Complex Arrhythmia Syndrome in a Knock-In Mouse Model Carrier of the N98S Calm1 Mutation

Abstract

Background:

Calmodulin mutations are associated with arrhythmia syndromes in humans. Exome sequencing previously identified a de novo mutation in CALM1 resulting in a p.N98S substitution in a patient with sinus bradycardia and stress-induced bidirectional ventricular ectopy. The objectives of the present study were to determine if mice carrying the N98S mutation knocked into Calm1 replicate the human arrhythmia phenotype and to examine arrhythmia mechanisms.

Methods:

Mouse lines heterozygous for the Calm1^N98S allele (Calm1^N98S/+) were generated using CRISPR/Cas9 technology. Adult mutant mice and their wildtype littermates (Calm1^+/+) underwent electrocardiographic monitoring. Ventricular de- and repolarization was assessed in isolated hearts using optical voltage mapping. Action potentials and whole-cell currents as well as [Ca²⁺]_i were measured in single ventricular myocytes using the patch-clamp technique and fluorescence microscopy, respectively. The microelectrode technique was employed for in situ membrane voltage monitoring of ventricular conduction fibers.

Results:

Two biologically independent knock-in mouse lines heterozygous for the Calm1^N98S allele were generated. Calm1^N98S/+ mice of either sex and line exhibited sinus bradycardia, QT_c interval prolongation and catecholaminergic bidirectional ventricular tachycardia. Male mutant mice also showed QRS widening. Pharmacological blockade and activation of β-adrenergic receptors rescued and exacerbated, respectively, the long QT phenotype of Calm1^N98S/+ mice. Optical and electrical assessment of membrane potential in isolated hearts and single left ventricular myocytes, respectively, revealed β-adrenergically induced delay of repolarization. β-adrenergic stimulation increased peak density, slowed inactivation and left-shifted the activation curve of I_Ca.L significantly more in Calm1^N98S/+ vs Calm1^+/+ ventricular myocytes, increasing late I_Ca.L in the former. Rapidly paced Calm1^N98S/+ ventricular myocytes showed increased propensity to delayed afterdepolarization-induced triggered activity, whereas in situ His-Purkinje fibers exhibited increased susceptibility for pause-dependent early afterdepolarizations. Epicardial mapping of Calm1^N98S/+ hearts showed that both reentry and focal mechanisms contribute to arrhythmogenesis.

Conclusions:

Heterozygosity for the Calm1^N98S mutation is causative of an arrhythmia syndrome characterized by sinus bradycardia, QRS widening, adrenergically mediated QT_c interval prolongation and bidirectional ventricular tachycardia. β-adrenergically induced I_Ca.L dysregulation contributes to the long QT phenotype. Pause-dependent early afterdepolarizations and tachycardia-induced delayed afterdepolarizations originating in the His-Purkinje network and ventricular myocytes, respectively, constitute potential sources of arrhythmia in Calm1^N98S/+ hearts.

Introduction

Mutations in the calmodulin (CaM)-encoding genes CALM1, CALM2 and CALM3 have been linked to ventricular arrhythmias manifesting with features of catecholaminergic polymorphic ventricular tachycardia (CPVT) and/or long QT syndrome.^1–7 All three CaM genes elaborate an identical, 148-amino acid, calcium binding protein. Calcium binding alters CaM interaction with target proteins, modifying their function. Nyegaard et al identified a de novo heterozygous missense mutation in exon 5 of the CALM1 gene, resulting in substitution of asparagine at position 98 with serine (N98S, start methionine denoted residue 1).⁸ The patient was a 4-year-old girl presenting with sinus bradycardia, stress-induced bidirectional ventricular ectopy, and sudden cardiac death.⁸ A de novo N98S mutation of a different CaM gene (CALM2) was identified in an unrelated 5-year-old boy diagnosed with long QT syndrome,¹ indicating that an identical amino acid substitution in two distinct CaM genes can give rise to divergent clinical phenotypes. The N98S mutation has been investigated in different in vitro models that demonstrated that it increases single-channel open probability of the cardiac ryanodine receptor (RyR2),⁹ attenuates L-type Ca²⁺ channel inactivation,¹⁰ and reduces activation of small conductance, Ca²⁺-activated potassium (SK) channels.¹¹ It has thus been inferred that arrhythmias develop as a consequence of Ca²⁺ cycling abnormalities (e.g., higher Ca²⁺ wave activity) and/or electrical disturbances (e.g., prolonged action potential duration).^5,9–12 However, experimental evidence which causally links this mutation to occurrence of arrhythmias is lacking. Accordingly, the goals of the present study were (1) to develop a knock-in mouse model carrier of a N98S Calm1 mutation that is the mouse equivalent of the human N98S CALM1 mutation,¹³ (2) to provide proof of the concept that this mutation is directly causative of the type of stress-induced ventricular arrhythmias seen in human patients, and (3) to investigate mechanisms contributing to the mutant phenotype.

Methods

A detailed description of the methods used including those for statistical analyses is provided in the Supplemental Data. The study was approved by the Institutional Biosafety Committee, Institutional Animal Care and Use Committee of the Indiana University and the Methodist Research Institute, Indianapolis, Indiana. All procedures were in accordance with the ethical standards of the responsible committees (institutional and national) on human experimentation.

Statistical analyses.

In all cases, data distribution was tested using Shapiro–Wilk’s normality test. The 2-tailed, independent or paired sample t-test was used for comparison of normally distributed data. The 2-tailed independent sample Mann–Whitney’s U-test was used for comparison of not-normally distributed data. One way Repeated Measures Analyses of Variance or Kruskal Wallis Mean Rank Sum test followed by appropriate post hoc analyses were used for comparisons of repeated measurements. Analyses of Covariance was used to test the effect of genotype on the linear dependence of ln QT on ln RR. A P value of less than or equal to 0.05 was considered significant. Values are presented as either mean ± SEM, or median with interquartile range. Experimenters were blinded to the genotype of the animals. SigmaPlot (Systat Software, Inc., San Jose, CA) was used for all statistical analyses.

Results

Generation of Calm1^N98S/+ mice

We used the CRISPR/Cas9 gene editing approach to generate knock in mice heterozygous for the N98S mutation in the Calm1 gene (Calm1^N98S/+ mice; see Supplemental Data and Figure I in the Supplement for details). Two independent mouse lines were generated, designated line 2 and line 18. Calm1^N98S/+ mice were born at expected Mendelian ratios, exhibited normal growth, and survived to adulthood. Structure and contractile function of Calm1^N98S/+ hearts were normal (Table I and Figure II in the Supplement). There was no evidence for fibrosis. Relative quantification of Calm1, Calm2 and Calm3 transcripts in left ventricle of adult Calm1^+/+ and Calm1^N98S/+ mice by use of the 2^−ΔCT method demonstrated that expression levels of each of the three Calm genes remained unchanged in Calm1^N98S/+ mice compared to their expression in littermate Calm1^+/+ mice (Table II in the Supplement), suggesting that the presence of a single N98S Calm1 allele did not significantly alter transcription of any of the three Calm genes in mouse heart.

