Development of ventricular trabeculae affects electrical conduction in the early endothermic heart

The ventricular trabeculae play a role, among others, in the impulse spreading in ectothermic hearts. Despite the morphological similarity with the early developing hearts of endotherms, this trabecular function in mammalian and avian embryos was poorly addressed.


| INTRODUCTION
The heart is essential for embryonic survival and one of the first functioning organs during embryogenesis.Its periodic contractions appear shortly after its formation to drive circulation and allow embryonic growth. 1 The cardiac movement is driven by a pacemaker positioned at the inflow region in birds 2 and mammals. 3First, the contraction is determined by the isotropic fashion of conduction. 3,4This slow movement progresses along the heart tube in a peristaltoid isotropic manner. 5Subsequently, the conduction pattern is diversified according to the particular heart segment. 6,7The conduction becomes anisotropic in the looping ventricle, ensuring fast and coordinated chamber activation.The primary interventricular ring (PIR) as an intermediate state is rapidly replaced by the apex-first (with right and/or left epicardial breakthrough) activation sequence.][10] In the same period, the developing trabecular network coarsens luminal the surface of the ventricle. 11In addition to their other functions, at least in part of ectotherms, trabeculae play a role in interconnecting the atrioventricular canal (AVC) to the ventricular apex, thus, providing a shortcut for activation wave.In the amphibians (namely Xenopus laevis) and fish (namely zebrafish and African lungfish), the apex-first activation pattern was observed and associated with the trabecular system. 12,13Despite the superficial morphological similarities between ectothermic and embryonal endothermic hearts, whether the trabecular design of early endothermic ventricles plays a similar role in the electrical conduction has not been directly addressed yet.This study analyzed intraventricular conduction by either simulation of impulse propagation or direct measurement of opened ED4 chick ventricle using high-speed optical imaging.In addition, we created a novel trabeculae-deficient chick model by pharmacologically targeting Neuregulin/ErBb pathway.This model, together with the mouse Nkx2.5 À/À heart, allowed us to investigate conduction patterns of trabeculae-deficient heart in both prototypic endothermic animals.We showed that trabeculae development allows faster electrical wave propagation and earlier apex activation.Both trabeculae-deficient models showed an isotropic activation pattern without progression to the more advanced anisotropic activation sequence.Moreover, the level of trabeculation was shown to be negatively correlated with the ventricular activation time, highlighting the role of trabeculation in the initiation of faster anisotropic ventricular conduction.

| The developing trabeculae allow earlier apex activation
To investigate the role of ventricular trabeculae in impulse spreading during early cardiac development of endotherms, we simulated impulse propagation in the three-dimensional (3D) reconstructed datasets of the trabeculae-deficient (induced by AG1478 treatment-an inhibitor of the ErbB1 receptor and the ErbB2/ ErbB3 heterodimer) and control looping chick heart at the Hamburger-Hamilton (HH) 14 stage 19.We used a model based on geodetic distance adapted to the embryonic heart. 15,16When the impulse was initiated at the AVC perimeter where the atrioventricular conduction is present, delayed apex activation in the trabeculae-deficient chamber (Figure 1C) compared to controls (Figure 1A) was observed.The internal view (Figure 1B, D) showed a difference in the intraventricular conduction by the trabecular system or the ventricular wall only.The delayed apex activation was observed in all analyzed trabeculaedeficient hearts.
We further approached the intraventricular conduction directly by imaging the endocardial aspect of the chick-looping heart (HH24; Figure 2).In this embryonic stage, the trabeculae are well developed but still not compacting, 17,18 enabling visualization of impulse propagation through the trabecular network.Using optical mapping, we observed that the first depolarization appeared in the atrioventricular area and the inner area of the ventral part of the ventricle (Figure 2A).This area corresponded with the presence of trabeculae in the mapped heart (Figure 2B).The depolarization wave subsequently propagated to the adjacent cardiac tissue to reach the ventricular apex (this activation pattern was observed in all analyzed hearts).When optical mapping was performed from the epicardial surface of the same Whereas in the trabeculae-deficient heart, the impulse is spread on the ventricular wall only (B), activation is spread along the trabeculae providing the shortcut to apex activation in the control hearts (D).n = 3 for both trabeculae-deficient and control hearts.A, atrium; OT, outflow tract; Tr, ventricular trabeculae.Scale bar 500 μm heart (before cutting, either anterior or posterior view; Figure 2D), the ventricular apex was activated later than in optical mapping performed from the intraventricular aspect (4.2 ± 0.4 ms from the outer aspect vs 3.4 ± 0.5 ms from inner aspect; Figure 2).In the epicardial maps, the nonconcentric shape of the isochrones directed to the apex was observed.This corresponded well with the prevailing trabecular orientation at this stage (Figure 2B 10,17 ) as well as with the prediction of the simulation model (Figure 1).

