The link between abnormal calcium handling and electrical instability in acquired long QT syndrome – does calcium precipitate arrhythmic storms?

Abstract

Release of Ca²⁺ ions from sarcoplasmic reticulum (SR) into myocyte cytoplasm and their binding to troponin C is the final signal form myocardial contraction. Synchronous contraction of ventricular myocytes is necessary for efficient cardiac pumping function. This requires both shuttling of Ca²⁺ between SR and cytoplasm in individual myocytes, and organ-level synchronization of this process by means of electrical coupling among ventricular myocytes. Abnormal Ca²⁺ release from SR causes arrhythmias in the setting of CPVT (catecholaminergic polymorphic ventricular tachycardia) and digoxin toxicity.

Recent optical mapping data indicate that abnormal Ca²⁺ handling causes arrhythmias in models of both repolarization impairment and profound bradycardia. The mechanisms involve dynamic spatial heterogeneity of myocardial Ca²⁺ handling preceding arrhythmia onset, cell-synchronous systolic secondary Ca²⁺ elevation (SSCE), as well as more complex abnormalities of intracellular Ca²⁺ handling detected by subcellular optical mapping in Langendorff-perfused hearts. The regional heterogeneities in Ca²⁺ handling cause action potential (AP) heterogeneities through sodium-calcium exchange (NCX) activation and eventually overwhelm electrical coupling of the tissue.

Divergent Ca²⁺ dynamics among different myocardial regions leads to temporal instability of AP duration and – on the patient level – in T wave lability. Although T-wave alternans has been linked to cardiac arrhythmias, non-alternans lability is observed in pre-clinical models of the long QT syndrome (LQTS) and CPVT, and in LQTS patients. Analysis of T wave lability may provide a real-time window on the abnormal Ca²⁺ dynamics causing specific arrhythmias such as Torsade de Pointes (TdP).

BACKGROUND

From a molecular perspective, the force generated during cardiac contraction is developed by the interaction between actin and myosin proteins and cross-bridge cycling in thousands of sarcomeres in each cardiomyocyte. For the heart to maintain blood circulation, individual sarcomeres have to contract in a synchronized pattern. Loss of temporal organization of this process results in circulatory arrest, as exemplified during the early phase of ventricular fibrillation, when the contractility of individual cardiomyocytes is still preserved.

At the organ level, the contraction of each myocyte is timed by a wavefront of cell membrane depolarization propagating through the cardiac chambers. For each myocyte, this electrical depolarizing signal is converted to a chemical signal, namely an increase in free Ca²⁺ concentration in the cytoplasm. Ca²⁺ binding to troponin C allows for cross-bridge formation between actin and myosin molecules, causing force generation along the longitudinal axis of individual sarcomeres and myocytes. Ca²⁺ enters the cytoplasm through voltage-gated L-type Ca²⁺ channels within approximately 10 milliseconds of membrane depolarization, but most of the Ca²⁺ ions responsible for cardiac contraction (80–85%)¹ are released from internal stores, called sarcoplasmic reticulum (SR) through a mechanism described as Calcium-Induced Calcium Release (CICR). In cardiac muscle, the initial Ca²⁺ influx occurring via L-type Ca²⁺ channels on the T-tubule membrane leads to allosteric opening of Ca²⁺-release channels or cardiac ryanodine receptors (RyR2), located on the SR membrane across a gap of 10–15 nm between the T-tubule membrane and the terminal cisternae membrane of the SR.² The RyR2 opening substantially amplifies the initial Ca²⁺ rise. After a contraction associated with elevation of cytoplasmic Ca²⁺, Ca²⁺ is removed from the cytoplasm to allow for myocardial relaxation and filling of cardiac chambers. Ca²⁺ released via RyR2 is transported back in the lumen of the SR against a concentration gradient by the cardiac Sarco (Endo) plasmic Reticulum Ca ATPase (SERCA2), a Ca²⁺ pump powered by ATP hydrolysis. Ca²⁺ influx from the extracellular via Ca²⁺ channels is returned to the extracellular space by the NCX1 countertransporter, a Na⁺/Ca²⁺ exchanger that exchanges 3 Na⁺ ions for each Ca²⁺ and is driven by the Na⁺ electrochemical gradient. Under steady state conditions, Ca²⁺ influx equals Ca²⁺ efflux in each subcellular compartment.³

Disturbances of electrical synchronization of cardiac contractions occur in many cardiac diseases and often lead to sudden cardiac death, a dramatic event which is usually caused by ventricular tachycardia (VT) or ventricular fibrillation (VF).⁴ A detailed understanding of the mechanisms triggering ventricular arrhythmias remains elusive. One reason for this is that ventricular arrhythmias can be produced by processes operating at different levels, ranging from point mutations in genes coding for ion channels,^{5, 6} which can cause repetitive firing in individual cells,^{7, 8} to reentry (continuous impulse propagation along a circular pathway), which requires spatially extended tissue and can occur around an anatomical obstacle such as a scar.^{9, 10}

CALCIUM HANDLING AND TRIGGERED ACTIVITY

Triggered activity refers to abnormal generation of an AP, which is initiated (“triggered”) by the preceding (normal or abnormal) AP. In contrast to reentry, it does not require spatially extended tissue and can be observed in single cardiomyocytes. Most clinically relevant arrhythmias may involve more than one mechanism – reentry around a scar initiated by a triggered ectopic beat being an obvious example. However, study of arrhythmia models involving mostly triggered activity offer important insights into arrhythmogenesis in general.

