Embryonic cardiomyocyte expression of endothelial genes

Abstract

Vertebrate precardiac mesoderm contains cells destined to become cardiomyocyte or endothelial cells. To determine the stability of these phenotypes freshly isolated embryonic day (E) 2.5–E6 chicken hearts were immunostained for myosin heavy chain (MyHC) to identify cardiomyocytes, and von Willebrand factor (vWF) and Flk-1 to identify endothelial cells. At E2.5–E3, 90% of cells express only MyHC and 6% express only vWF/Flk-1. However, 2% MyHC⁺ cells in E2.5–E3 hearts and 0.3% in E4–E6 hearts, also express vWF/Flk-1; and when cultured 3 days, >40% of the MyHC⁺ cells express vWF/Flk-1, but they do not express Vezf1, vascular endothelial cadherin, or Tie2. Thus, only a subset of endothelial genes are induced in cultured cardiomyocytes. While the subsequent developmental fate of embryonic heart cells exhibiting a vWF⁺/MyHC⁺ phenotype is unknown, analysis of this phenotype may provide information pertinent to mechanisms of cell phenotype stability, cellular transdifferentiation, and induction of stable cell types from embryonic stem cells. Developmental Dynamics 236:2512–2522, 2007. © 2007 Wiley-Liss, Inc.

INTRODUCTION

Formation of the tubular heart in vertebrates results from the fusion of bilateral precardiac mesenchyme migrating from the rostral half of the primitive streak on either side of Hensen's node (Rosenquist and DeHaan,1966; Garcia-Martinez and Schoenwolf,1993; Badorff et al.,2003). The early heart tube consists of two mesodermal epithelial layers: the endocardium and the myocardium (Manasek,1968). Studies in avians have shown that both cell types originate from the cardiac mesoderm and become distinct cell populations by Hamburger–Hamilton (HH) stages 5–8 (Linask and Lash,1993). Previous studies demonstrate that endothelial cells differentiate to mesenchymal cells early in chick heart development to form the cardiac cushions (Markwald et al.,1977,1996; Eisenberg and Markwald,1995) and later contribute to thickening of the intimal wall of the aorta (Arciniegas et al.,2000,2003a). Data from cell lineage studies suggest that cells in the heart field are committed to cardiomyocyte or endothelial cell fates before they migrate to form the tubular heart (Cohen-Gould and Mikawa,1996), perhaps as early as the primitive streak at HH stage 3 (Wei and Mikawa,2000).

The emergence of the endothelium is not limited to the developing heart. Several endothelial markers have been useful in identifying endothelial cells in developing vasculature. von Willebrand Factor (vWF) is synthesized by endothelial cells throughout the body (Jaffe et al.,1974) and by megakaryocytes (Nachman et al.,1977). As early as 12 hr after plating, cell cultures from 16-hr chick blastodiscs contain cells that show immunoreactivity with vWF-specific antibody and take up acetylated low-density lipoprotein (Yablonka-Reuveni,1989). vWF expression has also been detected in the aortic arch in embryonic day (E) 3–E21 chick embryos (Yablonka-Reuveni,1989; Arciniegas et al.,2000), as well as in the endothelium of blood and lymphatic vessels of E8 and E14 chick embryo chorioallantoic membranes (CAM; Ribatti et al.,1999). By in situ hybridization, transforming growth factor-β3 (TGFβ3) mRNA is first detectable in the precardiac mesenchyme, and then later in heart, smooth muscle progenitor cells of major arteries, myotome, visceral and parietal peritoneum, and mesenchymal cells in the limb of the chick embryo (Yamagishi et al.,1999). Tie2, the Type II TGFβ receptor, has been detected by immunostaining in 6- and 13-somite stage embryos in a reticular pattern corresponding to the vascular network, as well as within endothelial cells of the feather bud, limb, CAM, and vitelline membrane of E10 chick embryos (Brown et al.,1999). Finally, the vascular endothelial growth factor (VEGF) receptor (Flk-1) has been detected at low levels in unincubated quail embryo blastodics, and at higher levels in mesodermal tissue immediately after gastrulation at E1, but it is restricted to endothelial cells from E2 onward (Flamme et al.,1995).

Evidence indicating that endothelial cells and cardiomyocytes may also share a common lineage is based on gene expression experiments in which cells from the heart field express both cardiomyocyte and endothelial cell markers. For example, experiments with explant cultures of quail precardiac mesoderm identified small numbers of cells expressing both QH-1, an endothelial cell marker, and N-cadherin, a cardiomyocyte cell marker after 1 day in vitro (Linask and Lash,1993). In addition, the QCE-6 cell line, derived from stage 4 quail cardiomyogenic mesoderm, can be induced to form cells that express either endothelial cell or cardiomyocyte markers (Eisenberg and Bader,1995). In the presence of retinoic acid, basic fibroblast growth factor, TGFβ-2, and -3, approximately half of the QCE-6 cells react with the endothelial-specific monoclonal antibody (mAb), QH-1, and half express cardiomyocyte markers. Of interest, in vitro studies of quail endocardial endothelium formation reveal a small population of mesodermal cells that react with antibodies against both QH-1 and the cardiomyocyte marker sarcomeric myosin (Sugi and Markwald,1996). More recently, it was shown that Mesp-1, a basic helix–loop–helix transcription factor, is expressed in the myocardial precursors using a cross between a knockin mouse expressing Cre under the control of the Mesp-1 promoter–enhancer and a β-galactosidase transgenic reporter mouse (Saga et al.,1999). A subsequent examination of these mice revealed expression of Mesp-1 in the endocardium as well, indicating that both myocardium and endocardium were derived from Mesp-1–expressing cells (Saga et al.,2000). Similarly, lineage tracing studies in a knock-in mouse line demonstrate that Flk-1⁺ progenitor cells also contribute to cardiac muscle (Motoike et al.,2003). Brachyury⁺, Flk-1⁺ progenitor cells with cardiac muscle potential have also been isolated from mouse embryonic stem (ES) cells and head-fold stage mouse embryos (Kattman et al.,2006). Taken together with data presented in our studies, these experiments suggest that precardiac mesodermal cells have the potential to become either cardiomyocytes or endothelial cells and that some cells may either activate both phenotypes simultaneously or activate the cardiomyocyte phenotype and then activate one or more genes associated with the endothelial phenotype. The subsequent developmental fate of such cells is unknown.

