Author + information
- Received June 27, 2018
- Revision received December 30, 2018
- Accepted February 4, 2019
- Published online May 6, 2019.
- Rami Shinnawi, MSca,∗,
- Naim Shaheen, MSca,∗,
- Irit Huber, PhDa,
- Assad Shiti, MSca,
- Gil Arbel, MSca,
- Amira Gepstein, PhDa,
- Nimer Ballan, BSca,
- Noga Setter, BSca,
- Anke J. Tijsen, PhDa,
- Martin Borggrefe, MD, PhDc,d and
- Lior Gepstein, MD, PhDa,b,∗ (, )@TechnionLive
- aSohnis Research Laboratory for Cardiac Electrophysiology and Regenerative Medicine, Rappaport Faculty of Medicine and Research Institute, Technion-Israel Institute of Technology, Haifa, Israel
- bCardiolology Department of Rambam Medical Center, Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel
- cFirst Department of Medicine, Faculty of Medicine, University Medical Centre Mannheim, University of Heidelberg, Mannheim, Germany
- dDZHK (German Center for Cardiovascular Research), Partner Site, Heidelberg-Mannheim, Mannheim, Germany
- ↵∗Address for correspondence:
Dr. Lior Gepstein, Rappaport Faculty of Medicine, Technion–Israel Institution of Technology, POB 9649, Haifa, 31096, Israel.
Background The short QT syndrome (SQTS) is an inherited arrhythmogenic syndrome characterized by abnormal ion channel function, life-threatening arrhythmias, and sudden cardiac death.
Objectives The purpose of this study was to establish a patient-specific human-induced pluripotent stem cell (hiPSC) model of the SQTS, and to provide mechanistic insights into its pathophysiology and therapy.
Methods Patient-specific hiPSCs were generated from a symptomatic SQTS patient carrying the N588K mutation in the KCNH2 gene, differentiated into cardiomyocytes, and compared with healthy and isogenic (established by CRISPR/Cas9-based mutation correction) control hiPSC-derived cardiomyocytes (hiPSC-CMs). Patch-clamp was used to evaluate action-potential (AP) and IKr current properties at the cellular level. Conduction and arrhythmogenesis were studied at the tissue level using confluent 2-dimensional hiPSC-derived cardiac cell sheets (hiPSC-CCSs) and optical mapping.
Results Intracellular recordings demonstrated shortened action-potential duration (APD) and abbreviated refractory period in the SQTS-hiPSC-CMs. Similarly, voltage- and AP-clamp recordings revealed increased IKr current density due to attenuated inactivation, primarily in the AP plateau phase. Optical mapping of the SQTS-hiPSC-CCSs revealed shortened APD, impaired APD-rate adaptation, abbreviated wavelength of excitation, and increased inducibility of sustained spiral waves. Phase-mapping analysis revealed accelerated and stabilized rotors manifested by increased rotor rotation frequency, increased rotor curvature, decreased core meandering, and increased rotor complexity. Application of quinidine and disopyramide, but not sotalol, normalized APD and suppressed arrhythmia induction.
Conclusions A novel hiPSC-based model of the SQTS was established at both the cellular and tissue levels. This model recapitulated the disease phenotype in the culture dish and provided important mechanistic insights into arrhythmia mechanisms in the SQTS and its treatment.
The short QT syndrome (SQTS) is an inherited channelopathy, characterized by an abbreviated QTc interval on the electrocardiogram (ECG), increased propensity for atrial and ventricular arrhythmias, and sudden cardiac death (SCD) (1). The disease is classified into 6 subtypes according to the mutated gene, all leading to abbreviated repolarization (2). Short QT syndrome type 1 (SQTS1) is the most common form resulting from mutations (primarily the N588K missense mutation) in the KCNH2 gene (3,4), which encodes the hERG potassium channel responsible for the IKr current. Heterologous expression systems and in-silico models of the N588K mutated channel revealed gain-of-function IKr activity due to decreased rectification at plateau potentials, producing an augmented current density mainly during earlier phases of the action-potential (AP) (5). The predicted increase in the cardiomyocytes repolarizing forces shortens action-potential duration (APD), an in vitro surrogate for the ECG abbreviated QT interval (5).
One of the limitations of heterologous expression systems is the inability to reproduce the complexity of the whole cellular ion-channel milieu at the single-cell level and, even more so, the complex electrophysiological properties at the tissue level. Consequentially, there is a crucial need for human-based cardiomyocyte cellular and tissue models of the SQTS to identify new insights into the disease mechanisms and to serve as experimental platforms to evaluate therapeutic interventions.
