Disease Phenotypes and Mechanisms of iPSC-Derived Cardiomyocytes From Brugada Syndrome Patients With a Loss-of-Function SCN5A Mutation

Brugada syndrome (BrS) is one of the major causes of sudden cardiac death in young people, while the underlying mechanisms are not completely understood. Here, we investigated the pathophysiological phenotypes and mechanisms using induced pluripotent stem cell (iPSC)-derived cardiomyocytes (CMs) from two BrS patients (BrS-CMs) carrying a heterozygous SCN5A mutation p.S1812X. Compared to CMs derived from healthy controls (Ctrl-CMs), BrS-CMs displayed a 50% reduction of INa density, a 69.5% reduction of NaV1.5 expression, and the impaired localization of NaV1.5 and connexin 43 (Cx43) at the cell surface. BrS-CMs exhibited reduced action potential (AP) upstroke velocity and conduction slowing. The Ito in BrS-CMs was significantly augmented, and the ICaL window current probability was increased. Our data indicate that the electrophysiological mechanisms underlying arrhythmia in BrS-CMs may involve both depolarization and repolarization disorders. Cilostazol and milrinone showed dramatic inhibitions of Ito in BrS-CMs and alleviated the arrhythmic activity, suggesting their therapeutic potential for BrS patients.

Brugada syndrome (BrS) is one of the major causes of sudden cardiac death in young people, while the underlying mechanisms are not completely understood. Here, we investigated the pathophysiological phenotypes and mechanisms using induced pluripotent stem cell (iPSC)-derived cardiomyocytes (CMs) from two BrS patients (BrS-CMs) carrying a heterozygous SCN5A mutation p.S1812X. Compared to CMs derived from healthy controls (Ctrl-CMs), BrS-CMs displayed a 50% reduction of I Na density, a 69.5% reduction of Na V 1.5 expression, and the impaired localization of Na V 1.5 and connexin 43 (Cx43) at the cell surface. BrS-CMs exhibited reduced action potential (AP) upstroke velocity and conduction slowing. The I to in BrS-CMs was significantly augmented, and the I CaL window current probability was increased. Our data indicate that the electrophysiological mechanisms underlying arrhythmia in BrS-CMs may involve both depolarization and repolarization disorders. Cilostazol and milrinone showed dramatic inhibitions of I to in BrS-CMs and alleviated the arrhythmic activity, suggesting their therapeutic potential for BrS patients.

INTRODUCTION
Brugada syndrome (BrS), a genetic heart disease, is one of the major causes of sudden cardiac death in young people. The diagnosis of BrS is based on the changes in the electrocardiogram (ECG) with a transient or persistent ST-segment elevation in the right precordial leads and the right bundle branch block (Antzelevitch and Fish, 2006). In the past two decades, extensive research on BrS has revealed parts of its genetic basis. To date, 293 different mutations in SCN5A encoding the α-subunit of the cardiac sodium channel (Na V 1.5) have been implicated as possible causes of BrS (Kapplinger et al., 2010). The implantation of an automatic implantable cardiac defibrillator (ICD) or the ablation of the arrhythmogenic substrate in the right ventricular outflow tract area have been proven effective in treating ventricular tachycardia and fibrillation and preventing sudden cardiac death in BrS patients. In addition, drug therapy with low-dose quinidine has been shown to prevent arrhythmic events in BrS patients after arrhythmic storm and should be considered according to current guidelines (class IIa indication). However, prophylactic use in asymptomatic patients is controversially discussed and worldwide commercial availability is limited. Thus, a detailed understanding of arrhythmic mechanisms in BrS is required to develop novel drug therapies that are urgently needed.
For more than a decade, the underlying pathophysiological mechanism of BrS has been a matter of debate between the depolarization versus repolarization hypotheses (Wilde et al., 2010). The depolarization theory relies on right ventricular conduction slowing and involvement of structural abnormalities including increased epicardial and interstitial fibrosis, and the reduced gap junction expression (Nademanee et al., 2015). The repolarization theory is mainly supported by the late fractionated low-voltage potentials found in right ventricle epicardium but not in endocardium by using canine heart wedge model, which are associated with repolarization defects (Szel and Antzelevitch, 2014;Patocskai et al., 2017). Nonetheless, there is a general consensus that SCN5A mutations resulting in Na V 1.5 lossof-function either by the decreased expression of functional Na V 1.5 at the sarcolemma and/or by the altered channel gating properties lead to a diminution of I Na and contribute to the development of arrhythmias in BrS (Patocskai et al., 2017). Recently, single cardiomyocytes (CMs) derived from patientspecific and genome-edited induced pluripotent stem cells (iPSCs) were used to recapitulate the phenotypes of patients with BrS and to study the underlying mechanism (Liang et al., 2016;Veerman et al., 2016;Ma et al., 2018). iPSC-CMs from BrS patients who tested negative for mutations in the known BrSassociated genes do not exhibit clear cellular electrophysiological abnormalities (Veerman et al., 2016). Patient-specific iPSC-CMs with a double missense mutation (p.R620H and p.R811H) or with a deletion ( 1397) in SCN5A could recapitulate singlecell phenotype features of BrS, including the blunted I Na , an increased triggered activity, and an abnormal Ca 2+ handling (Liang et al., 2016).
Interestingly, a more recent study reported a remarked lower I Na density and an increased phase-1 repolarization at a slow pacing frequency (0.1 Hz) in BrS patient-specific iPSC-CMs with a compound SCN5A mutation (p.A226V and p.R1629X), leading to pro-arrhythmic action potential (AP) morphology (Ma et al., 2018). The dominant repolarizing currents in human CMs are the transient-outward potassium current (I to ) conducted by K V 4.2 and K V 4.3 channels which contribute to I to,fast , and by K V 1.4 which accounts for I to,slow and the rapidly and slowly activating delayed rectifier potassium currents (I Kr and I Ks ) conducted by HERG and K V LQT1/mink channels, respectively (Nerbonne and Kass, 2005). Previous study identified gain-of-function mutations in KCND3 encoding K V 4.3 in patients with BrS as a result of the increased I to (Giudicessi et al., 2011). These studies suggest that the underlying mechanism for BrS might be different in BrS patients with different mutations and both depolarization and repolarization mechanisms might coexist.
In the present study, we utilized iPSC-CMs from two BrS patients (BrS1-CMs and BrS2-CMs) with a heterozygous SCN5A point mutation p.S1812X causing a premature termination codon. We show that BrS-CMs from both patients reveal a reduced I Na and a delayed sodium channel activation, which result in a slowed AP upstroke velocity and a reduced field potential (FP) conduction velocity (CV), similar to the phenotype observed in BrS patients with SCN5A loss-of-function mutations. Importantly, BrS-CMs also exhibited the enhanced I to and an augmented I CaL (L-type calcium current) window current, which may contribute to the arrhythmic phenotype. Exposure of BrS-CMs to cilostazol and milrinone, two clinically used phosphodiesterase (PDE) blockers, lowered I to and alleviated arrhythmic activity, without affecting on the CV. These data indicate that the electrophysiological mechanisms underlying arrhythmia in BrS-CMs may involve an impaired coordination of I Na , I to , and I CaL and cilostazol and milrinone can reduce the arrhythmia but have no effect on the CV.

