Site and mechanism of BK channel inhibition and cerebral artery constriction by pregnenolone

Abstract

Circulating vasoactive steroids including estrogens, androgens, progesterone, and bile acids modulate the activity of voltage/Ca^{2 +}-gated, big conductance potassium (BK) channels, an action that requires regulatory BK β subunits. In contrast, pregnenolone at local/therapeutic concentrations has recently been shown to modulate BK channel activity in absence of such subunits. The site and mechanism of pregnenolone action remain unknown. Thus, we decided to test the hypothesis that BK channel-forming α subunits (slo1 proteins encoded by KCNMA1 or Slo1 in mammals) contain a specific residue(s) that recognizes pregnenolone and mediates modulation of smooth muscle BK channel activity and diameter of cerebral arteries by this steroid. We used microscale thermophoresis to demonstrate that pregnenolone at local/therapeutic concentrations binds to slo1 proteins (K_D=3.5 μM); these concentrations were reported to inhibit BK channel activity and decrease cerebral artery diameter. Based on a slo1 cryo-EM structure (PDB 6V38) computational modeling identified Tyr450, Hys409 and Asp896 as participants in pregnenolone recognition, with hydrogen-arene interactions between steroid and Tyr450. Consistently, slo1Y450F proteins failed to bind pregnenolone. Following heterologous expression, this mutant was resistant to pregnenolone-induced inhibition, an action that was coupled to Ca²⁺-driven gating and enabled by the RCK1 high-affinity Ca²⁺-binding site. In contrast to their wild-type counterparts, middle cerebral arteries from Kcnma1 Y450F knock-in mice failed to constrict to pregnenolone, whether probed ex vivo or in vivo. Thus, metabolic, circulating, or endothelial factors cannot override pregnenolone cerebrovascular action, which mainly results from direct interaction with Tyr450 within the proximal CRAC motif in the slo1 cytosolic tail domain. Our study provides a new paradigm in vasoactive steroid-BK channel interactions that can be used in developing pharmacophores/new agents to replace/regulate pregnenolone effects in clinical scenarios, and underscores that CRAC motifs may mediate ionotropic receptor regulation by steroids other than cholesterol.

1. Introduction

The modulation of physiological processes by direct interaction between lipid molecules and ion channel proteins and the contribution of such interaction to cell physiology constitute an expanding area of research [1–3]. Ca²⁺/voltage-gated, big conductance potassium (BK) channels regulate several physiological processes including myogenic tone and thus, arterial diameter [4,5]. Several circulating vasoactive steroids have been reported to increase BK channel activity through recognition by BK channel regulatory β subunits: β₁ for 17 β-estradiol [6] and bile acids [7,8]; β₁, β₂ and β₄ for progesterone and dehydroepiandrosterone [9,10]. Collectively, all these studies indicate that β subunits endow native BK channels with sensitivity to several circulating vasoactive steroids.

In sharp contrast, we demonstrated that the vasoactive neurosteroid pregnenolone (PREG) at concentrations found in physiological/therapeutic scenarios [11–16], inhibited BK channels in absence of steroid-sensing β subunits. Rather, channel-forming α subunits (slo1 channels, encoded by KCNMA1 or Slo1 in mammals) and a bare lipid microenvironment suffice to sustain PREG-induced channel inhibition [15]. Whether this PREG action results from direct interaction with slo1 channel proteins or disruption of the lipid microenvironment with secondary modification of channel function remains unknown. Here, we set to test the hypothesis that BK channel-forming slo1 proteins (slo1 channels) contain specific residues that recognize pregnenolone and mediate inhibition of BK channel activity and constriction of cerebral arteries by this steroid.

PREG modifies neurovascular function [15], an action that could contribute to the effects of this vasoactive steroid on conditions critically affected by cerebral blood flow, such as Alzheimer’s Disease, vascular dementia and other forms of cognitive decline [16–22]. For the implementation of novel, BK channel-targeting pharmacological interventions for these conditions, it is necessary to first identify the site and mechanism of PREG action on cerebrovascular BK channels. In this study, we combined microscale thermophoresis measurements (MST) of slo1 protein stability [23], computational modelling based on the full slo1 cryo-EM structure [24], point mutagenesis, single channel resolution electrophysiology on both recombinant channels reconstituted into artificial lipid bilayers and native cerebrovascular smooth muscle (SM) BK channels using patch-clamp methods, the Kcnma1Y450F knock-in mouse model, and evaluation of middle cerebral artery (MCA) diameter both ex vivo and in vivo, to demonstrate that PREG at physiological (local) and therapeutically relevant concentrations binds slo1 channels via specific docking onto the central Tyr of the proximal Cholesterol Recognition Amino acid Consensus (CRAC) motif present in the channel cytosolic tail domain (CTD), i.e., Tyr450. Our study also demonstrates that this residue contributes to PREG-induced MCA constriction, which is caused by PREG inhibition of cerebrovascular SM BK channels, with this inhibition linked to allosteric coupling of Ca²⁺-sensing by the high affinity divalent site present in the channel RCK1. Thus, this study identifies the critical importance of a single residue in PREG-slo1 channel direct interaction and the mechanism of this steroid cerebrovascular action at molecular, organ and organismal levels while underscoring that CRAC motifs may mediate biological effects of steroids other than cholesterol.

2. Methods and materials

2.1. Animals

The care of animals and experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committees of the University of Tennessee Health Science Center and the University of Pittsburgh, which are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International. In all studies, whether ex vivo or in vivo, 8- to 12-week-old mice of both sexes were used. Generation of Y450F knockin (KI) mice on a C57BL/6 J background was conducted at the University of Pittsburgh as follows: a gRNA binding site near the Kcnma1 Y450F KI mutation site was identified using CCTop [25]. This gRNA target site was used to produce an Alt-R CRISPR-Cas9 crRNA (IDT DNA, Coralville, IA) which was hybridized to a universal 67-mer Alt-R CRISPR-Cas9 tracrRNA (IDT DNA) to produce a gRNA. A 103 nt long PAGE purified Ultramer single stranded DNA Oligo (IDT DNA) with three phosphorothioate modifications on each end [26] that was homologous to the target locus and harbored the desired AC to TT KI mutations to change codon 450 from Tyr to Phe was used as a repair template. Single cell C57BL/6 J embryos were electroporated with gRNA (200 ng/μl), IDT Alt-R^® HiFi Cas9 Nuclease V3 protein (50 ng/μl), and repair template (200 ng/μl) using a Bio-Rad Gene Pulser Xcell electroporator and 1 mm-gap electrode with the following electroporation parameters: 25 V pulse × 5; 3 msec pulse time; 100 msec pulse interval. Following electroporation, surviving embryos were transferred to the oviducts of CD1 (Charles River) pseudo-pregnant recipient females. Offspring were genotyped using PCR (292 bp product of forward primer 5’ACACTTTTCTCCCTGGGCTG3’ and reverse primer 5’TGTGGAGAGGGAGAAGGTGT3’) followed by Sanger sequencing. Guide RNA off target sites were predicted and ranked using CRISPRater [27]. Nine of the top 14 predicted off target sites (list available upon request from GEH) were selected and successfully amplified from founder mouse DNA and Sanger sequenced. To establish the KI line, a male founder mouse with the KI mutation and no off-target mutations in the sites analyzed was bred to C57BL/6 J females. F1 offspring that inherited the KI allele were genotyped by PCR and Sanger sequencing. These offspring were shipped from the University of Pittsburgh to UTHSC to establish a breeding colony.

