Electrophysiological and Pharmacological Characterization of Human Inwardly Rectifying K ir 2.1 Channels on an Automated Patch-Clamp Platform

K_ir2.1 Cell Line

A Chinese hamster ovary (CHO) cell line stably expressing the human KCNJ2 gene product (GenBank NM_000891.2) under the control of a tetracycline-inducible promoter was obtained from a commercial source and cultivated according to the vendor instructions (Charles River; Cat. No. CT6127). The cells were grown in a humidified 95% air/5% CO₂ atmosphere in Dulbecco's modified Eagle's medium/F-12 Nutrient Mixture (ThermoFisher Scientific) supplemented with 10% fetal bovine serum and the appropriate selection antibiotics (0.01 mg/mL blasticidin and 0.4 mg/mL zeocin). For stock cultures, cells were grown in T175 flasks and passaged every 3–4 days below 80% confluence. Cell detachment was obtained by exposure to TrypLE Select^™ (Gibco, ThermoFisher Scientific) for 2–3 min at 37°C. Flasks prepared in view of electrophysiological recordings were typically seeded 2 days before the experiments with an inoculum containing about 2.5 million cells per T175 flask filled with 25–30 mL growth medium. Expression of the K_ir2.1 channels was obtained by overnight induction with 1 μg/mL doxycycline added to the growth medium. On the day of recording, detached cells were spun down, washed, resuspended to 5–8 million cells/mL in a glucose-containing extracellular buffer, and placed in the cabinet of a QPatch^® 48X workstation (Sophion Bioscience, Denmark). This standalone instrument comprises a robotic pipetting arm ensuring distribution of cell suspensions into disposable 48-well recording plates (QPlates^®) and the sequential application of drug solutions at final test concentration, while whole-cell patch-clamp is maintained without interruption. ³⁰

Automated Patch-Clamp

All recordings were performed at room temperature. Experiments aimed at characterizing the electrophysiological properties of the currents were conducted on biochips endowed with a unique pinhole orifice engineered at the bottom of each of the 48 wells of disposable measurement plates (i.e., single-hole QPlates). Currents were activated by application of a series of 500 ms-long square voltage pulses delivered every 90 s and incremented in 5 mV steps from a holding potential of −20 mV. Currents measured over 30 ms at the end of each step served to plot current-voltage (I–V) relationships. In some experiments, the external potassium [K⁺]_out concentration was changed, while maintaining internal [K⁺]_in constant to change the K⁺ equilibrium potential. In other single-hole recordings, the I–V relationships were established in an external buffer supplemented with various concentrations of alkaline or alkaline-earth metal ions.

Experiments aimed at characterizing the pharmacology of small molecules were conducted with 48-well measurement plates, each fitted with biochips perforated with 10 holes, so that each recording site sums up the signal from multiple cells (i.e., multi-hole QPlates). The voltage protocol used was a step-ramp protocol applied every 30 s, which first stepped the membrane from a holding potential of −20 mV down to −120 mV for 500 ms, then ramped it up to 0 mV over 1 s, and eventually stepped the membrane potential back to −20 mV, where the cells were held until the next voltage cycle. This protocol allowed for long-time (30 min) recordings comprising a stable baseline and ample space for the application of multiple drug concentrations in a row. Moreover, the voltage ramp permitted to check the recording quality over time by monitoring the stability of the zero-current reversal potential value. The size of the inward current measured at the end of the initial hyperpolarizing 500 ms-step to −120 mV was used to evaluate drug effect.

For all experiments, the intracellular recording buffer contained (in mM) 5.37 CaCl₂, 1.75 MgCl₂, 10 ethylene glycol tetraacetic acid, 10 HEPES, 120 KCl, and 4 Na₂-ATP (pH 7.2, 300 mOsm) and the standard extracellular recording buffer contained (in mM) 2 CaCl₂, 1 MgCl₂, 10 HEPES, 4 KCl, 145 NaCl, and 10 Glucose (pH 7.4, 310 mOsm).

