Ion channels are often modulated by changes in extracellular pH, with most examples resulting from shifts in the ionization state of histidine residue(s) in the channel pore. The application of acidic extracellular solution inhibited expressed KCa2.2 (SK2) and KCa2.3 (SK3) channel currents, with KCa2.3 (pIC50 of ∼6.8) being approximately fourfold more sensitive than KCa2.2 (pIC50 of ∼6.2). Inhibition was found to be voltage dependent, resulting from a shift in the affinity for the rectifying intracellular divalent cation(s) at the inner mouth of the selectivity filter. The inhibition by extracellular protons resulted from a reduction in the single-channel conductance, without significant changes in open-state kinetics or open probability. KCa2.2 and KCa2.3 subunits both possess a histidine residue in their outer pore region between the transmembrane S5 segment and the pore helix, with KCa2.3 also exhibiting an additional histidine residue between the selectivity filter and S6. Mutagenesis revealed that the outer pore histidine common to both channels was critical for inhibition. The greater sensitivity of KCa2.3 currents to protons arose from the additional histidine residue in the pore, which was more proximal to the conduction pathway and in the electrostatic vicinity of the ion conduction pathway. The decrease of channel conductance by extracellular protons was mimicked by mutation of the outer pore histidine in KCa2.2 to an asparagine residue. These data suggest that local interactions involving the outer turret histidine residues are crucial to enable high conductance openings, with protonation inhibiting current by changing pore shape.
The KCa2.2 (SK2) and KCa2.3 (SK3) potassium channels are characterized by their activation by intracellular calcium (Ca2+; half-maximal activation of 0.3–0.7 µM), voltage independence, and small single-channel conductance of 10–20 pS (Köhler et al., 1996; Hirschberg et al., 1998; Xia et al., 1998). Because of the distinct spatial distribution of the channel subtypes in the mammalian brain and their involvement in the generation of afterhyperpolarization currents, there has been considerable interest in developing subtype-selective pharmacological tools to study these channels (Stocker and Pedarzani, 2000; Shakkottai et al., 2001; Sailer et al., 2002, 2004; Stocker et al., 2004; Dilly et al., 2005). The predicted pore regions of KCa2.2 and KCa2.3 display high sequence homology. Each channel subunit possesses a histidine (H) residue that is predicted to reside within the extracellular-facing region of the outer pore (H337 in KCa2.2 and H491 in KCa2.3). In addition, the KCa2.3 subunit exhibits an additional H (H522) that is seen as asparagine (N368) in KCa2.2. Histidine residues in channels’ outer pore regions have been reported to be proton interaction sites that mediate inhibition of currents. For example, proton-induced inhibition is observed for several different potassium channels, including the voltage-gated Shaker-like plant channel ATK3 (Geiger et al., 2002), Kv1.5 (Steidl and Yool, 1999; Jäger and Grissmer, 2001; Kehl et al., 2002), Kv1.4 (Claydon et al., 2000), Kv7.2/7.3 (Prole et al., 2003), K2P2.1 (Cohen et al., 2008), and the inward-rectifying channel Kir2.3 (Coulter et al., 1995; Ureche et al., 2008). Several mechanisms have been proposed to explain this inhibition. For example, Kv1.5 macroscopic current block is mediated through modifying gating and enhancing inactivation without any effect on single-channel conductance (Kwan et al., 2006). Macroscopic inhibition of both ATK3 and Kir2.3 channel currents resulted from a reduction in single-channel conductance (Coulter et al., 1995; Geiger et al., 2002), whereas inhibition of Kv7.2/7.3 heteromeric channel currents resulted from both effects on gating and single-channel conductance (Prole et al., 2003).
We demonstrate that macroscopic KCa2.2 and KCa2.3 current is inhibited by extracellular protons, with KCa2.3 being more sensitive. Inhibition resulted from a reduction in single-channel conductance and, in particular for KCa2.3, a shift of inward rectification. We show that the outer pore histidine residues in the KCa2.2 and KCa2.3 channel act as proton sensors. The observed significant differences in modulation between the two subtypes provide insight into the role of such outer pore residues in channel function that might aid the development of subtype-selective modulators.
MATERIALS AND METHODS
Cell culture, expression of channels, and mutagenesis
Wild-type rat KCa2.2 (GenBank accession no. NM_019314) and human KCa2.3 (GenBank accession no. AF031815) channel DNA were subcloned into the mammalian plasmid expression vectors pcDNA3 (Invitrogen) and pFlagCMV2 (Sigma-Aldrich), respectively. Point mutations in KCa2.2 (KCa2.2(N368H), KCa2.2(H337N), and KCa2.2(H337N, N368H)) and KCa2.3 (KCa2.3(H522N) and KCa2.3(H421N)) were introduced using a site-directed mutagenesis kit (Quickchange-XL; Agilent Technologies) and confirmed by dye termination DNA sequencing.
