Sorcin ablation plus β-adrenergic stimulation generate an arrhythmogenic substrate in mouse ventricular myocytes

Abstract

Sorcin, a penta-EF hand Ca²⁺-binding protein expressed in cardiomyocytes, is known to interact with ryanodine receptors and other Ca²⁺ regulatory proteins. To investigate sorcin’s influence on cardiac excitation-contraction coupling and its role in the development of cardiac malfunctions, we generated a sorcin knockout (KO) mouse model. Sorcin KO mice presented ventricular arrhythmia and sudden death when challenged by acute stress induced by isoproterenol plus caffeine. Chronic stress, which was induced by transverse aortic constriction, significantly decreased the survival rate of sorcin KO mice. Under isoproterenol stimulation, spontaneous Ca²⁺ release events were frequently observed in sorcin KO cardiomyocytes. Sorcin KO hearts of adult, but not young mice developed overexpression of L-type Ca²⁺ channel and Na⁺-Ca²⁺ exchanger, which enhanced I_Ca and I_NCX. Consequently, spontaneous Ca²⁺ release events in sorcin KO cardiomyocytes were more likely to induce arrhythmogenic delayed afterdepolarizations. Our study demonstrates sorcin deficiency may trigger cardiac ventricular arrhythmias due to Ca²⁺ disturbances, and evidences the critical role of sorcin in maintaining Ca²⁺ homeostasis, especially during the adrenergic response of the heart.

1. INTRODUCTION

Sorcin (soluble resistance-related calcium-binding protein) is a 21.6 kDa protein that belongs to the penta-EF hand protein family.^1,2 The protein acquires its name because it was first found to be overexpressed in multidrug-resistant cells.^3–6 However, further studies find that sorcin has a broad expression in cardiomyocytes, neurons, pancreatic β-cells, and vascular smooth muscle,^7–10 clearly transcending its potential role in multi-drug resistance. The binding of Ca²⁺ to sorcin (K_d,Ca~1 μM) moves EF hand 1 and 3, leading to a ~33% increase of solvent-accessible surface areas. The exposure of hydrophobic residues enables sorcin to translocate from cytosol to cellular membranes and to reach its target proteins.^1,2,9,11,12 Immunostaining shows that in the presence of Ca²⁺, sorcin localizes at z-lines close to the T-tubules of adult mouse cardiomyocytes, which enables it to regulate proteins involved in excitation-contraction coupling (e-c coupling).^12,13 One key target protein of sorcin is ryanodine receptor 2 (RyR2), the sarcoplasmic reticulum (SR) Ca²⁺ release channel responsible for providing most of the Ca²⁺ necessary for cardiac contraction.⁷ In single channel recordings of RyR2, sorcin decreases RyR2 activity in a fast (onset of effect is less than ~15 ms),¹² reversible, and dose-dependent manner.¹⁴ Sorcin can also stimulate SR/ER Ca²⁺ ATPase (SERCA) activity by increasing its sensitivity to Ca²⁺ and enhancing the V_max of its Ca²⁺ pumping rate.¹⁵ and significantly increases the sarcolemmal Na⁺-Ca²⁺ exchanger (NCX) in rabbit cardiomyocytes.^16,17 Lastly, sorcin’s effect on the L-type Ca²⁺ channel (LTCC) is more complex: while Meyers et al. find that sorcin accelerates Ca²⁺-dependent inactivation of I_Ca leading to decreased I_Ca integral, Fowler et al. report that sorcin stimulates voltage-dependent inactivation but slows Ca²⁺-dependent inactivation.^13,18

The above in vitro experiments, most of which are conducted using purified exogenous sorcin protein, illustrate sorcin’s effects on its target proteins but fail to profile the integral role of sorcin on e-c coupling in intact cardiomyocytes and whole hearts. As sorcin not only regulates RyR2 channels, but also influences Ca²⁺ entry and SR Ca²⁺ load by interacting with LTCC, SERCA and NCX, it is difficult to predict sorcin’s global effects in intact systems. Previous research attempted to answer this question by using sorcin-overexpressing animals or cardiomyocytes, however, this approach may not be optimal because the method, time, and concentration used to expose sorcin to target proteins are critical determinants of its function. Also, a high concentration of sorcin or its long-term exposure is shown to be detrimental for cell function.^13,16,19,20 Thus, an alternative model is desired to provide novel critical insight into the role of endogenous sorcin in e-c coupling and heart function.

We generated a sorcin knockout (sorcin KO) mouse by ablating the exon 3 of Sri, the only sorcin-encoding gene. The sorcin KO mouse shows impaired insulin release due to ER stress in pancreatic β-cells,⁸ but the effects of sorcin ablation in cardiac cells have not been studied. The sorcin KO mice enabled us to study the role of sorcin on e-c coupling in intact cardiomyocytes, and to model the sorcin loss-of-function mutations found in patients/models with heart diseases.¹¹ By subjecting the sorcin KO mice to acute and chronic stress, we could determine the consequences of sorcin ablation and dysfunction under periods of Ca²⁺ overload. We found that sorcin KO mice were born without apparent phenotype and propagated at the expected Mendelian ratios, but displayed significantly higher incidence of cardiac arrhythmias and sudden death when subjected to stress. Hearts of adult sorcin KO mice (6-month-old), but not young mice (1-month-old) developed NCX and LTCC overexpression, which enhanced I_NCX and I_Ca density. Further, in isolated sorcin KO cardiomyocytes, isoproterenol stimulation triggered high incidence of delayed afterdepolarizations (DADs) accompanied with synchronized Ca²⁺ waves. We report for the first time that sorcin ablation plus β-adrenergic stimulation generate an arrhythmogenic substrate in ventricular myocytes that may trigger lethal arrhythmias due to global Ca²⁺ disturbances.

2. MATERIALS AND METHODS

2.1 Generation of sorcin KO mice

Portions of the murine Sri were obtained by screening the 129SV CITB BAC library (Invitrogen Inc.). A DNA fragment of Sri exon 1, 2, 3, 4 was cloned into pBluescriptSK (Stratagene Inc.) that contained the MC1-HSV-TK cassette. To remove axon 3, which presents in both sorcin isoforms (accession no. NM_001080974.2 and NM_025618.3), a mini targeting vector containing a loxP-flanked exon 3 and a loxP-flanked phosphoglycerol kinase (PGK) promoter-driven NEO-pGHpA cassette was generated. The mini targeting vector was transformed into the recombination-competent DY380 bacteria cells that previously transformed pBluescript-Sri-MC1-HSV-TK. Recombinants integrated the loxP-flanked exon 3 and loxP-flanked pGK promoter-NEO-pGHpA cassettes were selected as targeting vectors. The homologous integration of the mini targeting cassette into pBluescript-Sri-MC1-HSV-TK was confirmed by DNA sequencing.

