Transient Outward K+ Current (I_to) Underlies the Right Ventricular Initiation of Polymorphic Ventricular Tachycardia in a Transgenic Rabbit Model of Long QT Type 1

Abstract

Background

Sudden death in long QT syndrome type 1 (LQT1), an inherited disease caused by loss-of-function mutations in KCNQ1, is triggered by early afterdepolarizations (EADs) that initiate polymorphic VT (pVT). We investigated ionic mechanisms that underlie pVT in LQT1 using a transgenic rabbit model of LQT1.

Methods and Results

Optical mapping, cellular patch clamping, and computer modeling were used to elucidate the mechanisms of EADs in transgenic LQT1 rabbits. The results showed that shorter APD in the right ventricle (RV) was associated with focal activity during pVT initiation. RV cardiomyocytes demonstrated higher incidence of EADs under 50 nM isoproterenol. Voltage-clamp studies revealed that the transient outward potassium current (I_to) magnitude was 28% greater in RV associated with KChiP2 but with no differences in terms of calcium-cycling kinetics and other sarcolemmal currents. Perfusing with the I_to blocker 4-aminopyridine changed the initial focal sites of pVT from the RV to the LV, corroborating the role of I_to in pVT initiation. Computer modeling showed that EADs occur preferentially in the RV due to the larger conductance of the slow inactivating component of I_to, which repolarizes the membrane potential sufficiently rapidly to allow reactivation of I_Ca,L before I_Kr has had sufficient time to activate.

Conclusions

I_to heterogeneity creates both triggers and an arrhythmogenic substrate in LQT1. In the absence of I_Ks, I_to interactions with I_Ca,L and I_Kr promote EADs in the RV while prolonging APD in the LV. This heterogeneity of action potential enhances dispersion of refractoriness and facilitates conduction blocks that initiate pVTs.

Introduction

Long-QT syndrome type 1 (LQT1) is an inherited disease associated with prolongation of QT intervals in ECG recordings and an increased risk for developing polymorphic ventricular tachycardia (pVT) that underlies syncope and SCD.¹ The disease is caused by loss-of-function mutations in the KCNQ1 gene, which encodes the pore-forming α subunit of the slowly activating delayed rectifier repolarizing K⁺ current (I_Ks). LQT1 is the most common form of long QT syndrome and accounts for almost half of genotyped cases.² Arrhythmias and SCD in LQT1 are often precipitated by prolonged periods of exercise that sustains sympathetic activity and can in turn trigger early afterdepolarizations (EADs) that initiate pVTs.³

Experimental models using I_Ks blockers such as chromanol or HMR1556 to mimic LQT1 have been helpful in elucidating the roles of I_Ks in APD dynamics.^{4, 5} While chromanol perfusion produces limited APD prolongation, persistent sympathetic stimulation can both prolong APD significantly and enhance APD dispersion.⁵ Clinical data have also demonstrated that the T_peak – T_end intervals of surface ECGs were accentuated during exercise tests only in LQT1 patients but not in LQT2 patients,⁶ supporting the notion that APD dispersion plays significant roles in pVT initiation in LQT1. Although experimental models with chromanol demonstrated enhanced APD dispersion, chromanol often failed to induce EADs.⁷ In addition, non-specific action of chromanol such as blocking transient outward K⁺ current (I_to)⁸ limits the translation of these findings from the experimental models to congenital LQT1-related arrhythmias.

We have generated a transgenic rabbit model for LQT1 syndrome by overexpressing a pore mutant of KCNQ1 (KCNQ1-Y315S) in the heart.⁹ Our previous studies of LQT1 rabbits show that LQT1 rabbits develop EADs and pVTs following sympathetic stimulation,^{10, 11} and all LQT1 rabbits die within three weeks after slowing their heart rate with acute ablation of the AV node in vivo. Detailed mapping revealed that triggered activity mostly originated in the right ventricle (RV) despite shorter APD in RV compared to LV.¹⁰ Due to the formation of triggered activity in short-APD regions, its propagation often encountered conduction blocks and promoted reentry formation in LQT1 rabbits.¹⁰

The present study was designed to elucidate the molecular mechanisms underlying tissue heterogeneity and EAD formation in LQT1 rabbits. Here we combine electrophysiological studies, optical mapping, and computational modeling to delineate the mechanisms underlying paradoxical EAD formation from the RV despite shorter APD than in the LV. The present study demonstrates that the lack of I_Ks in LQT1 unmasks the critical role of I_to heterogeneity in promoting EADs preferentially in the RV by its effect on AP plateau voltage, I_CaL and I_Kr, and the initiation of arrhythmias.

