Phospholamban ablation rescues the enhanced propensity to arrhythmias of mice with CaMKII‐constitutive phosphorylation of RyR2 at site S2814

Abstract

Key points

• Mice with Ca²⁺–calmodulin‐dependent protein kinase (CaMKII) constitutive pseudo‐phosphorylation of the ryanodine receptor RyR2 at Ser2814 (S2814D^+/+ mice) exhibit a higher open probability of RyR2, higher sarcoplasmic reticulum (SR) Ca²⁺ leak in diastole and increased propensity to arrhythmias under stress conditions.

• We generated phospholamban (PLN)‐deficient S2814D^+/+ knock‐in mice by crossing two colonies, S2814D^+/+ and PLNKO mice, to test the hypothesis that PLN ablation can prevent the propensity to arrhythmias of S2814D^+/+ mice.

• PLN ablation partially rescues the altered intracellular Ca²⁺ dynamics of S2814D^+/+ hearts and myocytes, but enhances SR Ca²⁺ sparks and leak on confocal microscopy.

• PLN ablation diminishes ventricular arrhythmias promoted by CaMKII phosphorylation of S2814 on RyR2.

• PLN ablation aborts the arrhythmogenic SR Ca²⁺ waves of S2814D^+/+ and transforms them into non‐propagating events.

• A mathematical human myocyte model replicates these results and predicts the increase in SR Ca²⁺ uptake required to prevent the arrhythmias induced by a CaMKII‐dependent leaky RyR2.

Abstract

Mice with constitutive pseudo‐phosphorylation at Ser2814‐RyR2 (S2814D^+/+) have increased propensity to arrhythmias under β‐adrenergic stress conditions. Although abnormal Ca²⁺ release from the sarcoplasmic reticulum (SR) has been linked to arrhythmogenesis, the role played by SR Ca²⁺ uptake remains controversial. We tested the hypothesis that an increase in SR Ca²⁺ uptake is able to rescue the increased arrhythmia propensity of S2814D^+/+ mice. We generated phospholamban (PLN)‐deficient/S2814D^+/+ knock‐in mice by crossing two colonies, S2814D^+/+ and PLNKO mice (SD^+/+/KO). SD^+/+/KO myocytes exhibited both increased SR Ca²⁺ uptake seen in PLN knock‐out (PLNKO) myocytes and diminished SR Ca²⁺ load (relative to PLNKO), a characteristic of S2814D^+/+ myocytes. Ventricular arrhythmias evoked by catecholaminergic challenge (caffeine/adrenaline) in S2814D^+/+ mice in vivo or programmed electric stimulation and high extracellular Ca²⁺ in S2814D⁺/⁻ hearts ex vivo were significantly diminished by PLN ablation. At the myocyte level, PLN ablation converted the arrhythmogenic Ca²⁺ waves evoked by high extracellular Ca²⁺ provocation in S2814D^+/+ mice into non‐propagated Ca²⁺ mini‐waves on confocal microscopy. Myocyte Ca²⁺ waves, typical of S2814D^+/+ mice, could be evoked in SD^+/+/KO cells by partially inhibiting SERCA2a. A mathematical human myocyte model replicated these results and allowed for predicting the increase in SR Ca²⁺ uptake required to prevent the arrhythmias induced by a Ca²⁺–calmodulin‐dependent protein kinase (CaMKII)‐dependent leaky RyR2. Our results demonstrate that increasing SR Ca²⁺ uptake by PLN ablation can prevent the arrhythmic events triggered by SR Ca²⁺ leak due to CaMKII‐dependent phosphorylation of the RyR2‐S2814 site and underscore the benefits of increasing SERCA2a activity on SR Ca²⁺‐triggered arrhythmias.

Key points

• Mice with Ca²⁺–calmodulin‐dependent protein kinase (CaMKII) constitutive pseudo‐phosphorylation of the ryanodine receptor RyR2 at Ser2814 (S2814D^+/+ mice) exhibit a higher open probability of RyR2, higher sarcoplasmic reticulum (SR) Ca²⁺ leak in diastole and increased propensity to arrhythmias under stress conditions.

• We generated phospholamban (PLN)‐deficient S2814D^+/+ knock‐in mice by crossing two colonies, S2814D^+/+ and PLNKO mice, to test the hypothesis that PLN ablation can prevent the propensity to arrhythmias of S2814D^+/+ mice.

• PLN ablation partially rescues the altered intracellular Ca²⁺ dynamics of S2814D^+/+ hearts and myocytes, but enhances SR Ca²⁺ sparks and leak on confocal microscopy.

• PLN ablation diminishes ventricular arrhythmias promoted by CaMKII phosphorylation of S2814 on RyR2.

• PLN ablation aborts the arrhythmogenic SR Ca²⁺ waves of S2814D^+/+ and transforms them into non‐propagating events.

• A mathematical human myocyte model replicates these results and predicts the increase in SR Ca²⁺ uptake required to prevent the arrhythmias induced by a CaMKII‐dependent leaky RyR2.

Abbreviations

Adr

adrenaline

AE

anion exchanger

AP

action potential

APD50

AP duration at 50% repolarization

Caff

caffeine

CaMKII

Ca²⁺–calmodulin‐dependent protein kinase

CICR

Ca²⁺‐induced Ca²⁺ release

CPA

cyclopiazonic acid

CRU

Ca²⁺ release unit

CYTO

cytoplasm

DAD

delayed afterdepolarization

DC

dyadic cleft

EAD

early afterdepolarization

ECG

electrocardiogram

HF

heart failure

HR

heart rate

Ileak

SR diastolic Ca²⁺ flux

INa

Na⁺ current

ICaL

Ca²⁺ current through L‐type Ca²⁺ channels

INCXdir

Na⁺/Ca²⁺ exchanger direct mode current

INCXrev

Na⁺/Ca²⁺ exchanger reverse mode current

Iup

SR Ca²⁺ uptake flux

Irel

SR Ca²⁺ release flux

Iti

transient inward current

LCC

L‐type Ca²⁺ channel

LVFS

left ventricular fractional shortening

NCX

Na⁺/Ca²⁺ exchanger

PLN

phospholamban

PLNKO

PLN knock‐out

RyR2

cardiac ryanodine receptor

SAP

spontaneous action potential

S2814D^+/+

knock‐in mice with constitutive pseudo‐phosphorylation of RyR2 at Ser2814

S2814D^+/−

heterozygous S2814D mice

SD^+/+/KO

mice resultant from crossbreeding PLNKO and RyR2‐S2814D mice

SD^+/−/KO

mice resultant from crossbreeding PLNKO and RyR2‐S2814D^+/− mice

SD^+/+/KOsim and S2814D^+/+sim

simulated conditions of SD^+/+/KO and S2814D^+/+ mice

SERCA2a

SR Ca²⁺‐ATPase

SR

sarcoplasmic reticulum

TP

ten Tusscher–Panfilov

VT

ventricular tachycardia

WT

wild‐type

xfer

conductance parameter from dyadic cleft to cytosol

Introduction

Mechanical dysfunction and arrhythmias are hallmark features of heart failure (HF), a leading cause of morbidity and mortality worldwide (Cleland et al. 2002; Mozaffarian et al. 2007), and it is now well established that aberrant Ca²⁺ handling is a main cause of these two characteristic HF disarrays (Hasenfuss & Pieske, 2002; Pogwizd & Bers, 2004; Luo & Anderson, 2013). Indeed, a large fraction of ventricular arrhythmias in HF are thought to be initiated at the cellular level by focal triggered mechanisms such as abnormal spontaneous Ca²⁺ discharges from the sarcoplasmic reticulum (SR), which propagate as regenerative Ca²⁺ waves through cardiac cells (Laurita & Rosenbaum, 2008). Spontaneous Ca²⁺ waves are arrhythmogenic as a consequence of activating inward membrane currents, mainly the electrogenic Na⁺/Ca²⁺ exchanger (NCX) working in the forward mode (Pogwizd & Bers, 2004; Laurita & Rosenbaum, 2008; Luo & Anderson, 2013).

An enhanced SR Ca²⁺ leak occurs under conditions in which SR Ca²⁺ load exceeds a threshold that is largely determined by the particular state of the ryanodine receptor RyR2. For instance, RyR2 point mutations render the channel more prone to spontaneous SR Ca²⁺ release during adrenergic stimulation. Patients with this inherited anomaly exhibit catecholaminergic polymorphic ventricular tachycardia, a known cause of sudden cardiac death (Liu & Priori, 2008). Phosphorylation of RyR2 by Ca²⁺–calmodulin‐dependent protein kinase II (CaMKII) at the S2814 site has also been associated with an increase in SR Ca²⁺ leak and arrhythmogenesis in cardiac pathologies of different aetiology (Ai et al. 2005; Said et al. 2008, 2011; Chelu et al. 2009; Gonano et al. 2011). CaMKII is up‐regulated and more active in HF and different type of experimental evidence indicate that RyR2‐S2814 phosphorylation is a main underlying mechanism of arrhythmias in this disease (Ai et al. 2005). These results suggested a crucial role of RyR2 altered activity on triggered arrhythmias. In contrast, the effect of increasing SR Ca²⁺ uptake on cardiac triggered events is not clear and there is concern about whether the increase in SR Ca²⁺ uptake, which has been shown to be a useful therapy to revert the depressed cardiac contractility in human and experimental heart failure (Hajjar et al. 2008), is protective against Ca²⁺ triggered arrhythmias or exacerbates them.

This concern has experimental support since contradictory results have been obtained by either decreasing or increasing SR Ca²⁺ uptake (Lukyanenko et al. 1999; Davia et al. 2001; del Monte et al. 2004; Landgraf et al. 2004; Prunier et al. 2008; Stokke et al. 2010; Bai et al. 2013; Liu et al. 2015). The clue to explaining these inconsistent findings may rest, at least in part, in the opposite effects inherent to the augmented SR Ca²⁺ uptake itself on cytosolic Ca²⁺, i.e. increasing the rate of SR Ca²⁺ uptake would reduce cytosolic Ca²⁺ overload and the risk of cardiac arrhythmias, but would necessarily increase SR Ca²⁺ content, favouring RyR2 Ca²⁺ sensitization, improving diastolic SR Ca²⁺ leak and the risk of Ca²⁺ waves. This situation might be exacerbated if the increase in SR Ca²⁺ uptake coexists with an increase in the open probability of RyR2, as that produced by CaMKII‐dependent phosphorylation of the S2814 site (Ai et al. 2005), and may contribute to favouring a futile cycle of increased SR Ca²⁺ uptake and leak with an additional metabolic cost, due to SERCA2a ATP consumption. Thus, the beneficial effects of increasing SR Ca²⁺ uptake may turn out to be deleterious under conditions in which the balance between SR Ca²⁺ uptake and leak is lost.