Channels were transiently expressed in HEK293 cells for electrophysiology. Cells were dissociated for passage using an EDTA solution and maintained in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum and 1% penicillin/streptomycin (all from Invitrogen) at 37°C. Cells were plated onto 35-mm dishes (BD) 48 h before transfection for electrophysiology. Transient transfections of HEK293 cells were made using polyethylenimine (Johnson Matthey Plc) by combining channel plasmid DNA with enhanced green fluorescent protein DNA in a ratio of 1:1 to 1:10 (maximum plasmid content, 1 µg). Cells expressing enhanced green fluorescent protein were used for electrophysiology 12–24 h after transfection.
Expressed macroscopic currents were resolved using either whole cell or outside-out macropatch recording, with cells/patches being constantly superfused with external solutions of 160 or 5 mM potassium ([K+]o). The composition of the 160 mM K+ (high K+) solution was (in mM): 120 KCl, 10 HEPES, 10 EGTA, 1.44 MgCl2, and 6.19 CaCl2 (achieving a buffered concentration of 60 nM Ca2+ and 1 mM Mg2+), with the pH adjusted to 7.4 with ∼40 mM KOH. Extracellular Ca2+ and Mg2+ were set at this level to minimize concentration gradients for the divalent ions. Acidic solutions (pH <7.4) contained (in mM): 120 KCl, 10 KMes, 10 EGTA, 1.44 MgCl2, and 6.19 CaCl2, pH adjusted with KOH. The composition of the 5-mM K+ (low K+) solution was (in mM): 110 NaCl, 5 KCl, 10 HEPES(Na), 10 EGTA, 1.44 MgCl2, and 6.19 CaCl2 (achieving a buffered concentration of 60 nM Ca2+ and 1 mM Mg2+), with the pH adjusted to 7.4 with ∼23 mM NaOH. Acidic low K+ solutions contained (in mM): 110 NaCl, 5 KMes, 10 EGTA, 1.44 MgCl2, and 6.19 CaCl2, pH adjusted with NaOH.
Whole cell and macropatch recording electrodes were fabricated from KG-33 glass (Friedrich & Dimmock, Inc.) and filled with a solution of composition (in mM): 120 KCl, 10 HEPES, 10 BAPTA, 1.5 ATP(Na2), 2.34 MgCl2, and 8.98 CaCl2 (added to achieve 1 µM free Ca2+), pH 7.4, which gave resistances of 2–4 MΩ. Zn2+ solutions were prepared from a 500-mM stock (pH of ∼2), and the pH was adjusted to 7.4 with KOH. Zn2+ in solution can form the insoluble ZnOH2, therefore limiting the maximum solubility of Zn2+ at pH 7.4 to 4.5 mM.
Electrodes contained a high K+ solution for excised inside-out single-channel Ca2+ activation and noise analysis recordings of the following composition (in mM): 97 K aspartate or K methylsulfate (MeSO4), 30 KCl, 10 HEPES, 10 EGTA, 1.44 MgCl2, and 6.19 CaCl2 (achieving a buffered concentration of 60 nM Ca2+ and 1 mM Mg2+), with the pH adjusted to 7.4 with KOH. The intracellular side of excised inside-out patches was bathed in the same high K+ solution, with free Ca2+ buffered to 60 nM, 200 nM (adding 1.18 MgCl2 and 8.445 CaCl2), 300 nM (adding 1.12 MgCl2 and 8.9 CaCl2), 600 nM (adding 1.07 MgCl2 and 9.4 CaCl2), or 1 µM (adding 1.04 MgCl2 and 9.64 CaCl2). The effect of extracellular acidic pH on single-channel and noise analysis recordings was studied with the electrode containing the high K+ solution, which was buffered to the stated pH by replacing HEPES with Mes.
The liquid junction potential calculated between high K+ solutions was <2 mV, whereas that between low K+ external solutions and electrode solution was calculated to be ∼8 mV. Neither was corrected for. All standard chemicals were bought from Sigma-Aldrich.
Whole cell and macropatch currents were recorded with an Axopatch 200A (MDS Analytical Technologies). Capacitance and series resistance compensation (≥95%) was only used for whole cell recordings. Currents were low-pass filtered at 1 kHz (eight-pole Bessel; Frequency Devices, Inc.) and acquired at 20 kHz using Pulse (HEKA). All recordings were performed at room temperature (20–25°C). Extracellular solutions were exchanged by a whole bath perfusion system for pH experiments, with a rapid solution exchange system (RSC200; Biologic) used for noise analysis. A long-lasting current run-up was sometimes observed after current was inhibited by acidic extracellular pH (not depicted). Current amplitude was stabilized after returning to pH 7.4 and only accepted for analysis if the inhibition was reversible. Electrodes for single-channel recording were fabricated from quartz capillary (King Precision Glass, Inc.), with resistances of 7–12 MΩ. Single-channel currents were observed by continuous recording at all potentials. Channel currents were filtered at 1 kHz and acquired at 20 kHz using the Pulse program. Single-channel activity was analyzed with TAC (Bruxton) using the 50% threshold technique, and all transitions were visually inspected before being accepted. Single-channel amplitudes were estimated from Gaussian fits of amplitude histograms derived from events >1 ms to account for the rise time of the Bessel filter.
