Development of heart valve leaflets and supporting apparatus in chicken and mouse embryos

Authors

  • Joy Lincoln,

    1. Division of Molecular Cardiovascular Biology, Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Christina M. Alfieri,

    1. Division of Molecular Cardiovascular Biology, Children's Hospital Medical Center, Cincinnati, Ohio
    Search for more papers by this author
  • Katherine E. Yutzey

    Corresponding author
    1. Division of Molecular Cardiovascular Biology, Children's Hospital Medical Center, Cincinnati, Ohio
    • Division of Molecular Cardiovascular Biology, Children's Hospital Medical Center, Cincinnati ML 7020, 3333 Burnet Avenue, Cincinnati, OH 45229
    Search for more papers by this author

  • Figure 1 of this article was used for cover of the April issue of Developmental Dynamics Volume 229 Number 4

Abstract

Abnormalities in valvuloseptal development significantly contribute to congenital heart defects, yet the underlying causes are complex and poorly understood. Early cardiac regulatory genes are differentially expressed during valvuloseptal development, consistent with novel functions during heart chamber formation in chicken and mouse embryos. Distinct valve cell lineages were identified in the leaflets, chordae tendineae, and myotendinous junctions with the papillary muscles based on restricted expression of extracellular matrix molecules. Specific cell types within these structures demonstrate characteristics of chondrogenesis and tendon development, identified by scleraxis, type II collagen, and tenascin expression. In chicken embryos, valve remodeling and maturation accompanies a decrease in mitotic index indicated by reduced bromodeoxyuridine incorporation. Analysis of Tie2-cre × ROSA26R mice demonstrates that mature valve structures, including the atrioventricular and outflow tract semilunar valve leaflets, chordae tendineae, and the fibrous continuity that connects the septal leaflets of mitral and tricuspid valves, arise from endothelial cells of the endocardial cushions. Together, these studies provide novel insights into the origins and cell lineage diversity of mature valve structures in the developing vertebrate heart. Developmental Dynamics 230:239–250, 2004. © 2004 Wiley-Liss, Inc.

INTRODUCTION

Congenital heart defects (CHD) are among the most common birth defects in humans, occurring at an approximate rate of 1 in 100 live births in the United States (Hoffman and Kaplan, 2002). The majority of CHD are due to abnormalities in valvuloseptal development, and there is emerging evidence to suggest genetic contribution to cardiac structural abnormalities (Pierpont et al., 2000; Srivastava, 2001). During embryonic development, the initial events of valve formation are apparent in the induction of endocardial cushions of the atrioventricular (AV) canal and outflow tract (OFT) Eisenberg and Markwald, 1995; Barnett and Desgrosellier, 2003; Schroeder et al., 2003). The AV endocardial cushions fuse and condense into distinct mitral and tricuspid valve primordia, which consequently remodel to form the valve leaflets and supporting structures. While signaling pathways for initial endocardial cushion formation have been identified, later events of valve maturation and remodeling are not well defined.

The mature valves that form with the restructuring of endocardial cushion mesenchyme into organized leaflets and supporting apparatus are composed of diverse cell types and extracellular matrices (Wenink and Gittenberger-de Groot, 1986; Icardo and Colvee, 1995; Lamers et al., 1995; Wessels et al., 1996; Oosthoek et al., 1998). The tricuspid and mitral valve leaflets that form at the AV junction consist of a complex network of connective tissue and interstitial cells. The chordae tendineae support the valve leaflets and insert directly into the papillary muscles that project into the ventricles (Lamers et al., 1995). Histological and immunocytochemical analyses have been used to characterize the formation of mature AV valve leaflets and supporting structures during avian, mouse, and human development (Chin et al., 1992; Lamers et al., 1995; Wessels et al., 1996; Oosthoek et al., 1998; Webb et al., 1998). Initial cell labeling studies performed by injecting India ink or placing carbon particles on the endothelial surface of the endocardial cushions in avian embryos provided preliminary evidence that the AV valve leaflets and chordae tendineae are derived from cells of the endocardial cushions (De la Cruz et al., 1983; Oosthoek et al., 1998). However, the cell lineages and precise endothelial origins of specific valvuloseptal structures have not yet been clearly demonstrated.

Several critical regulators of early cardiac lineage induction and differentiation have been identified, but the specific roles of these signaling molecules and transcription factors in the later events of heart chamber formation are less well known (Bruneau, 2002; Yutzey and Kirby, 2002; Brand, 2003). In avian embryos, expression of bmp2, smad6, nkx2.5, gata4, tbx5, and tbx20 overlap in the heart forming region (Schultheiss et al., 1995, 1997; Ehrman and Yutzey, 1999; Yamada et al., 1999; Liberatore et al., 2000; Iio et al., 2001). Similar expression in the early mouse embryo is less clear; however, null mutations or reduced expression of these genes are associated with very early lethal heart defects in mouse and/or zebrafish embryos (Lyons et al., 1995; Zhang and Bradley, 1996; Molkentin et al., 1997; Kraus et al., 2001; Szeto et al., 2002). Initial evidence for later roles for these factors in heart chamber development is provided by the association of heterozygous mutations in TBX5, NKX2-5, and GATA4 with congenital heart defects and conduction system anomalies in the human population (Basson et al., 1997; Li et al., 1997; Schott et al., 1998; Garg et al., 2003). There is additional supporting evidence from gene targeting studies in mice for critical functions for Tbx5, Nkx2.5, bone morphogenetic protein (BMP) signaling, and Smad6 during valvuloseptal development in the AV canal and OFT (Biben et al., 2000; Galvin et al., 2000; Bruneau et al., 2001; Delot et al., 2003). The necessity for several early cardiac regulatory genes in later events of heart organogenesis may be indicative of novel functions for these genes in the diversification of specialized cardiac cell lineages and development of specific valvuloseptal structures.

