Low extracellular potassium prolongs repolarization and evokes early afterdepolarization in human induced pluripotent stem cell-derived cardiomyocytes

ABSTRACT Long QT syndrome (LQTS) is characterized by a prolonged QT-interval on electrocardiogram and by increased risk of sudden death. One of the most common and potentially life-threatening electrolyte disturbances is hypokalemia, characterized by low concentrations of K+. Using a multielectrode array platform and current clamp technique, we investigated the effect of low extracellular K+ concentration ([K+]Ex) on the electrophysiological properties of hiPSC-derived cardiomyocytes (CMs) generated from a healthy control subject (WT) and from two symptomatic patients with type 1 of LQTS carrying G589D (LQT1A) or IVS7-2A>G mutation (LQT1B) in KCNQ1. The baseline prolongations of field potential durations (FPDs) and action potential durations (APDs) were longer in LQT1-CMs than in WT-CMs. Exposure to low [K+]Ex prolonged FPDs and APDs in a concentration-dependent fashion. LQT1-CMs were found to be more sensitive to low [K+]Ex compared to WT-CMs. At baseline, LQT1A-CMs had more prolonged APDs than LQT1B-CMs, but low [K+]Ex caused more pronounced APD prolongation in LQT1B-CMs. Early afterdepolarizations in the action potentials were observed in a subset of LQT1A-CMs with further prolonged baseline APDs and triangular phase 2 profiles. This work demonstrates that the hiPSC-derived CMs are sensitive to low [K+]Ex and provide a platform to study acquired LQTS.


INTRODUCTION
Long QT syndrome (LQTS) presents as an acquired or inherited arrhythmic disease characterized by prolonged QT interval on electrocardiogram and is associated with the occurrence of syncope or cardiac arrest. A special type of ventricular tachycardia known as torsades de pointes may arise in LQTS, which may degenerate into life-threatening ventricular fibrillation and cause sudden cardiac death (Schwartz et al., 2013).
Congenital forms of LQTS typically result from mutations in the cardiac ion channel-encoding genes. LQTS can be divided into different subtypes with LQT1 being the most common subtype caused by mutations in the KCNQ1. The KCNQ1 encodes the αsubunit of the voltage-gated potassium channel and, assembled with auxiliary β-subunits encoded by KCNE1, conducts the slow delayed rectifier outward K + current (I Ks ) (Barhanin et al., 1996;Sanguinetti et al., 1996). The high prevalence (0.4%) of LQTS in Finland has been explained by four founder mutations in the Finnish population (Marjamaa et al., 2009). The most prevalent LQT1 causing founder mutations is the C-terminal KCNQ1 G589D missense mutation (Piippo et al., 2001). Evidence suggests that G589D is a moderate dominant-negative trafficking mutation with normal functioning but with hindered transport to the cell membrane (Aromolaran et al., 2014). Another prevalent founder mutation is a strong dominantnegative splice site mutation IVS7-2A>G (Fodstad et al., 2006). Both, the G589D and IVS7-2A>G mutation types result in reduced I Ks current when co-expressed independently with wild-type KCNQ1 and KCNE1 (Piippo et al., 2001;Fodstad et al., 2006).
Hypokalemia is one of the most common electrolyte disturbances and is characterized by low blood serum K + levels. In normokalemic conditions, K + concentrations range from 3.5-5.3 mM (Macdonald and Struthers, 2004). In moderate hypokalemia, the K + concentrations range from 2.5-3.0 mM; and in severe hypokalemia they are <2.5 mM (Unwin et al., 2011). Hypokalemia is known to delay ventricular repolarization, slow ventricular conduction, cause hyperpolarization of the resting potential in ventricular myocytes as well as cause abnormal ventricular automaticity increasing the risk of ventricular arrhythmia and sudden cardiac death (Macdonald and Struthers, 2004;Osadchii, 2010;Schulman and Narins, 1990).
