Ultrarapid Delayed Rectifier K+ Channelopathies in Human Induced Pluripotent Stem Cell-Derived Cardiomyocytes

Atrial fibrillation (AF) is the most common cardiac arrhythmia. About 5–15% of AF patients have a mutation in a cardiac gene, including mutations in KCNA5, encoding the Kv1.5 α-subunit of the ion channel carrying the atrial-specific ultrarapid delayed rectifier K+ current (IKur). Both loss-of-function and gain-of-function AF-related mutations in KCNA5 are known, but their effects on action potentials (APs) of human cardiomyocytes have been poorly studied. Here, we assessed the effects of wild-type and mutant IKur on APs of human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). We found that atrial-like hiPSC-CMs, generated by a retinoic acid-based differentiation protocol, have APs with faster repolarization compared to ventricular-like hiPSC-CMs, resulting in shorter APs with a lower AP plateau. Native IKur, measured as current sensitive to 50 μM 4-aminopyridine, was 1.88 ± 0.49 (mean ± SEM, n = 17) and 0.26 ± 0.26 pA/pF (n = 17) in atrial- and ventricular-like hiPSC-CMs, respectively. In both atrial- and ventricular-like hiPSC-CMs, IKur blockade had minimal effects on AP parameters. Next, we used dynamic clamp to inject various amounts of a virtual IKur, with characteristics as in freshly isolated human atrial myocytes, into 11 atrial-like and 10 ventricular-like hiPSC-CMs, in which native IKur was blocked. Injection of IKur with 100% density shortened the APs, with its effect being strongest on the AP duration at 20% repolarization (APD20) of atrial-like hiPSC-CMs. At IKur densities < 100% (compared to 100%), simulating loss-of-function mutations, significant AP prolongation and raise of plateau were observed. At IKur densities > 100%, simulating gain-of-function mutations, APD20 was decreased in both atrial- and ventricular-like hiPSC-CMs, but only upon a strong increase in IKur. In ventricular-like hiPSC-CMs, lowering of the plateau resulted in AP shortening. We conclude that a decrease in IKur, mimicking loss-of-function mutations, has a stronger effect on the AP of hiPSC-CMs than an increase, mimicking gain-of-function mutations, whereas in ventricular-like hiPSC-CMs such increase results in AP shortening, causing their AP morphology to become more atrial-like. Effects of native IKur modulation on atrial-like hiPSC-CMs are less pronounced than effects of virtual IKur injection because IKur density of atrial-like hiPSC-CMs is substantially smaller than that of freshly isolated human atrial myocytes.


