Article Information

• PDF (584 kb)
• Full Text with Large Figures
• Supporting Material

PubMed

• Articles by Hyun-Ju Kim
• Articles by Chul-Seung Park

Related Articles

• Crystal Structures of a Ligand-free MthK Gating Ring: Insigh...

  Crystal Structures of a Ligand-free MthK Gating Ring: Insights into the Ligand Gating Mechanism of K Channels Cell, Volume 126, Issue 6, 22 September 2006, Pages 1161-1173 Sheng Ye, Yang Li, Liping Chen and Youxing Jiang Summary MthK is a prokaryotic Ca-gated K channel that, like other ligand-gated channels, converts the chemical energy of ligand binding to the mechanical force of channel opening. The channel's eight ligand-binding domains, the RCK domains, form an octameric gating ring in which Ca binding induces conformational changes that open the channel. Here we present the crystal structures of the MthK gating ring in closed and partially open states at 2.8 Å, both obtained from the same crystal grown in the absence of Ca. Furthermore, our biochemical and electrophysiological analyses demonstrate that MthK is regulated by both Ca and pH. Ca regulates the channel by changing the equilibrium of the gating ring between closed and open states, while pH regulates channel gating by affecting gating-ring stability. Our findings, along with the previously determined open MthK structure, allow us to elucidate the ligand gating mechanism of RCK-regulated K channels. Summary | Full Text | PDF (1137 kb)

• Cytoplasmic gatekeepers of K-channel flux: a structural pers...

  Cytoplasmic gatekeepers of K-channel flux: a structural perspective Trends in Biochemical Sciences, Volume 29, Issue 1, 1 January 2004, Pages 39-45 Tarmo P Roosild, Khanh-Tuoc Lê and Senyon Choe Abstract Recently, rapid progress in our structural knowledge of K-selective channels has started to provide a basis for comprehending the biophysical machinery underlying their electrophysiological properties. These studies have begun to reveal how a diverse array of distinct, cytoplasmically positioned domains affect the activity of associated channels. Some of these establish functional diversity by selectively mediating channel assembly. More importantly, these cytoplasmic domains couple intracellular signals to the gating of their associated pore. New structural insights are providing a clearer understanding of the fundamental molecular mechanisms of these K channels that, in turn, partly underlie complex neurological phenomena. Abstract | Full Text | PDF (2284 kb)

• Voltage-gated K Channels - II

  Voltage-gated K Channels - II Biophysical Journal, Volume 94, Issue , 1 February 2008, Pages 184-194 Full Text | PDF (134 kb)

• …more

Copyright © 2008 The Biophysical Society. All rights reserved.
Biophysical Journal, Volume 94, Issue 2, 446-456, 15 January 2008

doi:10.1529/biophysj.107.108738

Channels, Receptors, and Electrical Signaling

Modulation of the Conductance-Voltage Relationship of the BK_Ca Channel by Mutations at the Putative Flexible Interface between Two RCK Domains

Hyun-Ju Kim^*, Hyun-Ho Lim^*, †, Seong-Hwan Rho^*, Lin Bao^‡, Ju-Ho Lee^*, Daniel H. Cox^‡, Do Han Kim^* and Chul-Seung Park^*, †, ,  

^* Department of Life Science, Gwangju Institute of Science and Technology, Gwangju, Korea
^† Center for Distributed Sensor Network, Gwangju Institute of Science and Technology, Gwangju, Korea
^‡ Department of Neuroscience, Tufts University School of Medicine, Boston, Massachusetts USA

Address reprint requests to Chul-Seung Park, PhD, Dept. of Life Science, Gwangju Institute of Science and Technology, 1 Oryong-dong, Buk-gu, Gwangju, Korea, 500-712. Tel.: 82-62-970-2489; Fax: 82-62-970-2484.