Molecular and cellular regulatory mechanisms that control cardiac valve formation were assessed in chicken and mice. Differential expression of early cardiac regulatory genes was observed in distinct valvuloseptal structures concurrent with heart chamber formation. In the differentiated valves, restricted expression of extracellular matrix molecules defines distinct cellular compartments in the leaflets and supporting structures. Regions of increased cell cycling during valvulogenesis in chicken embryos are consistent with a distal outgrowth mechanism of valve elongation and a reduction in cell proliferation in differentiated valve structures. In addition, the endothelial cell origins of the mature valve leaflets and supporting structures in mice were demonstrated. Together, these analyses provide novel evidence for specific cellular and molecular mechanisms by which the endocardial cushions remodel and mature into the valves of the heart.

RESULTS

Early Cardiac Regulatory Genes Are Differentially Expressed During Heart Valve Remodeling and Chamber Formation in Chicken and Mouse Embryos

Expression of transcription factors and signaling molecules important in early heart formation was examined at later stages of valvuloseptal development as an indicator of diversified regulatory functions in cardiac organogenesis. The gene expression patterns of gata4, nkx2.5, tbx5, tbx20, bmp2, and smad6 were examined in bisected chicken and mouse hearts relative to myosin heavy chain (MHC) protein expression. Expression analyses were performed on chicken hearts isolated at embryonic day 6 (E6), after the endocardial cushions have fused to form distinct valve primordia, and at E10, after remodeling of valve leaflets and chordae tendineae. Mouse hearts of comparable stages, E12.5 and E17.5, also were examined. Expression studies were performed on bisected hearts to visualize complex valvuloseptal structures in three dimensions.

MHC expression was detected using the MF20 antibody (Bader et al., 1982) in E6 and E10 sectioned chicken hearts (Fig. 1A,B). MF20-reactive cells were not detected in the valve primordia at E6 (Fig. 1A) or the mature mitral valve at E10 (Fig. 1B). In contrast, the avian mural tricuspid valve leaflet is composed almost completely of muscle (Fig. 1B). Nkx2.5 is expressed in the muscular mural tricuspid valve leaflet in addition to the atrial and ventricular myocardium (arrow Fig. 1E,F). However, nkx2.5 expression was not observed in the valve primordia at E6 (arrow Fig. 1E), nor the remodeled mitral valve at E10 (arrowheads Fig. 1F). Although GATA4 and Nkx2.5 are both transcriptional regulators of cardiac muscle gene expression, their expression patterns during heart chamber formation are distinct. Expression of gata4 was relatively undetected in myocyte cell lineages (Fig. 1C,D), but gata4 is expressed in endothelial cells of the valve primordia at E6 (Fig. 1C), and the fibrous mitral valve leaflets at E10 (Fig. 1D). Myocardial expression of tbx5 is predominant in the atria and left ventricle, but relatively little expression was detected in the valve primordia at E6 (Fig. 1H) or in the valve leaflets or chordae tendineae at E10 (Fig. 1H). However, both tbx5 and nkx2.5 are expressed in the specialized cardiac conduction system, including the left bundle branch (LBB), the right atrioventricular ring (RAR), and the retroaortic ring (RR) at E10 (Fig. 1F,H; Gourdie et al., 2003).

Figure 1.

Expression of early cardiac regulatory genes during chicken valvuloseptal development. A–N: Immunohistochemistry (A,B) and in situ hybridization (C–N) were used to examine the differential expression patterns of early cardiac regulatory genes in bisected chicken hearts. Expression was examined in the valve primordia (vp, arrow) and atrial septum (as, arrowhead) at embryonic day (E) 6 (A,C,E,G,I,K,M) and the tricuspid (tv, arrows) and mitral (mv, arrowheads) valves at E10 (B,D,F,H,J,L,N). A,B: Immunohistochemistry of MF20 in the chicken, at E6 (A) and E10 (B). C–J: In situ hybridization expression of gata4 (C,D), nkx2.5 (E,F), tbx5 (G,H), and tbx20 (I,J). K–N: Expression of bmp2 (K,L) and smad6 (M,N) in the atrioventricular canal region (AVC, concave arrows). The atrial surface of the mural tv is indicated (asterisk in D). Regions of the specialized conduction system are abbreviated retroaortic ring (RR), right atrioventricular ring (RAR), left bundle branch (LBB, white concave arrows).

Additional transcription factors and signaling molecules present in the heart forming region exhibit distinct expression patterns during valvuloseptal development. Tbx20 is expressed in the heart primordia and primitive heart tube of mouse and chicken embryos and has been identified as a repressor of tbx5 gene expression in the zebrafish primitive heart (Iio et al., 2001; Kraus et al., 2001; Szeto et al., 2002). In the developing chick, tbx20 is highly enriched in valvuloseptal structures (Fig. 1I,J). These structures include the AV and OFT valve primordia as well as AV canal and atrial myocardium at E6 (Fig. 1I) and on the atrial surfaces of the mature mitral and tricuspid valve leaflets (arrowheads, Fig. 1J) at E10. However, relatively little tbx20 expression was apparent in the ventricular myocardium or muscular tricuspid valve. The expression of tbx20 during valve development is consistent with an important, but yet unknown, role in valvuloseptal development. Colocalization of bmp2 and smad6 expression was observed in the endocardial valve primordia and myocytes of the AV canal (AVC) at E6 (Fig. 1K,M). Bmp2 expression also was detected in the forming atrial septum (as) at E6 (arrowhead, Fig. 1K), which could be indicative of novel myogenic inductive events in atrial septation. The expression patterns of bmp2 and smad6 support previously identified complex regulatory functions for this signaling pathway in multiple aspects of heart chamber formation.

