Modulation of hERG potassium channel gating normalizes action potential duration prolonged by dysfunctional KCNQ1 potassium channel

Abstract

Long QT syndrome (LQTS) is a genetic disease characterized by a prolonged QT interval in an electrocardiogram (ECG), leading to higher risk of sudden cardiac death. Among the 12 identified genes causal to heritable LQTS, ∼90% of affected individuals harbor mutations in either KCNQ1 or human ether-a-go-go related genes (hERG), which encode two repolarizing potassium currents known as I_Ks and I_Kr. The ability to quantitatively assess contributions of different current components is therefore important for investigating disease phenotypes and testing effectiveness of pharmacological modulation. Here we report a quantitative analysis by simulating cardiac action potentials of cultured human cardiomyocytes to match the experimental waveforms of both healthy control and LQT syndrome type 1 (LQT1) action potentials. The quantitative evaluation suggests that elevation of I_Kr by reducing voltage sensitivity of inactivation, not via slowing of deactivation, could more effectively restore normal QT duration if I_Ks is reduced. Using a unique specific chemical activator for I_Kr that has a primary effect of causing a right shift of V_1/2 for inactivation, we then examined the duration changes of autonomous action potentials from differentiated human cardiomyocytes. Indeed, this activator causes dose-dependent shortening of the action potential durations and is able to normalize action potentials of cells of patients with LQT1. In contrast, an I_Kr chemical activator of primary effects in slowing channel deactivation was not effective in modulating action potential durations. Our studies provide both the theoretical basis and experimental support for compensatory normalization of action potential duration by a pharmacological agent.

The biophysical characteristics of an action potential (AP) are controlled by contributions of different ionic currents. In the human cardiac system, synchronized firing and conduction of APs of electrically coupled cardiomyocytes give rise to rhythmic contraction, which is commonly monitored by the surface electrocardiogram (ECG). Rapidly activating K⁺ current (I_Kr) and slow delayed rectifier K⁺ current (I_Ks) are major components of AP repolarization and hence have roles in determining AP duration (APD) (1, 2). Reducing their contributions pharmacologically or genetically is reflected in a lengthening of the Q-T interval between the QRS complex and the T wave in the ECG. Long QT syndrome (LQTS) caused by prolongation of APD has a prevalence of 1 in 5,000 among the general population. As a predisposed condition, patients with LQT have a high incidence of sudden cardiac death. Among 12 different genes causal to LQTS, remarkably, 90% of patients harbor mutations in KCNQ1 or KCNH2 [human Ether-a-go-go–Related Gene (hERG)] genes, causing a reduction in either I_Ks or I_Kr currents. Therefore, understanding the contributions of repolarizing ionic currents and potential compensatory effects between them is of considerable interest.

Reduction of the I_Ks current due to mutations in KCNQ1 has been reported in 40–55% of patients with LQT and represents the most frequent cause of congenital LQTS (3). In both rabbit and guinea pig, the I_Ks component is present in adult cardiomyocytes. Indeed, reduction of KCNQ1 expression resulted in prolongation of the QT interval (4). However, several important features of human cardiac function including diversity of genetic mutations of I_Kr or I_Ks and drug-induced LQT phenotype cannot readily be reproduced or modeled in animal systems (5). R-L3 is an active compound that was reported to potentiate I_Ks in guinea pig and had similar effects on KCNQ1. However, it was not effective on the oligomeric KCNQ1/KCNE1 channel, the molecular determinant for I_Ks (6). Phenylboronic acid (PBA) activates KCNQ1/KCNE1 complexes at millimolar effective concentrations (7).

Using induced pluripotent stem cells (iPSCs) derived from members of a family affected by LQT syndrome type 1 (LQT1) and harboring an R190Q heterozygous missense mutation in KCNQ1, we generated and investigated functional cardiomyocytes. The patient-derived cells recapitulated the electrophysiological features of the disorder with noticeably prolonged APs (8). By recording autonomous beating cardiomyocyte clusters displaying ventricular features, we investigated the feasibility of quantitatively analyzing both control and LQT1 action potentials. On the basis of this model system, we then tested whether the specific perturbation of hERG channel gating would effectively rescue the LQT1 disease phenotype.

