It has been well established that the epicardium and coronary vascular system of all vertebrates arise from an extracardiac source (as reviewed by Manner et al., 2001). Morphologic and experimental studies point to a transitory structure, the proepicardium (PE), a grape-like cluster of mesothelial cells derived from the pericardial serosa that extends to and migrates over the myocardial surface during the looping, tubular stage of heart development (Manasek, 1969; Ho and Shimada, 1978; Viragh and Challice, 1981; Komiyama et al., 1987; Kuhn and Liebherr, 1988; Fransen and Lemanski, 1990; Manner, 1992; Mikawa and Fischman, 1992; Viragh et al., 1993; Munoz-Chapuli et al., 1994; Mikawa and Gourdie, 1996).
The translocation of PE cells across the pericardial coelom to the myocardium has been proposed to occur by either one of two mechanisms unique to different species (Manner et al., 2001). One mechanism, based on studies in fish and rodent embryos, suggests that cells from the pericardial serosa in the region of the septum transversum detach as free-floating, multicellular aggregates or vesicles that passively float across the pericardial cavity to eventually attach as independent patches on the myocardium (Komiyama et al., 1987; Kuhn and Liebherr, 1988; Munoz-Chapuli et al., 1994). On the myocardial surface, they flatten out, eventually coalescing as a simple squamous epithelial sheet. In the mouse, this process occurs between days 9 and 11 of embryonic development (Komiyama et al., 1987).
In contrast, avian and amphibian species exhibit a PE tissue bridge that extends across the coelomic cavity between the serosal and myocardial surfaces. In the chick, this PE anlagen develops on the ventral surface of the sinus venosus opposite the dorsal aspect of the heart tube. By an unknown mechanism, the PE reaches the atrioventricular (AV) junction of the looped tubular heart by Hamburger and Hamilton stage (HH) 16 (Ho and Shimada, 1978; Shimada et al., 1981; Hiruma and Hirakow, 1989; Manner, 1992) and then begins a continuous sheet-like migration over the myocardial surface, eventually covering the heart by HH23 in chicks (Hiruma and Hirakow, 1989).
Common to all vertebrate species, once the epicardial sheet has formed, some its cells undergo an epithelial–mesenchymal transformation that gives rise to connective tissue cells of the subepicardium and myocardium, and the angiogenic precursors of the coronary vessels (Mikawa and Fischman, 1992; Dettman et al., 1998; Vrancken Peeters et al., 1999).
In this study, we have addressed the potential mechanism(s) by which cells of the PE reach the myocardium. By using a variety of morphologic and cytochemical procedures, we have reexamined the structure of the PE and have discovered an extracellular matrix bridge (ECMB) between the pericardial serosa and the AV junction of the myocardium. Degradation of this matrix with heparinase is shown to impair formation of the epicardium. These data are compatible with a role for the ECMB in extension of the nascent PE to the myocardium. It remains to be seen whether this bridge serves a structural or regulatory role in epicardial morphogenesis.
Although development of the PE in chicken embryos has been described by both light and electron microscopy (Manasek, 1969; Ho and Shimada, 1978; Shimada et al., 1981; Hiruma and Hirakow, 1989; Manner, 1992), there remains uncertainty about how cells of the PE reach the myocardium and are specifically targeted to the AV junction. Close attention was paid to PE morphogenesis with a particular focus on the potential role of ECM components in PE cellular migration.
Morphogenesis of the PE
The PE was first detectable by light microscopy at stage HH14 (Figs. 1, 2). It arises as a dome-shaped outgrowth of cuboidal mesothelial cells on the surface of the right horn of the sinus venosus, just ventral and cranial to the developing liver, at the ventral lip of the anterior intestinal portal (Fig. 1). Its approximate dimensions at this stage were 150 μm in width (parallel to the body wall) and 120 μm in height (perpendicular to the body wall). The emergence of the PE was cotemporal with the formation of the cervical flexure and dextral rotation of the embryo on its craniocaudal axis. At this stage, the heart tube was dextrally looped and C-shaped and had well demarcated sinoatrial, ventricular, and outflow subdivisions. The AV junction of the myocardium was situated ∼25–75 μm from the outermost cells of the PE, across the coelomic cavity.
