A point mutation in the human Slo1 channel that impairs its sensitivity to omega-3 docosahexaenoic acid

Dietary fats in the form of triglycerides are broken down by lipase in the small intestine, and released free fatty acids are then absorbed into the body. Among the diverse fatty acids, long-chain polyunsaturated omega-3 fatty acids play particularly critical roles in human health (Uauy and Dangour, 2006). Dietary intake of these omega-3 fatty acids, enriched in oily fish, is postulated to have a wide array of health-promoting effects (Saravanan et al., 2010; Mozaffarian and Wu, 2011). For example, the omega-3 fatty acids docosahexaenoic acid (DHA; 22:6(ω-3)), with a 22-carbon chain, and eicosapentaenoic acid (EPA; 20:5(ω-3)), with a 20-carbon chain, may promote healthy cardiovascular function (Ramel et al., 2010; Saravanan et al., 2010; Liu et al., 2011), although recent studies have not yielded unequivocal results (Rizos et al., 2012; Roncaglioni et al., 2013), and some undesirable correlations have been reported in a different organ system (Brasky et al., 2013). The underlying mechanisms of the purported beneficial effects of omega-3 fatty acids are beginning to be investigated, and one of the critical tasks is to identify the molecular targets of these fatty acids. An early effort has revealed that the G protein–coupled receptor 120 involved in the inflammatory response (Oh et al., 2010; Yan et al., 2013) and weight control (Ichimura et al., 2012) is directly activated by DHA with an EC₅₀ of ∼10 µM (Oh et al., 2010). This interaction may contribute to the physiological role of omega-3 fatty acids in the inflammatory process (Im, 2012; Flock et al., 2013; Orr et al., 2013).

Large-conductance Ca²⁺- and voltage-gated K⁺ (Slo1 BK, maxiK, or K_Ca1.1) channels are allosterically activated by intracellular Ca²⁺ and membrane depolarization (Horrigan and Aldrich, 2002; Hoshi et al., 2013a). Slo1 BK channels play important roles in the regulation of numerous physiological processes, including regulation of vascular tone (Nelson and Quayle, 1995; Brenner et al., 2000b), and the channels are well known for their rich repertoire of modulation by multitudes of cellular signaling molecules (Hou et al., 2009), including fatty acids and lipid-derived messengers (Clarke et al., 2002; Sun et al., 2007; Vaithianathan et al., 2008; Lai et al., 2009; Wang et al., 2011). Recently, we showed that DHA applied to either side of the membrane potently activates vascular BK channels made of pore-forming Slo1 and auxiliary β1 or β4 subunits (Hoshi et al., 2013c). When examined in inside-out membrane patches, its EC₅₀ is estimated to be ∼500 nM, and the stimulatory effect has a fast onset and is reversible on wash-out (Hoshi et al., 2013c). This action on the Slo1 BK channel underlies the acute hypotensive effect of DHA observed when it is injected into anesthetized mice because the effect is absent in mice with the Slo1 gene disrupted (Hoshi et al., 2013c). Unlike DHA, its derivatives, 17-hydroxy DHA (17OH DHA) with a hydroxyl moiety in the tail group and DHA ethyl ester (DHA EE) with an ethyl ester moiety in the head group, are without effect on Slo1 BK channels (Hoshi et al., 2013c). The stimulatory effect of DHA is particularly noticeable when the pore-forming Slo1 subunit is coexpressed with the auxiliary subunit β1, as found in vascular BK channel complexes (Knaus et al., 1994; Wallner et al., 1995), or with β4, as frequently found in neurons (Brenner et al., 2000a; Meera et al., 2000), leading to an increase in macroscopic currents by up to ∼20-fold at some voltages (Hoshi et al., 2013b). When DHA is applied to Slo1 BK channels without heterologously expressed auxiliary subunits, a small current-enhancing effect of DHA is observed, strongly indicating that the presence of β1 or β4 potentiates the functional consequence of DHA binding to the Slo1 subunit (Hoshi et al., 2013b).

In this study, we sought to reveal the biophysical mechanism of the effect of DHA on the Slo1 channel and determine the molecular elements within Slo1 critical for the stimulatory effect. Prompted by the finding that DHA at a functionally saturating concentration for human Slo1 (hSlo1) is without effect on Drosophila melanogaster Slo1 (dSlo1), we used chimeric constructs encompassing dSlo1 and hSlo1 channels. Our electrophysiological measurements show that a single residue near the cytoplasmic end of S6 plays a critical role in the differential responses of hSlo1 and dSlo1 channels to select free long-chain omega-3 fatty acids.

