The ventricular conduction system represents the electrical wiring responsible for the co-ordination of cardiac contraction. Defects in the circuit produce a delay or conduction block and induce cardiac arrhythmias. Understanding how this circuit forms and identification of the factors important for its development thus provide insights into the origin of cardiac arrhythmias. Recent advances, using genetically modified mouse models, have contributed to our understanding of how the ventricular conduction system is established during heart development. Transgenic mice carrying different reporter genes have highlighted the conservation of the anatomy and development of the ventricular conduction system between mice and humans. Lineage tracing and retrospective clonal analysis have established the myogenic origin of the ventricular conduction system and determined properties of conductive progenitor cells. Finally, gene knock-out models reproducing human cardiac defects have led to the identification of transcription factors important for the development of the ventricular conduction system. These transcription factors operate at the levels of both conduction system morphogenesis and differentiation by controlling the expression of genes responsible for the electrical activity of the heart. In summary, defects in the ventricular conduction system are a major cause of arrhythmias, and deciphering the molecular pathways responsible for conduction system morphogenesis and the differentiation of conductive myocytes furthers our understanding of the mechanisms underlying heart disease.
The primary role of the heart as a pump relies on two main functions, conduction and contraction. Active contraction of cardiac muscle is ensured by the working myocardium, upon electrical stimulation. Efficient expulsion of blood depends on tightly co-ordinated sequential contraction of the atria and ventricles. The dual role of the cardiac conduction system is to produce this electrical activity and to propagate it in a co-ordinated manner.
The electrical activity of the heart is generated by pacemaker cells localized in the sino-atrial node (SAN) in the dorsal region of the right atrium. From the SAN, electrical activity spreads rapidly to both atria, provoking simultaneous atrial contraction. The atria and ventricles are electrically isolated by the annulus fibrosus, composed of non-myocardial cells.1 In order to reach the ventricles, the electrical impulse follows a specialized route termed the atrioventricular (AV) conduction system. Electrical activity from the atria converges on the atrioventricular node (AVN), where its conduction velocity is reduced to delay the spread of electrical activity and allow atrial contraction before ventricular activation. Electrical activity is then rapidly propagated to the ventricular apex through a specialized fast conducting system or ventricular conduction system (VCS), including the atrioventricular bundle (AVB), also known as the bundle of His, the right and left bundle branches (RBB, LBB) on either side of the interventricular septum, and a complex network of Purkinje fibres (PF) ramifying over the subendocardial ventricular surface. The VCS plays a critical role in co-ordinating the heartbeat by rapidly conducting the electrical impulse to the ventricular apex in order to activate contraction from the apex and maximize efficiency of expulsion of the blood through the great arteries at the base of the heart.
The first anatomical description of the human VCS was made by Tawara at the beginning of the 20th century.2 Conductive cells can be distinguished from working myocardium by the presence of a poor contractile apparatus with few sarcomeres, enriched glycogen content and reduced numbers of T tubules.3 The function of conductive cells has been investigated by electrophysiological experiments demonstrating specific electrical properties, including rapid conduction;4 however, it is the global anatomy of the VCS that orchestrates co-ordination of the heartbeats, and disturbances in VCS anatomy result in arrhythmias.5 The availability of genetically modified models makes the mouse a highly attractive species in which to study the anatomy and development of the conduction system. This review focuses on the murine VCS, with emphasis on transgenic mouse models that have provided insight into the origin and development of the VCS.
2. Mouse models for studying the ventricular conduction system
2.1 The asymmetrical anatomy of the VCS
In order to visualize the anatomy of the entire murine VCS, we generated a transgenic mouse line in which a green fluorescent protein (GFP) reporter gene was inserted at the Cx40 locus, encoding a connexin expressed throughout the VCS.6 Cx40 defines the adult VCS both by its pattern of expression and by its role in conduction. Cx40 forms gap junction channels with a high conductance and is expressed in all compartments of the fast conduction system, including the AVB, left and right bundle branches, and PF network, but not in working myocardium or slow-conducting compartments of the conduction system, such as the SAN and AVN.7Cx40GFP expression delineates the adult ventricular conduction system, revealing a network extremely similar to the drawings of Tawara for the human heart (Figure 1) and consistent with classical histological images of the mouse VCS.8 Three-dimensional imaging of Cx40GFP hearts reveals that the VCS represents only 1% of the volume of the ventricles.9 A precise three-dimensional image of the AV conduction axis was recently published by Aahnaanen and collaborators, by analysing gene expression patterns in the central conduction system.10 These data reveal the existence of specific subregions of the AV axis delineated by distinct molecular signatures. From the atria, two successive rings of cells form figure-of-eight-like structures around the AV junctions, described as the transitional AV ring and the nodal AV ring, that appear to play a role in slowing down atrial activation. The AV or central conduction axis is composed of a distinct compact AVN and an AVB, including lower nodal cells anatomically similar to that described using the Cx40GFP reporter line (Figure 2A and B).