Generation of iPSCs
The study was approved by the Institutional Ethics Committee of University Medical Center Göttingen (approval number 21/1/11) and of Technical University of Dresden (approval number EK 422092019) and carried out in accordance with the approved guidelines. Bone marrow aspirate (10 mL) and peripheral blood (15 mL) were taken from the BrS1 and BrS2 patients, respectively, after obtaining written informed consent. For the Ctrl1, bone marrow aspirate of a 45-year-old female (without known cardiac disease) left over from diagnostic purposes was used. For the Ctrl2, skin biopsy was taken. Bone marrowderived mesenchymal stem cells, peripheral blood mononuclear cells, and skin fibroblasts were cultured and reprogrammed into iPSCs using the STEMCCA lentivirus (BrS1-iPSCs, Ctrl1-iPSCs, Ctrl2-iPSCs) or Sendai virus system (BrS2-iPSCs), as described previously (Streckfuss-Bomeke et al., 2013;Cyganek et al., 2018). The generated iPSCs from the BrS patients and the healthy control were adapted to feeder-free culture conditions and cultivated on tissue culture plates pre-coated with Geltrex R (Thermo Fisher Scientific) in Essential 8 TM medium (Thermo Fisher Scientific).

Karyotype Analysis
Karyotyping of BrS-and Ctrl-iPSCs was performed using standard methodology. Prior to karyotyping, the iPSCs were cultured on feeder-free conditions. The cells were treated with 100 ng/mL colcemid (Thermo Fisher Scientific) for 16 h. The metaphases were prepared according to standard methodology and subjected to Giemsa staining before analysis with a light microscope (Zeiss Axio Imager.M2). The karyotype was analyzed and documented by using the software Case Data Manager 6.0 (Applied Spectral Imaging).

Genomic Sequencing
The genomic DNA of Ctrl-and BrS-iPSCs was isolated and purified using the automated Maxwell R 16 cell DNA purification kit (Promega) according to the manufacturer's instructions. DNA sequence comprising the SCN5A point mutation site in the generated Ctrl-and BrS-iPSCs was amplified by PCR. The primer set used is given in Supplementary Table 3. The PCR products were purified using the QIAquick R gel extraction kit (Qiagen) according to the manufacturer's instructions and sequenced by a commercial sequencing facility (Seqlab, Göttingen).

Spontaneous in vitro Differentiation
For spontaneous in vitro differentiation experiments, the cells were digested with 200 U/mL collagenase type 4 (Worthington Biochemicals) into big clusters and cultivated in suspension to form multicellular aggregates, known as embryoid bodies (EBs) for 8 days. The differentiation medium contained IMDM (Thermo Fisher Scientific), 20% fetal bovine serum, 1x non-essential amino acid (Thermo Fisher Scientific), and monothioglycerol (450 µM, Sigma-Aldrich). EBs were thereafter plated onto 0.1% gelatin-coated tissue culture dishes, cultivated up to 1 month and used for further analyses.
Total mRNA was isolated as described in the SV Total RNA Isolation System with on-column DNase digestion (Promega). DNase-treated RNA (200 ng) was used for the first-strand cDNA synthesis by using Murine Leukemia Virus Reverse Transcriptase and Oligo d(T) 16 (Thermo Fisher Scientific). One-tenth of the cDNA reaction was taken as PCR template and amplified for 25-35 cycles, denaturation at 95 • C for 15 s, annealing at 52-64 • C for 15-30 s depending on the primer melting temperatures, and elongation at 72 • C for 30 s. The primer sequences (forward and reverse), annealing temperatures, and the cycles used for RT-PCR analyses are listed in Supplementary Table 3.
Quantitative real-time PCR were carried out using SsoAdvanced Universal SYBR Green PCR super mix, Hard-Shell Optical 96-well plates, and the BIO-RAD CFX96 Real-time PCR system (Bio-Rad) according to the manufacturer's instructions. Initial denaturation was performed at 95 • C for 30 s, followed by 45 amplification cycles of 15 s at 95 • C and 1 min elongation at 60 • C. A melting curve was obtained for 65-95 • C. PCR products were confirmed by the melting temperature and agarose gel electrophoresis. Expression was quantified using Ct method with RPL32 as a reference. Sequences for RPL32, TNNT2, KCND2/3, and KCNA4 real-time PCR analyses are listed in Supplementary Table 3.

Teratoma Formation Analysis
To investigate their differentiation potential, the generated iPSCs were differentiated into derivatives of the three germ layers in vivo as previously described (Streckfuss-Bomeke et al., 2013). Briefly, the undifferentiated iPSCs were subcutaneously injected into 8-week-old recombination activating gene 2 and gamma C deficient (RAGC) mice. The mice were sacrificed 8-12 weeks after injection and the resulting teratomas were examined with hematoxylin and eosin staining.

Directed Differentiation of iPSCs Into CMs
All iPSCs were cultured on Geltrex R in Essential 8 TM medium as a monolayer with 80-100% confluence before the initiation of differentiation. At day 0, cells were cultured in RPMI1640 medium (Thermo Fisher Scientific) with Glutamax and HEPES (Thermo Fisher Scientific), 0.5 mg/mL human recombinant albumin, and 0.2 mg/mL L-ascorbic acid 2-phosphate and treated with 4 µM CHIR99021 (Merck Millipore), a highly selective inhibitor of GSK3β. After 48 h, CHIR99021 was removed and the cells were treated with 5 µM Wnt antagonist II (IWP2, Merck Millipore). Another 48 h later, the medium was replaced without IWP2. From day 8 on, the medium was replaced by RPMI1640 with Glutamax and HEPES containing 1x B27 supplement with insulin (Thermo Fisher Scientific). First beating CMs appeared around day 8. At day 25, the CMs were first treated with collagenase 2 (200 U/mL, Worthington) for 30 min so that beating clusters could be detached from the culture dish. These clusters were collected and digested with 0.25% trypsin/EDTA solution (Thermo Fisher Scientific) into single cells, which were replated onto Geltrex R -coated culture dishes as a monolayer. CMs were kept in culture for 3 months for maturation and further electrophysiological and gene and protein expression analyses.

Allele-Specific Expression Analysis of the SCN5A Gene
For the allele-specific expression analysis of SCN5A in BrS-CMs, the mRNA was sequenced with primers containing specific barcodes using the Ion Torrent semiconductor sequencing system (Thermo Fisher Scientific). The genomic DNA was used as control.
Total mRNA was isolated and reverse transcribed into cDNA. The primer set (forward: 5'-GAG AGC ACC GAG CCC CTG AGT GAG G-3' and reverse: 5'-CAC CAT GGG CAG GTC CAT GTT GAT G-3') was used to amplify the region containing the mutation site. The PCR product was diluted 1/200 and used for a second PCR run with 12 cycles using the same forward primer and different custom-designed reverse primers (Sigma-Aldrich), which contained a unique barcode to distinguish individual samples. The DNA concentration of each sample was measured with Qubit 2.0 Fluorometer (Thermo Fisher Scientific) and 250 ng of each sample was pooled and electrophoretically separated on a 2% agar gel. The specific product was extracted from the gel in a QIAcube system using the QIAquick R gel extraction kit (Qiagen) and subsequently purified with Agencourt AMPure XP PCR purification kit (Beckman Coulter) according to the manufacturer's instructions and eluted in low TE buffer (Thermo Fisher Scientific). The DNA quantity was determined by real-time PCR analysis in a 384-well plate using the GeneRead Library Quant kit (Qiagen). 400 µL of 10 pmol PCR product was used for clonal amplification onto Ion Sphere particles accomplished by an emulsion PCR in the Ion OneTouch (Thermo Fisher Scientific) system according to the manufacturer's instructions. The Ion Sphere particles coated with the amplified template DNA were applied to an Ion Torrent sequencing chip and placed on the Ion Personal Genome Machine (Thermo Fisher Scientific) for sequencing.

