Disruption of Ca²⁺_i Homeostasis and Connexin 43 Hemichannel Function in the Right Ventricle Precedes Overt Arrhythmogenic Cardiomyopathy in Plakophilin-2–Deficient Mice

Plakophilin-2 (PKP2) is classically defined as a protein of the desmosome, an intercellular adhesion structure. Recent studies demonstrated that in addition to cell-cell adhesion, PKP2 translates information initiated at the site of cell-cell contact into intracellular signals that maintain structural and electrical homeostasis.^1–4 Mutations in PKP2 associate with most cases of gene-positive arrhythmogenic right ventricular cardiomyopathy (ARVC), a pleiotropic disease that can manifest as mainly electrical, structural, or both depending on stage progression.⁵ As such, while ARVC is best recognized as a cardiomyopathy of right ventricular predominance, catecholaminergic sudden cardiac arrest is common during the subclinical, or “concealed”, phase of the disease.^5–7 The molecular/cellular mechanisms responsible for these arrhythmias remain unclear.

To study the role of cardiomyocyte PKP2 expression in cardiac function we used a cardiomyocyte-specific, tamoxifen (TAM)-activated, PKP2 knockout murine line (PKP2 fl/fl, αMyHC-Cre-ER(T2), dubbed PKP2cKO). We previously reported that adult PKP2cKO mice present, in a compressed time line, aspects of the history of human ARVC.³ Following TAM injection these hearts progress from a normal state, to an arrhythmogenic cardiomyopathy of right ventricular (RV) predominance and eventually biventricular dilated cardiomyopathy and end-stage failure. The time course of this process is reproducible, allowing us to study the molecular environment at a given phenotype stage.

The earliest functional event detectable in PKP2cKO mice is an increase in RV area (measured by echocardiography 14 days post-TAM injection) and a high propensity to Isoproterenol-induced lethal ventricular arrhythmias at day 16 post-TAM. These events occur in the setting of preserved left ventricular ejection fraction and in the absence of increased collagen abundance or other changes in ventricular wall morphology (ie, during a concealed stage preceding overt arrhythmogenic cardiomyopathy^3,5,7). Thus, for this study we focused on mice 14 days post-TAM, so as to capture early molecular/cellular events that can act as nascent arrhythmia substrates. Knowing that PKP2 deficiency first leads to a structural phenotype of RV predominance³ we separately studied myocytes from right versus left free ventricular walls. We focused on possible changes in intracellular Ca²⁺ cycling, given previous evidence showing PKP2-mediated transcriptional regulation of intracellular Ca²⁺ homeostasis in the cardiomyopathic stage.³ Overall, we found that loss of PKP2 creates not only an RV-predominant cardiomyopathy but also an RV-predominant arrhythmogenic substrate that precedes the cardiomyopathy and that is, at least in part, dependent on Connexin 43 (Cx43) expression/function. We speculate that disrupted [Ca²⁺]_i homeostasis in RV myocytes can be 1 of the triggers for gross structural changes in the RV at a later stage.

We used a commercially available C57BL/6 CX43^flox mice line (Jackson Laboratory B6.129S7-Gja1^tm1.1Dlg/J, stock number 008039; loxP sites flank exon 2 of GJA1; first reported by Liao et al⁸ and extensively studied in various systems; see, eg, Hulsmans et al⁹), crossed with PKP2flox/flox/αMHC-Cre-ER(T2) to obtain (CX43flox/flox)/(PKP2flox/flox)/Cre+, and CX43flox/wt)/(PKP2flox/flox)/Cre+ mice. All procedures conformed to the Guide for Care and Use of Laboratory Animals of the National Institutes of Health and were approved by the NYU-IACUC committee. For cardiac-specific PKP2 and/or GJA1 deletion, mice were intraperitoneally injected 4 consecutive days with TAM (0.1 mg/g body weight). Cre-negative flox-positive, TAM-injected littermates were used as controls. Separate controls with TAM-injected, Cre-positive PKP2flox/wt showed no effects on life expectancy or left ventricular ejection fraction in a prolonged time window (Figure I in the online-only Data Supplement), consistent with previous observations.¹⁰ Both genders were included and all animals were between 3 and 6 months-old. Further details in online-only Data Supplement.