Anatomy of the ventricular condution system in the mouse and human. (A) Green fluorescent protein (GFP) expression in the left ventricular conduction system of a Cx40GFP mouse heart. AVB, atrioventricular bundle; LBB, left bundle branch; LPF, left Purkinje fibres. (B) Drawing of the ventricular conduction system in the human left ventricle, reproduced from Tawara (1906).2
Right–left asymmetry of the mouse ventricular conduction system. (A and B) Three-dimensional reconstruction of the atrioventricular bundle (AVB) of an adult Cx40GFP heart showing a small right bundle branch (RBB; A) and large left bundle branch (LBB; B). The interventricular septum is coloured yellow. (C–E) Right–left asymmetry of the peripheral conduction system. (C) Transverse section of an adult Cx40GFP heart showing the large number of Purkinje fibres (PFs) on the right ventricular free wall (D′ arrowhead) compared with the right septal surface (D). In contrast, in the left ventricle PFs are mainly present on the septal surface (E) and absent between the two papillary muscles (*) in the left ventricular wall (E′). Left and right ventricular lumens are coloured grey. Three-dimensional reconstruction of the right (D and D′) and left (E and E′) Purkinje fibre network of a Cx40GFP adult heart. LS, left septum; LV, left ventricle; RS, right septum; RV, right ventricle.
One striking feature of the anatomy of the VCS is the asymmetrical morphology of the bundle branch and PF network between right and left ventricles (Figure 2). The proximal RBB is formed by a single fascicle, while the LBB is formed by numerous fascicles. Optical mapping of the adult interventricular septum (IVS) has shown that the electrical pathways in the proximal part of the VCS are also asymmetrical.6 The PF network is localized at the surface of the right ventricular free wall, while in the left ventricle the PF network covers the entire septal surface (Figure 2D–F). Modelling of the ventricular conduction system demonstrates a precise alignment of activation maps and PF anatomy.11 This consistency between anatomy and physiology illustrates how morphology underlies function in the VCS. In human patients, RBB block is more frequent than LBB block, which may be explained by the fact that a single branch is likely to be more sensitive to damage than multiple branches.12 In general, the pattern of electrical activation follows the direction of blood flow.13 The asymmetry of the peripheral VCS may thus be explained by the necessity of coupling differential patterns of activation in each ventricle with maximal haemodynamic efficiency.
2.2 Markers of the developing VCS
The first peristaltic contractions of the mouse heart appear by embryonic day (E) 8.5 and are rapidly replaced as the cardiac tube loops by sequential contraction of the atria and ventricles detected by a regular ECG.14,15 Based on histological criteria, Viragh and Challice carried out an exhaustive analysis of the development of the mouse conduction system and did not find specialized conductive cells at these early stages.3 Instead, sequential contraction of the primary heart tube is thought to be ensured by alternating regions of fast-conducting (future atria and ventricles) and slow-conducting myocytes [atrioventricular canal (AVC), and outflow tract (OFT)].16–18 By E9–E10, the primordium of the AV conduction pathway appears in the inner dorsal wall of the AVC and becomes increasingly compact as development proceeds to form the AVN.3 At the same time, a group of cells differentiates along the ridge of the IVS, extending from the primordium of the AVN to the middle portion of the rim of the interventricular foramen, where it divides into right and left branches.3 More recently, a number of transgenic mice carrying reporter genes have allowed visualization of these cells.15,19,20 In Mink-LacZ or CCS-LacZ embryos, the primordium of the AVN can be seen in the dorsal wall of the outer curvature of the heart, while reporter gene expression is observed in rings at the crest of the IVS and around the atrioventricular canal.21–23 A similar ring pattern is also observed for the T-box transcription factor, Tbx3, which identifies the developing central conduction system.24 The peripheral conduction system is not yet delineated at this stage of development. At E9.5, Cx40-positive trabeculae form at the subendocardial surface of both ventricles and are thought to play the role of a fast-conducting network in the embryonic heart.14,25 Optical mapping of embryonic hearts has shown that ventricular activation switches from an apex to base pattern at E10.5, to biventricular activation points by E12.5.23 A functional embryonic route equivalent to the His–Purkinje network is thus present prior to completion of ventricular septation.
