Ion sensing in the RCK1 domain of BK channels

Abstract

BK-type K⁺ channels are activated by voltage and intracellular Ca²⁺, which is important in modulating muscle contraction, neural transmission, and circadian pacemaker output. Previous studies suggest that the cytosolic domain of BK channels contains two different Ca²⁺ binding sites, but the molecular composition of one of the sites is not completely known. Here we report, by systematic mutagenesis studies, the identification of E535 as part of this Ca²⁺ binding site. This site is specific for binding to Ca²⁺ but not Cd²⁺. Experimental results and molecular modeling based on the X-ray crystallographic structures of the BK channel cytosolic domain suggest that the binding of Ca²⁺ by the side chains of E535 and the previously identified D367 changes the conformation around the binding site and turns the side chain of M513 into a hydrophobic core, providing a basis to understand how Ca²⁺ binding at this site opens the activation gate of the channel that is remotely located in the membrane.

Large conductance Ca²⁺-activated K⁺ (BK) channels open in response to membrane depolarization and the elevation of intracellular Ca²⁺ concentration ([Ca²⁺]_i). In neurons and muscle cells, membrane depolarization activates voltage-dependent Ca²⁺ channels, resulting in Ca²⁺ entry into the cell and subsequent activation of BK channels. The K⁺ efflux through BK channels repolarizes the membrane, which shuts Ca²⁺ channels, thereby providing a negative feedback mechanism to modulate membrane excitability and [Ca²⁺]_i. Because of this function, BK channels are important modulators of muscle contraction (1), neuronal spike frequency adaptation (2), neurotransmitter release (3), and circadian pacemaker output (4). BK channels are formed by four Slo1 subunits (5, 6). Each Slo1 contains a membrane-spanning domain, which comprises the pore-gate domain (PGD) and the voltage sensing domain (VSD), and a cytosolic domain (CTD) (7, 8), which is made of two regulating domains for K⁺ conductance (RCK1 and RCK2) (9, 10). Intracellular Ca²⁺ binds to the CTD to activate the channel by enhancing the open probability of the activation gate located in the membrane-spanning PGD.

Previous studies have identified two putative Ca²⁺ binding sites in the CTD of BK channels, one is the Ca²⁺ bowl located in the RCK2 domain (10 –13) and the other is located in the RCK1 domain including the residue D367 (14). The existence of two distinctively different high-affinity Ca²⁺ binding sites that are responsible for Ca²⁺-dependent activation of BK channels has been demonstrated in various experimental studies (15). These studies demonstrated that Ca²⁺ binding to the two sites activates channel independently with only a small cooperativity (14, 16, 17), and the two sites show differences in various properties including affinities for Ca²⁺ (14, 17), voltage dependence (17), and the molecular mechanisms of coupling to the activation gate (18). The distinction between the properties of the two putative Ca²⁺ binding sites may lead to different physiological roles of these sites. For instance, a mutation in Slo1 that is associated with epilepsy and dyskinesia in human (19) specifically enhances the coupling of the RCK1 site to the activation gate to increase Ca²⁺ sensitivity of channel activation (18).

Although previous studies showed that the Ca²⁺ binding site in RCK1 is important for physiological functions, its molecular identity is less certain than that of the Ca²⁺ bowl. In the Ca²⁺ bowl, previous mutagenesis studies (13) and a recently published X-ray crystallographic structure of the Ca²⁺-bound Slo1 CTD demonstrate that a Ca²⁺ ion is coordinated by the side chains of D898 and D900 and the main chain carbonyls from Q892 and D895 (10). On the other hand, besides D367, no other residues have been identified to be part of the putative RCK1 Ca²⁺ binding site. Surprisingly, the same structure of CTD in high Ca²⁺ concentration did not identify any second Ca²⁺ binding site although the residue D367 is shown (10). To gain a better understanding of how the RCK1 Ca²⁺ binding site contributes to physiological functions and to solve the discrepancy between the structural data and the results from functional studies, further studies of this binding site are needed.

