Metal-driven Operation of the Human Large-conductance Voltage- and Ca2+-dependent Potassium Channel (BK) Gating Ring Apparatus*♦

Large-conductance voltage- and Ca2+-dependent K+ (BK, also known as MaxiK) channels are homo-tetrameric proteins with a broad expression pattern that potently regulate cellular excitability and Ca2+ homeostasis. Their activation results from the complex synergy between the transmembrane voltage sensors and a large (>300 kDa) C-terminal, cytoplasmic complex (the “gating ring”), which confers sensitivity to intracellular Ca2+ and other ligands. However, the molecular and biophysical operation of the gating ring remains unclear. We have used spectroscopic and particle-scale optical approaches to probe the metal-sensing properties of the human BK gating ring under physiologically relevant conditions. This functional molecular sensor undergoes Ca2+- and Mg2+-dependent conformational changes at physiologically relevant concentrations, detected by time-resolved and steady-state fluorescence spectroscopy. The lack of detectable Ba2+-evoked structural changes defined the metal selectivity of the gating ring. Neutralization of a high-affinity Ca2+-binding site (the “calcium bowl”) reduced the Ca2+ and abolished the Mg2+ dependence of structural rearrangements. In congruence with electrophysiological investigations, these findings provide biochemical evidence that the gating ring possesses an additional high-affinity Ca2+-binding site and that Mg2+ can bind to the calcium bowl with less affinity than Ca2+. Dynamic light scattering analysis revealed a reversible Ca2+-dependent decrease of the hydrodynamic radius of the gating ring, consistent with a more compact overall shape. These structural changes, resolved under physiologically relevant conditions, likely represent the molecular transitions that initiate the ligand-induced activation of the human BK channel.

The large-conductance voltage-and Ca 2ϩ -activated K ϩ channel (BK, 4 MaxiK, Slo1) is an important regulator of cellular function including cellular excitability, neurotransmitter release, vascular tone, and hair cell tuning (1)(2)(3)(4)(5)(6)(7)(8). Its K ϩ conductance of ϳ250 picosiemens is an order of magnitude larger than that observed in typical voltage-gated K ϩ -selective channels (9), making the BK channel a powerful regulator of the cell membrane potential. A functional BK channel is composed of four identical ␣ subunits, each consisting of a short extracellular N-terminal tail, seven transmembrane segments (S0 -S6) (10,11), and a large intracellular C-terminal domain. The four ␣ subunits likely assemble around a central symmetry axis, forming the K ϩ -selective pore, surrounded by peripheral transmembrane voltage-sensing domains (12,13) that undergo conformational rearrangements upon membrane depolarization (14 -16) to facilitate pore opening (17)(18)(19)(20). Another pathway to BK channel activation is mediated by intracellular Ca 2ϩ . In the CNS, BK channels localize within Ca 2ϩ nanodomains generated at the immediate proximity of voltage-gated Ca 2ϩ channels as an efficient strategy to regulate BK channel activity (reviewed by Ref. 21). Intracellular Ca 2ϩ sensitivity is conferred by a large "gating ring" tetrameric superstructure, which is composed of two tandem C-terminal regulators of K ϩ conductance (RCK1 and RCK2) domains from each of the four channel ␣ subunits (22)(23)(24). The gating ring apparatus regulates BK channel activation by micromolar Ca 2ϩ and millimolar Mg 2ϩ (4,7,8,25) as it is also expected to undergo ligand-driven conformational rearrangements that in turn favor channel opening (26 -28). A region containing five consecutive aspartates (Asp 894 -Asp 898 ) termed the "calcium bowl" (29) constitutes a high-affinity Ca 2ϩ -binding site (24, 30 -33) located within the RCK2 domain (23,24,33). Consistent with evidence that the neutralization of the calcium bowl region reduces, but does not abolish, Ca 2ϩ sensitivity, a second high-affinity Ca 2ϩ -sensing region is thought to exist within the RCK1 domain, where residues aspartate 367, methionine 513, arginine 514, and glutamate 535 have a critical role in the Ca 2ϩ sensitivity of the channel (34 -38). The human BK channel mutation D434G (in RCK1), thought to pathologically enhance the Ca 2ϩ -sensing mechanism of the channel (39), has been linked to generalized epilepsy and paroxysmal dyskinesia (40). Low-affinity Mg 2ϩ sensing depends on coordinating residues within RCK1 (glutamines 374/399) (34,41) and the transmembrane voltage sensor domain (aspartate 99, asparagine 172) (4,7,42,43). The gating ring is thought to be the site of action of different cytosolic interacting partners (2,44) as well as small signaling molecules such as reactive oxygen species, CO, hemin, and H ϩ (4,5,45,46). BK channels have been visualized at 17-20 Å resolution (22) and, more recently, atomic structures (3 Å resolution) of the human BK gating ring have been reported (23,24). Thus, although the BK gating ring has a well defined structural identity, the functional properties linked to its ligand-sensing operation have not yet been investigated. We have probed the structural and functional properties of the recombinant human BK channel C terminus under physiologically relevant conditions, providing evidence that RCK1-RCK2 complexes assemble into a tetrameric gating ring structure that undergoes metal-dependent conformational rearrangements. The resolved molecular events are likely associated with the ligand-induced signal propagation that activates the channel pore. The gating ring superstructure appears to possess the structural and functional properties of a chemo-mechanical coupler able to transduce the free energy of ligand binding into mechanical work.