Human embryonic kidney tsA cells were transiently transfected with plasmid DNAs encoding Slo1 channels and monomeric enhanced green fluorescent protein (GFP) as a transfection marker using FuGene6 (Roche). The cells were grown in Dulbecco’s modified Eagle’s medium, 10% FBS, 1% penicillin/streptomycin at 37°C, and 5% CO₂, and used in experiments typically 1–3 d after transfection. The amino acid residue numbering of hSlo1 described here is based on GenBank accession number AAB65837, and that of dSlo1 is based on RefSeq accession number NP_001014653. In some experiments, the plasmids coding for hSlo1 β1 (hβ1; RefSeq accession no. NP_004128) fused with enhanced GFP at the C terminus or hSlo1 β4 (hβ4; GenBank accession no. AAF69805) fused with GFP at the C terminus (Hoshi et al., 2013b) were cotransfected with the hSlo1 plasmid (weight ratio of 1:1) as described previously (Hoshi et al., 2013b). Functional assembly of hSlo1 and hβ subunits in each patch was confirmed as described previously (Hoshi et al., 2013b). The presence of the C-terminal GFP tag does not alter the response of the hSlo1+hβ1 or hSlo1+hβ4 complex to DHA (Hoshi et al., 2013c).

Single–amino acid mutations were introduced into hSlo1 (pCI-neo) and dSlo1 (pRc/CMV) using a commercially available kit (QuickChange; Agilent Technologies), and the sequences were verified. The chimeric constructs were generated using the method of Yon and Fried (1989).

The experiments were performed at room temperature as described previously (Horrigan et al., 2005). For the pharmacological experiments described here, only those results obtained from membrane patches that formed high resistance seals very quickly, less than several seconds, without excess negative pressure were used. The membrane patches formed in this manner were devoid of much cytoplasmic material, based on visual inspection with bright field and/or fluorescent microscopy, and the channels contained therein responded quickly to DHA and other agents (Hoshi et al., 2013b). Large omega-shaped membrane patches, although they produced large ionic currents, were not used for the pharmacological experiments described. Macroscopic currents were recorded with pipettes with initial resistances of 0.5–2 MΩ, and 60–70% of the initial resistance was electronically compensated. Unless otherwise noted, capacitive and leak currents have been subtracted from the results shown. For single-channel measurements, the pipette diameter was adjusted to obtain desired numbers of channels. The output signal of the amplifier (Axopatch 200A or 200B; Molecular Devices) was filtered through the amplifier’s built-in 10-kHz filter and digitized.

The results were analyzed using IgorPro (WaveMetrics) as described previously (Horrigan et al., 2005). Normalized macroscopic G-V curves were estimated from extrapolated instantaneous tail current sizes and characterized by a Boltzmann function,

where G(V_m)/G_max represents the normalized conductance at the membrane potential V_m; I(0, V_m)/I_max (0) is the normalized extrapolated instantaneous tail current size after a pulse to V_m; and F, R, and T have their usual meanings. The half-activation voltage (V_0.5) and the number of apparent equivalent charges (Q_app) were used as data description parameters. Macroscopic ionic current kinetics was fitted with a single exponential. The time constant values were estimated from currents elicited by depolarization (triangles in the figures) and from tail currents (inverse triangles in the figures).

Single-channel ionic currents were recorded through the 500-MΩ feedback resistor of the amplifier, filtered through the built-in 10-kHz filter, and digitized at 83 or 100 kHz. Single-channel open probability (P_o) was inferred by generating all-point amplitude histograms, and the number of channels present in each patch was estimated from the macroscopic current size at 220 mV (Horrigan et al., 2005). In single-channel experiments with dSlo1, the number of channels present was not estimated. The single-channel open duration analysis was performed as described previously (Avdonin et al., 2003; Horrigan et al., 2005), and no left-censor correction was made.

