal contribution of the neural crest cells to the heart, this leaves the majority of the heart cells to EPDC.

1) Migration as a Contiguous Layer of Cells. In the vertebrate species investigated, the pro-epicardial vesicles may differ in size and number, but invariably they adhere to the pericardial surface of the myocardial tube. This initially takes place at the dorsal side of the heart close to the atrioventricular transition, because this location is adjacent to the proliferation site of the PEO vesicles. In some species, as described by Van den Eijnde and associates in the rat, ¹⁴ the vesicles detach from the PEO to float in the pericardial cavity and to adhere secondarily to the cardiac tube. In chicken and quail, this has not been observed. In the latter species, the vesicles adhere, flatten, and expand to cover the cardiac tube. In the chicken embryo, this process takes several days (between HH16–25) and follows a specified pattern. The atrioventricular and the interventricular groove are the 1st to be covered, and the ballooning chambers and the outflow tract are the very last. ³ The outflow tract is covered in a peculiar way. We have demonstrated this by ablation of the PEO in HH16 chicken embryos. ⁶ In the set of experimental embryos that did not survive beyond HH30 due to heart failure, an epicardial covering was present as a collar surrounding the outflow tract. Histologic analysis showed that a secondary epicardial covering, the arterial epicardium, derived from the mesothelium surrounding the arterial pole. Since the embryo did not survive, the length of this collar was limited to approximately the distal half of the outflow tract. Although coronary vessels could not have been established, due to the absence of the PEO and concomitant vessels, irregular endothelial profiles that resembled, in sections, vasa vasorum were connected to the lumen of the aortic root.

During completion of covering, both from the venous and arterial poles, the subepicardial space will be used by invading endothelial cells, as well as neural crest cells. At the venous pole, the endothelial cells and tubes derive from the microvasculature of the hepatic primordium and the sinus venosus. These extend in a lattice-like network, initially confined to the subepicardial space, but before HH25, endothelial cells are already found penetrating the myocardial wall. Some endothelial cells migrate across the complete thickness of the myocardium and might touch the endocardial lining, ⁷ without actually establishing anastomoses between the cardiac lumen and the growing coronary system.

In our laboratory, we have not been able to show that the epicardial cells function as stem cells for the coronary endothelial lineage. Using cytokeratin antibodies as an epicardial marker and endothelial-specific markers to characterize the endothelium, we have not observed double-stained cells. However, Muñoz-Chapuli and collaborators ¹⁵ recently showed that endothelial cells may also carry epicardial markers; therefore, the possibility that EPDC contribute to the coronary endothelium cannot be ruled out. At the arterial pole, irregular connections of the microvessels with the lumen are found only after ablation of the PEO, but we lack data to describe this convincingly because stages before HH29 have not been analyzed in depth. At stage 19, neural crest cells start to arrive at the arterial pole ¹⁶ to form parts of the inner framework of the heart, and at stage 25 they settle also in the subepicardial space to form the autonomous cardiac ganglia. ¹⁷

2) Epithelial–Mesenchymal Transformation. The transformation of an epithelium into mesenchymal cells (EMT) is a mechanism encountered in many stages of embryonic development, including ingression during germ-layer formation, detachment of neural crest cells from the neural plate, and cushion formation in the endocardium. Moreover, disease processes—such as metastasis from epithelial precursors and intimal thickening early in atherosclerosis—present with mechanisms of EMT. In regard to the epicardium, EMT has been observed over its entire surface to give rise to mesenchymal cells that ingress the subepicardial space. These cells move a step farther and invade the myocardial tube to differentiate into fibroblast-like cells and smooth muscle cells (SMC). This takes place in several waves during development. The last wave probably occurs in the atrioventricular groove and relates to the ensuing differentiation of coronary smooth muscle cells, preceding the onset of coronary arterial ostia formation in the aortic sinuses. During this period, the up-regulation of the matrix metalloproteinase MMP-9 is noteworthy, as has been shown by using subtractive hybridization, which compares premigration (stage 17) with postmigration (stage 29) epicardial cells. We are currently investigating whether MMP-9 is expressed only during migration or also earlier, during EMT, and possibly later, extending into the next phase of development (myocardial invasion).

