Therapeutic Potential of Small Molecules that Block receptor-induced Kv7/M-Current Suppression in Neuroprotection, Seizures, and Pain

Abstract

Background and Purpose:

Neuronal Kv7 channels generate low-voltage–gated potassium currents known as the M-current. The M-current is transiently suppressed by activation of Gq-coupled receptors, which is caused by a loss of an essential cofactor, phosphatidylinositol 4,5-bisphosphate (PIP2), from the Kv7 channel complex, thereby increasing neuronal excitability. M-current suppression–mediated neuronal overexcitation contributes to the pathogenesis of several disorders, including seizures, pain, and excitotoxicity. However, therapeutic strategies targeting M-current suppression remain unutilized. We hypothesized that small-molecule binding to a state-dependent cytosolic pocket of Kv7 channels stabilizes the ion-conducting conformation and mitigates M-current suppression.

Experimental Approach:

The Kv7 cytosolic pocket was used as a target site for in silico screening of 3D molecular libraries of clinically used compounds. Candidate compounds were functionally validated using patch-clamp recordings in CHO cells heterologously expressing Kv7 channels and human m1 muscarinic receptors. Selected model compounds were further characterized for effects on modulations to Kv7 channels, neuroprotection, and in vivo outcomes using pilocarpine-induced seizure and formalin paw inflammatory pain models.

Key Results:

Several compounds reduced oxotremorine-M–induced Kv7.2 current suppression. Dabigatran etexilate, carvedilol, and nebivolol were selected as model compounds for further study. These compounds (1) attenuated m1 muscarinic receptor–induced suppression across all Kv7 subtypes and endogenous M-currents, (2) protected neurons from glutamate- or hydrogen peroxide–induced neurotoxicity, and (3) reduced inflammatory pain responses. Dabigatran etexilate additionally mitigated pilocarpine-induced seizures.

Conclusion and Implications:

We identified a new class of Kv7 channel modulators that reduce M-current suppression and are efficacious in several disease models. This mechanism provides a rationale for developing novel therapeutics targeting this cytosolic site.

1. Introduction

Overexcitation of neurons contributes to various neurological pathologies, including seizures, pain, and excitotoxicity. These enhanced neuronal activities are often caused by various neurotransmitters and chemical mediators. One established mechanism to increase neuronal excitability is M-current suppression (Brown & Passmore, 2009; Greene & Hoshi, 2017). The M-current is generated by low-voltage-gated potassium channels in the KCNQ/Kv7 family, which includes Kv7.2, Kv7.3, Kv7.4, and Kv7.5, while Kv7.1 subtype is considered cardiovascular. Various combinations of heteromeric and homomeric Kv7 channels contribute to the M-current (Brown & Passmore, 2009; Hadley et al., 2003; Jentsch, 2000; Springer, Varghese, & Tzingounis, 2021). Since Kv7 channels are activated at low voltages near the resting membrane potential, the M-current counteracts depolarization below the threshold of action potential firing, thereby providing negative regulation on neuronal excitation. Consequently, when the M-current is suppressed by activation of Gq-coupled receptors, same current inputs cause larger depolarization thus increasing neural firing (Brown & Passmore, 2009; Greene & Hoshi, 2017). Conversely, attenuating M-current suppression can reduce neuronal overexcitation.

Extensive studies have identified three primary mechanisms mediating M-current suppression: 1) phosphorylation of the Kv7 subunit by channel-anchored protein kinase C (Hoshi et al., 2003), 2) changes in calmodulin binding configuration to Kv7 subunits induced by cytosolic calcium (Kosenko & Hoshi, 2013), and 3) depletion of PIP2 by phospholipase C at high agonist concentrations (Suh & Hille, 2002). All these pathways result in a loss of PIP2 from the Kv7 subunit (Greene & Hoshi, 2017). Since PIP2 is crucial for maintaining the ion-conducting conformation of Kv7 channels (Ma et al., 2023; Sun & MacKinnon, 2020; Zheng et al., 2022), its loss reduces the number of ion-conducting channels, leading to M-current suppression.

Our previous study demonstrated that dehydroepiandrosterone sulfate (DHEAS), an endogenous hormone, can stabilize the ion-conducting Kv7.2-PIP2 conformation, diminishing M-current suppression (Alhassen et al., 2023). This finding suggests that small molecules may be able to modify M-current suppression. However, since DHEAS is not permeable to the blood-brain barrier and is also a precursor of other steroid hormones, its use as a Kv7 channel modulator is limited. Therefore, this study aimed to identify small compounds that bind to the DHEAS binding site and exhibit similar effects, to evaluate therapeutic potential of this new class of Kv7 channel modulators.

2. Methods

2.1. In silico Screening.

Computational binding analyses were performed using Autodock Vina (Trott & Olson, 2010) in Pyrx software package (pyrx 0.8) (Dallakyan & Olson, 2015). The template structure for the functional Kv7.1 was obtained from the published structure of Kv7.1-KCNE3-CaM-PIP2(+) (pdb: 6v01) (Sun & MacKinnon, 2020). The template for inactive Kv7.1-PIP2(−) structure was obtained from pdb: 6v00 (Sun & MacKinnon, 2020). 3D molecular models were obtained from the Zinc database (https://zinc15.docking.org). Protein structures were visualized in MacPymol.

2.2. Reagents and Expression Plasmids.

Expression plasmids for KCNQ1,2,3,4,5 (Alhassen et al., 2023), rat KCNQ2(H558C) plasmid (Alhassen et al., 2023), calmodulin plasmids (Kang, Xu, Cooper, & Hoshi, 2014) have been described. cyt-CKAR construct is from Dr. Alexanda Newton (Gallegos, Kunkel, & Newton, 2006). CFP-PH construct was obtained from Dr. Tobias Meyer (Stanford University) through Addgene. eVSP construct was obtained from Dr. Yasushi Okamura (Osaka University, Japan) via addgene. Carvedilol, DHEAS, dabigatran etexilate mesylate, dabigatran were purchased from Cayman Chemical (Ann Arbor, MI). Dabigatran etexilate mesylate, nebivolol were purchased from Med Chem Express (Monmouth Junction, NJ).

