Neonatal atrial and ventricular rat cardiomyocytes were isolated with the neonatal rat cardiomyocyte isolation kit (Worthington, New Jersey, USA) (Toraason et al., 1989; Macgregor et al., 1995) and cultured at 37°C with 5% CO₂. Briefly, heart atria and ventricles were dissected from 1 day old Hsd:Sprague–Dawley rats, then digested overnight at 4°C with trypsin. Digestion continued 12 h later with collagenase for approximately 90 min at 37°C. Isolated cells were pre-plated for 120 min on uncoated cell culture dishes to remove fibroblasts, and subsequently, atrial and ventricular cardiomyocytes were either pooled or separately seeded on Matrigel-coated 96-Well Glass-Bottom Plates (SensoPlate™, Greiner Bio-One, North Carolina, USA) in high-serum media [DMEM/F12 (1:1), 0.2% bovine serum albumin, 3 mM sodium-pyruvate, 0.1 mM ascorbic acid, 4 mg l21 transferrin, 2 mM L-glutamine, 100 nM thyroid hormone (T3) supplemented with 10% horse serum and 5% fetal calf serum (FCS)] at a density of 4 × 10⁵ cells/well. After 24 h, media was changed to low-serum medium (the same as above but substituting 0.25% FCS for 10% horse serum and 5% FCS). Seventy two hours after seeding the cells, they were processed for voltage imaging. All animal handling and care followed the NIH Guide for Care and Use of Laboratory Animals. The experimental protocols were approved by Institutional Animal Care and Use Committees of Sanford-Burnham-Prebys Medical Discovery Institute and Stanford University.

Vials of both MyCell (LQTS3) and iCell (healthy individual) Cardiomyocytes (Cellular Dynamics International, Wisconsin, USA) were thawed according to the manufacturer's specifications (see Supplementary Table 1 for lot information). A single cryovial was removed from liquid nitrogen and placed on dry ice for no more than 3 min. The vial was transferred to a 37°C water bath and incubated for 4 min. Once completely thawed, the contents were slowly transferred to a 50 mL conical vial using a 1,000 μL pipette. 1 mL of iCell Cardiomyocyte Plating Media (iCCPM) was used to rinse the inside of the cryovial to gather any remaining cells. After rinsing the vial, the 1 mL of iCCPM was added dropwise to the 50 mL conical vial over a period of 2–3 min with continuous agitation for each drop. The additional volume for dilution was calculated to achieve a final concentration of 2.5 × 10⁵ cells/mL and depended on the specifications of each lot of cells. The additional media was added dropwise to the 50 mL vial. Once diluted, the cells were transferred into a reagent reservoir and 20 μL was added to each well with a 16 channel pipette (Integra Viaflo II) at a flow rate of 35 μL/s with 1 mix cycle prior to each addition in a pre-coated 384 well plate (Greiner Bio-One) with 0.1% (w/v) gelatin (Stem Cell Technologies). The plates were left standing at room temperature for 20 min to allow the cells to settle to the bottom of the plate before it was placed in a 37°C 5% CO₂ incubator. After 24 h, the plating media was diluted by adding 80 μL of iCell Cardiomyocyte Maintenance Media (iCCMM), supplemented with 5 mM D-glucose, and subsequently 50 μL of iCCMM was removed are replaced with 50 μL of iCCMM. This process was repeated two additional times, ultimately yielding a final volume of 100 μL/well. The plates were placed back in the 37°C 5% CO₂ incubator for 48 h. Media was exchanged every other day for 14 days prior to imaging by removing 50 μL of media and adding 50 μL of fresh iCCMM.

Briefly, hiPSCs (hiPSC line SCVI15; see Supplementary Table 1) were dissociated using 0.5 mM EDTA (ThermoFisher Scientific) in PBS without CaCl₂ or MgCl₂ (Corning) for 7 min at room temperature. The dissociated hiPSCs were plated at a density of 3 × 10⁵ cells per well of a Matrigel-coated 12 well plate in mTeSR1 media (StemCell Technologies) supplemented with 2 μM Thiazovivin (Selleck Chemicals). After 24 h, the media was replaced with mTESR1 without Thiazovivin and was replenished daily for 3–5 days until the cells reached ≥90% confluence to begin differentiation. Cardiomyocytes were differentiated by methods previously described (Burridge et al., 2014; Cunningham et al., 2017). At day 25, cells were dissociated and plated onto Matrigel-coated 384-well tissue culture plates (Greiner Bio-One) at a density of 20,000 cells/well. All experiments were performed day 28 post-differentiation.

