The Endocardium and Heart Valves

Abstract

Endocardial cells are specialized endothelial cells that, during embryogenesis, form a lining on the inside of the developing heart, which is maintained throughout life. Endocardial cells are an essential source for several lineages of the cardiovascular system including coronary endothelium, endocardial cushion mesenchyme, cardiomyocytes, mural cells, fibroblasts, liver vasculature, adipocytes, and hematopoietic cells. Alterations in the differentiation programs that give rise to these lineages has detrimental effects, including premature lethality or significant structural malformations present at birth. Here, we will review the literature pertaining to the contribution of endocardial cells to valvular, and nonvalvular lineages and highlight critical pathways required for these processes. The lineage differentiation potential of embryonic, and possibly adult, endocardial cells has therapeutic potential in the regeneration of damaged cardiac tissue or treatment of cardiovascular diseases.

The developing heart tube consists of an inner layer of endocardial cells and an outer layer of myocardium (Tian and Morrisey 2012). The endocardial cell layer not only provides a physical barrier to protect the inner cardiac tissue against the hemodynamic circulation, but also serves as a focus for dynamic signaling, as well as a source of different cardiovascular cell types (Zhang et al. 2018).

Upon molecular stimulation emanating from outer cardiomyocyte cells, a subset of inner endocardial cells within the atrioventricular canal (AVC) and outflow tract (OFT) regions undergo endothelial-to-mesenchyme transformation (EMT) to give rise to a population of precursor cells that later form the mature valve structures (Markwald et al. 1975, 1977; de Lange et al. 2004; Lincoln et al. 2004). Alternatively, “nonvalvular” endocardial cells primarily give rise to the endothelial lining within the heart including the atria, ventricles, and sinus venosus. Furthermore, fate-mapping studies demonstrate that these endocardial cells contribute to the majority of coronary vascular endothelial cells (Tian et al. 2015) and serve as novel progenitors for mural cells of the heart (Wang et al. 2016), potentially cardiomyocytes (Van Handel et al. 2012), and other lineages (Zhang et al. 2018).

During development, reciprocal molecular communication between the endocardial and myocardial cell layers is essential for valve and ventricular chamber formation, and improper cross talk leads to premature lethality. Alternatively, survivors may develop forms of congenital heart disease, or noncompaction of the myocardium (Grego-Bessa et al. 2007; de la Pompa and Epstein 2012; Bressan et al. 2014; D'Amato et al. 2016; Zhang et al. 2018). Some of these early signaling pathways active between the two cell layers are similar in developing valves and ventricular chambers (Notch), while more specialized pathways are additionally required for coronary artery development (Vegf), and EMT in developing valves (Tgfβ, Bmp, Wnt).

This review will focus on the embryonic contribution of the endocardium to the cardiac valves and myocardial lineages, and discuss the essential signaling pathways critical for these processes. In addition, the speculative role of endocardial-derived cells in the onset of congenital and adult cardiovascular disease will be discussed, with a view of developing potential endocardial-based therapeutics.

DEVELOPMENTAL ORIGINS OF ENDOCARDIAL CELLS

During presomite stages of embryogenesis, two progenitor cell populations form parallel horseshoe-shaped clusters in the splanchnic layer of the lateral plate mesoderm, then migrate and give rise to the primitive heart tube consisting of two layers of specified cell lineages: the outer layer of myocardium and the inner layer of endocardium. The two progenitor cell populations that give rise to the lineages are known as the first (or primary) and secondary heart fields (SHFs), and are the earliest cardiac progenitor cells (CPCs) to contribute to the developing heart. The primary heart field will later contribute to the left ventricle and majority of the atria, and the SHF will give rise to the mature right ventricle and OFT (Zhang et al. 2018). However, the fate of these progenitors also includes the endocardium, the conduction system, and the valves. As development continues, additional progenitors from alternative sources are added to the heart. This includes progenitors originating from the proepicardium that comprise the epicardium and differentiate into interstitial fibroblasts, vascular smooth muscle cells, and endothelial cells of the vasculature and some myocytes. Finally, cardiac neural crest cell progenitors migrate in and contribute to the smooth muscle and mesenchyme cell populations within the OFT region (Brade et al. 2013).

There are several lines of evidence using lineage tracing and differentiated embryonic stem cell experiments to suggest that the origin of endocardial and myocardial cells is common (Nakano et al. 2016). Tracking of Nfatc1 CPCs (Kattman et al. 2006), Flk1⁺ mesoderm (Ema et al. 2006), and cells of the SHF (Mef2-AHF^Cre, Isl1^Cre) (Cai et al. 2003; Verzi et al. 2005; Moretti et al. 2006) includes both myocardial and endocardial fates. However, their localization may be important for specification, as single-cell labeling studies in zebrafish have mapped CPCs to the lateral margins of the zebrafish blastula, with cardiac potential dissipating toward the dorsal region, and endocardial cells being more restricted to the most ventral region (Lee et al. 1994). Although during later stages, the endocardial precursor are positioned more dorsal to reside at the lateral regions of the embryo mixed with myocardial progenitors (Keegan et al. 2004). More recent studies using Mesp1^Cre lines (mesoderm) irreversibly labeled low numbers of cells and reported rare recombination events that give rise to either myocardial or endocardial clones (Lescroart et al. 2014), but not both in the first heart field, while clones in the SHF comprise both myocardial and endocardial lineages (Moretti et al. 2006). These studies suggest limited lineage commitment that might be stage-dependent.