Phenotypic characterization

To evaluate the effects of the N98S Calm1 mutation on cardiac electrical function in vivo, we examined the ECGs of conscious mice derived from continuous recordings via an implantable telemeter. Representative ECG examples are presented in Figure 1A, revealing sinus bradycardia, QT prolongation and QRS widening in the Calm1^N98S/+ mice. Twenty-four-hour plots of the mean RR, QT_c and QRS intervals were generated from long-term ECG recordings (see Supplemental Data and Figure III in the Supplement for a description of the method that was used to calculate QT_c intervals). As shown in Figure IV A in the Supplement, Calm1^N98S/+ mice exhibited increases in all three parameters for most of the 24-h period. Mean values of 24-h RR, QT_c and QRS intervals were significantly larger in Calm1^N98S/+ versus Calm1^+/+mice (Figure 1B) of either line. While the mean 24-h RR interval was significantly larger in line 18 vs line 2 Calm1^N98S/+ mice (left upper panel of Figure 1B), the differences for mean 24-h QT_c and QRS intervals between Calm1^+/+ and Calm1^N98S/+ mice were of similar magnitude (Figure 1B). Thus, Calm1^N98S/+ mice of two independently generated knock-in lines exhibited almost identical electrocardiographic phenotypes, characterized by sinus bradycardia, long QT and QRS widening.

Figure 1. Calm1^N98S/+ mice exhibit sinus bradycardia, QRS widening, QT_c prolongation and catecholaminergic ventricular tachycardia.

A, Top two rows show representative ECG traces recorded during the dark and light cycle in a littermate control (blue) and a Calm1^N98S/+ mouse (red) from line 2. Traces in the third and fourth row show signal-averaged ECGs obtained by overlaying all consecutive cycles recorded within a 5- to 10-s interval, using QRS maximum or QRS minimum for the alignment. Lower row shows superimposed QRS complexes at expanded time scales. B, Dot plots of RR, QRS and QT_c intervals from littermate Calm1^+/+ and Calm1^N98S/+ mice of line 2 and line 18. Each dot represents the 24-h average in a single animal. Horizontal lines superimposed on dots depict mean and SEM. P values by One Way ANOVA followed by Student-Newman-Keuls Method for multiple comparisons (RR and QT_c) or Kruskal-Wallis One Way ANOVA on Ranks followed by Dunn’s Method for multiple comparisons (QRS). Line 2: 8 Calm1^+/+and 11 Calm1^N98S/+mice; Line 18: 9 Calm1^+/+ and 10 Calm1^N98S/+ mice. C, Calm1^N98S/+ mice display bidirectional ventricular tachycardia (BVT). Representative telemetric ECG recordings from a conscious Calm1^+/+ and Calm1^N98S/+ mouse obtained before and shortly after intraperitoneal injection of epinephrine (2 mg per kg body weight) and caffeine (120 mg per kg body weight). In the Calm1^N98S/+ mouse, an episode of sustained BVT was preceded by ventricular bigeminy (*) and spontaneously reverted to sinus rhythm. No ventricular arrhythmias were observed in the Calm1^+/+ mouse. D, Complete dissociation of P waves (arrowheads) from QRS complexes during an episode of induced BVT. E, Polymorphic ventricular tachycardia in a male Calm1^N98S/+ mouse. * marks transition from BVT into polymorphic VT and ^# denotes conversion to sinus rhythm. F, Prevalence of sustained (>15 s) BVT induced by co-administration of epinephrine and caffeine. * P < 0.001 versus Calm1^+/+ by Fisher’s Exact test. G, Incidence, cumulative duration, cycle length and CI/RR ratio of episodes of sustained (≥ 15 s) BVT during a 30-min period of post-injection telemetric ECG recordings in conscious Calm1^N98S/+ mice from line 2 and line 18. Dot plots with mean ± SEM shown as horizontal lines. N = 7, line 2; N = 8, line 18; *P=0.041 by t test. VT indicates ventricular tachycardia. CI indicates coupling interval and RR indicates RR interval.

Calm1^N98S/+ mice are susceptible to inducible ventricular arrhythmia

Treadmill exercise followed by epinephrine injection did not evoke arrhythmias (see Supplementary Data). Combined administration of caffeine and epinephrine has been used to elicit ventricular arrhythmias in different mouse models of CPVT.^14,15 Accordingly, we administered epinephrine (2 mg per kg body weight i.p.) and caffeine (120 mg per kg body weight i.p.) to 13 mice of line 2 (6 Calm1^+/+ and 7 Calm1^N98S/+) and 15 mice of line 18 (7 Calm1^+/+ and 8 Calm1^N98S/+) while continuously monitoring their ECG via telemetry. Representative ECGs recorded from Calm1^+/+ and Calm1^N98S/+ littermates (line 2) before and following a single caffeine and epinephrine injection are shown in Figure 1C. The Calm1^+/+ mouse developed transient sinus tachycardia, but ventricular tachyarrhythmias were not observed during the 30-min period of post injection telemetric ECG recording. Overall, 6 out of 7 line 2 Calm1^+/+ mice receiving epinephrine and caffeine did not have ventricular tachyarrhythmias within the 30-min post injection intervals, while one Calm1^+/+ mouse exhibited repetitive episodes of ventricular bigeminy (duration ranging from 5 s to 324 s) post injection. In contrast, administration of epinephrine and caffeine to the Calm1^N98S/+ mouse triggered ventricular bigeminy which was followed by a 25-s episode of tachycardia displaying beat-to-beat alternations in the polarity of the QRS axis as well as atrio-ventricular dissociation (Figure 1C and D), consistent with bidirectional ventricular tachycardia (BVT), a hallmark of CPVT.¹⁵ In one Calm1^N98S/+ mouse, BVT transitioned into polymorphic VT (Figure 1E). Overall, 7 out of 7 Calm1^N98S/+ mice of line 2 exhibited at least one episode of pharmacologically induced, sustained BVT, indicating high penetrance (P < 0.001 versus Calm1^+/+; Figure 1F). All VT episodes reverted spontaneously to sinus rhythm. Identical results for arrhythmia inducibility were obtained in mice of line 18 (Figure 1F). Cumulative incidence, cycle length, and coupling interval ratio (see Figure IV B in the Supplement) of induced BVT episodes were indistinguishable between Calm1^N98S/+ mice of the two lines, whereas their cumulative duration was significantly longer in line 18 mutants (Figure 1G). Finally, the cumulative incidence of premature ventricular contractions (PVCs) per 24 h as well as the 24-h arrhythmia scores were significantly elevated in Calm1^N98S/+ versus Calm1^+/+ mice of either mouse line, but their respective mean values were similar between genotypically identical mice in the two lines (Figure IV C in the Supplement).

We also compared the electrical phenotype between conscious female and male mice of line 2. The results are summarized in Figure V in the Supplement. Like their male counterparts, female Calm1^N98S/+ mice exhibited sinus bradycardia and QT_c interval prolongation, but QRS duration was not prolonged in female Calm1^N98S/+ mice (Figure V A and B in the Supplement). Mean 24-h QT_c intervals in female Calm1^+/+ and Calm1^N98S/+ mice were significantly shorter than those in their male counterparts (Figure V B in the Supplement). Co-administration of epinephrine and caffeine induced sustained BVT in 7 out of 8 female and 7 out of 7 male Calm1^N98S/+ mice, but in none of the Calm1^+/+ males or females (Figure V C and D in the Supplement). Median incidence of BVT was significantly lower in female than male Calm1^N98S/+ mice, whereas mean values for cumulative duration, cycle length and coupling interval of BVT episodes were similar among male and female Calm1^N98S/+ mice of line 2 (Figure V E in the Supplement). The 24-h arrhythmia scores were not significantly different between Calm1^+/+ males and females or between Calm1^N98S/+ males and females. Together, these results indicate that the arrhythmia phenotype presents with extraordinary similarity between Calm1^N98S/+ mice from two independently generated knock-in lines, irrespective of sex. Accordingly, all following experiments utilized male mice from one line only (line 2).

