Ligand-Gating by Ca²⁺ Is Rate Limiting for Physiological Operation of BK_Ca Channels

Abstract

Large conductance Ca²⁺- and voltage-activated potassium channels (BK_Ca) shape neuronal excitability and signal transduction. This reflects the integrated influences of transmembrane voltage and intracellular calcium concentration ([Ca²⁺]_i) that gate the channels. This dual gating has been mainly studied as voltage-triggered gating modulated by defined steady-state [Ca²⁺]_i, a paradigm that does not approximate native conditions. Here we use submillisecond changes of [Ca²⁺]_i to investigate the time course of the Ca²⁺-triggered gating of BK_Ca channels expressed in Chinese hamster ovary cells at distinct membrane potentials in the physiological range. The results show that Ca²⁺ can effectively gate BK_Ca channels and that Ca²⁺ gating is largely different from voltage-driven gating. Most prominently, Ca²⁺ gating displays a pronounced delay in the millisecond range between Ca²⁺ application and channel opening (pre-onset delay) and exhibits slower kinetics across the entire [Ca²⁺]_i-voltage plane. Both characteristics are selectively altered by co-assembled BKβ4 or an epilepsy-causing mutation that either slows deactivation or speeds activation and reduces the pre-onset delay, respectively. Similarly, co-assembly of the BK_Ca channels with voltage-activated Ca²⁺ (Cav) channels, mirroring the native configuration, decreased the pre-onset delay to submillisecond values. In BK_Ca–Cav complexes, the time course of the hyperpolarizing K⁺-current response is dictated by the Ca²⁺ gating of the BK_Ca channels. Consistent with Cav-mediated Ca²⁺ influx, gating was fastest at hyperpolarized potentials, but decreased with depolarization of the membrane potential. Our results demonstrate that under experimental paradigms meant to approximate the physiological conditions BK_Ca channels primarily operate as ligand-activated channels gated by intracellular Ca²⁺ and that Ca²⁺ gating is tuned for fast responses in neuronal BK_Ca–Cav complexes.

Introduction

Large-conductance calcium- and voltage-activated K⁺ channels (BK_Ca) are key elements of signal transduction in the CNS that sharpen action potentials (Storm, 1987b; Shao et al., 1999; Edgerton and Reinhart, 2003), generate the fast phase of afterhyperpolarization (Lancaster and Nicoll, 1987; Storm, 1987a; Sah and Faber, 2002; Womack and Khodakhah, 2002), shape Ca²⁺ spikes in dendrites (Golding et al., 1999; Rancz and Hausser, 2006), and influence synaptic transmission (Robitaille et al., 1993; Pattillo et al., 2001; Raffaelli et al., 2004; Grimes et al., 2009).

All these actions critically depend on the properly timed activation of BK_Ca channels that is driven by two distinct stimuli, membrane depolarization and elevation in intracellular Ca²⁺ concentration ([Ca²⁺]_i) (Marty, 1981; Cui et al., 1997; Horrigan and Aldrich, 2002). Both stimuli are captured by distinct structural domains in the channel protein (Xia et al., 2002; Ma et al., 2006; Cui et al., 2009; Wu et al., 2010; Yuan et al., 2010) and trigger distinct conformational changes that converge onto the pore to open the channel (Yuan et al., 2012).

Classically, the allosteric coupling of the two gating stimuli has been documented in experiments where voltage steps were applied to BK_Ca channels in the presence of defined but constant levels of [Ca²⁺]_i (Marty, 1981; Cox et al., 1997; Cui et al., 1997; Lippiat et al., 2003). In this setting, BK_Ca channels are primarily gated by voltage and [Ca²⁺]_i acts as a modulator facilitating channel activation. Notwithstanding, a few experiments demonstrated that [Ca²⁺]_i may also act as the gating stimulus at constant transmembrane voltage. Thus, the steady-state open probability of BK_Ca channels largely increased when [Ca²⁺]_i was raised from 0 to 100 μm at membrane potentials keeping the voltage-sensor(s) at rest (at −120 mV; Horrigan and Aldrich, 2002) or when [Ca²⁺]_i jumps were applied at constant voltages (Markwardt and Isenberg, 1992).

As both, voltage and [Ca²⁺]_i, are able to independently activate BK_Ca channels the question remains, which stimulus is in command under physiological conditions where both can change in response to cellular activity? In fact, changes in [Ca²⁺]_i usually occur almost instantaneously with the change in voltage, at least with BK_Ca channels that are located in nano-domains around Cav channels (Neher, 1998; Fakler and Adelman, 2008) or that form stable channel–channel complexes with these channels (Berkefeld et al., 2006; Loane et al., 2007; Berkefeld and Fakler, 2008; Grimes et al., 2009). Accordingly, analysis of such channel–channel complexes showed that their K⁺-current output is closely controlled by the gating characteristics of the Cav subunit(s): BK_Ca channels only opened after Ca²⁺ ions were provided through activated Cav channels and BK_Ca channels assembled with Cav2.1 (P/Q-type) activated faster and at more negative potentials than those in complex with Cav 1.2 (L-type; Berkefeld and Fakler, 2008). While all these observations demonstrate that voltage-driven elevations in [Ca²⁺]_i are required for BK_Ca activity under physiological conditions, they cannot answer whether channel gating is predominantly driven by changes in voltage or [Ca²⁺]_i.