Classically, triggered APs are categorized as either delayed afterdepolarizations (DADs) or early afterdepolarizations (EADs). EADs can be triggered during long ventricular AP durations (APDs) or during short atrial APDs and large amplitude Ca²⁺ transients as may occur in atrial fibrillation triggered by sympathovagal activity.^11–14 DADs occur after completion of the preceding AP (i.e. the takeoff of the ectopic AP occurs from fully repolarized membrane potential values). EADs by definition occur before the completion of the preceding AP, from the plateau or repolarization phase. DADs and EADs are believed to occur in somewhat different settings. DADs typically develop in the setting of tachycardia or high extracellular Ca²⁺ concentration. Clinically, they have been linked to toxic levels of cardiac glycosides, adrenergic stimulation or catecholaminergic polymorphic ventricular tachycardia (CPVT). CPVT is a congenital arrhythmic syndrome associated with sudden cardiac death caused by mutations that destabilize Ca²⁺ release from the SR.^15–19 The dominant factor in genesis of EADs is prolonged AP duration, along with bradycardia and adrenergic stimulation. EADs occur in a wide variety of clinical settings, including hypokalemia, hypocalcemia, toxicity of multiple cardiac and non-cardiac medications, and occasionally in patients with congenital long QT syndrome, a group of hereditary arrhythmic conditions defined by delayed ventricular repolarization. The typical clinical manifestation of EAD activity is a characteristic polymorphic ventricular tachycardia labeled torsade des pointes (TdP).^20–22

The cellular mechanism of DAD genesis appears to be well understood. The conditions listed above in association with DADs all result in spontaneous SR Ca²⁺ release during diastole, which is not triggered by membrane depolarization. This spontaneous SR Ca²⁺ release is caused by an unusually high Ca²⁺ concentration in the junctional SR (jSR), in the case of digoxin toxicity, short cycle length or elevated extracellular Ca²⁺ which all promote RyR2 opening in the absence of a voltage trigger. This mechanism has been labeled Store-Overload Induced Calcium Release (SOICR).²³ In many cases of CPVT, the threshold for SOICR is decreased because of mutation in the gene encoding the RyR itself, or in a gene encoding an interacting protein such as calsequestrin, allowing DAD formation despite normal or even low jSR Ca²⁺ load.¹⁷

The initial SOICR in cardiac myocytes is often fairly localized and propagates through the myocyte cytoplasm as a Ca²⁺ wave: Ca²⁺ ions diffuse from the region of elevated concentration (wave “crest”) into surrounding regions, triggering opening of neighboring RyRs in the direction of wave propagation through CICR. The regions in the “wake” of the wave are refractory with respect to Ca²⁺ release because of depletion of junctional SR Ca²⁺ or intrinsic refractoriness of RyRs following the previous Ca²⁺ release.^{24, 25}

The Ca²⁺ released into cytoplasm that produce a Ca²⁺ wave is transported in part into the extracellular space by NCX1. Since this molecule transports one Ca²⁺ out of the cytoplasm in exchange for 3 Na⁺ ions into the cytoplasm in each transport cycle, its activity depolarizes cell membrane. In an isolated cell, AP is triggered if the threshold for activation of voltage-gated cardiac Na⁺ channels is reached. The question remains of how can Ca²⁺ waves occurring randomly in individual myocytes overcome the electrical load of the surrounding tissue to trigger a propagated AP? The question has been recently addressed by the elegant experiments of Wasserstrom et al.,^{16, 26} who used confocal microscopy of perfused rat hearts to study the mechanism of synchronization of Ca²⁺ waves. In their model of hypercalcemia and rapid pacing followed by a pause, both the coupling interval and timing dispersion of the Ca²⁺ wave appearance decreases with increasing degree of Ca²⁺ overload, eventually resulting in sufficient temporal overlap of depolarizations in individual myocytes to trigger tissue-wide depolarization.

Abnormal Ca²⁺ handling has been implicated in arrhythmias in acute myocardial ischemia,²⁷ which results in marked AP shortening and CaT prolongation (setting the stage for diastolic depolarization by NCX current). It also plays a role in arrhythmogenesis during reperfusion, heart failure²⁸ and atrial fibrillation. For example, simultaneous adrenergic and cholinergic stimulation of pulmonary vein cuffs causes triggered activity corresponding to EADs, which can be abolished by ryanodine administration, indicating role of intracellular Ca²⁺ handling. Interestingly, AP shortening rather than prolongation correlates with arrhythmia in this model.^{29, 30} Detailed discussion of these important topics is outside the scope of this review.

In contrast to DADs, the cellular mechanism of EAD generation remains controversial, although the involvement of Ca²⁺ handling is not disputed. The prevailing theory suggests that repolarization delay prolongs the time period which the L-type Ca²⁺ channels spend in the “window current” voltage range, when neither the activation nor the inactivation voltage gate of the channel is fully closed. The resulting increase in the inward current carried by L-type Ca²⁺ channels depolarize the membrane, further opening the activation gate and triggering the EAD. This scenario is supported by some mathematical models,³¹ and is consistent with experiments demonstrating that L-type channel opener promotes EAD formation, and L-type channel blockers suppress it.³² De Ferrari et al.³³ directly observed Ca²⁺ release in isolated myocytes using a Ca²⁺-sensitive dye and found that Ca²⁺ waves are associated with DADs, but that that the Ca²⁺ rise related to EAD occurs relatively synchronously across the cell, a finding more consistent with the L-type channel window current theory than with ‘spontaneous’ SR Ca²⁺ release. Marban et al.³⁴ reported that ryanodine (an alkaloid blocking RyR, the SR Ca²⁺ channel) failed to suppress EADs in ferret papillary muscle superfused with cesium, but that EADs could be suppressed by an L-type channel blocker.

