Morphological and molecular left–right asymmetries in the development of the proepicardium: A comparative analysis on mouse and chick embryos

Authors


Abstract

The proepicardium (PE) is an embryonic progenitor cell population that delivers the epicardium, the majority of the cardiac interstitium, and the coronary vasculature. In the present study, we compared PE development in mouse and chick embryos. In the mouse, a left and a right PE anlage appear simultaneously, which subsequently merge at the embryonic midline to form a single PE. In chick embryos, the right PE anlage appears earlier than the left and only the right anlage acquires the full PE-phenotype. The left anlage remains in a rudimentary state. The expression patterns of PE marker genes (Tbx18, Wt1) correspond to the morphological data, being bilateral in the mouse and unilateral in the chick. Bmp4, which is unilaterally expressed in the right PE of chick embryos, is symmetrically expressed in the sinus venosus wall cranial to the PE in mouse embryos. Asymmetric development of the chicken PE might reflect side-specific differences in topographical relationships to tissues with PE-inducing or repressing activity or might result from the PE-repressing activity of the right PE, which grows earlier. To test these hypotheses, we analyzed PE development in chick embryos, firstly, subsequent to experimentally induced inversion of PE topographical relationships to neighbouring tissues; secondly, in organ cultures; and, thirdly, subsequent to induction of cardia bifida. In all three experiments, only the right PE develops the full PE phenotype. Our results suggest that PE development might be controlled by the L–R pathway in the chick but not in the mouse embryo. Developmental Dynamics 236:684–695, 2007. © 2007 Wiley-Liss, Inc.

INTRODUCTION

The early embryonic heart is a tube-like organ built up by an inner epithelial layer of endocardial cells in contact with the blood, a middle layer of a cell-free extracellular matrix called the cardiac jelly (Davis, 1924), and an outer epithelial layer of myocardial cells in contact with the pericardial fluid. The early embryonic heart thus lacks well-known structural components of the mature heart such as the epicardium, the cardiac interstitium, and the coronary vasculature. Research from the past three decades has shown that the majority of non-endocardial and non-myocardial cells do not derive from the early embryonic heart tube, as originally thought (Kölliker, 1879; Mollier, 1906; DeHaan, 1965), but derive from primarily extracardiac sources (for reviews see Poelmann et al., 2002; Gittenberger-de Groot, 2003). One of these sources is the proepicardium (PE). The PE consists of an accumulation of villous or vesicular protrusions of the pericardial coelomic mesothelium that form close to the venous pole of the embryonic heart (Virágh and Challice, 1981; Männer, 1992; Virágh et al., 1993; Männer et al., 2001). It delivers the epicardium, the majority of the cardiac interstitium, and the coronary vasculature (for reviews see Männer et al., 2001; Muñoz-Chápuli et al., 2002; Wessels et al., 2004). PE-derived cells not only provide the material for different structural components of the mature heart but additionally they play important modulatory roles in myocardial development (for review see Männer, 2006).

Initial data suggested that the PE was a single unpaired structure lying in the embryonic midline (Kurkiewicz, 1909; Virágh and Challice, 1981; Komiyama et al., 1987; Männer, 1992; Virágh et al., 1993). This view was first challenged by data from dogfish embryos, which showed that the PE formed as a bilaterally paired anlage (Muñoz-Chápuli et al., 1997). Subsequent observations on chick embryos have suggested that in avian embryos, the PE might also form as a bilaterally paired anlage from which, however, only the right seems to develop the full PE-phenotype (Männer et al., 2001; Schlueter et al., 2006). A primarily paired PE anlage might also exist in mammalian embryos since Kuhn and Liebherr (1988) have noted that the PE of Tupaia belangeri embryos forms at the caudal border of the sinus venosus and on the cranio-lateral aspect of both of its sinus horns.

The above-mentioned findings suggested a bilaterally paired nature of the PE anlage. Exact data on left–right (L–R) development of the PE, however, were absent for the chick embryo as well as for the currently most frequently used mammalian model embryo, the mouse embryo. Thus, questions have remained unanswered as to whether the development of the paired PE anlagen in mice occurs in a bilaterally symmetric pattern, as suggested by data from other mammalian embryos (Kuhn and Liebherr, 1988), or in a bilaterally asymmetric pattern, as suggested by data from avian embryos (Männer et al., 2001; Schlueter et al., 2006). In the present study, we have, therefore, investigated the development of the PE in mouse and chick embryos. Our observations show that the development of the mouse PE proceeds in a bilaterally symmetric pattern, whereas the development of the chicken PE proceeds in a bilaterally asymmetric pattern. Our data, furthermore, suggest that the development of the PE might be linked to the specification of the L–R body axis in chick embryos, but seems to be independent in mouse embryos. Due to the fact that the development of the chicken PE proceeds in a bilaterally asymmetric pattern, the chick embryo might be a good model to identify the factor(s) responsible for the induction of the PE.

RESULTS

Visual examination of the PE in the intact embryo is normally hampered by the ventricular bend of the heart loop, which prevents free ventral views of the venous pole of the embryonic heart loop. The PE, therefore, is usually examined in right or left lateral views of intact embryos or in histological sections. Both approaches, however, can give only limited information about the exact number, topography, and three-dimensional morphology of the PE anlagen. In the present study, we have, therefore, examined embryos that have been stretched along their cranio-caudal axis. This leads to the displacement of the ventricular bend to a position cranial to the atria and, thus, facilitates free ventral views of the PE.