Aged-matched, WT C57BL/6 J mice of both sexes were purchased from The Jackson Laboratory (Bar Harbor, ME) and used as non-littermate controls. At UTHSC, 8- to 12-week-old Kcnma1 WT and Kcnma1 Y450F KI mice were maintained in-house on a C57BL/6 J background. In addition, both WT and KI littermates were obtained by backcrossing to produce heterozygous breeders to ensure that both genotypes originated from the same parental lineage.

2.2. Cell cultures

Chinese hamster ovary cells (CHO; ECACC Cat # 85050302, RRID: CVCL_0213) were grown to confluence and transiently transfected with construct-carrying pcDNA3.1 using X-tremeGENE HP (Sigma-Aldrich).

2.3. Immunoprecipitation and binding studies

Flag-tagged Slo1 protein product (cbv1 isoform; AY330293) [28] and engineered constructs were purified by immunoprecipitation, performed using Dynabeads Protein G kit (10007D by Invitrogen) following manufacturer’s instructions. Rabbit polyclonal anti-flag antibody (10 μg; Abcam, Cat# ab205606, RRID:AB_2916341) was cross-linked with Dynabeads. Transfected CHO cell lysate supernatant was divided into aliquots containing 20⁷ cells per column. Following precipitation, kit instructions for non-denaturing elutions were followed; one elution of 30 μL was collected for each column, and pH adjusted to 7.5 using UltraPure 1 M Tris-HCI buffer; pH 7.5 (cat# 15567027 by Fisher Scientific). PREG stock or vehicle control (DMSO) was added to the eluate to reach PREG final concentrations (0.001–30 μM), all dissolved in 1% DMSO. Capillaries were prepared and loaded in triplicates into Prometheus (Nanotemper) and heated from 15° to 95°C at a rate of 1°/min. The concentration response curve data obtained from MST were fitted to a Hill coefficient function of the type:

$$y=Start+(End-Start)\frac{(x^n)}{k^n-x^n}$$

using Origin 2020 software (OriginLab), n = 6–9 for each group. Validation of protein purification, reading, and conformational stability shown in SFig.1.

2.4. Computational modeling

Sequences and structures provided by 6V38 Cryo-EM structure of Ca²⁺-bound hslo1 channel (PDB 6V38) [24] were modeled using the modeling feature in MOE 2019.01 (Chemical Computing Group). The library of 10 energetically stable conformations of PREG was docked to all atoms of the hslo1 CTD using the Amber10:EHT force field and built-in function in MOE.

2.5. Bilayer electrophysiological recordings

Slo1 channels were reconstituted into planar lipid bilayers as previously described [29–31]. Succinctly, slo1 cDNA (cbv1 isoform; AY330293) cloned from smooth muscle (SM) cells freshly isolated from rat cerebral arteries [29,30] and inserted into pcDNA3.1 plasmid vector as described [31]. CHO cells were transiently transfected with slo1-pcDNA3.1 plasmid using Lipofectamine 2000 (Invitrogen), grown to confluence, pelleted, and re-suspended on ice in 10 ml of buffer solution (mM): 30 KCl, 2 MgCl₂, 10 HEPES, 5 EGTA; pH 7.2. A membrane preparation was obtained using a sucrose gradient as previously described [32] and aliquots were stored at −80 °C. Ionic currents at single channel resolution were recorded following slo1 channel reconstitution into planar lipid bilayers made of a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) and 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-l-serine (POPS) 3:1 (w/w) mixture. Lipid mixtures were dried under N₂ gas and re-suspended at 25 mg/ml of n-decane. Vertical bilayers (80–120 pF) were formed by painting the lipid mix across a 200 μm diameter hole in a deldrin cup (Warner Instruments). The slo1 protein membrane preparation of slo1 protein was added to the bilayer cis chamber, and fusion between the membrane preparation and the bilayer was promoted by osmosis, with the cis chamber recording solution being hyperosmotic to the trans. Solution for the cis chamber contained (mM): 300 KCl, 10 HEPES; pH 7.2, while solution for the trans chamber contained (mM): 30 KCl, 10 HEPES, pH 7.2. In all experiments, free [Ca²⁺] in these solutions was the same and adjusted to the desired nominal value by adding 1 mM CaCl₂ stock.

The trans chamber was held at ground (equivalent to the extracellular compartment) while the cis chamber (equivalent to the intracellular compartment) was clamped at the desired potential. Only channels with their intracellular Ca²⁺-sensors oriented toward the cis chamber were considered for analysis. Ion currents were obtained during 10-minute gap-free recording at the desired voltage using a Warner BC-525D amplifier, low pass-filtered at 1 kHz using the 4-pole Bessel filter built into the amplifier and sampled at 5 kHz with Digidata 1322 A/pCLAMP 8 (Molecular Devices). For proper comparisons with previous works from us [7,10,15,32] and others [33,34] using different slo1 isoforms, bilayer studies were conducted at room temperature (20–22 °C). PREG stock or vehicle control (DMSO) was added to the cis chamber to reach final concentration (0.01–30 μM), all dissolved in 0.1% DMSO; n = 3–4 for 0.001 and 1 μM PREG. These points were not used for comparative statistical analysis but included for the fitting of PREG inhibitory action; n = 5–10 for all other concentrations.

NPo was used as an index of channel steady-state activity, where N = number of channels in the bilayer (defined as a maximal number of opening levels) and Po = single channel open probability. Drug-induced NPo changes were determined by comparing “base” NPo (i.e., NPo with the bilayer perfused with bath solution) to the NPo from the same patch/cell exposed to agent (DMSO or PREG+DMSO) dissolved in bath solution. Concentration-NPo curve data obtained from the lipid bilayer experiments were fitted to a Boltzmann function of the type:

$$y=\frac{A1-A2}{1+e^{x-x0}}+A2$$

using Origin 2020 software (OriginLab; OriginLab Corporation).

2.6. Electrophysiology on cerebrovascular myocytes

SM cells were isolated from 8- to 12-week-old WT (C57BL/6) or Y450F KI on C57BL/6 J background male and female mouse MCAs as described [15,29]. Ionic currents were recorded in from excised, inside-out (I/O) patches clamped at +30 mV to evoke robust, yet sub-maximal currents, so the inhibitory effect of PREG could be better visualized. Bath and pipette electrode solutions contained (mM): 130 KCl, 5 EGTA, 1.6 HEDTA, 2.28 MgCl₂ ([Mg²⁺] free=1 mM), 15 HEPES; pH 7.40. In all experiments, free [Ca²⁺] in solution was adjusted as mentioned in Section 2.5. An agar bridge with Cl⁻ as main anion was used as ground electrode. Solutions were applied onto the cytosolic side of I/O patches using an automated, pressurized Octaflow system (ALA Scientific Instruments) through a micropipette tip with an internal diameter of 100 μm. Experiments were carried out at room temperature (20–22 °C). Ionic currents at single channel resolution were recorded using an EPC8 amplifier (HEKA) at 1 kHz. Data were digitized at 5 kHz using a Digidata 1320 A A/D converter and pCLAMP 8.0, and analyzed using Clampfit 10.7 (Molecular Devices). NPo was used as an index of channel steady-state activity and its modification by DMSO vs. PREG+DMSO was determined (please see Section 2.5). Within each group, data recordings were obtained from separate mice.