Expression of early cardiac regulatory genes also was examined in valve primordia and mature valvuloseptal structures in comparative stages of the mouse, E12.5 and E17.5. Unlike the chicken, MF20-reactive cells were not observed in either AV valves, demonstrating that both the mitral and tricuspid valves of the mouse are nonmuscular (Fig. 2A,B). The absence of nkx2.5 and tbx5 expression from the valve primordia and mature valves is consistent with the nonmuscular phenotypes of the mouse valves (Fig. 2E–H). Similar to the chicken, gata4 and tbx20 are both present in the undifferentiated valve primordia and tbx20 is highly expressed in the mature valves (Fig. 2C,I,J). However, unlike the chicken, enrichment of tbx5 and nkx2.5 expression in structures associated with the mouse cardiac conduction system was not observed (Fig. 2F,H). In the mouse, bmp2 expression was not observed in the AV valve primordia, but expression was restricted to the outflow tract cushions and surrounding myocardium at E12.5 (Fig. 2K). However, there was similar bmp2 expression in the AV canal myocardium in both mouse and chicken (Figs. 1K,L, 2L). Together these analyses demonstrate that regulatory genes involved in early cardiac development are differentially expressed in mouse and chicken embryos and that these differences may be associated with distinct cellular compositions of valve structures in the two species.

Figure 2.

Expression of early cardiac regulatory genes during mouse valvuloseptal development. A–L: Immunohistochemistry on 7-μm sections (A,B) and in situ hybridization of bisected mouse hearts (C–L) were used to examine expression patterns of early cardiac genes in the mouse, at the level of the valve primordia (vp, arrows) at embryonic day (E) 12.5 (A,C,E,G,I,K) and the mature tricuspid (tv, arrows) and mitral (mv, arrowheads) valves at E17.5 (B,D,F,H,J,L). A,B: Immunohistochemistry of myosin heavy chain, detected using MF20 at E12.5 (A) and E17.5 (B). C–L: In situ hybridization expression pattern of gata4 (C,D), nkx2.5 (E,F), tbx5 (G,H), tbx20 (I,J), and bmp2 (K,L).

Extracellular Matrix Molecules and Regulatory Factors Associated With Chondrogenic and Tendon Cell Lineages Are Differentially Expressed During Valvuloseptal Development

Early biochemical and ultrastructural studies demonstrated that heart valves are primarily composed of an organized matrix of connective tissue (Manasek, 1976). The developmental expression of individual extracellular matrix molecules was assessed to identify molecular markers for specific valve compartments, including the valve leaflets, chordae tendineae, and the myotendinous junction with the papillary muscles. Extracellular matrix proteins examined include type I collagen (the main component of fibrous tissue), type II collagen (predominant in cartilage), and tenascin (expressed in tendon and chondrogenic cell types; Linsenmayer and Hendrix, 1980; Chiquet and Fambrough, 1984; McDonald et al., 1986; Edom-Vovard et al., 2002). In addition, reactivity of valvuloseptal structures with antibodies C1 and C5, which recognize muscle fibroblast and chondrogenic precursors respectively in the somites and limb buds also was determined (George-Weinstein et al., 1988). Expression of the basic helix loop helix (bHLH) transcription factor scleraxis was examined to establish similarities in the molecular hierarchies that control heart valve extracellular matrix remodeling with other connective cell lineages, including tendon and cartilage (Schweitzer et al., 2001). These markers of chondrogenesis and tendon development were used to distinguish specific compartments of the mature valves and supporting structures in avian embryos and neonatal mice.

Distinct patterns of extracellular matrix molecule expression were identified in the valve leaflets and supporting structures of the chicken mitral valve and both AV valves of the mouse. Chicken hearts were examined at E10, when the primitive valves are apparent, and at E14, after maturation of the leaflets and supporting structures. At E10, scleraxis is expressed in the chordae tendineae adjacent to the papillary muscles but is absent from the valve leaflets (Fig. 3A). Restricted expression of the C1 and C5 antigens also was observed in the valve structures, with restricted reactivity in the myotendinous junction (Fig. 3B,C). At E10 and E14, type I collagen is strongly expressed in the fibrous valve leaflets, but is absent from the supporting chordae tendineae and myotendinous junctions (Fig. 3D,E). At E14, expression of type II collagen is observed in the valve supporting apparatus, consistent with the presence of chondrogenic cell types in this structure (Fig. 3H). In contrast, expression of type II collagen is not detectable in the valve leaflets (Fig. 3H). At E10, tenascin is expressed in the valve leaflets and its expression is increased throughout the valve supporting apparatus at E14 (Fig. 1J,K). The chicken mural leaflet of the tricuspid valve, which is composed primarily of muscle, is predominantly unreactive with chondrogenic and tendon lineage markers. However, expression of type I collagen, type II collagen, and tenascin was observed in the septal leaflet of the tricuspid valve as well as the atrial surface of the muscular tricuspid leaflet.

Figure 3.