INTRODUCTION
Worldwide, the prevalence of atrial fibrillation (AF) is around 1-2% (Potpara and Lip, 2011). Mutations in cardiac genes account for onset of 5-15% of AF cases (Darbar et al., 2003;Potpara and Lip, 2011). Mutations in KCNA5 are associated with AF, although rare (Feghaly et al., 2018). KCNA5 encodes the pore-forming α-subunit K v 1.5 of the channel carrying the ultrarapid delayed rectifier K + current (I Kur ) (Fedida et al., 1993;Wang et al., 1993). In the human heart, K v 1.5 and the mRNA encoding K v 1.5 are both highly expressed in the atria (Ellinghaus et al., 2005;Gaborit et al., 2007), whereas expression of K v 1.5 is very low in both endocardial and epicardial ventricular tissue (Mays et al., 1995;Gaborit et al., 2007) and expression of mRNA encoding K v 1.5 is also low (Kääb et al., 1998;Gaborit et al., 2007). Accordingly, in their voltage clamp experiments on isolated human atrial and subepicardial ventricular myocytes, Amos et al. (1996) could not observe an I Kur -like current in their ventricular myocytes, in contrast to their atrial myocytes. I Kur activates rapidly upon depolarizations to membrane potentials positive to −50 mV and is responsible for the early repolarization in human atrial action potentials (APs) Amos et al., 1996;Wettwer et al., 2004;Li et al., 2008).
Both loss-of-function and gain-of-function mutations in KCNA5 have been identified in patients with AF (Olson et al., 2006;Yang et al., 2009Yang et al., , 2010Christophersen et al., 2013;Hayashi et al., 2015;Tian et al., 2015). Loss-offunction mutations in KCNA5 are supposed to increase susceptibility to AF by prolonging the AP duration (APD) of atrial myocytes, which may eventually result in early afterdepolarizations (EADs) (Yang et al., 2009;Hayashi et al., 2015). Indeed, in vitro electrophysiological studies where I Kur was blocked, representing complete KCNA5 loss-of-function mutations, resulted in prolonged APDs and presence of EADs (Olson et al., 2006). EADs as a consequence of prolonged APDs have also been observed in in silico studies on loss-of-function KCNA5 mutations (Colman et al., 2017;Ni et al., 2017).
Gain-of-function mutations, on the other hand, are presumed to cause AF by shortening the effective refractory period (ERP) of the atrial AP, facilitating re-entry wavelets in the atria (Nattel, 2002;Christophersen et al., 2013). This hypothesis is supported by in silico studies, which demonstrated that increased I Kur density, representing gain-of-function mutations, resulted in a shortened APD and arrhythmogenesis in human atrial tissue (Colman et al., 2017;Ni et al., 2017).
Although the in silico studies are instrumental in determining the potential effect of both loss-of-function and gain-of-function mutations in KCNA5, detailed electrophysiological studies of the KCNA5 mutations in human cardiomyocytes are limited. Human induced pluripotent stem cell cardiomyocytes (hiPSC-CMs) have become a highly suitable tool to study cardiac ion channelopathies and their electrophysiology (Zhang et al., 2011;Hoekstra et al., 2012;Verkerk et al., 2017). Over time, the technique of cardiomyocyte differentiation has advanced, facilitating the generation of distinct atrial-and ventricular-like hiPSC-CM populations (Zhang et al., 2011;Devalla et al., 2015;Devalla and Passier, 2018). In the present study, we employed dynamic clamp to investigate the effects of loss-of-function and gain-of-function mutations in KCNA5 in both atrial-and ventricular-like hiPSC-CMs.

MATERIALS AND METHODS
hiPSC-CM Differentiation hiPSC-CMs were generated from the control LUMC0099iCTRL04 hiPSC line, which was derived from human fibroblasts extracted through skin biopsies from of a Caucasian woman. The LUMC0099iCTRL04 line is registered in the Human Pluripotent Stem Cell Registry (Seltmann et al., 2016), which contains all details pertaining to its generation and characterization (hPSCreg, 2019). hiPSC clones showing stem cell morphology were characterized for pluripotency marker expression and differentiation potential to hiPSC-CMs in BPEL medium (Ng et al., 2008) containing activin-A, BMP4, and CHIR99021 (Devalla et al., 2016). After 3 days, this medium was replaced by BPEL medium containing XAV939 (Tocris Biosciences) for ventricular differentiation (Ng et al., 2008;Devalla et al., 2016). To differentiate hiPSC-CMs to atrial-like hiPSC-CMs, 1 µM all-trans retinoic acid (RA) was added (Devalla et al., 2015). Twenty days after differentiation, hiPSC-CMs were dissociated with TrypLE Select (Life Technologies), and plated at a low density (≈7.5 × 10 4 cells) on Matrigel coated coverslips in BPEL medium (Devalla et al., 2016).