β-adrenergic stimulation induces long QT phenotype in Calm1^N98S/+ mice

Modest increases in the QT_c and QRS intervals were found in Calm1^N98S/+ mice (Figure 1), compatible with abnormalities in both ventricular repolarization and conduction. To quantify the extent to which heterozygosity for the Calm1^N98S allele alters these electrophysiological parameters, we next performed epicardial optical voltage mapping of the anterior surface of 11 Langendorff-perfused Calm1^N98S/+ and 11 Calm1^+/+ hearts. Representative color activation maps during right atrial pacing at a cycle length of 120 ms are presented in Figure 2A. In both genotypes, activation pattern is characterized by 2 epicardial breakthroughs that originate on the free walls of the right and left ventricles near the apex (asterisks). These breakthroughs occur simultaneously and form wave fronts that propagate towards the bases of the hearts, fully activating the field of view within 6 ms and 8 ms in the Calm1^+/+ and Calm1^N98S/+ hearts, respectively. To determine whether conduction defects within the ventricular myocardium contribute to QRS widening in the Calm1^N98S/+ mice, epicardial conduction patterns and velocities were determined. Figure 2B shows activation maps obtained from the anterior aspect of a Calm1^+/+ and a Calm1^N98S/+ heart while pacing the ventricles directly at a cycle length of 100 ms. The anisotropic conduction pattern in the Calm1^N98S/+ heart was similar to that of Calm1^+/+. Figure 2C shows dot plots of CV_max and CV_min for all Calm1^+/+ and Calm1^N98S/+ hearts tested. No statistically significant differences were found between the two genotypes. These data indicate that the QRS widening seen in Calm1^N98S/+ mice is unlikely to result from conduction disturbances in the ventricular myocardium.

Figure 2. Delayed repolarization in ventricles of Calm1^N98S/+ mice.

A, Isochrone maps of epicardial activation during atrial pacing at a cycle length of 120 ms. Asterisks denote epicardial breakthrough sites on the anterior surface near the apex of the left (LV) and right ventricle (RV). Depolarization spread from the apex to the base. Sequential depolarizations are depicted in different colors according to the color legend shown in the left. Isochronal lines are 1 ms apart. B, Epicardial ventricular conduction velocities. Ventricular activation isochrone maps of the anterior aspect of a Calm1^+/+ and Calm1^N98S/+ heart. The anisotropic propagation in the ventricle enables measurements of maximal and minimal conduction velocities (CV_max and CV_min). C, Dot plots of CVs. Numbers indicate mean and SEM in m/s from 6 Calm1^+/+ and 4 Calm1^N98S/+ hearts. No statistically significant differences between the two genotypes were found. D, Left: Representative optical action potentials from the anterior left ventricular epicardium of littermate Calm1^+/+ and Calm1^N98S/+ mice during atrial pacing at a cycle length of 120 ms. Middle three panels: exemplar maps of the repolarization time points at APD₃₀, APD₅₀ and APD₈₀. Colored scale bars are shown in the inserts. Right: superimposition of the same optical action potentials shown in the left panels. E, Dot plots showing APD₃₀, APD₅₀ and APD₈₀ of littermate control and Calm1^N98S/+ mice. Black horizontal lines depict mean ± SEM. * P ≤ 0.02 by t-test. N = 11 per genotype. F, Exacerbation of long QT phenotype upon exposure to epinephrine (E) and caffeine (C) combined. Exemplar ventricular repolarization (APD₈₀) isochrone maps of the anterior aspect of a Calm1^+/+ and a Calm1^N98S/+ heart at baseline, during E and C exposure, and following washout. Pacing cycle length, 120 ms. G, Scatter plots of APD₈₀. Black lines are mean ± SEM from 8 Calm1^+/+ and 6 Calm1^N98S/+ hearts. P values by Repeated Measures ANOVA followed by Tukey test for intragroup analyses; for intergroup analyses, a t-test (for comparisons at baseline and during washout) and Mann-Whitney Rank Sum test (for comparison during epinephrine) were used.

Exemplar color repolarization maps obtained from atrial paced hearts (120 ms cycle length) of either genotype are shown in Figure 2D. Ventricular action potential durations (APD) at 30%, 50% and 80% repolarization were longer in the Calm1^N98S/+ versus Calm1^+/+ heart. The mean values for APD₃₀, APD₅₀ and APD₈₀ were modestly, yet significantly, larger in Calm1^N98S/+ hearts (Figure 2E), in agreement with the modest QT_c prolongation observed in Calm1^N98S/+ mice in vivo. These results support the notion that abnormal ventricular repolarization underlies the long QT phenotype of Calm1^N98S/+ mice. While combined exposure of isolated hearts of either genotype to epinephrine (1.6 μmol/L) and caffeine (5 mmol/L) failed to induce ventricular arrhythmias, it reversibly prolonged APD₈₀ significantly more in Calm1^N98S/+ versus Calm1^+/+ hearts as shown in Figure 2F and G. Treatment of two Langendorff-perfused Calm1^N98S/+ hearts with the non-selective β-adrenergic receptor agonist isoproterenol (100 nM) quantitatively replicated the increases in APD₈₀ seen in the presence of epinephrine and caffeine (Figure VI in the Supplement), while it did not significantly affect ventricular activation patterns or conduction velocities (not shown). These data indicate that activation of cardiac β-adrenergic receptors mediates excessive APD prolongation in Calm1^N98S/+ hearts treated with epinephrine/caffeine.

We next recorded action potentials from single ventricular myocytes under current clamp. Exemplar recordings obtained at 1 and 7 Hz pacing rates show similar action potential shape and time course in Calm1^+/+ and Calm1^N98S/+ myocytes (Figure 3A). Figure 3B shows dot plots for APD₃₀, APD₅₀ and APD₉₀ for all cells tested. No statistically significant differences were found between the 2 genotypes. We also observed no differences in the resting membrane potential, action potential peak, or action potential upstroke velocity (Table III in the Supplement). These data raise the possibility that cardiomyocyte non-autonomous factors are responsible for the long QT phenotype of Calm1^N98S/+ mice. The optical mapping data obtained from Calm1^N98S/+ hearts suggested a critical role of altered responsiveness of the ventricular action potential to β-adrenergic receptor stimulation in mediating QT_c interval prolongation. Accordingly, we next compared isoproterenol (50 nmol/L)-induced changes in action potential properties of Calm1^N98S/+ ventricular myocytes with myocytes from Calm1^+/+ littermates. Representative action potential traces are presented in Figure 3C. While isoproterenol had no significant effect on the action potential repolarization in Calm1^+/+ myocytes, it markedly and reversibly delayed late repolarization (APD₉₀; Figure 3D) in Calm1^N98S/+ myocytes. Isoproterenol did not significantly change early repolarization (APD₃₀ and APD₅₀), resting membrane potential, action potential peak or maximal upstroke velocity in cells of either genotype (Table IV in the Supplement). Isoproterenol-induced, excessive APD prolongation impeded 1:1 stimulus-action potential responses in the majority of isolated Calm1^N98S/+ myocytes at 7 Hz pacing, preventing reliable APD measurements. We did not observe early afterdepolarizations in control or isoproterenol-treated myocytes of either genotype. In control conditions, none of 13 Calm1^+/+ and 14 Calm1^N98S/+ ventricular myocytes (distributed among 3 Calm1^+/+ and 4 Calm1^N98S/+ mice, respectively) developed delayed afterdepolarizations (DAD) at 1 Hz pacing, whereas one myocyte of either genotype generated one DAD each at 7 Hz pacing (P = 1.0 by Fisher Exact test). The fraction of isoproterenol-treated Calm1^N98S/+ myocytes exhibiting DAD-triggered action potentials at 7 Hz pacing (8 out of 10 cells distributed among 2 mice; Figure 3E) was significantly higher than that of isoproterenol-treated Calm1^+/+ myocytes (1 out of 7 cells; 2 mice; P = 0.015 by Fisher Exact test), whereas no significant differences were found between isoproterenol-treated myocytes of the 2 genotypes during 1 Hz pacing (1 out of 7 Calm1^+/+ myocytes versus 0 out of 10 Calm1^N98S/+ myocytes; P = 0.412 by Fisher Exact test).

