2019 Volume 44 Issue 12 Pages 859-870
We validated a motion field imaging (MFI) assay with human induced pluripotent stem cell-derived cardiomyocytes (hiPS-CMs) as a model to assess multiple cardiac liabilities by comparing the guinea-pig Langendorff heart with hiPS-CMs using 4 reference compounds and 9 internal compounds. We investigated repolarization duration, beating rate (BR), conduction speed, contractility, and inhibitory profile of three cardiac ion channels: hERG, Cav1.2, and Nav1.5. For repolarization, the contraction-relaxation duration (CRDc) of hiPS-CMs was generally consistent with the QTc interval of Langendorff heart. However, 2 internal compounds shortened CRDc despite QTc prolongation in Langendorff heart. Cardiac ion channel profiling revealed that hiPS-CMs could not be used to detect QTc prolongation when the value of Cav1.2 IC50 / hERG IC50 for a compound was between 1 and 10, whereas hiPS-CMs showed responses largely consistent with Langendorff heart when Cav1.2 IC50 / hERG IC50 was below 1 or above 10. The accuracy of hiPS-CMs for the BR was not high, mainly because the BR of hiPS-CMs was increased by an inhibition of Cav1.2. The hiPS-CMs were highly sensitive to conduction speed and contractility, able to detect QRS widening caused by Nav1.5-inhibition, as well as decreased LVdP/dtmax caused by the inhibition of Cav1.2 and/or Nav1.5. In conclusion, the MFI assay with hiPS-CMs would be useful for evaluating multiple cardiac liabilities. The ion channel profile helps to interpret the results of MFI assay and correctly evaluate cardiac risks. Therefore, an integrated cardiac safety assessment with MFI and ion channel profiling is recommended.
Human induced pluripotent stem cell-derived cardiomyocytes (hiPS-CMs) have emerged as a novel in vitro model to replace or complement existing cardiac toxicity assays for drug development. Fundamental cardiac functional mechanisms, such as electrophysiological properties that generate action potential (Honda et al., 2011; Ma et al., 2011), electrical connection of cardiomyocytes that conduct excitation waves (Kadota et al., 2013), and an excitation-contraction coupling mechanism that induces cardiac contraction (Kane et al., 2015) are shown to be inherent in hiPS-CMs, suggesting the great potential of hiPS-CMs for evaluating diverse functional cardiac toxicities, including proarrhythmic potential or contractile dysfunction.
In our previous research, we developed a method to measure the conduction, contraction, repolarization duration, and beating rate (BR) of hiPS-CMs in the same cells simultaneously using motion field imaging (MFI), which is a phase-contrast video imaging technology with high spatiotemporal resolution, able to analyze the motion of cardiomyocytes in a label-free and non-invasive manner (Isobe et al., 2018). We characterized the conduction and contraction properties of hiPS-CMs with MFI using typical cardiac functional modulators. However, to apply the MFI assay to the integrated assessment of cardiac functional toxicities, a comprehensive validation of the assay is needed. In particular, the repolarization of hiPS-CMs needs to be investigated with MFI to evaluate the risk of life-threatening arrhythmia associated with delayed ventricular repolarization.
Recent studies have revealed that reference compounds with potent or selective hERG inhibition prolonged the action potential duration (APD) or field potential duration (FPD) of hiPS-CMs and caused arrhythmic responses such as early after depolarization, which is consistent with clinical responses in humans (Blinova et al., 2017; Ando et al., 2017). However, some multichannel blockers shortened or had no effect on repolarization duration in hiPS-CMs though they did prolong QT interval in humans. These results indicate a limitation of hiPS-CMs, which suggests that the repolarization property needs further clarification.
Therefore, the purpose of this study is two-fold: to characterize the repolarization property of hiPS-CMs and to validate the MFI assay with hiPS-CMs as an integrated cardiac safety assessment model. For these purposes, we investigated the cardiac functional modulatory effects of internal compounds and reference compounds in both hiPS-CMs and in the ex-vivo whole hearts of guinea pigs perfused by the Langendorff method. The guinea-pig Langendorff heart is an established model for the integrated assessment of cardiac functions (Tabo et al., 2010; Guo et al., 2009). Hence we characterized the repolarization property of hiPS-CMs and validated the MFI assay by comparing the pharmacological responses in hiPS-CMs with those in the Langendorff heart.
