Crystal Structure of a Ba²⁺-bound gating ring reveals elementary steps in RCK domain activation

SUMMARY

RCK domains control activity of a variety of K⁺ channels and transporters through binding of cytoplasmic ligands. To gain insight toward mechanisms of RCK domain activation, we solved the structure of the RCK domain from the Ca²⁺-gated K⁺ channel, MthK, bound with Ba²⁺, at 3.1 Å resolution. The Ba²⁺-bound RCK domain was assembled as an octameric gating ring, as observed in structures of the full-length MthK channel, and shows Ba²⁺ bound at several positions. One of the Ba²⁺ sites, termed C1, overlaps with a known Ca²⁺-activation site, determined by residues D184 and E210. Functionally, Ba²⁺ can activate reconstituted MthK channels as observed in electrophysiological recordings, whereas Mg²⁺ (up to 100 mM) was ineffective. Ba²⁺ activation was abolished by the mutation D184N, suggesting that Ba²⁺ activates primarily through the C1 site. Our results suggest a working hypothesis for a sequence of ligand-dependent conformational changes that may underlie RCK domain activation and channel gating.

INTRODUCTION

Regulator of K⁺ conductance (RCK) domains are structurally-conserved ligand-binding domains that control the activity of a diverse array of K⁺ channels and transporters, including large-conductance Ca²⁺-activated K⁺ channels (BK channels), in response to binding of cytoplasmic ligands, such as nucleotides, Na⁺, Ca²⁺, Mg²⁺, and H⁺ (Bellamacina, 1996; Butler et al., 1993; Golowasch et al., 1986; Jiang et al., 2002a; Jiang et al., 2001; Kroning et al., 2007; Kuo et al., 2005; Parfenova et al., 2007; Salkoff et al., 2006; Schreiber et al., 1998; Xia et al., 2004; Yuan et al., 2003; Zhang et al., 2006a). Consequently, RCK-containing channels and transporters couple binding of ligands to regulation of membrane potential or K⁺ uptake. For example, the bacterial KtrAB complex couples nucleotide binding with K⁺ uptake, whereas BK channels link binding of Ca²⁺ and Mg²⁺ to K⁺ efflux to control smooth muscle contractilty and neuronal activity (Brenner et al., 2005; Brenner et al., 2000; Girouard et al., 2010; Imlach et al., 2008; Meredith et al., 2006).

MthK is a prototypical RCK-containing K⁺ channel whose crystal structure has provided insight toward mechanisms of channel gating by RCK domains (Hou et al., 2008; Jiang et al., 2002a, b; Niu et al., 2004; Pau et al., 2010; Pau et al., 2011; Ye et al., 2006). In MthK, binding of Ca²⁺ to an octameric ring of RCK domains (the “gating ring”), which is tethered to the pore of the channel, leads to a series of conformational changes that facilitates channel opening and K⁺ conduction (Figure 1A) (Jiang et al., 2002a; Li et al., 2007; Parfenova et al., 2006; Pau et al., 2010; Zadek and Nimigean, 2006). The Ca²⁺-activation mechanism of MthK involves three different Ca²⁺ binding sites per RCK domain, for a total of 24 Ca²⁺ sites per channel (Pau et al., 2011); thus a complete quantitative description of the gating mechanism could be complex. To dissect the mechanism and quantify energetic relationships among modulatory binding sites and the gate of the channel, it is useful to have access to experimental tools that permit channel activation through specific sites. In addition, it will be of value to understand the structural basis of ion coordination and selectivity at specific binding sites, as this information may provide insight toward ion coordination mechanisms in other proteins.

Figure 1. Comparison of Ca²⁺- and Ba²⁺-bound MthK gating rings.

A, C-α trace of the biological structure of the MthK channel (PDB accession no. 3RBZ) . Presumed boundary of the plasma membrane is shaded gray, with extracellular and cytoplasmic sides of the membrane indicated by ext and cyt, respectively. Subunits that form the pore region and their associated RCK domains are colored magenta (unresolved linker residues are represented by red dashes); associated RCK domains are colored yellow. Ca²⁺ ions are shown as green spheres. B, C, Side-by-side comparison of the Ca²⁺-bound (B) and Ba²⁺-bound gating ring (C; PDB accession number 4EI2), viewed from above; in B, the transmembrane domains have been removed for clarity. In C, Ba²⁺ ions are shown as purple spheres. Ions at the C1 sites are circled. D, Magnified view of the boxed region in C, showing a dimer of RCK domains with associated Ba²⁺ ions. Each dimer consists of an N-lobe, a cleft, and a C-lobe, radiating from the central cavity of the gating ring. Blue circled regions show the location of the C1 site. E, Ca²⁺ coordination at the C1 site. F, Structure of the Ba²⁺-bound C1 site. Structure is overlayed with NCS-averaged anomalous difference map contoured at 20σ (purple mesh). Ba²⁺-O distances, indicated by dashed lines, are: D184(Oδ1), 2.6 Å; D184(Oδ2), 4.0 Å; E210(Oε1), 3.8 Å; L185(O), 2.8 Å. See also Figure S1.