The advent of the human-induced pluripotent stem cell (hiPSC) technology (6) coupled with improved methods for cardiomyocyte differentiation (7) can overcome the aforementioned obstacles. Consequentially, several patient/disease-specific hiPSC-based models of a variety of inherited arrhythmogenic syndromes (8) were derived, such as for the long QT syndrome (9,10), arrhythmogenic right ventricular cardiomyopathy (11), Brugada and overlap syndromes (12), and catecholaminergic polymorphic ventricular tachycardia (13). The generated patient-specific hiPSC-derived cardiomyocytes (CMs) were demonstrated to recapitulate the different disease phenotypes, to provide insights into various disease pathogenesis, and to establish platforms for drug testing. Nevertheless, the vast majority of these studies focused on studying the electrophysiological properties of hiPSC-CMs at the single-cell level. Understanding more complex electrophysiological phenomena, such as conduction and re-entrant arrhythmias, however, can only be studied at the tissue level.
In this study, we aimed to take the field of hiPSC-based inherited arrhythmia modeling to the next level by combining patient-specific hiPSC-CMs, clustered regularly interspaced short palindromic repeats (CRISPR)–based genome editing, cellular electrophysiological phenotyping, 2-dimensional tissue-engineering models, and optical mapping to study the mechanisms underlying SQTS pathophysiology and treatment at both the cellular and tissue levels (Central Illustration).
Expanded methods are available in the Online Appendix.
hiPSC generation and cardiomyocyte differentiation
Dermal fibroblasts, obtained from an SQTS1 patient, were reprogrammed to generate the patient-specific hiPSCs by retroviral delivery of SOX2, KLF4, and OCT4 as described (10). A healthy-control hiPSC line (14) and CRISPR/Cas9-based gene-corrected isogenic hiPSC line served as controls. Cardiomyocytes differentiation was achieved using the monolayer-directed differentiation system (7,14). Beating monolayers (25 to 50 days of differentiation) were enzymatically dissociated and seeded as single cardiomyocytes for patch-clamp recordings or used to generate hiPSC-derived cardiac cell sheets (hiPSC-CCSs) (15).
Tissues were loaded with FluoVolt and imaged using a high-speed EM-CCD camera (Photometrics, Tucson, Arizona) mounted on a fluorescent macroscope. hiPSC-CCSs were excited using 475-nm peak wavelength LED. Emitted fluorescence was passed through a 495-nm long-pass dichroic mirror and filtered using 525/50-nm band-pass filter. Optical mapping data were acquired at 260 frames/s in 4 × 4 binning and analyzed using a custom-made software. Phase-maps and related rotor biophysical parameters were analyzed as previously described (16).
Derivation of patient-specific SQTS-hiPSCs
Dermal fibroblasts were obtained from a 73-year-old woman previously diagnosed with SQTS1. Clinically, the patient displayed palpitations, frequent ventricular and supraventricular ectopy, and shortened QT (270 ms) and QTc (310 ms) intervals on the ECG (Figure 1A). The patient had a 3-generation family history of SCD (Figure 1B) and was previously implanted with an ICD (3). Genetic workup revealed a heterozygous missense mutation (N588K) in the S5-P loop region of the KCNH2-encoded hERG channel in all affected family members (Figure 1C) (4).
Fibroblasts were reprogrammed to generate the patient-specific SQTS-hiPSCs by retroviral delivery of KLF-4, OCT4, and SOX-2. The generated hiPSC colonies displayed characteristic morphology; maintained a normal karyotype (Figure 1D); and expressed the pluripotency markers NANOG, OCT-4, TRA-1-60, and SSEA-4 (Figure 1E). Pluripotency was confirmed by presence of cell-derivatives of all 3 germ layers in differentiating embryoid bodies (Online Figure 1A) and by formation of teratomas in SCID-beige mice (Online Figure 1B). Finally, undifferentiated hiPSCs showed silencing of the 3 retroviral transgenes (Online Figure 1C) and reactivation of endogenous pluripotency genes NANOG, OCT4, and SOX2 (Online Figure 1D). Sequencing of the KCNH2 gene revealed the N588K mutation in SQTS-hiPSCs but not in healthy control hiPSCs (Figure 1F).
The SQTS-hiPSCs were coaxed to differentiate into hiPSC-CMs, which maintained retroviral transgene silencing (Online Figure 1C). The differentiated hiPSC-CMs showed positive staining for cTnI, α-actinin, and MLC2V (Figure 1G). Similarly, gene expression analysis of hiPSC-CMs demonstrated down-regulation of pluripotency genes (OCT4 and NANOG) and up-regulation of cardiac-specific genes (MYH-6, NKX2.5, MLC-2V, and MYH7) (Online Figure 1E, Online Table 1). Allele-specific qPCR (Online Figure 1F), targeting the mutation sequence, revealed similar expression levels of the mutated (48 ± 3%) and wildtype (52 ± 2%; p = NS) KCNH2 alleles. A previously created healthy-control hiPSC line (14) was used to obtain healthy-control hiPSC-CMs.