Figure 3. β-adrenergic stimulation induces a long QT phenotype in Calm1^N98S/+ mice.

A, Left panels: superimpositions of representative action potentials electrically recorded from single myocytes isolated from the left ventricular free wall of a Calm1^+/+ (blue) and a Calm1^N98S/+ (red) heart. Cells were paced at 1 and 7 Hz. Right panels: single action potentials at 1 and 7 Hz pacing at expanded time scales. B, Dot plots of APD₃₀, APD₅₀ and APD₉₀ in single ventricular myocytes paced at 1 and 7 Hz. Black lines denote mean ± SEM; 11 cells each from 3 Calm1^+/+ mice and from 4 Calm1^N98S/+ mice. * P = 0.004 vs. Calm1^N98S/+ 1 Hz by paired t-test. C, Superimpositions of representative action potentials recorded from a Calm1^+/+ and a Calm1^N98S/+ ventricular myocyte at baseline, during isoproterenol (50 nM; Iso) treatment, and during washout. Action potentials were evoked by 2-ms square current injections delivered at 1 Hz. D, Summary of the effect of isoproterenol (Iso) on APD₉₀ in Calm1^+/+ and Calm1^N98S/+ ventricular myocytes; 14 cells from 5 Calm1^+/+ mice and 10 cells from 4 Calm1^N98S/+ mice. * P < 0.01 by Repeated Measures ANOVA followed by Tukey test for pairwise multiple comparisons. Numbers denote mean ± SEM in ms. Base: baseline. E, Original recordings of a delayed afterdepolarization (*) and delayed afterdepolarization-triggered action potentials (arrows) in a Calm1^+/+ and a Calm1^N98S/+ ventricular cardiomyocyte, respectively. Traces show the terminal 6 paced action potentials from a 10-s train at 7 Hz in the presence of 50 nM isoproterenol. F, Summary of the effects of propranolol (1 mg/kg) or isoproterenol (0.17 mg/kg) on heart rate. Horizontal black lines denote mean ± SEM. Propranolol (Prop): N = 8, Calm1^+/+ mice; N = 14, Calm1^N98S/+ mice. Isoproterenol (Iso): N = 7, Calm1^+/+; N = 7, Calm1^N98S/+. P values by unpaired and paired t-test. Base: baseline. G, Representative QT interval changes in conscious male Calm1^+/+ and Calm1^N98S/+ mice following intraperitoneal injection of propranolol (1 mg/kg) or isoproterenol (0.17 mg/kg). Shown are signal-averaged ECGs obtained by overlaying all consecutive cycles recorded within a 5- to 10-s interval, ca. 20 min after injection. Arrows mark QT intervals. H, Summary of the effects of propranolol and isoproterenol on QT_c interval. Horizontal black lines denote mean ± SEM. Propanolol (Prop): N = 8, Calm1^+/+ mice; N = 14, Calm1^N98S/+ mice. Isoproterenol (Iso): N = 7, Calm1^+/+; N = 7, Calm1^N98S/+. P values by unpaired and paired t-test. Base: baseline.

Finally, we assessed the effects of pharmacological β-adrenergic receptor inhibition and activation on heart rate and durations of QRS and QT_c intervals in conscious mice implanted with wireless ECG transmitters. Administration of propranolol (a non-selective β-adrenergic receptor blocker; 1 mg/kg i.p.) or isoproterenol (0.17 mg/kg i.p.) significantly reduced and increased, respectively, mean heart rate in mice of either genotype, but significant heart rate differences between genotypes persisted in the presence of either drug (Figure 3F). Propranolol rescued the QT_c interval prolongation of Calm1^N98S/+ mice, while isoproterenol aggravated the relatively modest baseline differences in QT_c interval between Calm1^+/+ and Calm1^N98S/+ mice (Figure 3G and H). Neither drug significantly affected QT_c interval in Calm1^+/+ mice and QRS duration in either genotype (not shown). No ventricular tachycardias were observed in either genotype. Taken together, the in vitro and in vivo data indicate that manifestation of the long QT phenotype of Calm1^N98S/+ mice requires stimulation of cardiac β-adrenergic receptors.

Epicardial voltage mapping of ventricular arrhythmias

Our current-clamp measurements revealed delayed afterdepolarization-mediated triggered activity in rapidly paced Calm1^N98S/+ ventricular cardiomyocytes in the presence of isoproterenol suggesting involvement of focal mechanisms in the genesis of catecholaminergic arrhythmias in the mutant mice. The concomitant, isoproterenol-induced APD prolongation, if occurring in a spatially and/or temporally discordant manner, could also facilitate reentrant activity. We used epicardial optical voltage mapping of isolated perfused Calm1^+/+ and Calm1^N98S/+ hearts to investigate arrhythmia mechanisms associated with elevated extracellular calcium and/or catecholaminergic stimuli. Exposure to epinephrine and caffeine did not evoke ventricular arrhythmias in atrial paced (120 ms cycle length), blebbistatin-treated Calm1^N98S/+ (n=6) or Calm1^+/+ hearts (n=8). Other isolated, non-paced hearts were perfused with blebbistatin-free Tyrode’s solution containing 3.6 mmol/L CaCl₂ plus 100 nmol/L isoproterenol (4 hearts per genotype) and subjected to programmed electrical stimulation and burst stimulation. None of the Calm1^+/+ hearts showed spontaneous or inducible arrhythmias. In contrast, one Calm1^N98S/+ heart each had spontaneous or burst-induced episodes of VT as shown in Figure 4 and Figure VII in the Supplement, respectively. Each VT episode in Figure 4 is triggered by a ventricular ectopic beat, sharing the same morphology as the single ectopic beat (complex 2), suggesting a common focal origin. Indeed, voltage maps of the VT episodes revealed focal discharges on the right ventricular free wall which coincided with the emergence of the triggering ectopic beats on the ECG, as shown exemplarily for the first NSVT episode. Their concentric breakthrough patterns were similar to those emanating from the right ventricular free wall during sinus rhythm (Figure 4A). The RV focal discharges initiated NSVT for 2–4 cycles. The last QRS complexes of each episode were associated with the emergence of an RV (complex 6) or LV focal discharge, exhibiting breakthrough patterns similar to those seen during sinus rhythm. Figure VII in the Supplement shows an episode of a non-sustained VT following a 1-s train of stimuli at 100 Hz in another Calm1^N98S/+ heart. This VT episode also exhibited a combined mechanism of right ventricular focal trigger and macro-reentry. BVT was not observed in either one of the 2 Calm1^N98S/+ hearts. In addition, non-paced Calm1^N98S/+ hearts perfused with 200 nmol/L isoproterenol and elevated Ca²⁺ (n=4), 1.6 μmol/L epinephrine, 5 mmol/L caffeine and elevated Ca²⁺ (n=2), or 1.6 μmol/L epinephrine plus 5 mmol/L caffeine (n = 1), but not undergoing programmed or burst stimulation, showed no arrhythmias. Taken together, these observations suggest that focal discharges initiated reentry in catecholaminergic arrhythmias of Calm1^N98S/+ mice.

Figure 4. Ventricular arrhythmia in a Calm1^N98S/+ heart.

