This article was accepted for inclusion in Developmental Dynamics 235 #1 –Cardiovascular Special Issue.
In vivo and in vitro analysis of the vasculogenic potential of avian proepicardial and epicardial cells†
Article first published online: 2 FEB 2006
Copyright © 2006 Wiley-Liss, Inc.
Special Issue: Campos-Ortega Special Focus
Volume 235, Issue 4, pages 1014–1026, April 2006
How to Cite
Guadix, J. A., Carmona, R., Muñoz-Chápuli, R. and Pérez-Pomares, J. M. (2006), In vivo and in vitro analysis of the vasculogenic potential of avian proepicardial and epicardial cells. Dev. Dyn., 235: 1014–1026. doi: 10.1002/dvdy.20685
- Issue published online: 10 MAR 2006
- Article first published online: 2 FEB 2006
- Manuscript Accepted: 13 DEC 2005
- Ministerio de Sanidad y Consumo/ISCIII, Spain. Grant Number: PIO31159
- Ministerio de Ciencia y Tecnología, Spain. Grant Number: SAF2002-02651
- NATO. Grant Number: LST.CLG 980429
- EU, FP6. Grant Number: LSHH-CT-2005-018630
- NICHD. Grant Number: NO1-HD-2-3144
- heart development;
- avian embryo
Coronary vessel formation is a special case in the context of embryonic vascular development. A major part of the coronary cellular precursors (endothelial, smooth muscle, and fibroblastic cells) derive from the proepicardium and the epicardium in what can be regarded as a late event of angioblastic and smooth muscle cell differentiation. Thus, coronary morphogenesis is dependent on the epithelial–mesenchymal transformation of the proepicardium and the epicardium. In this study, we present several novel observations about the process of coronary vasculogenesis in avian embryos, namely: (1) The proepicardium displays a high vasculogenic potential, both in vivo (as shown by heterotopic transplants) and in vitro, which is modulated by vascular endothelial growth factor (VEGF) and basic fibroblast growth factor signals; (2) Proepicardial and epicardial cells co-express receptors for platelet-derived growth factor-BB and VEGF; (3) Coronary angioblasts (found all through the epicardial, subepicardial, and compact myocardial layers) express the Wilms' tumor associated transcription factor and the retinoic acid-synthesizing enzyme retinaldehyde-dehydrogenase-2, two markers of the coelomic epithelium involved in coronary endothelium development. All these results contribute to the development of our knowledge on the vascular potential of proepicardial/epicardial cells, the existent interrelationships between the differentiating coronary cell lineages, and the molecular mechanisms involved in the regulation of coronary morphogenesis. Developmental Dynamics 235:1014–1026, 2006. © 2006 Wiley-Liss, Inc.
Coronary vessels are essential for myocardial performance. This finding is dramatically illustrated by the devastating consequences of coronary artery disease, which often results in an extensive myocardial necrosis leading to cardiac failure. Notwithstanding their extreme importance, much is still to be known about multiple aspects related to the development of these blood vessels. As we will herein show, coronary vessels are special vessels in some regards, and many questions about their origin, differentiation, and spatial patterning remain without answer.
A classic controversy in this field has been that of the embryonic origin of the coronary cell precursors. Many hypotheses were raised about the endocardial (Virágh and Challice,1981), aortic endothelium (Rychter and Oštádal,1971; Hirakow,1983; Conte and Pellegrini,1984; Hutchins et al.,1988), or hepatic (Poelmann et al.,1993) origin of coronary endothelial cells, but recent reports have shown directly or indirectly that an important part of coronary endothelium is a proepicardial/epicardial derivative (Pérez-Pomares et al.,1998a, b, 2002a, b; Tevosian et al.,2000). However, no systematic in vivo or in vitro studies about the extracardiac vasculogenic potential of this tissue are available in the literature.
Not much was known about the origin of coronary smooth muscle either, but a series of papers from the 1990s (Mikawa and Fischman,1992; Mikawa and Gourdie,1996; Dettman et al.,1998; Gittenberger-de Groot et al.,1998; Landerholm et al.,1999; Vrancken-Peeters et al.,1999) and some more recent studies (Lu et al.,2001; Wada et al.,2003) have set a proepicardial/epicardial origin for the muscular cells of the coronary walls.
To give rise to all these cardiac cell lineages, the proepicardium and the epicardium undergo an Epithelial-to-Mesenchymal Transformation (EMT; Markwald et al.,1996; Pérez-Pomares et al.,1997, 1998a; Dettmann et al.,1998; Vrancken Peeters et al.,1999). It is a common belief that the proepicardial/epicardial cell type might be a sort of cardiac stem-like cell (Wessels and Pérez-Pomares,2004), and the pluripotent properties of the epicardial derivatives (Epicardially-Derived Cells, EPDCs) are thought to be directly linked to the molecular mechanisms controlling the epicardial EMT (Pérez-Pomares et al.,2002b).
All these studies and their conclusions have opened a discussion on the nature of coronary angioblasts in relation with the current scenarios about the embryonic origin of endothelial cells. The hemangioblast hypothesis postulates a common origin for endothelial and blood cells (reviewed in Eichmann et al.,2002). On the other hand, a hypothesis on a vascular bipotential progenitor (Yamashita et al.,2000) has suggested that precursor cells can give rise to both endothelium and smooth muscle. No definitive evidence has been provided for coronary angioblasts, either as hemangioblasts or bipotential cell precursors, but we have suggested elsewhere that coronary progenitors could be bipotential (Muñoz-Chápuli et al.,2002; Pérez-Pomares et al.,2002a), a feature that is likely to be unique for these vessels. Another feature supporting the idea of the coronary endothelium as a special one is the up-regulation of the Wilms' tumor protein (Wt1) in the adult coronary endothelium and smooth muscle under hypoxic conditions (Wagner et al.,2003). It is important to remark that Wt1 is expressed in the embryonic epicardium and its derivatives throughout development (Moore et al.,1999; Carmona et al.,2001), but its specific expression in early coronary endothelium has not been reported yet.
