Epicardial-like cells on the distal arterial end of the cardiac outflow tract do not derive from the proepicardium but are derivatives of the cephalic pericardium


  • José M. Pérez-Pomares,

    1. Department of Cell Biology and Anatomy, Cardiovascular Developmental Biology Center, Medical University of South Carolina, Charleston, South Carolina
    2. Department of Animal Biology, Faculty of Science, University of Málaga, Málaga, Spain
    Search for more papers by this author
  • Aimée Phelps,

    1. Department of Cell Biology and Anatomy, Cardiovascular Developmental Biology Center, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Martina Sedmerova,

    1. Department of Cell Biology and Anatomy, Cardiovascular Developmental Biology Center, Medical University of South Carolina, Charleston, South Carolina
    Search for more papers by this author
  • Andy Wessels

    Corresponding author
    1. Department of Cell Biology and Anatomy, Cardiovascular Developmental Biology Center, Medical University of South Carolina, Charleston, South Carolina
    • Department of Cell Biology and Anatomy and Cardiovascular Developmental Biology Center, Medical University of South Carolina, 173 Ashley Avenue, Basic Science Building, Room 648, P.O. Box 250508, Charleston, SC 29425
    Search for more papers by this author


A series of recent studies strongly suggests that the myocardium of the cardiac distal outflow tract (d-OFT) does not derive from the original precardiac mesoderm but, instead, differentiates from a so-called anterior heart field. Similar findings were also reported for the endocardium of the d-OFT. However, very little information is available on the origin of the epicardium of the OFT. To address this issue, we have performed a study in which we have combined experimental in vivo and in vitro techniques (construction of proepicardial chimeras, proepicardial ablation, OFT insertion of eggshell membrane pieces, and culture on collagen gels) with molecular characterization techniques to determine this origin and define the properties of d-OFT epicardium compared with proepicardially derived epicardium. Our results demonstrate that the coelomic/pericardial epithelium in the vicinity of the aortic sac (and not the proepicardium) is the origin of d-OFT epicardium. This “pericardially” derived epicardium and the proepicardially derived epicardial tissues differ in their morphologic appearance, gene-expression profile, and in their ability to undergo epithelial-to-mesenchymal transformation. We conclude that the heterogeneity in the epicardial cell population of the OFT could be a factor in the complex developmental remodeling events at the arterial pole of the heart. Developmental Dynamics 227:56–68, 2003. © 2003 Wiley-Liss, Inc.


The fusion of two bilateral heart fields of precardiac mesoderm results in the formation of a single tubular heart consisting of an outer myocardial sleeve and an internal lining of endocardial cells. Both myocardium and endocardium classically are regarded as primary derivatives of the embryonic heart fields (reviewed in Mikawa, 1999; Tam and Schoenwolf, 1999). These two tissue-layers are separated by a well-hydrated extracellular matrix commonly referred to as the cardiac jelly (reviewed in Mjaatvedt et al., 1999). The remnants of cardiac jelly in the atrioventricular (AV) junction and outflow tract (OFT) will eventually be invaded by a mesenchyme derived from epithelial-to-mesenchymal transformation (EMT) of the endocardium (Wessels and Markwald, 2000). These mesenchymalized endocardial cushion tissues will contribute to the formation of the cardiac valves and (mesenchymal) septal structures. Once primitive cardiac segments are formed and cushion morphogenesis has been initiated, an additional population of cells arrives at the surface of the heart. The development of this layer, the epicardium, is a relatively late event (Vrancken Peeters et al., 1995; Männer et al., 2001).

It is now commonly accepted that the epicardium derives from the proepicardium (Viragh et al., 1993), a cluster of coelomic/splanchnic mesothelial cells located between the sinus venosus and the liver primordium in avians (Hiruma and Hirakow, 1989; Männer, 1992, 1993; Viragh et al., 1993), and analogous regions in different vertebrates (Viragh and Challice, 1981; Komiyama et al., 1987; Kuhn and Liebherr, 1988; Fransen and Lemanski, 1990; Van den Eijnde et al., 1995; Muñoz-Chápuli et al., 1997). The proliferating proepicardium generates epicardial progenitor cells, which attach to the myocardium and spread over the heart in a well-defined spatiotemporal pattern, a process that was nicely demonstrated in a series of different morphologic and experimental studies (for a review, see Männer et al., 2001). After forming an outer epithelial layer, separated from the myocardial surface by a subepicardial extracellular matrix-rich space, a subpopulation of the epicardial cells transforms into mesenchymal cells through an EMT. This transformation generates a population of epicardially derived cells (EPDCs; Gittenberger-de Groot et al., 1998), which, in turn, differentiates into multiple cell lineages (Mikawa and Gourdie, 1996; Pérez-Pomares et al., 1998, 2002; Gittenberger-de Groot et al., 1998; Dettman et al., 1998; Männer, 1999; Vrancken Peeters et al., 1999).

As already indicated, several experimental studies (Argüello et al., 1975; De la Cruz et al., 1977, 1983, 1998) have shown that the atrial and ventricular components of the heart derive from the tissues found in the straight heart tube. In addition, these studies also suggest that the OFT does not directly derive from the primitive splanchnic precardiac epithelium (i.e., the lateral heart fields), a hypothesis that has gained considerable support from recent experimental and genetic studies that indicate an anterior/noncardiac origin for the OFT in avians (Noden, 1991; Mjaatvedt et al., 2001; Waldo et al., 2001) and mammals (Kelly et al., 2001). The embryonic OFT is one of the most controversial segments of the heart. Controversies exist about its origin, its fate, and about the descriptive terminology of its components during remodeling (Pexieder, 1995; Ya et al., 1998; Qayyum et al., 2001). The notion that myocardial and endocardial cells in the OFT derive from an “anterior” or “secondary” heart field, combined with the observation that the distal OFT (d-OFT) of experimentally manipulated avian hearts is populated with a subpopulation of epicardial-like cells that are of nonproepicardial origin (Männer, 1999; Gittenberger-de Groot et al., 2000), prompted us to investigate the origin, phenotypical characteristics, and fate of OFT epicardium in more detail.

