Reconstruction of the Patterns of Gene Expression in the Developing Mouse Heart Reveals an Architectural Arrangement That Facilitates the Understanding of Atrial Malformations and Arrhythmias
Firm knowledge about the formation of the atrial components and of the variations seen in congenital cardiac malformations and abnormal atrial rhythms is fundamental to our understanding of the normal structure of the definitive atrial chambers. The atrial region is relatively inaccessible and has continued to be the source of disagreement. Seeking to resolve these controversies, we made three-dimensional reconstructions of the myocardial components of the developing atrium, identifying domains on the basis of differential expression of myocardial markers, connexin40, and natriuretic precursor peptide A. These reconstructions, made from serial sections of mouse embryos, show that from the outset of atrial development, the systemic and pulmonary veins are directly connected to the atrium. Relative to the systemic junctions, however, the pulmonary venous junction appears later. Our experience shows that three-dimensional reconstructions have three advantages. First, they provide clear access to the combined morphological and molecular data, allowing clarification and verification of morphogenetic concepts for nonmorphological experts and setting the scene for further discussion. Second, they demonstrate that, from the outset, the myocardium surrounding the pulmonary veins is distinct from that clothing the systemic venoatrial junctions. Third, they reveal an anatomical and molecular continuity between the entrance of the systemic venous tributaries, the internodal atrial myocardium, and the atrioventricular region. All these regions are derived from primary myocardium, providing a molecular basis for the observed nonrandom distribution of focal right atrial tachycardias.
The atrial chambers represent one of the most complex morphological and electrical areas of the heart. These chambers are the origin of many abnormal rhythms1 and the seat of many congenital malformations.2 There are many reasons, therefore, why we need to understand their origin and development. Because of their rapid transformation and dorsal position, these aspects are difficult to study experimentally and many controversies persist regarding their development, ranging from the very existence of a sinus venosus,3 through whether the pulmonary vein terminates within the sinus venosus4–6 or the atrium,3,7 to the myocardium surrounding the pulmonary veins as a substrate for arrythmogenesis.8,9 It is our contention that these ongoing issues will only be resolved by making three-dimensional reconstructions of the dynamic and complex structure of the developing heart, associated with the patterns of relevant genes. In the present study, therefore, we used three-dimensional reconstructions of myocardial probes to visualize the atrial components in the mouse. We further subdivided this myocardium into different phenotypes on the basis of differential expression of the markers connexin40 (Cx40) and natriuretic precursor peptide (Nppa, also known as ANF). The expression of the gap junction protein Cx40 has been shown to delineate regions of fast-conducting myocardium as opposed to slow-conducting myocardium within the primary heart tube.10 The expression of Nppa is known to be conserved within vertebrates, qualifying it as a marker for chamber myocardium.10
Materials and Methods
Tissue Processing and In Situ Hybridization
We used mouse embryos ranging from embryonic day (E) 9.5 to E16.5, following our previously described protocols.11 Serial sections were consecutively stained (Figure 1A) with probes against Nppa, Cx40, and a myocardial marker consisting of a mixture of probes (750 ng/mL each) against α-myosin heavy chain, β-myosin heavy chain, and cardiac troponin I, as detailed in Soufan et al.12
Our protocol for acquisition and reconstruction of images has been recently described,12 although the use of consecutive staining of sections in this study demanded some adaptations in the thresholding and labeling protocol. So as to visualize and label the myocardium using Amira (version 3.1; TGS Template Graphics Software, http://www.tgs.com), the 8-bit grayscale sets were processed to a Label-Field file, which is a 3-D data set in which the voxels are assigned to the myocardial label. Duplicates of this myocardial set are then used as templates for the adjacent manually labeled Nppa and Cx40 images. The myocardium, Nppa, and Cx40 patterns were interpolated between the neighboring sections and merged to obtain six labels (see Definitions). The labels were corrected by manual closure of holes and smoothing in 3D to compensate for the jagging and surface breakage. The Label-Fields were then converted to three-dimensional meshes. In total, we made 15 reconstructions from hearts ranging in age from E9.5 to E16.5, using the embryos from E9.5 to 11.5 to visualize the developing atria. The full range of reconstructions has been used for volume quantification.
