cAMP-PKA signaling modulates the automaticity of human iPSC-derived cardiomyocytes

In recent years human-induced pluripotent stem cell–derived cardiomyocytes (hiPSC-CMs) have been used to model inherited as well as acquired cardiac diseases (Itzhaki et al, 2011, 2012; Chauveau et al, 2017; Ben Jehuda et al, 2018; Eisen et al, 2019). Further, hiPSC-CMs are employed to screen and characterize drugs and uncover mechanisms underlying cardiac diseases. However, before hiPSC-CMs can be employed as reliable experimental models, the physiological mechanisms controlling their normal function must be thoroughly explored. Spontaneously beating hiPSC-CMs have attracted enormous interest (Pozo et al., 2022; Morad and Zhang, 2017) because of their ability to generate rhythmic action potentials resembling the sinoatrial node (SAN) cells, which control heart rate. The SAN cell function is orchestrated by an ensemble of surface membrane channels, exchangers, and pumps (the so-called M clock) that are coupled to the sarcoplasmic reticulum (SR; the so-called Ca²⁺ clock). Spontaneous local Ca²⁺ release (LCR) from the SR, which activates the M clock to generate action potentials (Lakatta et al, 2010; Neco et al, 2012; Groenke et al, 2013; Yaniv et al, 2013; Tsutsui et al, 2018), and the cAMP/protein kinase A (PKA) activity which regulates mechanisms that drive both clocks (Yaniv et al., 2015; Behar et al., 2016), are the two main signaling cascades that couple these clocks. Thus, in mammalian pacemaker cells, the degree of coupling between the clocks depends on LCR characteristics and the level of cAMP/PKA signaling. Whereas it was shown that the two clocks coexist in human embryonic stem cell–derived cardiomyocytes (Zahanich et al., 2011), human SAN (Tsutsui et al., 2018), and spontaneously beating hiPSC-CMs (Mandel et al., 2012), it is unknown which signal controls their coupling. To investigate the mechanisms that couple these clocks, we tested three hypotheses in hiPSC-CMs: (1) normal automaticity of spontaneously beating hiPSC-CMs is regulated by LCRs and cAMP/PKA-dependent coupling of the Ca²⁺ and M clocks; (2) the LCR period indicates the level of crosstalk (i.e., the beat interval) within the coupled-clock system; and (3) perturbing the activity of even one clock, can lead to hiPSC-CM dysfunction due to diminished crosstalk within the coupled-clock system. Our main findings were (1) upregulation/downregulation of cAMP-dependent coupling between the clocks influence LCR properties, in turn affecting hiPSC-CMs automaticity; (2) clocks’ uncoupling by attenuating the pacemaker current I_f or SR Ca²⁺ kinetics decreased hiPSC-CMs beating rate and prolonged LCR period; and (3) LCR characteristics of spontaneously beating (at comparable rates) hiPSC-CMs and rabbit SAN are similar.

A skin biopsy was obtained from a 42-yr-old healthy female (clone 24.5) and a foreskin was obtained from a healthy neonatal (clone FSE-5m), as previously characterized and described (Yehezkel et al., 2011; Novak et al., 2015). All donors signed consent forms according to approval #3116 by the Helsinki Committee for experiments on human subjects at Rambam Health Care Campus, Haifa, Israel. hiPSCs were differentiated into cardiomyocytes (hiPSC-CMs) according to the directed differentiation by modulating Wnt/β-catenin signaling, as previously described (Lian et al., 2013).