RESULTS

Rare Cells in Early Chick Hearts Express Cardiomyocyte and Endothelial Cell Markers

To localize cells expressing cardiomyocyte and endothelial cell phenotypes in the developing heart, we examined frozen sections from E3 chick hearts that were immunostained to detect sarcomeric myosin, Flk-1, and vWF (Fig. 1). Flk-1 and vWF are expressed in cells localized to the endothelial lining of the endocardium of the heart (red staining in Fig. 1) and Flk-1 is also is expressed within epicardial cells (red staining in Fig. 1A,B). Sarcomeric myosin is expressed throughout the myocardium and not in the endocardium or epicardium (green staining in Fig. 1B,C,E,F). To verify that the vWF polyclonal antibody (pAb) stains chick embryo endothelial cells, the antiserum was tested on E14 coronary vessels and shown to immunostain cells located in the luminal endothelial layer (Fig. 1G). Of interest, in E3 hearts we observed numerous areas in 8 μm sections, which should generally encompass only single cell layers, where myosin and the endothelial markers appear to overlap (arrows in Fig. 1C,F), suggesting that some heart cells may express both cardiomyocyte- and endothelial-specific proteins. However, this technique does not definitively prove that regions immunostaining for both vWF and MyHC contained individual heart cells expressing both phenotypic markers.

Figure 1. Immunolocalization of endothelial cells in embryonic chick hearts. A–P: Frozen sections of embryonic day (E) 3 embryonic hearts (A–F and H–P) and E14 coronary artery (G) stained with polyclonal antibodies against Flk-1 (A–C) or von Willebrand Factor (vWF) (D–G and K–M). Cells reacting with these antibodies are red and occur in the endocardium (En) and epicardium (Ep) in C. Sections from E3 hearts are also shown in phase contrast (H–J), and after staining with DAPI (blue) to mark nuclei (A,D) and with the monoclonal antibody MF20 (green) to identify cells containing sarcomeric myosin heavy chain (MyHC) (B, C, E, F, and N–P), and to mark the myocardium (My) in C. The arrows in C, F, O, and P (high magnification views of the boxed inserts of A, B, and N, respectively) indicate areas in the myocardium where both striated muscle and endothelial cell marker expression overlaps. The white circles in L, M, O, and P indicate elongated vesicles known as Weibel-Palade bodies that are observed in both vWF⁺/MyHC⁻ and vWF⁺/MyHC⁺ areas. G: In later chick stages, the vWF (red) expressing endothelial cells are readily identifiable in regions lining smooth muscle (SM) actin (green) expressing vessels. The asterisk indicates the lumen of the vessel.

The distinct punctate and granular staining of vWF in the chick heart (Fig. 1F,G) is similar to what has been previously observed in immuonostaining for vWF in chick endothelial cells (Arciniegas et al.,2000,2003b) and bovine (Gospodarowicz et al.,1976), porcine (Booyse et al.,1977), and human umbilical vein endothelial cells (Booyse et al.,1981). The punctate components can be attributed to the storage of vWF in membrane-bound, elongated vesicles known as Weibel-Palade bodies (Wagner et al.,1982; Michaux et al.,2006). In addition to vWF, these rod-like endothelial granules also store P-selectin and other vascular modulator proteins (Lowenstein et al.,2005). Upon closer examination of E3 heart sections at higher magnification (Fig. 1L,M–P), we observed vWF⁺ rod-like subcellular structures in both MyHC⁻ and MyHC⁺ areas (white circles in Fig. 1L,M–P). Furthermore, the location of the vWF⁺/MyHC⁺ areas tended to be subendocardial (Fig. 1O, arrows).

To determine unambiguously whether individual MyHC⁺ cells from the embryonic heart also express an endothelial-specific marker, cells from trypsinized E3 hearts were spread onto glass slides, fixed immediately, and then immunostained for vWF with a pAb and with a sarcomeric myosin mAb (Fig. 2A–H). Greater than 90% of the cells express sarcomeric myosin and not vWF (Fig. 2B–D arrow), and approximately 6% of the cells express the endothelial cell marker (Fig. 2D arrowhead) and not myosin. However, among freshly isolated cells, a subpopulation of approximately 3% express both sarcomeric myosin and vWF (Fig. 2D, double-headed arrow). Collecting the freshly isolated cells onto glass slides using a cytospin protocol provided an additional technique for observing dual phenotypic heart cells. Although this protocol tended to flatten the nuclei (Fig. 2E), individual cells containing both myosin and vWF were detected (Fig. 2E–H). No immunostaining of Flk-1 in freshly isolated cells was observed, presumably due to destruction of these cell surface epitopes during the trypsin treatment. Notably, treatment of E3 hearts with trypsin caused disruption of the subcellular structural organization of myosin in myofibrils (Fig. 2B,F) as well as of vWF in Weibel-Palade bodies (Fig. 2C,G). This phenomenon has previously been observed in freshly isolated rat neonatal cardiomyoctyes where a marked disorganization of cellular structures is detected in the first 24 hr after trypsinization, followed by formation of new structures and a return of rhythmic beating by 48 hr (Perissel et al.,1980). Similar observations have been reported for trypinsinized cardiomyocytes obtained from embryonic chick hearts (Lin et al.,1989).

