Role of the transient outward potassium current in the genesis of early afterdepolarizations in cardiac cells

Abstract

Aims

The transient outward potassium current (I_to) plays important roles in action potential (AP) morphology and dynamics; however, its role in the genesis of early afterdepolarizations (EADs) is not well understood. We aimed to study the effects and mechanisms of I_to on EAD genesis in cardiac cells using combined experimental and computational approaches.

Methods and results

We first carried out patch-clamp experiments in isolated rabbit ventricular myocytes exposed to H₂O₂ (0.2 or 1 mM), in which EADs were induced at a slow pacing rate. EADs were eliminated by either increasing the pacing rate or blocking I_to with 2 mM 4-aminopyridine. In addition to enhancing the L-type calcium current (I_Ca,L) and the late sodium current, H₂O₂ also increased the conductance, slowed inactivation, and accelerated recovery from the inactivation of I_to. Computer simulations showed that I_to promoted EADs under the condition of reduced repolarization reserve, consistent with the experimental observations. However, EADs were only promoted in the intermediate ranges of the I_to conductance and the inactivation time constant. The underlying mechanism is that I_to lowers the AP plateau voltage into the range at which the time-dependent potassium current (namely I_Ks) activation is further slowed and I_Ca,L is available for reactivation, leading to voltage oscillations to manifest EADs. Further experimental studies in cardiac cells of other species validated the theoretical predictions.

Conclusion

In cardiac cells, I_to, with a proper conductance and inactivation speed, potentiates EADs by setting the AP plateau into the voltage range where I_Ca,L reactivation is facilitated and I_Ks activation is slowed.

1. Introduction

The transient outward potassium (K) current (I_to) has been shown to play important roles in the action potential (AP) morphology and AP duration (APD) alternans,^1–3 which has been linked to arrhythmogenesis in Brugada syndrome.⁴ In a study by Guo et al.,⁵ it was shown that the elimination of I_to in mouse ventricular myocytes increased the propensity of early afterdepolarizations (EADs). EADs are secondary depolarizations or oscillations during the AP plateau or repolarizing phase, which are associated with arrhythmogenesis in cardiac diseases, such as long QT syndrome.^6,7 It is well accepted that EADs are facilitated under conditions of reduced repolarization reserve,^8,9 in which APDs are prolonged by either increasing inward currents or decreasing outward currents, or both. It should be noted that although a reduced repolarization reserve prolongs the APD, simply lengthening the APD does not necessarily lead to EADs. In a recent theoretical study,¹⁰ Tran et al. showed that in addition to a reduced repolarization reserve, the activation and inactivation kinetics of I_Ca,L must be properly matched to cause the voltage oscillations underlying EADs. In addition, the activation time of the slow time-dependent K current (namely I_Ks) must be slow enough to allow the voltage oscillations to manifest during the plateau or repolarizing phase. More specifically, the membrane voltage must decline to <0 mV quickly enough before I_Ks has sufficient time to fully activate so that I_Ca,L can be reactivated while the repolarization reserve is still low. This leads us to hypothesize that I_to, by lowering the AP plateau voltage into the range at which I_Ks activation is slowed and I_Ca,L is available for reactivation, can lead to voltage oscillations to manifest EADs. In our previous study,¹¹ we showed that exogenous addition of H₂O₂ to myocytes induces EADs, which is probably mediated by the activation of Ca/calmodulin kinase II (CaMK II),¹² and the consequent activation of the L-type calcium current (I_Ca,L) and the late sodium current (I_Na).^13–16 In the present study, we used this H₂O₂-induced EAD model and confirmed our hypothesis via experimental and theoretical studies.

2. Methods

2.1. Patch-clamp experiments

This investigation conforms with the Guide for the Care and Use of Laboratory Animals, published by the National Institutes of Health (NIH Publication No. 85–23, Revised 1996). All animal experimental procedures 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, by the Ethical Committee of Xi'an Jiaotong University, and by the Institutional Animal Care and Use Committee at Columbia University.