On the other hand, Volders et al.,³⁵ argued that spontaneous (non-voltage-triggered) SR Ca²⁺ release may be the cause of EADs, based on the observation that the beginning of cell contraction precedes EAD takeoff in isolated myocytes. According to their hypothesis, repolarization delay results in SR Ca²⁺ overload, and secondary SR Ca²⁺ release triggers an EAD by activating NCX, i.e. through a mechanism similar to DAD generation.¹¹ The report by Milberg et al³⁶ that NCX block suppresses EADs in a rabbit model of long QT syndrome supports this mechanism.

DUAL OPTICAL MAPPING DATA

The dual wavelength optical mapping technique has been used by our team and others to study the mechanisms of arrhythmogenesis in the setting of repolarization delay. This technique allows simultaneous recording of V_m and Ca_i signals with excellent spatial and temporal resolution. Moreover, it can be naturally applied to study perfused heart at physiological temperature, and can thus provide insights that cannot be readily obtained from isolated cell experiments.¹

Simultaneous application of bradycardia, hypokalemia, hypomagnesemia, and block of I_Kr (the rapid component of the delayed rectifying K⁺ current) results in quick and reproducible TdP induction in a well-established rabbit heart model.^{37, 38} All these factors are known to promote TdP in humans.

The analysis of V_m and Ca_i signals simultaneously recorded during TdP using phase-plots suggested that at the site of EAD focus, that the rise of Ca_i rise precedes that of V_m, supporting the role of SR Ca²⁺ release as the driver of the arrhythmia.³⁹ More detailed results were obtained by evaluation of V_m and Ca_i dynamics during transition from slow paced rhythm (after acute AV node destruction) to onset of TdP (caused by pharmacological I_Kr blockade).⁴⁰ Invariably, gradual prolongation of AP duration was accompanied by change in Ca²⁺ dynamics, which preceded onset of EADs by minutes (Figure 1). Before the I_Kr block, the course of the Ca²⁺ transient (CaT) closely follows the course of AP. After I_Kr block, the APD prolongs markedly, but initially without the appearance of EADs. At this stage, CaT does not follow the AP by developing a smooth long plateau, but instead falls nearly as fast as it did before I_Kr block and then rises a second time during phase 2 or 3 of the AP, forming a second CaT peak which returns to baseline approximately at the same time as AP. The amplitude of this systolic secondary calcium elevation (SSCE) increases gradually in the span of a few minutes and eventually, electrical instabilities during the AP plateau develop to EADs. EADs are approximately synchronous with SSCEs, but are delayed by tens of milliseconds. The onset of SSCEs always precedes the onset of EAD at the earliest site of a propagated EAD.⁴¹ As the ectopic beat propagates away from the focus, AP upstroke overtakes the CaT upstroke, and the AP upstroke precedes the CaT upstroke as expected during the propagation of a normal AP (Figure 2). With further APD prolongation, the single SSCE is followed by multiple Ca²⁺ oscillations during the same AP plateau and each Ca_i peak tends to be accompanied by its own EAD.⁴¹ Finally, long AP plateaus with oscillations of both V_m and CaT are observed, which correspond to TdP runs on EKG tracing. Treatment with ryanodine and thapsigargin (SERCA inhibitor) prevented appearance of Ca oscillations and TdP following I_Kr block, although AP prolongation still developed. Treatment with K201 (1 μM to avoid off-target effects), a RyR stabilizer, had a similar effect.⁴² Taken together, these results provide compelling evidence that in this animal model, spontaneous systolic SR Ca²⁺ release, manifested as a SSCE, cause EADs. The observation that L-type Ca²⁺ channel blocker and low extracellular Ca²⁺ also prevent EAD appearance is not surprising, since both interventions are expected to deplete SR Ca²⁺ load.

Figure 1.

Development of SSCEs precedes the onset of TdP. Simultaneous recordings of membrane voltage (blue) and Ca²⁺ signals (red) are shown in all panels. A: During slow pacing before I_Kr block, the AP and CaT waveforms are monophasic and similar. B: After I_Kr block, but before the development of EADs, AP prolongation is accompanied by complex, biphasic CaT, with an SSCE following the initial rise of the CaT. C: With increasingly more robust I_Kr block, the amplitude of the SSCE increases and a corresponding EAD develops. Note that the rise of SSCEs precedes the EAD upstroke, and that a second SSCE of smaller amplitude develops during AP downstroke. D: Eventually, the SSCE increases in amplitude and develops its own EAD. E: Finally, long plateaus with superimposed oscillations of both membrane voltage and Ca²⁺ signal occur. They correspond to runs of polymorphic ventricular tachycardia on EKG (black). Adapted from⁴¹

Figure 2.

