Vessel development within the embryonic heart occurs via two processes: vasculogenesis and angiogenesis (Rongish et al.,1994; Risau,1995,1997). Vasculogenesis is the formation in situ of coronary vessels from endothelial cell progenitors (angioblasts) or angioblast migration to areas of vessel formation and their subsequent differentiation into vascular channels. Angiogenesis is the development of vessels from preexisting ones by capillary sprouting, intussusceptive growth, and remodeling (Risau,1997; Ratajska et al.,2003). The first morphological signs of vasculogenesis within an embryo are “blood islands,” which consist of erythroblasts and premature endothelial cells (angioblasts). Blood islands are encountered within the yolk sac of 7.5–13 dpc embryos. The first sings of vasculogenesis in heart development are blood island-like structures. This term was introduced due to the structures' morphological resemblance to the blood island of the yolk sac (Rongish et al.,1994). These structures assemble in clusters prior to their presumed coalescence and differentiation into vascular channels (Tomanek et al.,1996). Blood island-like structures are located within the myocardial wall and the subepicardium of 10 and 11 dpc mice (Virágh and Challice,1981). They are encountered within 13 dpc rat hearts (Rongish et al.,1994; Ratajska and Fiejka,1999). They are also found in the embryonic heart in later stages of development (until full term). Little is known about the derivation of the blood cell component of the blood island-like structures and their relationship to hematopoietic differentiation stages.
Heart vascularization immediately follows the epicardial expansion. Hearts devoid of the epicardium do not develop vascular channels (Kwee et al.,1995). The epicardium derives from an extracardiac source (Kurkiewicz,1909): proepicardial villi, which form on the ventral surface of the sinus venous horns as cellular protrusions growing toward the pericardial cavity and reaching the opposite surface of the looping heart (dorsal atrioventricular sulcus) (Ho and Shimada,1978; Virágh and Challice,1981; Kuhn and Liebherr,1988; Hiruma and Hirakow,1989; Männer,1992; Virágh et al.,1993; Männer et al.,2001). The proepicardium subsequently spreads over the surface of bare myocardium, giving rise to the epicardium and the subepicardial mesenchyme. Some cells of the epicardial mesothelium possess the ability to migrate to the subepicardial space undergoing epicardial-mesenchymal transformation. These mesenchymal cells are considered to be a source of smooth muscle cells that make up the tunica media of coronary vasculature and interstitial fibroblasts as well as fibroblasts of the adventitia (Mikawa and Gourdie,1996; Dettman et al.,1998; Gittenberger-de Groot et al.,1998; Vrancken Peeters et al.,1999; Wada et al.,2003). Subepicardial space, particularly at the atrioventricular and interventricular sulcuses, is filled with blood cells, angioblasts, mesenchymal cells, and extracellular matrix material (Virágh et al.,1993; Kálmán et al.,1995).
The origin of coronary vascular endothelial cells is still controversial. One theory claims that endothelial cells derive from liver primordium/sinus venosus region and are transported via proepicardium to the heart's surface (Poelmann et al.,1993,2002). According to the other theory, endothelial cells derive from mesenchymal cells after epithelial-mesenchymal transformation of the epicardial mesothelium (Mikawa and Gourdie1996; Muñoz-Chápuli et al.,1999,2002; Pérez-Pomares et al.,2002). Although both theories have accumulated some experimental evidence, the derivation of endothelial cells within the embryonic heart is ambiguous. There is also a possibility that coronary endothelial cells are of dual origin. Regardless of their origin, they arrive to the heart from the epicardial surface. Close association of the nucleated red blood cells (NRBCs) with primordial endothelial cells at early stages of vasculogenesis might indicate that either both kinds of cells derive from a common precursor, or blood cells derive from migrating angioblasts (Morabito et al.,2002). The existence of hemangioblast as the common progenitor of the angioblast and blood cell has been suggested before (Pardanaud et al.,1989). The current evidence for this opinion comes from common molecular markers in cells of endothelial/angioblastic and hematopoietic potential (Millauer et al.,1993; Lin et al.,1995; Shalaby et al.,1995; Young et al.,1995; Eichmann et al.,1997). In addition, the areas of embryos where hematoblasts appear are strictly related to hematopoietic and vasculogenic events (yolk sac) (Haar and Ackerman,1971; Tavassoli,1991). The issue of the relationship between hematopoietic cells and vasculogenesis in embryonic heart development has been raised in a current study by Kattan et al. (2004). We wanted to add some relevant data to this topic.
