2.1 Electrophysiology

Pipettes (1–6 MΩ) were filled with a solution of KCl (145 mM), MgCl₂ (1 mM), HEPES (10 mM), and EDTA (5 mM). The pH was adjusted to 7.2 with NaOH. Electrophysiological measurements were obtained using an AXON Multiclamp 700 B patch-clamp amplifier. The bath solution contained NaCl (144 mM), KCl (5 mM), MgCl₂ (1 mM), CaCl₂ (2 mM), and d-glucose (10 mM), with the pH adjusted to 7.4 using NaOH. The protocol involved an increase in voltage from −120 mV to 40 mV in increments of 10 mV to evoke a series of K_V7 currents (1000 ms per pulse) in the bath solution. For details on cell culture, please see Appendix S1 2.2.1.

2.2 Animals

Referencing prior studies that assessed the efficacy of epilepsy treatments, male rats or mice were consistently chosen for experiments [28, 29]. The animals used in this study were housed individually in separate cages and maintained under a 12-h light/dark cycle at an ambient temperature of 22°C–25°C. Prior to the commencement of the study, the rats were acclimated for 7 days in the animal facility to ensure adaptation to the environment. All procedures related to feeding and experimentation strictly adhered to the guidelines set forth by the Animal Welfare and Ethics Committee of the Research Center for Safety Evaluation of New Drugs, Hebei Medical University.

2.3 Electrophysiological Recordings in mPFC and Hippocampal Slices

The experimental procedures involving Sprague–Dawley (SD) rats aged 8 weeks adhered strictly to the guidelines set forth by the Institutional Animal Care and Use Committee. Animals were anesthetized using isoflurane, followed by decapitation to obtain rat brains promptly. The brains were then immersed in a cold solution maintained at 4°C, containing sucrose (220 mM), KCl (2.5 mM), CaCl₂ (2 mM), MgCl₂ (4 mM), NaH₂PO₄ (1.25 mM), NaHCO₃ (26 mM), and glucose (10 mM) at a pH range of 7.2–7.3. This solution provided optimal conditions for brain tissue preservation. Incubating isolated brain slices in Mg²⁺-free cell solution is an in vitro method of replicating the abnormal neuronal firing associated with epilepsy. Under physiological conditions, Mg²⁺ blocks NMDA receptors and enhances the activity of gamma-aminobutyric acid (GABA). Therefore, removing Mg²⁺ increases neuronal excitability [30].

For recording action potentials and resting membrane potentials, an internal solution was formulated with the following composition (in mM): K-gluconate 140, NaCl 5, CaCl₂ 1, MgCl₂ 2, EGTA 11, HEPES 11, Mg-ATP 2, and Na₂-GTP 0.5. Its pH was adjusted to 7.2–7.3. The electrode internal solution for recording sEPSCs and sIPSCs comprised (in mM): Cs-gluconate 117, CsCl 11, HEPES 10, EGTA 11, CaCl₂ 1, MgCl₂ 1, Mg-ATP 1, Na₂-GTP 4, and QX-314 2; pH was adjusted to 7.2–7.3. Stimulation was applied in 20 pA increments every 500 ms, starting from 0 to 180 pA.

M current protocol: The amplitudes of M-currents were measured as deactivating currents during 1000-ms test pulses to −60 mV, starting from a holding potential of −30 mV. In current-clamp experiments, the membrane potential was set to −70 mV. Testing was conducted after continuous perfusion for 5 min during the application of QO-83, RTG, and XE991. During sIPSC recordings, the AMPA receptor antagonist CNQX (20 μM) and the NMDA receptor antagonist D-AP5 (50 μM) were added to the freshly prepared Mg²⁺-free ACSF solution. For sEPSC recordings, the GABA_A receptor antagonist bicuculline (50 μM) dissolved in dimethyl sulfoxide (DMSO) was added to the Mg²⁺-free ACSF solution.

2.4 QO-83 Against Pilocarpine-Induced Status Epilepticus (SE) and Awake Electroencephalogram (EEG) Recordings

2.5 Effects of QO-83 on Acute and Chronic Epilepsy

2.6 Molecular Dynamics Simulation

The protein structure of K_V7.2 was obtained from the Protein Data Bank (https://www.rcsb.org/) using the PDB code 7CR2. Molecular docking was performed using the CDOCKER module in the DS 2020 software. Initial conformations of the QO-83-K_V7.2 and RTG-K_V7.2 complexes were derived based on docking scores and conformational comparisons. The parameters for the K_V7.2 protein were assigned using the Amber ff14SB force field, while the parameters for the ligand small molecules were assigned using the Amber GAFF force field. Partial charges for each atom of the ligand small molecules were initially calculated using Gaussian 09 at the B3LYP/6-31G* level of theory and further refined using the RESP charge fitting program in Amber.