N588K mutation correction using CRISPR/Cas9-mediated genome editing
To establish the relevant isogenic-control, we used CRISPR/Cas9-mediated genome editing (Figure 2A, Online Figure 2) to correct the N588K mutation in the SQTS-hiPSCs. Sequencing of the KCNH2 gene identified the N588K mutation in SQTS-hiPSCs and the wildtype sequence at the same site in gene-corrected isogenic-control hiPSCs (Figure 2B). Potential genome-editing related on-target or off-target mutagenesis were ruled out by sequencing the open reading frame of the KCNH2 gene (comprised of 9 exons) and by sequencing potential off-target sites (chosen based on their “cutting frequency determination” score), respectively (Online Table 2). The corrected hiPSC clones were propagated, demonstrated to maintain a normal karyotype (Figure 2C) and pluripotency properties (Figure 2D), and differentiated to generate isogenic-control hiPSC-CMs.
We next investigated the AP properties of the SQTS-hiPSC-CMs and compared them with healthy and isogenic control hiPSC-CMs. Whole-cell patch-clamp recordings of APs stimulated at 1 Hz (Figures 3A and 3B) revealed significantly shortened APD90 in the SQTS-hiPSC-CMs (APD90: 181 ± 5 ms; n = 18) when compared with healthy (285 ± 15 ms; n = 16; p < 0.001) and isogenic (319 ± 14 ms; n = 14; p < 0.001) control cardiomyocytes.
To evaluate the refractory-period (RP) of hiPSC-CMs, we applied repeated trains of 10 stimuli at 1 Hz (S1) followed by a premature stimulus (S2) (Online Figure 3A). By shortening the S1-S2 coupling interval until loss-of-capture, we determined the RPs (Online Figures 3A and 3B). The SQTS-hiPSC-CMs exhibited significantly shortened RP (292 ± 11 ms; n = 13) when compared with both healthy (445 ± 17 ms; n = 15; p < 0.001) and isogenic (522 ± 24 ms; n = 12; p < 0.001) control groups (Figure 3C, Online Figure 3B). Interestingly, although both control groups displayed markedly longer RPs than SQTS cells, we found small, but significant differences (p < 0.01) in RP values between healthy and isogenic-control hiPSC-CMs.
Another finding in the SQTS-hiPSC-CMs was the reduced rate-adaptation of APD when paced at increasing frequencies (Figures 3D and 3E). This was manifested by a lesser steep slope in the APD90/rate restitution-plot for the SQTS-hiPSC-CMs (slope: −26 ± 3 ms/Hz; n = 13) (Figure 3E) as compared with healthy (−68 ± 3 ms/Hz; n = 10; p < 0.001) and isogenic (−73 ± 7 ms/Hz; n = 9; p < 0.001) control hiPSC-CMs.
Characterization of IKr current
To study the mechanism underlying the shortened APD in SQTS1, we performed voltage-clamp studies to characterize IKr properties (defined as the E4031-sensitive current) in the SQTS-hiPSC-CMs. These studies revealed significantly higher IKr end-pulse current-density in SQTS cardiomyocytes when compared with both healthy- and isogenic-hiPSC-CMs (Figures 3F and 3G; p < 0.01), as well as reduced rectification (Figure 3H). Interestingly, the IKr tail-current density was lower than the IKr end-pulse current-density in SQTS-hiPSC-CMs, suggesting attenuated IKr inactivation. Finally, voltage dependence of the tail-currents did not differ between SQTS and both control hiPSC-CMs (Online Figure 3C).
To elucidate IKr contribution to AP waveform in SQTS, we used the AP-clamp technique and revealed significant augmentation of IKr density throughout the AP in SQTS-hiPSC-CMs when compared with both control cardiomyocytes (Figure 3I). Interestingly, as shown in both the absolute current-density (Figure 3I) and normalized current (Online Figure 3D) tracings, IKr density was especially augmented in SQTS-hiPSC-CMs during the plateau and early-repolarization phases.