SSCE precedes EAD at the site of origin of the corresponding ectopic beat. A: An activation map of a propagated ectopic beat with an origin at site 1; the depolarization wavefront proceeds from red to blue regions. B: Simultaneous CaT and AP signals from the beat focus (site 1) and a distant site (site 3). Secondary CaT upstroke precedes secondary AP upstroke at site 1, but trails it at site 3. Note the tiny SSCE is not followed by an EAD on the seconds beat at site 1 (vertical arrow). C: Similar to panel B, but at displayed with better time resolution. Note the shift in relative positions of the CaT and AP upstrokes between sites 1 and 3. Adapted from⁴¹

SPATIAL HETEOGENEITY OF CALCIUM TRANSIENT

When the EAD model described above is studied at better spatial resolution (100×100 pixels, 150 μm per pixel), a striking spatial heterogeneity of the normalized CaT signal is easily appreciated.⁴² This is largely absent at baseline, when spatial the amplitude of both CaT and AP is relatively even across the epicardial surface at all times, but increases dramatically with the appearance of Ca oscillations. Intriguingly, the areas of high calcium oscillation amplitude are highly irregular, do not appear to follow obvious anatomic regions and differ among different experiments, although they generally remain relatively stable between subsequent beats. In fact, regions of elevated Cai as small as 1 mm can be resolved (Figure 3). Although the amplitudes of the Ca_i and V_m are well correlated and the spatial heterogeneity of V_m increases along with Ca_i, V_m heterogeneity increase is less pronounced and V_m distribution over the epicardial surface is smoother than that of Ca_i (Figure 4). This would be expected if Ca_i heterogeneity drives the process, since Ca_i diffusion is much slower than charge diffusion. Still, the spatial heterogeneity of V_m leads to appearance of steep spatial voltage gradients, which would be expected to generate electronic currents through the tissue. Such currents would depolarize a portion of cell membrane and could trigger propagated action potential. This seems to be an important arrhythmogenic mechanism, since all the propagated ectopic beats where the focus could be localized originated from regions of steep voltage gradients.

Figure 3.

Spatial heterogeneity of CaT in LQTS prior to development of ectopic beats is demonstrated by 3 consecutive images of the anterior epicardial surface during a single paced heartbeat. A: Superimposed normalized CaTs from 3 pixels (a–c) indicate differences in Ca²⁺ concentration among closely spaced ventricular sites. B: The timing of the 3 images is indicated by the vertical lines in panel A. Note the highly irregular, non-contiguous “islands” of Ca²⁺ elevation which shift during the CaT. The color scale changes to facilitate visualization of the spatial heterogeneity. Adapted from⁴²

Figure 4.

Comparison of spatial heterogeneity of Ca²⁺ and membrane voltage signals during paced rhythm and I_Kr block. A: Simultaneous images of Ca²⁺ (top) and membrane voltage (bottom) signals from the anterior epicardial surface at the time of SSCE. Areas of elevated Ca²⁺ correspond to more positive membrane potentials, but the voltage map is much smoother or homogeneous than the Ca²⁺ map. B: Superimposed Ca²⁺ and membrane voltage signals from 3 closely spaced pixels (a–c) indicated in the previous panel. The vertical line marks the timing of the images in A. Adapted from⁴²

SEX-DIFFERENCES EXPLAIN SPATIAL HETEOGENEITIES OF CaTs and LOCATION OF EADs

An examination of the distribution of the sites that fired the first EADs produced a striking pattern. EADs could not be associated with specific anatomical features but were initiated around the base of the adult female rabbit heart.⁴³ In contrast, the same LQT model failed to elicit EADs and TdP in adult male rabbit hearts. This sex difference in arrhythmia phenotype was reversed in pre-pubertal rabbits where I_Kr blockade elicited EADs and TdP in pre-pubertal males but not females.^{43, 44} Sex-differences in the vulnerability to TdP have long been acknowledged in humans for both congenital and drug-induced LQT type 2 (LQT2).⁴⁵ Young adult women (pre-menopausal) are at greater risk of TdP compared to their male counter-part, but males under the age of 14 years old have a considerably greater risk of lethal arrhythmias in LQT2).⁴⁵ These sex differences and the switch between pre-puberty and adulthood has been attributed to the surge of testosterone and estrogen.⁴⁶

A comparison of the arrhythmia risk in LQTS was recently reviewed for various species (mouse, guinea pig, rat, rabbit and dog) and the New Zealand White rabbits most closely parallel human sex-differences in arrhythmia risk in adult and pre-puberty.⁴⁶ Based on extensive studies in man and rabbit,⁴⁶ the general consensus is that in LQT2, testosterone is protective of TdP and estrogen greatly enhances the risk of TdP. Ventricular myocytes were isolated from various regions (endo-epi, base and apex) of adult rabbit hearts and incubated with or without estrogen (1 nM) for 0 to 3 days. Estrogen treated myocytes from the base of the epicardium expressed increasing current densities of I_Ca,L and I_NCX as well as increasing levels of their mRNA and channel proteins (Cav1.2α and NCX1).^{47, 48} This genomic upregulation occurred during the first 2-days of estrogen treatment and only in female base myocytes isolated from the epicardium and not in male myocytes. Myocytes isolated from other regions of the ventricles did not respond to estrogen treatment and failed to exhibit EADs after I_Kr block.^{47, 48} In Langendorff rabbit hearts under LQT2 conditions, EADs were initiated near the base of the heart and from the epicardium because cryoablation of the ventricular chambers with liquid N₂ did not alter the incidence of EADs.³⁹ In these cryoablated hearts, only a 1 mm thick layer of epicardium survived which was the source of all EADs and provided compelling evidence that EADs did not originate from the conduction system, the endocardium or mid-myocardium.³⁹ These measurements are consistent with the localization of the estrogen genomic upregulation of I_Ca,L and I_NCX. The modulation of cardiac channel by estrogen is localized to the base of the epicardium in female hearts which could be due to regional expression of the two estrogen receptors, ERα and ERβ. However, Western blot analysis from different regions of male and female rabbit hearts revealed a uniform estrogen receptor expression (ERα and ERβ) in hearts from both sexes and ERα and ERβ could not explain regional these heterogeneities.⁴⁷ Table 1 summarizes sex and regional heterogeneities of I_Ca,L and I_NCX densities in rabbit hearts based on voltage clamp experiments. Differences in the transient outward current, I_t.o,. the rapid and slow delayed K⁺ rectifying currents, I_Kr and I_Ks and the inward rectifying K⁺ current, I_K1 were compared between male and female rabbits but the regional heterogeneities have not been reported.

