Plakophilin-2 Haploinsufficiency Causes Calcium Handling Deficits and Modulates the Cardiac Response Towards Stress

Human variants in plakophilin-2 (PKP2) associate with most cases of familial arrhythmogenic cardiomyopathy (ACM). Recent studies show that PKP2 not only maintains intercellular coupling, but also regulates transcription of genes involved in Ca2+ cycling and cardiac rhythm. ACM penetrance is low and it remains uncertain, which genetic and environmental modifiers are crucial for developing the cardiomyopathy. In this study, heterozygous PKP2 knock-out mice (PKP2-Hz) were used to investigate the influence of exercise, pressure overload, and inflammation on a PKP2-related disease progression. In PKP2-Hz mice, protein levels of Ca2+-handling proteins were reduced compared to wildtype (WT). PKP2-Hz hearts exposed to voluntary exercise training showed right ventricular lateral connexin43 expression, right ventricular conduction slowing, and a higher susceptibility towards arrhythmias. Pressure overload increased levels of fibrosis in PKP2-Hz hearts, without affecting the susceptibility towards arrhythmias. Experimental autoimmune myocarditis caused more severe subepicardial fibrosis, cell death, and inflammatory infiltrates in PKP2-Hz hearts than in WT. To conclude, PKP2 haploinsufficiency in the murine heart modulates the cardiac response to environmental modifiers via different mechanisms. Exercise upon PKP2 deficiency induces a pro-arrhythmic cardiac remodeling, likely based on impaired Ca2+ cycling and electrical conduction, versus structural remodeling. Pathophysiological stimuli mainly exaggerate the fibrotic and inflammatory response.


Introduction
Arrhythmogenic right ventricular cardiomyopathy (ACM) is an inherited heart disease characterized by fibrous or fibrofatty infiltration of the cardiac muscle, ventricular arrhythmias, PKP2-Hz hearts is likely based on impaired Ca 2+ cycling and a disturbed electrical conduction, instead of structural remodeling. Showing that PKP2 deficiency controls the cardiac response towards environmental modifiers via different mechanisms.

Characterization of PKP2-Hz Hearts Over Time
Since cardiac electrical and structural remodeling progresses with age, we compared three-monthto six-month-old WT and PKP2-Hz murine hearts [29,30]. Histological analysis by hematoxylin and eosin (H&E) and Picrosirius red revealed no structural modifications at both ages and genotypes ( Figure 2A,B), which was confirmed by RNA expression analysis of collagen-and inflammation-related genes ( Figure 2C). Timp2 expression was increased (0.83 ± 0.18 vs. 1.98 ± 0.40, p ≤ 0.05) in three-month-old PKP2-Hz mice, but this effect did not persist over time ( Figure 2C). The heart weight/body weight (HW/BW) index decreased over time, but did not differ between WT and PKP2-Hz mice ( Figure 2D and Table S1). Electrophysiological parameters were unchanged and arrhythmia incidence was rare in six-month-old WT hearts (10%; Figure 2E and Table S2). Western blot and immunohistochemistry revealed no difference in expression and localization of Ncad and Cx43 ( Figure 2F,G). PKG expression however was 49% lower in six-month-old PKP2-Hz hearts compared to WT (1.0 ± 0.13 vs. 0.51 ± 0.03, p ≤ 0.05; Figure 2C). Representative western blots (left) and average densitometry (right; n = 5 for all groups) of ankyrinB (AnkB), calsequestrin-2 (Casq2), Cav1.2, and SERCA2a, measured from wildtype and PKP2-Hz ventricular lysates of three-month-old mice. Left upper panels represent western blots, bottom panels according to ponceau staining used for quantification (mean ± SEM, * p < 0.05).