Concentration–inhibition relationships to extracellular protons (H+) were plotted as fractional current remaining (I/Icont) and were fitted with a Hill equation of the form:
where I was the recorded macroscopic current, Icont the current at pH 7.4, A1 the minimum fraction of current remaining, A2 the maximum, and IC50 the concentration in molar of protons that evoked 50% of the maximum inhibition. The IC50 concentration is also quoted as −log IC50 (pIC50) in the text. Unless stated otherwise, fitting was performed on data from individual experiments, and the fit parameters quoted are of the mean ± sem of the individual fits. For display purposes, figures show Eq. 1 fitted to mean data. Fitting was performed using Prism (GraphPad Software, Inc.).
Nonstationary noise analysis
Excised inside-out macropatches were held at −50 mV. Currents were acquired using Pulse (HEKA), and only sweeps without artifacts from the solution exchange system were accepted. Mean current (I) and variance (σ2) were calculated for each ensemble of current traces using Clampfit (MDS Analytical Technologies). The single-channel current (i) and the number of channels (N) in each patch were then obtained by fitting a plot of the variance against the mean current using Eq. 2 (Sigworth, 1980; Hartveit and Veruki, 2007):
Single-channel conductance (γ) and maximum open probability at the maximum mean I (P(o)) were calculated. Curve fitting of equations to the data were performed using Origin software (OriginLab Corp.). Statistical significance was evaluated using Prism (GraphPad Software, Inc.) by either the Student's t test or one- or two-way ANOVA with Bonferroni post hoc test as appropriate.
The validity of applying this analysis to KCa2 channel activity was tested by simulation of three different gating schemes using CSIM (MDS Analytical Technologies; see Fig. 4 for gating schemes). Simulated deactivation current relaxations (128) were generated for each gating scheme using 80 independent channels, except for the scheme in Fig. 4 B, which used 40 channels of high conductance and 40 channels of low conductance. Each simulation used the open channel as the initial state. Transition rates shown in Fig. 4 were obtained from Hirschberg et al. (1998). Data were analyzed in the same manner as experimentally derived data using Clampfit. Amplitude histograms were produced by running the simulation of each gating scheme with one channel, with data analyzed in the same way as experimentally derived data using TAC and TACFit.
Homology model construction
Putative locations of pore residues were predicted based on a model of the KCa2.2 pore region. The pore region of KCa2.2 spanning S5-Phelix-SF-S6 was modeled using the mammalian Kv1.2 pore (PDB: 2a79) (Long et al., 2005) as a template through the comparative modeling web server, Geno 3D (Combet et al., 2002).
Extracellular protons ([H+]o) inhibited KCa2.2 and KCa2.3 macroscopic currents
Ramp currents from outside-out macropatches with an intracellular Ca2+ concentration of 1 µM showed inwardly rectifying KCa2.2 and KCa2.3 currents, which were inhibited differentially by increasing concentrations of extracellular protons ([H+]o) (Fig. 1 A). KCa2.3 currents showed a greater sensitivity to [H+]o than the KCa2.2-mediated current, with the KCa2.3 current being inhibited by pH 6.6 while the KCa2.2 current was unaffected. The effect of [H+]o appeared to be voltage dependent, with inhibition being most apparent at potentials negative to 0 mV (Fig. 1 B). Inhibition at −60 mV by decreasing extracellular pH was plotted for both KCa2.2 and KCa2.3. The maximum inhibition by pH 5.5 was significantly greater for KCa2.3 currents (78 ± 0.1%; n = 4) than that observed for KCa2.2 currents (44 ± 2.0%; n = 3; P < 0.0005). Fitting the data with the Hill equation showed the higher sensitivity of KCa2.3 to [H+]o (IC50 = 158 nM and pIC50 = 6.8 nM; nh = −2.1 ± 0.1 and n = 4–11) compared with inhibition of the KCa2.2 current (IC50 = 63 nM and pIC50 = 6.2 nM; nh = −1.9 ± 0.1 and n = 3; pIC50, P < 0.0001; nh, not significant). The steep Hill slopes for KCa2.2 and KCa2.3 suggest that more than one proton binds to inhibit current and positive cooperativity exists for inhibition.
Although potential effects of extracellular pH on the determined levels of intracellular Ca2+ were controlled for by using intracellular BAPTA, it was possible that the reduction of macroscopic KCa2 current resulted from a change in the Ca2+-dependent gating of channels. Inside-out macropatches with electrodes containing either pH 7.4 or 5.9 were exposed to increasing intracellular concentrations of free Ca2+ to determine the effects on Ca2+-dependent gating. Extracellular acidosis had no significant effect (unpaired t test; n = 3–5 patches) on EC50 or Hill slopes for either KCa2.2 (pH 7.4, EC50 = 281 ± 29 nM and nh = 5.98 ± 1.27; pH 5.9, EC50 = 231 ± 15 nM and nh = 6.21 ± 0.96) or KCa2.3 (pH 7.4, EC50 = 257 ± 21 nM and nh = 5.22 ± 0.55; pH 5.9, EC50 = 284 ± 12 nM and nh = 5.86 ± 0.24) current, indicating that inhibition by protons was independent of changes in the mechanics of Ca2+ gating. These data suggested that inhibition of macroscopic current might have resulted from changes in single-channel properties.