3) Invasion of the Cardiac Tube. The cardiac tube will be invaded as well by endothelial cells derived from the omphalomesenteric veins that join the sinus venosus, and by epicardium-derived mesenchymal cells. The endothelial precursors and tubes give rise to the microvascular lattice (stages 26–32) encased between the cardiomyocytes. The primitive coronary lattice is connected only to the sinus venosus. The coronary system is filled by sinus venosus pressure, which is rather slow, but sufficient to serve the early stages of development. Definitive completion of the lattice also requires a period in which pruning takes place. ¹⁸ The coronary ostia and some connections with the right atrium are formed around stage 32, and from that moment on, the blood flow in the coronary system is reversed in direction: it pulsates from the aorta towards the atrium and the sinus venosus. Because this is aortic pressure and unidirectional, it will be fast and efficient for the stages to come. At present, we assume that oxygen demand of the cardiac muscle governs the ingrowth and patterning of the endothelial tubes, probably through a HIF-1α (hypoxia-inducible factor-1α), ¹⁹ a VEGF (vascular endothelial growth factor), ²⁰ and an angiopoietic pathway. ²¹ It is unclear what signals guide the endothelial cells to come into contact with the aorta (the normal point of ingression), rather than with the pulmonary trunk.

Upon connection of the coronary microvascular lattice to the aorta, the next phase of vessel-wall differentiation takes place. Epithelial-mesenchymal transformation in the epicardium covering the atrioventricular (AV) groove produces a large number of mesenchymal cells that are apparently recruited to the coronary ostia, where they start to differentiate into smooth muscle cells and fibroblasts. In quail-chicken chimeras, in which a quail-derived PEO was added to the otherwise normal chicken heart of the same stage, we observed quail cells that were positive for smooth muscle actin and still others positive for procollagen. These observations were made at the same stage (HH32) in which connection to the aorta is established. From an engineering point of view, this coincidence has the very practical effect of reinforcing the vascular wall with SMC and fibroblasts, which copes with the sudden increase in pulsatile blood pressure and flow.

The signaling pathway, however, is still unclear. The differentiation of the media probably depends on growth factors from the FGF (fibroblast growth factor) ²² and PDGF (platelet-derived growth factor) families. ²³ It has to be determined whether differentiation of the SMC along the coronary arterial wall is a matter of migration of SMC precursors along the basement membrane of the endothelial tube (starting at the coronary ostium and ending at the apex of the heart), or whether SMC precursors are locally recruited from EPDC that have locally invaded the heart tube. The results of experiments in chimeras favor the latter explanation. Chimeras always display a mosaic of quail- and chicken-derived areas. When quail cells are found as differentiated SMC in the wall of a vascular segment, the neighboring epicardium is likewise of quail origin. If SMC were to migrate distally over a vascular segment, the presence of these SMC clones would be unrelated to the species of the overlying patch of epicardium.

4) Local Differentiation into Several Cell Types. It is evident that EPDC differentiate into several cell types, including interstitial fibroblasts, coronary smooth muscle cells, vascular adventitial fibroblasts, and, as described above, perhaps even endothelial cells. As far as we know, they (normally) do not give rise to cardiomyocytes or to cardiac ganglia. This has been proved by physical marking of the PEO and EPDC in (for example) the chicken-quail chimera model. Immunohistochemical detection using antibodies against more or less specific epitopes helps to determine the stage of differentiation. It must be kept in mind, however, that antibodies and probes do not offer full proof as cell-lineage markers, because genes are expressed and switched off, and proteins may appear and disintegrate, suddenly rendering a once-positive cell population inconspicuous, or vice versa.

The epicardium-derived interstitial fibroblasts expressing pro-collagen can be encountered throughout the myocardium. They will take part in the formation of the fibrous elements of the heart. As EPDC have also been encountered inside the AV cushions, ⁵ these may very well form the fibrous skeleton of the heart. We assume that fibrosis in diseased hearts has its basis in the malfunction of EPDC. Fibroblasts also remain as the last survivors in ischemic parts of the heart, since they are less sensitive to oxygen deprivation.

The smooth muscle cells of the coronary vasculature will envelop both the arterial and the venous sides of the system. The determination that a vessel shall become arterial or venous probably depends on the Eph–ephrin signaling pathway, as it does elsewhere in the vascular system. ²⁴ A particular transition occurs in the proximal part of the venous media, where EPDC connect to the atrial myocardium. All of the major veins, including the coronary veins, are enveloped in an atrial myocardial sleeve. The transition from epicardial origin to SMC is sudden, yet a proper communication between SMC and cardiomyocyte will have to exist in order to develop functional drainage to the atrium.