In this work, we use Ba²⁺ to pan for native metal binding sites in the MthK RCK domain. Use of Ba²⁺ as a probe in X-ray crystallographic analysis has several advantages; Ba²⁺ is electron-dense, and thus in general can be distinguished from other protein atoms based on its strong electron density peak (Jiang and MacKinnon, 2000). Also, Ba²⁺ can yield anomalous scattering that is detectable with moderate resolution diffraction data, further enabling its unambiguous distinction from other cations, water or protein atoms.

A specific advantage in using Ba²⁺ to probe MthK RCK domain structure and function is that MthK contains native Ca²⁺ binding sites. Ba²⁺, although slightly larger than Ca²⁺ in ionic radius, can sometimes act as a functional surrogate for Ca²⁺, for example as a charge-carrying permeant ion in the analysis of Ca²⁺ channels, as well as activation of BK channels (DiPolo et al., 1983; Hagiwara and Ohmori, 1982; Hille, 2001; Lee and Tsien, 1983; Randall and Tsien, 1995; Todorovic and Lingle, 1998; Zhou et al., 2012). On the other hand, Ba²⁺ cannot effectively replace Ca²⁺ in triggering neurotransmitter release at the frog neuromuscular junction (Zengel and Magleby, 1977, 1980). Thus some Ca²⁺ binding sites may bind Ba²⁺, whereas other may exclude it.

Our goal in the present studies was to gain further insight toward properties of binding sites that may contribute to Ba²⁺/Ca²⁺ selectivity, and also to exploit potential selectivity to gain further insight toward channel activation through ligand binding to specific sites. Our results suggest that Ba²⁺ can distinguish among activation sites in the MthK RCK domain, and selectively binds to a single activation site in each RCK domain, facilitating opening of the MthK channel. Comparison with previous MthK gating ring structures suggests that the Ba²⁺-bound gating ring exists in an intermediate-activated conformation, and binding of additional ligands is required for maximal stability of the fully-activated conformation in the channel gating mechanism.

Results

Ba²⁺-coordination in the MthK gating ring

The MthK channel is formed by a pore domain that yields a pathway for conduction of K⁺ through the plasma membrane, and a tethered RCK domain complex (the “gating ring”; Figure 1A) (Jiang et al., 2002a; Pau et al., 2011). Biologically, the gating ring is comprised of eight RCK domains, four of which are directly tethered to the pore via a 19-residue linker (represented by dashed lines in Figure 1A), and four of which are expressed as soluble domains that are co-assembled with the tethered domains (Jiang et al., 2002a). Here we focus on the structure of the gating ring to determine conformations that may underlie activation of the channel by divalent cations, to gate transmembrane K⁺ conduction.

Using Ba²⁺ as a probe for divalent binding sites in the MthK RCK domain, we crystallized the MthK RCK domain protein in the presence of 100 mM barium chloride and obtained crystals that diffracted X-rays to 3.1 Å, in space group P2₁2₁2₁ (Table 1). The structure was solved by molecular replacement and refined to an R_work/R_free of 0.209/0.243. Two essentially identical gating rings (16 RCK domains) are present per asymmetric unit (Figures 1 and S1), which results in 16-fold non-crystallographic symmetry (NCS). The final refined electron density maps were of excellent quality owing in part to 16-fold NCS-averaging, and allowed building of the entire RCK domain for all 16 chains in the asymmetric unit. Ba²⁺ ions bound to the RCK gating ring were identified in the structure based on a combination of strong electron density peaks, coinciding peaks in the anomalous difference density map, and favorable local chemistry for Ba²⁺ coordination at the site (i.e. close contact with oxygen atoms, and lack of close contacts with basic or hydrophobic sidechains). Based on these data, we identified positions of several Ba²⁺ ions in each gating ring (Figure 1C), which we categorized into several classes based on their locations in each RCK domain.

Table 1.

Data collection and refinement statistics

Data collection
Space group	P212121
Cell dimensions
a, b, c (Å)	110.0, 136.4, 498.4
α, β, γ (°)	90, 90, 90
Resolution (Å)	30-3.1 (3.22-3.1)
No. total / unique reflections	3,132,635 / 131,081
R sym	0.118 (0.556)
I / σI	24.4 (4.2)
Completeness (%)	97.5 (96.4)
Redundancy	4.9 (4.4)
Refinement
Resolution (Å)	30-3.1
No. unique reflections (non-anomalous)	131,021
Rwork / Rfree$	0.209 / 0.243
No. atoms
Protein	27,716
Ligand/ion	57
Water	32
B-factors
Protein	141.2
Ligand/ion	257.4
Water	101.8
R.m.s. deviations
Bond lengths (Å)	0.006
Bond angles (°)	0.996

*Values in parentheses are for highest-resolution shell.

