The BK channel-NS1619 agonist complex reveals molecular insights into allosteric activation gating

Significance

BK channels are crucial regulators of physiological processes, including muscle tone and neurotransmitter release. Understanding their activation mechanisms is essential for therapeutic development. Using cryo-EM and molecular dynamics, we identified a binding site for the BK channel agonist NS1619, demonstrating how its interaction stabilizes the open conformation. Our findings provide a structural framework for designing more effective BK channel modulators, potentially aiding in the treatment of conditions linked to channel dysfunction.

Abstract

BK channels play essential roles in a wealth of physiological functions, including regulating smooth muscle tone and neurotransmitter release. Its dysfunction, often caused by loss-of-function mutations, can lead to severe phenotypes, including ataxia and sensory impairment. Despite the therapeutic potential of BK channel agonists, the molecular mechanisms by which they stabilize the pore’s open conformation remain unclear. Using cryoelectron microscopy and molecular dynamic simulations, we identified that NS1619, a synthetic benzimidazolone agonist, first described as a BK opener, binds within a pocket formed by the S6/RCK1 linker and the S4 transmembrane segment. Our simulations suggest that agonist binding promotes a twisting motion in the S6 segment, enabling critical interactions with residues K330, K331, and F223. These findings provide a molecular model for the mechanism of NS1619 and suggest that its binding site can accommodate other agonists, highlighting a promising target for therapeutic development.

By virtue of its dual activation by intracellular Ca²⁺ and depolarizing voltages, the voltage- and Ca²⁺-activated channel (BK) is the ultimate damper of excitatory stimuli. These characteristics make them essential in numerous physiological functions. In the nervous system, BK channels shorten the action potential by colocalizing with voltage-dependent Ca²⁺ channels, damp neurotransmitter release, and their association with β2 or β4 subunits regulates neuron excitability (1, 2). In smooth muscle, Ca²⁺ sparks produced by the liberation of Ca²⁺ from the sarcoplasmic reticulum promote the activation of BK channels, facilitating muscle relaxation. Hence, they regulate the arterial smooth muscle tone (3–5). In the kidney, the BK channel´s large single-channel conductance is essential to sustain high rates of urinary K⁺ secretion (6–8). Not surprisingly, malfunctioning of BK channels in these tissues due to gain- or loss-of-function (GOF/LOF) mutations leads to a variety of pathophysiological phenotypes.

BK channels have been implicated in epilepsy, fragile X syndrome, mental retardation, autism, and chronic pain (1, 9). Analysis of 37 KCNMA1 alleles (10) shows that the distribution between patients with GOG and LOF mutations shows the same tendency to seizures. However, paroxysmal nonkinesigenic dyskinesia is predominantly observed among patients with GOF alleles of the BK channel. Other movement disorders are observed in patients with LOF variants. Dysfunction of BK channels related to the loss of the β1 subunit in smooth muscle leads to hypertension (3, 4), an overactive bladder (10, 11), and erectile dysfunction (12, 13). Deleting the pore-forming BK α subunit in the kidney produces an absence of flow-induced K⁺ secretion and hypertension (7). Deletion of β1 leads to poor handling of K⁺ loads and is linked with a rise in plasma K⁺ and hypertension. Deleting β4, on the other hand, results in insufficient K⁺ handling and high plasma K⁺, but only mild hypertension (7).

Given the large number of LOF BK mutations, identifying BK channel activators with potential therapeutic applications is an important pharmacological target. The first characterized BK channel opener was the synthetic benzimidazolone NS1619 (14, 15), a potential therapeutic agent in inflammatory pain (16), pulmonary hypertension (17), erectile dysfunction (12), bladder instability (18), and migraines (19). NS1619 acts as an antagonist in the BK LOF mutant G354S rescuing BKG354S cells, the G354S mutant was found in a child with congenital and progressive cerebellar ataxia and located in the selectivity filter (20). NS1619 appears to activate BK channels via the S6/ RCK1 linker (21). However, elucidating the nature and location of the agonist binding site is of fundamental importance in the quest to define its mechanism of action and design drugs able to activate BK selectively and with high affinity. Here, we have determined the structure of the complex BK channel-NS1619 by single particle cryo-EM. NS1619 resides in a pocket lined by amino acids from the S6/RCK1 linker and the S4 transmembrane segment. We show evidence suggesting that this arrangement allosterically stabilizes BK’s open conformation by increasing the equilibrium constant that defines the closed-open equilibrium. Furthermore, docking and molecular simulations suggest that the NS1619 binding pocket could behave as a versatile druggable site, able to accommodate other agonists of a variety of different structures.

Results

Calcium-Bound Structure Reveals the Putative NS1619-Binding Site.

We determined the structure of the human BK channel in the presence of the NS1619 activator using cryoelectron microscopy. Since the effect of the activator is independent of the saturating concentration of Ca²⁺ (21) SI Appendix, Fig. S1, we followed established purification protocols to obtain the human BK channel in the open state (22), adding NS1619 to the sample before freezing under saturating conditions (1 mM final). The structure was solved at a global resolution of 3.4 Å, which allowed detailed modeling of the BK-NS1619 complex. SI Appendix, Fig. S2 shows the diagram of the global processing and refinement strategy. An additional Coulomb density was observed and attributed to NS1619 (Fig. 1A; red density); this density is absent in previous BK structures observed (23). As anticipated from the functional results, the overall structure of the BK-NS1619 complex remains unchanged when compared with its Ca-bound form, suggesting that NS1619 acts as an open-state stabilizer and confirms the existence of a binding site for NS1619 between RCK1 and the C-linker, which can interact with amino acids from the VSD, in line with molecular docking results. We have identified a set of amino acids that putatively belong to the binding site on the BK channel (all within a 5 Å): Y332, G334, K392, and E399 (Fig. 1 B–D). As shown below, molecular dynamic simulations indicate that this picture changes with time, implying a functional reorganization of the binding site that now includes amino acid residues K330, K331, and F223.