More recently, Christoffels and Moorman have published a recapitulative table of the expression pattern of known markers of conduction system development.15 Many of these molecular markers are not selective for the definitive conduction system and are also expressed in surrounding tissues that share a common developmental pathway. Tbx3, for example, is expressed in the AVC and in AVC-derived cells, including valve leaflets as well as the central conduction system, and broad expression of the CCS-LacZ transgene is observed in the IVS and valves.21,24Cx40, which as we have discussed above is a definitive marker of the mature VCS, is expressed in ventricular myocardium but not in the AVN and AVB before E14.5.7 In contrast, the CCS-LacZ transgene is present in the AVC and IVS at E10.5, thus differing in expression from the Cx40GFP allele in early development, despite a subsequently convergent expression pattern throughout the definitive VCS (Figure 3). These results reveal the complexity underlying the development of the VCS and suggest that the central and peripheral components of the conduction system develop independently prior to integration in a single conductive pathway. More recently, a new marker of the VCS, contactin 2 (cntn-2), encoding a cell adhesion molecule critical for neuronal patterning and ion channel clustering, has been identified by a transcriptome analysis of PFs purified from CCS-LacZ mice.26 Immunohistochemistry and electrophysiological recordings showed that cntn-2-GFP+ cells have a conductive phenotype. The expression pattern of contactin 2 during embryonic development remains to be described, and no cardiac phenotype has been reported in mutant mice.27,28 The absence of transgenic mice specifically delineating the entire VCS during development may stem from the multiplicity of VCS cell lineages.
Comparison of the expression pattern of CCS-LacZ and Cx40GFP VCS reporter lines during heart development. (A and B) The expression of CCS-LacZ reproduced with permission from Jongbloed et al. in Circ. Res.21 (A) and Cx40GFP (B) differs at E10.5. AVC, atrioventricular canal; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle. (C and D) Expression patterns of CCS-LacZ reproduced with permission from Rentschler et al. in Development23 (C) and Cx40GFP (D) converge at neonatal stages (P0). See main text for details. H, bundle of His; LBB, left bundle branch; LPF, left Purkinje fibres; RBB, right bundle branch.
3. The developmental origin and establishment of the VCS
3.1 The origin of conductive cells
A long-standing debate in the cardiac field concerns the origin of conductive cells. This debate reflected observations that conductive cells express certain neuronal genes and the heterogeneity of cardiac cell populations potentially contributing to the VCS, i.e. cardiomyocytes, neural crest cells (NCCs) and epicardially derived cells (EPDCs).29 Cardiomyocytes originate from two contiguous populations of mesodermal progenitor cells, the first and second heart fields.30 The first heart field constitutes the cardiac crescent at E7.5 and gives rise to the linear heart tube. The heart tube subsequently elongates by addition of second heart field cells in pharyngeal mesoderm to the venous and arterial poles. The first heart field participates predominantly in the development of the left ventricle and atria, while the second heart field contributes to the right ventricle and outflow tract, in addition to part of the atria. The outflow tract is also colonized by cardiac neural crest cells that drive septation and participate in valve formation.29 Epicardially derived cells give rise to coronary vessels and cardiac fibroblasts.31 In order to address the cellular origin of the VCS, Mikawa and co-workers carried out pioneering clonal analysis experiments in the avian primary heart tube using replication-defective retroviral infection.17 They discovered that conductive cells are always found in clones together with adjacent working myocytes.32,33 This was the first evidence of the myogenic origin of the conduction system using an experimental procedure independent of the expression of molecular markers. Recently, we have taken a similar approach in the mouse heart using the α-cardiac-actin-nlaacZ mouse line, which permits retrospective clonal analysis.9 This approach is based on a defective reporter gene carrying a duplication with a stop codon in the β-galactosidase coding sequence.34 Low-frequency spontaneous intragenic recombination leads to loss of the duplication and restores the nlacZ open reading frame. The cell in which this event occurs thus carries a functional nlacZ gene that will be stably transmitted to all daughter cells, allowing visualization of clusters of clonally related myocytes. In order to define the lineage relationship between cells of the murine VCS and the surrounding working myocardium, clusters of clonally related myocytes were screened for conductive cells using Cx40-driven GFP expression. The presence of mixed clusters, composed of conductive and working myocytes, revealed that both cell types develop from common progenitor cells. Mixed clusters were observed in all compartments of the VCS, including the AVB, right and left bundle branches, and PF network, demonstrating that the entire VCS shares a common cellular origin with surrounding working myocytes.9 However, these results do not exclude a minor participation of other cell lineages in the VCS. Such a possibility is suggested by a genetic lineage analysis identifying a small proportion of the VCS that originates from Mesp1-negative cells.35 A genetic lineage analysis of NCCs using a Wnt1-Cre allele suggested that NCCs may contribute to the central conduction system,36 although the participation of NCCs in the VCS was not clearly demonstrated. The contribution of epicardially derived cells to the VCS was evaluated by lineage tracing analysis using a Tbx18Cre allele. The T-box transcription factor encoding gene Tbx18 is specifically expressed in the sinus horn and epicardium during cardiac development.10Tbx18-derived cells were shown to contribute to the sheath insulation of the AV axis but not to cardiomyocytes of the AV conduction system. To conclude, both retroviral and retrospective clonal analyses clearly demonstrate the participation of myogenic progenitors in all compartments of the VCS. As yet, there is no convincing evidence that EPDCs or NCCs make cellular contributions to the VCS.