Methods

The authors declare that all supporting data and computer simulation source code are available within the article and its supplementary files.

Heart Preparation

LQT1 rabbits of either sex, averaging 16.5 months old / 4.2 kg body weight / 9.14 g heart weight, were euthanized with buprenorphene (0.03 mg/kg IM), acepromazine (0.5 mg.kg⁻¹ IM), xylazene (15 mg.kg⁻¹ IM), ketamine (60 mg.kg⁻¹ IM), pentothal (35 mg.kg⁻¹ IV), and heparin (200 U.kg⁻¹). This investigation conformed to the current Guide for Care and Use of Laboratory Animals published by the National Institutes of Health (National Academies Press, revised 2011) and approved by the Lifespan Animal Welfare Committee at Rhode Island Hospital. Detailed methods for optical mapping are available in the supplemental material.

Patch Clamping

Isolation of cardiomyocytes by standard enzymatic techniques and patch-clamp recordings were performed as described previously.⁹ RV and LV (septal region) myocytes were isolated from hearts (n=5 hearts each from LQT1 and LMC). Whole-cell recordings (11–18 cardiomyocytes per group) were obtained with an Axopatch-200B amplifier (Axon Instruments) with standard patch-clamp techniques (see supplemental material for detail). Our voltage-clamp data on I_to indicates that the inactivation of I_to is well fit by the sum of two exponentials

$${\mathrm{I}}_{\text{to},\text{total}}={\mathrm{I}}_{\text{to},\text{fi}}exp(-\mathrm{t}/{\mathrm{\tau }}_{\text{fi}})+{\mathrm{I}}_{\text{to},\text{si}}exp(-\mathrm{t}/{\mathrm{\tau }}_{\text{si}})$$ (1)

where I_to,total is the total I_to current, I_to,fi and I_to,si are the amplitudes of the fast-inactivating (fi) and slow-inactivating (si) components, respectively. It should be emphasized that I_to,fi and I_to,si represent the fast and slow inactivating components of I_to,total and in principle contain contributions from both I_to,f, (Kv4.2/Kv4.3) and I_to,s (Kv1.4) that inactivate on similar time scales,¹² but recover from inactivation on very different time scales.¹³ To determine the recovery kinetics from inactivation of I_to, a double-pulse protocol with variable interpulse intervals ranging from 50 ms to 15 seconds was used (see supplemental material for detail). Decay of each I_to current evoked by each 2^nd pulse was fit to a double exponential decay to determine the magnitude of the fast and slow inactivating components of the current. The amplitude of each component as a function of inter-pulse interval was fit to a double exponential recovery function to determine the magnitude of the fast and slow recovering components of each decay component.

Confocal Ca²⁺ recording

Cytosolic and intra-SR Ca²⁺ changes were monitored using a Leica TCS SP5 II confocal system in line-scan mode, and V_m were simultaneously recorded with the patch-clamp technique.¹⁴ Ca²⁺ transients were recorded in intact cells loaded with Fluo-3-AM at 0.25 Hz field stimulation in the presence of 50 nM isoproterenol. Detailed methods for confocal Ca²⁺ imaging are available in the supplemental material.

Western blots of I_to and Ca²⁺-handling proteins

Western blotting of Kv1.4, Kv4.2, Kv4.3, and KChIP2 were performed as described in ¹³. Membranes from RV and LV heart (atria dissected away) were isolated by homogenization. The antibodies for Kv1.4, Kv4.2, and KChIP2 were purchased from Alomone Labs (Jerusalem, Israel). Western blot analyses of Ca²⁺-handling proteins were performed as described in ¹⁴. Detailed methods are available in the supplemental material.

Computer modeling of rabbit myocytes

We used the rabbit ventricular myocyte model of Mahajan et al.¹⁵ with modified mathematical formulations of I_to, I_Kr, and I_Ca,L that quantitatively reproduce voltage-clamp data obtained from LQT1 rabbits under isoproterenol. Following the double-exponential form of Equation (1), I_to was modeled as the sum of I_to,fi and I_to,si with different conductances and kinetics in RV and LV. I_Kr and I_Ca,L were taken to be the same in RV and LV consistent with experimental data. We model I_Kr using the multi-state Markov model of Mazhari et al.¹⁶ with parameters fitted to our patch clamp measurements that reproduce the slow-activation kinetics of I_Kr in the V_m range relevant for EAD formation. In addition, we use a novel formulation of I_Ca,L gating to quantitatively fit voltage-clamp data under isoproterenol. Details of the ionic model are provided in the supplemental material.