Ca²⁺ images were recorded from 30–60-d-old spontaneously beating hiPSC-CMs. To this end, hiPSC-CM monolayers were dissociated by adding 2 ml/well trypsin-EDTA at room temperature (Biological Industries) followed by incubation at 37°C in 5% CO₂ (Forma series 3 water-jacketed CO₂ incubator) for 5 min. The contents of each well were collected in a conical tube to which 3 ml RPMI⁺ medium (RPMI 1640 [Gibco], B-27 [Gibco], and penicillin-streptomycin [Biological Industries]) was added, and was centrifuged (Spectrafuge 6C; Labnet) for 3 min at 800 rpm. The supernatant was aspirated, and the hiPSC-CMs were resuspended in 3 ml RPMI⁺ medium and reseeded onto the center of a collagen-coated glass-bottom culture dish (MatTek Corporation). The dishes were incubated (37°C, 5% CO₂) for a recovery period of 5 d before performing the Ca²⁺ imaging experiments.

Ca²⁺ cycling into and out of the cytosol was recorded by monitoring the fluorescence of the Ca²⁺ indicator dye Fluo-4 AM (Thermo Fisher Scientific) using a LSM880 confocal microscope, as previously described (Davoodi et al., 2017). Dishes with spontaneously beating hiPSC-CMs were loaded with 2.5 μM Flou-4 AM for 20 min in the dark at room temperature and then washed with Tyrode’s solution (at 37°C) containing (in mM) 140 NaCl, 5.4 KCl, 10 HEPES, 2 Na-pyruvate, 10 glucose, 1 MgCl₂, and 2 CaCl₂ (pH 7.4). Fluorescence was observed by exciting the sample with a 488 nm argon laser and measuring fluorescence emission with LP 505 nm. Ca²⁺ images were acquired using a 40×/1.2 water immersion lens with line-scan mode (1.22 ms per scan; pixel size, 0.01 µm) oriented along the cell, close to the sarcolemmal membrane.

To identify global and LCRs from the SR and quantify their characteristics, line-scan images were analyzed using a modified version of a semiautomatic Matlab graphic user interface named Sparkalyzer (Davoodi et al., 2017). Briefly, the signal (F) was normalized by the minimal value between beats (F₀). Ca²⁺ transients were semiautomatically detected, and LCRs were manually marked. LCRs were marked as a small increase in Ca²⁺ signal during the diastolic depolarization that lasted for at least 2 ms, 2 µm length, and at least 1.05 F/F₀ height due to the microscope resolution constraints. Only a certain amount of Ca²⁺ signal can initiate the electrical activity, thus more than one LCR appears at the same beat interval. Beat interval (referred to as the cycle length in Davoodi et al. [2017]) was calculated as the mean beat interval of each cell, and 50 and 90% relaxation time and LCRs parameters were automatically calculated by the software as previously described (Davoodi et al., 2017). The LCR period is referred to as the time between the Ca²⁺ spark and the former peak (the time elapsed between former AP-induced Ca²⁺ transient and LCR appearance; Fig. 1; Davoodi et al., 2017). The number of LCRs is the number of detected LCRs in the line-scan Ca²⁺ image by the user, scaled by the time interval and length of the scan line. The Ca²⁺ signal of individual LCRs indicating the amount of Ca²⁺ released by each LCR was calculated using LCR parameters for each LCR as follows: 50% LCR duration × LCR length × normalized amplitude (ms·µm·F/F₀).

Isoproterenol (ISO), carbachol (CCh), forskolin, N-[2-(p-Bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide (H-89), cyclopiazonic acid (CPA), and hydrochloroquine (HCQ) were purchased from Sigma-Aldrich. IVA was purchased from Toronto Research Chemicals.

Data are presented as mean ± SEM. Data were compared using regressions, paired or unpaired Student’s t test, and two-sample Kolmogorov–Smirnov test. Differences were considered statistically significant at P < 0.05 or P < 0.01, as indicated in the respective figures and tables.