Our purpose was to address derivation of blood cells that are constituents of blood islands of the embryonic heart at the stages before a connection between coronary vessels and the peripheral circulation is established. We wanted also to study a chronology of the appearance of NRBCs in the embryonic hearts and to characterize the relationship between vascular structure appearance and hematopoietic stage of NRBC. Proepicardium as a possible source of NRBC was also explored. All these studies were performed on mouse embryos at the stages since 9.5 through 13.5 dpc.
Although the embryonic hematopoiesis has been the topic of many experimental papers and recent reviews (Dzierzak et al.,1998; Dzierzak,2002,2003), the issue of the embryonic heart as a hematopoietic organ has been addressed (Virágh et al.,1990,1993; Kálmán et al.,1995) but has not been proven so far. Thus, red blood cells within early vascular structures of the embryonic heart may derive from common hemangioblastic precursors of endothelial cell and blood cell lineage or from independent hematoblastic or endothelial cell lines. They may also enter the heart directly from circulation at the established stage of erythroid differentiation.
MATERIALS AND METHODS
All procedures were performed according to requirements of the Ethical Animal Care Committee of Poland (which is also in accordance with the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals).
Animals (pregnant mice, 9, 9.5, 10, 10.5, 11, 11.5, 12, 12.5, 13.0 dpc) were sacrificed under anesthesia (narcotan and chloral hydride 100 mg/kg b./w. i.p.); fetuses were removed, decapitated, and fixed immediately in 2.5% glutaraldehyde/2% paraformaldehyde solution in 0.1 M cacodylate buffer, pH 7.6, for 3 hr. Subsequently, tissue was rinsed in 0.1 M cacodylate buffer, osmicated in 1% OsO4 for 45 min, dehydrated in graded series of ethanol solutions, and embedded in Spurr resin. Embryos were oriented for sagittal sectioning. Semithin sections were cut serially every 20–30 μm and were stained with toluidine blue for analysis under a light microscope.
Some embryos were fixed in zinc fixative to preserve antigenic determinants of CD34 (Beckstead,1994), or in a buffered formalin (to preserve the antigenic determinant of Ter/119 and CD45), dehydrated in increasing alcohol concentrations, cleared in xylene, and embedded in paraffin. Other set of embryos were frozen immediately for obtaining cryostat sections, suitable for anti-PECAM1 staining and some hematopoietic markers staining. At least three embryos for each of the above-mentioned time points were sacrificed and taken for analysis.
Paraffin sections were deparaffinized on a hot plate and 3 in changes of xylene and alcohol, 10 min each. For detection of CD45 antigen sections were treated with Antigen Retrieval solution according to manufacturer requirements (Pharmingen). Sections were then incubated with rat anti-CD34 (1:30; Pharmingen), anti-CD45 (1:30; Pharmingen), or anti-Ter/119 (1:30; Pharmingen) antibodies for 60 min in a humid chamber or overnight at 4°C, washed with TBS, and subsequently treated for 30 min with antirat IgG-biotin conjugate (Pharmingen). After washing in TBS, the sections were stained with streptavidin-peroxidase complex (Dako) and the color reaction was developed with DAB (Sigma). For double labeling, frozen sections were cut and stained with anti-Ter/119 (or anti-CD34) antibodies and the same sections were stained with goat anti-PECAM1 antibodies, (Santa Cruz Biotechnology), rinsed as above, and incubated with antigoat alkaline phosphatase complex (Sigma) and antirat-IgG-biotin conjugate. After rinsing with TBS, streptavidin-peroxidase complex was applied on sections. The color reaction was developed for horseradish peroxidase with (DAB) substrate and subsequently for alkaline phosphatase with NBT/BCIP substrate (Dako) or Fast Red substrate (Dako). The sections were finally stained with hematoxylin-eosin and mounted in glycerogel (Keiser).
Semithin sections stained with toluidine blue were analyzed under a light microscope. Selected samples chosen from semithin sections were cut for ultrathin sections for copper grids, counterstained with lead nitrate and uranium acetate according to Reynolds (1963), and examined in a transmission electron microscope (TEM).
To explore possible sources of NRBCs within the early vascular structures of mouse hearts, we examined initially the proepicardium (9.5–10 dpc) and the avascular heart (since 10.5 dpc) for the presence of hematopoietic progenitor cells.
