Optical Recording of Action Potentials in Human Induced Pluripotent Stem Cell-Derived Cardiac Single Cells and Monolayers Generated from Long QT Syndrome Type 1 Patients

Induced pluripotent stem cells (iPSCs) from type 1 long QT (LQT1) patients can differentiate into cardiomyocytes (CMs) including ventricular cells to recapitulate the disease phenotype. Although optical recordings using membrane potential dyes to monitor action potentials (APs) were reported, no study has investigated the disease phenotypes of cardiac channelopathy in association with the cardiac subtype at the single-cell level. We induced iPSC-CMs from three control and three LQT1 patients. Single-cell analysis using a fast-responding dye confirmed that ventricular cells were the dominant subtype (control-iPSC-CMs: 98%, 88%, 91%; LQT1-iPSC-CMs: 95%, 79%, 92%). In addition, LQT1-iPSC-ventricular cells displayed an increased frequency of early afterdepolarizations (pvalue = 0.031). Cardiomyocyte monolayers constituted mostly of ventricular cells derived from LQT1-iPSCs showed prolonged AP duration (APD) (pvalue = 0.000096). High-throughput assays using cardiomyocyte monolayers in 96-well plates demonstrated that IKr inhibitors prolonged APDs in both control- and LQT1-iPSC-CM monolayers. We confirmed that the optical recordings of APs in single cells and monolayers derived from control- and LQT1-iPSC-CMs can be used to assess arrhythmogenicity, supporting the feasibility of membrane potential dye-based high-throughput screening to study ventricular arrhythmias caused by genetic channelopathy or cardiotoxic drugs.


Introduction
Type 1 long QT syndrome (LQT1), a frequent type of congenital LQT syndrome (cLQTS) [1], is caused by a reduction of slow delayed rectifier K + current (I Ks ) and is associated with a loss-of-function mutation in the KCNQ1 gene [2]. KCNQ1 A341V is known as one of the most frequent and severe KCNQ1 mutations [3]. Its coexpression with wild-type KCNE1, a beta subunit of the I Ks channel, causes a pathological reduction of I Ks [4], suggesting the importance of building physiological conditions to recapitulate the pathology in vitro. Human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) were shown to remodel ion channels that regulate the electrophysiological activity of the human heart and to successfully recapitulate the LQT1 phenotype [5][6][7][8].
Previous papers related to cLQTS patient-specific hiPSC-CMs verified the disease phenotype by using either the manual patch clamp method alone [5,7] or in combination with the microelectrode array (MEA) system [6,8]. However, the field potential duration (FPD), which reflects QT on electrocardiograms (ECG), of ventricular hiPSC-CM monolayers is longer than that of atrial iPSC-CMs when using the MEA system [9], suggesting a high proportion of ventricular hiPSC-CMs is required to analyze ventricular arrhythmic disease. Patch clamp analysis is standard for classifying the cardiac subtype, but it requires much time and expertise [10]. Automated patch clamp techniques may resolve these drawbacks, but they cause cells to produce spontaneous action potentials (APs) that are erratic due to enzymatic treatment immediately before the analysis [11].
FluoVolt (FV) is a new membrane potential VF2.4.CI dye [12,13]. It modulates photo-induced electron transfer (PeT) from an electron donor to a fluorophore through a synthetic molecular wire [12,14] and responds on a microsecond timescale, which makes it faster than genetically encoded voltage indicators [15,16]. Moreover, it has a higher fluorescence ratio (about 20% ΔF/F per 100 mV greater) and lower phototoxicity than electrochromic dyes such as di-4-ANEPPS [12,17], enabling classification of the subtype of single cardiomyocytes [18].
FV and VF2.1.Cl have been used to analyze APs from iCell ® cardiomyocyte monolayers or type 3 LQTS-iPSC-CM monolayers and also drug effects on multiwell plate readers [19,20]. However, the proportion of ventricular-type cells in those monolayers was not examined. Since arrhythmogenic change occurs in ventricular cells in LQTS, it is crucial to investigate the electrophysiological properties of this specific subtype.
In the present study, we propose a new method that incorporates FV and combines optical measurements of the membrane potential of single cardiomyocytes and in cardiac monolayers to identify the subtype and associate the subtype with the cardiomyocyte properties. We show the effectiveness of this method for modeling LQT1 using patient hiPSC-CMs.

Generation of Human iPS Cells from Three Type 1 LQTS
Patients. This study was approved by the Ethics Committees of the Graduate School of Medicine Kyoto University and the Kyoto University Hospital. Written informed consent was obtained from the patients in accordance with the Declaration of Helsinki. LQT1-iPSCs were generated from three patients using episomal vectors as described previously [21]. Three iPSC lines derived from healthy donors (201B7 [22], 409B2 [21], and 692D2 [23]) were used as control iPSCs. Human iPSCs were maintained on STO feeder layers cultured with primate ES cell medium (ReproCell, Yokohama, Kanagawa, Japan), as previously described [22].