Previous studies demonstrated that ventricular myocytes from S2814D^+/+ mice, in which the Ser2814 CaMKII site on RyR2 is constitutively pseudo‐phosphorylated, are prone to arrhythmias (van Oort et al. 2010). In the present work, we crossbred phospholamban knock‐out (PLNKO) mice (PLN^−/−) (Luo et al. 1994) with RyR2‐S2814D^+/+ knock‐in mice to generate double‐mutant mice, PLN^−/−/RyR2‐S2814D^+/+ (SD^+/+/KO), to test the idea that the increase in SR Ca²⁺ uptake may rescue S2814D^+/+ mice from their typical arrhythmogenic susceptibility. With the aid of a mathematical model we further tested the necessary change in SR Ca²⁺ uptake required to prevent the arrhythmogenic effect of a CaMKII‐dependent leaky RyR2.

Methods

Ethical approval

All experiments involving mice were performed as per institutional guidelines and appropriate laws, and were approved by the Faculty of Medicine, University of La Plata Institutional Animal Care and Use Committee (CICUAL no. P03‐01‐15). The authors have read and understood the policies and regulations of The Journal of Physiology as outlined by Drummond (2009) and ensured that all experiments complied with these regulations.

Animals

Experiments were performed in 3‐ to 4‐month‐old male mutant mice with phospholamban ablation (PLNKO mice) (Luo et al. 1994), RyR2‐S2814D^+/+ knock‐in mice (van Oort et al. 2010), and mice resulting from crossbreeding PLNKO and RyR2‐S2814D^+/+ mice (SD^+/+/KO mice). Some experiments in perfused hearts (see Results) were performed in mice resulting from crossbreeding PLNKO and heterozygous RyR2‐S2814D⁺/⁻ mice. These mice were called SD^+/−/KO mice. C57BL/6 mice were used as controls (wild‐type (WT) mice). Mice genotype was confirmed by PCR analysis using mouse tail DNA and specific primers for each mutation.

Histology

A transverse section of the heart was fixed in 10% buffered formalin for 48 h. After paraffin embedding and sectioning, 5 μm sections were stained with Masson's trichrome and picrosirius red techniques. Slides were observed with an Olympus BX53 microscope (Japan) and transmitted to the computer using a digital camera (Olympus DP‐73, Japan). Images were captured using the automatic Multiple Image Alignment process of the cellSens Dimension v1.7 image analysis software (Olympus). The stained areas of collagen were calculated by measuring 10 areas of the septum and ventricular right and left free walls and expressed as percentage of total area measured. Cardiomyocyte transverse diameter was determined in the digital analysis system in cells sectioned at the nuclear level. Sections were treated by means of immunohistochemical technique with a mouse monoclonal antibody against CD34 (endothelial cells; QBend/10, BioGenex, San Ramon, CA, USA). This procedure facilitates myocyte border detection.

Echocardiographic examination

Baseline cardiac geometry and function were evaluated by two‐dimensional M‐mode echocardiography with a 13 MHz linear transducer in conscious mice (3 and 6 months old). The echocardiographic study was performed without anaesthesia, after a procedure habituation period of at least 3 consecutive days. Systolic and diastolic dimensions were assessed according to the American Society of Echocardiography's guidelines (Lang et al. 2005). Global LV systolic function was determined by calculating left ventricular fractional shortening (LVFS):

LVFS = (LVEDD – LVESD / LVEDD) × 100, where LVFS is expressed as a percentage, and LVEDD and LVESD are left ventricular end‐diastolic and end‐systolic diameters, respectively.

Electrocardiogram

Surface electrocardiogram (ECG) was recorded in conscious mice, for 30 min, using standard ECG electrodes (lead I) and a PowerLab 4ST data acquisition system (Gonano et al. 2011). Twelve mice (6 SD^+/+/KO and 6 WT) were also studied under basal conditions with ECG telemetry (Contreras et al. 2014).

Surface ECG was recorded for 10 min and then a catecholaminergic challenge (caffeine and adrenaline, 120 mg kg⁻¹ and 1.6 mg kg⁻¹, respectively) was applied to unmask the presence of arrhythmias under ‘stress’ conditions.

Ex vivo experiments: intact hearts. Intracellular Ca²⁺ and action potentials

Animals were anaesthetized with an intraperitoneal (i.p.) injection of ketamine–diazepam (100 mg kg⁻¹ and 5 mg kg⁻¹, respectively). Central thoracotomy and heart excision were performed immediately after phase III anaesthesia was reached, verified by the loss of pedal withdrawal reflex. Isolated hearts were perfused according to the Langendorff technique at constant temperature (37°C) and flow (2–3 ml min⁻¹) as previously described (Vittone et al. 2002; Said et al. 2008).

Rhod‐2, Mag‐Fluo‐4 and Di‐8‐ANEPPS (Invitrogen) were used to evaluate intracellular Ca²⁺ transients in the cytosol and the SR and transmembrane action potentials (APs), respectively, at the epicardial layer of intact mouse hearts, as previously described (Valverde et al. 2010; Ferreiro et al. 2012), using a custom‐made setup for pulse local‐field fluorescence (PLFF) microscopy (Mejia‐Alvarez et al. 2003). After mounting and loading, hearts were paced at 5 Hz for approximately 15 min and intracellular Ca²⁺ or transmembrane potentials were recorded according to the different protocols (see Results). The recordings were obtained by gently placing one end of the optic fibre on the tissue.

Intracellular Ca²⁺ measurements in intact isolated myocytes

Myocytes were isolated by enzymatic digestion as previously described (Palomeque et al. 2009). Isolated myocytes were loaded with Fura‐2/AM. Ca²⁺ fluorescence was measured in an Myocyte Calcium and Contractility System (IonOptix, Milton, MA, USA) (Palomeque et al. 2009).

Confocal imaging of intact cardiac myocytes

Freshly isolated mouse ventricular myocytes were loaded with 10 μm fluo‐3‐AM (Molecular Probes) in Tyrode buffer containing 2.5 mm Ca²⁺ for 20 min at room temperature, and mounted in a small chamber placed onto an inverted microscope equipped with a 63× objective as previously described (Palomeque et al. 2009). After stabilization (usually 3–5 min), confocal line‐scanning (512 × 512 pixels and 4.3 ms per line) was performed along the longitudinal axis of cells (avoiding nuclei), using the Zeiss LSM 210 confocal system in quiescent cells. The fluo‐3‐AM loaded myocytes were excited using the 488 nm argon laser and the fluorescence emission was recorded at 500–550 nm. This procedure was repeated under ‘stress’ conditions, i.e. in the presence of 6.0 mm Ca²⁺. Extracellular Ca²⁺ was increased stepwise (2.5, 4.0 and 6.0 mm) to avoid cell damage.

Image processing

Ca²⁺ sparks were measured using the ‘Sparkmaster’ plugin for ImageJ (Picht et al. 2007). Ca²⁺ waves and mini‐waves were visually counted and classified, by two different operators, according to the following definitions. A propagated Ca²⁺ wave was defined as a continuous wave front in the line scan image visualized as a robust fluorescent line that propagates across the full width of the myocyte without breaking. A mini‐wave was defined as a fluorescent line that breaks without propagating across the full width of the cell. Image processing software was developed using Python language. This program allowed us to recheck the results obtained by eye. The main library used for this task was scikit‐image (http://scikit‐image.org/) (van der Walt et al. 2014), which is a freely available open source collection of algorithms that allows image manipulation. The program was also used to estimate SR Ca²⁺ leak by multiplying Ca²⁺ spark frequency (number of events per second, normalized to 100 μm) by the total fluorescence of each spark (sum of fluorescence of each pixel).

Ca²⁺ wave velocity was calculated on a linear region of the record, dividing a given distance covered by the wave by the corresponding time, as previously described (Mattiazzi et al. 2015). The different individual Ca²⁺ wave characteristics (amplitude and duration) were analysed by division of the wave into 10 sections, 8 pixels wide. Plots of these sections were averaged to calculate peak fluorescence and rate constants of the decay phase of a wave.

Modelling

A modified ten Tusscher–Panfilov (TP) human ventricular myocyte model (ten Tusscher & Panfilov, 2006) was used, incorporating acid–base regulation, a reformed RyR2 receptor and CaMKII effects on flows based on experimental data (Lascano et al. 2013), and a recently reformulated contractile part (Negroni et al. 2015). In addition, as a global representation of Ca²⁺‐induced Ca²⁺ release (CICR) mechanism, two dyadic clefts (DCs) were considered, the first (DC₁) consisting of the Ca²⁺ release units (CRUs) (RyR2 forming part of the RyR2 cluster) facing the L‐type Ca²⁺ channel (LCC), and the second (DC₂) composed of the remaining CRUs not facing LCC (Inoue & Bridge, 2005). In DC₁, [Ca²⁺] elevation by LCC Ca²⁺ influx or from an overloaded SR‐induced Ca²⁺ leak activates CRU opening by CICR. This Ca²⁺ diffuses to DC₂ promoting regenerative CICR release from the neighbouring RyR2. Then, all Ca²⁺ in DC₂ (diffusing from DC₁ and/or released from DC₂) passes into the bulk cytoplasm (CYTO).

It was assumed that RyR2 activation (either spontaneously or by LCC Ca²⁺ influx) starts in DC₁ and is forced to pass through DC₂ before entering CYTO. To represent this concept, the original TP DC volume was postulated as a hexagon surrounded by six other hexagons of the same size. This involved seven CRUs for each LCC, within the range postulated by Izu et al. (2001) and Sato & Bers (2011). Considering that the original TP DC volume was 0.34% of CYTO volume, in this new approach total cleft volume was 2.42% of CYTO volume, similar to the one postulated in the O'Hara human myocyte model (2.94%) (O'Hara et al. 2011). Once the volumes were defined, the DC compartments were simplified as two concentric cylinders (see Fig. 9 B in Results for further details), allowing Ca²⁺ diffusion through their lateral surfaces (surface 1 and surface 2). To represent this concept, the conductance parameter from the single DC to CYTO (xfer) in the TP model gave rise to xfer1 and xfer2 parameters, imposing a total conductance equivalent to xfer1 and xfer2 arranged in series.

$$\frac{1}{\mathrm{xfer}}=\frac{1}{\mathrm{xfer}1}+\frac{1}{\mathrm{xfer}2}$$ (1)

Figure 9. Schematic diagram of the model .