Procedures used for RNAseq, cardiomyocyte dissociation, Ca²⁺ imaging, Western blots, single-molecule localization microscopy, serial block-face scanning electron microscopy, [³H]Ryanodine binding assays, mass spectrometry, echocardiography, patch clamp, and voltage-sensitive optical mapping followed those previously published and are detailed in the online-only Data Supplement.

For evaluation of membrane permeability in whole hearts, 2 fluorescent dyes, Lucifer Yellow (LY) and Rhodamine Dextran, were used to quantify the entry of a small molecular weight dye into Langendorff-perfused hearts. Hearts were washed, fixed, epicardium was manually removed, free walls separated and placed epicardial face down for observation by fluorescence confocal microscopy. Imaging was performed at several randomly chosen sites in each tissue sample. Details in the online-only Data Supplement.

Previously, we showed that PKP2cKO mice develop an arrhythmogenic cardiomyopathy of RV predominance 21 days post-TAM injection.³ To understand its cellular/molecular substrates, here we studied these mice at a time preceding the overt phenotype (ie, 14 days post-TAM) and looked for side-specific differences (RV vs LV). Given our previous data,³ and given that the phenotype once apparent is both structural and electrical, we focused on a cellular function at the center of both excitability and contraction, namely, regulation of [Ca²⁺]_i.

At 14 days post-TAM mice did not show cardiac fibrosis, cellular dimensions were similar to those in control, and left ventricular ejection fraction was within normal limits. Transcriptome differences were minor and levels of PKP2 were drastically down in both ventricles. Yet, we detected important differences in [Ca²⁺]_i handling in PKP2cKO RV myocytes, while PKP2cKO LV myocytes remained similar to control. These differences were sensitive to Cx43 reduced expression or Cx43 hemichannel block, and coincided with a Cx43-dependent increase in membrane permeability of the RV, an RV-selective increase in [Ca²⁺] sensitivity of RyR2, and an RV-selective phosphorylation of RyR2-Thr2809, an amino acid within the phosphorylation hotspot of RyR2 long known to modulate channel gating. The eagerness of RyR2 to trigger Ca²⁺ sparks was tempered by exposing cells to PKC inhibitors. Based on previous knowledge about: a) the relation between [Ca²⁺]_i alterations and arrhythmia susceptibility, b) the importance of regional heterogeneity on arrhythmia generation, and c) the relation between [Ca²⁺]_i and cardiac remodeling,^34,35 we propose that [Ca²⁺]_i overload and RyR2 dysfunction play a central role in the generation of the electrical and RV-predominant structural phenotype in PKP2-deficient mice.

The Cx43-dependent increase in membrane permeability reported here can explain the excess [Ca²⁺]_i. We speculate that PKP2 deficiency causes an excess of “orphan” Cx43 membrane hemichannels that can act as conduits for Ca²⁺ entry. Gap junction plaque formation is known to depend on proper intercellular adhesion.²⁷ It is also known that Cx43 hemichannels reside in the perimeter of the gap junctions (the perinexus)³⁶ and are capable, when undocked and in the membrane, to flicker with very low probability.^37,38 Further supporting the relation between PKP2 expression and Cx43-dependent permeable conduits is a recent study showing that PKP2 deficiency can increase cell membrane permeability to ATP, an event prevented by silencing Cx43 expression.³⁹ We postulate that loss of PKP2 expression, perhaps via weakened intercellular adhesion consequent to loss of desmosomal integrity, causes an increased population of orphan, flickering, Cx43 hemichannels that allow for excess entry of Ca²⁺ into the cell. It is worth noting that studies in cell culture systems and in human tissue from ARVC-affected patients have consistently reported that loss of expression or mutations in PKP2 can alter the structure and function of gap junctions.^40–42 We did not detect changes in Cx43 abundance or in gap junction structure, suggesting that functional dysregulation of Cx43 hemichannel activity, coincident with nanostructural changes in the intercalated disc, are among the earliest events following loss of PKP2 expression.