2.7. Cerebral artery diameter measurement ex vivo

WT (C57BL/6 J) and Y450F KI on C57BL/6 J background mice of both sexes were deeply anesthetized with isoflurane via inhalation in a tightly sealed jar. Resistance-size MCAs (~100 μm in external diameter) were dissected out of the brain after animal euthanasia with sharp scissors. For studies on de-endothelialized MCA, endothelium was removed by passing an air bubble into the vessel lumen for 90 s. This method has been consistently used by our group and validated using endothelium-dependent versus endothelium-independent vasodilators [8,35–39]. MCAs were cut into 5–10 mm-long segments under a microscope (Nikon SMZ645) and then each segment was cannulated at its ends, as previously described [8,36–39]. The MCA exterior was continuously perfused with physiologic sodium saline (PSS) of following composition (mM): 119 NaCl, 4.7 KCl, 1.2 KH₂PO₄, 1.6 CaCl₂, 1.2 MgSO₄, 0.023 EDTA, 11 glucose, 24 NaHCO₃, pH= 7.4, continuously bubbled with an O₂/CO₂/N₂ (21/5/74%) gas mixture, and maintained at 35–37°C. Artery external diameter was measured using the automatic edge-detection function of IonWizard (IonOptix) via a Sanyo VCB-3512T camera (Sanyo Electric Co.). Arteries were first incubated at an intravascular pressure of 10 mm Hg for 10 min. Then, intravascular pressure was increased to 60 mmHg and held steady throughout the experiment to induce development of myogenic tone and its maintenance [35,37]. Each MCA segment was exposed to PREG only once. Within each group, individual data points were obtained from separate mice and separate MCA segment. When required by experimental design, arterial contractility was probed with a high KCl solution of the following composition (mM): 63.7 NaCl, 60 KCl, 1.2 KH₂PO₄, 1.2 MgSO₄, 0.023 EDTA, 11 glucose, 24 NaHCO₃, 1.6 CaCl₂, pH= 7.4. Like PSS, the high-KCl solution was continuously bubbled with O₂/CO₂/N₂ (21/5/74%) gas mixture and maintained at 35–37°C. All solutions were applied at a constant rate of 3.75 ml/min using a peristaltic pump Dynamax RP-1 (Rainin Instrument, Inc.) [35–39].

2.8. Cerebral artery diameter measurement through cranial window in vivo

WT (C57BL/6) and Y450F KI on C57BL/6 background mice of both sexes were anesthetized with a mixture of xylazine/ketamine (12/100 mg/kg) and kept anesthetized for the duration of the experiment with subsequent ketamine doses (50 mg/kg of weight) as needed. A catheter was inserted into the internal carotid artery so that the infusion went directly to the brain, rather than towards the thoracic cavity. An area of the skull was cleared of tissue and thinned as previously described [10,15]. The exposed arteries were monitored using a Leica MC170 HD microscope with a mounted camera (Leica M125C) connected to a computer monitor. Drugs were diluted to their final concentration in sodium saline (0.9% NaCl) and administered via catheter at 0.1 ml/25 g of weight. Cranial window images before and after drug administration were acquired every 60 s for subsequent analysis; n = 5 for each group. Each n was obtained from a separate mouse. Changes in artery diameter obtained from cranial window experiments were determined using ImageJ software (ImageJ Downloaded from 9 1.52a, https://imagej.nih.gov/ij/download.html).

2.9. Chemicals

PREG was purchased from Abcam. POPE, and POPS were purchased from Avanti Polar Lipids. All other chemicals were purchased from Sigma Aldrich. PREG and was dissolved in dimethyl sulfoxide (DMSO) and diluted into saline, PSS solution, bath solution, or cis chamber-recording solution immediately before injection or application to the artery, myocyte membrane or bilayer, respectively. Each mouse, artery, myocyte membrane, or artificial lipid bilayer was exposed to PREG only once to avoid possible desensitization.

2.10. Design and statistical analysis

Statistical analysis was performed using InStat 3.05 (GraphPad) and Statistics Kingdom, a freely available to use website (https://www.statskingdom.com/). Unless otherwise stated, n represents the number of independent biological or experimental replicates rather than repeated readings from a same sample. For immunoprecipitation and binding studies, each measurement was performed using triplicate capillaries loaded with the same protein preparation. The triplicate capillaries were considered technical replicates and averaged to become a single data point before analysis. For planar lipid-bilayer recordings, each n represents an independently formed artificial lipid bilayer containing functional slo1 channels. For cerebrovascular myocyte electrophysiology, each n represents a recording obtained from a separate mouse. For ex vivo cerebral artery diameter measurements, each n represents a separate MCA segment obtained from a separate mouse. For the in vivo cranial-window experiments, each n represents a separate mouse. Each mouse, artery segment, myocyte membrane patch, or artificial lipid bilayer was exposed to PREG only once. When the number of observations in the groups under comparison exceeded 6, and the Gaussian distribution of the data was confirmed by the Kolmogorov-Smirnov test, analysis was performed using an unpaired Student's t-test. In all other cases, statistical analysis was conducted using the Mann-Whitney nonparametric test. For comparison of multiple experimental groups, SPSS statistics software was used (repeated measures ANOVA followed by Bonferroni correction). In all cases where we used two-tail t-test, significance set at P < 0.05. For all data, validity of the Hill or Boltzmann fit results were given by the corresponding reduced Chi-square and R-square values; the reduced Chi-square statistic was used to represent the goodness of fit testing [40] and the R-square results evaluated the scattering of the data points around the fitted regression line [41]. Data are expressed as means±S.E. Final plotting and fitting of data were conducted using the Origin 2020.

3. Results

3.1. Pregnenolone binds to slo1 proteins

Since BK channel-forming α subunits (slo1 channels) cloned from cerebral artery myocytes (cbv1 isoform; [28]) and expressed in a bare lipid environment are sufficient to sustain PREG-induced channel inhibition [15], we first investigated whether PREG bound to slo1 itself. For this purpose we chose microscale thermophoresis (MST), as this methodology is very sensitive to protein-drug interactions, allowing quantification of biomolecule interactions by the thermophoretic detection of time-dependent changes in conformation of a molecule as they are induced by a binding event [23]. We tested a concentration range of PREG that fully encompassed the concentrations used to construct concentration response curves (CRCs) to its inhibitory effect on native BK channels and its eventual constriction of cerebral arteries (0.1–100 μM; [15]). Thus, we measured changes in BK protein stability in a thermo-dependent manner in absence and presence of PREG. Comparison of PREG, i.e., slo1 in presence of 10 μM PREG-containing eluate, to control vehicle, i.e., slo1 in presence of dimethyl sulfoxide (DMSO)-containing eluate showed that PREG decreased the difference (Δ) in temperature between the start of thermal-induced unfolding (onset) and peak unfolding (inflection) events, which reflected destabilization of slo1 protein conformation by PREG and chemical binding of steroid to slo1 (Fig. 1A,B; SFig1). As experimental validation, the no-protein and boiled-protein control samples showed Δ values close to zero, indicating that the assay signal required intact Slo1 protein and was not generated in the absence of protein or after protein denaturation (SFig1). Moreover, the binding curve of PREG to purified slo1 was monotonic with an estimated K_D of 3.47 μM (Fig. 1C).