Figure 2. Early cardiac cells express both endothelial and cardiomyocyte phenotypes. A–D: Embryonic day (E) 3 chick hearts were trypsinized and the dissociated cells were smeared immediately on glass slides. E–H: Alternatively, the dissociated cells were collected onto slides using a cytospin. The cells were then fixed and reacted with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; A,E), sarcomeric myosin monoclonal antibody (mAb) (B,F), and von Willebrand Factor (vWF) polyclonal antibody (pAb) (C,G). The merged images of all three stainings are shown in D and H. Greater than 90% of the cells in E3 hearts exhibit only myosin heavy chain (MyHC) staining (arrow) and approximately 6% exhibit only vWF staining (arrowhead) but small numbers of cells exhibit both MyHC and vWF (double-headed arrow). I: Developmental changes in the percentage dual-phenotype cells are shown. Cells from embryos of various ages were dissociated, smeared on fibronectin coated glass slides, fixed and immunostained, and more than 1,000 cells were analyzed at each stage. Hearts from 12–24 embryos were examined at each age; bars indicate means and one standard deviation of vWF⁺ immunostaining within sarcomeric MyHC⁺ cells. E3 hearts (diamonds) contain the greatest percentage of dual-phenotype cells and the 82% increase from E2.5 hearts (crosshatched) is significant (P < 0.05). In older embryos, the dual-phenotype cells decrease significantly compared with E3 (P < 0.005, boxed) and E2.5 (P < 0.05, italicized).

Percentage of vWF⁺/MyHC⁺ Cells Changes During Heart Development

To determine whether the percentage of vWF⁺/MyHC⁺ cells changes during heart development, freshly isolated and immunostained cells from E2.5–E6 (HH 17–29) hearts were counted in preparations similar to those illustrated in Figure 2A–D. Analysis of more than 1,000 cells at each stage is shown in Figure 2I. At E2.5, 1.5% of the total myosin-expressing cells also expressed vWF. This percentage increases to 2.7% by E3, and then precipitously drops to levels of 0.5, 0.2, and 0.09% by E4, E5, and E6, respectively.

Percentage of Embryonic Heart Cells that Coexpress Cardiomyocyte and Endothelial Cell Markers Increases in Culture

To further investigate the behavior of dual-phenotype cells, freshly isolated E3 heart cells were placed in cell culture and the percentage of MyHC⁺ cells that express vWF was determined at various time points after plating (Fig. 3A). The percentage of E3 MyHC⁺ cells that also express vWF increased from 2.7% at the time of plating (Fig. 2I) to over 40% at day 3, before decreasing to approximately 30% at days 4 and 5. Either the culture environment or removal from an in vivo repression mechanisms, thus, induce a substantial increase in the population of cultured heart cells containing both endothelial and cardiomyocyte markers.

Figure 3. The percentage of dual-phenotype cells increases during cell culture. Uncultured and cultured cells were fixed and immunostained for von Willebrand Factor (vWF) and myosin. A: Embryonic day (E) 3 (Hamburger–Hamilton [HH] stages 18 to 20) cardiac cells were cultured for 1–5 days and scored for their dual expression of vWF and myosin heavy chain (MyHC). Cells cultured for 3 days show the highest expression of the dual-phenotype. B: The percentage of culture-induced dual-phenotype cells increases dramatically in cells derived from E2.5–E6 hearts when cultured for 3 days. Embryonic heart cells from E2.5 to E6 were isolated and cultured for 3 days and scored for vWF-positive cells as a percentage of myosin-positive cells (open bars). The fold increases in the percentage of dual-phenotype cells after 3 days of culture relative to the intact heart (Day 0) is represented by the black bars. More than 500 cells were analyzed for each data point.

To determine whether the percentage of heart cells that exhibit both cardiac and endothelial proteins in culture changes in relation to their initial embryonic age, heart cells from E2.5 (HH stage 17), E3 (HH19), E3.5 (HH21), and E6 (HH29) were isolated and cultured for 3 days (Fig. 3B). E3 cultures exhibit the highest percentage of MyHC⁺ cells that coexpress vWF at the time of isolation. While there appears to be a slight downward trend in the percentage of culture-induced vWF⁺/MyHC⁺cells at increasing ages, statistical analysis did not substantiate differences between any of these developmental stages (Fig. 3B, white bars). However, the proportional increase in such cells is considerably greater at increasing age; that is, E3-derived cells exhibit a 15-fold increase, whereas E6-derived cells exhibit a 370-fold increase in vWF⁺/MyHC⁺ cells (Fig. 4B, black bars). The large increases in such cells after only 3 days in vitro strongly suggests that they do not result from induction of MyHC expression within the relatively small initial populations of endothelial cells. Additionally, because equivalent percentages of endothelial, cardiomyocyte, and vWF⁺/MyHC⁺cells replicate in culture, as assessed by bromodeoxyuridine labeling (data not shown), their increase is not due to selective expansion of the initial vWF⁺/MyHC⁺ populations in E2.5–E6 hearts Rather, these cells must arise from the induction of vWF gene expression in more than 30% of the cardiomyocytes.

