The process of heart development can be broken down into a handful of basic steps according to major morphogenic landmarks. First is the formation of the moon-shaped cardiac crescent, followed by coalescence to create the linear heart tube, which then loops and subsequently undergoes formation of discrete cardiac chambers, myocardial trabeculation, and valvulo-septa morphogenesis (1, 2). While endocardium-myocardium crosstalk appears to be involved in all these developmental milestones, it is especially important during trabeculation and valve formation (1, 3–5). Improper endocardial-myocardial communication leads to a failure of cardiac growth and embryonic lethality or congenital heart diseases (6–10). This review attempts to encapsulate what is currently known about the unique, reciprocal communication between endocardial cells and cardiomyocytes during early cardiac development, including endocardial and myocardial crosstalk during their differentiation, myocardial trabeculation, and endocardial growth during trabeculation. The endocardial and myocardial signaling during cardiac endocardial-to-mesenchymal transformation and endocardial cushion formation, a well-studied process, has been comprehensively reviewed elsewhere (5, 11–13) and will not be discussed here. Similarly, the potential contribution of the endocardium to coronary vascular development and extracardiac organogenesis has been recently summarized (10) and is beyond the scope of this review.

During mouse embryogenesis, the endocardium is initially morphologically identifiable at the one- to two- somite stage as a “proendocardium layer” residing between the myocardial and endodermal layers (14). Molecularly, early endocardial cells can be distinguished by endocardial-specific expression of the transcription factor Nfatc1 (15). Recently the endocardial-specific molecular signature that differentiates them from other endothelial cells has been elaborated upon using single cell RNAseq analyses. This distinctive signature includes expression of markers such as natriuretic peptide receptor 3 (16), cytokine-like protein 1 (17), MEIS2, HAPLN1, FOXC1, LEPR, and TMEM100 (18) within the endocardium.

The embryonic origins of the myocardium have received much attention and is described in many excellent reviews (2, 19, 20). Extensive studies have primarily used avian models because of the easily accessible embryo. These studies have revealed that, within the heart-forming region, mesodermal differentiation into myocardium is governed by positive signals (BMP, FGF, TGF-β and Hedgehog pathways) and inhibitory signals including Wnts, Chordin, and Noggin (2, 21). However, the timing of endocardial specification, as well as the origins of endocardial vs. myocardial lineages, remain controversial (22, 23). For example, fate-mapping data, largely gleaned from chick and zebrafish models, provide evidence for separate origins. In these studies, cells were retrovirally tagged, which allowed cells to be tracked as they migrated from the primitive streak to the cardiac crescent. Results from these zebrafish studies demonstrated a separation of endocardial- and myocardial-destined cells during early gastrulation, with a progenitor cell only differentiating into one cell type or the other (24–26). Similarly, results from studies using chicken embryos provide data that myocardial and endocardial progenitors are independent from one another at the primitive streak stage or perhaps even earlier (27–30). Overall, these data support a model wherein pre-specification of mesodermal cells within the primitive streak dictates either an endocardial or a myocardial fate.

More recently, studies utilizing mouse models, cell lineage-mapping, and embryonic stem cells (ESCs) have shown that endocardial and myocardial precursors have common spatial and molecular characteristics, which ultimately supports the idea that they are derivatives of a common multipotent cell residing within the cardiac mesoderm (22, 26). For instance, lineage tracing in mice revealed a multipotent population of precardiac progenitors in the late primitive streak that express vascular endothelial growth factor receptor 2 (Vegfr2, or Flk1) and were capable of generating cardiomyocytes and endothelial cells (31). Substantiating that data are studies using Cre lines driven by cardiogenic promoters (Nkx2-5-Cre, Isl1-Cre, and the Mef2C-enhancer-Cre) to trace cell lineages, all of which led to the labeling of both endocardial cells and cardiomyocytes (32–34). Additionally, embryonic murine Isl1⁺ cardiac progenitors proved to be multipotent in vivo and in vitro, giving rise to endocardial, myocardial, and smooth muscle cells (35). Another link between endocardial and myocardial specification comes from studies showing that the cardiac transcription factor NK2 homeobox 5 (Nkx2-5) promotes expression of the ETS-related transcription factor gene Etv2 (Er71 or Etsrp71) (36), which is necessary for endocardial cell specification (discussed below) (23, 36, 37). Removal of Etv2 results in myogenic differentiation of cells that would otherwise have become endocardium (38). Finally, experiments using ESC differentiation not only validate that a common multipotent progenitor cell gives rise to myocardial, smooth muscle, and endocardial cells, but provide evidence that endocardial cell progenitors are derived from a distinct lineage when compared to other haematopoietically-derived vascular endothelial cells (31, 35, 39–41). Together these data illustrate a shared origin story beginning in a multipotent myocardial-endocardial progenitor cell during the late primitive streak stages.