Specification of the endocardium has been shown to be induced by the presence of endodermal cells, from which emanates a “cocktail” of mediators including bone morphogenetic proteins (BMPs) and Wnt inhibitors (Schultheiss et al. 1997; Marvin et al. 2001). In contrast, Nkx2-5 specifies the myocardial cells, although some endocardial tissue also expresses low levels of this transcription factor (Stanley et al. 2002; Paffett-Lugassy et al. 2013). As development progresses, the endocardium becomes molecularly distinguishable from the myocardium based on expression of markers such as VEGFR-2, Flk1 CD31, and VE-cadherin, that become down-regulated in specified myocardial cells (Yamaguchi et al. 1993). In addition, Etv2 has been shown to be necessary and sufficient for the induction of endothelial identity and deemed as the earliest restricted regulator of endothelial development (De Val et al. 2008; Palencia-Desai et al. 2011; Wong et al. 2012). Following specification and differentiation of the endocardial (and myocardial) cells, the cardiac heart tube forms consisting of the outer myocardium and inner endocardium, and this localization is maintained during looping and beyond. Figure 1 depicts this spatial relationship in the developing zebrafish heart at 2 and 4 days postfertilization, which is structurally distinct from mammalian development. From here, the endocardial cells receive diverse molecular cues dependent on their location within the looped heart tube that will drive their fate toward formation of the valves, development of the coronary vasculature, or maturation of the myocardium.

Figure 1.

Endocardial cells of the developing zebrafish heart. (A) Schematic to show the anatomical location of endocardial cells in the developing zebrafish heart at 2 days postfertilization. In the mouse, this process occurs at ∼E9.0 in a structurally distinct heart anatomy. (B) 4 days postfertilization in zebrafish development marks the development of trabeculae in the primitive ventricle. In the mouse, this process occurs at ∼E11.5 (figure kindly generated by Mandy Root-Thompson).

CELL-FATE ANALYSIS OF ENDOCARDIAL CELLS DURING EMBRYONIC DEVELOPMENT

Following looping of the heart tube, the endocardial cells give rise to several diverse lineages throughout the cardiovascular system. These include coronary endothelium, cardiomyocytes, and endocardial cushion mesenchyme cells, and these will be discussed in this review. However, in addition, it has been reported that endocardial cells also demonstrate plasticity toward liver vasculature, mural cells, adipocytes, fibroblasts, and hematopoietic cells, which are reviewed elsewhere (Zhang et al. 2018) and summarized in Figure 2.

Figure 2.

The fate of cardiac endocardial cells in development, disease, and regeneration. The cardiac endocardial cell has many fates. During early stages of cardiogenesis, “specialized” endocardial cells line sites of the primitive valve structures (termed nonvalvular) (pink) and myocardium (blue). Within these regions, these endocardial cells give rise to multiple, diverse cell fates during development (green arrows) and have been associated with processes involved in adult disease and regeneration (red arrows).

Heart Valve Development: Endocardial Cushion Formation

In the late 1970s, histological and ultrastructural analysis characterized the morphology of “swellings,” termed endocardial cushions in the OFT and AVC regions (Markwald et al. 1975, 1977). It was found that the AVC contains four cushions—the inferior and superior localized central and two lateral cushions—while the OFT cushions are divided into distal and proximal according to the curvature of the primitive vessel. Based on appearance, AV and OFT cushions were described as containing “mesenchymal cells beneath the endothelium that lack a basal lamina,” and longitudinal analysis of these structures suggested that they morph into the valvular structures (Markwald et al. 1977). Early experiments in chick and quail embryos were the first to provide evidence that mesenchyme cells of the cushion are endothelial-derived, based on positive immunoreactivity of the quail endothelial cell marker QH1, in cushion mesenchyme (Markwald et al. 1990), and invasion of cushion cells plated on collagen gel lattices (Hay et al. 1983). Additional electron microscopy further captured the transformation of endothelial cells into mesenchyme cells as noted by cell–cell separation and migration into the cushion space rich in hyaluronan matrix (Markwald et al. 1990). Several years later, genetic tracing using the Tie2-Cre transgene confirmed the original microscopic analyses and proved that the endothelium is the major source of mesenchyme cells within the endocardial cushions (de Lange et al. 2004; Lincoln et al. 2004). Furthermore, the utilization of this endothelial-specific transgene validated that the mechanism underlying the generation of the mesenchyme cell population at the cushion stage was via EMT.