During the last 2 y, we have searched residues other than D367 that may be part of the putative Ca²⁺ binding site in RCK1 by systematic mutagenesis. These experiments show that the mutations of E535 in the RCK1 domain produce nearly identical functional consequences on the Ca²⁺-dependent activation as the mutations of D367. Therefore, both E535 and D367 may be part of the Ca²⁺ binding site in RCK1. We found that mutations of M513, some of which have been shown to reduce Ca²⁺ sensitivity (20), result in a different pattern of functional consequences than those of E535 and D367, which suggests that M513 may not be part of the Ca²⁺ binding site. We have also investigated Cd²⁺-dependent activation of BK channels and found that Ca²⁺ and Cd²⁺ interact with different sets of residues to activate BK channels, but some of the residues important for Cd²⁺-dependent activation may overlap with part of the putative Ca²⁺ binding site in RCK1. These results identify a cluster of residues that are important for BK channel function and further differentiate their roles in controlling channel gating. The Ca²⁺ binding site formed by D367 and E535 is consistent with the recently solved structures of BK channel CTD (9, 10) and provides a basis for understanding how the Ca²⁺ binding site couples to the activation gate during Ca²⁺-dependent activation.

Results and Discussion

Ca²⁺ ions bound to proteins are usually coordinated by oxygen-containing side chains, main chain carbonyl groups, and water molecules (21). To search for an oxygen-containing residue that may be part of the Ca²⁺ binding site in RCK1, we mutated most Asp, Glu, Asn, Gln, Ser, Thr, and Tyr residues in RCK1 to Ala individually (Fig. 1A), which are possibly located close to D367 in structural models of RCK1 based on the structure of the RCK domain of MthK (22). We examined the change in Ca²⁺ sensitivity of the channel due to these mutations, and Fig. 1B shows the results of the mutations in the C-terminal half of RCK1. The results of the mutations in the N-terminal AC region have been shown (18). In response to increases in [Ca²⁺]_i, the conductance-voltage (G-V) relation of BK channels shifts to more negative voltage ranges (23) (Fig. 1C). Because the effect of voltage on Ca²⁺-dependent activation is weak (17, 24, 25), this property has been used as an effective measure of Ca²⁺ sensitivity of BK channels in most studies of BK channel function (26). Similarly, here we define Ca²⁺ sensitivity as the G-V shift in response to the [Ca²⁺]_i change from 0 to the saturating 100 μM, ΔV _1/2 = V _1/2 at 0 [Ca²⁺]_i – V _1/2 at 100 μM [Ca²⁺]_i, V _1/2 is the voltage where G-V is half maximum. Mutations alter Ca²⁺ sensitivity and the change in Ca²⁺ sensitivity, ΔΔV _1/2 = ΔV _1/2 ^mut – ΔV _1/2 ^wt for all of the Ala scan mutations described above is shown (Fig. 1B). Similar to reported (14), D367A reduces more than half of the total Ca²⁺ sensitivity, with ΔΔV _1/2 = −112.8 ± 2.3 mV (Fig. 1B). Of all other mutations, E535A reduces Ca²⁺ sensitivity similarly as D367A (Fig. 1B) (18), suggesting that E535 may play an equivalent role as D367 in Ca²⁺ binding.

Fig. 1.

The E535A mutation reduces Ca²⁺ sensitivity of mSlo1 BK channel activation. (A) Sequence alignment of the RCK1 domain of mouse (GenBank accession number, GI: 347143) and fly (Drosophila, GI: 7301192) Slo1. The residues in bold from the mSlo1 were mutated to Ala as shown in B and ref. 18. Lines above the sequence indicate secondary structures. The numbers at right indicate the sequence number of the last amino acid. (B) Effect of mutations on Ca²⁺ sensitivity. ΔV _1/2 and ΔΔV _1/2 are defined in the text. D367A and E535A result in the largest reductions in Ca²⁺ sensitivity. (C) Macroscopic current traces from inside-out patches expressing WT and E535A mutant channels. Currents were elicited in 100 μM [Ca²⁺]_i by voltage pulses from −150 to 190 mV at 20-mV increment. The voltages before and after the pulses were −50 and −80 mV, respectively. G-V curves for WT and E535A mutation channels in [Ca²⁺]_i of nominal 0 (≈0.5 nM, open symbols) and 100 μM (filled symbols). The solid lines are fittings of the Boltzmann relation. (D) Effect of individual and combined mutations on Ca²⁺ activation. Error bars in this and other figures show the SE of mean (n = 6–15).