The present work pursued this question by detailed analyses of the Ca²⁺-triggered gating in sole BK_Ca channels and in BK_Ca channels co-assembled with Cav2 channels, as they are in many neurons. For this purpose we used a combination of patch-clamp recordings in excised-patch and whole-cell configuration together with rapid piezo-driven changes in [Ca²⁺]_i.

Materials and Methods

Molecular biology.

Ion channels were heterologously reconstituted either in unmodified Chinese hamster ovary (CHO) cells or in CHO cells stably expressing Cav2.2 channels (Cav2.2α1, Cavβ1, α2δ1, γ4; a gift from Novartis). Transfection of cDNAs was done with the JetPEI transfection reagent (Biomol), cells were incubated at 37°C and 5% CO₂ and measured 2–3 d after transfection. GenBank accessions of the clones used are ADO63674 (BKα; a gift from Dr. L. Salkoff, Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, MO), NM_021452.1 (BKβ4), AF055477 (Cav2.2α1), NM_017346 (Cavβ1b), AF286488.1 (for transient transfection), and NM_012919 (in stably transfected cells) (α2δ). EGFP was simultaneously added as a transfection control. All cDNAs were verified by sequencing.

Electrophysiology.

Standard whole-cell patch-clamp recordings on CHO cells were done at room temperature (22–24°C) as previously described (Bildl et al., 2004). Briefly, currents were recorded with an Axopatch 200B amplifier, filtered at 3–10 kHz and sampled at 25 kHz; series resistance and capacitance (95%) were compensated.

Recording pipettes were made from quartz glass capillaries and had resistances of 1–5 MΩ when filled with intracellular solution containing the following (in mm): 137.5 KCl, 3.5 MgCl₂, 2.5 NaATP, 10 K₂EGTA, and 5 HEPES (pH adjusted to 7.3 with HCl). The extracellular solution was composed as follows (in mm): 144 NaCl, 5.8 KCl, 0.9 MgCl₂, 1.3 CaCl₂, 0.7 NaH₂PO₄, 5.6 d-glucose, and 10 HEPES (pH adjusted to 7.4 with NaOH).

Rapid solution exchange at excised inside-out patches was performed using a piezo-driven fast application system with a double-barrel application pipette (Schwenk et al., 2009). Speed and exact onset of solution exchange were monitored for each patch at the beginning of the respective experiment with a switch between intracellular solutions containing 150 and 60 mm K⁺ (K⁺ equally exchanged for Na⁺). Patches with exchange times < 1 ms (time constants of exponential functions fitted to the current change ≤ 0.3 ms) were used for evaluation. Solutions applied to the intracellular face of the patches contained the following (in mm): 137 KCl, 1.25 MgCl₂, 2 DiBrBAPTA, 5 HEPES, 5 NaCl (pH adjusted to 7.3 with KOH). Free [Ca²⁺] of 1, 3, 10, and 100 μm were obtained by adding 0.706, 1.232, 1.7, and 2.06 mm CaCl₂ (calculated with WEBMAXC 2.22). Free [Ca²⁺] was verified with a Ca²⁺-sensitive electrode (Berkefeld et al., 2006). Recording pipette solutions for inside-out patches was identical with the extracellular solution used for whole-cell measurements (see above).

Steady-state activation curves (see Fig. 3A) were determined as follows: for Ca²⁺ gating, [Ca²⁺]_i steps were applied 20 ms after stepping the membrane potential to constant values between −80 and 110 mV (10 mV increment); for voltage gating voltage steps as above were applied at constant [Ca²⁺]_i of values between 0 and 100 μm.

Figure 3.

Comparison of voltage- and Ca²⁺-triggered gating: similar steady-states, but distinct time courses. A, Relative steady-state activation of BK_Ca channels determined in inside-out patches and plotted along [Ca²⁺]_i and transmembrane voltage. Intense colors refer to steady-state activation as obtained with Ca²⁺-triggered gating, faint colors are steady-state activation resulting from voltage-triggered gating. Data points are mean ± SEM 6–11 (Ca²⁺ gating) or 5–7 (voltage gating) experiments. Lines are results of fits with a Boltzmann function to the activation curves in both Ca²⁺ and voltage dimension. B, Time constants obtained for voltage (faint colors) and Ca²⁺ gating (intense colors) in the experiments in A and plotted against voltage and [Ca²⁺]_i.

Data analysis.

Data analyses were done with Igor Pro 4.05A on a MacBook Pro. Time constants of channel activation (see Figs. 2B, 3, 4) were derived from monoexponential fits to the rising phase of the recorded K⁺ currents. For the fast application experiments, the delay of the current onset was defined as time between start of effective solution exchange and crossing of the signal threshold (set to 2 SDs from background noise; Figs. 2C, 4).

Figure 2.

Characterization of Ca²⁺gating: the time domain. A, Currents through BK_Ca channels elicited by [Ca²⁺]_i steps from 0 to 1, 3, 10, or 100 μm at membrane potentials of −50 mV, 0 or 50 mV (upper, middle, and bottom). Increase in amplitude upon withdrawal of the 100 μm Ca²⁺ solution reflects release of the pore block by Ca²⁺. B, Time constants of Ca²⁺ gating obtained from experiments as in A. Rising and declining phases of the currents were fitted with a mono-exponential function; closed symbols (τ_ON) represent fits to the rising phase ([Ca²⁺]_i stepped from 0 μm to 1, 3, 10, or 100 μm Ca²⁺), open symbols (τ_OFF) are results of fits to the decline of the current ([Ca²⁺]_i stepped from 100 μm to 0, 1, 3, or 10 μm Ca²⁺). Data points are mean ± SEM for 3–10 (τ_ON) or 3–16 experiments (τ_OFF). C, Left, Boxed periods of the traces in A at enlarged timescale. Arrows indicate start of effective solution exchange. Right, Pre-onset delay obtained from experiments as on the left and plotted versus the respective [Ca²⁺]_i. Data points are mean ± SEM 7–10 experiments.