The extracellular solution contained (mM): 140 KCl, 2 MgCl₂, and 10 HEPES, pH 7.2 with N-methyl-d-glucamine (NMG). The Ca²⁺-free internal solution contained (mM): 140 KCl, 11 EGTA, 0.02 18C6TA, and 10 HEPES, pH 7.2 with NMG. Other solutions used are described in the legends to the figures. The fatty acid solutions were prepared immediately before measurements as described previously (Hoshi et al., 2013b,c). DHA was applied after the ionic currents stabilized after patch excision. This was particularly important for the experiments using dSlo1 channels, which often showed noticeable “run-up.” Unless otherwise noted, DHA was typically applied at 3 µM because this concentration represents a nearly saturating concentration to stimulate BK channels (Hoshi et al., 2013c; also see Fig. S1). DHA, α-linolenic acid (ALA), and arachidonic acid (AA) were obtained from Sigma-Aldrich. 17OH DHA, DHA EE, and EPA were purchased from Cayman Chemical Company.

Equation parameters are presented as mean ± 95% confidence interval as implemented in IgorPro. Statistical results are presented in the text as mean ± SEM (n), where n is the number of independent measurements. The error bars in the figures represent SEM. Statistical comparisons of the results were performed using the two-sided Mann-Whitney or Wilcoxon test, with the significance level of 0.05 in IgorPro as appropriate. For comparisons of three or more groups, the significance level was corrected using the Bonferroni method.

Fig. S1 illustrates that increasing the concentration of DHA from 3 to 10 µM does not lead to any additional increase in currents through hSlo1 channels; 3 µM DHA is a saturating concentration. Fig. S1 also shows that the stimulatory effect of DHA is reversible. Fig. S2 presents G-V curves of hSlo1–dSlo1 chimeric channels in the absence of Ca²⁺ and DHA. Fig. S3 shows single-channel openings from hSlo1(1:327)–dSlo1(342:1164) where the entire C-terminal area of hSlo1 is replaced with that of dSlo1. DHA increases open probability in this chimeric channel, illustrating that the S0–S6 segments of hSlo1 possess the molecular elements required for the stimulatory action of DHA. Fig. S4 depicts the essential characteristics of the stimulatory action of 3 µM DHA in the absence of Ca²⁺ in hSlo1+hβ1; DHA increases currents by up to ∼20-fold (A and B) and shifts G-V by approximately −60 mV. Fig. S5 compares the effects of DHA on the kinetics of ionic current relaxation in hSlo1+hβ1 and hSlo1 Y318S+hβ1.

3 µM DHA applied to the intracellular side in the virtual absence of Ca²⁺ increased macroscopic K⁺ currents through hSlo1 channels in the absence of heterologously expressed auxiliary subunits (herein referred to as hSlo1 channels; Fig. 1). As found with Slo1+β1 channels (Hoshi et al., 2013c), 3 µM DHA is a saturating concentration in increasing Slo1 currents; 10 µM caused no further increase (Fig. S1 A). Greater fractional increases were observed at less depolarized voltages (Fig. 1, A and B), and the increase in macroscopic current peaked at ∼2.5-fold (Fig. 1 B). After DHA application, the macroscopic G-V curve shifted by −10.3 ± 1.5 mV (11) (Fig. 1, C and D; P < 0.0001) without a significant change in its steepness (Q_app after DHA/Q_app before = Q_app ratio = 0.94 ± 0.03 [11]; Fig. 1 D; P = 0.08). The shift in G-V V_0.5 (ΔV_0.5) in hSlo1 channels was modest and much smaller than that previously observed in hSlo1+hβ1 channels, approximately −58 mV (Hoshi et al., 2013c), but the effect was reversible (Fig. S1 B). The application of DHA also modestly altered the kinetics of ionic currents. Although the kinetics at extreme negative and positive voltages remained largely unaltered, the time course of current relaxation was noticeably faster at intermediate voltages (e.g., 160 mV, where G/G_max is ∼0.5) after DHA application (Fig. 1, E–G).

In hSlo1+hβ1 channels, we reported that activation of the voltage-sensor domain (VSD) was not required for the stimulatory effect of DHA because it increased the channel open probability (P_o) robustly (>20-fold) at very negative voltages in the absence of Ca²⁺ (Hoshi et al., 2013c). In hSlo1 channels, we found a similar but smaller (approximately fourfold) increase in P_o at negative voltages without Ca²⁺ (Fig. 2, A and B). The mean open durations remained relatively unaltered (1.3- to 1.6-fold increases; Fig. 2 B, bottom) after DHA application. Because P_o at these negative voltages is determined primarily by a single closed–open transition of the ion conduction gate (Horrigan et al., 1999), DHA most probably increases the rate constant of gate opening by 3- to 2.5-fold.