| Trabeculae-deficient models show thin ventricular walls and insufficient cardiac function
To confirm the role of the early ventricular trabeculae in conduction, we analyzed the activation patterns in ventricles without trabeculae.First, we adapted the previously described trabeculae-deficient model created by targeting Neuregulin/ErbB signaling for the endothermic heart.The AG1478 was applied adjacent to the developing heart before trabecular emergence.The mortality of treated embryos was 37% (18/49).Disruption of the Neuregulin/ ErbB pathway resulted in a severe deficit in trabeculation (Figure 3C), albeit only in a narrow window of its application (Table S1).This window coincided well with the period of early trabecular emergence in the chick. 17,18he optimum time frame for a successful trabecular inhibition by AG1478 was around 62 hours of incubation when the ventricle started the ballooning process.The earlier application could not block the signaling pathway, and during the later inhibition, the trabecular process was already established and thus no longer sensitive to inhibition (Table S1).
Both control and AG1478-treated hearts possessed tubular morphology in HH stage 17 (Figure 4A).While control hearts continued in normal development with trabeculae appearing first in the outer curvature of the ventricle and filling up a substantial part of the ventricle at HH18-19 (Figures 3A and 4A), we observed heart dilatation and a thin ventricular wall at HH18-19 in 14 out of 22 AG1478-treated embryos at HH18-19 (Figures 3C and  4C).Development of the trabeculae-deficient phenotype was accompanied with decreased heart rate.While in controls, heart rate continued to significantly rise (from 74 ± 12 bmp to 97 ± 16 bpm in HH17 and HH18-19, respectively; P < 0.05), a drop (P = 0.073) in beating frequency was observed in the treated group (from 89 ± 23 bpm to 65 ± 10 bpm in HH17 and HH18-19, respectively; Figure 3J).AG1478-treated hearts also showed irregular beating (not observed in controls), distended ventricles and compromised function with decreased contractility (Figure 4C, Supplementary Movie S1 and S2).This phenotype resulted in embryonic mortality past the stage HH19.

| Trabeculae-deficient models revealed isotropic and slow conduction pattern
Next, we addressed the conduction pattern in the trabeculae-deficient ventricle.During development, the heart undergoes a transition of activation sequence from slow isotropic to fast anisotropic.This switch to more advanced anisotropic patterns was observed in control hearts of both pharmacologically inhibited (chick AG1478) and genetically targeted (Nkx2.5 À/À mouse) models; however, both types of trabeculae-deficient ventricles showed an isotropic conduction sequence.Despite the continuing embryonic development, no sign of progression to the more advanced conduction pattern was observed in the chick AG1478-treated ventricles (Figure 4A).Both control and AG1478-treated hearts showed a primitive activation pattern with slow isotropic conduction along the ventricle in developmental stage HH17.In the HH18-19 control hearts, 93% of ventricles (24/26) were activated by the advanced PIR sequence (Figure 4A).On the contrary, 7/7 AG1478-treated hearts were activated by the primitive isotropic pattern (P < 0.001).

| Acceleration of ventricular conduction in early cardiogenesis correlates with the level of trabeculation in chick trabeculae-deficient hearts
We observed that AG1478-induced trabecular inhibition did not produce a ventricle with a completely smooth wall, rather creating a phenotype with rudimentary trabeculae (trabecular area: 28.8 ± 6.0% in AG1478 treated vs 50.8 ± 2.5% in control; Figure 3I, P < 0.0001).This finding is similar to an observation about the not entirely absent trabecular system, but rather arrested at the stage of their emergence as reported previously in AG1478-treated zebrafish or even in some of the neuregulin pathway-targeted mouse.][21][22] Therefore, we took an advantage of the phenotype variability and correlated the degree of trabecular development with ventricular conduction time.Analyzing both control and AG1478-treated hearts, we showed a significant negative correlation between the level of developed trabecular area and time of ventricular activation (Figure 4B; P = 0.0014).This implies that the less level of trabeculation, the slower ventricular activation was.We were not able to perform a similar correlation analysis in the mouse models since they do not possess enough level of phenotypes variation.Although the significant reduction of a trabecular layer of the ventricle was observed in both null mice compared with WT (trabecular area: 31.2 ± 6.1% vs 59.0 ± 6.3% for Nkx2.5 À/À and WT, respectively, P < 0.0001, Figure 3K; and 19.2 ± 2.0 vs 35.8 ± 5.3% for ErbB2 À/À and WT, respectively, P < 0.001), the development of trabeculae in the heterozygotes did not allow the correlation analysis.