^$R_free was calculated using 3,758 reflections selected in thin resolution shells

Most notably, each RCK domain contained a Ba²⁺ ion coordinated by sidechain oxygens from residues D184 and E210 and the main-chain carbonyl oxygen of residue L185, at a site that coincides in part with a Ca²⁺ binding site identified in previous Ca²⁺-bound structures of the MthK RCK domain, here termed the “C1” site (Dong et al., 2005; Jiang et al., 2002a). Comparison of ion coordination in the Ca²⁺-bound and Ba²⁺-bound RCK domain structures illustrates differences between Ca²⁺ and Ba²⁺ coordination at C1, principally the shorter metal-oxygen distances for Ca²⁺ (ranging from 2.4 – 2.7 Å) vs. Ba²⁺ (ranging from 2.6 – 4.4 Å) (Figure 1E and F). Mean Ba²⁺-O distances for residues D184, L185, and E210 were: D184(Oδ1), 2.7 Å; D184(Oδ2), 4.0 Å; E210(Oε1), 3.6 Å; and L185(O), 2.9 Å, which is within the range of Ba²⁺-O distances observed in structures of other proteins at comparable resolution (Chaptal et al., 2011; Inanobe et al., 2011). Unlike Ca²⁺ coordination observed in previous MthK structures, the E212 sidechain does not appear to participate directly in Ba²⁺ coordination in this gating ring structure; the mean distance between Ba²⁺ and the nearest sidechain oxygen from E212 was 4.5 Å (Figure 1E, F).

Aside from C1, Ba²⁺ ions were located at several other sites on the gating ring, though these other Ba²⁺ ions interacted directly with fewer protein oxygens than observed in C1 based on Ba²⁺-O distances at these sites (Figure 2). These sites were termed B2 (or the “cavity site”), formed by an electronegative pocket of E190 residues from adjacent subunits; B3, formed by the sidechain hydroxyl group of Y214 and the main chain carbonyl from E258, also in adjacent subunits; and B4, formed in part by the main chain carbonyl from D244. These sites were characterized by strong electron density, anomalous peaks, and favorable local chemistry, though they did not exhibit multiple close protein contacts as observed at the C1 site. For site B2, the mean distance to E190(Oε2) was 4.6 Å; for B3, mean distances to Y214(OH) and E258(O) were 4.6 and 4.5 Å, respectively; for B4, the mean distance to the only apparent close protein contact (D244(O)) was 3.2 Å. Although multiple close protein contacts (<4 Å) were not observed at these sites, each site was characterized by a relatively high density of oxygen-containing sidechains (Figure 2), which likely provides a favorable electrostatic environment for cation-binding.

Figure 2. Ba²⁺ sites in the MthK gating ring cavity and cleft region.

Locations of Ba²⁺ sites in the RCK domain are indicated by blue circles (left panels); representative sites are shown as ball-and-stick (right panels), overlayed with NCS-averaged 2Fo-Fc maps (blue mesh) and anomalous difference maps (purple mesh). A, Ba²⁺ site in the gating ring cavity determined by E190 residues from adjacent subunits (indicated by ′) Ba²⁺-O distances in Å: E190(Oε2), 4.9; E190′(Oε), 4.9. B, Ba²⁺ site determined by the Y214 sidechain and E258 carbonyl from adjacent subunits. Ba²⁺-O distances in Å: Y214(OH), 4.6; E258′(O), 4.5. C, Ba²⁺ site determined by the main-chain carbonyl of D244. Ba²⁺-O distance to D244(O), 2.6 Å. NCS-averaged 2Fo-Fc maps are contoured at 5σ; NCS-averaged anomalous difference maps are contoured at 20σ (A, C) or 14σ (B). See also Figure S2.

Additional weaker Ba²⁺ sites were modeled in the vicinities of residues D202, D287, and D284 (Figure S2). Because strong anomalous peaks were not observed at these sites in all RCK subunits, identification of these sites was mainly based on strong electron density peaks that could not be adequately modeled as water molecules or Na⁺ ions. At one of these sites (formed by D284/D287), Ba²⁺ ions may provide a countercharge to stabilize lattice contacts between gating rings (Figures S1, S2C). No density corresponding to Ba²⁺ was observed in the other previously-identified MthK Ca²⁺-activation sites, formed between residue E248 and E266, and between residues D305 and E326 (Pau et al., 2011).

Thus, the most well-defined Ba²⁺ site in the RCK domain was observed at site C1, which overlaps with a key Ca²⁺ activation site in the MthK channel (Jiang et al., 2002a; Pau et al., 2011).