Endothelial cell tracking showed that EMT gives rise to the majority of mesenchymal valve precursor cells particularly in the AVC from E9.0 (de Lange et al. 2004; Lincoln et al. 2004) and the process is critical, as defects lead to embryonic lethality. However, further examination of the fate of Tie2-Cre cells in valve structures showed that Cre recombination does not occur in all valve cells, suggesting additional cell sources may contribute to the valve structures during the course of development (Lincoln et al. 2004). At embryonic day (E)14.5 in the mouse, mesenchyme cells within the septal leaflets of the AVC (which presumably arise from the superior and inferior cushions) were labeled with the Tie2-Cre transgene; however, very few were positive in the parietal leaflets (originating from the lateral cushions) (Wessels et al. 2012). Interestingly, these nonseptal leaflets were populated by cells likely derived from the epicardial lineage, as indicated by Wt1-Cre expression (Wessels et al. 2012). In the OFT, the epicardial lineage is not involved, but EMT is also the mechanism of proximal cushion formation (Wu et al. 2011). In the distal OFT cushions, the cardiac neural crest cell lineage is most prominent, as well as SHF (Mef2c-Cre) cells (Li et al. 2000; Jiang et al. 2002; de Lange et al. 2004; Verzi et al. 2005; Jain et al. 2011). While these developmental studies are informative, the field has yet to delineate the purpose or function of differential cell lineage contributions to the primitive AV and semilunar valve structures.

Heart Valve Development: Regulators of Endocardial Cushion Formation

While endothelial-derived cells are not the only source of cushion mesenchyme cells, their contribution to cushions is the most well-studied process. The process of EMT is complex and each step is tightly regulated by a complex network of growth factors, transcription factors, and intermediate signaling molecules that cross talk between multiple cell types, particularly the myocardium and endocardium.

Transforming Growth Factor (TGF)-β Signaling

The Tgfβ superfamily consists of the Tgfβ and BMPs. The canonical signaling pathways function through Tgfβ or BMP ligand binding to a type II receptor that recruits and phosphorylates the type I receptor leading to complex formation with Smad proteins that translocate to the nucleus and function as transcription factors on target genes. Tgfβs, BMPs, and their downstream signaling mediators are the most well-characterized regulators of EMT and their signaling is required in multiple cell lineages for the initiation and progression of EMT. This has been well described in early studies using the chick model to show a requirement for myocardial-derived Tgfβ2 and Tgfβ3 to initiate EMT (Nakajima et al. 2000; Mercado-Pimentel and Runyan 2007). A summary of endocardial cushion defects reported from genetic studies of Tgfβ signaling mediators in mice are summarized in Table 1.

Table 1.

Endocardial cushion phenotypes in mouse models harboring mutations in Tgfβ signaling mediators

Target gene	Model	Endocardial cushion phenotype	References
ALK2	Tie2-Cre deletion	Small endocardial cushions at E10.5, remain impaired at E14.5	Wang et al. 2005
ALK3	Tie2-Cre deletion	Absence of AV endocardial cushions at E9.5	Ma et al. 2005
ALK5	Tie2-Cre, Gata5-Cre deletion	Hypoplastic endocardial cushions	Sridurongrit et al. 2008
Bmp2	Nkx2.5-Cre deletion	Impaired endocardial cushion formation at E9.5	Ma et al. 2005
Bmp4	Nkx2.5-re deletion	Reduced endocardial cushion growth at E11.5. Embryonic lethal at E13.5–E18.5	Liu et al. 2004
	Bmp4loxp-lacZ/tm1	ASD, partial AVCD	Jiao et al. 2003
	cTnT-Cre; Bmp4loxp-lacZ/loxp-lacZ	Complete AVCD due to underdevelopment of AV cushions, DORV	Saxon et al. 2017
Bmp5, 7	Global double knockout	Absence of endocardial cushions. Embryonic lethal at E9.5–E10.5	Solloway and Robertson 1999
Bmp6, 7	Global double knockout	Underdeveloped OFT endocardial cushions at E11.5. Embryonic lethal at E9.5–E15.5	Kim et al. 2001
Bmp7	Global knockout	Hypoplastic OFT endocardial cushions	Liu et al. 2004
Latent Tgfβ-binding protein 1	Global knockout	Impaired endocardial cushion formation	Todorovic et al. 2011
Noggin	Global knockout	Enlarged endocardial cushions	Choi et al. 2007
	Global transgenic	Hypoplastic endocardial cushions	Snider et al. 2014
Smad4	Tie2-Cre deletion	Disturbed endocardial cushion formation (decreased mesenchyme cell population)	Moskowitz et al. 2011
Smad6	Global knockout	Hyperplastic valves	Galvin et al. 2000
Tgfβ1	Global knockout	Disorganized valves. Embryonic lethal at E9.5–E11.5	Dickson et al. 1993; Letterio et al. 1994
Tgfβ2	Global knockout	Incomplete valve formation	Sanford et al. 1997; Bartram et al. 2001
TgfβRII	cTnT-Cre, cGata6-Cre, Mlc2v-Cre deletion	Common AV valve	Jiao et al. 2003, 2006; Robson et al. 2010

(AV) Atrioventricular, (ASD) atrial septal defect, (AVCD) atrioventricular canal defect, (DORV) double-outlet right ventricle, (OFT) outflow tract.