Extracellular protons inhibited KCa2 current by decreasing single-channel conductance
Extracellular protons were proposed to inhibit Kv1.5 current by reducing channel P(o) (Kwan et al., 2006). In contrast, a reduction in single-channel conductance was suggested to underlie the inhibition of inward-rectifying K+ current (Coulter et al., 1995; Geiger et al., 2002). We used a combination of single-channel recording and nonstationary noise analysis to resolve which channel property might be affected to underlie the inhibition of macroscopic KCa2 current. Excised inside-out recording of KCa2.2 and KCa2.3 channels showed in extracellular pH 7.4 that both channels exhibited two conductance levels of 15 and 11 pS for KCa2.2 channels (14.7 ± 1.1 and 10.8 ± 1.4 pS; n = 3) (Fig. 2 A) and 16 and 8 pS for KCa2.3 (15.8 ± 0.6 and 8.3 ± 0.6 pS; n = 5 and 3) (Fig. 2 D), respectively. Both amplitudes were observed in patches that contained single-level activity bathed in 200 nM Ca2+, and rare transitions between open states were resolved (Fig. 2 A). Recording single-channel activity at the extracellular pH that was close to the determined IC50 values showed that conductance was reduced for both channel subtypes by extracellular acidosis, with only a 10-pS (9.6 ± 0.07 pS; n = 3) and a 6.5-pS (6.4 ± 0.8 pS; n = 4) conductance level observed for KCa2.2 channels and KCa2.3 channels, respectively (Fig. 2, B and E). The observed conductances were not significantly different from the lower conductance class observed in pH 7.4, suggesting that extracellular protons promoted the lower conductance state at the expense of the higher conductance state. These data suggested that the promotion of low amplitude openings was a significant contributor to the reduction of macroscopic current.
Excised inside-out patches usually contained multiple channels that were revealed by observing channel activity in the presence of 1 µM Ca2+, which is close to the maximum for KCa2 activation and correlated with a P(o) of ∼0.6–0.8 (Hirschberg et al., 1998; Xia et al., 1998). Lower activity (giving single-level openings) was obtained by bathing patches in 200 nM free Ca2+, and the effect of extracellular pH on open-state kinetics was determined under these conditions to resolve whether the reduction of macroscopic current might be contributed by a decrease in channel open time. KCa2.2 and KCa2.3 channel open-state kinetics were best described by the sum of two exponentials, with τ of ∼1–2 and 11–17 ms at extracellular pH 7.4 (Fig. 3, A and D). Analysis of openings in pH 5.9 and 6.6 gave open-state histograms that were best fit with the sum of two exponentials of similar τ (Fig. 3, B and E). Plotting the mean values for each open-time component against membrane voltage showed that channel open-state kinetics were independent of voltage (Hirschberg et al., 1998), and that extracellular protons had no significant effect on open-state kinetics (one-way ANOVA; P > 0.05) (Fig. 3 F).
Channel P(o) was extremely difficult to reliably obtain with multichannel patches containing low amplitude openings promoted by lowering extracellular pH. We wanted to use nonstationary noise analysis to get estimates of P(o), together with weighted single-channel amplitude (i) and total number of channels active in a patch (N). It was important to determine whether this technique could be applied to a channel that displayed two amplitude levels by using simulated channel data derived from three gating schemes (Fig. 4). Estimates of channel conductance and P(o) were taken from three time points during the deactivation of a simple C-C-O scheme of conductance of 15 pS (Fig. 4 A). Channel conductance was reasonably well estimated with this technique, with most error occurring when P(o) was lowest. In contrast, channel conductance was not well estimated when noise was contributed by two classes of channels, with conductances of 10 and 15 pS (Fig. 4 B). Importantly, differences in estimated P(o) were noted between this scheme and the scheme in Fig. 4 A. Finally, the scheme in Fig. 4 C was used where a channel exhibited two amplitude levels to replicate that observed for both KCa2.2 and KCa2.3 channels. This gating scheme gave estimated P(o) values that were very similar to those seen with the scheme in Fig. 4 A. Analysis of these simulated data showed that nonstationary noise analysis could be applied to KCa2 channel activity to obtain estimates of P(o).