2.1.1. Single cell isolation

Myocytes were enzymatically isolated from the left ventricles of adult rabbit, rat, and mouse hearts in Langendorff fashion at 37°C with collagenase and protease.^11,17 Canine epicardial ventricular myocytes were isolated from the LV sections. Canine Purkinje cells were isolated from the Purkinje fibres of both right and left ventricles. The animals were euthanized with Euthasol (0.22 mL/kg, i.e. pentobarbital 86 mg/kg + phenytoin 11 mg/kg; iv in rabbits and ip in rats and mice) or anaesthetized with propofol (5–15 mg/kg; iv in dogs). Immediately after the cessation of breathing (in rabbits, rats, mice) or the induction into a surgical plane of anaesthesia (in dogs, the jawbone, pupil reflexes, paw relaxes, and tongue rigidity were continually monitored during the procedure), the chest was opened through a left thoracotomy. The heart was quickly excised from the chest. The detailed protocols for heart perfusion and single cell dispersion are described in Supplementary material online.

2.1.2. Patch clamp

Myocytes were patch-clamped using the whole-cell configuration in either current-clamp (for AP recording) or voltage (or AP)-clamp mode (for I_to or I_Ca,L recording) as in our previous publications.^11,18–20 To record I_to, the I_Ca,L blocker Cd³⁺ (0.3 mM) and Na current blocker TTX (10 µM) were added to Tyrode's solution. Cells were paced at a pacing cycle length (PCL) of 1 and 6 s to study the rate dependence of I_to and EADs. All experiments were carried out at 34–36°C.

2.1.3. Pacing protocols

A dynamic pacing protocol was used to measure APD restitution. The myocytes were paced at PCLs of 5, 2, 1 s, 700 ms for 6 beats each, and then at PCL 400 ms, decremented every 12 beats by 20 ms (from 400 to 300 ms), 10 ms (from 300 to 200 ms), or 5 ms (from 200 to 150 ms), or until a 2:1 block occurred. Recovery of I_to from the inactivation was investigated using a conventional two-pulse protocol: an inactivating pulse depolarizing to +40 mV for 400 ms (P₁) followed by a variable recovery interval (5, 10, 20, 50, 100, 300, 600 ms; 1, 3, 6, 10, 15 s, respectively) and subsequent +40 mV test pulse (P₂). The interpulse potential was set at – 80 mV.

2.2. Computer simulation and theoretical analysis

Simulations were carried out in single myocyte models with the following governing equation for voltage (V): C_mdV/dt = −I_ion+ I_sti, where C_m= 1 μF/cm², I_ion is the total ionic current density, and I_sti is the stimulation current density. I_ion formulation was taken from the rabbit ventricular myocyte AP model by Mahajan et al.²¹ with modifications. Modifications to the model were made to better fit the APD restitution and EAD properties observed in the experimental studies. The major changes include substitution of I_Ca,L by the Hodgkin-Huxley type of formulation as in the 1994 Luo and Rudy model (LR2)²² and I_to by the formulation modified from the one by Dong et al.²³ Therefore, the ionic current density of the rabbit ventricular model became: . The effects of H₂O₂ on ion channels were modelled by increasing I_Ca,L, late I_Na, Na-Ca exchange current (I_NCX), and I_to, based on previous^11,13,14,17 and present experimental studies. 4-AP is known to inhibit different components of I_to.²⁴ For simplicity, however, we modelled the 4-AP effect by only reducing the maximum conductance of I_to with the goal of reducing 4-AP sensitive currents. Details of the modifications and formulations of the ionic currents are presented in Supplementary material online, Methods and Figures S2–6.

To study the general mechanism of I_to promoting EADs and for the convenience of theoretical analysis, we used the 1991 Luo and Rudy (LR1) model²⁵ and added an I_to to the LR1 model, with I_to modified also from the formulation by Dong et al.²³ (see Supplementary material online, Methods and Figures S7–8). The total ionic current density was

. The rationale for choosing LR1 is because it is much simpler than the rabbit ventricular model and it has the minimum requirements for EADs to occur, as shown in the previous study.¹⁰ Using this simple model, we can follow the same method as in the previous study¹⁰ to perform bifurcation analysis to understand how I_to promotes EADs (see Supplementary material online, Figure S11).

