Cardiomyocyte-specific deletion of β-catenin protects mouse hearts from ventricular arrhythmias after myocardial infarction

Wnt/β-catenin signaling is activated in the heart after myocardial infarction (MI). This study aims to investigate if β-catenin deletion affects post-MI ion channel gene alterations and ventricular tachycardias (VT). MI was induced by permanent ligation of left anterior descending artery in wild-type (WT) and cardiomyocyte-specific β-catenin knockout (KO) mice. KO mice showed reduced susceptibility to VT (18% vs. 77% in WT) at 8 weeks after MI, associated with reduced scar size and attenuated chamber dilation. qPCR analyses of both myocardial tissues and purified cardiomyocytes demonstrated upregulation of Wnt pathway genes in border and infarct regions after MI, including Wnt ligands (such as Wnt4) and receptors (such as Fzd1 and Fzd2). At 1 week after MI, cardiac sodium channel gene (Scn5a) transcript was reduced in WT but not in KO hearts, consistent with previous studies showing Scn5a inhibition by Wnt/β-catenin signaling. At 8 weeks after MI when Wnt genes have declined, Scn5a returned to near sham levels and K+ channel gene downregulations were not different between WT and KO mice. This study demonstrated that VT susceptibility in the chronic phase after MI is reduced in mice with cardiomyocyte-specific β-catenin deletion primarily through attenuated structural remodeling, but not ion channel gene alterations.


Results
β-catenin knockout reduces the susceptibility to ventricular tachycardias after myocardial infarction. Successful knockout of β-catenin after tamoxifen treatment was confirmed by western blot showing marked reductions of β-catenin protein in the left ventricle of Ctnnb1 flox/flox ; αMHC-MerCreMer +/− (KO) mice as compared to Ctnnb1 flox/flox ; αMHC-MerCreMer −/− (WT) mice (Fig. 1C). Post-MI survival curve showed that there was no difference (p = 0.75) in the death rate between WT and KO groups ( Fig. 1D) with all the death (15% for WT; 19% for KO) occurred within the first week in both groups. The absence of animal deaths in the late stage after MI suggests no lethal arrhythmias in these mice. This reflects a known limitation of studying arrhythmias in mice-the low rates of spontaneous arrhythmias due to their fast heart rate, short action potential duration and smaller heart size, as compared to human and large animal models (e.g., monkeys and pigs) of heart disease.
To investigate the susceptibility of mice to ventricular arrhythmias, we used a protocol that we recently developed for rodent hearts 23,33 that combined adrenergic stimulation (with isoproterenol) and progressive programmed electrical stimulation (PES) in isolated, Langendorff-perfused hearts (Fig. 1E), while ex vivo ECG was recorded by placing electrodes around the heart. With this protocol, ventricular tachycardias (VT, defined as three or more consecutive ectopic ventricular beats) were successfully induced in 77% (10/13) post-MI WT mice at week 8, but in none (0/18) of the sham-operated WT mice (Fig. 1F,G) validating our VT-inducing protocol. No difference was observed between male and female mice in VT inducibility (4/5 in male and 6/8 in female, p = 0.84) or arrhythmia score (Fig. 1G) (3.40 ± 0.60 in male and 3.13 ± 0.58 in female, p = 0.76). In contrast, VT Figure 1. β-catenin knockout reduces the susceptibility to ventricular tachycardias after myocardial infarction. (A) To generate cardiomyocyte-specific β-catenin (Ctnnb1) knockout mice, Ctnnb1 flox/flox mice (with exons 2 to 6 floxed) were crossbred with αMHC-MerCreMer mice to obtain Ctnnb1 flox/flox ; αMHC-MerCreMer +/− mice, which were then treated with tamoxifen for deletion of exons 2-6 generating Ctnnb1 null allele due to the loss of the translational start site. Littermate Ctnnb1 flox/flox ; αMHC-MerCreMer −/− mice were used as control wild-type mice. (B) Experimental design. (C) Representative western blot for confirmation of reduced total β-catenin in both sham