EXPERIMENTAL PROCEDURES
Expression and Purification-Expression and mutagenesis corresponding to the wild-type and calcium bowl-neutralized (D894A,D895A,D896A,D897A,D898A) human BK C-terminal domain ( 322 IIE⅐⅐⅐ALK 1005 ) were performed as described previously (33,47). Briefly, protein fractions were solubilized in 20 mM Tris-HCl and 8 M urea (pH 8.0). The supernatant obtained after centrifugation was applied to a nickelnitrilotriacetic acid affinity column, and the protein fractions were eluted with (in mM) 250 imidazole and dialyzed against (in mM) 25 MOPS, 2 EGTA, 120 KCl, pH 7.2. The purity of the expressed proteins was analyzed using a 12.5% SDS-PAGE. Protein concentrations were determined using the Biuret-Lowry assay.
Mass Spectrometry (MS)-Enzyme digestion was performed as follows. Purified human Slo1 was separated by SDS-PAGE and in-gel digested prior to analysis by mass spectrometry. Gel plugs were dehydrated in acetonitrile and dried completely in a SpeedVac. Samples were reduced and alkylated with 10 mM dithiothreitol and 10 mM tris(2-carboxyethyl)phosphine solution in 50 mM NH 4 HCO 3 (30 min at 56°C) and 100 mM iodoacetamide (45 min in dark), respectively. Gel plugs were washed with 50 mM NH 4 HCO 3 , dehydrated with acetonitrile, and dried down in a SpeedVac. Gel pieces were then swollen in digestion buffer containing 50 mM NH 4 HCO 3 and 20.0 ng/l chymotrypsin (25°C, overnight).
MS Analyses and Database Searching-10 l of extracted peptides were analyzed by nanoflow LC/MS/MS on a Thermo Orbitrap with a dedicated Eksigent nanopump using a reversed phase column (75-m inner diameter, 10-cm BioBasic C18, 5-m particle size, New Objective). Spectra were acquired in data-dependent mode with dynamic exclusion where the instrument selects for fragmentation the top six most abundant ions in the parent spectra. Data were searched against an Escherichia coli database inserted with the human Slo1 BK channel amino acid sequence using the SEQUEST algorithm in the Bio-Works software program version 3.3.1 SP1. All spectra used for identification had deltaCN Ͼ 0.1 and met the following Xcorr criteria: Ͼ2 (ϩ1), Ͼ3 (ϩ2), Ͼ4 (ϩ3), and Ͼ5 (ϩ4). Mass tolerance was 2 Da for precursor and 1 Da for product ions.
Circular Dichroism (CD) Spectroscopy-CD spectra were obtained as described previously (33,47). Briefly, far-UV spectra of purified gating ring in 20 mM phosphate (pH 7.2) were recorded as an average of 7-9 scans. Secondary structure fractions were calculated using the CONTINLL, SELCON3, and CDSSTR algorithms with the SMP56 reference set of the CDPro software package (48). The numbers of ␣-helices and ␤-strands were calculated by dividing the numbers of residues included in the distorted ␣-helix and ␤-structure by a factor of four and two, respectively.
Size-exclusion Chromatography-Size-exclusion chromatography experiments were performed as described previously (47). Purified BK gating ring in (in mM) 50 Tris-HCl, 2 ␤-mercaptoethanol, and 2 EGTA (pH 8.4) or (in mM) 20 MOPS (pH 7.2) was loaded onto a Superdex 200 10/300 column equilibrated with the same buffer with a flow rate of 0.5 ml/min. Calibration was performed using protein standards (Sigma).