The results presented and discussed in this study strongly suggest that d-OFT epicardium is not derived from the proepicardium but rather is a derivative of the cephalic coelomic/pericardial epithelium covering the base of the aortic sac. Our morphologic and experimental data show that this tissue resembles that of other mesothelial tissues. The phenotypical characteristics of the d-OFT epicardial-like cells, is different from that of the epicardium found in the rest of the heart. This includes a difference in the level of expression of the retinoic acid converting enzyme RALDH2 in the OFT epicardial-like cells compared with the epicardium found in the other heart chambers (see Moss et al., 1998, Xavier-Neto et al., 2000; Pérez-Pomares et al., 2002), and differences in the expression levels of the intermediate filament proteins cytokeratin (CK) and vimentin (VIM). We also studied the in vitro behavior of these cells and found that, in contrast to proepicardially derived epicardial cells, under standard conditions the pericardially derived epicardial cells do not readily undergo epithelial-to-mesenchymal transformation. Finally, our results indicate that the two epicardial populations on the cardiac OFT might differentially contribute to the development of the segment.


Anatomical Description of the Outflow Tract During Remodeling

The terminology used in the literature to describe the different regions of the developing OFT (e.g., conus, truncus, and aortic sac) varies considerably, and consensus seems hard to reach (Pexieder, 1995, Qayyum et al., 2001). The events involved in the remodeling of the elongated tubular outflow tract (Thompson et al., 1985, 1987; Ya et al., 1998), as well as the clinical relevance of congenital defects of the cardiac outlet (Clark, 1986), are areas of open debate.

In this study, the subdivision of the OFT into different components will follow the dynamics of OFT morphogenesis, and will be based on histologic criteria (e.g., myocardial vs. nonmyocardial) as well as on external anatomic features (e.g., flexure or bending). Thus, we consider the outflow tract as the segment that connects the ventricular (i.e., trabeculated) portion of the heart to the aortic sac (i.e., the peribranchial, mesenchyme-embedded, extrapericardial cavity in which the OFT drains; see also Los, 1978; Orts-Llorca et al., 1982); therefore, the OFT extends from the trabeculated portion of the right ventricle to the pericardial reflections. As the epicardium will not spread over the myocardium of the OFT until Hamburger and Hamilton (1951) stage 18 (Vrancken Peeters et al., 1995), we will distinguish two stages in OFT development.


Starting at H/H16, the OFT can be subdivided into a proximal segment (p-OFT, analogous to the conus; for an extensive review, see Pexieder, 1995) and a distal segment (d-OFT). The lower boundary of the p-OFT is internally defined by the transition between the trabeculated right ventricle to the untrabeculated myocardial OFT, and the upper boundary is internally demarcated by the upper boundary of the proximal (or conal) cushions and externally by a characteristic “dog-leg” bend (see Qayyum et al., 2001). In the d-OFT, a lower and an upper part can be distinguished. The lower part has a myocardial component, which is internally lined with endocardium and endocardially derived tissues and, therefore, can be termed myocardial d-OFT. The upper part of the d-OFT (toward the aortic sac), which is quite a short segment, consists of a mesenchymal (nonmyocardial) belt, externally covered by an epithelium and internally lined by endocardium. We refer to this segment as the mesenchymal (nonmyocardial) d-OFT.


At these stages, the p-OFT is formed by epicardium, myocardium, endocardium, and endocardial-derived mesenchyme. For the d-OFT, the wall of the lower part of the segment remains myocardial and epicardial, whereas the wall of the upper part of the d-OFT, which is now a long segment connecting to the aortic sac, is a cylinder of mesenchyme externally covered by an epicardial-like epithelium. This upper d-OFT is analogous to the “arterial segment” as described by Qayyum et al. (2001). Due to the developmental complexity of this area, and following the example of others (Thompson and Fitzharris, 1985), we include this cardiovascular segment, which extends to the base of the aortic sac–pharyngeal arches, in the description of the OFT.

Histologic and Immunohistochemical Analysis of the Developing Outflow Tract

Stage H/H19.

At this stage, the OFT is a long cylinder consisting of an outer layer of myocardium and an inner layer of endocardium. The OFT connects the prospective right ventricle to the aortic sac and shows a “dog-leg” bend (described as a “change in the directionality in the sagittal and parasagittal planes” by Pexieder, 1995). This bend roughly divides the OFT in two halves, proximal and distal. The aortic sac is an extrapericardiac cavity located ventrally to the pharynx, connecting the lumen of the OFT with the developing aortic arches. The OFT myocardium extends from the ventricle to the pericardial reflections that mark the level of the aortic sac region. Only the very upper part of the d-OFT is devoid of myocardium (Fig. 1A). This area consists of a collar of cuboidal epithelium reminiscent of the adjacent pericardial epithelium. This epithelium covers a population of mesenchymal cells (Fig. 1A,B).

Figure 1.