Development of the PE in Mouse Embryos

In mouse embryos, a left and a right PE anlage became morphologically apparent on ED 8.5. Both PE anlagen appeared in the form of prominent oval-shaped areas of the ventro-caudal wall of the left and right halves of the sinus venosus (Figs. 1A,F, 2A, and 3A). The mesothelial cells covering these areas showed numerous bleb- and finger-like protrusions of their apical cell membranes whereas the surrounding mesothelial cells had smooth apical cell membranes (Fig. 2A). Whole mount ISH showed that the PE marker genes Tbx18 and Wt1 were expressed in these two areas in a bilaterally symmetric fashion (Fig. 3A; Wt1 expression data not shown).

Figure 1.

Formation of the PE in mouse embryos as shown by SEM micrographs. Ventral views of the opened pericardial cavity showing the venous pole of the heart loop and the developing PE on ED8.5 (A,F), ED9.0 (B,G), ED9.5 (C,H), 9.75 (D,I), and ED10.5 (E,J). The two PE anlagen of mouse embryos (marked by arrows) appear simultaneously on ED8.5. They move towards the embryonic midline where they start merging to form a single PE on ED9.0. The mature PE-phenotype is reached on ED9.5. It appears as a triangle-shaped accumulation of vesicular cell aggregates. On ED10.5, only a small number of PE vesicles is present on the ventro-caudal wall of the sinus venosus. The former area of the PE appears as a rhombus-shaped prominence. a, common atrium; aip, anterior intestinal portal; la, developing left atrium; ls, left sinus horn; lv, embryonic left ventricle; o, outflow tract; ra, developing right atrium; rs, right sinus horn; rv, embryonic right ventricle; st, septum transversum. Scale bar = 200 μm.

Figure 2.

Higher magnification views of the venous pole of mouse embryos from ED8.5 (A), ED9.0 (B), and ED9.5 (C). The initially separated PE anlagen (marked by dotted lines) of mouse embryos (A) move towards the embryonic midline (B) where they merge to form the mature PE (C). Note that the mesothelial cells of the two PE anlagen show numerous bleb- and finger-like protrusions of their apical cell membranes whereas the mesothelial cells of the mature PE have smooth apical cell membranes. Scale bar = 20 μm.

Figure 3.

Formation of the PE in mouse embryos as shown by whole mount ISH with probes for PE marker genes (Tbx18, A–C; Wt1, D). Ventral views of mouse embryonic hearts from ED8.5 (A), ED9.0 (B), ED9.5 (C), and ED10.5 (D). Abbreviations as in Figure 1. Scale bar = 200 μm.

Between ED8.5 and ED9.5, the two areas identified as the PE anlagen became morphologically more prominent and moved towards the embryonic midline where they merged to form a single PE (Figs. 1B,G, 2B, and 3B). The fully developed PE-phenotype was reached on ED9.5. At this time point, the PE appeared as an accumulation of vesicular cell-aggregates (Figs. 1C,H and 2C). This accumulation of PE vesicles covered a large triangle-shaped area of the pericardial surface of the septum transversum. The base of this triangle covered the caudal border of the ventral wall of the sinus venosus and its left and right sinus horns whereas the apex of this triangle pointed towards the midline of the ventral wall of the primitive pericardial cavity (Figs. 1C,D,H,I and 3C). In contrast to earlier stages, the apical cell membranes of the mature PE mesothelial cells now were devoid of bleb- and finger-like protrusions and, therefore, appeared as smooth surfaces (Fig. 2C).

Between ED9.5 and ED10.5, the majority of PE vesicular cell-aggregates were released into the free pericardial cavity. These PE vesicles adhered to the naked myocardial surface of the heart loop where they contributed to the formation of the primitive epicardium (not shown). On ED10.5, only a small number of the original PE vesicles were present on the ventro-caudal wall of the sinus venosus. The area originally harbouring the mass of PE vesicles now appeared as a rhombus-shaped prominence of the ventro-caudal wall of the sinus venosus and adjacent portions of the septum transversum (Figs. 1E,J). This area still expressed the PE marker genes Tbx18 (data not shown) and Wt1 (Fig. 3D).

Recent studies have shown that BMP signalling factors were important regulators of PE identity in the chick embryo (Schlueter et al., 2006; Kruithof et al., 2006). During the formation of the chicken PE, Bmp2 has been found to be expressed in a bilaterally symmetric pattern in the sinus venosus myocardium cranial to the PE, whereas Bmp4 has been found to be expressed in the PE (Schlueter et al., 2006; Kruithof et al., 2006). We have, therefore, analysed the expression patterns of both genes in mouse embryos during the phase of PE development. During the initial phase of formation of the two PE anlagen (ED8.5), only a very faint expression of Bmp2 was found in the wall of the future sinus venosus (Fig. 4A). A faint expression of Bmp4 was found in bilaterally symmetric domains medial to the two PE anlagen (Fig. 5A). During the phase of formation of the mature PE (ED9.0 to ED9.5), Bmp2 and Bmp4 were both found to be expressed in the ventral wall of the sinus venosus cranial to the area harbouring the PE vesicles (Figs. 4B,C,E,F,F′ and 5B,C,E,F,F'). Thereby, Bmp2 expression was found along the wall of both sinus horns, whereas the expression of Bmp4 was confined to a prominent domain in the midline between the two sinus horns. Subsequent to the disappearance of the mass of PE vesicles (ED10.5), expression of Bmp2 as well as Bmp4 was found along the border between the ventral wall of the sinus venosus and the developing atria (Figs. 4D,G and 5D,G). Thus, in contrast to the situation in chick embryos (Schlueter et al., 2006), we did not observe Bmp4 expression within the PE of mouse embryos.