Fig. S1 shows the classification of control spontaneously beating hiPSC-CMs into two groups based on the average beat interval. Fig. S2 shows average dose-dependent changes in the beat interval in response to forskolin. Fig. S3 shows the percent changes in the beat interval in response to either doses of IVA or CPA and their combination. Fig. S4 shows HCQ-induced changes in global and local Ca²⁺ properties. Fig. S5 shows a 3-D relationship between hiPSC-CMs, rabbit sinoatrial node cells (SANCs), and rabbit ventricular cells based on LCR properties. Table S1 summarizes the global and local Ca²⁺ properties of control spontaneously beating hiPSC-CMs. Table S2 summarizes the global and local Ca²⁺ properties of the two clones used for the experiments. Tables S3, S4, S5, S6, and S7 summarize the effect of ISO and CCh (Table S3), 1 µM ISO (Table S4), forskolin and H-89 (Table S5), IVA and CPA (Table S6), and HCQ (Table S7) on global and local Ca²⁺ properties in spontaneously beating hiPSC-CMs. Table S8 compares the global and local Ca²⁺ properties of hiPSC-CMs, rabbit SANCs, and rabbit ventricular cells.

To test the relationship between LCR characteristics and the basal beat interval, Ca²⁺ transients and LCRs were measured in spontaneously beating hiPSC-CMs. Fig. 1, A–C; and Table S1, respectively, show representative Ca²⁺ transients and LCRs detected over a wide range of beat intervals and their characteristics. The mean beat interval of hiPSC-CMs derived from Ca²⁺ line-scan images was 1,354.5 ± 55.7 ms (Fig. 1 D; n = 114), consistent with previously reported hiPSC-CMs beat intervals (Mandel et al., 2012). The high correlation between the LCR period and beat interval (Fig. 1 E; R² = 0.86) suggests that one is a readout of the other. The LCR period also correlated with 90% time of relaxation (T₉₀; Fig. 1 F), which indicates the 90% replenishment of SR Ca²⁺ following Ca²⁺-induced Ca²⁺ release (Vinogradova et al., 2010). The Gaussian distribution of the beating interval (a Gaussian fitting yielded a mean of 1,024.5 ms and an SD of 373.5 ms; Fig. 1 G) was similar to the LCR period distribution (Kolmogorov–Smirnov test, P < 0.05; Fig. 1 H). In contrast, the T₉₀ distribution (Fig. 1 I) differed from the LCR period and beat interval distributions. Note that no differences were found regarding the beat interval or Ca²⁺ characteristics between the two clones (Table S2).

To determine whether LCR characteristics are dependent on the beat interval, the data were divided into two groups: hiPSC-CMs with a mean beat interval >2,000 and ≤2,000 ms. The resulting populations (Fig. S1 A) differed in their average T₉₀ (Fig. S1 B), 50% LCR duration (Fig. S1 C), and LCR period (Fig. S1 D). The correlation between LCR period and beat interval (Fig. S1 E) and the LCR period and T₉₀ (Fig. S1 F) was weaker (0.46 and 0.18, respectively) among hiPSC-CMs with beat intervals >2,000 ms. Thus, the LCR period is correlated with the basal beat interval.

To determine the effect of altering cAMP/PKA signaling on beat interval and LCR characteristics, we used the non-selective β-adrenergic receptor agonist ISO and the cholinergic receptor agonist CCh. CCh was used at the concentration reported to cause a maximum increase in the beat interval (Nozaki et al., 2017), but that did not stop the spontaneous activity and the appearance of LCRs. ISO concentration was used at a concentration causing the same absolute change in the beat interval as CCh, leading to a decrease in the beat interval (Ben-Ari et al., 2014; Takeda et al., 2021). Of note, the basal beat intervals before drug application are from the same distribution and are not significantly different from each other. Fig. 2, A–D, shows representative examples of the effects of ISO and CCh on the Ca²⁺ transient. Specifically, 100 nM ISO decreased the beat interval by 22.6 ± 8.9% (n = 7, Fig. 2 E), while 100 nM CCh increased the beat interval by 28.8 ± 14.9% (n = 6, Fig. 2 E). Ca²⁺ transient characteristics time to peak, 50% relaxation (T₅₀), and T₉₀ were unaffected by ISO, but prolonged by CCh (Table S3). LCR characteristics were affected by ISO and CCh; the absolute changes in the LCR period (Fig. 3 A) and Ca²⁺ signals of individual LCRs (Fig. 3 B) were similar by either drug perturbation. ISO also increased the normalized LCR signal amplitude but decreased the number of LCRs (Table S3). In contrast, CCh decreased the normalized amplitude and length of LCRs. Finally, ISO and CCh did not alter the tight correlations between the LCR period and beat interval (Fig. 3 C; R² = 0.86), and between the LCR period and T₉₀ (Fig. 3 D; R² = 0.59). Of note, a 1 µM ISO affected the beat interval and LCR characteristics (Table S4).