We found increased Ca²⁺ sparks frequency in PKP2cKO-RV myocytes. In addition to increased SR Ca²⁺ load and higher diastolic [Ca²⁺]_i, this result may associate with increased sensitivity of RyR2 to Ca²⁺, as well as a change in the phosphorylation state of RyR2-T2809. This residue is located near serine residues well studied as PKA and CAMKII substrates that, when phosphorylated, modify RyR2 gating behavior.²⁰ We do not know the kinase phosphorylating T2809; however, sequence analysis and data showing that PKC inhibitors reduced Ca²⁺ sparks frequency at comparable levels of SR load support the hypothesis of PKC involvement. Interestingly, previous work has shown that PKP2 is necessary to recruit and retain PKC at the desmosome, and that loss of desmosomal integrity releases PKC, freeing it to phosphorylate unintended targets.² Consistent with this notion, our single-molecule localization microscopy data indicated reduced phospho-PKC cluster size and density at cell end, and increased phospho-PKC distance from N-Cadherin, a marker of localization of the adhesion complex. Yet, whether phosphorylation of RyR2 in PKP2cKO-RV myocytes is a determinant of RyR2 calcium release remains to be defined.

We observed that SERCA2a protein (but not transcript) levels were decreased, primarily in RV. The latter may play a role in the prolongation of the time constant of recovery in the Ca²⁺ transients. Moreover, we saw no difference in NCX function at fixed [Ca²⁺], though dynamic changes may occur. It is known that Cx43 hemichannels allow for entry not only of Ca²⁺ but also Na⁺. Increased [Na⁺]_i would retard the forward mode activity of the Na⁺-Ca²⁺ exchanger, further favoring Ca²⁺ accumulation in the intracellular space. Overall, the activity of NCX to extrude Ca²⁺ would be outmatched by persistent intake of external Ca²⁺ (and perhaps, Na⁺) through Cx43 hemichannels, reaching a steady-state at higher levels of [Ca²⁺]_i; this imbalance (or new balance) would be corrected by inhibition or ablation of Cx43. In terms of the relation between RyR2 abundance and function, reduced RyR2 levels may be compensated by the organization of RyR2 units into super-clusters that facilitate calcium release.^11–13 The latter change was not asymmetric, indicating that it may be a substrate that facilitates but is not sufficient to alter Ca²⁺ release.

Previous echocardiographic analysis at 14 days post-TAM showed increased RV area in PKP2cKO hearts, an early sign of changes in RV contractility.³ Partial ablation of Cx43 mitigated this event, likely through normalization of [Ca²⁺]_i; whether contractile dysfunction is consequent to Ca²⁺ dysregulation remains to be determined, though an association between Ca²⁺ homeostasis and contractility is well documented.⁴³ Alternatively, we speculate that the increased RV area is an early manifestation of impaired mechanical forces at the cellular level, combined with the differences in force distribution in various areas of the heart. At the cell level, Broussard et al have shown that impairing the association of desmosomes with intermediate filaments leads to decreased cell tension and stiffness.⁴⁴ One can surmise that at the tissue level, these changes would reflect on a weakened ventricular wall, leading to dilation. While the change in cell stiffness may (or may not) be generalized, the forces impacting the ventricular wall are not evenly distributed throughout the heart; recent studies show, for example, that the thin-walled RV (and in particular the RV outflow tract) is exposed to a particularly high wall shear stress.⁴⁵ Along these lines, echocardiographic analysis has documented impaired contractility and stiffness of the RV myocardium prior to LV dysfunction in subclinical ARVC patients that present with no evidence of fibrosis by late enhancement magnetic resonance imaging.⁴⁶ The latter observation, all major differences notwithstanding, is consistent with the increased RV area in the 14 day post-TAM mice, which precedes overt fibrosis. At the cellular level, these uneven physical forces may cause the RV cells to be first to separate from each other, allowing increased entry of Ca²⁺ that in turn, may further impair the contractile function of the cells. Yet, it is important to recognize the role of Wnt-signalling-dependent embryonic programming as potential substrate for asymmetric differences, particularly in case of desmosomal deficiency.⁴⁷

Altogether, we propose that loss of PKP2 expression occurs in both ventricles but first affects RV desmosomal integrity. Loss of desmosomal integrity triggers 2 relatively unrelated events: a) loss of stability of neighboring gap junctions, with consequent increase in orphan Cx43 hemichannels at the perinexus, and b) dislodgement of a kinase (likely PKC) from the intercalated disc, which translocates to phosphorylate an off-target substrate (RyR2-T2809), changing RyR2 gating properties. These 2 arms converge to facilitate development of potentially lethal arrhythmias and, as time progresses, alter the transcriptional program of the cells and myocyte integrity, leading to arrhythmogenic cardiomyopathy.