Expressed KCa2.2 or KCa2.3 channels were repeatedly activated by rapid application of 1 µM Ca2+ to excised inside-out macropatches. The variance was calculated from the deactivation relaxation evoked by the rapid removal of 1 µM Ca2+ (Fig. 5, A–C). Plotting the variance (σ) against mean current (I) produced a relationship fit with Eq. 2, yielding estimates of single-channel conductance (γ) (Fig. 5 C), P(o) (Fig. 5 D), and N. Decreasing the extracellular pH from 7.4 to 5.5 reduced γ of KCa2.2 channels from 23.6 ± 5.7 to 10.3 ± 0.6 pS (P < 0.05; n = 3) (Fig. 5 C). Similar decreases in γ by pH 5.5 were observed for KCa2.3 channels, where it decreased from 24.1 ± 4.9 pS at pH 7.4 to 6.04 ± 1.4 pS in pH 5.5 (P < 0.03; n = 3) (Fig. 5 C). These data are consistent with the estimates of single KCa2.2 and KCa2.3 channel conductances in the presence of both pH 7.4 and 5.5 (Fig. 2). Channel P(o) activated by 1 µM Ca2+ for both subtypes was estimated to be 0.7 ± 0.1 (n = 3) for KCa2.2 at pH 7.4 and 0.8 ± 0.09 at pH 5.5, and 0.81 ± 0.05 for KCa2.3 at pH 7.4 and 0.73 ± 0.04 at pH 5.5 (no significant difference) (Fig. 5 D), showing that extracellular protons did not affect channel P(o). These data show that neither changes in open-state kinetics nor channel P(o) underlie the inhibition of macroscopic current by extracellular protons, suggesting that the promotion of low conductance openings was the principal mechanism underlying inhibition.
A histidine residue in the S5-P helix outer pore region was required for current inhibition
The amino acid sequences of KCa2.2 and KCa2.3 channels were aligned, and the predicted extracellular pore regions of the channel were examined for residues that could be ionized by increasing extracellular [H+]o. Histidine (H) residues are weakly basic (pKa of ∼6), acquiring a single-positive charge by protonation across the pH range that has been examined (pH 7.4–5.5). Fig. 6 displays the pore region sequence alignment of the channels with the H residues highlighted. Both channels share an H residue in homologous positions in the outer pore region (H337 in KCa2.2 and H491 in KCa2.3) (Fig. 6, A–C), and it was hypothesized that the protonation of H337 of KCa2.2 and H491 in KCa2.3 underlay the inhibition of macroscopic current by extracellular protons. Mutation of H to the polar but unionizable asparagine residue (N) was done in both KCa2.2(H337N) and KCa2.3(H491N). The application of extracellular pH 5.5 to either mutant had very little effect on macroscopic current (Fig. 6, D and E). Full concentration–inhibition relationships for both mutants showed that they exhibited significantly reduced sensitivity to [H+]o over the pH range of 7.4 to 5.5 (Fig. 6 F). These data indicated that the outer pore H common to both channel subtypes was the molecular target for titration by protons.
Previous data showed that extracellular protons did not inhibit KCa2.2 or KCa2.3 currents by reducing the sensitivity of the channel to activation by intracellular Ca2+. Consistent with this was the finding that the sensitivity of KCa2.2(H337N) to activation by intracellular Ca2+ was not different from wild-type channels (KCa2.2(H337N), EC50 = 227 ± 8 nM; nh = 5.61 ± 0.34). Our data indicate the primary role of the outer pore H residue in mediating the inhibition of macroscopic KCa2 current, which is at least in part mediated by a reduction in single-channel conductance. Consistent with the above data, the mutant channel KCa2.2(H337N) that was insensitive to [H+]o inhibition, displayed only low conductance openings (10.7 ± 0.26 pS; n = 3) (Fig. 2 C), but with unchanged open-state kinetics (Fig. 3 C). This conductance state was very similar to both the lower conductance states of wild-type KCa2.2 channels of 10.8 ± 1.4 pS (Fig. 2 A) (P = 0.94; n = 3), with a small but significant difference between the conductance of the mutant and the wild-type channel extracellular pH 5.9 (9.63 ± 0.07 pS; P < 0.02; n = 3). These data suggested that an uncharged basic H residue in the outer pore of KCa2 channels was required to maintain a high conductance state configuration of the pore. This proposal was supported by finding that mutation of H337 to the positively charged arginine residue (H337R) produced a channel whose conductance was estimated by nonstationary noise to be 11.9 ± 1.9 pS (n = 6; not depicted), a conductance significantly different from wild-type KCa2.2 channels (P < 0.05).