VF2.1.Cl (the dye used in this study was synthesized by EWM, but is marketed as Fluovolt) dye loading solution and the compound dilutions were prepared prior to manipulation of the cells on the day of imaging. 1 μL of 2 mM VF2.1.Cl in DMSO was mixed with 1 μL of 10% Pluronic F127 (diluted in water) in a 1.7 mL tube by agitating and centrifuging 3 times. Mixing of VF2.1.Cl with Pluronic F127 prior to subsequent dilution in physiological buffer is critical for VF2.1.Cl loading. Separately, Hoechst 33258 was diluted into Tyrode's solution (136 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, 10 mM HEPES, pH 7.4) to a concentration of 4 μg/mL. 1 mL of the Hoechst/Tyrode's solution was added to the 1.7 mL tube containing the VF2.1.Cl/Pluronic F127 mixture and vortexed for 10 s. The contents of the 1.7 mL tube was added to 4 mL of Hoechst/Tyrode's solution. The mixture (5 mL of a 2x concentrated VF2.1.Cl dye loading solution) can be scaled to larger volumes if reagent ratios are maintained. Reference compounds isoproterenol (Tocris, Bristol, UK), propranolol (Tocris, Bristol, UK), verapamil (Tocris, Bristol, UK), tetrodotoxin (Tocris, Bristol, UK), ATX-II (Sigma Aldrich, Missouri, USA), mexiletine (Toronto Research Chemical, Toronto, Canada), ranolazine (Tocris, Bristol, UK), acetylcholine (Sigma Aldrich, Missouri, USA) and sotalol (Tocris, Bristol, UK) were diluted in water and formoterol (Tocris, Bristol, UK), isradipine (Tocris, Bristol, UK), BayK 8644 (Tocris, Bristol, UK), veratridine (Tocris, Bristol, UK), E-4031 (Tocris, Bristol, UK), and dofetilide (Tocris, Bristol, UK) were diluted in DMSO. Each compound was diluted in Tyrode's solution to a 2x concentrated stock and was warmed to 37°C using a dry heat block prior to application to the cells.

hiPSC-CMs were removed from a 37°C 5% CO₂ incubator and placed immediately into in a tissue culture cabinet on a dry heat block set to 37°C to prevent temperature fluctuation during the subsequent washing and dye loading steps. All subsequent manipulations were conducted on a 37°C dry heat block. Tyrode's solution was warmed to 37°C in a water bath. To wash the cells and remove the media, 50 μL of media was removed and replaced with 50 μL of Tyrode's solution five times using a 16 channel pipette at flow rate of 35 μL/s. After the fifth wash, 50 μL of the 2x VF2.1.Cl dye loading solution was added to each well. The plate was placed back in the 37°C 5% CO₂ incubator for 50 min. After the 50 min incubation, the cells were rinsed 4 more times by removing 50 μL of Tyrode/dye solution from the well and adding 50 μL of fresh Tyrode's solution to the well and repeating. After rinsing, the cells were placed back in a 37°C 5% CO₂ incubator for 10 min to recover. After recovery, 50 μL of solution was removed and 50 μL of 2x reference compound was added to the respective well and incubated at 37°C and 5% CO₂ for 5 min before image acquisition. Time series images were acquired automatically using the IC200 KIC instrument (Vala Sciences, California, USA) at an acquisition frequency of 100 Hz for a duration of 6.5 s or 33 Hz for 20 s, as indicated in the figure legends, with excitation wavelength of 485/20 nm and emission filter 525/30 nm using a 0.75 NA 20x Nikon Apo VC objective. A single image of the Hoechst was acquired after the time series. The optimized dye loading and imaging conditions were replicated using both a different high content imager, the ImageXpress Micro XLS platform (Molecular Devices), and hiPSC-CMs produced by a novel differentiation protocol (Cunningham et al., 2017; and Supplementary Figure 3).