Fig. 1.

Coulombic density of the hBK-NS1619 complex. (A) This panel displays the density map of the BK channel as obtained by cryoelectron microscopy (cryo-EM). The channel was purified using detergent in the presence of calcium, with NS1619 added just before freezing. The density attributable to NS1619 is highlighted in red. The average resolution is 3.44 Å. (B–D) Top (B), bottom (C), and side (D) views of the NS1619 density, highlighting amino acids within a 5 Å distance (Y332, K392, and E399) and the amino acids whose mutation to alanine decreases the activation effect (F223, K330, and K331).

Contribution of Binding Pocket Residues in NS1619 Coordination.

To determine the functional role of the amino acids forming the NS1619 binding site, we selected candidate residues for alanine-scanning mutagenesis using a dual approach: proximity to the ligand density in our cryo-EM structure and high-frequency interactions predicted by computational docking. This integrative analysis provided a list of potential key residues for functional testing.

Macroscopic currents were measured for the wild type (WT) and each mutant. Perfusion with NS1619 (≤30 µM) induced a leftward shift in the G-V curve of −36 ± 5 mV in WT BK (Fig. 2C). Of the tested mutants, noteworthy differences from WT were observed only in F223A, K330A, and K331A, with statistically significant shifts of −20 ± 3 mV (P = 0.014 vs. WT), −7 ± 2 mV (P = 0.014 vs. WT), and −26 ± 2 mV (P = 0.015 vs. WT), respectively, after Bonferroni correction for multiple comparisons (Fig. 2 D−F). In contrast, mutating other selected residues (R329, Y332, G334, F391, K392, and E399) resulted in a similar activation gating as that of WT-BK or subtle increases in the effect of NS1619, as shown in SI Appendix, Fig. S3.

Fig. 2.

Characterization of the NS1619 binding site on the BK channel and effects of specific mutations. (A) Membrane topology representation of an α subunit of the BK channel, highlighting regions where molecular docking identified amino acids potentially involved in the NS1619 binding site. Regions are colored according to their involvement: voltage sensor domain (VSD) in golden yellow, C-linker in green, and RCK1 in maroon red. (B) Percentage distribution of 50 different docking poses for a total of 24 frequently repeated amino acids, where 8.3% belong to the VSD, 37.5% to the C-linker, and 54.1% to the RCK1. (C) Normalized current/voltage curves for the wild type (wt) BK channel in the absence (solid line) and presence (dashed line) of 30 µM NS1619, showing the effect of the activator on channel activity. (D−F) Alanine scanning analysis for amino acids that decrease the effect of NS1619, specifically for F223A (D), K330A (E), and K331A (F), demonstrating alterations in current responses in the presence of the activator. Insets in panels (C−F) display representative macroscopic current traces, with the upper and lower sets in each panel corresponding to recordings in the presence and absence of 30 µM NS1619, respectively.

Notably, the critical role of K330 is consistent with a previous study by Gessner et al. (21), who found it to be relevant to the activation mechanism of Cym04, where mutations like K330N affected the allosteric gating (though it remains uncertain whether the S6/RCK linker itself represents a binding site for Cym04). This suggests a potential functional overlap with NS1619, a hypothesis supported in our system by the severe impact of the K330A mutation on drug sensitivity (Fig. 2E).

Dynamic Binding and Conformational Changes of the NS1619 Binding Pocket.

To determine whether the binding site of NS1619 indeed undergoes a conformational reorientation, we conducted 4.5 μs molecular dynamics simulations of the BK-NS1619 complex in both closed and open states, each simulated separately. In both closed and open states, NS1619 interacts with the residues that are mainly located in the predicted active site (Fig. 3 A, Left panel). As the trajectory advances in the open state, the S6 helix shifts, enabling interaction of NS1619 with the K330–K331 and Y332 complex (Fig. 3 A, Right panel), resulting in conformational shifts in this region compared to the initial closed state (Fig. 3B). The side chains of K330 and Y332 rotate following NS1619, while F315 of the S6 chain strengthens its interaction with NS1619 only in the open state of BK. This behavior is less pronounced in the closed state. Both lysines (K330 and K331) interact with NS1619 via van der Waals interactions involving the aliphatic carbons of their side chains, not with their amino groups. The activator finally interacts with F223, linking the S6 helix to S4 and thereby stabilizing the open conformation of the channel; the closer proximity between F223 and the C-linker residues in the open-state simulation is maintained over time (Fig. 3C), consistent with several molecular dynamic simulations.

Fig. 3.