Thawing Frozen iPSC-CMs.
For perforated patch clamp recording, the cryotubes were thawed in a 37°C water bath, centrifuged at 300 g for 5 min, and seeded on fibronectin-coated No. 2 cover glass (Matsunami, Osaka, Japan) filled with StemPro 34 SFM (Thermo Fisher Scientific) containing buffer and 10 ng/ml VEGF. Five to ten days after seeding, the cells were subjected to patch clamp experiments.   . One hour later, fibronectin was aspirated, and a 2 μl bead of thawed 4 × 10 4 iPSC-CM suspension was placed on the 35 mm dish or a 5 μl bead of thawed 5 × 10 4 iPSC-CM suspension was placed on the 96-well plate. After another one-hour incubation, the appropriate volume of media was added gently. The dish or plate was incubated at 37°C, 5% CO 2 . The composition of the media was the same on days 7 to 29. Medium was exchanged every 2 to 3 days.

Loading of FV.
When loading FluoVolt (FV; Thermo Fisher Scientific), the medium was exchanged with modified Tyrode's solution and FV (0.1% volume). Pluronic surfactants were not used in this study. Twenty minutes after loading, the medium was refreshed with modified Tyrode's solution and used in each assay.

Establishment of Three LQT1 Patient-Derived iPS Cell
Lines. We selected three LQT1 patients as donors for the iPSC derivation. One of the donors was a 50-year-old woman (II-2 in Figure 1(a)) who experienced presyncope several times when she was in junior high school and underwent recurrent syncope in her thirties. She showed prominent QT prolongation in resting ECG (Figure 1(b)) and exercise ECG. The other donors were her two daughters whose QT intervals were prolonged according to school medical examinations. Genetic testing diagnosed the mother and two daughters as type 1 long QT syndrome with KCNQ1 A341V mutation (c.1022C>T) (Figure 1(c)). The mutation is located at the transmembrane region in segment 6 near the pore of the I Ks channel (Figure 1(d)) and is reported as one of the severest types of LQT1 [3]. Medical therapy (beta-blockade) and lifestyle measures were sufficient to prevent recurrent events in the three patients. Five of the six family members positive for KCNQ1 mutation experienced syncope, and the sixth (III-1 in Figure 1(a)) did not. All carriers showed QTc prolongation on ECG. iPSCs were generated from the peripheral blood of the three patients using a nonviral method [21] and were differentiated into cardiomyocytes via EB formation [24,25] (Figure 1(e)). There was no significant difference in the expression of genes that affect APD between control-and LQT1-iPSC-CMs (Supplementary Figure 1).

Patch Clamp
Analysis. Ventricular-type cardiomyocytes derived from the three control-and three LQT1-iPSC lines were subject to current clamp recordings (Figure 2(a)). Although there was no significant difference in MDP or APA, there was a significant difference in APD 90 during pacing at 1 Hz between control-and LQT1-iPSC-CMs (Figure 2(b)) (p value, 0.0026). In addition, voltage clamp recordings revealed much smaller chromanol 293B-sensitive I Ks currents from LQT1-iPSC-CMs than controls (Figures 2(c) and 2(d)).

Action Potentials Recorded by FV in hiPSC-CM
Monolayers. We next measured the APDs during pacing in high-density cardiomyocyte culture (Figure 4(a)). Ten consecutive waves after dye loading were averaged (Figure 4(b) and Supplementary Movie). APD 90 from LQT1-iPSC-CMs was significantly longer than that from control-iPSC-CMs (Figure 4(c)) (p value, 0.000096).

APD Measured by FV on High-Throughput Plate Reader.
We next measured hiPSC-CM monolayers by FV in a higher throughput system consisting of a 96-well plate. The APs on monolayers were stable before and after 1 Hz pacing ( Figure 5(a)). The APD 90 averaged from 10 consecutive waves of LQT1B1-iPSC-CMs was longer than that from control-iPSC-CMs, and cisapride prolonged the APD 90 of cardiomyocytes derived from both iPSC groups ( Figure 5(b)). In addition, we administered a beta stimulant to mimic the conditions under exertional or emotional stress, in which the disease phenotypes are more prominent [5,6]. In the presence of 100 nM isoproterenol (ISP), we assessed the effect of several agents on drug-induced long QT syndrome (diLQTS). Erythromycin and cisapride prolonged APD 90 in both iPSC-CM groups under ISP, but the APD 90 of LQT-iPSC-CMs was consistently higher at all concentrations (Figures 5(c) and 5(d)). These findings support the applicability of the membrane dye system to high-throughput-based drug discovery and toxicology testing.