Cryopreserved cardiomyocytes derived from human iPS cells (iCell® Cardiomyocytes2; Cellular Dynamics International, Madison, WI, USA) were obtained and prepared according to the manufacturer’s protocol. The cells were immediately thawed in the iCell Cardiomyocytes Plating Medium (Cellular Dynamics International) and 30 μL of cell suspensions were re-plated on a small area (approximately 3 mm in diameter) coated with fibronectin (Corning, New York, NY, USA) on a 12-well plates (Thermo Fisher Scientific, MA, USA) or on a probe with 64-microelectrode arrays (MED probe; MED-P545A, Alpha MED Scientific Inc., Osaka, Japan) at a density of 1.5 × 105 cells/30 μL/well. Each well was then filled with the iCell Cardiomyocytes Maintenance Medium (Cellular Dynamics International) at 2 mL. The cells were incubated at 37°C under an atmosphere of 5% CO2 for 5 to 25 days until recording, during which time the medium was replaced every 2 to 3 days. On the days of recording, the culture medium in the wells of a 12-well plate or in the MED probe was exchanged at a volume of 2 mL before starting experiments.
The 12-well plate was set into a mini-incubator (INUG2P-SI8, TOKAI HIT Co., Ltd., Shizuoka, Japan) attached to the stage of the video-microscope system, in which cells were maintained at 37°C under an atmosphere of 5% CO2. The cells were cultured in the mini-incubator for more than 30 min for stabilization. Video images of beating hiPS-CMs were recorded as sequential phase-contrast images with a 4 × objective at a frame rate of 150 frame/sec and a resolution of 2048 × 2048 pixels using the SI8000 cell motion imaging system (Sony Corporation, Tokyo, Japan). After obtaining a baseline value for each well under treatment of medium, vehicle (0.1%-0.5% dimethyl sulfoxide: DMSO) or test articles were applied from the lowest to the highest concentration in a cumulative manner with each concentration treated for 10-12 min. Analysis points were set at the end of the treatment of each concentration of vehicle or the test articles, where video images were recorded for 10 sec.
The motion of beating hiPS-CMs was analyzed with the SI8000 system using a block matching algorithm, the details of which are described in the previous report (Hayakawa et al., 2014). The following parameters were analyzed from the motion waveforms: maximum contraction speed (MCS), contraction-relaxation duration (CRD), and BR. CRD was defined as the interval between the start of the contraction to the end of the relaxation. The end of the relaxation was identified as the point where motion speed returned to baseline. CRD corrected by Fidericia’s formula (CRDc) was calculated as follows: CRDc = CRD/[Inter-beat interval (msec)/1000]1/3. In this formula, the inter-beat interval was calculated as the reciprocal of the BR.
The conduction speed of an excitation wave on the hiPS-CMs sheet was calculated by the SI8000 system. In the microscopic view, the field of the hiPS-CMs sheet (2.8 mm × 2.8 mm) is divided into 64 (8 × 8) square compartments of equal size. The time when a contraction wave is elicited in each compartment is measured. Based on the time, the conduction speed was determined by the following calculation: conduction speed (m/sec) = [distance between A and B (m)]/[elapsed time from C to D (sec)]. A (or B) represents the center point of a compartment on the cell sheet where an excitation wave was first (or last) elicited during a synchronous contraction. C (or D) represents the time when an excitation wave was elicited in the A (or B) compartment.
The MED probe was put in a stage top mini-incubator (MED, TOKAI HIT) to simultaneously record the motion and the field potential of hiPS-CMs using MFI and MEA, respectively. In the incubator, cells were maintained at 37°C under an atmosphere of 5% CO2. Recording of the video images by MFI and that of field potential by MEA were synchronized using the external triggering options of the MEA system. Analogue field potential signals were acquired through a 0.1-Hz high-pass filter and a 5-kHz low-pass filter at the sampling rate of 20 kHz using the MED64 MEA system (Alpha MED Scientific), and were analyzed by Mobius Software (Alpha MED Scientific) to measure the FPD.
In this study, the data of vehicle, E-4031, CH-a, CH-b, CH-c, and CH-d in hiPS-CMs were obtained by simultaneous recording of MEA and MFI. Vehicle or the test articles were applied from the lowest to the highest concentration in a cumulative manner with each concentration of vehicle or test article treated for 12 min, during which pacing stimulation was applied from 10-12 min. Analysis points of sinus data and pacing date were 10 min and 12 min after the treatment, respectively, with the duration of 10 sec. Stimulation pulses were biphasic and rectangular in shape (0.9 msec, 30 μA in amplitude), and were applied at the cycle length of 1000 msec. FPD and inter-spike interval were analyzed with Mobius Software (Alpha MED Scientific). FPD was corrected for BR with Fridericia’s formula (FPDc = FPD/[inter-spike interval (msec)/1000]1/3).