2.3. Cell culture and Transfection.

Chinese Hamster Ovary-K1 cells (CHO-K1, RRID:CVCL_0214) stably expressing the human muscarinic m1 receptor (CHO hm1) (Buckley, Bonner, Buckley, & Brann, 1989) were grown in alpha minimum essential medium with 5% fetal bovine serum and 500 μg ml⁻¹ G418 sulfate. This CHO hm1 cell line produces consistent m1 responses across passages and has been used in our lab since 2003 (Hoshi et al., 2003). This continuity allows direct comparison with our historical data. TransIT-LT1 reagent (Mirus Bio, Madison, WI, USA) and expression plasmids were used for transient transfections (Smith & Hoshi, 2011). Superior Cervical Ganglion (SCG) neurons were prepared from adult mice and cultured for 2 to 4 days as described (Hoshi et al., 2003). For testing effects of valproate, CHO hm1 cells were pretreated with 500 μM valproate for 24 h before electrophysiological measurements since AKAP150-mediated effects of valproate requires overnight treatment as we reported (Kay, Greene, Kang, Kosenko, & Hoshi, 2015). Cortical neurons were prepared from newborn pups (P0 – P1), plated in 96 well plates at 5 × 10⁴ cells/well, and maintained as we described previously (Kosenko et al., 2012).

2.4. Electrophysiological Measurements.

Patch-clamp recordings were performed at room temperature on isolated CHO hm1 cells using an Axopatch 200B patch-clamp amplifier (Molecular Devices, Sunnyvale, CA, USA) as described previously (Greene, Kosenko, & Hoshi, 2018). Amphotericin-B based perforated-patch clamp technique is used for M-current recording from mouse SCG neurons (Greene et al., 2018). Series resistance compensation was set to 75–86%. Signals were sampled at 2 kHz, filtered at 1 kHz, and acquired using pCLAMP software (Version 7, Molecular Devices). For patch-clamp experiments, cells were perfused with a solution containing 144 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 0.5 mM MgCl₂, 10 mM glucose and 10 mM HEPES (pH 7.4). Patch pipettes (4 – 6 MΩ) were filled with intracellular solution containing 135 mM potassium aspartate, 2 mM MgCl₂, 1 mM EGTA, 0.1 mM CaCl₂, 4 mM ATP, 0.1 mM GTP and 10 mM HEPES (pH 7.2). Kv7 currents were evoked by two-step test pulses from a holding potential of −80 mV to 0 mV followed by −60 mV, with 500 ms duration for each step at a 10-s interval. Kv7 currents were measured at the end of the 0-mV step for time course experiments. Measured currents were normalized to reference current amplitudes to adjust current density variations among cells. For measuring activation curves, Kv7.2 tail currents at −60 mV were measured and normalized to the maximal current. For measuring the M-current in SCG neurons, cells were held at −30 mV and 500-ms steps to −50 mV were given every 10 s. Data were collected from multiple batches of neuron preparations derived from multiple animals. Primary neuronal cultures are prepared from 4–8 mice per batch. M-current amplitudes were measured as difference between averaged current amplitudes of a short segment (30–50 ms) following peak tail currents, and that of a short segment (30–50 ms) at the end of 500-ms pulse. To test the reliability of this measurement, several variations of M-current quantification were compared (Fig. S1). All methods provided comparable results. However, extracting the XE991-sensitive current by applying XE991 at the end of experiments had a drawback due to changes in hyperpolarization activated currents after oxo-M application, likely originate from calcium-activated chloride currents (Marsh, Trouslard, Leaney, & Brown, 1995; Clark, Jordt, Jentsch, & Mathie, 1998). Therefore, we chose to use a classic protocol of tail-current decays during 500-ms hyperpolarizing steps.

For eVSP experiments, 1:1 mixture of eVSP and Kv7.2 plasmids were used to transfection. After two days transfection, patch clamped cells were conditioned by series of 10-s steps to +60 mV until reaching complete suppression of Kv7.2 current, then 10-s steps to +60 mV were applied every 1 min thereafter, which gave consistent and stable recordings. A 100-ms voltage step to −60mV from −70 mV before the test 60 mV step was applied to calculate the linear leak as we described (Alhassen et al., 2023; Kosenko & Hoshi, 2013).

2.5. Live cell imaging.

All imaging procedures have been previously described (Kosenko & Hoshi, 2013; Kosenko et al., 2012). Transiently transfected CHO hm1 cells were plated onto 18-mm glass coverslips 24 hours after transfection. In post-transfection day two, Images were acquired with an inverted microscope IX-81 (Olympus, Tokyo, Japan) and an ImageEM CCD camera (Hamamatsu Photonics, Shizuoka, Japan) and processed with MetaMorph 7.6.3 (Molecular Devices, Sunnyvale, CA, USA). For total internal reflection flurorescence (TIRF) experiments for CFP-PH, the excitation source was a 445-nm diode laser (Coherent, Santa Clara, CA, USA) with an acousto-optical tunable filter.

Dual-emission images of cyt-CKAR experiments were obtained as follows. A xenon arc lamp (Lambda LS, Sutter Instrument, Novato, CA) equipped with an excitation filters ET436/20× (Chroma Technology) was used for excitation of CFP, and two types of fluorescent signals were detected through a dual-view module (Photometrics); ET480/40m for CFP (direct CFP, dCFP) and ET535/30m for YFP (indirect YFP, iYFP).

Signals from these two channels were used to calculate ratio of two channels as an indicator of fluorescence resonance energy transfer (FRET) signals as described previously (Hoshi, Langeberg, Gould, Newton, & Scott, 2010; Smith & Hoshi, 2011). All signals were normalized to the average values of first seven images corresponding to (t = 60 sec to −10 sec) to pool collective data.

2.6. Short trajectory Molecular Dynamic (STMD) Analysis.

To assess the quality of binding modes obtained from Autodock Vina, a short trajectory molecular dynamics (STMD) run was performed using OpenEye Scientific Software’s Orion platform (Sørensen et al., 2024). A structure Kv7.1 (PDB: 6v01) and Autodock Vina was used to determine starting binding poses to use in STMD runs. Input files required for docking with Autodock Vina including dabigatran etexilate, DHEAS, carvedilol, and nebivolol, were prepared with Meeko (https://pypi.org/project/meeko/). For the molecular dynamics run, each protein was parametrized with Amber FF14SB (Maier et al., 2015), and the ligands were parametrized with OpenFF-2.2.0 force field (Boothroyd et al., 2023). AM1-BCC ELF10 charges were applied for the ligands (Jakalian, Jack, & Bayly, 2002), and the ligand bound NPT production calculations were run for 4 nanoseconds each, and the MD engine in the backend was OpenMM (Eastman et al., 2024). Rest of the parameter defaults of the workflow remain unchanged. This workflow also provides ensemble average molecular mechanics/poisson–boltzmann surface area (MMPBSA) energies for each of the ligands.