Large-conductance calcium-activated potassium (BK_Ca) channels play an important role in modulating a number of physiological processes, such as neuronal excitability, smooth-muscle contraction, frequency tuning of hair cells, and immunity 1,2,3,4,5,6,7,8. The opening of these channels is promoted by membrane depolarization and elevated cytosolic free calcium via separate regions of the α-subunit (Slowpoke or Slo) 9,10,11,12. The positive charges on transmembrane domains (S1–S4) are responsible for voltage sensing 13,14,15,16,17,18,19, whereas the calcium-dependent activation of the channel is mediated by the bulky cytoplasmic carboxy terminus 20,21,22. The cytoplasmic C-terminus of Slo has been proposed to contain a high-affinity Ca²⁺-binding site (Ca²⁺-bowl) and a structural module known as the regulator of K⁺ conductance (RCK) domain 21,23,24,25,26,27,28. The Ca²⁺-bowl is composed of a series of Asp residues and binds Ca²⁺ with micromolar affinity. Mutations here have been shown to cause positive shifts in the Slo channel's conductance-voltage relationship at constant Ca²⁺ concentrations 21,22,25,26,29,30. Recognized initially in a Ca²⁺-activated K⁺ channel of Methanobacterium thermoautotrophicum (MthK), a double ring composed of eight RCK domains (gating ring) has been studied intensively as a model for Ca²⁺-dependent channel gating 31. Based on sequence homology and supporting experimental evidence, two RCK domains have been proposed to lie in tandem within the long C-terminus of the mammalian BK_Ca channel 23,32,33,34. The first RCK domain (RCK1), located at the proximal C-terminus of the Slo protein, has been characterized in detail using mutational analysis. RCK1 contains two divalent-cation-binding sites with high and low affinity for Ca²⁺, respectively 26,27,28. A gain-of-function mutation that causes epileptic seizures has also been localized to RCK1 35. In the exogenously expressed mutant channel, the calcium sensitivity was found to increase by three- to fivefold over the wild-type and the conductance-voltage (G-V) relationship was shifted toward more negative potentials by 57mV. The second RCK domain (putative RCK2), which is essential for Ca²⁺-dependent gating of the channel, is located in the distal region of the C-terminus and is immediately followed by the Ca²⁺-bowl. Recently, we reported that hydrophobic interactions between RCK1 and the putative RCK2 are essential for the gating of the channel. This interaction is reminiscent of the fixed (assembly) interface between the two RCK domains within the gating ring of MthK 23. It was also suggested that the dimeric pairs of RCK domains on a single subunit are responsible for cooperative activation of BK_Ca channels by calcium 36.

In a search for important amino acid residues for channel activation in the putative RCK2, we mutagenized several residues located in the region corresponding to the flexible interface as revealed in the crystal structure of the MthK gating ring. Among the mutant channel, an Asp substitution at Gly-803 greatly enhanced channel activity as evidenced by as much as an 80-mV shift in the G-V curve toward a more negative potential. This potentiation can be explained by a stabilization of the open conformation that neither leads to a change in Ca²⁺ affinity nor a change in voltage sensitivity. The residue is predicted to be located at the interface between RCK1 and the putative RCK2, suggesting that the region of the putative RCK2 that interfaces with RCK1 is important for determining the energetics of the conformational changes that lead to channel opening. This region may mediate the conformational coupling between downstream Ca²⁺ binding at the Ca²⁺-bowl and the upstream RCK1 and transmembrane domains.

All electrophysiological experiments, except for the gating-current measurements, were performed on Chinese hamster ovary (CHO)-K1 cells expressing the rat Slo channel gene (GenBank accession No. AF135265) 37. CHO-K1 cells were maintained in F-12K nutrient mixture, Kaighn's modification (GIBCO, Carlsbad, CA), supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA), in a humidified atmosphere at 5% CO₂ and 37°C. To obtain high-quality plasmid DNA for transient transfection, the plasmid DNA was prepared using a commercial kit (Qiagen, Valencia, CA).

To measure the gating current, wild-type and mutant rSlo channels were expressed in Xenopus oocytes. We used complementary DNA for rSlo subcloned into a modified pGH expression vector for high-level expression. Complementary RNAs for the rSlo channel were synthesized in vitro using T7 RNA polymerase (Ambion, Austin, TX) and plasmids were linearized with NotI. Approximately 150–200ng of cRNA were injected into oocytes. The injected oocytes were incubated at 18°C for 4–8 days in ND96 solution.

Silent mutations were introduced at amino acid positions 728, 729, and 1016 of rSlo (GenBank accession No. GI:4972782) using two sequential polymerase chain reactions to create AgeI (from ACGG to ACGG) and XhoI (from CTCGA to CTCGA) restriction sites. Cassette mutagenesis was performed to substitute specific amino acid residues. Mutations were generated by polymerase chain reaction using mutagenic primers. The amplified DNA fragments flanked by AgeI and XhoI were substituted for the wild-type rSlo gene cloned into the pcDNA3.1(+) vector using BamHI and XbaI restriction sites. To confirm the DNA sequence of each mutant channel, DNA sequencing was performed using an ABI 377 automatic DNA sequencer (PerkinElmer Life and Analytical Sciences, Foster City, CA).