On 9–9.5 dpc, the proepicardium consisted of loosely arranged mesenchymal cells (Fig. 1a) and of cells that exhibited epithelial morphology (Fig. 1b). In TEM, mesenchymal cells developed long cytoplasmic processes of irregular shapes, which were connected with adjacent cells. Mesenchymal cells contained rich Golgi complex and scattered filaments of an intermediate type. Epithelial cells of proepicardium formed primitive intercellular junctions (Fig. 1c). They formed blebs, which protruded toward the pericardial cavity. From the tips of these protrusions, vesicles were released to the pericardial cavity (Fig. 2a and b). The vesicles consisted of one or several epithelial cells. The blebs and vesicles were never positive with anti-CD34 or anti-CD45 antibodies on serial sections (Fig. 2c), which indicates the absence of hematopoietic precursors and endothelial cells in the proepicardium (Wood et al.,1997; Jaffredo et al.,1998). CD34+ endothelial cell precursors were, however, detected in the vicinity of the proepicardium within the septum transversum/primitive liver. The highest number of vesicles within the pericardio/peritoneal cavity was detected on 9.5 dpc. After the vesicles had reached the dorsal surface of the ventricles, they flattened and spread, forming the cover of the heart. The process of the epicardial cover development was accomplished on 11 dpc and after this stage the free-floating vesicles were not encountered within the cavity. Occasional proepicardial vesicles were detected in close association with the heart's surface of 11 dpc embryos.
In myocardium, neither nucleated blood cells nor their precursors (CD34+ or CD45+) were detected at the stages preceding heart vascularization (9.5–10 dpc), although endocardial endothelial cells expressed CD34 antigen (Wood et al.,1997).
Myocardium and Subepicardium
Beginning on 10.5 dpc, NRBCs were found in the wall of embryonic heart in association (in apposition) with angioblasts. They appeared as clusters of closely apposed cells or entrapped in loose assemblies within vascular vesicles. The latter were located in the subepicardium within the interventricular or atrioventricular sulcuses (Fig. 3). Clusters of NRBCs and angioblasts were recognized as blood island-like structures in TEM. The NRBCs residing in blood islands express Ter/119 antigen, whereas endothelial cells exhibited PECAM1 antigen. In addition to the previous TEM studies (Rongish et al.,1994; Ratajska and Fiejka,1999), which claimed that red blood cells were accompanied by angioblasts/endothelial cells, we demonstrated that some of the NRBCs were located in areas where angioblasts were absent. These solitary (free) NRBCs were mostly positioned in the subendocardium. However, some were found also in the subepicardium. The highest number of solitary NRBCs was demonstrated on 11.5–12 dpc. They were recognized on semithin sections (Figs. 4 and 5a) by the use of immunohistochemical labeling with anti-Ter/119 antibodies (Fig. 5b–d) and in TEM (Fig. 5e). They did not express CD34 antigen, indicating that they did not belong to cells of early hematopoietic lineage. Double labeling with anti-PECAM1 and anti-Ter/119 antibodies allowed us to confirm that Ter+ cells resided subendocardially and were not encircled by PECAM+ angioblasts. However, they adjoined PECAM+ endocardial endothelium (confirmed by TEM study; Fig. 5e). Single cells exhibiting Ter/119 antigen were also found to be located on the border of the endocardial endothelium, as if passing through it (Figs. 5d and 6). Some subepicardially located NRBCs adjoining the epicardial epithelium were also distinguished (Fig. 7). An ultrastructural study indicated that NRBCs moved to establish contact with angioblasts. Moving blood cells were recognized morphologically by the presence of pseudopods (Fig. 8a). Subsequently, NRBCs were being enveloped by angioblasts (Fig. 8). Pseudopods or long processes were also visible during angioblast recruitment to blood islands (Fig. 9).
In the subepicardium and myocardium of 11–13 dpc hearts, most NRBCs were accompanied by angioblasts/endothelial cells. This close vicinity of NRBCs and angioblasts represented blood island-like structures. Both circulating NRBCs and residing NRBCs (in blood islands) of the same stage of differentiation consisted of a population of cells that differed with respect to their maturation stages. However, among residing NRBCs, the early erythroblastic progenitors were never encountered. Ter/119+ NRBCs residing in the heart belonged to late erythroblastic stages of differentiation (a halo around a nucleus indicated the commencement of the nucleus shedding; Fig. 10). There were some singly scattered CD45+ cells in peripheral circulation and residing within the heart of 10.5–13 dpc embryos. Some residing NRBCs became enucleated by expulsion of the nucleus (not shown), whereas blood cells circulating in the peripheral blood stream maintained their nuclei longer (Fig. 10b; they lose their nuclei beginning on 15 dpc). Enucleated RBCs coexisted with nucleated ones in blood island-like structures beginning from 12 dpc. Thus, blood islands of the same stage of heart development differentiate and mature at different time points, not simultaneously.