2.7. Neuroprotection assay.

DIV7–13 cultured mouse primary cortical neurons were used for neuroprotection assays. Test compounds (20 μM) were added 10 minutes prior to treatment with either glutamate [50 μM for wild-type and 200 μM for Kv7.2(S559A) neurons] or 100 μM hydrogen peroxide. Cells were incubated for 30 minutes for glutamate from wild-type mice, 1 hour for glutamate from Kv7.2(S559A), or 1 hour for hydrogen peroxide. After treatment, cells were washed twice with culture medium. Following the second wash, culture medium with test compounds (20 μM) were added and cells were cultured overnight. The following day, cell viability was assessed using the CyQUANT MTT Cell Viability Assay (Thermo Fisher Scientific, MA, USA), according to the manufacturer’s instructions.

2.8. Experimental Animals.

All animal care and experimental procedures followed the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals. All protocols used in this study were approved by the Institutional Animal Care and Use Committee of the University of California, Irvine and the Institutional Biosafety Committee of the University of California, Irvine. All animals were randomly assigned to experimental groups and behavior was scored by treatment blinded observers. Animal studies are reported in compliance with the ARRIVE guidelines (Percie du Sert et al., 2020) and with the recommendations made by the British Journal of Pharmacology (Lilley et al., 2020). Adult C57BL/6 mice of both sexes were used in this study. Mice were group housed and maintained on a 12 h light/dark cycle with food and water available ad libitum. Any mouse that seemed to be in distress was excluded from the study.

Test compounds were injected into mice intraperitoneally; dabigatran etexilate mesylate (7 mg kg⁻¹), carvedilol (10 mg kg⁻¹), or nebivolol (2 mg kg⁻¹). Injection solutions were freshly prepared on the day of the from a stock solution in DMSO.

2.9. Pilocarpine-induced seizures.

Pilocarpine-induced seizures were induced as we described previously (Greene et al., 2018). Briefly, methyl scopolamine (1 mg kg⁻¹, ip) was administered 15 minutes before administration of pilocarpine (289 mg kg⁻¹, ip). For anti-seizure experiments, twenty minutes after pilocarpine injection, mice were injected with either Dulbecco’s phosphate buffered saline (DPBS), dabigatran etexilate (7mg kg⁻¹, ip), carvedilol (10mg kg⁻¹, ip) or nebivolol (2 mg kg⁻¹, ip) dissolved in DPBS. For delayed administration experiments, above test drugs were administered 2 h after pilocarpine injection together with diazepam (5 mg kg⁻¹) in 5% glucose solution. Mice were videotaped for 2 hours after pilocarpine injection, and behavior was scored by treatment blinded observers every 10 minutes according to the Racine scale: stage 1—immobility, lying low; stage 2—twitching/tremor/shaking of tail/head/body/limbs, not continuous; stage 3—continuous tremor/clonic seizures of body while retaining posture; stage 4—rearing/hyperexcitability/running, tonic extension/clonic seizures with loss of posture; and stage 5—Status epilepticus (continuous stage 4 seizures). Sample sizes were as follows: control, 11 males and 7 females; dabigatran etexilate, 7 males and 6 females; carvedilol, 3 males and 3 females; nebivolol, 2 males and 5 females; dabigatran etexilate + XE991, 5 males and 6 females. We did not detect any sex significance in 2-way ANOVA, thus combined results are used in this study.

For Fluoro-Jade C (FJC) assays, two days after pilocarpine injection, mice were deeply anesthetized and fixed with 4% paraformaldehyde perfusion and brains were collected. 30 μm coronal sections of brains were stained by 0.0001% Fluoro-Jade C (Histo-Chem Inc, AK) in 0.1% acetic acid for 20 min. Fluorescent images were collected by a fluorescent microscope (IX-81, Olympus) equipped with Image EM CCD camera (Hamamatse Photonics).

2.10. Formalin Paw Assay.

The formalin paw assay was performed as described (Alhassen et al., 2023). Mice were habituated to the testing arena for 7 days before tests. Either vehicle (PBS, ip), dabigatran etexilate (7 mg kg⁻¹, ip), carvedilol (10mg kg⁻¹, ip), or nebivolol (2mg kg⁻¹, ip) were administered 10 min prior to the formalin injection. 10 μL of 2% formalin solution was administered into the dorsal surface of the right hind paw using a Hamilton syringe with a 30G needle. After formalin injection, mice were placed in the testing arena and paw licking was observed for 50 min. Amount of time mice spent licking the injected paw during the first 15 minutes was defined as the early-phase, and the period between 20–50 minutes as the late phase. Paw licking duration was recorded by experimenters blinded to drug treatments. All animals subjected to the formalin paw assay were euthanized immediately after the assessment to alleviate inflammation induced distress. Sample sizes were as follows: control, 10 males and 9 females; dabigatran etexilate, 4 males and 8 females; carvedilol, 8 males and 4 females; nebivolol, 5 males and 6 females; dabigatran etexilate + XE991, 6 males and 4 females; carvedilol + XE991, 4 males and 4 females; nebivolol + XE991, 5 males and 5 females. We did not detect sex differences in 2-way ANOVA. Two mice treated with carvedilol and XE991 became flat and immobile, behaviors similar to stage 1 of pilocarpine seizure, and did not lick paw at all. We did not notice any difference in behavior before injection of formalin. These two mice with abnormal/distressed behaviors have been removed from the analysis.

2.11. Statistics.

All pooled values are expressed as mean ± SEM. One-way ANOVA followed by Dunnett’s multiple comparisons test was used when standard deviations among comparison groups are not significantly different in the Brown-Forsythe test. When distinct standard deviations were detected, the Kruskal-Wallis test followed by Dunn’s multiple comparisons test was used. For eVSP experiments, paired Student’ t-test was used to evaluate significance. Statistical significance was defined as two-tailed p < 0.05. P-values for statistical analyses were calculated using Prism 6 (GraphPad software, La Jolla, CA, USA). The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology (Curtis et al., 2025).

2.12. Data availability.

All data reported in this paper will be shared upon reasonable request. This study includes no data deposited in external repositories. The complete in silico screening result has been provided as supplementary file 1.

2.13. Nomenclature of Targets and Ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, and are permanently archived in the Concise Guide to PHARMACOLOGY 2023/24 (Alexander et al., 2023).