Other Cre lines have been used to follow the progeny of different myocardial progenitor cells populations during cardiac development. Observations with cGATA-6-enhancer-Cre mice suggested that the AVC contributes to the AVN and AVB;37 however, as this enhancer is also active in these structures later in development, these mice cannot be reliably used for lineage analysis of the AVC. Expression of the T-box transcription factor Tbx2 is restricted to the primary myocardium of the AVC and outflow tract.38 Using a Tbx2Cre allele, Aanhaanen et al. showed that AVC primary myocardium participates in formation of the AVN, but not of the AVB and bundle branches. These results indicate that the AVN and AVB do not develop from a single progenitor cell population, but that central conduction system components segregate early in development. Indeed, the AVN originates from an Isl1-positive progenitor population, suggestive of a second heart field origin, although Isl1 is also expressed in a subset of cells in the AVN.39 The Mef2c-AHF-Cre line has been used to label the anterior component of the second heart field, which provides the precursors of the outflow tract and right ventricle.40 Observations using this line have revealed a large contribution of labelled cells to the right ventricle and IVS, including the AVB. However, during cardiac development this transgene is continuously expressed in the IVS and so, again, it is not suitable to follow the contribution of second heart field-derived cells to the ventricles.10 However, analysis of the Mef2c-Cre lineage revealed the absence of a ventricular contribution to the AVN, thus confirming the Tbx2-Cre result. More recently, we have generated an inducible Cre allele expressed under transcriptional control of Cx40 regulatory sequences.41Cx40Cre mice allow precise temporal control of Cre recombination in the Cx40 expression domain. As mentioned earlier, Cx40 expression initiates at the onset of trabeculation and persists in trabeculae throughout cardiac development.42 Recombination at E10.5 demonstrates that Cx40 is expressed in cells giving rise to both conductive and working myocytes in each ventricle. In contrast, recombination at E16.5 reveals that Cx40-expressing cells at this stage contribute only to the VCS, including the AVB, right and left bundle branches, and PF network.9 These data illustrate the lineage relationship between Cx40-positive embryonic trabeculae and the peripheral PF network.
3.2 Biphasic development of the VCS
Two models have been proposed to explain the mode of development of the VCS. The ‘out-growth’ model is based on the expression pattern of specific markers describing populations of conductive progenitors organized in rings along the heart tube.15,19 However, as mentioned above, none of these markers is exclusive to the conduction system, and this model depends on approximating histological, immunological, and functional criteria. The alternative ‘in-growth’ or recruitment model was postulated based on lineage analysis experiments in the avian system.43 Using replication-defective retroviral labeling, Mikawa and colleagues demonstrated that conductive cells share common progenitors with working myocytes and that the VCS develops by a process of induction and recruitment of myocytes through endothelium-derived signals.17 According to this model, conductive myocytes are non-proliferative, and subsequent growth of the conduction system occurs by further myocyte recruitment.
The nlaacZ clonal analysis has provided insights into the mode of development of the murine VCS.9 Retrospective clonal analysis consists of studying the properties of clonally related cells, such as clone size, cell identity, and cell dispersion, to provide information on the events that take place during development of a particular structure.44 In double-transgenic α-cardiac-actin-nlaacZ/Cx40GFP hearts, two classes of conductive clusters were identified, containing only conductive cells (unmixed conductive clusters) or both conductive and working cells (mixed clusters; Figure 4A). While the presence of mixed clusters demonstrates that conductive and working myocytes share a common progenitor, as discussed above, unmixed conductive clusters indicate that cells retain their proliferative potential after restriction to a conductive fate.9 Analysis of mixed and unmixed cluster size reveals that development of the mammalian VCS is biphasic; conductive myocytes develop from common progenitors with working myocytes, followed by limited proliferative out-growth. This conclusion is supported by genetic fate mapping using an inducible Cre allele at the Cx40 locus.9
Biphasic development of the ventricular conduction system. (A) Retrospective nlaacZ clonal analysis of the ventricular conduction system demonstrating two types of clusters: mixed clusters composed of conductive and working myocytes, and unmixed clusters composed of either conductive or working myocytes. (B) Biphasic model showing differentiation from a common myogenic progenitor followed by limited proliferation of conductive myocytes.