3. Results

3.1. I_to promotes EADs in rabbit ventricular myocytes

EADs are usually promoted by bradycardia because during the longer diastolic intervals²⁶ the delayed rectifier K channels enter the states that are deeply closed, thus reducing outward currents and lengthening the APD. However, in rabbit ventricular myocytes, the APD first increases as the heart rate decreases, but then becomes shorter at very slow heart rates (i.e. the APD restitution curve is biphasic) (Figure 1A and B). This has been attributed to the slow recovery of I_to in the rabbit cells.²⁷ However, it is at these very slow heart rates (PCL > 3 s) that exposure to H₂O₂ induces EADs, but not at the intermediate heart rates at which the APD is normally longer (Figure 1C).^11,19 We assessed the role of I_to in the EAD formation by applying 4-AP to block I_to. As shown in Figure 1D and E, 4-AP (2 mM) alone prolonged APD at all PCL, to a greater extent at the long PCL (e.g. 6 s), and to a less extent at short PCLs (e.g. 1 s). Therefore, the restitution curve was transformed from a biphasic shape to a monophasic one. Despite prolonging the APD at very slow heart rates, 4-AP paradoxically prevented overt EADs in response to 200 μM H₂O₂ (Figure 1F). The same phenomenon was observed in 20 cells from 8 rabbits. It should be noted that some cells from rabbits showed a less negative slope in the biphasic restitution curve, suggesting a lower I_to level. However, these cells exhibited similar responses to H₂O₂ for inducing EADs and to 4-AP for attenuating EADs (data not shown).

Figure 1.

Dependence of APD and EAD genesis on I_to. (A) The APD restitution curve obtained from a rabbit ventricular myocyte showing a biphasic change of the APD. The expanded region between 0 and 300 ms is shown in the insets. DI, diastolic interval. (B) APs recorded at PCL 6 s (grey solid) and 1 s (dashed). (C) H₂O₂ (200 μM, perfusion for 6 min) induced EADs at PCL = 6 s. (D–F) Same as (A)–(C), but in the presence of 2 mM 4-AP. Note that 4-AP eliminated the EADs (F). Five consecutive APs are shown in panels (C) and (F).

3.2. Effects of H₂O₂ on I_to kinetics and conductance

Under our experimental conditions, H₂O₂ enhanced I_to in a time course comparable with that for the EAD onset.¹¹ This was attenuated by 4-AP (Figure 2A and B). H₂O₂ increased both the peak amplitude of I_to and the magnitude of the current at the late phase under the 400 ms-pulse clamp (we refer to the late component as I_late, including the slow and non-inactivating components) (Figure 2C). Finally, H₂O₂ slowed the I_to inactivation by decreasing the time constants of both the slow (τ_s,in) and fast components (τ_{f in}) (Figure 2D).

Figure 2.

Effect of H₂O₂ I_to in rabbit ventricular myocytes. (A–D) Effects of H₂O₂ on I_to amplitude and inactivation: I_to was elicited by depolarization to + 60 mV from a holding potential of −70 mV every 10 s. (A) The time course showing the effect of H₂O₂ and 4-AP on the I_to amplitude. The current values measured at 100 ms after the depolarization pulse [as indicated by the dashed line in (B)]. The current traces at point a, b, c are shown in (B). (C) Bar graph summarizing the effect of H₂O₂ on peak and late currents (measured at the end of 400 -ms pulse). (D) H₂O₂ slowed down the inactivation of both the fast and slow components of I_to. **P < 0.01 compared with control (n = 7). (E–H) Effect of H₂O₂ on the recovery from inactivation: (E) representative I_to traces following a standard two-pulse protocol (supplements) to measure the recovery from inactivation under control in the absence of H₂O₂ (Ctl). P₂/P₁ is the ratio of the test pulse current/inactivating pulse current amplitudes. The inactivated current amplitude was measured by the difference between the peak inactivating pulse current or the test pulse current and the current at the end of the inactivating pulse. The inter-pulse potential was set at –80 mV. (F) The same as (E), in the presence of 1 mM H₂O₂. (G) Summarized peak I_to values fitted to a two-term exponential function. (H) The effect of H₂O₂ on time constants for slow (τ_slow) and fast (τ_fast) components.