and LAD-ligated (MI) KO hearts. Protein samples were prepared from left ventricular free wall of sham mice and from the border zone in mice at 8-weeks post-MI (myocardial infarction). GAPDH was used as a loading control. The original uncropped gel images are included in Supplementary Fig. 1. (D) Survival curve showing no difference in survival between WT and KO mice after LAD ligation (p = 0.7473, analyzed by Log-rank (Mantel-Cox) test). Only the mice in the long-term (8-weeks after MI) study were included in this survival analysis. All mortality occurred within the first week. (E) Protocols for evaluation of the inducibility of ventricular tachyarrhythmias (VT). Left Panel: Mouse hearts were isolated and Langendorff-perfused with Tyrode solution containing isoproterenol, ex vivo ECG were continuously measured by placing recording electrodes around the heart as we previously described 23,54 . Programmed electrical stimulation (PES) was applied using a MyoPacer (IonOptix) via a pair of platinum electrodes placed on the left ventricular apex of the heart. Right Panel: the standard stimulation protocol consisted of 10 stimuli at 100 ms intervals (S1, 5V) followed by one extra stimulus (S2) starting at an interval of 80 ms which was then reduced by 2 ms until the effective refractory period (ERP) was reached. If VT or VF was not induced, a second extra stimulus (S3) was added at 80 ms after S2. The S3 interval was then reduced by 2 ms until the ERP was reached. Finally, a third extra stimulus (S4) was added 80 ms after S3 and was then decreased by 2 ms until the ERP was reached. If a heart failed to develop a VT or VF with 3 extra stimuli, the heart was deemed non-inducible. www.nature.com/scientificreports/ was induced in only 18% (2/11) of the post-MI KO mice at week 8 indicating reduced susceptibility to ventricular arrhythmias (p < 0.05 vs. WT, Fig. 1G,). Interestingly, a significant portion of the KO mice at week 8 after MI (55% vs. 15% in WT) exhibited isolated or coupled premature ventricular contractions (PVCs) that did not develop into VT even when stimulated with the maximum PES strength (Fig. 1G). At week 1 after MI when the scar tissue was still immature, the majority of WT and KO hearts only exhibited isolated or coupled PVCs with VT induced in only one WT heart (Fig. 1G), suggesting a low inducibility of VT at this subacute phase.
β-catenin knockout reduces the prolongation of QRS duration and QT interval after myocardial infarction. Surface ECG recording in live animals showed no difference between sham-operated KO and WT mice in any of the parameters analyzed (QT interval, QRS duration, PR internal and RR internal, Fig. 2). WT mice that received LAD ligation (MI) surgery exhibited prolonged QT interval (starting at week 1) and increased QRS duration (starting at week 3), but showed no changes in PR or RR intervals, suggesting that MI led to left ventricular remodeling without significantly affecting the function of the central conduction system (sinoatrial node and atrioventricular node), although it is not clear if the intrinsic heart rate (i.e. after blocking the effects of autonomic nervous system) is altered. The MI-induced QT interval prolongation and QRS duration increases were attenuated in KO mice compared with WT mice (Fig. 2), which is consistent with reduced VT susceptibility in KO mice.
β-catenin knockout attenuates chamber dilation and pump dysfunction after myocardial infarction. Echocardiography (Fig. 3A)  β-catenin knockout reduces scar sizes after myocardial infarction. Masson's trichrome staining of sham-operated hearts did not show any difference in gross morphology between KO and WT mice (Fig. 4A). Consistent with observations in echocardiogram, post-MI WT hearts showed LV chamber dilation and wall thinning (Fig. 4A). The scar size, as determined by the % fibrotic area of the LV wall, was smaller in post-MI KO (Scar size 38.5 ± 3.9%, n = 5 vs. 50.3 ± 3.1%, n = 6 in post-MI WT at 8-week, p < 0.05, Fig. 4B).