The different patterns of expression of Cx40 and Nppa permit us to divide the myocardium into three major types. We have labeled these primary, appendage, and mediastinal myocardium. We also identified a fourth, albeit minor, type that we call the spurious myocardium. The tissue types and corresponding color codes used in the figure, are shown in Figure 1. The distinct myocardial phenotypes may not represent lineages, as the observed temporal changes can result either from changes in gene-expression, in other words differentiation, or lineage, which is growth and migration.
This myocardium is negative for both Cx40 and Nppa. It is the slowly conducting myocardium of the primary heart tube.10
Appendage myocardium is positive for both Cx40 and Nppa and forms the expanding parts of the atrial chambers that become the appendages.10
Myocardium positive for Cx40, and negative for Nppa, was found in the dorsal wall of the atrium in conjunction with the dorsal mesocardium, surrounding the orifice of the pulmonary vein.
At stage E11.5, we observed a small area of myocardium that was negative for Cx40, but positive for Nppa. This was the area formed by the septum spurium and was, therefore, dubbed the spurious myocardium. As yet, we do not know whether, in later stages, this phenotype will remain confined to this septum.
We found some areas of myocardium in which we could not accurately reconstruct the expression. This was particularly the case for the right venous valve, a thin dual-layered structure comprising a layer of primary myocardium derived from the caudal fold of the orifice of the right sinus horn and a layer of mediastinal myocardium derived from the right atrial wall.13
We labeled all cardiac tissues not stained by the mix of myocardial-specific probes as mesenchyme.
Systemic Venoatrial Junction
We have used the name systemic venoatrial junction (SAJ) rather than sinuatrial junction to describe the junction between the systemic venous tributaries and the atrium because we have not observed any cardiac segment that could be distinguished as the “sinus venosus.”
Pulmonary Venoatrial Junction
The junction between the primordium of the pulmonary vein and the atrium is described as the pulmonary venoatrial junction (PAJ).
Collectively, we describe the SAJ and PAJ as the venous junctions.
To supplement our results, we have included a number of movie-clips to facilitate the understanding of the intricate 3-D morphology of atrial development (see online data supplement available at http://circres.ahajournals.org).
Resolution and Reproducibility
As shown in Figure 1, the section thickness of 12 μm allowed accurate reconstruction of thin-walled structures, such as the left systemic vein (LSV) and the left atrial wall. Because of the consecutive nature of staining, phenotypes could be recognized only at three times the thickness of the section. This constraint only hindered the phenotyping of the thin dual-layered leaflets of the venous valves. The sagittal section through the reconstruction of the E10.25 embryo displays high similarity with three consecutive sagittal sections of a nonreconstructed embryo of similar stage, validating the reproducibility of the reconstructions (Figure 2).
9.5 Days of Development
Online Movie 1 shows the position of the E9.5 heart in the embryo. This movie starts with a lateral view, and ends with a dorsal view of the venous junctions, showing the entire atrial myocardial component (Figure 3A). Most of the mesenchyme and dorsal mesocardium has been removed, revealing the lumen of the systemic veins. Figure 3A shows only the most craniodorsal tip of the dorsal mesocardium. This wedge-shaped structure is composed of loosely arranged mesenchyme, comprising the pulmonary venous primordium (PVP), albeit that at this earliest stage it lacks any patent connection with the atrial lumen. When viewed from the dorsal or ventral sides (Figure 3A and 3B), two ridges of mediastinal myocardium enclose the future PAJ. This is the myocardium we have labeled as being mediastinal.14 The ridges themselves are clearly seen morphologically (Figure 4) and have been described previously as the right and left pulmonary ridges (LPR and RPR).15 The arrangement is well seen in a section (Figure 4A). The loosely arranged cells of the dorsal mesocardium surround the developing pulmonary vein (Figure 4A), forming a continuum with the mesenchyme surrounding the foregut. The section reveals the bilateral nature of the PVP, showing regular cuboidal epithelium near the pulmonary ridge. Comparison of this section with the reconstruction (Figure 4B) and an scanning electron micrograph (Figure 4C) shows striking similarities.