Drugs

All test articles were from Sanofi's internal libraries, except ML133 (CAS Number 1222781-70-5; Wang et al. ³¹ ), which was obtained from a commercial source (Sigma-Aldrich, Saint-Quentin-Fallavier), and PA-6 (CAS Number 1199627-07-0, Takanari et al. ³² ), a pentamidine derivative, which was synthesized following published procedures and references therein. ^32,33 Figure 5 depicts the structural formula and IUPAC chemical names of ML133, PA-6, and the five other small-molecule inhibitors examined in this study. Compounds were dissolved as 10, 50, or 100 mM concentrated stocks in dimethyl sulfoxide (DMSO) using sonication as needed for complete dissolution, then aliquoted into capped polypropylene vials, and kept frozen at −20°C until use. On the day of study, stock solutions were diluted to the final concentrations intended for cell applications in standard 96-well microtiter plates (MTP) containing the extracellular buffer supplemented with 0.06% Pluronic F-68^™ (Gibco, ThermoFisher Scientific). The latter surfactant was added to help slow unseen precipitation (if any) of poorly soluble drugs in the MTP wells upon dilution. ³⁴ The final drug solutions also contained 0.3% residual DMSO from the dilution processes. The test article solutions were typically applied to the cells within 30 min of preparation. Concentration–response data were either obtained from the cumulative applications of six increasing concentrations to the same cells within a given recording well, each concentration being exposed for 3 min, or by applying a single different concentration into given wells with an exposure duration extended to 15 min.

Data Handling

Current amplitude measurements and normalization for the fitting of concentration–response curves were done with Sophion's dedicated analysis software (Odense 5.6.4). Time course of current amplitude was visualized offline and analyzed following optional rundown compensation as needed using built-in routines proprietary to this software (uncorrected rundown measured in N = 5 vehicle-treated cells did not exceed 1% per min in single application experiments). All recordings for pharmacological determinations were concluded by the application of a supramaximal concentration (3 mM) of barium as full block. Residual leak currents remaining under full block (if any) were subtracted from precedent data points. A professional graphing package (Prism 7.02; GraphPad Software) was used for drawing exported current traces, plotting I–V curves, and performing statistical calculations. Half-maximal inhibitory concentrations (IC₅₀) were obtained by fitting the average of the two or the five last current amplitude readings at the end of each drug concentration exposure period in a cumulative or noncumulative application mode, respectively. A classical four-parameter Hill equation following normalization with respect to predrug baseline readings was used for data fitting to a sigmoidal curve with minimum and maximum constrained to 0% and 100%, respectively. Estimated IC₅₀ and Hill coefficient (n_H) values are reported along with their estimated 95% confidence interval (CI). Mean of individual replicates is represented with their calculated standard deviations shown as error bars on graphs where appropriate.

Electrophysiological Characterization

The upper part of Figure 1A illustrates the typical appearance of whole-cell currents recorded from stably transfected CHO cells induced to express the human KCNJ2 gene product. Inward or outward currents reaching steady state within a 100 ms developed when the membrane was stepped tens of mV below or above the K⁺-equilibrium potential (E _K), respectively (Fig. 1A). At more depolarized potentials, the current–voltage (I–V) relationship displayed a strongly inward rectifying profile. This profile fully concurs with the I–V outline of native ventricular I _K1 currents. ³⁵ Specifically, the amplitudes of the inward currents below E _K grew linearly in proportion to the negativity of the membrane, reaching relatively large values at strong hyperpolarization (e.g., −2.3 ± 0.4 nA at −120 mV, N = 10, Fig. 1B), whereas the positive currents above E _K culminated at potentials around −60 to −50 mV (e.g., 0.31 ± 0.09 nA at −60 mV, N = 10, Fig. 1B). Virtually no currents were detected in induced cells at potentials more positive than −20 mV (Fig. 1B) nor at any potential in noninduced CHO cells (Fig. 1A bottom), even when the latter were stepped to very negative potentials (e.g., 0.07 ± 0.05 nA at −120 mV, N = 10), indicating the absence of endogenous I _K1-like currents in the host CHO cell line. The success rate of exploitable recordings obtained from single-hole QPlates routinely exceeded half of the 48 wells. For example, statistics derived from N = 3 single-hole QPlates indicated that, from 126 cells, which successfully entered whole-cell mode (i.e., 88% of the 144 theoretically possible), a total of N = 65 cells formerly exhibiting seal resistances well above 200 MΩ provided complete I–V curves after 30-min recording, demonstrating that 52% of the cells recovered satisfactorily from clamp engagement and remained stable during repeated voltage stepping cycles. Figure 1C illustrates the current size frequency distribution for this sample of 65 cells. The median inward steady-state current amplitude at −120 mV was −2.4 nA, along with an interquartile range spreading from −1.7 to −2.9 nA.

OPEN IN VIEWER

Fig. 1.