| Ventricle of ErbB2 À/À mouse expresses Cx40
To determine the potential role of the chamber myocardium differentiation on the observed slow conduction phenotype, we employed the ErbB2 À/À mice.Heterodimer F I G U R E 5 Trabeculae-deficient mouse hearts are activated in a slow and isotropic pattern.Panel A shows representative spatiotemporal maps of mouse electrical activation (LV is depicted by the red line in the grayscale image).The ventricular activation by PIR is typically encountered in the WT as well as Nkx2.5 +/À hearts, while a primitive activation pattern is present in the Nkx2.5 À/À ventricle.Quantification of ventricular activation patterns showed that while all WT and Nkx2.5 +/À hearts were activated by advanced conduction sequence (PIR or LAB), ventricles of Nkx2.5 À/À were activated exclusively by the primitive isotropic pattern, and no advanced activation pattern was observed (B).This was associated with a significant increase in the LV activation time in Nkx2.5 À/À compared to Nkx2.5 +/À and WT hearts (C; Kruskal-Wallis test).In all analysis, 10 to 14 animal per group were used.Example of the optical maps of the ErbB2 mutated mice showing a less pronounced electrophysiological phenotype compared with Nkx2.5 GFP mice (D).PA, primitive activation along the heart tube; PIR, primary interventricular ring; LAB, left activation breakthrough; LV, left ventricle.Data expressed as mean ± SD; **P < 0.001 and ***P < 0.0001 vs Nkx2.5 À/À .All pictures in panels A and D are at the same magnification (10Â water immersion objective).
ErbB2/ErbB4, as a neuregulin receptor, is a crucial component of the Neuregulin/ErbB signaling pathway. 20rbB2 À/À mice have a deletion of one of the nonredundant myocardial receptors for neuregulin.Unlike the other mice with deletion of neuregulin itself or its receptor ErbB4, which do not survive past ED9.5, these mice survive up to ED9.75 23 The embryos showed a drastic reduction in the trabecular layer of the ventricular wall compared to the WT littermates (Figure 6A, B).By crossing this line to the Cx40 GFP background, 24 we were able to study the expression of Cx40 (as a hallmark of the chamber specification into the faster conduction myocardium) even before possible Cx40 detection by immunohistochemistry (up ED10). 25The level of green fluorescence in the atria and outflow tract endocardium was similar between genotypes in all four cases of ErbB2 À/À ;Cx40 GFP/+ and comparable with the level observed in the WT;Cx40 GFP/+ hearts (Figure 6C, D).The intensity of Cx40 GFP fluorescence in the ventricle was not significantly different between ErbB2 À/À ;Cx40 GFP/+ and ErbB2 +/À ;Cx40 GFP/+ (98 ± 8 vs 95 ± 16 in ErbB2 À/À ; Cx40 GFP/+ and ErbB2 +/À ;Cx40 GFP/+ , respectively, P = 0.73).