A second main question in coronary development has been that of the cellular mechanisms accounting for coronary morphogenesis and vascular patterning. Not surprisingly, angiogenesis (growth of vessels from pre-existing ones, reviewed by Folkman,2003) has been claimed to be the main process driving coronary formation (Poelmann et al.,1993; Virágh et al.,1993; Vrancken Peeters et al.,1997). However, it is currently accepted that early coronary vessels develop through vasculogenesis (Mikawa and Fischman,1992; Pérez-Pomares et al.,1998a, b, 2002a, b; Tomanek et al.,1999, 2001a, b, 2002; Wada et al.,2001; Yue and Tomanek,2001; Kattan et al.,2004), i.e., by the in situ assembly of vascular progenitors (Risau and Flamme,1995), whereas the final growth of the preformed coronary tree occurs by angiogenesis (Tomanek et al.,2001a; Yue and Tomanek,2001). On the other hand, main coronary branches are known to remain in the subepicardium all through the adult life of both avians and some mammals (including humans), while in other mammals (such as rodents) larger coronary vessels will become surrounded by myocardium. We do not know if coronary vasculogenesis is restricted to the subepicardium, the myocardium becoming vascularized only through angiogenic growth of subepicardial vessels, or if intramyocardial vasculogenesis can also occur.
Other points about the fine features of the early process of cardiac vascularization still remain obscure. Coronary vasculogenesis takes place in the subepicardium, an extracellular matrix-rich space between the epicardial lining and the myocardium (Wessels and Pérez-Pomares,2004). Most subepicardial cells express retinaldehyde-dehydrogenase-2 (RALDH2), the main retinoic acid (RA) -synthesizing enzyme in the mesoderm (Berggren et al.,1999). This finding suggests that the vasculogenic process giving rise to the primary coronary network presumably occurs in an RA-rich environment, although we do not really know if retinoic signalling is needed for coronary development as it is for other embryonic vessels (Berggren et al.,2001; Lai et al.,2003) nor which ones are the precise molecular mechanisms triggered by this morphogen during coronary vasculogenesis.
Taking all these points into account, we have carried out a descriptive and experimental study of coronary vasculogenesis in avian embryos with a focus in the molecular identification and characterization of coronary angioblasts. Our work includes several in vivo and in vitro experiments specifically designed to analyze the vasculogenic potential of the proepicardium/epicardium, but also to define the specific features of the differentiation and morphogenesis of coronary vessels.
Heterotopic Vasculogenic Potential of Proepicardium
To test the vasculogenic potential of the proepicardium in vivo, quail-to-chick proepicardial chimeras were constructed using diverse non-cardiac organs as host tissues. Classic proepicardial chimeras (transplantation of quail proepicardia onto developing chick hearts; Männer,1999) were used as controls for this experiment. We also tested the effects of a stimulation of the proepicardium with Vascular Endothelial Growth Factor (VEGF) and basic Fibroblast Growth Factor (bFGF) before transplantation.
Quail proepicardia transplanted into the pericardial cavity of chick embryos generated a whole vascular bed formed by endothelial tubes of different sizes that expressed the QH1 vascular marker (Fig. 1A). Vascularization proceeded from the subepicardium of the dorsal aspects of the atrioventricular (AV) and conoventricular (CV) grooves to the lateral and ventral walls of the ventricles and then to the outer compact myocardial layers. This vascularization sequence includes (1) an early dispersion of isolated vascular precursors (angioblasts) in the subepicardium followed by a progressive infiltration of coronary angioblasts (single cells) into the myocardial layers; (2) their coalescence into small clusters or chords of cells and occasionally into blood island-like structures; and (3) a final fusion and self-assembly into tubularized blood vessels. These three steps have been defined after our own work (Pérez-Pomares et al.,1998a, 2002a, b) and after studies from other laboratories (Mikawa and Fischman,1992; Männer,1999; Tevosian et al.,2000; Tomanek et al.,2001b).
In these chimeras, a major portion of perivascular cells were proepicardial derivatives. A small portion of these cells expressed α-smooth muscle actin (α-SMA; not shown) even before the vascular network contacted the aortic root to become perfused by the systemic blood flow (Hamburger and Hamilton stage [HH] 31). The complete muscularization of coronaries, however, only takes place once the coronary circulation is effective (from HH31 onward). After that time point, the majority of the muscular cells as well as the outer fibroblastic-like (nonmuscular) population of cells that represents the adventitial vessel layer were also found to be proepicardial derivatives (Fig. 1B).
Heart vascularization is a primary consequence of epicardial formation from proepicardial progenitors. Interestingly, when preincubation of proepicardial tissue in VEGF+bFGF was performed before the transplantation, a massive local vascularization of the compact myocardium was induced compared with controls (Fig. 1C,D).
Transplantation of quail proepicardia into liver primordia caused the immigration of donor cells into the hepatic tissues. Scattered quail angioblasts (QH1-positive) were found to be integrated into the endothelium of the liver sinusoids (Fig. 1E). A similar result was obtained when proepicardial tissue was grafted into developing mesonephric ridges. In this case, proepicardial-derived angioblastic cells incorporated to the developing renal vascular bed (Fig. 1F). When transplanted to developing chick limb buds, quail proepicardia caused a coherent vascularization of the dermal and perichondral tissues of the limb primordium. These chimeric vessels were typical interconnected endothelial tubes of variable size (Fig. 1G). Quail proepicardial transplantation to hindgut regions led to the vascularization of the splanchnopleural mesenchyme covering the endoderm of the developing intestine. As described for the case of the limb buds, angioblasts also differentiated from proepicardial-derived cells and acquired a tubular vascular morphology constituting a local network of vessels (Fig. 1H). Both limb bud and intestinal chimeric vessels do normally integrate with the host vasculature (i.e., do not remain as an isolated and closed network of blood vessels).
Coronary Vasculogenesis In Vitro
Proepicardia transplanted onto collagen gels attached to the surface of the gel and spread to form a monolayer of cells (Fig. 2A). Mesenchymal cells soon invaded the collagen layers from the epithelial surface (Fig. 2B) and some of them differentiated into angioblasts. Angioblastic differentiation (revealed by QH1 expression) of quail proepicardial cells cultured on collagen gels was greatly enhanced by preconditioning the gels with VEGF and bFGF (Fig. 2D), although during the first 24 hr of culture coronary angioblasts spontaneously differentiated from the proepicardium when cultured in DMEM+1% chick serum (Fig. 2C). The number of angioblasts was higher in experimental (growth factor treated, Fig. 2F) than in control (Fig. 2E) cultures. This difference was statistically significant after an analysis of the surface covered by QH1-positive cells respect to the total area covered by cells (t = 6.49; P < 0.01). The angioblast density found in the collagen gels after the treatment with VEGF+bFGF supplements was found to be critical for the viability of the vascular cells in the culture, as a low angioblastic density caused early vascular cell death (not shown). Of interest, cell proliferation was quite similar in control and treated cultures as double QH1/anti-phospho-Histone H3-Ser10 immunostainings illustrate (Fig. 2G,H). Under the proper conditions, proepicardially derived angioblasts joined to form clusters and/or chords of cells (Fig. 2I,J), which only occasionally became hollow, constituting a tubular vascular structure (Fig. 2K). When endoderm-conditioned medium, obtained from chick embryonic hindgut culture, was added to the collagen gels, the number of isolated angioblasts was reduced in the cultures; free vascular precursors presented an extremely elongated shape and distributed in a parallel manner with respect to other neighboring angioblasts (QH1-positive). The great majority of QH1-positive cells did coalesce, giving rise to vascular structures of diverse complexity. Small groups of two to six angioblasts aligned to form chords (Fig. 2L–N). Many of these chords remained in close contact (Fig. 2L,M) or fused into tubes with a true lumen as revealed by confocal analysis (Fig. 2N). Some of these vascular structures distributed in the collagen layers following a polygonal pattern (Fig. 2O).