The presence of a second H residue in the pore region of KCa2.3 conferred extra sensitivity to [H+]o
It was postulated that the additional histidine residue (H522) located in the outer pore region of KCa2.3, but not KCa2.2, was responsible for the higher sensitivity of KCa2.3 current to extracellular protons. This was tested by mutating the outer pore regions of both KCa2.2 and KCa2.3 to mimic the sibling's pore. Therefore, N368 in KCa2.2 was mutated to a histidine (N368H) to mimic the KCa2.3 pore (KCa2.2(N368H)), and H522 in KCa2.3 was mutated to asparagine (H522N) to mimic the KCa2.2 pore (KCa2.3(H522N)). Both mutants produced functional current with properties consistent with KCa2 channel current (Fig. 7, A and B). The application of extracellular pH 6.6 to KCa2.2(N368H) showed a greater inhibition than wild-type KCa2.2 (Fig. 7 A). Conversely, KCa2.3(H522N) was less sensitive to extracellular protons than wild-type KCa2.3 current (Fig. 7 B). The generation of concentration–inhibition relationships for both mutants showed that KCa2.2(N368H) (n = 5) displayed a sensitivity to extracellular protons that mirrored (P = 0.91) wild-type KCa2.3 current (n = 11) (pIC50 of ∼6.9), shifting the pIC50 of KCa2.2(N368H) to 6.8 from 6.2 of wild-type KCa2.2 (Fig. 7 C). In contrast, the sensitivity of KCa2.3(H522N) to protons was significantly reduced (P < 0.0001; n = 3) (pIC50 of ∼6.4) when compared with wild-type KCa2.3 current (pIC50 of ∼6.9) (Fig. 7 C). These data strongly suggested that H522 in the outer pore region of KCa2.3 was responsible for the higher sensitivity to extracellular protons displayed by the channel.
The histidine–proton interaction sites do not form a high affinity Zn2+ binding site
Zinc ions (Zn2+) bind to histidine and cysteine residues that are in specific conformations with a high affinity (Vallee and Auld, 1990). This binding has been shown to mimic the effects of extracellular protons interacting with H residues in voltage-gated K+ channels (Kehl et al., 2002). The effect of extracellular Zn2+ on the KCa2.2 channel current was tested to determine whether the outer pore H residue could act as a binding site of the divalent cation. The KCa2.2 current was not significantly inhibited at −60 mV by Zn2+ (4.5 mM), whereas KCa2.3 was significantly inhibited 38 ± 3.4% (P < 0.01) (Fig. 8, A and B). The difference in sensitivity between channel subtypes resulted from H522 in KCa2.3, as KCa2.2(N368H) showed an increased sensitivity (26.9 ± 2.9% block by 4.5 mM Zn2+; P < 0.01) (Fig. 8 C) and KCa2.3(H522N) showed a significant reduction in block compared with the wild-type channel (0.4 ± 2.5% block by 4.5 mM Zn2+; P < 0.001) (Fig. 8 D). These data suggest that Zn2+ could not coordinate at the H337/H498 residue, located in the outer pore between the S5-P helix, but could access and interfere with conduction by binding to the H in the pore at the SF-S6 segment (H522 in KCa2.3) with very low affinity.
Extracellular [H+]o shifted inward rectification of KCa2.3 current to more depolarized potentials
The inhibition of macroscopic current by [H+]o was voltage dependent, with less inhibition observed for outward current evoked positive to 0 mV (Fig. 1, A and B). However, resolving the inhibition of outward current was difficult to achieve because of the small amplitude and clear rectification that was observed under high K+ conditions. The inhibition of both KCa2.2 and KCa2.3 current by extracellular protons was also observed in low K+. KCa2.2 and KCa2.3 current recorded under low K+ conditions showed clear inward rectification at membrane potentials positive to 0 mV. The application of extracellular protons inhibited both current types, but it also shifted the onset of inward rectification to more depolarized levels for the KCa2.3 current (Fig. 9, A and B). The concentration–inhibition relationships for the two channel subtypes showed that the sensitivity of current to protons was reduced in low K+, with a shift of pIC50 of 6.2 in high K+ to 5.6 in low K+ for KCa2.2, and from pIC50 from 6.8 to 6.6 in KCa2.3 (Fig. 9 C). The small amplitude of the KCa2.2 current in outside-out patches made it unsuitable to obtain pIC50 values from individual experiments, with the fit being imposed onto meaned data. This prevented the determination of significance of the shift in sensitivity of the KCa2.2 current to protons between high and low K+. However, the shift in sensitivity was significant for the KCa2.3 current (P < 0.002; n = 6–11).
It was clear that current inhibition was associated with a shift in inward rectification, especially for KCa2.3-mediated current (Fig. 9, A and B). The apparent voltage dependence of the rectification was characterized using a modified Goldman-Hodgkin-Katz (GHK) current equation. KCa2 channel currents do not display voltage dependence (Hirschberg et al., 1998), with inward rectification arising from block by intracellular Mg2+ ions (Soh and Park, 2001). Furthermore, KCa2 currents approach pure GHK rectification in the absence of intracellular Mg2+ (Soh and Park, 2001). Therefore, the current–voltage relationships of both KCa2.2 and KCa2.3 currents should follow the predicted deviation from the GHK equation due to rectification by internal divalent cations, including a term that describes the voltage dependence of block (Woodhull, 1973) by intracellular cations (Pusch, 1990) (Eq. 3):
where MgKd(0 mV) is the apparent equilibrium dissociation constant for Mg2+ at 0 mV. The effect of [Mg2+]i is represented as a first-order blocking mechanism dependent on voltage, whereas the fit parameter MgKd(0 mV) is a measure of the degree of rectification with higher affinity values correlating with a stronger rectification. Fig. 9 D shows that the mean normalized currents for KCa2.2 measured every 20 mV at pH 7.4 and 5.5 were well described by Eq. 3 (the intermediate external pH data points are left out for clarity), revealing that extracellular protons had little effect on δ (0.19 at pH 7.4 compared with 0.16 at 5.5) or MgKd(0 mV) (∼0.01 mM). This suggests that the interaction of protons with KCa2.2 channels has little impact on the Mg2+ binding site and consequent rectification. Fitting the normalized KCa2.3 currents at pH 7.4 and 5.5 with Eq. 3 showed that extracellular protons had little effect on δ, but the MgKd(0 mV) shifted from ∼0.15 mM in pH 7.4 to 1.8 mM in pH 5.5 (Fig. 9 E). These results suggest that proton binding to KCa2.3 channels reduces the affinity of Mg2+ binding to the rectification site without shifting the position of the site within the membrane field.