Conformation of the Ba²⁺-bound gating ring and its relation to MthK channel activation

Comparison of MthK gating ring structures shows that the Ba²⁺-bound gating ring is in a different overall conformation from the Ca²⁺-bound gating ring (Figure 1B, C). Because the Ca²⁺-bound gating ring is tethered to the apparently open MthK pore, it has been hypothesized that the Ca²⁺-bound form represents the “activated” gating ring conformation (Jiang et al., 2002a). Because the Ba²⁺-bound gating ring structure is in a different overall conformation, initially we hypothesized that this conformation may not be in an fully-activated form. We therefore tested whether Ba²⁺ could activate MthK channels by recording MthK currents with Ba²⁺ (instead of Ca²⁺) at the cytoplasmic side of the channel; in these experiments, the pore-blocking effect of Ba²⁺ was minimized by analyzing MthK currents recorded at -100 mV, at which little Ba²⁺ blockade would be expected to occur (Thomson and Rothberg, 2010).

Despite the apparent conformational differences between the Ca²⁺- and Ba²⁺-bound gating rings, we found that MthK channels can be activated by Ba²⁺ (Figure 3A, B). With 10 mM Ba²⁺, low levels of activity are observed, and increasing [Ba²⁺] at the cytoplasmic side of the channel increases Po, in a manner that was completely reversible (Figure S3). Increasing [Ba²⁺] also leads to decrease in single-channel current amplitude, which appears to arises from fast blockade, similar to that elicited by Ca²⁺ and other divalent cations in K⁺ channels (Ferguson, 1991; Li et al., 2007; Zhang et al., 2006b).

Figure 3. Ba²⁺-activates MthK channels.

A, Current traces illustrating activation of MthK channels at pH 7.3 with Ba²⁺ at the indicated concentrations. Traces are from the same bilayer, which contained 24 active channels. Open channel current levels are indicated by dotted lines; note that increasing [Ba²⁺] decreased the size of the single-channel current, through an apparent fast-blocking effect. B, Po vs. [Ba²⁺] relations determined at pH 6.5, 7.3 and 7.7. Curves represent Hill equation fits, with EC₅₀ = 60.5, 81.1, and 54.8 mM Ba²⁺, and n_H = 5.3, 5.4, and 3.5 for pH 6.5, 7.3, and 7.7, respectively. C, Means of log(EC₅₀) values determined from fitting Po vs. [Ba²⁺] relations of MthK wild-type channels from individual bilayers, plotted as a function of pH. Log(EC₅₀) values for Ca²⁺-activation of MthK wild-type channels (circles) are plotted on the same axes for comparison, and illustrate the steep pH-dependence for Ca²⁺ activation. Ca²⁺-activation of E212Q mutant channels (triangles) exhibits pH-dependence comparable to wild-type. Log(EC₅₀) values for Ca²⁺ activation correspond to EC₅₀ (in mM) of 16 ± 0.6, 3.0 ± 0.2, and 2.2 ± 0.1 for wild-type (pH 6.5, 7.3, and 7.7, respectively) and 9.9 ± 0.2 and 1.3 ± 0.1 for E212Q channels (pH 6.5 and 7.7). D, Means of Hill coefficients, also determined from fitting Po vs. [Ba²⁺] relations of individual bilayers, plotted with Hill coefficients for Ca²⁺-activation. See also Figure S3.

It was previously observed that Ca²⁺-activation of MthK is inhibited by cytoplasmic H⁺, through a mechanism in which H⁺ specifically inhibits Ca²⁺-dependent gating and not the intrinsic gating of the MthK pore (Pau et al., 2010). The effect of [H⁺] on Ca²⁺-activation of MthK is substantial, resulting in a 5.6-fold change in Ca²⁺ EC₅₀ per 10-fold change in [H⁺] over the range of 20 to 316 nM H⁺ (pH 6.5 to 7.7; Figure 3C). To determine whether Ba²⁺ activates MthK through a pH-sensitive mechanism, we assessed Ba²⁺-activation over a range of pH (6.5 to 7.7). Figure 3B shows that, interestingly, Ba²⁺ activation of MthK is much less H⁺-sensitive than Ca²⁺-activation, with an 11% decrease in Ba²⁺ EC₅₀ per 10-fold increase in [H⁺] over the range of 20 to 316 nM H⁺ (pH 7.7 to 6.5; Figure 3C).