Culture of proepicardial cells on plastic showed the highly proliferative activities of the epicardial progenitors (a monolayer from a single proepicardium with a diameter of approximately 2 mm can be expanded to confluence in a 100 mm Petri dish in approximately 2 weeks). Under high serum conditions (10% fetal bovine serum, 2% chick serum) around 10–15% of the cells in culture expressed the quail vascular marker QH1. Different levels of QH1 expression occurred in this expanding proepicardial populations, from the weak expression of some cells (Fig. 3A) to the much higher levels of other differentiating cells (showing a characteristic punctuated staining pattern, Fig. 3B). Many of these cells did spread and joined through their cell membranes to form a coherent cobblestone-like endothelium (Fig. 3C). Some other vascular cells in culture tightly aggregated and vacuolized forming ring-like structures (Fig. 3D).
Identification and Characterization of Coronary Angioblasts
VEGFR-2 or QH1 immunoreactivity was virtually absent in proepicardial cells before their attachment to the myocardium (not shown in Fig. 4; for more details, see Fig. 5). Once this contact was established, VEGFR-2 expression rapidly increased in the mesothelial lining and proepicardial mesenchymal cells (Fig. 4B,C) as well as in the developing primitive epicardial epithelium, where these VEGFR-2–positive cells were found to distribute in local patches of a few cells (from 2–3 to 10–15). Epicardial epithelial VEGFR-2–positive angioblasts also expressed other epicardial markers such as cytokeratin (CK) and RALDH2 (see Fig. 5). Many VEGFR-2–positive coronary angioblasts from the attached proepicardium, the developing primitive epicardium, and the subepicardium colocalized with the Platelet Derived Growth Factor Receptor β (PDGFRβ; Fig. 4B–F), which is commonly expressed by many EPDCs as shown by quail-to-chick proepicardial chimeras (Fig. 4A). VEGFR-2/PDGFRβ colocalization was especially evident in some small clusters of two or three cells found around the AV groove and the dorsal walls of the ventricles of HH21–HH27 embryos (Fig. 4E).
Coronary Vasculogenesis and RA Signalling
Coronary angioblasts (VEGFR-2–positive) are absent from the proepicardium before its attachment to the myocardium (Fig. 5A). Epicardial and subepicardial angioblasts expressed characteristic epicardial markers such as RALDH2 (Fig. 5A–E). RALDH2 expression was stronger in the cells of the outer epicardial layer, whereas the inner layer (i.e., cells closer to the myocardium) displayed a lower immunoreactivity (Fig. 5D). VEGFR-2/RALDH2 colocalization was typical of immature vascular coronary cell precursors (single cells or few-cell clusters, HH21–HH27), and it became less evident as vascular morphogenesis proceeded (HH31, Fig. 5E,F). No RALDH2 expression was found in intramyocardial vessels.
Retinoid XReceptor alpha (RXRα) immunoreactivity showed a similar pattern to that described for RALDH2, i.e., a higher immunoreactivity in epicardial cells, and a variable degree of staining in subepicardial mesenchymal cells, usually weaker in areas closer to the myocardium (Fig. 5G,H). Double RXRα/QH1 immunostainings also showed variability in RXRα expression in endothelial cells, as only endothelial cells from immature, small vessels, were positive. The endothelium from mature coronary vessels surrounded by perivascular cells was weakly stained or not stained at all.
Wt1 Expression in Coronary Angioblasts
Wilms' tumor transcription factor-1 (Wt1) was expressed in proepicardial, primitive epicardial and subepicardial cells. Wt1 immunoreactivity displayed a sharp gradient from the strongest staining found in proepicardial, epicardial, and outer subepicardial cells (i.e., those cells closer to the epicardial epithelium) to the weak staining found in the inner subepicardial cells closer to the myocardium (Fig. 6A–C). Using the quail vascular marker QH1 for colocalization, a significant number of subepicardial coronary angioblasts were found to express Wt1. Wt1 expression was obvious in both single and clustered angioblasts (Fig. 6B,C), whereas only traces of Wt1 expression were found in differentiated coronary endothelial linings, suggesting a down-regulation in the expression of the protein as the differentiation of coronary endothelium proceeded (Fig. 6B–D). The endothelium of those vessels with a forming media did not show Wt1 expression. However, perivascular cells of the blood vessel wall displayed a strong Wt1 immunoreactivity (Fig. 6D).
Intramyocardial single Wt1/QH1-positive cells were present in compact layer of the ventricle (but not in the atrium) starting at HH25. These cells were not in direct contact with the developing subepicardial vascular capillary plexus, as shown by the study of confocal serial sections (Fig. 5C).
It is evident that coronary development is strongly influenced by the epicardial origin of the coronary cell lineages from the coelomic epithelial cells of the proepicardium and the epicardium (Pérez-Pomares et al.,2002a, b; Wessels and Pérez-Pomares,2004). In this context, it is worthy to emphasize the special nature of the epicardium compared with other coelomic epithelia. All the developing visceral primordia (lungs, liver, kidneys, gonads, spleen, etc.) are covered ab initio by the coelomic epithelium and its associated (splanchnopleural) mesenchyme. The heart, however, is a direct derivative of the coelomic epithelium itself and, therefore is not originally covered by a mesothelial lining. The proepicardial transfer from the coelomic wall to the myocardial surface (an event common to all the vertebrate models hitherto studied) must be regarded as a developmental innovation to provide splanchnopleural mesoderm to the myocardial layers, which, at these embryonic stages, lack it (Pérez-Pomares and Muñoz-Chápuli,2002). This mesodermal mesenchymal supply is critical to provide the heart with vascular and connective tissue progenitor cells. It is possible that the specific nature of the cardiac (coronary) angioblasts can be attributed to this specific origin, but it is also possible that somehow “hidden” properties of embryonic visceral vasculogenesis become more “visible” in the heart, due to the late stage of differentiation of the cardiac angioblasts. Thus, some of the features of the cardiac angioblasts (for example Wt1 expression) would perhaps be attributable only to the timing of their differentiation rather than to an intrinsic property of these cells.