The E535A mutant channel retains part of Ca²⁺ sensitivity; the G-V relation of the channels shifts to negative voltages in response to an increase of [Ca²⁺]_i from 0 to 100 μM (Fig. 1 C and D). An additional mutation 5D5N, which substitutes the five consecutive Asp residues in the Ca²⁺ bowl with Asn, nearly eliminates the remaining Ca²⁺ sensitivity of E535A (Fig. 1D), indicating that E535A reduces Ca²⁺ sensitivity specifically derived from the Ca²⁺ binding site in RCK1. Consistent with this result, a double mutation D367A/E535A reduces Ca²⁺ sensitivity similarly as either D367A or E535A (Fig. 1D), indicating that E535A, similar to D367A, destroys Ca²⁺ sensitivity derived from the RCK1 site.

Previous studies measured Ca²⁺-dependent activation derived from either the RCK1 site or the Ca²⁺ bowl by mutating the other site and found that the two sites show different affinities for Ca²⁺ (14, 17). To further examine whether E535A destroys Ca²⁺ sensitivity derived from the RCK1 site, we measured G-V relations of the mutant channel at various [Ca²⁺]_i from 0 to 100 μM (Fig. 2A). Comparing the pattern of G-V shifts at various [Ca²⁺]_i among mutants D367A, E535A, and 5D5N, it is apparent that the E535A channels behave more similarly as D367A than 5D5N, with a larger reduction of Ca²⁺ sensitivity and less even distribution of G-V relations along the voltage axis (Fig. 2 A and B). Fitting the G-V relations of each mutant channel with a voltage-dependent Monod–Wyman–Changeux (MWC) model (27), we obtained apparent affinity of Ca²⁺ binding sites (Fig. 2C). These results indicate that 5D5N destroys the Ca²⁺ bowl that has a higher affinity for Ca²⁺, leaving the RCK1 site intact with a lower affinity for Ca²⁺. However, both E535A and D367A destroy Ca²⁺ sensitivity derived from the RCK1 site that has a lower Ca²⁺ affinity.

Fig. 2.

Ca²⁺ dependence of the E535A mutant channels is similar to that of D367A. (A) Normalized G-V relations of the D367A, E535A, and 5D5N mutant channels in [Ca²⁺]_i from nominal 0 (≈0.5 nM) (•), 1 μM (▼), 2 μM (), 5 μM (◆), 10 μM (▲), 30 μM (◆), and 100 μM (■). The solid lines are fittings with the MWC model. (B) V _1/2 versus [Ca²⁺]_i for the D367A, E535A, and 5D5N mutant channels, showing a similar steepness between E535A and D367A. (C) Parameters of the MWC model from fitting the G-V relations of D367A, E535A, and 5D5N.

Fig. 2 also shows that E535A and D367A affect BK channel gating with different details. First, in 0 [Ca²⁺]_i, the G-V relation of D367A is shifted to less-positive voltage ranges compared with that of the WT mSlo1 or E535A (Fig. 2 A and B). Second, in the intermediate [Ca²⁺]_i, the G-V relations of D367A show a steeper slope, which makes these G-V curves appear more shifted away from that in 0 [Ca²⁺]_i as compared with those of E535A (Fig. 2A). All these differences between the two mutant channels suggest that, in addition to affecting Ca²⁺-dependent activation, D367A also alters voltage-dependent activation of BK channels. This result is consistent with the location of D367 that is close to the VSD of the channel in the membrane (9, 10). Previous results have demonstrated that the cytosolic domain interacts with the VSD during BK channel activation (28, 29), and such interactions may result in a weak voltage dependence of Ca²⁺ sensitivity derived from the RCK1 site (17). Thus, it is possible that the mutation D367A also alters function of the VSD, resulting in changes of voltage-dependent activation.

