In most mammalian tissues, Ca2+i/voltage-gated, large conductance K+ (BK) channels consist of channel-forming slo1 and auxiliary (β1–β4) subunits. When Ca2+i (3–20 µM) reaches the vicinity of BK channels and increases their activity at physiological voltages, β1- and β4-containing BK channels are, respectively, inhibited and potentiated by intoxicating levels of ethanol (50 mM). Previous studies using different slo1s, lipid environments, and Ca2+i concentrations—all determinants of the BK response to ethanol—made it impossible to determine the specific contribution of β subunits to ethanol action on BK activity. Furthermore, these studies measured ethanol action on ionic current under a limited range of stimuli, rendering no information on the gating processes targeted by alcohol and their regulation by βs. Here, we used identical experimental conditions to obtain single-channel and macroscopic currents of the same slo1 channel (“cbv1” from rat cerebral artery myocytes) in the presence and absence of 50 mM ethanol. First, we assessed the role five different β subunits (1,2,2-IR, 3-variant d, and 4) in ethanol action on channel function. Thus, two phenotypes were identified: (1) ethanol potentiated cbv1-, cbv1+β3-, and cbv1+β4-mediated currents at low Ca2+i while inhibiting current at high Ca2+i, the potentiation–inhibition crossover occurring at 20 µM Ca2+i; (2) for cbv1+β1, cbv1+wt β2, and cbv1+β2-IR, this crossover was shifted to ∼3 µM Ca2+i. Second, applying Horrigan–Aldrich gating analysis on both phenotypes, we show that ethanol fails to modify intrinsic gating and the voltage-dependent parameters under examination. For cbv1, however, ethanol (a) drastically increases the channel’s apparent Ca2+ affinity (nine-times decrease in Kd) and (b) very mildly decreases allosteric coupling between Ca2+ binding and channel opening (C). The decreased Kd leads to increased channel activity. For cbv1+β1, ethanol (a) also decreases Kd, yet this decrease (two times) is much smaller than that of cbv1; (b) reduces C; and (c) decreases coupling between Ca2+ binding and voltage sensing (parameter E). Decreased allosteric coupling leads to diminished BK activity. Thus, we have identified critical gating modifications that lead to the differential actions of ethanol on slo1 with and without different β subunits.
Large conductance, Ca2+- and voltage-gated K+ (BK) channels are widely expressed in both excitable and nonexcitable tissues and thus play a critical role in several physiological processes, such as neurotransmitter release, hormone secretion, circadian rhythms, inflammatory responses, metastasis progression, and regulation of smooth muscle tone (Weiger et al., 2002; Lu et al., 2006; Salkoff et al., 2006; Dopico et al., 2012). Functional BK channels result from tetrameric association of channel-forming α subunits (encoded by the Slo1 or KCNMA1 gene), which are often referred to as slo1 channels. Each slo1 subunit consists of a transmembrane “core” (S0–S6) and a long cytosolic domain (CTD). In most mammalian tissues, however, slo1 proteins are associated with auxiliary β1–β4 subunits (encoded by KCNMB1–4; Knaus et al., 1994; Wallner et al., 1999; Xia et al., 1999; Behrens et al., 2000; Brenner et al., 2000a; Uebele et al., 2000). Unlike α subunits, BK β subunits do not form functional channels themselves. Rather, they usually alter several gating processes, which leads to modification of the channel’s response to physiological ligands, such as Ca2+i, and/or changes in macroscopic current kinetics (McManus et al., 1995; Brenner et al., 2000a; Cox and Aldrich, 2000; Nimigean and Magleby, 2000; Qian and Magleby, 2003; Orio and Latorre, 2005). For example, β1 subunit–induced changes in gating include modifications in intrinsic gating (i.e., gating in the absence of activating Ca2+i/voltage; Bao and Cox, 2005; Gruslova et al., 2012), stabilization of the slo1 voltage sensor in the active configuration (Bao and Cox, 2005; Contreras et al., 2012; Castillo et al., 2015), increased affinity for Ca2+ in the RCK1 domain high-affinity Ca2+-sensing site when the channel is open, decreased affinity for Ca2+ in the RCK2 domain Ca2+-sensing site when the channel is closed (Bao and Cox, 2005; Sweet and Cox, 2009), and increased allosteric coupling between Ca2+ binding and channel opening (reviewed in Hoshi et al. ) while not modifying equivalent gating charge (Bao and Cox, 2005; Contreras et al., 2012; Castillo et al., 2015). The β1-dependent gating modifications lead to a significant increase in the channel’s apparent binding affinity for Ca2+i, which is evident by a decrease in the voltage needed for reaching half-value in the conductance (G)/maximal conductance plot (V0.5) when ionic current behavior is evaluated at ≥1 µM Ca2+i (Cox and Aldrich, 2000).
It is noteworthy that the expression of the different KCNMB transcripts is highly tissue specific: thus, β1 mRNA is particularly abundant in vascular and nonvascular smooth muscle (Nelson and Quayle, 1995; Brenner et al., 2000b), wild-type (wt) β2 is the predominant transcript in adrenochromaffin cells (Wallner et al., 1999; Xia et al., 1999), β3 is abundant in testis, pancreas, skeletal muscle, and spleen (Behrens et al., 2000; Brenner et al., 2000a; Uebele et al., 2000), and β4 is abundant in central neurons (Behrens et al., 2000; Brenner et al., 2000a). Distinct expression patterns from one tissue to another may enable a given β subunit type to contribute to physiology in a tissue-specific manner. For example, β1 subunits are critical to fine tune the overall Ca2+ sensitivity of the BK channel complex (Brenner et al., 2000a; Cox and Aldrich, 2000), allowing channel activation by low µM Ca2+, which is reached in the vicinity of the BK channel during depolarization-induced myocyte contraction (Pérez et al., 2001). Local Ca2+ levels are usually generated by local signals termed “Ca2+ sparks,” which result from the activity of ryanodine receptors in the sarcoplasmic reticulum (Jaggar et al., 2000; Ledoux et al., 2006). Functional coupling between Ca2+ sparks and plasma membrane BK channels via β1 subunits generates outward K+ current in the vascular myocyte, which opposes depolarization-induced Ca2+ influx and smooth muscle contraction (Brenner et al., 2000b). Indeed, KCNMB1 knockout mice characteristically suffer from reduced spark–BK coupling and BK-mediated current in vascular smooth muscle (Brenner et al., 2000b).
Ethanol concentrations obtained in blood during moderate to heavy alcohol intoxication (50 mM) drastically modulate BK channel open probability (Po) and, thus, macroscopic current (Brodie et al., 2007; Martin et al., 2008; Wynne et al., 2009). The final effect of ethanol on BK channel currents depends on a variety of factors (reviewed in Brodie et al.  and Dopico et al. ), including posttranslational modification of slo1 proteins (Liu et al., 2006; Palacio et al., 2015), levels of Ca2+i (Liu et al., 2008), and the lipid environment of the BK channel complex, in particular, cholesterol levels (Crowley et al., 2003; Yuan et al., 2011) and type of phospholipid species (Crowley et al., 2005; Yuan et al., 2011) in the lipid bilayer where the slo1 channel resides. Several studies have documented different outcomes when ethanol action is evaluated on BK channels of different subunit composition, including homomeric slo1, slo1+β1, and slo1+β4; potentiation, inhibition, and ethanol refractoriness have all been reported (Crowley et al., 2003, 2005; Liu et al., 2004, 2008, 2013; Martin et al., 2008; Bukiya et al., 2009a). The use of different slo1 subunits, channel expression systems, lipid bilayer composition, and activating Ca2+ levels in these previous studies makes it impossible to address the precise contribution of regulatory β subunits to ethanol action on BK channels. Filling this knowledge gap acquires particular relevance because ethanol action on BK channels appears to be β subunit dependent; at physiological Ca2+i levels, neuronal and recombinant (slo1 and slo1+β4) channels are usually activated by ethanol, whereas vascular smooth muscle and recombinant slo1+β1 are usually inhibited (reviewed in Dopico et al. ). The relevance of this topic is compounded by the fact that differential actions of ethanol on BK channels have a direct impact on tissue physiology: ethanol (40–50 mM) activation of native BK channels thought to consist of α and/or α+β4 subunits leads to reduced action potential (AP) firing rate and neuronal excitability, as reported for nucleus accumbens medium spiny neurons (Martin et al., 2004, 2008) and dorsal root ganglia nociceptive neurons (Gruß et al., 2001). In contrast, inhibition of smooth muscle BK channels by 10–100 mM ethanol leads to cerebral artery constriction. Moreover, genetic ablation of BK β1 subunits totally blunts this alcohol action (Liu et al., 2004; Bukiya et al., 2009a).