In addition to targeting downstream transcription factors, Tgfβ cross talks with other signaling pathways during endocardial cushion development including β-catenin. Tie2-Cre deletion of β-catenin in endothelial and endothelial-derived cells inhibits Tgfβ-mediated induction of EMT in mice (Liebner et al. 2004), suggesting a requirement of the canonical Wnt signaling in this process. Tbx20 was later shown to regulate Lef1, a key transcriptional mediator of Wnt/β-catenin signaling, during early stages of valve formation (Cai et al. 2013). In zebrafish, overexpression of a secreted Wnt inhibitor (Dickkopf1) blocks cushion formation, while truncation mutants develop hyperplastic cushions as a result of increased cell proliferation (Hurlstone et al. 2003). The notion that Wnt signaling regulates VIC proliferation is continued in the avian system where the Wnt receptor Frzb and Wnt9a promote mesenchyme cell number in the AV cushions (Person et al. 2005). Consistently, expression patterns of other Wnt pathway genes including Wnt2a, LEF1, and Fzd2 in murine mesenchyme cells of the AVC and OFT cushions suggest conserved mechanisms across species.

Notch Signaling

Activation of Notch signaling requires interaction of a transmembrane receptor (Notch1-4) to a specific ligand (Jagged [Jag-]1, 2, Delta-like [Dll]1, 3, 4) presented on neighboring cells. Upon receptor–ligand interaction, Notch receptors undergo proteolytic processing and release the Notch intracellular domain (NICD). NICD is then translocated to the nucleus where it binds the DNA-binding corepressor RBPJ (recombination signal-binding protein for immunoglobulin κ J region), and recruits the coactivator MAML1 (Mastermind-like 1) to derepress and activate transcription of downstream target genes including HES and HEY families of transcription factors (Iso et al. 2003). Components of the Notch signaling pathway have been implicated in EMT, many of which are required for murine endocardial cushion formation in vivo, and these are shown in Table 2.

Table 2.

Endocardial cushion phenotypes in mouse models harboring mutations in Notch-signaling mediators

Target gene	Model	Endocardial cushion phenotype	References
Hey1, HeyL	Global double knockout	Impaired EMT in E9.5 AV canal explant cultures	Fischer et al. 2007
Hey2	Predicted null allele (Hey2lacz)	EMT occurs, but cushion maturation is impaired	Donovan et al. 2002
Jag1	VE-cadherin-Cre deletion	Hypoplastic endocardial cushions at E10.5, temporarily delayed EMT	Hofmann et al. 2012
MAML	Dominant negative (dn), Isl1-Cre deletion	Hypocellular OFT endocardial cushions at E10.5	High et al. 2009
MAML	Tet-inducible dn under control of VE-cadherin promoter	Tet-treatment at E10.5, impaired EMT at E11.5 and E12.5	Chang et al. 2014
NICD1	Mesp1-Cre transgenic	Ectopic cell masses in endocardial cushions at E9.5	Watanabe et al. 2006
Notch1	Global knockout	Impaired EMT in E9.5 AV canal explant cultures	Timmerman et al. 2004
	VE-cadherin-Cre deletion	Delayed EMT at E10.5	Hofmann et al. 2012
RBPjk	Global knockout	EMT is initiated but endocardial cushions lack mesenchyme cells at E9.5	Timmerman et al. 2004

(EMT) endothelial-to-mesenchyme transformation, (OFT) outflow tract, (AV) atrioventricular.

Vascular Endothelial Growth Factor (VEGF)