Most of the macroscopic ionic currents carried by wild-type and mutant rSlo channels were recorded in excised membrane patches of CHO-K1 cells with inside-out configurations using an Axopatch 200B amplifier (Axon Instruments, Sunnyvale, CA). All patch recordings were performed at room temperature 24–48h after transfection. Pipettes were prepared from thin-walled borosilicate glass (World Precision Instruments, Sarasota, FL) and fire-polished. Macroscopic currents of rSlo channels were activated by voltage pulses delivered from a holding potential of −100mV to test potentials ranging from −150 to 200mV in 10-mV increments. When filled with the solutions described below, the input resistances of electrodes for macroscopic currents were 2.5–3.5MΩ, whereas for single-channel recording they were >4.0MΩ. Average series resistance, ∼2.9MΩ, was used to compensate electronically for the macroscopic recordings. Signals were filtered at 1–2kHz using a four-pole low-pass Bessel filter, digitized at a rate of 10kHz using a Digidata 1200 (Axon Instruments), and stored in a personal computer. Commercial software packages such as Clampex 8.1 (Axon Instruments) and Origin 6.1 (OriginLab, Northampton, MA) were used for the acquisition and analysis of macroscopic data.

Single-channel opening events were obtained from patches containing one to hundreds of channels. Recordings were obtained for durations of 20 to hundreds of seconds. Unitary currents were sampled at 200kHz and filtered at 10kHz. Analysis was performed using the Clampfit program (Axon Instruments). In the limiting P_O measurements, open probabilities <10⁻³ were determined at 0 [Ca²⁺] with patches that contained hundreds of channels. The number of channels in a given patch (N) was obtained from the instantaneous tail current amplitude during maximal opening at saturating [Ca²⁺]_i divided by the unitary conductance for each channel at the tail voltage.

For the gating-current measurements, patch pipettes were made of borosilicate glass (VWR, West Chester, PA) and their tips were coated with sticky wax (Sticky Wax, Dharma Trading, San Rafael, CA) and fire-polished to a resistance of 0.5–1MΩ. Voltage commands were filtered at 7.5kHz. Data were acquired at 100kHz and filtered at 10kHz with an Axopatch 200B amplifier and a Macintosh-based computer system equipped with an ITC-16 hardware interface (Instrutech, Port Washington, NY) and pulse-acquisition software (HEKA Electronik, Southboro, MA). Data analysis was performed with Igor Pro graphing and curve-fitting software (WaveMetrics, Oswego, OR).

Solutions for macroscopic ionic-current recording were prepared according to Lim and Park 38. Pipette solutions contained 10mM HEPES, 2mM EGTA, 116mM KOH, and 4mM KCl. The intracellular solution for perfusing to the internal face of excised patches was the same as the pipette solution except for supplemental CaCl₂. To provide the precise free concentration of intracellular Ca²⁺ ([Ca²⁺]_i), the appropriate amount of total Ca²⁺ to be added to the intracellular solution was calculated using the program MaxChelator 39. The pH was adjusted to 7.2 with 2-[N-morpholino]ethanesulfonic acid. Gating-current recording solutions were prepared according to Bao and Cox 40. The composition of the pipette solution was 127mM triethylammonium-OH, 125mM HMeSO₃, 2mM HCl, 2mM MgCl₂, 20mM HEPES, pH 7.2 (adjusted with HMeSO₃ or triethylammonium-OH). The internal solution contained 141mM N-methyl-D-glucamine, 135mM HMeSO₃, 6mM HCl, 20mM HEPES, 40μM (+)-18-crown-6-tetracarboxylic acid (18C6TA), 5mM EGTA, pH 7.2 (adjusted with NMDG and HMeSO₃).

On the basis of the gating ring structure of MthK, we constructed a heterodimeric structural model of the Slo channel's putative flexible interface between RCK1 and RCK2 (Figure 1A, left). Although the flexible interfaces of the homomeric RCK dimers in MthK are formed by two helices from each RCK domain's N-terminal lobe (αF and αG), and also parts of their C-terminal lobes 23, our model of the Slo channel's heterodimeric RCK dimer predicts that RCK1 interacts with RCK2 mainly through αF helices. This is because of the apparent absence of an αG helix and a C-terminal lobe in the putative RCK2. We surveyed the amino acid residues at the contact sites in our model between RCK1 and RCK2, and residues of the putative RCK2 that appeared to be in close proximity to charged residues in RCK1 were identified (Figure 1A, right). We, then, substituted each interfacial residue of the putative RCK2 with a residue that had the counter charge of the one in RCK1 to create a charge pair across the interface.