Immunofluorescence Staining
Undifferentiated iPSCs or EB cultures at different differentiation stages as well as 3-month-old CMs were fixed with 4% paraformaldehyde (Sigma-Aldrich) in PBS at room temperature for 20 min and then blocked with 1% bovine serum albumin (Thermo Fisher Scientific) at 4 • C overnight. The primary and respective secondary antibodies used are listed in Supplementary  Table 4. Nuclei were counter-stained with 4-6-diamino-2phenylindole (DAPI, 0.4 µg/mL, Sigma-Aldrich). Samples were mounted with Fluoromount-G mounting medium (eBioscience). Antibodies staining against non-transmembrane proteins were additionally treated with 0.1% Triton X-100 (Sigma-Aldrich) in PBS to permeabilize the cell membrane. Fluorescent images were captured with a fluorescence microscope (Zeiss Observer.Z1 or Axio Imager.M2). For some images, the ApoTome modus (Zeiss) was used.

Western Blot
For analysis of protein levels, Ctrl-and BrS-CMs were solubilized in 0.5% digitonin, 1.5% NP-40, and 1.0% Triton X 100 solubilization buffer for 35 min at 4 • C. Protein concentrations were determined with a BCA assay kit. Equal amounts of protein were loaded on biphasic 6 and 10% sodium dodecyl sulfate polyacrylamide gel (SDS-PAGE) and run at low voltage. After blotting to PVDF membranes, membranes were incubated with Na V 1.5, Cx43, Ca V 1.2, K V 4.3, and cTnT antibodies overnight. The primary and respective secondary antibodies used for Western blots are listed in Supplementary Table 4. Western blots were developed with a peqlab Fusion FX Vilber Lourmat camera and analyzed using the FusionCapt Advance 16.12 software.

Patch-Clamp
Whole-cell clamp was used to measure membrane potential (current-clamp configuration), I Na , late I Na , I to , and I CaL (voltageclamp configuration), as previously described (Wagner et al., 2006;Streckfuss-Bomeke et al., 2013). All experiments were conducted with 3-month-old Ctrl-and BrS-CMs at room temperature. The pipette and extracellular solutions for various kinds of voltage and current patch-clamp recordings are listed in Supplementary Table 5.
Spontaneous APs were recorded immediately after rupture with the EPC10 amplifier (HEKA Elektronik) using the Patchmaster software (HEKA Elektronik). No current was injected into the cells. Signals were filtered with 2.9 and 10 kHz Bessel filters. At least 20 spontaneous APs in a row were analyzed and averaged using LabChart Pro software (ADInstruments) to determine V max , APA, and RMP.
For current measurements, all recordings started at least 1 min after rupture with the EPC10 amplifier using the Patchmaster software. Signals were filtered with 2.9 and 10 kHz Bessel filters and recorded with an EPC10 amplifier. Membrane capacitance (C m ) and series resistance (R s ) were compensated automatically after rupture, and currents were normalized to C m .
The pulse stimulation of sodium current recording was picturized in inset of Figures 3B,D-F. The current-voltage (I-V) relationship was determined by increasing the voltage stepwise from −95 mV to +5 mV in 5 mV steps at a holding potential of −100 mV. Each pulse lasted 50 ms. Steady-state inactivation was measured by a two-pulse protocol at a holding potential of −100 mV. The voltage of the first conditioning pulse (500 ms) was increased stepwise from −120 to −20 mV in 5 mV steps. The second pulse to −20 mV was used to measure I Na and normalized to the maximum I Na (which was usually after the first pulse). For the measurement of the recovery from inactivation of sodium channels, a two-pulse protocol was used, with an increasing delay between the two pulses ranging from 1 to 165 ms. The first conditioning pulse (1000 ms) is used to induce sodium channel inactivation and the second pulse is used to measure I Na generated by sodium channels that recover from inactivation. The I Na from the second pulse was normalized to the maximal I Na initiated by the conditioning pulse.
To study the effect of milrinone on I Na in Ctrl-and BrS-CMs, an automated patch-clamp system (Patchliner Quattro, Nanion Technologies GmbH) was used with low resistance NPC-16 chips and I Na was recorded at room temperature. iPSC-CMs were dissociated into single cells for automated patch-clamp analysis, as previously described (Li et al., 2019). The liquid junction potentials and series resistance were not compensated. The holding potential was −100 mV. I Na was recorded using voltage steps from −90 to +70 mV for 20 ms in 5 mV steps at an interval of 2000 ms. Currents were sampled at 25 kHz and low-pass-filtered at 2.9 kHz.
The pulse protocols for I to current were shown in inset of Figures 4B,C,F. The current-voltage (I-V) relationship was determined by increasing the voltage stepwise from −40 to +60 mV in 10 mV steps at a holding potential of −90 mV with a 20 ms prepulse at −35 mV to inactivate sodium channel. Each pulse lasted 600 ms. Steady-state inactivation was measured by using a two-pulse protocol at a holding potential of −90 mV. The voltage of the first conditioning pulse (600 ms) was increased stepwise from −90 to +40 mV in 10 mV steps after a 20 ms prepulse at −35 mV to inactivate sodium channels. The second pulse of 600 ms at +60 mV was used to measure I to and normalized to the maximum I to . For the measurement of the recovery from inactivation of I to , a two-pulse protocol was used, with an increasing delay between the two pulses ranging from 0 to 10,240 ms. The pulse was depolarized from −90 to +40 mV and last for 600 ms. The I to from the second pulse was normalized to the maximal I to initiated by the conditioning pulse. The pulse intervals were 15 s for all the stimulations.
The impulse protocol for I CaL was shown in inset of Figures 4H,I,K, as previously described (Luo et al., 2020). The current-voltage (I-V) impulse depolarized from −90 to −80 mV then every 10 mV as step to +70 mV with a duration of 600 ms. Steady-state inactivation was measured by using a two-pulse protocol at a holding potential of −90 mV. The voltage of the first conditioning pulse (600 ms) was increased stepwise from −90 to +10 mV in 10 mV steps. The second pulse of 600 ms at 0 mV was used to measure I CaL and normalized to the maximum I CaL . For the measurement of the recovery from inactivation of I CaL , a twopulse protocol was used, with an increasing delay between the two pulses ranging from 0 to 1460 ms. The pulse was depolarized from −90 to 0 mV and last for 600 ms. The I CaL from the second pulse was normalized to the maximal I CaL initiated by the conditioning pulse. The pulse intervals were 3 s for all the stimulations.
For the conductance (G/G max ) calculation, a custom builtin formula and add-in modular in excel were used. The same fit functions were used for analyzing the kinetics of I Na and I CaL . Steady-state activation and inactivation curves were fitted with a standard Boltzmann function: where V 1/2 is the half-maximal voltage of steadystate (in)activation and k ∞ is the slope factor of the voltage dependence of (in)activation. The development of the recovery from inactivation was fitted with single exponential function: where A is the amplitude and τ rec the time constant of recovery from inactivation. Persistent I Na was measured by integrating the current between 50 and 450 ms of a 1000 ms pulse from a holding potential of −100 to −20 mV. The probability of I CaL window current was calculated from steady-state (in)activation parameters through the following equation as previously described: (Huang et al., 2011).