The targeting vector was electroporated into mouse 129S1/SvImJ ES cells. ES cells integrated the targeting vector were selected by growth on G18. Neo^r, GANC^r clones were picked, expanded, and genomic DNA were isolated from each clone. Correctly targeted colonies, 1C4, 1D12, 1E1, 2B7, 2F5, 2H11, were identified by the appearance of a 2.6k-bp band by using the ³²P autoradiogram 5′ probe. One euploid clone, 1D12, was microinjected into the blastocyst to produce a chimeric founder. The male chimera was mated with 129S1/SvImJ female (F₀). The FloxNeo^−/+ mouse in F₁ generation was selected to mate with EIIa-Cre transgenic mice to excise Neo cassette as well as Sri exon 3 (F₂). Three primers: sorIn3Rev 5′-GAA GGC TGG CAT GGA GTG AAA GCA-3′, sorIn2For 5′-CTG ACC TCA GTC AAC CAG TAA GTA GG-3′, and Neo 5′-CGT TGG CTA CCC GTG ATA TT-3′ were used to determine genotyping of mice. The excision of Sri exon 3 as well as Neo cassette was confirmed by a 453-bp band in PCR, which included a 3′ loxP site, a remnant 168-bp insertion from the targeting vector and neighboring intron, while WT DNA containing exon 3 presented a 1070-bp band. The Sri^+/− mice was backcrossed to 129S1/SvImJ mice for 10 generations to eliminate Cre recombinase. Finally two Sri^+/− mice were crossed to get the Sri^−/− mice.

All animal experiments were approved by the University of Michigan Institutional Animal Care and Use Committee

2.2 Western blot

Expression of RyR2, LTCC, NCX and SERCA were measured in 6-month and 1-month-old sorcin KO mice as described before.²¹ Heart frozen in liquid nitrogen was pulverized by pestle, and homogenized in lysis buffer containing 0.9% NaCl, 10 mM Tris-HCl pH 6.8, 20 mM NaF, 2 μM leupeptin, 100 μM phenylmethylsulphonyl fluoride, 500 μM benzamidine, 100nM aprotinin at 4°C. The sample was centrifuged at 1000 ×g for 10 minutes, and the supernatant was collected. The protein concentration was measured by Bradford method. For Western blots, 50 μg of lysate suspended in Laemmlii buffer was separated by SDS-PAGE in 4–20% TGX or AnyKD precast gels (Bio-Rad). Proteins were transferred to PVDF membranes at 25V for 16–18h at 4°C. Then, membranes were blocked in PBS-T containing (mM) 3 KH₂PO₄, 10 Na₂HPO₄, 150 NaCl, pH 7.2–7.4, 0.1% Tween 20) plus 5% dried skim milk. Proteins were probed with the following primary antibodies: sorcin (1:2000 custom), SERCA (1:1000, ab2861, Abcam), NCX (1:200, ab6495, Abcam), Cav1.2 (1:200, ACC-003, Alomone), RyR2 (1:2000, MA3-925, ThermoFisher). After the membrane being washed three times by PBS-T, membranes were incubated with goat anti-mouse-HRP (1:1000, 31437, Thermo) or goat anti-rabbit-HRP (1:2000, 31463, Thermo). Protein-antibody reactions were detected by using SuperSignal Femto ECL reagent (Thermo), and imaged by the ChemiDoc MP apparatus (Bio-Rad). Band intensity was analyzed by the ImageLab software (Bio-Rad).

2.3 Confocal Ca²⁺ imaging

Ca²⁺ activities, including Ca²⁺ spark, field stimulation-stimulated Ca²⁺ transient, SR Ca²⁺ load, and RyR2-mediated diastolic Ca²⁺ leak, were recorded by the LSM510 Meta inverted confocal microscope (Carl Zeiss) with a 40×/1.2 N. A water immerse objective. Cardiomyocytes were incubated with 10μM Fluo-4 AM, a cell-penetrating Ca²⁺ indicator with Ca^{2 +} binding affinity (K_d) of ~335 nM, at 37°C for 5min. Then cells were washed and kept in fresh bath solution. Binding to Ca²⁺, Fluo-4 presents an increase in fluorescence, which is excited at the wavelength of 488nm and recorded at wavelength >505nm. Ca²⁺ images were collected by the one-direction line scan of the long axis of cell, at the speed of 3.072ms/line. For β-adrenergic stimulation, 300nM isoproterenol was applied in bath solution. Ca²⁺ sparks, RyR2-mediated diastolic Ca²⁺ leak, Ca²⁺ transients, SR Ca²⁺ load, SERCA rate and NCX rate were measured as described in data supplement. For simultaneous recording of action potential and Ca²⁺ transient, cardiomyocytes were dialyzed with 0.2mM fluo-4 pentapotassium salt via pipette solution. Ca²⁺ transients were recorded by the Olympus IX51 inverted microscopy system with a 40×oil immerse objective.

2.4 Patch clamp

Whole-cell patch clamp experiments were conducted by using an Axopatch 700B and a Digidata 1440A digitizer (Axon Instruments) at room temperature. I_Ca,, I_NCX, Action potential and action potential-triggered Ca²⁺ transient were measured as described in data supplement.

2.5 Electrocardiographic (ECG) recording

Mice were anesthetized by 5% isoflurane inhalation vaporized in oxygen at a flow rate of 0.8–1 L/min, and maintained at anesthetized state by 1.5–2% isoflurane. Mice were placed in a supine position on a heating pad of 37 °C, and needle ECG electrodes were placed under skin to record leads I and II ECG. After 10 min of baseline ECG recording, acute stress will be induced by intraperitoneal injection of epinephrine 2 mg/kg + caffeine 120 mg/kg. ECG was analyzed as described before.²² After recording, mice were sacrificed by cervical dislocation.

