Identifying the Evolutionary Building Blocks of the Cardiac Conduction System

The endothermic state of mammals and birds requires high heart rates to accommodate the high rates of oxygen consumption. These high heart rates are driven by very similar conduction systems consisting of an atrioventricular node that slows the electrical impulse and a His-Purkinje system that efficiently activates the ventricular chambers. While ectothermic vertebrates have similar contraction patterns, they do not possess anatomical evidence for a conduction system. This lack amongst extant ectotherms is surprising because mammals and birds evolved independently from reptile-like ancestors. Using conserved genetic markers, we found that the conduction system design of lizard (Anolis carolinensis and A. sagrei), frog (Xenopus laevis) and zebrafish (Danio rerio) adults is strikingly similar to that of embryos of mammals (mouse Mus musculus, and man) and chicken (Gallus gallus). Thus, in ectothermic adults, the slow conducting atrioventricular canal muscle is present, no fibrous insulating plane is formed, and the spongy ventricle serves the dual purpose of conduction and contraction. Optical mapping showed base-to-apex activation of the ventricles of the ectothermic animals, similar to the activation pattern of mammalian and avian embryonic ventricles and to the His-Purkinje systems of the formed hearts. Mammalian and avian ventricles uniquely develop thick compact walls and septum and, hence, form a discrete ventricular conduction system from the embryonic spongy ventricle. Our study uncovers the evolutionary building plan of heart and indicates that the building blocks of the conduction system of adult ectothermic vertebrates and embryos of endotherms are similar.


Introduction
The hearts of mammals and birds maintain high rates of contraction [1] that in concert with high systemic blood pressures accommodate their high rates of oxygen consumption due to their endothermic state [2]. The high heart rates, the timing of sequential atrial and ventricular contractions and the rapid spread of the activating impulse over the avian and mammalian ventricles are possible because of a specialized cardiac conduction system [3]. However, while the sequential activation of the cardiac chambers and appropriate matching of the atrial and ventricular contractions are similar across all vertebrate groups, there is no anatomical evidence for a specialized conduction system in hearts of reptiles or other ectothermic vertebrates [4,5]. Because mammals and birds evolved independently from reptilian ancestors, the evolutionary origin of their specialized conduction systems has remained unclear; either their conduction systems evolved independently or primordial components of the system were already present in the ancestral reptiles (Fig. 1).
Like the hearts of ectothermic vertebrates, embryonic mammalian and avian hearts also exhibit regulated sequential activation patterns in the absence of a morphological conduction system [2,6,7]. This suggests that the functional components for conduction system are established early in development and in evolution, but are not represented by anatomically distinguishable components as in the mature hearts of endothermic vertebrates. Instead, the components may reflect an intrinsic part of the building plan of the heart.
Tbx2 and Tbx3 belong to an ancient family of transcription factors [8] expressed in the embryonic atrioventricular canal from human to primitive fish [3,9]. In the embryo, the atrioventricular canal delays the impulse from atrium to ventricle. Tbx2/3 suppresses chamber genes including Nppa and Gja5, encodes for connexin40 that is required for fast conduction [10], and hence inhibits differentiation of the atrioventricular canal to fastconducting chamber myocardium [3,11]. Tbx3 remains expressed in the mature conduction system components of mammals, including the atrioventricular node that derives from the atrioventricular canal [12]. Bmp2/4 are expressed in the atrioventricular canal of early embryonic mammals, birds and fish, and are crucial for activation of Tbx2/3 (Fig. S1) [13][14][15][16][17]. Hence, Tbx2/3, Bmp2/4 and Gja5/Nppa represent evolutionary conserved positive and negative markers, respectively, that discriminate the embryonic slow-conducting atrioventricular myocardium and fastconducting chamber myocardium. The ventricular His-Purkinje system of mammals and birds is specifically marked by expression of Gja5 (and Nppa in mammals) [3,10].
Here, we carried out in situ hybridization analysis using evolutionary conserved genetic markers and provide a threedimensional reconstruction of the key components of the conduction system. The ventricular conduction pattern was visualized using optical imaging of activation. The cardiac expression and conduction patterns of a reptile were then compared to those of mammals, chicken, and other ectothermic vertebrates, frog and fish. We find an anatomic, genetic and physiologic conserved building plan where hearts of adult ectothermic vertebrates are similar to embryos of the endothermic mammals and birds. This indicates that primordial components of the cardiac conduction system were present in the ancestral reptiles.