Data Acquisition
Electrophysiological recordings were performed 4-13 days post dissociation from spontaneously beating single hiPSC-CMs. RA-treated hiPSC-CMs displaying a short, pulse-like beating pattern and non-RA-treated hiPSC-CMs with a contractionlike beating pattern were selected for data acquisition. APs and I Kur were recorded at 36-37 • C with the perforated patch-clamp technique using an Axopatch 200B amplifier (Molecular Devices, Sunnyvale, CA, United States). Data acquisition and analysis were performed with custom software. Signals were low-pass filtered with a cut-off frequency of 2 kHz and digitized at 40 and 5 kHz for AP and I Kur recordings, respectively. Cell membrane capacitance (C m , in pF) was calculated by dividing the time constant of the decay of capacitive transient when hyperpolarized by 5 mV from −40 mV in voltage clamp by series resistance. C m of atrial-and ventricular-like hiPSC-CMs was 16.4 ± 2.3 pF (mean ± SEM, n = 28), and 19.2 ± 2.5 pF (n = 27), respectively (t-test, N.S.). Patch pipettes with a resistance of ≈2.0 M were pulled from borosilicate glass (Harvard Apparatus) and filled with solution containing (in mM): 125 K-gluconate, 20 KCl, 5 NaCl, 0.44 Amphotericin-B, 10 HEPES; pH set to 7.2 (KOH). Cells were superfused with modified Tyrode's solution containing (in mM): 140 NaCl, 5.4 KCl, 1.8 CaCl 2 , 1.0 MgCl 2 , 5.5 Dglucose, 5 HEPES; pH set to 7.4 (NaOH). All potentials were corrected for the estimated liquid junction potential of −15 mV (Barry and Lynch, 1991).

Action Potential Recordings
APs were elicited at 1 Hz by 3-ms, ≈1.2 × threshold current pulses through the patch pipette. The AP parameters analyzed were resting membrane potential (RMP, in mV), maximum upstroke velocity (dV/dt max , in V/s), AP amplitude (APA, in mV), AP duration at 20, 50, and 90% repolarization (APD 20 , APD 50 , and APD 90 , respectively, in ms), and AP plateau amplitude (APPlatA, in mV), derived from the membrane potential (V m ) at 50 ms after the time of dV/dt max .

Native I Kur Recordings
Native I Kur was activated by 200-ms voltage clamp steps from −50 to +50 mV. A 50-ms prepulse to 0 mV was applied to activate and inactivate remaining transient membrane currents. Series resistance was compensated by ≥ 80%. I Kur was measured as the current sensitive to 50 µM 4-aminopyridine (4-AP) Caballero et al., 2010), and was normalized to C m to calculate current density (in pA/pF).

Dynamic Clamp
Although inward rectifier K + current (I K1 ) is not necessarily low in hiPSC-CMs (Horváth et al., 2018), hiPSC-CMs tend to lack I K1 , which is responsible for stabilizing the RMP of atrial and ventricular myocytes, and thus show spontaneous activity (Dhamoon and Jalife, 2005;Hoekstra et al., 2012;Verkerk et al., 2017). The RMP of our atrial-and ventricular-like hiPSC-CMs was stabilized and set at a regular hyperpolarized value using the dynamic clamp technique (Wilders, 2006). A virtual Kir2.1based I K1 , with a standard peak current density of 2 pA/pF, was injected into the hiPSC-CMs and this I K1 was computed in real time, based on the acquired V m , following the approach of Meijer van Putten et al. (2015). Accordingly, the mathematical equation for I K1 reads In this equation, in which the rectification properties of I K1 are implemented through a Boltzmann equation, I K1 is in pA/pF and V m is in mV. E K is the Nernst potential for potassium, which amounts to −86.9 mV in our experimental setting.
The effect of the injection of this virtual I K1 is illustrated in Figure 1, which shows the APs of typical atrial-like and ventricular-like hiPSC-CMs in the absence and presence of this virtual I K1 (top panels) and the associated injected current (bottom panels), which consists of this I K1 and a short inward stimulus current. A virtual Kir2.1-based I K1 , characteristic for human ventricular myocytes , was used in both atrial-and ventricular-like hiPSC-CMs because a more 'atriallike' I K1 in hiPSC-CMs results in a substantial current during early repolarization due to its reduced rectification (Meijer van Putten et al., 2015;Verkerk et al., 2017;Fabbri et al., 2019), and we wanted to prevent a prominent overlap and potential interference of I K1 and I Kur during the course of an action potential.
The dynamic clamp technique was also used to provide our atrial-and ventricular-like hiPSC-CMs with a virtual wild-type or mutant I Kur , as illustrated in Figure 2. Like I K1 , I Kur was computed in real time, based on the acquired value of V m . I Kur was formulated as detailed in Section "I Kur Equations" below. Virtual I Kur was injected into atrial-and ventricular-like hiPSC-CMs with a fully activated conductance of 12.5, 25, 50, and 75% of its wild-type value to mimic loss-of-function mutations, and 125, 150, 175, and 200% of its wild-type value to mimic gain-of-function mutations.