2.6 Statistics

All data are presented as mean±SEM. Student t test, rank-sum test, two-way repeated-measures ANOVA, Holm-Sidak test, χ² test, Kaplan-Meier test were carried when appropriate to determine statistical significance. p < 0.05 was considered statistically significant.

Detailed methods for ventricular cardiomyocytes isolation, Ca²⁺ imaging (Ca²⁺ sparks, Ca²⁺ transient, SR Ca²⁺ load, RyR-mediated diastolic Ca²⁺ leak, NCX rate, SERCA rate), patch clamp recording (I_Ca, I_NCX, action potential), echocardiographic recording, transverse aortic constriction, Langendorff perfusion, and histological staining are described in data supplement.

3. RESULTS

3.1 Generation of sorcin knockout mice

This is the first description of the generation of the sorcin KO mouse. Figure 1A shows the strategy used to generate the targeting vector and the resultant targeted allele. 820-bp from Sri, which include exon 3 and neighboring introns, were excised by Cre-Lox recombination. The knockout of exon 3 of Sri was confirmed by a 453-bp band in the PCR of sorcin KO tissues (included the 3′ LoxP site, a remnant 168-bp insertion from the targeting vector, and neighboring intron), instead of the 1070-bp band detected in WT samples (Fig. 1B). Western blot using a sorcin antibody¹² demonstrated complete absence of the 21.6 kDa sorcin band in sorcin KO hearts (Fig. 1C). Crossbreeding of sorcin heterozygous mice yielded offspring with the expected Mendelian ratios of WT (sorcin^+/+ 22%), heterozygous (sorcin^−/+ 52%) and homozygous (sorcin^−/− 26%) mice (Fig. 1D). Vertical planes of hearts indicated no structural alterations in sorcin KO hearts compared with WT (Fig. 1E), and echocardiographic parameters were all normal in adult (5~7-month-old) sorcin KO hearts (Suppl. Fig. 1). Thus, sorcin KO mice at baseline display no overt cardiac structural or functional alterations.

Figure 1. Generation of sorcin KO mice.

(A) Strategy for the generation of sorcin KO mice by homologous recombination. a. The sorcin targeting vector containing LoxP-flanked exon 3 of SRI, LoxP-flanked Neo-cassette and the HSV-TK cassette; b. WT SRI containing seven exons (four exons are shown in graph); c. Homologous recombination between the endogenous SRI and sorcin targeting vector results in a DNA carrying LoxP-flanked SRI exon 3 and Neo-cassette. d. Sorcin KO DNA is created by the Cre-excision of SRI exon 3 and Neo-cassette. (B) PCR confirmation of WT and sorcin KO mice. Instead of the 1070-bp band seen in WT, a 453-bp band is detected in sorcin KO allele. (C) Western blot with sorcin antibody confirms the presence of sorcin in 3 WT hearts (first three bands) and its absence in 3 sorcin KO mice. (D) Crossbreeding of sorcin heterozygous mice yields the expected proportion of WT (sorcin^+/+), heterozygous (sorcin^−/+) and homozygous knockout (sorcin^−/−) mice. (E) Vertical planes of hearts indicate no structural alterations in sorcin KO heart compared with WT.

3.2 Arrhythmias and sudden death in sorcin KO hearts and mice under acute or chronic stress

The absence of overt structural or functional alterations in the sorcin KO mice was surprising, given the prominent role of sorcin in modulating proteins of e-c coupling.^12,13 However, lifespan of non-manipulated, freely-fed sorcin KO mice was shorter than WT littermates (18-month survival rate: KO 70% vs. WT 92.5%, N=40 in each group; Kaplan-Meier test p<0.05) (Fig. 2A), suggesting there were concealed risk factors in sorcin KO mice uncovered by increase of age. To assess whether sorcin KO mice were in a pseudo-equilibrium state, we subjected them to chronic and acute stress. We first conducted transverse aortic constriction (TAC) to induce chronic stress in mice. After surgery, the survival rate in sorcin KO mice was significantly lower than that of WT (KO 52% vs. WT 88% after 3 weeks, N=24 in each group, Kaplan Meier test p<0.01,) (Fig. 2B). The substantial difference in survival rate indicated that the fragility of sorcin KO mice was exacerbated by TAC. Since no structural remodeling and difference in heart function were observed in sorcin KO mice after TAC (Suppl. Fig. 1), we assessed whether cardiac arrhythmias were part of the underlying mechanisms leading to accelerated death. We conducted acute stress tests in anesthetized 6-month-old mice, by i.p. injection of Epinephrine 2 mg/kg + caffeine 120 mg/kg.²³ Continuous recording of ECG activity showed that, whereas the WT group had few arrhythmic events, five out of 6 sorcin KO mice presented higher incidence of premature ventricular contractions (PVCs), bidirectional ventricular tachycardia (BVT), and ventricular tachycardia (VT) within 5 min after injection (Fig. 2C–E). One third of sorcin KO mice, but no WT, experienced sudden cardiac arrest during recording (Fig. 2F). The duration of ventricular arrhythmia was significantly longer in sorcin KO mice (356±151s) than WT (16.6±6.9s, p<0.05) (Fig. 2G), and two sorcin KO mice presented BVT for longer than 1000s. In agreement with in vivo test, freshly explanted and Langendorff-perfused sorcin KO hearts, although without significant difference in their response to isoproterenol (300 nM)-induced left ventricular developed pressure (Suppl. Fig. 2C), systolic (maximum) dp/dt (Suppl. Fig. 2D) and diastolic (minimum) dp/dt (Suppl. Fig. 2E), displayed increased incidence of PVCs and bigeminy/trigeminy compared to WT hearts (Suppl. Fig. 2A–B, F–H). Young (1-month-old) sorcin KO mice also presented higher incidence of premature contractions, bigeminy, and bidirectional ventricular tachycardia than WT littermates (Suppl. Fig. 3A–C). The duration of ventricular arrhythmias, however, was not different between young sorcin KO (19.15±4.44s) and WT mice (16.41±5.78s) (Suppl. Fig. 3D). These results showed that sorcin KO mice, while apparently normal under basal conditions, are thrown into tachyarrhythmias under acute sympathetic stimulation and fare worse than WT when subjected to chronic stress. Therefore, the shorter lifespan and increased TAC mortality of global sorcin KO mice may be due to cardiac alterations.