Figure 4.

Effects of co-assembled BKβ4 and the BKα(D369G) mutation on Ca²⁺ gating of BK_Ca channels. A, B, Time constants of Ca²⁺ gating measured at the indicated potentials in inside-out patches from CHO cells expressing wild-type BKα either alone or together with BKβ4 (A) or expressing the BKα(D369G) mutant (B). Symbols are as in Figure 2; data points for τ_ON are mean ± SEM 3–16 (BKα), 5–11 (BKα + BKβ4), and 6–9 (BKα(D369G)) experiments; data points for τ_OFF are mean ± SEM 3–16 (BKα), 6–12 (BKα + BKβ4), and 7–8 (BKα(D369G)). Note the distinct effects of BKβ4 and the D to G mutation on the gating characteristics. C, Pre-onset delay plotted versus [Ca²⁺]_i for BKα and BKα + BKβ4 (left) and BKα and BKα(D369G) (right). Data points are mean ± SEM 7–10 (BKα), 5–9 (BKα + BKβ4), or 3–9 (BKα(D369G)) experiments. Insets, Representative current traces recorded in response to application of Ca²⁺ (horizontal bar) and normalized to maximum. Note the marked reduction in pre-onset delay observed with the BKα(D369G) mutant channels.

For determination of activation curves (Fig. 3A) steady-state currents obtained through voltage gating as well as Ca²⁺ gating were analyzed. Along the voltage axis, V_1/2, slope factor k and conductance C were calculated with f(V) = (V − V_rev) * C/(1 + exp((V − V_1/2)/k)) with a solution-dependent V_rev of −82 mV. First-order Boltzmann curves were deduced from these parameters with I = I_leak + I_max/(1 + exp((V − V_1/2)/k)) where I_leak is voltage-independent leak current (set to 0), I_max is the amplitude of the fully activated current (set to 1), V is prepulse voltage, V_1/2 is voltage for half-maximal activation, and k is the slope factor. Along the [Ca²⁺]-axis, V and V_1/2 in the Boltzmann function were replaced by [Ca²⁺]_i and [Ca²⁺]_i at half-maximal activation [Ca²⁺]_i,1/2. Data are given as mean ± SEM throughout the manuscript.

The Kolmogorov–Smirnov test was used for testing the pre-onset delays and gating time constants recorded at distinct values of [Ca²⁺]_i and membrane potentials to be different; p values ≤ 0.05 are indicative for different behavior (at a confidence of >99%).

Results

Ca²⁺-triggered gating of BK_Ca channels

Classically, the gating properties of BK_Ca channels have been examined by applying voltage steps in the presence of constant [Ca²⁺]_i, a regime where channels are gated by voltage pulses (“voltage gating”) and [Ca²⁺]_i is considered a modulator (of gating time constants and steady-state open probabilities). Figure 1 shows application of such a paradigm to excised inside-out patches from CHO cells expressing homomeric BK_Ca channels. Thus, BK_Ca channels responded instantaneously to a voltage step from −120 to 0 mV with an increase in current whose time course strongly depended on the [Ca²⁺]_i set to defined values between 1 and 100 μm (Fig. 1B, left, middle, and bottom).

Figure 1.

Gating of BK_Ca channels by voltage and Ca²⁺ in separation. A, Left, Scheme of the piezo-driven fast application system; each of the two barrels could be perfused with several different solutions. Right, Solution exchange (150 mm K⁺ vs 60 mm K⁺, at 10 μm Ca²⁺) with this system determined either at an open pipette or an excised inside-out patch containing BK_Ca channels. Exponential fits (black lines) to rise and decline of the currents through inside-out patches yielded time constants of 0.30 and 0.28 ms, respectively. Current scales are 0.1 nA. B, Voltage and Ca²⁺ gating determined in an inside-out patch excised from a CHO cell expressing BKα. Left, Current response (middle, bottom, and boxed period at expanded timescale) to a voltage step from −120 to 0 mV recorded at the indicated (constant) [Ca²⁺]_i. Note the instantaneous rise in current at the voltage step. Right, Current response to [Ca²⁺]_i steps (indicated by the arrow) from 0 to 1, 3, 10, or 100 μm recorded at a (constant) membrane potential of 0 mV; gray bar denotes three times the value of the exchange time constant (∼0.9 ms). Note the initial delay before the current onset. The reduction in current amplitude at 100 μm Ca²⁺ reflects pore block by the divalent (Ledoux et al., 2008).