Cell lineage compartmentalization during valvuloseptal maturation. A–L: In situ hybridization (A) and immunohistochemistry (B–L) were carried out on bisected chicken hearts at embryonic day (E) 10 (A–D,G,J) and E14 (E,H,K) and on 7-μm sections of neonatal mouse hearts (mNeonate, F,I,L). A: In situ hybridization of scleraxis expression in the chordae tendineae (arrowheads) of the E10 chicken mitral valve. B,C: Immunohistochemistry of cells reactive to C1 and C5, respectively, at the myotendinous junction (arrows). D–F: Type I collagen expression in the tricuspid (tv, arrows) and mitral valves (mv, arrowheads) of the chicken heart, with additional expression detected on the atrial surface of the mural tv (+) at E10 (D) and E14 (E), and in the neonatal mouse heart (F). G–I: Type II collagen expression in the chicken heart at E10 (G) and E14, with the asterisk marking the chordae tendineae (H). I: Type II collagen reactivity in the neonatal mouse heart, including the fibrous continuity structure across the ventricular septum (pound sign). J–L: Tenascin reactivity in E10 (J) and E14 (K) chicken hearts and the neonatal mouse heart (L).

Restricted expression of extracellular matrix molecules also was observed in neonatal mouse hearts. This stage is comparable to E14 avian hearts when mature valve leaflets and supporting structures are apparent. As in the chick, type I collagen is strongly expressed in the mitral valve leaflets but not supporting chordae tendineae (Fig. 3F). In contrast to the chick, type I collagen expression was observed in the tricuspid valve leaflets of the neonatal mouse, consistent with the nonmuscular nature of this valve. Type II collagen expression in mouse hearts was apparent in the chordae tendineae, consistent with expression observed in the chick. Also similar to the chicken is the expression of tenascin in the murine valve leaflets and supporting structures (Fig. 3I). An additional region with strong expression of both type II collagen and tenascin, but not type I collagen, is the fibrous continuity between the septal leaflets of tricuspid and mitral valves that extends across the crest of the ventricular septum (Fig. 3F,I,L, indicated by pound signs). Together, these analyses demonstrate the restricted expression of chondrogenic and tendon lineage markers in distinct compartments of the mature valves and supporting structures.

Cell Proliferation Is Decreased During Differentiation of the Avian Valve Primordia

The mechanisms that induce outgrowth and remodeling of the un-differentiated valve primordia into mature valve leaflets, chordae tendineae, and myotendinous junctions are not yet defined. Elongation and remodeling of the primordia into mature valve structures is likely to be associated with regional changes in the cell proliferation of valve progenitor populations. Temporal and spatial regulation of cell proliferation in valve primordia and definitive valves was examined by using 5-bromo-2′-deoxyuridine (BrdU) incorporation as an indicator of DNA synthesis in cells in S-phase. Fertilized chicken eggs were injected with BrdU at valve primordium stages (E6 and E7) or during valve maturation (E10). Immunohistochemistry for BrdU incorporation suggested a higher degree of positive staining nuclei at distal ends of mitral (arrowheads) and tricuspid (arrows) valve primordia at E6 and E7 (Fig. 4A–F). However, fewer proliferating cells are apparent proximal to the ventricular septum, supporting a distal outgrowth mechanism for elongation of the valve primordia (Fig. 4D,E, asterisks). By E10, BrdU incorporation in tricuspid and mitral valves is obviously reduced (Fig. 4G–I), indicating that the remodeling and maturation of the valve primordia is accompanied by decreased proliferation.

Figure 4.

Temporal and spatial regulation of cell proliferation in valvuloseptal structures of chicken hearts at embryonic day (E) 6, E7, and E10. A–I: Embryos were exposed to bromodeoxyuridine (BrdU) for 6 hr at E6 (A–C), E7 (D–F), and E10 (G–I). BrdU incorporation in proliferating cells was assessed immunohistochemically in hearts that were paraffin-embedded, sectioned (5μm), and counterstained with eosin. A–F: BrdU incorporation in the valve primordia (vp) and ventricular septum (vs) at E6–E7. Increased cell proliferation is apparent in the distal regions of the valve primordia and definitive valves (arrows indicate tricuspid valves, arrowheads indicate mitral valves) and adjacent endocardium. G–I: Reduced BrdU incorporation was observed in the valve leaflets at E10 (arrows and arrowheads). Photomicrographs are shown at ×10 objective magnification (A,D,G), with boxed areas indicating higher magnification at ×20 objective (B,C,E,F,H,I) to visualize BrdU-positive cells in the tricuspid valve (arrows) and mitral (arrowheads) valves. The asterisk indicates areas of fewer proliferating cells in relation to the distal valve leaflets and ventricular septum.

The proliferation indices of cells in the mitral and tricuspid valve primordia and mature valves were calculated as the number of BrdU-positive nuclei/total number of nuclei (Fig. 5). High levels of proliferation (>40% of all cells) were observed in mitral and tricuspid valve primordia at E6 and E7. No significant differences in the percentages of BrdU-positive nuclei between the mitral and tricuspid valves were detected at any developmental stage examined. In the mature valves at E10, there was an approximate 75% reduction in cell proliferation compared with E5 and E6 (Fig. 5). The observed decrease in BrdU-positive cells is not due to general reduction in proliferation throughout the embryo, as the mitotic index in brain and lung tissue was unchanged at these stages (data not shown). These data demonstrate that the valve primordia are highly proliferative and that decreased proliferation is a feature of the differentiated and remodeled mature valves.

Figure 5.

Quantitative analysis of cell proliferation in avian valve primordia and mature valves. Mitotic indices were calculated based on the percentage of bromodeoxyuridine (BrdU) labeled nuclei in the tricuspid (TV, filled bars) and mitral (mv, open bars) valves of the chicken at embryonic day (E) 6, E7, and E10. Cell counts were performed in three fields on three hearts, and numbers are represented as number of BrdU-positive nuclei/total nuclei. Statistical significance of reduced proliferation at E10 (asterisks) was determined by Student's t-test with P < 0.05 (P = 0.019).