Inward and outward currents in CHO cells stably expressing the human KCNJ2 gene recorded with single-hole QPlates. (A) Family of representative current traces obtained by stepping the membrane from a holding of −20 mV to voltages ranging from −120 to 0 mV in a cell exposed overnight to doxycycline (top left) or in a noninduced cell (bottom left). (B) Mean steady-state currents plotted as a function of voltage for N = 10 induced (closed circles) and noninduced (gray squares) cells. Error bars represent standard deviation and are smaller than symbol size in noninduced cells. (C) Descriptive statistics of steady-state current size at −120 mV collected from N = 65 cells sampled from three single-hole QPlates. Frequency distribution was constructed at a binwidth of 0.25 nA. CHO, Chinese hamster ovary.

A prominent property of currents flowing through K_ir2.x channels is their activation by external potassium [K⁺]_out. This counterintuitive behavior (given the decreased electrochemical gradient when [K⁺]_out is increased) results from an increased open-channel conductance in response to elevation of external K⁺. ³⁶ In the experiments illustrated in Figure 2, [K⁺]_out was varied over four concentrations ranging from 2 to 20 mM, while internal potassium was kept constant at [K⁺]_in = 120 mM. The I–V relationships outlined by the four pairs of buffers exhibited a rightward shift proportional to [K⁺]_out, along with a concomitant increase in the size of both the inward and outward currents. The latter culminated at 0.12 ± 0.06, 0.26 ± 0.08, 0.37 ± 0.10, and 0.74 ± 0.20 nA when the external buffer contained 2, 4, 10, and 20 mM [K⁺]_out, respectively (N = 5–10 cells per condition, Fig. 2). Beyond these maxima, the I–V curves presented regions of negative slope conductance until complete extinction of the currents. Their steepness was also proportional to [K⁺]_out, and the superposition of the outward currents in these voltage regions of the I–V curves draw a series of typical “cross-over” patterns as previously described by others. ^35,37 In response to the increasing [K⁺]_out, the zero-current potential of each I–V relationship shifted rightward from −92 ± 1.2 mV at 2 mM [K⁺]_out to −75 ± 2.7 mV, −56 ± 1.5 mV, and −36 ± 1.2 mV for 4, 10, and 20 mM [K⁺]_out, respectively. Linear regression of these data as a function of log [K⁺]_out yielded a highly significant (r ²  = 0.992) fit (Fig. 2, inset) with an estimated 54 ± 3.4 mV shift of the zero-current potential per tenfold variation of external [K⁺]_out, in good agreement with the Nernst equation prediction of 58 mV per decade.

OPEN IN VIEWER

Fig. 2.

Potassium selectivity of the currents in KCNJ2 expressing CHO cells. The graph shows I–V relationships established in external recording buffer containing various [K⁺]_out concentrations. Steady-state currents were recorded in [K⁺]_out = 2 mM (closed circles), [K⁺]_out = 4 mM (open squares), [K⁺]_out = 10 mM (closed triangles), or [K⁺]_out = 20 mM (closed diamonds). Error bars represent standard deviation for data derived from N = 5–10 cells. The inset depicts a linear regression fit of the log₁₀ of [K⁺]_out concentration to the membrane potential values at which no current is observed (i.e., zero-current reversal potential).

Taken together, the electrophysiological features summarized in Figures 1 and 2 demonstrate rectification properties and activation by external potassium typical of K_ir2.1-mediated I _K1 currents. Of note, the high yield of decent currents collected simultaneously allowed to expedite the characterization of a number of electrophysiological features with minimal need of experiments.