Our study has provided several important observations regarding red blood cell characteristics and chronology of red blood cell occurrence in the blood island-like structures of the embryonic heart. One, at the onset of heart vascularization, NRBCs are found in the heart either accompanied or not accompanied by angioblasts. Two, the solitary (free) NRBC (i.e., not accompanied by angioblasts) are found to be located mostly in the subendocardium, on the border of endocardial endothelium, or within the subepicardium, whereas the ones accompanied by angioblasts/endothelial cells are found in myocardium and in the subepicardium. Three, free NRBCs establish contact with angioblasts. Four, premature erythroid cells and/or hematopoietic stem cells were not encountered in 10.5–13 dpc hearts.
Based on these observations, we postulate that NRBCs reach the heart via diapedesis through endocardial endothelium, whereas endothelial cells/angioblasts reach the heart from the epicardial surface (Muñoz-Chápuli et al.,2002; Poelmann et al.,2002), and that blood island-like structures are formed by angioblasts establishing contact with NRBCs, and angioblasts encircle NRBCs and assemble, forming primitive vascular vesicles or blood islands.
Some individual cells of erythroblastic lineage (which exhibit Ter/119 antigen) (Ikuta et al.,1990) were found to be located on the border of the endocardial endothelium (Figs. 5d, 6), as if pictured while moving across the endocardial endothelium. Although we have not studied the movement of these cells, our indirect morphological observations strongly suggest that NRBCs move through myocardium to come in contact with angioblasts (presence of cytoplasmic processes). They enter the heart at the stage of development corresponding to the established erythroid lineage, not at the stage of hematopoietic progenitor cells. This was confirmed by their more mature morphology (ultrastructure) as compared to the erythroblasts of yolk sac blood islands (Haar and Ackerman1971; Tavassoli,1991) and their Ter/119 antigen expression, the latter being specific to cells of erythroid/erythrocyte stage (Ikuta et al.,1990). A similar observation, namely, that commitment to hematopoietic lineage precedes the formation of blood island-like structures in the embryonic heart, has been suggested in a previous study by Kattan et al. (2004). Contrary to the authors' finding, we have not observed many CD45+ cells within the blood island structures and vascular vesicles of the 10.5–13 dpc hearts. This discrepancy might be caused by species-specific differences in CD45 antigen expression (quail versus mouse). NRBCs encircled by endothelial cells/angioblasts are still able to proliferate, which indicates that not all of these cells are at the terminal stage of erythroid differentiation.
The proepicardium is generally known as the precursor of the epicardial mesothelium, which, after mesenchymal transformation, gives rise to cellular components of the coronary vasculature (Mikawa and Gourdie,1996; Muñoz-Chápuli et al.,1996,2002; Dettmann et al.,1998; Gittenberger-de Groot et al.,1998; Pérez-Pomares et al.,2002; Poelmann et al.,2002). The proepicardium has also been acknowledged as having a potential to deliver endothelial cell precursors when transplanted to fetal liver (Pérez-Pomares et al.,2004). Since the presence of hematopoietic cells in the proepicardium had been previously suggested by Virágh et al. (1993), we wanted to verify this notion. Detailed ultrastructural analysis of the proepicardium allowed us to disprove the hypothesis that NRBCs are supplied by the proepicardium, the presumed source of angioblasts/endothelial cells. Neither red blood cells (bearing Ter/119 antigen) nor any cell of hematoblastic activity (bearing CD34 or CD45 antigen) (Wood et al.,1997) has been detected in the proepicardium. Thus, we come to the conclusion that the authors (Virágh et al.,1993) had considered the proepicardium as an organ that maintained its properties while attached to the heart and spread on its surface. According to current definition, the proepicardial organ is understood as a transient tissue that protrudes to the pericardial cavity (Männer,1999; Männer et al.,2001; Pérez-Pomares et al.,2002; Kattan et al.,2004). When the tissue attaches to the heart, it is no longer considered to be the proepicardium but the epicardium and its derivative, subendocardial mesenchyme. Thus, the presence of hematoblasts or NRBCs detected by Virágh et al. referred to their location in the subepicardium at the onset of heart vascularization (Virágh and Challice,1981; Virágh et al.,1993), which has been consistent with our observation.
A possibility that NRBCs enter the myocardium from the epicardial surface had been suggested before (Virágh et al.,1990). Blood cells have been assumed to be enveloped by endothelial cell progenitors and internalized to the subepicardium and subsequently to myocardium. We, however, were unable to detect any blood cells floating freely within the pericardial cavity when the pericardium was intact during isolation of the embryo (Fig. 2a). The only cells detected within the pericardial cavity formed vesicles deriving from the proepicardium and from the pericardial mesothelium.