When viewed dorsally (Figure 3A; online Movie 2), the almost perfect symmetry of the atrial chamber, apart from the size of the appendages, is striking. The myocardium of the left atrial appendage bulges slightly more than that of the right appendage, whereas a significantly broader cuff of primary myocardium is present around the right systemic venous tributary compared with the left. The mass of mesenchyme between the orifices of the right and left systemic venous tributaries has been called the sinus septum (Figures 3B and 4B).6,13,16 This mesenchyme is contiguous with the PVP, itself mesenchymal at this stage. The cranial view of the caudal segment (Figure 4B) displays the entirety of the most upstream region of myocardium, revealing that primary myocardium surrounds the SAJ, whereas mediastinal myocardium surrounds the PAJ. Thus, the venous junctions are composed collectively of primary myocardium, mediastinal myocardium, and a wedge of dorsal mesenchyme (Figure 5A). Taking the foregut as a midline reference point, the junctions initially occupy a midsagittal position. The systemic veins approach the heart tube dorsally and bend sharply in cranial direction just before entering the heart tube (online Movie 2). The PVP, in contrast, is connected dorsally to the atrium via the dorsal mesocardium (Figure 4A). Removing the appendage myocardium (online Movies 3 and 4) reveals that the primary myocardium forms a continuum from the SAJ, through the floor of the atrial chamber, to the atrioventricular canal (AVC), which is positioned ventral to the developing left atrium. The atria are also continuous with the systemic veins, it proving impossible at this early stage to recognize a “sinus venosus” as a discrete morphological entity.
10.25 Days of Development
Roughly half a day later, the mediastinal myocardium has come to enclose the entirety of the PVP, and now extends toward the SAJ (Figure 3C). The SAJ has shifted from its medial-caudal position to a more laterodorsal position (compare Figure 3A and 3B with 3C and 3D). Both systemic venous tributaries now enter dorsally into the cavity of the right atrium, becoming enclosed by the pericardial cavity as they become covered with primary myocardium (Figure 3C). The parts of the systemic veins still embedded in the body mass, in contrast, remain covered by mesenchyme rather than myocardium.
By this time, the right and left venous valves (LVV and RVV) have become evident, with the RVV being more prominent (Figure 3D). The valves merge cranially to form the septum spurium and reach caudally and medially to the most caudal part of the RPR. The venous valves are thin bilaminar structures (Figure 1) and their phenotype cannot accurately be displayed. We have shown them as undefined myocardium. The LPR and RPR have become even more pronounced, albeit that loosely arranged mesenchyme is still evident (online Movie 5, Figure 3D). The AVC is now positioned in the midline of the embryo (Figure 3D), as opposed to its left-sided position in the E9.5 embryo (online Movie 4). Using the dorsal mesocardium, and the future site of the pulmonary venous orifice, as the midline, both the entrance of the systemic venous tributaries and the AVC have shifted toward the right over the course of about half a day.
11.5 Days of Development
Another 30 hours later, the reconstruction shows a marked increase in primary myocardium enclosing the systemic veins (Figure 3E). Only at this stage has the PVP luminized, providing continuity between the lumens of the pulmonary vein and the atrial cavity. The vein itself is positioned even more caudally (compare Figure 3C with 3E), and is surrounded by mediastinal myocardium, albeit that the layer of mediastinal myocardium at the caudal side is very thin. A sheet of mesenchyme and mediastinal myocardium is observed in the atrium cranial to the RPR, separating the left atrium from the right (Figures 3F and 6, online Movie 6). The myocardial part of this sheet is the primary atrial septum (SP), whereas the mesenchyme forms its cap. The LPR and RPR are no longer distinguishable. Within this stage, the mesenchymal cap of the SP fuses ventrally with the inferior and superior cushions of the AVC (Figures 3F and 6, online Movie 6). A perforation in the cranial part of the primary septum then produces the so-called “ostium secundum.” The process of septation places the PAJ in the left, and the SAJ in the right atrium. The venous valves have now become more pronounced and the septum spurium now displays a phenotype never previously seen, being negative for Cx40 and positive for Nppa (Figure 3F).
Quantification of Myocardial Components
We have measured the growth of the three phenotypically distinct atrial components between E9.5 and E16.5 (Figure 7). From E9.5 through E10.5, the relative growth of the primary myocardium is less than that of the appendage myocardium and mediastinal myocardium. Beyond E10.5, all three atrial components show a similar growth rate.