| DISCUSSION
In this work, we focused on the function of the trabecular system in impulse spreading in the early developing endothermic heart.In the looping heart, using either simulation of the impulse propagation inside the ventricle or direct endocardial surface imaging, we showed that trabeculae development is associated with an anisotropic conduction pattern allowing earlier apex activation.
Despite the obvious morphological similarities, ventricles of ectotherms differ from embryonic mammalian and avian trabecular ventricle designs.Adult mammalian and avian hearts work under higher pressure than ectothermic ones, [26][27][28] and their highly trabeculated embryonic ventricles represent only an intermittent stage.While hemodynamic loading represents an important factor in trabeculae formation, 20,29 the question about the similar morphology to the functional relationship has been raised.1][32] Moreover, the initial steps in trabecular formation differ in the prototypical ectothermic (zebrafish) and endothermic (mouse and chick) heart. 17,33,34Variation also exists in the level of trabeculation.In the adult ectotherms, the trabecular proportions encompass around 80% of the ventricular wall instead of 40% to 50% in the case of ED4 chick heart. 27oncerning the role of ventricular trabeculae in the impulse spreading, interconnection of the trabecular system at the atrioventricular junction with the apical part of the ventricle in Xenopus laevis was reported.This connection was associated with the first epicardial activation near the ventricular apex and was interpreted as a functional analog of the specialized ventricular conduction tissue. 12,13Similar findings were reported in frog hearts. 35nterestingly, intraventricular conduction by trabeculae is not linked with univentricular design (common in the ectothermic heart) only since it was observed in reptile hearts with the complete ventricular septum (alligator, crocodile) as well. 36,37The potentially confusing fact that the apex-first activation is not universal in all ectothermic species [38][39][40] could be attributed to the high level of variation in the ventricular architecture and trabecular organization. 27,38,41,42he unresolved question is whether the trabeculae in endotherms also lead to the anisotropy of impulse conduction inside the early-developed ventricle.This assumption was proposed already by Viragh and Challice F I G U R E 6 ErbB2 À/À embryonic hearts with rudimentary trabeculae express Cx40.Representative image of WT (A) and ErbB2 À/À (B) mouse hearts at ED9.75 showing a dramatic reduction in the trabecular network (arrows) in the ErbB2 À/À hearts.The level of green fluorescence was comparable in the OT endocardium and atria between the WT;Cx40 GPF/+ (C) and ErbB2 À/À ;Cx40 GPF/+ (D) embryos.Quantification of GFP fluorescence in the left ventricle showing comparable level between ErbB2 À/À ;Cx40 GPF/+ and ErbB +/À ;Cx40 GPF/+ (n = 2-3 animals per group, E).AVC, atrioventricular canal; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle; OT, outflow tract; Tr, ventricular trabeculae.Scale bar 100 μm (same between panels A and B, and C and D, respectively).
based on morphological observation of trabecular arrangement. 43The first attempt at indirect confirmation of this assumption came from the study of de Jong and colleagues.Using bipolar electrode measurements in the HH31 chick embryonic heart showed fractionalized epicardial electrogram and considered the first deflection as depolarization of the trabecular network.This led them to suggest that trabecular activation precedes the ones of the outer compact ventricular wall. 7However, HH31 hearts are highly compacted and thus prevent a clear distinction between the trabecular and compact layer of the ventricular wall.In the present study, the employment of the earlier stages of the developing ventricle allowed us to localize activation waves more precisely within the trabecular network.We used a model of impulse propagation adapted to the embryonic ventricle. 16Simulated activation maps assuming a constant activation speed along the developing structures (both ventricular wall and trabecular network) showed earlier apex activation in the trabeculated dataset than in the trabeculaedeficient ventricle, implying that trabeculae provided a shortcut for the apex activation.This role of trabeculae was confirmed by earlier activation of the trabeculae compared to the ventricular wall in the opened chick ED4 heart.Activation in the trabecular network was clearly shown to precede the rest of the surrounding tissue.This signal could not be attributed to the wall activation since it was not observed in the uncut hearts.These findings suggest that in the early endothermic hearts, the depolarization wave was spread preferentially through the embryonic trabecular network, similar to some ectotherms.