We also showed earlier that the effect of DHA on hSlo1+hβ1 channels was maintained at [Ca²⁺]_i ≥ 100 µM (Hoshi et al., 2013c), where the Ca²⁺ sensors are largely saturated with Ca²⁺ (summarized in Hoshi et al., 2013a). In hSlo1 channels, DHA was also effective at 300 µM [Ca²⁺]_i, increasing currents by up to fourfold (Fig. 3, A–C) and causing a small but noticeable left-shift in G-V (Fig. 3 D).

The electrophysiological effects of DHA on hSlo1 are thus similar to those observed on hSlo1+β1. In both channel types, DHA promotes opening of the ion conduction gate without any need for activation of the VSDs or the Ca²⁺ sensors. However, the effect of DHA on the ion conduction gate is clearly much greater in the hSlo1+β1 channel than in the hSlo1 channel. The results are collectively consistent with the idea that the functional consequence of the interaction of DHA with the Slo1 protein is amplified by the presence of β1 (Hoshi et al., 2013c).

In humans, DHA is an essential fatty acid and plays critical roles in various organs, especially in the brain (Lauritzen et al., 2001; Horrocks and Farooqui, 2004). In contrast, the fruit fly Drosophila normally lacks DHA in the body (Shen et al., 2010). As an initial step toward identification of the molecular elements within the hSlo1 channel critical for the stimulatory effect of DHA, we examined how DHA affected hSlo1 and dSlo1 channels. Currents through dSlo1 channels were very slow to activate and required much greater and longer depolarization than that needed for hSlo1 channels; in the absence of Ca²⁺, the mean V_0.5 value for dSlo1 was 245.0 ± 1.0 mV (5) compared with the mean V_0.5 value of 156.2 ± 0.78 mV (11) for hSlo1 (Fig. 4, A and B). The slow kinetics and the right-shifted voltage dependence of steady-state activation made it impractical to systematically study the effects of DHA on dSlo1 G-V as performed with hSlo1 (see Fig. 1); the integrity of high resistance seals was easily compromised by long and large voltage pulses. Thus, we opted to examine the outward currents through dSlo1 only at 140, 170, and 200 mV, which correspond roughly to G/G_max of 0.02, 0.06, and 0.15 based on the results shown in Fig. 4 B. In hSlo1 channels, at the equivalent G/G_max voltages, DHA has clear current-enhancing effects (see Fig. 1). The application of 3 µM DHA to dSlo1 channels produced virtually no change in currents at these three voltages for up to 200 s or more (Fig. 4 C, left). In comparison, DHA increases currents through hSlo1 within 50 s of application (Fig. 4 C, right). The mean fractional change in the dSlo1 current size was at most only 1.19 ± 0.07 (17) (at 140 mV; Fig. 4 D). When the fractional changes in dSlo1 and hSlo1 currents were plotted against their estimated G/G_max values, the ineffectiveness of DHA on dSlo1 was made even more evident (Fig. 4 E). Additionally, single-channel measurements failed to detect any significant effect of 3 µM DHA at negative voltages where the VSDs are expected to be at rest (P = 0.94, 0.22, and 0.16 at −40, −80, and −120 mV, respectively; Fig. 4, F and G), and the relative changes in P_o in dSlo1 by 3 µM DHA were different from those found in hSlo1 at the same voltages (P = 0.001, 0.0003, and 0.0006 at −40, −80, and −120 mV, respectively). Drosophila possesses cytoplasmic auxiliary subunits that are structurally distinct from vertebrate β subunits (Schopperle et al., 1998); therefore, we did not test if the Drosophila auxiliary subunits alter the response to DHA.

