Revisiting the ionic mechanisms of early afterdepolarizations in cardiomyocytes: predominant by Ca waves or Ca currents?

Abstract

Early afterdepolarizations (EADs) have been implicated in severe cardiac arrhythmias and sudden cardiac deaths. However, the mechanism(s) for EAD genesis, especially regarding the relative contribution of Ca²⁺ wave (CaW) vs. L-type Ca current (I_Ca,L), still remains controversial. In the present study, we simultaneously recorded action potentials (APs) and intracellular Ca²⁺ images in isolated rabbit ventricular myocytes and systematically compared the properties of EADs in the following two pharmacological models: 1) hydrogen peroxide (H₂O₂; 200 μM); and 2) isoproterenol (100 nM) and BayK 8644 (50 nM) (Iso + BayK). We assessed the rate dependency of EADs, the temporal relationship between EADs and corresponding CaWs, the distribution of EADs over voltage, and the effects of blockers of I_Ca,L, Na/Ca exchangers, and ryanodine receptors. The most convincing evidence came from the AP-clamp experiment, in which the cell membrane clamp was switched from current clamp to voltage clamp using a normal AP waveform without EAD; CaWs disappeared in the H₂O₂ model, but persisted in the Iso + BayK model. We postulate that, although CaWs and reactivation of I_Ca,L may act synergistically in either case, reactivation of I_Ca,L plays a predominant role in EAD genesis under oxidative stress (H₂O₂ model), while spontaneous CaWs are a predominant cause for EADs under Ca²⁺ overload condition (Iso + BayK model).

early afterdepolarizations (EADs), delayed afterdepolarizations (DADs), and triggered activities (TAs) are implicated in arrhythmias and sudden cardiac deaths (4). Although afterdepolarizations and TAs have been extensively studied, their underlying mechanisms remain incompletely understood. While there is a consensus on the role of spontaneous sarcoplasmic reticulum (SR) Ca²⁺ release and Ca²⁺ waves (CaWs) in the generation of DADs (36), significant discrepancies exist regarding the mechanisms of EADs. For example, early experimental and computer model studies (14, 19, 44) suggested EADs are exclusively caused by the reactivation of L-type calcium current (I_Ca,L) without the involvement of CaWs. However, accumulating evidence obtained from recent experimental (33, 40) and computer simulation studies (13) support a mechanism involving CaW for EAD formation under Ca overload conditions. Thus it seems likely that EAD generation can be mediated by at least two mechanisms: 1) sarcolemma-dependent (or I_Ca,L-dominant) mechanism: enhancement of inward currents during repolarization, such as I_Ca,L; and 2) SR-dependent (or CaW-dominant) mechanism: Na/Ca exchange current (I_NCX) or transient inward current (I_ti) initiated by spontaneous SR Ca²⁺ release through the ryanodine receptor (RyR) under Ca²⁺ overload conditions. Furthermore, these two mechanisms (i.e., I_Ca,L and I_NCX) are considered to be highly interactive and function synergistically to induce EADs (41). However, disputes still remain on the relative contribution of different mechanisms to EAD generation in different models (9, 31, 40). It is unquestionable that a better understanding of the factors playing a primary role in EAD formation may be helpful for developing therapeutic approaches.

In our previous studies, we have established the following two different EAD models in isolated rabbit ventricular myocytes: 1) hydrogen peroxide (H₂O₂) model: EADs induced by reactive oxygen species (ROS) and H₂O₂ (42); and 2) Iso + BayK model: EADs induced by isoproterenol (100 nM) in addition to BayK 8644 (50 nM) (43). Here, to further clarify the mechanisms accounting for EADs under different conditions, we make systematic comparisons between these two types of EADs. We have provided more convincing evidence suggesting that I_Ca,L reactivation and CaW may account for distinct predominant ionic mechanisms under different conditions, although they may act synergistically to generate EADs.

MATERIALS AND METHODS

Cell isolation.

Single ventricular myocytes were enzymatically isolated from adult rabbit hearts. Briefly, the hearts were removed from adult New Zealand White rabbits (2–3 kg), anesthetized with intravenous pentobarbital sodium, and hearts were perfused retrogradely in Langendorff fashion at 37°C with nominally Ca²⁺-free Tyrode solution containing ∼1.4 mg/ml collagenase (type II; Worthington) and 0.1 mg/ml protease (type XIV; Sigma) for 25–30 min. After the enzyme solution was washed out, the hearts were removed from the perfusion apparatus and swirled in a culture dish. The myocytes were isolated from left ventricles without specific separation from different layers. The Ca²⁺ concentration was slowly increased to 1.8 mM, and the cells were stored at room temperature and used within 8 h. The use and care of the animals in these experiments were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Medicine and Dentistry of New Jersey–New Jersey Medical School.