How to reconcile the above, apparently opposing sets of data? The answer may lie with a prolonged developmental plasticity of the progenitor cells. For instance, as briefly mentioned above, endocardial cells can be reprogrammed into muscle cell lineages, including myocardium, through loss of Etsrp/Etv2 function (38, 42). Notably, using Mesp1^Cre to permanently label the earliest cardiovascular progenitor population within the early primitive streak stage (E6.25), Mesp1+ cells within the first heart field heterogeneously expressed endocardial and myocardial markers in an “either or” fashion. Interestingly, this changed during the late primitive streak stage (E7.25), when the Mesp1-expressing lineage in the second heart field more inclusively expressed endocardial and myocardial markers (29, 43, 44). These data from zebrafish and mammalian studies allow an integration of the above-mentioned, seemingly at-odds research by setting up a plausible scenario wherein multipotent endocardial-myocardial progenitors remain developmentally pliable longer than anticipated (23).

It is likely that there are interactions between the myocardial and endocardial cells from the earliest development stages, given that the myocardium develops concurrently with the endocardial layer and that the two populations of cells are intimately associated (Figure 1). Yet the paucity of available models to study myocardial differentiation without contributions from an endocardial population has made it difficult to further delineate the endocardial-myocardial signaling. Saint-Jean et al. (45) recently developed a novel in vitro mouse ESC system to determine how myocardial differentiation is dependent on the endocardium, focusing on the initial stages of cardiogenesis. Using regulatory elements of the endocardial-specific marker, Nfatc1, they created a mouse stem cell line that expresses the diphtheria toxin receptor (Nfatc1-DTR) only within differentiated endocardial cells. During early stages of in vitro differentiation, the NFATc1-DTR mouse embryoid bodies were treated with diphtheria toxin to ablate endocardial cells, while other endothelial populations remained intact. Loss of the endocardial cells was detrimental to cardiomyocyte function and differentiation, with less beating embryoid bodies and reduced expression of genes necessary for early and late myocardial differentiation. Bmp2 was found to partially rescue the myocardial function and gene expression, suggesting that initial stages of myocardial differentiation are mediated by endocardially-derived Bmp2. Supporting these data, Pasquier et al. (46) found that endothelial cells augmented cardiomyocyte maturation when co-cultured with human ESCs.

In turn, early endocardial differentiation is dependent on myocardial paracrine signaling. One recent study used differentiation of human pluripotent stem cell (hPSC)-derived cardiovascular mesoderm to investigate endocardial differentiation. In this model, cells were treated with BMP10 (a growth factor secreted by cardiomyocytes), NRG1 (an endocardial-expressed growth factor), or inhibitors including siRNA against BMP10 and NRG1 (18). They found that hPSC-derived endocardial cells require BMP10 for specification and maintenance, including expression of NKX2-5 and NRG1. The hPSC-derived cardiomyocytes likewise responded to endocardial-secreted NRG1, adopting a trabecular fate complete with BMP10 expression. These NRG1- and BMP10-mediated in vitro interactions between the cardiomyocytes and endocardial cells are similar to those that occur in the developing heart.

