Contribution of potassium channels to action potential repolarization of human embryonic stem cell‐derived cardiomyocytes

Abstract

Background and Purpose

The electrophysiological properties of human pluripotent stem cell‐derived cardiomyocytes (CMs) have not yet been characterized in a syncytial context. This study systematically characterized the contributions of different repolarizing potassium currents in human embryonic stem cell‐derived CMs (hESC‐CMs) during long‐term culture as cell monolayers.

Experimental Approach

The H9 hESC line was differentiated to CMs and plated to form confluent cell monolayers. Optical mapping was used to record the action potentials (APs) and conduction velocity (CV) during electrophysiological and pharmacological experiments. RT‐PCR and Western blot were used to detect the presence and expression levels of ion channel subunits.

Key Results

Long‐term culture of hESC‐CMs led to shortened AP duration (APD), faster repolarization rate, and increased CV. Selective block of I_Kr, I_Ks, I_K1, and I_Kur significantly affected AP repolarization and APD in a concentration‐ and culture time‐dependent manner. Baseline variations in APD led to either positive or negative APD dependence of drug response. Chromanol 293B produced greater relative AP prolongation in mid‐ and late‐stage cultures, while DPO‐1 had more effect in early‐stage cultures. CV in cell monolayers in early‐ and late‐stage cultures was most susceptible to slowing by E‐4031 and BaCl₂ respectively.

Conclusions and Implications

I_Kr, I_Ks, I_K1, and I_Kur all play an essential role in the regulation of APD and CV in hESC‐CMs. During time in culture, increased expression of I_Kr and I_K1 helps to accelerate repolarization, shorten APD, and increase CV. We identified a new pro‐arrhythmic parameter, positive APD dependence of ion channel block, which can increase APD and repolarization gradients.

Abbreviations

AP

action potential

APD

AP duration

APD₃₀

APD at 30% repolarization

APD₈₀

APD at 80% repolarization

CV

conduction velocity

hESC‐CM

human embryonic stem cell‐derived cardiomyocyte

hiPSC‐CM

human induced pluripotent SC‐CM

hPSC‐CM

human pluripotent SC‐CM

IQR‐APD₈₀

interquartile range of APD₈₀

What is already known

• The electrophysiological properties of human pluripotent stem cell‐derived cardiomyocytes (hPSC‐CMs) have not yet been characterized in a syncytial context.

• hPSC‐CMs express human ion channels similar to primary human CMs, but there is electrophysiological heterogeneity.

What this study adds

• The repolarization characteristics of hPSC‐CMs in terms of four major repolarizing potassium channels were systematically analyzed.

• Positive action potential duration dependence of ion channel block is a pro‐arrhythmic index that can augment action potential duration and repolarization gradients.

What is the clinical significance

• When hPSC‐CMs are used for drug screening, the electrophysiological heterogeneity and differences in drug response should be considered.

• Transplantation of hPSC‐CMs into impaired myocardium during regenerative therapies can result in a pro‐arrhythmic condition.

1. INTRODUCTION

Human pluripotent stem cell‐derived cardiomyocytes (hPSC‐CMs), including both induced pluripotent and embryonic stem cell‐derived cardiomyocytes (hiPSC‐CMs and hESC‐CMs, respectively) have great potential in drug screening because they are electrophysiologically similar to primary human CMs (Blazeski et al., 2012a; Mordwinkin, Burridge, & Wu, 2013; Van Den Heuvel, Van Veen, Lim, & Jonsson, 2014). Currently, all new drugs must be screened by I_Kr assay prior to FDA approval (International Conference on Harmonization, 2005) because block of this repolarizing current is the most common cause of arrhythmias associated with acquired long‐QT syndrome. Several studies have investigated I_Kr in single hPSC‐CMs (Gintant, Sager, & Stockbridge, 2016; Jonsson et al., 2010, 2012; Nalos et al., 2012; Navarrete et al., 2013), making these cells more relevant to the safety of pharmacology applications.

The electrophysiology of cardiomyocytes is determined by the activities of multiple cardiac ion channels (Liu, Laksman, & Backx, 2015; Nerbonne & Kass, 2005) and by the interactions within the population of cells. In adult cardiomyocytes, there are five major potassium currents involved in repolarization, which affect different phases of the action potential (AP): the rapid‐ and slow‐delayed rectifier voltage‐gated K⁺ currents (I_Kr and I_Ks, respectively), the inward‐rectifier K⁺ current (I_K1), the ultra‐rapid delayed K⁺ current (I_Kur), and the transient outward K⁺ current (I_to). The molecular components of ion channels carrying these currents (KCNH2, KCNQ1, KCNJ2, KCNA5, and KCND3 genes, encoding K_v11.1, K_ir2.1, K_v7.1, K_v1.5, and K_v4.3 subunits of the channels, respectively) except perhaps for I_to are expressed in hPSC‐CMs. The expression and functional role of K⁺ currents in hPSC‐CMs have been studied (Otsuji et al., 2010; Peng, Lacerda, Kirsch, Brown, & Bruening‐Wright, 2010; Sayed, Liu, & Wu, 2016), but some currents (I_Ks and I_Kur in particular) have been largely overlooked. Further, the role of K⁺ currents in syncytial populations of hPSC‐CMs and across different stages of differentiation, where the expression levels of these ion channels vary, has not been studied.

Thus, the goals of the present study were to (a) culture hESC‐CMs in centimetre‐sized confluent monolayers, so their electrophysiology could be studied in a large population and syncytial context, (b) systematically determine any alteration in the distribution of AP durations (APDs) within the monolayer, the average monolayer conduction velocity (CV) following block of four major repolarization potassium channels (I_Kr, I_Ks, I_K1, and I_Kur), and (c) identify the changes in electrical phenotype and expression of these repolarization channels over time in culture to delineate the utility and limitation of hPSC‐CMs in pharmacological studies on safety of a particular drug.

2. METHODS

2.1. Cell culture and differentiation

The H9 human ESC line was cultured and differentiated by a monolayer‐based protocol as described previously (Bhattacharya et al., 2014). Briefly, undifferentiated hESC monolayers were maintained in E8 media (ThermoFisher Scientific, Waltham, MA) on Geltrex (ThermoFisher Scientific)‐coated tissue culture plates prior to differentiation. A directed differentiation protocol was used to differentiate hESC into the cardiac lineage. At Day 0 of differentiation, media were switched to RPMI 1640 (ThermoFisher Scientific) with B27 without insulin (ThermoFisher Scientific) until Day 9. Media were supplemented with 6‐μM CHIR‐99021 (Selleckchem, Houston, TX) for 48 hr (Day 0 to Day 2) and then 5‐μM IWR‐1 (Sigma‐Aldrich, St. Louis, MO) for 48 hr (Days 3–5). On Day 9, media were switched to RPMI 1640 with B27 with insulin (ThermoFisher Scientific). On Days 10–12, cells were dissociated using 0.05% trypsin and re‐plated on Geltrex‐coated tissue culture plates at a density of 250,000 cells·cm⁻² to form confluent 1‐cm²‐diameter monolayers using a PDMS stencil. The hESC‐CM monolayers were maintained in culture for 15–70 days before electrophysiology and biology experiments were performed.