Next, we performed experiments that used concentration steps in [Ca²⁺]_i to trigger channel gating, while the membrane potential was held at constant values (“Ca²⁺ gating”). Experimentally, [Ca²⁺]_i steps were achieved by a piezo-driven rapid application system that allowed for a complete solution exchange at the cytoplasmic face of inside-out patches within <1 ms (Fig. 1A). This time course was determined from the exchange of high versus low K⁺ concentrations at open BK_Ca channels (Fig. 1A, inset; time constants of ≤0.3 ms for switches in either direction) and, due to its critical importance, was repeated at every excised patch before the actual experiment. Similarly, the exchange of Ca²⁺ was not rate limiting (see below; Figs. 4C, 7). Figure 1B (right) illustrates such a rapid application experiment performed on the same patch as before and held at a membrane potential of 0 mV: [Ca²⁺]_i steps to distinct concentrations between 1 and 100 μm elicited BK_Ca currents with amplitudes similar to those recorded upon voltage steps, but largely different kinetic properties. Thus, Ca²⁺-triggered BK_Ca currents displayed a prominent delay before their onset and rose with a significantly slower time course (Fig. 1B, right). The delay in current onset did not result from the solution exchange as it depended on [Ca²⁺]_i, and exceeded the period required for complete exchange at a saturating [Ca²⁺]_i of 100 μm (Fig. 1B, gray bar, right, bottom). Moreover, it should be noted that [Ca²⁺]_i of ≥1 mm largely abolished BK_Ca-mediated currents due to the pore block of the divalent (Ledoux et al., 2008; Fig. 2).

Figure 7.

Acceleration of the BK_Ca channel Ca²⁺ gating upon complex formation with Cav channels. Time course of BK_Ca activation recorded either with Cav-free BK_Ca wild-type and D369G mutant channels upon fast application of 10 and 100 μm Ca²⁺ (at 50 mV) or in Cav2.2-associated BK_Ca channels at 40 mV. Data for complexes are from Figure 6, data for free BK_Ca channels (black traces) are mean ± SEM 8 (10 μm Ca²⁺) and 4 (100 μm Ca²⁺) fast application experiments with excised patches. Note the largely decreased pre-onset delay observed for the BKα(D369G) channels and the Cav2.2-associated BK_Ca channels.

The Ca²⁺ gating of BK_Ca channels was further characterized by determining both the pre-onset delay and the gating time constants over broad ranges of physiologically relevant [Ca²⁺]_i and membrane potentials (Figs. 2, 3). The time constants for channel activation (opening, τ_ON) and deactivation (closing, τ_OFF) were obtained from mono-exponential fits of the rising and falling phases of the currents recorded upon application ([Ca²⁺]_i stepped from 0 to 3, 10, and 100 μm) and withdrawal of Ca²⁺ ([Ca²⁺]_i stepped from 100 μm to 3, 1, and 0 μm), respectively; the pre-onset delay was determined as the period between the start of effective solution exchange (Figs. 1, 2, arrowheads) and the time point when currents exceeded the background noise. As shown in Figure 2, pre-onset delay and channel gating exhibited marked dependence on [Ca²⁺]_i, with increasing [Ca²⁺]_i promoting faster activation (Fig. 2A,B) and slower deactivation kinetics, together with a shorter pre-onset delay (Fig. 2A–C, left). The gating time constants were distributed in a bell-shaped manner over the [Ca²⁺]_i range investigated (Fig. 2B). They were lowest at either [Ca²⁺]_i ≥10 μm or at [Ca²⁺]_i of 0 and 1 μm where τ_ON and τ_OFF reflected the mere activation or deactivation kinetics, respectively (Fig. 2B). At 3 μm [Ca²⁺]_i, where current amplitudes were about half-maximal and channel activation and deactivation occurred at balanced levels, the time constants displayed maximal values (Fig. 2B).

These observations were essentially independent of the membrane potential, although Ca²⁺ gating was modulated by voltage (Fig. 2). Thus, depolarization of the membrane potential (from −50 to 50 mV) accelerated Ca²⁺-triggered channel opening ∼2-fold, while channel closing was slowed down >3-fold (Fig. 2B). In addition, the pre-onset delay obtained at any [Ca²⁺]_i was significantly decreased by membrane depolarization with values down to ∼1.5 ms at a [Ca²⁺]_i of 100 μm and potentials ≥0 mV (Fig. 2C). Noteworthy, increasing [Ca²⁺]_i from 10 to 100 μm significantly decreased the pre-onset delay at all voltages, while τ_ON remained essentially constant at these [Ca²⁺]_i (Fig. 2B,C).

Comparison of Ca²⁺- and voltage-triggered gating

These results demonstrated that BK_Ca channels can be effectively gated by both [Ca²⁺]_i and voltage, consistent with previous studies (Markwardt and Isenberg, 1992; Rothberg and Magleby, 2000; Horrigan and Aldrich, 2002). To directly compare the two gating modes in the same BK_Ca channels, we determined activation curves for Ca²⁺ gating ([Ca²⁺]_i steps at constant potentials along the voltage-axis from −80 to 110 mV) and voltage gating (voltage steps at constant [Ca²⁺]_i between 0 and 100 μm) in 5–11 patches and plotted steady-state current amplitudes along the [Ca²⁺]_i and voltage axes. The resulting 3D-plot showed that activation of BK_Ca channels under steady-state conditions is independent of whether Ca²⁺ or voltage was used as a gating trigger or, equivalently, that steady-state activation does not discriminate between the two stimuli (Fig. 3A). Different from the steady-state condition, however, the time courses were quite distinct between voltage- and Ca²⁺-triggered gating (Fig. 3B, Table 1). Thus, only Ca²⁺ gating exhibited a delay between stimulus and current onset (Fig. 1B), and the time constants determined for Ca²⁺-triggered gating exceeded those for voltage gating to about the same extent across the entire [Ca²⁺]_i-voltage plane (Fig. 3B, Table 1).