Human myocyte model (C) including ion pumps and exchangers responsible for action potential, Ca²⁺ management, force development, pH_i and Ca²⁺–calmodulin‐dependent protein kinase II (CaMKII) regulation. Our previous model (Lascano et al. 2013) was modified by the addition of a second dyadic cleft: DC₁ and DC₂. The model is coupled to the myofilament force development model consisting of 5‐state troponin systems (TS) with Ca²⁺ binding at the myofilament level (Negroni et al. 2015). CYTO, cytosol; LTCC, L‐type Ca²⁺ channels; NCX, Na⁺/Ca²⁺ exchanger; NKA, Na⁺/K⁺‐ATPase; Cap, membrane Ca²⁺‐ATPase; NHE, Na⁺/H⁺ exchanger; NBC, Na⁺–HCO₃ ⁻ cotransporter; CHE, Cl⁻/OH⁻ exchanger; AE, anion exchanger; I represents different ionic currents (see text for further details). A, detail of the DC structure where DC₁ is surrounded by 6 DC₂. DC₁ represents the dyadic clefts where RyR2 are facing the L‐type Ca²⁺ channel (LCC) and DC₂ the dyadic clefts where RyR2 are not facing LCC. In the original TP model (ten Tusscher & Panfilov, 2006), the single dyadic cleft involves 14.3% of the present total cleft space (DC₁+DC₂). B, dyadic cleft structure assumed as two concentric cylinders. Surface 1: lateral surface of the cylinder with radius R, representing DC₁. Surface 2: lateral surface of the cylinder with radius R2, representing DC₂. The planar surfaces limiting the height of the cylinders are the sarcolemmal membrane on one side and the SR membrane on the other (see text for further details).

It was assumed that Ca²⁺ diffusion was perpendicular to cylinder surface and proportional to xfer; then, surface1 = k × xfer1 and surface2 = k × xfer2. The relationship between the two cylinder lateral surfaces provided the equation to find the numerical xfer1 and xfer2 values.

$$\frac{\mathrm{Surface}2}{\mathrm{Surface}1}=\frac{2\pi \times R_2\times \mathrm{height}}{2\pi \times R_1\times \mathrm{height}}=\frac{R_2}{R_1}=\frac{\mathrm{xfer}2}{\mathrm{xfer}1}.$$

$$\mathrm{Vo}{\mathrm{l}}_2=\pi \times R_2^2\times \mathrm{height}-\mathrm{Vo}{\mathrm{l}}_1=K\times \mathrm{Vo}{\mathrm{l}}_1$$

with K = 6

$$\frac{\pi \times R_2^2\times \mathrm{height}}{\pi \times R_1^2\times \mathrm{height}}=\frac{(K+1)\times \mathrm{Vo}{\mathrm{l}}_1}{\mathrm{Vo}{\mathrm{l}}_1}=K+1$$

$$\frac{R_2}{R_1}=\frac{\mathrm{xfer}2}{\mathrm{xfer}1}=\sqrt{K+1}$$

$$\mathrm{xfer}2=\sqrt{7}\times \mathrm{xfer}1$$ (2)

Solving eqn (1) and eqn (2) systems gives numerical values for xfer1 and xfer2. Lower case k is different from upper case K; the first is the proportionality constant between lateral surface and conductance, and K is the relationship between volume 2 and volume 1.

The DC₂ compartment was assumed to have 10% of the TP NCX (Shannon et al. 2004) with the rest in the CYTO membrane. Moreover, Ca²⁺ flow parameters involving DC₁ and DC₂ were adjusted to achieve total flows of the original TP model, resulting in adequate stabilized beats (see Results).

The four‐state RyR2 structure regulated by [Ca²⁺]_SR and [Ca²⁺] in the cleft postulated by Shannon et al. (2004) was used, modified to represent SR Ca²⁺ leak as Ca²⁺ flow through the RyR2 modulated by the resting (R) state of this channel (Lascano et al. 2013).

To simulate the behaviour of S2814D^+/+ and SD^+/+/KO mice, an increase in RyR2 Ca²⁺ release and in SERCA2a Ca²⁺ uptake was added as appropriate. These changes were represented as 50% increased conductance for RyR2 and 50% for SR Ca²⁺ uptake. These two values were obtained from experimental data, showing an increase in SR Ca²⁺ leak of 50% in S2814D^+/+ mice relative to WT mice (van Oort et al. 2010 and the present results) and an increase in SR Ca²⁺ uptake of approximately 50% from control values at pCa between 7.0 and 6.5 (Luo et al. 1994). Additionally, increased Ca²⁺ uptake was accompanied by 25% decrease in RyR2 conductivity to reproduce the reported compensatory mechanism in PLNKO mice (Chu et al. 1998).

To simulate a stress situation similar to the experimental conditions in isolated myocytes and intact hearts, the SR was first loaded with 4.2 mm external [Ca²⁺]_o at 150 stimuli min⁻¹ during 1 min; then the stimulation was stopped and spontaneous action potentials (SAPs) or early or delayed afterdepolarizations (EADs or DADs, respectively) were computed for 12 s. An adrenergic challenge, as the one used in intact mice, was simulated by applying the effect of 100 nm isoproterenol (Iso). Since CaMKII was incorporated into our model and its activity and effects adapt to [Ca²⁺]_i variations (Lascano et al. 2013), only the effects of protein kinase A (PKA) on the different Ca²⁺ fluxes were simulated based on experimental data, as previously described (Negroni et al. 2015): SERCA2a activity was increased by 40% (equivalent to 68% in SERCA2a K _m reduction); I _CaL and phospholemman sensitivity were increased by 40% and 20%, respectively. Other PKA targets were modified as previously described (Negroni et al. 2015). Iso effect on SERCA2a activity was not considered in SD^+/+/KO_sim, due to phospholamban (PLN) ablation. RyR2 conductance was not modified by PKA, considering that most of the evidence indicates that PKA does not appear to significantly influence RyR2 diastolic leakage (Li et al. 2002; Curran et al. 2007). Contraction frequency was modestly increased to 80 beats min⁻¹ after the stabilization period at 70 beats min⁻¹.

Statistics

Continuous variables were expressed as means ± SEM and were evaluated with either unpaired Student's t test or ANOVA followed by Tukey's post hoc test, to compare differences between groups. Spontaneous Ca²⁺ release events were expressed as nonparametric continuous data. The Mann–Whitney U test was used to compare non‐normal data distribution. P < 0.05 was considered significant. Categorical data were expressed as percentages and compared with Fisher's exact test.

Results

Structural characterization of PLNKO/RyR2‐S2814D^+/+ double‐mutant mice (SD^+/+/KO)

WT and SD^+/+/KO mice were genotyped by PCR analysis using mouse tail DNA and specific primers for WT, PLNKO and S2814D^+/+ mutations, as previously described (Luo et al. 1994; van Oort et al. 2010). Cardiac wall dimensions, assessed from haematoxylin–eosin‐stained histological sections of 3‐month‐old mice, were similar between WT and SD^+/+/KO hearts: left ventricular anteroposterior diameter 1.33 ± 0.19 vs. 2.01 ± 0.29 mm, left ventricular maximal wall thickness 1.20 ± 0.30 vs. 1.65 ± 0.20 mm, left ventricular minimal wall thickness 0.96 ± 0.09 and 1.05 ± 0.12 mm, septum thickness 1.30 ± 0.14 and 1.23 ± 0.19 μm, respectively (n = 3). Moreover, no significant differences were detected in either myocardial cell transverse diameter (17.40 ± 1.03 μm (WT mice,) vs. 15.50 ± 0.32 μm (SD^+/+/KO mice), n = 134 and 139 myocytes, respectively, from 3 hearts, or interstitial fibrosis 4.0 ± 0.5% (WT) vs. 4.5 ± 0.6 % (SD^+/+/KO) (n = 3). Table 1 shows that at baseline, cardiac structure and function were similar in young (3‐month‐old) WT and SD^+/+/KO mice when determined by echocardiography. Instead, at 6 months, SD^+/+/KO exhibited a significant increase in end‐diastolic diameter, systolic ventricular septum wall thickness and left ventricular posterior wall thickness vs. WT mice. An increase in these parameters was also previously observed in S2814D^+/+ mice at 12 months (van Oort et al. 2010).

Table 1.

Echocardiographic parameters of WT and SD^+/+/KO mice at 3 and 6 months of age

	3 months old		6 months old
	WT	SD+/+KO	WT	SD+/+KO
n	4	4	4	4
Heart rate (beats min−1)	651.82 ± 22.80	707.11 ± 17.03	618.82 ± 25.8	623.68 ± 7.89
Ejection fraction (%)	92.86 ± 0.38	94.54 ± 2.43	91.92 ± 1.00	94.70 ± 0.30
LVFS (%)	31.08 ± 0.49	34.51 ± 3.67	30.66 ± 1.01	34.19 ± 1.49
ESD (mm)	1.02 ± 0.03	0.95 ± 0.06	1.02 ± 0.02	1.00 ± 0.03
EDD (mm)	2.83 ± 0.07	3.02 ± 0.15	2.72 ± 0.09	3.13 ± 0.11*‡
IVSs (mm)	1.39 ± 0.04	1.37 ± 0.05	1.26 ± 0.09†	1.60 ± 0.07*‡
IVSd (mm)	0.75 ± 0.01	0.68 ± 0.02*	0.70 ± 0.01	0.79 ± 0.03
LVPWs (mm)	1.61 ± 0.08	1.59 ± 0.12	1.26 ± 0.09†	1.60 ± 0.07*‡
LVPWd (mm)	0.90 ± 0.04	0.74 ± 0.02*	0.68 ± 0.03†	0.85 ± 0.04*

Data are expressed as mean ± SEM. LVFS, left ventricular fractional shortening; ESD and EDD, end‐systolic and end‐diastolic diameter; IVSs and IVSd, intraventricular septum wall thickness in systole or diastole; LVPWs and LVPWd, LV posterior wall thickness in systole or diastole. *P < 0.05 WT vs. SD^+/+/KO; ^† P < 0.05, 3 months vs. 6 months for WT; ^‡ P < 0.05, 3 months vs. 6 months for SD^+/+/KO.