2.2. Electrophysiological studies

2.2.1. Samples

The monolayers for optical mapping were seeded under the same conditions, after which they were assigned a unique identifier number to keep track of the experiment results. Cells were seeded sequentially at similar times in wells in a culture plate under the same conditions.

2.2.2. Optical mapping

hESC‐CMs monolayers were optically mapped during early‐, middle‐, and late‐stage culture, which were Days 25–30, Days 40–45, and Days 75–80 post‐differentiation, respectively. The monolayers were placed in Tyrode's solution in a 35‐mm dish and stained with 10‐μM of the voltage‐sensitive dye di‐4‐ANEPPS (Sigma‐Aldrich) for 10 min. Then this solution was replaced with Tyrode's solution with 10‐μM blebbistatin (Sigma‐Aldrich) to inhibit contraction and motion artefacts. At least 10 min after the addition of blebbistatin, the monolayer was stimulated by point pacing, which was applied through a pair of platinum electrodes placed at the edge of the monolayer. A battery‐powered voltage stimulator (Digitimer, FL) was used to deliver 10‐ms rectangular voltage pacing pulses. The pacing rate for the baseline study was varied from 0.5 to 2 Hz and for the pharmacological experiments was 1 Hz. Temperature was controlled at 37°C throughout all experiments. Optical APs were recorded using a 100 × 100 pixel CMOS camera (MiCAM Ultima‐L, SciMedia, Costa Mesa, CA). The excitation light (2 × 150‐W halogen lights, MORITEX, Japan) was filtered by a 531/40‐nm bandpass, and the emission was filtered by a 610‐nm longpass. The optics are composed of a tandem combination of two 1.0× PLANAPO lenses (Leica), resulting a 1 cm × 1 cm field of view. The images were acquired at 500 frames·s⁻¹.

Mapping data were analysed using automated, unbiased, custom MATLAB (RRID:SCR_001622) scripts; therefore, blinding of the experimental condition was not necessary. Optical recordings were convolved with a 5 × 5 Gaussian filter to remove excessive noise. Low‐signal channels were excluded from analysis manually. Activation maps were constructed using the time of maximum upstroke rate of the AP. CV, APD at 30% and 80% repolarization (APD₃₀ and APD₈₀, respectively), and triangulation (time from APD₃₀ to APD₈₀) were calculated from the optical voltage signal.

2.3. Pharmacological experiments

The concentrations of each drug to be tested were predetermined and tabulated prior to commencement of the experiments for that drug. During the experiments, individual monolayers were given the identical sequence of drug concentrations, starting with a no drug condition.

E‐4031 (Tocris, Minneapolis, MN) was dissolved in DMSO at 10 mM as the stock solution and serially diluted in Tyrode's solution to 2 μM, 500 nM, 200 nM, and 20 nM as the test solutions. (−)‐[3S,4R]‐Chromanol 293B (Tocris) was dissolved in DMSO at 10 mM as the stock solution and serially diluted in Tyrode's solution to 60, 30, 10, 5, and 1 μM. DPO‐1 (Tocris) was dissolved in 100% ethanol at 10 mM as the stock solution and serially diluted in Tyrode's solution to 1 μM, 300 nM, 100 nM, 50 nM, and 20 nM. BaCl ₂ was dissolved in ddH₂O at 1 M as the stock solution and serially diluted in Tyrode's solution to 1 mM, 500 μM, 250 μM, 50 μM, and 10 μM. All stock solutions were stored at −20°C until use. Dose–response experiments were performed by exchanging the test solution into the mapping chamber to reach the desired test concentrations serially. Single‐dose experiments were not feasible due to the large number of cells required for each monolayer; therefore, blinding of drug concentrations was not possible. After each drug addition, 10 min were given to allow the drug binding to reach a relative steady state before recording. For each drug, a group size of n = 7 monolayers was used for each time point. However, some monolayers detached from the culture vessel (n = 1, 3, 3, and 1 for E‐4031, chromanol 293B, DPO‐1, and BaCl₂, respectively) and were excluded from the experiments. Some other monolayers became quiescent with application of certain concentrations of drug: For E‐4031, two monolayers became quiescent at the concentration of 500 nM, and one became quiescent at the concentration of 2 μM; and for BaCl₂, two monolayers became quiescent at the concentration of 250 μM. All five of these monolayers were excluded from data analysis. But otherwise, n = 7 for all other experiments. The mapping data were recorded as described above.

2.4. Western blot analysis

hESC‐CMs were lysed with RIPA buffer (ThermoFisher Scientific) supplemented with protease inhibitor (Sigma‐Aldrich) and phosphatase inhibitor cocktail (Sigma‐Aldrich) on Days 25–30, Days 40–45, and Days 75–80. For K_v1.4 and K_v4.3, lysate from porcine heart tissue was used as positive control. All lysates were centrifuged at 12,000 g for 15 min at 4°C. Total protein was quantified with a Pierce BCA assay (ThermoFisher Scientific). Thirty micrograms of protein lysates were resolved on 4–12% mini Bis‐Tris gels (Bio‐Rad, Hercules, CA) and transferred onto PVDF membranes for immunoblotting. Membranes were incubated with primary antibodies diluted in 5% milk at appropriate concentrations for K_v11.1 (1:200, Santa Cruz, sc‐377388, RRID: N/A), K_ir2.1 (1:200, Abcam, ab65796, RRID:AB_1140953), K_v7.1 (1:1,000, Santa Cruz, sc‐20816, RRID:AB_2131551), K_v1.5 (1:1,000, Alomone Labs, APC‐004, RRID:AB_2040156), K_v4.3 (1:1,000, NeuroMab, 75‐017, RRID:AB_10672856), K_v1.4 (1:500, Abcam, ab191052, RRID: N/A), and GAPDH (1:1,000, EMD Millipore, MAB374, RRID:AB_2107445) overnight at 4°C. Secondary antibodies (1:20,000, Li‐COR, 925‐32210, RRID: N/A, 925‐32211, RRID: N/A, 925‐68070, RRID: N/A) were used and imaged by Odyssey CLX (LI‐COR) as per the manufacturer's protocol. Detailed information on antibodies used in this study can be found in Table 1. Protein band intensities were quantified using ImageJ software (RRID:SCR_003070). All measurements and calculations were conducted without knowledge of the treatments. The immuno‐related procedures used comply with the recommendations made by the British Journal of Pharmacology.

Table 1.