Figure 4.

Expression of Bmp-2 at the venous pole of mouse embryonic hearts on ED8.5 (A), ED9.0 (B,E), ED9.5 (C,F,F'), and ED10.5 (D,G). Ventral (A–D) and left lateral views (E–G) of hearts subjected to whole mount ISH with a Bmp-2 probe. The insert (F') shows Bmp-2 expression in a sagittal histological section from a ED9.5 heart. On ED8.5, Bmp-2 is only weakly expressed in the region of the future sinus venosus (marked by asterisks). Between ED9.0 and ED10.5, Bmp-2 expression (marked by arrows) is found in the ventral wall of the sinus venosus cranial to the area harbouring the PE. Pe, proepicardium; st, septum transversum. Other abbreviations as in Figure 1. Scale bar = 200 μm.

Figure 5.

Expression of Bmp-4 at the venous pole of mouse embryonic hearts on ED8.5 (A), ED9.0 (B,E), ED9.5 (C,F,F'), and ED10.5 (D,G). Ventral (A–D) and left lateral views (E–G) of hearts subjected to whole mount ISH with a Bmp-4 probe. The insert (F') shows Bmp-4 expression in a sagittal histological section from a ED9.5 heart. On ED8.5, Bmp-4 expression (marked by arrows) is found in bilaterally symmetric domains medial to the region of the future sinus venosus (marked by asterisks). On ED9.0 and ED9.5, Bmp-4 expression (marked by arrows) in the sinus venosus is confined in a midline domain cranial to the area harbouring the PE. On ED10.5, Bmp-4 expression (marked by arrows) is found along the border between the sinus venosus and the developing atria. Abbreviations as in Figure 4. Scale bar = 200 μm.

Development of the PE in Chick Embryos

In chick embryos, the coelomic mesothelium covering the ventral wall of the right half of the sinus venosus started to form short villous protrusions in a rounded area at HH-stage 14 (Fig. 6A,F). This accumulation of mesothelial protrusions represents the right PE anlage. A morphologically similar accumulation of mesothelial protrusions, representing the left PE anlage, appeared on the ventral wall of the left half of the sinus venosus somewhat later at HH-stage 15/16 (Fig. 6B,C,G,H). During the initial stages of PE formation, some of the mesothelial cells forming the villous protrusions showed numerous bleb- and finger-like protrusions of their apical cell membranes.

Figure 6.

Formation of the PE in chick embryos as shown by SEM micrographs. Ventral views of the opened pericardial cavity showing the venous pole of the heart loop and the developing PE at HH-stage 14 (A,F), HH-stage 15 (B,G), HH-stage 16 (C,H), HH-stage 17/18 (D,I), and HH-stage 18/19 (E,J). The right PE anlage of chick embryos (right arrow) appears on HH-stage 14, whereas the left PE anlage (left arrow) appears on HH-stage 15/16. From the two PE anlagen, only the right one forms the mature PE. The left anlage remains in a rudimentary state and disappears on HH-stage 18/19. The right PE establishes a firm contact to the heart at HH-stage 17. The PE-derived epicardium spreads over the cardiac surface in the form of a continuous epithelial sheet (border marked by dotted line). Note that chick embryos lack a septum transversum. Contact between the sinus venosus and the body wall is established via the left and right mesocardium laterale. li, liver; lm, left mesocardium laterale; rm, right mesocardium laterale. Other abbreviations as in Figure 1. Scale bar = 200 μm.

From the two asynchronically appearing PE anlagen, only the right anlage, which appears earlier, underwent rapid growth and acquired the well-known phenotype of a cauliflower-like accumulation of mesothelial villi on HH-stage 16 (Fig. 6C,H). The left PE anlage, which appears later, remained in a rudimentary state and disappeared at HH-stage 18/19 (Fig. 6E,J). In contrast to mice, the two PE anlagen did not merge but remained separated from each other. The bilaterally asymmetric formation of the paired PE anlagen was reflected by corresponding expression patterns of the PE marker genes Tbx18 and Wt1, which were prominently expressed only in the right PE as previously reported (Fig. 7; Schlueter et al., 2006).

Figure 7.

Correlation between the morphological data, as shown by an SEM micrograph (A), and the expression pattern of the PE marker gene Tbx18, as shown by whole mount ISH (B). Ventral views of chick embryonic hearts at HH-stage 16. Abbreviations as in Figure 1. Scale bar = 200 μm.

Mesothelial villi of the right PE established firm contacts to the dorsal wall of the ventricular bend at HH-stage 17/18 (Fig. 6D,I). Thereby, a secondary tissue bridge was established that facilitated the transfer of PE cells to the heart. Subsequent to the establishment of this “secondary dorsal mesocardium,” PE-derived mesothelial cells spread over the naked myocardial surface of the heart in the form of a continuous epithelial sheet that represented the primitive epicardium (Fig. 6D,E,I,J).