Molecular dynamics simulations were conducted using the sander program in Amber20. The simulated systems were optimized in a vacuum and placed in a TIP3P water box model. Periodic boundary conditions were applied during the simulation analysis. To maintain the system's neutrality, sodium or chloride ions were randomly added to the simulated systems. During the simulation process, the solute was initially restrained, and the system was optimized using 5000 steps of steepest descent minimization followed by 5000 steps of conjugate gradient minimization. Subsequently, the selected systems underwent 100 ns of molecular dynamics simulation in the NPT ensemble, with a simulation time step of 2 fs. The SHAKE algorithm was employed to constrain the bond lengths involving hydrogen atoms, and the Particle Mesh Ewald (PME) method was used to calculate long-range electrostatic interactions. Finally, the root mean square deviation (RMSD) and binding/dissociation-free energies of each system were calculated, and representative conformations were extracted.

2.7 Statistical Analysis

Normality tests on the data were conducted using nonparametric analyses for those that do not follow a normal distribution. Statistical differences between multiple groups (mean ± SEM) were compared using one- or two-way ANOVA. Electrophysiological data were analyzed using SPSS version 21 and OriginPro (version 9.1.0). P Peak inward currents obtained from activation protocols were calculated using the following equation: G = I/(Vm-EK), where G is conductance, I is peak inward current, Vm is instantaneous membrane potential, and EK is equilibrium potential for the sodium channel. Conductance data were normalized using maximum conductance and further fitted to the Boltzmann equation: $$$ \mathrm{G}={\mathrm{G}}_{\mathrm{m}\mathrm{ax}}+\left({\mathrm{G}}_{\mathrm{m}\mathrm{in}}-{\mathrm{G}}_{\mathrm{m}\mathrm{ax}}\right)/\left(1+\exp \left[\left({\mathrm{V}}_{\mathrm{m}}-{\mathrm{V}}_{1/2}\right)/\mathrm{k}\right]\right) $$$, where G is conductance, $$$ {\mathrm{V}}_{1/2} $$$ is midpoint of activation, and k is slope factor. The “Spectrum” module of LabChart was used for data analysis of EEG signals, including the calculation of power density and averaging power spectral density (PSD).

3.1 QO-83 Influences the Opening Activity of K_V7.2-K_V7.5 Channels In Vitro

Details of the NMR spectra of synthesized QO-83 are presented in Figures S1 and S2. Using Rb⁺ jet high-throughput screening, we demonstrated that QO-83 had a strong ability to open K_V7.2/7.3 channels (EC₅₀ = 0.08 ± 0.04 μM), although it showed poor selectivity for K_V7.4 (EC₅₀ = 0.84 ± 0.27 μM) (Figure S3). Importantly, QO-83 did not trigger the opening of K_V7.1 channels (Figure S3).

As a further comparison of QO-83 apparent affinity for each channel subtype, we used the manual patch clamp technique to obtain EC₅₀ for opening K_V7.2, K_V7.2/7.3, K_V7.3, K_V7.4, and K_V7.5 channels. Respectively, the results were 0.56 ± 0.09 μM (Figure 1A), 0.07 ± 0.01 μM (Figure 1E), 0.22 ± 0.04 μM (Figure 1I), 0.67 ± 0.03 μM (Figure 1M), and 0.59 ± 0.08 μM (Figure 1Q). These values accordingly indicate that, compared with other K_V7 subtypes, QO-83 shows a pronounced selectivity for K_V7.2/7.3 channels.

We subsequently investigated the effects of QO-83 on the voltage-dependent activation of K_V7 channel currents. Cells were perfused with different QO-83 concentrations (0.01 μM, 0.05 μM, 0.1 μM, 0.5 μM, 1 μM, 5 μM, and 10 μM) for the recording. Corresponding EC₅₀ values of ΔV_1/2 were fitted as 0.25 ± 0.23 μM (K_V7.2, Figure 1B), 1.00 ± 0.10 μM (K_V7.2/7.3, Figure 1F), 0.82 ± 0.10 μM (K_V7.3, Figure 1J), 0.52 ± 0.27 μM (K_V7.4, Figure 1N), and 0.74 ± 0.18 μM (K_V7.5, Figure 1R). QO-83 shifts the V_1/2 of K_V7.2–K_V7.5 channels in the hyperpolarizing direction. At a concentration of 10 μM, QO-83 shifts the activation V1/2 of K_V7.2, K_V7.2/7.3, K_V7.3, K_V7.4, and K_V7.5 channels by −26.48 mV, −27.79 mV, −37.67 mV, −10.01 mV, and −26.28 mV, respectively (Table S1). QO-83 induces a concentration-dependent shift of V_1/2 in K_V7.2–K_V7.5 channels toward hyperpolarization, thereby promoting channel activation.