The proepicardium has been reported to have an obvious cardiac vascular potential in vivo. By using different cell tracing methods combined with in ovo transplantations, it was demonstrated that proepicardial tissue significantly contributes to coronary blood vessel development (Mikawa and Fischman,1992; Pérez-Pomares et al.,1998a; Männer,1999), but these studies could not determine whether the vasculogenic abilities of proepicardial cells strictly depend on a myocardial signal (being, therefore, restricted to heart territories), or whether the proepicardium is also sensitive to triggering signals from other tissues. Finally, we did not know if coronary vasculogenesis (from angioblast differentiation to endothelial tube formation) could also take place in vitro. To clarify all these points, we have analyzed the vascular properties of proepicardial cells from the point of view of their ability to differentiate into angioblasts and their competence to undergo vasculogenesis, both in vivo and in vitro.
In this study, we have demonstrated that the vasculogenic potential of the proepicardium is not cardiac-specific, as the heterotopic transplantation of proepicardia to several developing primordia has shown the contribution of donor-derived tissue to the vascularization of these organs. Vascularization of noncardiac embryonic structures using proepicardial tissue seems to be a “plastic” process in which vascular proepicardial cells adapt to the specific anatomical characteristics of the vascular bed of different organs, from the coherent formation of endothelial tubes in the endoderm and limb buds to the scattered integration in the hepatic sinusoidal lining or developing mesonephric vessels. The fusion of the donor-derived vessels with the host-derived ones indicates that proepicardial angioblasts can participate in vasculogenic processes occurring in quite different mesodermal and endodermal environments and, thus, that they are not special in this regard, taking into account that the mode of vascularization is basically controlled by each organ (Sherer,1991). Actually, some of the results included in this work indicate that the vasculogenic potential of the proepicardium (both in vivo and in vitro) can be greatly enhanced by the preincubation of the tissue with bFGF+VEGF before its transplantation. This finding suggests that the proepicardial tissue, before its attachment to the myocardium, is not directly exposed to bFGF+VEGF and that the profound changes experienced by the proepicardial tissue when it attaches to the outer surface of the heart are related to the high myocardial production of these two cytokines, which are capital to the differentiation, proliferation, and maturation of vertebrate angioblasts in many organs, including the heart (Risau and Flamme,1995; Tomanek et al.,1999, 2001a, b; Poole et al.,2001).
In our opinion, the singularity of coronary angioblasts relates more to the actual speed of coronary morphogenesis, its strict patterning, and the low oxygen tension of the developing myocardium (Tomanek et al.,1999; Ivnitski-Steele et al.,2004) than to the intrinsic vascular potential of these cells. In the case of coronary angioblasts, the speed of coronary morphogenesis and the need of a finely tuned process of vasculogenesis (blood vessels of quite different sizes will have to form following a strict spatiotemporal pattern) might explain the low in vitro proliferation levels found in coronary angioblasts. Actually, the observation that the numbers of coronary angioblasts are increased in VEGF+bFGF–treated cultures without a significant up-regulation of cell division seems to support the hypothesis of a direct and local epicardial differentiation into angioblasts under the requirements of the developing subepicardial coronary plexus (results included in this study, Fig. 2).
As shown for the first time by this study, in vitro cultured proepicardial tissue does retain (at least a part of) its in vivo vascular potential. The differentiation of proepicardial cells into angioblasts can be reproduced in two different in vitro assays (culture on plastic or on collagen gels) which also recapitulate various stages of in vivo proepicardial development, including the formation of epithelial monolayers, EMT, and angioblastic differentiation. Several important events of capillary morphogenesis can be recorded in these in vitro systems, including the vacuolation and formation of ring-like structures in proepicardial cell culture on plastic—a classic feature of active endothelial cells (Folkman and Haudenschild,1980)—and the coalescence of coronary angioblasts into clusters, chords, and occasionally, tubular structures (only in collagen gel cultures). This vascular tubularization event has been reported to be linked to embryonic endodermal induction (Vokes and Krieg,2002), and the process has ultimately been described as hedgehog-dependent (Vokes et al.,2004). Yet, we do not know if hedgehog(s) plays a similar function in the vascularization of the heart, where TGFβs have also been reported to modulate vascular tube formation (Holifield et al.,2004). A possible role of hedgehogs in coronary development and their interaction with other vasculogenic growth factors is currently under study in our laboratory. Finally, we would like to emphasize that our results indicate that the collagen gel assay is an excellent in vitro experimental setting to study and dissect the precise molecular and cellular mechanisms of coronary endothelial differentiation and the assembling of coronary angioblasts into true blood vessels, as well as one of the few available in vitro assays for vasculogenesis.