Effects of PLN ablation on altered Ca²⁺ dynamics of S2814D^+/+ mice

Figure 1 shows typical recordings (Fig. 1 A) and overall results (Fig. 1 B–H) of intracellular Ca²⁺ transients and Ca²⁺ dynamics in cardiac myocytes of SD^+/+/KO and S2814D^+/+ mice. Records and results of PLNKO and WT myocytes are also displayed for comparison. In SD^+/+/KO mice, diastolic Ca²⁺ was significantly decreased and twitch Ca²⁺ transient amplitude significantly enhanced with respect to S2814D^+/+ myocytes (Fig. 1 B and C), whereas the rate of rise of Ca²⁺ transient was higher than in S2814D^+/+ and WT myocytes (Fig. 1 D). SR Ca²⁺ load estimated by the caffeine‐induced Ca²⁺ transient amplitude – which was diminished in S2814D^+/+ with respect to WT myocytes, as expected from a leaky RyR2 – was significantly enhanced by PLN ablation in SD^+/+/KO myocytes (Fig. 1 E), whereas fractional Ca²⁺ release was more improved in SD^+/+/KO than in S2814D^+/+ myocytes (Fig. 1 F). Moreover, SD^+/+/KO myocytes exhibited a faster time constant of twitch Ca²⁺ transient decay and a slower time constant of caffeine‐induced Ca²⁺ transient decay than S2814D^+/+ (Fig. 1 G and H).

Figure 1. PLN ablation enhances Ca²⁺ dynamics of S2814D^+/+ myocytes .

Basal intracellular Ca²⁺ dynamics in electrically stimulated (1.0 Hz) cardiac myocytes of WT, S2814D^+/+, PLNKO and SD^+/+/KO mice. A, typical examples of twitch‐ and caffeine‐induced Ca²⁺ transients. B, diastolic Ca²⁺. C, twitch Ca²⁺ transient (Ca_iT) amplitude. D, maximal rate of rise of Ca_iT. E, caffeine‐induced Ca_iT amplitude. F, fractional Ca²⁺ release. G and H, twitch‐ and caffeine‐induced Ca_iT decay (tau). Caffeine‐induced Ca_iT decay, used to estimate NCX activity, was higher in PLNKO mice vs. WT. A similar increase was observed in SD^+/+/KO mice. In this and the following figures, the number of myocytes and animals (between parenthesis), are shown inside the bars. FU, fluorescence units. *P < 0.05 vs. WT; ^# P < 0.05 vs. S2814D^+/+.

Analogous changes in Ca²⁺ dynamics were observed in SD^+/+/KO mice when cytosolic and intra‐SR Ca²⁺ were measured at the epicardial layer of the intact beating heart (Fig. 2). Cytosolic and intra‐SR Ca²⁺ transients relaxed faster and restitution of Ca²⁺ transients was more rapid in SD^+/+/KO than in S2814D^+/+ and WT mice and similar to PLNKO mice, which is consistent with a decrease in RyR2 refractoriness produced by the enhanced SR Ca²⁺ uptake due to PLN ablation in PLNKO and SD^+/+/KO myocytes (Huser et al. 1998; Szentesi et al. 2004). Instead, no differences in action potential restitution curves were observed between SD^+/+/KO and S2814D^+/+ animals.

Figure 2. PLN ablation increased the cytosolic and SR Ca²⁺ relaxation and Ca²⁺ transient restitution of S2814D^+/+ intact hearts .

Basal intracellular Ca²⁺ dynamics at the epicardial layer of the intact beating heart of WT, S2814D^+/+, PLNKO and SD^+/+/KO mice. A and B, upper panel, typical examples of cytosolic Ca²⁺ transient (Rhod‐2 fluorescence) (A) and intra‐SR Ca²⁺ transient records (Mag‐fluo‐4 fluorescence) (B). Lower panel, overall results showing the time constant (tau) of cytosolic (A) and intra‐SR Ca²⁺ decay (B) in the four groups of hearts. Cytosolic and intra‐SR Ca²⁺ both decayed faster in SD^+/+/KO than in S2814D^+/+ and WT mice and similar to PLNKO hearts. C and D, restitution curves for cytosolic Ca²⁺ transient and action potential (above) and their respective exponential time constants (below). Whereas Ca²⁺ restitution curves increased faster in SD^+/+/KO than in S2814D^+/+ mice, action potential restitution curves were similar in both groups. The inset of the figure depicts the experimental protocol. *P < 0.05 vs. WT; ^# P < 0.05 vs. S2814D^+/+ and ^& P < 0.05 vs. PLNKO.

Taken together the results indicate that SD^+/+/KO mice combine, at the functional cellular level, the enhanced SR Ca²⁺ uptake produced by PLN ablation and the decrease in SR Ca²⁺ load (relative to PLNKO) produced by the enhanced SR Ca²⁺ leak typical of S2814D^+/+ mice.

PLN ablation diminished ventricular arrhythmias promoted by CaMKII phosphorylation of S2814 on RyR2

To evaluate the effects of increasing SR Ca²⁺ uptake on the arrhythmogenic propensity typical of CaMKII‐mediated RyR2 phosphorylation, we used 3‐ to 4‐month‐old S2814D^+/+ knock‐in and SD^+/+/KO mice to study ECG waveforms. S2814D^+/+ and SD^+/+/KO mice had normal heart rhythm at rest, with most of the electrophysiological parameters such as PQ and QRS being unaltered, although there was a modest but significantly prolonged QT interval, a finding also observed in PLNKO (Table 2). However, when challenged with an i.p. injection of caffeine/adrenaline, the ECG response was completely different between S2814D^+/+ and SD^+/+/KO mice. Whereas S2814D^+/+ mice exhibited a high incidence of premature ventricular complexes (PVCs), in agreement with previous findings (van Oort et al. 2010), PVCs were significantly less frequent in SD^+/+/KO mice (Fig. 3 A–C). Moreover the high incidence of sustained bidirectional ventricular tachycardia seen in S2814D^+/+ mice (4 out of 5), was completely offset in SD^+/+/KO animals (0 out of 6, P < 0.05) (Fig. 3 D).

Table 2.

ECG basal parameters in conscious mice

ECG parameter	WT	S2814D+/+	PLNKO	SD+/+/KO
HR (beats min−1)	598.0 ± 52.2	663.9 ± 14.7	543.5 ± 36.0†	649.9 ± 16.2
RR (ms)	111.4 ± 7.2	90.9 ± 2.6*	107.8 ± 10.1	90.4 ± 2.0*
PR (ms)	35.8 ± 1.2	35.4 ± 1.0	36.5 ± 1.2	32.9 ± 0.8
P (ms)	14.1 ± 2.0	11.2 ± 0.5	10.9 ± 0.8	11.0 ± 0.5
QRS (ms)	11.7 ± 0.8	11.0 ± 0.3	13.8 ± 0.7	11.5 ± 0.8
QT (ms)	47.1 ± 1.1	45.9 ± 1.3	70.0 ± 8.6*,†	53.1 ± 2.4‡
QTc (ms)	44.9 ± 1.8	48.1 ± 1.3	67.0 ± 4.9*,†	55.7 ± 2.1‡

HR, heart rate; RR, RR period duration; PR, PR interval duration; P, P wave duration; QRS, QRS complex duration; QT, QT interval duration; QTc, corrected QT. Of note, SD^+/+/KO mice showed a prolongation of the QT interval, similar to that observed in PLNKO mice. Although we did not explore the cause of this prolongation, it may result from the down‐regulation of outward K⁺ currents that has been described in the presence of similar chronic alteration of intracellular Ca²⁺ homeostasis, in mice overexpressing SERCA2a (Xu et al. 2005). *P < 0.05 vs. WT, ^† P < 0.05 vs. S2814D^+/+, ^‡ P < 0.05 vs. PLNKO (n = 4–11 for the different groups).

Figure 3. PLN ablation rescues the propensity to ventricular arrhythmias promoted by CaMKII constitutive CaMKII RyR2 phosphorylation of RyR2 at the S2814 site in vivo .

A and B, representative ECG tracings in conscious S2814D^+/+ knock‐in mice and SD^+/+/KO mice respectively, at rest and after i.p. injection of caffeine/adrenaline (Caff/Adr). Arrows in S2814D^+/+ tracings indicate premature ventricular complexes (PVCs). Bidirectional ventricular tachycardia (BVT) was observed in the majority of S2814D^+/+ mice after Caff/Adr challenge. PVCs were rare and BVT was completely absent in SD^+/+/KO mice. C and D, box plot showing the frequency of PVCs and incidence of VT, respectively, in S2814D^+/+ and SD^+/+/KO mice after Caff/Adr challenge. *P < 0.05 vs. S2814D^+/+.

Since homozygous S2814D^+/+ mice represent maximal CaMKII‐dependent RyR2 phosphorylation, we further assessed the impact of increasing SR Ca²⁺ uptake on the propensity of arrhythmias evoked by CaMKII pseudo‐phosphorylation of RyR2 at the S2814 site, in heterozygous S2814D mice (S2814D^+/‐ mice) with PLN ablation (S2814D^+/− /PLNKO^−/−). These mice are referred to as SD^+/−/KO mice. These experiments were performed in isolated perfused mouse hearts preloaded with the fluorophore Di‐8‐ANEPPS (Invitrogen) to evaluate transmembrane APs at the epicardial layer of intact mouse hearts (Mejia‐Alvarez et al. 2003). Under basal conditions the APs occurred regularly, following the imposed stimulation frequency. Moreover, AP configuration was similar in both types of mice, although AP duration at 50% repolarization (APD₅₀) tended to be prolonged in SD^+/−/KO myocytes relative to S2814D^+/−, 41.5 ± 4.7 ms (n = 6) vs. 30.1 ± 4.7 ms (n = 7), respectively, in agreement with the QT prolongation in homozygous SDKO animals shown in Table 2. S2814D^+/− and SD^+/−/KO hearts were then subjected to three successive 5 s bursts of high frequency pacing (12 Hz) at high extracellular Ca²⁺ concentration, followed by a pacing period at 1 Hz and high Ca²⁺, to compare cardiac susceptibility to ventricular ectopic activity of S2814D^+/− and SD^+/−/KO hearts. Figure 4 shows typical tracings (Fig. 4 A and C) and overall results (Fig. 4 B and D) of these experiments. During the first 3 min after the burst pacing protocol, S2814D^+/− hearts did not usually respond to the low pacing frequency imposed and exhibited a spontaneous high frequency (Fig. 4 A) and a significantly higher increase of premature ventricular beats versus SD^+/−/KO hearts and in the incidence of trains of ventricular tachycardia (VT) (S2814D^+/−, 63% vs. SD^+/−/KO, 0%) (Fig. 4 B and D). Taken together, the results indicate that ablation of PLN rescues the in vivo and ex vivo ventricular arrhythmic events promoted by maximal and submaximal CaMKII pseudo‐phosphorylation of RyR2 at the S2814 site.

Figure 4. PLN ablation rescues the propensity to ventricular arrhythmias promoted by constitutive phosphorylation of RyR2 at the S2814 site ex vivo .

A and C , representative action potential tracings at rest and after a high frequency train of stimuli in S2814D^+/− (A) and SD^+/−/KO mice (C). Arrows in A indicate the occurrence of a spontaneous beat. B, premature ventricular beat incidence (left panel) and frequency (right panel, box plot). D, ventricular tachycardia incidence (left panel) and duration (right panel). *P < 0.05 vs. S2814D^+/−.