Increased hERG channel protein expression in SQTS-hiPSC-CMs
We next investigated for additional molecular mechanism contributing to the abbreviated APD in SQTS-hiPSC-CMs. We performed flow-cytometry analysis to assess hERG-channel protein expression in the 3 hiPSC-CM groups (quantified as mean fluorescence intensity). We found that hERG-channel expression level was 1.47 ± 0.27 folds higher in the SQTS-hiPSC-CMs when compared with healthy control (p < 0.05) hiPSC-CMs and 1.77 ± 0.21 folds higher when compared with isogenic control (p < 0.01) hiPSC-CMs (Online Figures 4A and 4B).
We also performed similar experiments in HEK293 cells, transfected to express either wildtype or N588K-mutant KCNH2 transgenes. A similar phenomenon was noted, in which hERG-channel expression levels were 1.20 ± 0.15 folds higher in the N588K-HEK293 cells compared with wildtype-HEK293 cells (p < 0.05) (Online Figures 4C and 4D).
Taken together our results suggest that the augmented IKr current in the N588K-SQTS-hiPSC-CMs results from a combination of increased hERG-channel expression and altered biophysical properties of the channel.
Studying the SQTS at the tissue level
To study SQTS in a more clinically-relevant model, we utilized a novel 2-dimensional hiPSC-based cardiac tissue model (15). Differentiated hiPSC-CMs (containing ≥85% cardiomyocytes) were enzymatically dissociated and seeded as circular, high-density, large-scale (5-mm-diameter) hiPSC-derived cardiac cell sheets (hiPSC-CCSs) (Online Figure 5). The hiPSC-CCSs generated from all groups (SQTS, heathy and isogenic controls) were loaded with the voltage-sensitive dye FluoVolt (Thermo Fisher Scientific, Waltham, Massachusetts) and subjected to detailed optical mapping. These studies, performed during focal electrical stimulation, confirmed the formation of a functional syncytium in all hiPSC-CCSs with uniform AP propagation (Figure 4A, Online Video 1).
Similar to the single-cell studies, assessment of the optically derived AP properties revealed significant differences between SQTS-hiPSC-CCSs and control groups. As noted from the AP signals (Online Figure 6A), APD maps (Figure 4B), and APD restitution-plots (Figure 4C), the SQTS-hiPSC-CCSs displayed significantly shorter mean APD80 values when compared with both control tissues at all pacing rates (Figure 4C). For example, APD80 values at 1 Hz averaged 164 ± 4 ms (n = 37) in SQTS-hiPSC-CCSs versus 284 ± 9 ms (n = 27; p < 0.001) and 339 ± 15 ms (n = 20; p < 0.001) in healthy and isogenic control tissues, respectively (Figure 4C).
In contrast to the marked differences in the repolarization properties, we did not observe any significant changes in conduction in the SQTS-hiPSC-CCSs. Thus, mean conduction velocity (CV) values, measured from the optical-mapping derived activation maps (Figure 4A), did not differ between the SQTS-hiPSC-CCSs and both control specimens (p = NS) (Figures 4A and 4D).
As a consequence of the aforementioned results, the impulse wavelength, which can be approximated by the product of APD and CV, was significantly shorter in the SQTS-hiPSC-CCSs when compared to healthy and isogenic control tissues (Figure 4E, Online Figure 6B, Online Video 1) at all pacing frequencies (Figure 4E). For example, the wavelength at 1 Hz was 5.3 ± 0.3 mm in SQTS-hiPSC-CCSs (n = 37), which was significantly shorter than in healthy (9.3 ± 0.6 mm; n = 27; p < 0.001) and isogenic (11.0 ± 0.9 mm; n = 20; p < 0.001) control tissues (Figure 4E). This finding is important, because wavelength abbreviation is associated with increased risk for initiation and sustainment of re-entrant arrhythmias.
Characterizing arrhythmogenicity in the SQTS-hiPSC-CCS model
We next aimed to utilize the hiPSC-CCS model to detect, monitor, and study re-entrant arrhythmias. To this end, we attempted to induce arrhythmias in the different hiPSC-CCS groups through a systematic electrical programmed stimulation approach. Using this protocol, we could robustly induce sustained re-entrant arrhythmias in the SQTS-hiPSC-CCSs (Figures 5A and 5B, Online Video 2). These arrhythmias were manifested as spiral-waves (Online Videos 2 and 3, Figures 5B and 5C, Online Figure 7), which were highly stable, persisted for long periods, and could be monitored and analyzed using optical mapping.
Sustained rotors were successfully induced in the vast majority (81%; n = 48) (Figure 5A) of the SQTS-hiPSC-CCSs. In contrast, we could only induce rotors in 35% (n = 31; p < 0.001) (Figures 5A and 5B) of healthy-control hiPSC-CCSs and in none of the isogenic-control hiPSC-CCSs (n = 16; p < 0.001) (Figure 5A). To evaluate the rotors' biophysical properties, we performed phase-mapping analysis of the optical recordings (15,16) (Figure 5C, Online Figure 7, Online Videos 3 and 4). These studies revealed significant differences in the rotors' dynamic properties between SQTS and healthy control tissues (rotors were not induced in the isogenic-control hiPSC-CCSs).