The image analysis and physiological parameter calculation was conducted using commercially available Cyteseer (Vala Sciences) as previously described (Cerignoli et al., 2012; Lu et al., 2015). Briefly, the image output by Cyteseer Scanner were loaded into Cyteseer and the whole well cardiac time series algorithm was executed on the image files. The metrics analyzed by the whole well cardiac time series algorithm were beat rate, normalized area under the peak trace (normalized peak integral), and APD₂₅, APD₅₀, APD₇₅, and APD₉₀ (referring to the width of the cardiac AP at a point 25, 50, 75, and 90% respectively), from the peak to baseline amplitude. EADs were quantified automatically using the default noise threshold in the whole well cardiac time series algorithm. The algorithm determines EADs as peaks following a local minimum at a user-specified threshold above the diastolic interval minimum. Data tables were analyzed using Microsoft Excel 2013 and dose response curves and heat maps were calculated using GraphPad Prism 7 software.

After live imaging using VF2.1.Cl, the neonatal CMs were fixed using 4% paraformaldehyde. After fixation, the cells were incubated with a blocking solution [10% horse serum, 0.5% Triton X-100, 0.01% gelatin in phosphate buffered saline (PBS)] for 30 min at room temperature followed by incubation with anti-Myl2 (Proteintech, cat: 10906-1-AP) or anti-Myl7 (Synaptic Systems, cat: 311 011) for 1 h at room temperature. Following incubation with the primary antibody, the cells were washed 5 times with PBS before incubation with anti-mouse Alexa Fluor 568 (ThermoFisher Scientific, A-11057) or anti-rabbit Alexa Fluor 647 (ThermoFisher Scientific, A-31573) for 1 h at room temperature. After incubation with the secondary antibody, the cells were washed 5 times with PBS and stored in a 50% glycerol (v/v) solution in PBS for subsequent imaging using the KIC.

The single cell analysis was performed manually using ImageJ. Regions of interest (ROIs) were selected using the freehand selection tool and exported as .roi files. Using nuclear labeling from the live and fixed images, offsets for the image sets were manually determined. The ROIs were then applied to the time series images and used to generate fluorescent intensity values for each ROI in all images. The data was analyzed using Microsoft Excel 2013 and plotted using GraphPad Prism 7.

An R-script (R version 3.3.1) was developed that reads the CSV files generated by Cyteseer and amalgamates all the physiological parameters of interest into a single file. Using this dataset, we performed a meta-analysis of all experiments and determined the intra and inter thaw variability in APD₇₅ as well as the relationship between APD₇₅ and the beat rate. As a result of the non-linear relationship of APD₇₅ and the beat rate, statistical significance was determined using a sliding window (center value ± 3) to mirror experimental ranges and applying the Mann-Whitney test (using the Bonferroni criterion for multiple testing).

Cardiac ion currents were recorded from single cardiomyocytes using the whole-cell patch-clamp method. Briefly, coverslips with cardiomyocytes were transferred into electrophysiological perfused recording chamber (RC-25-F, Warner Instruments, Hamden, CT) mounted on the stage of an inverted Olympus microscope. Patch pipettes were pulled from thin-wall borosilicate glass capillaries (CORNING 7740, 1.65 mm) with a P-2000 laser pipette puller (Sutter Instruments, California, USA) and had electrode tip resistances between 1.5 and 5.5 MΩ with access resistance of < 8 MΩ for whole-cell patch recordings. Series resistance and cell capacitance were compensated to between 30 and 60% in some voltage-clamp recordings. Patch electrodes were filled with intracellular solution containing: 120 mM CsCl, 20 mM tetraethylammonium chloride (TEA-Cl), 10 mM Hepes, 2.25 mM EGTA, 1 mM CaCl₂, 2 mM MgCl₂, pH 7.4. All recordings were performed at room temperature in Tyrode's solution. Current response traces were acquired using the Axon 200B amplifier. Currents were digitally sampled at 10 kHz using Digidata 1322A digitizer hardware and pClamp 10.2 software (Molecular Devices, California, USA). Current responses were filtered using 8-pole Bessel analog low-pass filter at 1-2 kHz cutoff frequency. In order to record the late sodium component, the pipette-cell seal stability and access resistance was carefully monitored during the recording. Current amplitude at the end of the depolarizing step was corrected by the leak current and the corrected current amplitude was averaged for 10 cells. After the leak-correction, we computed the average value of the recorded current amplitude over the last 50 ms of the 150 ms depolarizing pulse of each 250 ms sweep and averaged the resulting value over 3 consecutive sweeps.