Pharmacological Characterization

Cations such as Cs⁺ and Ba²⁺ are blockers of native I _K1 currents in sheep Purkinje fibers, ³⁸ starfish eggs, ^39,40 or frog leg muscles, ⁴¹ abolishing the influence of [K⁺]_out on the I–V profile and blocking I _K1 in a concentration- and voltage-dependent manner. These cations are fast-acting open-channel blockers, reaching steady-state inhibition of the current flow within a few seconds after external exposure. This property was used to document K_ir2.1 channels when they were first isolated by expression cloning in Xenopus oocytes. Figure 3 illustrates the effects of micromolar Ba²⁺ concentrations on K_ir2.1-mediated I _K1 currents activated by a step-ramp voltage protocol in KCNJ2 expressing CHO cells. As previously described by Kubo et al., ⁴² external Ba²⁺ blocked the steady-state current more strongly toward the end of the hyperpolarizing pulses than the initial instantaneous current (Fig. 3A). Moreover, this time-dependent inhibition was also voltage dependent, the channel block increasing when the membrane was stepped to more negative potentials (Fig. 3B). Group data obtained from N = 3 single-hole recordings indicated that the inhibition of the inward current produced by 3 μM Ba²⁺ reached 17% at −120 mV, increasing to 48% at −150 mV. Similarly, the 58% inhibition of the current afforded by 10 μM Ba²⁺ at −120 mV rose to 77% at −150 mV. These properties result in characteristic changes in the I–V relationship profile in the presence of Ba²⁺ (Fig. 3B) and Cs⁺ ions (data not shown). Figure 3C depicts the time course of the effect of cumulative applications of increasing concentrations of Ba²⁺ into a multi-hole recording well. The K_ir2.1-mediated I _K1 currents stabilized within several minutes after whole-cell access. As expected from a fast-acting pore blocker, the inhibitory effect of each cumulatively applied concentration of Ba²⁺ fully developed within the first couple of voltage-protocol cycles immediately at the beginning of each 3-min application period. The IC₅₀ value (calculated from the average normalized inward current amplitudes collected at the two last voltage steps down to −120 mV of each incubation period) was estimated at 8.2 μM (95% CI: 7.6–8.8 μM, N = 6, Fig. 4). The Hill coefficient (n_H) associated to the sigmoidal fit of the normalized data was close to unity (n_H = 1.03; 95% CI: 0.96–1.1), consistent with a mechanism involving a single ion blocking each channel pore according to first-order kinetics. Notwithstanding the voltage dependency of the blocking effect of Ba²⁺, this value is in fairly good agreement with published values on K_ir2.1 currents recorded in HEK cells (IC₅₀ = 7.9 μM at −90 mV, Shen et al. ⁴³ ) or Xenopus oocytes (IC₅₀ = 16 μM at —120 mV, Schram et al. ²³ ). The time course and the inhibitory effects of Cs⁺ and Sr²⁺ ions using the same cumulative concentration–response application protocol were very similarly to Ba²⁺, exhibiting full inhibitory activity immediately at the beginning of each application period. These blockers were less potent than Ba²⁺, yielding IC₅₀ values of 833 μM (95% CI: 789–880 μM, N = 12) with n_H = 1.07 (95% CI: 1.0–1.1), and 4,585 μM (95% CI: 4,308–4,881 μM, N = 11) with n_H = 0.92 (95% CI: 0.87–0.98), respectively (Fig. 4). The potency obtained for Cs⁺ on human K_ir2.1 channels was one order of magnitude weaker than published data for mouse K_ir2.1 overexpressed in a murine erythroleukemia cell line (IC₅₀ = 91 μM at —100 mV, Abrams et al. ⁴⁴ ), raising the possibility that K_ir2.1 expression in a conspecific rather than ectopic cell environment affects voltage dependency of Cs⁺ blockade. We did not find quantitative pharmacological data with which to compare our potency determinations for Sr²⁺ ions.

OPEN IN VIEWER

Fig. 3.

Sensitivity of K⁺ currents in KCNJ2 expressing cells to external Ba²⁺ ions. (A) Representative current traces recorded in a multi-hole QPlate, which groups signal derived from up to 10 cells in response to the voltage protocol schematized above the traces. (B) Normalized I–V relationships derived from N = 3 cells recorded in a single-hole plate in the absence (closed circles) or presence of 3 μM Ba²⁺ (open squares), 10 μM Ba²⁺ (closed triangles), or 100 μM Ba²⁺ (open diamonds) showing concentration and voltage dependency of K_ir2.1 blockade by external Ba²⁺. Error bars represent standard deviations. (C) Plot of steady-state current amplitude normalized with respect to baseline. Cells were exposed to successive vehicle control or test-article liquid application episodes, each of 3-min duration as indicated by vertical dashed line. Currents were activated by repeated application of the same step-ramp protocol as in (A). Control experiment (open circles) consisted in nine repeated applications of vehicle alone. Treatment experiment (closed circles) consisted in applications of vehicle buffer containing BaCl₂ at the indicated concentrations. The average two measurements at the end of the last vehicle period and at the end of each drug period served for current amplitude normalizations. Recordings were terminated by application of a saturating concentration of Ba²⁺ (3 mM) as full block.

OPEN IN VIEWER

Fig. 4.