PLN ablation increases Ca²⁺ spark frequency but decreases the cell‐wide propagating Ca²⁺ waves produced by constitutive phosphorylation of RyR2 at the S2814 site

Next, we sought to elucidate the underlying mechanisms of the anti‐arrhythmogenic effects of increasing SR Ca²⁺ uptake by means of PLN ablation. We used confocal microscopy to assess SR Ca²⁺ sparks and waves in isolated ventricular myocytes. We created conditions for the transition of localized Ca²⁺ releases into propagating Ca²⁺ waves by increasing the cellular Ca²⁺ load via elevation of extracellular [Ca²⁺]. This procedure is known to increase [Ca²⁺] in both cytosolic and SR luminal compartments of cardiac myocytes (Wier et al. 1987). Figure 5 A–E displays typical examples and main characteristics of SR Ca²⁺ sparks at baseline for the four mouse strains (WT, S2814D^+/+, PLNKO and SD^+/+/KO). The amplitude and rate of rise of SR Ca²⁺ sparks was higher in S2814D^+/+, PLNKO and SD^+/+/KO with respect to WT myocytes, whereas full duration at half‐maximum was significantly reduced in PLNKO and SD^+/+/KO vs. S2814D^+/+ and WT myocytes. Figure 5 F and G shows that when S2814D^+/+ and SD^+/+/KO myocytes were submitted to 6.0 mm extracellular Ca²⁺, the frequency of Ca²⁺ sparks and the calculated SR Ca²⁺ leak was higher in SD^+/+/KO than in S2814D^+/+ myocytes, consistent with the overall higher SR Ca²⁺ load in SD^+/+/KO cells (see Fig. 1). We then examined the occurrence and characteristics of Ca²⁺ waves in S2814D^+/+ and SD^+/+/KO myocytes. Figure 6 A, upper and middle panels, shows typical images of cell‐wide propagating Ca²⁺ waves displayed by WT and S2814D^+/+ myocytes when exposed to high extracellular Ca²⁺. In contrast, in SD^+/+/KO myocytes, Ca²⁺ waves were mostly replaced by non‐propagating Ca²⁺ events, known as mini‐waves (Fig. 6 A lower panel). Figure 6 B shows expanded line scan images at high extracellular Ca²⁺ to further illustrate the predominance of Ca²⁺ waves in S2814D^+/+ myocytes and of Ca²⁺ mini‐waves in SD^+/+/KO myocytes. The overall results of these experiments are depicted in Fig. 6 C. At 6.0 mm extracellular Ca²⁺, S2814D^+/+ myocytes showed a significantly higher frequency of Ca²⁺ waves compared to WT and SD^+/+/KO myocytes. Thus, in spite of the higher SR Ca²⁺ leak observed in SD^+/+/KO myocytes vs. S2814D^+/+ (Fig. 5), fully propagating events paradoxically predominated in S2814D^+/+, whereas in SD^+/+/KO myocytes there was a significantly higher proportion of mini‐waves over waves. Figure 7 shows main characteristics of Ca²⁺ waves in S2814D and SD^+/+/KO mice at 6.0 mm extracellular Ca²⁺. Figure 7 A illustrates an example of a Ca²⁺ wave (red arrow) in a S2814D^+/+ myocyte, associated with a triggered Ca²⁺ transient and cell contraction (white arrow). This is a common finding in S2814D^+/+ but not in SD^+/+/KO myocytes, which strongly supports the notion that a Ca²⁺ wave is an arrhythmogenic event, triggering an action potential and the associated contraction. Figure 7 B and C shows Ca²⁺ wave propagation and typical time courses of wave fluorescence signals from S2814D^+/+ and SD^+/+/KO myocyte line scans. As indicated, Ca²⁺ waves propagated faster in SD^+/+/KO than in S2814D^+/+ mice. Figure 7 D shows the overall results of Ca²⁺ wave properties: whereas fluorescence intensity rose and decayed faster in SD^+/+/KO vs. S2814D^+/+ myocytes, full duration of fluorescence was shorter and the resultant peak fluorescence intensity showed a trend to be higher in SD^+/+/KO myocytes vs. S2814D^+/+, without reaching statistically significant levels.

Figure 5. Typical space–time plot (A) and main characteristics (B–E) of spontaneous SR Ca²⁺ sparks under basal conditions in the different groups of myocytes .

Spark amplitude (B) and the rate of rise of Ca²⁺ sparks (C) are higher in S2814D^+/+, PLNKO and SD^+/+/KO myocytes. Full duration at half‐maximum (FDHM, D) was decreased in PLNKO and SD^+/+/KO vs. S2814D^+/+ and WT and full width at half‐maximum (FWHM, E), was similar in all groups. F and G, overall results showing the frequency of Ca²⁺ sparks and the estimated SR Ca²⁺ leak, respectively, in S2814D^+/+ and SD^+/+/KO myocytes. FU, fluorescence units. *P < 0.05 vs. WT, ^# P < 0.05 vs. S2814D.

Figure 6. PLN ablation rescues the propensity to SR Ca²⁺ propagating events of S2814D^+/+ myocytes .

A, typical confocal linescan images obtained in isolated myocytes from WT, S2814D^+/+ and SD^+/+/KO myocytes at 2.5 and 6.0 mm extracellular Ca²⁺. Myocytes were stimulated at constant frequency (0.5 Hz). When stimulation was stopped, Ca²⁺ waves appeared in S2814D^+/+ mice. In contrast, Ca²⁺ mini‐waves were present at 2.5 mm Ca²⁺ and further increased at 6.0 mm Ca²⁺ in SD^+/+KO myocytes. In this example mini‐waves occurred simultaneously at multiple sites but they were soon interrupted. B, amplified confocal linescan images obtained at 6.0 mm Ca²⁺, showing Ca²⁺ waves (upper panel) and Ca²⁺ mini‐waves (lower panel) typically displayed by S2814D^+/+ and SD^+/+/KO myocytes, respectively. C, overall results of Ca²⁺ wave frequency (right panel) and Ca²⁺ mini‐wave frequency (left panel) at low and high extracellular Ca²⁺. Ca²⁺ waves frequency was significantly higher in isolated S2814D^+/+ myocytes vs. WT myocytes at 2.5 and 6.0 mm Ca²⁺. Ablation of PLN did not modify Ca²⁺ wave frequency in SD^+/+/KO myocytes at 2.5 mm Ca²⁺ but significantly decreased it at 6.0 mm Ca²⁺. Moreover, PLN ablation increased Ca²⁺ mini‐wave frequency at 2.5 and 6 mm Ca²⁺. *P < 0.05 vs. WT; ^# P < 0.05 vs. S2814D^+/+.

Figure 7. Main characteristics of Ca²⁺ waves in S2814D^+/+ and SD^+/+/KO myocytes .

A, typical confocal linescan images of isolated S2814D^+/+ and SD^+/+/KO myocytes at 6.0 mm extracellular Ca²⁺, to illustrate the association of a wave (red arrow) with a triggered Ca²⁺ transient and myocyte contraction (see indentation in fluorescence signal, white arrow). B, confocal linescan examples and average results of the velocity of propagation of Ca²⁺ waves at 6.0 mm extracellular Ca²⁺. Ca²⁺ waves are faster in SD^+/+/KO than in S2814D^+/+ myocytes. Broken lines indicate that the slope of front wave propagation was higher in SD^+/+/KO vs. S2814D^+/+ myocytes. C, Ca²⁺ wave profiles from a S2814D^+/+ and a SD^+/+/KO myocyte. D, analysis of Ca²⁺ wave profiles indicates that SD^+/+/KO Ca²⁺ waves rose and decayed faster and were shorter in duration than S2814D^+/+ Ca²⁺ waves, but they were similar in magnitude. *P < 0.05 vs. S2814D^+/+.

Taken together these results indicate that ablation of PLN was able to reverse the propensity of S2814D^+/+ myocytes to trigger arrhythmogenic Ca²⁺ waves under stress conditions by converting most of the propagating Ca²⁺ waves into non‐propagation Ca²⁺ events. The reduced duration of the remaining Ca²⁺ waves in SD^+/+/KO myocytes would also contribute to reduce the incidence of arrhythmias in these myocytes, by decreasing the time period of Ca²⁺ efflux from the cell (O'Neill et al. 2004).

Inhibition of SERCA2a sets back the protective effects of PLN ablation in SD^+/+/KO myocytes

As a proof of concept of the beneficial effects of increasing SR Ca²⁺ uptake on Ca²⁺ triggered SR Ca²⁺ waves due to CaMKII constitutive pseudo‐phosphorylation of RyR2 at the S2814 site, we performed experiments in SD^+/+/KO isolated myocytes treated with 1 μm of the SERCA2a inhibitor cyclopiazonic acid (CPA) and submitted to the same stress protocol described above. It has been previously shown that this CPA concentration inhibits SERCA2a activity by approximately 80% (Schwinger et al. 1997). In our conditions, 1 μm CPA decreased the time constant of Ca²⁺ transient decline by 79.6 ± 28.0 % (P< 0.05, n=6). Figure 8 A describes a typical experiment in an isolated SD^+/+/KO myocyte. Figure 8 Aa depicts paced Ca²⁺ transients at 6.0 mm extracellular Ca²⁺ concentration. Figure 8 Ab shows the typical mini‐waves displayed by these myocytes when stimulation was stopped. After addition of 1 μm CPA, Ca²⁺ mini‐waves shifted to fully propagating Ca²⁺ waves (Fig. 8 Ac). Paced transients of SD^+/+/KO in the presence of CPA (Fig. 8 Ad) show that transients are prolonged when compared to the absence of CPA (first panel), mimicking the behaviour of S2814D^+/+ myocytes. Figure 8 B shows the overall results of these experiments: after CPA treatment, the relationship Ca²⁺ mini‐waves/waves observed in SD^+/+/KO myocytes was significantly decreased. The main characteristics of Ca²⁺ waves in SD^+/+/KO myocytes at 6.0 mm extracellular Ca²⁺ before and after treatment with CPA are shown in Fig. 8 C and D. After treatment with CPA, the characteristic SD^+/+/KO Ca²⁺ waves shifted to those of S2814D^+/+ Ca²⁺ waves, i.e. they were significantly slower and longer in duration than but similar in amplitude to SD^+/+/KO waves before CPA treatment. These findings confirm the protective effect of PLN ablation on arrhythmogenic Ca²⁺ waves induced by constitutive phosphorylation of RyR2‐S2814.