Statistical Analysis

Optical mapping data of intact heart are presented as means ± SD and single cell data are presented as means ± SEM. For normally distributed values, we used Student’s t-test (paired and unpaired) to compare the means of two groups and ANOVA to compare multiple means. A 2-side p value of ≤ 0.05 was considered significant. Origin software (OriginLab, Northampton, MA) was used for Shapiro-Wilk normality test, Student’s t-test, and ANOVA. Fisher’s exact test was used for categorical variables using GraphPad software (San Diego, CA).

Results

Potential role of I_to in pVT initiation in LQT1 rabbits

Previous optical mapping study of LQT1 rabbits revealed that APD dispersion became pronounced under slow heart rate (> 1 sec) mostly due to shorter APD in RV than in LV.¹⁰ Paradoxically, most pVTs were started from premature ventricular complex originating from RV despite shorter APD in RV.¹⁰ We hypothesized that I_to plays a major role in APD gradient and pVT initiation between RV and LV. To test this hypothesis, we investigated the effect of the I_to blocker 4-Aminopyridine (4-AP) on arrhythmia initiation and APD gradient. Figure 1A shows sample traces and APD maps under 0 and 0.5 mM 4-AP. Control APD in RV (red) is shorter than the LV, but 4-AP markedly prolonged APD in RV, causing inversion of the APD gradient between RV and LV (Figure 1A) while LMC hearts did not (see supplementary Figure S2). The addition of isoproterenol induced VT/VF, but the initiation of VT/VF was from the LV rather than the RV (5 out of 7 VTs originated from the LV, n=5 hearts) (Figure 1B&C). These results strongly suggest that higher I_to density in RV plays an important role in shortening APD in the RV and initiating EADs from that region.

Figure 1.

A) APD maps under 4-AP. APD is longer in LV under normal conditions in LQT1 but 4-AP prolonged APDs (135% increase from 312 ± 63 to 729 ± 143 ms at 0.5 Hz) and flipped the APD gradient, resulting in longer APD in RV (n=4/5 hearts, Fisher exact test p < 0.05). B) RV initiation of pVT in LQT1 hearts (RV initiation 83% of pVT events, n=5/5 hearts, p < 0.05¹⁰). C) LV initiation of pVT in LV after 0.5 mM 4-AP (n=4/5 hearts, Fisher exact test p < 0.05). The black area indicates failed activation due to conduction block (Movies of activation are provided as supplemental material movie S1 and S2).

RV CMs origin of EADs in LQT1 rabbits

To verify the origin of EADs at the cellular scale, we isolated single myocytes from RV and LV and investigated their propensity to generate EADs under isoproterenol. Figure 2A shows V_m recordings from RV (red) and LV (blue) myocytes (n=12 per group) under 50 nmol/L isoproterenol. In line with the intact heart optical mapping study, current-clamp experiments on single cardiomyocytes (CMs) isolated from the RV of LQT1 rabbits showed frequent EADs, while myocytes from the LV did not show EADs (Figure 2A&B). CMs from LMC rabbits did not show EADs under 50 nmol/L isoproterenol (see Supplementary Figure S3). Note that despite the prolonged APD of LV CMs, V_m at the plateau did not oscillate to form EADs (blue traces in panel A). Detailed examination of V_m traces at expanded time scale (Figure 2C) suggests that the initial plateau V_m is substantially lower than that of LV CMs. Of note, this lower plateau V_m was associated with large-amplitude EADs. EADs were not induced under the same condition in LMC CMs (supplementary Figure S3).

Figure 2.

Frequent EADs despite shorter APD in RV myocytes. A) Representative V_m traces recorded from RV (red, n=12 cells) and LV (blue, n=11 cells) myocytes in 50 nM isoproterenol. B) EAD incidence. EADs were frequently in LQT1 RV myocytes (67% in RV vs. 0.18 % LV myocytes, Fisher’s exact test; p < 0.05). C) Detail traces of initial action potential repolarization phase in expanded time scale. Note that the initial rapid repolarization in RV cells is associated with EAD formation. D & E) APD₉₀ and APD₃₀/APD₉₀. APD₉₀ from LV cells showed prolonged APD despite lack of EADs. APD₃₀/APD₉₀ indicates that RV cells have lower V_m during plateau phase, showing an association between rapid initial repolarization phase and EAD formation, suggesting a potential role of I_to in EAD formation (ANOVA, p < 0.05).