Three distinct cell types could be identified by light microscopy. The pericardial surface of the emerging PE was composed of a simple cuboidal epithelium overlying an internal cluster of stellate mesenchymal cells suspended within a large, weakly stained, extracellular matrix (Figs. 2b, 3a). Deep to this internal matrix compartment was a stratified cuboidal cell layer, four to six cell layers thick, adjacent to the endothelial lining of the sinus venosus. Thin cellular extensions linked the mesenchymal cells with the overlying cuboidal epithelium and with the deep stratified epithelium. These intercellular contacts contained desmosomal adherent junctions (data not shown). Cells of the outermost layer of the PE were linked to each other by a series of interrupted tight junctions that collectively measured up to 1 μm in length (Fig. 3b,c). These tight junctions were found at the apical boundaries of their lateral cell borders. Basal to the tight junctions were desmosomes ∼ 0.3 to 0.4 μm in length, usually one on each lateral border (Fig. 3b). Intermediate filaments extended from the desmosomes into the cytoplasm. PE cells contained a rich content of free polysomes and rough-surfaced endoplasmic reticulum, and their nuclei were euchromatic, typically containing a single prominent nucleolus. Sparsely distributed microvillus-like projections extended into the coelomic cavity from the apical plasma membrane (Fig. 3c). On their basal surface, long cytoplasmic extensions coursed into the internal matrix of the PE and typically contacted the mesenchymal cells. A variety of membrane-bounded vesicles were observed within the internal matrix (Fig. 3a).
Over the next few hours, the PE enlarged significantly. By stage HH15, it resembled a grape-like outgrowth, measuring ∼150–200 μm in diameter. It protruded from the ventral surface of the sinus venosus toward the AV junction on the lesser curvature of the heart (Figs. 4, 5). The major changes of the PE from the previous stage were an increase in size, largely caused by an expansion of the internal matrix-filled cavity, and an elaboration of the apical villi further into the coelomic cavity toward the heart. At its caudal boundary, the PE abutted the developing liver (Fig. 4c,d). Infrequent grape-like projections, reminiscent of a nascent PE, were also observed at other sites on the coelomic wall (Fig. 4d). We have not observed them touching the heart, but it cannot be ruled out that these outgrowths of mesothelia might also contribute to the future epicardium of the heart.
Serial sections through the thoraces of embryos stained with ruthenium red revealed regions of fibrillar and granular staining at localized sites within the coelomic cavity (Fig. 4a–d). The most prominent of these sites was the gap between the PE and the AV junction (Fig. 4b,c), but some of this matrix material was also detected adjacent to the PE, between the sinoatrium and myocardium of the lesser curvature (Fig. 4a). The coelomic matrix material also stained intensely with Alcian blue (Fig. 5a,b). This positive reaction with Alcian blue and ruthenium red is suggestive of a proteoglycan within the matrix. Of most significance was the constant association of PE outgrowth with this matrix spanning the gap between PE and myocardium (Figs. 4b–d, 5a,b). We have termed it the ECMB.
Earliest contact between the PE and the myocardium was observed at HH16 (Fig. 6a). Villus extensions at the apices of the PE contacted the apical surfaces of the myocytes. Both of these cell surfaces stained intensely with ruthenium red by transmission electron microscopy (TEM; Fig. 6b,c). This densely stained layer exhibited a cobblestone-like pattern of globular material (∼50–100 nm in diameter). Projecting into the gap between the adjacent cells were straight filaments (∼20–24 nm diameter) that bridged between the cells (Fig. 6c,d). The filaments appeared to be bound to the globular material on the cell surface (Fig. 6d).
After the initial contacts of the myocardium at stage HH16, extensive migration of the PE over the myocardium began at HH17. Contact of the myocardium by the PE villi initiated a process reminiscent of a flowering rosebud (Fig. 7a). The PE cells were reflected on the apical surface of the myocardium; thus, the apical surface of each PE cell maintained its original orientation with its apex in direct contact with the coelomic cavity. The basal surface of each PE cell faced the apical side of the myocardium, separated from the myocytes by a thin matrix-filled cavity spanned by cell extensions from the PE. Opposing surfaces of myocytes and PE cells stained intensely with ruthenium red (Fig. 7a,b). During PE migration, the leading cell exhibited close contact with the apical surface of underlying myocytes (Fig. 7a,b), whereas the trailing cells of the migratory sheet were more loosely associated with the myocardium. In addition, the coelomic surface of the migrating PE sheet and neighboring PE villi appeared to be linked by matrix tracks of globular and filamentous material. (Fig. 7b–d). These tracks appeared to emanate from villi of PE cells not in direct contact with the myocardium (Fig. 7b,c).