Because Ca²⁺ activates MthK channels through other sites in addition to C1, our electrophysiological results suggest that pH modulation of MthK channels does not arise through direct modulation of C1, but instead through the other Ca²⁺ sites (Pau et al., 2011). However, the lack of pH modulation with Ba²⁺ activation could also arise from non-participation of the E212 sidechain, which contributes to Ca²⁺ coordination but not Ba²⁺ coordination. Therefore to determine whether protonation of E212 might contribute to pH modulation of Ca²⁺-dependent activation, we recorded activity from E212Q mutant channels over a range of [H⁺] and [Ca²⁺]. If pH-dependence arises from protonation of E212, then the E212Q mutation should abolish or substantially reduce the shift in EC₅₀ for Ca²⁺ activation. We observe, however, that E212Q channels retain pH-sensitivity comparable to MthK wild-type channels, with a 5.2-fold change in Ca²⁺ EC₅₀ per 10-fold change in [H⁺] over the range of 20 to 316 nM H⁺ (Figure 3C). This further supports the idea that pH modulation of MthK channels does not arise through titration of Ca²⁺-coordinating residues at C1 and may arise through effects on other sites, or effects on residues that couple these sites to channel opening.

Comparison of EC₅₀ values for Ca²⁺ and Ba²⁺ activation also illustrates that overall, higher [Ba²⁺] is required to activate the channel than [Ca²⁺] (Pau et al., 2010). This difference may reflect a lower binding affinity for Ba²⁺ at its activation sites and/or overall weak energetic coupling between Ba²⁺-binding and channel opening. In addition to the higher EC₅₀ for Ba²⁺-activation, we also observe generally lower Hill coefficients for Ba²⁺-activation compared to Ca²⁺-activation of the channel (mean Hill coefficients from data in Figure 3D were 5.3 ± 0.7 for Ba²⁺ activation, 9.5 ± 1.3 for Ca²⁺ activation, from 10 and 16 different bilayers, respectively), which may reflect a weaker energetic coupling between Ba²⁺ binding and channel opening in MthK compared with Ca²⁺-activation (Pau et al., 2010; Zadek and Nimigean, 2006).

Ba²⁺ activation occurs primarily through binding to the C1 site

Our identification of Ba²⁺ binding sites in the MthK gating ring and functional observations of steep relation between Po and [Ba²⁺] in reconstituted MthK channels are consistent with the idea that Ba²⁺ activation may arise through binding to a specific site in the MthK RCK domain. However, because Ba²⁺ appears to decorate many sites on the RCK domains (at 100 mM), it is alternatively possible that channel activation may occur through Ba²⁺ acting as a non-specific counter-charge at one or more of these identified sites. We next tested whether Ba²⁺ activation might arise from a general effect of increasing ionic strength, first by assessing the effects of other divalent ions on channel gating. Of the ions tested, we found that 100 mM Mg²⁺ elicited no detectable activation of MthK channels, while 100 mM Sr²⁺ or 0.5 mM Cd²⁺ resulted in near-maximal levels of channel activity, over pH ranging from 7.3 to 7.7 (Figure 4). The observation that Mg²⁺ is essentially ineffective, while Cd²⁺ appears to be an even more potent activator of MthK than Ca²⁺ under otherwise identical ionic conditions, is consistent with the idea that divalent cation activation of MthK arises through a set of selective binding sites. These results agree well with previous findings that the activation sites of MthK are highly-selective for Ca²⁺ over Mg²⁺ (Dong et al., 2005; Parfenova et al., 2007; Zadek and Nimigean, 2006). In contrast, Cd²⁺ is known to bind to at least two sites in the MthK RCK domain; the C1 site, as well as an additional site determined by residues D305 and E326, which binds Ca²⁺ but not Ba²⁺ (Dvir et al., 2010; Pau et al., 2011).

Figure 4. MthK channels can be activated by Cd²⁺ or Sr²⁺, but not Mg²⁺.

Representative current traces at -100 mV showing activation of wild-type MthK channels at pH 7.3 (left) and 7.7 (right) with Ca²⁺ at the cytoplasmic side of a bilayer, followed by replacement with a solution containing either Mg²⁺, Sr²⁺, or Cd²⁺, followed by replacement back to Ca²⁺ (at the indicated concentrations). At either pH, 100 mM Sr²⁺ or 0.5 mM Cd²⁺ was sufficient to maximally activate MthK channels (Po = 0.95 ± 0.02 for Sr²⁺; 0.92 ± 0.04 for Cd²⁺), whereas no activation was observed with 100 mM Mg²⁺ (bottom traces; 0 openings observed). Breaks in the recordings, indicated by “/ /”, were introduced manually and represent a gap of ~3 sec in each instance.

Coordination of Ba²⁺ at the C1 site led us to further hypothesize that Ba²⁺ activation may arise specifically through binding of Ba²⁺ at this site. To test this idea, we obtained recordings from D184N mutant channels with [Ba²⁺] sufficient to elicit near-maximal activation of wild-type MthK channels. The D184N mutation was previously shown to decrease Ca²⁺ activation of MthK, and apparently knock out Ca²⁺ binding at the C1 site, as indicated by crystallographic data (Pau et al., 2011). Figure 5 illustrates that no detectable Ba²⁺ activation is observed in D184N channels at [Ba²⁺] up to 100 mM, consistent with Ba²⁺ activation occuring primarily through the C1 site.