Summary of antibodies used in this study

Antibody name	Commercial source	Cat#	Species	Epitope	Isotype	RRID	Diluting buffer	Dilution
Kv11.1	Santa Cruz	sc‐377388	Mouse monoclonal	Amino acid 96–120 of HERG	IgG1	N/A	5% milk	1:200
Kir2.1	Abcam	ab65796	Rabbit polyclonal	Synthetic peptide corresponding to Human Kir2.1 (N terminal)	IgG	AB_1140953	5% milk	1:200
Kv7.1	Santa Cruz	sc‐20816	Rabbit polyclonal	Amino acid 547–676 of KCNQ1	IgG	AB_2131551	5% milk	1:1,000
Kv1.5	Alomone Labs	APC‐004	Rabbit polyclonal	Amino acid residues 513–602 of mouse Kv1.5	IgG	AB_2040156	5% milk	1:1,000
Kv4.3	NeuroMab	75‐017	Mouse monoclonal	Amino acids 415–636 (cytoplasmic C‐terminus) of Kv4.3	IgG1	AB_10672856	5% milk	1:1,000
Kv1.4	Abcam	ab191052	Rabbit polyclonal	Synthetic peptide corresponding to Human Kv1.4 aa 329–344	IgG	N/A	5% milk	1:500
GAPDH	EMD Millipore	MAB374	Mouse monoclonal	GAPDH from rabbit muscle	IgG1	AB_2107445	5% milk	1:500
IRDye® 800CW goat anti‐mouse IgG	LI‐COR	925‐32210	Goat	IgG	IgG	N/A	5% milk	1:10,000
IRDye® 800CW goat anti‐rabbit IgG	LI‐COR	925‐32211	Goat	IgG	IgG	N/A	5% milk	1:10,000
IRDye® 680RD goat anti‐mouse IgG	LI‐COR	925‐68070	Goat	IgG	IgG	N/A	5% milk	1:10,000

2.5. RNA isolation and RT‐PCRs

Total RNA of hESC‐CMs was isolated on Days 25–30, Days 40–45, and Days 75–80 using Trizol (ThermoFisher Scientific) according to the manufacturers protocol. RNA (2 μg) was reverse transcribed using cDNA Reverse Transcription Kit (ThermoFisher Scientific). The resulting cDNA (1 μl) was amplified in a 20‐μl reaction system containing 10‐μl 2× CR master mix (ThermoFisher Scientific), 7‐μl deionized water, and 2‐μl PCR primer. PCR was performed with the following cycle parameters: one cycle of denaturation at 94°C, 60 s, followed by 35 cycles of denaturation at 94°C, 30 s; annealing at 58°C, 30 s; and extension at 72°C, 30 s. Reaction products (12 μL) were loaded onto 1.2% agarose gels containing SYBR Safe DNA Gel Stain (ThermoFisher Scientific). PCR products were visualized with fluorescence illumination and photographed. Primer sequence and the size of the PCR products were as follows: KCND3 sense, CCTGCGCTTGCTTAGTGTTG, and KCND3 antisense, CATCCTGCCGCTTGTTCTTG, with a cDNA amplification product of 314 bp; KCNA4 sense, CTGCCAAACCCGAGTGATTC, and KCNA4 antisense, GCAAAGACCCTGGGAGGTAG, with a cDNA amplification product of 280 bp; and GAPDH sense, CCCACTCCTCCACCTTTGAC, and GAPDH antisense, CCACCACCCTGTTGCTGTAG, resulting in a 108‐bp product.

2.6. Data and statistical analysis

All monolayers were treated as independent biological samples. The particular drug treatment for each monolayer was preassigned regardless of any attributes of the monolayer other than its identifier number prior to the start of experiments.

Data are presented as mean ± SEM. Statistical analysis was performed using Prism 6 (GraphPad Software, Inc., RRID:SCR_002798). Student's unpaired two‐tailed t test was used for two‐group comparisons, and one‐way ANOVA followed by Tukey's post hoc tests was conducted for multiple comparisons, the significance and confidence level was .05 (95% confidence interval), homogeneity of variance was tested before one‐way ANOVA, and no variance inhomogeneity was observed (P > .05 for all data). Responses to drugs were normalized and compared with baseline via Student's two‐tailed t test. P < .05 was considered statistically significant. The data and statistical analysis comply with the recommendations of the British Journal of Pharmacology on experimental design and analysis in pharmacology.

2.7. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017).

3. RESULTS

3.1. Baseline electrophysiology and CV studies of hESC‐CM monolayers

All monolayers used in this study exhibited spontaneous activities in culture (typically <0.5 Hz). Early hESC‐CM monolayers were point stimulated at different rates, and APs and their propagation were recorded using optical mapping (Figure 1a). APs were normalized from 0 to 1, given that the optical signal is only indicative of relative and not absolute voltage. A representative histogram of the distribution of APD₈₀ for a monolayer containing 6,000 to 7,000 recording sites over an area of ~0.79 cm² is shown (Figure 1b). At 1 Hz, monolayers (n = 14) had ventricular‐like APs with average APD₈₀ of 345 ± 7 ms and APD₃₀ of 216 ± 6 ms. Average CV was 8.5 ± 0.2 cm·s⁻¹. At higher pacing rates of 1.5 and 2 Hz, APD₈₀ and average CV had negative rate dependencies (Figure 1c,d). Also, at these higher pacing rates, 2:1 block was sometimes observed, where the monolayer failed to maintain 1:1 capture to electrical pacing.

Figure 1.

Mapping of H9 hESC‐CM monolayers. (a) Sample activation map. (b) APD₈₀ distribution of a monolayer paced at 1, 1.5, and 2 Hz, showing shorting of APD₈₀. (c) Action potential traces for 1‐, 1.5‐, and 2‐Hz pacing. (d) Average conduction velocity at pacing rate between 0.5 and 2 Hz. n = 14

3.2. Changes in electrophysiological parameters in response to known blockers of repolarizing potassium channels

Next, we assessed drug‐induced effects on APD and CV of early hESC‐CM monolayers. The responses to ion channel blockers were tested at a constant pacing rate of 1 Hz. All data are summarized in Table 2.

Table 2.

Summary of drug‐induced effects of human embryonic stem cell‐derived cardiomyocyte monolayers at varying times in culture