Table 1.

Sex differences in Ion Channel Expression in New Zealand White Rabbits

Ionic current	ICa,L	INCX	It.o.	IKr	IKs	IK1
Male Epicardium Base	↔	↔	M=F	M>F	M>F	M>F
Male Epicardium Apex	↔	↔
Male Endocardium	↔	↔
Female Epicardium Base	↑	↑↑
Female Epicardium Apex	↔	↔
Female Endocardium	↔	↔
Pre-pubertal Male Epicardium Base	↑	↑↑	ND	ND	ND	ND
Pre-pubertal Male Epicardium Apex	↔	↔	ND	ND	ND	ND
Pre-pubertal Male Endocardium	↔	↔	ND	ND	ND	ND
Pre-pubertal Female Epicardium Base	↔	↔	ND	ND	ND	ND
Pre-pubertal Female Epicardium Apex	↔	↔	ND	ND	ND	ND
Pre-pubertal Female Endocardium	↔	↔	ND	ND	ND	ND

Summary of sex and regional heterogeneities of ionic currents

Voltage-clamp data on sex and regional differences in L-type Ca²⁺ current density and Na-Ca exchange current density (I_Ca.L and I_NCX) are compiled from^{43, 47, 48} and preliminary data. Sex differences in I_Kr, I_t.o and I_K1 currents are compiled from Liu et al., J. Pharmacology and Experimental Therapeutics, 1998⁷⁸ and I_Ks from Zhu et al., Pflugers Archives, 2013.⁷⁹ Each arrow, ↑ represent ~30% increase in current density compared to other regions of the same heart and of the opposite sex.

At the molecular level, estrogen treatment (17β-estradiol at 0.3–1.0 nM) promotes the induction of EADs by upregulating the L-type Ca²⁺ current and the NCX current (I_Ca,L and I_NCX) by a genomic mechanism but only at the base of the female rabbit epicardium. In adult female rabbit hearts, mRNA, protein (Cav1.2α) and current density of I_Ca,L were ~ 30% greater at the base of the epicardium than the apex or endocardium or adult male myocytes.⁴³ Treatment of freshly isolated female base, epicardial myocytes with an I_Kr blocker leads to cellular EADs when paced at long cycle lengths (1–2 s) but rarely elicited EADs in myocytes derived from other regions of the heart.⁴³ Taken together, the female sex is more vulnerable to LQT2-related TdP because of estrogen upregulation of Ca²⁺ influx (I_Ca,L) and Ca²⁺ efflux (I_NCX) mechanisms which more readily results in Ca²⁺ overload, EADs and TdP under conditions of delayed repolarization.

SUBCELLULAR Ca²⁺ DYNAMICS

Clearly, study of subcellular Ca²⁺ dynamics underlying EAD generation in the beating heart may fundamentally improve the understanding of the arrhythmogenic mechanisms. If SR Ca²⁺ release is indeed responsible for EADs, does it occur in the form of propagated Ca²⁺ waves as observed in the DAD models?

The subcellular Ca²⁺ dynamics in the beating heart has been studied with confocal microscopy by several research teams.^{16, 49–51} The advantage of confocal microscopy is its high z-axis resolution which comes at the price of low light levels due to confocal apertures, slow scan rates which limit these studies to line scans and the inability to correlate CaT events in different regions of a myocyte let alone adjacent myocytes. With conventional fluorescence microscopy and a high aperture, water-immersion 40X objective, allows excellent spatial and temporal resolution, with a z-axis resolution considerably smaller than the diameter of ventricular myocytes. The movie files from our subcellular experiments have spatial resolution of 1.5 μm per pixel (100×100 pixels) and a frame rate of 200 to 500 frames per second. Continuous perfusion with blebbistatin and custom-designed perfusion chamber were used to diminish contraction artifacts.

The subcellular optical recordings show that the secondary systolic Ca²⁺ oscillations corresponding to EADs occur synchronously throughout myocyte cytoplasm, in agreement with the report by De Ferrari et al.,³³ in isolated cells. This contrasted with propagated diastolic Ca²⁺ waves, which were clearly observed in many of the same experiments (Figure 5).

Figure 5.

Contrasting subcellular patterns of systolic and diastolic Ca²⁺ elevations in LQTS. A: Superimposed CaT tracings from 2 pixels of in the same myocyte. The SSCE occurs nearly simultaneously, and there is a low level of spatial heterogeneity of Ca²⁺ signal. B: In contrast, a wave of diastolic Ca²⁺ elevation propagates from top right to bottom left of the images, and can be easily seen in the left and middle image. Adapted from⁴²

Other, previously unreported forms of subcellular Ca²⁺ dynamics were also observed. These included brief diastolic Ca²⁺ elevations that did not propagate through the cell, low-amplitude Ca²⁺ oscillations (“ripples”) that occurred in on top of long CaT plateaus and in contrast to the typical Ca²⁺ oscillation, were not synchronous across the cell. On rare occasions, chaotic intracellular Ca²⁺ dynamics was observed, with multiple simultaneous wavefronts propagating through the cytoplasm of individual cell in various directions, with continuous collision, annihilation and generation of new wavefronts (Figure 6), with an overall pattern superficially resembling depolarization wavefront propagation through the ventricular myocardium during VF. These spatially dyssynchronous modes of subcellular of Ca²⁺ dynamics cannot be explained by activation of sarcolemmal Ca²⁺ channels.