I Kur Equations
To compute I Kur in our dynamic clamp system, I Kur equations of the comprehensive human atrial myocyte model by Maleckar et al. (2009) were used. These equations were also adopted by Grandi et al. (2011) in their human atrial action potential and Ca 2+ model and read: In these equations, the dimensionless Hodgkin and Huxley-type activation and inactivation gating variables, ranging between 0 and 1, are denoted by a ur and i ur , respectively, whereas I Kur (in pA/pF), g Kur (in nS/pF), V (in mV), E K (in mV), and t (in s) denote the ultrarapid delayed rectifier outward K + current, its fully activated conductance, the membrane potential, the K + reversal potential, and the time, respectively. The steady-state values of a ur and i ur are denoted by a ur,∞ and i ur,∞ , respectively, and the associated time constants by τ aur (in s) and τ iur (in s), respectively. As in the models by Maleckar et al. (2009) and Grandi et al. (2011), a default value of 0.045 nS/pF was used for g Kur . Of note, Maleckar et al. (2009) based this value on experimental data on I Kur density in human atrial myocytes.

Statistical Analysis
Data are presented as mean ± SEM. Statistical analysis was carried out with SigmaStat 3.5 software (Systat Software, Inc., San Jose, CA, United States). Native I Kur density of atrial-and ventricular-like hiPSC-CMs was compared with an independent samples t-test. Two-way repeated measures ANOVA followed by the Student-Newman-Keuls post hoc test was used for FIGURE 2 | Dynamic clamp setup. Ultrarapid delayed rectifier K + current (I Kur ) and inward rectifier potassium current (I K1 ) were computed in real time on a PC running the Linux operating system and Real-Time Experiment Interface (RTXI) software (Patel et al., 2017), based on the acquired membrane potential (V m ) of the human induced pluripotent stem cell-derived cardiomyocyte (hiPSC-CM). Data were recorded on an Apple Macintosh computer using custom software to visualize and control the experiment. The sum of the stimulus current (I stim ), I K1 , and I Kur resulted in the current that was injected into the hiPSC-CM (I in ). Sample rates were 30 kHz ( t 1 = 33.33 µs) and 20 kHz ( t 2 = 50 µs). comparing AP parameters of atrial-and ventricular-like hiPSC-CMs in absence or presence of 4-AP. One-way repeated measures ANOVA followed by the Student-Newman-Keuls post hoc test was used for comparing the effect of injecting virtual I Kur at various densities into atrial-and ventricular-like hiPSC-CMs. P < 0.05 was considered statistically significant.

