December 2009 | Vol 6 | N.º 12 | CNIC-15 [PDF (1.150 KB)]

Tissue-patterned embryonic interactions in heart development: cell versus non-cell-autonomous molecular signalling and the origin of CHD

José María Pérez-Pomares¹ and José Luis de la Pompa ²

Competing interests: The authors declare no competing interests.

Summary
Pattern formation is a key event in heart development. Activation of domain-specific genetic programs and definition of competent populations of cells will ultimately generate the diversity of tissue types that characterises the vertebrate heart. Development of the complex anatomical arrangement of all of these cell types is progressive and occurs in parallel to the molecular specification of cardiac domains. Both intrinsic (cell-autonomous) and non-intrinsic (non-cell-autonomous) signals are required to pattern the heart; the impact of the latter set of signals has been overlooked in the literature, and its enormous influence in the differentiation and maturation of cardiac tissues has been largely underestimated. In this review, we will provide a concise account of the events and molecular signals that contribute to define the anatomical structure of the heart, also relating them to the onset of congenital heart disease (CHD). Our views are supported by the research developed in our own laboratories as well as by data reported by other scientific teams.

Introduction
The highly sophisticated four-chambered architecture of the vertebrate heart is acquired after a deep morphologic remodelling of the embryonic cardiac outline, which originally is a straight tube located at the embryonic midline. From a morphogenetic perspective, the real challenge is to develop a structure able to fit the postnatal and adult blood circulation with a double, parallel flow circuitry (pulmonary/systemic) from a very simple organ rudiment (Horsthuis et al., 2009). This has to be accomplished while proper differentiation and maturation of cardiac tissues takes place and, most importantly, without impairing normal blood flow.

All of these changes are orchestrated by a complex transcriptional network that sets the basic genetic patterning of the heart (Davidson and Erwin, 2006). As a consequence of this informational layout, different regions of the heart are conferred a characteristic molecular identity, which in turn defines developmental fields of competence. Many of the signals involved in the process are not tissue-intrinsic (cell-autonomous) but are rather provided by different, adjacent tissues. This extremely important aspect is frequently overlooked by studies that attempt to analyse the molecular basis of congenital heart disease (CHD), sticking to the rationale “one cell type/one molecule”. Therefore, spatial coordination of the molecular crosstalk between distinct cardiac cell types is a key concept for understanding many cardiac malformations.

Embryonic heart cell types
For many members of the cardiovascular scientific community, “heart” is a synonym for “myocardium”. It is obvious that muscle cells define the nature of the adult heart, but it is no less evident that the homeostasis of the adult organ is guaranteed by non-muscular cells such as endocardial, coronary and fibrous interstitial cells (Eckert et al., 1998; Chapman and Spinale, 2004; Brunner et al., 2006). Moreover, these cells are not only important for the physiologic equilibrium of the heart, but also a frequent substrate or component of various acquired cardiac diseases. In the context of embryonic development, non-muscle cells are critical for the proper differentiation and maturation of a variety of myocardial populations and are absolutely indispensable for the morphogenesis of valvuloseptal or chamber (trabeculae) structures.

Myocardium and endocardium develop at the same time from two bilateral regions of the precardiac splanchnopleural epithelium known as “cardiac fields” (Buckingham et al., 2005; Abu-Issa and Kirby, 2007). Endocardial and myocardial cells physically segregate after the epithelial-to-mesenchymal transition (EMT) of the precardiac epithelium: endocardial primitive cells migrate and assemble over the basal lamina of the pharyngeal epithelium, whereas early myocardial precursor cells, retaining an epithelial profile, will cover the endocardial lining to form the two concentric tissue layers that constitute the straight heart tube (Sugi and Markwald, 1996). Although in terms of its molecular fingerprint the embryonic endocardium is almost indistinguishable from the vascular endothelium (with the exception of its characteristic NFATc, Sox4, and Rb expression as well as an early GATA5 transcriptional peak; see Schilham et al., 1996; Morrisey et al., 1997; de la Pompa et al., 1998; Wagner et al., 2001), the endocardium has been known to be heterogeneous with respect to its potential to undergo EMT. Endocardial EMT is the very first step in the morphogenesis of cardiac valves. Only restricted regions of the endocardium (atrioventricular, AV; and conoventricular, CV) are competent to transform into mesenchyme (Eisenberg and Markwald, 1995; Person et al., 2005) and they do so under segment-specific signals that initially emanate from the endocardium (see below).

In the ventricles, endocardial cells will not normally transform, but their active signalling will promote the formation of trabeculae, a series of endocardium-covered, ridge-like protrusions of the myocardium that will be remodelled into papillary muscles, the interventricular septum, part of the supporting valvular apparatus and components of the distal (Purkinje) conduction system. This endocardium-guided myocardial trabeculation is typical of the ventricular chambers but not of the inflow myocardium or the atria (Ben-Shachar et al., 1985; Grego-Bessa et al., 2007).