Specifically, we noted increased rotor rotation frequency in the SQTS-hiPSC-CCSs (3.59 ± 0.11 Hz; n = 37) when compared with healthy hiPSC-CCSs (2.58 ± 0.23 Hz; n = 11; p < 0.01) (Figure 5D). To compare the rotor's wave front curvature near its core, we measured the distance from the rotor's core to its wave front (an inversely proportional parameter to rotor wave front curvature ) (Online Figure 7B). Rotors induced in the SQTS-hiPSC-CCSs displayed smaller distances from their cores to their wave fronts than those induced in healthy-hiPSC-CCSs (1.13 ± 0.07 mm vs. 2.09 ± 0.18 mm; p < 0.001) (Figure 5E).
We also followed the rotor tips' trajectories over time (Online Video 4, Figure 5F left panel) and identified decreased rotor meandering (correlating with greater stability) in the SQTS-hiPSC-CCSs compared with rotors in healthy-hiPSC-CCSs (p < 0.01) (Online Video 4, Figure 5F right panel). Finally, arrhythmias induced in SQTS-hiPSC-CCSs were more complex than their control counterparts as defined by the percentage of specimens containing more than 1 active rotor simultaneously (Online Video 5, Figures 5G and 5H) (p < 0.05).
Quinidine can rescue the abnormal SQTS phenotype
We next evaluated the utility of the SQTS-hiPSC-CCS model for assessing drug treatments. Quinidine, a class IA antiarrhythmic agent, is considered the treatment of choice for SQTS (17,18). Application of 1 μmol/l quinidine significantly prolonged APD in both the healthy and isogenic control hiPSC-CCS groups (Online Figures 8A and 8B) and in the SQTS-hiPSC-CCSs, as depicted in the optical-mapping derived APD80 maps (Figure 6A) and optical signals (Online Figure 9A and 9B). Consequentially, quinidine significantly prolonged APD80 values in the SQTS-hiPSC-CCSs at all pacing rates (Figure 6C). Interestingly, quinidine also normalized the impaired rate-adaptation in the SQTS-hiPSC-CCSs (Slope: −0.77 ± 0.10 ms/beats/min [n = 14] vs. −0.15 ± 0.03 ms/beats/min; n = 14; p < 0.001) (Figure 6C).
Evaluating quinidine effect on conduction, we noted that quinidine treatment did not alter CV values (p = NS) (Figures 6B and 6D). As a result of its effects on APD and CV, quinidine treatment significantly increased the wavelength of excitation (Online Figure 9B, Online Video 6) at all pacing frequencies (Figure 6E). Finally, we evaluated the potential antiarrhythmic action of quinidine. To this end, we repeated the arrhythmia pacing induction protocols and noted that quinidine significantly reduced arrhythmia inducibility in treated SQTS-hiPSC-CCSs (induction rate of 29% compared with 93% in untreated SQTS-hiPSC-CCS; n = 14; p < 0.001) (Figure 6F).
Disopyramide but not sotalol can rescue the SQTS phenotype
We next evaluated the effects of alternative antiarrhythmic agents. Sotalol, a class III antiarrhythmic agent, is known to prolong cardiomyocytes' APD by blocking the IKr current. Surprisingly, in contrast to the significant sotalol-induced APD prolongation in healthy and isogenic control groups (Online Figures 8C and 8D), application of 5 μmol/l sotalol did not prolong APD in the SQTS-hiPSC-CCSs (Figures 7A and 7B). For example, APD80 at 1 Hz did not differ between sotalol-treated (166 ± 5 ms) and untreated (163 ± 4 ms) SQTS-hiPSC-CCSs (n = 16; p = NS) (Figure 7B). The APD restitution curve (Figure 7B) revealed that sotalol did not have any significant effect on the rate-adaptation slope either. Optical mapping showed that sotalol also did not alter CV, and consequentially did not change wavelength values (p = NS) (Figures 7A, 7C, and 7D). Finally, sotalol treatment did not affect arrhythmia inducibility in the SQTS-hiPSC-CCSs (94% in untreated vs. 88% in treated; n = 16; p = NS) (Figure 7E).