Figure 8. Results of the effects of SERCA2a inhibition on the ratio Ca²⁺ mini‐waves/Ca²⁺ waves and the characteristics of Ca²⁺ waves in SD^+/+/KO myocytes .

A, confocal images showing the sequential behaviour of a SD^+/+/KO myocyte after treatment with 1 μm of the SERCA2a inhibitor cyclopiazonic acid (CPA) at high extracellular Ca²⁺. Aa, confocal images of Ca²⁺ transients in a SD^+/+/KO myocyte when paced at constant frequency. Ab, high Ca²⁺‐induced SR Ca²⁺ mini‐waves (light blue arrows) after stimulation was stopped. Ac, high Ca²⁺‐induced SR Ca²⁺ waves (red arrows) in the presence of CPA. Ad, SD^+/+/KO myocyte in the presence of CPA, paced at constant frequency. Addition of CPA transforms Ca²⁺ mini‐waves, typical of SD^+/+/KO at high Ca²⁺, into Ca²⁺ waves, typical of S2814D^+/+ myocytes. B, inhibition of SERCA2a with 1 μm of CPA decreases Ca²⁺ mini‐waves and increases Ca²⁺ waves, so that the ratio Ca²⁺ mini‐waves/Ca²⁺ waves significantly decreases. C, Ca²⁺ wave profiles in a SD^+/+/KO myocyte before and after treatment with CPA. D, analysis of Ca²⁺ wave profiles. Note that SD^+/+/KO Ca²⁺ waves rose and decayed more slowly and were prolonged after treatment with CPA, similar to S2814D^+/+ (see Fig. 7). *P < 0.05 vs. S2814D^+/+.

Computational simulation predicts that increasing SR Ca²⁺ uptake diminishes the spontaneous arrhythmic events in a human myocyte model

The present experiments indicated that increasing SR Ca²⁺ uptake by PLN ablation was able to rescue the enhanced propensity to arrhythmias of mice with CaMKII‐constitutive phosphorylation of the RyR2‐S2814 site. An intriguing question that arises from these experiments is whether any increase in SR Ca²⁺ uptake, independently of its extent, is able to counteract the detrimental effect of a CaMKII‐dependent leaky RyR2. To address this question, we utilized an already tested integrated human cardiac myocyte model to gain further insight into the effects of increasing SR Ca²⁺ uptake on SR‐Ca²⁺ leak‐triggered arrhythmic events. The model encompasses electrophysiology, Ca²⁺ dynamics and contractile activity of the human myocyte (Lascano et al. 2013) (see Methods and Fig. 9). Figure 10 A shows typical steady state beats that simulate the distinctive characteristics of WT, S2814D^+/+, PLNKO and SD^+/+/KO cardiac myocytes in human normal conditions (2.0 mm extracellular Ca²⁺ and 70 stimuli min⁻¹). The main characteristics of intracellular Ca²⁺ dynamics displayed by the different modelled strains are shown in Fig. 10 B. Under these basal conditions, the myocyte models did not exhibit arrhythmogenic events and reproduced the intracellular Ca²⁺ transients and SR Ca²⁺ content pattern observed experimentally in the different mouse groups (compare with experimental results in Fig. 1). As a further model validation, it also exhibited the SR Ca²⁺ leak–SR Ca²⁺ content non‐linear relationship reported experimentally (Bers, 2014) (Fig. 10 C). Figure 11 A (left panel) shows the continuous SAPs observed after stimulation was stopped, in conditions that simulate S2814D^+/+ myocytes, when challenged with 4.2 mm extracellular Ca²⁺ and 150 beats min⁻¹. The middle panel of the figure shows that SAPs were not observed when the SD^+/+/KO conditions were simulated and only a few DADs appeared after stopping stimulation. To better appreciate the temporal sequence of these distinctive events, the left and middle panels show an expanded view of the different responses corresponding to the last stimulated beat and the following spontaneous events after stopping stimulation. In both S2814D^+/+ _sim and SD^+/+/KO_sim stimulated beats, the quasi‐simultaneous upstroke of AP, activation of I _Na, I _CaL and I _NCXrev preceded the increase in Ca²⁺ in the clefts ([Ca²⁺]_DC1–2) and the Ca²⁺ transient in cytosol ([Ca²⁺]_i). In the triggering mechanism of SAP and DAD, the sequence was reversed: in S2814D^+/+ _sim, the increased diastolic Ca²⁺ leak elicited a rise in Ca²⁺ in the dyadic cleft through CICR, and hence in the cytosol by diffusion therein. This increase – favoured by the enhanced diastolic Ca²⁺ – was high enough to augment the forward NCX mode (I _NCXdir) and to produce a membrane depolarization up to the level of fast Na⁺ channel opening (I _Na) and SAP generation. The higher SR Ca²⁺ uptake in SD^+/+/KO_sim myocytes, which provoked a low diastolic Ca²⁺, prevented the increase in cytosolic Ca²⁺ necessary to enhance I _NCXdir and the intracellular Na⁺ influx to depolarize the cell membrane up to the excitability threshold. Thus, the enhanced SR Ca²⁺ uptake in SD^+/+/KO_sim myocytes, precluded the triggering of a SAP, generating a DAD. This occurred in spite of a higher Ca²⁺ at the SR and the dyadic clefts in SD^+/+/KO_sim vs. S2814D^+/+ _sim myocyte. The inset in the figure shows an expanded image of the main events that explain the different behaviour of S2814D^+/+ _sim and SD^+/+/KO_sim. Cytosolic Ca²⁺ levels and I _NCXdir decisive participation in the type of arrhythmic events (SAPs or DADs) was confirmed by the reproduction of SAPs in SD^+/+/KO_sim, after introducing a threshold intracellular [Ca²⁺]_i pulse (2 nM Ca²⁺ in cytosol for 12 ms), during the end diastolic period of the last stimulated beat (not shown).

Figure 10. Model simulation of experimental results .

A, model‐derived time course of action potential, intracellular sarcoplasmic reticulum Ca²⁺ transient ([Ca²⁺]_SR), intracellular cytoplasmic Ca²⁺ transient ([Ca²⁺]_i), Ca²⁺ in dyadic cleft 1 and 2, and developed force (F _m) in WT, S2814D^+/+, PLNKO and SD^+/+/KO myocyte simulations. B, intracellular Ca²⁺ dynamics in the human myocyte model when the different characteristics of mouse myocytes were simulated (WT_sim, S2814D^+/+ _sim, PLKO_sim, SD^+/+/KO_sim). Notice the similarity with the behaviour of the experimental counterparts (see Fig. 1). C, relationship between diastolic Ca²⁺ leak (I _leak) and sarcoplasmic reticulum (SR) Ca²⁺ content ([Ca²⁺]_SR), in the different strains simulated. Note that for a given I _leak, SR Ca²⁺ load is lower in S2814D^+/+ _sim and SD^+/+/KO_sim myocytes than in WT_sim and PLNKO_sim myocytes, which is due to the higher open probability of RyR2 in these myocytes

Figure 11. Model simulation of experimental arrhythmias under stress conditions .

Aa and b, last steady state beat at high [Ca²⁺] before stopping high stimulation frequency (ST) and the events that followed after stopping stimulation, in simulated S2814D^+/+ (S2814D^+/+ _sim) and SD^+/+/KO (SD^+/+/KO_sim) cardiac myocytes. A train of spontaneous action potentials (SAP) immediately appeared after stopping stimulation in S2814D^+/+ _sim myocytes. In contrast, SD^+/+/KO_sim myocytes did not develop SAP but delayed afterdepolarizations (DAD). Ac–f, expanded view of different events of the last steady state beat and the first spontaneous events at high [Ca²⁺] after stopping stimulation. Ac and d, triggered AP after ST and SAP or DAD (V _m), and intracellular Ca²⁺ in the dyadic clefts [Ca]_c1 and [Ca]_c2, in the cytosol [Ca]_i and the SR ([Ca]_SR. Ae and f, SR Ca²⁺ uptake (I _up), SR Ca²⁺ release (I _rel), NCX current (I _NCX), L‐type Ca²⁺ current (I _CaL) and Na⁺ current (I _Na). Notice the higher SR Ca²⁺ content and uptake and the higher peak Ca²⁺ in the clefts in SD^+/+/KO_sim vs. S2814D^+/+ _sim. The inset to the right shows an amplified view of the main events leading to either SAP or DAD. The relative low rate of SR Ca²⁺ uptake (full line) (I _up) in S2814D^+/+ _sim allows an increase in cytosolic Ca²⁺ that activates NCX_dir to a level that triggers a SAP. In contrast, the higher SR Ca²⁺ uptake rate in SD^+/+/KO_sim precludes the necessary increase in cytosolic Ca²⁺ to activate NCX_dir and reach the membrane depolarization threshold to trigger a SAP. Note the similitude of the cytosolic Ca²⁺ profile of S2814D^+/+ _sim in the middle panel of the inset with the displayed profile of Fig. 7 A (middle panel). B, action potential and Ca²⁺ transients during ISO stimulation in WT, S2814D^+/+ _sim and SD^+/+/KO _sim showing no spontaneous events in WT, SAPs and DADs in S2814D^+/+, and EADs in SD^+/+/KO_sim. Note that whereas WT and SD^+/+/KO _sim myocytes followed the stimulation frequency (vertical mark, ST), in the S2814D^+/+ _sim myocyte, a SAP appeared before the triggered stimulus.

Since in the in vivo experiments we applied an adrenergic challenge, we also modelled the effects of β‐adrenergic stimulation in a cardiomyocyte, by simulating the effect of 100 nm isoproterenol (Iso) on the different Ca²⁺ fluxes, based on experimental data as previously described (Negroni et al. 2015). Figure 11 B shows that when the model was submitted to a β‐adrenergic challenge, S2814D^+/+ _sim cells displayed arrhythmic events which were not observed either in the wild‐type simulated model or in SD^+/+/KO_sim myocytes. Thus, although the in vivo conditions can be only partially mimicked in an isolated cell, the model reproduced the experimental finding that ablation of PLN prevents the arrhythmogenic effect of an adrenergic challenge.