Molecular dynamics simulations reveal NS1619 stabilizes the open state by bridging the C-linker and VSD. (A) Snapshots from MD simulations. Representative structures of the BK channel at the start (frame 0) and end of separate simulations initiated in the closed (PDB: 6V35, blue) and open (PDB: 6V22, red) states. The magnified Insets highlight the binding pocket, showing the key residues (labeled) and NS1619. (B) Conformational changes. Superposition of the channel structure from the beginning of the closed-state simulation (blue, frame 0) and the end of the open-state simulation (red, final frame), highlighting the conformational shifts in the C-linker and VSD region upon stabilization of the open state by NS1619. For visual reference, the yellow shaded area indicates where residues K330, K331, and Y332 are located. (C) Temporal evolution of the distance between the COM of residues K330-Y332 and F223. Plot showing the distance between the center of mass (COM) of the F223 residue (S4 helix) and the COM of residues K330, K331, and Y332 (S6/RCK1 linker) over the 4.5 µs simulation time for trajectories initiated from the closed (blue trace) and open (red trace) states. The COM for each group was calculated frame-by-frame using VMD, defined as the mass-weighted average of the coordinates of all atoms within the entire residues (including both backbone and side chains). The distance remains consistently shorter and shows less variability in the simulation initiated from the open state, indicating closer proximity and a stable interaction between the VSD (represented by F223) and the modulatory pocket in this conformation.

NS1619 Increases Gate Equilibrium Constant L_o.

Residues from both the voltage sensor and pore domain interact with NS1619 as part of its binding pocket (Figs. 1 and 2). To quantify the involvement of pore gating in the activation mechanism of the BK channel by NS1619, we took advantage of the BK allosteric gating model (23), where the open probability (P_o) in the absence of Ca²⁺ is given by the expression:

$$P_o=\frac{1}{1+\frac{(1+J{)}^4}{L(1+JD{)}^4}},$$ [1]

where, $J=J_0{e}^{\frac{\mathit{zjFV}}{\mathit{RT}}}$ defines the equilibrium constant of the resting-active equilibrium of the voltage sensor, J₀ is the equilibrium constant at 0 mV, z_J defines the voltage dependence of the voltage sensor activation, V is the applied voltage, and F, R, and T have their usual meanings. Eq. 1 becomes Po = 1/(1 + 1/L) when all the voltage sensors are resting, where $L=L_0{e}^{zLFV/RT}$. $L_o$ is the equilibrium constant defining the closed-open equilibrium at 0 mV, and z_L is the voltage dependence for channel opening. Therefore, if the number of channels in the patch is known, L can be determined by plotting the absolute Po at sufficiently negative voltages [limiting slope method (24, 25)]. The number of channels in the patch can be obtained using nonstationary noise analysis. (Eq. 4) z_L can be calculated in the absence of Ca²⁺ at very negative voltages using the relation

$$\underset{V\to -\infty }{\text{lim}}\frac{\mathit{dlnPo}}{\mathit{dV}}=\frac{z_{L}F}{\mathit{RT}}.$$ [2]

Fig. 4A shows representative unitary currents at −60 mV without the activator, and (Fig. 4B) in the presence of 30 μM NS1619, where the presence of the activator leads to an apparent increase in Po. Po data at negative voltages were plotted on a logarithmic scale (Fig. 4F), and the slope of the straight lines fitting the data defines z_L. Extrapolating this straight line helps determine L_o, represented by the log Po data at zero mV (Fig. 4 E and F). Under these conditions, L_o is 4 × 10⁻⁶ ± 2 × 10⁻⁷ and 2 × 10⁻⁵ ± 1 × 10⁻⁶ in the absence and the presence of 30 μM NS1619, respectively.

Fig. 4.

NS1619 modulates the equilibrium constant L between the open and closed states of the BK channel. (A) Representative BK currents recorded at −60 mV in the absence of NS1619 (0 µM). NPo = 6.018 × 10⁻⁴. The Inset shows representative channel openings from the same trace displayed at an expanded time scale. (B) BK currents from the same patch shown in (A) after addition of 30 µM NS1619. NPo = 5.092 × 10⁻³. The inset shows representative channel openings at an expanded time scale. The patch was estimated to contain 489 active channels, as determined by nonstationary noise analysis. (C and D) Logarithmic representation of Po vs. voltage for the BK channel in the absence of the activator (black, N = 6) and in the presence of 30 µM NS1619 (blue, N = 5). Solid lines represent the best fit using the HA model. L₀ = 4 × 10⁻⁶ ± 2 × 10⁻⁷ without the activator and 2 × 10⁻⁵ ± 1 × 10⁻⁶ with 30 µM NS1619. D = 17 ± 2 and 8.5 ± 1 in the absence and presence of the activator, respectively. (E) τ vs. voltage from patches recorded with 0 µM (black circles) or 30 µM NS1619 (blue circles). Lines represent fits using Scheme 1 with parameters listed in SI Appendix, Table S1.

Fig. 4E shows the time constant versus voltage in the absence (black circles) and presence (blue circles) of 30 µM NS1619, derived from macroscopic currents (SI Appendix, Fig. S4). Notably, at negative potential, where the time constant is given by 1/γ, the time constant in the presence of the activator (τ) is higher, indicating that the rate describing the transition from the open to the closed state of the channel (γ), (Scheme 1) decreases. Since L₀ = δ₀/ γ₀, where δ₀ is the forward rate constant, we conclude that all the change in L due to the agonist resides in the decrease in γ. To explore whether the change in γ alone is sufficient to explain the effect of NS1619 or if other parameters of the allosteric model [Scheme 1; (26)] are also involved, we fitted the logPo(V) data using Eq. 1. We found that the best fit (Fig. 4E, black and blue lines) is achieved by a decrease in both γ and the allosteric factor D (allosteric coupling between the PD and VSD), as shown in SI Appendix, Table S1. Importantly, the rates describing the voltage sensor (α and β) remain unchanged in the presence of the activator. This aligns with the findings of Gessner et al. (21), who demonstrated that the activator does not affect the channel’s Q/V relationship. Furthermore, the values obtained for δo and γo from τ are consistent with the Lo values derived from the limiting slope method, validating the hypothesis that NS1619 modulates this equilibrium constant.