Characterization of PKP2-Hz Hearts Over Time
Since cardiac electrical and structural remodeling progresses with age, we compared threemonth-to six-month-old WT and PKP2-Hz murine hearts [29,30]. Histological analysis by hematoxylin and eosin (H&E) and Picrosirius red revealed no structural modifications at both ages and genotypes (Figure 2A,B), which was confirmed by RNA expression analysis of collagen-and inflammation-related genes ( Figure 2C). Timp2 expression was increased (0.83 ± 0.18 vs. 1.98 ± 0.40, p ≤ 0.05) in three-month-old PKP2-Hz mice, but this effect did not persist over time ( Figure 2C). The heart weight/body weight (HW/BW) index decreased over time, but did not differ between WT and PKP2-Hz mice ( Figure 2D and Table S1). Electrophysiological parameters were unchanged and arrhythmia incidence was rare in six-month-old WT hearts (10%; Figure 2E and Table S2). Western Representative western blots (left) and average densitometry (right; n = 5 for all groups) of ankyrinB (AnkB), calsequestrin-2 (Casq2), Ca v 1.2, and SERCA2a, measured from wildtype and PKP2-Hz ventricular lysates of three-month-old mice. Left upper panels represent western blots, bottom panels according to ponceau staining used for quantification (mean ± SEM, * p < 0.05). blot and immunohistochemistry revealed no difference in expression and localization of Ncad and Cx43 ( Figure 2F,G). PKG expression however was 49% lower in six-month-old PKP2-Hz hearts compared to WT (1.0 ± 0.13 vs. 0.51 ± 0.03, p ≤ 0.05; Figure 2C).

Pro-Arrhythmic Cardiac Remodeling in PKP2-Hz Mice Exposed to Exercise
As endurance training decreases cardiac function and provokes ventricular arrhythmias [17,18], PKP2-Hz mice were subjected to one-month voluntary running on a treadmill. PKP2-Hz mice exposed to exercise training did not show increased levels of fibrosis ( Figure 3A, Figure S3), signs of hypertrophy ( Figure 3B,C, Table S1), affected cardiac contractility ( Figure 3F), or changes in electrocardiogram (ECG) parameters (Table S2), compared to controls. The arrhythmia incidence was clearly higher in trained PKP2-Hz mice compared to trained WT animals (PKP2-Hz vs. WT; 40% vs. 0%; Figure 3H). Impaired intracellular Ca 2+ dynamics (as shown in the three-and six-month-old mice) in combination with endurance exercise possibly increases the susceptibility towards arrhythmias in PKP2-Hz hearts [11].
To study the electrical cell-cell coupling in trained mice we analyzed Cx43 localization by immunofluorescence and measured the ventricular conduction velocity by epicardial mapping. PKP2-Hz mice exposed to voluntary exercise presented right ventricular lateral Cx43 expression ( Figure 3E) and longitudinal right ventricular conduction slowing, compared to WT mice (left ventricle-LV: 72.10 ± 9.97 vs. 67.43 ± 1.94, ns.; and right ventricle-RV: 59.98 ± 2.58 vs. 52.78 ± 6.11, p < 0.05; Figure 3G). The transversal ventricular conduction velocity did not differ between groups ( Figure 3G).
To further investigate the determinants of arrhythmogenicity in these mice, we performed tissue analyses for expression of cell-coupling proteins and interstitial fibrosis. Immunolabelling revealed no difference in expression and localization of NCad and Cx43 ( Figure 4F). Sirius red staining for collagen abundance presented significant higher levels of interstitial (but not subepicardial) fibrosis in PKP2-Hz TAC mice compared with sham-treated WT and PKP2-Hz mice (WT-Sham: 1.64% ± 0.1%; WT-TAC 2.83% ± 0.95%; PKP2-Hz Sham: 1.86% ± 0.29%; and PKP2-Hz TAC: 5.0% ± 0.86%, p ≤ 0.05 vs. WT-Sham and PKP2-Hz Sham; Figure 4G,H). To explore if the Wnt signalling pathway was involved in this pro-fibrotic response, we examined protein levels of β-catenin in ventricular lysates of all groups. Levels of β-catenin were significantly higher in ventricular lysates of TAC mice and this difference was more pronounced in PKP2-Hz mice compared to WT ( Figure S5).