VEGF A is a potent cytokine highly expressed by the myocardium and valve endothelial cells (VECs) prior to cushion formation. However, expression of the ligand and its receptors (VEGFRs) becomes restricted to the endocardium once cushion formation has been initiated (Miquerol et al. 1999; Dor et al. 2001). Given the role that VEGF plays in response to hypoxia in the vascular system, it has been suggested that this temporal and spatial change in VEGF expression during EMT correlates with oxygen levels in the developing embryonic heart (Dor et al. 2003). Studies in chick, mouse, and zebrafish demonstrate that endocardial VEGF signaling contributes to early valvulogenesis by promoting the VEC phenotype and proliferation, thereby maintaining a proliferative population of endothelial cells throughout EMT and cushion maturation (Carmeliet et al. 1996; Ferrara et al. 1996; Miquerol et al. 2000; Dor et al. 2001; Combs and Yutzey 2009). From these studies, it is apparent that VEGF levels of expression must be tightly controlled during endocardial cushion formation; too much VEGF inhibits EMT, while too little attenuates VEC proliferation, thereby decreasing VEC availability for transformation (Miquerol et al. 2000; Dor et al. 2001). Interestingly, it appears that VEGF signaling is required for complete EMT in the OFT, but has less important roles in AVC endocardial cushion formation (Stankunas et al. 2010). This observation has been attributed to the differential mediation of VEGF signaling through VEGFR1 in the OFT, and VEGFRII in the AVC (Stankunas et al. 2010). Further downstream, VEGF regulates the transcription factor nuclear factor of activated T cells cytoplasmic 1 (NFATc1) in both developing and mature VECs to promote VEC proliferation (Johnson et al. 2003; Combs and Yutzey 2009). While Nfatc1^−/− mice undergo successful endocardial cushion EMT (de la Pompa et al. 1998; Ranger et al. 1998; Lange and Yutzey 2006), a more recent role has emerged in fate decisions of VECs during transformation stages.

Nuclear Factor of Activated T Cells 1 (Nfatc1)

During a tightly regulated temporal window, VECs break cell–cell contacts, transform, and migrate into the underlying cardiac jelly. However, after ∼E13 in the mouse when cushion formation is complete, it is believed that VECs no longer transform, despite maintained expression of many EMT inducers. For many years, it has remained a challenge to identify the molecular cues that determine the fate of individual VECs during this window of events: which VECs transform and which VECs remain in the endothelium. In the heart, Nfatc1 is expressed in VECs during endocardial cushion formation (de la Pompa et al. 1998; Ranger et al. 1998; Lange and Yutzey 2006); however, a role during early stages of valve formation has remained largely elusive. A study by Wu et al. (2011) identified a transcriptional enhancer region (En) within Nfatc1 that may play an important role in regulating VEC fate during EMT. In contrast to Nfatc1 expression in all VECs, the study used reporter mice to reveal a small subpopulation of VECs that express the Nfatc1-En. VECs positive for Nfatc1-En expression do not appear to undergo EMT and remain part of the proliferative endothelial cell population surrounding the developing valve. Further, Nfatc1 inhibits EMT in a cell-autonomous manner by suppressing transcription of Snai1 and Snai2 (Wu et al. 2011). This regulation of VECs by Nfatc1 is the first to provide a mechanism for cell fate decisions of endothelial cells during endocardial cushion EMT.

Sox9

The SRY transcription factor Sox9 is activated during initiation of EMT in the AVC and OFT regions (Akiyama et al. 2004). Targeted deletion in the Tie2-Cre lineage leads to premature lethality at ∼E12.5, associated with small endocardial cushions and reduced mesenchyme cell proliferation (Lincoln et al. 2007), thereby suggesting a critical role for Sox9 in generating the valve precursor cell pool during early valvulogenesis.

Heart Valve Development: Post-EMT Maturation

Once the valve precursor pool of mesenchyme cells has been established (∼E14.5 in the mouse), be it from EMT or alternative cell migration, the endocardial cushions undergo extensive remodeling as they elongate and thin into valve primordia. Cell proliferation is significantly reduced at this time, although proliferative cells remain enriched at the tip (Hinton et al. 2006). Concurrently, precursor cells lose mesenchymal molecular markers including Twist1 but gain the activated myofibroblast marker, smooth muscle α actin (αSMA) (Horne et al. 2015). This phenotypic change is thought to reflect transition toward an activated valve interstitial cell (aVIC) that likely mediates physiological remodeling of the extracellular matrix (ECM) during this stage of maturation. This includes breakdown of primitive cardiac jelly and synthesis of new ECM components that will later form the fibrosa, spongiosa, and ventricularis layers within the mature leaflet. The molecular regulators of mid-to-late valve development are largely unknown, but pathways important for EMT, including Tgfβ, Bmp, Wnt, and Sox9, are also active during remodeling and have been shown to play differential roles at this stage (Lincoln et al. 2007; Azhar et al. 2011; Cai et al. 2013; Saxon et al. 2017). More recently, additional regulators have been reported, including hypoxia, cadherin-11 (cell adhesion), the chemokine receptor CXCR7, and the matrix remodeling enzyme ADAM17 (Yu et al. 2011; Wilson et al. 2013; Bowen et al. 2015; Amofa et al. 2017). While mouse models with targeted genetic disruptions that result in valve remodeling defects are viable, it is considered that defects at this stage could underlie congenital valve malformations present at birth, or potentially acquired disease manifested later in life.