Figure 2. Life span, ECG and survival rate of mice.

(A) Long-term follow up of lifespan of sorcin KO and WT mice. N=40/group. Kaplan-Meier test p<0.05. (B) Survival rate of 6-month-old mice after the conduction of transverse aortic constriction (TAC) N=24/group. Kaplan-Meier test p<0.01. (C) ECG recordings of 6-month-old anesthetized mice after the administration of epinephrine (2 mg/kg) + caffeine (120 mg/kg). First trace: representative ECG of a WT mouse after 5 min of injection; second trace: bidirectional ventricular tachycardia (BVT) in a sorcin KO mouse after 3 min of injection; third trace: premature ventricular contractions (PVCs) in a sorcin KO mouse after 5 min of injection; fourth trace: ventricular tachycardia (VT) in a sorcin KO mouse after 3 min of injection. Each trace is 1s recording. (D) Percentage of mice that had ventricular arrhythmia after injection. (E) Number of ventricular arrhythmia episodes in WT and sorcin KO mice, per arrhythmia cocktail challenge. (F) Percentage of mice that presented sudden cardiac arrest after injection. (G) Duration of arrhythmias in WT and sorcin KO mice. N=6 in each group. ×, average value; *, p<0.05 vs. WT, #, p=0.051 vs. WT.

3.3 Electrophysiological remodeling in adult sorcin KO cardiomyocytes

The increased vulnerability of sorcin KO mice to arrhythmias and death under stress was intriguing and warranted examination of underlying mechanisms. We first determined whether the loss of sorcin led to altered expression of other proteins involved in e-c coupling by using Western blots. NCX and LTCC (Cav1.2) were ~2.2-fold and 2.4-fold, respectively, increased in adult (6-month-old) sorcin KO hearts compared to WT, while the expression of SERCA and RyR2 was unaltered (Fig. 3A, B). Interestingly, young (1-month-old) sorcin KO hearts did not present overexpression of LTCC and NCX as adult hearts. Instead, the expression of NCX in young KO hearts was modestly lower than WT (Fig. 3C, D). No other changes were noted.

Figure 3. Expression of Ca²⁺ regulatory proteins.

(A–B) Representative Western blots (A) and quantification (B) for sorcin, SERCA2a, RyR2, NCX, and LTCC expression in 6-month-old WT and sorcin KO hearts. (C–D) Representative Western blots (C) and quantification (D) for protein expression in 1-month-old WT and sorcin KO hearts. N=5 in each group. *, p<0.05 vs. WT; ** p<0.01 vs. WT.

To assess whether the overexpression of NCX and LTCC in adult sorcin KO hearts had a functional correlation, we measured NCX (Fig. 4) and LTCC (Fig. 5) activities. NCX activity was first determined from the rate of decay of the caffeine-induced Ca²⁺ transient (K_caffeine), defined as 1/τ (where τ = the time constant of the decay phase). Adult sorcin KO cardiomyocytes had comparable K_caffeine as WT under basal condition, but exhibited faster K_caffeine under 300 nM isoproterenol stimulation, reflecting faster Ca²⁺ reuptake rate by NCX (Fig. 4A, B). To directly evaluate this parameter, we recorded NCX current (I_NCX) in patch-clamped cells, in which free [Ca²⁺]_i was set at ~210 nM by adding 6 mM Ca²⁺ and 10 mM ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA) in the pipette solution. NCX current,generated by exchanging 3 Na⁺ ions for 1 Ca²⁺ ion, was determined by subtracting the Ni²⁺-sensitive currents (Fig. 4C).^16,17 NCX inward current is the forward-mode depolarizing current (Ca²⁺ out, Na⁺ in) at membrane potential lower than −60 mV and turns to reverse-mode (Ca²⁺ in, Na⁺ out) during depolarization. Adult sorcin KO and WT cardiomyocytes had comparable I_NCX under basal condition (Fig. 4D) but KO cardiomyocytes had significantly larger outward current (clamped at 50~80 mV) and inward current (clamped at −120~−80 mV) under isoproterenol stimulation (Fig. 4C, E). The increased K_caffeine as well as I_NCX in sorcin KO cardiomyocytes suggested that although the stimulating effect of sorcin on NCX^16,17 was not present in sorcin KO cells, NCX activity was enhanced under isoproterenol stimulation. It is likely that the overexpression of NCX protein detected by Western blots allowed for a more robust [Ca²⁺]_i extrusion, resulting in enhanced NCX activity observed in isoproterenol-stimulated sorcin KO cardiomyocytes.

Figure 4. NCX activity.

(A) Representative Ca²⁺ imaging and plot of caffeine-induced Ca²⁺ transient in WT and sorcin KO cardiomyocytes. WT τ_{caf f}= 2369 ms, KO τ_caff= 1758 ms. (B) NCX rate (K_NCX). WT-basal n=10, KO-basal n=9; WT-iso n=12, KO-iso n=13. (C) Protocol for recording I_NCX and representative I_NCX of WT and sorcin KO cardiomyocytes under isoproterenol stimulation. (D) I–V curve of I_NCX under basal condition. WT-basal n=9, KO-basal n=8 (E) I–V curve of I_NCX under isoproterenol stimulation. WT-iso n=9, KO-iso n=9. *, p<0.05 vs. WT.

Figure 5. L-type Ca²⁺ current.

(A–B) I–V curve of I_Ca under basal condition (A), and under isoproterenol stimulation (B). (C) τ_fast of I_Ca inactivation at 0 mV under basal condition. Insert: representative curves of I_Ca at 0 mV under basal condition. (D) τ_fast of I_Ca inactivation at −10 mV under isoproterenol stimulation. Insert: representative curves of I_Ca at −10 mV under isoproterenol stimulation. WT-basal n=11, KO-basal n=14, WT-iso n=10, KO-iso n=10. *, p<0.05 vs. WT.

We then measured L-type Ca²⁺ current (I_Ca) of adult sorcin KO and WT cardiomyocytes. Sorcin KO cardiomyocytes presented modestly but statistically significant increased peak I_Ca compared to WT under basal condition (Two-way RM ANOVA p<0.05) (Fig. 5A). The robust stimulating effect of isoproterenol blurs most of the difference in peak I_Ca, although at −20 mV sorcin KO cardiomyocytes still displayed significantly higher I_Ca than WT (Holm-Sidak test p<0.05) (Fig. 5B). As ours and other groups’ research have shown sorcin does not influence amplitude of I_Ca,^12,13 the increased peak I_Ca in sorcin KO cardiomyocytes corresponded to LTCC overexpression.