Drug	Culture time (days)	Dose (μM)	CV (cm·s−1)	APD80 (ms)	APD30 (ms)	APD80–APD30 (ms)
E‐4031	25–30 (n = 7)	Baseline	9.2 ± 0.5	345.6 ± 8.6	232.7 ± 5.5	112.9 ± 3.8
		0.02	6.8 ± 0.2*	445.3 ± 17.9*	272.9 ± 7.3*	172.4 ± 11.7*
		0.2	3.6 ± 0.1*	625.6 ± 9.1*	276.7 ± 4.2*	348.9 ± 11.2*
		0.5	3.2 ± 0.2*	640.4 ± 8.5*	275.3 ± 5.4*	365.1 ± 9.5*
		2	3.1 ± 0.2*	637.2 ± 6.3*	271.2 ± 4.2*	366.0 ± 7.8*
	40–45 (n = 7)	Baseline	8.0 ± 0.7	256.7 ± 13.9	149.7 ± 6.1	107.0 ± 9.7
		0.02	7.3 ± 0.5	362.4 ± 25.5*	179.3 ± 6.6	183.1 ± 20.1*
		0.2	5.7 ± 0.3*	509.7 ± 28.1*	199.9 ± 5.1*	309.9 ± 23.8*
		0.5	5.1 ± 0.3*	548.3 ± 28.9*	209.1 ± 7.2*	339.1 ± 23.9*
		2	5.0 ± 0.4*	565.9 ± 29.6*	222.6 ± 8.3*	343.3 ± 25.3*
	75–80 (n = 6)	Baseline	8.6 ± 0.5	226.0 ± 12.2	151.3 ± 6.7	74.7 ± 6.2
		0.02	9.0 ± 0.7	293.2 ± 12.2*	176.8 ± 4.9	116.3 ± 8.7*
		0.2	7.7 ± 0.8	458.8 ± 28.3*	186.0 ± 5.4	272.8 ± 24.5*
		0.5	8.3 ± 0.9	487.4 ± 38.0*	191.4 ± 8.7	296.0 ± 32.0*
		2	7.3 ± 1.0	517.7 ± 25.1*	207.8 ± 13.9*	309.8 ± 24.9*
BaCl2	25–30 (n = 6)	Baseline	9.1 ± 0.3	386.4 ± 15.0	227.6 ± 6.5	169.9 ± 6.1
		10	8.9 ± 0.3	393.0 ± 8.1	221.7 ± 4.2	168.0 ± 4.5
		50	9.0 ± 0.3	440.9 ± 7.9*	256.0 ± 2.6*	182.1 ± 5.2
		250	8.1 ± 0.2*	450.3 ± 9.8*	280.7 ± 4.1*	168.7 ± 8.3
		500	7.2 ± 0.3*	464.9 ± 8.3*	295.6 ± 2.6*	169.1 ± 5.7
	40–45 (n = 7)	Baseline	8.4 ± 0.3	303.3 ± 13.2	126.7 ± 7.0	176.5 ± 7.6
		10	9.1 ± 0.1	338.3 ± 8.9*	138.7 ± 5.6	198.5 ± 9.1
		50	8.4 ± 0.3	385.5 ± 7.4*	148.2 ± 5.8	237.3 ± 10.1*
		250	7.1 ± 0.3*	463.8 ± 11.0*	159.2 ± 6.5*	305.0 ± 16.9*
		500	5.1 ± 0.2	490.0 ± 19.5*	163.0 ± 6.2*	328.3 ± 24.8*
	75–80 (n = 7)	Baseline	7.7 ± 0.4	305.3 ± 17.8	131.4 ± 4.3	172.9 ± 14.3
		10	7.9 ± 0.5	382.4 ± 12.4*	158.3 ± 3.6*	221.4 ± 13.2*
		50	7.6 ± 0.5	428.7 ± 12.4*	177.9 ± 4.9*	248.6 ± 15.3*
		250	6.2 ± 0.5*	418.1 ± 8.2*	194.0 ± 7.5*	251.1 ± 17.2*
		500	3.4 ± 0.3*	350.3 ± 8.5*	207.6 ± 7.2*	145.3 ± 3.9
Chromanol 293B	25–30 (n = 7)	Baseline	8.7 ± 0.3	344.6 ± 11.0	199.9 ± 7.3	140.6 ± 6.1
		1	8.8 ± 0.5	376.4 ± 15.4	221.1 ± 8.8	150.6 ± 7.7
		5	8.9 ± 0.5	393.7 ± 15.3*	235.6 ± 7.5*	154.1 ± 7.9
		10	9.1 ± 0.4	408.6 ± 12.4*	247.1 ± 5.2*	157.0 ± 8.1
		30	8.8 ± 0.2	421.1 ± 13.7*	262.3 ± 5.9*	157.7 ± 8.4
		60	8.3 ± 0.3	441.0 ± 16.4*	278.8 ± 7.6*	160.8 ± 10.7
	40–45 (n = 6)	Baseline	8.7 ± 0.9	250.3 ± 7.2	159.0 ± 8.6	91.3 ± 2.5
		1	9.7 ± 0.7	273.7 ± 12.9	175.5 ± 8.4	98.2 ± 8.9
		5	10.1 ± 0.9	304.8 ± 16.8*	200.2 ± 11.9*	104.6 ± 11.1
		10	10.2 ± 0.8	315.0 ± 17.1*	209.5 ± 12.7*	105.5 ± 10.7
		30	10.0 ± 1.0	350.8 ± 21.6*	240.4 ± 13.8*	110.4 ± 9.9
		60	9.5 ± 0.7	359.0 ± 19.7*	250.8 ± 14.5*	108.2 ± 8.0
	75–80 (n = 5)	Baseline	8.6 ± 0.6	186.2 ± 10.9	124.0 ± 7.7	62.2 ± 4.3
		1	9.7 ± 0.6	207.8 ± 5.8	140.0 ± 4.0	67.8 ± 2.0
		5	10.1 ± 0.6	223.2 ± 8.8*	152.0 ± 5.8*	71.2 ± 3.2
		10	10.5 ± 0.6	233.4 ± 10.9*	159.8 ± 8.2*	73.6 ± 3.4
		30	10.6 ± 0.6	253.8 ± 14.6*	175.6 ± 11.7*	78.2 ± 3.8
		60	10.8 ± 0.6	268.6 ± 16.1*	189.2 ± 13.4*	79.4 ± 4.2
DPO‐1	25–30 (n = 6)	Baseline	10.0 ± 0.8	340.7 ± 20.9	183.7 ± 23.3	155.0 ± 14.0
		0.02	9.5 ± 0.4	404.8 ± 26.9	241.0 ± 30.9	158.8 ± 14.7
		0.05	10.0 ± 0.6	432.7 ± 24.3*	256.3 ± 31.6*	169.5 ± 17.6
		0.3	10.6 ± 0.4	477.5 ± 24.0*	311.8 ± 33.9*	164.0 ± 17.0
		1	9.9 ± 0.5	518.8 ± 13.6*	356.2 ± 21.4*	162.2 ± 14.4
	40–45 (n = 6)	Baseline	12.3 ± 0.3	295.3 ± 9.7	148.0 ± 2.3	157.8 ± 11.1
		0.02	13.5 ± 0.5	374.0 ± 10.6*	174.5 ± 5.0*	200.0 ± 12.7*
		0.05	14.3 ± 0.6	399.2 ± 14.4*	189.8 ± 5.1*	213.0 ± 16.1*
		0.3	14.6 ± 0.7	415.0 ± 13.2*	203.8 ± 7.1*	215.5 ± 14.1*
		1	14.5 ± 0.6	436.5 ± 7.8*	236.2 ± 6.0*	200.2 ± 11.5*
	75–80 (n = 6)	Baseline	11.8 ± 1.0	245.5 ± 7.4	144.2 ± 8.1	102.2 ± 5.9
		0.02	12.7 ± 0.7	253.8 ± 7.5	166.3 ± 6.7	86.8 ± 1.2
		0.05	13.6 ± 0.5	271.8 ± 5.8*	177.7 ± 6.1*	91.7 ± 1.0
		0.3	14.2 ± 0.4	292.8 ± 6.4*	196.0 ± 5.5*	95.5 ± 1.9
		1	13.8 ± 0.3	345.0 ± 8.2*	246.7 ± 7.7*	97.1 ± 1.8

Note. Data are mean ± SEM. Blue colour and * indicate significant decrease, and orange colour and * indicate significant increase, with *P < .05 when compared with baseline.