Materials and Methods
All experimental procedures on adult material complied with national and institutional guidelines and were approved by Institutional Animal Care and Use Committee of the University of Amsterdam. The approval is registered as ''DAE101617'' for optical mapping of the ectotherms and ''DAE101532'' for optical mapping of developing mice. In The Netherlands experiments

Animals
Adult zebrafish were provided by the Hubrecht Laboratory, Utrecht, the Netherlands, and adult Xenopus laevis from Leiden University, the Netherlands. Mice and Xenopus laevis embryos were raised in the AMC. Green and brown anole (Anolis carolinensis and A. sagrei) eggs and adults and fertilized chicken eggs, were bought commercially in the Netherlands. Xenopus embryos were staged according to Nieuwkoop [18], Anolis embryos according to Sanger et al. [19], chicken according to Hamburger and Hamilton [20] and mice from days post coitus.

In situ Hybridization
All embryos and hearts were fixed in 4% paraformaldehyde for one day and then kept in 70% ethanol until imbedding in paraffin and then sectioned at 7-12 mm for in situ hybridization. Methodology of the non-radioactive in situ hybridization analysis has been described previously [22,23] and so has probes for Zebrafish [16], Xenopus [24,25], chicken [26][27][28][29][30] and mouse [26]. Probes for Anolis were made in house based on the following coordinates using UCSC Genome Browser on Lizard May 2010 (Broad AnoCar2.0/anoCar2) Assembly; Tnnt2  Evolutionary Basis of Cardiac Conduction a cDNA library with standard TRIzol RNA extractions [31] from freeze-fixed specimens of developmental stages 5-9 and GOI cDNA were obtained by PCR amplification and cloned into pBluescript SK_ (Stratagene, La Jolla, CA). Digoxigeninelabeled antisense mRNA were then produced by in vitro transcription according to the manufacturer's instructions (Roche, Mannheim, Germany).
3D-reconstruction Protocol 7, 10 and 12 mm serial sections were stained by in situ hybridization and 3D reconstructions were performed as described previously using AmiraH version 5.2 software [32]. The interactive 3D pdfs were created using Adobe Acrobat Pro ExtendedH version 9.3. The 3D pdf can be viewed with the freeware version: Adobe ReaderH (version 9.3 or higher) with JavascriptH enabled.