3. Results

3.1. In silico screening and functional validation for a new type of Kv7 channel modulators.

The ion-conducting structure of Kv7.1-KCNE3-CaM in the presence of PIP2 (Kv7.1-PIP2(+), pdb: 6v01) (Sun & MacKinnon, 2020) was chosen as the drug receptor template for in silico screening because it accurately predicted the binding properties of the prototypical compound, DHEAS, and excluded related steroid hormones showing null effects on Kv7.2-current suppression in our previous study (Alhassen et al., 2023). Notably, the cytosolic pocket that serves as the DHEAS binding site collapses in the Kv7.1-KCNE3-CaM structure lacking PIP2 (Kv7.1-PIP2(−), pdb: 6v00) (Sun & MacKinnon, 2020), which could be considered as an analogous conformation to Kv7 channels under M-current suppression (Figure 1A). Therefore, we hypothesized that small molecules binding to this pocket might protect it from collapsing, thus preserving the ion-conducting Kv7.1-PIP2(+) conformation. Based on this idea, we conducted in silico screening of 3D molecular libraries from the Zinc15 database (Sterling & Irwin, 2015) using AutoDock Vina (Trott & Olson, 2010) to identify molecules that bind to this cytosolic pocket. We screened subsets of the Zinc15 database containing FDA-approved drugs (1379 compounds) and drugs approved worldwide but not by the FDA (2068 compounds), because these compounds are readily available and can be tested in animals. Overall screening results showed a strong correlation in calculated binding free energies between Kv7.1-PIP2(+) and Kv7.1-PIP2(−) (r = 0.89, Figure 1B), with slightly higher affinities toward Kv7.1-PIP2(+) (Figure 1C), which is consistent with the presence of a state-dependent cytosolic pocket.

Figure 1. Target cytosolic drug binding site in Kv7.1 channel structures for in silico screening and functional verification using homomeric Kv7.2 channels.

A) Protein structures of Kv7.1 channel complexes in the presence of PIP2, Kv7.1-PIP2(+) from pdb:6v01, and its absence, Kv7.1-PIP2(−) from pdb:6v00. Protein structures are shown as grey ribbon models and cytosolic pockets for Kv7.1-PIP2(+) in cyan and for Kv7.1-PIP2(−) in magenta. The target drug binding site used for screening is shown as a black rectangle. B) Scatter plots of calculated binding free energy for Kv7.1-PIP2(+) in x-axis [ΔG-PIP2(+)], and that for Kv7.1-PIP2(−) in y-axis [ΔG-PIP2(−)] (n = 3193). Correlation co-efficient is indicated. C) Cumulative probability distribution of data shown in panel B, indicating stronger binding toward Kv7.1-PIP2(+) over Kv7.1-PIP2(−). D) Representative current traces of homomeric Kv7.2 current traces showing before [oxo(−) t = 0] and after 0.3 μM oxo-M-induced current suppression [oxo(+) t = 1 min] and its attenuation of response by dabigatran etexilate (dabigatran E/dabE) E) Time course of oxo-M-induced current suppression shown in D (n = 38 control, n = 21 dabE). Dashed lines indicate peak suppression at the end of oxo-M application (t = 1 min), and that of current recovery after wash out (t = 2 min). Black box indicates presence of 0.3 μM oxo-M. Error bars show SEM. F) Quantitative summary of peak Kv7.2 current suppression at t = 1 min including data shown in E. control (n = 38, 19 replicates), carvedilol (car, n = 17, 3 replicates), dabigatran etexilate (dabE, n = 21, 7 replicates), ebastine (eba, n= 17, 2 replicates) FAD (n = 17, 2 replicates), flutroline (FL, n = 11, 4 replicates), nebivolol (neb, n = 10, 4 replicates), dabigatran (dab(bare), n = 13, 4 replicates). Red bars indicate significantly different compounds from control. G) Quantitative summary of Kv7.2 current recovery at t = 2 min. Red bars indicate significantly different responses from the control. X represent individual cells. ns = not significant, * < p = 0.05, ** < p = 0.01, *** < p = 0.001 to control by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars represent SEM.

Taking advantage of the difference in state-dependent affinities, we added a secondary criterion to exclude high-affinity binding compounds to the ion impermeable Kv7.1-PIP2(−) conformation. This approach narrowed down candidates for subsequent electrophysiological verification (Figure S2). Functional verification experiments were conducted in CHO hm1 with transiently expressed homomeric Kv7.2 channels (Alhassen et al., 2023; Kang et al., 2014; Kosenko & Hoshi, 2013). During electrophysiological measurements, we observed that some compounds inhibited Kv7.2 currents at our screening concentration of 10 μM; candesartan (6.5 ± 11% of control, n = 5), ebastine (57 ± 17%, n = 8), nebivolol (49 ± 17%, n = 6), rolapitant (27 + 14%, n = 6). Compounds with moderate inhibition of Kv7.2 currents (nebivolol and ebastine) were tested with a reduced concentration at 1 μM in a range where they do not show inhibition. Muscarinic m1 receptor-induced Kv7.2 current suppression was measured to test whether candidate compounds attenuate Kv7.2 current suppression. After measuring baseline Kv7.2 currents as a reference, cells were treated with test compounds at 10 μM for 2 minutes with exception of nebivolol and ebastine (1 μM) and flavin adenine dinucleotide (FAD). FAD was directly added to the pipette solution at 1 mM because FAD is a cell-impermeable intracellular co-enzyme. Subsequently, 0.3 μM oxotremorine-M (oxo-M) was applied for 1 minute to induce current suppression, followed by wash out of oxo-M to observe current recovery (Figure 1D&E). This assay identified eight compounds that attenuate Kv7.2 current suppression from a shortlist of 22 candidates (Figure S2). The 8 identified compounds were antrafenine, carvedilol, dabigatran etexilate, ebastine, FAD, flutroline, nebivolol, and nilotinib (Figure S2). Example of attenuation of Kv7.2 current suppression by 10μM dabigatran etexilate is shown in Figure 1D&E. Most of identified compounds attenuated peak Kv7.2 current suppression (Figure 1F), but also facilitated recovery from current suppression (Figure 1G, Figure S2). Since dabigatran etexilate is a prodrug for dabigatran (Blech, Ebner, Ludwig-Schwellinger, Stangier, & Roth, 2008), we also tested bare dabigatran itself. However, bare dabigatran (10 μM) had no effects on Kv7.2 current suppression nor recovery (Figure 1F&G). For further characterization, carvedilol, dabigatran etexilate, and nebivolol were chosen as model compounds.