We next evaluated the effects of disopyramide, a class IA antiarrhythmic agent. Application of 10 μmol/l disopyramide caused significant APD prolongation in the SQTS-hiPSC-CCSs at all pacing frequencies (Figures 7F and 7G). For example, APD80 at 1 Hz was significantly longer in disopyramide-treated SQTS cultures (236 ± 6 ms) compared with untreated counterparts (160 ± 3 ms; n = 13; p < 0.001) (Figure 7G). In addition, the administration of disopyramide rectified the rate-adaptation slope (Slope: −0.52 ± 0.06 ms/beats/min vs. −0.12 ± 0.02 ms/beats/min; n = 13; p < 0.001) (Figure 7G). APD prolongation was also observed following disopyramide administration to healthy and isogenic control specimens (Online Figures 8E and 8F).
Since disopyramide did not alter conduction in the treated cultures (Figures 7F and 7H), the wavelength of excitation was increased in the treated tissues (Figure 7I). For example, the wavelength at 1Hz was 6.9 ± 0.6 mm (n = 13) in treated and 5.1 ± 0.4mm (n = 13; p < 0.05) in untreated SQTS-hiPSC-CCSs. More importantly, we found that disopyramide significantly reduced arrhythmia induction in the SQTS-hiPSC-CCSs (86% in drug-free vs. 55% in drug-treated; n = 22; p < 0.05) (Figure 7J).
We established a patient-specific hiPSC-CM model of the SQTS1 by reprogramming fibroblasts from a patient carrying the KCNH2-N588K mutation (Central Illustration). By studying differentiated SQTS-hiPSC-CMs at the single-cell level, we could: 1) recapitulate the clinical SQTS phenotype by showing shortened APD; 2) demonstrate shortened RP and impaired rate-adaptation of APD; and 3) identify augmented IKr current due to a combination of increased channel protein expression and attenuated inactivation.
Because SQTS-related arrhythmias are re-entrant and cannot be studied at the single-cell level, we utilized a novel 2-dimensional hiPSC-CCS modeling approach to study the tissue's electrophysiological properties. Our results demonstrate: 1) the ability of the SQTS-hiPSC-CCSs to recapitulate the disease phenotype, manifested by marked shortening of the tissue's APD and wavelength (without affecting CV); 2) increased susceptibility for induction of re-entrant arrhythmias (spiral waves) in the SQTS-hiPSC-CCSs; 3) distinct biophysical properties and increased stability of rotors induced in SQTS-hiPSC-CCSs, manifested by increased rotor frequency and curvature, decreased rotor-core meandering, and increased arrhythmia complexity; and 4) effective antiarrhythmic actions of quinidine and disopyramide (but not sotalol) in preventing arrhythmia induction through APD and wavelength prolongation.
Generation of CRISPR/Cas9-based isogenic-control hiPSCs
Among the limitations of hiPSC-based disease modeling is the potential for phenotypic variability between lines due to diverse genetic backgrounds. This genetic variability could mask modest differences between the studied and healthy control cells or alternatively reveal differences that are not related to the studied mutation. The use of isogenic-control lines, in which diseased cells differ from controls only at the mutation site may allow to “isolate” the causative role of the studied mutation. Here, we utilized CRISPR/Cas9 genome-editing to correct the KCNH2-N588K mutation. By comparing the resulting gene-corrected isogenic-control hiPSC-CMs with the SQTS-hiPSC-CMs, we identified electrophysiological abnormalities directly resulting from the KCNH2-N588K mutation. Interestingly, the aforementioned electrophysiological differences were greater than the differences observed between SQTS-hiPSC-CMs and healthy control hiPSC-CMs, highlighting the importance of using an isogenic-suitable control.
Another important implication of correcting the causative mutation is the potential for future therapeutic applications. Although many challenges still exist before CRISPR-based gene correction can be used clinically in the in vivo heart, the ability to completely rescue the abnormal SQTS phenotype in the different models studied is highly encouraging.
Cellular electrophysiological characterization of SQTS-hiPSC-CMs
Abbreviated cardiac repolarization is the hallmark of SQTS (2). Our patch-clamp experiments revealed shortened APD90 in the SQTS-hiPSC-CMs, similar to a recent report studying the same mutation (19). We also observed impaired cellular APD rate-adaptation, consistent with the abnormal QT rate-adaptation in SQTS1 patients (20). The shortened APD resulted in abbreviated RP in SQTS-hiPSC-CMs, correlating with similar RP shortening in a pharmacologically-induced SQTS model in the canine left ventricular wedge preparations (21) and in SQTS patients during electrophysiological testing (3).
SQTS1 is caused by gain-of-function mutations in the KCNH2 gene, with the most common N588K mutation leading to a lysine-asparagine substitution at the S5-loop region of the hERG-channel. In vitro heterologous expression studies (5) and a recent hiPSC model (19) assessing the N588K mutation revealed gain-of-function activity of IKr, positive shift in IKr inactivation voltage dependence, and diminished inward rectification (exhibiting rectification only above +60 mV).