Next, adenylate cyclase was activated by forskolin, and PKA was inhibited by H-89. We used the lowest forskolin concentration that led to maximal shortening of the beat interval (Fig. S2) and the H-89 concentration that caused a similar maximum absolute change in the beat interval. We used the lower forskolin concentration that led to maximal shortening of the beat interval (Fig. S2) and the appropriate H-89 concentration that caused a similar maximum absolute change in the beat interval. Of note, the basal beat intervals before drug application are from the same distribution and are not significantly different from each other. Fig. 4, A–D shows representative examples illustrating the effects of forskolin and H-89 on the Ca²⁺ transients; H-89 induced bradycardia and arrhythmias in three of the six cells studied. The SD increased for H-89 from 153 ± 56 to 229 ± 56 s. Forskolin decreased the beat interval by 21.9 ± 10.2% (n = 6, Fig. 4 E), while H-89 prolonged the beat interval by 22.3 ± 12.8% (n = 6, Fig. 4 E). Additionally, Ca²⁺ transient characteristics (time to peak, T₅₀, and T₉₀) and LCR characteristics were unaffected by forskolin and H-89 (Table S5); the LCR period (Fig. 5 A) was decreased by forskolin. The change in the Ca²⁺ signal of individual LCRs (Fig. 5 B) was increased twofold by forskolin compared with H-89. Forskolin also increased the normalized LCR signal amplitude, while H-89 decreased the normalized amplitude and the number of LCRs and increased LCR duration by 50% (Table S5). Finally, forskolin and H-89 did not affect the tight correlations between the LCR period and beat interval (Fig. 5 C; R² = 0.81), while the relationship between the LCR period and T₉₀ was reduced (Fig. 5 D; R² = 0.49).

It was demonstrated that the I_f blocker ivabradine (IVA; 3 µM) blocks only the HCN4 channel (Yaniv et al., 2012). At this concentration, IVA did not affect the T- or L-type or delayed outward potassium current densities (Bois et al., 1996) and SR Ca²⁺ content. Thus, IVA specifically perturbs the M clock with no direct effect on the Ca²⁺ clock. On the other hand, the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) inhibitor CPA has no direct effect on membrane currents (funny and L-type current densities), but only on SR Ca²⁺ content (Yaniv et al., 2013). The IVA concentration used was previously shown to induce a maximum decrease in beat interval without affecting other channels (T- or L-type or delayed outward potassium current densities) than HCN4 (Chauveau et al., 2017; Yaniv et al., 2013) or the Ca²⁺ clock (Yaniv et al., 2013). CPA was applied at a concentration causing a similar change in beat interval as IVA but was shown not to affect key channels such as L-type and funny current densities. Fig. 6, A–D, depicts representative examples of the effects of IVA and CPA on Ca²⁺ transients. 3 µM IVA or 3 µM CPA increased the beat interval by 45.6 ± 7.7% (n = 9, Fig. 6 E) and 48.2 ± 11.7% (n = 7, Fig. 6 E), respectively. Ca²⁺ time to peak, T₅₀, and T₉₀ were prolonged by both IVA and CPA (Table S6). LCR characteristics were also affected by IVA and CPA; the LCR period (Fig. 7 A) and the Ca²⁺ signal of individual LCRs (Fig. 7 B) were similarly decreased by IVA and CPA. IVA also decreased the LCR length (Table S6), while CPA decreased the normalized LCR amplitude, as well as the length and number of LCRs. Finally, IVA and CPA did not affect the tight correlations between the LCR period and beat interval (Fig. 7 C; R² = 0.79). However, the relationship between the LCR period and T₉₀ was not maintained (Fig. 7 D; R² = 0.54). Note that, when both IVA and CPA were used together, no further increase in the individual effect on the beat interval was measured (Fig. S3; n = 11).