The instrument response function was acquired using a nonfluorescent, light scattering sample (Ludox, colloidal silica). The instrument response function is a function of the detector characteristics and timing electronics and demonstrates the speed of the instrument and its detection properties. It is the shortest time profile that can be recorded by the instrument (49).
The fluorescence intensity decay data were fit to a sum of three exponential functions, using the DAS6 v6.4 software (Horiba Jobin Yvon) where I is the fluorescence intensity and ␣ and are the normalized pre-exponential factor and decay time constant, respectively. The average fluorescence lifetimes ( avg ) for threeexponential iterative fittings are calculated from the decay times and pre-exponential factors using the equation where f i is the fractional contribution of each decay time to the steady-state intensity, which is given by The goodness of the fit was determined from its 2 value and the variance of the weighted residual distribution (n ϭ 3, unless otherwise noted).
Steady-state Fluorescence Spectroscopy-Steady-state Trp fluorescence emission spectra of the gating ring in (in mM) 25 MOPS, 120 KCl, 2 EGTA, pH 7.2, were acquired in the 310 -400 nm range, ex ϭ 295 nm (5 nm excitation/emission slit width). Normalized fluorescence data were fit with a Hill function in the form where F is the fluorescence intensity at 340 or 350 nm for WT and calcium bowl mutant gating ring, respectively; [M 2ϩ ] is the concentration of Ca 2ϩ or Mg 2ϩ ; K1 ⁄ 2 is the apparent dissociation constant; and n is the Hill coefficient.
Dynamic Light Scattering (DLS)-DLS measurements were recorded using a NICOMP 380 submicron particle sizing system. The hydrodynamic properties of gating ring particles were determined by measuring the translational diffusion coefficient D T , which is related to the frictional coefficient f by the Einstein-Sutherland equation where K B is the Boltzmann constant and T is the temperature in K. The frictional coefficient of a spherical particle, f sph , is a function of the fluid viscosity, , and the radius of the particle, r sph . It is defined by the Stokes law.
f sph ϭ 6r sph (Eq. 6) Mean particle size distribution was analyzed in volumeweighted Nicomp-fit mode. The theoretical hydrodynamic radius (R theo H ) was calculated from the formula where M is the molecular mass of the gating ring (312,000), V s is the particle specific volume (0.73 cm 3 g Ϫ1 ), h is the hydration (0.35 g of H 2 O (g-protein) Ϫ1 ), and N A is Avogadro's constant (50). The Perrin, or shape factor, is the ratio of the measured fric- The polydispersity (P d ) was evaluated according to the following equation where is the standard deviation of the distribution and D H is the mean hydrodynamic diameter from DLS measurements.

Structural Organization of the BK Gating Ring in Solution-
The intracellular C-terminal domain (CTD, 684 amino acids) of the BK channel, which accounts for Ͼ60% of the whole channel and includes the two ligand-sensing modules RCK1 and RCK2 (supplemental Fig. S1), was expressed and purified. The CTD electrophoretic mobility in SDS-PAGE corresponds to an estimated M r ϭ 72,000 (Fig. 1B), close to the theoretical molecular weight of the purified protein (78,000). The identity of the purified BK CTD was confirmed by mass spectrometry, which identified 67% of the primary sequence (supplemental Fig. S1). The secondary and quaternary structures of the protein were investigated under physiologically relevant ionic conditions similar to the intracellular environment to ascertain its proper folding and native assembly, information necessary for the validity of the results of subsequent functional assays.
CD spectra were acquired to probe the secondary structure composition of the CTD in solution (Fig. 1C) and were subsequently analyzed using algorithms of the CDPro software (48). The experimental (black) and calculated (red, CONTINLL algorithm) spectra are shown superimposed. The CD spectral analysis revealed a CTD secondary structure composition of ϳ29% ␣-helix, ϳ22% ␤-strand, organized in 22 ␣-helices and 30 ␤-strands, and ϳ49% turn and unordered structure (supplemental Fig. S2A). Although a direct comparison of secondary structure composition across a crystal structure study is not straightforward because of differences in the regions purified and/or resolved, reasonable correlation exists (supplemental Fig. S2).