Display large version of this figure

Figure 1

Effects of charge mutations at the putative flexible interface of between RCK1 and the putative RCK2 in BK_Ca channel. (A) Structural model of RCK1-2 dimer interacting via putative flexible interface (left). RCK1 and the putative RCK2 domains are denoted yellow and green, respectively. Several charged residues in RCK1 and their adjacent residues in the putative RCK2 are shown with stick in enlarged dimeric interface (right). (B) Representative raw traces of wild-type and mutant channels. In mutant channels, each interfacial residue in the putative RCK2 was substituted with countercharged residues of adjacent residue in RCK1. Ionic currents were induced by voltage steps ranging from −20 to 140mV in 40-mV increments from the holding voltage of −100mV. (C) The conductance-voltage (G-V) relationships of the wild-type (open symbols) and several mutant channels (solid symbols) are shown. The symbols for each mutant channel are indicated in the inset. Each channel was recorded with 2μM [Ca²⁺]_i. The curves were fitted with the Boltzmann function. (D) The half-activation voltage (V_1/2) of the wild-type and mutant channels were determined at 2μM [Ca²⁺]_i. The V_1/2 of the wild-type is indicated as a dotted line. Each data point represents the mean±SE. Values differing from the wild-type by the paired Student's t-test at p<0.05 (*) or p<0.01 (**) are indicated.

The wild-type channel and each mutant were expressed in CHO-K1 cells, and their functional activities were investigated electrophysiologically. Figure 1B shows representative macroscopic currents from each channel type, evoked by a common voltage protocol in the presence of 2μM [Ca²⁺]_i. As is evident, the mutations showed varied effects. In Figure 1C, the steady-state extent of activation of the four mutant channels is plotted as a function of voltage and these plots compared with wild-type. The G-V relations of the G803D and N806K mutants were strongly shifted toward more negative voltages, whereas the G-V relation of M934D was slightly shifted in a positive direction. Compared to the half-activation voltage (V_1/2) of the wild-type channel, 79mV, the V_1/2 values of the G803D and N806K mutants were −3mV and 34mV, respectively (Figure 1D), which correspond to shifts of 82 and 45mV in the negative direction. Despite the fact that the mutants’ G-V curves were spread across an ∼90-mV range, the slope of each G-V curve representing the apparent voltage dependence of the channel activation was not altered significantly. The apparent gating charges determined from the slope of each activation curve are as follows: for wild-type, 1.31e±0.07; for G803D, 1.38e±0.06; for N806K, 1.47e±0.04; for N912K, 1.41e±0.08; for M934D, 1.38e±0.06.

As we were intrigued by the dramatic G-V shift generated by the Gly-to-Asp mutation at position 803, we scrutinized the mutant channel for insights into the mechanism involved. As shown in Figure 2A, the ionic currents of G803D were activated at more negative voltages and at lower concentrations of intracellular Ca²⁺ than were those of the wild-type channel. The G-V relationships of the wild-type channel (open symbols) and the mutant (solid symbols) are plotted at 0.5μM (rectangles), 2μM (circles), and 20μM (triangles) [Ca²⁺]_i in Figure 2B. At all three [Ca²⁺]_i the mutation negatively shifted the G-V curve by ∼80mV. We then quantified the effects of intracellular Ca²⁺ on the position of the mutant's G-V relationship by plotting V_1/2 vs. [Ca²⁺]_i for [Ca²⁺]_i ranging from 0.5 to 20μM. In the semilogarithmic plot shown in Figure 2C, the V_1/2 values of both the wild-type and mutant channels were found to be linearly related to [Ca²⁺]_i; however, the G-to-D mutation shifted the V_1/2 value of the each G-V curve by −78mV at all [Ca²⁺]_i tested.

Display large version of this figure

Figure 2

Functional analysis of the G803D mutant channel. (A) Representative macroscopic current traces of the wild-type and G803D mutant recorded at different [Ca²⁺]_i are shown. The concentrations of intracellular calcium are indicated. Ionic currents were induced by voltage steps ranging from −20 to 140mV at 20-mV increments from the holding voltage of −100mV. Current traces represent an average of three successive records. (B) The G-V relationships of the wild-type (n=14) and G803D mutant (n=11) are shown for 0.5μM (rectangles), 2μM (circles), and 20μM (triangles) [Ca²⁺]_i. The wild-type and G803D are indicated by open and solid symbols, respectively. Each data set was fitted with the following equation: G/G_max=1/[1+{(1+[Ca²⁺]/K_C)/(1+[Ca²⁺]/K_O)}⁴×exp(−QFV/RT)/L₀], where K_C is the dissociation constant of [Ca²⁺] in the closed state, K_O is the dissociation constant of [Ca²⁺] in the open state, Q represents the equivalent gating charge associated with this equilibrium, and L₀ is the [O]/[C] at 0 [Ca²⁺]_i. (C) Relationship between [Ca²⁺]_i and the half-activation voltages. Open and solid symbols denote the wild-type and G803D, respectively. Each data point represents the mean±SE.