Male guinea pigs ([Slc: Hartley], Japan SLC, Inc., Shizuoka, Japan) weighing 500-800 g were used in this study. The animals were provided with food and water ad libitum, and were pair-housed or group-housed on wood-chip bedding in stainless steel cages (pair housing) or modified polyphenylene oxide cages (group housing). In each cage, the animals were provided with a dumbbell-shaped toy made of polypropylene as an enrichment. The room was HEPA-filtered and environmentally controlled with a temperature range of 23 ± 3°C, a relative humidity of 55% ± 20%, and a 12:12-hr dark/light cycle. Animals were acclimatized for at least 1 week before experiments started. All procedures of animal use were reviewed and approved by the Institutional Animal Care and Use Committee in Chugai Pharmaceutical Co., Ltd., which is an institute accredited by AAALAC International.
The experimental method was described previously (Tabo et al., 2010). Briefly, guinea pigs were pretreated with heparin (1000 unit/kg, i.p.) and anesthetized by pentobarbital (100 mg/kg, i.p.). With the animals restrained in a supine position, a tracheotomy was performed, and the animals were artificially ventilated with room air. Then the heart was quickly removed and its aorta was cannulated for perfusion of the coronary artery according to the Langendorff method with constant pressure (60-70 mmHg). Throughout the experiments, the hearts were perfused at a temperature of 37 ± 0.5°C and a flow rate of 15 ± 2 mL/min with Krebs-Henseleit (KH) solution containing (in mM) NaCl (120), KCl (4.7), Na-Pyruvate (2.0), CaCl2 (1.8), MgSO4 (1.2), KH2PO4 (1.2), NaHCO3 (25), and glucose (11.1), which was continuously aerated with a mixture of 95% O2 and 5% CO2. The negative electrocardiogram (ECG) electrode was connected to the perfusion cannula, and the positive electrode was softly attached to the apex of the left ventricle. A water-filled latex balloon was placed into the left ventricle cavity to measure left ventricular pressure (LVP). The heart was stabilized for more than 30 min before recording started.
Following stabilization, test articles were applied to the isolated heart from low to high concentration. Each concentration of the test articles was applied for 30 min, at the end of which the data of 10 consecutive waveforms were analyzed. Data from 3 to 5 animals were obtained for each test article. Vehicle (0.1% DMSO) was applied for 90 min to obtain time-matched control data. The ECG and LVP signals were amplified with an ECG amplifier (Nihon Kohden, Tokyo, Japan, AC-601G) and a carrier amplifier (Nihon Kohden, AP-601G), respectively. The rate of increase in LVP (LVdP/dt) was measured with a pressure processor (Nihon Kohden, EQ-601G). All the signals were digitized and recorded with a software (Win VAS3 version 1.1, Physio-Tech Co., Ltd., Tokyo, Japan) at a sampling rate of 1 kHz. The software was used to analyze a series of ECG parameters: QRS duration, QT interval, and heart rate (HR) as the reciprocal of RR interval. QT interval corrected by Fidericia’s formula (QTc) was calculated as follows: QTc = QT/(RR/1000)1/3. As a contractility parameter, the maximum rate of increase in LVP (LVdP/dtmax) was analyzed with the software.
The human ether-a-go-go related gene (hERG), Cav1.2, and Nav1.5 currents were measured by a whole-cell patch clamp technique using an automated patch-clamp system, Qpatch HTX (Sophion Bioscience A/S, Ballerup, Denmark) at room temperature. CHO cells expressing hERG and CHL cells expressing hNav1.5 were established at and obtained from F. Hoffman-La Roche, Ltd (Basel, Switzerland). CHO cells expressing hCav1.2 were purchased from ChanTest Corporation (Cleveland, OH, USA).
The extracellular solution for the recording of hERG and Nav1.5 currents contained (in mM): NaCl (150), KCl (4), CaCl2 (1.2), MgCl2 (1), HEPES (10), glucose (10), pH adjusted to 7.4 with NaOH. The extracellular solution for Cav1.2 contained (in mM): CsCl (87.5), TEA-Cl (40), CaCl2 (5), MgCl2 (1), HEPES (10), glucose (10), sucrose (45), pH adjusted to 7.3 with TEA-OH. The intracellular solution for hERG contained (in mM): KCl (140), EGTA (5), HEPES (10), MgCl2 (1), MgATP (5), pH adjusted to 7.2 with KOH. The intracellular solution for Cav1.2 and Nav1.5 contained (in mM): CsF (30), CsCl (112), EGTA (11), HEPES (11), MgATP (5), TrisGTP (0.4), pH adjusted to 7.2 with CsOH.