Considering the role of the trabecular network in intraventricular conduction, we further analyzed the conduction in two examples of the trabeculae-deficient ventricle.In the first model, we targeted the Neuregulin/ ErbB signaling.This pathway was shown to be a key pathway for trabeculation in both ectotherms and endotherms. 33,44,45Disruption of Neuregulin/ErbB signaling by AG1478 was reported to create a phenocopy with loss of neuregulin signaling [46][47][48] and to produce trabeculaedeficient heart in zebrafish. 19,20Interestingly, a recent study showed that atrial cardiomyocytes of zebrafish, like ventricular cardiomyocytes, can respond to Nrg2a/Erbb2 signaling as well. 21Similar to our findings, this study showed timescale as an essential factor for proper Neuregulin/ErbB signaling disruption.Here we determined the window for the chick ventricle at around 62 hours of incubation.This time frame clearly corresponds with trabecular emergence in chick hearts. 17,18We have shown slow and isotropic conduction sequence with no progression to the advance activation patterns in the chick AG1478-induced trabeculae-deficient hearts.
The second model of the trabeculae-deficient ventricle was created by genetic ablation of Nkx2.5, a transcription factor governing the cardiomyocyte differentiation program.The heterozygous state led to the cardiac conduction system hypoplasia, while its complete deletion is embryonic lethal around ED10. 47 The weak and atypical contraction sequence was described in the ED9.5 Nkx2.5 À/À hearts. 49The authors correlated this with the observed broad AVC and possible retrograde blood flow. 49The role of Nkx2.5 in the development of pacemaker is controversial.Earlier studies thought that Nkx2.5 is excluded from the developing sinoatrial node (SAN) 50 and its expression in SAN or pulmonary sleeves prevents myocardium from pacemaker activity. 51,52However, a recent study has shown that the Nkx2.5 is expressed in the SAN junction 53,54 and plays a role in the development of a subpopulation of pacemaker cells. 55kx2.5 inactivation in SAN junction thus led to the SAN dysfunction. 55We and others observed decreased heart rate in the Nkx2.5 À/À embryos. 49Since the heart rate physiologically increases during early mouse development, 56,57 the decreased heart rate could be also the result of developmental delay observed in Nkx2.5 À/À embryos.Alternatively, decreased perfusion due to low cardiac output caused by a slower heart rate could lead to embryonic growth retardation and ultimate demise, as seen in the pharmacological intervention model (ivabradine) in the chick embryo. 58Considering the conduction pattern, similarly as in the AG1478-treated chick, Nkx2.5 À/À trabeculae-deficient ventricles failed to progress to the advanced activation pattern and showed a slow and isotropic conduction pattern.Accordingly, the decrease of ventricular activation time presented during physiological development concomitantly with activation pattern progression 9 was not observed.
Since the Nkx2.5 deletion was reported to dysregulate the cardiac connexins, 49,[59][60][61] their depletion could affect the conduction speed in Nkx2.5 À/À hearts.To define the role of the fast conduction myocardium maturation in the observed slow conduction of the trabeculae-deficient ventricles, the ErBb2 À/À ;Cx40 GFP/+ was subsequently analyzed.The ErBb2 À/À ;Cx40 GFP/+ ventricle slowed the presence of green fluorescence implying unaltered Cx40 expression.Thus, it can be assumed that the Neuregulin/ ErbB pathway does not affect Cx40 expression.Moreover, our simulation data considering spatial dimensions only (without any changes related to the connexins or ion channel expression), showed that the trabeculae network is sufficient to provide a shortcut to apex activation.Altogether, these findings suggest that the observed slow epicardial conduction of trabeculae-deficient hearts could be attributed to severely affected trabecular architecture rather than absent Cx40 expression.
In addition to the slow conduction pattern presented in both trabeculae-deficient models, in the AG1478-induced trabeculae-deficient chick hearts we showed a significant negative correlation between the total ventricular activation time and trabecular development level.Collectively, these findings strongly suggest that the early trabecular development forms a substrate for preferential impulse propagation and is essential for the proper activation patterning.Ventricular dilatation, probably due to insufficient wall stiffness observed in the trabeculae-deficient AG1478-treated hearts, resulted in altered contractility and ultimately led to heart failure and embryonic death.We cannot fully exclude the role of possible insufficient AVC valve function with consequent retrograde flow to the atria, similarly as reported in Nkx2.5 À/À animals, 49 in altered contractility observed of these hearts.On the other hand, in case of AG1478-induced trabeculae-deficient heart, no cardiac valve defect was reported in the zebrafish. 19Moreover, it is highly unlikely that the altered AV valve function with preserved ventricular contractility would result in important ventricular dilatation.Since we observed dilated ventricle with weak contraction, we attributed the altered contractile function in the AG1478-treated hearts to the inability of trabeculae-deficient hearts to develop a mature conduction system with a consequent inadequate adaptation of the pumping function.This together with a thin ventricular wall resulted in ventricular dilatation.Similar findings of impaired contractility after AG1478 treatment was reported in the zebrafish hypotrabeculaed model. 20