To create the automated platform, we developed methods for high throughput acquisition of high speed/high resolution (5.5 megapixel at 100 Hz) time series images as well as downstream image analysis to calculate physiological parameters that describe the AP kinetics of CMs in electrical syncytium (Figure 1A). To measure changes in membrane potential, the platform utilized VF2.1.Cl (Miller et al., 2012), which intercalates into the plasma membrane and senses changes in transmembrane potential by a novel photo-induced electron transfer (PeT) mechanism (Miller, 2016; VF2.1.Cl and related VSPs have been described in detail Woodford et al., 2015; Deal et al., 2016; Kulkarni et al., 2017, and can be obtained directly from EWM). VF2.1.Cl and structurally related probes generate a larger change in fluorescence in response to changes in voltage than conventional electrochromic dyes while maintaining sufficient time resolution to measure cardiomyocyte APs (Miller et al., 2012; Supplementary Figure 1A). We established a robust protocol to standardize cell culture conditions that consistently achieved synchronously contracting monolayers of hiPSC-CMs as well as dye loading and imaging conditions that produced optical traces with sufficient signal-to-noise and reproducibility (Supplementary Movie 1). 200 nM VF2.1.Cl and ~120 mW/cm² illumination intensity (excitation: 485/20 nm, emission: 525/30 nm) enabled us to achieve excellent signal-to-noise, while maintaining reproducibility (Supplementary Figure 1B). Excessive illumination or VF2.1.Cl concentration caused variability in the AP response manifested by a prolongation of the APD within successive peaks from a single recording of an individual well (Supplementary Figures 1C–K). This observation is consistent with previous reports for the VSP, di-4-ANEPPS (Schaffer et al., 1994). Thus, our optimization provided conditions for statistically robust analysis of synchronously contracting monolayers of hiPSC-CMs for high throughput screening applications.

We began development of our AP waveform analysis procedures using bona fide ventricular and atrial cardiomyocytes isolated from the neonatal rat heart (Figures 1B–H). Purified atrial and ventricular cardiomyocyte populations, cultured as an electrically coupled monolayer, exhibited characteristic AP morphologies using a whole well analysis (Figure 1C). However, chamber specific morphologies could not be discerned when atrial and ventricular cardiomyocytes were co-cultured in a 1:1 ratio, either at the whole well analysis, or by analyzing the waveforms of individual Myl2⁺ (ventricular) or Myl7⁺ (atrial) cardiomyocytes identified retrospectively (see Supplementary Figure 1L and Methods for overlaying waveform and immunostaining). Instead, the individual waveforms and whole well waveforms were indistinguishable (Figure 1D). In addition, no difference in AP morphology was observed in the whole well vs. individual cell waveforms of the hiPSC-CM cultures (Supplementary Figures 1M,N). These results show that AP morphology acquired from the whole well ROI represents that of all individual CMs, regardless of subtype identity, in a synchronously contracting monolayer. This result suggests that an effect on one cell type (i.e., atrial) might be masked by an opposing effect on another (i.e., ventricular) in an electrical syncytium of multiple cell types. However, whole well analysis greatly reduced the computational workload when compared to single cell analysis and enabled the rapid generation of physiological parameters. Using this approach, we could visualize characteristic atrial specific drug effects in pure populations of neonatal rat atrial and ventricular cardiomyocytes. Acetylcholine, a selective agonist of I_Ach shortens the APD in atrial but not ventricular myocytes (Wang et al., 1993; Koumi et al., 1994; Devalla et al., 2015) whereas E-4031, a selective blocker of I_Kr mediated by hERG, prolonged APD in ventricular CMs but not atrial CMs (Figures 1E–I). Atrial CMs utilize IK_ur mediated by K_v1.5 in addition to IK_r for repolarization, and are less sensitive hERG block (Roy et al., 1996). Thus, high throughput optical recording of VF2.1.Cl and whole well analysis was sufficient to distinguish chamber specific drug effects.