Under control conditions and consistent with APD shortening at PCL > 3 s, I_to displayed a slow recovery from inactivation, as shown in Figure 2E and G. I_to recovery from the inactivation was best fit with a double exponential equation (see Supplementary material online, Methods) where the averaged time constants for the fast (τ_{f, re}) and slow (τ_{s, re}) components are 832 ± 222 ms and 5.8 ± 2.1 s, respectively. H₂O₂ accelerated the recovery from the inactivation of I_to, mainly by decreasing τ_{f, re} to 240 ± 61 ms, while τ_{s, re} showed no significant change (Figure 2F–H). The amplitudes of the fast and slow components (A_f: 0.22 ± 0.08 and A_s: 0.79 ± 0.14 under control conditions) were significantly affected by H₂O₂, (A_f: 0.40 ± 0.05 and A_s: 0.61 ± 0.04) (P < 0.05, n = 6).

3.3. Effect of I_to on I_Ca,L during the AP during H₂O₂ exposure

Since I_Ca,L reactivation plays an essential role in EAD genesis, we compared the effects of an AP with or without I_to on I_Ca,L during an AP clamp. When the cells were clamped with the AP morphology corresponding to the presence of I_to (i.e. an AP pre-recorded at a PCL 6 s), I_Ca,L was significantly larger both in the absence and presence of H₂O₂, than those recorded during the AP morphology corresponding to the absence of I_to (i.e. rectangular AP pre-recorded at PCL 1 s) ( see Supplementary material online, Figure S1). This result indicates that the presence of I_to lowers the AP plateau voltage into a voltage range that produces a larger inward I_Ca,L, known to be a key factor in EAD formation.^28,29

3.4. I_to promotes EADs in an AP model of rabbit ventricular myocytes

To understand how I_to promotes EAD genesis, we carried out simulations using a rabbit ventricular AP model,²¹ modified to include the property of slow recovery of I_to observed experimentally. The effects of H₂O₂ were modelled by increasing the late I_Na, I_Ca,L, and I_to (see Supplementary material online, Methods and Figures S2–6). The APD restitution curve for the control (Figure 3A and B) was biphasic, similar to experimental data. When the effects of H₂O₂ on ionic currents were simulated, EADs occurred only at slow pacing rates (Figure 3C). When we removed I_to from the model to simulate the effects of blocking I_to with 4-AP, the APD restitution curve became monotonic (Figure 3D and E), and no EADs were observed (Figure 3F). These data agree well with the experimental observations shown in Figure 1.

Figure 3.

APD restitution and H₂O₂-induced EADs in the rabbit ventricular model in the presence of I_to. The plots are an analogy to Figure 1. (A) The APD restitution curve of the control cases. (B) Control APs for PCLs of 6 s (grey solid) and 1 s (dashed). (C) Simulated H₂O₂-induced EADs. (D–F) The same as (A)–(C), but in the presence of 4-AP and I_to blocked.