To demonstrate that LCR characteristics are affected by changes in the beating rate, hiPSC-CMs were treated with HCQ, which causes bradycardia (Capel et al., 2015). We recently showed that HCQ-induced bradycardia was caused by a direct effect on the M clock and an indirect effect on the Ca²⁺ clock (Segal et al., 2022). Fig. S4, A and B, depicts a representative example of the HCQ effect on Ca²⁺ transients. Of the three tested HCQ concentrations (1, 3, and 10 µM), only 10 µM increased the beat interval (71.8 ± 13.2%, n = 7; Fig. S4 C). Ca²⁺ time to peak, T₅₀, and T₉₀ were prolonged by HCQ (Table S7). Further, HCQ increased the LCR period (Fig. S4 D) but decreased the Ca²⁺ signals of individual LCRs (Fig. S4 E) and their normalized amplitude (Table S7). Finally, the tight correlations between the LCR period and beat interval (Fig. S4 F; R² = 0.54) and between the LCR period and T₉₀ (Fig. S4 G; R² = 0.01) were not preserved in the presence of HCQ.

To explore the similarity between hiPSC-CMs and SAN cell features, we compared LCR characteristics of hiPSC-CMs that beat spontaneously at a beat interval of 1,022.5 ± 10.1 ms with rabbit SAN cells that beat spontaneously at 1,010.3 ± 50 ms and rabbit ventricular cells paced at 1 Hz. Fig. S5 shows the 3-D representation of the relationship between the LCR period, number of LCRs, and 50% LCR duration. The hiPSC-CMs can be separated from the ventricular cells based on the combination of LCR characteristics. We did not compare the properties to atrial cells because we used 30–60-d-old, spontaneously beating hiPSC-CMs, previously shown to present a low occurrence of atrial-like cells (Ben-Ari et al., 2016). Thus, LCR characteristics of spontaneously beating (at comparable rates) hiPSC-CMs and rabbit SAN are similar, as was shown before (Giannetti et al., 2021).

In the present study, we investigated the mechanisms that couple the hiPSC-CM M- and Ca²⁺ clocks. Drug-induced alterations in Ca²⁺ or cAMP/PKA signaling affected the spontaneous beat interval and LCRs, supporting the first hypothesis that normal automaticity of spontaneously beating hiPSC-CMs is regulated by Ca²⁺-cAMP/PKA–dependent coupling of the M- and Ca²⁺ clocks. Drugs that affect the coupling by disturbing clock mechanisms or their interconnected signaling affected the LCR period and beat interval. The tight correlation between the LCR period and beat interval in response to these drugs supports the hypothesis that the LCR period indicates the level of crosstalk within the coupled-clock system. Finally, disrupting only the M- or Ca²⁺ clocks by IVA or CPA, respectively, increased the spontaneous beat interval and altered LCR properties, supporting the second hypothesis that perturbing even one clock activity leads to hiPSC-CM dysfunction by decreasing crosstalk within the coupled-clock system.

The LCRs measured in hiPSC-CMs were similar to those reported for other mammals (Lakatta et al., 2010), including humans (Tsutsui et al., 2018). Table S8 presents Ca²⁺ transient and LCR characteristics of hiPSC-CMs, rabbit SANCs, which beat spontaneously at a similar rate (Davoodi et al., 2017), and paced (1 Hz) ventricular rabbit cells. The T₉₀, the Ca²⁺ signal of individual LCRs, the LCR period, and the number of LCRs were comparable in hiPSC-CMs and rabbit SANCs. Similarities also existed between the LCR periods of human SANCs and hiPSC-CMs (Tsutsui et al., 2018). However, LCR properties were distinguished from paced ventricular rabbit cells. These results strengthen the hypothesis that spontaneously beating hiPSC-CMs have similar properties as human SANCs.