Upregulation of Wnt pathway genes in myocardium after myocardial infarction. Wnt pathways are fine-tuned by the soluble Wnt ligands and Wnt inhibitors on the extracellular side of the plasma membrane, as well as the receptors and co-receptors on the plasma membrane (Fig. 5A). Our qPCR analyses of gene transcripts of the myocardial tissues, which contained both cardiomyocytes and other cell types such as fibroblasts (Fig. 5B), showed that a variety of Wnt ligands, inhibitors and receptors are upregulated in the border zone and infarct region of the mouse hearts after MI ( Fig. 5 and 6).
Lrp6 also showed a 1.7-2.3 fold increase at week 1 (p < 0.05 in WT infarct and in KO border vs. shams), and a ~ twofold increase in infarct at week 8 (p < 0.05 in KO infarct vs. KO sham). But in the border zone at week 8 both Lrp5 and Lrp6 returned to near sham levels. Ror2 and Vangl2 are co-receptors required for non-canonical Wnt mediated activation of β-catenin-independent pathways. Ror2 and Vangl2 were increased in both border and infarct at week 1 by 4-9 fold and 2-4 fold (p < 0.05 vs. WT or KO shams), respectively, and they also showed a significant decline in border zone at week 8.  www.nature.com/scientificreports/ β-catenin protein.
In cells not stimulated with Wnt/β-catenin pathway ligands, β-catenin in the cytoplasm is phosphorylated at Ser33/Ser37/Thr41 which labels them for ubiquitination and degradation 36 . The level of β-catenin that is not phosphorylated at Ser33/Ser37/Thr41 (i.e., active β-catenin) is an index of the Wnt/βcatenin pathway activation in the cells 37 . As shown in Fig. 6C, active β-catenin was increased by 4.9 fold in the border zone myocardium of WT hearts at 1 week after LAD ligation (p < 0.01 vs. WT sham), suggesting activation of the Wnt/β-catenin pathway in the myocardial tissue. An increase in active β-catenin at 1 week (3.9 fold, p < 0.05) was found in KO hearts in which only the cardiomyocytes were deleted of β-catenin. This suggests activation of the Wnt/β-catenin pathway at 1 week in non-cardiomyocytes (such as fibroblasts and inflammatory cells). In addition, the increased total β-catenin in KO hearts at 1 week also suggests an increased proportion of non-cardiomyocytes (e.g., by fibroblast proliferation and infiltration of inflammatory cells) in the border zone myocardium.

Upregulation of Wnt pathway genes in ventricular myocytes after myocardial infarction.
In order to investigate if Wnt pathway genes are altered in cardiomyocytes after MI, single cardiomyocytes were isolated from the infarct and border tissues (or from the left ventricular free wall in sham-operated groups) after collagenase digestion of the hearts (Fig. 7A). Cardiomyocytes were purified by removing non-cardiomyocyte cells after settlement for 10 min (Fig. 7A). qPCR analysis of pure cardiomyocytes showed that many of the Wnt pathway genes are also upregulated after MI (Fig. 7B to D).
Wnt ligands (Fig. 7B). Wnt4 was increased in cardiomyocytes by 17-38 fold at week 1 (p < 0.05 in KO vs. sham), which is similar to its increase in myocardial tissue (~ 30 fold). Wnt10 showed a trend of 8-13 fold increase in cardiomyocytes at week 1 (p = 0.06 in KO vs. sham). Wnt2 showed a trend of 4-6 fold increase in cardiomyocytes at week 1 (p = 0.06 in both WT and KO vs. shams). Wnt1 was increased but to a lesser degree (by ~ 30%, p < 0.05) than in myocardial tissue (by 3-7 fold), while Wnt3a was not affected in cardiomyocytes (p > 0.25) after MI. Wnt7 was increased by 6-13 fold (p < 0.05, WT week 1 vs. WT sham), which is less than its increase in myocardial tissue (20 fold).
Wnt inhibitors (Fig. 7C) . Sfrp2 was increased by 52-68 fold in cardiomyocytes at week 1 (p < 0.05 in KO vs. sham) and by 16 fold at week 6 (p < 0.05 in KO vs. sham), which were less than its increase in myocardial tissues (97-256 fold at week 1). Sfrp1 and Sfrp3, which had ~ 30 fold increases in myocardial tissues at week 1, were increased in cardiomyocytes by 8-17 fold and 8-12 fold, respectively (p < 0.05 in both WT and KO vs. shams) but were not significantly affected at week 6. Dkk3 was increased by 7-13 fold at week 1 (p < 0.05 in both WT and KO vs. sham) which is similar to its increase in myocardial tissues (6-11 fold), while Dkk1 was not statistically affected (p > 0.09) after MI.