The primitive valve continues to grow and mature after birth, and in the mouse the three layers of predominant ECM components [collagen (fibrosa), proteoglycan (spongiosa), elastin (ventricularis/atrialis)] are apparent between postnatal day (PND) 7 and 10. At this time, cell proliferation is ∼16.3% in VECs, and ∼15.2% in VICs (based on 7-h pulse change of EdU) and cell division remains at this frequency until ∼PND4 (Hinton et al. 2006; Anstine et al. 2016). Concurrently, VICs lose αSMA but maintain Vimentin expression, suggesting a transition toward a quiescent (or nonactivated) fibroblast-like cell type (Horne et al. 2015). This quiescent phenotype is maintained throughout life in the absence of disease with cell proliferation estimated at a lower frequency of ∼2.0% VECs and ∼1.1% VICs (7-h pulse chase) (Anstine et al. 2016). This level of normal adult cell turnover in the valve might be considered high compared to other cardiac cell types (<1% in cardiac myocytes) (Yutzey 2017); however, the overall valve cell number does not appear to increase with aging (but matrix synthesis does) and, therefore, cell death likely occurs at a similar frequency. The mechanism for maintaining adult valve cell population during the normal wear and tear of aging not only relies on resident cell proliferation, but the contribution of extracardiac cells. Using mouse models to fate map CD45-positive cells, we and others have shown that, under homeostatic conditions, ∼2.3% of the valve cell population is derived from this lineage at postnatal stages, and up to 10.3% at 6 weeks (Visconti et al. 2006; Hajdu et al. 2011; Bischoff et al. 2016; Anstine et al. 2017). Similar to valve development, it is critical that the adult valve cell populations are maintained throughout life and it can be appreciated that an imbalance in VEC or VIC number or function might lead to perturbations in ECM homeostasis and subsequent biomechanical function.

Mature Heart Valve Structure–Function Relationships

Following valvulogenesis, the heart valve structures mature into dynamic structures that must open and close over 100,000 times a day to regulate unidirectional blood flow from the left ventricle to the rest of the body. Active developmental processes in the embryonic regions of the AVC results in the formation of the mitral and tricuspid valves that separate the atria from the ventricles. While in the OFT, the aortic and pulmonic semilunar valves separate the ventricles from the great arteries. Although the functional demand of each valve set is similar, their anatomies and, as discussed, cellular origins are different. The AV valves are situated in the AVC separating the atria from the ventricles. Structurally, these valves consist of two (mitral) or three (tricuspid) leaflets, with external supporting chordae tendineae that attach the leaflet to papillary muscles within the ventricles (Dutta and Lincoln 2018). In contrast, the semilunar valves located at the base of the aorta and pulmonary trunk are comprised of three leaflets termed cusps and lack external chordae and papillary muscles, although a unique internal support structure has been described (Hinton et al. 2006). It is the coordinated movement of these dynamic valvular structures that maintain unidirectional blood flow during the cardiac cycle.

Opening and closing of the valve leaflets or cusps is largely achieved by three organized layers of ECM arranged according to blood flow direction, and each provides a unique biomechanical property to withstand the complex hemodynamics experienced with every cardiac cycle (Hinton et al. 2006). The fibrosa layer is situated furthest away from blood flow and is largely composed of bundles of aligned fibrillar collagens that provide strength. Organized elastic fibers make up the ventricularis/atrialis (semilunar/AV) layer situated adjacent to blood flow. This matrix component allows for valve extension and recoil during each heartbeat (Vesely 1998). The proteoglycan-rich spongiosa layer is “sandwiched” between the fibrosa and ventricularis and provides compressibility in these load-bearing regions (Grande-Allen et al. 2004). As discussed, this intricate ECM arrangement is laid down during post-EMT stages of development and maintained in adults by a heterogeneous population of VICs that are largely considered quiescent under healthy conditions. In addition to the VIC population, a single-cell layer of VECs overly the leaflets and cusps and serve as a physical barrier against the hemodynamic environment, and emanate and receive signals with neighboring VECs and underlying VICs to maintain structural (and therefore functional) integrity throughout life (Dutta and Lincoln 2018).

Myocardial Chamber Maturation

In addition to valvular structures, the developing endocardium also gives rise to several other mature cell lineages within the myocardial chamber.