| CONCLUSION
Our results showed that, in addition to other functions in the developing heart, the trabeculae are essential in establishing a preferential conduction pathway inside the forming ventricle and conduction patterning.Lack of trabeculae then leads to failure of conduction parameters differentiation and decline in contractility (in AG1478-induced model), resulting in primitive peristaltoid ventricular activation with consequent embryonic heart failure.

| EXPERIMENTAL PROCEDURES
All procedures performed on animals were in accordance with the ethical standards of Charles University and were approved by the Animal Care and Use Committee of the First Faculty of Medicine.

| Mouse models
The Nkx2.5 GFP knock-in 62 at ED9.5 was used for genetic ablation of trabecular development.ErbB2 À/À mice at ED9.75 were used to produce hearts with substantial trabecular reduction targeting Neuregulin/ErbB pathway.The breeding pairs were caged overnight, and the noon of the day when the plug was discovered was considered ED 0.5.Time-pregnant females were sacrificed at the appropriate time, and the embryos were rapidly dissected.The Nkx2.5 +/À and Nkx2.5 À/À genotype was confirmed by detecting GFP using an epifluorescence microscope fitted with the appropriate filter set (Olympus, Tokyo, Japan).The genotype of ErbB2 À/À mice was confirmed by PCR as described. 23In addition, ErbB2 À/À mice were crossed with the Cx40 GFP line to visualize Cx40 expression. 24Cx40 GFP mice with Swiss albino background were used as a control.The CRL was measured directly after embryos dissection.

| Chick trabeculae deficient heart
In chick, the Neuregulin/ErbB pathway was targeted by applying the AG1478, an inhibitor of the theErbB1 receptor, and the ErbB2/ErbB3 heterodimer. 20We adjusted the protocol previously described for zebrafish 19 to the chick embryos.White Leghorn chicken eggs incubated for 62 hours were windowed on the blunt end, and AG1478 (Sigma Aldrich; 90 pg in 18 μL DMSO, diluted to 200 μL by physiological saline) was applied adjacent to the developing vasculature.Controls received DMSO with physiological saline solution.The shell was closed with the Scotch tape, and the eggs were re-incubated to the desired HH stage 17 and 18, 19, respectively. 14,63At HH18-19, the phenotype was fully developed (ventricular dilatation, weak contraction), and at the same time, embryos were still alive to enable electrical and mechanical function assessment.

| Simulation of impulse propagation
The chick trabeculae-deficient hearts were generated as described above.At HH19, AG1478 treated and control hearts were isolated, fixed with 4% paraformaldehyde in phosphate buffer saline (PBS), and scanned using an exvivo micro-CT (SkyScan 1272, Bruker Micro-CT, Belgium).Before scanning, specimens were X-ray contrasted by phosphotungstic acid according to the protocol described by Metscher. 64Each specimen was placed in a plastic tube with PBS and scanned with the following parameters: pixel size 2 μm, source voltage 60 kV, source current 166 mA, rotation step 0.4, no filter, 180 rotation.After pre-procession, the data volume was thresholded by a Gaussian filter using Amira 2020.1 software (Thermo Fisher Scientific, Waltham, Massachusetts, USA).The geodesic distance 15 from the AVC to the ventricle border was further calculated and visualized by color on the triangulated surface of the heart.Three trabeculae-deficient and control hearts were used for analysis.
Subsequent analysis was performed using the software BV_ANA bundled with the camera as described previously. 65,66We analyzed heart rate, spatiotemporal reconstruction of activation pattern, total ventricular activation time, and location of the first ventricular activation site.
Mapping of electrical activity in the endocardial aspects was performed in chick ED4 hearts (HH24; n = 5).After whole heart mapping, the ventricle was dissected in the frontal plane under the SZH 10 stereomicroscope (Olympus, Japan) to view the dorsal half of the heart with the endocardial trabeculae surface. 29Hearts were pinned endocardial side up in a custom-made Sylgard chamber in the position parallel with the objective lens to prevent heart edges to coil and enable mapping of the activation pattern of the whole endocardial surface trabeculae.All hearts were stable in beating frequency for the entire imaging time.Mapping and analysis were performed as described above.