Although I_Ca peak increased in adult sorcin KO cardiomyocytes under basal condition, the Ca²⁺-dependent fast inactivation of I_Ca (τ_fast) significantly accelerated at the same time (Fig. 5C) so the total I_Ca (Σ_ICa) was not significantly changed. On the other hand, under isoproterenol stimulation, τ_fast of I_Ca in adult sorcin KO cardiomyocytes was slower than that of WT, which favors more Ca²⁺ entry during e-c coupling (Fig. 5D).

3.4 Ca²⁺ transients and SR Ca²⁺ load in adult sorcin KO cardiomyocytes

We previously showed that sorcin (3 μM) dialyzed into cardiomyocytes via pipette significantly decreased intracellular Ca²⁺ transient amplitude and e-c coupling gain,¹² underscoring sorcin as a potent inhibitor of I_Ca-triggered RyR2 activity. To investigate the effect of endogenous sorcin on I_Ca-triggered Ca²⁺ release during contraction, we measured Ca²⁺ transients by pacing Fluo-4 loaded cardiomyocytes at 1Hz for 20 s (Fig. 6A). During pacing, the Ca²⁺ transient amplitude gradually decreased until SR Ca²⁺ reached a steady level. We used the averaged peak value of the last 10 Ca²⁺ transients of a train of stimulation to represent the Ca²⁺ transient amplitude.

Figure 6. Intracellular Ca²⁺ transients.

(A) Representative recordings of Ca²⁺ transients and SR Ca²⁺ load of WT and sorcin KO cardiomyocytes under basal condition and isoproterenol stimulation. Cardiomyocytes were paced 20 times and 10 mM caffeine was perfused to the cell after pacing to measure SR Ca²⁺ load. (B–D) Ca²⁺ transient amplitude (B), SR Ca²⁺ load (C), and SERCA rate (D) of WT and sorcin KO cardiomyocytes under basal condition and isoproterenol stimulation. WT-basal n=13, KO-basal n=14, WT-iso n=13, KO-iso n=16. *, p<0.05 vs. WT; ***, p<0.001 vs. WT.

Mirroring the increased I_Ca amplitude in sorcin KO cardiomyocyte and sorcin’s inhibitory effect on I_Ca-induced RyR2 activity, sorcin KO cardiomyocytes presented modest but statistically higher Ca²⁺ transient amplitude than WT under basal conditions, but the difference disappeared under isoproterenol stimulation, as higher diastolic Ca²⁺ leak in sorcin KO cardiomyocytes (see below) dampened Ca²⁺ release (Fig. 6B). The total SR Ca²⁺ load in resting cardiomyocyte, measured as caffeine-induced Ca²⁺ release after pacing, was not different between sorcin KO and WT cardiomyocytes (Fig. 6C). Sorcin KO cardiomyocytes presented slower Ca²⁺ pumping rate by SERCA under isoproterenol stimulation (Fig. 6D), which bodes well with previous results showing that sorcin stimulates SERCA function.¹⁵

3.5 Isoproterenol reveals increased Ca²⁺ leak in sorcin KO cardiomyocytes

We tested whether sorcin ablation leads to increased RyR2 activity by measuring Ca²⁺ sparks after pacing. After a 20-s train of field stimulation at 1Hz, both WT and sorcin KO cardiomyocytes had few Ca²⁺ sparks under basal conditions. Notably, however, sorcin KO cells exhibited higher frequency of Ca²⁺ sparks under isoproterenol stimulation (Fig. 7A, E). The spark amplitude, full duration at half maximum (FDHM), and half decay time were all significantly higher in sorcin KO cardiomyocytes (Fig. 7B, F–H). Since SR Ca²⁺ load of sorcin KO cardiomyocytes was not significantly changed (Fig. 6C), the result suggests increased SR Ca²⁺ cycling in KO cells, and RyR2 hyperactivity during diastole.

Figure 7. Ca²⁺ sparks and leak in cardiomyocytes.

(A) Ca²⁺ sparks of WT and sorcin KO cardiomyocytes under isoproterenol stimulation. Sparks were measured after 20 times of 1Hz pacing. (B) 3D plots of representative Ca²⁺ sparks of WT and sorcin KO cardiomyocytes. (C–D) Representative plots of RyR2-mediated SR Ca²⁺ leak of WT (C) and sorcin KO (D) cardiomyocytes under isoproterenol stimulation. Cardiomyocytes were perfused in 0Na⁺-0Ca²⁺ solution. Then, tetracaine (1 mM) was added to block Ca²⁺ leak through RyR2. After 10s, tetracaine was washed out by 0Na⁺-0Ca²⁺ solution. The fluorescence difference before and after tetracaine application (marked by red dashed line) reflects RyR-mediated diastolic Ca²⁺ leak. (E–H) Ca²⁺ spark frequency of cardiomyocytes under basal condition and isoproterenol stimulation (E), Ca²⁺ sparks amplitude (F), full duration at half maximum (FDHM) (G), and half decay time (H) under isoproterenol stimulation. WT-basal n=16, KO-basal n=16, WT-iso n=16, KO-iso n=32 cells. (I) RyR2-mediated diastolic SR Ca²⁺ leak under basal condition and isoproterenol stimulation. WT-basal n=11, KO-basal n=15; WT-iso n=12, KO-iso n=12 cells. *, p<0.05 vs. WT or sorcin KO; **, p<0.01 vs. WT under basal condition; ***, p<0.001 vs. WT under isoproterenol stimulation.

To ascertain more categorically whether RyR2 activity is increased in KO cells, we measured RyR2-mediated diastolic SR Ca²⁺ leak by an alternative method, as described by Shannon et al.²⁴ Cardiomyocytes were perfused in 0Na⁺-0Ca²⁺ solution, which stopped Ca²⁺ exchange between extracellular and intracellular compartments as LTCC and NCX were blocked. Then, tetracaine (1mM) was perfused to block Ca²⁺ leak through RyR2. The fluorescence difference before and after tetracaine perfusion reflected diastolic Ca²⁺ leak (Fig. 7C, D). Sorcin KO cardiomyocytes had diastolic SR Ca²⁺ leak comparable to WT under basal condition but displayed higher diastolic SR Ca²⁺ leak under isoproterenol stimulation (Fig. 7C, D, I). The difference was maintained after the Ca²⁺ leak was normalized to the SR Ca²⁺ load.