Fig. 1.

Pregnenolone binds to and alters the thermodynamic stability of slo1 (cbv1) proteins. A. Original trace records of the thermal unfolding of slo1 in the presence of vehicle control (DMSO; left) and PREG 10 μM (right). B. Averaged range of temperature between the onset and inflection point of slo1 thermal unfolding in the presence of PREG or vehicle control (DMSO). Individual recording points are indicated by hollow (DMSO) or black (PREG) diamonds. n = 6 triplicate capillary measurements per group. Here, and in all figures, deviation from averaged value is shown as standard error. C. Graphical depiction of PREG-driven, concentration-dependent changes in slo1 thermal unfolding, measured by the delta (Δ) of onset to inflection point. K_D: PREG dissociation constant obtained from Hill fit to the data points. *Statistically significant difference when compared to DMSO-containing eluate with [DMSO] being identical to that present in PREG-containing eluate (P = 0.004; two tail Mann Whitney U-test).

3.2. Pregnenolone inhibits slo1 channel activity in a concentration-dependent manner and in the absence of a native cellular environment

To determine whether PREG-slo1 biochemical binding could contribute to PREG electrophysiological effects, we examined the inhibitory action of PREG on channel activity in absence of a complex proteolipid environment. Thus, we applied PREG at a concentration range found to bind slo1 (Fig. 1C) to slo1 channels reconstituted into a 3:1 (w/w) 1-palmitoyl-2-oleoyl-glycero-3-phosphoethanolamine (POPE):1-palmitoyl-2-oleoyl-glycero-3-phosphoserine (POPS) planar bilayers, with trans and cis solutions containing 30 μM free [Ca²⁺]. Extending recent findings from our group [15], PREG within a few minutes of cis chamber bath application reduced slo1 steady-state activity (NPo) (Fig. 2A–C). This PREG action was monotonic, with EC50= 0.35 μM ± 0.058 and ECMax= 10.34 μM ± 0.05 (R-square=0.954; Reduced Chi-square=0.855). Results demonstrate that PREG reduces slo1 channel activity in the absence of the heterogenous proteolipid domains that exist in natural membranes around BK-forming proteins.

Fig. 2.

Pregnenolone inhibits the activity of homomeric slo1 channels in a concentration-dependent manner, in absence of the native cell environment. A. Representative slo1 channel activity recordings obtained after incorporation of slo1 (cbv1 isoform) into POPE/POPS (3:1 wt/wt) planar bilayers. Traces are displayed vertically in pairs with the top trace being baseline recordings corresponding to each bottom trace, after increasing concentrations of PREG (1nM-100μM) were added to the cis chambers. Channel openings are shown as upward deflections; black arrows indicate the baseline. Bilayers were bathed by recording solutions containing 30 μM free calcium; all records were obtained at +30 mV. Base conditions correspond to the channel activity in the presence of vehicle control used to dissolve PREG (i.e., DMSO) in bath solution. B. Averaged data showing NPo/NPo_base for each minute of recording, comparing each concentration of PREG to vehicle (DMSO) time-matched controls. A black arrow indicates the time at which PREG or DMSO was added. A horizontal dashed line at 1 (y-axis) indicates no change from baseline activity. Data were analyzed by repeated-measures ANOVA with time as the within-subject factor and treatment groups as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons; *p < 0.05, wt cbv1 10 μM PREG vs. vehicle control (DMSO). C. Comparison of the effects of each concentration of PREG (1 nM-100 μM) compared to vehicle control (DMSO). Black diamonds indicate individual points for PREG at each concentration; hollow circles indicate individual points for DMSO. A horizontal dashed line at 1 (y-axis) indicates no change from baseline activity. For this data, we used the responses evoked by PREG 6 min after initiating its application to the bilayer chamber highlighted by the red box around 6 min in panel B. n = 3–9 for each group; no group with less than 5 recordings was used in statistical analysis. *Statistically significant difference from each concentration of PREG (1 nM-100 μM) compared to vehicle control DMSO (p < 0.05, one-way ANOVA followed by Tukey’s post hoc test).

3.3. Pregnenolone-induced inhibition and binding require the full slo1 protein

The recording conditions and system used to examine PREG inhibition of BK channels indicate that slo1 proteins in a bare lipid environment suffice to sustain PREG action (Fig. 2A–C). Since PREG is a hydrophobic compound, we tested whether the transmembrane core of slo1 (S0-S6) was sufficient for PREG-induced channel inhibition. Thus, we applied 10 μM PREG on a truncated slo1 construct made with a stop codon at residue 323 (trS6; Fig. 3A). Following reconstitution into the same bilayer type and evaluation under conditions identical to those used to probe PREG on full-length slo1, trS6 channels characteristically displayed a unitary current amplitude much lower than that of their full-length counterparts (Fig. 3B) as previously reported [30]. The “BK” identity of the channels under study, however, was confirmed by their blockade in presence of 1 μM paxilline (Fig. 3C). More important, trS6 channels were much less sensitive to PREG inhibitory action than their full-length counterparts, with the maximum inhibitory effect of PREG on full-length slo1 being drastically larger than that on trS6: 66.1 ± 0.93% and 3.3 ± 0.81% inhibition, respectively (P < 0.05 by factor:10 μM PREG and P < 0.01 by factor:trS6; Fig. 3B–D). Matching trS6 electrophysiological refractoriness, MST measurements on trS6 showed no destabilization or changes in the temperature between the start of thermal-induced unfolding delta (Δ) induced by PREG when compared to those of vehicle controls (SFig. 2). Moreover, MST negative outcome also applied to PREG interaction with the isolated CTD (SFig. 2). Collectively, results indicate that the full slo1 structure is required for PREG binding and eventual contribution to inhibition of BK channel activity.

Fig. 3.