An increase or decrease in cAMP/PKA-dependent signaling affected the spontaneous activity of hiPSC-CMs. Two methods were used to manipulate the cAMP-dependent clock coupling: (1) autonomic stimuli (ISO or CCh) that affect automaticity by altering adenyl cyclase (AC)-cAMP/PKA–dependent signaling (Figs. 2 and 3) and (2) drugs (forskolin or H-89) that alter cAMP activity (Figs. 4 and 5). In general, ISO and forskolin increased the coupling between the M- and Ca²⁺ clocks through the increased cAMP-dependent activity of I_f and increased PKA-dependent activity of M clock–dependent ionic channels (e.g., the L-type Ca²⁺ channel and slow heart potassium current) and Ca²⁺ clock–related proteins (ryanodine receptor and phospholamban). Increased clock coupling enhances the interplay between the L-type Ca²⁺ channel and SR Ca²⁺ cycling, which lead to changes in LCR parameters (Lyashkov et al., 2018). The net increase in Ca²⁺ affects the membrane potential through changes in Na⁺–Ca²⁺ exchanger (NCX) current (I_NCX; Bogdanov et al., 2001) and Ca²⁺-activated SK4 and BK potassium currents (Weisbrod et al., 2013; Lai et al., 2014), which decrease the spontaneous beating interval. CCh and H-89 decreased the coupling between the clocks through a decrease in the interplay between L-type Ca²⁺ channels and SR Ca²⁺ cycling, thereby prolonging the spontaneous beat interval. Note, that although LCRs were measured in human SAN (Tsutsui et al., 2018), the mechanisms that control their activity and the LCR period that reports the level of crosstalk within the coupled-clock system were not shown.

This study demonstrated that the magnitude of drug-induced bradycardia is not solely driven by the activity of one channel or one specific clock mechanism, but rather, by the potent crosstalk within the coupled-clock system. The approach used to show this was similar to that applied for SANCs (Yaniv et al., 2013) by CPA, which inhibits SERCA2 Ca²⁺ pumping but not the M clock directly, or IVA, which inhibits I_f but does not affect the Ca²⁺ clock. Both compounds increased beat interval to the same extent and affected LCR characteristics similarly (Figs. 6 and 7). Ca²⁺ pumping into the SR via SERCA was suppressed by CPA, thereby reducing SR Ca²⁺ load and subsequently prolonging the LCR period. In turn, the spontaneous LCR Ca²⁺ signal to activate NCX and Ca²⁺-activated K channels (SK4 and BK) is reduced and arrived later during the diastolic phase. The resultant decreased feedback between the M- and Ca²⁺ clocks leads to an increase in the beat interval. In turn, the net Ca²⁺ influx is reduced, which decreases the interplay between the L-type Ca²⁺ channel and SR Ca²⁺ cycling (Lyashkov et al., 2018), leading to further reductions in SR Ca²⁺ cycling kinetics and prolongation of the LCR period. Similarly, I_f was inhibited by IVA (Ben-Ari et al., 2014), leading to an increased beat interval, which reduced the net Ca²⁺ influx and thus the interplay between the L-type Ca²⁺ channel and SR Ca²⁺. Reduction in the interaction between altered SR Ca²⁺ kinetics reduced and delayed the spontaneous LCR ensemble Ca²⁺ signal to the M clock to activate NCX and Ca²⁺-activated K channels, subsequently resulting in a further increase in the beat interval. In general, any HCN intervention that increased beat interval will affect the Ca²⁺ clock. Of note, we employed spontaneous beating hiPSC-CMs that have a higher density of I_f than young CMs (<30-d-old; Ben-Ari et al., 2016). Thus, we found a substantial effect of IVA on the spontaneous beat interval compared with other groups who used younger hiPSC-CMs (Zhang et al., 2015; Kim et al., 2015).