| Analysis of contractility
To assess the contractile function of HH19 hearts (n = 7 for AG1478 treated and n = 8 for controls), the in-ovo videomicroscopy was performed as described previously. 67The embryos in ovo were maintained at 37 C using a custom-made Styrofoam-insulated metal container filled with pre-heated Bath Armor pellets placed on a Torrey Pines Scientific chilling/heating plate.Tensecond movies were recorded with a Nikon D7000 camera (640 x 480 px, 30 fps) mounted on a Leica 125 dissecting microscope; for illumination, a 150 W halogen light source was fitted with a green interference filter to enhance blood contrast, was used.Ventricular dimensions during the whole cardiac cycle and contractility were subsequently assessed.

| Quantitative analysis
After optical mapping, the Nkx2.5 GFP hearts were fixed in 4% (wt/vol) in paraformaldehyde in PBS, optically cleared using CUBIC2, 68 and subjected to whole-mount microscopy with 4x or 10x objectives and 10 or 2.5 μm z-steps using FluoView FV1000 confocal system fitted on an upright BX61 microscope (Olympus, Tokyo, Japan).Chick hearts were imaged in the same way using endogenous autofluorescence instead of GFP. 69o analyze the area of the trabecular network, the maximal ventricular diameter was found in the 3D data set and analyzed in FIJI software (NIH, USA).The threshold was set to exclude blood and include the outer ventricular wall with the entire trabecular network.In the ROI delineated by the atrioventricular cushions on one side and the outflow cushions on the opposite side, the trabecular area was then calculated as a percentage of an outer ventricular wall from ROI. Hearts of ErbB2 À/À and WT littermates were processed for paraffin embedding and sectioned at 7 μm.Hematoxylin and eosin staining was further performed to reveal ventricular wall morphology.The extent of the trabecular network was analyzed as described above.In all quantitative analyses n = 3 to 7 per group.Whole heart fluorescence imaging of optically cleared ErbB2 mice at Cx40 background was performed to analyze GFP fluorescence as a proxy for connexin40 expression.Specific GFP fluorescence was measured after background subtractions.The mean intensity of green channel fluorescence was analyzed in five different areas of the left ventricular myocardium spanning the entire ventricle in FIJI software (NIH, USA) using a sampling square of 10 Â 10 pixels (n = 2-3 mice per group).

| Statistical analysis
GraphPad Prism 9 (GraphPad Software, Inc., San Diego, CA) was used for graphic presentation and statistical analysis.Normality of data distribution was tested by Shapiro-Wilk test.Unpaired Student's t-test or one-way ANOVA and subsequent Student-Newman-Keuls test were used to compare differences in normally distributed variables between groups.Mann-Whitney or Kruskal-Wallis test was used to compare differences in nonnormally variables between groups (analysis of ventricular activation time and CRL).The Chi-square test was used to compare activation patterns frequency among the groups.Pearson's correlation coefficient was used for correlation analyses.Differences were considered statistically significant when P < .05.

F I G U R E 1
Simulated activation of ventricular apex shows earlier apex activation chick trabeculated heart.The surface view of the simulation shows a delayed apex activation in the reconstructed chick trabeculae-deficient (A) compared with the control ventricle (C).Panels B and D show intraventricular views of the simulations.

F I G U R E 2
Ventricular trabeculae are the first activated area within the intraventricular surface.Representative image of activation wave propagation at the endocardial (A) and epicardial (C) aspect of HH24 chick heart (n = 5).The early activated area corresponds to the developing trabeculae in the intracardiac view (A, red arrow) and later activation of the same area in the epicardial view (C, red arrow).Grayscale images of the analyzed area are shown on panels B and D. AVC, atrioventricular canal; LA, left atrium; RA, right atrium; Tr, ventricular trabeculae

F
I G U R E 4 Development of ventricular trabeculae correlates with anisotropic conduction pattern and ventricular activation time.The optical map of chick HH17 and HH19 hearts shows the activation pattern transition from slow and isotropic (number and area of colored band per unit of tissue) at the HH17 to the advanced conduction by PIR in control HH18-19 hearts.The AG1478 hearts did not undergo this transition, and an isotropic pattern still activated them even in the HH19 stage (n = 10-13 animals per group at HH17 and 7-26 animals per group at HH18-19, A).The level of ventricular trabeculation at HH18-19 showed a significant negative correlation with the ventricular activation time (B; Pearson's correlation).Trabeculae deficient HH18-19 phenotype was accompanied by important ventricular dilatation compared with controls (red arrow, n = 7-8 per group, C).Scale bar 200 μm; PIR, primitive interventricular ring.