Although we could not observe each step of this morphogenetic process in real time, we interpret these movements in the following manner: the intercellular junctions between the foremost PE cells dissociate, creating cells with a free edge that then evert and spread radially as a monolayer from the initial contact point on the myocardium (Fig. 7a). The interior of the PE matrix cavity is then in direct confluence with the myocardial surface. It is into this subepicardial compartment that delamination of angiogenic progenitors will occur, followed by cellular invasion of the myocardium.
Immunocytochemical Analysis of Matrix Components
To identify the composition of the ECMB, we used a panel of antibodies against matrix components. These included antibodies to heparan sulfate, chondroitin sulfate, fibronectin, keratan sulfate, and laminin. Sections were double-stained with anti-cytokeratin to demarcate the PE cells. Parasagittal cryosections of stage HH15 embryos stained with anti-heparan sulfate antibody revealed strong reactivity with the internal matrix compartment of the PE (Fig. 8a,b). Before attachment of the PE, weak immunostaining for heparan sulfate was observed in the acellular matrix spanning the coelomic gap between PE and myocardium (Fig. 8a). The ECMB was also positive for fibronectin (Fig. 8c) but negative for laminin, keratan, and chondroitin sulfate. At this stage, the myocardium exhibited weak reactivity with the antibody to heparan sulfate (Fig. 8a). Fibronectin was restricted to the basal surface of the myocardium and the cardiac jelly (Fig. 8c). Chondroitin sulfate appeared to restricted to the cardiac jelly and keratan sulfate antibody exhibited little or no reactivity within the PE or heart tube (data not shown). After contact of the PE with the myocardium (HH17), there was increased staining for heparan sulfate at the myocardial–PE interface (Fig. 9a), but the myocardial surface not in contact with the PE had relatively little heparan sulfate staining.
To explore the role of heparan sulfate in growth and attachment of the PE to the myocardium, a cocktail of heparinase (heparitinase I, II, and III) was injected into the pericardial coelom before PE attachment (HH15). Control injections included a heat-inactivated enzyme solution (boiled for 2 min) and injections of the dilution buffer. Embryos were fixed either 24 or 72 hr after injection. Of 30 injected controls (heat inactivated and buffer injections), 21 survived for 24 hr with similar mortality in both groups (Table 1). The surviving embryos exhibited the normal appearance of embryonic day 4 embryos, including limb buds, head enlargement, a prominent cervical flexure, and normal neural tube and somite differentiation. Based on examination of living embryos under a dissecting microscope, the hearts appeared to pulsate normally and sustain the systemic circulation. In hematoxylin and eosin (H&E)-stained paraffin sections, the structure of the endocardium, myocardium, and epicardium appeared normal (Fig. 10).
Table 1. Results from Pericardial Injections
aHep, heparitinase; Inact, inactivated.
Hep I, II, III
Twenty-four hours after heparinase injections, 17 of 25 embryos were still alive. Of those 17 surviving embryos, five exhibited abnormal development, whereas the remaining 12 appeared unaffected. The most notable abnormality of the five that exhibited abnormal development was the absence of an epicardium (Fig. 11). Of two heparinase-treated embryos that were examined 72 hr after injection, one also exhibited a complete absence of the epicardium. In all the abnormal cases, the PE had enlarged with villous projections directed ventrally into the coelomic cavity. However, these did not attach to the myocardium. The heart exhibited enlargement of atria and ventricles, although the ventricular walls appeared thinner and contained fewer trabeculae than the control embryos (Figs. 10, 11). Additional features noted in the 72 hr heparinase-injected embryos were the failure of neural tube, mandible, and palatal fusion, plus reduced limb bud outgrowth. The mechanism of these extracardiac abnormalities is currently unknown. There may have been some enzyme leakage from the injection site, possibly leading to systemic distribution of the heparinase. However, that possibility should be the focus of another study.