It has been proposed that rectifying divalent cations bind to serine 359 in KCa2.2 channels, which is located at the intracellular mouth of the selectivity filter (Soh and Park, 2002). It is possible that protons might conduct through KCa2.3 channels and compete with intracellular Mg2+ ions for rectification. Values for MgKd(0 mV) obtained from the fitting of Eq. 3 across the pH range (7.4–5.5) vary with pH and are fitted by the Hill equation with a pIC50 of 6.1 (Fig. 9 F). This pIC50 is similar to the pKa of a histidine residue in solution as a single amino acid, which might suggest that the increase in MgKd(0 mV) is mediated through a titration of a histidine exposed to the extracellular solution by protons, and not by direct competition with Mg2+ for the rectification site. In addition, it suggests that the shift in rectification of the KCa2.3 current observed in low K+ by lowering extracellular pH might result from protons interacting with the H522 that is unique to KCa2.3 (Fig. 6 A).
The presented data demonstrated that the protonation of an outer pore H residue common to both KCa2 channel subtypes mediated inhibition of macroscopic current. Extracellular acidification or mutation of the H residue did not affect channel activation by Ca2+, indicating that protonation or mutation did not affect the structural requirements for gating. Inhibition of current was principally mediated by a decrease in single-channel conductance, associated with a depolarizing shift in current rectification in KCa2.3, and to a lesser extent in KCa2.2. This shift in rectification was produced by a reduction in the affinity of intracellular divalent ions for their proposed intracellular binding site (Soh and Park, 2002). These data indicated that the protonated outer pore H522 in KCa2.3 could be in the local electrostatic vicinity of the intracellular binding site (S359) for intracellular Mg2+ ions. Sensitivity to extracellular protons required the outer pore H residue common to both KCa2 channel subtypes, even though KCa2.3 possesses the additional H residue (H522). This was supported by the finding that the double-mutant KCa2.2(H337N, N368H) was also poorly sensitive to protons. This suggests that the first H residue acts to provide a specific molecular conformation that allows the second H to interact with protons to produce a shift in the rectification of KCa2.3. Finally, it was observed that KCa2.3(H522N) (a mutation to mimic the outer pore of KCa2.2) was more sensitive to extracellular protons than wild-type KCa2.2. This might suggest subtle differences in pore structure between the two wild-type channels.
It has been proposed that the pore structure of KCa2 channels shares close structural homology to Kv channels (Jäger and Grissmer, 2004). Superimposition of the pore sequence onto the crystal structure of Kv1.2 (Doyle et al., 1998; Long et al., 2005) suggested that H337 in KCa2.2 (within the S5-P helix loop) is a substantial distance away from the ion conduction pathway (Fig. 6 C). This would make it highly unlikely that the channel is subject to a direct occlusion/pore block by protonation of H337 (H491 in KCa2.3), and also unlikely to significantly electrostatically influence the conduction pathway. In addition, we observed that lowering external K+ significantly reduced the sensitivity to proton inhibition, which is contrary to the competitive interaction with K+ that would be expected if the H residue formed any part of the conduction pathway (Kehl et al., 2002; Fedida et al., 2005). This lack of competition also suggested that outer pore collapse (C-type inactivation), which is seen in Kv channels as the mechanism underlying inhibition, was not the mechanism for inhibition of KCa2 current, as this would also be antagonized by high K+ (Baukrowitz and Yellen, 1995). It seems likely that inhibition of macroscopic current in KCa2.2 results from an indirect structural perturbation of the channel outer pore through H337 protonation. The protonation of H337 could change the local salt bridges and hydrogen bonds (Thompson et al., 2008) in a manner that couples to the structure of the conduction pathway. This allosteric effect could cause a change in pore shape resulting in lowered single-channel conductance, as the wild-type channel with H positively charged only displayed a conductance of ∼11 pS, without affecting open-state kinetics. This hypothesis is supported by the reduced single-channel conductance that we observed for the KCa2.2(H337N) mutant channel. This is consistent with the proposal that a nonprotonated H residue in an outer pore position common to all KCa2 channels is necessary for maintaining a pore shape that permits high conductance openings, a pore shape not supported by a neutral or a positively charged amino acid. In KCa2.3, the extra sensitivity to protons is likely affected by the combination of structural perturbation through the protonation of S5-P helix H491 and the additional protonation of H522 causing further potential structural and electrostatic impairments to K+ flux. Molecular modeling studies examining electrostatic interactions have supported experimental evidence that protonation of H residues in the outer pore perturb the electrostatic interactions of K+ ions as they travel through the pore (Liu et al., 2007). The closer proximity of H522 than H491 to the KCa2.3 pore mouth might suggest that protonation of H522 could also contribute to inhibition through an electrostatic repulsion of K+ in the outer pore region and conduction pathway.