Multielectrode Array
The digested iPSC-CMs were seeded on Geltrex R -coated MEA chips. The final concentration was approximately 200,000 iPSC-CMs in 50 µL RPMI1640 medium supplemented with 1x B27. The medium was replenished to 1 mL after seeding for around 30 min. In this study, 60MEA200/30iR-Ti-gr chips were used (Multichannel Systems). The medium was changed one day after digestion and thereafter every two days until day 6, which performed the recording. All the recordings were carried out at 35-37 • C by using Cardio 2D software (Multichannel Systems). The sample rate was 10,000 Hz. The spontaneous beating frequency and inter-beat interval were automatically generated by Cardio 2D. At least five stable continuous CV values were averaged from spontaneous beatings. For FP metrics [peakto-peak amplitude (PPA), peak-to-peak duration (PPD), and peak-to-peak slope (PPS)] analysis, the averaged measurements (3 min) of the electrodes (Pin numbers: 22,33,44,55,66,and 77) were analyzed by Cardio 2D+ software. The rate of arrhythmia was quantified as standard deviation (SD) of the inter-beat interval, which was normalized to averaged inter-beat intervals.

Statistics
Data are presented as the mean ± standard error of the mean (SEM). Statistical analysis was performed with GraphPad Prism 7 using the two-tailed unpaired Student's t-test to compare differences between two independent groups, one-way ANOVA followed by Tukey's post hoc test to compare more than two groups, and the paired Student's t-test to compare differences between two dependent groups or the two-way ANOVA test for comparison of more groups and conditions. Results were considered statistically significant when the P value was <0.05 ( * P < 0.05; #P < 0.01; §P < 0.001).

Generation and Characterization of iPSCs
We recruited two BrS patients, a 50-year-old male (BrS1) and his biological sister (BrS2) in the study. The ECG recordings of both patients at rest displayed the BrS-typical pattern, a covedtype ST-segment elevation in the right precordial leads V 1 and V 2 and a more saddleback pattern in V 3 followed by a negative T wave (Supplementary Figure 1A). In addition, both patients revealed first-degree AV block. In long-term ECG recordings of patient BrS1, multiple episodes of polymorphic ventricular tachycardia appeared during stair climbing (Supplementary Figure 1B). Both patients received an ICD for the treatment. Genetic screening for possible mutations in several cardiacspecific genes (KCND3, KCNQ1, HERG, KCNE1, and SCN5A) revealed the heterozygous SCN5A point mutation C > A at position c.5435 in both patients (Schulze-Bahr et al., 2003). This non-sense mutation results in a premature termination codon causing the protein truncation at amino acid 1812 (p.S1812X). Genetic tests of their family members identified the mother as an asymptomatic carrier of the mutation, showing first-degree AV block, but no BrS-specific ECG pattern at rest. A brother died from sudden cardiac death at the age of 15. The father and another brother do not carry the mutation and show no BrS phenotype (Supplementary Figure 1C).
We generated iPSCs from these two patients (BrS1 and BrS2) by overexpression of the four Yamanaka factors SOX2, KLF4, OCT4, and c-MYC in somatic cells (Supplementary Figure 2). Three clones per individual were selected for further characterization (BrS1.1-3 and BrS2.1-3). In addition, we included control iPSC lines (Ctrl1.1-3 also known as MSC3-iPS1-3; and Ctrl2.1 and 2 also named as FB2-iPS1 and 2) from two independent healthy donors Ctrl1 and Ctrl2, which were described in our previous studies (Streckfuss-Bomeke et al., 2013;Cyganek et al., 2018). All generated BrS-iPSCs showed the typical morphology similar to Ctrl-iPSCs, were positive for pluripotency markers ( Supplementary Figure 2A), showed expression of endogenous pluripotency genes (Supplementary Figure 3A), and revealed normal chromosome numbers in more than 90% of cells (Supplementary Figure 2C). Genomic sequencing confirmed the presence of the SCN5A mutation c.C5435A (p.S1812X) in BrS1-and BrS2-iPSCs and its absence in both Ctrl1-and Ctrl2-iPSCs (Supplementary Figure 2B). Furthermore, Ctrl-and BrS-iPSCs were differentiated into derivatives of all three embryonic germ layers in vitro (Supplementary Figures 3B,C) and developed teratomas in immune-deficient mice (Supplementary Figure 2D). To reduce the phenotypic variability among different cell lines, all three lines of every individual were used for the following studies.
Reduced Na V 1.5 and Cx43 Expression in BrS-CMs We differentiated Ctrl-and BrS-iPSCs into ventricular CMs using the directed differentiation method, as described previously (Cyganek et al., 2018). This method gave rise to the differentiation efficiency with more than 90% of cells positive for cTnT at day 25 after the initiation of differentiation. No significant difference in cardiac differentiation efficiency was observed between Ctrland BrS-iPSCs (data not shown). Immunofluorescence analyses using antibodies against the cardiac-specific proteins cTnT and α-actinin revealed well-organized cross-striations in both 3month-old BrS-and Ctrl-CMs (Figures 1A-C). To study whether the p.S1812X mutation affected the cellular localization of Na V 1.5 in BrS-CMs, we applied an antibody that detects both fulllength and truncated proteins of Na V 1.5. We found a different distribution of Na V 1.5 in Ctrl-and BrS-CMs ( Figure 1B). Whereas equal distribution of Na V 1.5 along cell membrane was clearly observed in Ctrl-CMs, Na V 1.5 appeared irregular and intermittent on the cell surface. Similar to the findings of others (Malan et al., 2011;Davis et al., 2012), Na V 1.5 was also expressed in a diffused or a fine granular-like pattern in the cytosol of both Ctrl-and BrS-CMs. In some Ctrl-and BrS-CMs, we observed that Na V 1.5 was localized in a striated pattern, mostly overlapped with α-actinin (Figure 1B), as described in adult CMs (Mohler et al., 2004;Xi et al., 2009;Shy et al., 2014). The expression of connexin 43 (Cx43) was detected at the intercellular gap junctions in both Ctrl-and BrS-CMs (Figures 1C,D), indicating a cellto-cell coupling. However, the co-staining of Cx43 with Na V 1.5 revealed the strong overlap and interaction of Cx43 with Na V 1.5 in Ctrl-CMs, but less in BrS-CMs ( Figure 1D).
Western blot analyses using an antibody that only detects the full-length protein of Na V 1.5 showed a significantly reduced expression of Na V 1.5 in BrS-CMs by 69.5% compared to Ctrl-CMs (P = 0.0093; Figures 1E,F). Interestingly, Cx43 expression in BrS-CMs was also reduced by 34%, but not significantly, compared to Ctrl-CMs (P = 0.1443; Figures 1E,G). Furthermore, we assessed the expression of Ca V 1.2 and K V 4.3 in Ctrl-and BrS-CMs and did not observe significant differences between Ctrland BrS-CMs (Figures 1E,H,I). Analysis of the allele-specific SCN5A expression in BrS1-and BrS2-CMs revealed the balanced expression of SCN5A between the wild-type and mutated alleles (Supplementary Figure 3D).