To address which aspects of Slo-channel gating were altered by the G-to-D substitution in the putative RCK2 domain, we simulated our experimental data based on the simple voltage-dependent version of the Monod-Wyman-Changeux (MWC) model previously proposed as a gating model for the BK_Ca channel 41. In this model, there are several simplifying assumptions. The channel is a homotetramer containing four Ca²⁺-binding sites, one on each subunit. Ca²⁺ can bind to both the open and closed conformations but it favors the open conformation, and thereby promotes the opening. The voltage sensors from each subunit move in a highly concerted way such that their movement can be represented by a single voltage-dependent conformational change. Four different parameters obtained from the simulation were compared between the wild-type and G803D (Table 1). Three parameters were not significantly affected (i.e., the apparent gating charge (Q), and the affinities for Ca²⁺ in the closed state (K_C) and the open state (K_O)), however, the close-to-open equilibrium constant in the absence of bound Ca²⁺ at 0mV, L₀, was altered dramatically from 8.6×10⁻⁴ to 0.021. The results of our simulation suggest that the substitution of Asp at position 803 alters the equilibrium constant for channel opening by 24-fold.

The potentiating effects of the G803D mutation were then examined at the single-channel level. Single-channel current traces of the wild-type channel and G803D were compared in the absence and presence of 2μM [Ca²⁺]_i (Figure 3A). As predicted from our analysis of macroscopic currents, the mutant channel opened more frequently than did the wild-type channel, even in the absence of Ca²⁺. And, at 2μM Ca²⁺, where the wild-type channel exhibited significant opening, G803D always displayed a higher open probability (see the expanded timescale in Figure 3A).

Display large version of this figure

Figure 3

Effects of the Gly-to-Asp substitution on single-channel characteristics. (A) Representative current traces of single wild-type and G803D mutant channels recorded at two different voltages and [Ca²⁺]_i. The scales for both the current and time are the same for the wild-type and mutant channels. Portions of the current traces are presented in an expanded timescale. The current levels of the single open state (O) and closed state (C) are indicated. (B) The open probability of the wild-type (□) and G803D (■) were plotted against the membrane voltage at which the single-channel recordings were obtained. Intracellular calcium was set at 2μM throughout. Each data set was fitted by the Boltzmann function. (C) The single-channel current amplitudes were plotted against membrane voltage. Single-channel currents were obtained by fitting the all-point histograms to the Gaussian function. The unitary conductances estimated from the linear fit of each slope were 232±3 pS for the wild-type and 229±3 pS for G803D.

The open probabilities (P_O) of the wild-type and the mutant channel were plotted against membrane voltage in Figure 3B. The P_O of both channels increased as the membrane potential was set to more positive values. However, the wild-type channel was activated in the voltage range of 20–100mV, whereas the G803D channel was activated at far lower voltages, between −80 and 40mV. Thus, the mutation negatively shifts the rSlo channel's P_O versus voltage curve by 88.6mV (Figure 3B). This shift is similar to the G-V shift we observed with macroscopic current recordings, 82mV (Figure 2B). Despite its large effects on gating, the mutation did not affect the single-channel conductance of the rSlo channel. When we plotted the current amplitudes obtained from all-point histograms at different membrane voltages, the unitary conductances of the wild-type channel and G803D mutant were determined to be 232±3pS and 229±3pS, respectively (Figure 3C).

