Ca²⁺ dysregulation in cardiac stromal cells sustains fibro-adipose remodeling in Arrhythmogenic Cardiomyopathy and can be modulated by flecainide

Arrhythmogenic cardiomyopathy (ACM, also indicated as arrhythmogenic right ventricular cardiomyopathy, ARVC) is a rare cardiac disease of genetic origin characterized by a high incidence of sudden death in young people, due to malignant arrhythmic events, and by progressive degeneration of the ventricular myocardium, mainly the right ventricle, leading to heart failure. Mutations causing ACM are mostly found in genes encoding for desmosomal proteins, such as Plakophilin2 (PKP2), Plakoglobin (JUP), Desmoplakin (DSP), Desmoglein-2 (DSG2), Desmocollin-2 (DSC2) [1]. Mutated desmosomes are responsible for alterations in mechanical stability, electrical coupling between cells, and cellular signalling that likely lead to cardiomyocyte (CM) death and to the increase of adipogenic and fibrogenic genes expression for aberrant repair [2, 3]. The progressive loss of ventricular myocardium and the fibro-fatty infiltration contribute to determine clinical events such as ventricular arrhythmias and impaired ventricular systolic function. A desmosome-dyad axis dysfunction, mainly demonstrated in mouse models, is recognized as a mechanism impacting arrhythmic risk in ACM CM [4‐10].

Different studies revealed the important contribution of non-myocyte cells, such as Cardiac Mesenchymal Stromal Cells (C-MSC), to the ACM pathogenic mechanisms. C-MSC support the structural and functional integrity of the myocardium. They can be affected by ACM causative mutations based on the expression of desmosomal genes and they are the source of adipocytes and myofibroblasts in ACM patients’ hearts [11, 12].

MSC differentiation process can be influenced by spontaneous oscillations in intracellular calcium concentration ([Ca²⁺]_i) that are frequently detected in these cell types [13, 14]. The frequency of Ca²⁺ oscillations is different between undifferentiated MSC and MSC undergoing differentiation. Furthermore, the Ca²⁺ spiking pattern can vary depending on the differentiation level and type [13, 15, 16].

The mechanisms driving spontaneous Ca²⁺ oscillations were mainly investigated in adipose tissue- and bone marrow-derived MSC, in which they are sustained by both Ca²⁺ release from endoplasmic reticulum (ER) and Ca²⁺ inflow from extracellular space [17]. Endogenous Ca²⁺ is liberated from the ER through inositol-1, 4, 5-trisphosphate (IP3) receptors (IP3R), whereas ryanodine receptors (RYR) are not expressed in most of MSC. Depletion of the Ca²⁺ ER store results in the activation of “store-operated” Ca²⁺ channels (SOCC) on the plasma membrane (PM), which refills the ER with Ca²⁺ in a Sarco-Endoplasmic Reticulum Ca²⁺-ATPase (SERCA)-dependent manner [17]. SOCC depend on the physical interaction between STIM1-2, which serve as the ER Ca²⁺ sensor engaged upon ER Ca²⁺ depletion, and ORAI1-3, which form the Ca²⁺-permeable channel on the PM. In human MSC, SOCC impinge on STIM1 and ORAI1 and are required to maintain the spontaneous oscillations in [Ca²⁺]_i that regulate differentiation [18]. Additional pathways for extracellular Ca²⁺ entry are through L-type voltage-operated Ca²⁺ channels (VOCC) [19, 20]. Intriguingly, Store-operated Ca²⁺ entry (SOCE) is constitutively activated in several genetic muscle disorders, thereby enhancing resting [Ca²⁺]_i and increasing the ER Ca²⁺ load [21, 22]. The mechanisms that regulate spontaneous Ca²⁺ oscillations in C-MSC, their outcome on fibro-adipogenic differentiation and, therefore, their potential involvement in ACM, are currently unknown.

A molecular decoder of Ca²⁺ oscillations is the Ca²⁺/Calmodulin-dependent protein kinase II (CaMKII), a ubiquitous enzyme that mediates the Ca²⁺ effects on cellular targets involved in multiple cellular functions. CaMKII can act as a frequency decoder of Ca²⁺ oscillations in vitro.

This property is critically dependent on the effectiveness of individual Ca²⁺ spikes in stimulating autophosphorylation of the enzyme. These oscillations can be decoded according to their frequency, amplitude, duration and number [23].

The regulation of Ca²⁺-handling proteins by CaMKII [24, 25] leads to an increase in spontaneous Ca²⁺-release from the Sarcoplasmic Reticulum (SR) [26]. As consequence, in pathological scenarios, these events result in increased Ca²⁺ in the cytosol. Basically, CaMKII is a pro-arrhythmogenic protein in the heart, but its role in non-myocyte cells, e.g., C-MSC, is still unknown. Intriguingly, CaMKII also regulates fate decisions in multiple stem cell types [27, 28], including MSC [29], but the mechanistic link between intracellular Ca²⁺ oscillations, CaMKII activation, and stem cell differentiation is still unclear.