Coronary vasculogenesis can be studied, as shown in this study, not only in vivo but also in vitro. The availability of defined in vitro assays to study coronary morphogenesis is of extreme importance, because whereas the basic cellular events taking place throughout coronary vascular morphogenesis are relatively well known, not much information about the molecular mechanisms that regulate them is available. Needless to say, in vitro assays have been proved to be methodologically powerful to test the dynamics of molecular mechanisms defining, instructing, and modulating developmental processes. A critical aspect to understand the molecular regulation of coronary development is that of the lineage relation that exists between the diverse coronary cell types (endothelium, smooth muscle, and adventitial fibroblasts). Increasing evidence suggests that both coronary endothelial and SMCs are epicardial derivatives; results from proepicardial quail-to-chick chimeras have been the main support for this hypothesis (Männer,1999; Pérez-Pomares et al.,2002a, b). Nonetheless, the emerging question still is: are these two lineages already separated before the epicardial EMT, or do they derive from single, multipotent progenitors? The interest of this question has been boosted by the report by Yamashita et al. (2000) of a “bipotential” vascular progenitor derived from stem cells and able to differentiate into endothelium and smooth muscle under treatment with VEGF and PDGF-BB, respectively. Thus, hypothetical bipotential progenitors should be able to respond to both growth factors. In this regard, it is important to remark on the colocalization of VEGFR-2 and PDGFRβ in cells of the myocardially attached proepicardium and the developing primitive epicardium, even in cells that form part of the epicardial epithelial lining. Indeed, only clonal studies would be able to decide if these cells are truly bipotential. Previous studies seem to argue against this possibility, although we actually believe this evidence is compatible with the hypothesis of a coronary bipotential progenitor. Mikawa and Fishman (1992), in an elegant experiment, showed that proepicardia tagged with LacZ transfecting retroviruses gave rise to clones of β-Gal–positive coronary cells that were either endothelial or muscular. We think that this experiment can easily be reconciled with the hypothesis of a (pro)epicardial-derived “bipotential” coronary progenitor, taking into account the two-step model proposed elsewhere by our group (Muñoz-Chápuli et al.,2002; Pérez-Pomares et al.,2002a). In our model, the endothelial or muscular fate of the hypothetical “bipotential coronary precursor” is conditioned by the extracellular and cellular environment. In a first step, myocardially derived VEGF would preferentially drive endothelial differentiation of the earliest EPDCs invading the subepicardium. Then, these coronary endothelial cells would induce differentiation and recruitment of subsequent waves of EPDCs through PDGF-BB secretion. Thus, the fate of a cell clone would be ultimately determined by the timing of its differentiation and the cell context. Quite obviously, the “vascular bipotential precursor” concept has an evident interest for researchers in the field of cardiovascular regenerative medicine, as it offers a developmental explanation for smooth muscle cell differentiation from adult endothelial cells (Arciniegas et al.,2000), and might help to elucidate the molecular mechanisms involved in the commitment of mesodermal progenitors into one cell lineage or another. Several studies are being carried out in our lab to clonally analyze the differentiation of the previously described VEGFR-2/PDGFRβ–positive proepicardial/epicardial subpopulation of cells.
The “bipotential” hypothesis for the origin of coronary cell precursors offers a clear rationale that is very helpful to define complementary signaling pathways directly regulating the differentiation of EPDCs into one of the two major coronary cell types (endothelium or smooth muscle). Most interestingly, the results presented in this study seem to indicate that the primary coronary network develops from local clusters of angioblasts (or perhaps “bipotential precursors”), a fact that necessarily confers a great relevance to cell proliferation events. An often neglected feature of coronary development is that these vessels form in a cellular environment with a high expression of RALDH2 (a key enzyme that catalyzes the conversion of retinaldehyde into all-trans retinoic acid) and, thus, with a presumably high level of RA in their milieu. The importance of the role played by RA in vascular morphogenesis has been shown in other embryonic systems comprising a massive muscular differentiation (the developing limb bud; Berggren et al.,2001) or in close contact with the endoderm (e.g., the yolk sac; Bohnsack et al.,2004). In addition to that, the proliferation of endothelial cells (Lai et al.,2003) and mesenchymal cells of organs undergoing an extensive vascularization (e.g., the lungs; Liebeskind et al.,2000) seems to be under the control of RA. Actually, RA has been reported to be directly related to vascular development by modulation of endothelial cell proliferation (Bohnsack and Hirschi,2004). RALDH2 is expressed in the proepicardium, the epicardium, and EPDCs, showing a decreasing gradient from the outer, epithelial layer of the epicardium to the inner mesenchymal layers (i.e., those closer to the myocardium), something that probably reflects a down-regulation of this enzyme in the EPDC population as it differentiates. Colocalization of RALDH2 and RXRα with vascular markers has been reported in this study. As already indicated, this coexpression seems to be restricted to the very early steps of coronary angioblast/endothelial differentiation, suggesting that a possible role of the RA signalling system is to modulate proliferation or perhaps sustain the survival of these cells.
A final comment is deserved by our observations on Wt1 expression in coronary vascular development. This study includes the first report of the expression of Wt1 in the embryonic coronary endothelium, a feature that seems to be unique to these endothelial cells. Such a result could explain the abnormal expression of Wt1 in hypoxic coronary endothelial and smooth muscle cells (Wagner et al.,2003) as a reactivation of the normal molecular pathways involved in the embryonic development of coronary vessels, a vasculogenic event that happens to occur in a poorly oxygenated environment. Linked to this Wt1 expression described for coronary endothelial cells, the presence of QH1/Wt1–positive cells (Wt1-expressing angioblasts) inside the compact ventricular myocardial layers is also reported in our study. These Wt1-positive cells were isolated (i.e., not connected) from subepicardial coronary angioblasts as demonstrated by the study of confocal serial sections and, therefore, cannot be regarded as a part of an angiogenic process. This result suggests that single angioblasts can migrate intramyocardially. The fate of these cells cannot be ascertained precisely, because they could integrate into the endocardium (donor-derived endocardial cells can be observed occasionally in epicardial chimeras, as some results from our own unpublished work indicate), but also could contribute to different portions of the intramyocardial vascular system, including Thebesian vessels (Männer,2000). This explanation would open the possibility of an intramyocardial vasculogenic process, an issue that, as far as we know, had not previously received any experimental support.
A closing remark will allow us to briefly comment on the possible connection between Wt1 expression and RA signaling. In a previous study, we did hypothesize about a Wt1-RALDH2 system controlling the development of several cardiac tissues (Pérez-Pomares et al.,2002b); some recent and unpublished results from Wt1-null mice seem to credit this hypothesis, being the case that Wt1-deficient mice show a reduced RALDH2 expression in the epicardium and the EPDC population (Pérez-Pomares et al., manuscript in preparation). Thus, at least a part of the functions of Wt1 in cardiac development could be related to the maintenance of an active RA signaling system.
The animals used in our research program were handled in compliance with the international guidelines for animal care and welfare. Chick and quail eggs were kept in a rocking incubator at 38°C. Embryos were staged according to the Hamburger and Hamilton (1951) stages of chick development.