The oligomeric state of the BK CTD was assessed by sizeexclusion chromatography. A characteristic column elution and corresponding calibration curve is shown in Fig. 2, A and B, respectively. The BK CTD eluted with an M r ϭ 324,000 Ϯ 4,000, which is approximately four times the estimated molecular weight of the monomer (78,000), suggesting that in solution, the BK CTDs (RCK1-RCK2 complexes without transmembrane structures) self-assemble into tetramers, likely corresponding to the BK gating ring apparatus.
Ca 2ϩ and Mg 2ϩ Induce Conformational Changes in the BK Gating Ring-As a chemo-mechanical signal transducer, the gating ring is expected to regulate BK channel open probability by converting the free energy of ligand binding into conformational rearrangements that facilitate channel opening (5,25,26,28,(51)(52)(53).
The purified gating ring includes five Trp residues per subunit that serve as intrinsic environment-sensitive fluorophores reporting ligand-induced structural changes (49). We probed the gating ring for metal-induced conformational rearrangements by measuring the fluorescence lifetime of gating ring Trp residues ( ex ϭ 296 nm; em ϭ 340 nm) in solutions with different [Ca 2ϩ ] or [Mg 2ϩ ], using time-correlated single-photon counting (49). Fig. 3 shows superimposed data obtained at four different concentrations of Ca 2ϩ (Fig. 3A) or Mg 2ϩ (Fig. 3B), represented as single photon detection events binned by the time interval of detection after the excitation pulse; shorter Trp fluorescence lifetimes will result in fewer photons detected at long time intervals, accelerating the decay of the photon histogram. Each photon histogram was fit by the sum of three exponentials (black curves) to determine the mean fluorescence lifetime ( avg ) of the native gating ring Trp residues, under different [Ca 2ϩ ] or [Mg 2ϩ ].
The average fluorescence lifetime ( avg ) of the gating ring in nominal zero Ca 2ϩ (0.00058 M) was 2.6 Ϯ 0.010 ns. Increasing free [Ca 2ϩ ] accelerated the decay of fluorescence intensity, as shown by the progressive reduction of avg to a limiting value of 1.6 Ϯ 0.0027 ns at 35 M free Ca 2ϩ (Fig. 3A and supplemental Table S1), indicating a change within the immediate environment of one or more Trp residues.
We found that the Trp fluorescence lifetime of the gating ring was also sensitive to Mg 2ϩ (Fig. 3B) at concentrations in the millimolar range. As [Mg 2ϩ ] was incrementally raised from 0 to 12 mM, avg progressively decreased to the same limiting value obtained for Ca 2ϩ (1.6 Ϯ 0.015 ns; supplemental Tables S1 and S2). Thus, the binding of both Ca 2ϩ and Mg 2ϩ to the gating ring induced structural transitions at physiologically relevant concentrations, revealing the divalent metal cation-sensing properties of the isolated gating ring apparatus in solution.
The Structural Rearrangements of the Gating Ring Are Ion-specific-The following experiments were designed to estimate the apparent binding constants and assess the specificity of the BK gating ring for divalent cations, based on steady-state fluorescence spectroscopy of intrinsic Trp. The intensity of the fluorescence emission spectra of the gating ring ( ex ϭ 295 nm) was progressively reduced upon free [Ca 2ϩ ] elevation in the buffer solution up to 35 M, as shown in Fig. 4A. The [Ca 2ϩ ]dependent quenching of the fluorescence intensity at 340 nm was well described by the linear combination of two Hill functions with apparent binding constants K11 ⁄ 2 ϭ 0.29 Ϯ 0.0043 M (n 1 ϭ 3.4 Ϯ 0.16) and K21 ⁄ 2 ϭ 3.5 Ϯ 0.51 M (n 2 ϭ 2.1 Ϯ 0.35) (Fig. 4D), suggesting at least two Ca 2ϩ -binding sites and positive cooperativity. This apparent Ca 2ϩ affinity is relevant to BK channel activation (4,54). K11 ⁄ 2 likely relates to Ca 2ϩ binding to the RCK2 domain at the calcium bowl site, whereas the higher K21 ⁄ 2 value likely reflects Ca 2ϩ binding to the RCK1 domain, which has a lower apparent affinity in solution (33,47), also inferred from electrophysiological investigations (34 -37).
The apparent Mg 2ϩ affinity, K1 ⁄ 2 ϭ 154 Ϯ 20.7 M (n ϭ 2.02 Ϯ 0.44) was estimated by fitting to a Hill function the fluorescence intensity at 340 nm acquired at different [Mg 2ϩ ] (Fig. 4, B and  D). Saturating [Mg 2ϩ ] produced only an overall ϳ15% decrease in fluorescence intensity as compared with a 30% reduction observed with saturating [Ca 2ϩ ].