Results

Quantitative Analyses of Compensatory Effects Between I_Ks and I_Kr.

A detailed characterization of different ionic components in differentiated human cardiomyocytes from pluripotent stem cells has not been reported. We compared the expression of genes generating the cardiac I_Kr (HERG1a and HERG1b) and I_Ks (KCNQ1) currents in cardiac myocytes obtained from control and LQT1 iPSC lines, as well as in adult and fetal human heart. Quantitative real-time PCR (qRT-PCR) analyses revealed that HERG transcript levels were similar between control and LQT1 cardiomyocytes (Fig. 1A), which is consistent with the earlier data of similar I_Kr current densities in control and LQT1 cells (8). Therefore, the subsequent analyses use the same I_Kr value for both the healthy control and LQT1 in modeling.

Fig. 1.

Comparison between simulated and recorded ventricular-like action potentials from iPSC-derived cardiomyocytes from a healthy control and patient with LQT1. (A) qRT-PCR analysis of HERG1a, HERG1b, and KCNQ1 in cardiomyocyte clusters from control (bar with light shading) and LQT1 (open bar) iPSCs at 4-mo maturation compared with human adult (AH, solid bar) and fetal (FH, bar with dark shading) heart. Expression values are relative to those of AH, normalized to GAPDH, and presented as mean ± SD, n = 3. (B) Representative traces of the ventricular-like AP recorded from cardiomyocytes derived from either a healthy control or a patient with LQT1 with R190Q mutation as indicated. The experimentally recorded traces are shown on the Left, whereas the simulated ventricular-like APs based on a modified ten Tusscher model are shown on the Right.

To assess the contributions of different current components in a typical ventricular-like AP, we recorded spontaneous electrical activity of human cardiomyocytes derived from iPSCs. We then used the model first developed by ten Tusscher, which describes a variety of electrophysiological behaviors of adult human ventricular cardiomyocytes (9). One unique feature in differentiated cardiomyocytes is the autonomous contraction absent in acutely isolated human cardiomyocytes, consistent with the notion that differentiated myocytes in culture are not mature (10). Some components in the ten Tusscher model were modified to match the recorded spontaneous APs. Inclusion of a late sodium current component (I_NaL) and a funny current component (I_f) afforded a better cross-examination between model-based simulation and the recorded action potential (SI Experimental Procedures). With the modifications outlined in Table S1, the simulated AP traces agree well with those of the representative APs recorded in control cell clusters (Fig. 1B). The key parameters include APD₉₀ = 388.4 ms and action potential amplitude (APA) = 82.5 mV for simulated traces and APD₉₀ = 377.3 ± 14.0 ms and APA = 79.6 ± 1.6 mV for recorded traces (Table 1). To model LQT1 APs, we reduced I_Ks by 30% of control, the level of I_Ks measured in LQT1 iPSC-derived myocytes with the R190Q-KCNQ1 mutation (11). This procedure resulted in a simulated action potential of APD₉₀ = 523.6 ms and APA = 84.9 mV that matches well with the recorded trace of LQT1 cells with APD₉₀ = 530.6 ± 18.8 ms and APA = 82.8 ± 2.2 mV (Fig. 1). The agreement between experimental data and simulated data supports the applicability of the modified model for quantitative analyses.

Table 1.