Discussion
ACM is seen as a disease of the intercalated disc (ID) and human variants in PKP2 associate with most familial ACM cases. PKP2 is mainly a component of the desmosome, a cellular structure involved in cell-cell adhesion [5,11,32]. Besides PKP2, genetic variants in other desmosomal genes (Desmocollin2; DSC2, Desmoglein2; DSG2, and Plakoglobin; PKG), have been linked to ACM as well [33]. Since the environmental and genetic modifiers of ACM remain unknown, we examined the effect of exercise, cardiac pressure overload, and inflammation on the progression of a PKP2-related cardiomyopathy. We show that PKP2 deficiency modulated the cardiac response to environmental modifiers via different mechanisms. Exercise induced a pro-arrhythmic cardiac remodeling, likely based on a combination of impaired Ca 2+ cycling and electrical conduction. Upon pathophysiological stimuli, PKP2 haploinsufficiency mainly exaggerated the fibrotic and inflammatory response.
First of all, we found that PKP2 haploinsufficient mice displayed deficits in Ca 2+ -handling-related proteins, most critically of Ca V 1.2, SERCA2a, AnkB, and Casq2. This correlates with findings in PKP2 conditional global knock-out (PKP2cKO) mice, in which the expression of proteins involved in Ca 2+ i cycling is markedly reduced [11]. Altered Ca 2+ i transients and Cx43 disfunction potentially causes the ventricular arrhythmias in these mice [11,34]. In addition, PKP2cKO mice present biventricular mechanical dysfunction and fibrosis formation, both starting in the right ventricle [11]. Spontaneous ventricular arrhythmias were observed before the onset of LV dysfunction, followed by end-stage failure and sudden death [11]. In contrast, PKP2-Hz hearts did not display arrhythmogenic events, suggesting that a 50% reduction of PKP2 did not suffice to provoke spontaneous ectopic activity in the heart. It is tempting to speculate that a combination of factors was needed to induce pathological (electrical) remodeling upon PKP2 deficiency. In humans, PKP2 variants are transmitted in an autosomal-dominant manner and mutation carriers can proceed throughout life without developing any signs of cardiomyopathy, similar as observed in the PKP2-Hz mouse model [6,15,35].
In mice, heterozygous loss of PKP2 expression associates with ultrastructural defects, sodium current deficits and hampered sodium channel inactivation kinetics [15,16]. However, under baseline conditions PKP2-Hz mice appear farily nomal and lack an overt disease phenotype [15]. In line with the literature, three-or six-month-old PKP2-Hz mice displayed no structural or electrical abnormalities. This might be explained by the relatively young age of the mice. We acknowledged that six months of age is still mild ageing in mice and increasing the age towards 18-22 months would probably enhance fibrosis formation and arrhythmia incidence [36]. Senescent mice become arrhythmogenic by increased interstitial fibrosis levels and redistributed Cx43 [36]. However, whether extensive aging will discriminate between WT and PKP2-Hz mice remains to be determined.
In line with the literature we observed that PKP2-Hz hearts of mice exposed to one month of voluntary running displayed a higher vulnerability towards sustained ventricular arrhythmias, although lacking any signs of fibrosis [9,12]. While in humans ACM penetrance is low, exercise also greatly increases the risk for developing the cardiomyopathy and for its progression toward failure [17,19]. Life-threatening ventricular arrhythmias or sudden death often occur in the concealed phase of the disease, prior to overt structural damage [37]. The underlying mechanisms responsible for these arrhythmias remain under study but a disturbed electrical coupling could be an important factor [15]. Acute adrenergic stimulation of PKP2 null mice associates with malignant ventricular arrhythmias and sudden death, mimicking exercise during the concealed phase of the disease [11]. In accordance, exercise training in PKP2-R735 mutant mice appeared crucial for developing the ACM phenotype [18]. In addition, we show that a 50% reduction of PKP2 expression was not associated with exercise-induced contractile impairment, in contrast to the study of Cruz et al. [18]. Important to note is the difference in the PKP2 model (PKP2-Hz vs. AAV-induced PKP2-R735 mice), the extent of exercise training (one month of running vs. eight weeks of swimming), and the contractile parameters examined (left ventricular fractional shortening (LVFS) vs. right ventricular ejection fraction (RVEF)) [18]. For future studies, we recommend to explore RV contractility in PKP2-Hz mice exposed to exercise training. Sudden death upon exercise was not observed, and PKP2-Hz mice and their hearts lack any signs of hypertrophy. However, we could not control the intensity of training in our voluntary exercise experiments. Chronic isoproterenol infusion in PKP2-Hz mice would allow studying the effect of adrenergic stimulation on disease development in a more controlled experimental setting [38].