Regarding the voltage protocol, hERG currents were measured by a following voltage pattern: holding potential at −80 mV, baseline voltage step to −40 mV for 0.1 sec, prepulse voltage step to +20 mV for 1 sec, test pulse voltage step to −40 mV for 0.5 sec. The pulse pattern was repeated at an interval of 10 sec. Cav1.2 currents were measured using a voltage pattern of a test pulse to 0 mV for 0.2 sec from a holding potential of −80 mV, at a stimulation interval of 10 sec. Nav1.5 currents were measured using a voltage pattern of a test pulse to 0 mV for 0.2 sec from a holding potential of −120 mV, at a stimulation interval of 10 sec. Current signals of hERG, Cav1.2, and Nav1.5 were digitized at a sampling rate of 5 kHz, 10 kHz, and 25 kHz, respectively.
The test articles were sequentially applied to the cells from a low to a high concentration. The inhibition rates of each ion channel current was corrected by the vehicle control (0.1% DMSO) to obtain a % inhibition of the hERG, Cav1.2 or Nav1.5 current. The IC50 values of each ion channel were calculated by a logistic regression analysis using the % inhibition at the concentrations tested.
All the internal compounds were synthesized in Chugai Pharmaceutical Co., Ltd. Terfenadine and flecainide were purchased from Sigma-Aldrich (St. Louis, MO, USA). E-4031 and verapamil were purchased from Wako Pure Chemical Industries (Osaka, Japan). The test articles were dissolved in the DMSO to prepare stock solutions that were 1000-fold the final concentrations to be applied. As for the hiPS-CMs experiments, the stock solutions at a volume of 2 μL were applied to each well containing the medium at 2 mL. In the guinea-pig Langendorff assay and the automated patch-clamp assay, the stock solutions were diluted 1000-fold with the KH solutions and the extracellular solutions of each ion channel, respectively.
Statistical analysis was performed in the hiPS-CMs experiments and the Langendorff experiments. Data were represented as mean ± S.E., except for the baseline values of the parameters in hiPS-CMs and Langendorff heart which were represented as mean ± S.D. The F-test was used between the test article group and the vehicle control group to test for homogeneity of variance. Then, Student’s t-test and Welch’s test were applied when the variance between the two groups were homogenous and heterogeneous, respectively. Values of P < 0.05 were considered statistically significant. These statistics were calculated with JMP® (SAS Institute Inc., Cary, NC, USA).
For the validation of the MFI assay, the sensitivity, specificity, and accuracy of the assay were calculated as follows: Sensitivity (%) = B/A × 100. In this formula, A represents the number of compounds which caused an increase or decrease in the parameters of Langendorff heart. B represents the number of compounds which caused the same changes (increase or decrease) in the parameters of MFI as in the Langendorff heart. Specificity (%) = D/C × 100. In the formula, C and D represent the number of compounds which had no effect on the parameters of Langendorff heart and MFI, respectively. Accuracy (%) represents the proportion of the number of compounds which showed the same results in the MFI assay as in the Langendorff heart.
The pre-treatment baseline values of CRDc, BR, conduction speed, and MCS in hiPS-CMs were 0.56 ± 0.06 sec, 32.7 ± 5.0 bpm, 0.190 ± 0.040 m/sec, and 9.3 ± 0.9 μm/sec, respectively in this study (mean ± S.D., N = 67). The pre-treatment baseline values of QT interval, HR, QRS duration, and LVdP/dtmax in Langendorff heart were 190 ± 23 msec, 187 ± 23 bpm, 23 ± 4 msec, and 1173 ± 256 mmHg/sec, respectively in this study (mean ± S.D., N = 34-38). These baseline values were close to those in our previous study (Isobe et al., 2018).
To investigate the relationship of FPD and CRD, field potential and motion waveform of hiPS-CMs were simultaneously measured with MEA and MFI, respectively. Representative motion waveform and field potential waveform that are simultaneously measured using the same cells are shown in Fig. 1A. The correlations between FPD and CRD are shown in Fig. 1B, which represents that both parameters have a linear correlation with an R2 value of 0.92. Therefore, CRD can be used as a parameter to evaluate the drug-induced changes in the repolarization duration of hiPS-CMs.