Abbreviations: APD₃₀, action potential duration at 30% repolarization; APD₈₀, action potential duration at 80% repolarization; CV, conduction velocity.

3.3. Rapid delayed rectifier potassium current: I_Kr

I_Kr contributes prominently to the plateau and repolarization phases of the adult human AP. Four concentrations (0.02 to 2 μM) of the I_Kr blocker E‐4031 were tested (n = 7). At 0.02 and 0.2 μM, E‐4031 prolonged APD₈₀ by 29% and 81.3%, respectively. AP triangulation also increased by 53% and 209%, respectively. Increasing the dose to 0.5 and 2 μM produced no further increases of APD₈₀ or AP triangulation (Figure 2a–c(A)). No early afterdepolarizations (EADs) were induced in any of the seven monolayers, unlike in single hESC‐CMs (Gibson, Yue, Bronson, Palmer, & Numann, 2014; Izumi‐Nakaseko et al., 2014; Ma et al., 2011). CV also decreased by 26% and 61% by 0.02‐ and 0.2‐μM E‐4031, respectively (Figure 2d(A)), likely because the maximum diastolic potential is largely set by I_Kr, which then affects excitability (see Section 4).

Figure 2.

Responses to potassium channel block of hESC‐CM monolayers at increasing concentrations. (a) Representative optical APs before and after application of E‐4031 (A), chromanol 293B (B), BaCl₂ (C), and DPO‐1 (D). (b) APD₈₀ distributions of a monolayer paced at 1 Hz, before and after application of E‐4031 (A), chromanol 293B (B), BaCl₂ (C), and DPO‐1 (D). (c) Effect of various potassium channel blockers on APD₈₀, APD₃₀, and AP triangulation: (A) prolongation of APD₈₀, APD₃₀, and AP triangulation by E‐4031; (B) prolongation of APD₈₀ and APD₃₀ but little effect on triangulation by chromanol 293B; (C) prolongation of APD₈₀ and APD₃₀ but little effect on triangulation by BaCl₂; and (D) prolongation of APD₈₀ and APD₃₀ but little effect on triangulation by DPO‐1. *P < .05, significantly different as indicated. (d) Effect of various potassium channel blockers on conduction velocity: (A) slowing of conduction velocity by E‐4031, n = 7; (B) little effect on conduction velocity by chromanol 293B, n = 7; (C) decrease in conduction velocity by BaCl₂, n = 6; and (d) little effect on conduction velocity by DPO‐1, n = 6. *P < .05, significantly different as indicated

3.4. Slow delayed rectifier potassium current: I_Ks

I_Ks acts as a “repolarization reserve” (Varró & Baczkó, 2011) and backs up I_Kr as the repolarizing current. In hESC‐CM monolayers (n = 7), selective block of I_Ks by chromanol 293B (Bachmann, Quast, & Russ, 2001) caused concentration‐dependent (5–60 μM) slowing of repolarization (Figure 2a,b(B)), with significant prolongation of APD₈₀ and APD₃₀ by 14% and 18%, respectively, at a low dose of 5 μM (Figure 2c(B)), but with minimal effects on triangulation and CV even up to 60 μM (Figure 2c,d(B)). No EADs were observed.

3.5. Inward rectifier potassium current: I_K1

I_K1 contributes to late AP repolarization and setting of the resting membrane potential. I_K1 block using BaCl₂ (Sun, 1999; 10–500 μM) caused prolongation of APD₈₀ by 14%, 17%, and 21% at 50, 250, and 500 μM, respectively (n = 6; Figure 2a–c(C)), and no significant change was observed at 10 μM. AP triangulation was not significantly prolonged by BaCl₂ (Figure 2c(C)), and no EADs were induced. BaCl₂ also produced a decrease in CV at the higher concentrations of 250 and 500 μM (Figure 2d(C)).

3.6. Ultra‐rapid delayed rectifier potassium current: I_Kur

I_Kur contributes to early AP repolarization. DPO‐1, a selective blocker of I_Kur (Lagrutta, Wang, Fermini, & Salata, 2006), increased APD₈₀ (n = 6) in a concentration‐dependent manner (0.02–1 μM; Figure 2a,b(D)). At 0.02 and 1 μM, DPO‐1 prolonged APD₈₀ by 27% and 52% (Figure 2c(D)). There was not a significant change in AP triangulation (Figure 2c(D)) or CV (Figure 2d(D)), and no EADs were induced. We also tested against the possibility that the observed increase in APD by DPO‐1 was the result of non‐specific inhibition of other potassium channels. Pretreatment of the monolayers by E‐4031, chromanol 293B, or BaCl₂ did not prevent prolongation of repolarization by DPO‐1 (data not shown).

3.7. APD dependence of responses to repolarizing potassium current blockade

As shown in Figure 1b, we observed intrinsic variability in APs across hESC‐CM monolayers. To test whether regions with variable APs respond differently to the blockade of repolarizing potassium currents, we compared the change in APD across all 6,000–7,000 recording sites from each monolayer. Figure 3a shows the response of representative early‐stage hESC‐CM monolayers to blockage of I_Kr, I_Ks, I_K1, and I_kur (n = 7, 7, 6, and 6, respectively), where the changes in APD₃₀ (A–D) or APD₈₀ (E–H) at low (red) or high (blue) concentrations of drugs are plotted against APD in baseline conditions for each recording site. This analysis reveals three distinct types of APD dependence for repolarizing potassium current blockade: (a) negative APD dependence, where regions of shorter baseline APD have greater prolongation than regions of longer baseline APD, for example, APD₃₀ and APD₈₀ for high concentration (2 μM) of E‐4031 (blue clouds in Figure 3a(A,E)) and APD₈₀ for BaCl₂ (Figure 3a(G)) and DPO‐1 (Figure 3a(H)); (b) positive APD dependence, where regions of longer baseline APD have greater prolongation than regions of shorter baseline APD, for example, APD₃₀ and APD₈₀ for low concentration (20 nM) of E‐4031 (red clouds in Figure 3a(A,E)); and (c) absence of APD dependence, where APD prolongation does not correlate with baseline APD₃₀ or APD₈₀, for example, chromanol 293B (Figure 3a(B,F)) and APD₃₀ for BaCl₂ (Figure 3a(C)) and DPO‐1 (Figure 3a(D)). Strikingly, the APD dependence is opposite for low (20 nM) and high (2 μM) E‐4031 concentrations, being positive for the former and negative for the latter (see Section 4 for more detail).

Figure 3.