Figure 6.

An example of disorganized subcellular Ca²⁺ dynamics. A: Five consecutive subcellular Ca_i images approximately 500 ms apart illustrate the presence of multiple intracellular waves. These waves may terminate with collisions, but new waves continue to arise and a stable state of low Ca²⁺ concentration is never achieved. B: Superposition of 3 Ca_i signal tracings (right) from the 3 pixels marked in the microscopic image (left). Vertical lines indicate the timing of the images in A. Adapted from⁴²

Overall, these results indicate that SSCE is a necessary precondition for EADs in our rabbit model, and the reactivation of L-type Ca²⁺current contributes to EADs after the initial depolarization caused by I_NCX.

DISCUSSION

The results of the optical mapping experiments described above lead us to suspect that the mechanisms of EAD generation share many similarities with DAD arrhythmogenesis, namely cellular Ca²⁺ overload, SOICR and depolarization mediated by NCX activity. In contrast to situations associated with DADs, repolarization delay has not been firmly linked to SR overload. However, there are good reasons to believe that repolarization delay increases SR Ca²⁺ load, since the influx of Ca²⁺ into the cell during each AP increases compared to baseline because the incomplete inactivation of the L-type Ca²⁺ channels due to the component of voltage-dependent inactivation during prolonged depolarization.⁵² At the same time, the efflux of Ca²⁺ out of the cell by NCX is attenuated because of decreased driving force during the long AP plateau. During long APs, SERCA continues to pump excess of Ca²⁺ ions in the lumen of the SR. Ca²⁺ transported in the SR increases with prolonged APD. After the initial Ca²⁺ release from the junctional SR (jSR) caused by the normal voltage-triggered CICR, the jSR is gradually replenished because of finite speed of SERCA transport and the diffusion limitation between non-junctional SR (where most SERCA pumps are located) and jSR – the site of RyR and Ca²⁺ release. We believe that in LQTS, jSR is replenished to a point which triggers SOICR during the prolonged AP plateau, causing the calcium oscillation (Figure 7). The rise of Ca²⁺ concentration in the cytoplasm then activates NCX. The resulting membrane depolarization triggers EAD by increasing L-type open channel probability. It is likely that in electrically connected tissue, the electrotonic currents related to V_m gradients contribute to this final step. The often invoked argument that NCX cannot play a major role in EAD triggering because EAD takeoff voltage is relatively close to NCX equilibrium potential may not be valid, since the this potential depends on cytoplasmic Ca²⁺ concentration and moves to more positive values when cytoplasmic Ca²⁺ increases, as is likely to be the case in our EAD model.

Figure 7.

Proposed mechanism of Ca²⁺ handling underlying EAD generation in repolarization delay (panels D–H); normal Ca²⁺ handling is shown in panels (A–C). Each panel shows a cartoon of T-tubule and SR cisterna, with selected Ca²⁺ handling molecules (labeled in A; Ca²⁺ ions are schematized as red dots). The corresponding phase of cardiac cycle is indicated in the insets, which show simultaneous membrane voltage signals in blue and Ca²⁺ signals in red. A In diastole, Ca²⁺ concentration in cytoplasm is low, RyRs are closed and Ca²⁺ fills both junctional and nonjunctional SR. B Depolarization of sarcolemma opens L-type Ca²⁺, which in turn open RyRs through CICR mechanism, leading to emptying of junctional SR and increase in cytoplasmic Ca2+. C During AP phase 3, opening of K⁺ channels repolarizes cell membrane, resulting in gradual closure of L-type channel. RyRs have closed, and SERCA and NCX transport Ca²⁺ back to SR and extracellular fluid. D and E In repolarization delay, CICR follows AP upstroke and opening of L-type Ca²⁺ channels, similar to A and B. F During the prolonged plateau phase, deactivation of L-type Ca²⁺ channels is delayed, and inactivation is incomplete, leading to persistent Ca²⁺ influx into cytoplasm. NCX activity is attenuated because of decreased driving force at depolarized membrane potentials. This indirectly accentuates the SERCA-mediated transport of Ca²⁺ onto SR compartment. G Diffusion of Ca²⁺ ions from the nonjunctional SR eventually increases junctional SR load to a level sufficient to cause SOICR; this occurs before CaT and membrane potential return to diastolic levels. H The cell-wide Ca²⁺ release results in increase of inward current carried by NCX, depolarizing sarcolemma and reactivating L-type Ca²⁺ channels, i.e. triggering EAD.

Several questions about this process remain to be answered. One of them is the reason for the difference between the regime of SR Ca²⁺ release underlying DADs (propagated Ca²⁺ waves) and EADs (cell-synchronous release). This difference could be caused by the relatively high cytoplasmic Ca²⁺ concentration at the time of SSCE takeoff. The RyR2 channels can be opened either by Ca²⁺ binding in a cytoplasmic site (possibly formed by the EF hand motif at the central domain of the cytoplasmic portion of the channel),^{53, 54} or by the Ca²⁺ binding to the luminal site at the S6 transmembrane domain of the channel.^{55, 56} Although Ca²⁺ binding to either site can open RyR and trigger Ca²⁺ release, they are likely to interact, i.e. increased Ca²⁺ concentration in the cytoplasm decreases the threshold for SOICR. One could then expect that during diastole, when cytoplasmic Ca²⁺ is low, the SOICR threshold might be only reached at one or a few sites, resulting in a Ca²⁺ propagating relatively slowly from that focus. Prior to systolic Ca²⁺ oscillation, the cytoplasmic Ca²⁺ is much higher and SOICR threshold might be substantially lower and may thus can be simultaneously reached at many junctional sites at approximately the same time.