The slo1 cytosolic tail domain (CTD) confers BK channel sensitivity to pregnenolone-evoked channel inhibition. A. Cartoon representation of a single BK channel (slo1) pore-forming subunit. A truncation between Ile-322 and Ile-323 removes the CTD to render the construct ‘trS6’, indicated by the red arrow. B. Representative slo1 (cbv1 isoform) and trS6 channel recordings obtained after channel incorporation into POPE/POPS (3:1 wt/wt) bilayers. Top traces are baseline recordings corresponding to each construct. Bottom traces are representative of channel activity immediately after PREG (10 μM) was added to the bilayer cis chamber. The lower trace presents the channel activity in the truncated construct immediately after paxilline (1 μM) was added. Channel openings are shown as upward deflections; black arrows indicate the baseline. Bilayers were bathed by recording solutions that contained 30 μM free calcium; all records were obtained at +30 mV. C. Averaged data showing NPo/NPo_base for each minute of recording, comparing full-length cbv1 to trS6 construct in presence of either PREG (10 μM) or vehicle (DMSO), time-matched controls. A black arrow indicates the time at which PREG/vehicle control was added. A red box indicates when the BK channel blocker paxilline (1 μM) was added. A horizontal dashed line at 1 (y-axis) indicates no change from baseline activity. Data was analyzed by repeated-measures ANOVA with time as the within-subject factor and constructs/treatment groups as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. *p < 0.05, wt cbv1 PREG vs. trs6 PREG. n = 9–10 independent bilayer recordings per group. D. Comparison of the maximal effects of PREG (10 μM) or vehicle control (DMSO) for cbv1 and trS6. Hollow diamonds indicate individual points for DMSO-treated cbv1. Black diamonds indicate individual points for PREG-treated cbv1. Black squares indicate individual points for PREG-treated trS6 while hollow squares indicate individual records for DMSO-treated trS6. For this data, we used the responses evoked by PREG 6 min after initiating its application to the bilayer chamber. *Statistically significant difference compared: **p < 0.05 by two-way ANOVA (factor: 10 μM PREG treatment); p < 0.01 by two-way ANOVA (factor: trS6).

3.4. Pregnenolone binds to Tyr450 in the slo1 protein CTD

Since PREG-induced inhibitory effect and binding to BK α subunits were lost in both the CTD-missing trS6 and the isolated CTD (Fig. 3 and SFig. 2) and a recognition binding area for the PREG precursor cholesterol includes Tyr450 [42,43], we performed computational modeling to dock ten energetically stable conformations of PREG onto all atoms of the slo1 CTD in the vicinity of Tyr450. Indeed, PREG binding within a defined slo1 pocket involved Tyr450 via a H-arene interaction with the 16th carbon of the steroid backbone, in addition to H-bonding with His409, and acidic receptor-ligand contouring by Asp896 (Fig. 4A). Next, these three residues were substituted with either alanine (H409A, D896A) or phenylalanine (Y450F) and used to determine the effect of these substitutions on PREG binding (Fig. 4B,C). In contrast to data from their wt counterparts, protein conformation in H409A and Y450F point slo1 mutants in the presence of PREG (10 μM) were similar to those from the corresponding vehicle controls. In turn, D896A showed a small PREG associated increase in Δ; this change follows the opposite direction of that evoked by PREG in wt, and occurred against a markedly reduced vehicle control DMSO Δ value, suggesting that the D896A substitution itself markedly alters basal slo1 thermal unfolding. These data reflect the loss of PREG binding to the CTD of mutated slo1 subunits (Fig. 4C) and experimentally validate our PREG-slo1 CTD docking model.

Fig. 4.

Pregnenolone and its residence sites within the slo1 BK channel CTD. A. Representative snapshot of the most energetically favorable docking motif of PREG (pink) on the slo1 channel homotetramer. Each slo1 subunit is color-coded to depict the interaction interactions within the binding pocket and the corresponding subunits. Close-up depiction of a theoretical binding pocket for PREG, depicting hydrogen-arene interactions with Tyr450 and the 16th carbon of the PREG backbone, hydrogen donor from His409, with Asp896 providing electron exposure. B. Original trace records of the thermal unfolding of the point mutation Y450F in the presence of vehicle control (DMSO; left) and PREG (right). C. Comparison of PREG-driven changes in average (delta) range of temperature between the onset and inflection point of the thermal unfolding of slo1 and each point mutation construct (H409A, D896A, Y450F) in the presence of vehicle control (DMSO) and 10 μM PREG. Black diamonds represent individual recordings; n = 6–19 triplicate capillary measurements per group. *Statistically significant difference of each construct in the presence of vehicle control (DMSO) compared to in presence of 10 μM PREG (*p < 0.05; two tail Mann Whitney U-test).

3.5. Tyr to Phe substitution at position 450 in slo1 blunts both pregnenolone-induced channel inhibition and eventual cerebral artery ex vivo and in vivo constrictions

After demonstrating the binding of PREG to the slo1 CTD and identifying the key role of Tyr450 in PREG binding to BK subunits using both computational modeling and MST, we examined any possible contribution of this residue in PREG-induced inhibition of native, cerebrovascular smooth muscle (SM) BK channels. When 10 μM PREG was perfused over the cytosolic side of excised, I/O patches from MCA myocytes of Y450F KI mice of both sexes, there was no significant difference in channel activity when compared to the basal channel activity of vehicle-containing (DMSO) bath solution-perfused, time-matched, controls (Fig. 5A–C). We next validated the differential PREG responses of BK channels from WT vs. Y450F KI mice by examining PREG action in WT and KI animals from the same parents thus ensuring a same genetic background, age, and in utero environment: indeed, 10 μM PREG applied to MCA myocyte I/O patches from WT mice mildly decreased BK channel activity in WT mice and failed to do so when applied to their Y450F KI littermates (SFig. 3B). Collectively, results highlight a key role for a direct interaction between PREG and slo1 Tyr450 for this steroid to inhibit the activity of cerebrovascular BK channels in their native SM membrane.

Fig. 5.

Pregnenolone fails to inhibit the activity of native, cerebrovascular smooth muscle BK channels upon substitution of slo1 Tyr450 by Phe. A. Representative BK current records obtained in I/O patches excised from membranes of Y450F knockin (KI) male mouse middle cerebral artery (MCA) myocytes during bath perfusion (base; top left) and 10 μM PREG perfusion (bottom left). Horizontal dashed lines and black arrows indicate the baseline. Channel openings are shown as downward deflections. Here and in B-C, all records were recorded at +30 mV. B. Representative BK current records obtained in I/O patches excised from membranes of Y450F KI female mouse MCA myocytes during bath perfusion (base; top right) and 10 μM PREG perfusion (bottom right). Horizontal dashed lines and black arrows indicate the baseline. Channel openings are shown as downward deflections. C. Averaged data showing NPo/NPo_basal for each minute of recording and comparing the effect of 10 μM PREG on patches from MCA myocytes of male and female Y450F mice with the time-matched vehicle control (DMSO) records. Control records are represented as hollow squares and triangles for male and female, respectively. Averaged PREG records are represented by black squares and triangles for male and female, respectively; n = 4–5 for each group; no statistical analysis was performed on any group with less than five observations (DMSO Female 19–23 and PREG Female 20–23). Black arrows indicate changes in perfusion. A horizontal dashed line at y = 1 underscores no change from baseline. Individual data recordings were obtained from separate mice for each group. Statistical analysis was performed using repeated-measures ANOVA with time as the within-subject factor and treatment groups as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. No significant difference was detected between groups.