Local response to repolarizing potassium current blockade in hESC‐CM monolayers. (a) Change of APD₃₀ (A–D) and APD₈₀ (E–H) in response to potassium current blockade. Each point in the scatter plot corresponds to a single recording site within the hESC‐CM monolayer. Red indicates the lower concentration tested for each drug: 20 nM for E‐4031, 1 μM for chromanol 293B, 10 μM for BaCl₂, and 20 nM for DPO‐1. Blue indicates the higher concentration tested for each drug: 2 μM for E‐4031, 60 μM for chromanol 293B, 500 μM for BaCl₂, and 1 μM for DPO‐1. (b) Example of spatial APD₈₀ variability within two monolayers. APD₈₀ variability is plotted as the difference between local APD₈₀ and median APD₈₀ across all recording sites for each monolayer. Red/yellow indicates regions with APD₈₀ longer than the median, and blue/cyan indicates regions with APD₈₀ shorter than the median. (A) APD₈₀ variability with 0, 20 nM, and 2 μM of E‐4031. (B) APD₈₀ variability with 0, 1 μM, and 60 μM of chromanol 293B. (c) Relative change of APD₈₀ interquartile range in response to repolarizing potassium current blockade. Dashed line indicates IQR‐APD₈₀ without any drug (100%), and the two clusters for each drug correspond to the low and high concentrations for each drug as indicated in (a). Data are shown as mean ± SD. n = 7, 7, 6, and 6 for E‐4031, chromanol 293B, BaCl₂, and DPO‐1, respectively

Next, we investigated the impact of APD dependence on spatial AP variability. The deviation of APD₈₀ at each recording site from the median APD₈₀ of the monolayer was used to visualize AP variability across different monolayers (Figure 3b). Regions with APD₈₀ longer than its median are coloured red/yellow, while regions with APD₈₀ shorter than its median are coloured cyan/blue. Without any drugs, our hESC‐CM monolayers exhibit some baseline AP variability, often with clustering of regions having similar APDs, particularly at the edges and the centre of the monolayer (Figure 3b(A,B) baseline). The application of 20‐nM E‐4031 further increased the baseline AP variability because of positive APD dependence and, subsequently, increased the spatial gradient of APD₈₀, as indicated by the deeper red and blue in regions originally having longer and shorter APD₈₀, respectively. However, a higher concentration of 2‐μM E‐4031 eliminated the APD clustering and reduced the spatial gradients in APD₈₀ (Figure 3b(A)). Application of DPO‐1 to block I_Kur and BaCl₂ to block I_K1 also reduced APD₈₀ gradients, but not to the extent of eliminating the APD clustering. These observations are consistent with the negative APD dependencies observed in these cases, which act to reduce baseline AP differences. On the other hand, for drugs having little or no APD dependence, such as I_Ks blocker chromanol 293B, a change in the spatial pattern or gradient of APD₈₀ was not observed (Figure 3b(B)).

We also quantified APD variability by the interquartile range of APD₈₀ (IQR‐APD₈₀) across all recording sites (Figure 3c). Consistent with our earlier observations, a low concentration of E‐4031 increased IQR‐APD₈₀ in the cell monolayer, whereas a high concentration of E‐4031 reduced IQR‐APD₈₀ back to the baseline level. Chromanol 293B and BaCl₂ had little effect on IQR‐APD₈₀, whereas DPO‐1 at high concentration reduced IQR‐APD₈₀.

3.8. Comparison of repolarizing potassium currents contributing to APD and CV in hESC‐CM monolayers at longer culture times

To determine developmental changes in ion current expression, we recorded paced APs in monolayers at early‐stage (n = 14), middle‐stage (n = 12), and late‐stage (n = 11) culture using hESC‐CMs from the same batch of differentiated cells. The data are summarized in Table 2, containing measurements of CV, APD₈₀, APD₃₀, and AP triangulation before and after administration of different drugs at the three culture stages. With increasing time in culture, APD became shorter, and repolarization rate (reciprocal of triangulation) became faster (Figure 4a(A,B)). Although CV did not change over time at 1‐Hz pacing, at 1.5‐ and 2‐Hz pacing, it increased in middle‐ and late‐stage culture (Figure 4a(C)), suggesting an increasing level of excitability. Similar amounts of connexin 43 expression were found at all three stages (Figure 4b(A,D)), suggesting similar degrees of cell–cell coupling. In summary, late‐stage hESC‐CMs exhibited more mature AP characteristics with shortened duration, faster repolarization, and faster conduction at higher pacing rates.

Figure 4.

Comparison of electrophysiological properties and expression of repolarization potassium channel proteins of hESC‐CM monolayers at different times of culture. (a) Comparison of AP, CV property: (A) representative optical APs at different culture times; (B) average APD₈₀, APD₃₀, and AP triangulation at 1‐Hz pacing of hESC‐CM monolayers from different culture times; and (C) average CV at different pacing rates of monolayers from different culture times. Data from differentiation Days 25–30 (n = 14), 40–45 (n = 12), and 75–80 (n = 11). (b) Comparison of repolarization potassium channel proteins and connexin 43 expression at different culture times: (A) connexin 43; (B) HERG (KCNH2/K_v11.1) and K_ir2.1; (C) K_v1.5 and K_v7.1; and (D) relative protein expressions normalized to GAPDH. n = 5. Data from differentiation Days 25, 40, and 75. *P < .05, significantly different as indicated. (c) Gene and protein expression of I_to channel subunit at different culture times: (A) expression of K_v1.4 and K_v4.3 proteins; (B) expression of KCNA4 and KCND3; and (C) relative protein expressions normalized to GAPDH. n = 5. Data from differentiation Days 25, 40, and 75 and adult human left ventricle

We next compared the electrophysiological responses of the hESC‐CM monolayers at the three culture stages to blockers of I_Kr, I_Ks, I_K1, and I_Kur. E‐4031 produced significant prolongation of APD and slowing of repolarization rate (increased triangulation) at all three stages, with a saturating effect at 0.2 μM regardless of the culture stage. No EADs were observed at 1‐Hz pacing (Figure 5a(A)). With 0.2‐μM E‐4031, the relative prolongation of APD₈₀ was significantly larger in the late‐stage monolayers (Figure 5b(A)). Western blot analysis confirmed that expression of Kv11.1 was ~2× greater in late‐ versus early‐stage monolayers (Figure 4b(B,D)). In early‐stage culture, the APD dependence varied from positive to negative in the majority (five of seven) of monolayers as concentration increased from 20 nM to 2 μM and was always negative in the remaining two monolayers. In middle‐stage culture, most monolayers (five of seven) exhibited negative APD dependence at both low and high E‐4031 concentrations and changed from positive to negative in the remaining two monolayers. However, in late‐stage culture, APD dependence was absent in all monolayers at both low and high E‐4031 concentrations. Lastly, reduction of CV by E‐4031 was observed in early‐ and middle‐stage monolayers but not in late‐stage monolayers (Figure 5c,d(A)), which would be consistent with a diminished dependence of maximum diastolic potential on I_Kr and presumably greater dependence on I_K1, in older monolayers.

Figure 5.