Figure 5. The D184N mutation effectively abolishes Ba²⁺ activation of MthK.

Representative current traces at -100 mV showing MthK channel activation at pH 7.3 with 15 mM Ca²⁺ at the cytoplasmic side of a bilayer, followed by replacement with a solution containing 100 mM Ba²⁺ (with no added Ca²⁺), and then back to 15 mM Ca²⁺. The top series of traces shows that 100 mM Ba²⁺ was sufficient to maximally activate wild-type MthK channels (Po = 0.9), whereas no Ba²⁺ activation was observed with D184N mutant channels (0 openings observed, n=5 different bilayer experiments).

DISCUSSION

Working hypothesis for the physical basis of Ba²⁺ recognition

Here we have exploited structural analysis of MthK to dissect mechanisms underlying metal-activation of RCK domains. We observe principally that Ba²⁺ can bind to a site on each RCK domain that overlaps with a Ca²⁺-activation site (C1, determined by residues D184 and E210), and through binding to this site, Ba²⁺ can activate the RCK domain to facilitate opening of the channel.

Ba²⁺ has a slightly larger ionic radius than Ca²⁺, and apart from general chemical similarities, Ba²⁺ and Ca²⁺ are known to have quantitatively different characteristics. For example, Ba²⁺ water substitution rate has been estimated to be around an order of magnitude greater for Ba²⁺ than that of Ca²⁺ (Hille, 2001). It is known that Ba²⁺ permeates rapidly through voltage-dependent Ca²⁺ channels, and can bind to and partially activate calmodulin, whereas it seems to be a poor substitute for Ca²⁺ in other physiological processes (such as neurotransmitter release) (DiPolo et al., 1983; Kursula and Majava, 2007; Satoh et al., 1987; Zengel and Magleby, 1980, 1981). Our results suggest that Ba²⁺ can bind to and activate MthK channels through selective action at the C1 site, whereas Ba²⁺ is effectively excluded from binding at the other known MthK Ca²⁺ sites. Because Ba²⁺ coordination at the C1 site incorporates additional protein oxygens that do not participate in Ca²⁺ coordination, it is likely that the presence of these oxygens at favorable positions, and the absence of such oxygens at other sites, contributes to the apparent exclusion of Ba²⁺ from the other MthK Ca²⁺ sites.

In addition to its similarities to Ca²⁺, Ba²⁺ also is nearly identical in ionic radius to K⁺, and Ba²⁺ can flow through K⁺ channels (albeit much more slowly than K⁺)(Bello and Magleby, 1998; Neyton and Miller, 1988a, b; Rothberg et al., 1996). Because, in contrast to site C1, Ba²⁺ ions at sites B2 through B4 do not form multiple close contacts with protein oxygens and likely interact with the protein in near fully-hydrated states (Figure 2 and S2), one may speculate that under physiological ionic conditions with 150 mM (or more) K⁺ in the cytoplasm, K⁺ ions may occupy these sites as countercharges. Similar cation binding sites have been observed within the electronegative pockets in the cytoplasmic domain of the Kir channel (Xu et al., 2009).

Mechanism of Ba²⁺ activation and implications for the Ca²⁺ activation mechanism

We observed that the overall conformation of the Ba²⁺-bound gating ring is different from that of the fully-activated, Ca²⁺-bound full-length MthK channel (Figure 1B, C) (Jiang et al., 2002a, b). In previous work, it was found that MthK channels in which two of the three Ca²⁺ binding sites the RCK domain have been mutated can diminish Ca²⁺-activation, suggesting that MthK channels can be weakly activated through single Ca²⁺ binding sites (Pau et al., 2011). Thus it is possible that Ba²⁺ activates MthK channels through binding to the C1 site (Figures 1, 3 and 5), and hence that the Ba²⁺-bound gating ring represents an intermediate, singly-liganded conformation in the domain's activation pathway.

To further explore whether the Ba²⁺-bound gating ring might represent an intermediate, singly-liganded conformation, we compared the conformation of a dimeric unit from the Ba²⁺-bound gating ring with other previously-determined MthK RCK domain structures by aligning the structures to yield the minimal root-mean-square deviation (rmsd). We reasoned that comparisons yielding the lowest rmsd values would indicate structures in the same conformation, and higher rmsd values would indicate conformational differences.