Because in a systematic Ala scan of oxygen-containing residues only the mutations of D367 and E535 nearly eliminate Ca²⁺ sensitivity in addition to the Ca²⁺ bowl mutation 5D5N (Fig. 1), D367 and E535 are likely to be part of the Ca²⁺ binding site in RCK1 of BK channels where the side chain of the acidic residues may coordinate Ca²⁺. To further examine whether these residues have the properties of a Ca²⁺ coordinator, we measured Ca²⁺ sensitivity of the mutations of D367 to various amino acids and found that D367E retains a Ca²⁺ sensitivity that is significantly greater than that retained by D367A (Fig. 3A; see Fig. S1A for current traces). This result is consistent with the idea that the oxygen atoms from the side-chain carboxylate groups of D367 or E367 can coordinate Ca²⁺, but because of the difference in the size of side chains, D367E may alter the structure of the binding site and, thus, impair Ca²⁺ sensitivity. Likewise, E535D also retains more Ca²⁺ sensitivity than E535A (Fig. 3B and Fig. S1B), indicating that E535 has similar properties of a Ca²⁺ coordinator.

Fig. 3.

Properties of E535 and D367 but not M513 are consistent with calcium binding coordinators. Left are G–V relations of D367E (A), E535D (B), and M513D (C) in [Ca²⁺]_i from nominal 0 (≈0.5 nM) (•), 1 μM (▼), 2 μM (), 5 μM (◆), 10 μM (▲), 30 μM (◆), and 100 μM (■), fitted with the Boltzmann relation (filled lines), and Right are Ca²⁺ sensitivity of D367 (A), E535 (B), and M513 (C) mutated to various amino acids. ΔV _1/2 is defined in the text. *, Ca²⁺ sensitivity of the mutated channel is significantly larger than that of the mutation to Ala (P < 0.005).

A previous study showed that the mutation M513I in the RCK1 domain reduces Ca²⁺ sensitivity (20) (Fig. 3C and Fig. S1C). The double mutation M513I/D367A does not cause any reduction of Ca²⁺ sensitivity in addition to D367A, but M513I/5D5N reduces Ca²⁺ sensitivity more than 5D5N alone (Fig. 1D), indicating that M535I affects the Ca²⁺ sensitivity derived from the RCK1 site. Likewise, M513I/E535A does not cause any reduction of Ca²⁺ sensitivity in addition to E535A (Fig. 1D), further supporting that E535 is part of the Ca²⁺ binding site in RCK1. Ca²⁺ prefers to bind to hard oxygen-containing ligands; the soft sulfur in the side chain of Met residues usually is not found to coordinate Ca²⁺ in other Ca²⁺ binding proteins (21, 30). If M513 coordinates Ca²⁺ in BK channels, a change of M513 to an oxygen-containing residue is expected to retain or even enhance Ca²⁺ sensitivity. We mutated M513 to Ile, Cys, Asp, and Glu residues and found that M513D reduces Ca²⁺ sensitivity more than any other mutations, whereas M513C reduces Ca²⁺ sensitivity the least (Fig. 3C). These results are in contrast to the profile of mutational results on D367 or E535, indicating that M513 does not have the properties of a Ca²⁺ coordinator. Therefore, the mutations of M513 may reduce Ca²⁺ sensitivity of BK channel by altering the structure of the Ca²⁺ binding site in RCK1 (see below).