In this report, we describe the morphologic transitions of the PE from emergence on the pericardial serosa and extension to the AV junction of the myocardium, to its contact and initial migration over the myocardium. Evidence is presented for an ECM bridge extending between the PE and the myocardium. This bridge contains heparan sulfate and fibronectin, stains strongly with ruthenium red and Alcian blue and exhibits the ultrastructural characteristics of proteoglycans present in the extracellular matrix of other embryonic systems (Soto-Suazo et al., 1999). Infusion of the pericardial coelom with heparinase impairs epicardial morphogenesis. These data suggest a role for the ECM bridge in PE morphogenesis.
It is now well established that the PE gives rise to the following cell types in the mature heart: (1) all epicardial constituents, including mesothelial cells of the visceral pericardium and cells of the subepicardial connective tissue; (2) all cells of the coronary arterial system, including endothelium, smooth muscle, and perivascular fibroblasts; (3) cells of the myocardial connective tissue. Cell lineage analyses after retroviral injections of the PE at stage HH16 have demonstrated that clones formed from the labeled cells consist of a single cell type. Thus, lineages had already been specified in the PE before retroviral labeling (Mikawa and Gourdie, 1996). By light and electron microscopy, we can distinguish at least three cell types in the developing PE that differ significantly in morphology. The first is the outermost simple epithelium in direct with the coelomic cavity; the second is the internal mesenchymal population; and the third, the multilayered population of cells directly adjacent to the sinus venosus. It remains to be seen if any of these types correspond to the independent cell lineages established by retroviral labeling.
The mechanism(s) governing PE extension and guidance to the myocardium is still uncertain. Our results demonstrate that enlargement of the PE is accompanied by a significant expansion of an internal matrix compartment that contains heparan sulfate and fibronectin. Earlier studies have demonstrated that this compartment in mice is rich in hyaluronic acid (Kalman et al., 1995), a substance known to promote cell migration in many embryonic tissues (Markwald et al., 1981). That this matrix compartment becomes confluent with subepicardial space upon eversion of the PE epithelium on the myocardium may be important in the facilitation of cell migration on and invasion into the myocardium.
In this report, we present evidence for an ECM bridge spanning the gap in the pericardial coelom between the PE and the AV junction of the myocardium. Two possible functions of this bridge are (1) a structural link akin to a tether line guiding PE extension to the lesser curvature, or (2) a signaling center that might serve to chemically regulate PE extension. There is no persuasive evidence for either hypothesis, but the enzymatic destruction of the bridge by heparinase suggests a role for this matrix in formation of the epicardium. There is a large body of published data (as reviewed by Princivalle and de Agostini, 2002) indicating that matrix-containing heparan sulfate proteoglycans are a rich store of growth factors that may play a direct role in paracrine signaling. Heparan sulfate proteoglycans are a diverse group of macromolecules with at least one heparan sulfate glycosaminoglycan chain linked covalently to a core protein. These are typically found in the extracellular matrix, the basement membrane, or integral to the plasma membrane (Timpl, 1993). The membrane-associated heparan sulfate proteoglycans syndecan and glypican function as low affinity coreceptors for basic fibroblast growth factor (FGF) and this interaction is required for the high affinity binding of basic FGF to its cellular receptor (Aviezer et al., 1994). Thus, it is conceivable that the matrix bridge might bind growth factors, e.g., FGF and vascular endothelial growth factor, known to be involved in vasculogenesis (Kazemi et al., 2002; Tomanek and Zheng, 2002).
In the mouse, it has been speculated that the mesothelial lining of the PE is aspirated when the ventricle contracts and that this may be a physical inducer for local mesothelial cells to proliferate and form villus protrusions (Kuhn and Liebherr, 1988). These villus protrusions were thought to bud off as aggregates or vesicles that float freely in the pericardial cavity before attachment to the myocardium (Viragh and Challice, 1981; Komiyama et al., 1987). Recently, it has been shown that the α4 subunit of α4β1 integrin, a cell surface receptor known to bind fibronectin (Wayner et al., 1989; Guan and Hynes, 1990) and VCAM-1 (Osborn et al., 1989; Elices et al., 1990), is required for attachment of the PE vesicles to the heart in the embryonic mouse (Sengbusch et al., 2002). Whether these vesicles actually float freely within the pericardial cavity remains to be proven. In the chicken embryo, we have never observed freely floating vesicles of the PE in the pericardial cavity but there may be species-specific differences in PE morphogenesis.