Another puzzling observation is the unexpected degree of spatial heterogeneity in the amplitude of SSCE, with discontinuous, irregular and initially found in small “islands” of SSCE. Pre-existing differences in density of ion channels or expression of Ca²⁺-handling proteins most likely contribute these local effects. The findings of Sims et al., provide compelling evidence that only cells from specific locations on the heart exhibit EADs after I_Kr block.⁴³ Moreover, these cells prone to EADs are characterized by their higher levels of I_Ca,L and I_NCX.^{47, 48} Besides differences in ion channel expression, local differences in signaling could amplify small initial perturbations by a positive feedback mechanism. CaMKII phosphorylation could contribute to an increase the incidence of SSCEs through the phosphorylation of L-type Ca²⁺ channel, RyR2 and phospholamban and the increase in Cai could in turn augment CaMKII activity. Such a mechanism could explain why CaMKII inhibition suppresses EADs in rabbits with chronic heart block (prone to TdP).⁵⁷

Isolated cell experiments were used to study the mechanism of EAD generation. In a related paper by Horvath et al.,⁵⁸ experiments on myocytes with drug-induced LQT3 (gain of function of the late Na⁺ current) demonstrated that EADs are preceded by a rise of Ca_i. Moreover, EADs were eliminated when Ca_i was buffered with BAPTA even though AP duration remained prolonged. This study supports the primary role of non-voltage gated SR Ca²⁺ release as the cause of EADs (Figure 8).

Figure 8.

Patch-clamping experiments from a rabbit cell superfused with anemonetoxin II (ATX-II; an inhibitor of cardiac sodium channel inactivation) indicated that CaT oscillations are required to generate EADs. A: Simultaneous membrane voltage (top) and calcium signal (bottom) tracings in the whole cell, current clamp configuration, without calcium buffering. In the absence of ATX-II (gray tracings), both AP and CaT are short and monophasic. With ATX-II (black tracings), marked AP prolongation with multiple EADs occurs. The simultaneous CaT tracing demonstrates Ca2+ oscillations. Note that the takeoff of Ca²⁺ oscillations precedes the EAD rise. B: When the Ca²⁺ chelator BAPTA is present in the pipette solution, EADs are present in the perforated patch configuration (when BAPTA does not have access to the cytoplasm). With patch rupture and buffering of cytoplasmic Ca²⁺, EADs disappear, although AP duration remains long. Adapted from⁵⁸

The acquired long QT syndrome model we studied is likely to be clinically relevant – bradycardia and hypokalemia occur frequently in patients with a wide range of conditions, and scores of FDA-approved drugs block I_Kr to some degree. However, it is conceivable that EADs may form through a different mechanism in different situations. For example, in the study of isolated rabbit ventricular myocytes by Zhao et al.,⁵⁹ the EADs induced by a combination of isoproterenol and L-type Ca²⁺ channel opener and exhibited similar properties to those we observed in perfused hearts.⁵⁹ Significantly, systolic Ca²⁺ oscillations persisted even after forcing for a normal APD and morphology using the AP clamp technique, such that the highest CaT peak occurred after the termination of the clamped AP. These Ca²⁺ dynamics provide robust evidence against the Ca²⁺ “window current” mechanism as an EAD trigger.

Interestingly, EADs induced by oxidative stress with a high concentration of hydrogen peroxide exhibited a different behavior: the CaT rise followed V_m rise, and Ca²⁺ oscillations could be eliminated by an AP clamp to force a short AP, and not by blocking SR Ca²⁺ release. These properties implicate a Ca²⁺ “window current” mechanism. It has even been suggested that in mice (which have short AP duration and utilize different K⁺ currents from rabbits and humans to achieve ventricular repolarization), non-equilibrium gating of the cardiac Na⁺ current could play an important role in EAD generation.⁶⁰ Nevertheless, our results are mostly concordant with the data reported by Maruyama et al.⁶¹ in a similar rabbit model. Although these authors did not report subcellular data, they demonstrated highly irregular regions of CaT elevation during plateau and repolarization, and CaT rise preceding EAD takeoff.

It has been proposed^{62, 63} that AP prolongation gives rise to deterministic chaos in a certain range of cycle lengths and that electrotonic coupling of the cells operating in the “chaotic” regime results in spatial dispersion of APD and leads to ventricular arrhythmia. We have not tested this ingenious hypothesis in our experimental model, but we would argue that a mathematical framework based solely on dynamics of sarcolemmal currents, not including the intracellular Ca²⁺ handling dynamics, is likely incomplete. The inclusion of SOICR and SSCE in the whole-heart level mathematical models would certainly be interesting and probably quite challenging, since the relevant phenomena occur on both μm and cm scales.

TRANSLATIONAL ASPECTS

How do these findings fit into the clinical picture of TdP treatment, and what are the implications for arrhythmia management and SCD risk stratification in general? First, the model we propose predicts the protective effect of rapid pacing (~90 beats per minute), recommended in patients with runs of TdP. Regular and frequent “unloading” of jSR through the normal CICR mechanism expected during relatively rapid pacing may minimize the chance that jSR will fill over the SOICR threshold. On the other hand, a long pause may increase the chance of such an event, consistent with the “long-short” sequence which often initiates TdP. SSCEs of small amplitude typically cause a delay in CaT downstroke, can be observed in profound bradycardia in the absence of I_Kr block, increase APD dispersion and occasionally initiate arrhythmia.⁶⁴ High adrenergic tone could further increase SR Ca²⁺ load and promote arrhythmias during profound bradycardia in the clinical setting.