Previous investigations have shown that Ba 2ϩ cannot activate reconstituted BK channels (55,56); however, excluding Ba 2ϩ -dependent BK activation is not trivial due to the blocking action of Ba 2ϩ at micromolar concentrations (56) on the internal side of the BK channel selectivity filter (57,58). We probed whether Ba 2ϩ can bind and induce conformational changes to the isolated BK gating ring by measuring intrinsic Trp fluorescence while progressively increasing [Ba 2ϩ ]. The addition of Ba 2ϩ up to 13 mM did not alter the Trp fluorescence intensity of  .)), shown in B. The mean molecular weight estimated from 15 column elutions was 324,000 Ϯ 4,000. As a theoretical CTD tetramer would have M r ϭ 78,000 ϫ 4 ϭ 312,000, these data support the view that in solution, the purified CTD forms tetrameric structures, likely corresponding to the BK gating ring. the gating ring (Fig. 4, C and D). Nevertheless, when Ca 2ϩ was added at the end of the Ba 2ϩ experiment, the Trp fluorescence was quenched, confirming that protein activity was intact (Fig.  4C, red trace). Thus, the isolated gating ring exhibits cationic ligand selectivity, undergoing structural changes upon Ca 2ϩ and Mg 2ϩ binding with different apparent affinity, whereas not apparently responding to Ba 2ϩ .
Effect of Calcium Bowl Neutralization on Divalent Cation Sensitivity-BK channels possess a high-affinity Ca 2ϩ -binding site consisting of five consecutive aspartate residues (Asp 894 -Asp 898 ) called the calcium bowl (29). To investigate the role of the calcium bowl in the sensitivity of the isolated gating ring for cations, we expressed and purified a gating ring carrying the calcium bowl-neutralizing mutations D894A,D895A,D896A, D897A,D898A (supplemental Fig. S3A). Calcium bowl neutralization did not seem to alter the overall secondary structure of the gating ring, as suggested by CD spectroscopic analysis (supplemental Fig. S3B, supplemental Table S3). However, this gating ring mutant elutes from the size-exclusion column with an apparent molecular mass (Ͼ600,000) higher than that of a tetrameric structure. Time-correlated single photon counting revealed that calcium bowl neutralization did not completely abolish the Ca 2ϩ sensitivity of the mutant gating ring (Fig. 5A) as the average Trp fluorescence lifetime was decreased from avg ϭ 3.2 Ϯ 0.015 ns (0.00058 M Ca 2ϩ ) to avg ϭ 2.7 Ϯ 0.12 ns (at 35 M Ca 2ϩ ; fitting parameters in supplemental Table S4).
Saturating free [Ca 2ϩ ] resulted in a significantly smaller degree of steady-state Trp fluorescence quenching in the calcium bowl-neutralized gating ring (ϳ10%; cf. WT: ϳ30% quenching). These data were well fit by a single Hill function, with an apparent Ca 2ϩ affinity of 2.0 Ϯ 0.074 M and a Hill coefficient of n ϭ 2.4 Ϯ 0.39 (Fig. 5, B and D). This residual Ca 2ϩ sensitivity could be due to a high-affinity Ca 2ϩ -binding site in RCK1. Interestingly, the calcium bowl mutations completely abolished the gating ring sensitivity for Mg 2ϩ (up to 12 mM), suggesting that the Mg 2ϩ effects observed for the WT gating ring likely arise from Mg 2ϩ binding to the calcium bowl under nominal zero [Ca 2ϩ ] (Fig. 5, C and D).
The Apo and Ca 2ϩ -bound Conformations of the Gating Ring Have Different Hydrodynamic Radii-The results from fluorescence-based approaches provide evidence for physiologically relevant rearrangements in the BK gating ring induced by cation binding. However, they cannot provide information regarding the extent of the rearrangements involved because Trp fluorophores are reporting changes in their immediate microenvironment. Are these conformational changes associated with a detectable change in the overall gating ring shape? To gain direct insight on the Ca 2ϩ -induced conformational transitions, we have analyzed the hydrodynamic properties of gating ring particles in solution using the DLS technique, also known as photon correlation spectroscopy. DLS, which is based on the Brownian motion properties of particles in solution, provides a measurement of the particle hydrodynamic radius (R H ). Rather than a direct measure of particle size, R H refers to the radius of an ideal hard sphere with the same mass, density, and hydration properties as the investigated protein particles.