Scheme 1.

BK channel allosteric model.

These findings support our hypothesis that NS1619 activates BK by stabilizing the open conformation of the pore domain. Our data strongly support the idea that NS1619 influences the conformational stability of the PD, resulting in a longer channel open time and a decrease in the deactivation rate.

Discussion

This study integrates techniques from electrophysiology, molecular docking, cryoelectron microscopy, and molecular dynamics simulations to define the binding site and mechanism of action of the BK channel activator NS1619. Single particle cryo-EM and molecular dynamics analysis reveal that amino acids from the RCK1 and C-linker initially form the binding site. Upon NS1619 binding, the binding pocket reconfigures to involve an amino acid from S4 (VSD). This dynamic behavior allows NS1619 to stabilize the BK’s open conformation, highlighting a dual pathway of modulation that should help define new therapeutic targets.

Structural and molecular docking results point to a set of residues in the VSD, RCK1, and C-linker that putatively participate in NS1619 binding, including F223, R329, K330, K331, Y332, G334, F391, K392, and E399. However, side chain perturbations by mutagenesis indicate that only F223, K330, and K331 significantly alter the response to the activator compared to the wild type. This apparent discrepancy between structural data and functional results can be explained by considering the nature of these particular protein–ligand interactions. Although residues such as Y332, G334, F391, K392, and E399 appear structurally close to NS1619, mutating these to alanine does not alter the channel function in the presence of the activator, suggesting their role may be more related to the structural conformation of the binding site or stabilization of the electrostatic milieu, rather than directly participating in the interaction with the activator. Our molecular dynamics results (Fig. 3) show that twisting the S6 segment induced by the final conformation of NS1619 in the binding pocket allows direct interaction between the K330–K331 complex and the activator, which also interacts with F223. These results suggest a “foot-in-the-door” mechanism where NS1619 would remain bound to F223 of the VSD on one side and to K330–K331 of the S6-RCK linker on the other, stabilizing the open configuration of the channel (Fig. 5).

Fig. 5.

Postulated mechanism of allosteric activation of the BK channel by NS1619. (A) BK channel in the closed state (0 Ca²⁺). Top: Representation of the channel in its basal closed conformation. Key residues F223 (S4) and K330 (C-linker) are highlighted. Bottom: It is postulated that in the closed state, NS1619 binds dynamically, oscillating between two locations: NS1619 cyan color and the active site (NS1619 red color). The yellow arrows of equal thickness symbolize an unstable equilibrium between these two positions. (B) BK channel in the open state (saturating Ca²⁺). Top: Representation of the channel in its open conformation, stabilized by Ca²⁺ (yellow ions bound to RCK1). Bottom: It is postulated that in the open state, the activator binds preferentially to the active site (NS1619 red color), interacting with F223 and the C-linker residues (K330/K331). The thicker yellow arrow pointing to the right indicates that the equilibrium is shifted toward this stable binding pose, acting as a foot-in-the-door that stabilizes the open conformation. Allosteric model diagram (Center). As also illustrated in the center of the figure, the allosteric model by Horrigan and Aldrich is shown (Top) along with the postulated effect of NS1619 on its parameters (Bottom). The activator shifts the C↔O pore equilibrium toward the open state, representing an increase in the equilibrium constant L (thick red arrow). Additionally, the activator decreases the allosteric coupling factor D (indicated by the smaller black arrow).

The reduction in NS1619 facilitation by these mutants correlates with the impact of the mutation on the closed-open pore equilibrium, defined by the equilibrium constant Lo. We note that K330A increases Lo approximately 24-fold, an effect on Lo much larger than the induced by mutations F223A and K331A, which increase Lo about threefold compared to the wild type (27). These differences suggest that K330 may play a crucial role anchoring NS1619 to the binding site and in transducing the conformational change that favors the opening of the channel. Mutant K330A likely alters the interaction dynamics between the transmembrane and cytosolic domains, preventing an efficient transition to the open conformation stabilized by NS1619. In contrast, although mutants F223A and K331A disrupt activation by NS1619, they lead to minor changes in Lo, indicating that their contribution to the stability of the open state is significant but less critical than that of K330. This analysis highlights how specific structural modifications within the BK channel can dramatically influence its pharmacology, underscoring the importance of these residues in the modulation of the channel by NS1619 and providing a direct link between the molecular structure of the channel and its biophysical function. A detailed understanding of these mechanisms is crucial for the biological characterization of the channel and in the structural-based design of more effective and specific modulators.