3.6 Prolonged action potential duration and delayed afterdepolarizations in adult sorcin KO cardiomyocytes

Western blots and patch clamp recordings suggested electrophysiological remodeling of I_Ca and I_NCX in adult sorcin KO cardiomyocytes, which may play a key role in maintaining SR Ca²⁺ load and contractility in isoproterenol-stimulated sorcin KO cardiomyocytes. Although electrophysiological remodeling plays important compensatory effects, it may turn sorcin KO mice more susceptible to arrhythmias because: 1) The enhanced inward I_NCX during diastole can lead to membrane depolarization and further, may trigger arrhythmogenic DADs if fired by spontaneous Ca²⁺ release;²⁵ 2) both increased inward I_Ca and I_NCX may prolong AP repolarization and induce early afterdepolarizations (EADs).^26,27 We recorded action potentials (APs) and Ca²⁺ transients simultaneously²⁸ to investigate if the electrophysiological remodeling described above brings about arrhythmogenic events in sorcin KO cardiomyocytes. Under basal condition, while most sorcin KO cardiomyocytes had normal APs and Ca²⁺ transients (Suppl. Fig. 4A, B), 4 out of 12 sorcin KO cells presented random EADs, which were accompanied by spontaneous Ca²⁺ release events between pacing (Supp. Fig. 4C). In the aggregate, however, EAD frequency was not statistically different between sorcin KO and WT cardiomyocytes (p=0.128) (Suppl. Fig. 4D). With the appearance of EADs, 6 out of 12 KO cells had APD₉₀ longer than 300 ms (Suppl. Fig. 4E). APD at 70, 75, 80, and 90 percent of repolarization were significantly increased in sorcin KO cells compared to WT (p<0.001, WT n=11, KO n=12), although APD variability was large (Suppl. Fig. 4E, 4F). In summary, most of sorcin KO cardiomyocytes presented normal APs and Ca²⁺ transients under basal conditions. Although some sorcin KO cardiomyocytes presented EADs and extremely long APD, the incidence did not reach statistical significance and so the KO mice did not present ventricular arrhythmias nor long Q-T interval at baseline. Nonetheless, the phenomenon might be a sign of disturbed Ca²⁺ homeostasis in sorcin KO cardiomyocytes.

Under isoproterenol stimulation, sorcin KO cardiomyocytes had significantly higher frequency of spontaneous Ca²⁺ waves than WT (Fig. 8A–C). Seventy-eight percentage (47/60) of spontaneous Ca²⁺ waves in sorcin KO cardiomyocytes led to membrane depolarizations (i.e., DADs) that were of sufficient amplitude to reach threshold for I_Na activation, thus triggering full-blown APs (Fig. 8B insert, D). By contrast, only 35% (6/17) of spontaneous Ca²⁺ waves developed into triggered activities in WT cardiomyocytes, while the rest only led to mild (<20mV) membrane depolarization (Fig. 8A, D). As sorcin KO cardiomyocytes had higher incidence of spontaneous Ca²⁺ waves, and their Ca²⁺ waves led to APs more readily, the frequency of triggered activities in sorcin KO cardiomyocytes was dramatically higher than WT (Fig. 8E). Sorcin KO cardiomyocytes also presented significantly longer APDs at 70, 80, and 90 percent of repolarization under isoproterenol stimulation (Fig. 8F, G). DADs and EADs were fully prevented in KO cells dialyzed with10 mM fast Ca²⁺ chelator 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA), which completely quenched Ca²⁺-induced Ca²⁺ release during e-c coupling (Suppl. Fig. 5A, C). However, 37.5% of KO cells still presented DADs with coordinated “Ca²⁺ spikes” when dialyzed with 10 mM Ca²⁺ chelator EGTA, whose Ca²⁺ binding rate was ~100 times slower than BAPTA (Suppl. Fig. 5B, C). Both BAPTA and EGTA significantly decreased action potential duration of sorcin KO cardiomyocytes (Suppl. Fig. 5D). The results suggested that [Ca²⁺]_i is necessary to trigger electrical alterations in the sarcolemma of sorcin KO cardiomyocytes.

Figure 8. Action potential and Ca²⁺ transients under isoproterenol stimulation.

(A) Representative WT cardiomyocyte that presented spontaneous Ca²⁺ release events but not triggered activity under isoproterenol stimulation. Upper panel: action potential; arrows indicate the time when the electrical stimulus was applied; middle panel: fluorescent recording of Ca²⁺ transient; under panel: line plot of Ca²⁺ transient. (B) Representative sorcin KO cardiomyocyte that presented triggered activities under isoproterenol stimulation. Insert at right: expanded image of the spontaneous Ca²⁺ wave in the red frame, showing how a localized Ca²⁺ release (arrow) ignited a full Ca²⁺ transient and triggered activity. (C) Frequency of spontaneous Ca²⁺ waves in quiescent state. (D) Percentage of spontaneous Ca²⁺ waves that developed into triggered activity under isoproterenol stimulation: 3 out of 14 and 47 out of 60 spontaneous Ca²⁺ waves in WT and sorcin KO cardiomyocytes, respectively, developed into triggered activities. (E) Frequency of triggered activities in quiescent state. (F) Scatter plot of under isoproterenol stimulation. Each dot represents a different cell. (G) APD at APD₉₀ indicated percent of repolarization under isoproterenol stimulation. WT-basal n=11, WT-iso n=12, KO basal n=10, KO-iso n=14 cells. *, p<0.05 vs. WT; **, p<0.01 vs. WT.