It is axiomatic that knowledge of the spatiotemporal pattern of expression of key cardiac genes is fundamental to our understanding of the intricate processes involved in cardiac morphogenesis. Cardiac morphology itself, however, remains a somewhat esoteric subject, if we are to judge by the ongoing controversies. Opinions still diverge on the origin of the pulmonary veins3–7 and on the existence of internodal tracts.9,17 These controversies are often difficult to appreciate for “outsiders.” Indeed, much of the disagreement may merely be caused by miscommunication, because it is far from easy to form a mental image of any complex three-dimensional structure, even for experts. We have made several novel observations from our reconstructions. First, a “sinus venosus” does not exist as a discrete segment of the heart tube in the mouse. Second, from the outset of their development, the systemic and pulmonary tributaries are surrounded by fundamentally different types of myocardium. We will expand on these and other observations.
Development of the Venous Junctions
Our reconstructions show that the venous junctions themselves have two distinct parts, namely the SAJ and the PAJ (Figure 5). Both junctions are surrounded by myocardium and mesenchyme. The systemic component is exclusively surrounded by primary myocardium, a domain that expresses neither Cx40 nor Nppa, and by the dorsal mesenchyme of the so-called sinus septum,16 which separates the orifices of the left-sided and right-sided systemic venous tributaries as they enter the heart.13 From the outset, the pulmonary component is exclusively associated with mediastinal myocardium, the domain expressing Cx40, but not Nppa, and also with dorsal mesenchyme that is contiguous caudally with the mesenchyme of the sinus septum, and dorsally with the mesenchyme surrounding the foregut. We had already distinguished these types of myocardium using different markers and transgenes.14
The mesenchymal wedge within the mediastinal myocardial component acts as a conduit for entrance of the pulmonary vein3,4,15,18 (Figure 3). Comparison of the dorsal views of the reconstructions of E9.5, E10.25, and E11.5 shows that the orifices of both the left and right systemic venous tributaries progressively become surrounded by primary myocardium. This muscularization occurs at those places where the veins bulge into the pericardial cavity, suggesting recruitment from the adjacent mesenchyme into the myocardial lineage, a possibility also suggested by recent lineage studies.19 This recruitment of mesenchyme is also likely to involve the sinus septum, because at E10.25, no mesenchymal component was observed in the systemic venous orifice (Figure 3D). Only during E11.5 do the venous valves develop as bilaminar structures, and only at this stage is a systemic sinus venosus seen as a morphologically discrete entity.
Both the SAJ and the PAJ are initially positioned symmetrically in the body, but are molecularly distinct from one another from the outset (Figure 5). This molecular difference persists over the period in which, in half a day of development, the SAJ becomes exclusively connected to the right atrium, whereas the PAJ remains in the midline of the body, draining to the left atrium only because the atrial septum develops to the right of the PAJ (Figure 5B). Within this short period of half a day, the mesenchymal connection disappears between the SAJ and PAJ, mediastinal myocardium comes to surround the canalizing pulmonary orifice, and the wall of the left systemic venous tributary, a structure that becomes the coronary sinus in the human, acquires a primary myocardial phenotype at its face with the atrial myocardium. Although closely apposed, these two walls of myocardium can be clearly distinguished not only anatomically, but also molecularly (Figure 6, white asterisk).
The Sinus Venosus
A sinus venosus, the confluence draining the systemic veins to the heart,20 is always recognizable in the formed hearts of lower vertebrates like the fish. In these species, this segment is contained within the pericardial cavity, albeit that its wall contains sparsely distributed, if any, myofibrils. This configuration is not seen in the formed mammalian heart, with embryological textbooks generally contending that the sinus venosus has become incorporated into the right atrium during development, forming the so-called smooth-walled sinus venarum. It is still claimed by some that the developing pulmonary veins drain within an area derived from the sinus venosus.4–6 Our reconstructions show unequivocally that both the systemic venous tributaries, and the future pulmonary venous orifice, are connected directly and separately to a structure that is classically called atrium. Our current findings, therefore, confirm our previous description.3 It can be argued that, by definition, the confluence of the systemic venous tributaries should be called the sinus venosus. Be that as it may, our reconstructions show that, in the developing mouse heart, this confluence is an integral part of the atrium, and never possesses an independent existence.