Murine Mature Valve Leaflets, Chordae Tendineae, Myotendinous Junction, and Valvuloseptal Fibrous Continuity Are Derived From the Endocardial Cushions

The contribution of the endocardial cushion cells to the mature valves and supporting structures was determined in progeny of Tie2-cre (tek-cre) transgenic mice bred with ROSA26R reporter mice (Soriano, 1999; Koni et al., 2001). The Tie2(tek) promoter is specifically expressed in endothelial cells; therefore, the Tie2-cre transgene provides an effective mechanism for cell lineage analysis of endothelial-derived structures (Kisanuki et al., 2001). The ROSA26R locus contains loxP sites adjacent to the lacZ reporter gene and expresses β-galactosidase (β-gal) in the presence of Cre recombinase (Soriano, 1999). Progeny of Tie2-cre and ROSA26R mice were examined at E12.5 and neonatal stages to identify endothelial-derived cells in the valve primordia and mature valves. Genomic DNA isolated from embryonic yolk sac or neonatal tail clips was used to identify individuals with both the Tie2-cre transgene and the ROSA26R lacZ insertion. Bisected hearts were stained with X-gal to detect Tie2-cre–positive cells and their progeny in the AV canal and OFT at E12.5 (Fig. 6A,B) and in the neonatal valves and supporting structures (Fig. 6C–H). Tie2-cre/ROSA26R–positive embryos express β-gal throughout the endocardial cushions of the AV canal and proximal OFT at E12.5 (Fig. 6A,B). The high levels of recombination observed at E12.5 in AV and OFT endocardial cushions and valve primordia are indicative of the extensive endocardial contributions to these structures. Because all progeny of these cells will express β-gal, the fates of these primordial structures can be determined at later stages of valve maturation.

Figure 6.

Determination of endothelial origins of valve leaflets and supporting structures in Tie2-cre × ROSA26R mice. X-gal staining for Tie2-cre × ROSA26R analysis of endocardial cushion fate in the tricuspid valve (tv) and mitral valve (mv) primordia and outflow tract (oft) at embryonic day (E) 12.5 of the mouse (A,B), and the mature neonatal valves (C–H). A,B: β-Galactosidase expression was examined in bisected hearts showing the atrioventricular (AV) endocardial cushion and valve primordia at E12.5 (A) and proximal oft cushions (arrow, B). C: Whole-mount X-gal staining in the neonatal tricuspid valve (tv, arrow), mitral valve (mv, arrowheads), and chordae tendineae (asterisk). D–H: Five-micrometer section of post-fixed X-gal stained transgenic neonatal mice. D: Aortic (Ao) semilunar valve leaflets (arrows). E: The AV canal region, including the tricuspid (arrows) and mitral (arrowheads) valve leaflets, the chordae tendineae (asterisk), and the fibrous supporting apparatus (pound sign). F–H: Higher magnification of the indicated boxed areas in E, showing the fibrous supporting apparatus (F), the tricuspid valve leaflets (G), and the mural mitral valve leaflet (H). Arrowheads in H indicate regions of the valves that do not express β-galactosidase and, therefore, are not derived from Tie2-cre–expressing cells.

In the neonate, high levels of β-gal expression were detected in the mature valve leaflets, chordae tendineae, and myotendinous junctions, demonstrating their endothelial origins (Fig. 6D–H). The vast majority of cells in each of these structures are X-gal–positive, although localized areas of less staining were noted (Fig. 6H). An additional region of obvious β-gal expression is the supporting valvuloseptal fibrous continuity that spans the ventricular septum between the septal leaflets of the mitral and tricuspid valves (Fig. 6E,F indicated by pound signs). In contrast, β-gal expression was not detected in muscular components of the atrial or ventricular septa, suggesting alternative cellular origins for these structures. In the outflow tract, β-gal expression was observed in the mature aortic semilunar valve leaflets (Fig. 6D) but not in the septum dividing the pulmonary artery and aorta, which has been shown previously to be derived from cardiac neural crest (Jiang et al., 2000). These cell lineage analyses clearly demonstrate that the valve leaflets and supporting structures are primarily of endothelial origin.

DISCUSSION

Chicken and mouse hearts were examined during valvuloseptal remodeling and differentiation for compartmentalized expression of early cardiac regulatory genes and extracellular matrix molecules. In addition, the temporal and spatial control of cell proliferation and endothelial origins of the valves and supporting structures were determined. Restricted expression in distinct cell lineages and cardiac structures was observed for gata4, nkx2.5, tbx5, tbx20, smad6, and bmp2 in chicken and mouse embryos, which is in contrast to their overlapping expression patterns during the initial stages of cardiac lineage determination and differentiation. Specific compartments of the mature valves, including the leaflets, chordae tendineae, and myotendinous junctions were identified based on restricted expression of extracellular matrix molecules. The expression of several markers of chondrogenesis and tendon development, including the bHLH gene scleraxis, in the mature valves may indicate shared regulatory mechanisms between these connective tissue lineages and valvular supporting structures. Localized areas of cell proliferation in the valve primordia of the chick are consistent with a distal outgrowth mechanism of valve elongation and demonstrate that valve differentiation is accompanied by decreased proliferation. The endothelial origins of the mature valve leaflets and supporting structures, including a valvuloseptal fibrous continuity between mitral and tricuspid valves, were demonstrated in Tie2-cre x ROSA26R mice. Together these analyses provide new insights regarding the regulatory events that control the transformation of endocardial cushions into mature valves and their supporting structures.