Atrial-and Ventricular-Like hiPSC-CM APs
APs were recorded from single atrial-and ventricular-like hiPSC-CMs that showed spontaneous beating upon visual inspection, clearly indicating a healthy and myocardial status. APs were elicited at 1 Hz and virtual I K1 was injected into the cells, based on the approach of Meijer van Putten et al. (2015), to stabilize the RMP and set it at a regular hyperpolarized value. Figure 3B shows typical atrial-and ventricular-like hiPSC-CM APs. AP parameters, as illustrated in Figure 3A, are summarized in Table 1. Atrial-like hiPSC-CMs repolarize faster than ventricularlike APs, resulting in a significantly shorter APD 20 , APD 50 , and APD 90 . The ventricular-like APs have a prominent plateau phase at relatively positive potentials, in contrast with the atrial-like hiPSC-CMs that show a less prominent plateau phase at less positive potentials, if any plateau at all. Consequently, APPlatA was significantly smaller in atrial-like hiPSC-CMs and thus appeared a strong tool to distinguish between atrial-like and ventricular-like hiPSC-CMs. APA and dV/dt max did not differ between the atrial-and ventricular-like hiPSC-CMs, but RMP was less negative in ventricular-like hiPSC-CMs. Next, the cells were superfused with 50 µM 4-AP to block intrinsic I Kur . Figures 3C,D, top panels, show typical atrialand ventricular-like hiPSC-CMs in absence (red and blue lines,

Native I Kur
Native I Kur density in atrial-and ventricular-like hiPSC-CMs was quantified during 200-ms depolarizing voltage clamp steps as the current sensitive to 50 µM 4-AP. Figure 4A shows typical examples in an atrial-like (red trace) and a ventricular-like hiPSC-CM (blue trace). On average, I Kur density in atrial-like hiPSC-CMs was significantly larger than in ventricular-like hiPSC-CMs, with densities of 1.88 ± 0.49 (n = 17) and 0.26 ± 0.26 (n = 17) pA/pF, respectively ( Figure 4B). If the voltage clamp protocol of Figure 4A is repeated in computer simulations with the Maleckar et al. (2009) human atrial myocyte model (Wilders, 2018), an I Kur density of 5.45 pA/pF is obtained. We regard the latter as a realistic value for human atrial myocytes since Maleckar et al. (2009) based the characteristics of their model I Kur on experimental data on I Kur from isolated human atrial myocytes.

Effects of Baseline Virtual I Kur on APs of Atrial-and Ventricular-Like hiPSC-CMs
Next, we studied the effects of a virtual I Kur on APs of atrial and ventricular-like hiPSC-CMs using dynamic clamp. In the human heart, I Kur is highly atrial-specific (Ellinghaus et al., 2005;Gaborit et al., 2007). However, the dynamic clamp technique allowed us to inject a virtual I Kur in both atrial-and ventricularlike hiPSC-CMs and thus assess to which extent this made their action potential morphology become similar. In either case, 4-AP (50 µM) was present to ensure that any native I Kur was blocked  and from here on we name this condition 0% I Kur . First, we injected a virtual I Kur as implemented in the Maleckar et al. (2009) human atrial myocyte model, i.e., with the aforementioned 5.45 pA/pF density at +50 mV, which we here consider as 100% density. Figures 5A,B, top panels, show typical examples of APs recorded from atrial-and ventricular-like hiPSC-CM at 0% (red and blue lines, respectively) and 100% I Kur (black lines). The injected current, which now consists of I K1 , 0% or 100% I Kur , and a short stimulus current, is shown in the middle panels of Figures 5A,B. The average effects on the AP parameters of 11 atrial-and 10 ventricular-like hiPSC-CMs appear as the bars at 0% and 100% I Kur in Figures 6A-H, 7A-F, in which each of the AP parameters is expressed as a percentage of its value obtained at 100% I Kur . Injection of I Kur shortened the AP of both atrial-and ventricular-like hiPSC-CMs, while the AP plateau was suppressed (Figures 5, 6). dV/dt max was unaltered in both atrialand ventricular-like hiPSC-CMs, but in atrial-like hiPSC-CMs RMP was significantly more negative and APA significantly larger in absence than in presence of I Kur (Figure 7). The small 1.2% difference in RMP, equivalent to a 1.0-mV hyperpolarization, is likely a false positive because injection of various amounts of I Kur did not affect the RMP in either atrial-like or ventricular-like hiPSC-CMs (see below).
The phase plane plots of Figures 5C,D show the injected currents of the middle panels of Figures 5A,B plotted against the associated membrane potentials of the APs shown in the top panels of Figures 5A,B. The start and the end of the negative depolarizing current that flows during the stimulus are indicated by downward and upward vertical arrows, respectively. The black loops of the phase plane plots clearly show that I Kur is a repolarizing current that is already activated during the 3-ms stimulus and stays active until repolarization reaches −40 to −50 mV and the black traces 'fuse' with the red and blue traces of the action potentials without I Kur (horizontal arrows). At these negative potentials, I Kur becomes small because of both deactivation-rather than inactivation, which is much slowerand diminishing driving forces. The maximum I Kur during the atrial-like AP is slightly larger compared to the ventricular-like AP because in this particular example the atrial-like AP reaches a higher peak than the ventricular-like AP, which results in a larger driving force for I Kur .