Through these comparisons, a physical picture of conformational transitions that may underlie MthK gating emerges (Figure 6; Table S1). We first noted that the structure of the unliganded MthK gating ring (with the D184N mutation; PDB accession no. 2FY8; (Ye et al., 2006)) showed that the unliganded ring can exist in two different overall conformations (Figure 6A, (1) and (2)). Unliganded ring (1) contains component RCK domains in two different conformations, indicated in Figure 6A by “a” (red box) and “b”, whereas unliganded ring (2) contains all RCK domains in the b form (Figure 6A, blue dashed box). Our alignments revealed that form a is most similar to a wild-type unliganded RCK domain (PDB accession no. 2AEJ; rmsd = 0.77 Å; Figure S4A; (Dong et al., 2005)), whereas b is most similar to the Ba²⁺-bound conformation, indicated by b* in Figure 6A (rmsd = 0.74 Å; Figure S4B). In contrast, alignment of form a with the fully-liganded RCK domain (c*, green box in Figure 6A) yielded a higher rmsd value, consistent with a conformational difference between these structures (rmsd = 1.76 Å; Table S1). Based on these data, we hypothesize that for wild-type MthK channels, RCK domains within the gating ring may undergo sequential transitions from form a to b, where b can be stabilized by binding of Ba²⁺ (or potentially Ca²⁺) to yield b*, followed by additional Ca²⁺ binding to yield c* (Figure 6A; see Supplemental Discussion). The ligand-bound states (b* and c*) would underlie facilitation of channel opening through allosteric coupling.

Figure 6. Working hypothesis of elementary steps in MthK gating ring activation.

A, RCK domains undergo transitions from inactive (a) to partially-activated/singly-liganded (b / b*), then to fully-liganded/activated (c*). Detailed explanation provided in Discussion. B, Aligned and overlayed unliganded (1), Ba²⁺-bound (3) and Ca²⁺-bound gating rings (4). C, Expanded view of the region indicated by black dashed box in B. Spheres within the dashed box indicate the Cα of residue R116, which in turn would be linked to the transmembrane pore. Changes in position of R116 may indicate RCK domain movements coupled to channel opening. D, Expanded view of the boxed region from C. Dashed arrows indicate the sequence a→b*→c*. E, Aligned and overlayed structures, viewed normal to the plane of the membrane. F, Expanded view of the boxed region from E, indicating the sequence a→b*→c*. See also Figure S4 and Table S1.

Alignment of the complete gating ring structures (1), (3), and (4) reveals a potential sequence of positions for the N-terminal ends of RCK domains (Cα's of R116, indicated by spheres in Figure 6D and F). In the context of the full-length channel, these N-termini are linked to the pore and are thus likely to impact pore conformation. These alignments suggest that binding of Ba²⁺ may lead to a relative translation of the R116 Cα in a direction normal to the membrane (a→b*; Figure 6F), whereas binding of Ca²⁺ to additional activation sites may move the R116 Cα in a plane parallel to the membrane (b*→c*, Figure 6D, F), to further stabilize the open state (Pau et al., 2011; Ye et al., 2006). Further experiments will be required to directly assess the impact of these movements on pore structure.

Relation to gating mechanisms in eukaryotic channels

Activation by metal cations, including Ba²⁺, has been used previously to dissect gating mechanisms of eukaryotic BK channels, yielding insight toward the specificity of putative metal binding sites within the BK channel RCK domains, defined through mutational analysis (Shi et al., 2002; Xia et al., 2002; Zeng et al., 2005; Zhang et al., 2001; Zhou et al., 2012). Despite determination of the structure of the BK channel gating ring, however, the structural basis of metal binding is clear only in the case of Ca²⁺ binding at the “Ca²⁺ bowl site”, determined principally by residues D985 and D897 (in the human BK channel)(Yuan et al., 2011; Yuan et al., 2010). Thus the structural basis for preference of specific metals for putative activation sites within the BK channel RCK domain is yet unclear. Results obtained through structural analysis of MthK channels may continue to provide useful insights toward the physical basis of RCK domain activation, complementing ongoing analysis of BK channels.

EXPERIMENTAL PROCEDURES

Crystallization

Purification of the MthK RCK domain was performed as described previously (Pau et al., 2011). The crystal analyzed for this study formed in a 2 μl hanging drop consisting of equal volumes of protein solution (6 mg/ml of purified MthK RCK domain, 250 mM NaCl, 20 mM Tris, pH 8.0) and crystallization buffer (28% polyethylene glycol (PEG) 300, 100 mM BaCl₂, 100 mM Tris pH 8.5). This crystal was mounted in a nylon loop and rapidly frozen in liquid N₂ without additional cryoprotection, and then stored in liquid N₂ until the time of diffraction data collection.