In agreement with these studies, we detected a pronounced increase in the IKr end-pulse current density and reduced rectification of IKr in the SQTS-hiPSC-CMs. By conducting AP-clamp experiments, we dissected the phase-specific IKr current density increase occurring primarily during the AP's early phases 2 to 3. In contrast to heterologous expression, the maximal IKr current density in SQTS-hiPSC-CMs was in physiologically relevant ranges, comparable to native human cardiomyocytes (22). Similarly, the reduced rectification and AP-related changes in IKr current-density waveform were smaller than in heterologous expression systems. These findings may be attributed to the heterozygous nature of SQTS (possessing both wildtype and mutated allele products), a phenomenon recapitulated in SQTS-hiPSC-CMs but not in heterologous expression. Moreover, in contrast to heterologous expression, hiPSC-CMs express the studied protein at physiological levels and recapitulate the cardiomyocyte's native milieu by expressing multiple other ion channels and modulators.
Flow cytometry allowed further investigation into the mechanisms underlying SQTS1. These studies revealed increased expression of the hERG-channel protein in the SQTS-hiPSC-CMs, which can contribute to the augmented IKr current beyond the biophysical findings. These results are in agreement with the study by El-Battrawy et al. (19), which identified a similar increase in hERG-channel expression in SQTS cardiomyocytes by immunostaining.
Studying SQTS at the tissue level
Most efforts in studying inherited arrhythmogenic disorders have focused on the cellular properties of hiPSC-CMs, namely their AP properties and ionic-current profile. Studying more complex electrophysiological phenomena such as conduction and reentry in arrhythmogenic syndromes like SQTS requires the development of multicellular tissue models.
To address the aforementioned challenge, we established a unique hiPSC-CCS model. Optical mapping—confirming the formation of electrically-coupled syncytium in all hiPSC-CCSs groups—allowed analysis of relevant electrophysiological parameters (CV and APD). This analysis revealed abbreviated mean APD80 values, reduced rate-adaptation of APD, and shortened wavelength of excitation in SQTS-hiPSC-CCSs. The SQTS-hiPSC-CCSs displayed increased susceptibility to development of re-entrant arrhythmias following programmed electrical stimulation, in line with increased ventricular fibrillation inducibility in SQTS patients (3).
The arrhythmias generated in the SQTS-hiPSC-CCSs consisted of stable spiral waves. Because similar rotors are hypothesized to play an important role in cardiac fibrillation (23), the established model may provide unique mechanistic insights into their initiation and perpetuation, especially in inherited arrhythmogenic conditions. The spiral waves were analyzed as dynamic displays, activation, or phase maps, allowing detailed analysis of rotors' biophysical properties. This analysis revealed significant differences in the rotors' dynamics in SQTS-hiPSC-CCSs when compared with healthy-control tissues (arrhythmias could not be induced in isogenic-control specimens). Such differences, manifested by increased rotation frequency, increased rotor curvature, decreased singularity-point meandering, and presence of multiple rotors, may lead to more stable and complex re-entrant excitation, contributing to the increased clinical incidence and severity of arrhythmias in SQTS patients.
Finally, we used the new model to screen the effects of potential SQTS therapies. Drugs were tested at the tissue level, in contrast to previous efforts focusing on single cells. We discovered that both quinidine and disopyramide were able to suppress arrhythmogenicity in our model through prolongation of the tissue's APD and wavelength without affecting CV.
Quinidine is considered the treatment of choice for SQTS and can normalize the QTc interval and prevent VF inducibility acutely (20), while reducing life-threatening arrhythmic events in long-term follow-up (24). For disopyramide there are only anecdotal case reports (25). Our results suggest that it could become a promising drug candidate for SQTS, albeit being somewhat less effective than quinidine in our study.
Interestingly, sotalol, which is known to prolong APD (due to its IKr blocking activity) in several experimental models and different clinical scenarios, did not affect APD or the wavelength in the SQTS-hiPSC-CCSs and consequentially failed to prevent arrhythmias in this model. These results are in agreement with clinical findings in N588K-SQTS1 patients, where sotalol and additional hERG blockers failed to cause significant QTc prolongations (17,26). It is hypothesized that the N588K mutation interferes with drug binding to the channel leading to reduced potency of canonical hERG inhibitors (26). It is interesting to note that in contrast to lack of sotalol effect in N588K-SQTS-hiPSC-CCSs, it displayed robust APD prolonging effects in both healthy and isogenic control hiPSC-CCSs.