Figure 4. Embryonic heart cells exhibit endothelial and smooth muscle or cardiac muscle phenotypes in culture. A–D: Early cardiac cells express both endothelial and cardiomyocyte phenotypes. A–C: Dissociated cells from embryonic hearts were cultured for 5 days, fixed and reacted with DAPI (A), sarcomeric myosin monoclonal antibody (B), and von Willebrand Factor (vWF) polyclonal antibody (C). D: The merged images of all three stainings are shown. B,D: After several days in vitro, some cardiac cells expressing vWF and myosin become fibrous and eventually assemble striated myofibrils. E–H: Cells derived from the dorsal aorta of an embryonic day (E) 14 chick coexpress endothelial and smooth muscle cell markers in culture. E–G: Explanted cells of the dorsal aorta (E14) were cultured for 5 days, fixed and stained with DAPI (E), antibody specific for smooth muscle actin (SMA; F), and vWF (G). H: Cells expressing both SMA and vWF are shown by the arrows in the merged image. The arrowheads indicate endothelial cells expressing vWF alone.

Subsequent culturing of total heart cell preparations for 5 days revealed that vWF⁺/MyHC⁺ cells form filamentous myosin and in some cases, these cells form well-striated myofibrils (Fig. 4B,D). The immunolocalization of vWF in cultured heart cells is similar to the punctate pattern observed in bona fide endothelial cells from dorsal aorta cultures (Fig. 4G,H). However, the distinct cross-striated myofibrils within the vWF⁺/MyHC⁺ cells detected in embryonic heart cultures clearly distinguishes these cells from vascular cells expressing both smooth muscle and endothelial cell proteins. This is demonstrated by comparison to cultures from E14 aortas which are known to contain mixtures of smooth muscle cells, identified by immunostaining for α-smooth muscle actin (Fig. 4F), endothelial cells (Fig. 4G), and as previously demonstrated (Arciniegas et al.,2000), some cells expressing both smooth muscle and endothelial proteins (Fig. 4H).

Embryonic Heart Cells Coexpressing Flk-1 and MyHC Increase During Culture

To determine whether cultured embryonic chick cardiomyocytes express other endothelial cell markers, cultures were immunostained for Flk-1. Flk-1 was of interest because in situ hybridization studies with Flk-1 cRNA by other investigators had indicated high levels of Flk-1 expression in the endothelial cells of myocardial and epicardial blood vessels of E10 chick hearts (Sugishita et al.,2000), as well as a much lower hybridization signal in ventricular cardiomyocytes. This result suggested that some chick embryonic cardiomyocytes express Flk-1 mRNA in vivo. To determine whether MyHC⁺ cells express Flk-1 protein, cultured E3 heart cells were immunostained for Flk-1 and sarcomeric myosin (Fig. 5). Forty-eight-hour cultures contained cells that express both Flk-1 and filamentous myosin (Fig. 5BαD). To determine how the percentage of heart cells expressing both markers changes over time in culture, the number of cells coexpressing MyHC and Flk-1 was determined after 1–5 days in vitro (Fig. 5E). After 1 day, the percentage of E3-derived Flk-1⁺/MyHC⁺ cells is 21%. The Flk-1⁺/MyHC⁺ population then increases to over 50% by day 2 and remains at that level before decreasing to 35% by day 5. The increase and subsequent decrease in Flk-1⁺/MyHC⁺ cells over time in culture is similar to the pattern observed for vWF⁺/MyHC⁺ cells. These results show that cultured embryonic MyHC⁺ heart cells can express at least two endothelial cell markers, vWF and Flk-1.

Figure 5. Flk-1 and myosin heavy chain (MyHC) are coexpressed in cultured embryonic heart cells. A–E: Isolated embryonic heart cells were cultured for 1 day (A–D) as well as for longer times (E). A–C: Fixed cultures were stained with DAPI (A), Flk-1 polyclonal antibody (B), and sarcomeric MyHC monoclonal antibody (C). D: The merged image of all three stainings is shown. After 1 day in culture, cells could be identified that contain only MyHC (arrowhead), or only Flk-1 (arrow), and approximately 21% that exhibited both MyHC and Flk-1 (double-headed arrow). E: When embryonic day (E) 3 cardiac cells were cultured for 1–5 days and scored for Flk-1 and MyHC expression, the percent of cells exhibiting the dual-phenotype rises to over 50% by day 2. More than 1,000 cells were analyzed for each data point in E. Data are reported as means ± one standard deviation.

von Willebrand Factor but not Vascular Endothelial Zinc Finger-1, VE Cadherin, or Tie2 Gene Expression Is Activated in Embryonic Heart Cultures

The large increase in MyHC⁺ cells containing vWF and Flk-1 during cell culture (Figs. 3, 5) suggested that some embryonic cardiomyocytes might initiate a global endothelial cell gene expression program. If so, endothelial cell transcription factors should also be expressed in the dual-phenotype cells. This possibility was investigated by identifying a chicken expressed sequence tag (EST) clone corresponding to mouse vascular endothelial zinc finger-1 (Vezf1) transcription factor. Vezf1 expression is known to be restricted to endothelial cells and their precursors during mouse embryogenesis (Xiong et al.,1999; Aitsebaomo et al.,2001). Polymerase chain reaction (PCR) primer sets were designed to detect chick Vezf1 and vWF transcripts and changes in these mRNA levels were normalized to the levels of chick hypoxanthine-guanine phosphoribosyltransferase (HPRT) mRNA. Quantitative reverse transcriptase (RT) -PCR of chick heart samples at day 0 (fresh hearts), and after 1 day of culture reveal a 27-fold increase in vWF gene expression; however Vezf1 transcript levels remained the same. Similarly, we found no increase in transcript levels of two other endothelial cell-specific markers, vascular endothelial cadherin (VECAD) or Tie2, an endothelial cell-specific receptor tyrosine kinase.