Comparison of simulated AP parameters with experimental values

	APA, mV	APD50, ms	APD90, ms	dV/dtMax, V/s	MDP, mV	APD90/APD50
Simulated control	82.5	339.7	388.4	8.4	−68.5	1.14
Recorded control (n = 14)	79.6 ± 1.6	330.7 ± 13.3	377.3 ± 14.0	16.3 ± 1.2	−64.5 ± 1.4	1.14 ± 0.01
Simulated LQT1	84.9	444.6	523.6	8.3	−67.9	1.18
Recorded LQT1 (n = 26)	81.1 ± 1.3	463.5 ± 16.9	530.6 ± 18.8	13.4 ± 0.9	−63.4 ± 1.0	1.15 ± 0.01
Simulated LQT1 + compound	84.0	323.0	360.4	18.1	−79.3	1.12
Recorded LQT1 + compound (n = 9)	82.3 ± 2.2	340.8 ± 12.6	379.7 ± 13.3	12.0 ± 1.0	−66.2 ± 1.5	1.12 ± 0.01

Key parameters are summarized of a simulated or recorded action potential from a healthy control and a patient with LQT1. For the simulation study, the values were obtained in the steady state. For the experimental study, the values were based on the real recording from small beating clusters of myocytes.

Therefore, we examined the robustness of the modified model by testing how reduction of I_NaL or increase in I_Kr would affect action potential in control or LQT1 cardiomyocytes. In adult human cardiomyocytes I_NaL is ∼0.055 nS/pF (11). Fig. 2A shows that when I_NaL is enhanced approximately sixfold (0.33 nS/pF) as seen in F1473C mutation in the SCN5A gene that leads to LQT3 (12), the control displays clear prolongation of APD. Specific sodium channel block therefore reduces this component and brings the duration back to 400 ms, similar to control values (Fig. 2A). However, in the LQT1 model, where I_Ks is reduced to 30%, the reduction of the normal I_NaL level was less effective on APD (Fig. 2B), consistent with an earlier report (13). Similarly, the I_NaL reduction had no significant effect on APD of the control either. This model-based prediction is in agreement with that of the experimental data, where specific sodium channel blockers are more effective to correct the defects in patients with LQT3 (14). In addition to I_NaL, I_Kr is another component critical for determining APD. Therefore, we examined whether an increase in repolarizing I_Kr could normalize the prolonged APD in LQT1 cardiomyocytes. The hERG potassium channel is the primary molecular determinant responsible for the I_Kr. A biophysical property of hERG absent in KCNQ1 is the rapid inactivation during depolarization. Thus, changing in channel inactivation by shifting its voltage sensitivity (V_1/2) to a more positive value would slow the inactivation and increase channel availability during the repolarizing phase of the AP. To examine this, we progressively shifted the V_1/2 from −87 mV to −70 mV, −40 mV, and −10 mV. The model predicts progressive shortening of APD in control cells (Fig. 2C). Importantly, in LQT1 cells the rightward shift of inactivation V_1/2 could also progressively shorten the APD into the normal range (Fig. 2D). Another way to increase I_Kr density during the AP is to slow down the deactivation rate of hERG current. Thus, we simulated the effect of decreasing I_Kr decay rate on APD by increasing the time constant of the activation (deactivation) gate, τx_r1. This change is not sufficient to shorten prolonged APDs in LQT1 cardiomyocytes. Instead, the model predicts a flattering of phase 4 diastolic depolarization and a reduction of firing frequency (Fig. 2E). This prediction is consistent with an earlier study demonstrating that I_Kr decay rate is one of the key determinants of the speed of pace-making depolarization (15). The simulation, despite the fact that some values used were based on empirical estimation or on earlier reports, is consistent with the experimental evidence that a sodium channel blocker is more effective in patients with LQT3, indicative of gene-specific rescue. In contrast, it argues an increase of I_Kr would be able to correct the APD in LQT1 cardiomyocytes. Importantly, the model specifies that a rightward shift of I_Kr inactivation V_1/2 and not a slower deactivation rate is effective in restoring the prolonged APD in LQT1 patients to a normal range.

Fig. 2.