The contrasting effects of DHA on hSlo1 and dSlo1 prompted us to generate chimeric constructs encompassing the two Slo1 channel types, and some of the chimeric channels examined are shown in Fig. 5. Because many of the chimeric channels did not express well and/or had right-shifted voltage dependence (Fig. S2), their G-V parameters were difficult to estimate accurately. Thus, we used fractional changes in peak outward current as the dependent variable. DHA noticeably increased currents through hSlo1(1:631)–dSlo1(648:1164), in which essentially the RCK2 segment of hSlo1 (residues 631 through 1113) is replaced with that of dSlo1 (residues 648 through 1164; Fig. 5 B, top). Substitution of the entire C-terminal cytoplasmic area of hSlo1 containing both the RCK1 and RCK2 domains with the corresponding area of dSlo1 (hSlo1(1:327)–dSlo1(342:1164)) drastically decreased the expression efficacy and also shifted the voltage dependence of activation to the positive direction (V_0.5 = 270.8 ± 2.6 mV [4]); however, the currents recorded were enhanced by DHA (Fig. 5 B, second from top, and Fig. S3). In contrast, substitution of the S0–S6 segments of hSlo1 with the corresponding segments of dSlo1 (dSlo1(1:341)–hSlo1(328:1113)) obliterated the stimulatory effect of DHA (Fig. 5 B, third from top), potentially signifying the importance of the transmembrane segments. Importantly, DHA clearly enhanced currents through dSlo1(1:249)–hSlo1(235:327)–dSlo1(342:1164), in which the S5-P-S6 segments of dSlo1 are replaced with those of hSlo1 (Fig. 5 B, bottom). The fractional increases in peak outward currents of this chimeric construct resembled those from hSlo1 channels (Fig. 5 B, bottom right).

The results from the hSlo1–dSlo1 chimeric constructs suggested the importance of the pore domain encompassing the S5, P, and S6 segments in mediating the current-enhancing effect of DHA observed in hSlo1. To further delineate the structural elements required for the DHA action, we introduced human-to-Drosophila point mutations in the pore (S5-P-S6) domain of the hSlo1 channel. The hSlo1 channel was used as the “background” channel because it expressed better than dSlo1 and its voltage dependence was much more amenable for electrophysiological measurements. The extracellular-facing S5-P linker segment, which is not well conserved among Slo1 channels from different species (e.g., Drosophila vs. Periplaneta americana), was not included in this mutagenesis strategy. Because the stimulatory effect of DHA on hSlo1 channels without any auxiliary subunit is small and difficult to study, producing only a ΔV_0.5 of about −10 mV (Fig. 1), we initially compared the effects of DHA on the hSlo1 channels with human-to-Drosophila mutations when coexpressed with hβ1. In wild-type hSlo1+hβ1 channels, DHA causes a readily discernible shift in G-V, of approximately −60 mV (Fig. S4). The strategy here was to identify those human-to-Drosophila mutations in hSlo1 that altered ΔV_0.5 by DHA when coexpressed with β1 and then to test the Slo1 mutant identified alone without β1. The Drosophila-to-human point mutations in hSlo1 effected a variety of changes to hSlo1+β1 (Figs. 6–8 and Table 1). Some of the mutations shifted G-V to the negative direction (V236A in S5 [Fig. 6, top] and L280F in P [Fig. 7, top]), whereas others shifted G-V to the positive direction. In particular, the mutations F252I in the S5 segment (Fig. 6) and M282I in the P segment (Fig. 7) markedly moved V_0.5 to ∼250 mV, similar to the mean V_0.5 value of dSlo1. In these extreme right-shifted mutants, accurate determinations of the G-V parameters were difficult. The Drosophila-to-human mutations I308L and G310V in S6 individually rendered the resulting channels electrophysiologically nonfunctional. However, the double mutant hSlo1 I308L:G310V was functional. In almost every mutant channel complex examined, DHA increased currents and caused a clear shift in V_0.5 to the negative direction (Figs. 6–9). The only exception was hSlo1 Y318S+hβ1. In this channel complex bearing a mutation toward the C-terminal end of S6, DHA caused negligible changes in current size and V_0.5 (Figs. 8 and 9). DHA markedly accelerates the activation kinetics and slightly slows the deactivation kinetics in wild-type Slo1+hβ1 (Hoshi et al., 2013c). In contrast, DHA only moderately accelerated the current relaxation kinetics at all voltages in hSlo1 Y318S+hβ1 (Fig. S5). We verified that hSlo1 Y318S functionally assembles with hβ1 by confirming the characteristically slower activation and deactivation when compared with hSlo1 Y318S alone (Fig. 10). Furthermore, the voltage dependence of activation of hSlo1 Y318S+hβ1 at a wide range of [Ca²⁺]_i without DHA closely resembled that of wild-type hSlo1+hβ1 (Fig. 10 B).