Conduction Slowing in BrS-CMs
Clinically, both BrS patients experience severe arrhythmias and reveal first-degree AV block in the ECGs. To investigate whether the arrhythmic predisposition is exhibited in the monolayer culture of iPSC-CMs, we applied multi-electrode array (MEA) to compare the conduction properties of spontaneous electrical excitations in 3-month-old Ctrl-and BrS-CMs (Figure 2). The differences between Ctrl-and BrS-CMs in the FP metrics PPD (ms), PPA (mV), and PPS (V/s) were distinct after enlargement (Figure 2A). The beating frequency of BrS1-CMs (0.46 ± 0.07 Hz, n = 23) and BrS2-CMs (0.44 ± 0.02 Hz, n = 19) was comparable with Ctrl-CMs (0.45 ± 0.03 Hz, n = 37, Figure 2C). However, BrS-CMs revealed a smaller PPA and PPS but a prolonged PPD when compared to Ctrl-CMs (Figures 2D-F). In Figure 2B, typical excitation patterns and local activation times in reference to the trigger point were shown for both Ctrl2-and BrS1-CMs. The excitation of Ctrl2-CMs spread from one corner (the lower left corner as the trigger point) of the MEA to the other in around 10 ms ( Figure 2B) whereas BrS1-CMs needed around 45 ms for the same distance ( Figure 2B), indicating pronounced conduction slowing in BrS-CMs when compared to Ctrl-CMs. Quantitative analysis revealed that the CV of Ctrl-CMs was 15.47 ± 1.05 cm/s (n = 37), whereas 4.7 ± 0.44 and 4.4 ± 0.49 cm/s were observed in BrS1-CMs (n = 23, P < 0.001) and BrS2-CMs (n = 19, P < 0.001), respectively ( Figure 2G). This dramatic difference can be vividly observed in our recorded movies (Supplementary Videos 1,  2). Moreover, BrS-CMs exhibited increased inter-beat interval variability compared to Ctrl-CMs, as demonstrated by the SD of the inter-beat interval normalized to the mean of the inter-beat intervals ( Figure 2H).

Reduced Peak I Na and Delayed Steady-State Activation of Sodium Channels in BrS-CMs
To assess the effect of the SCN5A mutation p.S1812X on the electrophysiological properties of BrS-CMs, I Na density and gating properties of sodium channels were compared in 3-month-old BrS-CMs versus Ctrl-CMs using the wholecell voltage-clamp technique (Figure 3). We observed no differences in the current density, voltage dependence, and gating properties of sodium channels between Ctrl1-and Ctrl2-CMs (Supplementary Figure 4). For a better overview and comparison of the data, we pooled the data of Ctrl1-and Ctrl2-CMs, as shown in Figure 3. Representative examples of I Na recordings were shown in Figure 3A. Compared to the Ctrl-CMs, the I Na densities were significantly reduced in BrS1-and BrS2-CMs ( Figure 3B). The maximal peak I Na in Ctrl-CMs (n = 60) at −35 mV showing −38.3 ± 1.8 A/F was reduced by more than 50% in BrS-CMs showing −18.1 ± 1.5 A/F in BrS1-CMs (n = 37, P < 0.001) and −16.7 ± 1.4 A/F in BrS2-CMs (n = 19, P < 0.001). There was no significant difference in the currentvoltage relationship and the I Na densities between BrS1-and BrS2-CMs ( Figure 3B). However, current-voltage relationships normalized to the maximum peak I Na showed a shift of peak  , and normalized SD of inter-beat interval (H) in Ctrl-CMs, BrS1-CMs, and BrS2-CMs. Gray triangle represents Ctrl1 and gray rectangle represents Ctrl2. Data are mean ± SEM. One-way ANOVA followed by Tukey's post hoc test was used for statistical analysis: * P < 0.05, §P < 0.001. PPA, peak-to-peak amplitude; PPD, peak-to-peak duration; PPS, peak-to-peak slope; CV, conduction velocity; SD, standard deviation. (G) Average current-voltage relationships of the I Na in Ctrl-CMs with and without 2.5 or 5 µM TTX treatment. Protocol is shown as inset. (H) Average steady-state voltage dependence of activation in Ctrl-CMs treated with and without 2.5 µM TTX. Protocol is shown as inset. (I) V max in Ctrl-CMs with and without 2.5 or 5 µM TTX treatment. Data are presented as mean ± SEM. Two-way repeated measures ANOVA (B-H) and one-way ANOVA followed by Tukey's post hoc test were used for statistical analysis: * P < 0.05, #P < 0.01, §P < 0.001. current toward the positive direction in BrS-CMs compared to Ctrl-CMs ( Figure 3C).
To study the changes in channel kinetics, we compared the steady-state (in)activation and the recovery from inactivation of the sodium channel in BrS-CMs versus Ctrl-CMs. The steadystate activation curve was significantly shifted in a depolarizing direction in BrS-CMs compared to Ctrl-CMs ( Figure 3D). V 1/2 was −39.1 ± 0.2 mV for Ctrl-CMs (n = 60), −32.4 ± 0.2 mV for BrS1-CMs (n = 37, P < 0.001), and −33.7 ± 0.3 mV for BrS2-CMs (n = 25, P < 0.001). Slope factor k ∞ was 6.1 ± 0.2 mV for Ctrl-CMs, significantly smaller than 6.9 ± 0.2 mV for BrS1-CMs (P < 0.001) and 6.7 ± 0.2 mV for BrS2-CMs (P < 0.05) (Supplementary Table 1). The steady-state inactivation of sodium channels in BrS-CMs did not differ significantly from Ctrl-CMs ( Figure 3E and Supplementary Table 1). Next, we determined the fraction (A) of sodium channels recovered from an inactivation pulse (1000 ms at −20 mV) and the time constant (τ rec ) for this recovery. We found a slight but not significant slowing of the recovery from the inactivation in BrS-CMs versus Ctrl-CMs ( Figure 3F and Supplementary Table 1). However, no difference in persistent I Na was detected in our study (Supplementary Figures 4F,G).
To assess whether the reduced I Na density in BrS-CMs is related to the positive shift of the steady-state activation, we treated Ctrl-CMs with the Na + channel blocker tetrodotoxin (TTX; 2.5 and 5 µM, Figures 3G-I). The treatment with 5 µM TTX resulted in a robustly reduced I Na , decreasing maximal I Na to 5% of the basal level ( Figure 3G). The peak I Na density at −35 mV in Ctrl-CMs treated with 2.5 µM TTX (−9.7 ± 1.2 A/F, n = 14) showed a significant reduction compared to nontreated Ctrl-CMs (−38.3 A/F, n = 60) (Figure 3G). We also observed a significant positive shift in the steady-state activation, as demonstrated by the change in V 1/2 in Ctrl-CMs treated with 2.5 µM TTX (−32.7 ± 0.3 mV, n = 14) compared to non-treated Ctrl-CMs (−39.1 ± 0.2 mV, n = 60, P < 0.001) ( Figure 3H). These data suggest that the reduced I Na density in BrS-CMs is associated with the positive shift of the steady-state activation.
To study whether the blockage of Na + channel by TTX can mimic the FP properties observed in BrS-CMs, we treated Ctrl-CMs with 2.5 µM TTX for 10 min. We found that the application of TTX in Ctrl-CMs led to a significantly reduced PPA, PPS, and CV, but a prolonged PPD (Supplementary Figure 5). In addition, TTX reduced beating frequency (Supplementary Figure 5). Taken together, these data indicate that BrS-CMs recapitulate the conduction abnormalities seen in patients with loss-of-function sodium channel mutations, and the conduction slowing can be evoked in Ctrl-CMs by TTX and is associated with the diminution of I Na .