The current work also demonstrated that the LCR period indicates the level of crosstalk within the coupled-clock system, regardless of the degree of clock coupling perturbation. Longer beat intervals due to spontaneous activity (Fig. S1) and reduced AC-cAMP/PKA activity were associated with a longer LCR period. In contrast, shorter beat intervals resulting from increased AC-cAMP/PKA activity were associated with shorter LCR periods. The relative changes in the LCR period were tightly correlated (R² = 0.97) to the relative change in the spontaneous beat interval (Fig. 8 A), as was found in rabbit (Vinogradova et al., 2010) and human (Tsutsui et al., 2018) SANCs. It was shown that T₉₀ is an indicator of the rate of SR Ca²⁺ pumping into the SR (Vinogradova et al., 2010). The fact that the relative changes in the LCR period are tightly correlated (R² = 0.86) to the relative change in T₉₀ (Fig. 8 B) further strengthens the conclusion that the LCR period indicates the degree of interplay between the L-type Ca²⁺ channel and SR Ca²⁺ cycling (Fig. 8 C). Note, the drug that induced bradycardia (forskolin, IVA, CPA, and H-89) decreased the degree of correlation between LCR period and T₉₀.

In addition to IVA, the effect of HCQ (a drug that causes bradycardia; Capel et al., 2015) was determined. HCQ was shown to reduce I_f, the L-type Ca²⁺ current, and the rapidly activating potassium current in isolated guinea pig SAN cells (Capel et al., 2015). We found here (Fig. S4) that HCQ-induced bradycardia was associated with a prolonged LCR period and reduced Ca²⁺ signal of individual LCRs. These results suggest that HCQ affects the spontaneous beating of hiPSC-CMs through changes in the signaling cascades that couple the Ca²⁺ and M clocks. Note that HCQ reduced the degree of clock coupling to a level where the LCR period is not an indicator of the spontaneous beat interval. Similar phenomena were observed when the physiological spontaneous beat interval was long.

The drug perturbation experiments were conducted on hiPSC-CMs that beat at a rate similar to that of the human heart. We also showed here that the LCR characteristics were comparable between human SAN, which beats at 1–1.5 Hz, and spontaneously beating hiPSC-CMs. It was previously shown that the I_f density (Giannetti et al., 2021) and action potential characteristics (Ben-Ari et al., 2016) were comparable between human SAN cells and spontaneously beating hiPSC-CMs. However, the exact cardiac origin (i.e., ventricular, sinoatrial, or atrial) of the hiPSC-CMs used in this study was not determined. Future use of SAN cells derived from hiPSCs (Protze et al., 2017) will enable further clarification of the similarities between these cell sources.

David A. Eisner served as editor.

The work was supported by Israel Science Foundation grants ISF 330/19 (Y. Yaniv) and ISF 824/19 (O. Binah), The Rappaport Family Institute for Research in the Medical Science grant 01012020RI (O. Binah), Niedersachsisches Ministerium: Medizinischen Hochschule Hannover grant 11-76251-99-16/14 (O. Binah), and The US-Israel Binational Science Foundation (O. Binah). The funders had no role in study design, data collection or analysis, decision to publish, or preparation of the manuscript.

The authors declare no competing financial interests.

Author contribution: Y. Yaniv conceived and designed the research. S. Mazgaoker performed the experiments and analysis. S. Mazgaoker, I. Weiser-Bitoun, and I. Brosh generated the cells. O. Binah contributed reagents. Y. Yaniv drafted the manuscript. S. Mazgaoker, O. Binah, and Y. Yaniv edited and revised the manuscript. S. Mazgaoker, I. Weiser-Bitoun, I. Brosh, O. Binah, and Y. Yaniv approved the final version.