Normal distal OFT development (Hamburger and Hamilton stage [H/H] 19; H/H21; H/H26). A–D: Different developmental aspects of the OFT in an H/H19 chick embryo. The sections in A–D are immunostained for myosin heavy chain with MF-20 (green) and for cytokeratin (CK; blue) and counterstained with propidium iodide to reveal nuclei (red). The boxed area in A is magnified in B. The most proximal portion of the distal OFT myocardium is not yet covered by epicardium. The more distal (cephalic) region of the OFT lacks myocardial tissue and consists of a compact mesenchyme covered by a thick cuboidal epithelium (arrowheads in B–D). The left box in B is magnified in C; the right box is magnified in D. Arrows in C and D indicate the distal limit of the myocardium (MF-20–positive), and double arrowheads show the absence of epicardium over this myocardium. E,F: Demonstrate vimentin (VIM; blue) and CK (green) expression in the OFT of a stage H/H21 chick embryo. In E, VIM is strongly expressed in the OFT endocardium and the endocardial cushion mesenchyme (asterisks in E). VIM expression is very low in the OFT myocardium (compare with atrial myocardium, which is clearly VIM-positive). CK expression, however, is noticeable in the myocardial tissue of the OFT (double arrowheads). Epicardial tissue is CK-positive as shown by the green staining (arrow). Proliferations of the epithelium covering the distal part of the OFT are indicated with arrowheads. Higher magnification in F shows the strong CK/VIM coexpression in the epicardium that covers the distal OFT myocardium (indicated by a dashed line). This coexpression (arrow, light blue staining) is similar to that found in the epicardium of other heart chambers (compare with atrial epicardium). Epicardium of the distal regions does not cover myocardial tissue and has a lower VIM expression level and high CK expression level (greenish staining, arrowheads). Proliferations of the epicardial-like tissue of the distal OFT are indicated by a double arrowhead). G: Expression of the retinoic acid converting enzyme RALDH2 in the OFT of the same embryo. In the more distally located OFT, patches of epicardium are seen with a significantly lower level of RALDH2 expression (arrowheads) as compared with the level of RALDH2 expression in the epicardium of the proximal OFT (arrow) and the rest of the heart. H–K: The OFT in a stage H/H26 chick embryo. The section in H (enlarged in I) is stained for myosin heavy chain with MF-20 (green) and cytokeratin (blue). In H, the distal rim of the OFT myocardium is indicated by an arrow. The epicardium covers the myocardium of all the heart chambers. In the OFT, flattened epicardium (arrowheads) is seen in association with the underlying myocardium. More distally, the OFT consists of mesenchymal cells (MF-20 negative) covered by a cuboidal epithelium (double arrowheads). In I, the boundary between the flattened epicardium and the cuboidal epicardium is indicated with an arrow. Some faint MF-20 labeling can be found in the loose tissue distal to the cephalic border of the OFT myocardium (arrowheads). J and K show H&E stainings of the more distal OFT epicardium and proximal OFT epicardium respectively. Note the characteristic cuboidal appearance of the epicardial cover directly lying over a mass of compact mesenchyme in J (arrowheads). In K, the epicardium is basically flattened (arrows) and covers a subepicardial space with dispersed mesenchymal cells (asterisk). A, atrium; AEP, atrial epicardium; AM, atrial myocardium; CVC, conoventricular canal; AoS, aortic sac; EN, endocardium; EP, epicardium; OFT, outflow tract; PE, pericardium; Ph, pharynx. Scale bars = 100 μm in A,E,H, 50 μm in B,F,G, 25 μm in C,D, 30 μm in J,K, 15 μm in I.

At this point, the lower part of the p-OFT is partially covered by a thin flattened epicardium. The myocardium of the upper part of the proximal OFT, as well as the myocardium of the entire d-OFT, is not yet covered by epicardium (Fig. 1C,D).

The epicardial cells and the epithelial component of the upper part of the d-OFT and aortic sac express cytokeratin (CK). The myocardium expresses myosin heavy chain, as indicated by the staining with MF-20. Coexpression of CK and MF-20 is frequently observed in a population of cells of the upper rim of the d-OFT myocardium (Fig. 1A–D).

Stage H/H21.

The overall morphology of the OFT at stage H/H21 is reminiscent of that described for stage H/H19. The major difference is the increase in length of the mesenchymal (upper) part of the d-OFT.

The p-OFT is now completely covered by epicardium. Large areas of the myocardial distal OFT, however, remain uncovered by epicardium. The epicardium shows a flattened morphology that makes it clearly distinguishable from the cuboidal epithelium covering the upper part of the d-OFT. An interesting feature at this stage is that the epithelium that covers the upper part (nonmyocardial portion) of the d-OFT at the junction with the pericardial reflections forms protrusions that give the region a ruffled aspect. These protrusions have the appearance of small “proepicardia” (Fig. 1E,F).

The epicardium and all mesothelial cells (upper d-OFT epithelial cells, coelomic epithelium, and pericardium) are CK-positive. Double-labeling with CK and vimentin (VIM) antibodies delineates cushion mesenchyme from epicardial/mesothelial tissue (Fig. 1E,F). The epicardium covering the atria, atrioventricular junction, ventricles, and the p-OFT coexpress both CK and VIM. This epicardium also expresses RALDH2 (see also Pérez-Pomares et al., 2002). The expression of vimentin in the “epicardial-like” cell layer of the d-OFT is relatively low compared with the epicardium of other heart chambers (Fig. 1E,F). The level of expression of RALDH2 in the epicardial-like cells is considerably lower than that of the epicardial cells covering the rest of the heart (Fig. 1G).

Stages H/H26–29.

At stage H/H26, the OFT is still an elongated structure. The characteristic bend that indicates the boundary between distal and proximal OFT now approximates 90 degrees (Fig. 1H).

From stage H/H26–27 onward, the entire heart is covered by epicardium, mostly with a flattened appearance, as well as by an “epicardial-like tissue,” located at the upper (mesenchymal) distal OFT. This latter epithelial tissue consists of relatively large cuboidal cells (Fig. 1J,K).

Immunohistochemistry shows that the distal end of MF-20–stained tissue (the junction between myocardial and nonmyocardial components of the d-OFT) approximately determines the distal-most margin of the flattened epicardium (CK-positive) with its related subepicardial space containing isolated CK-expressing mesenchymal cells (Fig. 1I). The upper (mesenchymal) d-OFT consists of a densely compact mesenchyme. This mesenchyme is strongly CK-positive and, generally, does not express MF-20. Some isolated mesenchymal cells close to the myocardium, however, express little MF-20 (Figs. 1I, 2C). The entire nonmyocardial d-OFT is covered by a cuboidal CK-positive epithelium. The proepicardial-like structures close to the pericardial reflections are especially prominent between stages H/H26–29 and are expressing CK (Fig. 1F) and RALDH2 (not shown). It is important to note that the “normal” epicardium of the lower portion of the OFT and the “epicardial-like” cells at the upper part of the d-OFT are in continuity at these stages (Fig. 1I).

Figure 2.