Simulation study of different compound effects on iPSC-derived cardiomyocytes from a healthy control and a patient with LQT1. (A) Simulated AP from a healthy control without compound treatment (I_NaL = 0.055 nS/pF) and a presumed patient with LQT3 without compound treatment (I_NaL = 0.33 nS/pF) or with I_NaL blocker treatment (I_NaL = 0.070 nS/pF). (B) Simulated action potential from a healthy control without compound treatment (I_NaL = 0.055 nS/pF, scale factor of G_Ks = 1), a patient with LQT1 without compound treatment (I_NaL = 0.055 nS/pF, scale factor of G_Ks = 0.3), a healthy control with I_NaL blocker treatment (I_NaL = 0.014 nS/pF, scale factor of G_Ks = 1), or a patient with LQT1 with I_NaL blocker treatment (I_NaL = 0.014 nS/pF, G_Ks =0.3). (C) Simulated APs from a healthy control (scale factor of G_Ks = 1, x_r2 V_1/2 = −87 mV) and a healthy control with a progressive shift in x_r2 V_1/2 = −70 mV, −40 mV, −10 mV, respectively, to show different doses of compound effect on inactivation V_1/2. (D) Simulated APs from a patient with LQT1 without compound treatment (scale factor of G_Ks = 0.3, x_r2 V_1/2 = −87 mV) and a patient with LQT1 with progressive shift in x_r2 V_1/2 = −70 mV, −40 mV, or −10 mV, respectively. (E) Simulated AP from a patient with LQT1 without compound treatment (scale factor of tx_r1 = 1) and a patient with LQT1 with progressive decrease in I_Kr decay. Scale factor of tx_r1 = 2 or 3, respectively, to mimic different doses of compound effect on deactivation.

Identification of a Specific Gating Modulator.

Using automated electrophysiology, we have conducted a large compound library screen to identify and characterize small molecule modulators for hERG potassium channels (16). Although a vast majority of modulators were inhibitors, a small number of activators have been identified in the primary screen. On the basis of the model, we examined and selected hits that have biophysical characteristics consistent with affecting voltage sensitivity of inactivation. One scaffold with unique core structure was confirmed with a resynthesized compound. We then purchased or synthesized various derivatives to examine both potency and efficacy, using automated electrophysiology. ML-T531 (or CFK332-B) is one potent activator with highest efficacy (EC₅₀ = 3.13 ± 0.47 μM (Table 2) and was selected for further investigation.

Table 2.

Structure–activity relationship (SAR) of ML-T531

To experimentally characterize the specificity of ML-T531 on I_Kr and I_Ks, we performed manual voltage clamp recording of hERG currents, using a standard protocol with a Chinese hamster ovary (CHO) cell line stably expressing hERG channels. For comparison, recordings were made from a CHO cell line coexpressing cDNAs of KCNQ1 and KCNE1, the molecular determinants for I_Ks (17, 18). Fig. 3A shows that in the absence of compound, hERG currents display a current–voltage relationship in agreement with the literature. In the presence of ML-T531, a dramatic increase in steady-state current was observed. In contrast, under the same conditions, KCNQ1/E1 coexpression exhibited no potentiation, but a slight current reduction was observed (Fig. S1). In addition, ML-T531 has no noticeable effects on other cardiac channels including Cav1.2, Kir2.1, Nav1.5, and Kv4.3. These channels are the molecular determinants for I_Ca, I_K1, I_Na, and I_to (Fig. S2).

Fig. 3.

ML-T531 effects on expressed hERG and KCNQ1/KCNE1 channels. (A) ML-T531, when tested at 10 μM, significantly potentiates hERG steady-state currents and tail currents in CHO cells. The holding potential was −80 mV. The steady-state currents were examined from −80 mV to +80 mV in 20-mV increments, whereas tail currents were elicited at −50 mV. (B) Steady-state current density of hERG either in the absence or in the presence of 10 μM ML-T531, plotted against the voltage. (C) ML-T531, when tested at 10 μM, has no significant effect on hERG activation V_1/2. Tail currents were elicited at −50 mV and the G-V curve was fitted by the Boltzmann equation. (D) ML-T531, when tested at 10 μM, slows hERG deactivation rates. Tail current deactivation was examined from −130 mV to −80 mV and was fitted by a standard double-exponential equation. (E) Inactivation of hERG in the presence and the absence of 10 μM ML-T531 was measured at different voltages using the indicated voltage protocol.