Even without coexpression of β1, 3 µM DHA caused no measurable change in peak outward current size, G-V, or kinetics in hSlo1 Y318S (Fig. 11, A–F). The application of a greater concentration of DHA, 10 µM, also failed to increase currents through hSlo1 Y318S and, in fact, often decreased currents (Fig. 11), potentially suggesting that DHA may exert multiple effects on the hSlo1 channel depending on the concentration.

We showed previously that DHA markedly increases currents through vascular hSlo1+hβ1 channels as well as through predominantly neuronal hSlo1+hβ4 channels, causing a ΔV_0.5 of approximately −60 mV (Hoshi et al., 2013b). The human-to-Drosophila mutation Y318S virtually eliminated the DHA sensitivity in hSlo1 when coexpressed with hβ4 (Fig. 12). We confirmed that hSlo1 Y318S functionally assembled with hβ4 based on the characteristically slow activation kinetics (Fig. 12 A); however, neither the peak outward current size nor the G-V characteristics in hSlo1 Y318S+hβ4 was altered by 3 µM DHA (Fig. 12, B–D).

Currents through hSlo1+hβ1 channels are enhanced by omega-3 and omega-6 fatty acids of different types to various degrees (Hoshi et al., 2013c). Our previous structure–activity experiments suggested the importance of both the polar “head” group and the long hydrophobic “tail” group. DHA EE with an ethyl ester group in the head group is essentially ineffective in modulating in hSlo1+hβ1 channels. DHA with a 22-carbon tail is more effective in shifting V_0.5 than the shorter omega-3 fatty acids EPA, with a 20-carbon tail, and ALA (18:3(ω-3)), with an 18-carbon tail. The presence of an OH moiety in the tail group in 17OH DHA also greatly impairs the effectiveness. AA (20:4(ω-6)), an omega-6 fatty acid with a 20-carbon chain, is less effective than DHA. We compared the effectiveness of these fatty acids (Fig. 13 A) on hSlo1 Y318S+hβ1 channels by measuring the ΔV_0.5 and Q_app ratio values. To facilitate comparison, the results obtained with wild-type hSlo1+hβ1 channels (Hoshi et al., 2013b,c) are illustrated with gray bars in Fig. 13 B, and the differences in ΔV_0.5 observed between wild-type hSlo1+hβ1 and hSlo1 Y318S+hβ1 channels are summarized in Fig. 13 C. When corrected for multiple comparisons, only the ΔV_0.5 values for DHA and EPA are significantly different between wild-type hSlo1+hβ1 and hSlo1 Y318+hβ1 (P < 10⁻⁷). It is noteworthy that ALA, the shortest omega-3 fatty acid tested, and AA, an omega-6 fatty acid, remained equally effective in wild-type hSlo1+hβ1 and hSlo1 Y318S+hβ1 (Fig. 13, B, left, and C). The effect of the mutation Y318S on hSlo1 may be preferential to DHA and EPA, two long-chain omega-3 fatty acids.

We introduced the Drosophila-to-human mutation S332Y into dSlo1, the converse of the human-to-Drosophila mutation hSlo1 Y318S, and examined whether the mutation introduced any noticeable DHA sensitivity to the dSlo1 channel. The voltage dependence of activation of dSlo1 S332Y remained shifted to the positive direction compared with that of hSlo1 (Fig. 14, A and B), and we therefore compared ionic current sizes at three different voltages (Fig. 14 C). DHA did not markedly alter the currents through dSlo1 S332Y (Fig. 14, C and D).