OFT mapping (MF-20 whole-mounts and quail-to-chick proepicardial chimeras). A–C: MF-20–immunostained whole-mount hearts of Hamburger Hamilton stage (H/H) 30 chick embryos (green). A and B illustrate ventral and right lateral aspects, respectively. Note the absence of MF-20 immunoreactivity in the upper (mesenchymal) part of the d-OFT. C shows a detail of the border between the myocardial (lower) and mesenchymal (upper) portions of the d-OFT. Arrowheads show the intricate architecture of the distal-most myocardial rim (arrowheads) as well as the presence of isolated groups of myocardial cells that are not directly in contact with the lower d-OFT myocardial mass (arrow). D,E: Expression of QCPN (green) and retinoic acid converting enzyme RALDH2 (red) in a stage H/H29 quail-to-chick proepicardial chimera. In D, almost all the epicardial cover is formed by donor (quail-derived) cells, as indicated by the QCPN-positive nuclei (green dots) found in the chimeric epicardial mesothelium (arrowheads) and subepicardial epicardially derived cells (EPDCs; arrows). RALDH2 expression (red), characteristic of epicardial tissue, is also conspicuous in the epicardial epithelium as well as in part of the subepicardial space of the ventricle and atrioventricular and conoventricular grooves. Double arrowheads in D indicate the upper boundary of the donor-derived (QCPN-positive) epicardium on the OFT. The more distal/cranial epicardium is not of quail origin and contains groups of cells with low RALDH2 expression (boxed areas). In E, a different section of the same chimera is presented to show some details of the chimerization process. EPDCs immigrate into ventricular myocardium (arrow) but not into the atrial and OFT myocardium. Arrowheads point to the distal border of the proepicardially derived (QCPN-positive) epicardium. RALDH2 expression is low in the more cephalic (non-proepicardially derived, QCPN-negative) OFT epicardium (double arrowhead). The asterisk indicates the presence of a non-quail–derived mesenchyme underlying this epithelial tissue. QCPN staining in F illustrates several aspects of the donor-derived/host-derived epicardial interface. The distal rim of myocardium is indicated with a dashed line. Host (QCPN-negative) and donor (QCPN-positive) epicardium integrate without discontinuities (arrow). Host-derived epicardial cells have a characteristic cuboidal phenotype (arrowheads) that can be compared with the flattened aspect of donor-derived epicardial cells. Scale bars = 400 μm in A, 300 μm in B, 100 μm in C–E, 25 μm in F. A, atrium; AoS, aortic sac, AVC, atrioventricular canal; AVE, atrioventricular epicardium; CV, conoventricular groove; d-OFT, distal OFT; EP, epicardium; LA, left atrium; ld-OFT, lower distal OFT; LV, left ventricle; OFT, outflow tract; OFTE, outflow tract epicardium; PE, pericardium; p-OFT, proximal OFT; RA, right atrium; RV, right ventricle; SE, subepicardium; ud-OFT, upper distal OFT; V, ventricle.

Proepicardial chimeras.

Quail-to-chick proepicardial chimeras were constructed to obtain insight into the contribution of proepicardially derived cells to the OFT. In the chimeras, epicardial development is complete around stage H/H26–27 (Pérez-Pomares et al., 2001), which coincides with the normal end point for the same process in control embryos (Vrancken Peeters et al., 1995). Thus, at stage H/H29, the atria (with the exception of small patches of host tissue in the atrial roof and walls), the atrioventricular region, the ventricles, and the proximal part of the OFT are all covered by QCPN-positive, quail-derived, epicardium. The upper distal-OFT, however, is covered by host (i.e., chick), QCPN-negative epicardial-like cells. Almost all of the subepicardial mesenchyme in the atrioventricular junction, ventricles, and p-OFT is of donor (quail) origin. No subepicardial mesenchyme is found in the atria. In addition, subsets of EPDCs have started to invade the myocardial layers of the ventricle at this stage (H/H29). EPDCs are scarce, however, in the atrioventricular myocardium and atrioventricular cushions at this stage and are completely absent from atrial and OFT myocardium and from the conal cushions (Fig. 2D–F).

Immunolabeling with QCPN and RALDH2 reveals differences in RALDH2 expression between the quail proepicardially derived epicardium and the chick-derived epicardial-like cells of the d-OFT. The proepicardially derived epicardium and associated subepicardial mesenchyme show a strong and equally distributed expression. The OFT epicardial-like cells show a more diffuse expression of RALDH2, with a few patches of cells expressing higher levels of RALDH2. These latter cells correspond to the areas where coelomic proliferations/proepicardial-like structures are found and are found on the mesenchymal (upper) portion of the d-OFT close to the roof of the pericardial cavity (base and walls of the aortic sac) and on the ventral aspect of the d-OFT (Fig. 2D). This pattern of expression is similar to the one described in the control embryos (see above).

The border between quail- and chick-derived cells in the chimeras is demarcated by QCPN immunostaining and is consistently found at the junction between myocardial and nonmyocardial tissues of the d-OFT (approximately from H/H29 onward). The donor and host-derived epithelial cells seem to form a continuous sheet of cells (Fig. 2F).

Characterization of proepicardially ablated embryos.

Proepicardial ablations were performed to obtain additional information about the origin of the “epicardial-like” tissues at the upper aspect of the d-OFT. Proepicardial ablations result in a clear delay in epicardial development. Of interest, as reported before (Pérez-Pomares et al., 2002), the procedure does not completely prevent epicardial spreading over the heart. Ablation of proepicardium actually seems to induce a secondary outgrowth of coelomic mesothelium in the area from which the proepicardium has been removed. This compensatory tissue eventually contacts the myocardium in the sinus venosus region and subsequently migrates over the cardiac surface/generating a late epicardial epithelial-like tissue that strongly stains with CK and RALDH2 (Fig. 3A). Although normal epicardial covering of the heart is completed between stages H/H26–27, stage-matched experimental (ablated) embryos still show numerous bare areas (not covered by epicardium) in the ventral regions of the ventricles and OFT (Fig. 3A). Discontinuities in epicardial development are also seen in the more proximal segment of the OFT, where the delayed epicardium, growing over the proximal region, has not made contact with the epicardial-like cell population of the d-OFT. This finding is especially evident in H/H28–29 embryos after CK or RALDH2 staining. The distal part of the OFT in the ablated embryos is completely covered by an epicardial-like tissue that forms a “collar” covering the upper area of the d-OFT (Fig. 3A,C,D). These cells closely resemble the pericardial epithelium surrounding the basal aortic sac with a characteristic cuboidal morphology. Proliferations of coelomic epithelium are often found in this area both in normal as well as in ablated embryos. It appears, however, that these proliferations are more prominent in the ablated specimens (Fig. 3E). These structures, that are CK/RALDH2-positive, are characteristically found close to the pericardial reflections around the base of the aortic sac. They have a heterogeneous morphology, with finger-like protrusions, and are reminiscent of the normal proepicardium located between the sinus venosus region and the liver primordium. This d-OFT epicardial-like tissue is associated with an underlying, relatively compact, mesenchymal cell population (Fig. 3F).

Figure 3.