3.2. Functional characterization of identified Kv7 channel modulators.

Since homomeric Kv7.2 channels represent a minor fraction of the endogenous M-current (Hadley et al., 2003; Springer et al., 2021), we next examined effects on endogenous M-currents. Endogenous M-currents were measured using the perforated patch recording from adult mouse SCG neurons (Figure 2). Three model compounds showed similar efficacy in attenuating peak M-current suppression induced by oxo-M (1μM) (Figure 2B&C). To further characterize the subtype specificity of these compounds, we tested other Kv7 subtypes in CHO hm1 cells. All three model compounds attenuated oxo-M–induced peak current suppression of heteromeric Kv7.2/Kv7.3 channels as expected, which constitute the major component of the M-current in SCG neurons (Figure 3A) (Brown & Passmore, 2009; Greene & Hoshi, 2017; Jentsch, 2000). Kv7.1 is a cardiac subtype, which was used as the template for our in-silico screening. The weaker suppression of homomeric Kv7.1 current at t = 1 min (Figure 3B) compared to that of Kv7.2 + Kv7.3 current (Figure 4A) aligns with the absence of the PKC-mediated suppression pathway in Kv7.1 (Hoshi et al., 2003) and model compounds attenuated suppression to equivalent degree to other Kv7 subtypes. Both homomeric Kv7.4 and Kv7.5 current suppressions were attenuated by three model compounds (Figure 3C&D).

Figure 2. Effects of model compounds on current suppression of the endogenous M-current in mouse SCG neurons.

A) Representative current traces of M-currents and 1 μM oxo-M-induced current suppression and their modulations by the indicated compounds. B) Time courses of relative M-current amplitudes in response to 1μM oxo-M. Indicated test compounds were applied 2 min before application of oxo-M. Test compounds and sample numbers are indicated. C) Quantitative summary of panel B showing peak current suppression at t = 1 min. Data were collected from 2–3 replicates from multiple animals. *** < p = 0.001, **** < p = 0.0001 by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars show SEM.

Figure 3. Effects of model compounds on current suppression of indicated Kv7 subtypes in CHOhm1 cells.

A) Effects of three model compounds on heteromeric Kv7.2/Kv7.3 current suppression: black open circles = control, red = dabigatran etexilate, purple = carvedilol, and blue = nebivolol. Color codes are same throughout this figure. Left panel shows time course and right panel shows quantitative summary of peak suppression at t = 1 min. Sample numbers are also indicated. Black box indicates presence of 0.3μM oxo-M. B) Effects on homomeric Kv7.1 current suppression. C) Effects on homomeric Kv7.4 current suppression. D) Effects on homomeric Kv7.5 current suppression. Results were obtained from 2–6 replicates. * < p = 0.05, ** < p = 0.01, *** < p = 0.001, **** < p = 0.0001 by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars show SEM.

Figure 4. Characterization of model compounds on Kv7.2 current modulation.

A) Kv7.2 current responses to 2 min exposure to indicated concentrations of nebivolol (left) and a derived dose-response curve (right). Data from 2 replicates. B) Pooled voltage-activation curves of homomeric Kv7.2 currents with following compounds: black open circles = control, red = dabigatran etexilate, purple = carvedilol, and blue = nebivolol. Color codes are same throughout B–F. C) Quantitative summary of effects on half-activation voltage on Kv7.2 current from experiments shown in B. D) Time course of Kv7.2 current suppression induced by 10 μM oxo-M with the indicated test compounds. (E) Quantitative summary of peak suppression at t = 1 min from dataset shown in D. (F) Quantitative summary of current recovery at t = 3 min shown in D. Data from 2–3 replicates. ns, not significant, ** < p = 0.01, *** < p = 0.001, **** < p = 0.0001 by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars show SEM.

Our binding model assumes one drug binding site per subunit, thus four binding sites per channel. Because nebivolol showed Kv7.2 inhibition in addition to attenuation of Kv7.2 current suppression, we measured cooperativity of nebivolol inhibition on homomeric Kv7.2 channels to assess stoichiometry. Steady-state dose-inhibition responses of Kv7.2 current was best fitted to the Hill equation with IC50: 4.4 ± 0.6 μM (n = 6 –11) with Hill coefficient (h): 1.3 ± 0.19 (Figure 4A). A Hill coefficient of 1.3 supports multiple molecular bindings of nebivolol per Kv7.2 channel. Furthermore, since attenuation of oxo-M induced Kv7.2 suppression occurred at 1 μM nebivolol, a concentration that did not inhibit the Kv7.2 current, it is likely that binding of a single molecule per channel is sufficient to attenuate Kv7.2 suppression.

Next, we examined effects on the activation voltage of Kv7 channels because existing Kv7 channel activators, such as retigabine, alter the activation voltage of Kv7 channels. None of the three model compounds shifted the half-activation voltage (V_1/2) of Kv7.2 channels (Figure 2B&C).

To further test efficacy of these identified compounds, we evaluated effects under a saturating concentration of oxo-M (10 μM) (Alhassen et al., 2023). In controls, 10 μM oxo-M reduced Kv7.2 current to ~9% of the reference amplitudes, and the test compounds did not significantly alter peak suppression (Figure 2D&E). However, cells treated with either of test compounds exhibited a facilitated recovery from current suppression after wash out (Figure 2F). This suggests that facilitated recovery is a more sensitive indicator for their pharmacological effects. On the other hand, valproate, which attenuates M-current suppression by inhibiting palmitoylation of AKAP79/150, a Kv7.2-anchored scaffold protein (Kay et al., 2015), suppressed peak Kv7.2 current suppression but had negligible effects on current recovery after wash out (Figure S2). Together, these results suggest that a facilitated recovery from current suppression is a unique pharmacological feature of these new Kv7 channel modulators identified in this study, which also supports our presumed mechanism, stabilizing the functional Kv7-PIP2 conformation (Alhassen et al., 2023).

Alternatively, facilitated recovery may result from other mechanisms such as interference of phospholipid metabolism. To evaluate this possibility, a fluorescent probe that binds to PIP2 and IP3, CFP-PH (Hirose et al., 1999) was used. m1 muscarinic receptor in CHO hm1 cells was first activated by 0.3 μM oxo-M at t = 0 min, followed by 150 μM pirenzepine at t = 1min to terminate receptor activity. Membrane localization of CFP-PH location via PIP2 binding was monitored by TIRF signal as we described (Alhassen et al., 2023). In control condition, oxo-M reduced TIRF signal to 0.46 ± 0.03 (n = 9) of the reference intensity (Fig. 5A&B), corresponding to translocation of CFP-PH from the plasma membrane to the cytoplasm due to hydrolysis of PIP2. Addition of test compounds at t = −1 min did not change either TIRF signals, translocation after oxo-M, or recovery (Fig. 5B–D), suggesting these compounds do not interfere with PIP2 turnover. In support of this view, cytosolic PKC activity measured by cyt-CKAR (Gallegos et al., 2006) was not affected by these model compounds (Fig. 5E&F). These results suggest that these model compounds do not interfere with PIP2 metabolism.

Figure 5. Effects of model compounds on PIP2 turnover in CHOhm1 cells.