To show how I_to interacts with other currents to potentiate EADs, we plotted voltages along with I_to, I_Ca,L, I_Ks, and I_NCX under different conditions in Figure 4. In the control case (Figure 4A and B), I_to was larger during slower pacing (PCL 6 s) due to its slow recovery, which lowered the voltage at the plateau phase. This in turn caused a larger peak I_Ca,L, a larger I_NCX, and a very small I_Ks; however, the APD was shorter. When the effects of H₂O₂ were simulated (Figure 4C and D), the AP and the ionic currents at PCL = 1 s were similar to those of the control at PCL = 1 s. At PCL = 6 s (Figure 4D), EADs were induced by H₂O₂. In this case, a larger I_to resulted in a lower AP plateau voltage that caused a larger peak I_Ca,L and a larger I_NCX. Concomitantly, the activation of I_Ks was much slower due to the lower AP plateau voltage (here the activation time constant of I_Ks vs. voltage is reproduced in Supplementary material online, Figure S9), which also played an important role in allowing the inward currents to reverse repolarization and cause EADs. When I_to was partially blocked (Figure 4E), the AP plateau voltage was elevated, resulting in a smaller peak I_Ca,L, a smaller I_NCX, and a much larger I_Ks, which prevented EADs from forming.

Figure 4.

I_to-induced ionic current changes in the rabbit ventricular model. AP, I_to, I_Ca,L, I_Ks, and I_NCX were plotted under different conditions. The dashed line in the top row indicates the plateau level of AP under the control condition. (A and B) The cell was paced at PCLs of 1 and 6 s, respectively, in the absence of H₂O₂. (C and D) The cell was paced at PCLs of 1 and 6 s, respectively, in the presence of H₂O₂. Note that EADs were induced by H₂O₂ at PCL 6 s. (E) H₂O₂-induced EADs were eliminated by partially blocking I_to.

3.5. Mechanisms by which I_to promotes EADs

As shown in Figure 4, I_to did not directly cause EADs; rather, its effects on voltage caused complex changes in other currents, making it difficult to sort out the general underlying mechanism. To simplify the analysis, we used the LR1 model with modifications to reduce repolarization reserve (see Supplementary material online for detailed changes). Under this (control) condition, the APD is long with no EAD presence (Figure 5A). Consistent with the results from myocyte experiments and the rabbit ventricular AP model, adding an I_to to the control APinduced EADs (Figure 5B). In this case, I_Ca,L was increased and the activation of the time-dependent K current (I_K, similar to I_Ks in the rabbit ventricular AP model) was slowed due to the lowering of the plateau voltage, similar to the more detailed rabbit ventricular AP model. When I_to was increased to a higher value, the APD became very short and no EADs could form (Figure 5C).

Figure 5.

Effects of I_to on EAD genesis in the LR1 model. APs, I_to, I_Ca,L, and I_K were plotted for three different conditions. (A) Control, no I_to. (B) A proper I_to added to the control model induced EADs. (C) A larger I_to shortened the APD and eliminated EADs. (D) Phase diagram showing the EAD region (coloured) in the I_to conductance (G_to) and τ_h0 parameter space. τ_h0 is the constant component of the inactivation time constant τ_h of I_to, i.e. τ_h= τ_h0+ τ_h(V) [see equation (17) in the Supplementary material online, Methods]. (E) The APD distribution in the I_to conductance and the τ_h0 parameter space.

In Figure 5D and E, we scanned the parameters of I_to conductance (G_to) and its inactivation time constant for EADs and the APD, respectively. EADs only occurred in the intermediate ranges of conductance and the inactivation time constant, whereas small or large conductance and slow or fast inactivation suppressed EADs. When I_to inactivation was fast, an increase in I_to conductance first increased the APD (and EADs) but then caused a sudden transition to a short APD when the I_to conductance was greater than a critical value (colour changes from green or red to blue suddenly in Figure 5E). When I_to inactivation was slow, the APD decreased continuously as the I_to conductance increased. These effects of I_to on the APD agree with those observed in previous studies.^23,30 In fact, a very slow-inactivating I_to component suppresses EADs (see Supplementary material online, Figure S10).

To further understand how I_to promotes EADs, we used non-linear dynamics and bifurcation theory following the method in a previous study by Tran et al.¹⁰ Details of this analysis and discussion are presented in Supplementary material online (Section 2.3 and Figure S11). The major finding is that I_to, with a proper conductance and inactivation speed, can bring the membrane voltage into the window to undergo the bifurcation that is required for voltage oscillations and thus EADs.