Another major gap in the current knowledge on ethanol–BK channel interactions results from the fact that all previous studies have evaluated ethanol action on steady-state macroscopic current and/or steady-state single-channel open probability (Po; reviewed in Brodie et al.  and Dopico et al. ). This descriptive approach falls short of identifying the gating mechanisms targeted by ethanol, which leads to ethanol-induced perturbation of BK current and possible dependence on the different regulatory β subunits.
In this study, we combined patch-clamp electrophysiology and analysis of BK channel gating by using an allosteric model to (a) determine the distinct contribution of each β subunit type to ethanol action on BK channels and (b) identify the gating mechanisms targeted by ethanol and their modulation by BK β subunits. Thus, we expressed the same slo1 isoform (cbv1) alone or with β1, wt β2, and β2 with its NH2 end deleted to remove inactivation (β2-IR), β3 (d variant), or β4 and obtained single-channel or macroscopic current recordings in the absence and presence of intoxicating levels of ethanol (50 mM) across a wide range of Ca2+i (0–100 µM). Then, data were fitted with the Horrigan–Aldrich (HA) allosteric model of BK channel gating. Our study demonstrates that (a) ethanol differentially modulates BK currents that result from channel complexes containing different β subunits under identical recording conditions and using the same slo1 channel. At physiological levels of Ca2+i, two phenotypes are revealed: slo1 like, which includes slo1+β3 and slo1+β4 channels, and slo1+β1 like, which includes slo1+wt β2 and slo1+β2-IR, the former phenotype being activated and the latter inhibited by ethanol. We also demonstrate that (b) ethanol action on BK channels is allosterically coupled to Ca2+-driven gating and (c) both qualitative and quantitative differences in ethanol action on specific gating parameters contribute to the differential overall effect of ethanol (activation vs. inhibition) on BK channels when evaluated within physiological levels of Ca2+i.
Materials and methods
Expression of recombinant BK channel subunits
cDNA coding for BK channel–forming α subunits (cbv1; AY330293; the protein product has a 99% primary sequence conservation with the mslo1 mbr5 variant, NCBI) and β1 were cloned from rat cerebral artery myocytes as described previously (Jaggar et al., 2005; Liu et al., 2006). cbv1 cDNA was cleaved from the pBlueScript cloning vector by BamHI (Invitrogen) and XhoI (Promega) and directly reinserted into pOX vector for expression in Xenopus laevis oocytes. Linearization of pOX-cbv1 was done with NotI (Promega) and transcribed in vitro by using T3 polymerase. BK β1 subunit cDNA inserted into the EcoRI–SalI sites of the pCI-neo vector was linearized with NotI and transcribed in vitro by using T7 polymerase. Human wt β2 cDNA in pXMX vector and hβ3d in pCAP vector were gifts from C. Lingle (Washington University School of Medicine, St. Louis, MO). These plasmids were linearized with MluI (Invitrogen) and transcribed with T7. β2 subunit with the inactivation ball removed (β2-IR), β3 (human d variant), and human β4 were gifts from L. Toro (University of California, Los Angeles, Los Angeles, CA). β2-IR in pXMX vector was linearized with NotI and transcribed with T7. β3 cDNA inserted into the pOX vector was linearized with SalI (Invitrogen) and transcribed with T7. β4 cDNA was subcloned into pOX vector and linearized by NotI and transcribed using T3. In all cases, the mMESSAGE-mMACHINE kit (Ambion) was used for in vitro transcription. All cDNA sequences were confirmed by automatic sequencing at the University of Tennessee Health Science Center Molecular Research Center.
Care of animals and experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at the University of Tennessee Health Science Center, an Association for Assessment and Accreditation of Laboratory Animal Care–accredited institution. Oocytes were removed from Xenopus (Xenopus Express) and prepared as described elsewhere (Dopico et al., 1998). cRNA was dissolved in diethyl polycarbonate–treated water at 50 (cbv1) and 150 (β1–4) ng/µl, with 1-µl aliquots being stored at −80°C. cbv1 cRNA (2.5–10 ng/µl) was coinjected with β1, wt β2, β2-IR, β3, or β4 (7.5–30 ng/µl) cRNA, giving β/cbv1 molar ratios ≥6:1 to ensure that all cbv1 subunits were saturated with βs (Dopico, 2003; Liu et al., 2013). cRNA injection (23–46 nl/oocyte) was performed using an automated nanoinjector (Drummond Scientific Co.). The interval between injection and patch-clamp recordings was 48–72 h.
Detection of coexpression of cbv1 and β3d proteins on the cell membrane surface by biotinylation and Western blotting
The presence of cbv1 and β3d proteins on the membrane surface of Xenopus oocytes was verified by biotinylation, using the Pierce Cell Surface Protein Isolation kit (Thermo Fisher Scientific) and following the manufacturer’s instructions. Immediately before the biotinylation-based labeling and separation of membrane surface proteins, the oocyte’s follicular layer was removed to allow access of kit reagents to the cell membrane. The purified surface protein fraction was analyzed by Western blotting, as follows.
Purified surface protein fraction for biotinylation (50 µg/lane) was separated on a 4–15% SDS–polyacrylamide gel and transferred onto polyvinylidene difluoride (PVDF) membranes. The membranes were then blocked with 5% nonfat dry milk made in Tris-buffered saline (TBS) containing 0.1% Tween 20 for 2 h. Membranes were then incubated with appropriate primary antibodies overnight at 4°C in TBS with 0.1% Tween 20 (TBS-T) and 5% nonfat dry milk. Membranes were then incubated with appropriate horseradish peroxidase–conjugated secondary antibodies (1:10,000 dilution; EMD Millipore) for 1–2 h at room temperature. Proteins were then visualized using the SuperSignal West Pico Chemiluminescent Substrate kit (Thermo Fisher Scientific). A mouse monoclonal anti-KCNMB3 antibody (1:500 dilution; ab57219; Abcam) and mouse monoclonal anti-slo1 antibody (1:1,000 dilution; University of California, Davis/National Institutes of Health NeuroMab) were used to detect β3d and cbv1 proteins, respectively.
After oocyte preparation for patch-clamp electrophysiology, either single-channel or macroscopic currents were recorded from inside-out (I/O) patches, as previously described (Dopico, 2003; Liu et al., 2008; Kuntamallappanavar et al., 2014). Both bath and electrode solutions contained (mM) 135 K+ gluconate, 5 EGTA, 2.28 MgCl2, 15 HEPES, and 1.6 HEDTA, pH 7.4. In all experiments, the desired free [Ca2+]i was obtained by adding CaCl2, with free Mg2+ being kept constant at 1 mM. Free Ca2+i and Mg2+ were calculated with MaxChelator (WEBMAXC; Stanford University) and validated experimentally using Ca2+i-selective and reference electrodes (Corning) as described elsewhere (Dopico, 2003). For experiments in nominal zero Ca2+i, Ca2+ buffering was achieved by including 10 mM EGTA in the solution; neither Ca2+i nor HEDTA was added to this solution. Free [Ca2+]i in this nominal zero Ca2+i solution is ∼0.5 nM (Cox and Aldrich, 2000).
Patch-recording electrodes were pulled from glass capillaries and treated as described previously (Dopico et al., 1998). When filled with high K+ solution (see above composition of electrode solution), the vast majority of tip resistances were ∼2 MΩ, with a few reaching 5 MΩ. Series resistance was electronically compensated up to 80% by the EPC8 amplifier. An Ag/AgCl electrode with gluconate as main anion was used as ground electrode. Experiments were performed at room temperature (20–22°C).
Macroscopic currents were evoked from I/O macropatches held at −80 mV by 100-ms-long, 10-mV depolarizing steps from −150 to 150–200 mV. A P/4 leak subtraction routine was applied using a built-in function in pCLAMP. Currents at single-channel resolution were acquired from I/O patches in gap-free recording at Vm = −20 to −40 mV for 30 s under each experimental condition. Data were acquired using an EPC8 amplifier (HEKA), low-passed at 1 kHz by using an eight-pole Bessel filter (902LPF; Frequency Devices), and digitized at 5 kHz by using a Digidata 1320A A/D converter and pCLAMP 8.0 (Molecular Devices). For HA analysis, data were low-pass filtered at 10 kHz and sampled at 50 kHz. The number of channels in the patch (N) was estimated by switching the perfusion solution from nominal zero Ca2+i to ≥10 µM Ca2+i, which drives the individual opening probability (Po) close to unity, and then measuring peak macroscopic current amplitude (I) size and unitary current amplitude (i) at 20–90 mV. Thus, N = I/i. As an index of channel steady-state activity, we used the product NPo, which was obtained using a built-in option in Clampfit 9.2 (Molecular Devices) from ≥30 s of continuous recording under each experimental condition. This built-in algorithm uses a half-amplitude threshold method for event detection.