Analysis of the temporal–spatial expression of early cardiac regulatory genes during valvuloseptal development revealed association with diverse cell types in distinct cardiac compartments. Nkx2.5 expression in the atrial and ventricular myocyte lineages and in the muscular mural tricuspid valve of the chicken embryo is consistent with its function as a transcriptional regulator of cardiac muscle gene expression (Bruneau, 2002). In contrast to nkx2.5, gata4 is expressed in chicken and mouse valve primordia and mature valves, which may be indicative of novel functions in nonmuscle cell types during valvuloseptal development. Similarly, mouse gata5 is expressed in the endocardium where it regulates endothelial cell differentiation and endothelin-1 gene expression in conjunction with NFATc1 (Nemer and Nemer, 2002). The colocalized expression of chicken tbx5 and nkx2.5 in a subpopulation of cells associated with the specialized conduction system is consistent with conduction system anomalies associated with mutation of human NKX2.5 and TBX5 (Basson et al., 1997; Schott et al., 1998). This colocalization of tbx5 and nkx2.5 was not observed in embryonic mouse hearts; however, mice with heterozygous mutations of tbx5 exhibit conduction system anomalies (Bruneau et al., 2001). In both chicken and mouse embryos, tbx20 expression is predominant in the valve primordia and mature valves; however, its regulatory function in valvuloseptal development is unknown. While the expression of all of these transcription factors overlaps early in heart development, it seems likely that their regulatory functions have diverged during the later events of heart chamber specification.

There is extensive evidence for the importance of BMP signal transduction in valvuloseptal development, but the precise regulatory functions and cell types involved have not been completely determined (Delot, 2003). In several studies, BMP signaling has been associated with decreased proliferation, migration, differentiation, and apoptosis of endocardial cushion cells during valve maturation and remodeling (Nakajima et al., 2000; Abdelwahid et al., 2001; Jackson et al., 2003). In mice, mutations in BMP ligands, receptors, or signaling proteins lead to valve and septal defects in the AV canal and outflow tract (Galvin et al., 2000; Kim et al., 2001; Jiao et al., 2003; Delot et al., 2003). Unlike the chicken, where bmp2 is expressed in the AV cushions, mouse bmp2 expression appears more restricted to the outflow tract. However, other BMP ligands, including BMP4, are expressed in the mouse AV canal and are required for normal AV valvuloseptal development (Jiao et al., 2003). Additional BMPs, including BMP5, -6, and -7 also are expressed in mouse and chicken AV valvuloseptal structures and likely regulate aspects of cell growth, differentiation, migration, and death (Kim et al., 2001; Yamagishi et al., 2001). The role(s) of BMP2 in the AV valve primordia are complex but may include inductive functions in specific valvular cell lineages as well as control of morphogenetic events.

Analysis of regulatory gene expression and extracellular matrix molecule distribution demonstrates species-specific differences in the cell types of the mature valve leaflets and supporting structures of chicken and mouse embryos. In the chicken, the mural leaflet of the tricuspid valve has a unique muscular composition and is not supported by insertion of fibrous supporting structures into ventricular papillary muscles (this study, Chin et al., 1992; Sedmera et al., 2000). The avian mitral valve is not muscular and is composed of distinct extracellular matrix phenotypes in the valve leaflets, chordae tendineae, and myotendinous junction. In the mouse, similar expression patterns of type I collagen, type II collagen, and tenascin were observed in mitral and tricuspid valve leaflets and supporting structures. The primarily muscular tricuspid valve without ventricular fibrous support in the chicken appears to be unique to avians, because fibrous valve leaflets and supporting structures are apparent in AV valves of Xenopus, zebrafish, mouse, and human embryos (Wessels et al., 1996; Sedmera et al., 2000). An additional feature of human AV valves is the close apposition of valve leaflet and chordae tendineae progenitors with muscular supporting structures in the valve primordia (Oosthoek et al., 1998). The association of cushion-derived fibrous valve structures with a differentiated muscle cell layer was not observed in the mitral valve of the chick or in either AV valves of the mouse. Therefore, the delamination of the fibrous valve leaflets from underlying myocardium may be a unique feature of human valve development (Wessels et al., 1996; Oosthoek et al., 1998). Together these histological analyses demonstrate that, while the early events of cardiogenesis are strikingly conserved across species, the later events of valve maturation and remodeling have become specialized among vertebrates.

Molecular markers of chondrogenesis and tendon development are present in the mature valve leaflets and supporting structures in both E14 chicken and neonatal mouse hearts (Fig. 7). Type II collagen is characteristic of cartilage and is expressed in the chordae tendineae adjacent to the valve leaflets (Chimal-Monroy et al., 2003). Scleraxis, a transcriptional regulator important in development of tendon progenitors of the limb and somites, also is expressed in the junction of the valve supporting apparatus and papillary muscles of the avian mitral valve (Schweitzer et al., 2001). Tenascin and C1 and C5 antigens, which are present in tendon and cartilage development in the limb and somites, also are expressed in the valve supporting apparatus and cardiac myotendinous junctions (George-Weinstein et al., 1988; Edom-Vovard et al., 2002). In the developing limb and somites, tendon development is determined by fibroblast growth factor signaling and Scleraxis activity regulating tenascin expression (Edom-Vovard et al., 2002). Limb chondrogenesis is characterized by BMP induction of Sox9, a transcription factor that activates type II collagen gene expression (Ng et al., 1997; Chimal-Monroy et al., 2003). In the developing avian heart, the critical regulatory components of both tendon lineage determination and chondrogenesis are temporally and spatially regulated in the valve primordia and mature valves (this study; Karabagli et al., 2002; Montero et al., 2002; Sugi et al., 2003). However, the precise functions of these regulatory pathways in the differentiation and patterning of the mature valves and supporting structures have not yet been determined.