To determine whether the 27-fold increase in vWF mRNA detected in E3-derived heart cultures was due to activation of vWF expression in cardiomyocytes, as opposed to increased vWF gene expression in endothelial cells, we performed in situ hybridization with vWF-specific riboprobes on MyHC immunostained heart cultures (Fig. 6). In situ hybidization with a vWF antisense probe revealed high signals in both MyHC⁺ and MyHC⁻ cells (Fig. 6D), presumably due to vWF gene expression in cardiomyocytes as well as endothelial cells; whereas the sense probe exhibited uniformly low background staining of all cells (Fig. 6F). Quantitation of the vWF mRNA⁺/MyHC⁺ cells in 1-day cultures indicated that 22.6 % ± 2.2 of the MyHC⁺ cells contained vWF transcripts, and in 2-day cultures 34.4% ± 1.0 of MyHC⁺ cells contained vWF mRNA. These results are consistent with the hypothesis that the increase in cultured MyHC cells containing vWF protein, is due to a concurrent induction of vWF message in the same cells. However, this increase does not result from increased expression of an endothelial transcription factor, Vezf1.

Figure 6. von Willebrand Factor (vWF) mRNA is expressed in cultured myosin heavy chain–positive (MyHC⁺) heart cells. A–H: In situ hybridization of 3-day-old embryonic heart cultures with Alexa 488–labeled vWF antisense (A–D) and sense (E–H) riboprobes was followed by DAPI staining (A,E), immunostaining with anti–Alexa 488 polyclonal antibody (B,F), and anti–sarcomeric myosin monoclonal antibody (C,G). D,H: The merged images of all three stainings are shown. The circle and arrowheads point to a MyHC⁻ cell with a strong vWF hybridization signal and to several MyHC⁺ cells with no detectable vWF signal, respectively. The arrows indicate cells positive for vWF transcript and MyHC protein. F,H: Hybridization with sense probe shows low nonspecific binding.

DISCUSSION

The studies reported above indicate that embryonic chick hearts contain a small population of cardiac cells that express both cardiomyocyte genes and a subset of endothelial genes: vWF, a protein usually restricted to endothelial linings of the endocardium and coronary vessels (Fig. 1E,G), and Flk-1 (Fig. 5), a marker of early endothelial cell differentiation (Yamaguchi et al.,1993). The in vivo percentage of cells expressing both myogenic and endothelial genes during heart development is as high as 3% in E3 (HH stage 19), declining to 0.09% by E6 (Fig. 2I). However, when embryonic chick heart cells are cultured, the percentage of vWF⁺/MyHC⁺ and Flk-1⁺/MyHC⁺ cells increases 15-fold or more, depending on the initial embryonic age. Although it has not been feasible to coimmunostain the same cells for MyHC and both endothelial markers, it seems likely that vWF⁺/MyHC⁺ cells are also Flk-1⁺, because virtually the same percentage of cultured MyHC⁺ cells exhibit each endothelial marker (Figs. 2A, 5E). Taken together, these studies, thus, suggest that Flk-1⁺/vWF⁺/MyHC⁺ cells are due to the de novo expression of at least two endothelial genes in 40–50% of the cardiomyocytes.

No evidence in our study, such as the occurrence of binucleated cells, suggests that cell fusion accounts for the occurrence of Flk-1⁺/vWF⁺/MyHC⁺ cells. Although spontaneous fusion between cardiomyocytes and endothelial cells has been previously demonstrated in cardiomyocyte-endothelial cell cocultures, resulting in cells containing markers from both cell types (Matsuura et al.,2004; Koyanagi et al.,2005; Gruh et al.,2006; Welikson et al.,2006), far fewer than 1% of the cells in these studies were proven to result from fusion.

At HH stage 15, the heart is composed of only cardiomyocytes and endothelial cells (Manasek,1968,1970). By stage HH19, epithelial cells from the epicardial organ are just beginning to migrate to the heart (Mikawa and Fischman,1992) and, therefore, the heart is still primarily composed of endothelial cells and cardiomyocytes. It is not until later in development, stage 29 or E6, that the epicardial cells give rise to coronary epithelial and smooth muscle cells as well as perivascular and intermyocardial fibroblasts (Dettman et al.,1998). The vWF⁺/MyHC⁺ cells detected in our study are, therefore, likely to have originated from the precardiac mesoderm. One possibility is that these cells originate from a common progenitor capable of giving rise to either endothelial cells or cardiomyocytes and that the few MyHC⁺ cells seen in vivo that coimmunostain for vWF or Flk-1 (Fig. 1) may be an intermediate stage of cardiomyogenic development. These cells may be analogous to either Flk-1–expressing mouse ES cells that can be induced to differentiate into spontaneously beating cardiomyocytes when cultured on a layer of stromal cells (Yamashita et al.,2005), or to Bry⁺, Flk-1⁺ mouse ES cells that can generate heterogeneous colonies of endothelial cells and beating cardiomyocytes (Kattman et al.,2006). However, in the embryonic chick cardiomyocyte system it appears that trypsinization and cell culture, activates a subset of endothelial gene expression rather than endothelial cells activating expression of cardiomyocyte genes.