Patch-clamp methods.

Myocytes were patch clamped using the perforated whole cell configuration of the patch-clamp technique in the current-clamp or voltage-clamp mode. Recording pipettes (resistance, 2–4 MΩ) were filled with internal solution containing the following (in mM): 110 K⁺-aspartate, 30 KCl, 5 NaCl, 10 HEPES, 0.1 EGTA, 5 MgATP, 5 Na₂-creatine phosphate, and 0.05 cAMP (pH 7.2, adjusted with KOH). Myocytes were superfused with normal Tyrode solution containing the following (in mM): 136 NaCl, 5.4 KCl, 0.33 Na₂PO₄, 1.8 CaCl₂, 1 MgCl₂, 10 glucose, and 10 HEPES (pH 7.4, adjusted with NaOH). Action potentials (APs) were elicited with 2-ms, 2- to 4-nA square pulses at various pacing cycle lengths (PCLs). In some cells, Ca²⁺ transients/waves were recorded under current-clamp or AP-clamp (voltage-clamp with a fixed AP waveform) conditions. Electrical signals were measured with an MultiClamp 700A patch-clamp amplifier, controlled by a personal computer using a Digidata 1322A interface, driven by pCLAMP 10 software.

Intracellular Ca²⁺ concentration measurement.

Myocytes were loaded with the Ca²⁺ indicator fluo-4 by incubating them for ∼30 min in bath solution containing 4 μM fluo-4 AM (Molecular Probe), after which the cells were washed and placed in a heated chamber on an inverted microscope. Intracellular Ca²⁺ concentration ([Ca²⁺]_i) fluorescence was recorded using an Andor Ixon charge-coupled device camera (Andor Technology) operating at ∼100 frame/s with a spatial resolution of 500 × 400 pixels. Fluorescence intensity was measured as the ratio of fluorescence (F) over the basal diastolic fluorescence (F₀).

Chemicals.

SEA0400 was synthesized by Taisho Pharmacutical (Saitama, Japan). Nifedipine and BayK 8644 were first dissolved in DMSO as stock solutions before being diluted into the bath solution to the final concentration. The maximum DMSO concentration was <0.2% (vol/vol). Chemicals and reagents were purchased from Sigma unless indicated. All experiments in the present study were carried out at 35–37°C.

Statistical analysis.

The incidence of EADs within groups of 20 consecutive AP recordings was analyzed by Fisher's exact test. All other data are shown as means ± SD. Statistical differences were evaluated with Student's paired or unpaired t-tests. P < 0.05 is considered as statistically significant.

RESULTS

Rate dependence of EADs in different models.

To compare systematically the properties and determine the mechanisms of EADs under different conditions, we have established the following two EAD models using different pharmacological interventions: 1) isoproterenol (Iso; 100 nM) + BayK 8644 (BayK; 50 nM); and 2) H₂O₂ (200 μM).

We first assessed the PCL dependence of EADs in these two models. In the H₂O₂ model, EADs were consistently observed at a PCL of 6 s after the cell was exposed to 200 μM H₂O₂ for >6 min, while no EADs were observed when the PCL was shortened to 1 s (Fig. 1A). Conversely, EADs were apparently induced by Iso + BayK at a PCL of 1 s but disappeared at a PCL of 6 s (Fig. 1B). The incidence of EAD was examined by counting the number of EADs within 20 consecutive APs and the PCL dependence of EAD incidence in the different models plotted in Fig. 1C. It was clearly shown that the EAD incidence rate was higher at a long PCL (or slow rate) in the H₂O₂ model but at a short PCL (or fast rate) in the Iso + BayK model. These results suggest that H₂O₂-induced EADs are slow-rate dependent, consistent with our previous report (29), while Iso + BayK-induced EADs are fast-rate dependent.

Fig. 1.

Pacing cycle length (PCL) dependence of early afterdepolarization (EAD) induction in different models. A: action potentials (APs) recorded from rabbit ventricular myocytes treated with 200 μM H₂O₂ (for 8 min in this case). Representative APs at PCL of 1 s (top) and 6 s (bottom) are shown. EADs were observed at long PCL of 6 s (arrows) but not at PCL of 1 s. B: APs recorded from myocytes treated with 100 nM isoproterenol and 50 nM BayK 8644 (Iso + BayK). EADs (arrows), delayed afterdepolarizations (DADs; d), and triggered AP (T) were induced at PCL of 1 s (top) but not at PCL of 6 s (bottom). C: summarized bar graphs showing the incidence of EADs within 20 APs at various PCLs in the 2 models (n ≥ 4 cells). The EAD incidence rate was higher at a low pacing rate (or long PCL) in the H₂O₂ model but at fast pacing rate (or short PCL) in Iso + BayK model. ***P < 0.0001, Fisher's exact test vs. no incidence of EADs.