Figure 7.

Compartmentalization of mature valve leaflets and supporting structures. A: The four-chambered avian heart at embryonic day (E) 14 with tenascin expression throughout the fibrous valves and supporting apparatus. B: The E14 avian mitral valve structures highlighted in the boxed area are shown schematically. The valve leaflets (yellow) predominantly express type I collagen. The valve leaflet is attached to the papillary muscles by means of chordae tendineae (blue). The chordae tendineae show further diversification between the junctions with the valve leaflet, marked by type II collagen (dark blue), and scleraxis in the region adjacent to the myotendinous junctions (light blue). The myotendinous junctions (red) with the papillary muscle are marked by C1 reactivity.

The derivation of mature valve structures from the undifferentiated endocardial cushions has been an area of controversy (De la Cruz et al., 1983; Lamers et al., 1995; Wessels et al., 1996; Oosthoek et al., 1998). The use of the Tie2-Cre × ROSA26R genetic approach in mice unambiguously labels cells derived from endothelial lineages. Previously, this system was used to show that the endocardial cushion mesenchyme at E12.5 is derived entirely from endothelial progenitors (Kisanuki et al., 2001). In the neonatal mouse, the AV and OFT valve leaflets and supporting chordae tendineae of the AV valves are primarily derived from the cushion endothelial lineages. In addition, a striking fibrous continuity between the septal leaflets of the tricuspid and mitral valves across the ventricular septum also is of endothelial cushion origin. This valvuloseptal fibrous continuity is apparent in human embryonic hearts and likely supports the septal leaflets of the AV valves throughout life (Oosthoek et al., 1998). In the chick, a population of epicardium-derived cells has been identified that contributes significantly to the AV valves (Gittenberger-de Groot et al., 1998). Therefore, the incomplete derivation of mural leaflets of the AV valves from endothelial cells in the Tie2-cre × ROSA26R hearts could be due to epicardial-derived contributions to the mitral and tricuspid valves in the mouse. In the OFT, the Tie2-cre × ROSA26R expression is complementary to that observed with the neural crest lineage marker, wnt1-cre, demonstrating distinct cellular origins of OFT septal and semilunar valve structures (Jiang et al., 2000). Together, these studies demonstrate that the vast majority of the cells in both semilunar and AV valves of the mouse are endothelial in origin.

Abnormal valve formation and morphology are features of many congenital valve anomalies (Bartram et al., 2001). Heart valve malfunction is often due to abnormalities in valve leaflet structure or ineffective supporting apparatus, including the chordae tendineae and papillary muscles. There is increasing evidence for genetic causes of congenital heart defects, including valve malformations associated with DiGeorge syndrome, trisomy 18, trisomy 21, and mutations in transcription factor and extracellular matrix molecule genes (Johnson et al., 1997; Pierpont et al., 2000; Bruneau, 2002; Hoffman and Kaplan, 2002). The identification of the cellular origins of the valve leaflets and supporting structures as well as the definition of the molecular regulatory pathways that control valvulogenesis will likely provide important insights into the etiology of congenital valve malformations.

EXPERIMENTAL PROCEDURES

Embryo Collection

Timed FVBN mouse matings were established with the morning of a vaginal mucous plug defined as E0.5. Embryonic hearts were dissected in phosphate buffered saline (PBS) at E12.5, E17.5, and 1 day postnatal and fixed overnight in 4% paraformaldehyde (PFA)/PBS at 4°C. White Leghorn chicken eggs (Spafas, Inc., Roanoke, IL) were incubated at 38°C under high humidity, and embryos were collected at days 6, 10, and 14. Hearts were dissected in PBS and fixed overnight in 4% PFA/PBS. After fixation and dehydration through a graded methanol/PTW (PBS, 0.01% Tween 20) series (25%, 50%, 75%, 95%), chicken and mouse hearts were either stored at −20°C in 100% methanol, embedded in paraffin wax and sectioned for immunostaining, or bisected with a clean razor blade for whole-mount immunohistochemistry or in situ hybridization analysis. All animal procedures were approved and performed in accordance with institutional guidelines.

Whole-Mount In Situ Hybridization

Whole-mount in situ hybridizations were performed as described by Wilkinson (1993) with reported modifications (Ehrman and Yutzey, 1999). Bisected hearts were treated with proteinase K for 12 min, and the color reactions were incubated from 1 to 5 hr using nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP; Roche). Hearts were fixed in 4% PFA after color development and stored in PBS at 4°C.

Digoxigenin UTP-labeled antisense RNA probes were generated specifically for both mouse and chicken. The generation of antisense riboprobes specific for chicken gata4, nkx2.5, tbx5, and bmp2 have been described previously (Laverriere et al., 1994; Schultheiss et al., 1995; Ehrman and Yutzey, 1999; Liberatore et al., 2000). The chicken scleraxis probe was a kind gift from Dr. Frederique Edom-Vovard (Edom-Vovard et al., 2001). The chicken tbx20 sequence was amplified from E3 heart cDNA by reverse transcriptase-polymerase chain reaction (PCR) using degenerate primers: forward, 5′-TGCTGRAAGTARTGRTG-3′; and reverse, 5′-GTGGAYAAYAAGAGATA-3′; and confirmed by sequencing corresponding to GenBank accession no. AB070554 (Iio et al., 2001). The 820-bp fragment was subcloned into pBluescript-SK and antisense riboprobe was synthesized with T3 polymerase from plasmid linearized with XhoI. Chicken smad6 was isolated from chicken E7 whole heart cDNA by using primers described previously (Yamada et al., 1999). The 554-bp sequence was subcloned into pGEM-T vector (Invitrogen), the plasmid was linearized with NcoI, and riboprobe was transcribed with SP6 polymerase.