Proepicardial Quail/Chick Chimeras
Donor quail embryos were incubated until stages HH16–HH17, excised, and washed in sterile EBSS (Gibco). Proepicardia were carefully dissected using tungsten needles, small iridectomy forceps, and scissors. Proepicardia were then transplanted onto chick embryo hosts incubated for 60 hr-HH16–HH17 (heart and liver grafts); 72 hr-HH18 (hindgut grafts); or 84 hr-HH21 (limb bud grafts and mesonephros). To optimize the donor (quail) epicardial covering of the host chick heart, quail-to-chick proepicardial chimeras were prepared as described by Männer (1999). To hold proepicardial grafts next to liver primordia, small pieces of the eggshell membrane were used. In a different set of experiments, proepicardia were cultured in a hanging drop system (Rudnicki and McBurney,1987) using fresh M199 medium supplemented with 100 ng/ml bFGF (Peprotech), 10 ng/ml VEGF (Peprotech), 1% chick serum (Sigma), and 100 IU penicillin/streptomycin (Gibco) for 12 hr and then transplanted to the pericardial cavity of HH16–HH17 host chick embryos. Proepicardia for control experiments were cultured in the absence of VEGF and bFGF.
Chimeras were fixed in modified Amsterdam's fixative (methanol, acetone, distilled water 2:2:1) or Dent's fix (methanol, dimethyl solfoxide [DMSO] 4:1), dehydrated in a graded series of ethanol, cleared in butanol, carefully oriented, and embedded in Histosec (Merck). Finally, 5-μm serial sections were mounted on microscope slides (Menzel-Gläser).
Immunohistochemical characterization of the chimeric tissues was performed using the QCPN antibody (quail pan-nuclear marker) or the QH1 antibody (quail endothelial and hemopoietic cells) as donor tracers. Double immunostainings were performed as described below.
Culture of Proepicardial Explants on Collagen Gels
Quail proepicardia (HH16–HH17) were cultured on 1.5-mg/ml drained collagen gels (rat tail type I collagen, Collaborative Research) in four-well plates (Nunc). The explants were allowed to attach to the collagen for 3–4 hr at 37°C, 5% CO2. Then, DMEM (Gibco) supplemented with 1% chick serum (Sigma), 10 ng/ml VEGF, 100 ng/ml FGF, 100 IU penicillin/streptomycin (Gibco), and insulin/transferrin/selenium (Gibco) was added. Control cultures were no supplemented with growth factors. In some cases, the explants were cultured in endoderm conditioned medium. This medium was prepared by culturing hindgut pieces from HH29–HH30 chick embryos in DMEM, 1% chick serum (Sigma), and 100 IU penicillin/streptomycin (Gibco). After 2 days of incubation (37°C, 5% CO2), the medium was centrifuged (5 min, 2,000 rpm) and the supernatant was frozen (−20°C) until its use.
All the explants were cultured for 2–3 days at 37°C and 5% CO2 and inspected daily by using an inverted microscope with phase contrast optics (Leica). At the end of the culture period, the explants were fixed in 4% paraformaldehyde in PBS, extensively washed in PBS, blocked in a 1% bovine serum albumin–phosphate buffered saline (BSA-PBS) solution and incubated overnight in a 1:200 dilution of the QH1 antibody diluted in PBS. The cultures were washed at least three times (1 hr each) in PBS and incubated overnight (4°C) in a secondary tetrarhodamine isothiocyanate (TRITC) -conjugated anti-mouse IgG. A fluorescein isothiocyanate (FITC) -conjugated anti-mouse IgG (Sigma) was chosen as a secondary antibody for those cultures selected for cell density studies (after propidium iodide nuclear counterstaining) and for quail angioblastic proliferation assays (after QH1 staining, proepicardial cells were incubated in a rabbit polyclonal anti–phospho-histone H3 followed by an TRITC-conjugated anti-rabbit IgG; Sigma). The samples were analyzed under a TCS-NT laser confocal microscope (Leica).
Analysis of confocal images was performed with Image J software. Confocal images were split in red and green channels. The surface of labeled areas was measured, and an index of total surface labeled in the green channel (i.e., QH1 immunoreactivity) with respect to the sum of the areas labeled in both channels was obtained. Mean indexes obtained for control and growth factor-treated cultures were compared using a Student t-test.
Proepicardial Cell Expansion In Vitro
Proepicardia from HH16–HH17 quail embryos were isolated as described. Two proepicardia per well originally were cultured in four-well plates (Nunc). DMEM (Gibco) supplemented with 10% fetal bovine serum (PAA), 2% chick serum (Sigma), 100 IU penicillin/streptomycin (Gibco), and Plasmocin (Invitrogen) was used as culture media (complete DMEM). After 4 days of incubation (37°C, 5% CO2), the cells were passed to 35-mm Petri dishes, cultured for 4 days, and finally transferred to 100-mm Petri dishes for an additional week. Cells were fixed in a 1:1 methanol:acetone solution, hydrated through a 70%, 50%, 30% methanol series, extensively washed in PBS, and permeabilized in 1% Triton X-100 in PBS. Samples were then blocked in a 1% BSA-PBS solution (overnight, 4°C) and incubated overnight in a 1:200 dilution of the QH1 antibody in PBS. The cultures were washed at least three times (1 hr each) in PBS and incubated again in a secondary TRITC-conjugated anti-mouse IgG (Sigma) at 4°C (overnight). After final washes in PBS, the cell nuclei were counterstained with propidium iodide (Sigma) and the samples were analyzed under a TCS-NT laser confocal microscope (Leica). For the estimation of the average number of QH1-positive cells in these cultures, five microscopic fields (×16) per sample were analyzed.
For single immunoperoxidase, endogenous peroxidase activity was quenched by incubating the sections for 30 min with 3% hydrogen peroxide in Tris-PBS (TPBS). After washing, non-specific binding sites were saturated for 30 min with 16% sheep serum, 1% bovine serum albumin, and 0.5% Triton X-100 in Tris-PBS (SBT). Endogenous biotin was blocked with the avidin–biotin blocking kit (Vector). The slides were incubated overnight at 4°C in the primary antibody (QH1). Then, the slides were washed in TPBS (3 × 5 min), incubated for 1 hr at room temperature in biotin-conjugated goat anti-mouse IgG (Sigma), washed again and incubated for 1 hr in extravidin–peroxidase complex (Sigma). After washing, peroxidase activity was developed with Sigma Fast 3,3′-diaminobenzidine tablets (Sigma) according to the indications of the supplier.
Single QH1 immunofluorescence was performed by blocking non-specific binding sites with SBT, incubating the slides in the primary antibody (overnight at 4°C). Then, the slides were washed in TPBS (3 × 5 min), incubated for 2 hr at room temperature in FITC- or TRITC-conjugated anti-mouse goat IgG (Sigma). After washing in TPBS, an optional nuclear propidium iodide counterstaining was performed for some FITC-labeled samples. Finally, the slides were mounted and analyzed under a TCS-NT laser confocal microscope (Leica).