Formation processes of EADs in different models.

Different pacing rates may affect the level of [Ca²⁺]_i. For example, pacing a myocyte at a rapid rate is thought to facilitate [Ca²⁺]_i overload (7, 13). Thus, in the following experiments, we simultaneously recorded [Ca²⁺]_i transients (CaTs) and APs and compared the different behaviors of CaTs during EAD formation between the two models.

In the H₂O₂ model, EADs normally emerged 5–10 min after H₂O₂ (200 μM) perfusion. As shown in Fig. 2A, H₂O₂-induced spontaneous Ca transient (SCaT) occurred only when EADs were also observed in the corresponding APs. A significant increase of CaT amplitude was observed after the first EAD and SCaT occurred, suggesting sarcolemmal Ca²⁺ influx and SR Ca²⁺ content were enhanced during EADs, probably due to reactivation of I_Ca,L. Consistent with this notion, our and other's previous studies (32, 42) have clearly demonstrated that H₂O₂ activates I_Ca,L under either square pulse voltage clamp or AP waveform clamp.

Fig. 2.

Formation process of EADs in different models. A: simultaneous recordings of whole cell (global) Ca²⁺ transients (CaTs) and APs at 5 min after treatment with 200 μM H₂O₂. Note the increase of CaT amplitude (F/F₀) following the emergence of EADs. B: same as A, except that the myocyte was treated with Iso + BayK (10 s after treatment). Note the Ca²⁺ accumulation before the emergence of EADs. Spontaneous Ca²⁺ transients or Ca²⁺ waves (SCaTs/CaWs) are indicated by ○ and *, which correspond to APD prolongation and EADs (arrows), respectively. Spontaneous CaT (S) and triggered APs (T) are also shown.

In the Iso + BayK model, apparent changes in CaT and AP were caused with a very fast time course. In the case shown in Fig. 2B, both diastolic Ca²⁺ level and CaT amplitude gradually increased at ∼10 s after Iso + BayK treatment. Particularly, apparent SCaTs were observed from the sixth CaT with their amplitudes increasing progressively, correlating to the gradual prolongation of APDs at first, and then frank EADs (2B, bottom) when the amplitudes of corresponding SCaT were higher. These results suggest Iso + BayK treatment leads to a progressive accumulation of [Ca²⁺]_i before EAD emergence. The emergence of CaWs/SCaTs preceded Iso + BayK-induced EADs, indicating CaWs were generated from spontaneous SR Ca²⁺ release rather than I_Ca,L reactivation. CaWs then caused I_ti, which contributed to EAD genesis in the Iso + BayK model.

Temporal relationship between EADs and corresponding SCaTs/CaWs.

To further determine the primary cause (CaW vs. I_Ca,L) of EADs in each model, we next examined temporal correlation between EADs and SCaTs/CaWs by comparing the onsets of EAD upstrokes and the initiation of corresponding SCaTs/CaWs. It is conceivable that if an EAD is generated by a I_Ca,L-dominant mechanism, the corresponding SCaT will be the consequence of the EAD and therefore will occur subsequent to the EAD upstroke. On the other hand, if the EAD is generated by an SR-dependent mechanism, the spontaneous Ca release/CaW will be the cause of the EAD and would therefore precede the EAD upstroke. This rationale can be clearly discriminated, as illustrated by the results shown in Fig. 3.

Fig. 3.

Temporal relationship between EAD and DAD corresponding SCaTs/CaWs in different models. Whole cell CaTs and APs were simultaneously recorded to compare the initiation time of an EAD upstroke (arrows) and corresponding SCaT/CaW (dashed lines). A: representative recording from H₂O₂ model. EAD was 130 ms ahead of corresponding SCaT. B: in Iso + BayK model, the start of a CaW preceded the upstroke of the EAD with a time difference (ΔT) of 50 ms in this representative cell. C and D: temporal relationship between DAD and CaW in H₂O₂ and Iso + BayK models, respectively. While the temporal relationship between EAD and SCaT remained the same as A and B, the DADs (d) and DAD-triggered APs (T) in both models seemed to always occur (as indicated by ▴) behind the initiation of their corresponding CaW (#), suggesting Ca overload and Na/Ca exchange current (I_NCX) as a common mechanism for the DADs in both models.

In the H₂O₂ model (Fig. 3A), the start of the EAD upstroke was significantly earlier than the initiation of the corresponding SCaT. The difference in initiation time (ΔT) was 130 ms for this specific cell, and 113.5 ± 31.1 ms on average (means ± SD, n = 8). In the Iso + BayK model (Fig. 3B), however, the initiation of SCaT preceded corresponding EAD upstroke. The ΔT was 50 ms for this specific cell, and 44.2 ± 8.7 ms on average (n = 9).