Based on these premises, we have hypothesized a correlation between intracellular Ca²⁺ oscillations and the fibro-adipogenic process in the ACM pathogenesis. We exploited a multidisciplinary approach to perform in vitro experiments on primary human C-MSC obtained from ACM patients.

Our findings show for the first time that the increased frequency of spontaneous Ca²⁺ oscillations, as well as the resulting CaMKII activation, are responsible for the ACM aberrant lipid/fibrotic accumulation and provide the first proof-of-concept that this signalling pathway can be targeted for therapeutic purposes.

Over the last few years, the idea that ACM is a disease not limited to the contractile component but instead a pathology that actively involves non-contractile cells become widely accepted [11, 12, 47‐49]. Despite this, a full knowledge of the molecular players involved and a full understanding of the molecular mechanisms leading to ACM pathogenesis are still lacking.

Of note, common ACM triggers (such as adrenergic stimulation and mechanical stretch) regulate Ca²⁺-dependent processes, and crucial ACM phenotypes are determined by Ca²⁺-related processes, ranging from modulation of myocyte excitability and fate (apoptosis) to maturation/differentiation of stromal cells [34, 46, 47, 50‐54].

Here we demonstrate for the first time that in ACM the Ca²⁺ handling machinery is dysregulated also in non-cardiomyocytes cells, i.e. in stromal cells, and that it contributes to the fibro-adipose degeneration process (Fig. 8). Our data, therefore, further expand the notion that Ca²⁺ dysregulation is a crucial determinant in the ACM phenotype [11, 12]. Intracellular Ca²⁺ signalling is instrumental to finely regulate proliferation and differentiation in human MSC isolated from multiple tissues [13, 17], but it is still unknown whether and how they modulate C-MSC fate.

The present study provides the first comprehensive characterization of the Ca²⁺ signalling machinery in human C-MSC and demonstrates that, in patients suffering from ACM, the Ca²⁺ toolkit and CaMKII undergo a complex dysregulation that ultimately results in stimulation of fibro-adipogenic differentiation. Indeed, our data show that spontaneous Ca²⁺ oscillations are present in cultured human C-MSC and occur in a significantly higher percentage and at enhanced amplitude and frequency in C-MSC isolated from ACM patients compared to what seen in HC. This, in turn, results in the hyperactivation of CaMKII, a pivotal effector of intracellular Ca²⁺ oscillations [23, 38].

Our results highlight the prevalence of CAMK2G and CAMK2D isoforms in human C-MSC, as it occurs in other cardiac cells [55]. Moreover, CaMKII activation was higher in ACM than in HC C-MSC. In agreement, an increase in Ca²⁺ oscillations can effectively boost CaMKII activation in ACM C-MSC, as shown by in vitro studies [23], cellular investigations [56], and in silico modelling [57]. The spontaneous Ca²⁺ oscillations finely tuning the differentiation outcome of bone marrow-derived MSC are shaped by rhythmic cycles of ER Ca²⁺ release through IP3R. Extracellular Ca²⁺ entry through SOCC is required to replenish the ER with Ca²⁺ in a SERCA-dependent manner, whilst VOGE is minimally contributory [14, 17]. Our data indicate that IP3R and SOCE are the major drivers of spontaneous Ca²⁺ oscillations also in ACM C-MSC. We provide the following evidences: i) the pharmacological blockade of SOCE with two selective ORAI1 inhibitors, i.e., PYR6 and BTP-2, reversibly prevented the spiking Ca²⁺ activity. Blocking VOGE through L-type CaV1.2 channels only weakly affected the Ca²⁺ transients. ii) spontaneous Ca²⁺ oscillations in ACM C-MSC were abrogated in most of the cells by blocking IP3R with the selective inhibitor, XeC, or by preventing the basal PLC-mediated production of IP3 with U73. Conversely, caffeine failed to mobilize intracellular Ca²⁺, since RYR2 was not expressed in ACM C-MSC. iii) spontaneous Ca²⁺ oscillations were suppressed by preventing Ca²⁺ reuptake into the ER with CPA, a selective inhibitor of SERCA activity, which further causes depletion of the ER Ca²⁺ store. Therefore, we can conclude that each Ca²⁺ transient is initiated by ER Ca²⁺ release through IP3R followed by ER Ca²⁺ refilling via SERCA. SOCE activation is required to maintain, and possibly trigger, the intracellular oscillations and downstream Ca²⁺-dependent signalling pathways.