Species-Specific Differences in the Topographical Relationships of the Two PE Anlagen

The two PE anlagen of chick embryos did not only show differences in their growth behaviour but additionally showed differences in their topographical relationships to neighbouring tissues that, interestingly, evolved shortly before the formation of the PE villi on the right sinus horn. Originally, the ventral side of chick embryos, including the areas of the sinus venosus harbouring the future PE anlagen, lies opposite to the yolk sac (not shown). From HH-stage 12 onward, however, chick embryos normally undergo a rightward rotation around their cranio-caudal axis that is mechanically linked to the rightward-looping of the heart tube. Axial rotation of chick embryos starts in the head region (HH-stage 12/13) from where it proceeds in a caudal direction towards the tail region (HH-stage 20). Due to the rightward rotation of the chick embryo, the right sinus horn including the right PE-anlage normally shifts away from the yolk sac and, as a consequence of a simultaneous caudal displacement of the ventricular region of the heart loop, comes to lie opposite to the myocardial surface of the developing ventricles. The left PE-anlage, on the other hand, retains its initial topographical relation to the yolk sac (Fig. 8A,B).

Figure 8.

Species-specific differences in the topographical relationships of the two PE anlagen. Transverse histological sections from normal chick (A,B; HH-stage 17) and mouse (C,D; ED9.5) embryos. A,B: In chick embryos, the right sinus horn including the right PE-anlage, normally lies opposite to the myocardial surface of the developing ventricles, whereas the left PE-anlage lies opposite to the yolk sac. C,D: In mouse embryos, the two PE-anlagen never lie directly opposed to the yolk sac and they both develop opposite to the myocardial surface of the developing ventricles. c, peritoneal portion of the coelomic cavity; fg, foregut; lpe, left PE-anlage; pc, pericardial portion of the coelomic cavity; rpe, right PE-anlage; st, septum transversum with liver anlage; sv, sinus venosus; v, ventricular bend; ys, yolk sac; y, yolk. Scale bars = 100 μm.

Due to species differences in the development of the pericardial cavity (initially opened pericardial cavity in chick embryos vs. initially closed pericardial cavity in mouse embryos) and due to differences in cardiac looping, the two PE-anlagen of mouse embryos are never directly opposed to the yolk sac and they both develop opposite to the myocardial surface of the developing ventricles (Fig. 8C,D).

Development of the PE in Chick Embryos Subsequent to Experimental Changes of the Topographical Relationships of the Two PE Anlagen

The observation that only the right PE anlage of chick embryos normally develops the mature PE phenotype poses the question as to which signalling mechanism(s) might be responsible for either a left-sided suppression or a right-sided stimulation of PE development. The fact that the left PE anlage normally develops opposite to the yolk sac and the right PE anlage normally develops opposite to the ventricular myocardium suggest that the asymmetric formation of the chicken PE normally might be controlled by factor(s) secreted by neighbouring tissues. The yolk sac mesoderm, for example, might be suspected to secrete factor(s) with PE suppressing activity whereas the ventricular myocardium might be suspected to secrete factor(s) with PE stimulating activity. To test this hypothesis, we carried out two different sets of experiments. In the first experiment, we induced inversion of the normal embryonic body rotation and cardiac looping in whole embryo cultures at HH-stage 12. This was achieved by a purely mechanical displacement of the c-shaped heart loop towards the left side of the embryo. After 24 hours of in vitro culture, all experimental embryos (n = 12; HH-stages 16/17) showed complete inversion of the normal body rotation as well as inversion of the normal topographical relationships of their two PE anlagen. The left PE, thus, developed opposite to the ventricular myocardium and the right one developed opposite to the yolk sac (Fig. 9C,D). Despite this fact, we found that all experimental embryos still had developed the normal pattern of PE asymmetry (Fig. 9A–D).

Figure 9.

Formation of the PE in chick embryos subsequent to mechanically induced inversion of the normal embryonic body rotation and cardiac looping as shown by SEM (A), whole mount ISH with the PE marker gene Tbx18 (B), and transverse histological sections (C,D). After 24 hr of whole embryo culture, experimental embryos (HH-stages 16/17) showed inversion of the normal topographical relationships of their two PE anlagen. The right PE (large arrow) developed opposite to the yolk sac (C) and the left one (small arrow) developed opposite to the ventricular myocardium (D). Despite this fact, all experimental embryos had developed the normal pattern of PE asymmetry. a, primitive atrium; aip, anterior intestinal portal; rs, right sinus horn; ls, left sinus horn; pc, pericardial portion of the coelomic cavity; y, yolk; ys, yolk sac. Scale bars = 200 μm.

In the second set of experiments, we isolated the developing PE of influences from the yolk sac by explanting tissue blocks carrying the heart and the two PE from HH-stage 12/13 embryos. The tissue blocks were mounted in a stretched state on pieces of the egg shell membrane to prevent the caudal displacement of the ventricular bend toward the region of the right PE anlage during subsequent development in in vitro culture. After 24 hr of in vitro culture, all specimens (n = 5) showed beating hearts with an immature position of their ventricular bends cranial to the venous pole of the heart. With respect to PE development, we found the normal HH-stage 16/17 pattern of PE asymmetry in every specimen (Fig. 10).

Figure 10.

Formation of the PE in organ culture as shown by SEM micrographs. Ventral views of a chick embryo heart (explanted at HH-stage 12) after 24 hr of organ culture. A: Low-magnification view showing the heart on the egg shell membrane. B: Higher magnification view of the boxed area marked in A. The large arrow points to the right PE and the small arrow points to the left PE. C: High-magnification view showing the right and left PE on the ventral wall of the sinus venosus. In cultured hearts, the development of the PE proceeded in the normal asymmetric pattern. Abbreviations as in Figure 6. Scale bars = 500 μm (A), 100 μm (B), 50 μm (C).