Fitting data to the HA model
In the absence of Ca2+i, the probability of channel opening as a function of voltage is given by the equation
In contrast, in the absence of voltage sensor activation, Ca2+i binding strongly affects channel opening (Horrigan and Aldrich, 2002; Sweet and Cox, 2008, 2009). In addition, Po is very small at extreme negative voltages. Thus, Po can be expressed as
According to Horrigan and Aldrich (2002), to determine Kd and C independently of N, the ratio (R0) of NPo in the presence and absence of Ca2+ was obtained from
where L0 = closed to open conformation equilibrium constant at V = 0, and no Ca2+i bound to the channel; zL = gating charge associated with channel opening; Vh(J) = voltage at which the voltage sensors are half the time active when the channel is closed; zJ = gating charge associated with each voltage sensor’s movement; Kd = dissociation constant when the channel is closed and all voltage sensors are in the resting state; C = allosteric coupling factor between Ca2+i binding and channel opening; D = allosteric coupling factor between voltage sensor activation and channel opening; E = allosteric coupling factor between Ca2+i binding and voltage sensor activation; R0 = the NPo ratio in the presence and absence of Ca2+i and independent of L and N (see above). Therefore, R0 depends only on Kd and C (Horrigan and Aldrich, 2002). In Eqs. 1, 2, 3, 4, and 5, R, T, and F have their usual meaning.
Chemicals and drug perfusion
Unless indicated otherwise, all chemicals were purchased from Sigma-Aldrich. On the day of the experiment, ethanol (100% purity; American Bioanalytical) was freshly added to the bath solution immediately before recordings. After excision from the cell, the cytosolic side of the membrane patch was exposed to a bath solution delivered from a pressurized, automated DAD12 system (ALA Scientific Instruments) via a micropipette tip having an internal diameter of 100 µm. Bath solution with urea isoosmotically substituting for ethanol was used as control perfusion, with osmolarity of all solutions ranging from 301 to 322 mOsM (Wescor Vapro).
Electrophysiology data were analyzed with pCLAMP 9.2 (Molecular Devices). G (conductance)/Gmax-V curves obtained from macroscopic ionic current at steady-state level were fitted to a Boltzmann function of the type G(V) = Gmax/(1 + exp[−V + V0.5)k]). Boltzmann-fitting routines were run using a Levenberg–Marquardt algorithm to perform nonlinear least squares fits. From G/Gmax-V plots, we obtained the voltage required to achieve half-maximal conductance (V0.5). For inactivating currents (cbv1+wt β2), V0.5 was determined from the peak current immediately before the onset of inactivation. Macroscopic current activation and deactivation data were fitted to standard exponential functions using a Chebyshev approximation. Time constant for current activation (τact) was measured at the voltage at which the channel reached maximal steady-state activity (Vmax), whereas the deactivation time constant (τdeact) was measured after voltage reached Vmax and then stepped down to −80 mV (Kuntamallappanavar et al., 2014). Further analysis, plotting, and fitting were conducted using Origin 8.5 (OriginLab) and InStat 3.0 (GraphPad Software). Data are expressed as mean ± SEM; n = number of patches. Each patch was obtained from a different oocyte. Fitted parameters are expressed as mean ± 95% confidence intervals. Statistical analysis was conducted using Student’s paired t test to compare the paired dataset (for instance, data before and after ethanol treatment). Multiple sets of data were compared using one-way ANOVA followed by Tukey’s multiple comparison test (Glantz, 2001) to determine statistical significance of differences between individual means; significance was set at P < 0.05.
Online supplemental material
Fig. S1 demonstrates the surface expression of β3 variant d and the single-channel phenotype that results from coexpressing cbv1 with this β subunit. Fig. S2 shows ethanol’s lack of effect in channels gated by voltage-Mg2+i in the absence of Ca2+i. Fig. S3 shows ethanol action on macroscopic currents originated from cbv1+wt β2.
Coexpression of different auxiliary β subunits with slo1 cloned from cerebral artery myocytes renders different macroscopic current phenotypes
Before probing ethanol on BK channels of different subunit composition (cbv1±βs), we assessed functional coexpression of the different β subunits (β1, wt β2, β2-IR, β3 variant d, and β4) with the channel-forming cbv1 subunit. Thus, using the oocyte expression system, macroscopic recordings were obtained from I/O patches over a wide range of Ca2+i concentrations (nominal zero to 100 µM). To identify the ion current phenotype from the different BK channel complexes, we obtained G/Gmax-V plots according to well-established acquisition voltage protocols (see Materials and methods). Representative ionic currents obtained at 10 µM free Ca2+i are shown in Fig. 1 A. From G-V plots, we obtained the voltage required to achieve half-maximal conductance (V0.5) and, eventually, V0.5-[Ca2+]i plots. From each macroscopic ionic current recording, time constants for current activation (τact) and deactivation (τdeact) were measured as stated in Materials and methods.
β1 subunits increased cbv1 macroscopic current as Ca2+i was raised above micromolar levels. This is evident as a decrease in V0.5 in the V0.5-Ca2+ plot. Moreover, these plots show that the decrease in V0.5 evoked by the presence of β1 becomes progressively larger as Ca2+ increases (Fig. 1 B). These results are in agreement with previously published data from BK complexes made of slo1+β1 other than cbv1 (Brenner et al., 2000a; Cox and Aldrich, 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Contreras et al., 2012). The shift in V0.5 has been widely conceptualized as a β1 subunit–induced increase in the apparent Ca2+i sensitivity of BK channel. In addition, β1 increased τact and τdeact from their cbv1 values: 1.40 ± 0.10 and 1.13 ± 0.13 ms to 5.16 ± 0.93 and 6.76 ± 0.67 ms; all values were obtained at 10 µM Ca2+i (P < 0.05 for both constants; Fig. 1, C and D). This slowing down of macroscopic cbv1 current activation and deactivation kinetics by BK β1 subunits is also in accordance with data from slo1 other than cbv1 (Brenner et al., 2000a; Cox and Aldrich, 2000; Bao and Cox, 2005; Contreras et al., 2012). Therefore, macroscopic currents obtained from coexpression of β1 with cbv1 display a current phenotype characteristic of slo1+β1 currents.
When compared with β1, wt β2 subunits possess an additional inactivation domain in their N terminus (Wallner et al., 1995, 1999; Xia et al., 1999, 2003; Uebele et al., 2000). Indeed, coexpression of wt β2 with cbv1 channels consistently rendered macroscopic ionic currents that displayed fast inactivation (Fig. 1 A, bottom left). This result matches findings from other groups that coexpressed wt β2 with slo1 other than cbv1 (Wallner et al., 1999; Xia et al., 1999, 2003; Uebele et al., 2000). In contrast to wt β2, β2-IR has been engineered to lack the inactivation domain (Brenner et al., 2000a; Orio and Latorre, 2005; Sun et al., 2013). As found with β1 subunits, coexpression of β2-IR with cbv1 resulted in an increase of the channel’s apparent Ca2+i sensitivity (Fig. 1 B). In addition, when compared with homomeric cbv1 channels, β2-IR–containing BK complexes rendered macroscopic ionic currents with very slow activation and deactivation kinetics: at 10 µM Ca2+i, τact and τdeact changed from cbv1 values of 1.40 ± 0.10 and 1.13 ± 0.13 ms to 4.73 ± 0.74 and 6.22 ± 0.82 ms (P < 0.05 for both constants; Fig. 1, C and D). These cbv1+β2-IR versus cbv1 comparisons are consistent with previous data obtained by expressing β2-IR with slo1s other than cbv1 (Brenner et al., 2000a; Orio and Latorre, 2005; Sun et al., 2013).