The epicardium is the third tissue layer of the heart. It develops once the cardiac tube is formed and has initiated its characteristic bending and torsion (looping) (Männer et al., 2001; Wessels and Pérez-Pomares, 2004). Findings from the past two decades have shown that the epicardium is not a passive tissue but one that is also able to initiate an EMT that generates a population of mesenchymal epicardium-derived cells (EPDCs). These EPDCs contain the building blocks of the coronary vascular system and progenitors of part of the cardiac interstitium and even cardiac valves (Pérez-Pomares et al., 2002). Most interestingly, the epicardial EMT is restricted to the ventricular, atrioventricular and conoventricular domains of the heart, as the atrial epicardium is refractory to EMT. Segmental restrictions also apply to the extent of the migration of EPDCs into the myocardium: AV EPDCs invade the compact and trabecular layers to populate the primordium of the developing valves (the so-called endocardial cushions); ventricular EPDCs mainly distribute in the compact myocardial layers, but some of them also reach the trabeculae, whereas CV EPDCs do not apparently invade the outflow tract myocardium and remain confined into the subepicardium (Männer, 1999; Pérez-Pomares et al., 2002; Wessels and Pérez-Pomares, 2004).

Both epicardial and EPDCs have potent signalling properties and are required for the outer compact ventricular myocardium to proliferate. Genetic ablation of epicardial-specific genes, such as the Wilms’ tumour suppressor Wt1, have a direct effect on the thickness of the compact ventricular layer (Moore et al., 1999), and it has been proven that a retinoic acid (RA)-dependent component of the epicardial secretome acts as the paracrine signal that mediates the epicardial impact on the myocardium (Chen et al., 2002).
The relationship that the epicardial lineage bears to the endocardium and myocardium remains unclear; some studies have shown that epicardial progenitor cells display early expression of some myocardial transcriptional specifiers, such as Nkx2.5, but not others, such as Isl1 (Ma et al., 2008). Embryonic and/or adult epicardial cells do not express myocardial markers, but surprisingly, embryonic epicardial cells have myocardial differentiation potential that can be triggered under certain circumstances (Kruithof et al., 2006).

Cardiac looping and spatial rearrangement of heart domains
Cardiac looping provides the major topological changes needed to align cardiac domains, so that previously distant regions of the heart approach in space. With the progression of looping, the heart tube changes from a C-shaped to an S-shaped conformation, and, at the same time, the myocardium at the ventral midline of the straight heart tube is relocated to the right, just on the most external limit of the heart bend or cardiac outer curvature (Männer, 2000, 2009). Cells at this position will soon fully activate a genetic program that leads to the development of the atrial and ventricular chambers. On the other hand, the myocardium at the opposite inner curvature retains the genetic features of the primitive myocardium of the straight heart tube. This non-chamber myocardium is located at the inflow, AV and CV regions of the looped tube, thus separating the developing atria and ventricles (Christoffels et al., 2000).

Molecular anatomy of the heart: intrinsic signals

Some of the earliest regional specification events taking place in the heart relate to the cranio-caudal (A/P) patterning of the heart tube. The expression of the classic cardiac specifiers in the primary and second cardiac fields (the transcription factors Nkx2.5, SRF, Mef2c, GATA4 and Isl1) is soon followed by the localised, regional expression of other transcription factors along the heart tube (including the caudal restriction of Tbx5 expression and a cranio-caudal expression gradient of GATA4, 5 and 6). A good candidate signal to initiate and regulate this transcriptional activity is RA, a morphogen that has been shown to directly activate the Tbx-5 and GATA transcription factors (Jiang et al., 1998; Bruneau et al., 1999). RALDH2, the main RA-converting enzyme in the embryo, is expressed in a pattern that suggests that caudal high levels of RA are developmentally read as an instruction to initiate an atrial program (Hochgreb et al., 2003). Mice deficient for RALDH2 do not develop atrial chambers, and increased levels of RA signalling expand prospective atrial domains in the heart tube (Niederreither et al., 2001).

Cardiac chambers apparently grow from the outer curvature (the ventricles at cranio-ventral positions and the atria at dorsolateral locations), and therefore, the dorsoventral specification of heart tissues is also a relevant aspect of cardiac development. Some identified genes could have a role in differentiating the ventral ventricular domains; among them is the transcription factor Hand1/eHand. Hand1 is originally expressed by cardiomyocytes at the ventral side of the cardiac tube and is expressed later on by these same myocytes at the outer curvature of the looped heart. Hand1-null mice present impaired looping as well as reduced ventricular domains without affecting ANF expression (Riley et al., 1998), whereas Hand1 knock-in mice (with Hand1 under the control of the MLC2v promoter) show an expanded ventricular domain and abnormal expression patterns for Chisel, ANF and Hand2/dHand (Togi et al., 2004). These results suggest that Hand1 is not a “master ventricular gene” but rather an important player in D/V specification of the heart tube.
A/P and D/V patterning signals intersect at the formation of cardiac chambers (atria and ventricles). The definition of chamber versus non-chamber cardiac domains depends on the active repression of the chamber phenotype program by certain transcription factors. Interestingly enough, some T-box transcription factors (Tbx2,3) with expression restricted to the inflow tract, AV canal, inner curvature and outflow tract, actively compete with other members of the same family (Tbx5) for effective binding to their potential transcriptional targets (the chamber markers ANF, Chisel and Cx40) in order to block chamber development (Christoffels et al., 2004).