Representative motion waveform (upper) and field potential waveform (lower) of a single beat of hiPS-CMs (A). Correlation of CRD and FPD was investigated with 188 data points derived from 26 preparations (B). Each data point represents pre-treatment baseline value or post-treatment values of each concentration of test articles or vehicle. All these data were obtained by simultaneous recording of MFI and MEA.
To characterize the repolarization property of hiPS-CMs, the effects of the internal and the reference compounds on CRDc were compared to those on the QTc interval in the Langendorff whole heart model. Representative results are shown in Fig. 2. CH-b caused dose-dependent prolongation in QTc in Langendorff heart and in CRDc of hiPS-CMs (Fig. 2A). CH-f caused dose-dependent shortening in QTc and CRDc (Fig. 2B). These results show that the pharmacological responses in the repolarization duration of hiPS-CM were consistent with those of Langendorff heart. As for CH-d, however, QTc was prolonged in Langendorff heart, while CRDc was shortened in hiPS-CMs, representing a discrepancy in repolarization between hiPS-CMs and Langendorff heart (Fig. 2C).
Effects of CH-b (A), CH-f (B), and CH-d (C) on CRDc of hiPS-CMs and QTc interval of guinea-pig Langendorff heart. Data represent the mean ± S.E. (N=4-5 for CRDc, N=3 for QTc). Statistically significant differences compared to time-matched vehicle control are represented with (*) for CRDc and (†) for QTc interval (p < 0.05). The arrows represent the prolongation (↑) and shortening (↓) of CRDc or QTc interval.
The effects of each compound on the repolarization of hiPS-CMs and Langendorff heart are summarized in Table 1 including reference data. Six compounds (E-4031, terfenadine, CH-a, flecainide, CH-b, and cilostazol) prolonged QTc interval in Langendorff heart and prolonged CRDc in hiPS-CMs, though terfenadine and flecainide shortened CRDc at their higher concentrations. Seven compounds (dronedarone, bepridil, CH-c, azithromycin, chlorpromazine, amiodarone, and CH-d) prolonged QTc interval in Langendorff heart or in humans, but shortened or had no effect on CRDc in hiPS-CMs. The other 10 compounds caused QTc shortening or had no effect on QTc in Landendorff heart, and caused CRDc shortening or had no effect on CRDc in hiPS-CMs. In total, 19 compounds caused prolongation or shortening of QT/QTc interval, among which 12 compounds showed the same response in the CRDc of hiPS-CMs (sensitivity: 63%). Four compounds had no effect on QTc interval, 2 of which showed the same response in hiPS-CMs (specificity: 50%). Hence the accuracy of CRDc of hiPS-CMs was 61%.
The criterion for the prolongation/shortening in CRDc and QTc was set to 10%. Although ΔCRDc in the CH-e group and ΔQTc in the CH-a, flecainide, verapamil, and CH-i groups were slightly less than 10%, these changes were considered to be physiologically significant because they were dose-dependent and/or statistically significant.
To investigate the reason for the discrepancy observed with the 6 compounds that caused QTc prolongation in Langendorff heart or in human but shortened or had no effect on CRDc in hiPS-CMs, we profiled IC50 values of cardiac ion channels, hERG, Cav1.2, and Nav1.5 for all compounds, and calculated the relative inhibitory potency of the two ion channels: Cav1.2 IC50 / hERG IC50 and Nav1.5 IC50 / hERG IC50. The results showed that the compounds with values of Cav1.2 IC50 / hERG IC50 between 1 and 10 prolonged QT/QTc in Langendorff heart or in humans, but shortened or had no effect on CRDc/FPDc in hiPS-CMs, with the exception of only a compound, cilostazol. Regarding the compounds whose values of Cav1.2 IC50 / hERG IC50 are below 1 or more than 10, pharmacological responses of the CRDc were consistent to those of the QTc, with the exception of only 2 compounds, CH-g and lidocaine. It seemed that Nav1.5 IC50/hERG IC50 values did not have any correlation to the pharmacological responses of CRDc or QTc.