In contrast to all other β subunits (including β4; see below in this paragraph), functional coexpression of β3 (variant d) with cbv1 was characterized by lack of modulation of cbv1’s apparent Ca2+i sensitivity, activation, or deactivation kinetics. At 10 µM, τact and τdeact values were 1.40 ± 0.10 and 1.13 ± 0.13 ms versus 1.91 ± 0.73 and 0.76 ± 0.11 ms for cbv1 versus cbv1+β3, respectively (Fig. 1, A–D). These data are in agreement with previous studies on slo1±β3 other than cbv1 (Behrens et al., 2000; Brenner et al., 2000a; Uebele et al., 2000; Zeng et al., 2008). Membrane surface coexpression of β3d subunit with cbv1 was confirmed by conducting biotinylation followed by Western blotting (Fig. S1 A). In addition, β3d coexpression resulted in longer channel openings (Fig. S1 B), as previously reported (Bukiya et al., 2013). In contrast to β1- and β2-IR–induced modifications in ionic current phenotype, expression of β4 increased cbv1 V0.5 at 0.3–10 µM Ca2+i while decreasing V0.5 at 30–100 µM Ca2+i (Fig. 1 B). In addition, at 10 µM Ca2+i, β4 significantly increased τact from its cbv1 value, 1.40 ± 0.1 ms to 4.25 ± 0.83 ms (P < 0.05), while increasing τdeact from its cbv1 value (P < 0.05; Fig. 1, C and D). Thus, the β4-introduced changes in V0.5, τact, and τdeact over cbv1’s values are consistent with those previously reported for β4 and slo1 channels other than cbv1 (Behrens et al., 2000; Brenner et al., 2000a; Orio et al., 2002). Collectively, results from this section established that BK channel complexes that result from expressing the same slo1 (cbv1) subunit when each type of BK β subunit known so far renders macroscopic current phenotypes that are similar to those reported with matching β subunits and different Slo1 gene products. These similarities in V0.5, τact, and τdeact together with our previous pharmacological data obtained with cbv1±β subunits that matched published studies on slo1+β subunits other than cbv1 (Bukiya et al., 2007, 2009b, 2013; Kuntamallappanavar et al., 2014) underscore the major role of BK auxiliary β proteins in determining the final BK ionic current phenotype and its pharmacology.
Ethanol direct modulation of homomeric cbv1-mediated current is Ca2+i dependent but not Mg2+i dependent
First, we evaluated ethanol direct action (i.e., independently of cytosolic signaling or alcohol metabolism) on homomeric cbv1 channels using excised, cell-free patches under continuous perfusion with divalent-buffered solutions. Thus, macroscopic currents were obtained over a wide range of activating Ca2+i (0–100 µM) in the presence and absence of 50 mM ethanol. Ethanol increased cbv1 channel activity at low [Ca2+]i i.e., 0.3–10 µM, which was evident as a shift to the left in the G/Gmax-V plot and eventual decrease in V0.5 as shown in the V0.5-[Ca2+]i plot (Fig. 2, A, B, E, and F). However, ethanol did not modulate cbv1 channel activity at zero Ca2+i (Fig. 2, D and F). Moreover, as [Ca2+]i increased (30–100 µM), ethanol-induced channel activation turned to inhibition, which was evident as a small right shift in the G/Gmax-V plot and, eventually, a small increase in V0.5 (P < 0.05; Fig. 2, C and E). Thus, ethanol-induced activation to inhibition crossover occurred at ∼20 µM (Fig. 2 E). These results demonstrate that ethanol action on cbv1 channels not only depends on Ca2+i presence but is also a function of [Ca2+]i. This result extends previous findings with slo1 channels cloned from mouse brain (mslo1, mbr5 variant) and heterologously expressed in the same expression system (Liu et al., 2008).
Previous studies have clearly established that BK channels can be gated by millimolar levels of Mg2+i, independently of Ca2+i-driven gating (Shi et al., 2002). Thus, we next obtained macroscopic current recordings from Mg2+i-gated cbv1 channels over a wide range of activating [Mg2+]i (0–100 mM) in the presence and absence of 50 mM ethanol (Fig. S2 A). Remarkably, ethanol failed to modulate cbv1-mediated current across all Mg2+i levels, which was evident from the lack of significant differences in G/Gmax-V and V0.5-[Mg2+] plots between control and ethanol-containing solutions (Fig. S2, B–D). These findings demonstrate that, when cbv1 channels are gated by Mg2+i, their sensitivity to high, intoxicating levels of ethanol (50 mM) is lost. The result underscores the specificity of Ca2+i over other ions in determining the sensitivity of BK channel gating to ethanol (Liu et al., 2013).
BK β1 subunit facilitates ethanol-induced inhibition of β1-containing BK channel complexes
After demonstrating that β1 and ethanol modulation of cbv1 current were both Ca2+i dependent (Figs. 1 B and 2 E), we next addressed the functional interactions between β1 and ethanol that resulted in regulating the cbv1 channel function. 50 mM ethanol activated cbv1+β1 channels at low [Ca2+]i, i.e., 0.3 µM, which was evident as a leftward shift in the G/Gmax-V plot (Fig. 3, A and B) and thus a decrease in V0.5 value in the V0.5-[Ca2+i] plot (P < 0.01; Fig. 3, D and E). As [Ca2+]i increased (3–100 µM), activation turned to increasingly significant inhibition (P < 0.05 at 30 µM and P < 0.01 at 100 µM), which was evident as a rightward shift in the G/Gmax-V plot (Fig. 3 C) and eventual increase in V0.5 values, as seen in the V0.5-[Ca2+]i plot (Fig. 3, D and E). Furthermore, ethanol-induced activation to inhibition crossover shifted to ∼3 µM (Fig. 3 D) from that of cbv1 channels (∼20 µM), that is, almost one log unit decrease. Our data demonstrate that ethanol-mediated inhibition of recombinant BK channels made of subunits (cbv1+β1) cloned from rat cerebral artery myocytes occurs at physiological Ca2+i (4–30 µM) and voltage (−60 to −20 mV) found near native BK channels during contraction of rat cerebral artery myocytes (Pérez et al., 2001).
Ethanol responses of wt β2- and β2-IR–containing BK channel activity are similar to that of β1 subunit–containing BK channels
After characterizing the effect of ethanol on the Ca2+i dependence of macroscopic current from cbv1+β1 versus cbv1 channels (previous paragraph) and having demonstrated that the Ca2+i dependence of macroscopic current was similarly modified when cbv1 channels were coexpressed with β1 or β2-IR subunits (Fig. 1 B), we next determined whether β1-induced modification of ethanol action on the V0.5-[Ca2+]i plot of cbv1 channels was unique to this subunit or, rather, matched by coexpression of wt β2 or β2-IR with cbv1. Coexpression of wt β2 with cbv1 evoked rapidly inactivating macroscopic currents, as initially shown in Fig. 1 A. Then, we evaluated ionic currents at single-channel resolution obtained from I/O patches in gap-free recordings at Vm = −20 to −40 mV and over a wide range of Ca2+i for 30 s under each experimental condition (control vs. 50 mM ethanol). Ethanol activated cbv1+wt β2 channels at low [Ca2+]i, i.e., 0.3 µM, which was evident as a robust increase in NPo (P < 0.01; Fig. 4, A and C). As [Ca2+]i increased, this activation turned into inhibition, which was evident as a significant reduction in NPo (P < 0.01; Fig. 4, B and C). Furthermore, as found with cbv1+β1 (Fig. 3 D), ethanol-induced activation to inhibition crossover of cbv1+wt β2 occurred at ∼3 µM (Fig. 4 C). These experiments of ethanol action on cbv1+wt β2 complexes at single-channel resolution were matched by experiments using macroscopic recordings (Fig. S3). Likewise, ethanol activated cbv1+β2-IR channels at low [Ca2+]i, i.e., 0.3 µM, which was evident as a marked decrease in V0.5 values in the V0.5-[Ca2+]i plot (P < 0.01; Fig. 5, A and B). As [Ca2+]i increased, ethanol-induced potentiation of ionic current turned to inhibition, which was evident as a significant (P < 0.05) increase in V0.5 values in the V0.5-[Ca2+]i plot (Fig. 5, A and B). As previously found with cbv1+β1 and cbv1+wt β2, ethanol-induced activation to inhibition crossover of cbv1+β2-IR currents occurred at ∼3 µM (Fig. 5 A). The similarity of ethanol action on cbv1+wt β2 versus cbv1+β2-IR strongly suggests that the inactivation domain of wt β2 does not play a significant role in ethanol action on these channels. However, the similarity of ethanol’s overall effect among cbv1+β1, cbv1+wt β2, and cbv1+β2-IR constructs raises the speculation that ethanol action on heteromeric BK channels is somewhat linked to channel gating processes commonly affected by these regulatory subunits, very likely gating transitions that contribute to determine the apparent Ca2+i sensitivity of the channel complex.