3.6. The role of I_to in EAD formation in cardiac cells of other species

It is well known that the molecular composition, current density, and kinetics (e.g. inactivation and recovery from inactivation) of I_to vary in different species and locations in the heart. To validate the general predictions from the general theoretical analysis (Figure 5D–E), we carried out additional experiments in ventricular cells from rats, mice, and canine epicardium, as well as canine Purkinje cells. Experimental results are shown in detail in Supplementary material online, Figures S12–15 and summarized and compared qualitatively with simulations in Figure 6.

Figure 6.

Roles of I_to in EAD genesis in different species. (A) The EAD distribution in the I_Ca,L and I_to conductance (G_Ca and G_to) space, using the same model as in Figure 5. The inactivation time constant of the I_to is the intermediate value (τ_h0= 200 ms) as revealed by Figure 5D. The EAD occurrence region is marked in grey. (B) APs recorded in different species as indicated (a–d). The black and coloured symbols mark the putative locations of the APs in the distribution diagram (A). (C) Summarized bar graphs showing the incidence of EADs within 20 APs (n ≥ 3 cells). ns, not significant; **P < 0.01 by Fisher's exact test. (D) The same as (A), except that the inactivation time constant of the I_to is fast (τ_h0= 20 ms), which is similar to that of canine ventricular myocytes. (E). APs recorded from canine ventricular myocytes. The symbols mark the putative locations of the APs in the distribution diagram (C).

Since we showed that H₂O₂ increases both I_Ca,L and I_to and in order to compare with experimental results, we first scanned the I_to conductance and I_Ca,L conductance on EAD formation in the LR1 model for slow and fast I_to inactivation (Figure 6A and D). Figure 6A was obtained for an I_to inactivation time constant in the intermediate range as shown in Figure 5D. In our model, EADs occurred in large I_Ca,L conductance and intermediate I_to conductance (grey region). For rabbit (Figure 6Ba and C) and rat ventricular myocytes (Figure 6Bb), H₂O₂ increased I_Ca,L and I_to, caused the AP to exhibit EADs. Blocking I_to by 4-AP (at either a low or high concentration) eliminated EADs. This is illustrated in Figure 6A by the transition from the black open circle (no EAD region) to the red closed circle (EAD region) and then to the blue open circle (no EAD region). For mouse ventricular myocytes (Figure 6Bc) and canine Purkinje cells (Figure 6Bd and C); however, H₂O₂ did not promote EADs. But partially blocking I_to by a low dose of 4-AP (50–100 µM) promoted EADs, which were eliminated by a higher dose of 4-AP (2 mM). This indicates that either I_to was already high in the control and/or H₂O₂ activated too much additional I_to. This scenario is illustrated in Figure 6A by the transitions from black open triangle (no EAD region) to the open red triangle (no EAD region), then to the red closed circle (EAD region), and finally to the blue open circle (no EAD region). In canine epicardial myocytes, the inactivation of I_to is very fast (∼19.9 ms). Based on our theoretical analysis, the propensity for EADs is greatly reduced for fast I_to inactivation (Figures 5D and 6D), and indeed we were not able to induced EADs in canine ventricular myocytes by H₂O₂ (Figure 6E). This is illustrated by the transitions of Figure 6D. In summary, the experimental results agree qualitatively well with the theoretical predictions for all the species and cell types analysed. Thus I_to with a proper conductance and inactivation time constant can promote EADs in cardiac myocytes, independent of species.