4. DISCUSSION

Sorcin, a 21.6-kDa Ca²⁺ binding protein originally discovered in multi-drug resistance cells, is also expressed in ventricular cardiomyocytes of mammalian species, including humans.^1,29 This is a basic observation that compels a fundamental question: what does sorcin do in cardiac cells? We previously found that the majority of sorcin is localized in bands of high intensity running along the width of the cardiomyocyte, interspaced at about ~1.7 μm, and overlapping with RyR2 located in z-lines.^12,13 Thus, a great portion of sorcin appears localized in the dyadic junction, where crucial steps of e-c coupling occur. Also, using single RyR2 channels or isolated ventricular cells, we found that sorcin rapidly binds to RyR2 and directly inhibits single channel activity, translocates from soluble to membrane-bound protein targets in a Ca²⁺-dependent manner, and attenuates Ca²⁺ sparks and Ca²⁺ transients in saponin-permeabilized cardiomyocytes.¹² These and other findings affirmed a role for sorcin in modulation of Ca²⁺ release by RyR2s, but its integral role in Ca²⁺ homeostasis and e-c coupling of the heart remained unclear. Here, we generated a mouse line with genetic ablation of Sri, the gene encoding for sorcin in humans and animal. These mice were born without apparent phenotype, propagated at the expected Mendelian ratio, and most survived into adulthood (Fig. 1). However, hearts from young (1-month-old) sorcin KO mice, although showing no apparent electrical remodeling, were nonetheless thrown into transient tachyarrhythmia episodes upon β-adrenergic stimulation (Supp. Fig. 3), indicating that sorcin absence per se predisposes mice to arrhythmias. As sorcin KO mice developed into adulthood, an increase in LTCC and NCX expression (Fig. 3) and activity (Figs. 4, 5) suggests cardiac contractions become increasingly dependent on sarcolemma Ca²⁺ entry and extrusion. Thus, the prolonged absence of sorcin plus its resultant electrical remodeling appear to lead the adult sorcin KO mice to a precarious condition, because a) their survival rate was lower than WT, even under resting condition (Fig. 2A), b) aortic banding that was tolerated by WT mice caused significantly more attrition in sorcin KO mice (Fig. 2B), c) stress tests promptly induced long-lasting tachyarrhythmias and sudden cardiac arrest (Figs. 2C–G, Suppl. Fig. 2). All these results, therefore, support the hypothesis that sorcin is a stabilizing component of the e-c coupling machinery of the heart, independent of or in addition to other roles that it may play in other cells.

4.1 Sorcin KO phenotype and resemblance to CPVT

Sorcin ablation brings about significant alterations in Ca²⁺ homeostasis of ventricular myocytes, and this becomes more apparent under β-adrenergic stimulation: isolated cardiomyocytes stimulated with isoproterenol display Ca²⁺ sparks, Ca²⁺ oscillations, aftercontractions, and DAD-elicited triggered activity (Figs. 7, 8). Bidirectional ventricular tachycardia, characteristic of Ca²⁺-overloaded cells and a hallmark of CPVT,^30,31 is also present in sorcin KO mice (Fig. 2, Suppl. Fig. 3). The general features of the sorcin KO phenotype are thus strikingly similar to those of catecholaminergic polymorphic ventricular tachycardia (CPVT), which is typically associated with RyR2 hyperactivity. In both cases, a cardiac phenotype is “silent” functionally and structurally and goes unnoticed until sympathetic stimulation triggers catastrophic Ca²⁺ release and life-threatening arrhythmias (Figs. 2, 7, 8).³⁰ We postulate, therefore, that RyR2 channels are among the most prominent molecular targets of sorcin in ventricular cardiomyocytes and the most affected by its absence.

4.2 Amplification of sorcin effect by sympathetic stimulation

During normal β-adrenergic stimulation, RyR2s are especially active due to increased I_Ca and higher SR Ca²⁺ load. Ca²⁺ release from a cluster of RyR2s into the dyadic nanospace could theoretically increase local [Ca²⁺] up to ~100 μM,^32,33 which is sufficient to translocate sorcin to SR membrane targets. By decreasing RyR2 activity, sorcin conceivably attenuates Ca²⁺-induced Ca²⁺ release and helps prevent Ca²⁺-triggered membrane depolarization. On the other hand, sorcin has been shown to be PKA-phosphorylated at residue S178,^2,14,15 and others¹⁵ and we (unpublished data) have shown that phosphorylation facilitates translocation of sorcin from cytosolic to membrane compartments. Therefore, the high [Ca²⁺] in the dyadic space and sorcin phosphorylation during β-adrenergic stimulation amplify sorcin’s regulatory effect on RyR2, its key target.

4.3 Hypothesis for an integrated role of sorcin in e-c coupling

While initial experiments to elucidate sorcin’s role in the heart demonstrated that sorcin inhibits Ca²⁺ release by RyR2 channels, equally compelling are the findings that sorcin increases NCX activity,^16,17 accelerates LTCC inactivation,^13,34 and stimulates SERCA Ca²⁺ uptake.¹⁵ Therefore, sorcin forms macromolecular complexes with at least four key players of e-c coupling. By summing the individual effects exerted by sorcin on the above-mentioned Ca²⁺ transporters, it seems obvious that the unified role for sorcin in cardiac cells is to restrict I_Ca-triggered RyR2 activity and avoid Ca²⁺ accumulation in cytosolic compartments. However, what is not immediately obvious is that such an integrated effect is not possible in the long run, because it would eventually deplete cardiac cells of Ca²⁺ by restricting Ca²⁺ entry and release and stimulating Ca²⁺ extrusion. In a delimited and finely regulated system such as a ventricular cell, inward and outward Ca²⁺ fluxes must be equal, and in this zero-sum game, small disturbances are rapidly counteracted, lest profound alterations in electrical excitability, e-c coupling and signal conduction ensue.³⁵ Therefore, a process that mitigates adrenergic-induced Ca²⁺ buildup is necessary, and we propose that sorcin is an integral part of such a process. Sorcin’s Ca²⁺ removal role may be especially active during β-adrenergic stimulation of the heart, right after cardiac cells have transitioned to Ca²⁺ overload due to a rapid surge of plasmalemmal Ca²⁺ entry. It is during this time window that Ca²⁺ fluxes are briefly unbalanced because the β adrenergic-induced Ca²⁺ overload, while maximizing contractions, promotes also the generation of DADs (by spontaneous Ca²⁺ release) and EADs (by prolonging APD). If removal of adrenergic-induced Ca²⁺ buildup by sorcin is indeed necessary, then sorcin KO mice should develop arrhythmias mainly during β-adrenergic stimulation, or under stress. In favor of this hypothesis, others and we found that adrenergic stimulation of ventricular cells increases translocation of sorcin from cytosolic to membrane-bound targets¹⁵ and that sorcin KO hearts and living mice are readily thrown into tachyarrhythmias under sympathetic simulation. Thus, the overall hypothesis that emerges from this and previous studies is that sorcin “sweeps” cytosolic compartments from excess Ca²⁺ during periods of Ca²⁺ overload.