Ethanol responses of β3- and β4-containing BK currents are similar to that of currents evoked by homomeric cbv1 channels
When compared with β1 and β2-IR subunits, β3 and β4 subunits have a milder effect on the apparent Ca2+i sensitivity of slo1 channels, as documented by several groups, including our laboratory (Fig. 1 B; Xia et al., 1999; Behrens et al., 2000; Brenner et al., 2000a; Uebele et al., 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Bukiya et al., 2009a,b, 2013; Sun et al., 2013; Kuntamallappanavar et al., 2014). If the hypothesis raised at the end of the previous section is true, ethanol action on β3- or β4-containing BK channels should not greatly differ from ethanol action on homomeric cbv1. Thus, cbv1+β3 and cbv1+β4 were coexpressed and probed with 50 mM ethanol under experimental conditions similar to those of the cbv1±β1/wt β2/β2-IR experiments. Ethanol activated cbv1+β3 and cbv1+β4 channels at low [Ca2+]i, i.e., 0.3 µM, which was evident as a marked decrease in V0.5 from the V0.5-[Ca2+]i plot (P < 0.01; Fig. 6, A–D). As [Ca2+]i increased, this activation turned into mild inhibition, which could be noticed by a significant increase (P < 0.05) in V0.5 from the V0.5-[Ca2+]i plot (Fig. 6, A–D). In both heteromeric BK complexes, ethanol-induced activation to inhibition crossover occurred at ∼20 µM (Fig. 6, A and C). Thus, β3- and β4-containing BK channel responses to ethanol did match that of homomeric cbv1 channels, indicating that β3 and β4 subunits do not noticeably modulate ethanol action on ionic currents mediated by homomeric cbv1 channels.
Ca2+i-independent gating processes of homomeric BK channels remain unaffected by ethanol
Once we established that (a) ethanol differentially modulated BK channels containing different β subunit compositions, having identified two basic phenotypes: (1) ethanol action on homomeric cbv1 (matched by cbv1+β3 or cbv1+β4 complexes) and (2) ethanol action on cbv1+β1 (matched by cbv1+wt β2/β2-IR), and (b) ethanol direct action on macroscopic current depended on Ca2+i presence and was modulated by this divalent, yet not by Mg2+i, we decided to address which specific BK channel gating mechanisms could explain ethanol action, with a focus on selective allosteric coupling between ethanol and Ca2+i. Thus, we evaluated ethanol action on cbv1±β1 channels under a wide variety of experimental conditions to fit experimental data to the 70-state HA allosteric gating model (Horrigan and Aldrich, 1999, 2002; Horrigan et al., 1999; Orio and Latorre, 2005).
It has been previously demonstrated that, by obtaining the Po-V relationship of BK channels at very negative potentials and nominal zero Ca2+i, it is possible to directly estimate two parameters associated with the channel’s closed to open conformational change (or intrinsic gating), i.e., L0 and zL, which indicate closed to open equilibrium constant at V = 0 when no Ca2+i is bound to the channel and gating charge associated with channel opening, respectively (Horrigan and Aldrich, 1999, 2002; Horrigan et al., 1999). For cbv1 channels (Fig. 7 A), the best-fit values were L0 = 2.3 × 10−6 ± 6.3 × 10−7 and zL = 0.34 ± 0.054 (Fig. 7 B), matching previously published data from slo1 channels other than cbv1 (Horrigan and Aldrich, 1999, 2002; Bao and Cox, 2005; Orio and Latorre, 2005; Sun et al., 2013). Ethanol, however, failed to alter these parameters, L0 = 2.1 × 10−6 ± 2.8 × 10−7 and zL = 0.35 ± 0.04 both in the presence of 50 mM ethanol (Fig. 7, B and C), indicating that ethanol at concentrations that increase BK Po does not modulate the conformational change associated with the closed to open transition of the channel.
After establishing that ethanol did not modify cbv1 L0 or zL, we next evaluated ethanol action on parameters Vh(J), zJ, and D, which indicate half-activation voltage of voltage sensors when the channel is closed, gating charge associated with voltage sensor movement, and allosteric interaction of voltage sensor activation with channel opening, respectively. To perform this, we combined Po-V data with G/Gmax-V data to obtain channel activity–V curves across the necessarily wide voltage range (Fig. 8, A and B). The resulting curves were fitted with Eq. 3. While fitting the data obtained from cbv1 channels in the absence of ethanol, Vh(J), L0, and zL were constrained to 155 (Horrigan et al., 1999), 2.3 × 10−6, and 0.34 (see Fig. 7), respectively, whereas D and zJ were allowed to vary. For ethanol data, L0 and zL values were constrained to 2.1 × 10−6 and 0.35, respectively, whereas Vh(J), D, and zJ were allowed to vary. In the absence of ethanol, fitting results rendered D = 19.8 ± 1.4 and zJ = 0.61 ± 0.05 for cbv1 channels (Fig. 8 A). These results are quite similar to the values obtained by different groups using slo1 channels other than cbv1 (Horrigan and Aldrich, 1999, 2002; Cox and Aldrich, 2000; Bao and Cox, 2005; Orio and Latorre, 2005).
Vh(J), D, and zJ from cbv1 channels remained unmodified by 50 mM ethanol: Vh(J)= 154.1 ± 2.7, D = 19.6 ± 0.6, and zJ = 0.6 ± 0.05 (Fig. 8 B). These data indicate that ethanol at concentrations obtained during binge drinking, while altering BK channel steady-state activity in the presence of Ca2+ (Dopico et al., 1998; Liu et al., 2008), does not alter the movement of voltage sensors when the channel is closed, the gating charge associated with the voltage sensor movement, or the coupling factor between voltage sensor activation and channel opening. Collectively, all the ethanol data shown so far indicate that, at physiological Ca2+i levels, ethanol markedly increases cbv1 channel activity without significantly modifying any of the Ca2+i-independent gating processes under study.
Ethanol enhances the apparent Ca2+i binding affinity of homomeric BK channels
Next, we addressed ethanol action on major Ca2+i-dependent gating processes of homomeric cbv1 channels. Several studies have demonstrated that two Ca2+i-dependent gating parameters (i.e., Kd and C, Ca2+ binding affinity and allosteric factor coupling Ca2+i binding with channel opening, respectively) can be estimated from the Ca2+ dependence of Po at very negative voltages (Horrigan and Aldrich, 1999, 2002; Horrigan et al., 1999; Sweet and Cox, 2008, 2009). Thus, we obtained R0 (i.e., the NPo ratio in the presence and absence of Ca2+i; see Materials and methods) at very negative voltages (−80 mV) in the absence and presence of 50 mM ethanol, and data were fitted with Eq. 5. Best-fit values were Kd = 9.02 ± 1.4 and C = 6.07 ± 0.1, matching previously published data for slo1 channels other than cbv1 (Horrigan and Aldrich, 1999, 2002; Bao and Cox, 2005; Orio and Latorre, 2005; Sun et al., 2013). Remarkably, ethanol decreased Kd from 9.02 ± 1.4 to 1 ± 0.07 (∼90% decrease). In contrast, ethanol decreased C from 6.07 ± 0.1 to 4.7 ± 0.12 (∼20% decrease; Fig. 9, A and B). Therefore, ethanol-induced potentiation of cbv1 current is basically the result of a nine-times increase in the channel’s apparent Ca2+ binding affinity, in spite of an ∼20% reduction in the efficacy of coupling between Ca2+ binding and channel opening.
Next, we examined ethanol action on parameter E, which indicates the allosteric interaction of Ca2+ binding with voltage sensor activation. Estimation of the value E from ionic current measurement alone is somewhat difficult. To determine E, however, we obtained G/Gmax-V relationships at various Ca2+i concentrations from nominal zero to 100 µM in the presence and absence of 50 mM ethanol and fitted the curves to Eq. 1 by minimizing least-squares with a mixture of Marquardt–Levenberg and Simplex iterations (Orio and Latorre, 2005). Vh(J), L0, zL, zJ, D, Kd, and C were constrained to the following values for control versus ethanol data (as previously determined from data shown in Figs. 7, 8, and 9): Vh(J) = 155 mV versus 154.1 mV; L0 = 2.3 × 10−6 versus 2.1 × 10−6; zL = 0.34 versus 0.35; Kd = 9.02 versus 1; and C = 6.07 versus 4.7, whereas E was allowed to vary. Several fits were conducted with different sets of initial parameters, and the fits with lower χ2 values were chosen. Best-fit results for cbv1 in the absence of ethanol rendered E = 6.15 ± 2.1 (Fig. 10 A), which is in reasonable agreement with previously obtained best-fit values for slo1 channels other than cbv1 (Horrigan and Aldrich, 1999, 2002; Orio and Latorre, 2005). Ethanol, however, did not alter E (Fig. 10, A and B), underscoring that the alcohol does not modify the allosteric interaction between Ca2+ binding and voltage sensor activation of homomeric cbv1 channels. Collectively, our results indicate that most of ethanol’s effect on homomeric BK (cbv1) channels is caused by modifying drug-induced increase in Ca2+i binding affinity and allosteric interaction between Ca2+i binding and channel opening, thereby modulating overall channel activity.