Wnt receptors and co-receptors (Fig. 7D). Fzd2, which was increased by 8-13 fold in myocardial tissues at week 1, was increased in cardiomyocytes by 4-5 fold at week 1 (p < 0.05 in both WT and KO vs. shams). Similarly, Fzd1, which was increased by 5-7 fold in myocardial tissue, was increased in cardiomyocytes by 2-3 fold at week 1 (p < 0.05 in KO vs. sham). Fzd7 was increased by 90% in KO at week 1 (p < 0.05 vs. sham). Fzd3, Fzd8 and Fzd9 were not statistically affected by MI (p > 0.57, p > 0.09 and p > 0.33, respectively). Fzd5 was reduced by 40% in WT at week 1 (p < 0.01 vs. WT sham) and reduced by 54% in KO at week 6 (p < 0.05 vs. KO sham). Lrp5 showed a trend of increase in cardiomyocytes at week 1 but was not statistically significant (p = 0.08 in KO vs. sham), while Lrp6, which had a ~ 2 fold increase in myocardial tissue, was reduced in cardiomyocytes by 45% at week 1 (p > 0.05 in WT vs. sham). Ror2 was increased (by 5-8 fold, p < 0.01) to a similar level in myocardial tissue (4-9 fold) at week 1. Vangl2, which was increased by 2-4 fold in myocardial tissue, was increased in cardiomyocytes only by 54% in KO at week 1 (p < 0.05 vs. KO sham).

Ion channel gene expressions after myocardial infarction. Ion channels genes for ventricular de-
polarizing currents (Fig. 8A). In the border zone at week 1, the cardiac Na + channel gene transcript (Scn5a, encoding Na v 1.5) was reduced in WT hearts (0.544 ± 0.07, n = 7 vs. WT sham 1.0 ± 0.07, n = 8, p < 0.01), but it was not reduced in β-catenin KO hearts (0.76 ± 0.15, n = 5 vs. KO sham 0.90 ± 0.04, n = 9, p > 0.99). Because it is known that Wnt/β-catenin signaling reduces Scn5a level in cardiomyocytes 22,23,25 these observations suggest that the Wnt/β-catenin pathway likely plays a role in Scn5a downregulation in the border zone of post-MI hearts.   www.nature.com/scientificreports/ Cacna1c in the border zone has returned to near sham levels. The high level of Cacna1c in infarct may reflect the expression of this channel in fibroblasts 38 which are abundant in infarct tissues.
Ion channels genes for ventricular repolarizing currents (Fig. 8A). All the major ventricular K + channel genes, including I to (Kcnd2, Kcnd3), I Kr (Kcnh2), I Ks (Kcnq1) and I K1 (Kcnj2), were reduced (p < 0.05) in border and infarct tissues at both week 1 and 8 as compared to sham groups. Consistent with our previous findings that Wnt signaling does not affect I K1 or action potential repolarization 22 , no differences were seen between KO and WT groups in these K + channel genes after MI.
Gap junctions and HCN channels (Fig. 8B). In the border zone, transcript of Cx43 (Gja1), the predominant gap junction isoform in ventricular myocardium, was reduced (p < 0.05) at week 1 but it returned to near sham levels at week 8 (Fig. 8B). However, immunostaining of Cx43 protein in border zones of both KO and WT hearts at week 8 showed that Cx43 is more diffusively distributed and is no longer restricted to cell-cell junction regions (intercalated disks) as seen in sham groups (Fig. 8C). Cx40 (Gja5), which is expressed in the Purkinje fibers, was not affected by MI. Among the genes of HCN channels that, if increased, will enhance myocyte automaticity, Hcn1 was increased while Hcn2 and Hcn4 were reduced at week 1 but they returned to near sham levels at week 8 (Fig. 8B). No differences were seen between KO and WT hearts in these gene transcript levels.