We used the tag “sinus venosus” in figure of our previous publications simply to denote the midpoint of a venous lumen lying upstream from an atrium, following classical convention. The same holds true for the tag “atrioventricular canal,” which we have depicted as being positioned between the atrium and the ventricle, albeit that it is impossible to distinguish this area in molecular terms from the floor of the atrium (online Movie 6). For simple description, such tagging should not cause problems. If we are to make functional and developmental correlations with our tags, however, they need to be molecularly defined. For example, some have used staining of expression of the HNK-1-epitope,5,6 or the expression of CCS-LacZ,9 to draw the inference that the pulmonary veins originate from the sinus venosus. This then lead inexorably to the conclusion that “embryonic conduction tissue” surrounded the pulmonary venous orifice. This inference was then used as an explanation for the arrhythmias known to originate in the myocardium surrounding the pulmonary veins. Our reconstructions clearly show that such an inference is unfounded. From the outset, the pulmonary venous orifice in the mouse is surrounded by Cx40-positive mediastinal myocardium, this pattern of expression being characteristic of fast-conducting rather than primary myocardium.10 Neither the nodal tissues, nor the myocardium of the embryonic heart tube, express Cx40.21,22 In the human, the nonuniform anisotropic orientation of the working pulmonary myocytes in itself is sufficient to provide a substrate for the rapid circular reentry now known to be the substrate of atrial fibrillation originating in the pulmonary veins.23 The claimed existence of specialized individual conduction cells in the myocardial sleeves surrounding these pulmonary veins in patients with atrial fibrillation,24 remains to be functionally proven. Our current findings, showing that the pulmonary myocardium has never had a nodal phenotype, cast still further doubt on this possibility.
Atrial Building Plan and Arrhythmias
Our reconstructions have revealed a complex arrangement of the different myocardial components of the walls of the atrial chambers. In molecular terms, the “roof” and lateral walls of the atrial chambers are made of appendage and mediastinal myocardium, both positive for Cx40, whereas the “floor” consists of Cx40-negative primary myocardium. A similar morphological configuration was described by Keith and Flack.25 According to them, systemic venous blood enters the heart through they sinoauricular junction and exits through the AVC. We have shown both of these junctions to be primary myocardium. Although all types of myocardium increase in volume during development, over the period E9.5 to E10.5, the relative growth of the primary myocardium is less than that of the chamber-specific and mediastinal myocardial domains. After E10.5, growth is proportional (Figure 7). Assuming that the growth of each component between E9.5 and E10.5 can be attributed to proliferation of cardiomyocytes only, the cell-cycle lengths that correspond to these increases in volume would be 17.2, 9.9, and 5.5 hours for the primary, appendage, and mediastinal myocardium, respectively. The extremely short cell-cycle for the mediastinal myocardium is very unlikely. Its growth in this period is surely partly attributable to recruitment from neighboring pools, such as the primary or appendage myocardium, or from dorsal mesenchyme. Recruitment from dorsal mesenchyme is in line with the observations of Cai et al.19 Obviously, the myocardial phenotypes do not necessarily represent lineages. Temporal changes can also result from differentiation, as opposed to growth and migration.
Because of its initially slow growth rate, the primary myocardium surrounding the entrance and exit of the atrial chamber appear as constrictions or rings, as has first been described by Benninghoff.26 This is in line with proliferation studies.27,28 These rings at the entrance and exit of the atrium have been termed the sinuatrial and atrioventricular rings, respectively, and have been associated with the developing conduction system.29 We have recently shown that the transcriptional repressor Tbx3 is present in both these areas, as well as in the internodal region.30 Online Movie 7 shows a reconstruction of the pattern of expression of this transcription factor in a heart from E9.5. This pattern is strikingly similar to the arrangement of primary myocardium of the E9.5 heart presented in this study (online Movies 4 and 8). In normal development, only the sinus and atrioventricular nodes persist as anatomically recognizable derivatives of this primary myocardium. It is conceivable, nonetheless, that other vestiges of this more extensive primary myocardium form the substrates for focal right atrial tachycardias, abnormal rhythms known to demonstrate a characteristic nonrandom distribution. Such focal origins are more frequently observed along the long axis of the terminal crest, around the atrioventricular bundle, the orifice of the coronary sinus, and in the myocardium forming the vestibules of the tricuspid and mitral valves.1 All of these areas initially consisted of primary myocardium, and also enclose the regions of the highly contentious internodal tracts.17 It is highly likely, therefore, that remnants of the primary myocardium provide the substrates for arrhythmogenesis described within these regions.