To further probe the functional role of these key residues, we performed more targeted mutagenesis (SI Appendix, Fig. S6). The F223Y mutation, which preserves aromaticity while adding a hydroxyl group, rendered the channel hypersensitive to NS1619, more than doubling the activatory shift. This large leftward shift of the conductance-voltage curve suggests that an aromatic residue capable of hydrogen bonding at this position enhances the drug–channel interaction. Furthermore, our results unveiled distinct roles for the two critical lysines. The charge-neutralizing K330N mutation completely abrogated the response to NS1619, confirming that a positive charge at this position is an absolute requirement for the agonistic effect. In stark contrast, the K331Q mutation also resulted in a dramatic potentiation of the activator’s effect, indicating that at position 331, the interaction can be significantly enhanced by a polar, hydrogen-bonding residue. Together, these results reveal a nuanced and complex mechanism beyond the initial alanine scan.

A Versatile Modulatory Pocket for Diverse BK Activators.

Our molecular docking results suggest the intriguing hypothesis that the interface between the VSD, C-linker, and RCK1 domains forms a versatile pocket capable of accommodating a diverse range of activators, including NS004, NS1608, GoSlo-SR-5-6, and Cym04 (SI Appendix, Fig. S5). While this suggests a shared general modulation location, a detailed comparison of our functional data for NS1619 with published results for other compounds, such as Cym04, reveals important mechanistic differences. For instance, NS1619 activation is associated with a decrease in the allosteric factor D, whereas Cym04 was proposed to increase it. Furthermore, the mutational sensitivity profiles are distinct: Our data show the K330A mutation significantly impairs NS1619 activation, while this was not observed for Cym04 (21). Conversely, mutations at Y332 are known to affect Cym04 but did not significantly impact the action of NS1619 in our hands. Despite these functional differences, the consistent involvement of residues within this S6-C-linker region supports the idea of a common modulatory hotspot.

Furthermore, our findings align with those of Webb et al. (28), who proposed a hydrophobic binding pocket for GoSlo-SR-5-6 in a similar region between the S4/S5 and S6 loops of the BK channel. Therefore, this region is best described not as a single, conserved binding site with a common mechanism, but as a versatile modulatory pocket where different activators can engage distinct subsets of residues to produce specific effects on channel gating. Future mutagenesis and electrophysiology studies will be essential to confirm whether this site is indeed a common hotspot for BK channel modulation by different families of activators.

Activation Mechanism of the BK Channel by NS1619.

Molecular dynamics, limiting slope, and time constant analysis all suggest that NS1619 activates the BK channel by shifting the closed-open equilibrium toward the open configuration, increasing the equilibrium constant Lo, and decreasing the allosteric factor D. It has been established that NS1619 does not affect the resting-active equilibrium of the voltage sensor and that the activation effect is independent of Ca²⁺ binding (21). Limiting slope fits shows an increase in the equilibrium constant between the closed and open states in the presence of the activator. When we evaluate the time constant at very negative potentials where τ is equivalent to 1/γ (describing the transition from open to closed), we observe a reduction in γ indicating that the channel closures slow down in the presence of NS1619 (Fig. 4F). Since the rate constant describing the transition from closed to open (δ) does not change in the presence of NS1619, the increase in Lo is entirely explained by the decrease in γ. Molecular dynamics calculations and our alanine scanning results suggest that the activator interacts with phenylalanine 223 in the VSD, an observation supported by the observed effect on the allosteric factor D.

Furthermore, molecular dynamics also show that once NS1619 binds to the final conformation of the binding site, a likely movement in S6 leads to an interaction with lysine 330. The fact that the K330A mutation eliminates the activating effect of NS1619 and the critical role of the interaction between NS1619 and K330 in stabilizing the channel in its open state leads us to propose a foot-in-the-door mechanism. In this scheme, the activator binds to its binding site. It undergoes a reconfiguration that enables NS1619 to interact with F223 of the VSD on one side and with the K330–K331 complex of the C-linker on the other, effectively stabilizing the open state of the channel and acting as a crucial foot-in-the-door (Fig. 5). The present results are also aligned with the idea that other activators might share this binding site and mechanism, either partially or fully. For example, Webb et al. (28) found that GoSlo-RS-5-6 activates the channel by increasing Lo, Jo, and reducing D. Gessner suggests that Cym04 and NS1619 are modulators of Lo and Jo, shifting the equilibrium toward the open and active configuration respectively (21). However, Hoshi and Heinemann (29) argue that there is some ambiguity regarding Jo, as part of the effect could be due to changes in the coupling between the VSD and the ion conductance gate. As Gessner shows, the Q/V of the channel does not change in the presence of the activator, implying that the activator does not modify the voltage sensor resting-active equilibrium. Thus, the mechanism would be through an increase in Lo and a decrease in D, which aligns with our results. Furthermore, Rockman et al. propose that NS11021 activates the BK channel through the PD by slowing the rate constant between the open and closed states of the channel, concluding that NS11021 stabilizes an open state (30).

Conclusion.

In this work, we have enhanced our understanding of the activation mechanism of the BK channel by the activator NS1619, identifying a binding site whose active conformation stabilizes the channel´s open state, thereby preventing closure. This finding sheds light on the regulation of ion channels and suggests a modulation pathway involving both the voltage-sensing and cytosolic domains, which could be therapeutically exploited. The involvement of specific residues in channel modulation, coupled with the proposed foot-in-the-door mechanism, highlights the complexity of ion channels as pharmacological targets and underscores the need for a thorough understanding of their modulation to design future drugs.

Furthermore, the characterization of this site as a versatile modulatory pocket opens avenues for developing innovative modulators. Understanding the features of this pocket, including its key hydrophobic interactions, provides a roadmap for the rational design of more specific BK channel activators.