Effects of I Kur Loss-of-Function Mutations in Atrial-and Ventricular-Like hiPSC-CMs
Next, we studied the effects of loss-of-function mutations in KCNA5, resulting in a decrease in I Kur . Therefore, we decreased the fully activated conductance of the virtual I Kur conductance to 75, 50, 25, and 12.5% of its control value. in AP parameters are shown in Figures 6, 7. In both atrialand ventricular-like hiPSC-CMs, APD 20 , APD 50 , and APD 90 significantly increased upon a reduction in I Kur (Figures 6A-F). However, while the increase in APD 90 in atrial-like APs is only present upon severe I Kur reduction, APD 90 prolongation in ventricular-like APs is already present at a mild reduction (Figures 6E,F). APPlatA was significantly increased in both atrial-and ventricular-like APs (Figures 6G,H). RMP and dV/dt max were unaffected (Figures 7A-D), whereas a slight increase in APA was observed, but only in atrial-like hiPSC-CMs at severe reductions of I Kur (Figures 7E,F).

Effects of I Kur Gain-of-Function Mutations in Atrial-and Ventricular-Like hiPSC-CMs
Finally, we studied the effects of gain-of-function mutations in KCNA5, resulting in an increase in I Kur . Therefore, we increased the fully activated conductance of the virtual I Kur conductance to 125, 150, 175, and 200% of its control value. Figure 9 shows typical examples of the effects on the APs of atrial-and ventricular-like hiPSC-CMs. The average changes in AP parameters are shown in Figures 6, 7. In both atrial-and ventricular-like hiPSC-CMs, APD 20 and APD 50 significantly shortened, but only when I Kur was strongly increased (Figures 6A-D). APD 90 was significantly reduced in ventricularlike, but not in atrial-like hiPSC-CMs (Figures 6E,F). APPlatA only showed a significant decrease in ventricular-like hiPSC-CMs (Figures 6G,H). Other AP parameters were unaffected upon increases in I Kur (Figure 7).