About 40% of the patients on thiazide diuretics have been reported to suffer from hypokalemia (Gennari, 1998) and a tenfold increase in hospital mortality was found in hypokalemic patients (Paltiel et al., 2001). Potassium electrolyte disturbance in serum is recognized as a risk factor among LQTS patients. The mortality rate of hospitalized hypokalemic patients was tenfold higher than in generalized hospital population illustrating the potentially lifethreatening consequences of hypokalemia (Paltiel et al., 2001). Other studies on hypokalemia have reported prolonged QTintervals, torsades de pointes and ventricular arrhythmias in LQT1 patients (Schulman and Narins, 1990;Roden, 1997). Under hypokalemic conditions, LQTS patients may be more susceptible to exaggerated QT-interval prolongation than healthy subjects because the repolarization reserve in LQTS patients is already compromised due to mutation(s) in the cardiac ion channel genes (Roden, 2004;Varró and Baczkó, 2011).
In this study, we have utilized a hiPSC-based model (Kiviaho et al., 2015) to study the effects of low [K + ] Ex in control CMs a well as in LQT1-specific CMs carrying Finnish founder mutations G589D or IVS7-2A>G. The electrophysiological properties of control subject (WT), and LQT1A (carrying G589D), LQT1B (carrying IVS7-2A>G) lines were assessed by determining the field potential durations (FPDs) and action potential durations (APDs), which correspond to surrogates for the QT-interval of the heart. The prolongation of APDs, FPDs and early afterdepolarizations (EADs) were assessed with low [K + ] Ex using ventricular-like CMs from WT, LQT1A and LQT1B lines.

Effect of [K + ] Ex on beating cardiomyocyte clusters
Baseline electrophysiological properties of healthy control (WT) and patient-specific LQT1A and LQT1B CM clusters were assessed by multielectrode array (MEA) measurements at 5.33 mM extracellular K + ([K + ] Ex ). Representative field potential traces from baseline (upper panels) of WT, LQT1A and LQT1B, respectively are depicted in Fig. 1A-C. Table 1A-C shows the FPD and corrected field potential durations (cFPDs) using two different correction formulae; Bazett and Fridericia. The Fridericia's correction formula was used to describe the extracted MEA data and a comparison of baseline properties revealed that cFPDs of LQT1A and LQT1B were more prolonged compared to WT (P=1.8E-4, one-way ANOVA) and in the same test no difference in cFPDs between LQT1A and LQT1B were found (P=0.839, one-way ANOVA).
Following the investigation of baseline properties on the CM clusters, the effect of sequential reduction to 3, 2 and 1 mM [K + ] Ex were studied. Representative field potential traces from 1 mM [K + ] Ex (lower panels) of WT, LQT1A and LQT1B, respectively are depicted in Fig. 1A-C. A concentration-dependent increase in cFPD correlated with sequential reduction in [K + ] Ex (Table 1A-C). Table 1A shows that low [K + ] Ex prolonged cFPD in WT with significant difference (P=0.045, paired t-test) at 1 mM [K + ] Ex .
In contrast, the cFPD was significantly prolonged in LQT1A (P=0.049, paired t-test) and LQT1B lines (P=0.043, paired t-test) already at 3 mM [K + ] ex (Table 1B,C). The absolute cFPD prolongation in response to lowered [K + ] Ex was calculated (Table 1D) but no statistically significant differences were found between the lines.

Effect of [K + ] Ex on single beating cardiomyocytes
Similar to MEA, the electrophysiological properties of single CMs were compared among WT, LQT1A and LQT1B in current clamp. Table 2A-C shows extracted data from action potentials (AP) consisting of: action potential duration (APD), action potential amplitude (APA) and maximum diastolic potential (MDP). APD 90/50 ratios were <1.30 indicating ventricular-like CM. The baseline APDs of LQT1A and LQT1B were significantly longer than WT (P<0.008, one-way ANOVA) and in the same test the APDs showed that LQT1A was significantly longer than LQT1B (P=0.007, n=22 in each group, one-way ANOVA). The baseline APD 50/10 ratio was higher in LQT1A compared to WT and LQT1B, which indicates that phase 2 of the APs in LQT1A were significantly more triangular in shape (ratios: LQT1A 2.95±0.13**, P<0.002: LQT1B 2.45±0.09 and WT 2.28±0.07, P=0.438, one-way ANOVA, n=22, Table 2A-C).