Table 1.

Time constants of Ca²⁺ and voltage gating obtained for BK_Ca channels in excised patches

Voltage gating		Ca2+ gating
[Ca2+]i	Voltage step	Vm	[Ca2+]i step
	−80 to −50 mV (τON)	−50 mV	0 to [Ca2+]i (τON)
100	2.20 ± 0.23 (3)	100	5.85 ± 1.08 (7)
10	3.10 ± 0.78 (6)	10	6.49 ± 0.50 (10)
3	4.18 ± 0.57 (8)	3	5.28 ± 0.72 (3)
	100 to −50 mV (τOFF)		100 μm to [Ca2+]i (τOFF)
1	0.76 ± 0.08 (4)	1	2.96 ± 0.28 (8)
0	0.47 ± 0.02 (3)	0	2.23 ± 0.57 (8)
	−80 to 0 mV (τON)	0 mV	0 to [Ca2+]i (τON)
100	1.71 ± 0.11 (4)	100	4.44 ± 0.88 (7)
10	1.72 ± 0.24 (9)	10	5.01 ± 1.90 (9)
3	6.13 ± 0.47 (8)	3	15.00 ± 1.09 (8)
	100 to 0 mV (τOFF)		100 μm to [Ca2+]i (τOFF)
1	3.77 ± 0.46 (10)	1	7.39 ± 1.07 (8)
0	0.86 ± 0.17 (8)	0	4.19 ± 1.06 (8)
	−80 to 50 mV (τON)	50 mV	0 to [Ca2+]i (τON)
100	0.68 ± 0.09 (3)	100	3.24 ± 1.03 (7)
10	1.09 ± 0.30 (7)	10	3.95 ± 0.68 (10)
3	5.56 ± 0.60 (8)	3	11.34 ± 3.05 (8)
1	14.57 ± 1.76 (9)	1	18.17 ± 2.01 (3)
	100 to 50 mV (τOFF)		100 μm to [Ca2+]i (τOFF)
0	3.28 ± 1.01 (4)	0	10.02 ± 3.29 (8)

All values are mean ± SEM of (n) experiments.

Together these results indicated that while both Ca²⁺ and voltage converge on the opening mechanism of BK_Ca channels, their actual coupling is quite distinct. Ca²⁺-triggered gating likely involves additional steps including Ca²⁺ binding, which engender the pronounced pre-onset delay.

Selective effects of BKβ4 and an epilepsy-associated mutation in BKα on Ca²⁺ gating

The time course of Ca²⁺ gating was further characterized by analyzing alterations induced by co-assembly of BKα with BKβ4, the most abundant auxiliary subunit of brain BK_Ca channels (Brenner et al., 2000; Berkefeld et al., 2006), or by a point mutation in the cytoplasmic region linking S6 to the “regulator of K⁺ conductance” (RCK) domain of the BKα protein (D369G; Du et al., 2005). Both have been reported to influence the gating of BK_Ca channels by shifting steady-state activation, decelerating (BKβ4), or accelerating (D369G) channel opening and by increasing the apparent sensitivity to Ca²⁺ (D369G) (Ha et al., 2004; Wang et al., 2006; Yang et al., 2010). Figure 4 summarizes the results obtained from analysis of currents recorded upon [Ca²⁺]_i steps at three distinct voltages for channels assembled either from BKα and BKβ4 (Fig. 4A,C) or from BKα(D369G) proteins (Fig. 4B,C). The two alterations exerted quite distinct effects on the dynamics of the Ca²⁺ gating: while BKβ4 increased the τ_OFF values (measured at 0 and 1 μm [Ca²⁺]_i) by up to an order of magnitude without significant effects on τ_ON (measured at [Ca²⁺]_i ≥10 μm; p >0.05, Kolmogorov–Smirnov test; Fig. 4A), the D to G exchange in the linker region resulted in marked decrease of τ_ON, but left τ_OFF virtually unaffected (Fig. 4B). Thus, co-assembly of BKα with BKβ4 predominantly influenced channel deactivation, whereas the (D369G) mutation selectively targeted Ca²⁺-triggered activation of the channels. In line with this interpretation, the pre-onset delay measured in BKα(D369G) channels was significantly decreased at all activation-triggering [Ca²⁺]_i and at all three membrane potentials (Fig. 4C, right) with values down to <1 ms measured at 100 μm [Ca²⁺]_i (0.77 ± 0.21 ms (−50 mV), 0.53 ± 0.19 ms (0 mV), 0.42 ± 0.18 ms (50 mV), n = 9; Fig. 4C). BKβ4 did not induce any obvious changes at [Ca²⁺]_i ≥10 μm, while a slight increase in the pre-onset delay was observed for [Ca²⁺]_i of 3 μm (p > 0.01, Kolmogorov–Smirnov test; Fig. 4C, left).

Together, these results indicated that BKβ4 and the D369G mutation selectively promote distinct gating processes underlying deactivation and activation of Ca²⁺-triggered gating, respectively. In fact, the observed decrease in pre-onset delay together with the accelerated channel activation are consistent with the enhanced allosteric coupling proposed for the D369G mutation by Cui et al. (Yang et al., 2010).