Pre-existing controversies can therefore be reconciled by a sequential model of development, according to which ‘out-growth’ follows ‘in-growth’ during development of the mammalian VCS (Figure 4B). The recruitment model is based on the arrest of proliferation of differentiating conductive cardiomyocytes.32 However, our clonal analysis suggests that conductive cardiomyocytes continue to proliferate, albeit limited to four or five cell divisions. This discrepancy may originate in the time of label induction in the mouse and avian experiments. Labelling is restricted to the cardiac tube stage in the chick model, whereas it can occur at any stage prior to observation in the nlaacZ experiment. Another discordant point concerns the nature of the progenitor cell from which the VCS differentiates, for example, whether it was a contractile or conductive myocyte.14,43 The nlaacZ experiment reveals the existence of common progenitors, but does not address their phenotype. These cells are likely to be embryonic myocytes that are distinct from mature contractile and conductive cardiomyocytes.
Analyses of mixed nlaacZ-labelled clusters provides information concerning the populations of progenitor cells in different compartments of the VCS (Figure 5). The percentage of conductive cells per cluster tends to be small in the PF network and large in the AVB, suggesting that progenitors of the central components of the VCS give rise to limited numbers of working myocytes.9 This conclusion is consistent with the broad expression pattern of reporter genes from the different mouse models described above.15 These data are also consistent with birth-dating studies showing that differentiation of the central conduction system precedes that of the peripheral conduction system.32 Moreover, this is also consistent with the early specification model recently proposed by Moorman and Christoffels, according to which a pool of Tbx2/Tbx3-positive cells maintains an undifferentiated phenotype and low proliferation rates compared with chamber or ventricular myocardium.15 In the peripheral conduction system, the nlaacZ clonal analysis revealed a striking difference in the potential of these progenitor cells between right and left ventricles. While the relative number of conductive cells is equivalent in both ventricles, the number of working cardiomyocytes is greater in mixed clusters in the left compared with the right ventricle. Using the Cx40Cre allele to genetically label trabecular myocardium, we observed a more extensive contribution to the left than to the right ventricular myocardium in the same proportion as nlaacZ clusters (Figure 5). These data suggest that while PF progenitor cells in both ventricles are Cx40 positive, the mode of VCS development differs in each ventricle. This difference may underline the origin of these progenitor cell populations, which derive from the first heart field for the left ventricle and second heart field for the right ventricle.29 Moreover, the T-box transcription factor Tbx5, both by its pattern of expression,45,46 and by its role of transcriptional activator of the Cx40 gene, may be one of the major genes involved in the development of these progenitors. By E16.5, Cx40-expressing cells are restricted to ventricular Purkinje fibres.9 The fast-conducting phenotype is thus broadly present in the embryonic ventricle prior to establishment of the specialized VCS and the mass of working chamber myocytes. Trabeculated hearts therefore do not require a fully specialized Purkinje fibre network for co-ordinated contraction. This conclusion has implications for both the ontogeny and the evolution of the VCS.14
Model of development of the VCS from two populations of progenitor cells. (A) The percentage of working myocytes in nlaacZ mixed clusters is higher in the left than in the right ventricle. This right–left discordance is also observed after Cx40-Cre induction at E10.5, suggesting that VCS progenitor cells are Cx40 positive. By E16.5, Cx40-derived cells give rise only to conductive cardiomyocytes. (B and C) At E10.5, the central conduction system derives from AV progenitors (pink) that proliferate at low rates and express markers such as Tbx3 or CCS-LacZ. nlaacZ clusters in the central conduction system contain a high percentage of conductive cells per mixed cluster. The peripheral conduction system is derived from a distinct population of Cx40-positive cells that is more extensive in the left than in the right ventricle. (C) At E16.5, the results from Cx40-Cre lineage analysis suggest that a differentiation step has definitively occurred and that both the central and peripheral VCS are derived from Cx40-expressing cells.