The [K + ] Ex in the perfusate was then lowered from 4.8 mM at baseline to 3, 2 or 1 mM and the effects were studied on the WT, LQT1A and LQT1B CMs. The APDs and APAs were increased and MDPs decreased at low [K + ] Ex (Table 2A- Fig. 2D and the APD 90 of LQT1B CMs were found to be more prolonged compared to the LQT1A and WT CMs at 1 mM [K + ] Ex (P=0.025, one-way ANOVA). The absolute APD 90 prolongation in response to lowered [K + ] Ex was calculated in the same data sets and is presented in Table 2D. This data reveals a more pronounced absolute APD 90 prolongations with 1 mM in LQT1A and LQT1B lines compared to WT. This suggests that LQT1A is more prolonged in absolute terms while LQT1B is more sensitive to low [K + ] Ex in relative terms.
Single LQT1A cardiomyocytes exhibit arrhythmia Low [K + ] Ex did not evoke arrhythmias in WT, LQT1A and LQT1B CM clusters in MEA recordings. Similarly, in current clamp recordings no arrhythmias were found in WT and LQT1B. However, LQT1A exhibited characteristics shown in Fig. 3A,B. The APDs were further prolonged and exhibited early afterdepolarizations (EADs) evoked only with 1 mM [K + ] Ex (Table 3A). In some LQT1As, EADs were already present in baseline conditions (Table 3B). The total occurrence of LQT1A CMs exhibiting EADs was 37.5%. In baselines without EADs (but with EADs at low [K + ] Ex ), the APD 90/50 were similar (<1.30) to CMs without EADs at low [K + ] Ex . However, in the former (EADs at low [K + ] Ex ) the baseline APD 50/10 ratios were higher than in the latter (CMs without the EADs) (P=1.3E-6, unpaired t-test, APD 50/10 ratio: 4.83±0.68 Table 3A and 2.95±0.13 Table 2B, see also Figs 2B and 3A). Note, the CMs analyzed in Table 3A exhibited EADs only with 1 mM [K + ] Ex . CMs with EADs occurring at baseline had an increase in EADs and APD prolongation with lower [K + ] Ex (Table 3B, Fig. 3B, arrows). The EAD 'take off' in baselines or with 1 mM [K + ] Ex occurred solely in phase 2 and were found to range from −15 mV to −27 mV. EAD 'take off' in baselines were −17.8±1.7 mV (n=5) and when evoked with 1 mM [K + ] Ex −20.3± 1.3 mV (n=7), suggesting a role of L-type calcium current due to the fact that the triangular phase 2 is driven towards more negative membrane potential during further APD prolongation in these LQT1A CMs.

DISCUSSION
This study demonstrates the effect of low [K + ] Ex on WT, LQT1A and LQT1B CMs using MEA and current clamp. The baseline APD and FPD of LQT1A and LQT1B CMs were longer compared to WT. Prolonged repolarization was evoked in a concentrationdependent fashion by low [K + ] Ex in all types of CMs with the LQT1B CMs being most sensitive to low [K + ] Ex . In current clamp we found that MDP was decreased (hyperpolarized) and APA was higher with low [K + ] Ex . EADs were occasionally observed in LQT1A at single cell level and these CMs had further prolonged APDs already at baseline compared to the majority of LQT1A CMs. An interesting finding was that the phase 2 triangularity (APD 50/10 ratio) was more pronounced in the LQT1A CMs at baseline when EADs were present. In support of our data, previous current clamp characteristics recorded at physiological potassium conditions at baseline are similar to WT, LQT1A and LQT1B CMs described in this study (Kiviaho et al., 2015). This is the first study where low potassium concentration induces prolonged repolarization in hiPSC-derived WT and LQTS CMs. Additionally, EADs were observed only in CMs with G589D  point mutation (LQT1A), but not in those carrying the IVS7-2A>G mutations (LQT1B). The appearance of EADs has been linked to the hyperpolarization of cell membrane in low [K + ] Ex conditions (Zaza, 2009). In the present study, we show that EADs similarly emerge from more hyperpolarized MDPs at low [K + ] Ex . This is supported by an altered K + conductance and electromotive force during acute and chronic low [K + ] Ex and is further strengthened by similar findings with data from animal ventricular cardiomyocytes (White and Terrar, 1991;Ruiz Ceretti et al., 1982;Kishida et al., 1979;Zaza, 2009), in transgenic rabbit LQT1 and LQT2 models (Liu et al., 2012) and in transgenic mouse LQT2 model (Teng et al., 2004). Our data with hiPSC-derived CMs thus correlates well with previous studies even though previously reported data were conducted with transgenic or non-human CMs.