Discussion
Although Wnt/β-catenin signaling has been studied in animal models of ischemic heart disease, this is the first study to investigate the post-MI arrhythmogenesis using β-catenin knockout mice. The reduced susceptibility to VT in β-catenin KO mice after MI is likely due to attenuated structural remodeling. The low VT inducibility at week-1 in both WT and KO mice when the scar was immature and at week-8 in KO mice that had a smaller scar highlights the importance of structural remodeling in VT induction. Previous studies in patients have also demonstrated a positive relationship between scar sizes and the risk for VT 39 . In addition, large scar sizes are associated with a longer cycle length for monomorphic VT 40 , further supporting the reentry as the predominant mechanism for VT in post-MI hearts 41 .
The increased QRS duration, reflecting electrical dyssynchrony, is commonly found in heart failure patients and has been demonstrated to be an independent predictor of mortality in these patients 42,43 . Consistent with this, in the present study the increased QRS in post-MI WT hearts is associated with a greater VT inducibility. The increased QRS in these mice is likely secondary to the structural remodeling (scar and chamber dilation), as well as possible reductions in the conduction velocity of Purkinje fibers (arborization block) which is common in ischemic heart failure 44 . The QT interval and its dispersion are positively associated with the scar size in patients with MI and their increases are a risk factor for VT 45 . The attenuated increases in both QRS duration and QT intervals in β-catenin KO mice are consistent with their reduced VT inducibility and a smaller scar size.
The upregulation of Wnt pathway genes (such as Wnt1 and Wnt4) in the myocardial tissue of post-MI hearts is consistent with previous studies 15,46 . In addition, the present study provided direct evidence for the upregulation of Wnt genes in the cardiomyocytes. The similar levels of Wnt4 increase (~ 30 fold) in myocardial tissues and purified cardiomyocytes suggest that it is upregulated at the same degree in both cardiomyocytes and other cell types (such as fibroblasts, vascular cells and inflammatory cells). In contrast, other Wnt genes that are significantly increased in myocardial tissues are either upregulated at a lower level (such as Wnt1 and Wnt7) or not increased (such as Wnt3a) in cardiomyocytes, suggesting differential levels of upregulation in these Wnt genes between cardiomyocytes and other cell types. This finding may be helpful for future designing of therapeutic strategies for cell type-specific interventions of the Wnt pathways in ischemic heart disease.
While previous studies have reported conflicting results regarding the role of β-catenin in post-MI structural remodeling, our observations that β-catenin KO mice had attenuated structural remodeling after MI are consistent with the study by Zelarayan et al., which also found a smaller scar size in β-catenin KO mice at 4 weeks after MI (LAD ligation) 32 . The study by Zelarayan et al. also suggested that enhanced cardiac regeneration via activation of a cardiac progenitor population in the infarcted region may be a mechanism for the reduced scar in β-catenin KO hearts 32 . However, future studies are needed to investigate if other mechanisms, such as reduced cardiomyocyte deaths (in the form of necrosis, apoptosis, necroptosis and pyroptosis) or enhanced cardiomyocyte survival, are also involved. In addition to its role as the intracellular mediator of the Wnt/β-catenin signaling, β-catenin is also expressed in the cytoplasmic side of the plasma membrane in the cell-cell junction region, where it interacts with other membrane proteins such as N-cadherin 47,48 . Future studies are needed to investigate if the deletion of β-catenin in cardiomyocytes alters cell-cell interactions 46 within the ischemic myocardium as a potential mechanism for the attenuated post-MI structural remodeling.