Our ionic current measurements presented here show that the mechanism of the action of DHA on the hSlo1 channel without any β subunit, at least qualitatively, resembles that observed in the hSlo1+hβ1 channel (Hoshi et al., 2013c). In both hSlo1 and hSlo1+hβ1, DHA increases P_o at very negative voltages in the virtual absence of Ca²⁺. Thus, DHA most probably increases L₀, which reflects the closed–open equilibrium of the ion conduction gate in the model of Horrigan and Aldrich (2002) (“HA model”), consequently shifting macroscopic G-V to the negative direction in a relatively Ca²⁺-independent manner. Of the two rate constants contributing to L₀, we propose that DHA predominantly increases the forward closed-to-open rate constant (δ₀ in the HA model) at extreme negative voltages where the VSDs are at rest, because the mean open durations at such voltages are not appreciably altered. If similar changes are maintained when the VSDs of the channel are activated (δ_1–4 in the HA model), it would account for the acceleration of the activation kinetics by DHA at intermediate voltages in both hSlo1 and hSlo1+hβ1. At very positive voltages where G/G_max is nearly saturated, the activation kinetics is not significantly faster with DHA (Fig. 1, F and G) (Hoshi et al., 2013b,c); the effects of DHA on the forward rate constants may not be uniform, and may depend on activation of the VSDs. It should be noted that the confidence on the estimate of the effect of DHA on L₀ may be somewhat limited by the brief nature of the channel openings at extreme negative voltages, often ∼100 µs (Fig. 2). Furthermore, our measurements do not necessarily preclude any action of DHA on other aspects of the Slo1 channel, such as the VSD function, although we expect it to be relatively minor if any. The underlying biophysical mechanisms may be similar qualitatively; however, DHA clearly exerts a much greater effect in hSlo1+hβ1 than in hSlo1 alone. In hSlo1+hβ1, the P_o values at negative voltages without Ca²⁺ increase by >20-fold (Hoshi et al., 2013c), but the increases are limited only to approximately threefold in hSlo1 alone. Thus, as proposed earlier (Hoshi et al., 2013c), coexpression of hβ1 may be considered to “amplify” the functional consequences of the interaction of DHA with the hSlo1 protein itself. The physicochemical nature of this amplification process remains to be investigated, but our previous study shows that a possible electrostatic interaction in the N terminus of some hβ subunits is critical (Hoshi et al., 2013b). Our measurements thus demonstrate that the principal effector of DHA is the ion conduction gate of the hSlo1 BK channel. In Shaker K⁺ channels, however, DHA (at >10 µM) electrostatically interacts with the VSDs (Börjesson et al., 2008); the effects of DHA may vary depending on the channel type.

The lack of responsiveness to DHA of Slo1 channels from Drosophila, which does not normally contain DHA in the body (Shen et al., 2010), has allowed us to deploy a chimeric strategy to identify the molecular component critical for the differential effects of DHA on dSlo1 and hSlo1 channels. Mutation of Y318 located most probably at the cytoplasmic end of S6 in hSlo1 to S, as found in dSlo1, impairs the response to DHA in hSlo1, hSlo1+hβ1, and hSlo1+hβ4. Our results did not identify any other statistically significant contributions. Importantly, in the absence of DHA, the gating characteristics of the mutant channel complexes containing the mutation Y318S otherwise closely resemble those of the wild-type channels. The observation that Y318 in S6, near the ion conduction gate located somewhere near the selectivity filter (Wilkens and Aldrich, 2006; Tang et al., 2010; Chen and Aldrich, 2011; Zhou et al., 2011; Thompson and Begenisich, 2012), plays a critical role may be consistent with the conclusion of the biophysical measurements that DHA alters the equilibrium constant L₀ in the HA model, representing the intrinsic stability of the ion conduction gate. The converse mutation S332Y in dSlo1 fails to introduce any DHA sensitivity to dSlo1; multiple structural components must be missing in the gating machinery of dSlo1 to respond to DHA.

It is clear that the mutation Y318S impairs the response of the hSlo1 channel to DHA. However, the results presented here do not specify the exact functional role of Y318 in the DHA sensitivity of hSlo1. This residue could be a structural component of the DHA-binding site itself or a component of the coupling mechanism that bridges the binding site located elsewhere to the effector site, probably the ion conduction gate near the ion selectivity filter (Wilkens and Aldrich, 2006; Tang et al., 2010; Chen and Aldrich, 2011; Zhou et al., 2011; Thompson and Begenisich, 2012). A similar interpretational issue was raised by Zhou et al. (2010) regarding the role of hSlo1 G311 in mediating the inhibitory effect of paxilline, a tremogenic alkaloid, on the Slo1 channel. An experimentally determined atomic structure of the Slo1 transmembrane domain is not yet available, and homology models based on the Kv1.2/2.1 structure may be of limited usefulness (Zhou et al., 2011); thus, the exact location of Y318 in hSlo1 is unknown. However, based on its location in the primary structure, Y318 is expected to be near the C-terminal end of S6 and may be exposed to the cytoplasmic side such that DHA can gain access to the residue through the narrow gap between the transmembrane domain and the cytoplasmic gating ring domain (Yang et al., 2008, 2013). Alternatively, it is conceivable that the Slo1 channel could have lateral openings through the transmembrane segments as found in voltage-independent two-pore domain K⁺ channels (Brohawn et al., 2012; Miller and Long, 2012), which are modulated by AA (Fink et al., 1998), and also in the bacterial Na⁺ channel NavAb (Payandeh et al., 2011). It is interesting to note that Y318 is three and six residues away from F315 and L312, respectively, which together may mediate coupling of the ion conduction gate with the Ca²⁺ sensors (Wu et al., 2009).