DISCUSSION
Overall, the APs of our atrial-like hiPSC-CMs were substantially shorter and had a lower AP plateau than those of our ventricularlike hiPSC-CMs, in qualitative agreement with previous studies on atrial-and ventricular-like hiPSC-CMs (Marczenke et al., 2017;Verkerk et al., 2017;Argenziano et al., 2018;Cyganek et al., 2018;Lemme et al., 2018;Veerman et al., 2019). There are some quantitative differences in AP parameters with previous studies, but these are likely due to differences in cell lines, differences in differentiation protocols, absence or presence of I K1 injection, and a different definition of AP plateau amplitude. The differences in AP parameters of our atrial-and ventricular-like hiPSC-CMs would have been even larger if we had supplied our atrial-like hiPSC-CMs with a more atrial-specific I K1 , as not only observed in human heart , but also in canine, murine and sheep heart (Dhamoon et al., 2004;Panama et al., 2007;Cordeiro et al., 2015). Of note, Fabbri et al. (2019) recently published a detailed in silico study of the effects of several I K1 formulations on AP duration of hiPSC-CMs.
Maximum sustained native I Kur density was larger in our atrial-like hiPSC-CMs than in our ventricular-like hiPSC-CMs. Yet, with a value of 1.88 ± 0.49 pA/pF at +50 mV, the I Kur density of our atrial-like hiPSC-CMs was small in comparison with that of freshly isolated human atrial myocytes, for which Amos et al. (1996) observed a density of 5.1 ± 0.3 pA/pF for peak I Kur and 4.7 ± 0.2 pA/pF for late I Kur during a 300-ms voltage clamp step to +40 mV at 22 • C (114 cells, 32 hearts). Therefore, we decided to block native I Kur and use dynamic clamp to study the effects of I Kur , using a virtual I Kur with characteristics, including its density, based on observations made in freshly isolated human atrial myocytes.
Due to the relatively low native I Kur density of our atrial-and ventricular-like hiPSC-CMs, it was not surprising that blockade of I Kur by 4-AP had only minor effects on AP parameters. When native I Kur was replaced with virtual I Kur with 100% density, similar to I Kur density in freshly isolated human atrial myocytes, more pronounced effects on AP parameters were observed. In atrial-like hiPSC-CMs, APD 20 shortened substantially, whereas APD 50 shortened only moderately and APD 90 even less so. In ventricular-like hiPSC-CMs, on the other hand, not only  APD 20 , but also APD 50 and APD 90 shortened substantially upon injection of virtual I Kur . The more pronounced effect on APD in ventricular-like hiPSC-CMs is likely related to the longer and more positive AP plateau potentials leading to more functional consequences of I Kur . The observed decrease in APD 20 was accompanied by a lowering of the AP plateau in both atrial-and ventricular-like hiPSC-CMs.
We only performed experiments at a pacing rate of 1 Hz and not at higher pacing rates. Therefore, we were unable to confirm that the relative contribution of I Kur to AP repolarization increases with increasing pacing rate (Ford et al., 2016;Aguilar et al., 2017). Aguilar et al. (2017) carried out comprehensive computer simulations with the Courtemanche et al. (1998) human atrial myocyte model, in which the I Kur formulation was updated in accordance with the experimental observations on I Kur inactivation by Feng et al. (1998). They found that I Kur did not inactivate significantly at high pacing rates and, consequently, the contribution of I Kur to repolarization was mainly determined by its (fast) activation kinetics. Accordingly, rate-dependent changes in I Kur were largely determined by changes in action potential morphology. In computer simulations with the Maleckar et al. (2009) model, on which we based our I Kur formulation, we made similar observations (data not shown). We aim to test the rate dependence of the effects of I Kur on AP repolarization in future experiments on hiPSC-CMs.
In both atrial-and ventricular-like hiPSC-CMs, simulation of loss-of-function mutations through lowering of the virtual I Kur density from 100% to 12.5-75% of its control value, resulted in prolongation of the AP and raise of its plateau, in line with the differences in AP parameters that were observed between 0 and 100% I Kur . Marczenke et al. (2017) found that knock-out of KCNA5 in hiPSC-CMs, representing a complete loss-of-function, may result in the development of EADs, which, however, were not observed in the present study, likely due to our higher pacing frequency. At I Kur densities > 100%, simulating gain-offunction mutations, effects on AP parameters were somewhat less pronounced. In both atrial-and ventricular-like hiPSC-CMs, APD 20 was only significantly decreased upon an increase in I Kur density to 200%. A significant lowering of the AP plateau, together with AP shortening, was only observed in ventricularlike hiPSC-CMs.