Epicardial-like tissue at the distal outflow tract of proepicardially ablated embryos. A: Expression of the retinoic acid converting enzyme RALDH2 (red) and MF-20 (green) in a Hamburger and Hamilton stage (H/H) 27 ablated chick embryo. RALDH2 expression is only detected in the few epicardial cells (arrows) found on the heart after the ablation procedure. In the distal OFT, strong RALDH2 expression is observed in the epicardial-like tissue covering the upper rim (arrowheads). Coelomic epithelial proliferations are indicated with an asterisk. B: A hematoxylin and eosin (H&E) -stained section of a proepicardial ablated H/H27 chick embryo. Arrowheads point to the lower boundary of the “collar” of epicardial-like tissue around the most distal part of the OFT, the arrows point to the upper boundary of OFT myocardial tissue. C: Detail of the distal OFT of an ablated embryo (same as shown in A). RALDH2 expression is shown in red, and MF-20 expression in green. The panel shows that, even after proepicardial ablation, the distal OFT epicardium still forms and strongly expresses RALDH2. The arrowheads mark the lower limits of the collar of distal OFT epicardium, the asterisk the presence of coelomic epithelial proliferations in the cephalic pericardium (both RALDH2-positive), and the arrows point to the upper boundary of the myocardium in the OFT (MF-20 staining). D: A different section of the same specimen is shown after MF-20 (green) and cytokeratin (CK; red) staining. Again, the arrowheads indicate the lower boundary of the collar of epicardial-like tissue. The arrows in this panel point to CK-positive cells in the compacting layers of the developing great arteries. E: The H&E staining shows the proliferations in the cephalic coelomic (pericardial) epithelium in a stage H/H27 ablated embryo. The arrowheads indicate the mesothelial cover and the arrows the associated mesenchyme. F: Detail of one of these proliferations is presented. Mesothelial (arrowheads) and underlying mesenchymal cells (double arrowheads) express CK (red). Note the formation of finger-like protrusions (arrows). A, atrium; AVC, atrioventricular canal; d-OFTE, distal outflow tract epicardium; MYO, myocardium; OFT, outflow tract; OFTC, outflow tract cushions; V, ventricle. Scale bars = 200 μm in A, 120 μm in B, 100 μm in C,D, 50 μm in E, 25 μm in F.

Eggshell membrane insertion.

To further characterize the nature of the “proepicardial-like” tissues at the uppermost aspect of the d-OFT, we designed an experimental strategy to isolate these cells. A piece of eggshell membrane was inserted between the inner curvature of the heart and the junction between OFT and aortic sac (pericardial reflections) in normal and proepicardially ablated hearts at H/H17 (Fig. 4A). The eggs were then reincubated for approximately 12 hr. In both sets of experiments, this strategy resulted in the growth of an “epicardial-like” tissue on the distal (cephalic) tip of the membrane. Hematoxylin and eosin (H&E) staining of the embryos shows how the epicardial-like tissue spreads over the surface of the membrane, acquiring an epithelial morphology. Importantly, this procedure does not interfere with the development of epicardial-like cells on the d-OFT. The more proximal part of the membrane (i.e., the part inserted into the inner curvature) becomes covered by proepicardially derived epithelial cells. The central part of the membrane remains devoid of any epithelial overgrowth. The epithelial cells at both ends of the membrane express CK (Fig. 4B,C) and RALDH2. At the distal end, the RALDH2 expression is most intense close to the site of insertion at the pericardial reflection, the expression being weaker in the epicardial-like cells that have migrated away from the point of insertion (Fig. 4D).

Figure 4.

Collecting epicardial-like tissues using an eggshell membrane insertion technique. A: The insertion of a piece of the eggshell membrane (arrow) between the inner curvature of the heart and the base of the aortic sac in a Hamburger and Hamilton stage (H/H) 19 chick embryo. B: Cytokeratin (CK) staining in an immunolabeled sister section of the same embryo shown in the boxed area in A. Note the insertion of the tip of the eggshell membrane into the coelomic surface in the roof of the pericardial cavity (arrow). C: The same area (CK staining) at higher magnification. The upper edge of the membrane is highlighted with a dashed line. The epicardial-like tissue growing over the membrane is indicated with arrowheads. D: Membrane insertion in a stage H/H20 chick embryo is shown after staining with the retinoic acid converting enzyme RALDH2. Note the strong staining in the pericardial reflection (asterisk) and the staining of the epicardial-like tissue on the eggshell membrane (arrowheads). The level of RALDH2 expression is highest in the distal part of the growing epicardial tissue, i.e., the regions closer to the coelomic/pericardial epithelium. Coelomic mesothelial cells display a strong RALDH2 expression (arrow), compared with the tissue spreading over the more caudal areas (lower part) of the membrane. Scale bars = 300 μm in A, 50 μm in B,D, 25 μm in C. A, atrium; AVC, atrioventricular canal; EM, eggshell membrane; OFT, outflow tract; Ph, pharynx; V, ventricle.

Collagen cultures and immunohistochemistry.

To obtain more insight into the differences in migration and transformation properties between true proepicardially derived cells, pericardially derived cells, and cells derived from the “proepicardial-like” proliferations at the junction of OFT and aortic sac, we performed in vitro collagen gel assays. The d-OFT epicardial-like tissue spreads over the surface of the gels, forming an epithelial monolayer (Fig. 5A). These cells are CK- and RALDH2-positive and have an elongated shape (Fig. 5B). Their expression patterns as well as their morphology resemble that of the cells derived from pericardial explants (Fig. 5D). The CK and RALDH2 expression found in the proepicardially derived cells is comparable to that of OFT epicardial-like cells and pericardial cells. The proepicardially derived cells, however, show a characteristic rounded-polygonal shape, which distinguishes them from the elongated-shaped pericardial and epicardial-like OFT cells (Fig. 5E,F). Distal OFT epicardial-like cells on the surface of the gel typically loose their epithelial context after 24 hr of incubation and acquire a spindle shape morphology. Only a few cells, however, invade the collagen matrix (Fig. 5B,C). The pericardially derived epithelium basically shows the same behavior (Fig. 5D).

Figure 5.