First, hiPSC-CMs exhibit relatively immature fetal-like phenotype, manifested by slower AP upstroke velocities, depolarized maximum diastolic potential, and slow CV in the hiPSC-CCSs. This slow conduction was actually beneficial for our study, facilitating reentry formation in smaller tissues and allowing downscaling of the generated hiPSC-CCSs. Second, the generated hiPSC-CMs are composed of mixed population of ventricular-, atrial-, and nodal-like cells, with the majority of cells having ventricular phenotype. By utilizing recently reported chamber-specific cardiomyocyte differentiation protocols (27), we anticipate generating tissue models of atrial and ventricular fibrillation in arrhythmogenic syndromes like SQTS. Third, recent studies demonstrated the potential for on- and off-target mutagenesis following CRISPR/Cas9-based genome-editing (28). DNA sequencing analysis, however, failed to identify such on- or off- target events in the gene-corrected isogenic-control hiPSCs. Finally, several gene mutations can be causative for SQTS, and our results may only be specific to the N588K-KCNH2 mutation.
This study demonstrated the ability to recapitulate the SQTS disease phenotype at both the cellular and tissue levels using patient-specific hiPSC-CMs and to provide new insights into the mechanisms underlying arrhythmia initiation, perpetuation, and treatment in this syndrome. Moreover, the importance of CRISPR-based gene editing was demonstrated by generating the optimal isogenic controls and as a potential future therapy. Finally, the ability to study arrhythmogenesis at the tissue level can advance the field of hiPSC modeling to the next level, providing unique opportunities for modeling inherited arrhythmogenic disorders, for gaining insights into different forms of reentry, and unique experimental platforms for drug development and testing.
COMPETENCY IN MEDICAL KNOWLEDGE: Confluent 2-dimensional cardiac cell sheets generated from patient-specific hiPSC-CMs carrying a specific SQTS gene mutation were used to study conduction, repolarization, and arrhythmogenesis in this syndrome at the tissue level. At the cellular level, SQTS-hiPSC-CMs exhibit shorter APDs and abbreviated refractory periods. Electrophysiological abnormalities were normalized after exposure to quinidine and disopyramide, suppressing also arrhythmia induction.
TRANSLATIONAL OUTLOOK: The 2-dimensional hiPSC-CCS model may be useful for patient/disease-specific drug testing and for developing specific gene correction therapies for inherited arrhythmia syndromes.
↵∗ Drs. Shinnawi and Shaheen contributed equally to this work.
This study was partially funded by the European Research Council (ERC-2017-COG-773181-iPS-ChOp-AF), by the Leducq foundation (18CVD05), by the BIRAX initiative (04BX14CDLG), by the Kamin grant (Israel innovation authority), and by the CROWN foundation. Dr. Tijsen received Rubicon grant 825.13.007 from the Netherlands Organisation for Scientific Research. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Listen to this manuscript's audio summary by Editor-in-Chief Dr. Valentin Fuster on JACC.org.
- Abbreviations and Acronyms
- action-potential duration
- clustered regularly interspaced short palindromic repeats
- conduction velocity
- human-induced pluripotent stem cell
- human-induced pluripotent stem cell–derived cardiac cell sheet
- human-induced pluripotent stem cell–derived cardiomyocytes
- short QT syndrome
- Received June 27, 2018.
- Revision received December 30, 2018.
- Accepted February 4, 2019.
- Patel C.,
- Yan G.X.,
- Antzelevitch C.
- Gaita F.,
- Giustetto C.,
- Bianchi F.,
- et al.
- Brugada R.,
- Hong K.,
- Dumaine R.,
- et al.
- Shaheen N.,
- Shiti A.,
- Gepstein L.
- Davis R.P.,
- Casini S.,
- van den Berg C.W.,
- et al.
- Itzhaki I.,
- Maizels L.,
- Huber I.,
- et al.
- Shinnawi R.,
- Huber I.,
- Maizels L.,
- et al.
- Shaheen N.,
- Shiti A.,
- Huber I.,
- et al.
- Hou L.,
- Deo M.,
- Furspan P.,
- et al.
- Gaita F.,
- Giustetto C.,
- Bianchi F.,
- et al.
- Giustetto C.,
- Schimpf R.,
- Mazzanti A.,
- et al.
- El-Battrawy I.,
- Lan H.,
- Cyganek L.,
- et al.
- Patel C.,
- Antzelevitch C.
- Pandit S.V.,
- Jalife J.
- Mazzanti A.,
- Maragna R.,
- Vacanti G.,
- et al.
- Laksman Z.,
- Wauchop M.,
- Lin E.,
- et al.