Thus, we postulate that NRBCs enter the heart independently from angioblasts. Subsequently, both kinds of cells move toward each other and assemble to form blood island-like structures, which are initially located in the subepicardium. Since the distance from the endocardium to the epicardium at the onset of heart vascularization is short (due to the very thin myocardial wall consisting of two to three layers of cardiocytes with deep trabecular invaginations) (Manasek,1968), the movement of erythroblasts to the heart surface does not take a long time.
Another possibility of RBC derivation in embryonic heart at the onset of vascularization might be in situ differentiation from progenitor cells that had migrated into the heart before. The embryonic heart has been postulated to be the hematopoietic organ in previous papers by Virágh et al. (1990) and Kálmán et al. (1995). There are several reports indicating that hematopoiesis within the embryo is strictly related to the areas of vasculogenesis (Pardanaud et al.,1989). We were unable, however, to demonstrate the presence of hematopoietic stem cells in the embryonic heart neither by the use of TEM or by immunohistochemical staining with antibodies to HSC antigens. We also could not find any blood island exhibiting a pattern of cellular assembly similar to a yolk sac blood island (Haar and Ackerman,1971). Our study indicates that the embryonic heart supplies only new erythroblasts owing to their proliferative capacity within the primitive vascular vesicles at the time before coronary vessels are connected to the aorta (Fig. 3). It is doubtful that the embryonic heart possesses a hematopoietic activity since this activity is always associated with production of many descendent cells, as has been demonstrated in the fetal yolk sac (Haar and Ackerman,1971; Tavassoli,1991), aorta-gonads-mesonephros (AGM) region of the embryo, and fetal liver (Dzierzak et al.,1998; Dzierzak,2002,2003).
Since formation of blood island-like structures occurs throughout the prenatal life (Rongish et al.,1994), it is possible that red blood cells (nucleated or enucleated) enter the embryonic heart also at later stages of development. We were unable, however, to study the mode of their entrance after the formation of the connection between the coronary system and the peripheral circulation, i.e., after 13.5 dpc. A possible mode of red blood cell entrance to the heart at later stages of coronary system development might be simply via diapedesis from capillary plexus.
Our study suggests that NRBCs enter the heart from the endocardium, whereas studies by other authors (Poelmann et al.,1993) have decisively indicated that the endocardial endothelial cells do not take part in the formation of coronary vasculature. Although the endocardial endothelium does not have a potential to differentiate into endothelial cells of coronary vasculature, it may exhibit properties of route via which NRBCs migrate to the myocardial wall.
Thus, it is highly possible that NRBCs arriving from the endocardium assemble together with angioblasts coming from the epicardium. Studies by Tomanek et al. (1996,1999) have demonstrated that there is an epicardial-endocardial gradient of density of newly formed vessels, as well as angiogenic growth factors such as VEGF. Precursors of coronary vasculature arrive to the heart from the epicardial surface (Muñoz-Chápuli et al.,1996,2002; Gittenberger-de Groot et al.,1998; Pérez-Pomares et al.,2002; Poelmann et al.,2002) and settle initially in the subepicardium, which is consistent with the highest density of new vessels in this area. These statistic measurements are not affected by single scattered NRBCs without angioblasts found in the subendocardium.
Thus, we postulate the subsequent sequence of events: erythroblasts are generated within the hematopoietic foci of the 9–13 dpc embryo: yolk sac, AGM (Muller et al.,1994; Medvinsky and Dzierzak,1996; Jaffredo et al.,1998; de Bruijn et al.,2002), placenta (Alvarez-Silva et al.,2003), and liver (Dzierzak et al.,1998), and circulate freely within the peripheral circulation. Subsequently, some of them adhere to endocardial surface and enter myocardium via diapedesis and finally they move toward the nearest angioblasts. Angioblasts arrive from the epicardium, as has been postulated previously by several authors (Poelmann et al.,1993; Pérez-Pomares,2002). The entrance of NRBCs to myocardium takes place concomitantly with angioblasts, i.e., after 10.5 dpc (after the epicardium had covered the majority of the heart's surface).
It is not known whether the presence of nucleated/enucleated red blood cells within the embryonic heart before the coronary system is connected with the aorta is of any significance. NRBCs found in blood island-like structures and within the vascular vesicles of the embryonic heart are certainly a source of erythrocytes within the peripheral circulation from the time point when blood starts to circulate within the coronary system. They may also be a source of certain paracrine substances to the adjacent cells (angioblasts) in vasculogenic events.
The technical assistance of Maria Michniewska and Anna Podbielska is greatly appreciated.