An interesting difference between birds and mammals, concerning the origin of coronary arterial smooth muscle cells, may exist. The Wnt1 neural crest reporter mouse ⁹ was re-investigated with respect to the vascular and cardiac contribution. As expected, many patterns well known from the earlier studies in birds were also found in this model. This includes the participation of neural crest (NC) cells in the formation of the media of the aorta and pulmonary trunk, the aortopulmonary septum and semilunar valve leaflets, and the cardiac ganglia. However, the presence in the myocardium of a multitude of apparently (because they were blue) NC-derived cells has not been described in avian embryos, nor in the other NC-reporter mice, such as Connexin-43 ²⁵ and Pax-3. ²⁶ On closer examination, many of these NC cells surround coronary arteries (Fig. 3). Although the sections did not enable exact determination of their precise association with the coronary endothelium, it was likely that the NC cells were outside the endothelium. Some cells were very close to the endothelium of the main arteries, which is reminiscent of smooth muscle cells. At embryonic day 15.5 (developmentally comparable to stage 31–33 chicken embryos), we encountered a mosaic of blue cells surrounding both coronary arteries. The presence of a mosaic suggests that not all smooth muscle cells derive from the neural crest. The remaining cells probably derive from the epicardium. From this material, we conclude that, in the Wnt1 reporter mouse, NC cells accompany the coronary arteries, although we cannot yet determine the differentiation pathway of these cells.

Fig. 3 Two sections of the base of a 15.5-day transgenic mouse heart at the level of the aortic semilunar valve. The Wnt1-Cre-lacZ neural crest reporter is visualized (dark cells) after staining for the presence of β-galactosidase. Neural crest (more ...)

5) Putative Interaction of EPDC with Cardiomyocytes. Epicardium-derived cells not only differentiate into various cell types, but interact with cardiomyocytes in both a functional (as described above) and instructive manner. After ablation of the PEO in a stage 16 chicken embryo, survival is limited to stages 29–30. The main reason for death of the embryo is probably not lack of coronary perfusion, because connection of the coronary lattice with the aorta does not take place before stage 32. Rather, the myocardial architecture is very abnormal after PEO ablation: the compact myocardial wall is extremely thin, the ventricular trabeculations are decreased in number and size, and the AV cushions are nearly absent. This failing heart leads eventually to death. ⁶ Apparently, EPDC are instructive for proper myocardial organization. We are currently investigating the basis of the interaction between EPDC and cardiomyocytes.

Furthermore, EPDC are present in the neighborhood of the differentiating conduction system. This holds for the subendocardially located Purkinje cells, ⁵ as well as for the periarterial fibers. ^27,28 Although the cardiac conduction system is considered to be a separate lineage of cardiomyocytes, ²⁹ the differentiation of cardiomyocytes into Purkinje fibers probably depends on instruction by EPDC, via secretion of (for example) the vascular active substance endothelin-1. ²⁸

Epicardium-derived cells play a very important role in cardiac development and differentiation, the extent of which becomes more and more apparent. Specific genes have been characterized, such as bves ³⁰ and capsulin, ^31,32 whereas knockouts of other genes have resulted in epicardial phenotypes, such as FOG-2, ³³ VCAM, ³⁴ and α4 integrin. ³⁵ Still other gene products of the epicardium give rise to potent signaling molecules, such as RALDH2 (retinoic acid-synthetic enzyme), responsible for retinoic acid production. ^36,37 The potentials of the EPDC in embryonic stages seem to be enormous in view of their differentiation and signaling capacities. It is tempting to consider these cells as a cellular resource to repair ischemic and failing hearts, delivering endothelial, SMC, and fibroblastic cells to sites of injury that need additional sustenance. It is even more tempting to consider the use of adult epicardial cells from the patient's own heart, thereby avoiding rejection of transplanted cells. It has to be kept in mind that stem cell repair requires not only “new” cardiomyocytes for adequate contraction, but the assistance of numerous noncardiomyocytes for adequate perfusion of the heart muscle and for structural integrity of the myocardium.