4. Molecular mechanisms regulating VCS development
4.1 Transcription factors
A panel of different transcription factors has been implicated in the differentiation and development of the VCS.15,47,48 Concerning central components of the VCS, AVN development is highly dependent on the transcription factor Nkx2.5.49 In Nkx2.5 null embryos, the lack of Mink-LacZ expressing cells in the AVC at E9.0 suggests the absence of an AVN primordium. In Nkx2.5 heterozygous adult hearts, the AVN is significantly reduced in size, being comprised only of a core of Cx40 and Cx45-positive cells.49 Using the nomenclature of Aanhaanen et al., this corresponds to the loss of the compact AVN and maintenance of lower nodal cells.10Nkx2.5+/− mice display AV conduction defects, as shown by a prolonged P–R interval on ECG recordings, consistent with absence of the AVN.49,50 Conditional loss of Nkx2.5 in the ventricles or at perinatal stages induces the development of a hypoplastic AVN and progressive AV block.51,52 The progression of the phenotype is attributed to the selective degeneration of the AVN after birth.52 These data demonstrate that Nkx2.5 plays a role in the maintenance of the AVN in the postnatal heart. In man, mutations in NKX2.5 cause a variety of cardiac anomalies frequently associated with conduction blocks.53 Histological examination of post-mortem hearts has shown that progressive AV block is associated with degeneration of the AVN and replacement by fibrosis and adipose tissues.52
Several transcription factors have been implicated in the specification, development, and maturation of the AVB and bundle branches. Tbx3, encoding a T-box-containing transcriptional repressor, plays a role in the specification of the AVB and bundle branches. Tbx3 null embryos die between E12.5 and E15.5, and present cardiac malformations, including ventricular septal defect with a short and blunted septum.54 In the absence of Tbx3, a number of working myocardial markers are expressed at the crest of the septum, and Nkx2.5, normally over-expressed in the developing AVB, is present at the same level throughout the septum. These molecular changes are accompanied by the absence of cell cycle exit of cells at the crest of the septum. Tbx3 thus appears to induce specification of AV conduction system progenitors by repressing expression of working myocardial genes. The early death of Tbx3 null embryos has prevented investigation of Tbx3 function in later steps of VCS development; nevertheless, heterozygous mice display a normal cardiac phenotype structurally and functionally.54 This is not the case for two other transcription factors, Nkx2.5 and Tbx5, for which conduction defects are observed when only one copy of the gene is removed.46,49 As mentioned above, a hypoplastic AVB is observed in adult Nkx2.5 heterozygote hearts or ventricular restricted Nkx2.5 knock-out mice.49,52 These mice display specific conduction defects in the AVB, as shown by intracardiac electrophysiological recordings. Tbx5 is also expressed in the developing AVB and bundle branches at the crest of the IVS.46 In Tbx5 heterozygous mice, the RBB is absent or severely abnormal, correlating with AV conduction defects. The AV conduction system develops in a ring structure that is maintained after birth in Tbx5+/− but not wild-type hearts. Tbx5 is thus required for the morphological maturation of the AVB and LBB and essential for patterning and function of the RBB. These observations underscore the link between defective patterning of the developing conduction system and functional abnormalities of the mature conduction system. Mutations in human TBX5 cause Holt–Oram syndrome, characterized by congenital heart defects and conduction system anomalies, including variable degrees of AV block.55 Moskowitz and colleagues have shown that Tbx5 and Nkx2.5 co-operate during development of the VCS.56 This co-operation is mediated by the expression of Id2, encoding a helix–loop–helix transcriptional repressor. In Id2−/− mice, the AVN develops normally, but the development of the AVB and bundle branches is impaired, correlating with conduction defects, such as long QRS and LBB block. In Tbx5+/−/Nkx2.5+/− compound heterozygous mice, Id2 is not expressed, and cells at the crest of the IVS fail to exit the cell cycle. These data suggest that Id2, like Tbx3, acts through the repression of myogenic genes and promotes VCS differentiation.15 However, in Tbx3−/− embryos, the expression of Tbx5 and Id2 are unchanged, indicating that these factors are not sufficient to promote VCS differentiation.54 Together, it appears that differentiation and maturation of the AVB and BB requires a combinatorial effect of different transcription factors, including Tbx3, Tbx5, Id2, and Nkx2.5.