We then asked the model the outcome of gradually increasing SR Ca²⁺ uptake from basal to PLNKO levels, on the occurrence of triggered SAP in a cardiomyocyte which simulates the characteristics of S2814D^+/+ cells. Since 100% PLN ablation produces 25% decrease in RyR2 expression (Chu et al. 1998), we simulated the gradual proportional decrease in RyR2 expression as a reduction in RyR2 conductance, for each incremental increase in PLN suppression. Table 3A depicts the results of this type of simulation. An increase in SERCA2a activity as modest as 10% was sufficient to drastically reduce SAPs from 21 to 5. When the effect of PLN ablation on RyR2 abundance (conductance) was not applied (Table 3B), the SR Ca²⁺ uptake required to reduce SAP, increased to 15%. This finding suggests that the reduction of RyR2 expression when the true SD^+/+/KO myocyte was simulated plays a negligible role in arrhythmia decline. Interestingly, the model shows that spontaneous beats declined with the gradual increase in SERCA2a activity despite the gradual rise of SR Ca²⁺ content. These results support the experimental findings indicating that the increase in SR Ca²⁺ uptake protects against triggered arrhythmias. Moreover, the model predicts that SR Ca²⁺ uptake below basal levels (Table 3B) also rescues from arrhythmic events, in agreement with previous experiments by Stokke et al. (2010). These authors demonstrated that reduction of SERCA2a abundance decreases the propensity for Ca²⁺ wave development in cardiomyocytes. The most probable reason for these results is the decrease in SR Ca²⁺ content produced by decreasing SR Ca²⁺ uptake (Stokke et al. 2010). Notice in Table 3B that the decrease in SR Ca²⁺ uptake was associated with a decrease in I _NCX and the level of membrane depolarization (V _m,max) necessary to reach the excitability threshold. Either no changes or increases in SR Ca²⁺ uptake up to 10% were associated with an increase in I _NCX and in V _m,max and a maximum increase in SAP frequency. Further increases in SR Ca²⁺ uptake up to 30% diminished the frequency of SAP, in spite of the increase in I _NCX. The increase in I _NCX allows membrane depolarization to reach the threshold to produce a single SAP. In the following spontaneous events, there was a progressive decrease in I _NCX (data not shown) to values which prevents the occurrence of successive SAPs, inducing only DADs. Thereafter, I _NCX as well as V _m,max fluctuated within values lower than those required for a SAP to occur.

Table 3.

Effect of percent changes in SERCA2a Ca²⁺ uptake on arrhythmia generation by enhancing RyR2 conductivity as in S2814D^+/+ myocytes

A. Progressive increase in SERCA2a Ca2+ uptake associated with a ‘compensatory’ prorated decrease in RyR2 conductance
% participation	SERCA2a activity	RyR2 Ca2+ conductance	[Ca2+]SR (μm)
of PLNKO	(% change vs. WT)	(% increase vs. WT)	(last st. beat)	I leak (μm s−1)	SAP	DAD/EAD
0 (S2814D+/+)	0	50.00	3726	20.52	21	0 / 0
10	5	46.25	3909	21.43	21	0 / 0
20	10	42.50	4091	22.12	5	3 / 0
30	15	38.75	4271	22.85	2	3 / 0
40	20	35.00	4447	23.60	1	3 / 0
50	25	31.25	4619	24.44	1	3 / 0
60	30	27.50	4787	25.22	1	3 / 0
70	35	23.75	4950	26.34	0	3 / 1
80	40	20.00	5106	27.65	0	3 / 1
90	45	16.25	5256	29.53	0	3 / 1
100 (SD+/+/KO)	50	12.50	5399	32.43	0	4 / 0

B. Progressive increase and decrease in SERCA2a Ca2+ uptake without ‘compensatory’ changes in RyR2 conductance
SERCA2a activity	[Ca2+]SR (μm)
(% change vs. WT)	(last st. beat)	I leak (μm s−1)	I NCX (μA μF−1)	V m,max (mV)	SAP	DAD/EAD
−40	2693	14.42	0.093	−90.0	0	0 / 0
−35	2798	14.82	0.090	−90.0	0	0 / 0
−30	2911	15.48	1.602	−76.6	0	1 / 0
−25	3035	16.07	1.760	−75.1	0	1 / 0
−20	3168	16.82	1.909	−73.6	0	1 / 0
−15	3306	17.65	2.054	−71.9	0	1 / 0
−10	3446	18.67	2.193	−70.0	0	2 / 0
−5	3587	19.57	2.323	−67.5	0	2 / 0
0	3726	20.52	2.438	Threshold*	21	0 / 0
5	3860	21.70	2.537	Threshold*	21	0 / 0
10	3988	22.85	2.618	Threshold*	21	0 / 0
15	4108	24.07	2.681	Threshold*	2	3 / 0
20	4218	25.41	2.725	Threshold*	2	3 / 0
25	4318	26.87	2.752	Threshold*	1	3 / 0
30	4407	28.61	2.759	Threshold*	1	3 / 0
35	4482	30.77	2.257	−69.1	0	3 / 2
40	4540	34.07	2.290	−68.6	0	4 / 2
45	4578	39.39	2.222	−69.8	0	4 / 2
50	4586	48.33	2.404	−65.7	0	4 / 0

A, enhanced RyR2 open probability was simulated by increasing RyR2 Ca²⁺ conductance and PLN inhibition by increasing SERCA2a activity. Data are expressed as percentage change with respect to WT. Simulations span from S2814D^+/+ to SD^+/+/KO conditions. The decrease in expression of RyR2 in PLNKO (Chu et al. 1998) was mimicked, i.e. the 25% decrease in expression of RyR2 in PLNKO was prorated according to the proportion of PLN ablation. The conditions of S2814D^+/+ (first row) and of SD^+/+KO (last row), were used in the experimental protocol shown in Fig. 11. B, SERCA2a activity was stepwise increased and decreased above and below basal levels, respectively. Compensatory modifications of RyR2 were not performed. Therefore, enhanced RyR2 open probability (increased RyR2 Ca²⁺ conductance) was kept augmented by 50% above basal levels all over the range of SERCA2a activities explored. SAP, spontaneous action potentials; DAD, delayed afterdepolarization; EAD, early afterdepolarization; I _NCX, maximum Ca²⁺ current by the forward mode of the NCX. V _m,max, maximum membrane potential depolarization produced by I _NCX current. *Sodium channel threshold was reached at −55.6 mV. Last st. beat: last stimulated beat.

Finally we modelled the effect of either increasing or decreasing SR Ca²⁺ uptake as shown in Table 3B, at different percent increases of SR Ca²⁺ leak (RyR2 conductance). This approach gave rise to the data depicted in the heat plot of Figure 12. The figure indicates that the arrhythmogenic zone extends from an increase in RyR2 conductance of approximately 0.5 up to 6 times over basal values, with different levels of arrhythmogenicity. The hot arrhythmogenic area starts at RyR2 conductances higher than 150%. Values of RyR2 conductance higher than 570% abruptly lessen SAP occurrence. This reduction can be attributed to the reduced SR Ca²⁺ content due to the increase in RyR2 conductance. Indeed with an increased RyR2 conductance of 570%, SR Ca²⁺ content diminished by 65% with respect to no increase in RyR2 conductance. The arrhythmic effect of increasing RyR2 conductance was completely cancelled by either decreasing or increasing SR Ca²⁺ uptake below and above basal values, respectively. As previously discussed, the increase in SR Ca²⁺ uptake provoked a decrease in cytosolic Ca²⁺ during diastole, preventing the enhancement of Ca²⁺ efflux through the NCX_dir and the required Na⁺ influx to depolarize the cell membrane up to the excitability threshold (see Fig. 11 and Table 3B), in accordance with our experimental conclusions. Conversely, the decrease in SR Ca²⁺ uptake reduced SR Ca²⁺ content and leak (see Table 3B), preventing the necessary Ca²⁺ efflux and Na⁺ influx from reaching the excitability threshold (see Table 3B).

Figure 12. 2D‐plot showing SAP occurrence as a function of increased or decreased SR Ca²⁺ uptake and RyR2 increase in conductance .

The figure indicates different levels of arrhythmogenicity. The higher arrhytmogenic area, in red, extends from an increase in RyR2 conductance between approximately 150% and 550% and from increased or decreased SR Ca²⁺ uptake that varies from a maximum of ±15% at the lower RyR2 conductances, to ±5% at the highest RyR2 increases in conductance (550%). Higher increases in RyR2 conductances diminished the arrhythmia propensity due the decreases in SR Ca²⁺ content.

Taken together, the model shows that, similarly to the experimental results, Ca²⁺‐triggered arrhythmias due to an increase in SR Ca²⁺ leak (S2814D^+/+ _sim mice) may be rescued by an increase in SR Ca²⁺ uptake (SD^+/+/KO _sim). The model further shows that a decrease in SR Ca²⁺ uptake can rescue from arrhythmias triggered by an increase in SR Ca²⁺ leak, offering an explanation for previous contradictory results (Stokke et al. 2010). Thus, the model indicates that the success of varying SR Ca²⁺ uptake in rescuing the propensity for arrhythmic events rests on a tight balance between the degree of Ca²⁺ leak and uptake by the SR.

Discussion

The present study demonstrates that PLN ablation reduces the propensity to ventricular arrhythmias caused by constitutive pseudo‐phosphorylation of the Ser2814 site on RyR2, both at the intact animal and the isolated perfused heart levels. Given the correlation between Ca²⁺ waves in single cells and the occurrence of arrhythmias, our results also suggest that the underlying mechanism by which PLN ablation blocks arrhythmias in S2814D^+/+ animals is the abortion of fully propagating SR Ca²⁺ waves which were transformed into non‐propagating, non‐arrhythmogenic events, e.g. mini‐waves. The clinical relevance of these findings is obvious with respect to arrhythmias in HF and reperfusion arrhythmias, since CaMKII expression, activity and phosphorylation of RyR2 at S2814 are all enhanced in patients and animals with HF (Ai et al. 2005; Sossalla et al. 2010) and during reperfusion after cardiac ischaemia (Vittone et al. 2002; Salas et al. 2010; Di Carlo et al. 2014).

Previous experiments in the CaMKII phosphomimetic S2814D^+/+ mice revealed the central role played by CaMKII phosphorylation of RyR2 in causing stress‐induced arrhythmogenic events at the whole‐animal level (van Oort et al. 2010). In the present work these events were associated with an increase in SR Ca²⁺ sparks and leak at the myocyte level (van Oort et al. 2010). We further showed that S2814D^+/+ mice increased not only Ca²⁺ spark frequency but also SR arrhythmogenic Ca²⁺ waves, providing a mechanistic link between CaMKII‐dependent RyR2 phosphorylation and ventricular Ca²⁺ arrhythmias in S2814D^+/+ mice. The increase in SR Ca²⁺ waves would lead to enhanced Ca²⁺ extrusion via the electrogenic NCX. This electrogenic transport generates a depolarizing current (I _ti or transient inward current), which, when sufficiently large, leads to delayed afterdepolarizations (DADs) and potentially triggers ectopic APs and ventricular arrhythmias (Spencer & Sham, 2003; Fujiwara et al. 2008). The present experiments further showed that, as expected, the increase in SR Ca²⁺ uptake produced by PLN ablation dramatically enhanced SR Ca²⁺ leak and sparks in SD^+/+/KO myocytes. In spite of this increase, Ca²⁺ waves were significantly reduced. An increase in SR Ca²⁺ leak associated with a decrease in propagating Ca²⁺ waves indicates a limitation in cytosolic Ca²⁺ diffusion. Our findings clearly show that PLN ablation interrupts cell‐wide propagating Ca²⁺ waves, converting them into non‐propagated events, like mini‐waves or groups of Ca²⁺ sparks. These results support the conclusion that by increasing SR Ca²⁺ uptake, PLN ablation decreases cytosolic Ca²⁺, which would increase cytosolic Ca²⁺ buffer capacity, hampering Ca²⁺ wave propagation and preventing the arrhythmogenic susceptibility produced by constitutive CaMKII‐dependent pseudo‐phosphorylation of the RyR2‐Ser2814 site. Further support to this idea is given by the experiments in which decreasing SR Ca²⁺ uptake by the SERCA2a inhibitor, CPA, converts non‐propagating mini‐waves into full propagating Ca²⁺ waves.