A, Volume-conducted ECG and epicardial activation map in sinus rhythm. Asterisks denote concentric breakthroughs on the anterior right and left ventricular free walls. B, Volume-conducted ECGs and epicardial voltage maps in the presence of 100 nM isoproterenol and 3.6 mmol/L extracellular Ca²⁺. QRS complex numbers on ECG correspond to voltage map sequences with the same numbers. The ECG shows a junctional escape rhythm with AV dissociation and a QRS morphology similar to that during SR, a single premature ventricular complex (PVC) and 4 episodes of non-sustained VT (NSVT). Maps show epicardial breakthrough pattern during junctional escape rhythm (sequence #1), RV focal discharge (white arrowheads) coinciding with the emergence of an R-on-T ectopic ventricular beat (sequence #2) in the ECG, RV focal discharge (green arrowhead) initiating NSVT with a large wave front traveling across the epicardial surface (sequences #3 - #5), followed by an RV focal discharge (yellow arrowhead) coinciding with the last beat of the arrhythmia episode (sequence #6). Map sequences of the other 3 VT episodes similarly showed RV focal discharges initiating NSVT. Times denote intervals after the onset of epicardial breakthroughs.

Enhanced L-type Ca²⁺ current contributes to β-adrenergic stimulation-induced prolongation of action potential repolarization in Calm1^N98S/+ mice

A previous study has demonstrated that overexpression of N98S CaM in guinea-pig ventricular myocytes diminishes Ca²⁺/CaM-dependent inactivation (CDI) of L-type Ca²⁺ channels,¹⁰ suggesting that impaired L-type current inactivation contributes to the long QT phenotype of Calm1^N98S/+ mice. Accordingly, we next sought to examine the properties of whole-cell L-type Ca²⁺ currents, I_Ca.L, in single ventricular myocytes isolated from Calm1^+/+ and Calm1^N98S/+ mice. Measurements were performed under whole-cell voltage clamp with high intracellular Ca²⁺ buffering by 10 mM BAPTA and in the presence of ryanodine (5 μM) in the pipette solution, to restrict CDI to that driven by Ca²⁺ entering through individual L-type Ca²⁺ channels.¹⁰ These measurements revealed attenuation of CDI and faster I_Ca.L recovery from inactivation in Calm1^N98S/+ cardiomyocytes, but no significant differences for peak I_Ca.L density or voltage-dependence of steady state activation and inactivation between the 2 genotypes (see Supplementary Data, Figure VIII and Table V in the Supplement).

Impaired CDI of L-type channels in combination with their accelerated recovery from inactivation in Calm1^N98S/+ ventricular myocytes would be expected to enhance L-type Ca²⁺ current during the action potential, delaying its repolarization. Since we did not, however, observe significant prolongation of action potential duration in isolated Calm1^N98S/+ ventricular myocytes under baseline conditions, the magnitude of changes in L-type channel gating was either insufficient to significantly lengthen action potentials or experimental conditions used for action potential recording blunted N98S CaM-mediated alterations of L-type channel gating. Measurements of macroscopic L-type currents were obtained with high intracellular Ca²⁺ buffering in the presence of ryanodine, while action potentials were recorded under more physiological Ca²⁺ buffering (0.5 mM EGTA), allowing global elevations of Ca²⁺. Accordingly, we next obtained the equivalent whole-cell patch-clamp data using a ryanodine-free pipette solution containing 1 mM EGTA. Results are summarized in Figure 5, Figure IX in the Supplement and Table VI in the Supplement. We observed no significant differences in the peak I_Ca.L density-voltage relations (Figure 5A and B), voltage-dependence of steady-state activation and inactivation (Figure 5E), time course of I_Ca.L inactivation (Figure 5G, H), and recovery from inactivation (Figure IX A and B in the Supplement) between the two genotypes under baseline conditions. In sharp contrast, isoproterenol (50 nM) increased peak I_Ca.L density (Figure 5C and D), left-shifted the steady-state activation curve (Figure 5F), increased the non-inactivating pedestal of the inactivation curve (Figure 5F), and slowed I_Ca.L inactivation (Figure 5G and H) significantly more in Calm1^N98S/+ versus Calm1^+/+ myocytes (Table VI in the Supplement). Isoproterenol also modestly reduced maximal I_Ca.L recovery from inactivation but did not significantly affect the recovery time constant (Figure IX C,D and Table VI in the Supplement). A more pronounced left shift of the steady-state activation curve combined with augmentation of peak I_Ca.L is expected to significantly increase the L-type window or late current (late I_Ca.L) within a voltage range that extends from threshold to potentials above the V_peak of the I-V relation (inset Figure 5F). Accordingly, we next measured total L-type Ca²⁺ current under action potential – clamp conditions both in the presence and absence of 50 nM isoproterenol. L-type Ca²⁺ currents were elicited by ‘typical’ action potential waveforms that had been pre-recorded from control and isoproterenol-treated Calm1^+/+ and Calm1^N98S/+ ventricular myocytes during steady-state pacing at 1 Hz. Results are summarized in Figure 6 and Figure X in the Supplement. Switching between action potentials pre-recorded in control Calm1^+/+ and Calm1^N98S/+ myocytes did not significantly change mean amplitudes of peak I_Ca.L or total I_Ca.L (∫I_Ca.L) in myocytes of either genotype in the absence of isoproterenol (Figure X in the Supplement). In its presence, however, magnitudes of both mean peak I_Ca.L and mean ∫I_Ca.L using the waveforms obtained from isoproterenol-treated myocytes of either genotype were significantly smaller in Calm1^+/+ myocytes than the ones that resulted when the same waveforms were imposed on Calm1^N98S/+ myocytes (Figure 6). Expanded views of the action potential waveforms and I_Ca.L traces (Figure 6C and D) demonstrate that I_Ca.L remains elevated throughout action potential repolarization in isoproterenol-treated Calm1^N98S/+ versus Calm1^+/+ myocytes. Taken together these results indicate that heterozygosity for the Calm1^N98S allele potentiates β-adrenergic effects on L-type channel gating, leading to increases in both peak and late I_Ca.L during the ventricular action potential.

Figure 5. Calm1^N98S heterozygosity potentiates β-adrenergic receptor-mediated effects on cardiac I_Ca.L.

A and C, Exemplar whole-myocyte I_Ca.L traces evoked by 300-ms step depolarizations to voltages from −50 to +60 mV from a holding potential of −70 mV in control (A) and during exposure to isoproterenol (C). Interpulse interval was 5 s. Top: voltage-clamp protocol. B and D, Peak I_Ca.L density – voltage relations in Calm1^+/+ and Calm1^N98S/+ ventricular myocytes in the absence (B) and presence of isoproterenol (D). Values are mean ± SEM. Control: 11 cells each from 5 Calm^+/+ mice and 5 Calm1^N98S/+ mice. Isoproterenol: 24 cells from 5 Calm1^+/+ mice and 29 cells from 5 Calm1^N98S/+ mice. * P < 0.03 versus Calm1^+/+ by t-test. E and F, Voltage-dependence of steady-state I_Ca.L activation and inactivation in the absence (E) and presence (F) of isoproterenol (50 nM). Values are mean ± SEM. Solid lines are best fits of the data to a Boltzmann function (fit parameters in Supplemental Table 6). Half-activation voltage in the presence of isoproterenol was significantly more negative in Calm1^N98S/+ versus Calm1^+/+ myocytes (−16.8 ± 0.8 mV versus −13.9 ± 0.8 mV; P = 0.018 by t-test). Insert: Calculation of the predicted window I_Ca.L density as a function of voltage. Amplitude of window I_Ca.L density was estimated as SS-act x SS-inact x I_Ca.L (at V_peak). Shaded area marks the range of membrane potential typically occurring during the action potential plateau of isoproterenol-treated Calm1^N98S/+ ventricular myocytes. SS-act and SS-inact indicate steady state activation and steady state inactivation, respectively. G, Exemplar whole-myocyte I_Ca.L traces elicited by 300-ms step depolarizations to +10 mV in the absence (top) and presence of 50 nM isoproterenol (bottom). Solid black lines are best fits of I_Ca.L decay to a double exponential: y(t) = A_fast exp(−t/τ_fast) + A_slow exp(−t/τ_slow) + y₀. H, Summary of I_Ca.L inactivation time constants, τ_fast and τ_slow, as a function of test potential. Values are mean ± SEM. Control: 11 cells from 5 Calm1^+/+ mice and 12 cells from 4 Calm1^N98S/+ mice; isoproterenol: 23 cells from 5 Calm1^+/+ mice and 25 cells from 6 Calm1^N98S/+ mice. Numbers above brackets indicate P values by One way ANOVA followed by multiple comparisons using the Bonferroni t-test (for τ_slow at 10 and 25 mV) or Kruskal Wallis One Way ANOVA on Ranks followed by Dunn’s Method for multiple comparisons (for all other voltages).