The chronotropic effects of a series of compounds were investigated to evaluate the correlations between the hiPS-CMs and the Langendorff heart (Table 2). Nine compounds (E-4031, terfenadine, flecainide, lidocaine, diltiazem, isoproterenol, CH-b, CH-c, and CH-d) decreased or increased the HR of Langendorff heart, 5 of which (flecainide, lidocaine, isoproterenol, CH-c, and CH-d) showed consistent responses in the hiPS-CMs (sensitivity: 56%). Of note, terfenadine and diltiazem increased the BR of hiPS-CMs despite the decreases in the HR of Langendorff heart. Eight compounds (verapamil, carbenoxolone, CH-a, CH-e, CH-f, CH-g, CH-h, and CH-i) had no effect on the HR of Langendorff heart, 3 of which (carbenoxolone, CH-a, and CH-h) showed a consistent response in the hiPS-CMs (specificity: 38%). In total, the responses in the hiPS-CMs were consistent with those in the Langendorff heart in 8 out of 17 compounds (accuracy: 47%).
The criterion for the increase/decrease in BR and HR was set to 20%. Although ΔBR in the CH-c, CH-d, and CH-g groups were slightly less than 20%, these changes were considered to be physiologically significant because they were dose-dependent and statistically significant.
The effects of a series of compounds on the conduction speed in the hiPS-CMs were compared to those on the ventricular conduction of the Langendorff heart (Table 3). Six compounds (flecainide, lidocaine, carbenoxolone, isoproterenol, CH-a, and CH-d) decreased or increased the ventricular conduction speed in Langendorff heart, 5 of which (flecainide, lidocaine, carbenoxolone, isoproterenol, and CH-d) showed consistent responses in the hiPS-CMs (sensitivity: 83%). Eleven compounds (E-4031, terfenadine, verapamil, diltiazem, CH-b, CH-c, CH-e, CH-f, CH-g, CH-h, and CH-i) had no effect on the ventricular conduction of Langendorff heart, 6 of which (E-4031, verapamil, CH-b, CH-c, CH-e, and CH-h) showed a consistent responses in the hiPS-CMs (specificity: 55%). In total, the responses in the hiPS-CMs were consistent with those in the Langendorff heart in 11 out of 17 compounds (accuracy: 65%).
The criterion for the increase/decrease in the conduction speed of hiPS-CMs and Langendorff heart was set to 20%. Although Δconduction speed in hiPS-CMs slightly exceeded 20% in the verapamil group, the change was considered to be physiologically non-significant because it was varied and statistically non-significant.
The effects of a series of compounds on MCS of hiPS-CMs were compared to those on the LVdP/dtmax of Langendorff heart (Table 4). Fifteen compounds (terfenadine, flecainide, verapamil, lidocaine, diltiazem, carbenoxolone, isoproterenol, CH-a, CH-b, CH-d, CH-e, CH-f, CH-g, CH-h, and CH-i) decreased/increased the LVdP/dtmax of Langendorff heart, 12 of which (terfenadine, flecainide, verapamil, lidocaine, diltiazem, isoproterenol, CH-d, CH-e, CH-f, CH-g, CH-h, and CH-i) showed consistent responses in the hiPS-CMs (sensitivity: 80%). CH-c had no effect on the LVdP/dtmax of Langendorff heart and on the MCS of the hiPS-CMs (specificity: 100%). In total, the responses in the hiPS-CMs were consistent with those in the Langendorff heart in 13 out of 16 compounds (accuracy: 81%).
The criterion for the increase/decrease in MCS and LVdP/dtmax was set to 10%. Although ΔMCS in the CH-e and CH-h groups were slightly less than 10%, these changes were considered to be physiologically significant because they were dose-dependent and statistically significant.
In this study, we characterized a repolarization property of hiPS-CMs with MFI and validated the MFI assay as a model for evaluating multiple cardiac liabilities using hiPS-CMs by comparing the pharmacological effects of a series of internal or reference compounds on repolarization duration, BR, conduction speed, and contractility between hiPS-CMs and the guinea-pig Langendorff heart model. The results in hiPS-CMs were generally consistent with Langendorff heart, suggesting that the MFI assay with hiPS-CMs can be a useful model for evaluating multiple cardiac liabilities.