To further characterize ML-T531 effects on hERG channels, steady-state current density was plotted against voltages from −80 mV to +40 mV (Fig. 3B). The enhancement of current was much more pronounced at positive compared with negative voltages. To examine whether ML-T531 affects the conductance–voltage (G-V) relationship, the relative conductance at −50 mV was plotted against step voltages (Fig. 3C). ML-T531 at 10 μM caused no detectable change in the activation V_1/2. Further characterization of ML-T531 effects on inactivation and deactivation was performed by measuring the time constants in the presence or absence of 10 μM ML-T531. Fig. 3D shows that ML-T531 slowed down the deactivation rate at all voltages tested, indicating its capability to increase occupancy of the channel open states (Fig. S3). Furthermore, it caused a rightward shift of inactivation V_1/2 by 28 ± 1.0 mV (n = 8) (Fig. 3E and Fig. S4). On the basis of the results obtained from heterologously expressed channels, we simulated the effects in LQT1 cells (Table 1).

Action Potential of iPSC-Derived Control and LQT1 Cardiomyocytes.

To examine the effects of ML-T531 on native currents, we performed single-cell voltage clamp experiments measuring native I_Kr and I_Ks currents from healthy human iPSC-derived cardiomyocytes. Using a published protocol for human iPSC-derived cardiomyocytes (19), the combined potassium currents mainly contributed by I_Ks and I_Kr were isolated (SI Experimental Procedures and Fig. 4). In the presence of 10 μM ML-T531, the current density consistently increased at all voltages tested for both steady-state and tail currents (Fig. 4 A and B). Moreover, under these conditions ML-T531 enhanced voltage-dependent potassium current amplitudes in the presence of an I_ks-specific inhibitor chromanol 293B. The activation kinetics and rectification properties of the enhanced current are consistent with those of I_Kr (Fig. S5). These results combined with specificity tests (Fig. S2) argue for I_Kr augmentation being the primary effect of ML-T531 in native human cardiomyocytes. To examine whether the LQT1 phenotype could be rescued by the compensatory effect of increasing I_Kr, we recorded the spontaneous action potentials of human cardiomyocytes derived from both control and LQT1 iPSCs (8). The experiments were carried out using spontaneously beating cardiomyocytes with ventricular-like action potentials. These cells have APA greater than 70 mV, maximum diastolic potential (MDP) more negative than −50 mV, and APD₉₀/APD₅₀ smaller than 1.25. Consistent with earlier reports (8), Fig. 5B shows that LQT1 patient-derived cardiomyocytes displayed significantly longer APD values (APD₉₀ = 530.6 ± 18.8 ms, n = 26) compared with those derived from the healthy control individual (APD₉₀ = 377.3 ± 14.0 ms, n = 14). Next, we investigated the effect of 10 μM ML-T531 on spontaneous APs of LQT1 cardiomyocytes. ML-T531 significantly shortened the APD of the diseased cardiomyocytes and normalized the APD values comparable to those recorded from the healthy control (Fig. 5 A and C and Table 1). In addition, ML-T531 caused a hyperpolarizing shift in the MDP values (Fig. 5C and Table 1). The above two major effects are consistent with our modeling results for a compound that causes a rightward shift in inactivation V_1/2 and thereby enhances potassium currents during repolarization. Indeed, when the compound effects on recombinant hERG were incorporated into the model, the predicted AP parameters during compound treatment correlated well with those obtained experimentally (Table 1). Furthermore, ML-T531 shortened the APD₉₀ of LQT1 cardiomyocytes in a dose-dependent manner (Fig. 5D). When fitted by the Hill equation, the dose–response curve gives rise to an EC₅₀ value of 5.71 ± 1.42 μM (Fig. 5E), which is comparable to the EC₅₀ (3.13 ± 0.47 μM) obtained in a CHO recombinant system. The similarity of EC₅₀ values measured in LQT1 human cardiomyocytes to that in hERG-CHO cells supports that I_Kr or hERG is the major target of this compound in native cells. Taken together, our results demonstrate the possibility of normalizing the LQT1-related disease phenotype by targeting the voltage dependence (V_1/2) of inactivation of hERG channels.