As previously mentioned, DADs were also induced in both models but with different time courses. While DADs or DAD-induced TAs always occurred in association with EADs in the Iso + BayK model (see Figs. 2–7), H₂O₂-induced DADs were only observed occasionally, required prolonged treatment with H₂O₂, and arose much later than the emergence of EAD. In both the H₂O₂ and Iso + BayK models (Fig. 3, C and D), it is apparent that the DADs were initiated after the beginning of corresponding CaWs/SCaTs, suggesting that Ca²⁺ overload-induced I_NCX is most likely to account for the DAD generation in both cases.

Fig. 7.

Effects of ryanodine (10 μM), a ryanodine receptor inhibitor, on EADs and corresponding SCaTs/CaWs. A: CaT and AP recording under control condition (left), after H₂O₂ treatment (middle) and the effect of ryanodine (Rya) on H₂O₂-induced EADs (arrows) and CaTs (*; right). B: CaT and AP recording under control condition (left), after Iso + BayK treatment (middle), and the effect of ryanodine (right). Iso + BayK-induced CaWs (* and #, corresponding to EAD and DAD, respectively), EADs (arrows), and DAD-triggered action potential (T) are indicated.

Distribution of EAD take-off potentials.

Since some ionic currents (e.g., I_Ca,L) highly depend on membrane voltages, the take-off potentials (TOPs) at which the EADs occurred were also compared between these two models. In the H₂O₂ model, EADs were consistently aroused at the end of phase 2 or early phase 3 of APs. The TOPs were within a narrow range (from +20 to −30 mV; Fig. 4A), which was consistent with that mediating the window current of I_Ca,L (14).

Fig. 4.

Distribution of take-off potentials (TOPs) of EADs in H₂O₂ and Iso + BayK models. A: representative recording of H₂O₂-induced EADs (top; stimulation marks are indicated below the trace) and a histogram graph showing distribution of their TOPs. Width of each bar is 3 mV. Data showed in the histogram were from 9 cells. B: same as A, except for Iso + BayK-induced EADs. Data showed in the histogram were from 7 cells.

On the other hand, the Iso + BayK-induced EADs occurred over a broader voltage range to a more negative level. As shown in Fig. 4B, the TOPs distributed over the whole AP repolarization phase (phases 2 and 3; from +20 to −60 mV). In addition, DADs and DAD-induced TAs (at phase 4) always appeared together in the same cell. These results suggest that in the Iso + BayK model, the charge carriers mediating EADs (as well as DADs) can be activated over a broad range of membrane potentials, in agreement with the voltage dependence of I_NCX.

Relation between SCaTs/CaWs and EADs revealed by AP-clamp experiments.

As pointed out previously, a SCaT/CaW might be either the consequence (i.e., sarcolemma-dependent mechanism) or the cause (i.e., SR-dependent mechanism) of an EAD. The following rationales were tested in a setting of AP-clamp experiments: If a SCaT is the consequence of an EAD, it should be suppressed by preferentially eliminating the EAD. Conversely, if a SCaT/CaW is the cause of an EAD, it should remain even if the EAD is preferentially eliminated.

In either model, EADs and SCaTs/CaWs were initially elicited in a cell under current-clamp mode. Thereafter, the myocyte was switched to voltage-clamp mode and the membrane potential was clamped under a normal AP waveform without EAD, which was recorded previously from a control myocyte. In the H₂O₂ model, the SCaT was completely eliminated by switching current-clamp to AP-clamp mode with a normal AP morphology (without EAD; Fig. 5A), indicating that the SCaT was caused by the corresponding EADs. In other words, the sarcolemma-dependent (or I_Ca,L-dominant) mechanism plays a predominant role in EAD genesis in H₂O₂ model.

Fig. 5.

Behaviors of SCaTs/CaWs under AP-clamp condition. After EADs were induced under current-clamp configuration (left), the recording from the same cell were switched to voltage-clamp configuration under an AP morphology without EAD (right) in H₂O₂ (A) and Iso + BayK (B) model. AP was recorded previously from a control cell. Note the SCaTs/CaWs were completely eliminated under AP-clamp condition in H₂O₂ model, while they persisted in Iso + BayK model.

In the Iso + BayK model (Fig. 5B), however, the SCaTs/CaWs persisted even after the cell was switched to voltage-clamp under a normal AP without EADs, suggesting that SCaTs/CaWs occurred independently from membrane potential and seemed to be the primary cause for EAD formation under [Ca²⁺]_i overload condition. These results have provided the most convincing evidence on the cause-effect relationship between EADs and SCaTs/CaWs under different conditions.

SR Ca²⁺ load and SCaT/CaW properties in the two EAD models.