Next, we examined the importance of Tyr450 in PREG-induced MCA constriction, a steroid effect that is endothelium-independent but dependent on SM BK channel inhibition [15]. Remarkably, at any concentration tested (0.001–100 μM) PREG failed to constrict de-endothelialized MCA in Y450F KI mice of both sexes, a concentration-dependent action of PREG that was observed in non-littermate WT controls (Fig. 6A–C). Of note, vehicle control (DMSO) alone did not significantly alter the diameter of de-endothelialized MCAs, and the responses to vehicle control did not significantly differ between WT and Y450F KI MCAs in both sexes (Fig. 6B–C). In male WT arteries, PREG-induced constriction was significantly different from the DMSO control at all concentrations of PREG. In male Y450F KI arteries, the response to the steroid was significantly different only at 10 μM PREG. In female MCAs, PREG-induced changes differed significantly only at 10 μM PREG in both WT and Y450F KI mice. As done when evaluating PREG actions on BK activity, the MCA diameter differential responses were also validated in WT vs. Y450F KI littermates: indeed, 10 μM PREG mildly but significantly decreased MCA diameter in WT mice while failing to do so in their Y450F KI littermates (SFig. 3D). Moreover, 10 μM PREG-induced constriction of MCA in WT animals reversed to mild dilation in Y450F mice (Fig. 6A–C). This finding is consistent with published data from Kcnma1^−/− mice (i.e., slo1-missing) showing that PREG acting on classic steroid receptors results in mild MCA dilation [15]. Therefore, current data demonstrate that the Y450F substitution ablates the endothelium-independent MCA constriction evoked by a direct PREG-slo1 channel interaction.

Fig. 6.

Pregnenolone-induced cerebral artery ex vivo constriction is ablated upon substituting slo1 Tyr450 with Phe. A. Diameter trace showing the response to the application of 10 μM PREG to a de-endothelialized pressurized middle cerebral artery (MCA) segment from C57BL/6 J wt and Y450F knock in (KI) male mouse. The arrow points at the time when 10 μM PREG was introduced into the bath, and the extension of the horizontal line indicates duration of PREG in the bath before being washed out. B. PREG-driven percent changes in de-endothelialized MCA of male Y450F KI mouse, represented by grey bars, compared to PREG-driven percent changes in de-endothelialized MCA of male C57BL/6 J mouse, represented by hollow bars. Vehicle control DMSO-driven percent change in wt and Y450KI MCA are also shown in the bar plots. Each hollow square symbol indicates individual points; n = 5 or 6 for each concentration C. PREG-driven percent changes in de-endothelialized MCA of female Y450F KI mouse, represented by grey bars, compared to PREG-driven percent changes in de-endothelialized MCA of female C57BL/6 J mouse, represented by hollow bars. Vehicle control DMSO-driven percent change in wt and Y450KI MCA are also shown in the bar plots. Each hollow diamond indicates individual points; n = 5 or 6 for each concentration for Y450F KI female mouse records. For all data shown in this figure, each record was obtained from a separate arterial segment. Within each group, individual data points were obtained from separate mice. *Statistically significant difference compared to genotype (p < 0.05; two-tail Mann Whitney U-test). Statistically significant difference compared to DMSO vehicle control (^#p < 0.05, ^##p < 0.01; one-way ANOVA followed by Tukey’s HSD post-hoc test).

To determine whether the effect of the Y450F substitution on PREG-induced ex vivo vasoconstriction remains when evaluated in vivo, we applied a bolus infusion of 10 μM PREG on anesthetized, Y450F KI mice of both sexes. Consistent with the ex vivo MCA diameter results from Y450F KI mice, the PREG-induced in vivo constriction previously reported in C57BL/6 J mice of both sexes [15] was lost (Fig. 7A–D). These data underscore that circulating, local or other regulatory factors that remain functional in vivo cannot override the key importance of Tyr450 for PREG to constrict cerebral arteries via direct interaction with SM BK channels.

Fig. 7.

Substitution of slo1 Tyr450 to Phe suppresses pregnenolone-induced cerebral artery in vivo dilation while rendering mild vasodilation. A. Cranial window images of male Y450F knockin (KI) mouse pial arteries branching from MCA immediately prior to drug infusion (3 min) and 3 min after 10 μM PREG infusion via catheterized carotid artery (6 min). Here and in C, dashed lines are positioned along external cerebral artery edges to highlight diameter changes by PREG. B. Averaged fold-changes in male cerebral artery diameter for each minute of recording compared to baseline diameter determined from the image taken immediately prior to any infusion. A horizontal dashed line points at 1 (y-axis), i.e., point of no change in artery diameter. A red arrow indicates the beginning of infusion; n = 4–5 for each group, each n was obtained from a separate mouse (C57/BL6J data adapted from [15]). Data analyzed by repeated-measures ANOVA with time as the within-subject factor and treatment/genotype group as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. *p < 0.05, Y450 KI 10 μM PREG vs C57BL/6 10 μM PREG. C. Cranial window images of femaleY450F KI mouse (top) pial arteries branching from the MCA immediately prior to drug infusion (3 min) and 3 min after 10 μM PREG infusion via catheterized carotid artery (6 min). D. Averaged fold-changes in female cerebral artery diameter for each minute of recording compared to baseline diameter determined from the image taken immediately prior to any infusion. A horizontal dashed line points at 1 (y-axis), i.e., point of no change in artery diameter. A red arrow indicates the beginning of infusion; n = 4–5 for each group, each n was obtained from a separate mouse (C57/BL6J data adapted from [15]). Data analyzed by repeated-measures ANOVA with time as the within-subject factor and treatment/genotype group as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. *p < 0.05, Y450 KI 10 μM PREG vs C57BL/6 10 μM PREG.

3.6. Pregnenolone’s mechanism of action

Representative snapshots of the most energetically favorable docking motif of PREG on the slo1 homotetramer revealed that the steroid binding pocket is in close vicinity to the high-affinity Ca²⁺-recognition site located within the RCK2 domain of the same slo1 subunit, with the Ca²⁺-recognition site within RCK1 being nearby positioned on the slo1 neighboring subunit (Fig. 8A). Given the vicinity of the PREG-sensing site and the high-affinity Ca²⁺ sites involved in slo1 channel activation, we next examined the inhibitory action of PREG (10 μM) on reconstituted slo1 channels under voltages spanning 0–60 mV and Ca²⁺ concentration= 0.05 μM, that is, below the activatory threshold for slo1-sensing of this divalent via its high affinity binding sites. Under these conditions, slo1 channel activity is largely independent of ${\mathrm{C}\mathrm{a}}_{\mathrm{i}}^{2+}$ and primarily driven by intrinsic/voltage-gating [44]. Data show that PREG inhibitory action, while robust at 30 μM ${\mathrm{C}\mathrm{a}}_{\mathrm{i}}^{2+}$, was lost in the presence of subμM Ca²⁺ (Fig. 8C,D), underscoring that PREG inhibitory action is coupled to Ca²⁺-activation of the channel. Fig. 8B shows the docking of cholesterol in the vicinity of the PREG-slo1 Tyr450 region. Remarkably, it has been recently shown that cholesterol inhibition of slo1 channels is also dependent on ${\mathrm{C}\mathrm{a}}_{\mathrm{i}}^{2+}$-gating [42] and upon interaction with Tyr450 [43]. Consistent with competition of two steroidal ligands on a common docking region and targeting of a common gating-mechanism, PREG-induced inhibition of slo1 channel activity was blunted (but not fully ablated; see Discussion) by the addition of 33 mol% cholesterol to the lipid bilayer expressing slo1 channels (SFig. 4 A,B).