An intricate and interdependent relationship between endocardial and myocardial cells during in vivo differentiation has been recently highlighted in zebrafish. In fish, Hedgehog (Hh) signaling is required for development of the myocardium (47). Hh signaling, along with vascular endothelial growth factor (Vegf) and Notch pathways, also has roles in vascular endothelial differentiation, migration, and branching (48). However, only Hh has a role in early endocardial differentiation (49). Genetic or pharmacological inhibition of Hh results in near loss of the endocardial marker, nfatc1, while leaving the differentiation of vascular endothelial populations unperturbed. Thus, Hh signaling may be of paramount importance specifically for endocardial development. Consistent with the zebrafish data, decreased Vegfa in mice results in irregular vascular endothelium formation, while the endocardium appears normal (50). These data suggest that, while Vegfa is required for general vascular development, it is not required for early endocardial morphogenesis. Additional work using myocardial-deficient hand2 mutant zebrafish embryos revealed an endocardial-specific defect in differentiation (51, 52). Moreover, in the aftermath of genetically ablating zebrafish cardiomyocytes, endocardial cells lose their identity through loss of nfatc1 expression to become more vascular-like (52, 53). Interestingly, in the context of myocardial-cell depletion, overexpressing Bmp2b can partially rescue endocardial progenitor cell identity by increasing Nfatc1 expression. This places BMP signaling at the crux of cardiogenic differentiation (23). Altogether, these data tell us that endocardial and myocardial differentiation are inextricably linked through a definitive and reciprocal paracrine signaling program.

Myocardial trabeculation is a unique and life-essential morphological landmark of ventricular chamber development wherein cardiomyocytes proliferate, differentiate, and form a network of protrusions extending into the lumen of the heart. Trabeculae increase cardiac muscle mass and, prior to coronary vascularization, they permit the cardiomyocytes access to oxygen and nutrients (68). To date, all mouse models that have stunted trabeculation have a hypoplastic left ventricular wall and fail to survive past midgestation (E14.5) (9, 10). It therefore appears that trabeculation is necessary to sustain embryonic life and failure in the process precludes ventricular compaction. Indeed, this is substantiated clinically, as there are no reports of human cardiomyopathies caused by a lack of trabeculation. Mouse models have provided insight into inter-cellular communication necessary for trabeculation. These studies show that trabeculation is contingent on the endocardium, myocardium, and cardiac extracellular matrix (ECM) for proper endocardial and myocardial proliferation and differentiation (69, 70).

During cardiac trabeculation, endocardial and myocardial cells support and interact with each other to stimulate cardiomyocyte proliferation and endocardial growth. Disrupting this endocardium-myocardium communication results in embryonic lethality due to failed cardiac growth (5, 70). Even as studies continue to highlight the essential roles of the endocardium, the regulatory mechanisms involved in endocardial proliferation and development remain largely unknown. Historically, research has focused on the transcriptional regulation of myocardial differentiation in the early stages of cardiac development. Less is known about factors that regulate endocardial ontogeny during heart development.

As discussed above, significant myocardial trabeculae growth begins after cardiac looping at E9.0, with rapid growth and expansion of the trabeculae from E9.5 to E13.5. Concomitantly, once the endocardial cells are specified, they must quickly proliferate to keep pace with the trabecular myocardium, leading to an expansion of endocardial cells near the end of trabeculation around E13.5 (Figure 3). Interestingly, despite the observation that the early endocardium uniquely expresses Nfatc1, the transcription factor is not necessary for regulating the specification or proliferation of endocardial cells, although it is later necessary for the formation of cardiac valves (22, 101, 102). For the most part, regulation of endocardial proliferation is uncharacterized.

Tie2 has dual roles in regulation myocardial proliferation in a paracrine manner, as discussed above, and it has a central autocrine effect on endocardial growth and proliferation that lays the foundation for normal trabeculation. Tie2 is dispensable for pan-endothelial differentiation and the first stages of vascular assembly (103), but it is required for endocardial sprout formation and growth. Utilizing the same endocardial-specific Nfatc1^Cre line to conditionally remove Tie2 from the endocardium as described above, Qu et al. (72) demonstrated that Tie2 signaling is a prerequisite to trabeculation. Loss of endocardial Tie2 results in impaired embryonic endocardial sprouting and growth as early as E9.0 due to decreased endocardial cell proliferation and impaired migration. Thus, Tie2 promotes endocardial proliferation and development of the angiogenic extensions, or sprouts, involved in initial stages of trabeculation. Interestingly, endocardial loss of Tie2 interfered with the endocardial gene signature, with decreased expression of Sox7, Arap3 and Tmem100 (72). Global knockout of these genes in mice leads to in utero demise with impaired trabeculation involving disturbed endocardial network formation. However, the cardiac defects are likely secondary, because when these genes are deleted specifically from the endocardium using Nfatc1^Cre, all lines display a normal endocardial network and proper trabeculation (72, 104). Therefore, more studies are needed to better understand how Tie2-Ang1 signaling regulates the endocardial proliferation necessary to support trabeculation.