To directly evaluate the SR Ca²⁺ content, the amplitudes of Ca transients induced by rapid exposure to 10 mM caffeine were measured. We observed SR Ca²⁺ was enhanced rapidly by Iso + BayK, while there was no significant change in the H₂O₂ model (Table 1), suggesting an intracellular Ca²⁺ overload status in the Iso + BayK model. Consistently, our results also revealed a significant elevation of CaT amplitude by Iso + BayK treatment but not by H₂O₂ treatment. The average amplitude of SCaT (or CaW) in the Iso + BayK model (F/F₀ = 1.65 ± 0.24 was higher than that in H₂O₂ model at F/F₀ = 1.26 ± 0.09; P < 0.01). It is also very interesting that while the SCaTs always had a lower amplitude compared with their preceding AP-elicited CaTs in H₂O₂ model, the amplitude of SCaTs in Iso + BayK model exhibited a rather wide range and were even higher than their preceding AP-elicited CaTs at certain beats (e.g., Figs. 6 and 7). These results suggest that the SCaT in the H₂O₂ model occur as a secondary event (triggered by reactivated I_Ca,L) when the SR Ca was partially depleted by the previous AP-elicited CaTs. On the contrary, the SCaTs/CaWs in the Iso + BayK model most likely represent primary SR Ca releases.

Table 1.

Comparison of Ca²⁺ handling properties in the two EAD models

Model	CaT at Control	CaT After Treatment	Spontaneous CaT/CaW	SR Content at Control	SR Content After Treatment
H2O2	1.56 ± 0.27 (48)	1.54 ± 0.32 (42)	1.26 ± 0.09† (27)	1.67 ± 0.57 (19)	1.52 ± 0.35 (19)
Iso + BayK	1.44 ± 0.20 (44)	1.75 ± 0.37* (49)	1.65 ± 0.24 (32)	1.62 ± 0.35 (26)	1.99 ± 0.54‡ (24)

Values are means ± SD; numbers of calcium transients (CaTs) measured from multiple cells are indicated in the parenthesis. Amplitude (F/F₀) of CaT, spontaneous CaT/calcium wave (CaW), and 10-mM caffeine-induced calcium release [sarcoplasmic reticulum (SR) content] are listed.

EAD, early afterdepolarization; Iso, isoproterenol; BayK, BayK 8644.

^*P < 0.01, compared with CaT at control;

^†P < 0.01, compared with CaT after treatment;

^‡P < 0.01, compared with SR content at control.

Fig. 6.

Effects of I_NCX inhibition on EADs and corresponding SCaTs/CaWs. A: effect of SEA0400, a selective I_NCX blocker, on H₂O₂-induced EADs (arrows) and CaTs (*). B: Iso + BayK-induced CaWs (* and #, corresponding to EAD and DAD, respectively), EADs (arrows), and DAD-triggered action potential (T; left) and the effect of SEA0400 (right). DADs (d) are indicated at right.

Effects of the I_NCX inhibitor SEA0400.

All above observations strongly suggest the contribution of I_Ca,L vs. I_NCX may vary (i.e., one of them is likely to play a predominant role) in EAD generation under different conditions. To determine directly the involvement of I_Ca,L and I_NCX, the effects of I_Ca,L and I_NCX inhibitors on the EADs and [Ca²⁺]_i handling were evaluated in the two different EAD models.

SEA0400 has been used as a tool inhibitor of I_NCX (5, 20). In the H₂O₂ model (Fig. 6A), SEA0400 not only abolished EADs, but also eliminated SCaTs, resembling the AP-clamp experiment shown in Fig. 5A. This result indicates that, although the primary current accounting for EAD formation is suggested to be the reactivation of I_Ca,L in the H₂O₂ model, the inhibition of I_NCX may reduce the inward current that is necessary for the generation of EADs.

In the Iso + BayK model, EADs were also efficiently abolished by SEA0400 (2 μM). However, SEA0400 exerted less effect on [Ca²⁺]_i behavior such that the SCaTs/CaWs persisted (Fig. 6B), suggesting SEA0400 functioned at the downstream targets (i.e., NCX) of CaWs.

SEA0400 is the most potent Na⁺/Ca²⁺ exchanger blocker available, although it is not completely specific and also inhibits L-type Ca²⁺ channels (5). While the level of SEA0400 we used (2 μM) may cause the maximal blockade of inward I_NCX (5), it only caused a small degree of inhibition of I_Ca,L (∼15%, from 1.56 ± 0.0.58 nA to 1.32 ± 0.38 nA; n = 7) , but this decrease was not significant by statistical standards (P = 0.13). Therefore, we assume that the effect of SEA can be mostly attributed to Na⁺/Ca²⁺ exchanger in the setting of our present study, although we cannot completely rule out the contribution of I_Ca,L (5). The different behaviors of SEA0400 in the two different EAD models (Fig. 6), particularly on CaWs, are consistent with its major role on I_NCX.