Two different methods were used for double immunofluorescence: (1) Avian VEGFR-2 or QCPN immunohistochemistry was carried out by using a tyramide amplification system (TSA, Perkin-Elmer) following the indications of the supplier. Briefly, the sections were incubated in undiluted mouse anti–VEGFR-2 or -QCPN hybridoma, and a pre-amplification step was developed with a regular avidin–biotin method. Tyramide amplification followed, being FITC-conjugated Extravidin (Sigma, 1:150 in PBS) used as the fluorochrome. After washing in PBS, the slides were incubated in the RALDH2, CK, or PDGFR-β antibodies (4°C, overnight), washed in PBS, incubated again in a Cy5-conjugated donkey anti-rabbit IgG (Jackson Laboratories) at room temperature 2 hr, and washed in PBS and mounted; (2) A regular avidin–biotin method was used to amplify Wt1, RXRα, and CK signals, and then the second antigen (QH1) was localized using a secondary FITC- or TRITC-conjugated secondary antibody as already described. All the samples were analyzed under a TCS-NT laser confocal microscope (Leica).
For all these protein expression studies, the embryos were routinely fixed in modified Amsterdam's fix (methanol:acetone:water, 2:2:1), with the exception of those selected for VEGFR-2 immunohistochemistry (fixed in Dent's fix -methanol:DMSO, 4:1- at −20°C for 12 hr), the ones used for RXRα staining (fixed in 4% paraformaldehyde, overnight at 4°C), and those used for Wt1 immunohistochemistry (fixed in 1:1 methanol:acetone solution after cryoprotection and sectioning, see Pérez-Pomares et al.,2000b, for details).
The monoclonal antibody anti-QH1 labels all quail endothelial and hemopoietic cells, whereas the QCPN antibody stains all quail nuclei. Both were obtained from the Developmental Studies Hybridoma Bank and were used at a 1:200 dilution and undiluted, respectively. The monoclonal anti-smooth muscle cell α-actin (clone 1A4, Sigma) was used at a 1:100 dilution. The polyclonal anti-bovine epidermal cytokeratin (Z622, Dako) was used at a 1:100 dilution. The affinity-purified anti-WT1 polyclonal antibody (sc-192, Santa Cruz) was used at a 1:50 dilution. The polyclonal anti-RXRα (sc-553, Santa Cruz) was used at a 1:80 dilution. The polyclonal anti–phospho-histone H3 (#06-570, Upstate) was used at a 1:150 dilution. Secondary TRITC, FITC, and Cy5-conjugated antibodies (from Sigma and Jackson Laboratories, respectively) were diluted 1:100 in TPBS. Secondary biotin-conjugated antibodies (from Sigma) were diluted 1:100 in TPBS. FITC- and peroxidase-conjugated Extravidins were diluted 1:150 in TPBS.
The QH1 and QCPN monoclonal antibodies were obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Pharmacology and Molecular Sciences, John Hopkins University School of Medicine, Baltimore, MD 21205, and the Department of Biological Sciences, University of Iowa, Iowa City, IA 52242, under contract NO1-HD-2-3144 from the NICHD. The 5H6 (avian anti–VEGFR-2) monoclonal and the anti-RALDH2 polyclonal were kind gifts from Dr. A. Eichmann (Institute de Biologie Paris-CNRS, France) and Drs. Peter McCaffery and Ursula Dräger (Eunice Kennedy Shriver Center-UMASS, USA), respectively. This work is supported by the European Union's Framework Programme 6, LSHH-CT-2005-018630, the Ministerio de Sanidad y Consumo/ISCIII, and the Ministerio de Ciencia y Tecnología (to both J.M.P.-P. and R.M.Ch.) and NATO (to J.M.P.-P.).
- 2000. Characterization of non muscle cells present in the intima of normal adult bovine pulmonary artery. Anat Rec 258: 262–268. , , .
- 1999. Differential distribution of retinoic acid synthesis in the chicken embryo as determined by immunolocalization of the retinoic acid synthetic enzyme, RALDH-2. Dev Biol 210: 288–304. , , , .
- 2001. Expression and regulation of the retinoic acid synthetic enzyme RALDH-2 in the embryonic chicken wing. Dev Dyn 222: 1–16. , , , .
- 2004. Red light, green light. Signals that control endothelial cell proliferation during embryonic vascular development. Cell Cycle 3: 1506–1511. , .
- 2004. Signaling hierarchy downstream of retinoic acid that independently regulates vascular remodeling and endothelial cell proliferation. Genes Dev 18: 1345–1358. , , , .
- 2001. Localization of the Wilms' tumour protein Wt1 in avian embryos. Cell Tissue Res 303: 173–186. , , , .
- 1984. On the development of the coronary arteries in human embryos, stages 14–19. Anat Embryol (Berl) 169: 209–218. , .
- 1998. Common epicardial origin of coronary vascular smooth muscle, perivascular fibroblasts, and intramyocardial fibroblasts in the avian heart. Dev Biol 193: 169–181. , , , .
- 2002. Vasculogenesis and the search for the hemangioblast. J Hematother Stem Cell Res 11: 207–214. , , , .
- 2003. Fundamental concepts of the angiogenic process. Curr Mol Med 3: 643–651. .
- 1980. Angiogenesis in vitro. Nature 288: 551–556. , .
- 1998. Epicardium-derived cells contribute a novel population to the myocardial wall and the atrioventricular cushions. Circ Res 82: 1043–1052. , , , , .
- 1951. A series of normal stages in the development of the chick embryo. J Morphol 88: 49–92. , .
- 1983. Development of the cardiac blood vessels in staged human embryos. Acta Anat 115: 220–230. .
- 2004. TGF-β1, β2 and β3 cooperate to facilitate tubulogenesis in the explanted quail heart. J Vasc Res 41: 491–498. , , , .
- 1988. Development of the coronary arteries in the embryonic human heart. Circulation 77: 1250–1257. , , .
- 2004. 2,3,7,8-tetrachlorodibenzo-p-dioxin reduces myocardial hypoxia and vascular endothelial growth factor expression during chick embryo development. Birth Defects Res 70: 51–58. , , .
- 2004. Formation and remodeling of the coronary vascular bed in the embryonic avian heart. Dev Dyn 230: 34–43. , , .
- 2003. Retinoic acid regulates endothelial cell proliferation during vasculogenesis. Development 130: 6465–6474. , , , .
