HEK293 wild-type cells (HEK-WT; CRL-1573, ATCC) and HEK293 cells stably expressing the human Kv10.1 potassium channel (HEK-Kv10.1) (generously provided by Dr. Walter Stühmer from the Max Plack Institute) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (12800-017, Gibco) containing 10% fetal bovine serum (26140087, Gibco) and 1% Penicillin/Streptomycin (15140122, Gibco). HEK-Kv10.1 were supplemented with Zeocin (30 μg/mL) (R25001, Invitrogen) as a selection antibiotic. All cells were cultured at 37°C in a 5% CO₂ incubator.

FLIPR^® membrane potential assay BLUE (R8034, Molecular Devices) was performed following the manufacturer’s protocols. Flat-bottom 96-well plates (3,599, Costar) were plated with HEK-WT or HEK293-Kv10.1 at a density of 20,000 cells/well in 100 µL of supplemented DMEM and incubated for 24 h at 37°C in a 5% CO₂ incubator. After 24 h, 100 µL of the FMP buffer was added for 30 min at 37°C in a 5% CO₂ incubator; at this point, the molecules of interest were mixed in the FMP buffer and pre-incubated in the selected wells. All the molecules were tested at 100 μM; except astemizole, which was tested at 50 µM and DMSO (0.1%). After pre-incubation, 96-well plates were transferred to a FlexStation3 microplate reader (Molecular Devices) controlled by the SoftMax Pro 7 software. FLIPR dye was excited at 530 nm, and the emitted fluorescence was recorded at 565 nm. Data were acquired at 0.5 Hz for 120 s: the first 20 s represent the basal fluorescence, and then, 50 µL (5x) of the high potassium solution was added (that resulted in a final K⁺ concentration of 60 mM), and the fluorescence signal was recorded for another 100 s. The fluorescence responses were normalized by F=F/F₀, where F represents the fluorescence at any given time, and F₀ is the mean basal fluorescence obtained in the first 20 s of recording. Once normalized, the amplitude was calculated as the difference between the basal fluorescence and the maximal amplitude at 120 s. The maximal rate of fluorescence increase is obtained on the first derivative of the fluorescence signal.

Cells were plated on 18 mm circular glass coverslips previously treated with poly-l-lysine for patch-clamp recordings. Coverslips were transferred to a recording chamber (RC-26G, Warner Instruments, Hamden, CT, United States), and whole-cell voltage- and current-clamp recordings were performed at room temperature and under continuous perfusion (at a flow rate of 2 mL/min) with a standard bath solution. The standard bath solution contained (in mM): 137 NaCl, 5.4 KCl, 2 CaCl₂, 1.3 MgCl₂, 10 HEPES, and 10 D-glucose (300 mOsm/kg, pH 7.4 adjusted with NaOH). Intracellular patch-pipettes had resistances of 2–3 MΩ and were filled with an internal solution containing (in mM): 140 KCl, 1 MgCl₂, 10 EGTA, 10 HEPES (300 mOsm/kg, pH 7.2 adjusted with KOH). Ionic currents were recorded with a setup composed of Multiclamp 700B/Digidata 1,550/pCamp10 (amplifier/analog-digital converter/software, all from Molecular Devices), filtered at 5 kHz, and digitally sampled at 10 kHz. 40%–60% of the series resistance was electronically compensated. Drugs tested on Kv10.1 currents were evaluated from a holding voltage (Vh) of −70 mV, and currents were evoked with 250 ms duration voltage steps from −70 to +30 mV (for BL-1249) or +50 mV (for fluoxetine and miconazole) applied every 5 s. I-V curves were constructed using a protocol of 250 ms voltage steps from −100 mV to +50 mV in 10 mV increments and a Vh of −70 mV; the amplitude of Kv10.1 currents was measured at the end of voltage steps (average of the last 3 ms) and plotted versus voltage. Conductance was assessed as G = I_K/(V-V_K), where V_K is the reversal potential for K⁺ flux, I_K is the Kv10.1 current, and V is the membrane potential. The corresponding normalized G (G/Gmax) is plotted versus voltage; data were fit in Origin 2019 with a single Boltzmann distribution: G/Gmax = 1/(1+e^−(V−V1/2^)/s), where V1/2 and s are the arithmetic means of half-maximal activation potentials and slope factors, respectively.

We performed whole-cell current clamp recording in the gap-free mode to test the effect of BL-1249 on the resting membrane potential of HEK-WT and HEK-Kv10.1 cells. Cells were held without current injection (0 pA), and the protocol consisted of 60 s in the standard bath solution, followed by 60 s in BL-1249 or DMSO, and 60 s for washout.