The inhibition by protons displayed an apparent voltage dependence with less block at depolarized membrane potentials, an effect most pronounced for KCa2.3 current in low K+. The fraction of inhibition at different voltages as a function of extracellular pH could not be fully described by the Woodhull model, which depicts a simple block of the permeation pathway by a charged ion (Woodhull, 1973). This might not be surprising, considering the arguments against protons acting as pore blockers discussed above. The lack of primary voltage dependence is also consistent with the prediction that both H residues would occupy positions in the outer pore region in areas not considered to be directly within the membrane field. We propose that protonated H522 interacted with the intracellular rectifying Mg2+ that is bound within the membrane field (Soh and Park, 2001). The affinity of the block by intracellular divalent ions to cause rectification varies with transmembrane voltage, [Ca2+]i, and [K+]o, suggesting the blocking ion is located in the conduction pathway (Soh and Park, 2001). To account for this, the currents were fit with a GHK equation modified to account for the voltage dependence of block by intracellular ions (Eq. 3) (Pusch, 1990). Our results suggested that the apparent voltage dependence of block by protons in KCa2.3 is a result of reducing affinity of intracellular divalent ions for the rectification binding site, with the estimated MgKd(0 mV) from the fit shifting from 0.1 mM at pH 7.4 to 1.8 mM at pH 5.5. The electrical distance δ stayed about the same at ∼0.2, suggesting that the location of the binding site in the field was not significantly shifted. The protonation-dependent change in pore shape, in addition to an electrostatic repulsive effect originating from the protonated H522, might perturb the configuration of the binding site for the divalent ion and lower affinity. Finally, the shift in rectification caused by extracellular protons was more pronounced in low K+ than high K+. It has been reported that increasing [K+]o lowers the apparent affinity of the Mg2+ for the intracellular binding site through a mechanism commonly referred to as “trans-Knockoff” (Soh and Park, 2001). This proposal is consistent with the shift of rectification being more pronounced when the affinity for rectifying ions is higher (e.g., in low K+). It is likely that the pore shape differs between KCa2.2 and KCa2.3 channels, which partly underlies the different sensitivity to extracellular protons. It is noteworthy that acidic pH is the only condition described so far that clearly inhibits KCa2.3 more than KCa2.2 channel currents. This knowledge may give hints for the development of future preferential KCa2.3 channel blockers.
Transient changes in extracellular pH occur during periods of physiological neuronal activity, such as synaptic activation (Chesler, 1990) and pathological events, such as epileptic seizures and cerebral ischemia (Chesler and Kaila, 1992; Tombaugh and Sapolsky, 1993), falling to below pH 6.5 during ischemia (von Hanwehr et al., 1986). Inhibition by acidosis would have effects dictated by channel subunit location. KCa2.2 and KCa2.3 channel subunits display distinct but overlapping expression patterns in the brain (Sailer et al., 2002, 2004). For example, KCa2.2 and KCa2.3 subunits are present in the hippocampus, with KCa2.2 present in dendrites and dendritic spines and KCa2.3 subunits localized in mossy fiber terminals (Sailer et al., 2002, 2004; Lin et al., 2008). Inhibition of spine-located KCa2.2 channel current during synaptic activation would aid the induction of long-term potentiation (Lin et al., 2008), whereas inhibition of channel current located in dendrites would increase excitability (Kelly and Church, 2004). In contrast, the inhibition of mossy fiber terminal KCa2.3 channel current would affect transmitter release and the induction of long-term potentiation at that synapse. It is clear that the differences in proton modulation of these channels could reflect a functionally relevant separation that might contribute to the physiological roles of the KCa2 channel subtypes in different brain regions.
We wish to thank Mr. Guillaume Drion, Prof. Andy Randall, and Dr. Jon Brown for helpful discussions on data analysis.
This work was funded by a grant from the Belgian Science Policy (IAP 6/31 to V. Seutin and N.V. Marrion). V. Seutin was also supported by grant 9.4560.03 from The National Fund for Scientific Research (Belgium). C. Lamy was supported by a FIRST-DEI grant (516131) from the Walloon region of Belgium.
Christopher Miller served as editor.
V. Seutin and N.V. Marrion contributed equally to this paper.