Effects of the I_Ca,L blocker nifedipine.

Next, we examined the effect of nifedipine, a selective I_Ca,L blocker on EADs and SCaTs/CaWs elicited in both H₂O₂ and Iso + BayK models. Nifedipine (10 μM) completely abolished both EADs and SCaTs in both models (data not shown). These results are conceivable since nifedipine can both suppress inward I_Ca,L and attenuate Ca_i overload by reducing Ca²⁺ influx.

Effects of the RyR blocker ryanodine.

Since spontaneous Ca²⁺ release from SR is mediated by the Ca release channel or RyR, we assessed the effect of ryanodine at 10 μM, a concentration that is thought to selectively inhibit RyR activities (15). As shown in Fig. 7, ryanodine suppressed CaTs in both H₂O₂ and Iso + BayK models, confirming its inhibitory effect on RyR and Ca²⁺-induced Ca²⁺ release. However, different effects on EADs were observed in the two models. In the H₂O₂ model, EADs were still present, although APDs were further prolonged, presumably due to reduced I_Ca,L inactivation (Fig. 7A). In the Iso + BayK model, however, both EADs and DADs were eliminated (while the corresponding CaWs were also removed) by ryanodine treatment.

DISCUSSION

EADs are abnormal voltage oscillations occurring during the repolarizing phase of cardiac APs and are thought to cause cardiac arrhythmias. It has been suggested that EADs occur under conditions of reduced repolarization reserves (3, 26). Thus either increased inward currents or reduced outward currents, or both, promote EAD generation. For example, activation of inward late sodium current (I_Na), I_Ca,L , I_NCX, chloride current (I_Cl), as well as blockage of outward I_Kr, I_Ks, or I_K1 have all been reported to mediate EAD genesis. Among these ionic mechanisms, CaW-mediated I_NCX (SR-dependent mechanism) and reactivation of I_Ca,L (sarcolemma-dependent mechanism) have been suggested as two major contributors for EAD genesis. However, their relative contributions underlying different pathological conditions are still under debate. In the present study, we conducted systematic comparison of two cellular models and focused on investigating the relative role of I_Ca,L vs. CaW-induced I_NCX in generating EADs by simultaneously recording APs and CaTs. We have provided important clues to identify different EAD mechanisms. For example, one significant result was that abnormal SCaT accompanying EADs were abolished when the EADs were eliminated (by voltage clamp) in the H₂O₂ model. However, this was not the case for Iso + BayK-induced SCaT/CaW and EADs, i.e., CaWs persisted even when EADs were eliminated (Fig. 5). Our results provide convincing evidence for the relative contributions of CaW-induced I_NCX and I_Ca.L in different EAD models. These two mechanisms are not mutually exclusive. They function coordinately to cause EADs; however, one mechanism may play a predominant role under a certain pathological condition.

H₂O₂ model: I_Ca,L-predominant mechanism for EAD.

ROS (including H₂O₂) have been suggested to mediate disturbances in the cardiac rhythm under various circumstances, such as heart failure, aging, and ischemia/reperfusion (22, 37). In contrast to Iso + BayK, H₂O₂ seems to elicit EADs primarily by modulating membrane ionic channels. For example, we and others (32, 42) have shown that H₂O₂ (0.2–1 mM) increased both peak and plateau currents of I_Ca,L and late I_Na in rabbit ventricular myocytes. The activation of CaMKII, either directly via oxidative stress, or indirectly via elevated Ca_i, is critical for these effects, which contribute to the arrhythmogenic potential of oxidative stress.

Consistent to classical EADs caused by sarcolemma-dependent mechanism, H₂O₂-induced EADs have been shown to be dependent on slow pacing rate (bradycardia). A previous computer modeling study (16) suggests that the delayed rectifier potassium current (I_Ks), which has a long activation time course and slow deactivation process, plays a key role in facilitating EAD generation at a slow pacing rate. In our most recent study (Zhao Z, Xie Y, Wen H, Xiao D, Allen C, Fefelova N, Dun W, Boyden PA, Qu Z, Xie LH, unpublished observations), we have demonstrated that I_to is also an essential determinant for the slow rate dependence of EADs in rabbit ventricular myocytes due to its very slow recovery from inactivation. I_to, despite being a pure outward current, can potentiate EADs by lowering the AP plateau quickly into more negative voltages to allow I_Ca,L reactivation before I_Ks is fully activated to repolarize the myocyte, causing voltage oscillations in the plateau phase and thus EADs.