Of note, we acknowledge that the mechanisms of arrhythmias may include additional events not yet detected by our experiments. Though we see cellular events that can be arrhythmogenic, and we do see arrhythmias, we do not argue that the arrhythmias are only and exclusively resulting from Ca²⁺ alterations.

We focused on PKP2 because of its importance in cardiac biology. Mutations in PKP2 associate with a pleiotropic disease (ARVC) that presents an electrical phenotype sometimes preceding and sometimes concurrent with a cardiomyopathy. It is worth noting that sudden death is the first disease manifestation in 11% of ARVC probands,⁵ and that many ARVC cases present mostly an electrical phenotype while the structural disease is concealed^5–7 or not present.⁴⁸ This coincides with the observation in our mice, indicating alterations in Ca²⁺ homeostasis well-known as potentially arrhythmogenic³ as first detectable manifestation, prior to collagen accumulation. It is tempting to speculate that elevation of [Ca²⁺]_i can be a trigger for subsequent structural disease and the modified transcriptional program of the heart. Under that scenario, the elevated [Ca²⁺]_i would exert first an arrhythmogenic effect, followed by changes that take more time to manifest, namely, the transcriptional and structural events. Interestingly, despite the elevated [Ca²⁺]_i we did not detect signs of hypertrophy. This and other observations have led us to postulate the hypothesis (yet to be tested) that loss of PKP2 blunts activation of the transcriptional program necessary for compensatory hypertrophy.

We recognize that weaknesses remain with regard to the mechanisms proposed. We have not studied quantitatively the relation between the Cx43-dependent permeability observed in our tissue preparations (demonstrated by the passage of LY) and the [Ca²⁺]_i as it is measured in the single cells. It is important to note though that experimental conditions differ. In particular, the LY experiment is performed in the absence of external Ca²⁺ added. This manipulation allows us to see the increased Cx43-dependent membrane permeability within the time frame of our recordings. But the natural time course of the increased [Ca²⁺]_i as it would occur in the heart, remains to be determined. Also, related to the point above, we lack a quantitative description of the magnitude of the Ca²⁺ influx taking into account the limited permeability and the presumably low number of Cx43 hemichannels. In this regard, it is worth noting the study of John et al.⁴⁹ These authors calculated that in a cardiomyocyte, 10 open Cx43 hemichannels producing a 100 pA current at −80 mV would increase Na⁺ influx by 75%. This would still represent a small fraction of a total of 2.6×10⁶ junctional connexins estimated to be present in the myocyte, and yet the increase in intracellular Na⁺ may favor an increase in [Ca²⁺]_i. The amount of open hemichannels at 1 particular snapshot in time would be minimal, not causing a major impact to the cardiomyocyte membrane potential. Yet, the long-term effect of this Cx43 flickering would be the accumulation of Ca²⁺ in the intracellular space. Overall, our data strongly support the notion that a Cx43-dependent membrane conduit is necessary for the increased [Ca²⁺]_i reported; nonetheless, how the various Ca²⁺ and Na⁺ fluxes balance, through the various membrane compartments, to yield high [Ca²⁺]_i levels remains to be determined.

While our observations are consistent with those in humans with ARVC, any extrapolation needs to be taken with caution, as it is clear that animal models do not reproduce the conditions of a patient with the disease. This is in fact true for all animal or cellular models of ARVC that have been published, and an obvious consequence of the limitations inherent to the study of a human disease. The significance of our studies is not in having reproduced conditions that lead to human ARVC, but in having studied a molecule that, when mutated, causes the disease. Whether the molecular steps described here occur in human hearts with PKP2 mutations remains undefined. Yet, it is worth mentioning that agents that block release of Ca²⁺ through RyR2 channels, such as flecainide, may be a possible combination therapy for ARVC patients,⁵⁰ and that recent reports indicate that patients with first diagnosis of CPVT have been later identified as carriers of PKP2 mutations.⁴⁸ The relation between ARVC and Ca²⁺ mishandling is emphasized by the fact that mutations in phospholamban, a known contributor to Ca²⁺ homeostasis, can cause ARVC.^5,7 We show a functional intersection of PKP2 and Ca²⁺ homeostasis and propose a model as to the steps by which one event (loss of intercellular adhesion at the intercalated disc) affects another (Ca²⁺ release), even though they are functions structurally separate in the myocyte space.

The authors acknowledge the excellent support of Joseph Sall and Chris Petzold for EM sample preparation and image acquisition.