Several internal compounds and reference compounds caused shortening or no changes in CRDc of hiPS-CMs despite the QTc prolongation in Langendorff heart. The discrepancy between hiPS-CMs and clinical results were reported in previous studies, in which the compounds that prolong QT interval and have a known risk of causing Torsade de Pointes (TdP) in humans (bepridil, chlorpromazine, terfenadine, amiodarone, azithromycin, dronedarone, and cilostazol) either shortened or had no effect on the FPDc of the hiPS-CMs, iCell® (Blinova et al., 2017; Ando et al., 2017). Including the results of these compounds, the discrepancy in the repolarization duration in this study was accounted for by the relative inhibitory potency on the two cardiac ion channels, hERG and Cav1.2. In general, cardiac APD or QT interval is prolonged by an inhibition of hERG and shortened by an inhibition of Cav1.2, counter-balanced by these ion channels. The gene expression level of Cav1.2 (CACNA1C) in hiPS-CMs is higher than that of adult cardiomyocytes, while the expression levels of hERG (KCNH2) and Nav1.5 (SCN5A) are lower than those of adult cardiomyocytes (Blinova et al., 2017), which suggests a greater contribution of the Ca2+ current to the APD in hiPS-CMs compared to adult cardiomyocytes. Hence the repolarization duration of hiPS-CMs would be predisposed to be shortened by the inhibition of Cav1.2. In support of this view, terfenadine and flecainide prolonged CRDc at lower concentrations, but shortened CRDc at higher concentrations in this study. These two compounds inhibit Cav1.2 and hERG with Cav1.2 IC50 being higher than the hERG IC50, suggesting that the shortened CRDc of hiPS-CMs observed at high concentrations of terfenadine and flecainide could be due to the inhibition of Cav1.2.
We found that hiPS-CMs could not detect QTc prolongation in the Langendorff heart when the value of Cav1.2 IC50 / hERG IC50 was between 1 and 10, whereas hiPS-CMs showed almost consistent responses with QTc interval in Langendorff heart when the value of Cav1.2 IC50 / hERG IC50 was below 1 or above 10. The results clearly defined a limitation of hiPS-CMs, suggesting that results from a hiPS-CMs assay alone could underestimate the risk associated with QT prolongation. To complement the interpretation of the repolarization response in hiPS-CMs and correctly evaluate the QT risk, cardiac ion channel profiling, especially hERG and Cav1.2, would be necessary. Although we investigated possible involvement of Nav1.5-inhibiton in the discrepancy in repolarization between hiPS-CMs and Langendorff heart, we could not find any evidence for this in this study.
A limitation that should be mentioned is that the cardiac ion channel data in this study, including reference data, were all obtained at room temperature using a patch-clamp method, except the data for azithromycin, the Cav1.2 IC50 of dronedarone, and the hERG IC50 of lidocaine. Under physiological temperature, cardiac ion channels may show different gating kinetics and drug binding mode, as is the case with the hERG channel (Vandenberg et al., 2006). However, it has been demonstrated that there is not much difference between the IC50 values of hERG, Cav1.2, and Nav1.5 at room temperature or at physiological temperature with regard to 16 reference compounds (Crumb et al., 2016). Room temperature makes it easier and more stable for recording the ion channel current by the patch-clamp method than physiological temperature, allowing us to achieve a high throughput using an automated patch-clamp system, which is an advantage for a screening assay at the drug discovery stage. Hence in this study we obtained the ion channel data at room temperature.
The sensitivity or specificity of hiPS-CMs on the BR was not high. A major reason for this was that, unlike in the heart, several compounds increased the BR of hiPS-CMs, and this was probably because the BR is increased by inhibition of the Ca2+ channel, which is a property specific to the hiPS-CMs (Kitaguchi et al., 2016; Lu et al., 2015; Guo et al., 2011). A recent study showed that Ca2+ channel blockers decreased the spontaneous BR of hiPS-CMs when the available Na+ current was reduced by a selective Na+ channel blocker (TTX) or by elevating membrane potential, which indicates that the Na+ current drives the action potential of hiPS-CMs, thereby regulating the BR (Zeng et al., 2019). Consistent with this, several internal compounds that inhibit Nav1.5 decreased or did not increase the BR of hiPS-CMs, even if they inhibited Cav1.2 in this study. Since the mechanism that regulates the spontaneous beating of hiPS-CMs is different from that of the heart, it can be difficult to evaluate the chronotropic effects of compounds simply based on the hiPS-CMs BR. However, understanding the inhibitory properties of compounds on Cav1.2 and Nav1.5 would be helpful to interpret responses in the BR of hiPS-CMs.