Previous studies indicated a general association between prolongation of APD and risk for EADs. In LQT1, APD₉₀ was substantially longer in LV CMs vs. RV CMs (Figure 2D), yet EADs were more frequent in RV CMs than LV CMs. Further analysis of APD₃₀/APD₉₀ (Figure 2E) show that the action potentials of RV cells have rapid repolarization during the early phase of the action potential (phase 1 period). Thus, these data show that in the case of LQT1 rabbits, EADs originate from a region (RV) with a shorter APD, both at the cellular scale and the tissue scale (reference¹⁰ and Figure 1).

I_to is larger in RV CMs

Given the shape of the action potential and the rapid initial repolarization, we hypothesized that the main difference between RV and LV LQT1 CMs is I_to. The ionic currents were measured using voltage clamp under isoproterenol (50 nmol/L). The results show a substantially higher peak I_to amplitude in RV CMs than LV CMs (Figure 3A–C).

Figure 3.

Amplitude and inactivation kinetics of I_to are markedly different between RV and LV cells. Sample voltage clamp traces for I_to from RV (A) and LV cells (B). The inactivation kinetics of I_to showed two-exponential decays, fast-inactivation component (fi) and slow-inactivation component (si), see methods for detail. C) The total amplitude of I_to is 28% larger in RV than LV (17.1 ± 2.1 pA/pF in RV vs. 12.4 ± 1.4 pA/pF in LV at 50 mV, ANOVA p<0.05, n=16 and 15 cells for RV and LV from n=5 hearts). D) I_to,fi and I_to,si from RV (n=16) and LV cells (n=15 from n=5 hearts). Both fast and slow components are greater in RV. E) The fast-inactivation time constants (τ_fi) are the same in RV and LV cells, whereas the slow-inactivation time constant (τ_si) is longer in RV (see supplemental Table S1). F) mean activation and inactivation curves of I_to,_total from LQT1 (solid lines) and LMC (dotted lines) myocytes. After normalized to the maximum value, the data were fitted to the Boltzman function and used for computer modeling.

I_to inactivation kinetics from our voltage-clamp measurements could be best fitted with two exponential decays (see Equation (1) and Figure 3A–B). We separated these two exponential decays (see Methods for details): a fast-inactivation component marked as I_to,fi with a time constant of τ_fi and a slow-inactivating component as I_to,_si with a time constant of τ_si. Detailed analysis of two components revealed that I_to,fi and I_to,si are 30% and 83% greater in RV (g_to,fi = 0.068±0.005 mS/μF and g_to,_si = 0.033±0.005 mS/μF) than LV CMs (g_to,fi = 0.052±0.005 mS/μF and g_to,_si = 0.018±0.002 mS/μF, see panel D), respectively. The τ_fi were not significantly different, but the τ_si of the slow component was 25% slower in RV CMs (τ_si = 73±6.7 ms in RV vs. 59±6.9 ms at high voltage, see panel E).

It is generally thought that I_to in rabbits recovers slowly from inactivation (I_to,s) and makes only a limited contribution to APD. We investigated I_to recovery kinetics using a double-pulse protocol (Figure 4A). We found that rabbit CMs have a substantial magnitude of I_to,f (τ_fast = 19.5 ms), which was larger in RV compared to LV myocytes (I_to,f = 8.9±1.7 and 5.3±1.6 pA/pF for RV and LV respectively, Figure 4B–C). I_to,s (τ_slow = 3.41 sec) was similar in RV and LV. In addition, the recovery of I_to,si was analyzed by separating fast- and slow-inactivation components of I_to during the double-pulse I_to recovery protocol. The slowly inactivating component of I_to was significantly larger in RV vs. LV (8.6±1.5 and 3.2±0.76 pA/pF for RV and LV respectively, p < 0.05). This results in a significant difference in the amount of available I_to,si when the interpulse interval is >600 ms (Figure 4B).

Figure 4.

I_to recovery from inactivation. A) The recovery kinetics was tested by a double-pulse protocol with interpulse time varying from 50 ms to 15 sec (n=12 RV and 7 LV cells from n=3 hearts). B) The amplitudes of the slow and fast inactivating components of I_to (I_to,si and I I_to,fi) as a function of inter-pulse interval were determined by fitting the time course of I_to decay during the second pulse to a double exponential function. The x-axis of inter-pulse intervals is in a logarithmic scale. C) The amplitudes of I_to,fi and I_to,si from RV and LV. Fast and slow-inactivating components (I_to,fi and I_to,si) of each I_to,f and I_to,s were calculated as described in Methods and represented as a stacked column plot. D) Western blots of Kv4.2, Kv1.4, and KChIP2 from LQT1 hearts. E). The accessory unit of I_to, KChIP2, known to affect inactivation and recovery kinetics, was twofold higher in RV (ANOVA, p < 0.05). The currents were measured under 50 nM isoproterenol. Additional Kv4.2 and Kv1.4 gel analyses are provided in the supplemental material.