Ca²⁺ gates BK_Ca channels in BK_Ca–Cav complexes

Next, we investigated the Ca²⁺ gating in BK_Ca channels that are coexpressed with Cav channels. Under these circumstances, the two channel types are co-assembled into macromolecular complexes similar to the association of BK_Ca and Cav channels in the mammalian brain (Robitaille et al., 1993; Prakriya and Lingle, 2000; Berkefeld et al., 2006; Loane et al., 2007; Berkefeld and Fakler, 2008). This configuration places the BK_Ca channels within ∼10 nm from their Ca²⁺ source, effectively delivering [Ca²⁺]_i within microseconds after voltage-dependent pore opening (Neher, 1998; Fakler and Adelman, 2008).

We performed whole-cell patch-clamp recordings under physiological ion conditions on CHO cells that coexpressed BK_Ca and Cav2.2 channels (N-type channels) and were dialyzed with a high concentration (10 mm) of the Ca²⁺ buffer EGTA to eliminate currents of Cav-free BK_Ca channels (Fig. 5A; Berkefeld et al., 2006; Berkefeld and Fakler, 2008). The current-output evoked in BK_Ca–Cav complexes by voltage commands reflects the combination of K⁺ and Ca²⁺ currents (Berkefeld et al., 2006). Therefore, a modified tail-current protocol was used to precisely trigger Ca²⁺ gating and to selectively focus on the resulting K⁺-current component (Berkefeld and Fakler, 2008). This tail-current protocol consisted of three voltage steps (Fig. 5B,C, top): (1) a depolarizing prepulse to 70 mV for 10 ms fully activating Cav channels without eliciting Ca²⁺ currents due to the largely reduced driving force for Ca²⁺ ions, (2) a conditioning pulse to potentials between −40 and 40 mV promoting Ca²⁺ influx (increase in driving force for Ca²⁺ ions of 30 to 110 mV) for a defined period of time (varied between 0.06 and 15.4 ms, details in Figure 5 legend), and (3) a test pulse to 70 mV selectively monitoring the BK_Ca-mediated K⁺ currents.

Figure 5.

Operation of Cav-associated BK_Ca channels by their Ca²⁺ gating. A, Experimental conditions for analysis of the Ca²⁺ gating in BK_Ca–Cav2.2 complexes. Channel–channel complexes were reconstituted in CHO cells upon coexpression of BKα, BKβ4, Cav2.2α1, α2δ, and Cavβ1b; whole-cell recordings were done with 10 mm EGTA in the pipette to eliminate currents through Cav-free BK_Ca channels (Berkefeld et al., 2006). B, C, Representative currents through BK_Ca–Cav2.2 complexes recorded in whole-cell configuration in response to the indicated “scanning-tail-current protocol” (details in the text). Step to 70 mV activated Cav channels without eliciting Ca²⁺ influx (effective reversal potential for Ca²⁺ slightly below this potential), steps to −40 mV (B; termed conditioning pulse), and 20 mV (C) triggered Ca²⁺ influx (increase in driving force) for defined periods (0.06, 0.12, 0.24, 0.48, 0.96, 1.92, 3.84, 7.68, and 15.36 ms). The final step back to 70 mV switched off the Ca²⁺-inward currents and elicited K⁺-outward currents. Traces in red are responses recorded with a conditioning pulse duration of 0.96 ms (B) or 7.68 ms (C); 0 time point is start of the conditioning pulse. Note the similarity in amplitudes of the two BK_Ca outward currents despite the largely different periods of Ca²⁺ influx. Bottom, Amplitudes of BK_Ca currents recorded 0.2 ms after the final pulse step (I _{BK,instantaneous}, filled symbols; “time” is duration of conditioning pulse plus 0.2 ms) or of BK_Ca peak currents (I _BK,ipeak, open symbols) plotted over the period of Ca²⁺ influx (start of the conditioning pulse). D, BK_Ca currents from an experiment as in B and C with a conditioning pulse of 0.96 ms to −40 mV (blue) and 20 mV (black). E, I_{BK,instantaneous}-time relation from the experiment in B and C at expanded timescale (time points are conditioning interval plus 0.2 ms). Note that the BK_Ca current is larger in amplitude at −40 mV than at 20 mV, and that the time course of channel activation is markedly faster at −40 mV than at 20 mV.

Figure 5B–D illustrates representative current traces recorded during such a “scanning tail-current protocol” with two distinct conditioning pulse potentials of −40 and 20 mV. In either case, the Cav-mediated Ca²⁺ influx effectively activated BK_Ca channels as visualized by the transient outward K⁺ currents recorded during the test pulse (see also Figure 5 legend). However, the time course of the BK_Ca currents was largely different for the two conditioning pulses tightly mirroring the distinct gating kinetics of the Cav channel(s) at the respective potentials (Fig. 5B,C, inset). Thus, BK_Ca currents rapidly rose and fell at −40 mV as a consequence of the pronounced Cav channel deactivation (time constant of ∼0.5 ms), while they steadily increased over time at 20 mV promoted by the only moderate inactivation of the Cav channels (time constant of ∼5 ms). This distinction was independent of whether BK_Ca currents were measured as a peak outward current (I_BK,peak) or as a tail current at the test step (I_{BK,instantaneous}), which directly monitored the Ca²⁺-driven gating during the conditioning pulse (Fig. 5B,C).