4.4 General model to explain arrhythmias in young and adult sorcin KO mice

From the above, we propose a model that integrates the major mechanisms governing the induction of arrhythmias in young and adult sorcin KO mice. First, the model must consider fundamental differences in e-c coupling between immature (neonatal and young mice) and adult ventricular cells, with the former relying more on external Ca²⁺ entry due to incompletely formed t-tubules and dyads,³⁶ while the latter obtaining from the SR (by Ca²⁺-induced Ca²⁺ release) most of the Ca²⁺ needed for contractions.³⁷ Second, the model must also consider data showing increased LTCC and NCX density in adult, but not in young, sorcin KO cardiomyocytes, as observed here (Fig. 3). In both cases, however, β-adrenergic stimulation increases contractility by increasing Ca²⁺ cycling. Hence, in the absence of sorcin, a modulator that normally decreases Ca²⁺ entry, promotes Ca²⁺ extrusion, stimulates SR Ca²⁺ reuptake, and decreases SR Ca²⁺ fluxes, the most plausible mechanism to induce ventricular arrhythmias in the immature hearts is one in which the adrenergically-stimulated cytosolic Ca²⁺ surges generate an inward depolarizing current of sufficient magnitude to bring membrane potential to threshold, generating DADs and subsequent triggered activity. This mechanism would be akin to that postulated to occur in CPVT³⁰ and does not require electrical remodeling (normal expression of e-c coupling proteins). On the other hand, abnormal expression of e-c coupling proteins in the mature sorcin KO hearts suggests gradual adaptation of the e-c coupling machinery to the absence of sorcin: the increased SR Ca²⁺ leak and decreased SERCA pumping may eventually prompt sorcin KO cardiomyocytes to become more dependent on external Ca²⁺ to activate contractions and reload the SR, hence, sorcin KO cells must increase their pool and/or their activity of LTCC (Figs. 3, 5). The increased Ca²⁺ influx (Ca²⁺ in, Na⁺ out) through overexpressed NCX (Figs. 3, 4) during the upstroke of action potential also makes contribution to maintaining SR Ca²⁺ release. Since a larger Ca²⁺ entry and SR Ca²⁺ leak cannot be sustained without a comparable Ca²⁺ extrusion mechanism, the 2.2-fold increase in the density of sarcolemmal NCX (Figs. 3, 4), who works in forward mode (Ca²⁺ out, Na⁺ in) during diastole, represents such concomitant Ca²⁺ extrusion mechanism. In this hypothetical scheme, adult sorcin-null ventricular cells appear to tilt the balance between sarcolemmal and SR Ca²⁺ fluxes to maintain a pseudo-equilibrium that is stable under basal conditions but favors arrhythmias under adrenergic stimulation.

4.5 Limitations of our study

Our general hypothesis above, while viable and accommodating of our major observations here, requires further testing and experimentation that go beyond the scope of this study. First, as it happens with most transgenic animals, ablation of a single gene seldom results in “clean” knockout of the target protein, because such targeted gene and protein interact with a variety of (known and unknown) partners that constitute an indivisible functional network. In the case of the sorcin KO, we performed a targeted analysis of what we considered the most relevant sorcin-interacting proteins and found alterations in some, but not all, binding partners. Wider changes in protein expression would not be surprising, and they may even force key revisions of our hypothesis. Second, sorcin is encoded by a single gene (Sri) and expressed in several tissues, therefore, it is possible that the cardiac arrhythmias or the decreased lifespan of sorcin KO mice reported here may have extra-cardiac origins. For example, we previously reported that the absence of sorcin in pancreatic β-cells triggers ER stress,⁸ and it is possible that the resultant glucose intolerance may have impacted cardiac function.³⁸ Even though our experiments using isolated, Langendorff-perfused hearts indicate that cardiac arrhythmias are present in the absence of autonomic nerve system or hormonal input, we cannot completely discard this possibility. Lastly, the majority of sorcin localizes in immunocytochemical experiments to the dyadic space and near RyR2,^12,13 but a small portion (~18%) does not. The first study describing expression of sorcin in cardiac cells⁷ found anti-sorcin reactivity in mitochondria. The potential impact of mitochondria dysfunction to the cardiac arrhythmias described here, if any, requires an in-depth study of its own.

4.6 Summary

In summary, data here and published studies make it increasingly evident that sorcin modulates e-c coupling in the heart, at the very least through its well-established effect on RyR2 channels, but also through newly-found molecular targets. Although our general hypothesis may require further testing, data obtained here with the sorcin KO mouse does lay some solid foundations to postulate a unified role for sorcin in cardiac cells: sorcin is a Ca²⁺-dependent molecular switch that is activated during cytosolic Ca²⁺ surges, especially during β-adrenergic stimulation and, by binding to at least four key e-c coupling proteins, it removes Ca²⁺ from cytosolic compartments and avoids Ca²⁺-dependent arrhythmias.

Highlights.

• Sorcin KO mice present arrhythmias and sudden death under acute or chronic stress.

• Adult sorcin KO cardiomyocytes develop electrophysiological remodeling.

• Isoproterenol reveals spontaneous Ca²⁺ release events in sorcin KO cardiomyocytes.

• Sorcin KO cardiomyocytes present high incidence of triggered activities.

GLOSSARY

AP

action potential

BAPTA

1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid

BVT

bidirectional ventricular tachycardia

CPVT

catecholaminergic polymorphic ventricular tachycardia

DAD

delayed afterdepolarization

EAD

early afterdepolarization

e-c coupling

excitation-contraction coupling

EGTA

ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid

FDHM

full duration at half maximum

LTCC

L-type Ca²⁺ channel

NCX

Na⁺-Ca²⁺ exchanger

PVC

premature ventricular contraction

RyR2

ryanodine receptor type 2

SERCA

SR/ER Ca²⁺ ATPase

Sorcin KO

sorcin knockout

SR

sarcoplasmic reticulum

TAC

transverse aortic constriction

VT

ventricular tachycardia