4. Discussion

I_to has been shown to play important roles in AP morphology and APD alternans.^1–3 In this study, we carried out both experiments and simulations to study the effects of I_to on EAD genesis in cardiac cells exposed to H₂O₂. In rabbit ventricular myocytes, EADs occurred only at slow pacing rates and were eliminated by either shortening PCL or blocking I_to with 4-AP. The experimental observations were confirmed in computer simulations of an AP model of rabbit ventricular myocytes. We have shown in general that I_to with a proper conductance and inactivation time constant can bring membrane voltage into the window allowing I_Ca,L reactivation before I_Ks is fully activated to repolarize the myocytes. In addition, this lowering of the AP plateau also brings I_Ks into a voltage range where its activation is the slowest, further decreasing the repolarization reserve. These two effects together potentiate voltage oscillations and thus EAD generation. When I_to is weak or inactivates too quickly, the AP is long but no EADs can be induced. When I_to is too strong or inactivates too slowly, the AP is short and no EADs occur either. This general prediction of the effects of I_to on EAD genesis was validated in cardiac cells isolated from other species.

4.1. Reduced repolarization reserve and EAD genesis

The repolarization reserve is a concept articulated originally by Roden³¹ stating that normal ventricles have a redundancy in total repolarizing currents (mainly K currents), ensuring rapid repolarization. However, when this reserve is reduced, the cell loses the ability of normal repolarization and exhibits an excessive prolongation of APD. Based on the dynamical analysis by Tran et al.¹⁰ and the present study (see Supplementary material online, Figure S11), we show that membrane voltage needs to decrease ‘fast’ enough to enter an oscillatory window before slowly activating repolarization currents, such as I_Ks, grow too large. In the study by Tran et al.,¹⁰ EADs were induced by increasing the maximum conductance of I_Ca,L and slowing the activation of the time-dependent K current to reduce repolarization reserve. In the present study, we did not slow the activation of the repolarizing currents directly, but rather added I_to, which accelerated the early repolarization phase of the AP, bringing the voltage to the window range to undergo the bifurcation to promote EADs. However, it should be noted that when I_to becomes very large (such as would occur with a specific I_to activator), it causes sufficient AP shortening to suppress EAD formation. Therefore, strong repolarization (enhanced repolarization reserve) is needed to bring the voltage into the window for I_Ca,L reactivation, but once the voltage is in this window, weak repolarization (reduced repolarization reserve) is needed to maintain the voltage at this window for oscillations and thus EADs. Notably a recent study has shown that coupling to fibroblasts can also speed up the early repolarization phase of a myocyte AP and promote EAD generation.³²

4.2. Roles of different I_to components in EAD genesis

At least two major I_to components, i.e. fast (I_to,f) and slow (I_to,s) components are present in adult myocytes.³³ It has been suggested that pore-forming α subunits for Kv4.2 and Kv4.3 encode I_to,f, whereas Kv1.4 encodes I_to,s. They differ in the kinetics of inactivation and, in particular, the kinetics of recovery from inactivation. I_to,f recovers very rapidly with time constants in the range from tens to hundreds of milliseconds, while I_to,s recovers slowly with time constants in the range of seconds.³⁴

The I_to components and their density are different in different species as well as different regions in the same heart. The kinetic properties of I_to in rabbit ventricular myocytes are substantially different from those of canine/human ventricular myocytes. The rabbit ventricular I_to is at least 10 times slower to inactivate and recover from inactivation than canine/human ventricular I_to.³⁴ The epicardial myocytes from a canine left ventricle only express the high level of I_to,f.³⁵ In contrast, both I_to,f and I_to,s are present in canine and human Purkinje cells.^36–39

Different components of I_to have different effects on AP properties. For example, in canine ventricular cells that only have the fast component, the presence of a moderate I_to prolongs AP, whereas a high I_to substantially shortens AP, resulting in a sudden discontinuous change in the APD as I_to increases.⁴⁰ In rabbit myocyte and canine Purkinje cells, I_to inactivates slowly with a non-inactivating steady-state component,³⁸ whose presence always shortens the AP. This was demonstrated in Figure 1 in which the rabbit APD was shortened as the pacing rate decreased due to the slow recovery (τ_{slow, re} = 5.8 ± 2.1 s) of I_to. The 4-AP-induced APD prolongation (Figure 1E and F) was at least partially due to the blockage of this non-inactivating component.