In vitro culture of distal outflow tract (d-OFT) epicardium. A: A d-OFT explant in a collagen gel assay (imaged using Hoffman Modulation Contrast optics). The tissue grows as an epithelial monolayer from the eggshell membrane (EM; arrows). In the borders of the culture, cells tend to detach and to migrate as isolated cells over the surface of the collagen gel (arrowheads). B,C: An epicardial d-OFT explant on collagen gel, immunofluorescently stained for cytokeratin (CK) expression (green). All the cells on the surface of the gel are CK-positive. Below the surface of the gel, no mesenchymal cells can be detected (C). D: A pericardial explant on collagen gel. The pericardial epithelium of this explant also expresses CK and is phenotypically very similar to that observed in the d-OFT explants (compare with A,B). It differs, however, from the epithelium seen in proepicardial explants (compare with E,F). E–H: The results from proepicardial/d-OFT epicardium cocultures on collagen gels. Quail proepicardial cell nuclei are QCPN-positive (green) in E–H. In E, both proepicardial (QCPN-positive) and d-OFT tissue (QCPN-negative) express the retinoic acid converting enzyme RALDH2 (red). The boxed area presents an area of the border between the two explants and is magnified in F. Both proepicardial and OFT tissues integrate normally but do not seem to mingle (arrows). G and H show details of another coculture of quail proepicardium with chick distal OFT epicardium. Both tissues have been stained with a CK antibody (red). Quail (pro)epicardial cells (QCPN-positive, green nuclei) remain in the surface of the collagen gel (arrow in G). Almost all the mesenchymal cells found in the depth of the collagen gel (H) are from quail proepicardial origin (QCPN-positive, arrowheads). In both cases, the area covered by the eggshell membrane is indicated with a dashed line. EM, eggshell membrane; PEe, proepicardial explant; OFTe, distal outflow tract explant. Scale bars = 100 μm A–E, 50 μm F–H.

In the proepicardial (quail)/d-OFT (chick) cocultures, the epithelial sheets derived from both tissues grow over the surface of the gel often making contact with each other. Many quail epicardially derived cells (QCPN-positive) undergo epithelial-to-mesenchymal transformation and migrate into the collagen matrix (Fig. 5G,H).


Our results demonstrate for the first time the presence of two distinct epicardial populations in the normal developing avian heart. Our experimental studies show that, although the majority of the epicardial OFT cells (the proximal population) is proepicardially derived, the most distal epicardial population is derived from coelomic pericardial proliferations at the base of the aortic sac. We show that the difference in the origin of these two populations is reflected in morphologic, molecular, and functional characteristics.

The pericardial proliferations, in the vicinity of the so-called pericardial reflections, have a characteristic ruffled appearance and closely resemble the coelomic epithelium at the caudal end of the heart in the early stages of proepicardial development. Unlike the proepicardium, however, the pericardial proliferations never develop into a true cauliflower-shaped cell cluster. The epicardium at the d-OFT that derives from these proliferations is morphologically different from the proximal, proepicardially derived, epicardium. At the d-OFT, the cells have a cuboidal morphology, closely resembling that of the coelomic epithelium, whereas the epicardium on the p-OFT is flat and has a thin lamellar or squamous phenotype. A similar difference in phenotypical characteristics was also observed in the epicardial population on the OFT of quail-to-chick chimeras. Compared with the epicardium in the rest of the heart, the d-OFT epicardium is characterized by a relatively low expression of RALDH2 and vimentin. The pericardial proliferations themselves, however, express a relatively high level of RALDH2.

To obtain more information on the characteristics of the pericardial proliferations, we performed in ovo microsurgical experiments where pieces of eggshell-membrane were inserted between the floor of the aortic sac and the inner curvature of the heart in normal and proepicardially ablated hearts. It was found that, after 12 hr of reincubation, the cephalic portion of the membrane was covered by epithelial tissue expressing moderate levels of CK and relatively high levels of RALDH2. This high RALDH2 expression in the cephalic portion of the membrane illustrates that the epithelium has developed in close proximity of the pericardial proliferations (“proepicardial-like structures”). Cells that have migrated away from this site express considerably lower levels of RALDH2.

The dual origin of OFT epicardium explains the characteristic patchy expression of RALDH2 seen in the d-OFT epicardium of control and chimeric embryos. It is important to emphasize that, in the case of the chimeras, the heterogeneity in RALDH2 expression does not reflect species-specific differences, but rather reflects the tissue origin of the cells (proepicardium vs. pericardial proliferations) as shown by control (nonchimeric) chick embryos. It is likely that the strong CK/VIM coexpression in the flattened p-OFT epicardial tissue (but not in the d-OFT epicardium) is related to the ability of this population to transform into epicardially derived cells (Pérez-Pomares et al., 1997, Morabito et al., 2001) as already shown in other systems (Fitchett and Hay, 1989). This ability to generate a mesenchymal cell population (composed of epicardially derived cells, EPDCs) through EMT has been reported by different authors (Dettman et al., 1998; Gittenberger-de Groot et al., 1998; Pérez-Pomares et al., 1998; Männer, 1999).

Because the migration of EPDCs into myocardial layers and their subsequent differentiation into a variety of cell types has been suggested to be essential in the proper development of the myocardium and coronary vessels (Pérez-Pomares et al., 1998, 2002; Dettman et al., 1998; Gittenberger de Groot et al., 1998; Vrancken Peeters et al., 1999), we determined the potential of the nonproepicardially derived d-OFT epicardium to transform into mesenchyme in a collagen gel assay. The in vitro experiments showed that proepicardial tissue initially forms an epicardial monolayer over the collagen gel. As a result of an EMT, the monolayer subsequently produces a population of invasive mesenchyme. Cells isolated from the d-OFT also spread over the surface of the gel to form an epithelial monolayer, but, in contrast to the proepicardially derived epithelium, they only generate very few invasive mesenchymal cells. The epithelial cells in isolated d-OFT cultures usually arrange in parallel rows. The integrity is generally lost after 24 hr, a phenomenon also seen in most of the pericardial control explants. Thus, this supports the histologic observations that the d-OFT epicardium is a derivative of coelomic-pericardial tissue. It also indicates that the two epicardial populations on the developing OFT have different potentials to undergo EMT, similar to what has been demonstrated before for the endocardial populations in the respective segments of the developing heart (Mjaatvedt and Markwald, 1989).