Materials and Methods

Channel Expression for Electrophysiology.

BK channels were expressed in Xenopus laevis oocytes. The cDNA for the wild-type human BK α-subunit (Genbank U11058) was kindly provided by L. Toro (University of California, Los Angeles, CA). Mutants were created using standard molecular biology techniques and verified by sequencing. mRNA was synthesized in vitro using the mMESSAGE mMACHINE kit from Ambion. Oocytes were then injected with 50 ng of mRNA and incubated in ND96 solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and HEPES, pH adjusted to 7.4 with NaOH) at 18 °C for 3 to 7 d before electrophysiological analysis.

Electrophysiological Recordings.

All current recordings were acquired using the patch-clamp technique in the inside-out configuration. Data were acquired with an Axopatch 200B (Molecular Devices) amplifier and the Clampex 10 (Molecular Devices) acquisition software. The voltage command and current output were filtered at 10 kHz with an 8-pole Bessel low-pass filter (Frequency Devices). Current signals were sampled with a 16-bit A/D converter (Digidata 1550B; Molecular Devices), using a sampling rate of 500 kHz. Unless otherwise stated, linear membrane capacitance and leak subtraction were performed based on a P/-8 protocol (Bezanilla and Armstrong, 1977). Borosilicate capillary glasses (1B150F-4, World Precision Instruments) were pulled in a horizontal pipette puller (Sutter Instruments). After fire-polishing, pipette resistance was typically 1 to 2 MΩ. The ground electrode was an agar bridge containing 1 M KMeSO₃. All experiments were performed at room temperature (20 to 22 °C).

Macroscopic Currents.

Potassium currents were elicited by pulse protocols with a holding potential of −100 mV followed by a test pulse to different voltage steps from −100 to 300 mV in 10 mV increments, separated by intervals of 10 ms, and ending with a voltage pulse of −100 mV. The pipette and bath solutions were the same and contained 110 mM KOH, 10 mM HEPES, 2 mM KCl, 1 mM EGTA (free Ca²⁺ ∼0.5 nM), and the pH was adjusted to 7.4 with methanesulfonic acid. All data analyses were conducted using Clampfit 11.0.3 (Molecular Devices), Python, and Excel 2013 (Microsoft). The voltage dependency of the macroscopic tail currents was quantified using a Boltzmann function.

$$\frac{I}{I_{\mathit{max}}}=\frac{1}{1+e^{\left(\frac{-z_{\delta }F\left(V-V_{0.5}\right)}{\mathit{RT}}\right)}},$$ [3]

where I_max is the maximum current, V_0.5 is the voltage at which I/I_max equals 0.5, F is Faraday’s constant, R is the gas constant, T is the temperature in Kelvin, and z_δ is the voltage dependency.

Nonstationary Noise Analysis.

These experiments were conducted using the same solutions described for macroscopic currents, under nominally Ca²⁺-free conditions (1 mM EGTA in the internal solution). A nonstationary noise analysis was applied, deriving the mean and variance (${\sigma }_i^2)$ from isochronal segments of 200 current traces at 200 mV. We determined the unitary current (i) and the number of channels (N) by fitting the variance against the mean current using the expression

$${\sigma }_i^2=i<I>-\frac{<I{>}^2}{N},$$ [4]

where <I> is the mean current. Variance and mean current were determined using analysis software, which accounts for the gradual rundown of the current during experiments (31, 32).

This method also allowed us to calculate the maximal Po (Po max) at 200 mV, using the maximum mean current obtained divided by iN (all channels open).

$$Pomax=\frac{<I{>}_{\mathit{max}}}{\mathit{iN}}.$$ [5]

Limiting Slope.

Limiting slope experiments were performed using macropatches with hundreds of channels. N, i, and Po max were determined as described in the previous section. After obtaining N, macroscopic currents induced by voltage pulses from −100 to 250 mV in increments of 10 mV were recorded. Voltage pulses were separated by 10 ms intervals. Unitary currents were recorded over 2 to 10 s, applying voltages ranging from −20 to −200 mV in steps of 20 mV. Both macroscopic and unitary current recording protocols were replicated on the same patch after perfusion with a solution containing 30 μM NS1619.

Histograms were created for all data points to obtain the NPo from single-channel current recordings. These histograms were normalized by dividing the ordinate values by the integral along the membrane current (I) axis to derive the probability density function (PDF), which delineates the likelihood of a given current level occurring based on the data distribution (33). Assuming that a maximum of nine channels could be open concurrently, the theoretical PDF was constructed from the sum of ten Gaussian distributions along the current axis.

$$PDF=\sum _{k=0}^9\frac{P_k}{{\sigma }_k\sqrt{2\pi }}{e}^{-\frac{(I-ki{)}^2}{2{\sigma }_k^2}},$$ [6]

where k is the number of simultaneously open channels, and i is the unitary current. Each Gaussian distribution is centered around the expected current level, ki, with a variance of ${\sigma }_k^2$. The variance for each current level accounts for both instrumental noise and the noise from the open channels. The overall variance ${(\sigma }_k^2)$ for the level with k open channels is taken as ${({\sigma }_c+k{\sigma }_o)}^2$, where ${\sigma }_c^2$ is the variance of the closed level, and ${\sigma }_o^2$ is the variance from the noise of an open channel. P_k is the relative weight of each current level and was calculated using the expected probability of finding k open channels simultaneously on a membrane containing N channels, with an opening probability (Po), as described by a Poisson distribution (33).

$$P_k=\frac{(NPo{)}^k}{k!}{e}^{-NPo}.$$ [7]

NPo was derived by fitting the theoretical PDF to the experimental PDF using the Solver add-in in MS-Excel. The adjustable parameters included NPo, ${\sigma }_c^2$, ${\sigma }_o^2,$, and i. The Po value was obtained by dividing by the number of channels determined from the nonstationary noise analysis conducted on the same patch.