Since the simulation of the macroscopic current recordings suggested a significant shift in the intrinsic gating equilibrium of the G803D mutant channel, we directly determined the equilibrium constants for the intrinsic close-to-open conformational change for both channels by measuring open probability (P_O) at hyperpolarized membrane potentials. According to the Horrigan et al. model 43,44,45, at extreme negative voltages, where no voltage sensors are active, P_O=L₀exp(z_LFV/RT), which can be rearranged to log (P_O)=0.4342z_LFV/RT+log(L₀) 40,46, where L₀ represents the equilibrium constant between closed and open at 0mV in the absence of Ca²⁺ when no voltage sensors are active, z_L is a small amount of gating charge associated with the closed-to-open conformational change, and F, R, and T have their usual meanings. This equation can be interpreted to mean that, as the membrane voltage becomes more and more hyperpolarized, a plot of log(P_O)-voltage will reach a limiting slope that is less than its maximum slope and reflects the voltage dependence of just the close-open conformational change, z_L. That is, in this voltage range, the slope of the log(P_O)-voltage relation will be determined only by z_L, and the position of the curve in the vertical axis will be determined only by L₀. Thus, we can directly estimate z_L and L₀ by determining the P_O versus voltage relation at far negative voltages. Here, it is important to note that the L₀ value is defined differently from the L₀ simulated with the voltage-dependent MWC model (Table 1). Whereas the L₀ parameter in the voltage-dependent MWC model represents the equilibrium constant between open and closed at 0mV in the absence of Ca²⁺, the L₀ value obtained from limiting slope measurements represents the equilibrium constant between open and closed at 0mV in the absence of Ca²⁺ if no voltage sensors are active 42. Thus, the L₀ parameter in the simple voltage-dependent MWC model includes L₀ as defined by Horrigan and Aldrich but also factors that depend on voltage sensor activation 43,44,45.

We measured P_O for the wild-type channel and the G803D mutant over a wide range of voltages at 0 [Ca²⁺]_i and determined the effects of the mutation on intrinsic and voltage-dependent gating. For the wild-type channel, P_O values between −150 and +60mV were measured using single-channel recordings, and P_O values between +70 and +200mV were obtained from macroscopic recordings. In the case of the G803D mutant, single-channel recordings were used for voltages between −150 and +20mV and macroscopic recordings were used for voltages between +30 and +200mV. Examples of single-channel recordings are shown in Figure 4A. Notice at each voltage the G803D mutant opens more frequently than does the wild-type channel. In Figure 4B, log(P_O)-voltage relations for the wild-type and mutant channels are compared. As the voltage becomes more negative, both plots deviate from linear. The wild-type curve deviates at around −30mV, and reaches a limiting slope by −110mV. By fitting just the most negative part of this curve, L₀ and z_L were calculated to be 8.69×10⁻⁷ and 0.3e, respectively. The slope of the mutant channel's log(P_O)-voltage relation deviates from linear at around −70mV and approaches a limiting slope near −140mV. The values of L₀ and z_L for the mutant channel were by the same analysis determined to be 2.04×10⁻⁵ and 0.64e, respectively. Thus, this analysis suggests that the Gly-to-Asp mutation stabilized the open state of the channel by 23-fold, a number very similar to that which we obtained with voltage-dependent MWC modeling, 24-fold (Table 1). Thus, we can assume that most of the mutation's effects on gating come from its effect on the intrinsic energetics of channel opening, rather than on Ca²⁺ binding or voltage sensor movement.

Display large version of this figure

Figure 4

Estimation of L₀ and z_L from open probability (P_O) at negative voltages. (A) Unitary currents were recorded at 0 Ca²⁺ from the wild-type and G803D channels. Each patch contains 2,768 channels of wild-type and 2,664 channels of G803D channels. (B) Mean log(P_O)-voltage relations are shown for the wild-type (□) and G803D (●) channel. Linear fits of log (P_O)-V relation at the “steep phase” (dashed line) indicate that the measurements either have reached or are approaching the limiting slope. The dashed line represents best fits to the Boltzmann function. The solid line at the bottom of the data point was fitted with the following function: log(P_O)=log(L₀)+0.4342z_LFV/RT. The resulting parameters were L₀=8.69×10⁻⁷ and z_L=0.3e for the wild-type, and L₀=2.04×10⁻⁵ and z_L=0.64e for G803D.

As a member of the voltage-gated ion-channel superfamily, the BK_Ca channel senses the membrane voltage by moving its voltage sensor within the S2–S4 segment of the α-subunit 19. Membrane depolarization drives the outward movement of the positively charged voltage sensor and evokes outward or “on-gating” currents, whereas membrane hyperpolarization generates inward or “off-gating” currents. To check whether the G803D mutation brings about any changes in other measurable gating parameters, we examined directly the effect of the mutation on voltage sensor movement by measuring gating currents from both wild-type and G803D channels. The ON- and OFF-gating currents of both channel types were measured at membrane voltages between −60 and +320mV with 1-ms voltage steps (Figure 5A). Gating charge displacement was determined by integrating the ON-gating current of each trace, and plots of charge displaced (Q) versus voltage are displayed in Figure 5B normalized to their maxima. As is evident the Q-V curves of the wild-type channel and the G803D mutant are superimposable, with half-activation voltages of 154mV (wild-type) and 158mV (G803D), and apparent gating charges per voltage senor of 0.52 (wild-type) and 0.50 (G803D). This result indicates that the G-to-D mutation in the putative RCK2 domain insignificantly affects voltage sensor movement in response to changes in voltage and therefore that the negative shift in the rSlo G-V (and P_O-V) relationship caused G803D at all [Ca²⁺]_i is not due to changes in voltage sensing. Furthermore, combined with the data in Fig. 2 and, discussed above, it strongly suggests that G803D alters predominately, if not exclusively, the intrinsic equilibrium constant for channel opening.