Exercise training of PKP2-Hz mice associates with RV conduction slowing, potentially caused by a combination of impaired mechanical coupling, reduced peak sodium current and redistributed Cx43 [15,18,36]. The absence of structural remodeling in our PKP2-Hz mice suggests that exercise-induced pro-arrhythmic remodeling in PKP2-Hz hearts is caused by an impaired cardiac electrical conduction. The reduced expression of Ca 2+ -handling-related proteins in trained PKP2-Hz hearts is likely not pro-arrhythmic in itself. Though it is tempting to speculate that a combination of impaired intracellular Ca 2+ cycling and impaired intercellular electrical coupling serves as a pro-arrhythmic substrate in exercise-exposed PKP2-Hz hearts. This needs to be confirmed by in vivo recordings of ventricular arrhythmias in PKP2-Hz mice during exercise, for example by telemetry.
Cardiac pressure overload is a well-known contributor to electrical and structural remodeling of the heart and a risk factor for cardiac arrhythmias [39,40]. In line with the literature we observed that TAC surgery induced hypertrophy, impaired cardiac contractility, and increased susceptibility for arrhythmias; however, PKP2-Hz mice presented similar effects as WT mice [21,39]. TAC surgery increased collagen abundance predominantly in PKP2-Hz hearts. This implicates that PKP2 is involved in regulating the cardiac fibrotic response, as PKP2 haploinsufficiency might lead to activation of pro-fibrotic pathways, inflammation, and apoptosis [40,41]. Enhanced ß-catenin levels in PKP2-Hz TAC mice might confirm this hypothesis, since a loss of desmosomal integrity alters the Wnt-signalling pathway, a well-known contributor to adipo-and fibrogenensis and ACM pathology [42]. Imbalance of the intracellular Ca 2+ homeostasis and hypertrophic signals possibly increase calcineurin activation in the heart and subsequent inflammatory-driven fibrosis in PKP2-Hz TAC mice [43]. Additional studies are needed to confirm these hypotheses.
In this study we exposed PKP2-Hz mice to an experimental auto-immune myocarditis and studied the acute and innate immune response of the heart. Whole heart collagen levels were not affected by the intervention, although PKP2-Hz mice developed severe levels of subepicardial fibrosis and myocardial damage six weeks after EAM onset. Inflammation was a common finding in ACM. The inflammatory response could be caused by cell death, viral infection, or be a consequence of defective desmosomes [44,45]. Endomyocardial biopsies of ACM patients show features of myocarditis-like inflammation or fibrofatty replacement, possibly reflecting different stages of the disease [27,46]. Acute myocarditis reflects an active phase of ACM and is a superimposed phenomenon during the natural history of the disease [27]. Cardiac fibrofatty replacement in desmosomal mutation carriers is mainly located in the posterolateral wall of the heart, confirming the observations we saw in PKP2-Hz mice exposed to EAM [47]. The molecular mechanism underlying this epicardial response is still under debate. Paracrine signaling between PKP2-positive and PKP2-negative cells in the epicardial region of PKP2-Hz mice is a likely causative factor, activating a TGF-β1/p38 MAPK-dependent fibrotic gene expression [48]. Inflammatory infiltrates were already present at three weeks post-EAM onset in PKP2-Hz mice, suggesting an inflammation-driven fibrotic response. We concluded that PKP2 deficiency altered the cardiac inflammatory and fibrotic response to EAM. This study fitted the hypothesis that ACM might be triggered by infection, as desmosomal mutations compromised the resilience of the heart to inflammation. Inflammation itself may also be arrhythmogenic as previous data suggest that cytokines can alter sodium current function [49]. Future studies are needed to unravel the exact arrhythmic potential of EAM in the PKP2 haploinsufficient heart.
In conclusion, reduced expression of PKP2 in the murine heart affects Ca 2+ -cycling-related protein levels and modulates the cardiac reponse towards environmental modifiers. Cardiac pressure overload and auto-immune myocarditis drive (subepicardial) pro-fibrotic mechanisms in the PKP2 happloinsufficient heart. Exercise-induced pro-arrhythmic cardiac remodeling in hearts with reduced expression of PKP2 is based upon impaired electrical conduction instead of structural remodeling. Our data presented potential important disease mechanisms in a PKP2-related cardiomyopathy.