Augmented I to Current and Altered Window Current of I CaL in BrS-CMs
Previous studies reported that in the presence of weak I Na , the unopposed outward K + current I to cause accentuation of the AP notch in the right ventricular epicardium, resulting in accentuated J wave and ST-segment elevation associated with the Brugada pattern (Szel and Antzelevitch, 2014). To test whether I to is pivotal in generating the disease phenotype in BrS, we recorded I to from single iPSC-CMs (Figure 4). The representative I to traces of Ctrl1-CMs and BrS1-CMs were shown in Figure 4A. Surprisingly, the BrS1-CMs and BrS2-CMs exhibited significantly larger I to in comparison to Ctrl-CMs ( Figure 4B). The maximum peak currents at +60 mV in BrS1-CMs (14.91 ± 2.27 A/F, n = 27, P < 0.001) and BrS2-CMs (12.14 ± 1.42 A/F, n = 38, P < 0.001) were 2.4 and 1.9 times bigger than those in Ctrl-CMs (6.14 ± 0.85 A/F, n = 73) ( Figure 4B). When the cells were treated with 4-aminopyridine (4-AP), an I to inhibitor, 60.9% reduction of the currents was observed in Ctrl-CMs (from 5.29 ± 0.67 to 2.07 ± 0.24 A/F, paired, n = 29; Figure 4D), which is comparable to 65.9% reduction in BrS-CMs (from 13.97 ± 1.8 to 4.8 ± 0.84 A/F, paired, n = 34; Figure 4E), indicating the measured currents are the 4-AP-sensitive I to . Nevertheless, we did not detect any significant differences in V 1/2 for the steadystate inactivation between Ctrl-CMs (−30.41 ± 1.6 mV, n = 14) and BrS-CMs (−30.42 ± 0.72 mV, n = 24) (Figure 4C and Supplementary Table 1). Recovery from the inactivation of I to in Ctrl-CMs showed fast (fast τ rec , 279.7 ± 121.1 ms) and slow phases (slow τ rec , 2854 ± 581.7 ms), which were comparable to the fast (fast τ rec , 211.9 ± 135.8 ms) and slow phases (slow τ rec , 2773 ± 462.6 ms) in BrS-CMs, respectively ( Figure 4F and Supplementary Table 1). Using real-time PCR technique, we assessed the expression of I to -related genes KCND2/3 and KCNA4. It turned out the expressions of KCND2 and KCND3 were significantly higher in BrS1-and BrS2-CMs when compared to Ctrl-CMs (Figures 4G,H). However, no difference regarding the expression of KCNA4 was found between Ctrl-and BrS-CMs ( Figure 4I). The TNNT2 expression was normalized to reference gene RPL32 like the other three genes (Figure 4J).
We further checked the I CaL and did not detect significant differences in the peak I CaL at 0 mV in BrS1-CMs (−13.99 ± 1.18 A/F, n = 18) or BrS2-CMs (−13.71 ± 0.8 A/F, n = 32) compared to Ctrl-CMs (−12.88 ± 0.66 A/F, n = 46) (Figures 4K,L). The curve of the steady-state activation was shifted to more depolarizing voltage in BrS2-CMs (V 1/2 = −12.51 ± 0.50 mV, n = 32, P < 0.001) but not in BrS1-CMs (V 1/2 = −14.45 ± 0.84 mV, n = 19, P > 0.05) when compared to Ctrl-CMs (V 1/2 = −15.54 ± 0.38 mV, n = 46) ( Figure 4M and Supplementary Table 1). The steady-state inactivation curves were also significantly shifted toward more depolarizing voltage for BrS1-CMs (V 1/2 = −36.83 ± 0.38 mV, n = 18, P < 0.001) and BrS2-CMs (V 1/2 = −34.61 ± 0.40 mV, n = 31, P < 0.001) when compared to Ctrl-CMs (V 1/2 = −39.18 ± 0.22 mV, n = 46) ( Figure 4M and Supplementary Table 1). Notably, the shifts in the voltages for half-(in)activation V 1/2 and a slight increase in slopes of (in)activation k ∞ resulted in the significant increases in the window current (area under the curves in Figure 4N). After calculating with a window current probability formula, we found the probabilities of the I CaL window current increased 1.59 times for BrS1-CMs and 1.56 times for BrS2-CMs when compared to Ctrl-CMs ( Figure 4N). As window currents are steady-state currents, bigger window currents indicate the increased Ca 2+ influx that may contribute to the formation of afterdepolarization and arrhythmia in BrS-CMs. The maximum ratio of peak I CaL [P 2 /P 1 ] of the recovery from the inactivation for Ctrl-CMs is comparable to those for BrS1-CMs and BrS2-CMs ( Figure 4O and Supplementary Table 1). Taken together, our results demonstrate that BrS-CMs exhibit distinct anomalies not only in Na + channel function but also in K + and Ca 2+ channel functions compared to Ctrl-CMs, which may result in the abnormal AP phenotype.

Reduced V max and Increased Irregular Repolarization in APs of BrS-CMs
According to the following stringent criteria, the majority of 3month-old iPSC-CMs (81%) reveal ventricular-like APs by using the C-clamp mode of patch-clamp technique (Figure 5 and Supplementary Table 2). Ventricular-like APs exhibit a relatively negative resting membrane potential (RMP; <−60 mV), a rapid maximal upstroke velocity (V max ), a prominent plateau phase, and an AP amplitude (APA) over 95 mV. Atrial-like APs reveal similar properties as ventricular-like APs but lack the plateau phase. Nodal-like APs exhibit a more positive RMP (≥−55 mV), The pulse protocol is shown as inset. Data are presented as mean ± SEM. Two-way repeated measures ANOVA was used for I to and I CaL statistical analysis. One-way ANOVA was used for gene expression analyses. * P < 0.05, #P < 0.01, §P < 0.001. a slow V max (≤5 V/s), and an APA < 85 mV. Overall, no significant differences in RMP and APA were observed in BrS-CMs compared to Ctrl-CMs (Supplementary Table 2). Since sodium channels are responsible for the fast upstroke of the atrial and ventricular APs, as expected, the V max was significantly slower in spontaneously beating ventricular-and atrial-like BrS1-and BrS2-CMs compared to Ctrl-CMs (Figure 5A). Similarly, Ctrl-CMs after the treatment with TTX showed a slower V max (Figure 3I). We also observed a significantly slower V max in BrS2-CMs than in BrS1-CMs (Figure 5B), which is consistent with the more pronounced I Na reduction in BrS2-CMs than in BrS1-CMs. Additionally, we found a pronounced AP notch in BrS-CMs compared to Ctrl-CMs (Figure 5C), which is in line with the augmented I to density in BrS-CMs. Notably, we detected prominent irregularities in spontaneous AP recordings in BrS-CMs, including early afterdepolarization (EAD), EAD-triggered activities, delayed afterdepolarization (DAD), and DAD-triggered activities ( Figure 5C). We observed the irregularities in 40% of BrS1-CMs (n = 43; P < 0.001, Fisher's exact test) and 42% of BrS2-CMs (n = 31; P < 0.001) compared to 11% of Ctrl-CMs (n = 74, Figure 5D). Taken together, these results demonstrate that BrS-CMs exhibit distinctly abnormal AP phenotype compared with Ctrl-CMs.