Regarding the conduction speed, the hiPS-CMs was highly sensitive in detecting the widening and shortening of QRS duration in the Langendorff heart. An internal compound, CH-d, caused QRS widening in Langendorff heart at 3 μM and above. Since the Nav1.5 IC50 of CH-d was 2.5 μM, the ventricular conduction delay would be caused by the inhibition of Nav1.5. Hence the conduction speed in hiPS-CMs would also be decreased by the inhibition of Nav1.5. Moreover, typical Na+ channel blockers, flecainide and lidocaine, elicited QRS widening in the Langendorff heart and decreased conduction speed in hiPS-CMs. These results suggest that the conduction of hiPS-CMs is dependent on Na+ current. Therefore, the hiPS-CMs would be useful for evaluating the risk associated with ventricular conduction delay caused by inhibition of the cardiac Na+ current. Terfenadine, diltiazem, CH-f, CH-g, and CH-i decreased conduction speed in hiPS-CMs, but did not cause a significant change in the QRS duration in Langendorff heart, which led to the insufficient specificity of hiPS-CMs. Terfenadine and diltiazem had inhibitory effect on Nav1.5 with an IC50 of 2.0 μM and 22.4 μM, respectively (Kramer et al., 2013). The Nav1.5 IC50 of CH-f and CH-g were above 10 μM, but the IC20 was 8.8 μM and 3.4 μM, respectively. These results demonstrate that these compounds have the potential to inhibit the Na+ current, thereby possibly decreasing the conduction speed in hiPS-CMs. The reason why the QRS duration was not altered in Langendorff heart is unclear, but it may be because hiPS-CMs might be highly sensitive to conduction delay due to the inhibition of Na+ current.
The hiPS-CMs showed high accuracy in contractility. Most of the internal compounds decreased LVdP/dtmax in Langendorff heart at 0.3-10 μM and the MCS of hiPS-CMs at 0.3-30 μM, and had inhibitory effects on Cav1.2 and/or Nav1.5, with IC50 values ranging between 1-10 μM. All these results suggest that the decreased contractility in Langendorff heart and in hiPS-CMs was due to the inhibition of Cav1.2 and/or Nav1.5. Hence it was confirmed that hiPS-CMs is highly sensitive in detecting a decrease in contractility caused by an inhibition of the cardiac Ca2+ and Na+ channels, and would therefore be useful for the assessment of cardiac contractility.
For cardiac safety assessment, the guinea-pig Langendorff heart has proven to be a useful model for evaluating multiple cardiac liabilities including the QT risk (Hamlin et al., 2004; Guo et al., 2009). The potential limitation of the Langendorff heart model is associated with species differences; the cardiac transient outward potassium current (Ito) in humans is absent in guinea pigs, and that the slow component of the delayed rectifier potassium current (IKs) is much more highly expressed in guinea pigs than in humans (Zicha et al., 2003). Although hiPS-CMs is expected to reflect human cardiac physiology and overcome species difference, most of the currently available hiPS-CMs have immature functional and structural properties, such as the relatively depolarized resting membrane potential and the absence of the t-tubule (Honda et al., 2011; Gherghiceanu et al., 2011). These lines of evidence suggest that the current hiPS-CMs are not necessarily more translational to humans or more predictive of clinical responses than the Langendorff heart. Instead, the main advantage of the hiPS-CMs analyzed by the MFI is that it enables us to evaluate multiple cardiac safety liabilities in the very early stages of drug development with much higher through-put than the Langendorff heart. Since the characteristics of the MFI assay are dependent on the cardiomyocytes under analysis, increasing the availability of human-relevant cardiomyocytes with mature properties is important to further improve the clinical predictability of the MFI assay.
In summary, this study revealed an important repolarization property of hiPS-CMs. A few previous studies demonstrated that hiPS-CMs could not detect the QT prolongation and proarrhythmic potential of compounds that have a multichannel blocking profile (Blinova et al., 2017; Ando et al., 2017). Most of the previous studies investigated the repolarization of hiPS-CMs using reference compounds that generally have high selectivity for a specific cardiac ion channel or show the clear-cut inhibition of cardiac ion channels, thus the limitations of hiPS-CMs repolarization observed with multichannel blockers has been difficult to investigate. In this study, by using internal compounds with the multichannel blocking profile of the three main cardiac ion channels, we clearly defined the limitation of hiPS-CMs. Based on the results validating comprehensive cardiac functions, we think that MFI assay with hiPS-CMs can be a useful model for evaluating multiple cardiac liabilities using hiPS-CMs. Several of the compounds induced different repolarization durations and BR in the hiPS-CMs and Langendorff heart, but these could be explained by the cardiac ion channel profile of the compounds. Therefore, the combination of an MFI assay and ion channel profiling would be recommended for an integrated cardiac safety assessment in the early stages of drug development.
The authors declare that there is no conflict of interest.