Study Limitations
The data presented in this study suggest that decreased levels of Ca 2+ -handling proteins played a maladaptive role in the development of different cardiac phenotypes upon environmantal modifiers in PKP2-Hz mice. Levels of Ca 2+ -handling-related proteins were not examined in mice exposed to cardiac pressure overload and autoimmune myocarditis, so whether cardiac remodeling and Ca 2+ -handling disturbances were directly linked in these models remains uncertain. In addition, no intracellular Ca 2+ measurements were executed. More in-depth in vitro studies need to be performed to confirm the definite effect of PKP2 haploinsuffiency on cardiac Ca 2+ dynamics. Another limitation of this study was the lack of arrhythmia score in PKP2-Hz hearts exposed to autoimmunce myocarditis. Histological examination showed that pathophysiological stimuli mainly exacerbated the fibrotic and inflammatory response, but pro-arrhythmic remodeling in PKP2-Hz hearts exposed to autoimmune myocardits could not be excluded

Mouse Model
Mice heterozygous-null for the PKP2 gene (PKP2-Hz) was generated and genotyped as described previously [7]. Except where noted, experiments were conducted in PKP2-Hz and WT littermate

Transverse Aortic Constriction Procedure
Wildtype and PKP-Hz 12-week-old mice were TAC or sham operated as explained previously [50]. Briefly, mice were anesthetized by isoflurane (mean 2.5% in oxygen), intubated with a 20 G polyethylene catheter, and ventilated (200 µL, 160 strokes/min) with a rodent ventilator (Minivent, Hugo Sachs Electronics, March-Hugstetten, Germany). The thoracic cavity was accessed through a small incision at the left upper sternal border in the second intercostal space. A 7-0 silk suture was passed around the aorta between the right innominate and left common carotid arteries. Constriction of the transverse aorta was performed by tying against a 27 G needle, which was subsequently removed. Sham-operated mice underwent the same procedure without aortic binding. Eight weeks after surgery echocardiograms and electrocardiograms were recorded, the mice were sacrificed by cervical dislocation and the hearts were removed for histological studies.

Exercise Protocol
Wildtype and PKP2-Hz 12-week-old mice were housed solitary in a cage with a treadmill for one month, where they were exposed to daily voluntary exercise training. After one-month voluntary exercise training, echocardiograms and electrocardiograms were recorded, mice were sacrificed by cervical dislocation and the hearts were removed for histological and electrophysiological studies.