Discussion
The proportion of a cardiac subtype in iPSC-CMs or cardiac monolayers and the AP parameters were reported to vary depending on the culture duration [29], suggesting the importance of a high proportion of the ventricular subtype when modeling the disease phenotypes of ventricular arrhythmias. In this study, we have proposed a simple method that uses FV to measure both single cardiomyocytes and cardiac monolayers derived from normal and LQT1-iPSCs. Optical recordings of single cardiomyocytes identified the cardiac subtypes of each cells and confirmed their variable electrophysiological properties (Figures 3(a)-3(c)), which was consistent with previous reports of a genetically encoded membrane potential sensor [15,16]. High electrophysiological variability of single iPSC-CMs warrants assessment of a large population of ventricular cells for modeling ventricular arrhythmia. Our results demonstrated the feasibility of optical recording to identify cardiac subtypes and assess the electrophysiological properties of a large number of iPSC-CMs simultaneously.
The monolayered cells successfully reduced the variations of APD when synchronized with electrical stimulation at a constant frequency. As previously reported, FPD, an electrophysiological parameter highly correlated with APD [30], of ventricular hiPSC-CM monolayers was longer than that of atrial iPSC-CM monolayers [9], suggesting that the high proportion of the ventricular subtype would be appropriate for analyzing the APs of LQT-iPSC-CM monolayers and diLQTS models. Combined with single cell-based identification of cardiac subtypes, our optical recordings confirmed that iPSC-CM monolayers composed mainly of ventricular cells recapitulated the prolonged AP duration of cLQTS and diLQT. In addition, this system was applicable to high-throughput analysis to investigate the response to drugs.
Although conventional styryl voltage-sensitive dyes, such as di-4-ANEPPS and di-8-ANEPPS, have the ability to respond quickly to voltage changes, they have low sensitivity and nonnegligible phototoxicity [17,31]. FV and di-4-ANBDQBS, a near-infrared fluorescent voltage-sensitive dye, can precisely track the transmembrane voltage with lower photodynamic damage than di-4-ANEPPS [12,32,33]. Furthermore, FV has been used for a high-throughput screening [19]. ArcLight is a genetically encoded sensor of membrane potential that has brighter fluorescence and negligible phototoxicity and has been used to study cardiomyocytes derived from human embryonic stem cells [15] and LQT2 iPSC-CMs [16]. However, the recorded APs do not agree with those from patch clamp techniques due to the slow temporal response of ArcLight [34]. VSFP-CR [35], a derivative of VSFP2.3 [36], was previously combined with subtype-specific marker genes (MLC2v, SLN, and SHOX2) to analyze the APs of three iPSC-CM subtypes [37]. Although genetically encoded voltage indicators enabled promotor-specific voltage response, it takes much time to establish the genetically modified cell lines stably. On the other hand, FV can be rapidly used with many cell lines, as it needs only 20 min loading. diLQTS, which is an acquired LQTS, is mainly caused by blockage of the I Kr channel and is more frequent than cLQTS. It is well known that the mutation of genes responsible for cLQTS could contribute to an increased risk of diLQTS [38,39]. diLQTS has a similar mutation rate as cLQTS for the three major genes responsible for cLQTS [40], leading to the hypothesis that type 1 cLQTS patient-derived iPSC-CMs are appropriate for studying drugs that cause QT prolongation by blocking I Kr current. In the present study, we used FV to identify drugs that prolonged QT in LQT1-iPSC-CM and control-iPSC-CM monolayers in a high-throughput manner ( Figures 5(b)-5(d)).
Despite our encouraging results, it should be noted that FV has some limitations because the measured APs are based on a relative scale. Nor do the APs provide MDP or APA values, unlike patch clamp analysis. Furthermore, hiPSC-CMs are heterogeneous in electrophysiological properties, including their APDs and automaticities, making it difficult to measure the APs of paced cardiomyocytes simultaneously. Thus, in FV experiments, APDs in single cells were obtained from unpaced cells with different beating frequencies. Heterogeneity of the induced cardiomyocytes may also cause variations in the recordings. Studies have investigated ways to mature hiPSC-CMs [41,42], which could reduce the heterogeneity and thus contribute to more robust and precise highthroughput screening.
Although many reports have used the AP morphologybased classification of iPSC-CM subtypes [27,28,43,44], it has recently been argued that AP morphology is not a reliable indicator [45,46]. Future methods that can distinguish cardiomyocyte subtypes induced from pluripotent stem cells would contribute to measuring long APD in LQT-iPSC-CMs more precisely.

Conclusions
In summary, using an FV-based optical measurement system, we successfully identified ventricular-type cells as the major population in cardiac subtypes and found that they have a higher frequency of APD prolongation and EAD when derived from LQT1 patient iPSCs. Further, monolayers with this major population were suitable for analyzing two-dimensional AP in LQT-iPSC-CMs and diLQTS models in a high-throughput manner.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
Dr. Yoshida owns stocks in iPS Portal, Inc. Although Takeda Pharmaceutical Company Ltd. was not a sponsor of this work, some data collection was conducted by using equipment leased by the company. Also, Takeda Pharmaceutical Company Ltd. is paying the salary of Dr. Takaki independently of this work. All other authors have no conflicts of interest.