Results and Discussion
In adult lizards, the sequential chamber contractions and an atrioventricular delay are well-established [33], but we found no insulating plane or insulated atrioventricular node in Anolis (Fig. 2). Instead, the atrioventricular canal was entirely myocardial (Fig. 3). This differs from the adult hearts of mammals and birds, where the atrioventricular myocardium has largely disappeared and an insulating plane of fibrous-fatty tissue has ingressed between the atria and ventricles except at the atrioventricular node and His bundle, which provide the sole electrical communication between the atria and the ventricles [12,34,35]. To explain the atrioventricular delay in reptiles, we hypothesized, therefore, that the atrioventricular canal of Anolis has a molecular phenotype that differs from that of the chambers.
In embryos of mammals, birds, frog and fish, Nppa and/or Gja5 mark the rapid propagating atrial and ventricular chamber myocardium, whereas the atrioventricular canal is negative for these markers [3,15] (Nppa is lost in birds and all reptiles except turtles [36,37]). Focusing on the atrioventricular gene programme during Anolis development, we found Tbx3 and Bmp2 to be expressed in the developing atrioventricular canal myocardium, exactly complementary to Gja5 in the adjacent chambers ( Fig. 4A-D). Tbx5, known to promote differentiation into Nppaand Gja5-expressing chamber myocardium [38], was also present in the atrioventricular canal and the chambers (Fig. S2) [39].
We then examined the adult atrioventricular region in Anolis, and observed that Bmp2, which in mammals and birds is expressed in the atrioventricular canal only at embryonic stages, remained expressed throughout ontogeny (Fig. 4B,F). Tbx3, which marks the cardiac conduction system in mature mammals [12], was found within the same Bmp2-positive atrioventricular domain (Fig. 4C,G). Three-dimensional reconstructions of the expression patterns of Tbx3 in early developing hearts of Anolis, chicken and mammals revealed a striking resemblance (Fig. 5, Fig. S3). In all species, a similarly-shaped atrioventricular ring was observed. The Tbx3 expression domain extended into the sinus venosus, marking the sinus node primordium. This expression pattern did not substantially change in lizards just prior to hatching, whereas in mammals and birds, it became more complex with further development as the morphology of the heart, and particularly the sinu-atrial region, changed (Fig. 5).
We then pursued this building plan of the atrioventricular canal to older vertebrate classes, represented by Xenopus, an amphibian and thus a non-amniotic member of tetrapods, as well as zebrafish. In both species, the atrioventricular canal is composed of myocardium in continuity with the atrium and ventricle (Fig. 3). An insulating plane and an insulated atrioventricular node were not found (Fig. 2). In adult zebrafish and stage 48 Xenopus, we found Tbx2 in the atrioventricular canal and Nppa (whose spatiotemporal pattern strongly resembles that of Gja5 in mammals) in the atria and ventricles, the patterns resembling those of Tbx3 and Gja5, respectively, in the amniotic vertebrates (Figs. 6, 7). No cardiac Tbx3 expression was found (Fig. S4).
Next, we examined the His-Purkinje system in Anolis. Gja5 was used as marker for the mature His-Purkinje system conserved in mammals and chicken [10]. The developing His bundle does not express Gja5 until late stages in mammals [40] and birds [41], but can nonetheless be identified very early by the expression of Tbx3 [26]. In mouse [40] and chicken, the Tbx3-positive and Gja5negative myocardium of the developing His bundle extends from the atrioventricular canal ventrally and dorsally into the ventricular myocardium and unto the crest of the ventricular septum (Fig. 5). The region formed by the dorsal and ventral extension Evolutionary Basis of Cardiac Conduction and the crest is referred to as the primary ring [7]. The Anolis ventricle is not septated, but shows Tbx3 expression into the ventricle ventrally and dorsally, indicating the presence of a primordial, but incomplete, primary ring that lacks the septal crest component (Fig. 5).
The ventricular wall in mammalian and avian embryos is composed of a trabecular inner layer and a thin compact outer layer. Initially, both of these layers express natriuretic peptides (Nppa and Nppb) and Gja5 [27,42], but halfway through development, Nppa and Gja5 expression ceases in the strongly expanding compact layer. After birth (or hatching in birds) the expression of Nppa and Gja5 is limited to the His-Purkinje network that eventually constitutes only a small fraction of the ventricular mass (Fig. 8) [10,43]. The ventricular wall of fish, amphibian and reptilian hearts does not display such an overt distinction in expression pattern between an inner trabecular wall and a compact outer layer. Their ventricular wall typically is composed of a spongy or trabecular type of myocardium (Figs. 2, 3). In adult Anolis hearts, Gja5 was homogenously expressed throughout the trabecular ventricular wall. This suggests an absence of tracts of preferential conduction leading to the ventricular apex and such condition resembles that of early mammalian and avian embryos (Fig. 9). Further back in evolution, as represented by Xenopus and zebrafish, we observed homogenous Nppa expression in their trabecular ventricular wall (Fig. 9A,D), while Tbx3 did not identify a primordial atrioventricular bundle (Fig. S4).
Mammalian and avian hearts have an elaborate Gja5/Nppaexpressing ventricular conduction system that activates the ventricles from apex to base [42,44,45]. The homogenous Gja5/ Nppa expression patterns in Anolis, Xenopus and zebrafish trabecular ventricles suggest that the electrical activation front may spread from the vicinity of the atrioventricular canal, i.e. the ventricular base, and reach the apex later. Such an activation pattern would be reminiscent of early embryonic mammalian and avian ventricles [45]. We used optical mapping to measure epicardial activation patterns in Anolis. The first point of activation always occurred in the cranial third of the ventricle, i.e. the base, and later at the apex (Fig. 9H). This activation pattern is consistent with most previous ECG and electrode investigations on reptiles and very similar to the activation patterns of chamber-forming hearts of mouse (E8-10) and chicken (E2-5) ( Table 1) [45][46][47]. At these stages in mammals and birds, a morphologically distinct conduction system has yet to form and ventricular septation is only starting to take place [48,49]. Assuming that the dorsal and ventral activation patterns share the same time point of activation at the apex, we could synchronize the activation patterns and infer that the dorsal base is activated prior to the ventral base in the ectothermic vertebrates (Fig. 9B,E). Dorsal activation of the ventricular base has been reported in chicken hearts prior to septation and seemingly occurs in embryonic mouse as well [42,[45][46][47][48][49]. In Xenopus and zebrafish, the activation front travels from the dorsal base to the apex. In Anolis, Xenopus and zebrafish the location of the first point of activation varied within the dorso- Evolutionary Basis of Cardiac Conduction basal region (Fig. S5). The activation maps and expression data indicate that the trabeculated ventricular wall of the ectothermic vertebrates function essentially as an isotropic conduction network. Consistently, conduction on the luminal surface of the adult mammalian and avian ventricles through the Gja5-positive His-Purkinje system also proceeds from base-to-apex [49,50]. On the epicardial side, however, the activation front reaches the apex first and then the base (Fig. 9J-L) [51]. In mammals and birds, the developmental change in activation pattern to apex-to-base observed at the epicardial side coincides with the development of the Gja5-negative compact ventricular wall, which therefore may contribute to this developmental change in activation pattern.
Hearts of ectothermic species and of embryos of endothermic species do not have anatomically marked conduction system components. In this study we used expression patterns of conserved genetic markers and identified molecular conduction system components in developing and adult lizards. We found them to be similar to the components in embryonic mammals and birds, indicating they constitute an integral part of the building plan of the heart. Therefore, the conduction systems found in mature mammals and birds most likely evolved from the components of this shared building plan, and did not evolved independently.
In mature birds and mammals, left-over traces of the atrioventricular canal muscle in addition to the atrioventricular node can be found. Birds have a well-developed right-sided atrioventricular ring bundle that communicates with the ventricle anteriorly through the so-called recurrent branch [4,52]. The mammalian heart also maintains a molecularly distinct atrioventricular ring bundle above the insulating plane [12]. Interestingly, in congenital corrected transposition of the human heart, the insulating plane disrupts the normal posterior atrioventricular communication, whereas the anterior communication is abnormally maintained [53,54]. The anterior part, then, resembles the recurrent branch of the bird heart.
The adult lizards and ectothermic vertebrates in general maintain important aspects of the embryonic vertebrate building plan. The Bmp2/4-Tbx2/3-positive, Gja5/Nppa-negative atrioventricular canal myocardium is maintained in adult ectothermic vertebrates. This provides an electrical insulation between atrium and ventricle in these hearts that lack an insulating plane of connective tissue. The developing hearts of mammals and birds have great tolerance to ischemia and regenerative potential, which  is lost around birth [55,56]. Interestingly, many ectothermic vertebrates (e.g. newts and zebrafish) retain the regenerative capacity and ischemia tolerance throughout life [6]. It is therefore tempting to speculate that the maintenance of important aspects of the embryonic programme in adult ectothermic vertebrates may be involved in the retention of these capacities.
Our study provides a plausible scenario of the evolution of the hearts of mammals and birds. The spongy myocardium of ectothermic adult vertebrates, as well embryonic mammals and birds allows for high ejection fractions and also serves to conduct the ventricular depolarization (Fig. 10). However, a transition to compact myocardium was necessary when pressure and heart rate In the formed hearts of mammals and birds, epicardial ventricular activation is from apex to base. 1 Left-to-right activation is only reported in broad-hearted turtles. 2 Arbel et al (1977) using ventrally placed electrodes find that as the apex becomes activated the current spreads towards the base. As they could not evaluate if the base was activated earlier, and since the zebrafish outflow region is indeed activated later than the apex their results may be in agreement with the present study. *Species where first point of activation definitely was neither base nor apex. From [33,. doi:10.1371/journal.pone.0044231.t001 increased. This rendered the early trabecules secondary on force generation, but available to differentiate into fibres of poor contractility and high propagation speeds. Furthermore, mammals and birds develop a compact ventricular septum whereby the early trabecules come to drape the septal surfaces and thus form the characteristic bundle branches of the His bundle (Fig. 10). Our study, therefore, suggests that the parallel evolution of virtually identical conduction systems and cardiac designs in birds and mammals can be traced back to the existence of a primordial conduction system of the ancestral reptile heart. Figure S1 Gene program of the developing atrioventricular canal in chicken. Tbx5, known to induce Gja5, is present in the atrioventricular canal but Gja5 is absent where Bmp2 and Tbx3 are expressed. la, left atrium; lv, left ventricle; ra, right atrium; rv, right ventricle. (TIF) Figure S2 Gene program of the developing atrioventricular canal in Anolis. Despite expression of Tbx5 the atrioventricular canal does not initiate chamber program and expresses the transcription repressor Tbx3 along with Bmp2. avv, atrioventricular valves; la, left atrium; ra, right atrium; t, trachea (positive for Tbx3).