Particles with minimal surface-area-to-volume ratio (i.e. spheres) with a mass of 312,000 and density and hydration properties relevant to proteins would exhibit a theoretical R H (R H_Theo ) of 5.1 nm. The crystallographically resolved Ca 2ϩ -  JUNE 10, 2011 • VOLUME 286 • NUMBER 23 free gating ring (23) has a torus-like shape, with radius ϳ4.5-6 nm, height ϳ5 nm, and a 2-nm internal aperture (Fig. 6A). As this shape is not spherical, gating ring particles in solution are expected to exhibit a larger R H than R H_Theo . Indeed, the dominant distribution (71 Ϯ 4.3%) of purified gating ring particles in nominally Ca 2ϩ -free solution was estimated to have a mean R H of 10 Ϯ 0.41 nm (Fig. 6B). The particles outside the dominant distribution had R H Ͼ 60 nm (most of them around 400 nm) and probably correspond to protein aggregates that were not investigated. The polydispersity of the dominant distribution (i.e. the standard deviation divided by mean hydrodynamic diameter) was 12 Ϯ 0.53%, within the range (10 -15%) of homogeneous, monodisperse solutions (59).

Metal-sensing Properties of the BK Channel Gating Ring
Notably, when [Ca 2ϩ ] was elevated to 35 M, the gating ring R H was reduced to 7.6 Ϯ 0.14 nm (Fig. 6B), which corresponds to a 27 Ϯ 3.5% R H decrease (Fig. 6D). This result suggests that the Ca 2ϩ -bound gating ring particles diffuse faster in solution; their shape has less surface-area-to-volume ratio, i.e. they are more compact. This Ca 2ϩ -induced structural compaction is reversible; decreasing free [Ca 2ϩ ] from 35 to 0.07 M by the addition of EGTA almost fully restored the gating ring hydrodynamic radius (9.4 Ϯ 0.47 nm). Mg 2ϩ induced a much smaller R H decrease (7.0 Ϯ 3.5%; Fig. 6, C and D).

DISCUSSION
Recently resolved atomic structures of the human BK channel intracellular ligand-sensing apparatus have revealed that it is a hetero-octameric assembly of RCK domains (a pair of RCK1 and RCK2 domains from each subunit) arranged to form a gating ring structure (23,24). This work, which capitalizes on these structural data, is the first attempt to shed light on the molecular mechanisms by which the free energy from ions binding to the human BK channel gating ring structure is transduced into conformational rearrangements required to open the pore. The presumption of this study is that ligands binding to the gating ring induce detectable conformational changes. We have probed for and resolved ligand-induced conformational rearrangements under physiologically relevant conditions. The significant findings of this investigation are as follows. 1) Purified human BK C termini self-assemble under physiologically relevant conditions into tetramers, likely corresponding to gating rings. 2) Ca 2ϩ and Mg 2ϩ specifically induce conformational changes in the isolated gating ring, whereas Ba 2ϩ does not. 3) Neutralization of key residues within the calcium bowl significantly reduces the Ca 2ϩ effect on the gating ring rearrangements and eliminates Mg 2ϩ sensitivity. 4) Upon Ca 2ϩ binding, gating ring particles undergo a fully reversible reduction in their hydrodynamic radii, consistent with a compaction of their structure, whereas Mg 2ϩ has a modest effect on the gating ring shape. Crucially, the structural and functional features of gating ring parallel the Ca 2ϩ -dependent activation properties of the native BK channel (4,7,8,25,27,34,52,53).
The Cationic Ligand Selectivity of the Gating Ring-Ca 2ϩ , Mg 2ϩ , and Ba 2ϩ were used to probe the metal-sensing properties of the purified gating ring using its intrinsic Trp fluores- cence. Ca 2ϩ and Mg 2ϩ , known regulators of the BK channel, significantly affected the average lifetime of Trp fluorescence (Fig. 3, A and B and supplemental Tables S1 and S2), suggesting conformational rearrangements. These results are also supported by steady-state fluorescence measurements, suggesting at least two distinct high-affinity Ca 2ϩ -binding sites with K1 ⁄ 2 0.29 and 3.5 M (Fig. 4, A and D), in agreement with the electrophysiologically inferred apparent Ca 2ϩ affinity of whole BK channels (4,34,35,37).