Assessment of Potential Pharmacotherapies Using BrS-CMs
Given that two clinically used PDE inhibitors cilostazol and milrinone have been shown to suppress the hypothermia-induced ventricular tachycardia/ventricular fibrillation by reversing the repolarization abnormalities (Gurabi et al., 2014), we tested their effects on our iPSC-CMs at therapeutically relevant concentrations (Szel et al., 2013). After 10 µM cilostazol treatment of BrS-CMs for half an hour, we observed a 50.9% reduction of I to in BrS1-CMs (from 14.91 to 7.3 A/F) and a 35.4% reduction in BrS2-CMs (from 12.14 to 7.8 A/F) at +60 mV pulse stimulation, whereas only a slight but not significant I to reduction of 14.4% was observed for the Ctrl-CMs (from 6.14 to 5.25 A/F) (Figures 6A-C). The similar pattern was found when we tested the treatment with 2.5 µM milrinone at the same condition: I to was significantly reduced from 14.91 to 9.5 A/F in BrS1-CMs (36.3% reduction) and from 12.14 to 7.3 A/F in BrS2-CMs (40% reduction), but only a 10.9% reduction in Ctrl-CMs (from 6.14 to 5.7 A/F) (Figures 6A-C).
Moreover, we assessed whether these two drugs can alleviate the arrhythmic beating degree of BrS-CMs. We found that the treatment of BrS-CMs with cilostazol led to the decrease of cells with arrhythmic beating from 40.5% (n = 74) to 21% (n = 14) (Figures 6D,E). We detected the similar effect of milrinone on BrS-CMs and the arrhythmic beating degree of BrS-CMs dropped to 17% (n = 12) (Figures 6F,G). Both cilostazol and milrinone treatment had no obvious effect on Ctrl-CMs (Figures 6D-G).
The most interesting thing is that no EAD-or EAD-triggered activities were observed in these AP recordings.
We further studied the effect of milrinone on I Na using the automated patch-clamp system and found no significant changes in I Na in both Ctrl-and BrS-CMs after 2.5 µM milrinone treatment for half an hour (Figures 6H-J). In line with these data, there were no significant effects of cilostazol and milrinone on the CV in Ctrl-, BrS1-, and BrS2-CMs (Figures 6K,L), but interbeat interval variability was slightly reduced (Supplementary Figures 4H-K).