The robust increase in spontaneous Ca²⁺ oscillations prompted us to hypothesize that the Ca²⁺ handling machinery was remodelled in ACM C-MSC. We found that SOCE was constitutively activated and resulted in a larger background Ca²⁺ entry in ACM as compared to HC C-MSC. In accordance, STIM1 and ORAI1 were both expressed in C-MSC but a significantly higher expression of STIM1 was found in ACM compared to the HC C-MSC. Although STIM2 is commonly regarded as the primary regulator of basal Ca²⁺ entry [58], constitutive activation of STIM1 enhances background Ca²⁺ influx and, thereby, increases resting [Ca²⁺]_i and/or ER Ca²⁺ levels in several genetic disorders. In agreement, both basal Ca²⁺ levels and the ER Ca²⁺ load are enhanced in ACM as compared to HC C-MSC [21, 22]. Similar to ACM C-MSC, adult CM isolated from transgenic mice overexpressing STIM1, displayed spontaneous Ca²⁺ oscillations and enhanced resting, i.e., diastolic [Ca²⁺]_i [59].

The increase in constitutive SOCE could per se lead to an increase in ER Ca²⁺ load in ACM C-MSC. In accordance, CPA-evoked intracellular Ca²⁺ release was also remarkably increased in ACM as compared to HC C-MSC. Furthermore, the pharmacological blockade of SOCE hampered both CPA- and ATP-induced endogenous Ca²⁺ mobilization from the ER, thereby confirming that the increase in basal Ca²⁺ permeability makes ACM C-MSC eager to release more Ca²⁺ during each IP3R-mediated Ca²⁺ transient. In addition, both SERCA and IP3R proteins are up-regulated, which, respectively, further increase ER Ca²⁺ loading and the ER Ca²⁺ releasing ability in patients-derived cells. Conversely, CaV1.2 is down-regulated and less active in ACM as compared to HC C-MSC, and, therefore, plays a minor role in spontaneous Ca²⁺ oscillations.

Arrhythmic events occurring in cardiomyopathies and channelopathies may be mostly associated to Ca²⁺ dysregulation in CM [60]. Research on human iPSC-CM revealed that Ca²⁺ imbalance seems mainly attributable to plasma membrane Ca²⁺ channels and transporters alterations. Changes in the action potential duration observed in DSG2 iPSC-CM were demonstrated to be sustained by altered Na⁺/Ca²⁺ exchanger (NCX) currents [61]. A PKP2-mutated ACM model of hiPSC-CM showed slower cytoplasmic [Ca²⁺]_i relaxation with a prolonged relaxation time constant [62]. Electrophysiological studies in obscurin-mutated hiPSC-CM, highlighted that L-type Ca²⁺ current (ICaL) is increased in ACM- hiPSC-CM and this imbalanced ICaL may be linked to arrhythmic state [63]. In the CM context, flecainide administration represents an efficient strategy, based on its ability to suppress the spontaneous SR Ca²⁺ release in mice and humans, thus exhibiting an antiarrhythmic effect [43, 46]. For instance, flecainide prevents arrhythmias in a mouse model of Catecholaminergic Polymorphic Ventricular Tachycardia (CPVT) by inhibiting cardiac RYR2–mediated Ca²⁺ release [43, 64, 65] and is now considered a valid therapeutic approach in CPVT patients non-responsive to betablockade alone. The efficacy of flecainide has also been proven in ACM models. Data from a CM-specific tamoxifen-activated Pkp2 knock-out mouse (Pkp2-cKO) model demonstrate that flecainide treatment abolishes adrenergic arrhythmias [46]. Moreover, it is able to normalise spontaneous Ca²⁺ transients in hiPSC-CM with missense mutation in the DSC2 gene [66]. Despite the historic hesitation in using class IC anti-arrhythmic in patients with structural heart disease [67], different recent evidence supported safety in the use of Flecainide in ACM [42, 68‐70]. Currently, a pilot double-blind cross-over clinical trial (ClinicalTrial.Gov NCT03685149) is evaluating the effectiveness of flecainide for arrhythmic burden reduction in ACM patients. Our data demonstrated that flecainide reduced spontaneous Ca²⁺ oscillations in ACM C-MSC. Flecainide is known as a NaV1.5 and RYR2 blocker [44, 45], which are absent in C-MSC. Here we show for the first time that flecainide inhibits constitutive SOCE, thereby preventing ER Ca²⁺ refilling and interfering with rhythmical ER Ca²⁺ release through IP3R. Therefore, besides interfering with the Ca²⁺ handling machinery in CM, flecainide exerts an etiological beneficial effect on the stromal component of the ACM heart. The main pharmacological therapy currently recommended in the clinical practice is the use of β-blockers and eventually amiodarone [71]. These drugs have either direct or indirect effect on Ca²⁺ content and channel activity. However, to date, specific studies on their effect on ACM C-MSC are not available.