Inhibition of K_ir2.1-mediated I _K1 currents by alkali or alkaline-earth channel blockers. Ba²⁺ (circles), Cs⁺ (squares), and Sr²⁺ (triangles) ions were tested as chloride salts in multi-hole QPlates. Group estimates were derived from at least N = 3 wells from overlapping concentration ranges applied cumulatively to the same cells (see legend to Fig. 3C and text for details and estimated IC₅₀ values with 95% confidence intervals). IC₅₀, half-maximal inhibitory concentrations

There are currently no selective small-molecule K_ir2.1 inhibitors with nanomolar potency described in the literature. ²² Even in the realm of toxins isolated from the venom of poisonous species, which are a rich source of high affinity ion channel modulators, no K_ir2.1 selective peptide inhibitors have been described so far. ⁴⁵ We therefore went on to characterize the pharmacology of the K_ir2.1-mediated I _K1 currents expressed in CHO cells using seven small-molecule inhibitors displaying micromolar potencies in manual patch-clamp experiments (Fig. 5). We started with the aromatic diamidine derivative PA-6, ³² the methoxybenzyl-methanamide ML133, ³¹ and the antimalaria drug chloroquine ⁴⁶ using the six-point cumulative application paradigm adopted with Ba²⁺ and the other ionic pore blockers. However, the active concentrations of these weak amines did not reach full inhibition within the 3-min exposure periods. Previous studies comparing the inhibitory activity of these compounds in the inside-out and whole-cell patch-clamp configurations evidenced an accelerated inhibition rate and increased potency against K_ir2.1 currents when excised patches were assayed. These data pointed to cytoplasmic sites of actions for these three compounds, and further studies with mutants established the involvement of negatively charged glutamate and aspartate residues at the level of the cytoplasmic K_ir2.1 pore domain. ^31,32,46 In the automated patch-clamp configuration, the time course for inhibition observed for such weak amine test articles applied externally results therefore from a combination of association kinetics at their internal sites of action and membrane permeation of their uncharged molar fraction at neutral pH. Entry by organic cation transporters has been shown for pentamidine, ⁴⁷ but it remains to be established whether such carriers are significantly expressed in CHO cells. We therefore adopted a noncumulative cell treatment protocol for the evaluation of PA-6, ML133, and chloroquine inhibitory potency. Cells were exposed to single concentrations of each compound in individual wells recorded in parallel. Figure 6A illustrates the time course of the effects of PA-6 when applied at concentrations ranging from 1.25 to 10 μM. Extension of the exposure duration to 15 min proved sufficient to reach stable inhibition for PA-6, ML133, and chloroquine. In contrast to Ba²⁺, the inhibitory action of PA-6 on the K_ir2.1 current traces indicated no time dependency during the hyperpolarizing pulse, displaying identical inhibition on its instantaneous and steady-state components (Fig. 6B). The inhibitory potency of PA-6 calculated on the average current amplitudes collected over the five last voltage steps to −120 mV under drug exposure was IC₅₀ = 1.3 μM (95% CI: 1.2–1.4 μM, N = 3–5, Fig. 6B) along with a Hill slope value near unity (n_H = 1.2, 95% CI: 1.2–1.4, Table 1). When tested with the same prolonged 15-min exposure protocol, ML133 and chloroquine yielded micromolar potencies (Table 1). Hill coefficients were close to 1, except for ML133, suggesting a cooperative mechanism of action for this compound independent of its slow kinetics of action as proposed by others. ³¹ These inhibitory activities are consistent with published data obtained using traditional manual patch-clamp. Thus, whole-cell K_ir2.1 currents elicited at −110 mV in KCNJ2-tranfected HEK cells were inhibited by 68% in the presence of 3 μM external PA-6, ⁴⁸ whereas ML133 showed an IC₅₀ value of 1.8 μM at −100 mV. ³¹ Similarly, in feline ventricular myocytes, 3 μM chloroquine decreased I _K1 currents at −100 mV by 67%. ⁴⁹

OPEN IN VIEWER

Fig. 5.

Chemical structure of disclosed compounds with K_ir2.1 inhibitory properties evaluated in this study: PA-6, N,N′′-((pentane-1,5-diylbis(oxy))bis(4,1-phenylene)) dibenzimidamide ³² ; ML133, N-(4-methoxybenzyl)-1-(naphthalene-1-yl) methanamine ³¹ ; Chloroquine, 4-N-(7-chloroquinolin-4-yl)-1-N,1-N-diethylpentane-1,4-diamine ⁴⁶ ; Chloroethylclonidine, N-[2,6-dichloro-4-[[2-chloroethyl(methyl)amino]methyl]phenyl]-4,5-dihydro-1H-imidazol-2-amine ⁵⁰ ; Chlorpromazine, 3-(2-chlorophenothiazin-10-yl)-N,N-dimethylpropan-1-amine ⁵¹ ; SKF-96365, 1-[2-(4-methoxyphenyl)-2-[3-(4-methoxyphenyl)propoxy]ethyl]imidazole ⁵² ; Terikalant, 1-[2-[(4S)-3,4-dihydro-2H-chromen-4-yl]ethyl]-4-(3,4-dimethoxyphenyl)piperidine. ⁵³

OPEN IN VIEWER

Fig. 6.