Molecular anatomy of the heart: crossing tissue boundaries
Endocardial-myocardial interactions are likely to be initiated from the very early steps of heart development, but are only evident with the onset of heart valve morphogenesis at AV and CV regions and the trabeculation of ventricular chambers. The case of AV valves (tricuspid and mitral) is especially significant because activation of the endocardial EMT providing the first components of the valvuloseptal mesenchyme depends on AV myocardial signals; this is also significant because AV valve formation takes place in a cardiac region that is limited by two large chamber formation domains (atria and ventricles). As we indicated above, AV myocardium is considered primitive and its endogenous transcriptional profile obviously influences its signalling activities. AV myocardium expresses high levels of BMP2 and TGFb2, two molecules regarded as the main triggers of the neighbouring AV endocardium EMT (Eisenberg and Markwald, 1995; Person et al., 2005). On the other hand, the Notch signalling pathway is active in most AV endocardial cells. Notch has been shown to promote endocardial EMT by activating expression of the cadherin repressor Snail1 (Timmerman et al., 2004). Some well-known Notch targets, including Hey1/HRT1 and Hey2/HRT2, are active not only in the endocardium, but also in the myocardium, where they seem to confine BMP2 expression to AV myocardium (Kokubo et al., 2005). The crosstalk between Notch and BMP2 during cardiac valve development is worth exploring as it may generate knowledge of interesting therapeutic implications.

In the developing ventricles, trabeculation is also dependent on endocardial Notch signalling. In the trabeculae, Notch promotes basal trabecular proliferation via BMP10, while at the same time favouring trabecular myocardium differentiation and maturation by activating endocardial EphrinB2 (directly) and NRG1 (indirectly). Endocardially secreted NRG1 acts as a paracrine signal over the ErbB2/4-expressing cardiomyocytes (Grego-Bessa et al., 2007).

Epicardial-myocardial interactions take place later in development. They are important because myocardial signals are likely to be required to sustain epicardial EMT, but epicardial cells and their derivatives (including coronary vessel progenitors) are also required to signal to promote the thickening of the compact ventricular working myocardium (Moore et al., 1999). Transcription factors such as Tbx18 and the Wilms’ tumour suppressor Wt1 are expressed by epicardial cells and EPDCs but not by ventricular myocytes (Zhou et al., 2008). Wt1 has been intensively studied as a possible master regulator of epicardial biology, including epicardial signalling activities, which have been reported to be RA-dependent, although RA itself is unable to promote ventricular myocardial proliferation or to rescue the myocardial defects caused by defective epicardial signalling (Chen et al., 2002). It is still not known how such control is articulated. A possible link is the suspected role that transcriptional forms of Wt1 could have in the synthesis of RA, which characterises epicardial embryonic cells.

The detailed role of EPDCs in coronary development, the contribution of this lineage to the cardiac interstitium (including the relevance of the interaction of this cell type with other cell populations homing between cardiac muscle fibres), and the possible impact of these cells in the morphogenesis of cardiac valves still need extensive research.

Abnormal heart patterning and the onset of congenital heart disease (CHD)
The vast majority of CHD is thought to be multifactorial. However, a variety of monogenetic alterations can also yield a CHD phenotype (Garg, 2006). As can be inferred from the preceding sections, gene deficiencies affecting a cardiac tissue type can deeply impact the development of adjacent tissues. Many specific CHD cases are in the spectrum of AV defects; the malfunction of genes regulating AV endocardial EMT or the proliferation of the valvuloseptal mesenchyme can generate abnormal valves and alter AV myocardium remodelling and, secondarily, the structure of the cardiac conduction system. Furthermore, endocardial cushion cells are critical for completing the septation of the chambers. Therefore, abnormal cushion development can also develop into a membranous interventricular septal defect. On the other hand, endocardial ventricular defects can affect the morphogenesis of the trabeculated myocardium at different levels. As a result, interventricular septal defects and myocardial non-compaction can occur.

Defects in the working myocardium can also arise as a consequence of abnormal epicardial signalling. The characteristic thinning of the compact ventricular myocardium, which is often associated with RA deficiency, is in many cases reproducible by simply altering epicardial development at different developmental checkpoints (Wessels and Pérez-Pomares, 2004). Moreover, abrogation of RA signalling through the RXRa receptor only affects ventricular myocardial mass if the deletion is epicardial (RXRa myocardial deletion shows no compact myocardium phenotype) (Chen et al., 1999).

In conclusion, we suggest that some adult cardiovascular diseases can arise by the abnormal persistence of embryonic signals in adulthood. As such, possible effects on the myocardial structure or function may be expected, especially for those defects identified in non-muscular tissues. Further research will be required to identify signals that could be the substrate for innovative therapies and for the early diagnosis of cardiovascular diseases.