In addition to the difference in rate dependence, a difference in the TOPs of EAD upstrokes was also found between the two models. H₂O₂-induced EAD developed within a narrow voltage range (−30 to +20 mV), which is consistent with the I_Ca,L window current. Other evidence such as CaT amplitude increase after EAD emergence, the temporal precedence of EAD upstrokes, and elimination of CaWs by clamping cell membrane potential with a normal AP (without a EAD) all support the primary role of I_Ca,L-dominant mechanism in EAD generation in H₂O₂ model.

I_NCX functions predominantly in the forward mode, generating inward current during most of the AP repolarization. This is likely to lengthen the APD and form a “conditioning phase,” which facilitates reactivation of I_Ca,L and/or I_Na and generation of EAD upstroke (14, 40). It has been shown that various I_NCX blockers (e.g., SEA0400, SN-6, and inhibitory peptide) shorten APDs under regular condition (1, 24, 35). Our present study showed that SEA0400 also eliminate H₂O₂-induced (I_Ca,L-predominant) EADs, as well as accompanying SCaTs (different from Iso + BayK model), which further support the “conditioning phase” mechanism. The SCaTs in the H₂O₂ model were most likely formed by Ca²⁺ entry through reactivated I_Ca,L and SR Ca²⁺-induced Ca²⁺ release triggered by the I_Ca,L reactivation.

Iso + BayK model: CaW-predominant mechanism for EAD.

It is well accepted that I_NCX driven by CaW accounts for DAD formation (27, 34, 36). Recent experimental and simulation studies (13, 33) have demonstrated that CaW is also capable of evoking EADs. Pure β-adrenergic stimulation (Iso alone) has been used to cause EADs and DADs in dog myocytes (25, 39), and these could be facilitated by caffeine-enhanced SR Ca²⁺ leak (23). In rabbit myocytes, however, Iso alone cannot efficiently induce EADs/DADs. This may suggest a high threshold for SR Ca²⁺ spontaneous release in rabbit myocytes. In the present study, we were able consistently to generate frequent CaW and EADs/DADs by the application of Iso in the presence of BayK, as we demonstrated in our previous study (43). In this model, the SR Ca²⁺ content may be increased by the activation of both I_Ca,L and sarco(endo)plasmic reticulum Ca²⁺-ATPase activity. In addition, BayK has also been reported to increase RyR Ca²⁺ leak (17), perhaps through a functional linkage between the sarcolemmal dihydropyridine receptor and the SR ryanodine receptor (30). This may decrease the threshold for generation of CaW. In addition, the activation effect of BayK on I_Ca,L (via changing both voltage dependence and single channel gating; Refs. 2, 28) may also lead to additional Ca²⁺ overload. Therefore, spontaneous SR Ca²⁺ release/CaW-induced elevation of cytosolic Ca²⁺ may depolarize myocytes by the electrogenic I_NCX, which is most likely to primarily mediate EAD generation in Iso + BayK model.

Since intracellular Ca²⁺ overload is exaggerated by fast pacing, EADs caused by spontaneous Ca²⁺ release should be aggravated by fast pacing and tempered by slow pacing. This is proved to be true when we compared the EAD genesis at PCL of 6 vs. 1 s. It should be noted that Iso + Bay-induced EADs may be suppressed by further shortening PCL (i.e., from 800 to 300 ms), probably due to less I_Ca,L recovery from inactivation when pacing rate is so fast.

Several other lines of evidence, such as CaT changes before or after EAD emergence, the temporal precedence of SCaT/CaW, less voltage dependence of EAD distribution, and suppression of EADs and DADs by inhibiting RyR with ryanodine, support the primary role of CaWs in EAD generation under Iso + BayK condition. The most convincing evidence came from the AP-clamp experiment (see Relation between SCaTs/CaWs and EADs revealed by AP-clamp experiments), in which the CaWs remained to appear even if the cell membrane APs were clamped without EADs. In addition, Ca handling (e.g., SR Ca²⁺ content and SCaT amplitude) analysis provided direct evidence (see SR Ca²⁺ load and SCaT/CaW properties in the two EAD models) for intracellular Ca²⁺ overload in Iso + BayK model. Thus it is most likely that SCaT/CaWs-activated I_NCX function as a primary underlying mechanism for both EAD and DAD formation in the Iso + BayK model, which may share a similar ionic mechanism with that of catecholaminergic polymorphic ventricular tachycardia (6, 10, 23).

Synergy between I_Ca,L and I_NCX on EAD genesis.

Based on the above analysis, we suggest that the relative contributions of I_Ca,L and I_NCX may vary (i.e., one of them may play a primary role) in EAD generation under specific conditions. However, it is conceivable that the synergistic interactions between I_Ca,L and I_NCX are also present in the process of EAD generation [refer to a comprehensive review by Weiss et al. (41)]. For instance, reactivation of the I_Ca,L during the plateau is likely to be the primary cause of EADs in H₂O₂ model. Meanwhile, enhanced window I_Ca,L may trigger additional Ca²⁺ release from the SR and increase I_NCX. Conversely, under the Ca overload condition (e.g., Iso + BayK model), spontaneous Ca²⁺ release from the SR occurs to promote the forward mode of I_NCX, causing delay in repolarization, namely a conditional phase (44). This allows longer time for I_Ca,L to recover from inactivation, which in turn favors EAD formation.