DISCUSSION
In this study, we applied patient-specific iPSC-CMs carrying the non-sense mutation p.S1812X in SCN5A and investigated the pathophysiological phenotypes and possible pharmacological therapy. Our results demonstrate the ability of patient-specific BrS-CMs to recapitulate the loss-of-function of Na V 1.5 resulting from the truncation of SCN5A as evidenced by the significant reduction of peak I Na , the significantly slower of the steadystate activation and the 69% reduction of Na V 1.5 α-subunit protein expression. In line with these findings, we also observed the decreased AP upstroke velocity and the conduction slowing in BrS-CMs. Notably, we also found the increased I to and the augmented window current of I CaL in BrS-CMs when compared to Ctrl-CMs. We believe that the impaired interplay between I Na , I to , and I CaL may be involved in the generation of arrhythmia in BrS-CMs. Furthermore, we showed the positive effect of PDE inhibitors cilostazol and milrinone on the reduction of the arrhythmia in BrS-CMs.
In this study, we, for the first time, applied the MEA technology and demonstrated the conduction slowing in BrS-CMs. It is well known that the I Na and the electrically coupling degree of gap junctions are the key determinants of cardiac conduction (Kucera et al., 2002). We demonstrated that the I Na density was reduced in BrS-CMs, which is consistent with other observations in iPSC-CMs with BrS-related mutations in SCN5A (Liang et al., 2016;Ma et al., 2018). Sodium channels play an important role in phase 0 of the cardiac AP and determine the upstroke velocity (Satin et al., 2004). In BrS-CMs, V max was significantly lower compared to Ctrl-CMs, consistent with observed reduction in peak I Na density. These findings were also demonstrated in patient-specific iPSC-CMs harboring the SCN5A 1795insD mutation (Davis et al., 2012). Our findings that the I Na activation curve shifted toward more positive potential with an increased slope factor are in accordance with the later activation of sodium channels resulted from loss-of-function mutations in SCN5A as previously reported for BrS patients (Wilde and Amin, 2018). Additionally, the diminution of I Na and the altered gating properties are clearly associated with the expression reduction of full-length Na V 1.5 protein and the disrupted localization at the cell surface due to the non-sense mutation p.S1812X in SCN5A in BrS-CMs. Our data are in line with the previous study demonstrating that in HEK293 cells overexpressing the wild-type and the BrS-associated mutant p.V2016M SCN5A together, the surface expression of Na V 1.5 is reduced due to its disturbed interaction with SAP97 via a PDZ domain (the last three residues SIV of Na V 1.5), with a subsequent decrease of I Na (Shy et al., 2014).
In line with previous studies (Kucera et al., 2002;Edokobi and Isom, 2018), we found the co-localization of Na V 1.5 and Cx43 at the sites of cell-to-cell appositions in both Ctrl-and BrS-CMs, however, in BrS-CMs the co-localization of Na V 1.5 and Cx43 seems altered. Physical and functional interactions of sodium channel complex with gap junction and desmosomal components at intercalated discs in CMs are essential for cardiac excitability and conduction (Delmar, 2012;Cerrone and Delmar, 2014). Previous study demonstrated that loss of the gap junction protein Cx43 resulted in diminished Na V 1.5 expression at intercalated discs (Jansen et al., 2012). In our study, we observed a reduced expression of Cx43 in BrS-CMs, this brings up an open question: how the reduced expression of Na V 1.5 regulates the Cx43 expression, which should be clarified in future studies. Nonetheless, our data indicate that BrS-CMs with loss-of-function mutation p.S1812X in SCN5A reveal some pathophysiological phenotypes including the conduction slowing, supporting the "depolarization disorder" theory as the underlying mechanism of BrS.
Notably, we show that I to is present in 3-month-old iPSC-CMs and, to our surprise, the I to was significantly larger in BrS-CMs than in Ctrl-CMs. However, genetic analysis in both patients did not find mutations in KCND3, KCNQ1, HERG, and KCNE1 (Schulze-Bahr et al., 2003). The 4-AP sensitivity experiments confirmed that the currents we recorded were I to . Moreover, we observed a prominent AP notch in BrS-CMs. Previous studies showed that I to is pivotal in generating the disease phenotype in BrS (Di Diego et al., 2002;Giudicessi et al., 2011). The direct relationship between I to and BrS was first reported by applying the I to activator NS5806 in isolated canine ventricular wedge preparations. NS5806 could increase phase 1 and notch amplitude of the AP in the epicardium, but not in the endocardium, and accentuate J-wave in the ECG (Calloe et al., 2009). It is well accepted that the orderly sequence of repolarization in the heart is linked to heterogeneous distribution of the fast component of I to in right ventricle epicardium and endocardium. The highest density of I to is seen in epicardial CMs, whereas the lowest density is observed in endocardial CMs (Litovsky and Antzelevitch, 1988;Wettwer et al., 1994;Costantini et al., 2005). The changes in repolarization might lead to the socalled phase 2 re-entry and polymorphic ventricular tachycardia. One limitation of our study is that the in vitro BrS-CM model cannot recapitulate the in vivo heterogeneous distribution of I to in right ventricular epicardium and endocardium. Nonetheless, our data demonstrate that the increased I to in BrS-CMs is associated to the disease phenotype, and are in line with the previous studies (Delpon et al., 2008;Giudicessi et al., 2011;Perrin et al., 2014). Gain-of-function mutations in KCND3 encoding K V 4.3 were found in patients with BrS, leading to the loss of the AP dome in the modified Luo-Rudy II AP model as a result of the increased I to (Giudicessi et al., 2011). A mutation in KCNE3 encoding a β-subunit of the K V 4.3 channel was found in a patient with BrS, which resulted in the reduction of its inhibitory effect on the K V 4.3 channel, leading to an increase in I to (Delpon et al., 2008). Additionally, the mutation p.D612N in KCND2 coding for K V 4.2 was identified responsible for the increased J-wave manifestation (Perrin et al., 2014). On the contrary, two recent studies reported some different findings. Different (G) Percentages of Ctrl-and BrS-CMs with rhythmic and arrhythmic APs after 2.5 µM milrinone treatment. Average I-V curves of I Na (automated patch-clamp) in Ctrl-CMs (H), BrS1-CMs (I), and BrS2-CMs (J) with and without 2.5 µM milrinone treatment for half an hour. No statistically significant differences were observed. (K) Quantitative analysis of the conduction velocity in Ctrl-, BrS1-, and BrS2-CMs with and without cilostazol (10 µM) treatment. No statistically significant differences were observed. (L) Quantitative analysis of the conduction velocity in Ctrl-, BrS1, and BrS2-CMs with and without 2.5 µM milrinone treatment. No statistically significant differences were observed. Two-way repeated measures ANOVA (A-C,H-J) and two-tailed paired Student's t-test (K,L) were used for statistical analysis: * P < 0.05, #P < 0.01, §P < 0.001. with our increased mRNA expression of KCND2/3, a significant decrease in the gene expression of KCND3 was revealed in CMs derived from iPSCs generated from two BrS patients carrying the SCN5A R620H-R811H mutation and SCN5A 1397 mutation, respectively (Liang et al., 2016). Moreover, the I to current density remained unaltered in BrS iPSC-CMs with SCN5A A226V-R1629X variant (Ma et al., 2018). Although our data showing the increased I to in BrS-CMs support the "repolarization disorder" theory of BrS, further studies need to investigate how the SCN5A mutation p.S1812X affects the transient-outward potassium channels, especially when we did not observe changes in K V 4.3 protein expression in BrS-CMs. We postulate that the augmented I to may result from the post-translational modification of Kv4.3 channel and impaired protein-protein interaction. It is worth mentioning that there is an inverse relationship between the maximum velocities of depolarization and repolarization, as demonstrated in murine isolated ventricular CMs as well as in HEK293 cells overexpressing both K V 4.3 and Na V 1.5 (Portero et al., 2018). Therefore, it is of vital importance to further study the entire molecular complexity, in which Na V 1.5 is naturally embedded in human CMs and to assess the functional interactions of different ion channels.
Moreover, in our study higher arrhythmic tendencies in BrS-CMs were observed. Spontaneous AP recordings revealed a higher percentage of BrS-CMs with arrhythmogenic events, and FP recordings displayed the increased SD of the inter-beat intervals. Unlike type 3 long QT syndrome, in which most of the SCN5A mutations cause a significantly enhanced persistent I Na contributing to the life-threatening ventricular arrhythmias (Malan et al., 2011;Ma et al., 2013), the mutation p.S1812X in our study did not alter the persistent I Na in BrS-CMs. To our surprise, the window current probability of I CaL was significantly increased in BrS-CMs compared to Ctrl-CMs, which is associated with the alterations in the steady-state activation and steady-state inactivation of the L-type calcium channels. Previous studies demonstrated that the increased I CaL window current contributed to the EAD formation and EAD-mediated arrhythmias (Madhvani et al., 2015). In iPSC-CMs derived from BrS patients without identified mutations, no significant changes in the steady-state activation and steady-state inactivation of the L-type calcium channels were observed, but the probability of I CaL window current was not calculated (Veerman et al., 2016). Future studies should investigate how the different channels in CMs, including the channels responsible for I CaL , I to , and I Na , co-operate and regulate the arrhythmia in BrS-CMs.
Based on the I to augment in our BrS-CMs, we speculate that the alleviation of AP arrhythmia could be possibly achieved through the reduction of I to currents. Cilostazol and milrinone are two oral PDE3 inhibitors, which might be potentially antiarrhythmic drugs for BrS patients (Tsuchiya et al., 2002). Previous study reported that the ventricular fibrillation of a 67-year-old patient with BrS was prevented by oral cilostazol administration (Tsuchiya et al., 2002). In the setting of an arterially perfused right ventricular wedge, cilostazol and milrinone could suppress arrhythmogenesis associated with BrS (Szel et al., 2013;Szel and Antzelevitch, 2014). The effects of cilostazol and milrinone may be related to an elevation of intracellular cyclic AMP concentration via inhibition of PDE activity, and consequently to suppression of I to and to an increase in I Ca , as previously discussed (Tsuchiya et al., 2002;Szel et al., 2013;Szel and Antzelevitch, 2014). Patocskai et al. (2016) demonstrated that both cilostazol and milrinone reduced I to in canine single left ventricular CMs, and addition of cilostazol or milrinone to the coronary perfusate restored the AP dome at all epicardial sites, reduced epicardial and transmural dispersion of repolarization, decreased J point and ST segment elevation, and terminated all arrhythmic activity in an experimental model of early repolarization syndrome. In our study, inhibition effects of cilostazol and milrinone on I to and arrhythmogenic events in BrS-CMs suggest a new therapeutic potential to alleviate BrS probably via I to reduction. Nevertheless, cilostazol and milrinone do not have effects on I Na or CV.

CONCLUSION
By using iPSC-CMs from two BrS patients harboring the SCN5A mutation p.S1812X, we demonstrate here that the mutation results in the reduced I Na but augmented I to and increased I CaL window current probability as well as conduction slowing, indicating that both repolarization and depolarization disorders coexist in BrS-CMs. Our findings indicate that the electrophysiological mechanisms underlying conduction slowing and arrhythmia in BrS-CMs involve not only the Na V 1.5 loss-offunction but also an impaired coordination of I Na , I to , and I CaL . Moreover, pharmacological treatments with the PDE3 inhibitors cilostazol and milrinone reduce the I to and proarrhythmic events in BrS, suggesting their therapeutic potential for BrS patients.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Institutional Ethics Committee of University Medical Center Göttingen (approval number 21/1/11) and of Technical University of Dresden (approval number EK 422092019) and carried out in accordance with the approved guidelines. The patients/participants provided their written informed consent to participate in this study.

FUNDING
This work was supported by the Deutsche Forschungsgemeinschaft (DFG) grants to KG (SFB 1002 TP A04, TP S, GU 595/3-1, and project number 288034826-IRTG 2251), and by the Free State of Saxony and the European Union EFRE (SAB project "PhänoKard" and "PhenoCor") to AE-A and KG, and by the Anschubsfinanzierung from University Medical Center Göttingen to KG and MiS. SW was funded by DFG grants WA 2539/4-1, 5-1, and 7-1. LSM was funded by DFG grants MA 1982/5-1 and 7-1. SW and LSM were also funded by the DFG SFB 1350 grant (project number 387509280, TPA6) and supported by the ReForM C program of the faculty. We acknowledge the support of the Open Access Funding by the Publication Fund of the TU Dresden.