Contributions to Coronary Endothelial Cells

The primitive heart tube does not require coronary vessels and the thin layer of myocardial cells obtain oxygen and nutrients by diffusion (Riley 2010; Pérez-Pomares and de la Pompa 2011). However, when the heart increases in size, coronary blood vessels start to emerge on the ventricle surface as immature vascular plexus. They undergo branching and expand to cover and infiltrate the myocardium to support the increasing oxygenation demand (Waldo et al. 1990; Zeini et al. 2009; Red-Horse et al. 2010). The coronary vessel structures are composed of an inner single layer of endothelial cells, surrounded by mural cells including smooth muscle cells and an outermost adventitia layer containing fibroblasts and other cell types (Passman et al. 2008; Armulik et al. 2011; Stenmark et al. 2013; Baeyens et al. 2016; Sharma et al. 2017). The origin of the endothelial cell component of coronary vessel development has been an area of active interest over the last two to three decades. Findings from early studies using quail chick transplant experiments showed that donor-derived epicardium gives rise to coronary vessel endothelial cells (and coronary smooth muscle cells and fibroblasts). However, later studies in mice using Cre recombination (Wt1-Cre, Gata5-Cre, Tbx18-Cre) showed that epicardial cells give rise to myocardial stroma and vascular smooth muscle cells, but few, if any coronary endothelial cells (Merki et al. 2005; Cai et al. 2008; Zhou et al. 2008). Subsequent histological and clonal analysis in mice showed that coronary arteries (capillaries and veins) are derived from venous endothelial sprouts of the sinus venosus that express high levels of the receptor protein for Apelin (APJ) and Elabela (ELA) (Red-Horse et al. 2010; Chen et al. 2014). These endothelial sprouts migrate along the subepicardium and invade into the underlying myocardium, forming vessels over the surface of the heart starting at E11.5 (in the mouse), a process that requires VEGFC and the ELA-APJ signaling axis, as well as epicardial-derived CXCL12, which interacts with CXCR4 in the sinus venosus endothelium (Tian et al. 2013; Chen et al. 2014; Cavallero et al. 2015; Sharma et al. 2017).

The sinus venosus origin of the majority of coronary endothelial cells has been challenged, and in 2008, Wu et al. reported that intramyocardial coronary arterial endothelial cells are derived from ventricular endocardial cells that directly migrate outward into the compact myocardium to form ∼E15.5, a process shown to require VEGFA, HIF-1a, and Sox17 (Wu et al. 2012; Sharma et al. 2017). Worthy of mention, aside from the sinus venosus origin, the Krasnow group did report a similar, but small contribution from the endocardial lining of the chamber myocardium (Red-Horse et al. 2010). The idea of contribution from ventricular endocardial cells was later questioned when data supporting the idea that migrating subepicardial endothelial cells contribute to the majority of coronary blood vessels in the embryonic ventricular free wall, particularly in the ventricular septum, which are largely void of sinus venosus–derived cells (Tian et al. 2013; Chen et al. 2014; Cavallero et al. 2015). The complexity in defining the mechanisms that regulate endocardial-to-vascular endothelial cell transition remain an area of interest.

Contributions to the Cardiomyocyte Population

The transcription factor Scl is required for differentiation of hemogenic endothelial cells from mesoderm (Lancrin et al. 2009), and hemogenic activity has been reported in a subset of endocardial cells in the developing OFT and atria (Nakano et al. 2013). Interestingly, loss of Scl^−/− in the developing endocardium leads to expression of cardiac transcription factors and structural proteins, and results in beating cardiomyocytes (Van Handel et al. 2012). However, the direct contribution of endocardial cells to the cardiomyocyte lineage in the developing embryo has not been shown, although lineage tracking has suggested contribution from 2 weeks after birth (Fioret et al. 2014).

In addition to the speculations that endocardial cells have the potential to directly contribute to the cardiomyocyte lineage, there is a wealth of data to show that they are necessary for cardiomyocyte maturation during ventricular chamber development. Trabeculation is the process by which ventricular cardiomyocytes form protrusions and grow toward the ventricular lumen. In mice, trabeculation requires cell division and studies have suggested that this is largely polarized in cardiomyocytes aligned perpendicular to the heart lumen (Passer et al. 2016). However, in zebrafish, trabeculation occurs via delamination of cardiomyocytes from the abluminal surface into the trabecular layer (Staudt et al. 2014).

Trabeculation is tightly regulated by cross talk with the overlying endocardial cells, which are separated by cardiac jelly (Fig. 1B), and several signaling pathways have been implicated. One of the most well studied is Notch, which functions in the endocardial cells to molecularly communicate with the ventricular cardiomyocytes (D'Amato et al. 2016). Genetic abrogation of the receptor Notch1 or downstream mediator Rbpjk leads to embryonic lethality at E10.5 associated with a hypoplastic ventricle and impaired trabeculation (Grego-Bessa et al. 2007). This phenotype is due to the known role that Notch plays in molecularly regulating cardiomyocyte proliferation and trabecular differentiation via Efnb2, Nrg1/ErbB, Bmp10, and Hand2 (Meyer and Birchmeier 1995; Grego-Bessa et al. 2007; VanDusen et al. 2014; Samsa et al. 2015). More recently, it was also shown that Notch-mediated regulation of the ECM within the cardiac jelly is also critical for normal trabeculae formation (Del Monte-Nieto et al. 2018).