Although the sustained native I Kur density at +50 mV was small in our atrial-like hiPSC-CMs (1.88 ± 0.49 pA/pF, n = 17), it was still significantly larger than in our ventricular-like hiPSC-CMs (0.26 ± 0.26 pA/pF, n = 17). Within our atrial-like hiPSC-CM population we noted cells lacking I Kur (Figure 4B), although atrial-like hiPSC-CM generation using retinoic acid has been shown to generate 90-95% atrial-like hiPSC-CMs   (Cyganek et al., 2018), with the rest being sinus-or ventricularlike hiPSC-CMs. Since hiPSC-CMs display an immature phenotype, it is possible that not all atrial-like hiPSC-CMs have developed I Kur densities large enough to be detected as 4-AP sensitive current in a voltage clamp setting. Our recorded I Kur densities are lower than those of the only other known quantification of I Kur density in atrial-and ventricular-like hiPSC-CMs (Kaplan et al., 2016). In the study by Kaplan et al. (2016), which has only been published in abstract form, the sustained I Kur density at +50 mV amounted to 3.71 ± 0.55 pA/pF (n = 5) in atrial-like hiPSC-CMs, which was significantly larger than that of ventricular-like hiPSC-CMs (1.00 ± 0.10 pA/pF, n = 16). To distinguish between the two types of hiPSC-CMs based on I Kur densities would require further investigation, although the available data suggest a trend of a significantly larger I Kur density in atriallike hiPSC-CMs.
Of note, all AP parameters of our atrial-and ventricularlike hiPSC-CMs except dV/dt max and APA were not only different under control conditions, but also upon blockade of I Kur by 4-AP, indicating that the two types of hiPSC-CMs are not only different in their level of K v 1.5 expression, as determined by I Kur density, and suggesting that differences in membrane currents other than I Kur also contribute to the observed differences in AP parameters. This result is in line with previous findings by both Marczenke et al. (2017) and Lemme et al. (2018), who found that knock-out of KCNA5 or I Kur blockade by 4-AP in atrial-like hiPSC-CMs did not result in completely ventricular-like APs. These findings are, however, to some extent at odds with those by Kaplan et al. (2016), who noticed that the APs of their atrial-like hiPSC-CMs took on a ventricular-like shape when treated with 4-AP, which strongly suggested that I Kur is the major determinant of atrial action potential morphology. Conversely, they observed that injection of a virtual I Kur in ventricular-like hiPSC-CMs, employing the dynamic clamp technique using oocytes expressing a cloned K v 1.5 current, resulted in APs similar to those of atrial-like hiPSC-CMs. Apart from the studies by Kaplan et al. (2016), Marczenke et al. (2017), and Lemme et al. (2018), data on I Kur in hiPSC-CMs are limited and the electrophysiology of I Kur in atrial-and ventricular-like hiPSC-CMs remains largely unknown.
Apart from demonstrating a link between altered I Kur density and changes in AP parameters, I Kur has now been quantified in both atrial-and ventricular-like hiPSC-CMs. Thus, the present study provides additional data toward a complete characterization of individual membrane currents in hiPSC-CMs. Moreover, our study illustrates the potentials of dynamic clamp experiments on hiPSC-CMs, allowing manipulation of characteristics of the injected current in real time, thus facilitating direct, systematic, and efficient testing of changes in those characteristics. In the context of studying drug effects, including effects of anti-AF drugs, dynamic clamp may prove useful in the identification of potential drug targets and in testing modelbased hypotheses (Ortega et al., 2018). For instance, dynamic clamp experiments on atrial-like hiPSC-CMs with I Kur based on specific loss-or gain-of-function mutations in KCNA5 can be utilized to assess the cellular effects of these mutations as well as effects of dedicated pharmacological treatment through modulation of I Kur . Ultimately, this may lead to mutation-specific treatment of AF.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
SH designed and performed the experiments, analyzed the data, and drafted the manuscript. HD cultured the hiPSC line and developed the procedures to generate atrial-like and ventricular-like hiPSC-CMs. LB prepared the hiPSC-CMs used for electrophysiology in the present study. AV and RW designed the study, interpreted the data, and drafted, edited, and approved the manuscript.