Comparison of responses to E‐4031 and BaCl₂ of hESC‐CM monolayers at different times of culture. (a) Representative optical APs at different culture times, before and after application of E‐4031 (A) and BaCl₂ (B). (b) Prolongation of APD₈₀ or APD triangle in hESC‐CM monolayers at different culture times by E‐4031 (A) and BaCl₂ (B). *P < .05, significantly different as indicated. (c) Activation maps before and after application of drugs: (A) E‐4031 caused conduction slowing in hESC‐CM monolayers of Days 25–30, and (B) BaCl₂ caused conduction slowing in hESC‐CM monolayers of Days 75–80. (d) Change of CV response to E‐4031 (A) and BaCl₂ (B) over time. Data from differentiation Days 25–30 (n = 7), 40–45 (n = 7), and 75–80 (n = 6) in E‐4031 experiments and from differentiation Days 25–30 (n = 6), 40–45 (n = 7), and 75–80 (n = 7) in BaCl₂ experiments. *P < .05, significantly different as indicated

Pursuing this notion further, we tested the sensitivity of hESC‐CM monolayers to the I_K1 blocker BaCl₂. A concentration‐dependent prolongation of APD₈₀ was observed at all stages, but increased triangulation was only observed in middle‐ and late‐stage monolayers (Figure 5a,b(B)). With 500‐μM BaCl₂, CV decreased by 11%, 39%, and 56% for early‐, middle‐, and late‐stage monolayers, respectively (Figure 5d(B)), indicating increased dependence on I_K1. Furthermore, Western blot analysis confirmed that expression of K_ir2.1 was ~3× greater in late‐ versus early‐stage monolayers (Figure 4b(B,D)).

Next, application of chromanol 293B resulted in a significant dose‐dependent increase in APD₃₀ at all three stages (Figure 6a(A)). The relative increase in APD₃₀ was larger in middle‐ and late‐stage monolayers. At 30‐μM concentration, APD₃₀ was significantly prolonged by 24% in early‐stage monolayers, 33% in middle‐stage monolayers, and 29% in late‐stage monolayers (Figure 6b(A)). K_v7.1 was expressed in hESC‐CM monolayers and increased slightly in late‐stage culture (Figure 4b(C,D)). CV was unaffected at all three stages (Figure 6c(A)).

Figure 6.

Comparison of responses to chromanol 293B and DPO‐1 of hESC‐CM monolayers at different times of culture. (a) Representative optical APs at different culture times, before and after application of chromanol 293B (A) and DPO‐1 (B). (b) Prolongation of APD₃₀ in hESC‐CM monolayers at different culture times by chromanol 293B (A) and DPO‐1 (B). *P < .05, significantly different as indicated. (c) Change of CV response to chromanol 293B (A) and DPO‐1 (B) over time. Data from differentiation Days 25–30 (n = 7), 40–45 (n = 6), and 75–80 (n = 5) in chromanol 293B experiments and from differentiation Days 25–30 (n = 6), 40–45 (n = 6), and 75–80 (n = 6) in DPO‐1 experiments. *P < .05, significantly different as indicated

Finally, application of DPO‐1 had a large prolonging effect on APD₃₀ (Figure 6a(B)). At a dose of 0.3 μM, APD₃₀ increased by 72% in early‐stage monolayers, 38% in middle‐stage monolayers, and 37% in late‐stage monolayers (Figure 6b(B)), while CV was unaffected at all three stages (Figure 6c(D)). The expression of K_v1.5 was similar among all stages (Figure 4b(C,D)).

One other set of experiments was directed at confirming the absence of I_to in the H9 hESC‐CMs. Expression levels of molecular subunits of cardiac I_to channels and their transcripts were assayed. The results of the Western blot analysis are shown in Figure 4c(A) and show virtually no expression of K_v4.3, although expression of K_v1.4 (a second, slow component of I_to) was similar to that in adult human left ventricle (n = 5). Further, we used RT‐PCR to confirm the gene expression of KCND3 and KCNA4, where no expression of KCND3 could be observed in any monolayers of different ages (n = 5, Figure 4c(C)).

To summarize the changes in repolarizing currents over time in culture, late‐stage monolayers had shorter AP duration and faster repolarization rate compared with early‐stage monolayers. Protein analysis of ion channel subunits revealed significant increases in expression of KCNH2 and K_ir2.1, indicating that increased I_Kr and I_K1 are likely responsible for the observed AP changes. Furthermore, we found a diminishing sensitivity of CV to I_Kr block and increasing sensitivity to I_K1 block with time in culture, consistent with an increasing role of I_K1 to hyperpolarize and stabilize the resting potential, thereby increasing sodium channel availability that in turn increases CV. Lastly, I_Kur supplements I_Kr, I_K1, and I_Ks as one of the major repolarizing currents of the AP in H9 hESC‐CMs at all times in culture.

4. DISCUSSION

hESC‐CMs are heterogeneous in their electrophysiology, even within individual cell clusters (Zhu, Millrod, Zambidis, & Tung, 2016). Interestingly, within individual monolayers, their APs tended to be similar and mostly ventricular like, although still with some baseline variability. This AP variability may mediate the tissue‐level response to block of repolarizing potassium currents (Figure 3) and is manifested as negative, positive, or neutral (little or no) APD dependence.

Negative APD dependence may indicate a saturating effect of channel block. Cells with shorter APs can have more of a given outward current than those with longer APs, so that their APDs would be prolonged more by high degree channel block. This could account for the negative APD dependence of I_Kr block at high E‐4031 concentration, I_K1 block (seen in nine of 11 monolayers), I_Kur block (Figure 3a), and to a lesser degree, I_Ks block. Spatial gradients in APD were reduced with block of any of the currents and were eliminated by high degree I_Kr block (Figure 3b(A)), suggesting that the spatial clustering of APD observed in monolayers at baseline was primarily due to regional differences in I_Kr expression.

Positive APD dependence can be explained in terms of when during the AP, the outward current is being blocked. If the block is during early repolarization, there will be prolonged dwell time within a transmembrane voltage range corresponding to the “window” region of the inward L‐type Ca²⁺ current (I_CaL) in which the current can be sustained (Hirano, Moscucci, & January, 1992), thereby augmenting the prolongation. This effect is absent if the current block is during the AP plateau where the transmembrane voltage lies outside the Ca²⁺ current window. Indeed, at moderate levels of channel block, positive APD dependence was only observed with E‐4031, which affected APs during early repolarization (Figure 2a(A)), and not with chromanol 293B, BaCl₂, or DPO‐1, which mainly elevated the plateau with little change in repolarization (Figure 2a(B–D)). Interestingly, the APD dependence of E‐4031 can become negative or neutral with increased drug concentration or culture time.

Overall, negative APD dependence reduces APD gradients, while positive APD dependence increases them. APD gradients (reflecting cellular responses), in turn, contribute to repolarization gradients (a tissue response) after conduction times are factored in, and it is known that intensified repolarization gradients can be proarrhythmic (Akar, Laurita, & Rosenbaum, 2000). These tissue‐level AP characteristics, along with CV measurements, are manifested in the syncytial monolayer models used in this study, potentially making them a more comprehensive and faithful in vitro representation of human cardiac electrophysiology and drug response compared with single cells.