In the presence of β1 subunits, Ca2+i-independent gating processes of BK channels remain unaffected by ethanol
In contrast to cbv1 data, log (Po)-V plots from cbv1+β1 channels did not reach the limiting slope (Fig. 11, A and B), as reported by several groups using slo1 other than cbv1 (Bao and Cox, 2005; Orio and Latorre, 2005; Sun et al., 2013). Therefore, for cbv1+β1 channels, we determined the effect of ethanol on voltage sensor movement (Vh(J)), parameters associated with closed to open conformation change (L0 and zL), and voltage dependence of gating (zJ and D) by fitting the Po-V data with Eq. 3 (Fig. 11, A and B). For control data, Vh(J) was constrained to a previously published value, i.e., 80 mV (Bao and Cox, 2005), whereas L0, zL, D, and zJ were allowed to vary freely. Our fitting results obtained for cbv1+β1 channels in the absence of ethanol resulted in L0 = 1.2 × 10−7 ± 4.4 × 10−8, zL = 0.41 ± 0.06, D = 30.2 ± 3.7, and zJ = 0.6 ± 0.03 (Fig. 11 B). These values reasonably match the values obtained by different groups for slo1 other than cbv1 when coexpressed with β1 (Cox and Aldrich, 2000; Orio and Latorre, 2005; Sun et al., 2013). As found for cbv1 channels, ethanol failed to modify any of these parameters: Vh(J) = 80.8 ± 1.6, L0 = 1.3 × 10−7 ± 3.9 × 10−8, zL = 0.42 ± 0.3, D = 29.3 ± 2.6, and zJ = 0.59 ± 0.02 in the presence of 50 mM ethanol (Fig. 11 B). In summary, when cbv1+β1 channels are probed with ethanol, the drug does not alter (a) the equilibrium constant between closed to open conformation, (b) the gating charge associated with such equilibrium constant, (c) the gating charge associated with the voltage sensor movement, or (d) the allosteric interaction between voltage sensor activation and channel opening. Collectively, these results indicate that ethanol, as found for homomeric cbv1 channels, does not modify the Ca2+i-independent gating processes of cbv1+β1 channels.
Ethanol modulates Ca2+i-dependent gating processes of β1-containing BK channels
To determine any possible effect of ethanol on the equilibrium governed by Ca2+i binding affinity (Kd) and its related allosteric factors (C and E), we used log (R0)-V and G/Gmax-V relations for cbv1+β1 channels in the presence and absence of ethanol at various Ca2+i concentrations from nominal zero to 100 µM. Then, log (R0)-V relations were fitted with Eq. 5, and G/Gmax-V relations were fitted with Eq. 1. Our best-fit values obtained for cbv1+β1 channels in the absence of ethanol were Kd = 19 ± 2.4, C = 27.2 ± 2.3 (Fig. 12 A), and E = 9.5 ± 1.6 (Fig. 13 A). As found for homomeric cbv1 channels, ethanol decreased Kd (from 19 ± 2.4 to 11.3 ± 1.2; Fig. 12 B). This decrease (∼40%) in Kd corresponds to an approximately two-times increase in the apparent Ca2+ binding affinity, which is markedly smaller than the nine-times increase in Ca2+ binding affinity evoked by ethanol on homomeric channels (Fig. 9 and Tables 1 and 2). As found with homomeric cbv1 channels, ethanol decreased C (in this case from 27.2 ± 2.3 to 13.5 ± 0.45; Fig. 12 B). This decrease (∼50%) in allosteric coupling between Ca2+ binding and channel opening is larger than the decrease evoked by the alcohol on homomeric channels (∼20%). Finally, ethanol markedly decreased E by half: from 9.5 ± 1.6 to 4.06 ± 0.6 (P < 0.05; Fig. 13, A and B). It should be underscored that ethanol fails to modify this parameter when homomeric cbv1 channels are probed with the alcohol (Fig. 10 and Tables 1 and 2). Therefore, ethanol targets both common and distinct gating mechanisms in homomeric cbv1 versus cbv1+β1 channels, the former represented by Kd and C and the latter by E. Regarding the ethanol-sensitive common processes, a smaller increase in Ca2+ binding affinity and a larger decrease in allosteric interaction between Ca2+i binding and channel opening would favor decreased (or at least reduced potentiation) cbv1+β1 channel activity by ethanol when compared with cbv1 channels. These modifications on gating are likely compounded by the fact that ethanol reduces E of cbv1+β1 heteromers but does not modify this parameter in cbv1 homomers (Tables 1 and 2). Thus, (a) quantitative differences on common gating processes and (b) unique disruption of coupling between Ca2+i binding and voltage sensor activation together very likely explain the fact that the steady-state channel activity of cbv1+β1 channels is inhibited, whereas that of cbv1 is activated in response to ethanol when evaluated at 3–20 µM Ca2+i (see also Discussion).
In spite of recent advances in our understanding of how ethanol at concentrations found in blood during moderate-heavy alcohol intoxication (50–100 mM) modulates BK channel activity and thus cell physiology, the gating mechanisms underlying such modulation remain unknown. All previous studies that addressed the response of BK currents to a brief (few minutes) exposure to ethanol in an alcohol-naive system, whether using recombinant slo1 proteins or native channels in their natural membrane environment, agreed that ethanol modulation of BK current occurred in the absence of perturbations in ion conduction or channel membrane expression (Brodie et al., 2007; Dopico et al., 2014). Thus, acute ethanol action on slo1 channel–mediated ionic current seems to be limited to that of a gating modifier. However, all studies previous to the present work have evaluated ethanol’s overall effect on slo1 macroscopic current, open channel probability, and/or dwell time distributions under a limited range of stimulus conditions. Previous studies offer no clue on the specific gating processes targeted by ethanol that lead to modification of slo1 current.
In addition, several groups have documented that different auxiliary BK β subunits condition the overall response of the BK channel complex to intoxicating levels of ethanol. Thus, at physiological Ca2+i levels, homomeric slo1, heteromeric slo1+β4, and native neuronal channels thought to consist of slo1 and/or slo1+β4 subunits (Dopico et al., 1998; Martin et al., 2004; Liu et al., 2008; Wynne et al., 2009) are activated by 50–100 mM ethanol. In contrast, heteromeric slo1+β1 and native cerebral artery myocyte BK channels thought to consist of slo1+β1 subunits are usually inhibited under similar voltage and Ca2+i conditions (Liu et al., 2004; Bukiya et al., 2009a). Moreover, ethanol-induced inhibition of cerebral artery myocyte BK channels is totally blunted by genetic ablation of β1 subunits (Bukiya et al., 2009a). Like ethanol, β subunits are gating modifiers; the actions of several of these accessory proteins on specific aspects of BK (slo1) channel gating (intrinsic gating–, voltage-, and Ca2+i-dependent parameters) are an expanding field of investigation (Cox and Aldrich, 2000; Bao and Cox, 2005; Orio and Latorre, 2005; Sweet and Cox, 2009; Contreras et al., 2012; Sun et al., 2013; Castillo et al., 2015). As found for studies of ethanol action on slo1 channels themselves, all previous studies on ethanol modulation of slo1+β heteromers have been limited to the descriptive electrophysiological approaches mentioned above. More importantly, these studies have been conducted using different slo1 isoforms, Ca2+i levels, expression systems, and/or membrane lipid composition, all factors that perturb slo1 channel gating by themselves (reviewed by Brodie et al., 2007; Dopico et al., 2012, 2014; Bettinger and Davies, 2014). This heterogeneity in experimental conditions makes it impossible to even determine the actual contribution (if any) of a given β subunit type to the overall effect of ethanol on BK current, not to mention identification of the underlying gating processes leading to such effect. To resolve these uncertainties and gaps in knowledge on ethanol–BK channel interactions, we systematically addressed the following questions: (a) whether ethanol indeed differentially modulates BK channels that differ only in β subunit type composition (i.e., cbv1±β1/wt β2/β2-IR/β3 d variant/β4) when studied under identical conditions using the same expression system and (b) which specific gating processes that control BK channel activity are distinctly modified by ethanol. This exploration was conducted within the framework of the HA allosteric model of channel gating (Horrigan and Aldrich, 1999, 2002; Horrigan et al., 1999; Bao and Cox, 2005; Orio and Latorre, 2005) by using a wide range of voltages and Ca2+ concentrations. We focused this analysis on homomeric cbv1 versus heteromeric cbv1+β1 channels because (a) ethanol exerts opposite effects on these currents (activation vs. inhibition) when evaluated within 3–20 µM Ca2+i and (b) the consequences of these differential ethanol effects have been well-documented: inhibition of neuronal firing in nucleus accumbens medium spiny neurons (Martin et al., 2004, 2008) and nociceptive DRG neurons (Gruß et al., 2001) versus constriction of cerebral arteries (Liu et al., 2004). Thus, addressing the distinct gating processes differentially targeted by ethanol in these constructs may likely provide mechanistic insights on alcohol-induced modification of excitable tissue physiology; (c) all other heteromeric BK channel combinations seem to follow the overall ethanol response of either homomeric cbv1 or heteromeric cbv1+β1 (see below). Throughout our experiments, we used an ethanol concentration (50 mM) reached in blood during moderate to heavy alcohol intoxication and widely reported to alter BK current and thus cell physiology (Liu et al., 2008; Bukiya et al., 2009a).