Activation and Deactivation Kinetics.

The activation kinetic characteristics were obtained from macroscopic currents evoked in the inside-out configuration from −100 mV up to 300 mV in 10 mV steps of 10 ms duration with a prepulse of −100 mV and returning to −100 mV under symmetric potassium conditions. The protocol to assess deactivation consisted of a depolarizing pulse of 150 mV followed by test pulses from 0 to −300 mV in 10 mV steps and 5 ms duration. Both protocols were repeated after perfusing the internal side of the patch with a solution containing 30 µM NS1619. Current traces from activation and deactivation were quantified using a simple exponential from which the experimental time constant (τ) was derived. τ data were fitted using a 10-state voltage-dependent model (Scheme 1) that utilized microscopic rate constants instead of equilibrium constants, allowing for the calculation of activation and deactivation time constants for BK channel gating. This approach is analogous to the methods previously (26, 30, 34) using the equation:

$$\tau =\frac{1}{\sum ({\gamma }_{i}p{O}_i+{\delta }_{i}p{C}_i)},$$ [8]

where $p{O}_i$ and $p{C}_i$ are the conditional occupancies of the open and closed states, respectively, ${\gamma }_i$ is the rate describing the transition from the open to the closed state and ${\delta }_i$ is the rate describing the transition from the closed to the open state.

In Scheme 1, rate constants were defined by rates for VSD activation and deactivation (α and β, respectively), PD closing and opening (γ and δ), a gating valence for each of these rate constants (zα, zβ, zγ, and zδ), and allosteric coupling between the VSD and PD (D) using the additional simplifying assumption D = f⁴ (30).

Molecular Docking.

Using the electron microscopy structure of human BK (PDB: 6V35), we focused on the entire protein and employed the SiteMap module within the Schrödinger suite to identify potential ligand-binding sites. The BK protein was preprocessed and analyzed using SiteMap’s default settings to assess properties such as exposure, pocket volume, and hydrophobicity. This analysis generated contour maps (site maps), which we evaluated using Site Score values. For each identified site, we generated docking receptor grids with 20 Å spacing on a side.The most promising site, yielding the highest scores, was located near residues R329, K330, K331, Y332, and K392 on a single chain. These residues were chosen because the SiteMap analysis indicated a high potential for ligand binding in this region. The α-β4 complex was used due to its higher resolution compared to structures lacking β4; according to Tao and MacKinnon (22) there are no significant structural differences between complexes with or without β4. We centered the grid on the centroid of each site map. Ligand docking was performed using the Glide program from the Schrödinger suite (2022-2) with default settings and the OPLS4 force field. The ligands were initially sketched in 2D and prepared using the LigPrep suite, which assigns protonation states automatically based on a user-defined pH. We conducted docking in standard precision (SP) mode, which allows flexible ligand sampling. The poses were scored and sorted using the default Glide Docking Score function, and the configurations with the most favorable docking scores were selected, confirming this site as druggable (Table 1).

Table 1.

Docking scores for diverse BK channel activators

compound	Score Kcal/mol
NS1608	−5.3
NS004	−5.0
SR-5-44	−4.8
CyM04	−3.2
NS11021	−3.1
SR-5-6	−4.6
NS1619 (6V22, open)	−6.1
NS1619 (6V35, closed)	−4.4

Molecular Dynamics Simulations.

We used the 3.5 Å resolution Cryo-EM structures of BK channels, with entry codes 6V35 (closed state) and 6V22 (open state) (23). To complete and refine small loops within these structures, we employed AlphaFold (35). The proteins were integrated into a POPC (1-palmitoyl-2-oleoyl-phosphatidylcholine) membrane system comprising 785 lipids, 125,611 water molecules, and ions (338 K⁺, 398 Cl⁻), with a salt concentration of 0.15 M. The final system dimensions were 180 × 180 × 181 ų. Molecular dynamics (MD) simulations were conducted using the AMBER software package (36). The protein structures were initially constrained in the loops with a harmonic restraint of 10 kcal·mol⁻¹Å⁻². The system underwent a relaxation process, where the restraints were progressively reduced by applying decreasing force constants in the following order: 10, 5, 2.5, 1.0, 0.5, and finally 0 kcal·mol⁻¹ Å⁻². This stepwise reduction facilitated a smooth transition to an unrestrained state, ensuring the stability and equilibrium of the system. Two separate MD simulations were performed in the presence of NS1619: one for the closed state (6V35) and another for the open state (6V22), both with a duration of 4.5 μs and a timestep of 2 fs. Before the MD simulations, molecular docking of the ligand was performed at the active sites of each protein subunit, and the resulting docked poses were used as the initial configurations for the subsequent MD simulations.The β4 subunit was not included in either simulation. Van der Waals interactions were calculated using a 1.0 nm cutoff, and a dispersion correction was applied for both energy and pressure (37). The particle–mesh Ewald (PME) method was used to compute electrostatic interactions (38).The simulations employed the Amber19SB force field for proteins (39), the LIPID21 force field for lipids (40), the OPC water model (37), previously described ion parameters (41). Temperature was maintained at 310 K using the velocity rescale (v-rescale) thermostat (42), while the semi-isotropic Berendsen barostat was used to maintain a pressure of 1 atm and 1 bar (43). The Visual Molecular Dynamics (VMD) software was used to construct the system and generate images (44). Histograms of angles, distances, and displacements from the MD simulations were generated using the Qtiplot program.