Based on all of these considerations, we would like to put forward a tentative model of PE growth, attachment to, and eversion on the myocardium (Fig. 12). Step 1: Upon expansion and differentiation of the PE (∼HH14–HH15), an extracellular matrix bridge is laid down between the leading edge of the PE and the myocardium. It is unknown if this is a product of the PE or myocardium. This matrix might store and present growth factors involved in directed outgrowth of the PE to the AV junction of the myocardium. Alternatively, the bridge might serve a structural role guiding PE extension to the myocardium. Step 2: Villous extensions of the PE epithelium contact the apical surface of the myocardium and this is accompanied by breakdown of the intercellular junctions between leading PE epithelial cells (∼HH16). Step 3: Cells of the PE exhibit an eversion process placing their basal surfaces in contact with the apical myocardium. This results in a direct confluence of the internal matrix compartment of the PE with the subepicardial compartment (∼HH17), which is followed by a rapid spreading of the PE cells over the myocardial surface.
Fertile White Leghorn chick eggs were purchased from SPAFAS (Charles River, NJ) and incubated at 38°C under humidified conditions. Embryos were staged according to the number of hours of incubation and by the staging system of Hamburger and Hamilton (Hamburger and Hamilton, 1951).
Chick embryos, ranging from stages HH14 to HH18, were fixed in ovo with 2% glutaraldehyde and 3% paraformaldehyde (EMS, Electron Microscopy Services, Ft. Washington, PA) in phosphate-buffered saline (PBS) by dropwise addition of fixative onto the embryo. Care was taken to not disrupt the integrity of the heart region and embryonic membranes during dissection. To promote infiltration of solutions into the pericardial coelom, small openings were cut in the pericardial membrane. At this point, the embryos were divided into three groups. One group was used for routine transmission electron microscopy, one was used for ruthenium red staining, and the last group for Alcian blue staining. For the second and third groups, whole embryos were immersion fixed for 12 hr at 4°C with either 0.1% ruthenium red (EMS), 3% glutaraldehyde (EMS) in 0.1 M cacodylate buffer (pH 7.4), or 0.05% Alcian blue (Sigma, St. Louis, MO), 2.5% glutaraldehyde, and 2% paraformaldehyde (EMS) in 0.1 M cacodylate buffer (pH 7.4), respectively. Samples were then washed with cacodylate buffer containing the same dye concentration (3 × 20 min). Alcian blue samples were subsequently dehydrated through ascending ethanols (50, 70, 85, 95, 100%) containing dye. The ruthenium red–stained samples were post-fixed with 1% osmium tetroxide in cacodylate buffer containing 0.1% ruthenium red, washed with distilled water (3 × 20 min), and en bloc stained with 2% aqueous uranyl acetate. The samples for routine TEM were fixed in 1% OsO4 and en bloc stained in 2% uranyl acetate. All tissues were then dehydrated in ascending ethanol concentrations, immersed in a 50:50 mixture of Spurr's epoxy resin (EMS) and 100% ethanol for 1 hr, and then in pure resin overnight. Embryos were embedded in rubber molds with fresh resin and polymerized for 12 hr at 60°C. Serial cross-sections at 50-μm intervals were cut through the thoracic region and thick sections collected on glass slides, stained for 30 sec at 60°C with 1% toluidine blue, washed with distilled water, and cover-slipped with Cytoseal 60 (Stephens Scientific, Kalamazoo, MI). Sections were examined by light microscopy and photographed with a color digital camera. For TEM, blocks were trimmed, thin-sectioned (90 nm thick), and collected on copper grids. Sections were stained with 2% uranyl acetate and lead citrate for 10 min each and examined with a JEOL-100X transmission electron microscope (Tokyo, Japan). Images were recorded on large format Kodak negatives, scanned at 900 dpi on a flatbed scanner, and imported into Photoshop (Adobe) for image processing.