The reduced Scn5a transcript level in WT hearts, but not in KO hearts, at 1 week after MI is consistent with previous observations by us and others that Wnt/β-catenin signaling reduces Scn5a levels in cardiomyocytes [22][23][24][25][26][27][28] . Our finding that Scn5a transcript returned to near sham levels in WT hearts at 8 weeks after MI is also consistent with the reduced Wnt gene levels at this timepoint as compared to 1 week. In addition, the similar levels of K + channel gene downregulations in both WT and KO hearts are consistent with our previous observation that Wnt/β-catenin signaling does not affect the resting membrane potential or the repolarization phase of action potentials in rat cardiomyocytes 22 . HCN channels have been shown to play a role in the arrhythmogenesis of dilated cardiomyopathy, and HCN2-overexpressing hearts have increased VT susceptibility 49 . In the present study, the selective upregulation of Hcn1 gene is consistent with previous studies using a mouse model of cardiac hypertrophy, although Hcn2 and Hcn4 are more abundant than Hcn1 50 . One limitation of the present study is that ion channel genes were not analyzed in purified cardiomyocytes and fibroblasts. Of note, the cardiac fibroblasts are known to express voltage-gated K + channels and transient receptor potential (TRP) channels 51 , and these cells www.nature.com/scientificreports/ can couple with cardiomyocytes in the myocardium to regulate electrical properties and arrhythmogenesis 51 . Future studies are needed to investigate the potential alterations in fibroblast ion channels and their contributions to the reduced arrhythmogenesis in β-catenin KO hearts after MI.
In summary, this study demonstrated that deletion of β-catenin reduces ventricular tachyarrhythmias in post-MI hearts primarily through attenuated structural remodeling. However, this study only examined the ischemic heart failure mouse model, and future studies are warranted to investigate the Wnt/β-catenin pathway activation in cardiomyocytes and its role in arrhythmogenesis in other types of heart failure, such as those induced by pressure overload or angiotensin II [17][18][19] .

Methods
All procedures were approved by the institutional animal care committee at the University of Ottawa (Protocol #: HI-2602) and all methods were performed in accordance with the ARRIVE guidelines 2.0 52 . All methods were carried out in accordance with relevant guidelines and regulations.

Experimental myocardial infarction. Myocardial infarction was induced in mice by permanent ligation
of the left anterior descending (LAD) coronary artery as we previously described 53 . Mice were injected with buprenorphine (0.05 mg/kg; subcutaneous) 1 h prior to surgery and twice daily thereafter for 3 days. During the surgery, mice were incubated, anesthetized using 2% isoflurane and maintained under physiologic temperature. The heart was exposed after a left thoracotomy, the LAD was identified under a dissection microscope (Topcon Corporation, Model No.: OMS-75) and ligated at 0.3 mm distal to the atrioventricular junction using a Prolene 7.0 suture. Successful LAD ligation was confirmed by the pale color of the affected myocardium before chest closure with a Prolene 6.0 suture. Surface ECG. ECG with limb Lead I and II were recorded in mice at one day before and at week 1, 3, 5 and 8 after the LAD ligation using ADInstruments small animal ECG system (Powerlab 8/35 and Animal Bio Amp) and data were analyzed using LabChart 8.0. During ECG recording, mice were anesthetized with 1.5-2.0% isoflurane and body temperature was kept at 37 °C using a heating pad. Mouse ECG parameters were analyzed in LabChart 8 software (version 8.1.5, ADInstruments) software. In brief, automated averaging of 10 consecutive beats was done by the software. Three consecutive "averaged cycles" were analyzed, and these values were finally averaged. To manually correct the auto-detected intervals if they were incorrectly placed: the PR interval was measured from onset of the P wave to the onset of the QRS complex (in cases where the Q wave was not very visible, the onset of the R wave was used). The QRS duration was measured from onset of the QRS complex to the return to baseline voltage following the S wave (or if there was no return to baseline, the peak positive deflection following the S wave was used). The QT interval was measured from onset of the QRS complex to the return to baseline voltage following the T wave.