X-ray data collection and analysis

Diffraction data were collected at beamline X6A at the National Synchrotron Light Source, at -180°C under a nitrogen stream, at a wavelength of 1.1 Å. Data were processed and scaled to 3.1 Å using HKL2000, and phasing was done by molecular replacement with Phaser (McCoy et al., 2007; Otwinowski and Minor, 1997). The best scoring molecular replacement solution used the structure of the “closed” MthK gating ring (PDB accession no. 2FY8). This solution contained two complete gating rings (16 RCK domain chains) per asymmetric unit. Examination of electron density maps and additional model building was performed using COOT (Emsley and Cowtan, 2004). The model was refined by positional and isotropic B-factor refinement in Phenix (Adams et al., 2010). To increase the observation/parameter ratio, 16-fold torsion-angle non-crystallographic symmetry (NCS) restraints were included in refinement and found to significantly decrease the R_free. Additionally, from the early stages of model building, the model was divided into 50 Translation/Libration/Screw (TLS) groups, and TLS refinement was performed using Phenix. Bias in the assignment of the R_free test set (arising from the high NCS) was minimized by use of thin resolution shells in test set selection, also performed in Phenix.refine. The final model contains 27,716 protein atoms (3,557 residues), 57 Ba²⁺ ions, and 32 water molecules and was refined to final working and free R values of 20.9% and 24.3%, respectively, at 3.1 Å resolution (Table 1). Ramachandran plot analysis showed 92.2% in most favored regions, 7.2% in additional allowed regions, 0.6% in additional allowed regions, and 0.0% in disallowed regions. Figures were prepared using PyMOL, and pairwise structural alignments were performed using the “align” command in PyMOL, as described in Supplemental Methods.

Channel purification and reconstitution

The full-length MthK channel was expressed in E.coli XL-1 Blue cells on induction with 0.4 mM IPTG, and purified by metal affinity chromatography as described previously (Pau et al., 2010). Channels were reconstituted into liposomes composed of E.coli lipids (Avanti) (Heginbotham et al., 1998), and the proteoliposomes were rapidly frozen in liquid N₂ and stored at -80°C until use. Mutations were generated using the QuickChange Site-directed Mutagenesis Kit (Stratagene) and confirmed by DNA sequencing.

Electrophysiology

Recordings were obtained using planar lipid bilayers of POPE:POPG (3:1), in a horizontal bilayer chamber, at 22-24°C. Solution in the cis (top) chamber contained 200 mM KCl and 10 mM HEPES, pH 7.0; solution in the trans (bottom) chamber contained 200 mM KCl, with either 10 mM HEPES (to buffer solutions to pH 7.3 or 7.7) or 10 mM MES (to buffer solutions to pH 6.5), and MgCl₂, CdCl₂, CaCl₂, SrCl₂ or BaCl₂ at the indicated concentrations. Recordings analyzed for this work were obtained at -100 mV transmembrane voltage, though currents through each bilayer were typically observed at additional voltages to confirm the reversal potential and other MthK channel properties. Within each bilayer, multiple solution changes were performed using a gravity-fed perfusion system, and to ensure completeness of solution changes, the trans chamber was washed with a minimum of ten chamber-volumes of solution during each solution change.

Currents were amplified using a Dagan PC-ONE patch clamp amplifier with low-pass filtering to give a final effective filtering of 1 kHz (dead time of 0.18 ms), and sampled by computer at a rate of 10 kHz. Currents were analyzed by measuring durations of channel openings and closings at each current level by 50% threshold analysis, using pClamp9. These measurements were used to quantify channel activity as $\mathrm{NPo}={\sum }_{i=1}^{n}i{P}_i$, where i is the open level and P_i is the percent of time at that level. The mean channel open probability (Po) is then calculated by dividing NPo times N, which is the number of active channels in the bilayer, determined by recording under conditions where maximal channel opening is observed. Data are presented as mean ± SEM. Electrophysiological results are based on observations from a minimum of three different experiments for each recording condition, and includes a total of 264 individual data sets from 45 different lipid bilayer patches.

Because we could observe the effect of fast Ba²⁺ blockade (i.e. Ba²⁺-dependent reduction of single-channel current amplitude) in our recordings, we wondered whether slow Ba²⁺ blockade might also occur. Slow Ba²⁺ blockade can produce long closed events of several seconds in duration, which would have a substantial impact on Po (Rothberg et al., 1996; Thomson and Rothberg, 2010). To estimate the potential impact of slow Ba²⁺ blockade, we activated MthK channels to Po of >0.9 using 15 mM Ca²⁺, and then changed the solution to include both 15 mM Ca²⁺ and 50 mM Ba²⁺ (Figure S5). In three different experiments of this type (ranging from 2.5 to 8 min of continuous recording for each experiment), we observed 0 closings >1 sec, suggesting that slow Ba²⁺ blockade has a nominal impact on Po in recordings at -100 mV.

Research Highlights.

• We present the structure of a Ba²⁺-bound RCK gating ring at 3.1 Å.

• Ba²⁺ binds to the RCK domain at a site in common with a known Ca²⁺-activation site.

• Ba²⁺ can substitute for Ca²⁺ to activate reconstituted MthK channels.

• These results suggest a working hypothesis for ligand-activation of MthK channels.