- 1999. A role for serum response factor in coronary smooth muscle differentiation from proepicardial cells. Development 126: 2053–2062. , , , , , .
- 2000. Retinoic acid stimulates immature lung fibroblast growth via a PDGF-mediated autocrine mechanism. Am J Physiol Lung Cell Mol Physiol 27: L81–L90. , , , .
- 2001. Coronary smooth muscle differentiation from proepicardial cells requires rhoA-mediated actin reorganization and p160 rho-kinase activity. Dev Biol 240: 404–418. , , , , , , , , .
- 1999. Does the subepicardial mesenchyme contribute myocardioblasts to the chick embryo heart? A quail-chick chimera study tracing the fate of the epicardial primordium. Anat Rec 255: 212–226. .
- 2000. Embryology of congenital ventriculo-coronary communications: a study on quail-chick chimeras. Cardiol Young 10: 233–238. .
- 1996. Epithelial-mesenchymal transformations in early avian heart development. Acta Anat 156: 173–186. , , , , .
- 1992. Retroviral analysis of cardiac morphogenesis: discontinuous formation of coronary vessels. Proc Natl Acad Sci U S A 89: 9504–9508. , .
- 1996. Pericardial mesoderm generates a population of coronary smooth muscle cells migrating into the heart along with ingrowth of the epicardial organ. Dev Biol 174: 221–232. , .
- 1999. YAC complementation shows a requirement for Wt1 in the development of epicardium, adrenal gland and throughout nephrogenesis. Development 126: 1845–1857. , , , , .
- 2002. Cellular precursors of the coronary arteries. Texas Heart Inst J 29: 243–249. , , , , , .
- 2002. Epithelial-mesenchymal transitions: a mesodermal cell strategy for evolutive innovation of Metazoan. Anat Rec 268: 343–351. , .
- 1997. Contribution of the primitive epicardium to the subepicardial mesenchyme in hamster and chick embryos. Dev Dyn 210: 96–105. , , , .
- 1998a. The origin of the subepicardial mesenchyme in the avian embryo: an immunohistochemical and quail-chimera study. Dev Biol 200: 57–68. , , , .
- 1998b. Immunolocalization of the vascular endothelial growth factor receptor-2 in the subepicardial mesenchyme of hamster embryos: identification of the coronary vessel precursors. Histochem J 30: 627–634. , , , .
- 2002a. Origin of coronary endothelial cells from epicardial mesothelium in avian embryos. Int J Dev Biol 46: 1005–1013. , , , , , .
- 2002b. Experimental studies on the spatiotemporal expression of Wt1 and RALDH2 in the embryonic avian heart: a model for the regulation of myocardial and valvuloseptal development by epicardially-derived cells (EPDCs). Dev Biol 247: 307–326. , , , , , , .
- 1993. Development of the cardiac coronary vascular endothelium, studied with antiendothelial antibodies, in chicken-quail chimeras. Circ Res 73: 559–568. , , , , .
- 2001. The role of FGF and VEGF in angioblast induction and migration during vascular development. Dev Dyn 220: 1–17. , , .
- 1995. Vasculogenesis. Annu Rev Cell Dev Biol 11: 73–91. , .
- 1987. Cell culture methods and introduction of differentiation of embryonal carcinoma cell lines. In: RobertsonEJ, editor. Teratocarcinomas and embryonic stem cells: a practical approach. Oxford: IRL Press. p 19–49. , .
- 1971. Mechanism of the development of coronary arteries in chick embryo. Folia Morphol (Praha) 19: 113–124. , .
- 1991. Vasculogenic mechanisms and epithelio-mesenchymal specificity in endodermal organs. In: FeinbergRN, ShererGK, AuerbachR, editors. The development of the vascular system. Basel: Karger. p 37–59. .
- 2000. FOG-2, a cofactor for GATA transcription factors, is essential for heart morphogenesis and development of the coronary vessels from epicardium. Cell 101: 729–739. , , , , , , , .
- 1999. Vascular endothelial growth factor expression coincides with coronary vasculogenesis and angiogenesis. Dev Dyn 215: 54–61. , , , , .
- 2001a. Vascular endothelial growth factor and basic fibroblast growth factor differentially modulate early postnatal coronary angiogenesis. Circ Res 88: 1135–1141. , , , , , .
- 2001b. Multiple growth factors regulate coronary embryonic vasculogenesis. Dev Dyn 221: 265–273. , , , , , .
- 2002. Role of VEGF family members and receptors in coronary vessel formation. Dev Dyn 225: 233–240. , , , , .
- 1981. The origin of the epicardium and the embryonic myocardial circulation in the mouse. Anat Rec 201: 157–168. , .
- 1993. Early development of quail heart epicardium and associated vascular and glandular structures. Anat Embryol (Berl) 188: 381–393. , , , .
- 2002. Endoderm is required for vascular endothelial tube formation, but not for angioblast specification. Development 129: 775–785. , .
- 2004. Hedgehog signaling is essential for endothelial tube formation during vasculogenesis. Development 131: 4371–4380. , , , , , , .
- 1997. The development of the coronary vessels and their differentiation into arteries and veins in the embryonic quail heart. Dev Dyn 208: 338–348. , , , , , .
- 1999. Smooth muscle cells and fibroblasts of the coronary arteries derive from epithelial-mesenchymal transformation of the epicardium. Anat Embryol (Berl) 199: 367–378. , , , .
- 2001. Bves: prototype of a new class of cell adhesion molecules expressed during coronary artery development. Development 128: 2085–2093. , , .
- 2003. Epicardial/mesothelial cell line retains vasculogenic potential of embryonic epicardium. Circ Res 92: 525–231. , , , , .
- 2003. Oxygen-regulated expression of the Wilms' tumor suppressor Wt1 involves hypoxia-inducible factor-1 (HIF-1). FASEB J 17: 1364–1366. , , , , , , .
- 1994. The expression of retinoid X receptor genes is regulated by all-trans- and 9-cis-retinoic acid in F9 teratocarcinoma cells. Exp Cell Res 210: 56–61. , , .
- 2004. The epicardium and epicardially derived cells (EPDCs) as cardiac stem cells. Anat Rec 276: 43–57. , .
- 2000. Flk1-positive cells derived from embryonic stem cells serve as vascular progenitors. Nature 408: 92–96. , , , , , , , , .
- 2001. Effects of VEGF (165) and VEGF (121) on vasculogenesis and angiogenesis in cultured embryonic quail hearts. Am J Physiol Heart Circ Physiol 280: H2240–H2247. , .