The yolk sac mesoderm and the ventricular myocardium might not be the only possible sources for asymmetric signals with PE-inducing or -repressing activity. The asynchronic formation of the two PE anlagen in chick embryos, for example, led us to speculate that the formation of the left PE normally might be suppressed by factors secreted by the right PE, which grows earlier. To test this hypothesis, we carried out a third set of experiments. In these experiments, we produced cardia bifida in ovo and analyzed the fate of the left and right hearts until HH-stage 25, when the heart loop is normally completely covered with epicardium. In the cardia bifida embryos, the PE was still formed only on the right side (Fig. 11A,B) and only the right heart became covered with epicardium (Fig. 11C–F).

Figure 11.

Formation of the PE and primitive epicardium in cardia bifida chick embryos as shown by whole mount ISH with a probe for Tbx18 (A,B) and immunohistochemical staining with an anti.RALDH2 antibody (C–F). A: Ventral view of the two hemi-hearts in a HH-stage-15 embryo. B: Transverse histological section through the sinus venosus regions of the embryo shown in A. Note the asymmetric expression of the PE marker gene Tbx18, which is prominently expressed on the right half of the sinus venosus (arrow) but only weakly expressed on the left half of the sinus venosus. C: Ventral view of the two hemi-hearts in a HH-stage-22 embryo. D–F: Transverse histological sections through the ventricles of the two hemi-hearts from the embryo shown in C. Note that only the right hemi-heart is covered with RALDH2-positive epicardium, whereas the left one shows a naked myocardial surface (red arrowheads). E, primitive epicardium; LH, left hemi-heart; LS, left sinus venosus; LV, ventricle of the left hemi-heart; PE, proepicardium; RH, right hemi-heart; RS, right sinus venosus; RV, ventricle of the right hemi-heart.

DISCUSSION

Our present data confirm recent evidence suggesting that vertebrate embryos do not form a single unpaired PE anlage, as was suggested by initial morphological data (Kurkiewicz, 1909; Virágh and Challice, 1981; Komiyama et al., 1987; Männer, 1992; Virágh et al., 1993), but form a bilaterally paired PE anlage (Muñoz-Chapuli et al., 1997; Männer et al., 2001; Schlueter et al., 2006). The failure of earlier studies to uncover and clearly document the presence of bilaterally paired PE anlagen might be explained by several facts. Firstly, a free ventral view of the developing PE is normally impeded by the ventricular bend of the heart loop, so that, in most previous studies, the PE was usually examined from left or right lateral views. Such views, however, cannot give exact information about the number and position of PE anlagen. Secondly, chick embryos, which were the first embryos in which the function of the PE was identified (Kurkiewicz, 1909), normally undergo torsion around their cranio-caudal body axis. This torsion brings the poorly developing left PE anlage out of sight and favours a view of the right PE anlage. Thirdly, many previous studies have documented only the fully developed PE and have not paid attention to earlier stages of PE development. As shown by our present data, however, the fully developed PE does not show its originally paired character and appears as a single structure in mouse as well as in chick embryos.

Because of the above-mentioned technical problems in visualizing the PE, we decided to study PE development in embryos that had been fixed in a stretched state without body torsion. This approach facilitated a free ventral view of the developing PE so that we were able to document, for the first time, the development of the paired PE anlagen in mouse and chick embryos during a period that started with the first morphological appearance of the two PE anlagen and ended with the initial phase of formation of the primitive epicardium. We, thereby, found interesting differences in PE development between the two species. Firstly, in the mouse embryo the two PE anlagen appeared simultaneously on ED8.5, whereas in the chick embryo the right PE anlage appeared some hours earlier than the left PE anlage (HH-stage 14 vs. HH-stages 15/16). Secondly, in the mouse embryo the two PE anlagen merged with each other in the embryonic midline and contributed equally to the formation of the mature PE, whereas in the chick embryo, the left PE anlage disappeared and only the right PE anlage formed the definitive PE. Development of the PE, thus, followed a bilaterally symmetric pattern in the mouse embryo and a bilaterally asymmetric pattern in the chick embryo.

Are the Observed Differences in PE Development Related to the Use of Different Mechanisms for the Transfer of PE Cells to the Heart?

It is tempting to speculate that the above-mentioned species differences in PE development might be related to the already-known differences in the mechanism of PE cell transfer to the heart, which is accomplished via free-floating cell vesicles in mammals (Komiyama et al., 1987; Kuhn and Liebherr, 1988; Männer et al., 2001) and via the formation of a secondary tissue bridge in birds (Männer, 1992; Männer et al., 2001; Nahimey et al., 2003). It is conceivable that a large number of the released PE cell vesicles do not reach the heart and become lost in the coelomic cavity. The PE cell transfer via free-floating vesicles might, therefore, be less effective than the direct immigration of PE cells via a secondary tissue bridge. The purpose of the symmetric growth of both PE anlagen in mammalian embryos then might be to guarantee the production of an excessive number of PE cells, in order to compensate for the possible loss of free-floating PE vesicles. In avian embryos, on the other hand, the growth of only one of the two PE anlagen might normally prevent excessive production of immigrating PE cells. This idea, however, conflicts with data from dogfish embryos. These embryos also produce free-floating PE vesicles instead of a secondary tissue bridge (Muñoz-Chápuli et al., 1997; Männer et al., 2001). In contrast to mice, however, they do not show a bilaterally symmetric but an asymmetric pattern of PE development. As in chick embryos, the right PE anlage of dogfish embryos has been found to appear earlier than the left anlage (Muñoz-Chápuli et al., 1997) and the latter also does not reach the full size of the mature right PE (R. Muñoz-Chápuli, personal communication).