Cloning, Expression, and Purification for Cryo-EM Experiment.

A human Slo1 channel was cloned into the pFastbac vector containing a C-terminal 3C protease site, eGFP. P0 baculovirus was generated using the Bac-to-Bac method (Invitrogen) using Cellfectin II as the transfection reagent (Thermo Fisher Scientific, 10362100). The P0 virus was amplified once to yield the P1 baculovirus, which was used to infect HEK293S GnTI- cells a ratio of 1:10 (v/v). After 20 to 24 h of incubation at 37 °C, 10 mM sodium butyrate was added to the cells and the culture was transferred to 30 °C. Cells were pelleted, washed with PBS, rapidly frozen in liquid nitrogen, and stored at −80 °C until use. For purification, all steps were performed at 4 °C and frozen cell pellets were thawed, diluted, and detergent-extracted in 20 mM Tris-HCl pH 8, 320 mM KCl, 10 mM MgCl₂, 10 mM CaCl₂, 2 mM TCEP, 1% LMNG for 90 min. The solubilized supernatant was isolated by ultracentrifugation and incubated for 2 h with 1 ml CNBR-activated Sepharose beads (GE Healthcare) coupled with 4 mg high-affinity GFP nanobodies (38). Beads were collected by low-speed centrifugation, washed in batch mode with 10 column volumes of main buffer 20 mM Tris-HCl pH 8, 320 mM KCl, 10 mM MgCl₂, 10 mM CaCl₂, 2 mM TCEP, 1 μL/mL aprotinin, 1 μL/mL pepstatin, 10 μL/mL Soybean Trypsin Inhibitor and 100 μL/10 mL 1-palmitoyl-2-oleoyl- glycero-3-phosphocholine (POPC):1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE): 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphate (POPA) 5:5:1(PC:PE:PA) diluted in GDN, collected on a column by gravity, and washed with another 10 column volumes. Proteins were cleaved by the HRV 3C protease (39) overnight, concentrated, and analyzed using SEC on a Superose 6, 10/300 GE column (GE Healthcare), with running buffer containing 20 mM Tris-HCl pH 8, 320 mM KCl, 10 mM MgCl₂, 10 mM CaCl₂, 2 mM TCEP, 1 μL/mL aprotinin, 1 μL/mL Pepstatin, 10 μL/mL soy tripsin and 50 μL/10 mL PC:PE:PA. Peak fractions were collected and concentrated using a 100 kDa molecular mass cutoff centrifugal filter (Millipore concentrator unit) to 3.5 mg/mL. The concentrated protein was immediately used for the cryo-EM grid-freezing step.

Cryo-EM Sample Preparation and Imaging.

Quantifoil 200-mesh 1.2/1.3 grids were plasma-cleaned for 30 s in an air mixture in a Solarus Plasma Cleaner (Gatan). Purified hBK-NS1619 samples were applied onto the grids and frozen in liquid-nitrogen-cooled liquid ethane using a Vitrobot Mark IV (FEI) and the following parameters: sample volume 3.5 μL, blot time 4 s, blot force 1, humidity 100%, temperature 22 °C. Grids were imaged on a Titan Krios with a K3 detector (in superresolution mode) and GIF energy filter (set to 20 eV) at a nominal magnification of ×81,000, corresponding to a superresolution pixel size of 0.53 Å. The videos were acquired at 1 e−/ Ų per frame for 50 frames.

Single-Particle Cryo-EM analysis.

Image processing and three-dimensional map reconstruction were performed using the cryoSPARC program. Initially, individual particles were selected from the cryo-EM images and aligned and classified into 2D classes. Subsequently, 3D classification and refinement were carried out to reconstruct the Coulomb density map. Finally, a nonuniform refinement was performed applying C1 and C4 symmetry, achieving a final resolution of 3.83 Å and 3.44 Å, respectively. SI Appendix, Fig. S2 shows a diagram of the overall processing and refinement strategy.

Model Building.

The model of hSlo1 (PDB 6V22) was used to build atomic models for the hSlo1–NS1619 complex into our density maps. The initial models were built via iterative rounds of manual model building on COOT (40) and real space refinement in Phenix (41). The final refined atomic model was obtained using interactive, flexible fitting using ISOLDE (42). All structural analyses and figures were generated using UCSF ChimeraX (43).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Cryo-EM density maps and atomic coordinates of the hSlo1 channel in complex with NS1619 have been deposited in the Electron Microscopy Data Bank under accession code: EMD-75024 [https://www.ebi.ac.uk/pdbe/entry/emdb/EMD-75024] (hSlo1-NS1619) (45). The Protein Data Bank under accession code:10AD [https://doi.org/10.2210/pdb10AD/pdb] (hSlo1-NS1619) (46).