Display large version of this figure

Figure 5

Gating currents of the wild-type and G803D mutant channels. (A) Representative traces of gating currents recorded for the wild-type and G803D mutant channels in the absence of internal Ca²⁺ are shown. (B) Normalized averaged Q-V relations for the wild-type (average of 12, ○) and G803D (average of 6, ●). Each curve was fitted with the Boltzmann function. The fit parameters were as follows: for the wild-type, z=0.519±0.013e, V_1/2=154±1.8mV; for G803D, z=0.499±0.013e, V_1/2=158±1.5mV. Each data point represents the mean±SE.

To find out how other amino acid residues at position 803 affect the gating properties of the Slo channel, we replaced Gly-803 with various amino acids. As expected all mutant channels evoked K⁺ currents in response to membrane depolarization or increases in intracellular Ca²⁺, but there were large differences in expression level among the mutant channels (Figure 6A). Whereas some mutants, i.e., G803D and G803E, produced macroscopic currents comparable to wild-type, others, including G803P, evoked much smaller currents. Substitution of Gly-803 with hydrophobic residues, for example, G803F and G803L, did not generate enough K⁺ current for accurate and reproducible recordings. Moreover, different substitutions at G803 gave rise to markedly different effects on channel activation. In Figure 6B, the G-V relationships of the wild-type channel (open symbols) and several mutants (solid symbols) are compared. The effects of each mutation were quantified by examining V_1/2 values (Figure 6C). The mutations could be categorized into four subgroups. Three polar amino acids (Thr, Ser, and His) had no effect. Three others (Ala, Cys, and Glu) produced small negative G-V shifts of <40mV. And three others (Pro, Asp, and Lys) produced large G-V shifts. The substitution of Gly-803 with Pro or Asp negatively shifted the Slo G-V relationship by 119.3 and 80.9mV, respectively. In contrast, the substitution to Lys shifted the Slo G-V relation 30.0mV in the positive direction. In all channels tested, the slopes of the G-V curves were not altered significantly by the mutation. These results indicate that the voltage-dependent activation of the BK_Ca channel is highly sensitive to the residue at position 803, and that single substitutions at this position can modulate the range of channel activation by as much as 147mV without affecting the channel's voltage sensitivity.

Display large version of this figure

Figure 6

Functional effects of amino acid substitutions at Gly-803 in the putative RCK2. (A) Representative raw traces of the wild-type and four different mutant channels are shown. Ionic currents were induced by voltage steps ranging from −20 to 140mV in 40-mV increments from the holding voltage of −100mV. Each current trace represents an average of three successive records. Due to the marked differences in channel expression, the scale bar for each mutant channel was adjusted accordingly. (B) The G-V relationships of the wild-type (open symbols) and several mutant channels (solid symbols) are shown. The symbols for each mutant channel are indicated in the inset. Each channel was recorded with 2μM [Ca²⁺]_i. The curves were fitted with the Boltzmann function. (C) The V_1/2 of the wild-type and nine different mutant channels were determined at 2μM [Ca²⁺]_i. The V_1/2 of the wild-type is indicated as a dotted line. Each data point represents the mean±SE. Values differing from the wild-type by the paired Student's t-test at p<0.05 (*) or p<0.01 (**) are indicated.

We wondered which aspect of the R-group is responsible for the perturbation of the gating equilibrium so sensitively. We tried to correlate the functional effects of the mutations with the biochemical and biophysical properties of the R-groups 47,48,49. The zV_1/2 value of each mutant channel was plotted against the acid-ionization constant (pK_a), polarity, hydrophobicity, and volume of the R-group (Fig. 7). pK_a showed a statistically significant positive correlation with the G-V shifts of the mutant channels (correlation coefficient=0.7 (n=7)). The correlation coefficients (r) for the other comparisons however (−0.345 for polarity (n=10), 0.316 for hydrophobicity (n=10), and −0.092 for volume (n=10)) did not indicate significant correlation.

Display large version of this figure

Figure 7

Correlation analysis of the zV_1/2 values and biochemical and biophysical properties of the R-group at amino acid position 803. The zV_1/2 values of each mutant channel were plotted against the pK_a (A), polarity (B), hydrophobicity (C), and volume (D) of the R-group at position 803. Correlation coefficient (r) values for each property are shown in the inset.