The ACM heart typically undergoes progressive replacement of the ventricular wall by fibrotic and/or adipose tissue resulting in impaired contractile function and providing a nonconductive substrate, a source for re-entrant arrhythmias [72]. Previous studies have highlighted that C-MSC contribute both to adipogenic remodelling and myocardial fibrosis in ACM hearts [11, 12]. Specifically, C-MSC exposed to a pathological microenvironment (i.e. determined by an excess of TGFβ and oxidized-LDL) contribute to induce collagen and fat accumulation [12, 48]. The induction of differentiation programs in MSC involves an alteration of spontaneous Ca²⁺ oscillations [13, 14] whose frequency can either increase or diminish depending on the differentiation outcome. Herein, we observed a significant increase in the oscillation frequency in C-MSC undergoing adipogenic differentiation that was further enhanced in patients-derived cells. Lipid accumulation and collagen production in ACM C-MSC is reduced by the pharmacological blockade of CaMKII with KN-93, which suggests that the oscillatory Ca²⁺ signal in these cells increases during adipogenesis to support CaMKII activation. Of note, CaMKII has been shown to promote adipogenesis in porcine BM-MSC by stimulating the PI3K/Akt-Fox01 signalling pathway to increase the expression of two crucial adipogenic transcription factors [29], such as peroxisome proliferator activated receptor γ (PPARγ) and CCAAT/ enhancer binding protein α (C/EBPα) [73]. In addition, the inhibition of CaMKII is able to reduce cardiac fibroblast proliferation, the secretion of TGF-β1 and TNF-α, and can revert the upregulation of MMPs and collagen, confirming its involvement in extracellular matrix regulation [74]. Of note, CaMKII activity interferes the angiotensin II-mediated differentiation of cardiac fibroblasts into myofibroblasts by inducing collagen deposition thus contributing to cardiac remodelling [75].

Remarkably, we obtained substantial results by either using compounds acting on known SOCE targets or flecainide that regulated SOCE activity in our cells.

We finally evaluated the possible role of PKP2 in the remodelling of the Ca²⁺ toolkit observed in ACM C-MSC [76], based on the recent reports highlighting a link between PKP2 and the control of the Ca²⁺ cycling machinery in CM [46]. Specifically, by using the Pkp2-cKO mouse model, Cerrone and co-workers found that several components of Ca²⁺ handling machinery were down-regulated in CM, causing a reduction in SR Ca²⁺ leakage, thereby leading to an increase in SR Ca²⁺ content [46]. Similar to ACM C-MSC, but through a different mechanism (e.g., RYR2 vs. IP3R), the enhanced SR Ca²⁺ load results in an increase in amplitude and frequency of spontaneous Ca²⁺ sparks in mouse Pkp2-cKO CM [46]. The relationship between PKP2 and the Ca²⁺ machinery in CM was also pointed out by a bioinformatics approach, based on transcriptomic information from human hearts, that confirmed that at lower expression of PKP2 corresponds to a lower expression of the genes encoding for RYR2, Ankyrin-B and CAV1.2 [5]. Interestingly, our data, obtained by mimicking PKP2 haploinsufficiency in C-MSC, demonstrated that the alterations in the intracellular Ca²⁺ cycling in ACM C-MSC and CaMKII activation can be linked to PKP2 levels in the cells. Indeed, both spontaneous Ca²⁺ activity and CaMKII activation were remarkably enhanced in HC C-MSC infected with shRNA PKP2. Moreover, as observed in ACM C-MSC, genetic downregulation of PKP2 led to an increase in the expression levels of STIM1 and SERCA proteins, and in the reduction of CAV1.2 protein. However, our experiments included C-MSC collected from a pool of ACM patients, not necessarily bearing PKP2 mutations, which suggests that increased Ca²⁺ load and Ca²⁺ homeostasis dysregulation could be a common mechanism in the setting of different genetic forms of the disease. As an example, data on DSG2-mutated CM also showed Ca²⁺ imbalance and increased SR load [61, 77].

In conclusion, our results provide evidence that Ca²⁺ imbalance in ACM extends beyond the CM compartment and involves stromal cardiac cells and the fibro-adipose degeneration process (Fig. 8). Moreover, we provided evidence about a novel mode of action of flecainide in ACM, which is effective in attenuating C-MSC fibro-fatty differentiation through the newly described molecular target SOCE. Our results add further rationale about the use of flecainide as pharmacological treatment of ACM and await efficacy confirmation on cardiac remodelling in a clinical trial in ACM patients.