The expression of K⁺ channel interacting protein (KChIP2) was significantly higher in RV (Figure 4D–E) but the western blots of Kv4.2 and Kv1.4 of RV and LV CMs (panel D and Figure S6) were not significantly different. Since KChIP2 modulates I_to,f channel expression and kinetics, higher KChIP2 expression in RV likely provides molecular mechanisms underlying differences of I_to,f magnitude and kinetics between RV and LV myocytes.

Of note, there were no significant differences in I_Ca,L or its inactivation properties as well as I_Kr and I_K1 between RV and LV myocytes under 50 nmol/L isoproterenol (Figure 5A–C). These results strongly suggest that I_to is responsible for APD gradient and RV-specific EAD formation in LQT1 rabbits.

Figure 5.

I_Ca, (A), I_Kr (B), and I_K1 (C) did not differ between RV and LV cells. All currents were measured under 50 nM isoproterenol.

Ca²⁺ handling between RV and LV CMs

Ca²⁺ handling and its coupling with voltage have frequently been suspected of promoting EADs.¹⁷ Our previous studies of EAD mechanisms in LQT2 rabbits also demonstrated that abnormal Ca²⁺ dynamics and leaky ryanodine receptors increased the forward mode of the electrogenic Na⁺/Ca²⁺ exchanger (I_NCX), slowing down repolarization and allowing reactivation of the I_Ca,L window current to produce EADs.¹⁴ We therefore investigated whether differences in Ca²⁺ handling between RV and LV myocytes in LQT1 rabbits underlie preferential EAD formation in RV using confocal Ca²⁺ imaging. Figure 6 shows an example of simultaneous recordings of V_m, cytosolic and SR Ca²⁺ during action potentials. Panels A&B show typical examples of V_m and Ca²⁺ from RV and LV CMs. The amplitude and the fractional release of Ca²⁺ from SR did not differ between RV and LV myocytes. We found no significant difference in Ca²⁺ handling between RV and LV (panel C–E), suggesting that Ca²⁺ handling does not underlie preferential EAD formation in RV myocytes. Western blots of Ca²⁺-handling proteins from RV and LV were not significantly different (see supplementary Figure S7).

Figure 6.

Ca²⁺ handling is not responsible for preferential EAD formation from RV cells. A & B) Simultaneous recordings of V_m, Ca²⁺ transients (Ca_Cyto), and SR Ca²⁺ contents (Ca_SR) from single RV and LV cells. Note that higher Ca²⁺ level during action potential plateau in LV did not trigger EADs. C) Amplitudes of Ca²⁺ transients. D) Ca²⁺ leak from RyR. E) Ca²⁺ uptake. Ca²⁺ handling was not markedly different between RV and LV cells.

I_to facilitates EAD formation: Computer-modeling studies

We examined whether computer modeling could reproduce the AP waveforms in RV and LV under isoproterenol, only through varying I_to. Figure 7A shows that when I_to conductance and inactivation kinetics are comparable to those measured in LV myocytes (g_to,si = 0.025 mS/μF, g_to,fi = 0.08 mS/μF, τ _si = 50 ms), the AP produced a long plateau at high V_m (solid blue curve) with long APDs (> 700ms). In contrast, when I_to conductance and inactivation kinetics are comparable to those measured in RV myocytes (g_to,si = 0.050 mS/μF, g_to,fi = 0.08 mS/μF, τ_si = 80 ms), we see an AP waveform similar to that found in experiments, with long APD due to many EADs (Figure 7A solid red curve). When parameters are altered to include the I_Ks current to model LMC1 RV and LV, AP traces show much shorter APD without EAD formation in both cell types (blue and red dashed lines in LV and RV, respectively). To further elucidate the role of I_to in EAD formation, we distinguish in Figure 7B the contributions of the fast- and slow-inactivating components of I_to, I_to,fi (FI) and I_to,si (SI), respectively. Two main differences of I_to properties between the LV and RV can be distinguished. Firstly, the larger peak I_to in the RV produces a faster initial repolarization during phase 1 of the AP (i.e. a larger notch). Secondly, the larger amplitude of the slow-inactivating component of I_to (I_to,si) in the RV makes a dominant contribution to the subsequent repolarization during phase 2 of the AP (t<200 msec), thereby bringing the voltage into the window range for reactivation of I_Ca,L and promoting EADs. Figure 7C shows that re-activation of I_Ca,L coincides with EADs in RV, while in LV repolarization is complete before I_Ca,L sufficiently recovers from inactivation to initiate an EAD during late phase 3 of the AP. The faster rate of repolarization during late phase 3 in LV is due to I_Kr (Figure 7D), which activates significantly more in LV than RV due to the long AP plateau at high V_m. Figure 7E directly compares I_Kr at the same V_m and shows that I_Kr contributes twice as much to repolarization in LV than RV in the −20 to +10 mV range relevant to EAD onset.