In this study, low [K + ] Ex prolonged the FPD and APD in all our CMs (WT, LQT1A and LQT1B). This data is supported by previous findings with various animal studies as well as in isolated ventricular CMs (White and Terrar, 1991;Akita et al., 1998;Guo et al., 2011;Pezhouman et al., 2015;Chan et al., 2015;Osadchii, 2010). The prolongation of repolarization in hypokalemia (low [K + ] Ex ) has also been documented in cardiac patients receiving diuretic therapy (Stewart et al., 1985). It is well established that low [K + ] Ex suppresses the K + currents such as I Kr contributing to delayed repolarization thereby prolonging the QT-interval (Yang et al., 1997;Sanguinetti and Jurkiewicz, 1992;Scamps and Carmeliet, 1989;Guo et al., 2011;Osadchii, 2010). Thus, low extracellular K + is recognized as a factor reducing the repolarization reserve (Varró and Baczkó, 2011).
The repolarization phase involves primarily I Kr and I Ks in ventricle myocytes. The LQT1 lines used in this study had mutation in the KCNQ1 depressing the I Ks current, which is presented as FPD and APD prolongations at baseline. Thus, this suggests prolongation in FPD and APD to increase in low [K + ] ex partly due to the lack of I Ks repolarization reserve. The deficiency in I Ks current in our LQT1 lines reproduces our previous findings where we show no or only a marginal effect on the APDs with a partial I Ks blockade (JNJ303) in the LQT1 lines but robust effects in the WT line (Kiviaho et al., 2015). As expected, a partial I Kr blockade (E4031) resulted in a prolongation of FPD and APD in the WT, LQT1A and LQT1B lines (Kiviaho et al., 2015;Pradhapan et al., 2013).
In a transgene expression system with the G589D point mutation, a reduced I Ks current has been reported (Piippo et al., 2001). The G589D mutation has been found to be a trafficking defect (Aromolaran et al., 2014). The IVS7-2A>G splicing mutation is a 'loss-of-function' defect (Fodstad et al., 2006). Based on earlier expression studies in co-expression with wild-type KCNQ1 and KCNE1, the IVS7-2A>G mutation decreases I Ks current more than the G589D mutation (Piippo et al., 2001;Fodstad et al., 2006). Thus, these earlier findings support our data with difference in repolarization times, but do not explain the presence of EADs. This  refers to complexity of K + homeostasis and additional mechanisms in CMs. The baseline FPD and APD in our data demonstrate that the repolarization time of LQT1A is longer than that of LQT1B, but LQT1B is more sensitive to low [K + ] Ex despite MDP and APA being similar in LQT1A and LQT1B CMs. Since I Kr is a major player involved in repolarization, the simplest explanation would be that the I Kr contribution to the repolarization reserve pool is larger in LQT1B than in LQT1A. The repolarization reserve pool of LQT1B would thus show greater K + sensitivity due to suppressed K + conductance, increased electromotive force when acutely exposed to low [K + ] Ex (Zaza, 2009) and no I Ks functionality. This would also explain why LQT1B has shorter FPD and APD baseline values compared to LQT1A in this study. Moreover, this would be in line with our previous finding that EADs in LQT1B are more pronounced compared to LQT1A during an I Kr blockade with E4031 at normal physiological K + conditions (Kiviaho et al., 2015). As I Ks is diminished in our LQT1 lines, the I Kr functionality will become important for the repolarization reserve during development of hypokalemia. The acute and chronic K + homeostasis is complex with effects in a variety of parameters. Based on our studies we can exclude I Ks as a major contributor in the repolarization reserve in the LQT1 lines. This is supported by the fact that only LQT1A CMs present a prolonged phase 2 