Fig. 8.

Pregnenolone-induced inhibition of slow channels is coupled to calcium-gating. A. Representative snapshot of the most energetically favorable docking motif of PREG (pink) on the slo1 homotetramer. Each subunit is color-coded to depict the interaction interactions within the binding pocket and the corresponding subunits. High-affinity calcium recognition sites are identified by bolus, orange-coded atoms. RCK2 is depicted closest to the PREG binding pocket (left); RCK1 is positioned on the neighboring subunit (right). B. Representative snapshot of cholesterol (orange) docking onto the slo1 homotetramer. Each subunit is color-coded to depict the interactions within the binding pocket and the corresponding subunits. The RCK2 high-affinity calcium recognition site is identified by bolus, orange-coded atoms. C. Averaged data showing NPo/NPobase for each minute of recording, comparing the inhibition elicited by PREG (10 μM) on full-length cbv1 in presence of activatory (30 μM) free Ca²⁺ and subactivatory (0.05 μM) free Ca²⁺ in both cis and trans bilayer chambers (red and black; respectively). Averaged data showing NPo/NPobase for each minute of recording, comparing the effect of vehicle control DMSO on full length cbv1 in presence of 30 μM free Ca²⁺ (gray). A horizontal dashed line at 1 (y-axis) indicates no change from baseline activity. Data were analyzed by repeated-measures ANOVA with time as the within-subject factor and treatment groups as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons; *p < 0.05, 30 vs. 0.05 μM free ${\mathrm{C}\mathrm{a}}_{\mathrm{i}}^{2+}$, both cond_itions in presence of 10 μM PREG; n = 5–10 for each. D. Comparison of the maximal effects of vehicle control for cbv1 in the presence of 30 μM free Ca²⁺ and PREG (10 μM) for cbv1 in the presence of 30 μM free Ca²⁺ and 0.05 μM free Ca²⁺. Each symbol indicates individual records. *Statistically significant difference compared to 10 μM PREG with 30 μM free Ca²⁺ (p < 0.05; one-way ANOVA followed by Tukey’s post hoc multiple comparison test).

Lastly, we examined the inhibition by PREG on mutations impairing each high-affinity Ca²⁺ recognition site: the RCK1 (D362A,D367A) and the RCK2 Ca²⁺ bowl (D899N,D900N,D901N,D902N,D903N) sites, and the combination of nullifying both high-affinity sites (D362A,D367A, D899N,D900N,D901N,D902N, D903N). Consistent with the idea that Ca²⁺ binding to its high-affinity sites in the slo1 CTD is needed for PREG action, PREG-induced channel inhibition, demonstrated under identical conditions with WT slo1 channels (Fig. 1C), was lost when both high-affinity Ca²⁺ sites were mutated (Fig. 9A,B). We next probed the steroid in individual Ca²⁺-site mutants. Remarkably, the RCK2 mutant (i.e., when Ca²⁺-activation was driven by RCK1) supported PREG-induced reduction of NPo (Fig. 9C,D). In contrast, the RCK1 mutant (i.e., when Ca²⁺-activation was driven by RCK2) was not sensitive to PREG-induced inhibitory action (Fig. 9E,F). Therefore, PREG inhibition of slo1 channels is allosterically coupled to Ca²⁺-recognition by the high-affinity Ca²⁺-binding site found in RCK1.

Fig. 9.

Pregnenolone-induced inhibition is driven by either/the RCK(1) high-affinity Ca²⁺-recognition site(s). A. Example records of the vehicle control (DMSO; top) vs. 10 μM PREG (bottom) on the double high-affinity mutant construct (D362A,D367A,D899N,D900N,D901N,D902N, D903N) at 60 mV. Channel openings are shown as upward deflections; black arrows indicate the baseline. B. Averaged unitary current amplitude values in DMSO-containing versus PREG-containing bilayers at varying transmembrane voltages at 30 μM free Ca²⁺ (left). Averaged data showing NPo/NPo_base as a function of transmembrane voltage in DMSO-containing versus PREG-containing bilayers at 30 μM free Ca²⁺ (right); data are shown as mean±SEM; n = 5 for each group. Data analyzed by repeated-measures ANOVA with voltage as the within-subject factor and treatment as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. *p < 0.05, vehicle control DMSO vs. 10 μM PREG. C. Example records of the vehicle control (DMSO; top) vs. 10 μM PREG (bottom) on the RCK2 high-affinity mutant construct (D899N, D900N,D901N,D902N, D903N) at 30 mV. Channel openings are shown as upward deflections; black arrows indicate the baseline. D. Averaged unitary current amplitude values in DMSO-containing versus PREG-containing bilayers at varying transmembrane voltages at 30 μM free Ca²⁺ (left). Averaged data showing NPo/NPo_base as a function of transmembrane voltage in DMSO-containing versus PREG-containing bilayers at 30 μM free Ca²⁺ (right); data are shown as mean±SEM; n = 5 for each group. Data analyzed by repeated-measures ANOVA with voltage as the within-subject factor and treatment as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons. *p < 0.05, vehicle control DMSO vs. 10 μM PREG. E. Example records of the vehicle control (DMSO; top) vs. 10 μM PREG (bottom) on the RCK1 high-affinity mutant construct (D362A,D367A) at 30 mV. Channel openings are shown as upward deflections; black arrows indicate the baseline. F. Averaged unitary current amplitude values in DMSO-containing versus PREG-containing bilayers at varying transmembrane voltages at 30 μM free Ca²⁺ (left). Averaged data showing NPo/NPo_base as a function of transmembrane voltage in DMSO-containing versus PREG-containing bilayers at 30 μM free Ca²⁺ (right); data are shown as mean±SEM; n = 5 for each group. Data analyzed by repeated-measures ANOVA with voltage as the within-subject factor and treatment as the between-subject factor, followed by Bonferroni-corrected pairwise comparisons.

4. Discussion

Our study demonstrates that Tyr450 in the BK slo1 protein CTD is essential for PREG-induced BK channel inhibition and subsequent MCA constriction whether evaluated ex vivo or in vivo. Using computational modeling, we first identified His409, Asp896 and Tyr450 for PREG docking onto the slo1 CTD and advanced a key role for a H-arene interaction between this residue and C16 of the PREG backbone for stable docking of this ligand (Fig. 4). Results from MST and lipid bilayer electrophysiology confirmed the key role of this residue in both PREG binding and direct inhibition of channel activity (Figs. 4,5). Remarkably, the K_D determined by MST (Fig. 1C) is in the same order of magnitude (low μM) as the PREG concentrations that are effective to exert both electrophysiological and MCA diameter actions, as documented in our recent [15] and present study (Figs. 5–7). Moreover, the K_D for PREG determined by MST is almost identical to the IC₅₀ for PREG inhibition of BK channels: 3.47 vs. 3 μM, respectively (Fig. 1A and [15]). This underscores the association between our biochemical and functional results. More important, the Y450F substitution completely ablated ex vivo and in vivo MCA constrictions evoked by physiological/therapeutic concentrations of PREG, which highlights the essential role of the Tyr450-PREG direct interaction in steroid modulation of cerebrovascular diameter (Figs. 6,7 and SFig. 3D).