The presence of a bilaterally asymmetric pattern of PE development, thus, neither seems to correlate with the use of a certain mechanism for PE cell transfer nor does it seem to represent a specific feature of avian embryos. The fact that this pattern occurs in such distantly related species such as the dogfish and the chick rather suggests that it might be evolutionary and poses the questions as to whether the bilaterally symmetric pattern of PE development found in mouse embryos might be an evolutionary young pattern specific to mammalian embryos. In view of the current lack of data from other vertebrate embryos, especially from frequently used model organisms such as the zebrafish and Xenopus, an answer to this question cannot be given at the present time but must await the results of future studies.

Is the Development of the Chicken PE Directly Linked to the Specification of Left- and Right-Sided Body Identities?

It is a well-known fact to physicians, morphologists, and experts in several biomedical disciplines that the internal organs of human beings and other vertebrates display species-specific morphological and topographical left–right (L–R) asymmetries. Our present data disclose a hitherto neglected L–R asymmetry of the cardiovascular system that is only transiently present during a short period of embryonic development. At the present time, it is largely unclear what functional relevance the L–R asymmetries of internal organs might have for the developing and mature organism. As discussed above, the same holds true for the asymmetric pattern of PE development presented here. During the past few years, however, fascinating insights have been gained into the embryological background of vertebrate L–R asymmetries. Molecular signalling cascades have been identified that regulate the morphogenesis of L–R asymmetries via the specification of left- and right-sided body identities during early phases of embryo development (for reviews see Raya and Belmonte, 2004; Levin, 2005). Thus, the question arises as to whether the asymmetric development of the chicken PE might be regulated either by signals directly linked to the recently identified L–R signalling cascades or by signals not directly linked to these signalling pathways.

In the present study, we have found that the left and right PE anlagen of chick embryos do not only show striking differences in gene expression and growth behaviour but additionally show different topographical relationships to neighbouring tissues during the phase of PE development. The left PE anlage has been found to develop opposite to the yolk sac whereas the right PE anlage develops opposite to the ventricular bend of the heart loop. We have additionally found that the two PE anlagen of mouse embryos do not show striking differences in their topographical relationships to neighbouring tissue. These findings suggest that the bilaterally asymmetric growth of the chicken PE might be a consequence of its different topographical relationships to tissues that might have PE-inducing (ventricular myocardium) or repressing (yolk sac mesoderm) activity. To test this hypothesis, we analyzed the development of the chicken PE in two sets of experiments: firstly, in organ cultures in which the two PE anlagen were physically isolated from the suspected PE inducer and repressor, and, secondly, in whole embryo cultures in which the normal topographical relationships of the two PE had been inverted by the mechanical induction of inversion of cardiac looping and embryonic body rotation. Both experiments were started at HH-stages 12/13 and, thus, were performed a long time after the establishment of the above-mentioned molecular L–R identities, which takes place between HH-stages 5 and 8 (Levin et al., 1995; Logan et al., 1998; Ryan et al., 1998). The fact that the development of the two PE anlagen proceeds in the normal asymmetric pattern in both sets of our experiments excludes the possibility that the ventricular myocardium and the yolk sac mesoderm have strong influences on PE development and suggests that the bilaterally asymmetric development of the chicken PE is not the consequence of side-specific differences in the topographical relationships of the two PE anlagen to neighbouring tissues. The yolk sac mesoderm and the ventricular myocardium, however, might not be the only possible sources for asymmetric signals with PE-inducing or -repressing activity. Our present results have shown that, in chick embryos, the formation of the two PE anlagen proceeds in an asynchronic fashion in which the right one appears earlier than the left one. This led us speculate that, in chick embryos, the development of the left PE might be suppressed by factors secreted by the right PE, which grows earlier. To test this hypothesis, we have carried out a third set of experiments. In this experiment, we produced cardia bifida embryos and analyzed the fate of their left and right hearts until HH-stage 25, when the heart normally is completely covered with epicardium. Cardia bifida were produced by a pure mechanical intervention at HH-stages 7–8 and, therefore, did not interfere with the specification of L–R body identities. In the cardia bifida embryos, the PE still formed only on the right side and only the right heart became covered with epicardium. This suggest, firstly, that the right PE anlage does not suppress the development of its left counterpart and, secondly, that the different fates of the two PE anlagen might be laid down before HH-stage 7.

To summarize, we can say that the data from the above-mentioned three experiments suggest that the asymmetric development of the chicken PE does not result from tissue interactions occurring after the phase of specification of the molecular L–R identities. We, therefore, speculate that the asymmetric development of the chicken PE might be directly linked to signalling pathways that specify the left- and right-sided body identities. The fact that mouse embryos do not show an asymmetric but a bilaterally symmetric pattern of PE development must not necessarily speak against this idea since it is well known that L–R development in mice differs in several respects from that in chicks (for review see Schlueter and Brand, 2007). To test our hypothesis, it would be necessary to analyze the development of the PE in chick embryos subsequent to experimental changes of L–R specification. During the past few years, several reliable methods have been developed to induce predictable changes in the specification of left- and right-sided body identities in early chick embryos. These interventions, however, only work in whole embryo cultures, which usually do not facilitate the survival of early explanted embryos beyond HH-stage 12. Since the growth of the PE starts at HH-stage 14, it is impossible to analyze the possible effects of changes in L–R specification on PE development in whole embryo cultures. Thus, in ovo experiments are needed to address this question. At the present time, however, we have no reliable methods to change the L–R specification of chick embryos in ovo. The answer to the question as to whether the asymmetric development of the chicken PE is directly linked to the recently identified L–R signalling cascades, thus, awaits the invention of such methods.