As to the roles of the multiple components of I_to in EAD genesis, we showed that EADs only occurred in a certain range of I_to conductance and inactivation speed (Figure 5). In fact, a very slow-inactivating I_to component suppresses EADs (see Supplementary material online, Figure S10). This may also explain why a moderate decrease in I_to density results in the elevation of the AP plateau and the enhancement of EADs in mouse myocytes and canine Purkinje cells (Figure 6 and see Supplementary material online, Figures S13 and 14). On the other hand, if the I_to inactivation is too fast, such as in canine ventricular myocytes, it is not sufficient to promote EADs and is cardioprotective.

4.3. Role of H₂O₂ in EAD genesis

It has been shown that H₂O₂ activates late I_Na and increases I_Ca,L; both are considered to be the major ionic mechanisms for a H₂O₂-induced EAD.^11,13–15 Here, we show that H₂O₂ also enhances the conductance of I_to, slows its inactivation, and accelerates its recovery from inactivation. This resembles the effect of CaMKII on the mammalian A-type K⁺ channel K_V1.4.⁴¹ The findings of the present study indicate that H₂O₂-mediated activation of I_to also plays an important role in EAD genesis. Han et al.³⁶ showed that H₂O₂ only activated I_to in canine Purkinje cells, but not in ventricular myocytes. This explains why H₂O₂ first caused further suppression of the low plateau in Purkinje cells before EAD generation. H₂O₂ only generated a dome in canine ventricular myocytes presumably by purely reactivation of I_Ca,L.

While more direct evidence is needed to validate the involvement of ROS in EAD genesis in vivo, a growing body of evidence has comprehensively established the link between an elevated ROS level and high propensity for cardiac arrhythmias in a number of circumstances, such as coronary heart disease, heart failure, and ageing.^42–46 Moreover, a recent whole-heart study has shown that H₂O₂ causes EADs and focal ventricular arrhythmias in the aged fibrotic hearts.⁴⁷

4.4. Limitations

It should be noted that in either experiments or simulations, I_to, an outward current per se cannot induce EADs. Other conditions (such as applying H₂O₂ in the experiments to activate inward currents, i.e. I_Na and I_Ca,L) must be satisfied. In addition, the mechanisms by which I_to contributes to EADs may be more complicated than those revealed in the simple model. For example, lowering the AP plateau promotes the Ca influx into the cell due to the reactivation of I_Ca,L, elevating intracellular Ca concentration and thus I_NCX, which could be the primary factor potentiating EADs. The elevation of Ca concentration may also affect CaMKII signalling, which affects many targets (such as I_Ca,L, late I_Na, and RyR).^11,14,48 In fact, the prolonged treatment by H₂O₂ occasionally induced both DADs and EADs, or even sustained the depolarization of the membrane potential, most likely due to intracellular Ca overload.¹¹ In a recent study⁴⁹ we have demonstrated that the ionic mechanisms for EAD generation can be dominated by either I_Ca,L reactivation or spontaneous Ca release (Ca waves)-induced increase of I_NCX in different models. This study focuses on EADs primarily caused by the reactivation of I_Ca,L. We would expect that I_to would play a less important role in Ca wave-induced DADs or EADs.

Nevertheless, by a combination of experiments and simulations, we show here that the presence of I_to plays an important role in EAD genesis under the condition of reduced repolarization reserve. This may provide useful insights into the development of novel therapeutic strategies for EAD-related arrhythmias. Our results suggest that, in general, blocking I_to helps to prevent EADs under the condition of reduced repolarization reserve, such as the clinical settings of the long QT syndrome and heart failure. However, it should be noted that complex changes (remodelling) occur in heart failure and many changes in Ca cycling are arrhythmogenic.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Funding

This study is supported by a Postdoctoral Fellowship from American Heart Association Western states affiliate (Y.X.), HL 66140 (P.A.B.), NIH/NHLBI P01 HL078931 and the Laubisch endowments (Z.Q.), and NIH/NHLBI R01 HL97979 (L.H.X.).