A) Live cell monitoring of PIP2. Membrane localization of a PIP2/IP3 probe, CFP-PH, is monitored by TIRF microscope. After addition of indicated test compounds at t = −1 min, oxo-M (final 0.3 μM) was add at t = 0 to activate PLC, followed by pirenzepine (100 μM) at t = 1 min to terminate m1 muscarinic receptor activity to measure recovery of PIP2. Data from 2 replicates. B) Quantitative summary of A at 1 min showing model compounds did not interfere with PLC activity. C, D) Quantitative summary of A at 2 min (C) and 3 min (D) showing model compounds did not interfere with recovery of PIP2. E) Live cell monitoring of protein kinase C activity by cyt-CKAR. Data from 2–3 replicates. F) Quantitative summary of E. ns, not significant by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars show SEM.

Another hypothesis is that binding of these compounds to the presumed binding pocket preserves Kv7 channel at ion-conducting conformation. To evaluate this hypothesis, we measured Kv7.2 current in PIP2 depleting condition using eVSP, an engineered voltage-sensitive phosphatase that catalyzes PIP2 dephosphorylation at very positive potentials (Kawanabe et al., 2020) (Fig. 6). Co-expression of eVSP with Kv7.2 channel induced depolarization-induced current decay during 10-s pulses to +60 mV, which was not observed when Kv7.2 channel is expressed alone (Fig. 6A&B), supporting PIP2 depletion during depolarization steps. Such current decays were consistent when pulses were applied at 1-min interval (Fig. 6B). Test compounds were applied after control responses were recorded. Application of these compounds diminished current decay (Fig. 6C–E). Given that these compounds do not interfere with PIP2 metabolism (Fig. 5), these results suggest that the compounds stabilize an ion-conducting conformation of the Kv7.2 channel under low PIP2 conditions.

Figure 6. Effects of model compounds on Kv7.2 current at PIP2 depleting condition.

A) Pooled scaled Kv7.2 currents during 10-s pulse to +60 mV. Quantitative summary of pulse-end currents of individual cells are shown in the right. Linked Xs are from same cells. B) Kv7.2 currents during eVSP-induced PIP2 depletion. C-E) eVSP-induced Kv7.2 current decays before and after addition of 10μM dabigatran etexilate (dabE , C, n = 11), 10 μM carvedilol (car, D, n = 11), or 1μM nebivolol (neb, E, n = 11). Data were obtained from 4 replicates. Error bars show SEM. ns, not significant, * < 0.05, ** < p = 0.01 by paired t-test.

3.3. Insight into the drug binding site and binding poses.

Next question was whether the presumed binding site is indeed the binding site for these three compounds. The presumed binding site consists of two distinct segments from the same Kv7 subunit, one from an adjacent Kv7 subunit, and one segment from calmodulin (CaM) (Figure 7A&B). Interestingly, the RQKH motif at the boundary between S6 and helix A, which acts as a hinge that remains extended in the presence of PIP2 but bends upon PIP2 loss (Sun & MacKinnon, 2020), is part of the lining segments (Figure 7B). Since we used Autodock vina for our screening, we examined predicted binding poses in this program. However, predicted binding poses by Autodock Vina fluctuate depending on small differences in initial conditions and were not reliable. Thus, we run short trajectory molecular dynamic (STMD) analysis to estimate thermodynamically stable binding poses (Fig. 7C). Analyses showed, all model compounds demonstrated stable binding to the target presumed binding pocket in Kv7.1 structure (Fig. 7C). In addition, all model compounds as well as DHEAS, a prototypical compound for this class, bridges arginine residue (R360) in the RQKH motif and the C-terminal tail of Kv7.1 or calmodulin. To further validate these binding poses, we made mutations to corresponding to Kv7.1(H363) in the RQHK and Kv7.1(R555) in the helix C, namely rat Kv7.2, rKv7.2(H328C) and rKv7.2(R564A) respectively. Unfortunately, these mutants did not form functional channels. Instead, we analyzed mutant Kv7.2 and CaM within the presumed binding site that show functional channels. We tested rKv7.2(H558C), corresponding to Kv7.1(H549), which we identified for critical for DHEAS (Alhassen et al., 2023), along with two mutations of CaM, CaM(E88K), and CaM(D94K) each combined with wild-type rKv7.2 (Figure 5B–E) (Kang et al., 2014). In control conditions, all three mutant conditions exhibited a similar degree of 0.3 μM oxo-M–induced current suppression as the wild-type Kv7.2 channel (Figure 5C–E). rKv7.2(H558C) canceled the effects of dabigatran etexilate and DHEAS but not carvedilol or nebivolol (Figure 5C). rKv7.2(wt) + CaM(E88K) canceled dabigatran etexilate and nebivolol, but not carvedilol and DHEAS (Figure 5D). rKv7.2(wt) + CaM(D94K) canceled dabigatran etexilate, carvedilol, and DHEAS, but not nebivolol (Figure 5E). Binomial distribution of effects in CaM mutants is likely due to degree of overexpression of mutant CaM relative to endogenous CaM. These distinct compound-dependent cancellation profiles in each mutant combination suggest different modes of interaction depending on compounds.

Figure 7. Subunit composition of the presumed binding pocket, predicted binding poses, and effects of mutation on drug effects of model compounds.

A) Subunit composition of the cytosolic drug-binding pocket, illustrated with color-coded structures. Lower panel shows the pocket from the same orientation with color codes as the ribbon model. B) Kv7 subtype alignments of each lining segment. Color-coded boxes correspond to those linings shown in A. C) Binding poses of indicated compounds (upper) and interacting residues (lower) predicted by STMD analysis in Kv7.1 structure (pdb: 6v01). D) Summary of 0.3 μM oxo-M-induced peak current suppression of rKv7.2(H558C) currents. E) Summary of oxo-M-induced peak current suppression of rKv7.2(wt) + CaM(E88K). F) Summary of oxo-M-induced peak current suppression of rKv7.2(wt) + CaM(D94K). * < p = 0.05, ** < p = 0.01, *** < p = 0.001 by one-way ANOVA followed by Dunnett’s multiple comparisons test. Error bars represent SEM.

We conducted Autodock Vina analyses on an available Kv7.2 structure (Ma et al., 2023). However, the reported structure (PDB: 8j01) did not show state-dependent high affinity binding nor binding to the target binding site in AutoDock Vina (supplementary File 1) due to disruption of binding site by a loop connecting Helix B–Helix C (supplementary figure 3).