Figure 7.

Computer modeling study of EADs reproduces I_to-dependent EADs in LQT1 myocytes under isoproterenol stimulation. A) Computed V_m traces for different I_to model parameters fitted to voltage-clamp measurements of I_to in RV and LV cells of LQT1 and LMC rabbits. Only RV (red) from LQT1 demonstrated EADs. B) I_to traces broken into fast-inactivating component (red in RV, dark blue in LV) and slowly inactivating component (orange in RV, light blue in LV) for action potentials in panel A. C) I_CaL from RV (red) and LV (blue) during action potentials in panel A. D) I_Kr during action potentials in panel A. E) Dynamic I–V curves showing I_Kr during repolarization as a function of V_m. Note that in the range of V_m oscillations during EADs, I_Kr from RV (red) is markedly lower than LV (blue). F) EAD formation in the parameter space of I_to,si conductance (g_to,si, Y axis) vs. I_Ks conductance (g_Ks, X axis) with and other parameters g_to,fi = 0.074 mS/μF, τ_si = 69 ms, and τ_fi = 8.5 ms. Representative AP traces within each region are shown, with parameters indicated by green diamonds. The right-most green diamond corresponds to parameters representative of LMC cells.

Figure 7F shows that, with addition of I_Ks into the model, EADs become fewer with increasing I_Ks conductance and vanish for a normal value of g_Ks corresponding to LMC myocytes. Figure 7F also shows the importance of varying I_to conductance in absence of I_Ks. When I_to conductance is small (g_to,si<0.04 ms/μF), as in LQT1 LV CMs, the high V_m AP plateau prevents EAD formation, despite APD prolongation. For greater I_to conductance, as in LQT1 RV CMs, multiple EADs occur (green square on the y axis for g_Ks = 0). In this case, I_to conductance is large enough for V_m to traverse the critical window for reactivation of I_Ca,L before sufficient activation of I_Kr, but small enough to avoid complete repolarization, allowing EADs.

To further demonstrate the important role of slow I_Kr activation kinetics in EADs formation, we repeated the simulations represented in Figure 7F using a Hodgkin-Huxley (HH) formulation of I_Kr with parameters fitted to our voltage-clamp data (Figure S10). The results shown in Figure S10 reproduce the RV and LV AP phenotypes obtained with the multi-state Markov I_Kr model (compare Figures 7A–D with Figures S10A–D). Furthermore, in Figures S10E–F, we compare the steady-state activation/inactivation curves and activation-time constant for the new I_Kr HH formulation fitted to our voltage-clamp data and the Zeng et al HH I_Kr formulation¹⁸ commonly used in rabbit ventricular myocyte ionic models. This comparison underscores the importance of accurate modeling of I_Kr activation kinetics to elucidate arrhythmogenic mechanisms in the setting of LQT1 where I_Ks is absent.

Discussion

LQT1 is the most common form of long QT syndrome and accounts for almost half of genotyped cases.² Pore missense mutations that confer loss of function with dominant negative effect are associated with a more severe form of the disease with a higher frequency of cardiac events.¹⁹ This study highlights the crucial role of I_to in repolarization when I_Ks is absent and elucidates the molecular underpinning for the malignant phenotype of LQT1. At the organ level, we found that the heterogeneity of I_to underlies pVT initiation by creating APD dispersion and EAD formation that propagates unidirectionally to the RV region to form a reentry. Consistent with those observations, we found that cellular EADs are present in myocytes isolated from RV, but absent from LV myocytes despite much prolonged APD in LV. Computer modeling, combined with detailed voltage-clamp and confocal Ca²⁺ imaging, revealed that heterogeneity in I_to, both in terms of amplitude and inactivation kinetics, causes membrane voltage to traverse the window for reactivation of I_Ca,L sooner, before I_Kr has had sufficient time to activate and prevent the depolarization that underlies EAD formation and pVTs.