It was demonstrated recently that proepicardial ablation results in the formation of a “compensatory mesothelial collar” in the truncal (distal) OFT region (Gittenberger-de Groot et al., 2000). Our current study supports this finding. Moreover, we demonstrate that, in the normal as well as in the quail-to-chick proepicardial explant-model, the epicardial population on the OFT also has a dual origin. The epicardium on the surface of the proximal OFT is proepicardially derived, the cephalic pericardium is the tissue source for the “epicardial-like” cells on the distal OFT. Inhibition of proepicardial development apparently creates a “permissive” environment for the pericardially derived cell population, resulting in an expansion of this population toward the proximal end of the OFT. The extent of this expansion in our ablation experiments and in those described by Gittenberger-de Groot et al. (2000), however, are not identical. This finding is not surprising as the methods used to delay epicardial growth were different (eggshell membrane block and direct ablation of the proepicardium respectively, see also Pérez-Pomares et al., 2002). Therefore, we conclude that the prominent presence of epicardial-like tissue on the distal OFT in proepicardium ablated specimens reflects an expansion of a pericardially derived tissue, which in normal development can also been found on the d-OFT, rather than a compensatory growth under abnormal conditions.

Recently, we showed that, in the ventricles, proper development of the epicardium is crucial for the differentiation and maturation of the underlying myocardium (Pérez-Pomares et al., 2002). Here, we show the presence of two “epicardial” cell populations associated with the OFT. Although it remains to be established whether—and if so, to what extent—these cell populations contribute to the regulation of OFT morphogenesis, the spatiotemporal relationship between the different tissues of the OFT suggests that the interface between the two epicardial populations might be playing an important role in the developing OFT. The mechanisms involved in OFT morphogenesis are contentious. Specifically, controversies exist regarding two pivotal events, i.e., arterialization of the distal part and the apparent torsion and retraction of the conal myocardium (Thompson et al., 1987). Although during early development a part of the d-OFT myocardium is covered by pericardially derived epicardium, eventually the OFT myocardium is completely covered by proepicardially derived epicardium, the pericardially derived cells ending up covering the roots of the great vessels. This apparent shift of the “proepicardial/pericardial” interface with respect to the distal myocardial rim of the OFT leads to a few possible models: (1) the mature epicardium moves freely over the myocardial surface; (2) the nonmyocardial (upper) portion of the d-OFT grows in a distal-to-proximal manner (cephalocaudal direction), displacing the distal myocardial edge proximally as a result of mechanical forces with a synchronic local regulation of cell division (Thompson et al., 1995); or (3) the myocardium in the d-OFT disappears as a results of transdifferentiation of myocardial cells (Argüello et al., 1978; Ya et al., 1998) and/or myocardial death (Watanabe et al., 1998), possibly involving EPDC-regulated apoptosis (Rothenberg et al., 2002). At this point, we do not have any data to determine the exact relationship between the formation of the interface between the epicardially derived and pericardially derived epithelial cells on the surface of the OFT and the spatiotemporal changes that occur at the myocardial/mesenchymal junction in the OFT. The observation that, in the setting of proepicardial ablations, the pericardially derived epicardium can occupy a relatively large segment of the myocardial OFT (see also Gittenberger-de Groot et al., 2000), resulting in a change of the relationship between the above-mentioned tissue interfaces, indicates that the myocardial/mesenchymal boundary itself is not the determining factor in establishing the extent of the migration of the pericardially derived cells over the OFT. Future studies on the development of the myocardial OFT in animal models with perturbed epicardialization (e.g., microsurgical intervention, knock-outs) will undoubtedly provide more insight in this matter.



Chick and quail eggs were kept in an incubator at 37°C. The embryos were staged according to Hamburger and Hamilton (1951).

Normal embryos for immunohistochemical characterization.

Chick eggs were incubated until stages H/H19–29. The embryos were fixed in Amsterdam fixative, embedded in Paraplast plus, sectioned in a microtome (5 μm), and guide series were stained with H&E. Selected slides were immunostained using antibodies against VIM, the retinoic acid-synthetic enzyme RALDH2, or CK. RALDH2 and CK were used as markers for epicardium (Pérez-Pomares et al., 2002). The MF-20 antibody was used to specifically label the myocardium (Van den Hoff et al., 1999). Cell nuclei were counterstained with propidium iodide.

Proepicardial chimeras.

Quail-to-chick proepicardial chimeras were prepared as described by Männer (1999) with the following modifications. Host chick embryos were incubated until stages H/H16–17. The eggs were windowed, and sharp tungsten needles were used to create small openings through the vitelline and chorionic membranes exposing the pericardial cavity. For each embryo, a small piece of the eggshell membrane was cut with iridectomy scissors and made to fit exactly between the sinoatrial sulcus and the caudal vitelline veins. Then, stage H/H16–17 quail embryo donors were excised and perfused with Earles Balanced Salt Solution (EBSS; GIBCO). The heart was removed by cutting it through the outflow tract and the sinoatrial sulcus. The previously prepared eggshell membrane was introduced through the omphalomesenteric vein of the quail embryo and pushed until it reached the cardiac lumen, so that the sinus venosus formed a cuff around the membrane, holding the donor proepicardium in its surface. The membrane carrying the quail (donor) proepicardium was inserted facing the ventricular heart surface (Fig. 6B). After the operation, the eggs were sealed with Scotch tape and reincubated for 96–108 hr (H/H25–26), 132–144 hr (H/H28–29), or 180 hr (H/H32), then fixed in modified Amsterdam's fixative (methanol:acetone:water = 2:2:1) and embedded in Paraplast (Paraplast Plus, OXFORD Labware, St. Louis, MO). Finally, 5-μm serial sections were mounted on microscope slides (Superfrost/Plus, Fisherbrand, Fisher Scientific, Pittsburgh, PA) and immunostained by using the QCPN antibody to localize the donor (quail) cells. The anti-RALDH2 and anti-CK antibodies were basically used to characterize mesothelial tissues.

Figure 6.