Echocardiography. Echocardiography was recorded using a VEVO3100 system (VisualSonic Inc., Toronto) with the MS400 transducer (VisualSonic Inc., Toronto), and images were analyzed in VevoLab software (v.3.2.0 VisualSonic Inc., Toronto). During recording, animals were anaesthetized under 1.5-2% isoflurane with body temperature maintained at 37 °C using a heating pad. To minimize the influence of heart rate on pump function, echocardiogram was recorded when heart rates were in the 400-500 bpm range. M-mode tracing of the left ventricle (LV) was recorded in the short-axis view at the mid-papillary level. Three consecutive cardiac cycles, measured in VevoLab, were averaged to determine ejection fraction % (EF), fractional shortening % (FS), end-diastolic volume (EDV), end-systolic volume (ESV), left ventricular internal diameter (LVIDs/LVIDd), and posterior wall thickness. The EF was calculated using the following equation: (LVIDd 3 -LVIDs 3 )/LVIDd 3 × 100. FS was calculated using the following equation: (LVIDd-LVIDs)/LVIDd × 100. B-mode tracing of the LV was recorded in the parasternal long-axis view and analyzed using the LV trace function in VevoLab to determine EF, FS, EDV, ESV, and LV volume.
Programmed electrical stimulation (PES) in isolated mouse hearts. At 1 and 8 weeks after MI, mouse hearts were isolated and Langendorff-perfused with Tyrode solution (37 °C) containing 1 µM isoproterenol (Sigma, Catalogue No.: I6504), ex vivo ECG were continuously measured by placing recording electrodes around the heart as we previously described 23,54 (Fig. 1E). PES was applied using a MyoPacer (IonOptix) via a pair of platinum electrodes placed on the left ventricular apex of the heart. The standard stimulation protocol (Fig. 1E) consisted of 10 stimuli at 100 ms intervals (S1, 5V) followed by one extra stimulus (S2) starting at an interval of 80 ms which was then reduced by 2 ms until the effective refractory period (ERP) was reached. If VT or VF was not induced, a second extra stimulus (S3) was added at 80 ms after S2. The S3 interval was then www.nature.com/scientificreports/ reduced by 2 ms until the ERP was reached. Finally, a third extra stimulus (S4) was added 80 ms after S3 and was then decreased by 2 ms until the ERP was reached. If a heart failed to develop a VT or VF with 3 extra stimuli, the heart was deemed non-inducible.
Masson's Trichrome staining. Mouse hearts were fixed by retrograde perfusion via the aorta with 4% paraformaldehyde (diluted in PBS) for 10 min at room temperature. The hearts were then incubated in increasing concentrations of sucrose dissolved in PBS (10% for overnight, 20% for 8 h, and 30% for overnight) on a horizontal rotator at 4 °C before being embedded in TissueTek OCT compound. Hearts were then cryosectioned at 10-µm slices using a Leica vibrating microtome (Model No.: CM3050S) at five different levels of the heart from the ligation site to the apex at 300-µm intervals. Ten consecutive slices were made for each level. Heart sections were then stained using a Masson's Trichrome Staining kit (Sigma, Catalogue No.: HT15) according to the manufacturer's instructions. Fibrosis areas were analyzed with ImageJ software using the Colour Deconvolution for Masson's Trichrome Stain tool. The scar size for each the 5 different levels was calculated as the percentage of fibrotic area to the total area of the left ventricle section. The heart's total scar size was then calculated as the mean of the scar sizes at the 5 levels and reported in Fig. 4.
Immunohistostaining. Cryosectioned heart slices, as prepared above, were blocked and permeabilized www.nature.com/scientificreports/ of the gels was assessed by re-probing the membrane with rabbit anti-GAPDH antibody (1:2000, Cell Signaling, #2118). Band densities were quantified using the "Gel Analyzer Protocol" function of ImageJ (https:// imagej. nih. gov/ ij/ docs/ menus/ analy ze. html# gels), and presented as values after normalization to GAPDH in the same samples. All the original uncropped gel images are included in Supplementary Fig. 1.

Statistical analysis.
Statistical analysis adhered to the Journal Guidelines. Data are expressed as mean ± SEM with p < 0.05 considered significant. Sample number indicates the number of biological replicates (each individual mouse is considered one sample). Differences between two means were evaluated by two-tailed Student's t-test. Differences among multiple means were assessed by one-way or two-way analysis of variance (ANOVA). When significance was detected by ANOVA, differences among individual means were evaluated post hoc by Bonferroni's test.