EADs formation has been traditionally associated with an instability of V_m dynamics driven by reactivation of I_Ca,L during the plateau phase of the AP.^{20, 21} While I_Ca,L is a necessary depolarizing current to induce EADs, several other membrane currents have been shown experimentally or computationally to modulate EAD formation, including I_NCX^{14, 22}, I_Ks²³, and I_to.²³ The present study highlights heterogeneity in I_to amplitude and inactivation kinetics as major factors in the initiation of EADs and reentry in LQT1 rabbits.

The role of I_to in the genesis of EADs has also been highlighted in previous experimental and computational studies in isolated rabbit ventricular myocytes exposed to oxidative stress^{23, 24} or hypokalemia.²⁴. A main finding of those studies is that, even though I_to helps repolarization, blocking this current can (counter-intuitively) suppress EADs. This paradoxical effect is linked mechanistically to the fact that Ito can lower the AP plateau voltage into a range where activation of IKs is slowed sufficiently to allow reactivation of ICa,L and induction of EADs.^{23, 24} Importantly, I_to is also implicated in arrhythmogenesis in Brugada syndrome. By contrast, deletions of I_to in several models in mice were not associated with sudden cardiac death.²⁵ Our results highlight a similar mechanism of Ito in LQT1 rabbits, but with the slow activation of IKr allowing reactivation of ICa,L when the lower AP plateau voltage is reached rapidly.

Two factors may precipitate pVT initiation in LQT1 rabbits: i) APD gradient between RV and LV and ii) EAD formation from the short-APD region. Enhanced APD dispersion has been well recognized as a substrate for reentry in LQTS.^{4, 26} However, in order to initiate reentry, the propagating AP must encounter a conduction block. In LQT1, because EADs form in RV where APD is shorter, the propagation of EADs eventually reaches the LV region where APD is longer, causing conduction block and initiation of reentry (Figure 1). Therefore, this study emphasizes the importance of I_to heterogeneity, which allows EAD formation in the short-APD region, and forming unidirectional conduction and reentry in LQT1 rabbits.

I_to heterogeneity between RV and LV^{27, 28}, and also epi- to endocardium^{13, 29}, has been well recognized. In our transgenic rabbit model of LQT1, I_to heterogeneity between RV and LV may have greater impact on arrhythmogenesis but in large animals including human, transmural heterogeneity of I_to can also have significant impact on EADs and reentry formation due to sufficient transmural dimension to form reentry. Two major transient outward (I_to) currents are likely responsible for this current in large mammals: I_to,s (Kv1.4), and I_to,f (Kv4.2 and Kv4.3), which differ in their recovery-time constants^{12, 30} but both of which have multi-exponential inactivation kinetics¹². Even though previous work suggests that the main component in rabbits is I_to,s, our double-pulse voltage-clamp study demonstrated that I_to,f in rabbit RV constitutes ~28% of total I_to and is the major component that underlies a large I_to in RV (Figure 4). Our western blot study demonstrated that I_to heterogeneity is associated with KChIP2 expression level (Figure 4E). KChIP2 has been reported to increase trafficking and expression of Kv4.2 and Kv4.3 that underlie I_to,f.³¹ KChIP2 slows I_to inactivation kinetics³², which may partly explain the molecular mechanisms underlying greater I_to in RV. However, further studies are needed to clarify the origin of the greater amplitude of the slow-inactivating component of I_to (I_to,si) in RV (Figure 3 and 4) and its potential transmural heterogeneity, which is potentially linked to different phosphorylated states of Kv1.4 and Kv4.2/4.3 channels in RV and LV^{12, 30}

Our computer modeling study showed that I_to,si plays a crucial role in determining the plateau V_m and initiating EADs (Figure 7 and Figure S12). I_to,si recovers to a larger amplitude in rabbit RV than LV at moderately slow heart rate > 600 ms (Figure 4). This may explain why LQT1 rabbits show dramatically more SCDs after bradycardia caused by AV node ablation.³³