We have compared the tail current activation curves of 10 μM RTG and 10 μM QO-83. The normalized current activation curve after drug perfusion was subtracted from that before drug perfusion. We observed that QO-83 caused a greater hyperpolarization shift in the activation curve of K_V7.2 than RTG (Figure 1B,C). Additionally, QO-83 induced a greater increase in the tail current of K_V7.2 than RTG (Figure 1D, Table S2). Shifts in the activation curves for K_V7.3, K_V7.2/7.3, and K_V7.5 were not significantly different between QO-83 and RTG, although V_1/2 was more hyperpolarized under the former than the latter (Figure 1). In contrast, QO-83 did not trigger greater activity than RTG in K_V7.4 (Figure 1O,P). When stimulated with 10 mV, QO-83 and RTG have no effect on the activation time (τ) of K_V7.2–K_V7.5. At −120 mV, QO-83 and RTG both significantly extended the deactivation time (τ) of K_V7.2–K_V7.5 (Table S2). In summary, compared to RTG, QO-83 exhibited greater selectivity and efficacy in opening K_V7.2 and K_V7.2/7.3 channels.

3.2 QO-83 Modulates Neuronal Excitability and sIPSC/sEPSC Currents in Adult SD Rats In Vitro

Mg²⁺ blocks NMDA receptors and also enhances the activity of gamma-aminobutyric acid (GABA). Therefore, removing Mg²⁺ from the medium increases neuronal excitability, which is a common method to observe abnormal electrical activity in neurons on brain slices [30]. At a concentration of 10 μM, both QO-83 and RTG significantly inhibited AP firing (Figure 2B,C). However, at 1 μM, while QO-83 still significantly reduced AP firing in mPFC and hippocampus pyramidal neurons, RTG had no significant effect (Figure 2A,B). Overall, QO-83 demonstrated a stronger inhibitory effect on neuronal firing than RTG at lower concentrations.

Under the stimulation of 180 pA current, QO-83 significantly increased the rheobase and threshold of mPFC and hippocampal pyramidal neurons, while significantly decreasing the RMP of these neurons (Figure 2D–K). Neuronal transmission depends on the frequency, mode, and timing of the pulse output, with after-hyperpolarization (AHP) shaping all of these variables. QO-83 significantly shifted the AHP of mPFC and hippocampal pyramidal neurons toward depolarization (as shown in Figure 2E,I). At 10 μM, RTG had the same effect as QO-83, but at 1 μM, RTG had less impact on threshold and RMP than QO-83. Interestingly, QO-83 and RTG exhibited a stronger inhibitory effect on the excitability of hippocampal pyramidal neurons than on cortical pyramidal neurons, an outcome likely related to the high K_V7.2 expression in the hippocampus. In conclusion, QO-83 exerts membrane-stabilizing effects by influencing the threshold, rheobase, resting membrane potential, and AHP.

Our results indicated that QO-83 regulates ectopic discharge of hippocampal pyramidal neurons. In the context of neurological disorders, such as epilepsy, hippocampal neurons often exhibit abnormal excitability and a relative imbalance between excitatory and inhibitory neurotransmitters. Comparing the sEPSC and sIPSC provides a quantitative measure of the synaptic input balance, offering insights into the active inputs to these neurons. Without affecting sEPSC frequency, QO-83 significantly inhibited sEPSC amplitude from 74.06 ± 8.08 pA to 23.20 ± 2.11 pA in the Mg²⁺-free group (Figure 2L–P). Additionally, QO-83 significantly reversed the effect of Mg²⁺-free conditions, increasing sIPSC amplitude (with no effects on frequency) from 30.08 ± 3.02 pA to 81.15 ± 7.57 pA (Figure 2Q–U). In contrast, 1 μM RTG had no effect on either sEPSC or sIPSC.