3.4. Neuroprotective effects by identified Kv7 channel modulators

Our previous study showed Kv7-mediated neuroprotection after pilocarpine-induced status epilepticus (Greene et al., 2018). Thus, we next tested whether our new Kv7 channel modifiers have neuroprotective effects. Primary cultured cortical neurons were prepared from newborn mice and cell viability was evaluated by the MTT assay. Glutamate toxicity was induced by 30-min exposure to 50 μM glutamate. Control wild-type neurons showed significant reduction after glutamate treatment in MTT signal (Figure 8A). Either 20 μM DHEAS, 20 μM dabigatran etexilate, 20 μM carvedilol or 2 μM nebivolol were added 10 min prior to glutamate exposure, which significantly improved survival of neurons (Figure 8A). Interestingly, cortical neurons prepared from Kv7.2(S559A) mice, which exhibit reduced M-current suppression due to disruption of the PKC pathway (Greene et al., 2018), showed significant resistance to glutamate-induced neurotoxicity. Detectable glutamate toxicity was observed when glutamate concentration was increased to 200 μM and exposure to 1 h (relative MTT value 0.82 ± 0.01, n = 20, Figure 8B). In Kv7.2(S559A) neurons, DHEAS, dabigatran etexilate, and carvedilol did not show further neuroprotection (Figure 8B) suggesting occlusion of shared pathway. However, nebivolol showed additional neuroprotection in these neurons suggesting contribution of other mechanisms. To further confirm involvement of Kv7 channels in neuroprotection, we tested a Kv7 channel selective inhibitor XE991 in wild-type neurons. XE991 reduced neuroprotective effects of DHEAS and dabigatran etexilate (Figure 8C). We then tested protective effects induced by 200 μM hydrogen peroxide. DHEAS, dabigatran etexilate, and nebivolol increased survival of neurons demonstrating Kv7-mediated neuroprotection is valid for oxidative stress (Figure 8D).

Figure 8. In vitro neuroprotection in primary cortical neurons and pilocarpine-induced seizures.

A) Primary cortical neurons cultured from wild-type mice were treated with 50 μM glutamate for 30 minutes and MTT assays were conducted at the following day. The following test compounds were applied both before and after glutamate washout: 20 μM DHEAS, 20 μM dabigatran etexilate (dabE), 20 μM carvedilol (car), and 2 μM nebivolol (neb). Concentration of test compounds were same throughout this figure. X represents an individual well and sample numbers are indicated. B) Primary cortical neurons from Kv7.2(S559A) mice were treated with 200 μM glutamate to induce neuronal damage. Test compounds did not further promote neural survival except nebivolol. C Reduction of neuroprotective effects of 20 μM DHEAS and 20 μM dabigatran etexilate by 10 μM XE991. D) Neuroprotective effects of indicated compounds against oxidative stress induced by 200 μM hydrogen peroxide. E) Delayed neuroprotection. Application of test compounds after 30-min glutamate challenge (50 μM) showed equivalent protective effects compared to pre-treatment shown in A. F) Anti-seizure effects tested in pilocarpine-induced seizures. The right panel show quantification of area under the curve of each animal shown in the left panel. Number of animals is also indicated. ns, not significant, **** < p = 0.00001 by one-way ANOVA followed by Dunnett’s multiple comparisons test. G) Damaged neurons assessed by Fluoro-Jade C staining 48 h after pilocarpine seizures. Test drugs were administered 2 h after pilocarpine injection. * < p = 0.05 by Kruskal-Wallis test followed by multiple comparisons test. Error bars represent SEM

Our previous study suggested a potential for post-event or delayed therapy (Greene et al., 2018). To examine this, test compounds were added after the 30-min glutamate treatment and cell viability was measured in a following day. All four test compounds promoted survival of neurons to a similar degree to pretreatments (Figure 8A&E).

3.5. Anti-seizure effects of identified Kv7 channel modulators

Our previous studies using a genetic model for attenuation of M-current, Kv7.2(S559A) mice, identified anti-seizure phenotype, neuroprotection, and analgesic effects (Alhassen et al., 2023; Greene et al., 2018). Therefore, we examined these therapeutic outcomes for the model compounds in vivo.

Chemoconvulsant-induced seizures were induced by pilocarpine (Greene et al., 2018). Twenty minutes after induction of seizures, either PBS control, dabigatran, or carvedilol was injected. Cumulative Racine scores during the observation window (0 – 2 h) were used to evaluate anti-seizure effects (Figure 6F). Interestingly, only dabigatran etexilate showed mitigated behavioral seizures. Anti-seizure effects of dabigatran etexilate was negated by co-administration of XE991. We also examined whether delayed administration of model compounds can prevent pilocarpine SE-induced neuronal death at the hippocampus. We injected either PBS, dabigatran etexilate, nebivolol, or carvedilol 2 h after pilocarpine injection. FJC staining was used to label damaged neurons. FJC positive cells in the CA1 region of the hippocampus were significantly lower in dabigatran etexilate-injected mice, but not in carvedilol or nebivolol injected mice. An efficacy similar to anti-seizure effects.

3.6. Analgesic effects of identified Kv7 modulators

Lastly, analgesic effects of model compounds were examined by the formalin paw test. For this test, PBS or test compounds were injected into mice 10 min prior to formalin injection. Treated mice showed unaffected early-phase pain responses, but milder late-phase pain responses for all three compounds, a profile similar to DHEAS, or Kv7.2(S559A) mice (Alhassen et al., 2023) (Figure 9). In addition, these effects were negated by co-administration of XE991.

Figure 9. Analgesic effects of model compounds in the formalin paw test.

Left panel show quantitative\summary of the early-phase pain responses. Right panel show the late-phase pain responses. In indicated groups, XE991 was co-administered with test compounds. ns = no significance, ** < p = 0.01, *** < p = 0.001 by one-way ANOVA followed by Dunnett’s multiple comparisons tests. Each symbol represents individual animal. Error bars show SEM.

4. Discussion

This study tested the hypothesis that small molecules can modify the responsiveness of Kv7 current suppression induced by Gq-coupled receptors. We identified several clinically used compounds that act as off-target Kv7 channel modulators, attenuating current suppression. Our findings established a novel class of Kv7 channel modulators, which can mitigate seizures, exhibit neuroprotection, and analgesia, underscoring the therapeutic value of this mechanism in reducing pathological neuronal firing.

Our two-step in silico screening strategy resulted in improved functional verification success rates, reaching around 50%. This resulted in identifying eight functionally verified compounds out of 3,447 candidates. Identified compounds had a wide range of molecular diversity, while showing strict chemical requirement within the same chemical backbones. For instance, only DHEAS, but not related steroid hormones, exhibits Kv7 current modulation (Alhassen et al., 2023). Similarly, dabigatran etexilate but not bare dabigatran attenuated Kv7.2 current suppression (Figure 1). Further analyses are required to determine chemical requirements essential for reducing responsiveness of receptor-induced Kv7 current suppression.