The Role of Adult Endocardial Cells in Cardiovascular Pathology

Embryonic Endocardial Defects in the Onset of Human Disease

Alterations in embryonic heart development can lead to premature lethality, or if those affected survive, there is a risk of congenital malformations. While the utilization of animal models has identified several critical signaling pathways important for endocardial maturation toward formation of the four-chambered heart, few are confirmed to be causative of human disease. However, one of the most well described is the Notch pathway, which as discussed is critical for ventricular development. Systemic or endocardial-specific deletion in mice (Rbpjk, Notch1, Mib1, Jarid2) frequently result in impaired trabeculation similar to human forms of cardiomyopathies (for review, see D'Amato et al. 2016). Indeed, sequencing of NOTCH-related genes (MIB1) has revealed mutations in familial pedigrees of left ventricular noncompaction patients. Furthermore, to date, NOTCH1 remains one of the few known genes to underlie congenital heart valve malformations, including the bicuspid aortic valve, whereby affected patients are born with two, rather than three cusps (Garg et al. 2005). These studies demonstrate a direct causative role of altered Notch signaling with structural abnormalities derived from the embryonic endocardium in human patients.

Endocardial Cell Plasticity after Birth in the Myocardium

The contribution of endocardial cells to the embryonic heart has been discussed and more recent studies have shown that neonatal endocardial cells continue to give rise to vascular endothelial cells within the coronary vessels after birth (Tian et al. 2014). However, the endocardial contribution to arteriogenesis following injury or disease in the adult is still not fully understood. In adult zebrafish, endocardial-mediated signaling (Notch) is required for repair and regeneration following both ventricular apex amputation and cryoinjury (Kikuchi et al. 2011; Münch et al. 2017), supporting a positive role for the endocardium in repair of the damaged heart. In mice, work by Miquerol et al. (2015) further suggests that endocardial cells respond to injury, as lineage mapping studies led to the conclusion that the endocardium could be a source of vascular endothelial cells following myocardial infarction (MI). However, more recent studies with the adult endothelium–specific Npr3-CreER mice have shown that adult endocardial cells have a very small contribution to vascular endothelial cells after MI, as well as other types of injury (Zhang et al. 2016). However, it appears that adult endocardial cells trapped in the injured myocardium may have the potential to give rise to new blood cells (Zhang et al. 2016).

Endocardial Cell Plasticity after Birth in Heart Valves

In comparison to the myocardium, the contribution of embryonic endocardial cells to adult disease and the plasticity of adult endocardial cells is not so advanced in the field of heart valve biology. Other than NOTCH1, and known disease-causing mutations in syndromic disease, the etiology of congenital (or adult) valve disease is wide open. However, emerging studies including those from our group have highlighted a critical role for valve endocardial cells in preventing valve disease (Bosse et al. 2013; Hjortnaes et al. 2015; Huk et al. 2016). In vitro, the presence of VECs (as well as human umbilical vein endothelial cells) is sufficient to prevent abnormal activation, and calcification of VICs under pathological stimulus (Bosse et al. 2013; Hjortnaes et al. 2015; Huk et al. 2016). Complementary in vivo studies further showed that this was due to Tgfβ1 (Huk et al. 2016) and nitric oxide (Bosse et al. 2013) signaling emanating from the VECs to the underlying VICs. In addition, we (Anstine et al. 2016) have reported morphological and functional changes in VECs with aging that might indicate that calcification in those aged 65 or over are influenced by limited nitric oxide bioavailability, reduced metabolism, and increased permeability of the endothelium. It is also considered that the repair or regenerative potential of the VEC population is impeded by age. While this has not been examined, clonal analyses of ovine VECs has demonstrated some degree of plasticity given their potential to undergo TGFβ-induced EMT, and differentiate toward chondrogenic- and osteogenic-like cell types in vitro (Shapero et al. 2015). However, direct cell fate mapping in vivo has yet to show that adult VECs exhibit similar plasticity (Kim et al. 2018). Given these interesting but conflicting observations, there is a critical need for future studies to determine the regenerative potential of the adult VEC population.

CONCLUDING REMARKS

The developing endocardium demonstrates diverse plasticity given its potential to differentiate and transform into many cell types that contribute to the valves and myocardial chambers during embryogenesis. After birth, there is mounting evidence that the postnatal and adult endocardium continues to possess a degree of plasticity, at least in the myocardium following injury. In adult heart valves, similar potential of the endocardium has been less explored largely because of the limited availability of suitable model systems to study valve-specific injury. To date, therapeutic approaches to treat diseases of cardiac structures arising from the endocardium are limited. Repair or replacement of dysfunctional heart valves remains the number one need for surgery and more than 110,000 procedures are performed in the United States each year, costing in excess of $2 billion. Similarly, pharmacological treatment of coronary artery disease can treat risk factors and help dilate blood vessels, but the most effective approach to improve or restore blood flow requires more aggressive surgical approaches that are not ideal, particularly in the aging population. Therefore, there is an increasing critical need to explore alternatives. To address this, research has focused on delineating the molecular mechanisms and biological process that underlie cell-fate decisions and survival of the endocardium. From here, we can better understand the etiology of congenital heart defects, and also develop new therapeutics that target the endothelium to treat both pediatric and adult heart disease in the future.