3.3 QO-83 Significantly Reverses Pilocarpine-Induced Epileptic Status and Reduces Epileptoid Electroencephalogram Activity in Adult SD Rats

KCNQ channel openers have been evaluated as potential treatments for status epilepticus (SE), the most extreme form of epileptic seizure that can result in neuronal damage and dysfunction [31]. EEG was recorded in rats with systemic injection of pilocarpine to induce epileptic states (PILO-SE). Recording brain activity while observing epileptic behavior revealed that QO-83 significantly prolonged the latency of seizures from 21.64 ± 8.89 min to 56.79 ± 26.29 min (Figure 3A). QO-83 reduced the seizure grade from 4.70 ± 0.12 to 2.33 ± 0.24 (Figure 3B). 1 mg/kg RTG could not rescue the epileptic behavior in mice, but 10 mg/kg RTG significantly prolonged the latency of seizures and reduced the seizure grade in the PILO model mice.

Pilocarpine-induced increase in EEG activity involved the entire spectrum of frequencies analyzed (0.5–50 Hz), particularly in the higher range, such as the alpha (8–12 Hz) and beta (12–30 Hz) bands (Figure 3C–G). In pilocarpine-treated rats, 1 mg/kg QO-83 and 10 mg/kg RTG significantly inhibited the increase in EEG power spectral density until it returned to baseline across the entire frequency range (Figure 3H). The administration of QO-83 resulted in a significant inhibition of PILO-induced power enhancement in δ, θ, α, and β waves (Figure 3I–M). Notably, at 1 mg/kg, RTG did not exhibit effects equivalent to QO-83 in either behavioral or epileptic EEG aspects. Higher RTG doses were required to achieve therapeutic efficacy. In summary, QO-83 significantly suppressed epileptiform EEG activity, thereby preventing seizure and mitigating seizure behavior in pilot model rats.

3.4 Effects of the QO-83 and XE991 Combination on KV7 Current In Vitro and Seizure Activity in PILO Model SD Rats

Based on the strong inhibitory effects of QO-83 on neuronal excitability and epileptiform EEG activity, we investigate whether QO-83 exerts its regulatory effects through the activation of K_V7 channels. We examined the in vivo and in vitro activity of QO-83 in combination with XE991. It was found that QO-83 significantly increased the K_V7.2 channel current (Figure 4A). The application of 20 μM XE991 significantly inhibited the increase in K_V7.2 channel current induced by QO-83 (Figure 4B). However, the co-application of 20 μM XE991 + 1 μM QO-83 did not further increase the K_V7.2 channel current (Figure 4A,B). This indicates that XE991, by binding to the K_V7.2 channel, prevents QO-83 from interacting with the K_V7.2 channel, thus blocking its activation effect. We further observed that QO-83 significantly opened the M-current in mPFC neurons (Figure 4C,D). After the perfusion of XE991, 20 μM XE991 + 1 μM QO-83 failed to further enhance the M-current, suggesting that XE991 blocks the interaction of QO-83 with the M-channel. These results indicate that the compound QO-83 can significantly increase the K_V7.2 channel and M-current density, likely exerting its inhibitory effect on neuronal excitability through the activation of the M-channel.

The administration of 2 mg/kg XE991 significantly reduced the seizure latency from 22.73 ± 6.03 min to 8.52 ± 4.73 min. However, co-administration of 2 mg/kg XE991 with 1 mg/kg QO-83 resulted in a seizure latency of 14.01 ± 7.10 min, which did not significantly prolong the latency (Figure 4E,F). No significant differences in seizure grades were seen between the model group, 2 mg/kg XE991, and 2 mg/kg XE991 + 1 mg/kg QO-83 (Figure 4G). It is worth noting that, compared to the PILO group, 2 mg/kg XE991 significantly increased the power density within the 0–50 Hz frequency range in SD rats, as well as the power density in the δ, θ, α, and β bands (Figure 4H–P). Interestingly, when 2 mg/kg XE991 was combined with 1 mg/kg QO-83, although there was a reduction in power density compared to 2 mg/kg XE991 alone, there was no statistically significant difference. These results suggest that the preferential administration of 2 mg/kg XE991 likely blocks K_V7 channels, thereby hindering the anticonvulsant activity of QO-83.