Figure 6. β-adrenergic receptor stimulation enhances peak and late I_Ca.L during the action potential in Calm1^N98S/+ ventricular myocytes.

A and B, L-type Ca²⁺ currents (lower panels) elicited by “typical” action potential waveforms (upper panels) that had been pre-recorded from isoproterenol (50 nM) - treated Calm1^+/+ and Calm1^N98S/+ ventricular myocytes. Action potential waveforms were delivered at 1 Hz steady-state frequency in the presence of isoproterenol (50 nM). Each data point in an I_Ca.L trace represents the mean ± SEM of 20 and 22 cells isolated from 4 Calm1^+/+ and 4 Calm1^N98S/+ hearts, respectively. C and D, The same action potential waveforms and I_Ca.L traces shown in A and B at expanded time scales. E and F, Dot plots of total I_Ca.L (∫I_Ca.L; E) normalized to cell capacitance and peak I_Ca.L density during the ventricular action potential in isoproterenol-treated cardiomyocytes. AP: action potential. Horizontal lines indicate mean ± SEM; 20 cells from 4 Calm1^+/+ mice and 22 cells from 4 Calm1^N98S/+ mice. P values by ANOVA followed by Bonferroni test for post hoc comparison.

Both Ca²⁺-free and Ca²⁺-bound calmodulin modulate properties of cardiac voltage-gated Na⁺ channels via direct or indirect (through Ca²⁺/CaM kinase II) interactions.¹⁶ We found no differences for peak I_Na density, steady-state activation or inactivation, recovery from inactivation, or inactivation kinetics between Calm1^+/+ and Calm1^N98S/+ ventricular myocytes (Figure XI A–E and Table VII in the Supplement), in agreement with a lack of a significant difference for ventricular conduction velocity between hearts of the 2 genotypes (Figure 2B and C). We also found no significant differences between the two genotypes for late Na⁺ current in the absence or presence of isoproterenol (Figure XI F and G in the Supplement). These findings support the idea that L-type Ca²⁺, but not Na⁺, channel dysregulation contributes to β-adrenergic receptor stimulation-induced prolongation of ventricular action potential repolarization and thus appears to be critical for manifestation of the long QT phenotype of Calm1^N98S/+ mice.

Cardiomyocyte Ca²⁺ cycling

Previous in vitro evidence led to the hypothesis that heightened propensity of ventricular myocytes for abnormal diastolic Ca²⁺ release events resulting from RyR2 disinhibition contributes to catecholamine-induced arrhythmias associated with the N98S CaM mutation.⁹ To examine RyR2 function, we next analyzed Ca²⁺ sparks in single, intact ventricular myocytes.¹⁷ Figure 7A and B shows that resting spark frequency in the absence of isoproterenol was decreased by ca. 50% in Calm1^N98S/+ myocytes, indicating reduced RyR2 channel activity, but rose to similar levels in both genotypes following isoproterenol exposure. Ca²⁺ spark amplitude (peak F/F₀; Figure 7C) was indistinguishable at baseline in cells of either genotype but significantly increased in Calm1^+/+, but not Calm1^N98S/+, myocytes following isoproterenol (1 μmol/L) treatment. We observed no significant isoproterenol effect on the spatial spread (FWHM) of Ca²⁺ sparks in myocytes of either genotype (Figure 7D). The spark duration (FDHM; Figure 7E) was longer in Calm1^N98S/+ myocytes both without and with isoproterenol, and the duration of the rising phase (time to peak) was significantly prolonged in Calm1^N98S/+ myocytes in the presence of isoproterenol (Figure 7F), compatible with an increase in the number of RyR2 channels recruited during a spark and/or delayed termination of SR Ca²⁺ release flux. The changes in Ca²⁺ spark properties may promote the development of spontaneous Ca²⁺ aftertransients. Accordingly, we next assessed the response of action potential-evoked whole-cell [Ca²⁺]_i transients to isoproterenol. Isoproterenol (1 μmol/L) similarly increased steady-state twitch [Ca²⁺]_i amplitude in the 2 genotypes and caused plateau-like [Ca²⁺]_i elevations lasting ~100 ms in Calm1^N98S/+ myocytes, delaying return of [Ca²⁺]_i to baseline (Figure XII in the Supplement). We found no evidence for increased propensity of Calm1^N98S/+ ventricular myocytes to develop early or delayed Ca²⁺ aftertransients under baseline conditions or during isoproterenol (1 μmol/L) treatment at pacing frequencies ranging from 1 to 3 Hz (Figure XIII in the Supplement).

Figure 7. Resting Ca²⁺ spark properties in Calm1^+/+ and Calm1^N98S/+ ventricular myocytes.

A, Confocal line-scan images of diastolic Ca²⁺ sparks recorded in a Calm1^+/+ and a Calm1^N98S/+ cell in the absence (top) and presence (bottom) of 1 μM isoproterenol. B, Box and whisker plots of (B) Ca²⁺ spark occurrence, (C) spark amplitude, (D) full width at half maximum (FWHM), (E) full duration at half maximum (FDHM), and (F) time to peak (TtP). The ends of the box are the upper and lower quartiles. The medians are marked by the horizontal lines inside the boxes, whereas the whiskers extend to the 10% and 90% quartiles. Black dots mark 5% and 95% percentiles. Numbers above horizontal lines denote P values by Kruskal-Wallis One Way ANOVA on Ranks followed by pairwise comparisons using Dunn’s method; Calm1^+/+: 443 sparks in 127 cells from 11 hearts without isoproterenol, 546/32/5 with 1 μM isoproterenol; Calm1^N98S/+: 274/125/9 without isoproterenol, 517/37/4 with 1 μM isoproterenol.