The template plasmids for mouse nkx2.5, gata4, bmp2, and tbx5 have been described previously (Lints et al., 1993; Niswander and Martin, 1993; Chapman et al., 1996; Molkentin et al., 1997). The mouse tbx20 sequence was isolated from E10.5 ventricle cDNA using primers: forward, 5′-CCCAGTTCCGCTTTGCTTGCTCTC-3′; and reverse, 5′-CCCCACTTCCCACCCACCCTACTT-3′. The 1,500-base pair sequence corresponding to GenBank accession no. AF306667 was subcloned into pBluescript-SK (Kraus et al., 2001). The plasmid was linearized with XbaI, and T7 polymerase was used to synthesize the tbx20 antisense riboprobe.

Immunohistochemistry

Embryonic chicken hearts were isolated at days 10 and 14, fixed, and bisected as described above. Mouse monoclonal antibodies directed against collagen(pro-) type I (M-38; McDonald et al., 1986), type II collagen (C11C1; Holmdahl et al., 1986), myosin heavy chain (MF20; Bader et al., 1982), tenascin (M1-B4; Chiquet and Fambrough, 1984), and C1 and C5 antigens (George-Weinstein et al., 1988) were purchased from the Developmental Studies Hybridoma Bank developed with the National Institute of Child Health and Human Development (NICHD) and maintained by The University of Iowa. Antibodies were obtained as partially purified Ig and used at 1:200 dilution in blocking solution (1% goat serum/PBS). Immunolabeling was performed by using the Vectastain kit, and immunodetection was carried out according to manufacturer's recommendations (Vector Laboratories). Detection of antibody binding was visualized by using diaminobenzidine (DAB) substrate, including nickel (Vector Laboratories) for 2–30 min.

For histological sections, mouse hearts were fixed and dehydrated through an ethanol/PTW series, treated with xylene and embedded in paraffin wax. Seven-micrometer sections were cut and mounted onto superfrost slides (Fisher Scientific). After deparaffinization and hydration through a graded alcohol series, slides were treated with 3% hydrogen peroxide for 30 min and blocking solution as described previously (Kruithof et al., 2003). Primary antibodies were applied to mouse tissue sections for 2 hr at room temperature and washed in PBS thoroughly overnight at 4°C. Further detection was carried out as for immunohistochemistry in the chicken, with the incubation time of the secondary antibody extended to 1 hr.

BrdU Incorporation and Quantitation of Cell Proliferation

E6, E7, and E10 chicken embryos were labeled for 6 hr with 250 μl of BrdU labeling solution (Zymed), by means of injection into the egg airspace followed by subsequent sealing. Embryonic hearts, brains, and lungs were dissected in PBS and fixed for 2 hr in ice-cold Methacarn (60% methanol, 10% glacial acetic acid, 30% chloroform), dehydrated through a graded alcohol series, and embedded in paraffin. Five-micrometer sections were cut and BrdU-positive nuclei were identified by immunohistochemistry by using the BrdU detection kit (Zymed). BrdU incorporation was detected by using biotinylated mouse anti-BrdU primary antibody treatment followed by streptavidin-peroxidase–conjugated secondary antibody. Colorimetric reactions were performed by using DAB precipitation for 2 min and sections were counterstained with eosin. The proliferative index was determined as the number of positive staining BrdU labeled nuclei in the AV canal region, brain, and lung, divided by the total number of nuclei within that section. This was repeated by using three independent BrdU-labeled sections, from three different embryos.

Analysis of Tie2-cre Transgenic Mice

Tie2-cre transgenic mice (Koni et al., 2001) were purchased from The Jackson Laboratory (B6-Cg-Tg Tek-Cre) and crossed with ROSA26 lacZ reporter mice (Soriano, 1999). PCR analysis of genomic DNA was used to genotype all adult and embryonic mice. Genomic DNA for genotyping was derived from tail clips or embryonic yolk sacs. The lacZ transgene was detected in litters by using specific DNA primers: forward, 5′-TGGGGAATGAATCAGGCCACGG-3′; and reverse, 5′-GCGTGGGCGTATTCGCCAAGGA-3′ (Searcy et al., 1998). The cre transgene was detected using specific primers, as described by The Jackson Lab: forward, 5′-GCGGTCTGGCAGTAAAAACTATC-3′; and reverse, 5′-GTGAAACAGCATTGCTGTCACTT-3′. Tie2-cre/ROSA26R mice were collected at E12.5 and 1 day postnatal. β-Galactosidase (β-gal) expression in the transgenic mice was analyzed in bisected hearts by using X-gal detection of β-gal activity as previously described (Searcy et al., 1998). For histology, X-gal stained embryos were dehydrated through a graded isopropanol/PBT series and paraffin wax embedded. Twelve-micrometer sections were cut, dehydrated, and mounted in Cytoseal 60 (Richard-Allen Scientific) for microscopic analysis.

Acknowledgements

We thank Paul Bushdid; Christine Liberatore; Alexander Lange; Timothy Plageman, Jr.; Woody Benson; Rob Gourdie; and members of the Division of Molecular Cardiovascular Biology for technical support and scientific advice. Artwork was generated with the help of Andreas Lange. K.E.Y. was funded by an Established Investigator Award from the American Heart Association and NIH grant HL66051.

Ancillary