Embryos between the stages of HH14 and HH18 were carefully excised and immersed in chilled periodate-lysine-paraformaldehyde (2%) fixative (McLean and Nakane, 1974) for 2 hr, washed with cold PBS, and cryoprotected with 5% sucrose in PBS for 1 hr and then 20% sucrose for 2 hr at 4°C. Samples were infiltrated with a 50:50 mixture of 20% sucrose and TBS tissue freezing medium (Triangle Biomedical Sciences, Durham, NC) for 1 hr. The 50:50 freezing mixture containing the embryos was transferred to a tin foil mold and frozen by slowly immersing in isopentane cooled with liquid nitrogen. Serial cryosections (8 μm thick) were collected on Superfrost slides (Fisher Scientific) and stained within two days of sectioning. Slides not stained immediately were stored at −20°C until use. For immunostaining, slides were first preincubated in PBS, blocked with 1% bovine serum albumin (BSA) in PBS for 1 hr, and incubated overnight with a panel of monoclonal antibodies to ECM molecules. These included anti-heparan sulfate 10E4 epitope (IgM or biotinylated form) diluted 1:50 in PBS BSA, anti-keratan sulfate 5-D-4 (1:50, IgG), anti-chondroitin sulfate Ly111 (1:100, IgM), all obtained from Seikagaku, Tokyo, Japan; and anti-fibronectin B3/D6 (1:50, IgG) and anti-laminin 31 or 31-2 (1:50, IgG), both developed by Dr. Douglas M. Fambrough and provided by the Developmental Studies Hybridoma Bank (under the auspices of the NICHD at The University of Iowa, Department of Biological Sciences, Iowa City, IA). Sections were double-stained with rabbit anti-cytokeratin (Biomedical Technologies, Inc., Stoughton, MA) diluted 1:200 to reveal PE cells. Slides were washed with PBS BSA (3 × 15 min), and secondary antibodies were applied for 1 hr at RT. These included goat anti-mouse IgM-Texas Red (Sigma; 1:200) for anti-heparan sulfate and anti-chondroitin sulfate, goat anti-mouse Ig (H+L) Alexa 594 conjugate (1:200) for the IgG primary antibodies and goat anti-rabbit Alexa 488 (Molecular Probes, Eugene, OR) for anti-cytokeratin (1:200). Sections were stained with 1 μg/ml 4′6-diamidino-2-phenylindole (DAPI; Sigma) in PBS for 5 min to detect nuclei, washed extensively with PBS BSA (3 × 20 min), and cover-slipped with an aqueous mounting medium (Gelvatol) containing 100 mg/ml of 1,4-diazobicyclo-[2.2.2]octane (Sigma) as anti-bleaching agent. Sections were examined with a Nikon TE-200 epifluorescence microscope (Tokyo, Japan), and images were captured with a Hamamatsu Orca ER digital camera (Hamamatsu, Japan).
A total of 58 stage HH15 embryos were used to study the effects of heparinase administration on PE growth and attachment to the heart tube. Eggs were opened and the heart region exposed by cutting a small flap in the thick overlying membrane. A volume of 1 μl of either active heparinase, heat inactivated heparinase (boiled for 2 min), or dilution buffer was injected into the pericardial coelom with a glass needle and microinjection system. The dilution buffer consisted of 100 mM NaCl, 50 mM HEPES, 1 mM CaCl2, 0.05% BSA (all from Sigma). A total of 30 mU of a combination of heparitinase I, II, and III (Seikagaku, Tokyo, Japan) in a final volume of 0.5 ml was prepared and used for injections. A trace of trypan blue was included to visualize the injections. Injections of 1 μl of dilution buffer into the pericardial coelom did not impair cardiac development. Total active heparinase per injection was 0.1 mU. After injection, the overlying membrane flap was restored and the eggs were resealed with Parafilm and reincubated. At 24 hr or 72 hr after injection, the embryos were removed and evaluated under a dissecting microscope for morphologic abnormalities. Embryos were then fixed in 2% PFA in PBS, dehydrated, embedded in paraffin, serial-sectioned through the heart region at 8 μm, and stained with H&E. Sections were then examined for myocardial and epicardial histogenesis by light microscopy.
We thank Dr. R. Rosenberg of Harvard Medical School for his helpful suggestions of anti-heparan sulfate antibodies and heparinase experiments. We also thank L. Cohen-Gould for assistance with the electron microscopy.