As discussed above, Etv2 and Vegfr2 are both specifically expressed in endothelial cells and are critical for their specification. Loss of either gene results in embryonic mortality with global depletion of all endothelial cells, so it is plausible to argue that both are required for endothelial and endocardial growth. However, conditional knockout mice of Etv2 with Tie2^Cre (active in the endothelial lineage) or Nfatc1^Cre (active in endocardial cells from E9.0, about 1 day later than Tie2^Cre) (71) were healthy (72, 105). This is consistent with the observation that Etv2 has transient endocardial/endothelial expression from E7.75-E9.5 (36). These results support a role for Etv2 during a restricted developmental window, when it activates haematopoietic and endothelial differentiation (54). Similarly, endocardial-specific deletion of Vegfr2 with Nfatc1^Cre also led to no apparent endocardial defects, although most null embryos died from E16.5 to E18.5 with abnormalities in coronary angiogenesis and myocardial vessel formation (97).

Other studies emphasize the importance of biomechanical cues in regulating endothelial proliferation, specifically during later stages of ventricular chamber morphogenesis. A recent report using a zebrafish model showed that endocardial cell proliferation during the endocardial ballooning phase and chamber morphogenesis is not controlled by Vegf signaling, rather it depends on hemodynamics and myocardial-derived Bmp (53). Bornhorst et al. (106) also used zebrafish to further investigate the biomechanics involved in cardiogenic regulation. They demonstrated that the expansion of the myocardium creates tension on the endocardium, where the biomechanical transducer VE-Cadherin responds by upregulating endocardial proliferation via Yap1 nuclear translocation. Taken together, these studies suggest that endocardial specification is Etv2- and Vegfr2- dependent, whereas the subsequent endocardial growth and expansion during trabeculation are not.

Lastly, it is evident that the Notch1-Dll4 signaling pathway is required for endocardial growth during trabeculation, as endocardial-specific deletion of Notch1 or its ligand, Dll4, with Nfatc1^Cre leads to impaired endocardial sprouting and a simplified endocardial network (9, 71). However, additional studies are required for further confirmation and to determine how Notch1-Dll4 signaling modulates endocardial proliferation during the critical trabeculation timeframe. Moreover, potential interactions between Notch1-Dll4 and Tie2-Ang1 signaling pathways remain to be explored.

The endocardium and myocardium interact through dynamic paracrine signaling pathways in all stages of cardiac development. Endocardial cells secrete signaling mediators such as Nrg1 that modulate cardiomyocyte development and survival. Reciprocally, cardiomyocytes promote endocardial cell proliferation, assembly, and survival through Angiopoietin-1. In this review, we focus on these critical reciprocal exchanges between the myocardium and endocardium in early cardiac development, including the origins of the endocardial and myocardial lineages, endocardial and myocardial crosstalk during differentiation and trabeculation, and signaling essential for endocardial growth during trabeculation. While much attention has been focused on the transcriptional regulation of myocardial development during early cardiogenesis, with the role of the endocardium being relegated to the process of EMT during valvulogenesis (11), it is now quite clear that the endocardium plays a central role in choreographing most of the major morphological processes required for ventricular chamber formation. Specification of the cardiac lineages as well as the cellular proliferation and differentiation necessary for valve and chamber morphogenesis, all rely on a complex interplay between multiple pathways such as Notch, Wnt, Bmp, Nrg1-ErbB2/4, and EphrinB2/EphB4. Endocardial Tie2 plays complimentary roles during ventricular chamber trabeculation by promoting endocardial proliferation and sprouting, while preventing myocardial hypertrabeculation through suppression of retinoic acid signaling within cardiomyocytes. Beyond this, endocardial cells have roles in coronary vascular development and in haematopoiesis, which suggests that they have roles during heart regeneration and could provide novel progenitors for mural cells of the heart (10, 107). Thus, the endocardium is a promising target for therapeutic intervention both to maintain homeostasis within the heart and to stimulate cardiomyocyte replenishment during cardiac disease or injury.

XQ and HB wrote the manuscript and generated the figures. XQ, HB, and CH revised the manuscript. All authors contributed to the article and approved the submitted version.

We are grateful for funding support from National Institutes of Health HL R01 DK125895-01.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.