Microsurgical procedures to alter/study epicardial development. A: The normal process of proepicardial attachment to the myocardial wall. B–D: Three microsurgical procedures that provide different and independent information about outflow tract (OFT) epicardium development. In A, the arrow indicates the attachment of the proepicardium (asterisk) to the inner curvature of the heart. B presents the proepicardial quail-to-chick transplantation as described by Männer (1999, see their materials and methods for a careful description). In C, ablation of the proepicardium is illustrated. The proepicardial cell mass (arrowhead) is carefully removed from the coelomic surface using tungsten needles and iridectomy scissors and then extracted with thin forceps. In D, a piece of eggshell membrane (depicted as a double dashed line, see arrow) is inserted between the inner curvature of the heart and the base of the aortic sac. When used to explant the distal OFT epicardial tissue onto collagen gels, only the more distal portion of the membrane is dissected (arrowhead). A, atrium; AoS, aortic sac; OFT, outflow tract, d-OFT, distal outflow tract; p-OFT, proximal outflow tract; V, ventricle.

Proepicardial ablation.

For proepicardial ablations, chick embryos were incubated until stages H/H16–17. Openings in the vitelline and chorionic membranes were used to enter the pericardial cavity. The proepicardium was carefully ablated by using fine dissecting tungsten needles (Fig. 6C). The embryos were fixed in modified Amsterdam's fixative (see above), embedded in Paraplast, sectioned (5 μm), and either stained with H&E or immunostained with RALDH2, CK, and/or MF-20 as described below.

Immunohistochemistry in normal and experimental embryos.

For both single antigen localization (RALDH2, CK) or colabeling (RALDH2/QCPN; RALDH2/MF-20; CK/MF-20), paraffin sections (5 μm) were used. The sections were dewaxed and hydrated in a graded series of ethanol, and nonspecific bindings sites were blocked with 5% normal goat serum, 1% bovine serum albumin(BSA), and 0.5% Triton-X 100 for 1 hr at room temperature (RT). The incubation with polyclonal antibodies (CK, Dakopatts, diluted 1:100 in phosphate buffered saline [PBS]; RALDH2, gift from Drs. P. McCaffery and U. Drägger, diluted 1:5,000 in PBS, both incubated overnight at RT) was followed by an AlexaFluor-568 goat anti-rabbit immunoglobulin G antibody incubation (Molecular Probes), 1:400 in PBS, 2 hr at RT. The sections incubated with monoclonal antibodies (QCPN [undiluted], MF-20 diluted 1:20 in PBS, and anti-vimentin [Amf-17b] diluted 1:75 in PBS [all from DSHB]) were incubated overnight at RT, followed by incubation with AlexaFluor-488 rabbit anti-mouse antibody (Molecular Probes, 1:400 in PBS, 2 hr at RT). In the case of colabeling, the method for localizing the polyclonal antibodies was followed by the labeling procedure for monoclonal antibodies. When this method was applied in combination with a propidium iodide counterstaining (red fluorescence), AlexaFluor-568 was substituted by Cy5-conjugated donkey anti-rabbit secondary antibody (Jackson Laboratories). In a subset of experiments Cy5-conjugated donkey anti-mouse secondary antibody (Jackson Laboratories, 1:100 dilution in PBS, 4 hr of incubation at RT) in combination with AlexaFluor-488 goat anti-rabbit secondary antibody were applied. Some sections were also counterstained with propidium iodide. After staining, slides were cover-slipped by using a 1:1 PBS/glycerol solution and analyzed under a Bio-Rad MRC 1024 laser scanning confocal microscope. Whole-mount MF-20 stainings were performed as described elsewhere (Van den Hoff et al., 1999).

In ovo and in vitro characterization of epicardial-like tissue of the distal OFT.

To isolate precursor tissue of the epicardial-like cells of the distal OFT, H/H16–17 chick embryos were used. A small piece of eggshell membrane was cut with iridectomy scissors to fit exactly between the inner curvature of the heart and the base of the aortic sac (Fig. 6D). After insertion of the membrane, the embryos were sealed and reincubated for 12 or 24 hr. A subset of these embryos was excised, fixed in Amsterdam's fixative, embedded in Paraplast, sectioned, and processed for H&E staining or immunohistochemistry (CK or RALDH2). The inserted eggshell membranes from other experimental specimens were isolated, and the segments closest to the aortic sac were used for culture on 1.5 mg/ml drained collagen gels (rat tail type I collagen, Collaborative Research, 500 μl/well in four-well NUNC plates) in a CO2 incubator at 37°C. In some cases, these “explants” were cocultured with small pieces of chick pericardium (stages H/H16–17) and/or with quail proepicardia (H/H16–17). In these coculture experiments, explants were typically placed 2–3 mm apart. The explants were allowed to attach to the collagen for 5 hr. Then, M199 medium (GIBCO) supplemented with 1% chick serum (SIGMA) and insulin/transferrin/selenium (ITS, Collaborative Research) was added. All the explants were cultured for 1–2 days at 37°C and 5% CO2 and inspected daily by using an inverted microscope with Hoffman Modulation Contrast optics or an Olympus microscope coupled to a SPOT digital camera and photographed. At the end of the culture period, the explants were fixed in 4% paraformaldehyde, extensively washed in PBS, blocked in a 1% BSA-PBS solution and incubated overnight in a 1:100 dilution of the CK antibody or a 1:2,500 of the RALDH2, both in PBS. The cultures were washed at least three times (1 hr each) in PBS-BSA and finally incubated overnight again in 1:400 dilution in PBS secondary Alexa-conjugated anti-rabbit antibody (AlexaFluor-568 or AlexaFluor-488, Molecular Probes). In quail–chick tissue cocultures CK or RALDH2 antibodies were used to characterize the mesothelial tissue (see above), and QCPN was used to localize quail cells. In brief, the gels were incubated in undiluted QCPN supernatant, extensively washed and incubated in a 1:400 dilution in PBS of the secondary Alexa-conjugated anti-mouse antibody (AlexaFluor-488 or AlexaFluor-568, Molecular Probes), washed, and mounted. The cultures were analyzed in a Bio-Rad MRC 1024 scanning laser confocal microscope.


We thank Tanya Rittmann for creative assistance in preparing cartoon illustrations. The authors also thank Dr. Peter McCaffery and Dr. Ursula Dräger for their kind gift of the RALDH2 antibody. The MF-20, vimentin (Amf-17b), QCPN, and QH1 monoclonal antibodies were obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD, and the Department of Biological Sciences, University of Iowa, Iowa City, IA, under contract NO1-HD-2-3144 from the NICHD. A.W. and J.M.P.-P received funding from the NIH and the AHA.