Experimental Autoimmune Myocarditis
Myocarditis was induced by immunization with cardiac α-myosin heavy chain (AnaSpec Inc., Fremont, CA, USA) as explained previously [51]. 100 µg cardiac α-myosin heavy chain was dissolved in 100 µL sterile phosphate buffered saline (PBS) and mixed with 100 µL of complete Freund's adjuvant (Sigma, F5881, st. Louis, MO, USA). This mixture was vortexed for 1.5 h to create an emulsion of antigen in complete Freund's adjuvant. On day 0, 200 µL of emulsion was injected subcutaneously at two flanks of wildtype and PKP2-Hz mice. For this study, mice were backcrossed onto the BALB/c background as it is permissive for myocarditis [51]. In addition, the mice were injected intraperitoneally with 500 ng of pertussis toxin (List Biological Laboratories, Campbell, CA, USA) dissolved in PBS. The mice were monitored for at least one week following the injections. A mouse was euthanized if it showed unusually large granulomas, ulcerations, or signs of being moribund. After three or six weeks, echocardiograms and electrocardiograms were recorded, the mice were sacrificed by CO 2 euthanasia and the hearts were removed for histological studies.

Echocardiography
Transthoracic echocardiography was performed using a Vevo2100 Imaging System (VisualSonics Inc., Toronto, Canada) with a 30 MHz probe. Briefly, after induction of anesthesia in a chamber containing isoflurane 4%-5% in oxygen, the mouse was positioned supinely on a heat pad in order to maintain a body temperature at 37-38 • C and anesthesia was maintained with 1.5% isoflurane in 700 mL O 2 /minute via a nose cone. Recordings were obtained in parasternal long and short axis views. Quantitative measurements were assessed offline using the Vevo2100 analytical software. A B-mode parasternal long axis view was used for left ventricular ejection fraction. Left ventricular fractional shortening was calculated from the parasternal short axis view (M-mode) [52].

Electrocardiograms Recordings
Mice were anesthetized with 1.5% isoflurane in 700 mL O 2 per minute via a nose cone (following induction in a chamber containing isoflurane 4%-5% in oxygen). The rectal temperature was monitored continuously and maintained at 37-38 • C using a heat pad. Three lead ECG (leads I, II, and III) were recorded from sterile needle electrodes inserted subcutaneously in each forelimb and hindlimb. The signal was then acquired and analyzed using a digital acquisition and analysis system (Power Lab; AD Instruments, Oxford, UK; LabChart 7Pro software version). ECG parameters were quantified after 1-2 min from anesthesia induction, in order to stabilize the trace. The QT interval was defined as the time elapsed from the beginning of the major deflection representing the QRS to the end of the secondary slow deflection, as described by Danik et al. [53]. QT intervals were corrected for RR interval by the equation (QTc = QT/(RR/100)1/2), according to Mitchell et al. [54]. Analysis was performed on lead II electrocardiograms.

Epicardial Activation Mapping in Langendorf-Perfused Hearts
For epicardial activation mapping experiments, mice were anesthetized (4% isoflurane in oxygen) and the hearts were quickly excised, rinsed, and placed on a Langendorf column for retrograde coronary perfusion. Hearts were continuously perfused with a Tyrode solution containing: (in mmol/L): NaCl 116, KCL 5, MgSO 4 1.1, NaH 2 PO 4 0.35, NaHCO 3 27, glucose 10, mannitol 16, and CaCl 2 1.8 at 37 • C. Solution was continuously gassed with 95% O 2 and 5% CO 2 . Electrograms were recorded using a 247-point multiterminal electrode (19 × 13 mm grid, 0.3 mm spacing) placed over both the LV and RV, as described before [36,50]. Recordings of the LV and RV were made during stimulation (2 ms duration, 2× diastolic stimulation threshold) from the center of the grid at a basic cycle length (BCL) of 120 ms. The moment of maximal negative dV/dt in the unipolar electrograms was selected as the time of local activation and determined using customized software [55]. Conduction velocity parallel (CV L ) and perpendicular to fiber orientation were determined from activation maps generated from BCL-pacing. Activation times of at least four consecutive electrode terminals along lines perpendicular to intersecting isochronal lines were used to calculate CVs.