A particularly interesting effect of the human-to-Drosophila mutation Y318S is that the response of hSlo1 Y318S+hβ1 to DHA is virtually eliminated, but those to other fatty acids, such as the shorter omega-3 fatty acid ALA and also the omega-6 fatty acid AA, remain unaltered. Thus, it may be interpreted that the mutation Y318S alters the stimulus rank order of the hSlo1+hβ1 channel complex and probably that of the hSlo1 channel without any auxiliary subunit. In wild-type channels, the stimulus rank order based on the mean ΔV_0.5 values is DHA > AA = EPA > ALA > 17OH DHA > DHA EE. With Y318S, the order changes to AA > ALA = EPA > DHA EE > DHA > 17OH DHA. Changes in the stimulus/ligand rank order are often observed when binding sites themselves are altered (e.g., Drake et al., 1997). Therefore, the altered rank order observed in this study may suggest that hSlo1 Y318 is within or near the free fatty-acid sensor of the hSlo1 channel. The Y318S-independent effect of AA probably represents the same phenomenon reported by Clarke et al. (2002) using native channel complexes from pulmonary smooth muscles cells, most probably Slo1+β1.

Crystal structures of some proteins with DHA bound are available: for example, brain fatty acid–binding protein (Protein Data Bank accession no. 1FDQ; Balendiran et al., 2000), nuclear retinoid X receptor (Protein Data Bank accession no. 1MV9; Egea et al., 2002), and cyclooxygenase-2 (Protein Data Bank accession no. 3HS7; Vecchio et al., 2010). None of these proteins are transmembrane proteins, and how relevant they are to Slo1 is unclear. Nevertheless, DHA in these structures assumes a semicircular omega shape, and the carboxylic acid head group interacts with the side chain of an arginine (Protein Data Bank accession no. 1MV9; Egea et al., 2002) or of a tyrosine (Protein Data Bank accession nos. 1FDQ and 3HS7; Vecchio et al., 2010). Thus, the idea that hSlo1 Y318 directly interacts with DHA may be feasible.

The observation that the response of hSlo1+hβ1 to AA, an omega-6 fatty acid, is unaltered by the mutation Y318S could be explained at least in two ways. First, AA may bind to the same site as DHA, but the mutation Y318S somehow does not interfere with the AA-induced conformational change. Second, AA may increase the channel activity through a different mechanism unaltered by the mutation Y318S. These possibilities are not well discriminated by the results presented here, but our observations clearly illustrate the complexity of lipid-based modulation.

Although the direct action of free DHA on the hSlo1 channel complex critically depends on hSlo1 Y318, DHA probably exerts other effects on the channel complex independently of Y318. As suggested by the current-diminishing effect of 10 µM DHA on hSlo1 Y318S, DHA may inhibit or block the channel at higher concentrations. Inhibitory effects of free fatty acids at micromolar levels were also observed in other K⁺ channels (Guizy et al., 2005). Additionally, DHA is capable of regulating native BK channel complexes indirectly through its metabolic breakdown products involving the enzyme cytochrome P450 epoxygenase (Ye et al., 2002; Lai et al., 2009; Wang et al., 2011). One of the challenges is therefore to identify and delineate the multiple lipid-based pathways converging on BK channel complexes.

In summary, our study demonstrates that DHA facilitates opening of the ion conduction gate of the Slo1 BK channel, and that a single residue at the cytoplasmic end of S6 accounts for the differential effects of DHA on hSlo1 and dSlo1. The human-to-Drosophila mutation Y318S impairs the direct current enhancing effect of DHA and EPA preferentially and represents an excellent tool to dissect out the complex molecular mechanisms underlying modulation of BK channels by fatty acids.

This work was supported by the public in part through the National Institutes of Health (R01GM057654), German Research Foundation (DFG, HE 2993/8), National Natural Science Foundation of China (31271217), and the Key Project of Shanghai Science and Technology Commission (contract no. 11JC1406400). The authors declare no conflict of interest.

Kenton J. Swartz served as editor.