Concentration-inhibitory curves were fitted with the Hill equation using the Levenberg–Marquardt method implemented in Origin 2019 software: Inhibition = Bmax*cⁿ/(IC50ⁿ + cⁿ), where Bmax is the maximum block, IC50 is the concentration of half-maximal inhibition, c is the concentration of the molecules, and n is the Hill coefficient.

For BL-1249, the concentration-effect curve was fitting using the method implemented in GraphPad Prism 8 software: Potentiation = 100/(1 + 10^{(LogEC50−X)}), where EC50 is the concentration of half-maximal potentiation. This model assumes data is normalized, forcing the curve to run from 0% to 100%.

Amitriptyline (A-155), bupivacaine hydrochloride (B-125), celecoxib (C-190), citalopram (C-195), flupirtine maleate (F-150), fluoxetine hydrochloride (F-155), ICA-11081 (I-160), L-ascorbic acid (L-140), linopiridine dihydrochloride (L-156), loperamide hydrochloride (L-100), miconazole nitrate (M-206), rosiglitazone (R-125) and 2,5-dimethylcelecoxib (D-150) were selected of a screening package for ion channel from Alomone Labs. BL-1249 (B2186), astemizole (A2861), and DMSO (D2650) were purchased from Sigma-Aldrich. Most stock solutions were prepared in 100% DMSO or water at a concentration of 100 mM. All molecules were diluted in a standard bath solution before starting every experiment. In patch-clamp experiments, drugs were delivered using a perfusion system VC-6 coupled to a fast-step solution charger (SF-77B, Warner Instruments, Hamden, CT, United States).

All data were analyzed using Origin 2019 (OriginLab, United States) and GraphPad Prism 8 (GraphPad Software, United States). Results are presented as mean ± SEM of at least three independent experiments. When two means were compared, statistical significance (p < 0.05) was assessed by Student’s t-test. For multiple comparisons, statistical significance (p < 0.05) was assessed by 1-way analysis of variance using the Dunnet post hoc test. Significant differences are expressed in figures: *p < 0.05, **p < 0.01, ***p < 0.001.

First, we tested the fluorescence responses to depolarization from HEK-WT and HEK-Kv10.1 cells using the FMP BLUE formulation. Adding a high K⁺ solution increased the fluorescence signals in both cell lines. Nonetheless, the fluorescence intensity recorded at 120 s in HEK-Kv10.1 cells was three times larger than in the HEK-WT, probably due to the expression of Kv10.1 channels (Figure 1A; p = 0.0006, unpaired t-test). Using the first derivative of the fluorescence responses, we observed that the maximum rate of change in HEK-Kv10.1 cells was around three times faster than in HEK-WT (Figures 1B, C; p = 0.0391, unpaired t-test). HEK-Kv10.1 cells that were not incubated with the FMP dye did not show a fluorescence signal either during the basal recording or the stimulation with a high potassium solution (Figure 1A). These results indicated that FMP assays are appropriate for studying Kv10.1 channel activity. Before starting the pharmacological screening, we evaluated the RMP on wild-type and Kv10.1-expressing cells. We measured the RMP in I = 0 mode within 30 s of achieving the whole cell configuration in patch-clamp, and recordings showed that HEK-WT cells have more depolarized values than HEK-Kv10.1, of −5.7 ± 1.0 mV (n = 31) and −44.4 ± 1.5 mV (n = 78), respectively (p < 0.001, unpaired t-test).

To validate the FMP assay, we pre-incubated HEK-Kv10.1 with previously described channel blockers, including astemizole, loperamide, and amitriptyline. As expected, astemizole, loperamide, and amitriptyline significantly decreased the fluorescence responses induced by the high potassium depolarization by 34.8%, 31.9%, and 25.5%, respectively (Figure 2A) (p = 0.0394, 0.0054, and 0.0192, respectively, 1-way ANOVA). DMSO (0.1%) did not produce any change in the responses (Figure 2A) (p = 0.8274, 1-way ANOVA). These results confirmed the method’s reliability. Next, we evaluated 12 small molecules and found that fluoxetine and miconazole nitrate significantly decreased the responses by 21.7% and 27.7%, respectively (Figures 2A–C) (p = 0.0361, and 0.0028, respectively, 1-way ANOVA). Interestingly, BL-1249 seems to increase the Kv10.1-mediated responses by 30% (Figure 2A); however, this effect was not statistically significant. Nevertheless, when we observed the basal responses (before normalization), we noticed that the basal fluorescence decreased statistically in the presence of BL (insert on Figure 2D) (p = 0.0189), suggesting BL could have activity on Kv10.1 channels.