During electrophysiological experiments, we observed that some compounds inhibited Kv7.2 currents at higher concentrations, consistent with direct binding to Kv7.2 channels. The positive cooperativity observed in the dose–inhibition curve for nebivolol is consistent with the presence of multiple binding sites, in line with four subunits per channel. Such inhibitory effects may contribute to the adverse effects of these compounds. Moreover, these observations suggest that this intracellular space is functionally coupled to ion conduction, as occupancy of the site by multiple molecules can inhibit ion conduction.

Known Kv7 channel modulators such as retigabine shift the half activation voltage of Kv7.2 channels (Wuttke, Seebohm, Bail, Maljevic, & Lerche, 2005). Similarly, increased intracellular PIP2 induces a left-shift of the activation voltage (Kim, Duignan, Hawryluk, Soh, & Tzingounis, 2016). In contrast, the compounds identified in this study are distinct in that they did not alter the activation voltage. Because enhanced PIP2 binding is associated with shifts in activation voltage, absence of such shifts suggests that pharmacological effects of these compounds arise from a different mechanism. We believe that stabilization of a functional conformation of Kv7 channel, rather than modulation of Kv7-PIP2 binding affinity, is a likely mechanism.

In support of this hypothesis, compound-treated Kv7.2 channel showed resistance to PIP2 depletion induced by eVSP (Fig. 6), while they did not interfere with PIP2 turnovers (Fig. 5). In addition, STMD analysis used to estimate compound-specific interaction profiles revealed that all model compounds interact with residues in the RQKH motif in extended S6 helix as well as residues in the C-terminal tail. Seemingly validated compounds bridge the extended S6 helix to proximate to the Helix B-C loop. However, since STMD analysis could not fully explain results from our mutation analyses, likely due to mutation-induced altered peptide backbone coordination that are not captured by STMD, further studies are required to determine precise mechanism for stabilization of the ion-conducting Kv7.2 conformation.

Since Kv7.2 is the key subtype of the M-channels, it is reasonable to use available Kv7.2 structure. However, we could not detect high affinity bindings of our compounds to available Kv7.2 (pdb: 8j01) (Ma et al., 2023) or in Kv7.4 (pdb: 7vnp) (Alhassen et al., 2023; Zheng et al., 2022) structures in Autodock Vina. This difference appears to depend on whether the loop connecting helices B and C forms a stable conformation in Kv7.2 and Kv7.4, in contrast to its non-structured state in Kv7.1 (Supplemenetary Figure 3). This disrupting loop at the putative drug binding may be a reason why Autodock Vina could not detect high affinity bindings in Kv7.2 and Kv7.4 structures.

In the present study, we demonstrated neuroprotective effects of our model compounds in vitro assays. In addition, dabigatran etexilate showed delayed neuroprotection after pilocarpine-SE. Together, these findings suggest that modulation of M-current suppression may be utilized as a post-event therapeutic strategy for neural insults. Notably, all three model compounds have previously been reported to exert neuroprotective effects through other mechanisms. Dabigatran has been shown to confer neuroprotective effects by inhibiting thrombin-induced inflammatory responses (Kandil, Sayed, Ahmed, Abd El Fattah, & El-Sayeh, 2018). Carvedilol has been reported to reduce ischemic brain injury (Savitz et al., 2000; Wong & Escayg, 2024), and nebivolol shows neuroprotection through nitric oxide dependent pathways (Heeba & El-Hanafy, 2012). While these studies suggest that our model compounds can penetrate the blood brain barrier, the present study further provides evidence for an additional, Kv7-dependent mechanism underlying their neuroprotective effects.

That said, pilocarpine-induced seizures or subsequent neuronal damages were not prevented by carvedilol or nebivolol. Since pilocarpine-induced seizures involve overexcitation of both sympathetic and parasympathetic nervous system (Devinsky, 2004; Moseley, Bateman, Millichap, Wirrell, & Panayiotopoulos, 2013), beta-adrenergic blockade during seizures may lead to fatal imbalance of autonomic regulation.

One of our initial hopes was a possibility of repurposing existing drugs as Kv7 channel modulators. However, repurposing dabigatran etexilate as a Kv7 channel modulator would be unsafe in human patients because it is rapidly converted to bare dabigatran in the liver and can cause iatrogenic hemophilia. In addition, repurposed beta blockers did not demonstrate anti-seizure or neuroprotective effects in vivo. Therefore, the development of specific compounds targeting Kv7 channels will be essential to safely and effectively harness this mechanism for therapeutic use in human.

In summary, we have identified a novel class of Kv7 channel modulators with promising therapeutic potential. These compounds attenuate M-current suppression and offer a new pharmacological strategy for conditions such as epilepsy, neurodegeneration, and chronic pain. Further development of selective molecules targeting this mechanism could address significant unmet clinical needs and broaden the clinical utility of Kv7 channel modulation.

Bullet point summary.

What is already known?

• Kv7/M-current suppression by activation of Gq-coupled receptors increase neuronal firing.

• An endogenous steroid hormone, DHEAS, attenuates Kv7/M-current suppression.

What does this study add?

• Diverse group of small molecules can function as Kv7 channel modulators attenuating receptor-induced Kv7/M-current suppression.

• Attenuation of receptor-induced Kv/M-current suppression by these small compounds show neuroprotective, anti-seizure, and analgesic effects.

What is the clinical significance?

• Attenuation of receptor-induced M-current suppression by small compounds provides a novel way to control pathological overexcitation of neurons.

Abbreviations:

CaM

calmodulin

CFP

cyan fluorescent protein

CHO

Chinese Hamster Ovary

CHO hm1

Chinese Hamster Ovary cells stably expressing the human muscarinic m1 receptor

DHEAS

dehydroepiandrosterone sulfate

DIV

days in vitro

DPBS

Dulbecco’s phosphate buffered saline

eVSP

enhanced voltage-sensing phosphatase

FAD

flavin adenine dinucleotide

FJC

Fluoro-Jade C

FRET

fluorescent resonance energy transfer

MMPBSA

molecular mechanics/poisson–boltzmann surface area

MTT

3-[4,5-dimethylthiazol-2-yl]-2,5 diphenyl tetrazolium bromide

oxo-M

oxotremorine-M

PIP2

phosphatidylinositol 4,5-bisphosphate

rKv7.2

rat Kv7.2

SCG

superior cervical ganglion

STMD

short trajectory molecular dynamic

TIRF

total internal reflection fluorescence

V1/2

half-activation voltage

Data availability statement

The data that support the study findings are available upon reasonable request.