triangularity and EADs, but not LQT1B CMs which, however, have less I Ks current based on gene transfection experiments (Piippo et al., 2001;Fodstad et al., 2006). Action potential shape has been suggested to play a crucial role in occurrence of pro-arrhythmic events (Hondeghem et al., 2001;Osadchii and Olesen, 2009). Phase 3 triangularity is shown to be a marker for pro-arrhythmia in monophasic action potentials of guinea pig and rabbit hearts (Hondeghem et al., 2001;Osadchii and Olesen, 2009). I Kr deficiency has also been associated in phase 3 triangularity (Hondeghem et al., 2001); however, in this study, we did not find EADs evoked in phase 3 but in phase 2 suggesting a reopening of L-type calcium channels (I Ca,L ), i.e. in the LQT1A CMs with EADs, these repeatedly occurred between -15 mV and -27 mV. It has previously been suggested that phase 2 EADs are carried by I Ca,L following an APD prolongation towards a more negative membrane potential while we cannot exclude that the Na + / Ca 2+ -exchanger works in concert with the I Ca,L (Guo et al., 2007;Sipido et al., 2007;Banyasz et al., 2012). It is currently unknown what causes the further prolongation and triangularity in phase 2 giving rise to EADs found in our LQT1A CMs. It should be noted that no arrhythmia was detected and no phase 2 triangularity was observed in LQT1B and WT CMs.

Potential limitations of the study
The effects of low [K + ] Ex on hiPSC-derived CM may not completely reflect the symptoms of hypokalemia in vivo. hiPSCderived CMs are still more fetal-like and this has to be considered when translating these results to the clinics. I Ks and I Kr have not been quantified in our study but their functionality has been analyzed with the use of specific ion channel blockers in our previous study (Kiviaho et al., 2015). Further understanding the interaction between ion channels and exchangers is needed for revealing the underlying mechanisms of prolongation and arrhythmia observed in the LQT1 lines used in this study. Here, we only have an acute expose and long term exposure to low [K + ] Ex on hiPSC-derived CMs may reveal different characteristics.  In conclusion, we have shown that low [K + ] Ex delays the repolarization in hiPSC-derived CMs at single cell as well as at multicellular level. The LQT1 CMs are more sensitive to potassium electrolyte disturbances than WT CMs, thus confirming previous clinical studies. In LQT1 CMs with G589D mutation phase 2 triangularity in EADs were occasionally observed at baseline or could be further evoked by low [K + ] Ex . However, phase 2 triangularity or EADs could not be evoked in LQT1 CMs with IVS7-2A>G mutation or in WT CMs. The effects of low [K + ] Ex has not previously been investigated in hiPSC-derived CMs. Thus, this work demonstrates that the hiPSC-derived CMs provide a platform for studying the effects of low extracellular K + and also provide a platform to study acquired LQTS.

Ethical approval
The study was approved by the Ethics Committee of Pirkanmaa Hospital District to establish, culture, and differentiate hiPSC lines (R08070). Participants donating skin biopsies signed an informed consent after receiving oral and written descriptions of the study. Skin biopsies for hiPSC establishment were received from the Heart Hospital, Tampere University Hospital, Tampere, Finland.

Long QT patient characteristics
The LQT patient characteristics have been previously described elsewhere (Kiviaho et al., 2015). In brief, the LQT1 patient carrying G589D mutation is a 46-year-old female with corrected QT interval (QTc) of 464 ms. The other LQT1 patient carrying IVS7-2A>G is a 51-year-old female with QTc value of 489 ms. These patients are symptomatic with episodes of syncope.