Fig. 4.

ML-T531 potentiates native potassium currents in iPSC-derived cardiomyocytes. (A) A pair of representative traces from a single differentiated cardiomyocyte in the absence and the presence of 10 μM ML-T531 as indicated. The holding potential was −60 mV. Steady-state currents were examined from −10 mV to +50 mV in 20-mV increments, whereas tail currents were elicited at −30 mV. Except for I_Kr and I_Ks, other major ionic components including I_Na, I_K1, I_to, and I_CaL were pharmacologically suppressed under the recording conditions (Experimental Procedures). (B) Summary of the compound effects on steady-state current (Upper) and tail current (Lower) at the indicated voltages (n = 11). (*P < 0.05; **P < 0.001).

Fig. 5.

ML-T531 normalizes the disease phenotype of cardiomycytes derived from a patient with LQT1. (A) A train of action potentials recorded from cardiomycytes derived from a patient with LQT1. The time window of ML-T531 application is indicated. Representative action potentials from indicated areas were expanded to appreciate the difference in APD. (Lower) Parameters for individual action potentials are shown in heatmap format. Data are z-score normalized to the first 20 potentials in the train, with color intensity representing the number of SDs of change from this baseline. The plotted parameters are as indicated on the left. (B) Comparison of APD₉₀ recorded from healthy control cells and cells from a patient with LQT1. (C) Overlay of representative action potential traces of cardiomyocytes from a patient with LQT1 at baseline (black), in the presence of 10 μM ML-T531(red), and after washout (blue). (D) ML-T531 effects on action potential durations (APD₉₀) of cardiomyocytes derived from a patient with LQT1 at the indicated concentrations (n = 3 or more). (E) Dose–response curve of ML-T531 on shortening APD₉₀ in LQT1 cardiomyocytes. The curve was fitted by a Hill equation with a Hill coefficient of 2.0.

Discussion

The modified ionic components in the current model relative to the ten Tusscher model are consistent with the notion that in vitro differentiated cardiomyocytes display electrophysiological characteristics that are closer to the fetal counterparts (10). These characteristics include a lower I_K1 component, reduced value of dV/dt during the initial stage of action potential, and more depolarized MDPs. The revised model enables several valuable predictions testable by experiments using human cells. Almost all model parameters were based on published work with a few modifications (9). The conductance of the I_K1 value was reduced to 3% of the normal value. This change is supported by the autonomous beating commonly seen in these cells and agrees with results of RT-PCR studies that suggest a noticeable reduction of normal levels of I_K1 (20). The I_CaL component was doubled in the simulations to account for the consistently longer APDs recorded from control cells compared with the original model. The revised values, even though some have qualitative support, were assigned artificially to achieve closer correlation between modeled and recorded action potential traces, either in the presence or in the absence of the compound treatments (Table S1). Future directions may include quantification of individual current components so that a more refined model may be established. Because the initial variables for this quantitative analysis were directly adapted from the original ten Tusscher model, clearly our simulated action potentials tend to gradually reach a steady state and gain some immature phenotypes, such as a depolarized MDP and a slower dV/dt_max (Fig. S6 A–F). Under steady state, while we note that LQT3 phenotype become less severe, which is likely due to reduced I_NaL availability under relatively deporlarized MDP, the conclusion remains essentially true; i.e., modulation of voltage dependence of inactivation is more effective than slowing of deactivation (Fig. S6G). To examine whether the general conclusion could be applicable to adult cardiomyocytes, we also tested the idea using original ten Tusscher values and obtained similar conclusion (Fig. S6H).