One could argue that since PLNKO myocytes contain fewer ryanodine receptors than WT cells (approximately 25%; Chu et al. 1998), the decrease in arrhythmogenic events observed in SD^+/+/KO mice can be at least partially explained by this reduction in SR Ca²⁺ channels. However, the present and other studies (Santana et al. 1997; Huser et al. 1998) also showed that Ca²⁺ spark frequency, Ca²⁺ spark amplitude and SR Ca²⁺ leak, are higher in PLNKO and SD^+/+/KO than in WT and S2814D^+/+ mice, a result that indicates that the decrease in SR Ca²⁺ waves observed in SD^+/+/KO mice can barely be attributed to the diminished RyR2 abundance.

Most of our studies were performed in SD^+/+/KO mice obtained from homozygous S2814D^+/+ and PLNKO mice. This model enabled us to elucidate the specific effects of PLN ablation on the enhanced SR Ca²⁺ leak typical of RyR2 phosphorylation. We are aware, however, that SD^+/+/KO mice represent a rather unique situation in which RyR2 is maximally and permanently phosphorylated by CaMKII at the S2814 site and SR Ca²⁺ uptake is increased up to the level produced by PLN ablation. This is unlikely to be the situation in the diseases in which CaMKII increases its activity, like ischaemia–reperfusion injury or heart failure. For instance, we know that RyR2 and PLN are phosphorylated by CaMKII only for the first few minutes of reperfusion after ischaemia (Vittone et al. 2002; Salas et al. 2010; Di Carlo et al. 2014). Also, in patients or animals with heart failure, the level of Ser2814 phosphorylation increased by approximately 50–100 % (Ai et al. 2005; Chelu et al. 2009; Sossalla et al. 2010). Previous experiments described that heterozygous S2814D^+/− mice have an arrhythmic pattern similar to homozygous mice (van Oort et al. 2010). Similar results were observed by us in our ex vivo experiments in the intact heart (Fig. 4). In these experiments, PLN ablation was able to rescue the arrhythmic pattern of heterozygous S2814D^+/− mice, suggesting that our data may be also relevant as a model of RyR2 phosphorylation observed in diseases in which CaMKII‐dependent RyR2 phosphorylation is less germane than in S2814D^+/+ mice.

Whereas the crucial role of increasing RyR2 activity in the development of triggered arrhythmias is rather obvious, the role of increasing SR Ca²⁺ uptake has remained uncertain and there is evidence describing both beneficial and detrimental effects of different manoeuvres used to increase SR Ca²⁺ uptake. For instance, and in full agreement with our present findings, it has been shown that improving intracellular Ca²⁺ handling by overexpression of SERCA2a restores contractile function and reduces ventricular arrhythmias in different models of ischaemia–reperfusion injury (del Monte et al. 2004; Prunier et al. 2008). Moreover, SERCA2a overexpression decreases the incidence of aftercontractions in rabbit ventricular myocytes (Davia et al. 2001). Conversely, a recent paper showed that super inhibition of SERCA2a by the human mutation of PLN (R25C‐PLN) is arrhythmogenic (Liu et al. 2015). In conflict with these findings, there is also evidence showing that diminishing SR Ca²⁺ uptake contributes to abrogation of triggered arrhythmias (Lukyanenko et al. 1999; Landgraf et al. 2004). Similarly, reduction of SERCA2a abundance decreases the propensity for Ca²⁺ wave development and ventricular extrasystoles (Stokke et al. 2010). The discrepancy of these results may lie, at least in part, in the fact that the increase in SR Ca²⁺ uptake may regulate SR Ca²⁺ wave development and the possibility of arrhythmia generation in different and even opposite ways: increasing SR Ca²⁺ leak through the increase in SR Ca²⁺ content favours the likelihood of arrhythmogenic Ca²⁺ waves. Moreover, experimental evidence indicates that increasing local SR Ca²⁺ uptake by SERCA2a may facilitate wave transmission via luminal sensitization of RyR2 (the ‘fire–diffuse–uptake–fire’ mechanism) (Maxwell & Blatter, 2012). However, other experiments indicate that increasing SR Ca²⁺ uptake hinders Ca²⁺ wave propagation by altering cytosolic Ca²⁺ buffer capacity (Huser et al. 1998; Bai et al. 2013). These mechanisms may not be necessarily exclusive, making the resultant outcome difficult to predict. To address this problem, we utilized an already proven integrated human cardiac myocyte model to mimic the experimental situation of S2814D^+/+ and SD^+/+/KO cardiac cells. The model was not only able to reproduce the functional behaviour of isolated myocytes under basal conditions, but also to replicate the development of spontaneous events, when either S2814D^+/+ myocytes or intact mice were submitted to stress. Furthermore, the simulation of PLN ablation was able to rescue the arrhythmogenic propensity of S2814D^+/+ myocytes, i.e. the increase in SR Ca²⁺ uptake rescued the arrhythmic consequences associated with enhancement of SR Ca²⁺ leak due to increases in the RyR2 conductance. According to our model, the increase in SR Ca²⁺ uptake is able to rescue the propensity to arrhythmias evoked by a leaky RyR2, by decreasing diastolic Ca²⁺ and therefore limiting the Ca²⁺ efflux through the NCX_dir. This would in turn reduce the Na⁺ influx necessary to depolarize the cell membrane up to the excitability threshold. The model provides in addition a much‐needed understanding for the disparate results seen in the literature, not only those which demonstrated, in agreement with our results, that overexpressing SERCA2a decreased the probability of aftercontractions or of reperfusion arrhythmias (Davia et al. 2001, del Monte et al. 2004, Prunier et al. 2008), but also those showing that the decrease in SR Ca²⁺ uptake can also rescue from triggered arrhythmias (Stokke et al. 2010). Our model supports the view of Stokke et al. indicating that a main cause of the decrease in arrhythmogenic Ca²⁺ wave development after decreasing SR Ca²⁺ uptake rests on a decrease in SR Ca²⁺ load, i.e. the decrease in SR Ca²⁺ uptake, by decreasing SR Ca²⁺ content diminishes SR Ca²⁺ leak to levels below those required for enhancing NCX_dir and the Na⁺ influx necessary to depolarize the cell membrane and to evoke an ectopic beat as depicted in Table 3B.

Limitations of the present study

Our experimental model is limited by the rather unique situation in which RyR2 is maximally and permanently pseudo‐phosphorylated by CaMKII at the S2814 site and SR Ca²⁺ uptake is increased at the level produced by PLN ablation. We addressed this limitation by using heterozygous S2814D^+/− mice and through the development of a human myocyte model that simulates our experimental conditions and allows predicting the effects of different degrees of increases in SR Ca²⁺ uptake on the arrhythmogenic pattern produced by the RyR2 enhanced conductance.

Regarding the mathematical model limitations, we acknowledge that although it is true that the incorporation of two dyadic clefts (DC₁ and DC₂), representing both the L‐type Ca²⁺ channel (LCC) and regenerative RyR2 activation, provides better insight into the CICR mechanism at the dyadic cleft, the model does not contemplate Ca²⁺ propagation throughout the myocyte by the regenerative mechanism of CICR. However, the effects of increasing SR Ca²⁺ uptake on the abrogation of the full propagation of Ca²⁺ waves observed experimentally has its model counterpart represented by the insufficient Ca²⁺ diffusion from the dyadic cleft to the cytosol and the decrease in diastolic Ca²⁺, both produced by the enhanced I _up. Thus, our simplified representation of CRU activation, involving Ca²⁺ propagation from DC₁ to DC₂, did not affect model reproduction of the experimental SAPs in S2814D^+/+ and DADs in SD^+/+/KO double mutants. Another potential limitation of the work is that a human myocyte model, and not a mouse model, was used to simulate experimental results in transgenic mice. This apparent limitation does not seem to involve large discrepancies in model structure, as the general approach used in the most representative mouse model (Bondarenko et al. 2004) is similar to the one presented here. Moreover, although ion current velocities are much shorter in the mouse model due to the elevated heart rate, their shapes are similar to the ones obtained with the present model (Ferreiro et al. 2012; Lascano et al. 2013). Furthermore, the ability of the human myocyte model to adequately reproduce transgenic mouse results would support the conclusions from experiments that cannot be performed in humans.

Additional information

Competing interests

The authors declare that they have no competing interests or conflicts of interest with respect to this work.

Author contributions

This work was done in the laboratory of A.M. A.M, E.L., J.N., C.A.V. and J.P. were involved in the conception and design of the experiments and interpretation of the data. G.M., L.S., J.P., J.I.F., M.N.D.C. and P.C. performed the experiments and were involved in collection, analysis and interpretation of the data. D.F., C.A.V., L.S. and J.I.F. were involved in image analysis. P.G. and M.N.D.C. were involved in the histologic studies. D.S. and M.D.McC. were involved in mutant mice production. X.W. and E.G.K. provided the knock‐in mice and PLNKO mice, respectively, and were involved in critically revising the manuscript. J.N. and E.L. built the model and J.N. developed the MATLAB code. A.M., C.A.V., J.P., L.S. G.M., E.L. and J.N. were involved in drafting the article and/or revising it critically for important intellectual content. All authors approved the final version of the manuscript. All persons designated as authors qualify for authorship and all those who qualify for authorship are listed.

Funding

This work was supported by grants from the National Research Council (Argentina), PIP0890 and PICT 2014–2524 to A.M., HL26057 and HL64018 (to L.K.), National Institutes of Health R01‐HL089598, R01‐HL091947, R01‐HL117641, and R41‐HL129570, the Muscular Dystrophy Association and the American Heart Association 13EIA14560061 (to X.H.T.W.) and K08‐HL130587 to M.D.McC.