If SOICR and NCX activation are required to generate EADs, could NCX blockade and RyR stabilization have a role in TdP management? It is certainly possible, but the effect at the level of the whole organism may be difficult to predict. For example, partial NCX blockade might initially protect against EADs, but eventually aggravate SR overload unless AP duration shortens, with unpredictable effects. Similarly, stabilizers of RyR are potentially promising antiarrhythmic agents, and K201 suppressed TdP in our model, but it should be kept in mind that RyR2 mutation which profoundly decreases SOICR is actually associated with ventricular fibrillation in affected patients,^{65, 66} probably because it increases jSR load during β-adrenergic stimulation.

The abnormal Ca²⁺ handling in LQTS is reflected by changes on a surface EKG. The local Ca²⁺ handling dynamics – i.e. the amplitude, precise timing and number of SSCEs – differ among different ventricular regions before arrhythmia onset. The SSCE pattern changes over the course of several seconds, presumably reflecting variable capacity of different ventricular regions to respond to increasing SR load. Although different myocardial regions are electrically coupled, resulting in partial smoothing of CaT through the AP effects on Ca²⁺ handling, the local heterogeneities of Ca²⁺ handling eventually overcome the tissue coupling. This dynamic interplay between CaT and AP before TdP onset is reflected in beat-to-beat changes of local CaT and AP, and globally as beat-to-beat changes of the T wave morphology. Beat-to-beat repolarization variability increases prior to TdP onset in LQTS⁴⁰ and produces non-alternans patterns of repolarization or T-wave lability. T-wave lability has been reported prior to TdP onset in patients with congenital LQTS (Figure 9),⁶⁷ in a canine model of acquired LQTS involving chronic complete heart block,⁶⁸ and was found to be predictive of arrhythmia in a rabbit heart exposed to a wide array of repolarization-prolonging agents.⁶⁹ In congenital LQTS patients, aperiodic T wave lability during catecholamine infusion correlates better with clinical status than corrected QT interval duration.⁶⁷ Interestingly, non-alternans repolarization lability has been also seen in a murine CPVT model,⁷⁰ indicating that it may not be unique to long QT syndrome. The question whether aperiodic T wave lability precedes arrhythmia onset in more common cardiac conditions and might provide a general risk-stratification tool for SCD remains unanswered at this time, but non-alternans repolarization variability prior to onset of ventricular arrhythmias has been reported in patients with structural heart disease and left ventricular dysfunction in ambulatory EKG tracings⁷¹ and in intra-cardiac electrograms stored by implantable defibrillators.⁷² In principle, repolarization lability might allow noninvasive, real-time monitoring of the arrhythmogenic process minutes before arrhythmia onset at least in some clinical settings.

Figure 9.

Non-alternans repolarization lability precedes TdP onset in a patient with congenital LQTS during dobutamine infusion. A: Before dobutamine infusion, sinus rhythm without obvious repolarization lability is recorded. B: During dobutamine infusion (dose of 10 μg/kg/minute), sinus rhythm is still present without ectopic beats, but the terminal component of the T-wave fluctuates dramatically in amplitude (red arrows) without an alternans pattern. C: With increased dobutamine dose (20 μg/kg/minute), TdP develops, preceded by polymorphic ventricular ectopy. D: The repolarization lability can be easily appreciated when 4 consecutive sinus beats are superimposed and color-coded. At baseline (top left), the repolarization lability is minimal. Striking T wave lability is evident during dobutamine infusion (top right and bottom left). This disappears after electrical cardioversion of TdP, cessation of dobutamine infusion and β-blocker administration (bottom right). Similar repolarization lability is observed in the rabbit LQTS model, when rises along with lability of Ca_i and V_m signals (not shown). Adapted from⁶⁷

A most interesting question is whether arrhythmia mechanisms involving SOICR play a role in ventricular arrhythmias associated with coronary ischemia or heart failure, which are much more complex than pure repolarization delay. However, various forms of repolarization impairment and abnormal Ca²⁺ handling are invariably present in heart failure models. One of the better defined pathways involves activation of CaMKII and phosphorylation of β_2a subunit of the L-type Ca²⁺ channels (augmenting Ca²⁺ influx),⁷³ RyR2 (promoting Ca²⁺ release)^{74, 75} and the Nav1.5 Na⁺ channel (augmenting the late sodium current and prolonging repolarization).^{76, 77} At this moment, the extent of similarities in arrhythmogenic mechanisms between LQTS and heart failure remains uncertain.

CONCLUSION

Normal cardiac function involves propagation of electrical signals through the cardiac chambers; local action potential leads to Ca²⁺ entry into the cell and release from SR in individual cardiomyocytes, resulting in synchronized contraction. The main mechanism of Ca²⁺ removal from myocytes is the electrogenic NCX exchanger. Therefore, disturbance of myocyte Ca²⁺ handling can in turn affect membrane potential via changes in NCX current and cause arrhythmia. This appears to be the case in long QT syndrome, as prolonged action potential increases the amount of Ca²⁺ entering SR compartment during each cardiac cycle. Multiple forms of abnormal Ca²⁺ handling regimes occur on the subcellular level in this setting. It possible that arrhythmogenic mechanisms involved in long QT syndrome play some role in sudden cardiac death associated with more common cardiac conditions.