For binding studies, we chose MST because this methodology is highly sensitive to slight changes in conformation and protein stability, known biomarkers of ligand-protein binding events [23]. Protein conformations during MST measurements rely heavily on the pH and the presence of hydrophobic compounds [23], such as DMSO (vehicle control) and PREG itself. Interestingly, PREG binding to the CTD requires the slo1 hydrophobic core domain (S0-S6), suggesting that this core is critically important for slo1 conformation and eventual PREG recognition, even though trS6 (the construct consisting of S0-S6) itself neither binds PREG nor is PREG-sensitive (SFig. 2 and Fig. 3B–D). Conceivably, a hydrophobic compound such as PREG would have a greater affinity (e.g., vs. DMSO) to the hydrophobic transmembrane domains that conform trS6. Thus, PREG inefficiency to bind and reduce the channel activity is, indeed, remarkable; these results further underscore the key role of a cytosolic residue (Tyr450) for PREG binding and inhibition of channel activity (Figs. 4,5).

Present findings demonstrating that Tyr450 in slo1 is a key contributor to the inhibitory action of PREG on BK channels drastically contrast with all previous studies dealing with direct interactions between vasoactive steroids and BK channels. First, PREG-BK interaction results in decreased channel activity while increased activity results from channel exposure to 17 β-estradiol, dihydroepiandrosterone, progesterone, or bile acids and related cholanes [6,7,9,10,45]. Second, β subunits embolden native BK channels with sensitivity to circulating vasoactive steroids other than PREG, which does not need these subunits to modulate BK activity (present data and [15]). Moreover, actual docking sites for cholanes, 17 β-estradiol and, more recently, progesterone, have been identified and characterized in BK regulatory β subunits [6,10,45,46]. Third, 17 β-estradiol’s mechanism of action has been linked to a shift in the voltage sensor from resting to active states and/or increased allosteric coupling between voltage-driven and intrinsic (i.e., in absence of activating voltage/Ca²⁺) gating [6]. In contrast, our data demonstrate that when the channel is probed in absence of Ca²⁺-driven channel activation, PREG becomes inefficacious (Fig. 8C,D). Moreover, data from slo1 channels where both RCKs have been rendered nonfunctional by mutagenesis show PREG becoming inefficacious (Fig. 9A,B), yet disruption of RCK2 sustains PREG action (Fig. 9C–F). Of note, the PREG-sensing site identified in our study does not topologically overlap with either RCK1 or RCK2 (Fig. 8A). Collectively, we can conclude that PREG and Ca²⁺ are heterotropic ligands on slo1, and PREG action is allosterically coupled to Ca²⁺-sensing by the RCK1 Ca²⁺-high affinity site in the slo1 CTD. Unexpectedly, voltage-sensing seems to play a role when PREG-induced channel inhibition is driven by Ca²⁺-sensing by RCK2 (Fig. 9D). It is possible to speculate that PREG is primarily coupled to the channel X-X_Ca transition at any RCKs and thus could Ca²⁺-dependent allosteric coupling parameters as described in the Horrigan-Aldrich gating model [44] while allosteric coupling between Ca²⁺-binding and voltage sensor activation would play a significant role only for RCK2 involvement in PREG-induced channel inhibition.

Tyr450 has been found to be important for cholesterol docking and chemical binding onto slo1 channels and for a partial contribution to cholesterol inhibition of channel activity, which is also Ca²⁺-dependent. From different slo1 constructs, Singh et al. [30] have interpreted that cholesterol action is related to Tyr450 functioning as the central Tyr that is characteristic of all cholesterol recognition amino acid consensus (CRAC) motifs [47,48]. Moreover, recent data have documented that Tyr450Phe slo1 constructs fail to bind cholesterol while their resulting channel activity is reduced by cholesterol in a significantly reduced degree when compared to their WT counterparts [43]. Thus, present results raise the hypothesis that different steroids acting upon recognition with Tyr450 may inhibit channel activity upon allosteric coupling to Ca²⁺-driven gating. Cholesterol and PREG do share a similar steroid backbone, yet slo1 channel inhibition by cholesterol via CRAC involvement is extremely specific: it is supported by neither cholesterol’s 3β isomer nor enantiomer-cholesterol [32]. Our computational data reveal that the interaction of Tyr450 with PREG vs. cholesterol is significantly different: Tyr450 directly interacts with C16 of the PREG backbone, whereas this residue interacts with cholesterol side chain via Van der Waals forces, with Ser474 serving as a sidechain acceptor of cholesterol’s hydroxyl, and Lys453 being the residue that primarily interacts with the steroid backbone (Fig. 8A–B). Moreover, in cholesterol-containing bilayers PREG-induced inhibition of slo1 channels is reduced but not totally ablated, consistent with the fact that these two steroids do not share identical binding sites. Noteworthy, our model shows that PREG establishes Van der Waals interactions with Asp896, which has a very close proximity to the high-affinity “Ca²⁺ bowl” (Asp899,900,901,902,903) (Fig. 8A). This site is important for the regulation of both Ca²⁺ activation and cholesterol inhibition of slo1 [42, 49].

The MCA diameter ex vivo and in vivo responses to PREG reported here (Figs. 6,7) are consistent with previously published data from Kcnma1^−/− mice showing that PREG, in absence of BK channels, evokes mild vasodilation, an action that involves canonical steroid receptors [15]. However, in biological systems expressing BK channels, PREG does cause vasoconstriction, which underscores that the Tyr450-PREG direct interaction unveiled here prevails over steroid receptor-mediated actions of PREG. Given the critical role of cerebral artery function in brain physiology, memory and cognitive function, the direct action of PREG on BK channels and its consequence on cerebrovascular diameter in our study should be taken into account when analyzing the effects of PREG levels on cognitive dysfunction and memory disorders [11–14,20,21,50,51], as well as in PREG-driven therapeutic intervention in substance use disorders and Alzheimer’s/dementia. PREG-induced cerebral artery constriction could potentially be considered as a contraindication for those who may be at a greater risk of cerebral ischemia or stroke, such as during vascular dementia or alcohol use disorders [16–22,52,53–55]. The increased resolution on direct PREG-BK channel interaction brought by MST binding assays and computation docking supplemented with functional validation in our current study allow for a better understanding of how steroids, their precursors and, eventually, future synthetic analogs, can be used to treat prevalent brain and vascular disorders.

Abbreviations:

BK

Ca²⁺-/voltage-gated potassium channel of large conductance

CHO

Chinese hamster ovary

CRAC

cholesterol recognition amino acid consensus

CTD

cytosolic tail domain

DMSO

dimethyl sulfoxide

MST

microscale thermophoresis

PREG

pregnenolone

RCK

regulator of conductance for potassium

SM

smooth muscle

TMs

transmembrane domains

TrS6

construct truncated at serine 323 of the sixth transmembrane domain

Y450F KI

Kcnma1-Tyr450Phe knockin

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.phrs.2026.108409.

Data availability

Data will be made available on request.