Human induced pluripotent stem cell generation and culture
Patient-specific hiPSCs were generated as described earlier (Takahashi et al., 2007). The LQT1-specific hiPSCs were derived from patients carrying G589D or IVS7-2A>G mutation in the KCNQ1 as previously described (Kiviaho et al., 2015). Healthy control hiPSCs (WT) were derived from skin fibroblasts of a healthy 55-year-old female (Ahola et al., 2014). hiPSC lines were cultured in knockout serum replacement medium using mouse embryonic fibroblasts (Millipore, Billerica, MA, USA) as feeders. The following components were included in the knockout serum replacement medium: knockout-DMEM (Invitrogen, Carlsbad, CA, USA) containing 20% knockout-serum replacement (Invitrogen, Carlsbad, CA, USA), nonessential amino acids, GlutaMAX, penicillin/streptomycin, 0.1 mM 2mercaptoethanol and 4 ng/ml fibroblast growth factor 2 (R&D Systems Inc., Minneapolis, MN, USA). The medium was refreshed three times a week and the hiPSCs were passaged weekly using collagenase IV (Invitrogen, Carlsbad, CA, USA). (Hamill et al., 1981;Rae et al., 1991). Action potentials were acquired with an Axon Series 200B amplifier and a Digidata 1440 AD/DA converter (Molecular device, LTD, USA). Dissociated CMs were plated on 5 mmØ coverslips and recorded on the 6th or 7th day. In brief, coverslips were transferred to a RC-24N recording chamber (Harvard Instruments UK, Warner Instruments, USA) and placed on an inverted Olympus IX71 microscope. The perfusate was maintained at 35-36°C using an SH-27B inline heater (Harvard Apparatus Ltd., Kent, UK). The HEPES based extracellular solution (HBS) contained (in mM): 143 NaCl, 1.8 CaCl 2 , 1.2 MgCl 2 , 5 glucose and 10 HEPES. For experiments KCl was added to the HBS to final concentrations of 1, 2, 3 and 4.8 mM K + ( pH was set to 7.4 with NaOH and osmolarity set to 300-302 mOsm with sucrose). HBS with 4.8 mM K + was used in baseline recordings. The pipette solution contained (in mM): 122 KMeSO 4 , 30 KCl, 1.2 MgCl 2 , 1 CaCl 2 ( pH was set to 7.2 with KOH and osmolarity set to 290-292mOsm with sucrose). The pipette resistance was ∼3MΩ after filling with pipette solution. The action potentials from spontaneously beating CMs were recorded in current clamp mode. Current clamp recordings were digitally sampled at 20 kHz and filtered at 2 kHz using low pass Bessel filter on recording amplifier. AP duration at 10, 50 and 90% repolarization (APD 10 , APD 50 and APD 90 ), APA, and MDP were extracted from recorded action potentials using in house developed analysis modules running on the Origin 9 platform (Microcal Origin TM , Northampton, Massachusetts, USA). Baseline data in a CM was extracted from a minimum of 15 single APs and summed for statistics. When EADs were present in a CM, a minimum of 5 single APs were obtained in regions with maximal effect of [K + ] Ex and summed for each concentration for statistics. Acute application of [K + ] Ex was tested in sequential application without wash out, with wash out and with single applications from baselines to either 3, 2 or 1 mM [K + ] Ex . Maximal effect was observed in seconds and recordings were fully reversible, thus data is pools of these series. The data presented from current clamp was from ventricular-like CMs. These were grouped as ventricular-like when baseline APs had APD 90/50 <1.3, APA >100 mV and MDP <−60 mV. The Phase 2 AP triangularity was calculated as APD 50/10 .

Statistical analyses
One-way ANOVA followed by Tukey's post hoc test was used to assess differences in means of groups of three pairs. Paired t-test was used to access difference between the means in baseline and the corresponding effect of 3, 2 or 1 mM K + in the same recording, while unpaired t-test was uses to access differences in means of two similar recordings (Microcal Origin™ 9.1, Northampton, Massachusetts, USA). Significant difference in the tables is presented as *P<0.05, **P<0.01 and ***P<0.001 and data is presented as mean±s.e.m.