Although the ultimate developmental fate of the cardiac Flk-1⁺/vWF⁺/MyHC⁺ cells is not known, it is apparent that, as a percentage of the total embryonic heart cells, the levels of such cells decrease substantially during development (Fig. 2I). This finding suggests that the 2.7% vWF⁺/MyHC⁺ cells detected in E3 hearts either stop proliferating or stop expressing either the cardiomyocyte or endothelial cell markers. By E6, this results in embryonic hearts with only approximately 0.09% vWF⁺/MyHC⁺cells. The decrease in such cells is not, however, observed when embryonic heart cells are isolated and cultured, suggesting that gene expression patterns in vitro may be more malleable and/or less restricted than those in vivo.

The number of E3 cardiomyocytes coexpressing myosin and vWF increases 15-fold from 2.7% of the cells in freshly dissociated E3 heart cell suspensions to 40% after 3 days in culture, whereas in cultures from E6 hearts, the percent of such cells increases nearly 400-fold, from approximately 0.09% to 33% (Fig. 4B). The increase in vWF⁺ cells in cultured E3 heart cells is preceded by a 27-fold increase in vWF mRNA levels within the first day of culture; but the increase in vWF transcripts is not accompanied by increased expression of either the endothelial transcription factor, Vezf1 or by transcripts of the VE cadherin or Tie2 genes. Thus, the entire endothelial program is not activated in Flk-1⁺/vWF⁺/MyHC⁺ cells. It is not known why only some MyHC⁺ cells activate vWF and Flk-1 expression when cultured, or what causes these and possibly other endothelial genes to be expressed. If, as might seem plausible, only the most recently differentiated cardiomyocytes are susceptible to the induction of vWF and Flk-1 expression when placed in culture, then the cultures generated from E3 hearts should exhibit a greater fold increase in dual-phenotype cells than cells from E6 hearts in which essentially all cardiomyocytes have been differentiated for several days. However, we found the opposite to be true (Fig. 3B).

Regulation of vWF and Flk-1 gene expression in cell culture depends on multiple factors that act positively or negatively on their promoters. Analysis of the human vWF promoter (−487 to +247) showed that the GATA6 transcription factor interacts with vWF promoter sequences and that this interaction is necessary for promoter activation in endothelial cells (Jahroudi and Lynch,1994). GATA is also critical for Flk-1 gene expression. Mutations of the GATA binding site in an Flk-1 intron-1 enhancer reporter gene in transgenic mice rendered the transgene completely inactive (Kappel et al.,2000). In addition, mutational analysis of a palindromic GATA site in the 5′ untranslated region of the human Flk-1 gene showed that the GATA binding site also plays a role in maintaining basal transcriptional activity in cultured bovine aortic endothelial cells (Minami et al.,2001). Because GATA6 is present in cardiomyocytes (Laverriere et al.,1994), it may play a role in the expression of vWF and Flk-1 in cultured chick cardiomyocytes, even in the absence of activating a more complete endothelial cell gene expression program.

A remaining complexity regarding vWF and Flk-1 expression by cardiomyocytes is that the regulation of most cell type-specific structural genes is controlled by multiple cell type-specific transcription factors. Because expression of at least one such transcription factor, Vezf1, is not up-regulated in chick heart cultures, vWF and Flk-1 gene expression in cardiomyocytes may not be controlled in the same manner as the Vezf1-regulated endothelial cell specific, endothelin I gene, in endothelial cells (Aitsebaomo et al.,2001). An additional enigma concerns the observation that vWF and Flk-1 expression, both in vivo and in culture, occurs in only a subset of cardiomyocytes. While not yet understood, these observations suggest a heretofore-unrecognized heterogeneity within the population of embryonic cardiomyocytes. Our data may also be pertinent to ES cell studies designed to investigate mechanisms required to induce specific cell types (Bost et al.,2002; Parisi et al.,2003; Kouskoff et al.,2005; Lowell et al.,2006), because the emergence of Flk-1⁺/vWF⁺/MyHC⁺cells from embryonic cardiomyocytes suggests that heterologous cell type-specific genes may be expressed by differentiated cells. However, although cardiomyocytes can modulate and express a broader program of genes, this may or may not have anything to do with an endothelial cell lineage relationship. Therefore, our use of the term “Flk-1⁺/vWF⁺/MyHC⁺cells” should be considered an operational term describing cardiomyocytes that also express several endothelial markers rather than as “dual phenotype” cells that function both as cardiomyoctes and endothelial cells, or as a proven transitional developmental state.