Present data demonstrate that the overall ethanol effect on BK channels that differ in β subunit composition when evaluated in the same system under the same recording conditions and including the same slo1 subunit (cbv1) does differ. At first glance, however, steady-state currents mediated by all constructs under study are markedly increased by ethanol when studied at low Ca2+i concentrations (<1 µM). This potentiation gradually decreases as Ca2+i increases until inhibition of current is reached (at >30 M Ca2+i). However, a closer inspection reveals two different phenotypes of overall ethanol effect on BK channel activity: the homomeric slo1 type (shared by cbv1, cbv1+β3 variant d, and cbv1+β4) and the slo1+β1 type (shared by cbv1+β1, cbv1+wt β2, and cbv1+β2-IR). The former is defined by a crossover in overall ethanol effect on current from activation to inhibition that occurs at ∼20 µM Ca2+, whereas the latter is defined by such crossover being shifted to 3 µM Ca2+i. This change in crossover along the Ca2+i axis is critical because it implies that at activating levels of Ca2+i reached in excitable tissues in the vicinity of the BK channel (3–20 µM; Pérez et al., 2001), ethanol will increase BK current in the former while inhibiting current in the latter. Indeed, these outcomes have been reported with native BK channels thought to consist of slo1+β4 versus slo1+β1 subunits (Liu et al., 2004; Martin et al., 2004, 2008; Bukiya et al., 2009a).
A second major finding from probing ethanol on the steady-state current mediated by the different constructs is that ethanol’s overall effect is identical in wt β2– and β2-IR–containing BK channels, indicating that the inactivation domain of wt β2 does not play a major role in the overall effect of the drug on steady-state activity. It should be stressed, however, that under our recording conditions (steady-state activity), the Po of slo1+wt β2 channels is dominated by the equilibrium between inactivated states and noninactivating states, and thus, obtaining gating parameters straightforwardly from the HA model is not tenable for this construct.
Gating analysis and fitting to the HA model of our data uncovered several important findings on ethanol actions on slo1 channels in the presence and absence of regulatory β subunits. First, ethanol’s overall effect on BK current, whether defined by the slo1 or the slo1+β1 prototype described above (third paragraph of Discussion; see also Figs. 2 and 3), occurs in the absence of modification of the conformational change associated with the closed to open transition of the channel (e.g., L0 and zL remain unmodified). Thus, intrinsic gating remains conserved in the presence of ethanol levels that modify steady-state current. Second, ethanol fails to modify cbv1 Vh(J), a parameter that has been widely used as an estimate of voltage sensor movement when the channel is closed (Horrigan and Aldrich, 1999; Horrigan et al., 1999; Orio and Latorre, 2005). In addition, we demonstrated that the gating charge associated with each voltage sensor’s movement (zJ) and the allosteric coupling factor between voltage sensor activation and channel opening (D) remained unchanged for both the slo1 and slo1+β1 prototypes. Finally, previous descriptive electrophysiology on mslo1 channels showed that 50–100 mM ethanol (a) failed to alter the rate-limiting slope of the ln NPo-V relationship and, thus, the effective valence (z; Dopico et al., 1998) and (b) similarly affected the V0.5-Ca2+ versus V0.5 × Q-Ca2+i plots of mslo1 channels (Liu et al., 2008). Collectively, present determinations from HA modeling and previous findings strongly argue that ethanol perturbs BK channel function in the absence of noticeable changes in voltage sensor function.
Our current data also showed that at Ca2+i <2 µM (which includes physiological Ca2+i found in excitable tissues), ethanol’s response of the slo1 channel prototype is potentiation of steady-state current. This effect is related to a drastic increase in the channel’s apparent Ca2+i binding affinity (Kd for Ca2+i decreases nine times in the presence of ethanol). The HA analysis also reveals that the ethanol effect on cbv1 current is an ∼20% decrease in allosteric coupling between Ca2+ binding and channel opening (C). At physiological Ca2+ levels (≤20 µM), the overall ethanol effect is potentiation of slo1 channel activity, which is explained by the drastic decrease in Kd caused by the alcohol. In a previous study, we used an ad hoc, very simple eight-state single-channel model (at constant voltage) to postulate that ethanol-induced inhibition of mslo1 current was caused by ethanol-induced facilitation of channel entry into a low Po mode triggered by nonphysiological, high Ca2+ (Liu et al., 2008). The specific gating parameter or parameters leading to this phenomenon were undetermined. From fitting data to Eq. 5 (Fig. 9), however, we now clearly demonstrate that the mild but significant decrease in BK Po evoked by ethanol at high Ca2+i is caused by the 20% decrease in C evoked by the alcohol.
Extending an early finding on mslo1 channels probed with ethanol in the presence of 1 mM Mg2+i and nominal zero Ca2+i (Liu et al., 2008), our current data unequivocally demonstrate that, in the absence of Ca2+i, cbv1 channels are insensitive to ethanol (Fig. S1) even when Mg2+i was present at sufficient levels (100 mM) to directly gate slo1 channels and increase BK current in the absence of Ca2+i (Fig. S1; Shi et al., 2002). These results, the fact that only Ca2+-dependent parameters in the HA model (Kd and C) are ethanol-sensitive, and previous findings documenting the inability of ethanol to modulate ionic current mediated by other members of the slo channel family whose gating is driven by ions other than Ca2+i (i.e., slo2 and slo3) while the Ca2+-gated MthK channel retains ethanol sensitivity (Liu et al., 2013) indicate that ethanol’s overall effect on slo1 currents is specifically linked to Ca2+i-driven gating.
The combination of electrophysiology and computational methods provides some structural insights on the Ca2+ dependence of ethanol effect on slo1 currents. Combined amino acid substitutions that abolish Ca2+i sensing by the two high-affinity sites mapped to the slo1 protein CTD (i.e., the RCK2 high-affinity site and the D362,D367 site) abolish the channel’s ethanol sensitivity (Liu et al., 2008). Equivalent mutations on a human slo1 gene introduced to Caenorhabditis elegans blunt ethanol-induced intoxication (Davis et al., 2015). In contrast, the same mutations on each high-affinity Ca2+ site, whether in mslo1, cslo1, or hslo1 in C. elegans, fail to blunt ethanol action on slo1 channels (Liu et al., 2008; Davis et al., 2015). Therefore, slo1 channels remain activatable by ethanol as far as Ca2+i is able to interact with either of the Ca2+i high-affinity sites in the slo1 CTD (for further discussion, see Liu et al. ).
We have recently identified in the mslo1 CTD a pocket site of discrete dimensions where ethanol docks and increases Po. This site is fully conserved in cbv1 (which has a 99% identity with mslo1) and includes K361, which must hydrogen bond with ethanol for the alcohol to increase channel activity (Bukiya et al., 2014). Thus, the ethanol recognition site is adjacent to the RCK1 high-affinity Ca2+i-binding site determined by D362 and D367. Ligand-binding sites for ethanol and Ca2+i in slo1 channels are nearby but do not overlap (Liu et al., 2008; Bukiya et al., 2014), which makes alcohol and Ca2+i heterotropic ligands of BK channel–forming proteins. The close spatial location of ethanol and Ca2+ recognition, however, suggests that channel conformational changes upon binding of these two ligands (Ca2+i and ethanol) interact or converge to modulate overall channel activity. Remarkably, the ethanol pocket includes R514, a residue that participates in Ca2+i sensing (Schreiber and Salkoff, 1997; Shi et al., 2002; Zhang et al., 2010; Bukiya et al., 2014). Thus, R514 may serve as a structural link for ethanol action on slo1 channels being allosterically coupled to Ca2+i binding. Consistently, mslo1-K361N, R514N, mslo1-R514N channels display both resistance to ethanol and disrupted Ca2+i sensitivity. Moreover, in the Ca2+i-unbound state, ethanol is unable to hydrogen bond with K361, a chemical interaction that is necessary for ethanol to increase mslo1 channel activity; in the Ca2+-free condition, R514 is shifted far away from the ethanol interaction, making it impossible for this residue to stabilize the K361–ethanol interaction (Bukiya et al., 2014). This model provides an explanation for the requirement of Ca2+i for ethanol to modulate slo1 activity as shown here with cbv1 (Fig. 2 and Fig. S1) and previously with mslo1 channels (Liu et al., 2008, 2013). The overall ethanol response of BK channels may also be determined by long-range interactions between the ethanol and Ca2+-binding sites and other regulatory slo1 domains. For example, CamKII-dependent phosphorylation of Thr107 in the S0–S1 cytosolic linker of bslo1 channels (from bovine aorta) blunts ethanol activation and favors ethanol inhibition of channel activity (Liu et al., 2006). The structural bases underlying functional cross-talking upon ethanol/Ca2+ binding to the slo1 CTD and CamKII interaction with bslo1 Thr107 remain unknown. It should be noted, however, that Thr107 is substituted by nonphosphorylatable residues in mslo1, hslo1, and rslo1, including cbv1 (used in our current study).