Increased propensity of the His-Purkinje system of Calm1^N98S/+ hearts for pause-dependent early afterdepolarizations

BVT has been previously suggested to originate from afterdepolarization-induced triggered activity generated in the His-Purkinje network.^18,19 To examine whether the His-Purkinje system of Calm1^N98S/+ hearts is susceptible to the development of pro-arrhythmic afterdepolarizations, electrical recordings were obtained from the right bundle branch as well as from the web and the septal endocardium in the right ventricles at baseline, during treatment with 100 nM isoproterenol, and following washout. Representative action potentials are shown in Supplemental Figure 14A and action potential parameters are summarized in Supplemental Table 8. Baseline APD₉₀ was significantly longer in Calm1^N98S/+ versus Calm1^+/+ conduction fibers. Isoproterenol prolonged APD₃₀, APD₅₀ and APD₉₀ in myocytes of either genotype, while significant differences for APD₉₀ remained (Figure XIV B and C in the Supplement). Interruption of steady-state pacing resulted in the development of pause-dependent phase 2 or phase 3 early afterdepolarizations (EADs; Figure 8A and B) in 34 of 67 Calm1^N98S/+ conduction fibers, but in only 6 of 55 Calm1^+/+ fibers (P < 0.001 by Chi-square test; Figure 8C). Isoproterenol did not significantly alter the incidence of EADs compared to baseline in cells of either genotype. Take-off potentials of a total of 202 and 23 EADs in Calm1^N98S/+ and Calm1^+/+ fibers, respectively, ranged from −75 mV to −29 mV and from −65 mV to −43 mV. Mean values for take-off potential and amplitude of EADs were similar between the 2 genotypes, whereas the mean diastolic interval preceding EAD-containing action potentials was shorter in Calm1^+/+ cells (Figure 8C). Delayed afterdepolarizations were never observed in wedges from either genotype. Taken together, these findings demonstrate that heterozygosity for the Calm1^N98S allele predisposes right ventricular conduction fibers to the development of pause-dependent EADs. Activation of β-adrenergic receptors did not promote EAD formation.

Figure 8. Pause-dependent early afterdepolarizations in the His-Purkinje network of Calm1^N98S/+ hearts.

A and B Top: original traces of membrane potential obtained from His-Purkinje myocytes in Calm1^N98S/+ hearts. The traces shown include the last 5 action potentials from a train of 30 action potentials at 5 Hz pacing, followed by >1-s pauses and spontaneous action potentials. Middle: Magnified views of the final paced action potential. Bottom: Magnified views of the boxed regions in the top panels, illustrating pause-dependent repetitive early afterdepolarizations (A) and an early afterdepolarization (B) arising from a spontaneous action potential. Neither cell developed diastolic (delayed) afterdepolarizations. Note the presence of automaticity in the fiber shown in B. C, Properties of early afterdepolarizations. N and n are the number of hearts and conduction fibers, respectively. EAD, early afterdepolarization; mean values for take-off potential, EAD amplitude and diastolic interval preceding the EAD-containing cycle were averaged from 202 and 23 EADs in Calm1^N98S/+ and Calm1^+/+ wedges, respectively. * P < 0.001 versus Calm1^+/+ by Chi-square test. ^# P = 0.006 versus Calm1^+/+ by Mann-Whitney Rank Sum test.

Discussion

This study provides, to the best of our knowledge, the first proof that the presence of N98S CaM is able to induce sustained ventricular arrhythmias on exposure to caffeine and epinephrine. Specifically, the arrhythmia phenotype of the murine model that we developed presents remarkable similarity with the clinical manifestation of the patient carrier of the N98S mutation in terms of arrhythmia morphology and inducibility by stress.⁸ Furthermore, Calm1^N98S/+ mice exhibited sinus bradycardia which was also present in the human patient. Besides these phenotypic commonalities, QT_c interval prolongation was not reported for the human carrier of the N98S CALM1 mutation. Manifestation of the long QT phenotype in Calm1^N98S/+ mice requires stimulation of β-adrenergic receptors. Because plasma norepinephrine levels in mice are typically more than an order of magnitude above values reported for humans,^20,21 enhanced basal sympathetic tone in the Calm1^N98S/+ mice may have contributed to the discrepancies between the human and murine phenotype.

Although the CRISPR/Cas9-mediated knock-in procedure may cause off-target mutagenesis which can confound interpretation of the genotype-phenotype relationship,²² we do not believe this is the case here. Calm1^N98S/+ mice generated from two independent integration events were essentially phenotypically identical. In addition, animals were analyzed after multiple back cross generations into a new inbred background, thus greatly diluting any contribution of an off-target event in the original targeted genome. Thus, contributions of off-target sites to the phenotypes observed in our mice are unlikely, strengthening our conclusion that heterozygosity for the Calm1^N98S allele is causative of the arrhythmia syndrome. We further found no evidence that the cardiac phenotype developed secondary to cardiac structural abnormalities or contractile dysfunction.

Mechanisms of QT prolongation

Our data indicates a key role of β-adrenergic induced I_Ca.L dysregulation in the pathogenesis of the long QT phenotype of Calm1^N98S/+ mice. β-adrenergic receptors signal through the G_s protein-adenylyl cyclase-cAMP-PKA pathway to regulate L-type channels.²³ Functionally, β-adrenergic receptor stimulation elevates the activity of L-type channels as a result of an increase in the channel’s open probability.²⁴ As a consequence, macroscopic I_Ca.L amplitude increases, voltage-dependence of steady-state activation shifts to more negative potentials, and the inactivation rates accelerate. We observed that isoproterenol both increased peak I_Ca.L and left-shifted the activation curve significantly more in Calm1^N98S/+ versus Calm1^+/+ ventricular cardiomyocytes (Figure 5), supporting the notion that N98S CaM potentiated β-adrenergic effects on L-type channel gating, increasing the magnitude of window I_Ca.L in mutant cardiomyocytes. Inactivation rates remained diminished in isoproterenol-treated Calm1^N98S/+ myocytes (Figure 5H). Taken together, these findings suggest a mechanism wherein attenuation of I_Ca.L inactivation pairs with β-adrenergic enhancement of window I_Ca.L to give rise to elevated ‘late’ I_Ca.L in Calm1^N98S/+ ventricular myocytes, prolonging their action potential duration. A role of adrenergic – induced L-type channel dysregulation in delaying ventricular repolarization is directly supported by our action potential-clamp data, showing augmented Ca²⁺ influx extending throughout action potential repolarization of isoproterenol-treated Calm1^N98S/+ myocytes. Overall, our data is compatible with the concept that N98S CaM-dependent attenuation of I_Ca.L inactivation in concert with potentiation of β-adrenergic-mediated effects on L-type Ca²⁺ channels is responsible, at least in part, for the adrenergic induced long QT phenotype of Calm1^N98S/+ mice.

L-type channels and small-conductance, Ca²⁺-gated K⁺ (SK) channels form a negative feedback system wherein Ca²⁺ influx through L-type Ca²⁺ channels directly activates SK channels, preventing excessive repolarization prolongation.²⁵ We have previously demonstrated that N98S CaM impedes activation of heterologously expressed SK channels,¹¹ raising the possibility that inadequate SK channel activation contributes to manifestation of the long QT phenotype of Calm1^N98S/+ mice.

Both Ca²⁺-free and Ca²⁺-bound calmodulin, via direct and indirect interactions, regulate gating of cardiac, voltage-activated Na channels. Specifically, calmodulin has been shown to control the magnitude of late I_Na, an important determinant of ventricular action potential duration.^16,26 β-adrenergic receptor stimulation also enhances late I_Na.²⁷ However, we found no evidence for a role of elevated late I_Na in prolonging ventricular repolarization in control or isoproterenol-treated Calm1^N98S/+ ventricular cardiomyocytes.

Calmodulin also regulates gating of voltage-activated K⁺ channels, via direct interaction or indirectly via Ca²⁺/CaMKII.²⁸ We did not, however, observe significant differences in the time course of early repolarization nor in resting membrane potential between isoproterenol-treated Calm1^+/+ and Calm1^N98S/+ ventricular cardiomyocytes, suggesting that activities of transient outward K⁺ currents and inward rectifier K⁺ currents, respectively, remained largely unaffected. Although calmodulin has been shown to regulate delayed rectifier K⁺ currents (I_Ks),²⁹ ventricular cardiomyocytes of adult mice do not elaborate significant I_Ks, excluding I_Ks as a long QT-causing candidate in our model.