3.5 QO-83 Effectively Alleviated Seizures in Both Acute Epilepsy Model Mice and Chronic Kindling Model Rats

The maximal electroshock seizure (MES) model is widely used to screen anticonvulsant drugs. We tested QO-83 at doses of 0, 1, 1.5, 2, 2.5, and 3 mg/kg; these treatments yielded electroshock protection rates of 0%, 10%, 50%, 60%, 80%, and 100% (Figure 5A). The anticonvulsant effect of QO-83 was dose-dependent, with an ED₅₀ of 2.27 ± 1.29 mg/kg (Figure 5B). In contrast, treatment with 15 mg/kg of RTG led to 90% protection.

In the pentylenetetrazol maximal seizure (MMS) test, 85 mg/kg PTZ was subcutaneously injected into mice, inducing myoclonic jerks and generalized tonic seizures. QO-83 at 1.0 mg/kg and 3.0 mg/kg significantly prolonged seizure latency from 6.92 ± 1.67 min to 14.45 ± 2.54 min and 19.43 ± 2.42 min, respectively. Under the same concentrations, RTG showed a similar effect on prolonging seizure latency (Figure 5C). Seizure protection rates of QO-83 at 1 and 3 mg/kg were 25% and 45%, respectively, whereas 15 mg/kg RTG had a protection rate of 40% (Figure 5D).

In the penicillin(PNC)-induced acute epilepsy model of SD rats, oral QO-83 administration of 5 mg/kg and 10 mg/kg significantly prolonged seizure latency from 10.63 ± 2.39 min to 35.00 ± 4.61 min and 36.25 ± 6.29 min, respectively (Figure 5E). At a far higher concentration of 40 mg/kg, RTG also significantly increased seizure latency. Additionally, 5 mg/kg and 10 mg/kg QO-83 reduced seizure grades to 3.33 ± 0.42 and 3.12 ± 0.47, respectively, while RTG reduced it to 2.66 ± 0.33 (Figure 5F).

On the 30th day of continuous administration of 35 mg/kg PTZ, changes in seizure parameters were observed in each group of rats. QO-83 at 1 mg/kg and 10 mg/kg RTG significantly reduced seizure grades from 4.68 ± 0.62 to 2.12 ± 0.22 and 1.75 ± 0.31, respectively (Figure 5G). QO-83 and RTG can both significantly prolong the latency period in the PTZ chronic kindling model in rats (Figure 5H). Furthermore, QO-83 at 1 mg/kg significantly increased rat body weight, but RTG did not (Figure 5I).

In summary, QO-83 has demonstrated potent anti-convulsant effects in both acute and chronic epilepsy models. Notably, while RTG is administered at a dose 10 times higher than that of QO-83, the latter achieves significant therapeutic effects at a much lower dose.

3.6 Comparison of the Binding Mechanisms of QO-83 and RTG

To compare the binding modes of QO-83 and RTG in K_V7.2, molecular dynamics (MD) simulations lasting 100 ns were performed. The QO-83/K_V7.2 complex remained relatively stable throughout the simulation (Figure 6). The RMSD values for the residues at the binding site were consistently within 2 Å. Furthermore, QO-83 exhibited relatively low fluctuations compared to RTG (Figures 6A and 5B). Next, we calculated the binding free energies of both compounds using MD trajectories for the last 50 ns. QO-83 had greater target affinity with K_V7.2 (−52.95 kcal/mol) than RTG (−45.51 kcal/mol), indicating a strong interaction between QO-83 and K_V7.2 active residues.

We compared the binding conformations of QO-83 and RTG to K_V7.2. Superimposition of the initial docking poses clearly revealed a high spatial overlap between QO-83 and RTG (Figure 6C). The amount of space overlap differed between distinct QO-83 and RTG conformations (Figure 6D). Moreover, the trifluoromethyl-substituted benzene ring on QO-83 differed from the corresponding benzene ring on RTG; however, other benzene rings significantly overlapped between the two compounds. Both QO-83 and RTG formed hydrogen bonds with W236 and F305 (Figure 6E,F).

Repositioning identified residues revealed that amino acids in chains A, C, and D were the main energy contributors to binding. Residues on chain A contributed −6.01 kcal/mol and −7.65 kcal/mol of energy to QO-83 and RTG, respectively, while residues on chain D contributed −13.27 kcal/mol and −10.93 kcal/mol (Figure 6G). Notably, W236 was a significant contributor to the binding energies of both QO-83 and RTG. Chain C differed significantly in energy contributions to QO-83 and RTG (6.00 kcal/mol vs. 1.00 kcal/mol), explaining the former's stronger binding affinity to K_V7.2. The chain C residues (F104, L107, and V108) we identified offer valuable insights for future receptor-based drug design.