In this study, we investigated the functional importance of amino acid residues located in the putative second RCK (RCK2) domain of the BK_Ca channel's α-subunit. We focused our attention on residues predicted by homology modeling to interact with RCK1 via the putative “flexible interface”. We did this to test for the presence of RCK2 and as well to perhaps gain experimental evidence in support of the arrangement of RCK1 and RCK2 in the homology model. Initially, we located four potential charge pairs across the interface and replaced the neutral residues on the putative RCK2 individually to either Lys or Asp, so that the mutated residues could interact electrostatically with countercharges on amino acids of RCK1. By examining the activation characteristics of the mutant channels, we initially found three mutations that altered the position of the G-V relation significantly. It was intriguing to observe that the mutations at different positions shifted the rSlo G-V relation in both the negative and positive directions, which suggests that they can stabilize either the open or the closed conformation. To examine whether mutations on the putative RCK2 domain alter the intrinsic equilibrium between open and closed, we performed further mechanistic studies on one such mutation, G803D.

This Gly residue (Gly-803 of rat Slo) is conserved, not only in the putative RCK2 of Slo homologs from Caenorhabditis elegans to human, but also in mammalian Slo paralogs such as Slick and Slo3 (Supplementary Fig. 1 , highlighted in blue). Based on the known structure of the MthK RCK domain and our secondary-structure prediction, Gly-803 is predicted to be located within the first loop between βA and αA of the N-terminal Rossmann-fold. The significance of this position in the gating ring of octameric RCK domains may be considered from many points of view. Gly-803 is predicted to be close to the Ca²⁺-binding sites of homologous RCK domains. When aligned with the amino acid sequence of RCK1, Gly-803 is only two residues away from one of the Asp residues (Asp-428 of rat Slo) known to affect the calcium sensitivity at low concentration of calcium 27 (Supplementary Fig. 1 , highlighted in yellow). In addition, the sequence alignment also predicts that the first loop of the MthK RCK domain, where Gly-803 is located in the putative RCK2, is physically close to the bound Ca²⁺ with the closest distance of only 6.2Å from polypeptide chain. 23. Since the relative movement of the flexible interface has been shown to be the main conformational change induced by Ca²⁺ binding in MthK 23, it is conceivable that the residues facing the flexible interface would be sensitive to any mutation. Due to its positioning at this strategically critical region of the RCK heterodimer, the chemical nature of Gly-803 may determine the equilibrium between the open and the closed conformations and thus determines the position of the G-V relationship of BK_Ca channel.

Based on the positive correlation between zV_1/2 and pK_a of R-group at position 803 (Figure 7A), some electrostatic influence at this position on the shift in the gating equilibrium can be expected. In fact, we probed the charge pair corresponding to mutagenized Asp at position 803 of the putative RCK2 and Lys-556 of RCK1 to understand the nature of the interaction. Using the Ala substitution at Lys-556 in the background of the G803D mutation, a double mutant cycle analysis gave a coupling energy of 3.0kJmol⁻¹ between Lys-556 of RCK1 and Asp-803 of RCK2 (data not shown). The magnitude of the coupling energy was rather weak, suggesting that the stabilization of the open conformation by the G803D mutation is more likely to be due to a weak through-space electrostatic attraction between the two residues rather than to a short-range salt bridge. It is also worth mentioning that no other countercharges other than Lys-556 were found in proximity to the amino acid position 803 in the structural model of the heterodimeric interface (Figure 1A). It is intriguing that Pro was such a stabilizing mutation for the open conformation and that it deviated the most from the correlation plots. It is conceivable that an amino-acid-like proline at this position adopts a local conformation that favors the open state. It remains to be seen whether the Gly-to-Pro mutation will also cause a shift in the intrinsic gating equilibrium.

In summary, we identified amino acid residues on the putative RCK2 domain of the BK_Ca channel's C-terminus that are important for channel activation. The most dramatic residue, Gly-803, modulated the channel's G-V relationship over a wide range of transmembrane voltages by altering the intrinsic gating equilibrium. These results are consistent with the structural model proposed for the putative RCK2 domain in which this residue is located on the flexible interface between RCK1 and RCK2. Thus, a conformational change in the gating ring, similar to that of bacterial Ca²⁺-activated K⁺ channels, may underlie the Ca²⁺-induced activation of the BK_Ca channel, and the putative RCK2 may be involved in the coupling between the Ca²⁺-bowl and transmembrane domains via RCK1.