Closer investigation of I_BK (I_{BK,instantaneous}) recorded at conditioning pulses of short length (between 0.06 and 1.2 ms; see also Fig. 5 caption) revealed two further peculiarities of BK_Ca activation in the BK_Ca–Cav complexes. First, activation at 20 mV occurred only after an initial time lag reminiscent of the pre-onset delay seen in fast application experiments (Fig. 5E) and, second, activation was markedly faster at −40 mV than at 20 mV (Fig. 5D,E). The latter observation is counterintuitive with voltage-dependent gating, but consistent with channel activation driven by Ca²⁺ whose delivery at −40 mV largely exceeded that at 20 mV as reflected by the amplitudes of the Ca²⁺-inward current (Fig. 5B–D). In fact, analysis of the time course of BK_Ca activation on an extended voltage range from −40 to 40 mV further corroborated this finding: the speed of BK_Ca activation steadily decreased with depolarization of the membrane potential during the conditioning pulse (Fig. 6A).

Figure 6.

Determination of the K⁺-current output of BK_Ca–Cav complexes by the properties of BK_Ca Ca²⁺ gating. A–C, I_{BK,instantaneous}-time relations determined for BK_Ca–Cav complexes of the indicated composition in experiments as in Figure 5 but for the conditioning pulse potentials between −40 and 40 mV given by the inset in A. Currents were normalized to the maximum recorded at a conditioning pulse potential of −40 mV; data points are mean ± SEM 13 (BKα), 10 (BKα+ BKβ4), and 6 (BKα(D369G)) experiments. Note the slowing of BK_Ca-channel activation with increasing depolarization and the lack of delay obtained with the BK_Ca(D369G) mutant. D, Overlay of data from A to C recorded at a conditioning pulse potential of 20 mV.

These results strongly suggested that in BK_Ca–Cav complexes Ca²⁺ dominates over voltage as the rate-limiting stimulus for activation of the BK_Ca channels and prompted experiments probing the effect of altered Ca²⁺ gating on the complexes' K⁺-current response. For this purpose, the experiments in Figures 5 and 6A were repeated with the BKα(D369G) mutant or with BKβ4 co-assembled with wild-type BKα. As illustrated in Figure 6B–D the K⁺ currents in either type of macromolecular BK_Ca–Cav complex increased in amplitude with hyperpolarization of the conditioning pulse; however, while the D to G exchange promoted a significant acceleration of the activation time course together with a reduction in the initial lag time (in particular at voltages ≥0 mV; Fig. 6B,D), the changes induced by the co-assembled BKβ4-subunit were rather moderate (Fig. 6D). Both observations fit the results obtained from the fast application experiments on BK_Ca channels expressed alone (Fig. 2) and emphasize that activation of Cav-associated BK_Ca channels showed the hallmark properties of Ca²⁺ gating. In addition, direct comparison of the activation time courses obtained from Cav-free and Cav-associated BK_Ca channels revealed a surprising difference (Fig. 7): while the pre-onset delay in isolated BK_Ca channels were in the range of 1–2 ms (measured for 10–100 μm Ca²⁺ at 50 mV), it was as short as ∼300 μs for Cav-associated channels (at 40 mV; Fig. 7). This reduction in lag time is very similar to that obtained in the D369G mutant (∼400 μs; Figs. 4C, 7) and thus suggests that it results from altered gating induced by direct interaction between the BK_Ca and the Cav channel proteins (Berkefeld et al., 2006) rather than from a high local [Ca²⁺]_i.

Together, these results demonstrated that in macromolecular complexes with the Cav channels, BK_Ca channels are primarily activated by Ca²⁺ ions suggesting that, under physiological conditions, they are operated as (voltage-sensitive) ligand-gated channels similar to their SK_Ca counterparts (Xia et al., 1998; Berkefeld et al., 2010).

Discussion

The central finding of the present work is that BK_Ca channels in macromolecular complexes with Cav channels behave as ligand-operated channels with Ca²⁺ gating determining channel activation. Accordingly, in this configuration depolarization serves primarily to activate the Cav subunits which, in turn, provide the elevation in [Ca²⁺]_i that drives the Ca²⁺ gating of the BK_Ca channels. In addition, the co-assembly with Cav channels appears to allosterically tune BK_Ca channels for fast responses by markedly reducing the pre-onset delay of their Ca²⁺ gating.

Ca²⁺ gating: characteristics and mechanism

We used voltage-clamp recordings on patches excised from CHO cells together with a rapid piezo-driven solution exchange system to obtain precise and fast manipulations of either of the two distinct input stimuli (Fig. 1). In particular, the quality of the [Ca²⁺]_i steps defining the time resolution of Ca²⁺ gating was ensured by determining the time constants of solution exchange at any inside-out patch before use (Fig. 1A). Respective estimates taken from exchange of K⁺ concentrations at open BK_Ca channels yielded values ≤0.3 ms (for either direction; Fig. 1A), and effective Ca²⁺ exchange was evidenced by the minimal pre-onset delay obtained with BKα(D369G) mutant channels that was as low as 0.4 ms (Figs. 4, 7).