The ability of hiPSC-CMs to recapitulate drug effects observed in adult CMs is critical for their use in both disease modeling and proarrhythmia evaluation. Therefore, we evaluated the expression and functionality of ion channels and adrenergic receptors with a panel of reference compounds (Figure 2A). To calculate EC₅₀ values, the cells were acutely treated with each compound (n = 3 wells per dose) and the average whole well fluorescent intensity was plotted and the waveforms were used to calculate the APD₇₅ for each well. Mexiletine and ranolazine are known to have an off-target inhibition of hERG, and indeed, the compounds caused a dose dependent APD prolongation and EADs at doses similar to those reported (EC₅₀ of 41.1 μM and 3.68 μM, respectively) (Mitcheson and Hancox, 1997; Terrenoire et al., 2013; Du et al., 2014; Gualdani et al., 2015; Figures 2B–E, Supplementary Figures 2A,B). The non-selective β-adrenergic agonist, isoproterenol, as well as the β2-selective agonist, formoterol, both induced positive chronotropic responses at expected doses (EC₅₀ of 18 nM and ~2.6 nM) (Wills et al., 2012; Sirenko et al., 2013; Figures 2F–I, Supplementary Figures 2C,D). The non-selective β-adrenergic antagonist, propranolol, prolonged the APD as well as slowed the beat rate considerably also as previously reported (EC₅₀ of 51.7 μM) (Sirenko et al., 2013; Figures 2J,K, Supplementary Figure 2E). The L-type calcium channel inhibitors verapamil and isradipine shortened the APD as expected and increased the beat rate (EC₅₀ of ~8.06 μM and ~232 nM) (Figures 2l–O, Supplementary Figures 2F,G) while the L-type calcium channel agonist, BayK 8644, significantly prolonged the APD (EC₅₀ of ~68.5nM) (Figures 2P,Q, Supplementary Figure 2H). To evaluate the presence and functionality of sodium channels in the hiPSC-CMs, the cells were treated with specific sodium channel inhibitor, tetrodotoxin (TTX), and two sodium channel agonists, anemonia sulcata toxin (ATX-II) and veratridine, which bind to different sites on the sodium channel and open the channel by slightly different mechanisms (Stevens et al., 2011). TTX caused a cessation of beating (EC₅₀ of 4.96 μM) at doses comparable to the IC₅₀ for peak sodium current (I_NaP) inhibition in CMs (>1 μM) indicating a reliance on sodium channel currents for the generation of APs (Figures 2R,S, Supplementary Figure 2I). However, dependence of the AP on sodium current varied across batches of CMs, suggesting inconsistency in the maturity of the hiPSC-CMs. The two sodium channel agonists prolonged the APD without inducing EADs (426 nM for veratridine and 198 nM for ATX-II) (Figures 2T–W, Supplementary Figures 2J,K), possibly reflecting either the maturation state of the hiPSC-CMs or genetic variation in the hiPSCs from different individuals.

To assess cross-platform reproducibility, we tested TTX, isoproterenol and E-4031 using Molecular Devices ImageXpress Micro XLS (Supplementary Figure 3, Supplementary Movies 2, 3) using iPSC-CMs prepared in-house (Cunningham et al., 2017), revealing similar dose responsiveness (EC₅₀ for isoproterenol 77 nM, EC₅₀ for TTX 4.85 μM, EC₅₀ for E-4031 >13.1 nM, EC₅₀ for ATX-II 1.03 μM). In summary, optical recording of hiPSC-CM AP waveform kinetics in response to reference compounds reveals the expected functional expression of cardiac ion channels (see Supplementary Table 2).

The appearance of EADs and voltage alternans are the cellular manifestations of T-wave prolongation and PVCs seen on the ECG that predispose patients to VT and TdP. Therefore, we analyzed the incidence of EADs in hiPSC-CMs treated with inhibitors of I_Kr and an agonist of the late sodium current (I_NaL). The development of these morphologies can be better assessed over longer periods of recording (Mohammad et al., 1997; Song et al., 2008). Thus, time series images were acquired for 20 s at 33 Hz to provide improved signal to noise as well as recording duration, and coupled with an automated image analysis to determine the effect of the hERG blockers dofetilide, E-4031 and sotalol. All three molecules caused extensive dose-dependent prolongation of APD, induction of EADs and, at some doses, voltage alternans (Figures 3A–C, see also Supplementary Figure 3). Representative movie clips for dofetilide illustrate the presence of EADs as compared to untreated hiPSC-CMs (Supplementary Movies 4, 5), and treatment with 33 and 100 nM dofetilide caused rapidly firing subthreshold APs that did not induce contraction (Supplementary Movies 6, 7). The appearance of these AP morphologies seemed analogous to tachyarrhythmia or fibrillation. To determine an EC₅₀ dose for prolongation of the APD, dose response curves were generated from the APD₇₅ values (Figures 3E–H, see also Supplementary Figure 3) in addition to the automatic detection of EADs (Figure 3I). Each of the hERG inhibitors induced significant APD prolongation as well as induced EADs at expected EC₅₀ values (2.79 nM for dofetilide, 59.0 nM for E-4031 and ~80 μM for sotalol). Interestingly, EADs were not detected in the ATX-II condition in the CDI hiPSC-CMs, although the compound prolonged the APD more than 7-fold (EC₅₀ of 438.5 nM), which was comparable to the extent of prolongation by dofetilide and sotalol. However, we have noted EADs in other hiPSC-CM populations, most likely reflecting cell line or differentiation protocol-specific differences (Supplementary Figure 3D). In conclusion, optical recording visualized drug induced prolongation of the APD, induction of EADs as well as tachyarrhythmia in vitro.