Female mice of either genotype had significantly shorter QT_c intervals compared to genotype-matched males. Previous reports revealed that QT_c intervals of adult female mice are shorter than those of age-matched males during stages of the estrous cycle when estrogen levels are low (which occurs typically for 3.5 days of the 4-day cycle).³⁰ Although we did not ascertain the estral stage during telemetry, the significantly shorter QT_c intervals seen in female mice suggest that they were under low-estrogen conditions during ECG monitoring. Their shorter QT_c intervals, however, did not protect female Calm1^N98S/+ mice from inducible arrhythmias.

Arrhythmogenic mechanisms

Our microelectrode studies revealed a markedly increased propensity of the right ventricular His-Purkinje network of Calm1^N98S/+ hearts to generate pause-dependent early, but not delayed, afterdepolarizations. Isoproterenol did not augment the incidence of early afterdepolarizations, suggesting that β-adrenergic activation promotes their transmission to the surrounding myocardium rather than their formation in the conduction system. Voltage mapping revealed a right ventricular focal origin of the VT-triggering ectopic beats in Calm1^N98S/+ hearts perfused with elevated Ca²⁺ and isoproterenol. Intriguingly, these ectopic beats manifested as concentric epicardial breakthrough patterns that were similar to those emanating from the right ventricular free wall during normal sinus rhythm, suggesting a focal origin in the His-Purkinje network. Thus, our findings support the possibility that focal discharges arising from early afterdepolarization-induced triggered events within the ventricular conduction system contribute to catecholaminergic arrhythmogenicity in Calm1^N98S/+ hearts. However, whether BVT in our model is mediated by early afterdepolarization-induced triggered activity originating in Purkinje fibers needs to be determined. EAD take-off potentials were observed to occur over a range of membrane potentials wherein I_CaL reactivation is limited, suggesting other ionic mechanisms also contribute to EAD formation, possibly including Na/Ca exchanger current and/or I_Na.³¹

DAD-induced triggered responses occurred in isoproterenol-treated Calm1^N98S/+ ventricular myocytes at high pacing rates, suggesting their involvement in arrhythmogenesis. Transition from BVT into PVT as observed in one Calm1^N98S/+ mouse after combined epinephrine and caffeine administration may be attributed to focal discharges mediated by DADs arising at multiple locations within the working myocardium. Thus, EADs in the His-Purkinje network and DADs in the working myocardium may serve complementary roles in initiating and maintaining, respectively, arrhythmias in our model. The similarly low prevalence of delayed Ca²⁺ aftertransients in isoproterenol-challenged ventricular myocytes of both genotypes is at odds with the increased susceptibility of Calm1^N98S/+ myocytes to the development of DAD-induced triggered responses at high pacing rates. We cannot readily explain this discrepancy. It is possible that the higher maximum pacing rate used in the current-clamp (7 Hz) versus imaging (3 Hz) experiments is critical to detect genotype-specific differences in the susceptibility to DAD-causing, spontaneous SR calcium release events. For example, a higher integrated Ca²⁺ influx through adrenergically activated L-type channels during rapid pacing may lead to a larger SR Ca²⁺ load of Calm1^N98S/+ myocytes, promoting RyR2 opening. However, the contribution of SR Ca²⁺ load versus that of other mechanisms (N98S CaM-mediated RyR2 disinhibition⁹, differences in cytosolic [Ca²⁺]) to arrhythmogenic Ca²⁺ release events remains to be determined.

Role of bradycardia in catecholaminergic BVT

Like the human carrier of the N98S CALM1 mutation, Calm1^N98S/+ mice exhibited chronic sinus bradycardia. Sinus node rates increased in mice of either genotype in response to β-adrenergic receptor stimulation but remained lower in Calm1^N98S/+ than in Calm1^+/+ mice. Such chronotropic incompetence paired with adrenergic induced QT_c interval prolongation may promote EAD-induced triggered discharges in the ventricular conduction system.

The persistence of sinus bradycardia following pharmacological blockade and activation of β-adrenergic receptors in Calm1^N98S/+ mice suggested a change in the intrinsic firing rate of sinus node myocytes through modulation of proteins involved in pacemaking (e.g., ryanodine receptor, Na⁺/Ca²⁺ exchanger),^32,33 either directly via binding to them or indirectly via changes in Ca²⁺/CaMKII activity.^33,34 Impulse transmission from the sinus node cells to the atrial myocardium is mediated by different voltage-gated sodium channel isoforms³⁵ which are regulated by CaM³⁶ as well as Ca²⁺/CaMKII and can thus be impaired following N98S-CaM binding.

QRS widening

QRS widening, indicative of ventricular conduction slowing, was seen in conscious male Calm1^N98S/+ mice but was not reported for the human female carrier. Using optical voltage mapping of Langendorff-perfused hearts, we found no evidence for ventricular conduction slowing in control or isoproterenol-treated Calm1^N98S/+ hearts, in agreement with the similar properties of voltage-gated fast Na⁺ current in ventricular cardiomyocytes of the two genotypes. We also observed no statistical differences for action potential amplitude or maximal phase 0 upstroke velocity between Calm1^N98S/+ and Calm1^+/+ His-Purkinje myocytes in situ, neither in control nor during β-adrenergic stimulation, suggesting that impulse propagation within the ventricular conduction system of ex vivo Calm1^N98S/+ hearts was intact. Thus, we cannot readily explain the mechanism underlying QRS complex widening in male Calm1^N98S/+ mice. A potential role of autonomic innervation in decelerating conduction in the in situ Calm1^N98S/+ heart that may be absent or attenuated in the isolated heart cannot be discarded.

Conclusions

Our data provide proof of the concept that heterozygosity for the Calm1^N98S mutation is causative of a complex arrhythmia syndrome characterized by sinus bradycardia, adrenergically induced QT_c interval prolongation, and susceptibility to the development of catecholaminergic BVT. Heterozygosity for the Calm1^N98S allele attenuates Ca²⁺/CaM-dependent I_Ca.L inactivation and potentiates β-adrenergic effects on cardiac L-type Ca²⁺ current, leading to elevated late I_Ca.L which prolongs ventricular repolarization. Pause-dependent EADs and rapid pacing-induced DADs originating in the His-Purkinje network and ventricular myocytes, respectively, constitute a potential source of catecholaminergic BVT.

Clinical Perspective.

What Is New?

• Knock-in mice heterozygous for the p.Asn98Ser mutation in the Calm1 gene encoding calmodulin exhibit sinus bradycardia, QT_c interval prolongation and catecholaminergic, bidirectional ventricular tachycardia.

• β-adrenergic receptor activation-induced increase in ‘late’ L-type calcium current contributes to the long QT phenotype.

• Pause-dependent early and delayed afterdepolarizations originating in the His-Purkinje network and working ventricular myocytes, respectively, are involved in the initiation and/or maintenance of catecholaminergic arrhythmias in knock-in mice.

What Are the Clinical Implications?

• Our results suggest the ventricular myocyte late L-type calcium current as a potential target in long QT syndrome caused by calmodulin mutations.

• Anti-bradycardia pacing may constitute a viable approach to suppress arrhythmias triggered by afterdepolarizations.

Non-standard Abbreviations and Acronyms

APD

action potential duration

BVT

bidirectional ventricular tachycardia

CaM

calmodulin

CDI

Ca²⁺/CaM-dependent inactivation

CPVT

catecholaminergic polymorphic ventricular tachycardia

CV

conduction velocity

DAD

delayed afterdepolarization

EAD

early afterdepolarization

FDHM

full duration at half maximum

FWHM

full width at half maximum

PVT

polymorphic ventricular tachycardia

QT_c

corrected QT interval

RyR2

ryanodine receptor type 2

SK channel

small conductance, Ca²⁺-activated potassium channel