Next, we performed whole-cell voltage-clamp recordings in HEK-Kv10.1 cells to determine whether fluoxetine and miconazole act on Kv10.1 channels.

Fluoxetine is a selective serotonin reuptake inhibitor used as an antidepressant (Wille et al., 2008). Our results showed that fluoxetine induced a dose-dependent and reversible inhibition of Kv10.1 currents. At 100 μM, fluoxetine produced a 96.1% ± 2.9% inhibition of Kv10.1 currents (n = 5, p < 0.0001; paired t-test; see Figures 3A–C). Figure 3D shows the dose-concentration inhibition effect of Fluoxetine on Kv10.1 currents; these data were fitted with a Hill equation, which yielded an IC50 of 11.2 ± 6.2 µM and a Hill coefficient of 0.8 ± 0.1 (Figure 3D).

MN is an antifungal agent that has also been reported to have Kv blocker activity (Kikuchi et al., 2005). As shown in Figure 4A, MN produced a time and dose-dependent irreversible inhibition of Kv10.1 currents (Figure 4A). At 100 μM, MN inhibited Kv10.1 currents by 93% (n = 10, p < 0.0001; paired t-test) (Figure 4C). The inhibition percentage at each concentration tested were fitted with a Hill equation, which yielded an IC50 of 23.6 ± 5.0 µM and a Hill coefficient of 1.5 ± 0.3 (Figure 4D).

BL-1249 has been described as a negatively charged activator of different types of K⁺ channels (Schewe et al., 2019). Here, we observed that BL potentiates in a dose-dependent and reversible manner the Kv10.1 currents. Due to the solubility of BL, the maximum concentration tested was 100 µM. As shown in Figures 5A–C, rapidly increased by 90% the Kv10.1 currents measured at +30 mV (n = 9; p = 0.0001; paired t-test). Since we could not test higher concentrations of BL, the maximal potentiating concentration is unknown. However, we decided to perform a trend analysis to evaluate whether the dose-dependent effect followed a direct relationship (p = 0.0013; trendy test). Posterior, the dose-concentration fitting yielded an EC50 of 36.5 µM and a hill slope of 1.4. Analysis of Kv10.1 conductance showed that BL shifted their V_1/2 activation by 44 mV to the left, from 0.6 ± 1.1 mV in control to −44.9 ± 6.2 mV in the presence of 100 µM BL (n = 6). These results suggest for the first time the potentiation effect of a molecule on Kv10.1 channels.

Next, we are interested in evaluating whether BL could hyperpolarize the resting membrane potential (RMP) of HEK-Kv10.1 cells. We performed current-clamp recordings to measure the RMP without current injection. HEK-Kv10.1 cell had a RMP of −56.8 ± 3.1 mV (n = 11). In the presence of DMSO (applying for 1 min), the RMP did not show any statistically significant change, from −56.1 ± 8.7 mv in control to −56.3 ± 8.9 mV with DMSO (Figures 6A,C,E) (n = 4; p = 0.345; paired t-test). In contrast, BL (100 µM) hyperpolarized the RMP of HEK-Kv10.1 cells by 20 mV, from −57.3 ± 1.9 mV in control to −77.2 ± 1.2 mV in the presence of BL (Figures 6B,D,F) (n = 7; p < 0.0001; paired t-test). Interestingly, during the washout, the RMP suffered a transitory rebound to more depolarized values.

We also evaluated whether BL could potentiate the activity of the endogenous currents of HEK-WT cells. BL did not show a significant effect on currents measured at +50 mV from 162.4 ± 18.4 pA in control to 131.2 ± 15.0 pA in the presence of 100 µM BL (n = 10; p = 114, paired t-test) (Figures 7A–C). Likewise, RMP in HEK-WT cells did not change when BL (100 µM) was added, yielding values from −7.9 ± 1.2 mV in control to −6.6 ± 0.7 mV with BL (n = 6; p = 1754, paired t-test) (Figures 7D, E).

We tested next whether it was possible to counteract the inhibitory effect of some of the new antagonists described here with BL. As is shown in Figure 8, fluoxetine (30 µM) induced a rapid and statistically significant decrease (∼83%; n = 6; p = 0.0029, paired t-test) in the Kv10.1 currents measured at +30 mV. Surprisingly, the combined application of fluoxetine plus BL (100 µM) potentiated the inhibitory effect of fluoxetine to 93% (n = 6; p = 0.0034, paired t-test). The inhibitory effect of fluoxetine was reversible when the molecule was retired from the bath solution. Then, BL tested alone still could potentiate the Kv10.1 currents (p = 0.021, paired t-test). These results open new possibilities for pharmacological studies on the Kv10.1 channel.