Cardiomyocytes have several potassium channels. These channels, although permeable to potassium, typically display distinct gating properties. Therefore, it is possible that compensatory rescue as described in this study requires specific gating modulation achievable by modifying specific biophysical properties of the targeted channel. Indeed, it was noted that 10 μM ML-T531 decreased AP firing frequency (Fig. 5A), which is consistent with our modeling results, because ML-T531 slows down the I_Kr deactivation (Fig. 3D). To examine this hypothesis more closely, we tested ginsenoside (Rg3), which potentiates hERG current by predominantly slowing the deactivation kinetics (Fig. S7 A and B) (21). Indeed, we found that when this compound is applied to the beating cardiomyocytes, the AP firing frequency was greatly reduced (Fig. S7 D and G). However, there was minimal change in shape or duration of action potentials (Fig. S7 C and F). This result provides the direct support for use of a specific mechanism of gating modulation to achieve compensatory rescue.

The two potassium currents, I_Kr and I_Ks, play a critical role in repolarizing the membrane to terminate an AP. Reports suggest a possible interaction between a mutant KCNQ1 channel and hERG, resulting in change of hERG trafficking to the membrane (22). However, in our LQT1 iPSC-derived cardiomyocyte model, expression levels of genes coding for I_Kr were similar to those of control cells (Fig. 1A) as well as I_Kr current density, as previously reported (11). Reduction of I_Ks current has been modeled both pharmacologically using a KCNQ1 inhibitor (23) and genetically with transgenic rabbits overexpressing loss-of-function pore mutants of KCNQ1 (4). Using these animal or pharmacological models, the attempts to rescue prolongation of APDs have been carried out using either an I_Kr activator (24) or an opener of ATP-sensitive potassium channels (25). Pharmacological interventions aiming at influencing the APD have been in use for many years, such as class III antiarrhythmics for APD prolongation. The notion of compensatory rescue of prolonged APDs has been previously tested by enhancing K_ATP current (26). Unlike the prolongation of the APD, the pharmacological shortening of a pathologically prolonged APD has been investigated only in very few studies using patient samples that are not readily available (27). Our results using patient iPSC-derived myocytes with disease phenotypes provide an important proof of principle of a therapeutic intervention to shorten APD.

The hERG activator ML-T531 has a distinct chemical structure compared with the previously described compounds that potentiate hERG currents (28–33). The reported activators may be classified according to the modes of action, including slowing of deactivation and modulation of voltage dependence of inactivation. Not all of them have effects on APD in tested animal models. ML-T531 has a pronounced effect on voltage sensitivity of hERG inactivation similar to that reported for ICA-105574. Interestingly, an earlier study has shown that ICA-105574 was capable of shortening APD in isolated cardiomyocytes from guinea pig (28). In contrast, RPR260243, which primarily acts in slowing hERG deactivation, has pronounced effects on T-wave amplitude in guinea pig hearts but almost no effects on QT duration even at 30 μM, the highest concentration tested (30). These experiments are in agreement with our results that reduction of voltage dependence for inactivation is more effective in modulating APD.

Whereas I_Kr may involve more than the hERG channel, it is thought that the hERG subunits are the primary determinants for the pharmacology of I_Kr (34). We noted that ML-T531 has nearly identical potency (EC₅₀) measured by shortening of action potential in cardiomyocytes and by steady current increase in heterologously expressed hERG channels. This correlation argues for the compound specificity on the native hERG channel in human cardiac cells. Our experiments demonstrate an application of human iPSC technologies by testing compound effects on correcting an abnormal functional phenotype due to an inherited cardiac disorder. Future optimization via structure-activity studies may afford improved potency and specificity, thereby increasing its utility.

Experimental Procedures

The general method of modeling is similar to that used in the ten Tusscher model (9) and SI Experimental Procedures (35). Human iPS cell lines from the patient with long QT1 syndrome and the healthy control (8) were maintained in feeder-free mTeSR1 medium. For whole-cell patch-clamp recordings, beating clusters 2–3 mo old were microdissected and disaggregated as previously reported. Traditional whole-cell voltage-clamp recording was performed at room temperature. Additional details are described in SI Experimental Procedures.