The Chick Embryo Might Be a Good Model to Identify Signals Inducing and Stimulating the Formation of the PE

For a long time, nothing was known about the tissues and factors involved in the induction and stimulation of PE formation (Männer et al., 2001). Recent studies have started to shed light on this topic by showing that BMP and fibroblast growth factor (FGF) signalling play important regulatory roles in the commitment of pericardial mesoderm to the PE identity in chick embryos (Schlueter et al., 2006; Kruithof et al., 2006). Our present study provides information that might further stimulate the search for PE inducers. Due to the fact that the development of the chicken PE proceeds in an asymmetric pattern, candidate inducers as well as repressors might be identified by corresponding asymmetric expression patterns in the wall of the sinus venosus. The functional significance of the candidate signalling molecules can ideally be tested in the chick embryo by the application or expression contralateral to their normal expression domain in the sinus venosus.

EXPERIMENTAL PROCEDURES

Mouse Embryos

Pregnant mice (NMRI) were obtained from a commercial supplier (Harlan Winkelmann, Borchen, Germany). Embryos were collected on embryonic days (ED) 8.5, 9.0, 9.5, and 10.5. Pregnant mice were sacrificed by cervical dislocation and the embryos were obtained by laparotomy and uterotomy.

Chick Embryos

Fertilized chicken eggs (White Leghorn, Gallus gallus) were obtained from a commercial supplier (Lohmann Tierzucht, Cuxhaven, Germany). Eggs were incubated at 38°C and 75% relative humidity until they had reached developmental stages 14 to 19 (3rd and 4th incubation day) according to Hamburger and Hamilton (1951).

Preparation and Fixation of the Embryos for Scanning Electron Microscopy (SEM) and Whole Mount In Situ Hybridization (ISH)

An exact determination of the number and topography of PE anlagen requires a free ventral view of the venous pole of the heart where the PE forms. This is physically impeded by the ventricular bend of the heart loop, which lies in front of the PE. In most previous studies, the PE, therefore, was usually examined in left or right lateral views, which could give only limited information about the number and left- or right-sided identities of the PE. To achieve a free ventral view of the PE, we fixed our mouse and chick embryos in a stretched state. Due to the stretching of the embryo along its cranio-caudal axis, the ventricular bend is displaced to its early embryonic position cranial to the atria. Fixation of the embryos in a stretched state was performed as follows: subsequent to their removal from the uterus or the egg, embryos (freed from extraembryonic membranes) were transferred to a Petri dish filled with Locke's solution. The bottom of the Petri dish was covered with a layer of wax. Insect needles were inserted into the mouths and caudal trunks of the embryos to physically fix them to the bottom of the Petri dish in a stretched state. Concomitantly with stretching, we removed the physiological body torsion around the embryonic cranio-caudal axis. The pericardial portion of the coelomic cavity then was opened and the still beating heart loop was externally rinsed with a calcium-free Locke solution of 20 mmol/l manganese chloride (via a micropipette) to induce cardiac arrest in general dilation (Asami, 1979). After cardiac arrest, the embryos were externally rinsed (via a micropipette) with a 25% solution of glutaraldehyde (specimens for SEM) (Männer, 1992) or a 4% solution of paraformaldehyde (specimens for whole mount ISH) for a short time (1–2 min). These treatments led to a rapid chemical prefixation of the embryos in the stretched state and prevented unintentional morphological alterations caused by subsequent handling of the embryos (e.g., removal and change of solutions in the Petri dish). Final fixation of the embryos for SEM or ISH was carried out according to established protocols (Männer et al., 1996; Andrée et al., 1998).

Scanning Electron Microscopy (SEM)

Embryos fixed for SEM were dehydrated in the usual manner and dried by the critical point method. The dried specimens were mounted on aluminium taps with conducting silver and sputter-coated with gold-palladium to a layer of about 40 nm (“Cool” Sputtering System Type E 5100, Polaron Equipment). Embryos were examined and photographed in a Zeiss DSM 960 scanning electron microscope.

In Situ Hybridization (ISH)

Whole mount in situ hybridization was carried out as described (Andrée et al., 1998). We analyzed expression of Bmp-2/-4, Tbx18, and Wt1 with digoxigenin-labelled riboprobes. For the detection of Tbx18 in chicken embryos, a 1.0-kb partial cDNA clone (ChEST861E19) was identified in the ChickEST Database (Boardman et al., 2002) and obtained from the MRC geneservice, while the Wt1 probe was kindly provided by T. Kudo, Ibaraki, Japan. The constructs for detection of mouse Bmp-2 (1.2-kb partial cDNA clone; Dickinson et al., 1990) and mouse Bmp-4 (fragment size 1.6 kb; Jones et al., 1991) were received from B. Hogan, Nashville, TN. For the detection of mouse Tbx18 and Wt1, we used a 2.0-kb Tbx18 cDNA probe kindly provided by A. Kispert (Kraus et al., 2001), and a 1.5-kb Wt1 probe kindly provided by H. Scholz (Wagner et al., 2005). For sectioning, the embryos were infiltrated overnight at 50°C with a 7.5% gelatine/15% sucrose solution and snap-frozen in dry ice cooled isopentane. Sections (20 μm) were obtained in a cryostat.