Slow inhibition of K_ir2.1-mediated I _K1 currents by the pentamidine derivative PA-6. (A) Four representative plots of steady-state current amplitude normalized with respect to baseline from multi-hole QPlates recordings are displayed. Induced CHO cells were first exposed to vehicle alone for the time period shown before the vertical dashed line. Then, in control experiments (open circles), the cells were exposed to vehicle alone, while in drug treatment recordings, the cells were exposed to vehicle containing 1.25 μM (closed squares), 5 μM (closed triangles), or 10 μM (closed diamonds) PA-6 (twofold dilution series). The inhibitory activity was allowed to develop over 15 min. The average two measurements at the end of the last vehicle period and the average five measurements at the end of a drug treatment period served for current amplitude normalization. A final episode in each experiment consisted in the application of a saturating concentration of Ba²⁺ as full block. (B) Representative multi-hole current traces in response to the voltage protocol shown above the traces. (C) Concentration–response curve for PA-6 inhibition using a threefold dilution series: IC₅₀ was estimated from at least N = 3 wells per concentration applied individually. See Table 1 for estimated value and 95% confidence interval.

Only a few small-molecule K_ir2.1 inhibitors have their molecular mechanism of action deciphered, and this was not an objective of this study. We rather extended our efforts to assess structurally diverse K_ir2.1-preferring or multichannel inhibitors to generate a high-quality patch-clamp data set for further pharmacological assay or protocol comparisons. Chloroethylclonidine, ⁵⁰ chlorpromazine, ⁵¹ SKF-96365, ⁵² and the class III antiarrhythmic agent terikalant ⁵³ are known to affect K_ir2.1-mediated I _K1 currents in recombinant or native cell systems at micromolar concentrations. We first screened them with the cumulative concentration application mode and observed slow inhibition kinetics for all of them, suggesting common actions at the cytoplasmic side of the K_ir2.1 channels. We therefore used the single-concentration noncumulative parallel application mode also for these four compounds; results of these quantitative evaluations are reported in Table 1. Only chloroethylclonidine appeared to reach fast-enough steady-state inhibition fitting a cumulative mode evaluation, which doubled its apparent potency with an IC₅₀ value of 25 μM (95% CI: 23−28 μM, N = 3) compared to the noncumulative application mode (data not shown).

Together with I _Kr and I _Ks currents flowing through the voltage-gated K_v11.1 and K_v7.1 channels, respectively, I _K1 currents contribute to the “repolarization reserve” of the cardiac AP, ^2,54 suggesting K_ir2.1 as a possible off-target for drug-induced cardiac side effects. Several screening assays for K_ir channels have previously been reported based on other readouts and formats. For example, a ratiometric fluorescence resonance energy transfer assay using cells in which K_ir-mediated potential changes are triggered by elevation of external [K⁺]_out could be miniaturized to a 1,536-well format. ⁵⁵ However, although well suited for high-capacity screening, assays based on fluorescence readouts tend to generate significant numbers of false positives. Another screening format utilized yeast strains rendered unable to grow in low [K⁺]_out media by genetic cancellation of their endogenous K⁺ transporters. Rescue of this phenotype by expression of mammalian K_ir2.1 and subsequent optical monitoring of yeast growth inhibition allowed the identification of a novel, although weak, catechol inhibitor. ⁵⁶ However, expression of mammalian proteins in prokaryotic hosts could lead to folding alterations or yeast-specific glycosylations, which may potentially alter the channel pharmacology. Unfortunately, assays based on cardiomyocytes induced from human pluripotent stem cell cannot be used for the detection of K_ir2.1 inhibitors since they express insufficient amounts of I _K1 to prevent them from spontaneously depolarizing and automatically beating. ⁵⁷ Thus, the assay we describe in this study provides a useful tool for cardiovascular safety screening. Furthermore, it provides a platform for the discovery of novel K_ir2.1 channel ligands as drug candidates.