Congenital and pharmacological hiPSC-CM models of arrhythmia are of enormous interest as tools to decipher arrhythmogenic mechanisms. Moreover, an individual's genetic makeup (particularly polymorphisms in voltage-gated ion channels) can predispose patients to drug-induced arrhythmia and might underlie the sporadic nature of this phenomenon in the population (Yang et al., 2002). Therefore, visualizing drug-induced arrhythmia in hiPSC-CM models from healthy and diseased individuals is potentially a powerful new tool to personalize treatment as well as to predict liability of drug candidates across “high risk” individuals. To test this hypothesis, we evaluated a previously characterized LQTS3 patient harboring a F1473C missense mutation in the cardiac sodium channel, Na_v1.5 (Terrenoire et al., 2013). Whole cell patch clamp recording of hiPSC-CMs derived from the LQTS3 patient confirmed increased I_NaL (I_NaL was 1.2% ± 0.604 of I_NaP in the LQTS3 hiPSC-CMs, n = 10, and no detectable late current in the normal hiPSC-CMs) (Figure 4A, Supplementary Movie 8), which is a hallmark of LQTS3, indicating that the cells were a bona fide model of the congenital channelopathy. We used the high throughput platform to evaluate the reproducibility of APD across multiple preparations of LQTS3 hiPSC-CMs. Intra-experimental APDs of spontaneously beating hiPSC-CMs were very consistent with an average coefficient of variation (CV) = 0.069 ± 0.019 for untreated wells across multiple plates from the same thaw. The inter-experimental average APD from vials thawed on different days had more variability (CV = 0.098) (Figure 4B). We plotted the average APD for every untreated, control well across all experiments and a statistically significant difference in APD between the normal and LQTS3 hiPSC-CMs was discernible only above a spontaneous beat rate of 37 beats per minute (BPM) based on the Bonferroni corrected p < 0.05 (Figure 4C) which is consistent with prior studies showing a phenotypic variability of the LQTS3 phenotype (Ma et al., 2013; Paci et al., in press). Nonetheless, when probed pharmacologically with mexiletine, a sodium channel inhibitor with selectivity for I_NaL over peak sodium current (I_NaP) (Wang et al., 1997), LQTS3 hiPSC-CMs showed a stereotypic, dose-dependent AP shortening, effectively reversing the LQTS3 phenotype (EC₅₀ = 1.65 μM) (Figures 4D,E). This reversion was highly consistent across 4 different thaws of LQTS3 hiPSC-CMs spanning 38–53 BPM, as seen when baseline APD is normalized for each dose (average EC₅₀ = 2.67 μM and average E_max of 0.88) indicating a robust reversion of the disease phenotype that was independent of the variation in APD and beat rate (Figure 4F).

Pharmacological models of LQTS are commonly used to probe drug effects on APD and arrhythmia (Shimizu and Antzelevitch, 1999). LQTS3 can be modeled pharmacologically by treatment with ATX-II, which phenocopies the disease by specifically inhibiting fast inactivation of cardiac sodium channels. ATX-II-induced APD prolongation can be effectively reverted by 2.5 μM mexiletine (Figures 5A,D) and another I_NaL selective inhibitor ranolazine (3.7 μM) (Figures 5B,E). Finally, we created a pharmacological model of LQTS2 using the I_Kr inhibitor, dofetilide. APD prolongation caused by 3 nM dofetilide was effectively reverted by 185 nM isradipine, an L-type calcium channel inhibitor (Figures 5C,F) (as expected because blocking the inward calcium current counteracts the block to the outward potassium current). Thus, the automated platform facilitates the characterization of diseased hiPSC-CMs and pharmacological models of disease in normal cells, and in both cases, can detect the therapeutic effect of small molecule drugs.

JP is a founder and MM is on the scientific advisory board of Vala Sciences, and both have equity in the company, which manufactures a high content instrument used in these studies. The other author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer SC and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.