Immunohistochemistry (IHC)

Embryos were fixed and prepared according to established protocols (Männer et al., 2005). For immunohistochemical detection of epicardial cells, we used an anti-RALDH2 antibody (Xavier-Neto et al., 2000) kindly provided by Peter McCaffrey and Ursular Dräger. Embryos were washed three times in PBS and fixed in a methanol/DMSO mixture (4:1) at 4°C overnight. Endogenous peroxidase was blocked by incubating the embryos with methanol/DMSO/30% H2O2 mixture (4:1:1) for 2 hr at room temperature. Unspecific binding sites were blocked by incubating the embryos 2× 1 hr in 2% BSA, 0.1% Triton X-100 in PBS at room temperature. Embryos were incubated overnight in a 1:400 dilution of the Raldh2 antibody, which had been preincubated with chick embryo powder. After 5 washes for 1 hr with TBST (0.8% (w/v) NaCl, 0.02% (w/v) KCl, 25 mM Tris-Cl, pH7.5, 1% Tween-20), the embryos were incubated with a 1:200 dilution of a horseradish peroxidase conjugated horse anti-mouse antibody (Vector Laboratories, Burlingame, CA). After 5 washes for 1 hr with TBST, the immunoreaction was color developed using diaminobenzidine. For sectioning, the embryos were infiltrated overnight at 50°C with a 7.5% gelatine/15% sucrose solution and snap-frozen in dry ice cooled isopentane. Sections (20 μm) were obtained in a cryostat.

Whole-Embryo Culture of Chick Embryos

Our method used for chick whole-embryo culture is a modification of the procedure originally described by Chapman et al. (2001). Eggs were opened at the 3rd incubation day (HH-stage 11/12) and their content was put into a sterile Petri dish without damage to the yolk. The albumin on the surface of the yolk was carefully removed with sterile absorbent paper. A sterile filter paper ring (inner/outer diameter 22/26 mm) then was placed on top of the yolk, so that the embryo was located in the centre of its opening. The paper ring containing the embryo and its extra-embryonic membranes was excised and transferred to a 35-mm dish where it was placed with its ventral side up on an agar medium. A second filter paper ring was placed on the first ring, so that the extra-embryonic membranes were sandwiched between the two rings. This prevented loosening and displacement of the embryo from its framing filter paper ring during subsequent culture. A thin liquid film of Ham's F-10 was created over the endodermal surface to prevent drying of the embryo. Cultures were incubated under sterile conditions in a humidified atmosphere at 38°C.

Induction of Inversion of the Normal Topographical Relationships Between the PE Anlagen and Neighbouring Tissues

The two PE anlagen of chick embryos show side-specific differences in their topographical relationships to neighbouring tissues (ventricular myocardium, yolk sac mesoderm). These differences are the consequences of the rightward rotation of the embryo around its cranio-caudal axis and the right-sided position of the embryonic heart loop. To test whether the different topographical relationships to suspected PE inducers (ventricular myocardium) or PE repressors (yolk sac mesoderm) might cause the asymmetric formation of the two PE anlagen of chick embryos, we produced mechanical inversion of cardiac looping and body rotation according to a technique originally developed in our laboratory (Steding and Seidl, 1980). For this purpose, embryos were removed from the eggs before the commencement of body rotation (at HH-stages 11/12) and prepared for whole-embryo culture as described above. The ventricular bend of the heart loop was gently displaced from the right to the left side of the body using blunt forceps. Thereby, the normal D-loop was converted into a false L-loop (Männer, 2004). Creation of false L-loops usually leads to the inversion of the normal body rotation during subsequent development, demonstrating that the embryonic body rotation is mechanically linked to cardiac looping. Subsequent to the production of false L-loops, embryos were cultured under the conditions described above. After 24 hr incubation, embryos were visually examined under a dissecting microscope and prepared for further examinations (SEM, ISH, conventional histology).

PE Organ Culture

To test whether the two PE anlagen of chick embryos would also form in an asymmetric pattern when grown in distance or isolation from suspected inducers/repressors, we have developed a new organ culture technique. Tissue blocks carrying the embryonic heart and the two venous limbs surrounding the anterior intestinal portal were explanted from HH-stage-12 chick embryos and transferred to a Petri dish with sterile Locke's solution. Here, the tissue blocks were mounted in a stretched state on a rectangular piece of egg shell membrane. Stretching of the tissue blocks prevented the normal caudal displacement of the developing ventricles towards the region of the PE anlagen during subsequent development in culture. The pieces of the egg shell membrane carrying the tissue blocks then were transferred into dishes of 4-well plates containing 200 μl of Ham's F-10 medium and cultured under sterile conditions at 38°C. After 24 hr of in vitro culture, the tissue blocks were examined under a dissecting microscope and prepared for further examinations (SEM, ISH).

Production of Cardia Bifida Embryos In Ovo

Fertilized eggs were incubated until embryos reached HH-stage 7–8. The embryos were made accessible through a window in the egg shell, and black ink (Pelican; diluted 1:10 in Pannett-Compton buffer) was used as a contrast medium. The anterior fused heart fields were divided mechanically by cutting a longitudinal slit with a tungsten needle through the midline of the cardiac crescent. The egg shell was re-sealed and operated embryos were incubated until HH-stage 12 to 25. Subsequently, these embryos were processed for ISH and immunohistochemistry (IHC), respectively.

Acknowledgements

We thank Mrs. Kirsten Falk-Stietenroth and Mr. Hannes Sydow for technical and photographic assistance and Mrs. Cyrilla Maelicke for correcting the English manuscript.

Ancillary