Modulation of ethanol effect on slo1-mediated currents by Mg2+i may have pathophysiological implications as alcoholics have reduced levels of Mg2+i (Marrero et al., 2015). A very recent study demonstrated that ethanol’s effect on hslo1 currents expressed in HEK cells, as well as on native BK current in isolated hippocampal neurons, is modulated by Mg2+i. However, this modulation is lost in the presence of cytochalasin D, an actin destabilizer (Marrero et al., 2015). These data are consistent with the hypothesis that Mg2+i levels do not participate in directly modulating ethanol actions on slo1 channels. Indeed, our present data obtained across a wide Mg2+i range (0–100 mM) unequivocally demonstrated that Mg2+i, even when probed at levels that can gate slo1 channels in the absence of Ca2+i, does not empower the cbv1 channels with ethanol sensitivity as Ca2+i does. Consistently, previous data have demonstrated that mutations of the low-affinity Mg2+-sensing mslo1 site (Glu374 and Glu399, which are conserved in cbv1) do not prevent ethanol from potentiating mslo1 channel activity (Liu et al., 2008). Collectively, data discussed so far indicate that ethanol modulation of slo1 channel gating is coupled to Ca2+ binding to this ion’s high-affinity sites in the slo1 protein CTD but not to ion binding to the slo1 channel’s low-affinity site for divalents.
As found for homomeric cbv1 channels, the HA model reveals that the gating processes of cbv1+β1 channels targeted by ethanol are linked to Ca2+-binding gating, whereas intrinsic gating itself and major parameters of voltage gating under study remain unaltered in the presence of the alcohol (see Results section and Figs. 7, 8, 9, 10, 11, 12, and 13). However, within a wide Ca2+ range (3–20 µM) that includes physiological levels of Ca2+i in contracting vascular myocytes (Pérez et al., 2001), ethanol’s overall effect is inhibition of steady-state current for constructs that follow the cbv1+β1 prototype. Thus, HA analysis unveils key distinctions in the gating processes targeted by ethanol in cbv1+β1 when compared with the homomeric slo1 prototype. First, ethanol also increases the channel’s apparent Ca2+i sensitivity, yet this increase (two times when compared with the value in the absence of ethanol) is significantly smaller than that observed with homomeric cbv1 (nine times). Second, ethanol-induced decrease in C is larger for cbv1+β1 than for cbv1: −33% versus −20%. In addition, ethanol significantly decreases the E parameter in cbv1+β1 (−55%) yet does not modify E in homomeric cbv1. Collectively, a reduced increment in Ca2+i binding affinity of the unliganded channel in addition to reduced coupling between Ca2+i binding and channel opening as well as between Ca2+i binding and voltage sensor activation (Table 2) are gating determinants leading to decreased slo1+β1 channel activity in the presence of ethanol.
The HA analyses of our data underscore the limitation of descriptive electrophysiology in interpreting ethanol action and mechanisms on BK channels: an inspection of V0.5 versus Ca2+i plots seems to indicate that at Ca2+i >2 µM, ethanol and β1 have opposite effects on cbv1 currents (Fig. 1 B vs. Fig. 3 D). Thus, ethanol could have been described as having an “anti–BK β1 action,” with possible mechanistic speculations (e.g., the drug alters slo1–β1 coupling). Rather, HA analyses of our data demonstrate that (a) for both slo1 and slo1+β1 channels, intrinsic gating remains unperturbed by ethanol; (b) ethanol targets common gating parameters in slo1 and slo1+β1 channels (Kd and C), yet to a different degree: ethanol quantitative differences on the gating of these constructs very likely contribute to the differential ethanol effect on these channels when evaluated at physiological levels of Ca2+i; (c) ethanol’s differential effect on slo1 and slo1+β1 channels also results from a qualitative difference in ethanol modification of channel gating: the alcohol significantly decreases the E parameter in the latter but not in the former.
The structural bases of β1 modulation of ethanol action on slo1 channels remain speculative mainly because neither the structural basis of slo1 modulation by ethanol (Bettinger et al., 2012; Bettinger and Davies, 2014; Bukiya et al., 2014) nor the structural bases of slo1 and β1 subunit interaction (Wallner et al., 1996; Orio and Latorre, 2005; Morrow et al., 2006; Wu et al., 2009, 2013; Gruslova et al., 2012; Kuntamallappanavar et al., 2014; Castillo et al., 2015) are fully resolved. Thus, we can speculate on several, albeit not mutually exclusive, possibilities: (a) ethanol interaction with the newly identified ethanol-sensing site (slo1 “activation site”; Bukiya et al., 2014) in slo1 CTD itself facilitates ethanol-induced inhibition when β1 subunits are present; (b) slo1 contains additional ethanol recognition sites that mediate a decrease in Po and become increasingly functional in the slo1-β1–coupled heteromer. Interestingly, the nicotinic acetylcholine receptor provides a precedent for a single protein containing distinct n-alkanol recognition sites: one mediating activation and the other inhibiting (Borghese et al., 2014). (c) The β1 subunit itself possesses an ethanol recognition/interaction site or sites, which favor channel inhibition upon ethanol binding; (d) finally, we cannot exclude the possibility that the presence of β1 subunits alters the channel’s lipid microenvironment, known to regulate ethanol action on BK channels (reviewed in Dopico et al. ).
BK β subunits have been widely reported to determine the basal BK current response to modulators, and several pharmacodynamics patterns have been described (Latorre and Contreras, 2013; Dopico et al., 2014; Torres et al., 2014). The chemotherapeutic agent tamoxifen and some endogenous lipids (17β-estradiol, cholane steroids, leukotrienes, and docosahexaenoic acids) require the presence of β1 subunits to increase BK channel steady-state activity (Valverde et al., 1999; Dick et al., 2001). In contrast, β1 subunits are not necessary for but amplify channel steady-state enhancement by phosphatidylinositol bisphosphate (PIP2; Vaithianathan et al., 2008). Accessory subunit-driven pharmacological selectivity has also been reported: corticosterone potentiates β4-containing BK channels but not slo1+wt β2 heteromers, whereas dehydroepiandrosterone potentiates the activity of slo1+wt β2 but not that of slo1+β4 channels (King et al., 2006). Current data underscore that accessory β subunits are not necessary for ethanol to modulate slo1-mediated currents, as previously shown (Liu et al., 2006, 2008; Martin et al., 2008). However, a new and more complex pharmacodynamics pattern emerges, in which β1 and wt β2 subunits favor a decrease in steady-state activity of BK current by ethanol and thus alter the prevalent ethanol response of homomeric slo1, whereas β3 or β4 subunit–containing heteromers respond to ethanol as homomeric slo1 does. Thus, different β subunits condition different qualitative responses of the BK channel to the same pharmacological agent.
In conclusion, our study documents for the first time that, at physiological Ca2+i levels, different β subunits are responsible for different ethanol responses of the same slo1 channels studied in the same expression system and under identical recording conditions. Thus, we unveiled two channels’ ethanol response prototypes: slo1 type and slo1+β1 type, which are, respectively, activated and inhibited when probed with ethanol at physiological levels of Ca2+i. Moreover, HA analysis revealed the specific gating parameters that contribute to the differential ethanol effect of ethanol on these two BK channel prototypes. Our functional approaches, together with recent (Bukiya et al., 2014) and upcoming work addressing the structural bases of ethanol action on BK channels, may lead to designing new agents that target distinct regions in slo1 and/or BK β subunits and/or ethanol-sensitive, specific gating processes in order to oppose ethanol-induced modification of physiology.
Authors thank Bangalore Shivakumar and Maria T. Asuncion-Chin for excellent technical assistance, Jianxi Liu and Phanindra Velisettyfor troubleshooting with some molecular biology procedures, and Anna N. Bukiya (University of Tennessee Health Science Center, Memphis, TN) for helpful discussion.
This work was supported by the American Heart Association Predoctoral Fellowship 13PRE14690014 (to G. Kuntamallappanavar) and National Institutes of Health grants R37AA11560 and R01HL104631 (to A.M. Dopico).
The authors declare no competing financial interests.
Richard W. Aldrich served as editor.