In this investigation, we have used three-dimensional reconstructions to illustrate the development of the atrial chambers of the heart, our results providing a bridge between meticulous anatomic descriptions provided almost a century ago25,26 and postmillennium demonstrations of molecular patterns of gene expression.10,30 We recognize that our reconstructions of the distinct myocardial phenotypes do not necessarily represent distinct cardiac lineages, but this does not detract from their intrinsic value. To prove that these myocardial phenotypes also represent lineages would require a separate study. It has been proposed that the heart is composed of a limited number of genetic modules.31 Following this line of reasoning, it is intriguing to speculate that the mediastinal myocardium is a novel component added to the heart during evolution concomitant with the need to establish a pulmonary circulation. The reconstructions show that a wedge of myocardium, including the atrial septum, interposes between the left and right appendages. The building plan of the atrial chambers resulting from our reconstructions will hopefully make the area more readily understood by nonmorphologists. It should also provide clues to the understanding of the development of the atrial components of the anatomically recognized conduction system and into the origin of focal atrial tachycardias.
M.J.B.v.d.H is supported by the Netherlands Heart Foundation grant M96.002. R.H.A. and S.W. are supported by the British Heart Foundation.
Original received August 17, 2004; revision received November 3, 2004; accepted November 4, 2004.
Gruber PJ, Epstein JA. Development gone awry: congenital heart disease. Circ Res. 2004; 94: 273–283.
Blom NA, Gittenberger-de Groot AC, de Ruiter MC, Poelmann RE, Mentink MM, Ottenkamp J. Development of the cardiac conduction tissue in human embryos using HNK-1 antigen expression: possible relevance for understanding of abnormal atrial automaticity. Circulation. 1999; 99: 800–806.
Moorman AFM, Christoffels VM. Cardiac chamber formation: development, genes and evolution. Physiol Rev. 2003; 83: 1223–1267.
Moorman AFM, Houweling AC, de Boer PAJ, Christoffels VM. Sensitive non-radioactive detection of mRNA in tissue sections: novel application of the whole-mount in situ hybridization protocol. J Histochem Cytochem. 2001; 49: 1–8.
Soufan AT, Ruijter JM, van den Hoff MJB, de Boer PAJ, Hagoort J, Moorman AFM. 3D reconstruction of gene expression patterns during cardiac development. Physiol Genomics. 2003; 13: 187–195.
Franco D, Campione M, Kelly R, Zammit PS, Buckingham M, Lamers WH, Moorman AFM. Multiple transcriptional domains, with distinct left and right components, in the atrial chambers of the developing heart. Circ Res. 2000; 87: 984–991.
Janse MJ, Anderson RH. Specialized internodal atrial pathways: fact or fiction? Eur J Cardiol. 1974; 2: 117–136.
de Ruiter MC, Gittenberger-de Groot AC, Poelmann RE, VanIperen L, Mentink MM. Development of the pharyngeal arch system related to the pulmonary and bronchial vessels in the avian embryo: with a concept on systemic-pulmonary collateral artery formation. Circulation. 1993; 87: 1306–1319.
Romer AS. Circulatory system. In: The Vertebrate Body, 3rd ed. London, UK: W.B. Saunders Company; 1962: 410–457.
Verheule S, Wilson EE, Arora R, Engle SK, Scott LR, Olgin JE. Tissue structure and connexin expression of canine pulmonary veins. Cardiovasc Res. 2002; 55: 727–738.
Perez-Lugones A, McMahon JT, Ratliff NB, Saliba WI, Schweikert RA, Marrouche NF, Saad EB, Navia JL, McCarthy PM, Tchou P, Gillinov AM, Natale A. Evidence of specialized conduction cells in human pulmonary veins of patients with atrial fibrillation. J Cardiovasc Electrophysiol. 2003; 14: 803–809.
Benninghoff A. Über die beziehungen des reizleitungssystems und der papillarmuskeln zu den konturfasern des herzschlauches. Verh Anat Gesellsch. 1923; 57: 185–208.
Thompson RP, Lindroth JR, Wong YMM. Regional differences in DNA-synthetic activity in the preseptation myocardium of the chick. In: Clark EB, Takao A, eds. Developmental Cardiology: Morphogenesis and Function. Mount Kisco, NY, Futura Publishing Co; 1990: 219–234.
Hoogaars WMH, Tessari A, Moorman AFM, de Boer PAJ, Hagoort J, Soufan AT, Campione M, Christoffels VM. The transcriptional respressor Tbx3 delineates the developing central conduction system of the heart. Cardiovasc Res. 2004; 62: 489–499.