Few studies have systematically characterized the electrophysiological changes that occur in hPSC‐CMs over time in culture, particularly in a tissue context (Lee et al., 2010; Trieschmann et al., 2016). We systematically blocked four major potassium currents and analysed individual ion channel subunit expression. Such an understanding is important if these cells are used to assess pharmacological safety profile (Farkas & Nattel, 2010; International Conference on Harmonization, 2005). We omitted a study of I_to because (a) the notch during the early phase of the AP (a hallmark of I_to) was never observed in our hESC‐CMs; (b) the expressions of KCND3 and Kv4.3 were absent at all three stages of culture (Figure 4c(A–C)); and (c) although KCNA4 and Kv1.4 were expressed (Figure 4c(A–C)), this slow component of I_to requires very long periods to recover from inactivation (Liu, Kim, London, Morales, & Backx, 2011) and would not contribute substantially to repolarization at the pacing rates used in this study.

I_Kr contributes prominently to phases 2 and 3 and early phase 4 of the AP and is present in hPSC‐CMs with varying expression over time (Otsuji et al., 2010). We found that K_v11.1 expression doubled between 25 and 75 days in monolayers. E‐4031 caused a significant prolongation of APD₈₀ that was approximately 82% in early‐stage monolayers and even greater (100%) in late‐stage monolayers. Furthermore, CV was particularly sensitive to I_Kr block among early‐stage hESC‐CM monolayers but much less so in late‐stage monolayers. This difference may be explained by a greater influence of I_Kr in early‐stage culture on maximum diastolic resting potential, which regulates the availability of sodium channels that act to propagate the AP (Bett et al., 2013; Doss et al., 2012; Gibson et al., 2014).

In adult atrial and ventricular CMs, the resting potential is determined primarily by I_K1 (Miake, Marbán, & Nuss, 2003; Silva & Rudy, 2003), but low I_K1 and lack of channel expression are hallmarks of the immaturity of hPSC‐CMs (Goversen, van der Heyden, van Veen, & de, 2018). K_ir2.1 could be detected in our hESC‐CMs after 25 days in culture, with a significant increase of expression at 75 days. Furthermore, BaCl₂ prolonged APD₈₀, decreased CV, and increased spontaneous beating in hESC‐CM monolayers, consistent with the role of I_K1 to regulate late repolarization and resting potential. Additionally, the effect of BaCl₂ increased over time in culture, indicating an increased function of I_K1 that correlated well with increased Kir2.1 expression, and can explain why CV was less affected by I_Kr block in older monolayers. One potential caveat of using BaCl₂ as I_K1 blocker is its off‐target effect on I_CaL. The highest BaCl₂ concentration was limited to 0.5 mM (lower than the 1.8‐mM Ca²⁺ concentration in the Tyrode's solution), and we verified that 0.5‐mM Ba²⁺ does not significantly alter the Ca²⁺‐dependent inactivation of I_CaL (data not shown).

In adult mammals, I_Ks is regulated by the sympathetic nervous system and produces AP abbreviation when heart rates increase (Jost, Papp, & Varro´, 2007). Blocking the channel in isolated preparations generally has limited effect on APD at normal heart rates so long as there is significant I_Kr (repolarization reserve). I_Ks is present in hESC‐CMs (Otsuji et al., 2010; Wang et al., 2011), but few studies have delineated its functional role in the repolarization of hPSC‐CMs, with or without β‐adrenoceptor stimulation. APD₅₀ and APD₉₀ of SA002 hESC‐CMs were prolonged by 25% by 100‐μM chromanol 293B (Jonsson et al., 2012), in line with a previous report on H7 hESC‐CMs (Peng et al., 2010). We similarly observed 25% prolongation of APD₃₀ and APD₈₀ by 60‐μM chromanol 293B (Figure 2c(B)). However, in another study, I_Ks block had only minor effects on field potential durations of hiPSC‐CMs, unless I_Kr was also blocked (Braam et al., 2013). These observations suggest that the balance of repolarizing currents differs among cell lines, with I_Ks having a primary role in the H9 line. Furthermore, the relative increases in APD₃₀ and APD₈₀ at high levels of I_Ks block were greater at 75 days, although expression of K_v7.1 protein increased just slightly (Figure 4b(C)). This expression level change is in line with a previous study where KCNQ1 expression was not significantly different between 1 and 2 months of culture (Otsuji et al., 2010).

I_Kur is an important modulator of the plateau amplitude and APD in adult CMs (Ravens & Wettwer, 2011) and is present in both human atria and ventricles (Li, Feng, Yue, Carrier, & Nattel, 1996; Wettwer et al., 2004). The functional role of I_kur has not been well defined in hPSC‐CMs, where a heterogeneous mix of ventricular‐like, atrial‐like, and nodal‐like cells is expected (Blazeski et al., 2012b). One study showed that 4‐AP prolonged APD in hESC‐derived atrial‐like but not ventricular‐like cardiomyocytes (Devalla et al., 2015). Interestingly, almost all regions of the monolayers in our study showed significant prolongation of APD with DPO‐1, despite having mostly a ventricular‐like phenotype. K_v1.5 was expressed in our early‐stage monolayers and increased slightly in late‐stage monolayers.

In conclusion, we have presented a systematic analysis of repolarization characteristics in hESC‐derived cardiomyocytes in terms of four major repolarization potassium channels. This study demonstrates that I_Kr, I_Ks, I_K1, and I_Kur all participate in cardiac repolarization in hESC‐CMs to varying degrees. Pharmacological and molecular evidence indicate that these currents evolve at different rates during the maintenance of hESC‐CMs as monolayers. Furthermore, drug responses of these cells, which are relatively homogeneous within individual monolayers, functionally manifest positive, negative, or neutral APD dependence that in turn can vary with time in culture. Positive APD dependence of ion channel block is potentially a new pro‐arrhythmic index, because it can augment APD and repolarization gradients in tissue.

4.1. Limitations

We studied cardiomyocytes derived from hESCs because hESCs are still considered the gold standard in terms of pluripotency and differentiated cardiomyocytes. The H9 line is a widely studied hESC line especially favourable for differentiation into cardiomyocytes (Sepac et al., 2012). Because of the complexity and time span involved in our experiments, we were not able to include other cell lines. We investigated the electrophysiological changes in these cells over time in culture, but further studies aimed at their cellular and molecular signatures are needed to determine whether the observed changes are reflective of a maturation process and/or changes in cell composition. Also, Western blots were conducted on total protein as opposed to the isolated membrane fraction, which was insufficient in quantity for analysis. Another limitation is that compared to single cell patch clamp studies, optical APs do not measure the absolute transmembrane potential, or provide direct measurements of individual ionic currents, although they have the advantage of hundreds of measurements from very large populations of intact cells in which the cells are unaffected by patch pipette solution and their structure remains unperturbed.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Y.W. and L.T. designed the study. Y.W. performed hESC culture and differentiation, electrophysiological experiments, and molecular biology assays. Y.W. and R.Z. analysed the data. Y.W., R.Z., and L.T. wrote the manuscript. All authors reviewed and approved the manuscript.

DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design & Analysis, and Immunoblotting and Immunochemistry, and as recommended by funding agencies, publishers, and other organizations engaged with supporting research.

Wang Y, Zhu R, Tung L. Contribution of potassium channels to action potential repolarization of human embryonic stem cell‐derived cardiomyocytes. Br J Pharmacol. 2019;176:2780–2794. 10.1111/bph.14704