Besides Ca²⁺, other divalent cations, including Cd²⁺, also activate BK channels (31). Previous studies suggested that Cd²⁺ might interact with the Ca²⁺ binding site in RCK1 but not the Ca²⁺ bowl because the double mutation D362A/D367A nearly abolished Cd²⁺ sensitivity (32) (Fig. 4; see Fig. S2 for current traces), whereas mutations of the Ca²⁺ bowl had no effect on Cd²⁺ sensitivity (11, 32). To examine whether the same residues are important for both Ca²⁺- and Cd²⁺-dependent activation, we examined the effect of mutations D362A, D367A, M513I, and E535A on Cd²⁺-dependent activation. The results show that the effect of these mutations on Cd²⁺ sensitivity is not correlated with the effect on Ca²⁺-dependent activation (Fig. 4 A and B). First, none of the individual mutations abolishes Cd²⁺ sensitivity, although D367A and E535A nearly abolish the Ca²⁺ sensitivity derived from the RCK1 site (Fig. 1). Second, D362A reduces Ca²⁺ sensitivity by the smallest fraction among all individual mutations but reduces Cd²⁺ sensitivity by the largest fraction. On the other hand, M513I and E535A have large effects on Ca²⁺ sensitivity but small effects on Cd²⁺ sensitivity. Thus, only D367 is important for both Ca²⁺- and Cd²⁺-dependent activations, whereas other residues are important for either Ca²⁺- or Cd²⁺-dependent activation. Therefore, Cd²⁺ does not seem to bind to the same site as Ca²⁺. Additionally, because D362A and D367A both reduce approximately half of Cd²⁺ sensitivity (ΔV _1/2 ^mut − ΔV _1/2 ^wt = −35.5 ± 5.1 mV and −32.8 ± 8.1 mV, respectively; Fig. 4B) and the effects of the two individual mutations add up to that of the double mutation D362A/D367A (ΔV _1/2 ^mut − ΔV _1/2 ^wt = −84.6 ± 6.6 mV), D362 and D367 seem to affect Cd²⁺ sensitivity independently and neither seem to be necessary for Cd²⁺ binding.

Fig. 4.

The effects of mutations on Cd²⁺ sensitivity are not correlated with that on Ca²⁺ sensitivity. (A) G-V relations in 0 (open circles) and 100 μM (filled circles) [Ca²⁺]_i, and 0 (open triangles) and 100 (filled triangles) [Cd²⁺]_i, fitted with the Boltzmann relation (filled lines). (B) Effects of Ca²⁺ (filled bars) and Cd²⁺ (open bars) on the WT and mutant channels. (Left) ΔV _1/2 = V _1/2 at 0 [Ca²⁺]_i (or [Cd²⁺]_i) − V _1/2 at 100 μM [Ca²⁺]_i (or [Cd²⁺]_i). (Right) ΔΔV _1/2 = ΔV _1/2 of mutations − ΔV _1/2 of WT.

Recently, two crystal structures of the gating ring from the human BK channel were experimentally determined, representing two different conformations of the gating ring: One is the Ca²⁺-bound conformation crystallized in the presence of 50 mM [Ca²⁺]_i (PDB ID code: 3MT5) (10), and the other is the Ca²⁺-free structure (PDB ID code: 3NAF) (9) (Fig. 5A). However, interestingly, only one Ca²⁺ is present at the Ca²⁺ bowl in the RCK2 domain of the Ca²⁺-bound crystal structure, whereas no Ca²⁺ is observed at the putative D367/E535 site. Based on the Ca²⁺-free structure, it was predicted that D367 and E535 could form a Ca²⁺ binding site (9). However, the structure is in the Ca²⁺-free conformation, and the side chains of D367 and E535 in the structure are not close enough to coordinate a Ca²⁺ ion. Comparisons between the RCK1 domains of the Ca²⁺-bound and Ca²⁺-free structures show that, although the overall architectures of the two conformations are similar, the conformations differ significantly around the D367/E535 binding site. This flexible region includes three segments: the loop between αA and βB, the loop between αG and αH, and the linker between αH and βG (Fig. 5A). The conformational changes suggest flexibility around the D367/E535 binding site, which binds and releases Ca²⁺ during the functional cycle of the BK channel. In other words, in addition to the two conformations shown by the crystal structures, there should be a third conformation of RCK1 that allows binding of Ca²⁺ at the D367/E535 site. Here, we model the conformation of RCK1 bound with Ca²⁺ near D367/E535 based on the Ca²⁺-free crystal structure (9), in which the side chain of the critical residue D367 points outward to the solvent and, thus, better resemble the Ca²⁺-bound conformation because Ca²⁺ ions stay in the solvent before binding.