To date, Nkx2.5 is the only transcription factor implicated in the differentiation and maturation of the PF network. Ventricular restricted inactivation of Nkx2.5 provokes a massive trabecular overgrowth and a prolonged QRS on ECG.52 Transcriptional profiling revealed up-regulation of BMP10 and conduction-specific genes in mutant hearts, suggesting a fundamental defect in ventricular cell lineage maturation. In addition, the peripheral PF network is severely hypoplastic in Nkx2.5 haplo-insufficient mice.49,57 Using Cx40GFP mice, we have shown that while development of trabeculae in Nkx2.5+/− fetuses is normal, postnatal development of the PF network fails (Figure 6). Furthermore, chimeric analyses demonstrated that Nkx2.5 plays a cell autonomous role during the perinatal period in the normal formation of the PF network.57
Hypoplasia of the Purkinje fibre network in Nkx2.5 haplo-insufficient mice. GFP expression of the left VCS shows a PF network organized of dense ellipses in Cx40GFP mice, while the PF network is reduced and fails to organize in elliptic structures in Nkx2.5+/− mice. AVB, atrioventricular bundle; LBB, left bundle branch; LPF, left Purkinje fibre network.
In addition to these transcription factors involved in the differentiation and maturation of the VCS, several other factors have been implicated in the expression of specific ion channel and gap junction protein encoding genes without overtly affecting VCS morphoplogy. In Hf1b−/− or Hop−/− mice, conduction defects were found to be secondary to dysregulation of the gap junction gene Cx40 in the absence of structural abnormalities of the conduction system.58,59 Ventricular restricted knock-out of Hf-1b causes electrophysiological abnormalities, including fatal ventricular arrhythmias.60 This phenotype is correlated with abnormalities in Cx43 levels, myocyte size, activation spread, and coronary arterial structure and function. The Iroquois homeobox gene Irx5 is detected in the sub-endocardial myocardium of the ventricles, and Irx5−/− mice are susceptible to tachyarrythmias.61 This phenotype is associated with the loss of the transmural potassium current Ito, which is important for cardiac repolarization.
Two categories of transcription factors can thus be distinguished based on their roles in the cardiac conduction system. Cardiac conduction disturbances may originate from defects in either the function and/or the structure of the conduction system. In the first category, several transcription factors regulate expression of genes encoding ion channel or gap junction proteins. Ion channels are essential for the acquisition of electrophysiological properties of conductive cardiomyocytes and gap junctions for the conduction of electrical activity. The second category contains transcription factors that specify cell lineages destined to become part of the cardiac conduction system. Defects in specification entail abnormal morphological development of the conduction system. Nkx2.5 can be classified in both categories and represents the major transcription factor playing a role in differentiation and development of the VCS, as the development of all compartments of the VCS requires a dose- and time-dependent expression of Nkx2.5.
4.2 Extrinsic signalling pathways
Development of the VCS also depends on extrinsic parameters, which in turn induce the transcriptional pathways directly involved in the differentiation and maturation of conductive cells. As described in section 3, the heart is comprised of diverse cell populations: cardiomyocytes, endothelial cells, smooth muscle cells, and fibroblasts. These cells originate from different progenitor populations, including the first and second heart fields, NCCs and EPDCs. As discussed previously, while NCCs and EPDCs may not contribute directly to the VCS, these cells play a role in VCS development. After ablation of NCCs in chick embryos, the electrical activation switch from apex to base does not occur but is maintained from base to apex in the IVS.62 This phenotype results from the absence of compaction and electrical isolation of conduction system bundles. NCCs migrate to the crest of the IVS in close vicinity to the AV conduction system.36 NCC-restricted deletion of Hf1b led to atrial and atrioventricular dysfunction resulting from deficiencies in the neurotrophin receptor trkC.63 Moreover, the inhibition of EPDCs in chick embryos disturbs the development of the PF network.64 Together, these results suggest that NCCs and EPDCs are required for normal VCS development.65
Cross-talk between endothelial and myocardial cells plays a major role during cardiac development. Endothelin-1 (ET-1) and neuregulin are two factors secreted by endothelial cells that play important roles in the development of ventricular trabeculae. Endothelin receptor and neuregulin null embryos die early during embryogenesis due to cardiac defects and absence of trabeculae.66,67 Gourdie and co-workers have demonstrated that an endothelial signal induces perivascular PF differentiation in chick hearts.68,69 Avian and murine embryonic cardiomyocytes express conductive markers when cultured in the presence of ET-1. In contrast, ET-1 does not up-regulate the expression of the CCS-LacZ transgene during in vitro culture of embryonic mouse hearts, but neuregulin is able to do so during a restricted time window.70 The modification of CCS-lacZ expression by neuregulin treatment is associated with conduction disturbances detected by optical mapping. ET-1 and neuregulin have been shown to direct the differentiation of embryonic murine cardiomyocytes into conductive cardiomyocytes.71 Together, these data identify an early role of the endothelium, mediated by neuregulin or endothelin, in trabeculae formation and consequently in VCS development.