PCR Validation of Gene Expression
Total RNA was extracted from the heart tissue of PKP2-Hz and control mice using an RNeasy Mini Kit (QIAGEN, Hilden, Germany). The complement DNA (cDNA) was generated by reverse transcription (RT)-PCR with SuperScript VILO cDNA Synthesis Kit (Thermo scientific, Landsmeer, Netherlands). For RT-qPCR on ventricular tissue, TaqMan gene expression assays (all from Applied Biosystems by Life Technologies Corp., Carlsbad, CA, USA) were used as described earlier [57]. As an internal control, the geometric mean of the TATA-binding protein (TBP) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used. Relative mRNA expression levels were determined for collagen 1α1 (Col1α1), collagen 1α2 (Col1α2), metalloproteinase 9 (MMP9), metallopeptidase inhibitor 1 (Timp1), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κß), and interleukin-6 (IL-6). Assay IDs are listed in Table S3.

Immunofluorescence Labeling
For the detection of protein localization by immunofluorescence, hearts were rapidly frozen in liquid nitrogen and stored at −80 • C. Hearts were sectioned (thickness: 10 µm) parallel to the long axis of the heart. Sections displaying a four-chamber view were used for immunostaining as previously described [56]. Samples were exposed to the primary antibodies Cx43 (1:250, BD transduction, 610062), N-cadherin (1:1000, Sigma, C3678), chicken anti-vimentin (1:400, EMD Millipore Corporation (Burlington, MA, USA), AB5733), and myosin (1:200, Enzo Life Sciences Inc. (Bruxelles, Belgium), ALX-BC-1150-S-L001) for primary labeling. Afterwards, Cx43-and Ncad-labeled sections were incubated with FITC conjugated anti-rabbit whole IgG or Texas Red-conjugated anti-mouse whole IgG. Vimentin-and myosin-labeled sections were incubated with secondary antibodies Alexa Fluor 488 goat anti-chicken IgG (1:500, abcam Inc., ab150169) and Alexa Fluor 568 goat anti-mouse IgG (1:500, Thermo scientific (Landsmeer, Netherlands)), A-11031). After labeling, sections were mounted in Vectashield (Vector Laboratories, Orton Southgate, Peterborough, UK) and visualized by conventional epifluorescence and confocal microscopy (Nikon optiphot-2). To examine the non-myosin area, pictures of 40 × magnification of the most damaged area in five heart areas; apex, left-ventricular base/mid region, and right-ventricular base/mid region, were taken. The perimeter of empty area continuous to the epicardial layer was manually selected (in triplet), excluding myocytes and vessels. The non-myosin area continuous to the epicardium was normalized to the length of the epicardium.

Immunohistochemistry
For detection of inflammatory infiltrates and gap junctions by immunohistochemistry, hearts were fixed with 4% paraformaldehyde in PBS. Hearts were sectioned (thickness: 10 µm) parallel to the long axis of the heart. Labeling was performed using the ImmPRESS anti-rat Ig Reagent kit, peroxidase (Vector Laboratories Inc., MP-7444), and the ImmPACT NovaRED Peroxidase substrate kit (Vector Laboratories Inc., SK-4805) to reduce background staining. Antigen retrieval was performed by incubating the sections in a citrate buffer (Biogenex, HK-080-9K) and heating this for 25-30 min using a steamer (Deni, Keystone Inc., Grand Island, NY, USA). Subsequently the sections were incubated with 0.5% hydrogen peroxide (Fisher Scientific (Landsmeer, Netherlands), H323-500) and incubated in 0.3% normal goat serum (Sigma, G9023) in PBS. After blocking, the samples were incubated with primary antibodies (Cx43, 1:250, Thermo Scientific 71-0700; anti-Ly-6G/Ly-6C, 1:200, Biolegend (San Diego, CA, USA) 108402) diluted in ready-to-use 2.5% normal goat blocking serum. Next, sections were incubated for 30 min with ImmPRESS (anti-rat and anti-rabbit) reagent and incubated in peroxidase substrate working solution. Finally, sections were incubated with Vector Hematoxylin QS nuclear counterstain, Gill's Formula (Vector Laboratories Inc., H-3404) and mounted with Vectashield (Vector Laboratories, H-1000). Quantification of inflammatory infiltrates was performed by computed analysis via Image J. RGB images were converted to stack images and red, blue, and green channels were separated. Red intensity should be at least 110% of the blue intensity, surface area and percent neutrophil positive area was quantified.