EXPERIMENTAL PROCEDURES

Cell Culture

Chick cardiac cell cultures were prepared from embryos in ages ranging from E2.5 to E6 (HH stages 17 to 29). Hearts were collected and stored in Petri dishes at room temperature in complete growth medium DMEM, containing 10% fetal bovine serum, 1% chick embryo extract, 50 U/ml penicillin, 50 μg/ml streptomycin (pen/strep), and 250 ng/ml fungizone. The hearts were then rinsed twice in phosphate buffered saline (PBS) and incubated in 0.5 ml 0.05% trypsin–ethylenediaminetetraacetic acid for 5 min at 37°C in a 5% CO₂ humidified incubator. The cells were dispersed by triturating ∼ 10 times with a P200 pipetman tip and the partially dissociated tissues were then incubated for an additional 5 min. After the second incubation, the tissue pieces were again triturated approximately 10 times and the suspension was then checked under a microscope to confirm that the cells were dissociated. An equal volume of complete culture medium was added to the suspension, and the cells were then plated onto tissue culture plates or chamber slides coated with 20 μg/ml fibronectin (Sigma, St. Louis, MO). The approximate density of cells added to the cultures was 5 × 10⁴ per cm². Cardiac muscle cultures from E6 embryos were prepared using a modification of previously described procedures (Lin et al.,1989). Tissue fragments were incubated in 4 ml of trypsin solution (0.125% trypsin in PBS containing 1 mM ethyleneglycoltetraacetic acid) for 5 min at room temperature. The supernatant containing dispersed cardiac cells was removed and temporarily stored in 25 ml of growth medium (10 % fetal calf serum, 10 U/ml penicillin, 10 mg/ml streptomycin in MEM with L-glutamine and Earle's salts; GIBCO BRL). The remaining tissue fragments were incubated in 4 ml of fresh trypsin solution, for two additional 5-min periods as above, and the dispersed cells from each harvest were added to the original suspension. The cells were then filtered by syringe through sterile lens paper held in a Swinnex 25 filter holder (Millipore, Billerica, MA) to remove large cell clumps, and then centrifuged for 5 min. The cell pellet was resuspended in growth medium and cultured as above.

Indirect Immunofluorescence Microscopy

Cultured cells were rinsed twice with PBS, fixed with 2% paraformaldehyde (PFA) in PBS for 10 min and permeabilized with 1% Triton X-100 in PBS for 10 min. After washing in PBS, cultures were blocked with 1% bovine serum albumin (BSA), 0.05% tween-20 in PBS (blocking solution) for 20 min and incubated with primary antibodies for 1 to 2 hr at room temperature. After three 5-min washes with blocking solution, the cultures were incubated with secondary antibodies and washed again with blocking solution for 0.5 to 1 hr at room temperature. After a final wash with PBS, the specimens were rinsed in distilled water and mounted in gelvatol (Air Products) with 100 mg/ml DABCO (Sigma). Indirect immunofluorescence microscopy was performed using a Zeiss Axioplan microscope using plan-neofluar 40× (N.A. = 0.73) and 100× (N.A. = 1.30) objectives or a Zeiss Axiovert 200 using a plan-neofluar 10× (N.A. = 0.3) 40× (N.A. = 0.6). Images were acquired using a DAGE-MTI 3CCD camera and imported into Adobe Photoshop using the Scion Series 7 import plug-in module (V. 1.4) and a Zeiss Axiocam MRm camera and captured using Zeiss Axiovision software (V. 4.5).

Cell Preparations for Single Cell Immunochemistry

Cells for single cell immunoanalysis were isolated as described above for chick cardiac cell cultures from E2 to E6 embryos. A total of 50 to 100 μl of cells were then placed on one side of a fibronectin-coated Superfrost-Plus glass slide (VWR), and a second slide was used to spread the cell drop. The cells were then air-dried for 10 min, fixed with 2% PFA, and processed for indirect immunofluorescence microscopy as above. Alternatively, freshly isolated heart cells were collected onto glass slides using a cytospin (Southern Instruments, Inc., Sewickley, PA) at 100 × g for 10 min at room temperature. The cells were then immediately fixed for indirect immunofluorescence microscopy.

Cryosections

E3 chicken hearts were cryoprotected in equal volumes of 10% (wt/vol) sucrose in PBS and O.C.T. (Tissue Tek, Miles, Inc.) for 1 hr at 4°C. They were embedded in O.C.T. and frozen in a bath with ethanol and dry ice. Frozen sections (5 to 10 μm) were mounted on Superfrost-Plus glass slides (VWR) and air-dried and stored at −20°C. Sections were rehydrated with PBS, treated with blocking solution for 20 min at room temperature and then incubated with primary antibody for 2 hr at room temperature or overnight at 4°C. The sections were then washed three times for 5 min in blocking solution and incubated with secondary antibodies at room temperature for 90 min. The incubation with secondary antibody was followed by three washes for 5 min with blocking solution and two washes with PBS each at room temperature. Coverslips were mounted with gelvatol containing DABCO.

Antibodies

Hybridoma supernatant containing anti–sarcomeric myosin heavy chain mAbs MF20 (gift of D.A. Fischman, Cornell Univ. Med. College) and F59 (gift of Frank Stockdale, Stanford Univ.) were used at dilutions of 1:100 and 1:10, respectively. Endothelial cells were identified using either anti–Flk-1 mAb (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:40 or anti-vWF pAb (Sigma) at 1:200. Smooth muscle cells were identified by immunostaining with anti–α smooth muscle actin (clone 1A4) from Sigma at 1:200. Secondary antibodies conjugated to Alexa fluorophores (Molecular Probes) were used at dilutions of 1:100. When necessary, anti-mouse IgG2b secondary antibody conjugated to Alexa 594 and anti-mouse IgG1 Alexa 488 were used in samples reacted with MF20 (IgG2b) in conjunction with other IgG1 mAbs.