The gating ring exhibited significantly less affinity for Mg 2ϩ (K1 ⁄ 2 ϭ 154 M, Fig. 4, B and D) and was apparently insensitive to Ba 2ϩ (Fig. 4, C and D), defining the cationic ligand selectivity of the BK metal-sensing apparatus. The lack of detectable Ba 2ϩinduced conformational changes is consistent with previous work on reconstituted BK channels that did not show evidence for Ba 2ϩ -dependent BK channel activation (55,56).
However, the well characterized block effect of Ba 2ϩ at the BK channel pore (55)(56)(57)(58) could confound attempts to characterize an alternative effect of Ba 2ϩ on channel activation (56). Because we did not resolve Ba 2ϩ -driven conformational changes in the isolated gating ring, we speculate that if Ba 2ϩ can activate BK channels, other protein domain(s), in addition to the gating ring, may be required.
How Many High-affinity Ca 2ϩ -binding Sites Exist within the Gating Ring?-A recent crystal structure of the BK gating ring in the presence of Ca 2ϩ (Protein Data Bank code 3MT5) definitively demonstrated that the calcium bowl region in RCK2 is a Ca 2ϩ -binding site (24). This finding was in agreement with previous electrophysiological investigations (29,34,35) as well as biochemical investigations of the calcium bowl region (30,32,60) and the whole RCK2 domain (33). Although it has yet to be confirmed in atomic structures, electrophysiological investigations support the existence of a second micromolar-affinity Ca 2ϩ -bindng site in the RCK1 domain (34 -38); indeed, purified RCK1 domains in solution assemble into homo-octameric gating ring-like structures that sense micromolar Ca 2ϩ in solution (47).
The Ca 2ϩ -induced conformational changes of the purified gating ring detected spectroscopically in this work exhibit a composite Ca 2ϩ dependence with K1 ⁄ 2 ϭ 0.29 and 3.5 M (Fig. 4,  A and D), which implies the existence of more than one type of high-affinity Ca 2ϩ -binding site. Furthermore, neutralization of the five consecutive aspartates that comprise the calcium bowl significantly reduced the overall extent of the detectable fluorescence changes (Fig. 5, B and D) but did not completely eliminate Ca 2ϩ sensitivity; calcium bowl gating ring mutants exhibit a single transition with an apparent Ca 2ϩ affinity of 2 M (Fig. 5D).
We propose that the transition with higher affinity (K11 ⁄ 2 ϭ 0.29 M) is relevant to Ca 2ϩ binding to the RCK2 calcium bowl site, whereas the second transition (K21 ⁄ 2 ϭ 3.5 M, Fig. 4D) arises from Ca 2ϩ binding to the RCK1 domain, which has less apparent affinity (34,35,37). Supporting this view, the homo-RCK1 gating ring exhibited an apparent Ca 2ϩ affinity of 1.7 M  (23), which has a toroidal structure with a diameter ϳ9 -12 nm, a height ϳ5 nm, and a central aperture of ϳ2 nm. Residues known to be involved in high-affinity Ca 2ϩ sensitivity in the BK gating ring are depicted in red (Asp 362 /Asp 367 in RCK1 and the calcium bowl in RCK2) or orange (Met 513 in RCK1). B, gating ring particle distributions obtained from a characteristic dynamic light scattering experiment. Only the dominant (70%) distributions below 60 nm are shown. The mean R H of the particle distribution in nominally 0 Ca 2ϩ (blue) was 10.1 nm. Increasing the free [Ca 2ϩ ] to 35 M shifted the distribution toward smaller R H values, with a mean of 7.8 nm (red), consistent with a transition to a more compact conformation. Reducing free Ca 2ϩ from 35 to 0.07 M by the addition of EGTA (green) practically restored the original distribution (9.5 nm), demonstrating that the Ca 2ϩ -induced rearrangements of the gating ring are reversible. C, a characteristic DLS experiment on the gating ring before (R H ϭ 9.70 nm; blue) and after (R H ϭ 8.65 nm; orange) the addition of 12 mM Mg 2ϩ . D, overall R H reduction by the addition of saturating ligand concentration to gating ring particles in solution, expressed as the percentile change from the mean ligand-free R H (R H_Apo ). Error bars indicate S.E. (47), similar to the lower Ca 2ϩ affinity component of the wildtype gating ring (3.5 M) and the residual Ca 2ϩ sensitivity of the calcium bowl mutant (2 M). Electrophysiological investigations have highlighted residues Asp 362 , Asp 367 , Met 513 , Arg 514 , and Glu 535 as candidates for Ca 2ϩ coordination in RCK1 (34 -38). Consistent with this view, the Ca 2ϩ binding properties of isolated RCK1 domains in solution suggest that an RCK1 Ca 2ϩbinding site could exist with multiple Ca 2ϩ -coordinating residues (47).