Fig. 5.

Molecular modeling for the mechanism of Ca²⁺-dependent activation of BK channels. (A) Comparison of the RCK1 domains between the Ca²⁺-bound crystal structure (PDB ID code: 3MT5, red) and Ca²⁺-free crystal structure (PDB ID code: 3NAF, cyan). Three flexible segments (green) show significant conformational discrepancy around the residue D367 (in stick mode). (B) Two Ca²⁺ ions (yellow spheres) bound on the modeled RCK1 (blue) and RCK2 (magenta) domains of the BK channel. The Ca²⁺ on RCK1 is coordinated by residues D367, R514, and E535 (Inset). (C) Comparison of the Ca²⁺ binding sites in RCK1 between the model (blue) and two crystal structures (PDB ID codes: red: 3MT5; cyan: 3NAF), in which the important residue M513 is shown in stick mode. (D) Details of the hydrophobic core around M513. The bound Ca²⁺ ion is shown for reference. Note: Y527 does not belong to the hydrophobic core.

Fig. 5B shows two Ca²⁺ ions binding to the modeled RCK1-RCK2 structure of the BK channel. It can be seen that the Ca²⁺ ion that binds at the D367/E535 site is coordinated by one main-chain carbonyl oxygen atom of R514 and four oxygen atoms from the side-chain carboxylate groups of D367 and E535, which is consistent with the previous study (14) and our experimental results (Figs. 1–3) (also see SI Results and Discussion and Fig. S3). Because only the main-chain carbonyl oxygen atom of R514 coordinates the binding of the Ca²⁺ ion, it is expected that mutants of this residue would not have much effect on atomic coordination for Ca²⁺ and, therefore, the Ca²⁺ sensitivity of the BK channel.

Comparisons between our modeled RCK1 conformation and the two crystal structures (9, 10) show that our RCK1 model experiences larger conformational changes around the D367/E535 binding site and better coordinates the Ca²⁺ ion (Fig. 5C). In addition to the directly interacting residues D367, R514, and E535 of Ca²⁺, another notable residue involved in the conformational change is M513. In the crystal structures, the side chain of M513 points toward the Ca²⁺ binding pocket. However, in our modeled structure, the side chain of M513 adopts a completely different orientation and points toward the protein due to the conformational change induced by Ca²⁺ binding in the loop where D367 locates (Fig. 5C). Further examinations on the surrounding environment of M513 in our model reveal that M513 would contribute to Ca²⁺ binding in the following way: The hydrophobic side chain of M513 points toward the interface between RCK1 and RCK2, forming a hydrophobic cluster core with the residues A508, F511, Y528, and V532 from RCK1 and residues Y904 and F909 from RCK2 (Fig. 5D). This hydrophobic core would be critically important for stabilizing the interface between RCK1 and RCK2 and the subdomain from αG to βG that contains the coordinating residues R514 and E535 of Ca²⁺. In other words, to allow binding of Ca²⁺, it would be necessary for the side chain of M513 to face the hydrophobic core at the interface between RCK1 and RCK2 so as to maintain the local structural integrity of the Ca²⁺ binding site, as shown in our modeled dimeric structure (Fig. 5D). Although both Met and Ile are hydrophobic residues, the side chain of Ile is shorter and bulkier than that of Met, which may affect the crowded hydrophobic core that contains multiple aromatic residues. On the other hand, the hydrophobic Cys residue better resembles Met despite its shorter length than Met. Therefore, the Ca²⁺ binding site would be affected more by charged mutations M513D, M513E, and M513I than M513C (Fig. 3). Fig. 1B shows that Y528, one of the critical residues in the above hydrophobic core (Fig. 5D), is also important to Ca²⁺ binding because Y528A significantly reduces Ca²⁺ sensitivity of the BK channel. In contrast, the neighboring residue Y527, which points to the solvent and is not part of the hydrophobic core (Fig. 5D), has little effect on the Ca²⁺ sensitivity when it is mutated to Ala (Fig. 1B).