RT-PCR

Total RNA was isolated from E3 chick hearts, cultured cells, and E17 chick hearts and aortas respectively. It was then treated with RNase-free DNase I, amplification grade (Invitrogen, Carlsbad, CA). The 2 μg of total RNA was reverse transcribed in a volume of 80 μl as previously described (Kastner et al.,2000). As a negative control, reverse transcriptase was omitted in duplicate reverse transcription reactions. For real-time PCR, 2 μl of cDNA was added to 18 μl of master mix, containing 1× SYBR Green (Applied Biosystems, Foster City, CA) and 300 μM sense and antisense primers. Triplicates of the cDNA were amplified for 40 cycles with the Opticon I Real-Time thermal cycler (MJ Research, Inc., Waltham, MA). The PCR product level is expressed as Ct, the amplification cycle at which the emission intensity of the product rises above an arbitrary threshold level. The product sizes were verified with gel electrophoresis.

To investigate whether the endothelial transcription factor Vezf1 mRNA is expressed in the early chick heart and in chick cardiomyocyte cultures, we used in silico techniques to identify a chicken EST clone (accession no. CR385256) that showed high homology to the mouse Vezf1 cDNA (accession no. AF104410). PCR primers that spanned two introns were designed for mouse and chick Vezf1 cDNAs. Using cDNA from E17 chick heart, we amplified a 335-bp amplicon, which was cloned into pCRII-Topo TA Vector (Invitrogen) according to the manufacturer's protocol. Sequence analysis was performed in sense and antisense directions, which confirmed an 89% match over the entire 335 bp between mouse Vezf1 and the chicken amplicons.

The chicken primers designed for this study are as listed here: HPRT (fw): 5′ccatgactgtggacttcattagg 3′, HPRT (rv): 5′gtactgcttaagtagagacagcaacg 3′; Vezf1 (fw): 5′ agcccttcgaatgtccgatt 3′, Vezf1 (rv): 5′ gctctgcccgtgtgtcttta 3′; vWF (fw): 5′ttcactgaacaagcctcagga 3′, vWF (rv): 5′tcttgaagtgagtgcagcag 3′; ventr. MyHC (fw): 5′ gctgaagcacaagccaatct 3′, ventr. MyHC (rv): 5′ ccgcacttatctcctctgcat 3′; VE Cadherin (fw): 5′ tcttggctccagcattctct 3′, VE Cadherin (rv): 5′ ccatcgtacccttgcacttt 3′; VE Tie2 (fw): 5′ cctcatcagaatggcagaaa 3′, VE Tie2 (rv): 5′ gcagtgggagctacaggaag 3′.

In Situ Hybridization

The sense and antisense probes for vWF fluorescence in situ hybridization (FISH) analysis were generated from E17 chick heart cDNA. A 525-bp amplicon corresponding to nucleotides 7957 to 8481 of the vWF cDNA was generated by PCR using forward (5′TCACTTCAGGATGGCTGTGA3′) and reverse (5′GGAGAGAAAATGAGGCTTGC3′) primers. The PCR product was cloned into pCRII-Topo TA Vector (Invitrogen) according to the manufacturer's protocol and recombinant plasmids were isolated with the vWF cDNA in both forward and reverse directions (pCRII-vWF-F and pCRII-vWF-R). Sequence analysis was performed in sense and antisense directions on both plasmids and showed 100% match to the chicken vWF.

Alexa 488–labeled vWF sense and antisense cRNA probes were generated using T7 polymerase and the FISH Tag DNA Green Kit (Invitrogen) and hybridized to cultured E3 heart cells using the mRNAlocator Kit under RNase-free conditions. Before hybridization, heart cells were fixed with 4% PFA in PBS for 10 min, permeabilized in absolute methanol for 10 min, and washed in 50 mM Tris buffer pH 7.5. After hybridization, the cells were washed at 55°C in 4× standard saline citrate (SSC), 1 mM dithiothreitol (DTT) for 5 min and then 2× SSC, 1 mM DTT for 30 min. The cells were then treated with RNase A (provided in the mRNAlocator kit) at 37°C for 30 min with gentle agitation, washed at 55°C in 2× SSC, 1 mM DTT for 30 min and then with 0.1× SSC for 30 min.

vWF mRNA⁺ and MyHC⁺ cells were identified by immunostaining with anti–Alexa Fluor 488 and anti–sarcomeric myosin antibodies. In brief, hybridized and washed cultures were blocked in 1% BSA, 0.05% tween-20 in 2× SSC (SSC blocking solution) for 20 min and incubated with anti–Alexa Fluor 488 pAb (Invitrogen) and MF20 for 2 hr at 1:100 dilutions at RT. After three 5-min washes with SSC blocking solution, the cultures were incubated with anti-rabbit and anti-mouse secondary antibodies conjugated to Alexa 488 and 594, respectively, and washed again with blocking solution for 0.5 to 1 hr at room temperature. After a final wash with 2× SSC, the cells were rinsed in distilled water and mounted in gelvatol (Air Products) with 100 mg/ml DABCO (Sigma). Images were obtained as described above.

Acknowledgements

We thank members of the Hauschka lab for their insightful discussions and for critical comments on the manuscript. Our cherished colleague Stephanie Kaestner died from cancer during the course of these studies. Her enthusiastic and highly skilled contributions are greatly missed. This work was supported by National Institutes of Health (BioEngineered Autologous Tissue [BEAT] Grant) and postdoctoral training grants from the American Heart Association, and Cardiovascular Research Training Program at the University of Washington, Experimental Pathology of Cardiovascular Disease.