Since strong synergies are present between I_Ca,L and I_NCX in the generation of EADs, reducing either current may be effective in suppressing EADs. This has been proven to be true in both the H₂O₂ and Iso + BayK models. Inhibition of I_Ca,L by nifedipine may suppress EADs in either model either by directly reducing the underlying inward Ca current or by secondarily attenuating the [Ca²⁺]_i overload and thus I_NCX. SEA0400, the most selective agent available to inhibit I_NCX, suppressed both EADs and SCaTs in the H₂O₂ model, suggesting that I_NCX plays a facilitating role (conditional phase) in H₂O₂-induced EAD induction. The facilitated reactivation of I_Ca,L represents the ion carrier for both the formation of EAD and the trigger of SCaT in the H₂O₂ model. In the Iso + BayK model, however, SEA0400 inhibited EADs without affecting CaWs, suggesting a primary causal role of spontaneous SR Ca²⁺ release on the EAD generation under the Ca²⁺ overloaded condition. Several studies (11, 21) have shown that SEA0400 is effective to prevent/treat triggered activities and arrhythmias.

Relevance of EAD models.

One may raise concerns that the EAD models we used in the present study might be too complex. We have tried to establish other “simple” models. Unfortunately, neither elevating extracellular Ca²⁺ concentration (4–8 mM) nor Iso (up to 2 μM) alone successfully induced sustained EADs, although they might occasionally generate DADs. These results were consistent with previous reports showing that Iso alone was less efficient in causing sustained spontaneous Ca release and EADs/DADs (23, 38), since the reduction of SR threshold for spontaneous Ca²⁺ release is also required (10). Therefore, we consider that the two EAD models used in the present study are the most ideal and stable approaches that enable us to make systematic comparisons for the relative contribution of I_Ca,L vs. CaW.

It should be noted that similar approaches have been utilized in the whole heart setting to induce EADs and triggered arrhythmias (22, 23), indicating the relevance of the models. In addition, it seems that the H₂O₂ model is more relevant to some pathological conditions. The H₂O₂ level in human blood plasma may reach as high as ∼35 μM (12). It is also well known that ROS levels can increase under certain oxidative stressed conditions, such as chronic heart failure and ischemia-reperfusion (by as much as 100-fold; Ref. 8). Thus the concentrations (200 μM) used in most of our experiments are reasonably within its pathological range in situ.

H₂O₂-induced EADs were slow-rate dependent and were often observed at PCL >2 s. Accordingly, ROS-induced ventricular arrhythmias may be more readily implicated in the clinical setting of bradycardia (18), such as sinus-node dysfunction and atrial-ventricular conduction disturbances. It should be noted a heart rate at 10 beats/min is thought to be a very deep bradycardia and mostly incompatible with life.

Limitation.

One limitation is the complexity of the EAD models we used in the present study. While it is impossible to dissect one single current component that exclusively contributes to the generation of EAD, we have provided convincing evidence showing that either I_Ca,L or I_NCX may play a primary role in one specific EAD model vs. another.

Conclusion.

Our results have provided more convincing evidence for the heterogeneous mechanisms and their synergies for EAD generation. While EADs involve the complex interplay of several mechanisms, one mechanism (e.g., I_Ca,L reactivation vs. CaW-induced I_NCX) may play a primary role under a certain pathological condition.

The classification (or nomenclature) of afterdepolarizations have been largely based on the timing they appear on the different phases during APD, i.e., EADs occur at phase 2 or 3 of an AP and DADs at phase 4 (after complete repolarization). Since afterdepolarizations (in particular EADs) may result from distinct cellular and ionic mechanisms, it may be more meaningful to further classify them based on their predominant mechanisms. As for the EADs induced in H₂O₂ and Iso + BayK models shown in the present study, we call them sarcolemma-dependent and SR-dependent EADs, respectively.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: Z.Z., H.W., N.F., C.A., and L.-H.X. performed experiments; Z.Z., H.W., N.F., C.A., and L.-H.X. analyzed data; Z.Z., A.B., T.M., and L.-H.X. interpreted results of experiments; Z.Z., H.W., N.F., C.A., and L.-H.X. prepared figures; Z.Z. and L.-H.X. drafted manuscript; Z.Z. and L.-H.X. edited and revised manuscript; Z.Z., H.W., N.F., C.A., A.B., T.M., and L.-H.X. approved final version of manuscript; L.-H.X. conception and design of research.