ET-1 signalling acts through the activity of the endothelin-converting enzyme ECE-1, which is responsible for cleavage of the precursor Big-ET into active ET-1.72 ECE-1 is expressed in the heart in close apposition with the developing VCS in both chick and mouse.73 Over-expression of ECE-1 in chick hearts induces development of ectopic peripheral PF.72 ECE-1 thus plays a key role in defining the time and the location of PF differentiation within the embryonic myocardium. The expression of ECE-1 depends on haemodynamic parameters.74 Pressure overload induces the expression of ECE-1 associated with precocious emergence of the mature apex-first activation pattern. Sedmera and collaborators have shown that mechanical loading is required for normal differentiation of the conduction system as well as for acquisition of the apex-to-base activation pattern underlying the functional His–Purkinje system.75,76 These findings identify haemodynamics as a key epigenetic factor in development of the cardiac conduction system. However, these studies were all performed in chick embryos with emphasis on periarterial PFs, and there is as yet no in vivo evidence that ET signalling is necessary for differentiation of the VCS in the mouse.
In addition to the murine and avian models, new insights into the genetic and epigenetic control of conduction system development stem from mutagenesis approaches in the zebrafish. Indeed, the power of the zebrafish experimental model is revealed in a recent study using optogenetics to delimit and control the electrical activity of the heart in vivo.77 This work opens up the possibility for developmental biologists to investigate the onset of cardiac function more precisely. In addition, analysis of the effect of disruption of cardiac conduction in the absence of contraction in the zebrafish has recently demonstrated that electrical activation per se influences cardiomyocyte architecture and overall cardiac morphogenesis.78 Feedback between form and function is thus essential for development of the VCS and normal heart development.
5. Concluding remarks
The VCS is responsible for the synchronization of the heartbeat, and physiologists have extensively studied the automatic and rhythmic activity of these specialized cardiomyocytes. In this review, we have seen that, despite important discrepancies between human and mouse cardiac physiology, mouse VCS anatomy is close to that of the human heart, making the mouse an attractive model for developmental biologists. Indeed, murine genetics, including the use of VCS reporter lines, together with experimental manipulation in avian systems, have addressed the cellular origin of the VCS and defined a number of the genes and signalling molecules regulating establishment of the VCS. Moreover, mouse models of congenital heart defects emphasize the role of VCS morphology in cardiac function. These data show that transcription factors play roles in differentiation of conductive cells as well as in the development and maturation of the network. Convergent genetic and physiological studies in different model systems will further define the relationship between structure and function in the VCS and cardiac rhythm and contribute to new therapeutic approaches for cardiac arrhythmia.
Supported by the European Community's FP6 contract Heart Repair (LSHM-C7-2005-018630) and FP contract CardioGeNet (Health-2007-B-223463) and the Association Française contre les Myopathies.
We are grateful to Vincent Christoffels for discussion and comments on the manuscript.
Conflict of interest: none declared.
This article is part of the Spotlight Issue on: Cardiac Development
. Development of the right ventricular inflow tract and moderator band: a possible morphological and functional explanation for Mahaim tachycardia. Circ Res 2005;96:776-783. doi:10.1161/01.RES.0000162000.03997.65.
. A GATA-6 gene heart-region-specific enhancer provides a novel means to mark and probe a discrete component of the mouse cardiac conduction system. Mech Dev 2001;108:105-119. doi:10.1016/S0925-4773(01)00500-7.
. Isl1 identifies a cardiac progenitor population that proliferates prior to differentiation and contributes a majority of cells to the heart. Dev Cell 2003;5:877-889. doi:10.1016/S1534-5807(03)00363-0.
. The right ventricle, outflow tract, and ventricular septum comprise a restricted expression domain within the secondary/anterior heart field. Dev Biol 2005;287:134-145. doi:10.1016/j.ydbio.2005.08.041.
. Nkx2-5 pathways and congenital heart disease; loss of ventricular myocyte lineage specification leads to progressive cardiomyopathy and complete heart block. Cell 2004;117:373-386. doi:10.1016/S0092-8674(04)00405-2.
. Hemodynamic-dependent patterning of endothelin converting enzyme 1 expression and differentiation of impulse-conducting Purkinje fibers in the embryonic heart. Development 2004;131:581-592. doi:10.1242/dev.00947.