Supporting Information
(TIF) Figure S3 Tbx3 expression shown in interactive 3D pdfs. Tbx3 expression (yellow) and lumen cast (red) of all models used in Figure 5. ( ) Figure S4 Tbx3 expression (in-situ hybridization) in developing and adult Xenopus. Tbx3 was only found outside the heart (e.g. developing trachea, t) of stage 40 embryos (top row).
No Tbx3 was only found in the heart of adults (lower row). a, atrium; avc, atrioventricular canal; c, conus; ven, ventricle; t, developing trachea (positive for Tbx3).
(TIF) Figure S5 Summary of individual ventricular activation maps. Red marks the earliest epicardial breakthrough of the activation front which was consistently at the ventricular base. Earliest and latest activation (red and white dots respectively) from each specimen is projected onto one specimen. (TIF) Figure 10. The trabecular myocardium is activated from base to apex in all vertebrates. (A-C) The trabecular myocardium gives rise to the His-Purkinje system in mammals and birds and remains activated from base to apex. (C) On the epicardial surface of septated and thick-walled ventricles, as in the formed hearts of mammals and birds, activation is seen to occur from apex to base and the luminal base-to-apex activation is obscured. doi:10.1371/journal.pone.0044231.g010 Evolutionary Basis of Cardiac Conduction