The Calcium Bowl Binds Mg 2ϩ with Low Affinity-A lowaffinity Mg 2ϩ -binding site known to modulate channel activity (61,62) is likely located at the interface between the RCK1 domains (Glu 374 , Glu 399 ) and the voltage sensors (Asp 99 , Asn 172 ) (4,7,34,(41)(42)(43), enhancing the efficiency of the voltage sensor-gate coupling (63). Because the purified gating ring does not include transmembrane domains, the effects observed in this study (Fig. 4, B and D) likely arise from Mg 2ϩ binding to alternative sites. We found that calcium bowl neutralization completely eliminated the Mg 2ϩ effect on the endogenous Trp fluorescence of the gating ring (Fig. 5, C and D), suggesting that the calcium bowl can bind Mg 2ϩ with much less affinity than Ca 2ϩ . The addition of Mg 2ϩ to the gating ring in the presence of 35 M free Ca 2ϩ induced no further spectroscopically resolvable conformational changes (supplemental Fig. S4), supporting the view that in the isolated gating ring, Mg 2ϩ acts through the calcium bowl site as an inefficient Ca 2ϩ mimetic, rather than an allosteric ligand. These results recapitulate previous electrophysiological experiments (4,61,62,64), where it was demonstrated that Mg 2ϩ competes with Ca 2ϩ for the highaffinity Ca 2ϩ -binding sites.
The Gating Ring Apparatus Operates as a Selective Chemomechanical Transducer of Metal Binding-In this work, we have demonstrated Ca 2ϩ -dependent structural rearrangements of the whole human BK channel gating ring apparatus, resolved by the fluorescence emission of native Trp residues and, from a whole-particle perspective, by dynamic light scattering. The latter showed that the gating ring reshapes in response to [Ca 2ϩ ] binding, toward a "more spherical" or compact conformation characterized by a reduced surface-area-tovolume ratio. As these conformational rearrangements were induced by [Ca 2ϩ ] relevant to BK channel activation, we propose that they likely constitute the prime molecular events that propagate to the BK channel pore, to increase open probability. In contrast, Mg 2ϩ binding produced much smaller changes in the shape of the gating ring (Fig. 6, C and D), which suggests that the Mg 2ϩ -induced perturbations in the microenvironment of Trp residues (ϳ15% maximal decrease in fluorescence intensity as compared with a 30% reduction observed with saturating [Ca 2ϩ ]) do not tightly correlate with the global structural rearrangements of the gating ring. Possibly, Trp residues report changes to their microenvironment, which may not reflect the extent of macroscopic structural rearrangements. The binding of Mg 2ϩ to the Ca 2ϩ -binding sites does not activate the channel (61,62); therefore, the modest compaction of the gating ring caused by saturating [Mg 2ϩ ] is likely to be insufficient to facilitate pore opening. Finally, the lack of spectroscopic responses of the gating ring to Ba 2ϩ further defined its selectivity for divalent metal cations.
In this work, we probed the divalent metal-sensing properties of the BK channel gating ring under physiological conditions using a combination of spectroscopic (time-resolved and steady-state spectroscopy) and particle-scale (DLS) optical approaches. We have demonstrated that the gating ring responds to elevations in [Ca 2ϩ ] and, with much less affinity, [Mg 2ϩ ]. Micromolar Ca 2ϩ can induce robust microscopic and macroscopic conformational rearrangements compatible with a compaction of the gating ring superstructure, which likely facilitate pore opening in the whole BK channel. The micromolar sensitivity of the gating ring to Ca 2ϩ persists after neutralization of the calcium bowl site, confirming the existence of more high-affinity Ca 2ϩ -binding sites. In contrast, millimolar Mg 2ϩ can induce changes to a smaller extent than Ca 2ϩ , and these changes are probably insufficient to facilitate channel activation in vivo. Finally, mirroring the selectivity of the channel to divalent metal agonists, the gating ring underwent no resolvable conformational rearrangements upon [Ba 2ϩ ] elevation up to millimolar levels. Thus, we have demonstrated that the BK gating ring operates as a selective chemo-mechanical coupler that transduces the free energy of ligand binding into mechanical work that ultimately activates the channel.