Previous studies indicated that E374 and E399 in the cytosolic RCK1 form a Mg²⁺ binding site with D99 and N172 in the membrane-spanning VSD in BK channels (29). The Mg²⁺ ion bound to the interface between the VSD and RCK1 activates the BK channel by an electrostatic repulsion to the S4 segment (28). These results suggest that RCK1 is located close to the VSD and the two domains interact intimately, which is corroborated by an electron cryomicroscopy (cryo-EM) structure of BK channels (33). D367 is situated in a loop between αA and βB in RCK1 (Fig. 5A) that is located close to the membrane. It is possible that the conformation of the D367/E535 Ca²⁺ binding site is affected by the membrane-spanning domain of the BK channel. Consistent with this idea, it has been shown that mutations in the intracellular loop between the transmembrane segment S0 and S1 alter Ca²⁺ sensitivity of channel activation (34). Because the Ca²⁺-bound crystal structure of the BK channel CTD was solved in the absence of the membrane-spanning domain (10), it is possible that the absence of a Ca²⁺ at the D367/E535 site and the misorientation of M513 side chain are due to the lack of the influence of the membrane-spanning domain on the conformation of the site.

Our recent study indicated that the N-terminal half of RCK1 from βA to αC (the AC region; Fig. 1A) is important in mediating Ca²⁺ binding to the site in RCK1 to the opening of the activation gate located in the membrane-spanning domain (18). Such an allosteric coupling is also affected by changes in intracellular viscosity, suggesting that molecular dynamics of the CTD are important in the coupling to activate the BK channel (18). This mechanism is consistent with the formation of the Ca²⁺ binding site by D367 and E535, which come from the AC region and the peripheral domain in RCK1, respectively, and bridge across the flexible interface between RCK1 and RCK2 that is formed by αF and αG from RCK1 and αS and αT from RCK2 (10) (Fig. 5B). Thus, the binding of Ca²⁺ to this site may pull and stabilize the AC region and turn the M513 side chain into the hydrophobic core formed at one side of the flexible interface, which may affect the AC region at the other side of the flexible interface (Fig. 5 B and D), thereby opening the channel via the AC region.

Materials and Methods

All mutations were generated from the mbr5 splice variant of mouse Slo1 (mSlo1) (8) by using overlap-extension PCR (35) and verified by sequencing. Xenopus laevis oocytes were injected with 0.05–20 ng of cRNA/oocyte, and currents were recorded in 2–4 d.

Macroscopic currents were recorded from inside-out patches. The data acquisition and analyses, solutions for Ca²⁺-dependent activation, and model fitting are the same as described (36) and can be found in SI Materials and Methods. In experiments on Cd²⁺-dependent activation, the internal solution contained 150 mM KF, 20 mM Hepes, and 2 mM MgCl₂ with either 0 or 100 μM CdCl₂ as the 0 or 100 μM [Cd²⁺]_i solution. CaF₂ precipitates from the solution so that the solution contains no free Ca²⁺ (31, 32).

We modeled the conformation of Ca²⁺-bound RCK1 based on 3NAF (9) as follows. First, the RCK1 domain of 3NAF was separated from the rest crystal structure. Then, the three flexible segments around the D367 binding site were remodeled by sampling possible conformations using the LOOPY modeling program (37). To further construct a model for Ca²⁺-bound RCK1–RCK2 complex, the modeled Ca²⁺-bound RCK1 and the experimental Ca²⁺-bound RCK2 of 3MT5 (10) were docked together by using our protein-protein docking program MDockPP (38). The docked RCK1–RCK2 complex was further optimized/minimized by using Amber force fields (39). Thus, we obtained a modeled structure of two RCK domains of the BK channel, which contains two bound Ca²⁺ ions: one is at the Ca²⁺ bowl on RCK2, and the other is at the D367 site on RCK1. Specific details of the modeling can be found in SI Materials and Methods.