Using this configuration we first showed that BK_Ca channels can effectively be gated by [Ca²⁺]_i and thus operate as ligand-activated channels (Markwardt and Isenberg, 1992; Horrigan and Aldrich, 2002), analogous to the SK-type Ca²⁺-activated K⁺ channels (Xia et al., 1998; Adelman et al., 2012). Different from SK_Ca channels, however, the Ca²⁺ gating of BK_Ca is sensitive to changes in the transmembrane voltage (Fig. 2) as a result of the channels' allosteric coupling of both Ca²⁺ and voltage gating branches through the pore-forming domains of the channel protein (Cui et al., 1997; Rothberg and Magleby, 2000; Horrigan and Aldrich, 2002). Direct comparison of the two gating branches showed that Ca²⁺ gating differs from voltage gating in mainly two aspects: (1) a prominent delay between Ca²⁺ application and pore opening (termed pre-onset delay; Figs. 1, 2) and (2) a slower activation time course to about the same extent over the entire [Ca²⁺]_i-voltage plane (Figs. 2, 3).

These two characteristics suggest that the conformational dynamics responsible for Ca²⁺ gating are different from those underlying voltage-driven gating. In this sense, the pre-onset delay and its dependence on [Ca²⁺]_i most likely reflect concentration-dependent binding of Ca²⁺ ions to respective site(s) in all four BKα-subunits and the subsequent structural rearrangements preceding channel opening. Importantly, the delay period measured at gating-saturating [Ca²⁺]_i of 100 μm was ∼1.5 ms and thus considerably longer than the period required for complete solution exchange at the cytoplasmic face of the channels (Figs. 1A, 4C, 7). Indeed, 100 μm is about an order of magnitude higher than what the BK_Ca channels encounter when co-assembled with Cav channels, as seen by their activation kinetics (Berkefeld et al., 2006; Berkefeld and Fakler, 2008). The slower speed of Ca²⁺-gated BK_Ca channel activation may reflect a pore-opening mechanism distinct from the conformational changes underlying voltage-dependent gating. Thus, while voltage gating in BK_Ca, similar to that in Kv channels (Perozo et al., 1999; Long et al., 2007; Pathak et al., 2007), is envisaged to operate through the moving voltage sensor (S2–4 helices) transferring energy to the pore-forming S5/6 helices via the S4/5 linker, Ca²⁺ gating implies coupling of the cytoplasmic agonist binding to the pore via the RCK-S6 linker (16 aa in length) and the S6 helix.

In fact, recent crystallographic work on the Ca²⁺ “gating ring” of BK_Ca channels in both the apo and Ca²⁺-bound form put forward a structural model for Ca²⁺ gating supporting this notion (Wu et al., 2010; Yuan et al., 2010, 2012). Accordingly, binding of Ca²⁺ to the four RCK2 domains of the gating ring triggers conformational changes in the N-terminal lobes of the adjacent RCK1 domains that, via a petal-like widening, directly enforce the S6 helices to undergo structural rearrangements underlying pore opening (Yuan et al., 2012). Moreover, in the framework of this model, the speed of the force transfer should be directly affected by the structural constraints of the RCK-S6 linker, in line with the strong effect observed for the pre-onset delay with the BKα(D369G) mutation (Figs. 4, 7; Yang et al., 2010).

Ca²⁺ gates BK_Ca channels in Cav–BK_Ca complexes

Experiments in whole-cell configuration examined the gating of BK_Ca channels in macromolecular BK_Ca–Cav complexes, reflecting the co-assembly of both channels in the mammalian brain (for review, see Berkefeld et al., 2010). In BK_Ca–Cav complexes, the K⁺-current output is tightly controlled by the gating properties of the particular Cav subtype co-assembled with BK_Ca (Fig. 5B,C; Berkefeld and Fakler, 2008).

Application of a pulse protocol separating voltage-dependent gating of the Cav channels from Ca²⁺ influx and subsequent increase in [Ca²⁺]_i (Fig. 5B,C) showed that (1) the speed of BK_Ca activation increased upon hyperpolarization, but decreased with depolarization of the membrane potential (Figs. 5E, 6A–C); (2) BK_Ca activation displayed an initial delay, in particular at membrane potentials ≥0 mV (Figs. 5E, 7); and (3) both activation speed and initial delay were affected by the D to G mutation in the RCK-S6 linker (Fig. 6D).

All these characteristics are consistent with BK_Ca channels being activated by [Ca²⁺]_i rather than by voltage and, consequently, Ca²⁺ gating dictating the time course of the K⁺-current output in BK_Ca–Cav complexes. In this context, despite the inherent voltage sensitivity of Ca²⁺ gating, changes in membrane potential primarily serve as the gating trigger for the Cav channels which, in turn, provide the increase in [Ca²⁺]_i required for effective activation of BK_Ca. The latter is most effective at highest [Ca²⁺]_i as promoted by hyperpolarization of the membrane potential due to the resulting increase in driving force (Figs. 5, 6).

This “sequential cascade” derived for activation of Cav-associated BK_Ca channels that reflect the configuration in many CNS neurons should be particularly effective during an action potential (Berkefeld and Fakler, 2008). The depolarizing upstroke initiates activation of the Cav channels, which pass Ca²⁺ into the cytosol mainly during the repolarization phase of the action potential (Li et al., 2007) when the driving force for Ca²⁺ is largest. The resulting increase in [Ca²⁺]_i serves as an effective input stimulus to the BK_Ca Ca²⁺ gating, which is further speeded by the reduction in the pre-onset delay established through the direct protein–protein interactions between Cav and BK_Ca channels (Fig. 7). Together, these precisely timed processes should engender effective and rapid activation of the BK_Ca channels under physiological conditions.