I_to heterogeneity is no greater in LQT1 than in LMC (supplemental figure S4). However, this heterogeneity has a greater impact on pVT initiation in LQT1 than in LMC (supplementary Figure S2–3), where I_Ks abolishes the arrhythmogenic effect of I_to at a cellular level (Figure 7F). Previous computer modeling studies of LQT1 found critical roles of I_Ks on APD prolongation³⁴ and EAD formation under reduced I_Kr condition.³⁵ In particular, the study by O’Hara and Rudy³⁵ showed that, in addition to isoproterenol, partial block of I_Kr was needed to evoke EADs when I_Ks kinetics is altered by the Q357R KCNQ1 mutation while in the present study isoproterenol suffices to evoke EADs without block of I_Kr in the absence I_Ks. Since I_to is heterogeneously expressed even in normal myocardium between RV and LV as well as epi- and endocardium,^{12, 13, 36} our study suggests that I_Ks has an important protective role beyond repolarization reserve by acting as a safety mechanism that counters the arrhythmogenic effect of I_to heterogeneity revealed by the present study. Hence, the main consequence of the loss of function of I_Ks is to unmask the arrhythmogenic role of I_to heterogeneity, which is masked by I_Ks in LMC hearts.

Regarding Ca²⁺-handling proteins, the hyperphosphorylation of RyR was recently shown to play an important role in the genesis of EADs in myocytes isolated from LQT2 transgenic rabbits carrying a loss-of-function mutation of the HERG potassium channel encoding the I_Kr channels.¹⁴ Experiments and computational modeling demonstrated that EADs are promoted by aberrant late Ca²⁺ releases by hyperactivity of RyR Ca²⁺ channels. The Ca²⁺-handling machinery under isoproterenol in LQT1, however, is not markedly different between myocytes forming EADs (RV) vs. myocytes not forming EADs (LV), indicating that EAD formation in RV cannot be ascribed to differences in Ca²⁺ homeostasis in RV vs. LV (Figure 6). However, it is still possible that sustained Ca²⁺ release during AP plateau can influence APDs and further studies are needed for potential roles of Ca²⁺ handling in LQT1.

Our voltage-clamp data from both LMC and LQT1 rabbits demonstrated the existence of both fast and slow components of I_Kr activation, with the slow component being dominant for V_m up to 10 mV. Unless V_m is high for a long duration (such as in LQT1 LV CMs), this dominant slow component will prevent I_Kr from significantly activating, which is key for EAD formation (supplemental Figure S9 and S10). Since I_Kr activation and deactivation kinetics are similar in rabbits and humans,³⁷ our results suggest that I_Kr kinetics is likely to play an important role in LQT1 EAD formation.

Limitations

This study focuses on the role of I_to in the initiation of EADs through determining the trajectory of V_m during phase 1 of early repolarization, but did not study the potential effects of sustained I_to that may also have contributed to the maintenance of multiple EADs. I_to,s is the major component of I_to in rabbits, and EADs in different species (including humans) may occur at faster heart rates. In optical mapping, 4-AP may have non-specific action on I_Kr and further prolong APD. Chromanol is known to alter I_to at higher concentrations, and I_to measurements in LMC myocytes may have been underestimated.

Conclusion

Using a transgenic rabbit model of LQT1 and computer modeling, we demonstrated that I_to plays a key role in arrhythmogenesis, creating both triggers and an arrhythmogenic substrate for reentry. When I_Ks is absent, complex interactions between I_to, I_CaL, and I_Kr transduce the marked heterogeneity of I_to properties into a marked heterogeneity of the AP phenotype leading to pVT initiation. This study suggests that I_to is a major contributor to LQT-related arrhythmias and that its heterogeneity may underlie regional EAD formation and initiation of pVTs.

WHAT IS KNOWN?

• Long QT syndrome type 1 (LQT1), an inherited disease caused by loss-of-function mutations in KCNQ1, is associated with sudden cardiac arrest triggered by early afterdepolarizations (EADs).

• EADs have been traditionally associated with an instability of V_m dynamics driven by reactivation of the L-type Ca²⁺ current (I_Ca,L) during the plateau phase of the action potential that initiates polymorphic VT (pVT).

WHAT THE STUDY ADDS?

• It uses a transgenic rabbit model of LQT1 to shed light on the mechanisms of EADs formation and pVT initiation at the cellular and organ scales.

• It finds that the transient outward K⁺ current (I_to) is both significantly larger and inactivates more slowly in the right ventricle (RV) than the left ventricle (LV), and links mechanistically this regional variation of I_to properties to the observed regional variation of EADs frequency, abundant in the RV but absent in the LV.

• It shows that larger I_to in the RV causes V_m to traverse the critical window range for reactivation of I_Ca,L earlier during the AP plateau before sufficient activation of I_Kr promoting EADs formation in the short-APD region. These EADs propagate uni-directionally to form reentry and polymorphic ventricular tachycardia (VT).