Picrosirius Red Staining
To evaluate the extent of fibrosis, sections were fixed with 4% paraformaldehyde in PBS, stained with Picrosirius red and examined by light microscopy. Briefly, paraffin embedded sections were deparaffinated and rehydrated. After repeated washes, slides were incubated in Picrosirius red working solution pH 2.0 (Picric acid, Sigma-Aldrich (Zwijndrecht, Netherlands), 74069; Sirius Red, Polysciences, Inc. (Hirschberg an der Bergstrasse, Germany) C.I. 35780) for 60 min. Sections were directly transferred to 0.01 M HCl solution and incubated under constant movement. Sections were dehydrated again and a coverslip was placed using Entallan (Merck, Schiphol-Rijk, Netherlands, 107960). Stained sections were scanned at a 40× magnification on a Leica SCN400F Whole Slide Scanner. Quantification of healthy myocardium was performed by computed analysis of the percentage red stained tissue with Image J (NIH), which refers to collagen. Regions of interest were selected manually by lining the tissue of both ventricles and septum. The subepicardial layer was defined as 0.2 mm tissues underneath the epicardial layer. By means of a macro the fraction of red staining to the total area was calculated.

Hematoxylin and Eosin Staining
To evaluate the extent of myocardial damage, sections were fixed with 4% paraformaldehyde in PBS, stained with hematoxylin and eosin according to the manufacturer's instructions, and examined by light microscopy. Briefly, paraffin embedded sections were incubated at 60 • C for 30 min. Slides were incubated in hematoxylin 2 (Thermo Scientific, 7231) for two minutes and directly transferred to clarifier 2 (Thermo Scientific, 7402) for 15 s. After repeated washes, slides were incubated in a Bluing reagent (Thermo Scientific, 1931423) for one minute. After a wash, slides were exposed to eosin Y (Thermo Scientific, 71204) for 1.5 min. After this step the sections were dehydrated again and a coverslip was placed using Permount Toluene solution. Stained sections were scanned at a 40× magnification on a Leica SCN400F Whole Slide Scanner. Two independent researchers assessed the subepicardial damage in HE-stained hearts, by determining the ratio of the non-myocyte area-continuous of the epicardial layer-to total tissue area. This was done by computed analysis via Image J (NIH) in five windows of 0.59 mm × 0.31 mm over the ventricular wall; apex, left-ventricular base/mid region, and right-ventricular base/mid region. The perimeter of the empty area continuous to the epicardial layer was manually selected (in triplet), excluding myocytes and vessels.

Statistical Analysis
Statistical analysis and drawing of graphs and plots were carried out in GraphPad Prism (version 6 for Mac OS X, GraphPad Software, San Diego, CA, USA) and SigmaStat 3.5 software. Normality and equal variance assumptions were tested with the Kolmogorov-Smirnov and the Levene's median test, respectively. Differences between two groups were analyzed using the paired two-tailed Student's t-test, comparisons between experimental groups were analyzed by a one-way ANOVA for non-parametric variables with a Tukey's post-test for intergroup comparisons. All data is presented as mean ± SEM, and p < 0.05 was considered significant. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001, and n.s. p > 0.05. n denotes the number of mice used per dataset.

Conflicts of Interest:
The authors declare no conflict of interest.