Structural Snapshots of Intermediates in the Gating of a K⁺ Channel

Abstract

Regulation of ion conduction through the pore of a K⁺ channel takes place through the coordinated action of the activation gate at the bundle crossing of the inner helices and the inactivation gate located at the selectivity filter. The mechanism of allosteric coupling of these gates is of key interest. Here we report new insights into this allosteric coupling mechanism from studies on a W67F mutant of the KcsA channel. W67 is in the pore helix and is highly conserved in K⁺ channels. The KcsA W67F channel shows severely reduced inactivation and an enhanced rate of activation. We use continuous wave EPR spectroscopy to establish that the KcsA W67F channel shows an altered pH dependence of activation. Structural studies on the W67F channel provide the structures of two intermediate states: a pre- open state and a pre-inactivated state of the KcsA channel. These structures highlight key nodes in the allosteric pathway. The structure of the KcsA W67F channel with the activation gate open shows altered ion occupancy at the second ion binding site (S2) in the selectivity filter. This finding in combination with previous studies strongly support a requirement for ion occupancy at the S2 site for the channel to inactivate.

Graphical Abstract

Introduction

K⁺ channels selectively conduct K⁺ ions across biological membranes [1, 2]. They play important roles in key physiological processes such as electrical excitability, maintenance of the membrane potential, secretion of hormones and sensory transduction [2]. The ion pathway in a K⁺ channel is contained within the pore domain, which shows a conserved architecture in the K⁺ channel family (Fig. 1 A). Selectivity for K⁺ and regulation of ion flux, factors that underlie the biological roles of K⁺ channels both take place at the pore domain [1]. Selectivity for K⁺ is accomplished at the selectivity filter which corresponds to the narrowest part of the ion pathway (Fig. 1A) [3, 4]. The selectivity filter consists of four K⁺ coordination sites (labelled S1-S4 from the extracellular side) that are built using the main chain carbonyl oxygen atoms and the threonine side chain of the conserved sequence TVGYG. The structure of the selectivity filter is highly conserved in the K⁺ channel family [5].

Figure 1. Activation and Inactivation in the KcsA channel.

A) Structure of the wild-type (WT) KcsA channel (PDB: 1k4c). Two opposite subunits of the KcsA channel tetramer are shown in cartoon representation with the selectivity filter (residues T75–G79) depicted as sticks and K⁺ ions bound to the selectivity filter shown as spheres. Inset: Close-up of the pore helix and the selectivity filter. The side chains of W67, E71 and D80 and H-bond interactions of these side chains are shown. Side chains of other residues in this region are not shown for clarity. B) Sequence alignment of the pore helix and selectivity filter region of the KcsA (gi: 61226909), K_vAP (gi: 14601099), Shaker (gi: 13432103) and the K_v1.2 channel (gi: 1235594). C) Inactivation on the KcsA channel. Macroscopic currents for the KcsA channels elicited by a rapid change in pH from 8.0 to pH 3.0. D) The fraction of the current remaining at 10 sec into the pH pulse (I_f) to the peak current (I_o) is shown as a bar graph. E) Activation of the KcsA channel. An expanded time scale of the macroscopic currents of the KcsA channels show a rapid increase in current due to channel activation by a change in pH from 8.0 to 3.0. F) The time constant for activation, determined by a single exponential fit to the rising phase of the current, plotted as a bar graph. Current traces shown in C and E were recorded under symmetrical 200 mM K⁺ at +80 mV. For panels D and F, the values of individual measurements are displayed as circles, error bars correspond to SD and number of independent patches is indicated in parentheses.

Regulation of ion flux takes place through two process, activation and inactivation. Activation initiates ion flux through the channel while inactivation terminates the ion flux [6]. These processes take place at distinct sites in the pore domain. The activation gate is towards the cytoplasmic side and is formed by the bundle crossing of the inner helices [7–9]. In the closed state, the bundle crossing obstructs the movement of ions. During channel activation, there is a hinged movement of the inner helices that widens the bundle crossing and permits the flow of ions. The inactivation gate is at the selectivity filter. During inactivation, there are structural changes that convert the selectivity filter from a conductive to a non-conductive conformation [5, 10]. The activation gate and the selectivity filter are allosterically coupled with the opening of the activation gate favoring inactivation at the filter and closure of the activation gate favoring recovery of the filter from inactivation [11].

The process of channel activation and inactivation have been extensively investigated in the family of voltage gated K⁺ (K_v) channels [6, 10, 12]. Understanding these processes has also been greatly aided by studies on KcsA, a bacterial K⁺ channel. KcsA is a pH gated channel that only consists of a pore domain [3, 13]. Activation and inactivation in the KcsA channel share functional similarities to the corresponding processes in K_v channels [14]. The KcsA channel provides a tractable system for obtaining structural and dynamic information on the gating states of the K⁺ channel pore. Structures are presently available for the KcsA channel with the activation gate closed or open and for the KcsA channel with the selectivity filter in the conductive or in a constricted conformation [4, 15]. It has been proposed that the constricted state of the selectivity filter corresponds to the inactivated state [16–18], but other studies have suggested that the constricted state corresponds to a deep-inactivated state that the selectivity filter transitions to from the inactivated state [19, 20].

In this study, we provide structural information on two intermediate states in the gating of the KcsA channel. These studies were carried out on a W67F mutant of the KcsA channel. W67 is in the pore helix of the KcsA channel and forms an H-bond with D80 of the selectivity filter (Fig. 1A). The W67 residue is highly conserved in K_v channels (Fig. 1B) [21]. Studies on the Shaker K⁺ channel have shown that substitution of this Trp residue by Phe (W434F) dramatically increases the rate of C-type inactivation [22, 23]. The W434F mutant has been extensively used in studies that require a non-conducting channel such as gating charge measurements in the Shaker channel [24]. The substitution of W67 by F in the KcsA channel also perturbs inactivation however results in reduced inactivation unlike the rapid inactivation seen for the equivalent substitution in Shaker and other eukaryotic K_v channels [25]. The difference in the effect of the W to F substitution on inactivation in the KcsA and the Shaker K_v channels suggest possible differences in the mechanism of inactivation in these channels.

Here we report structures of the W67F KcsA channel in two distinct states, a pre-open state and a pre-inactivated state. We also report on the functional effects of the W67F substitution and the pH dependence of channel opening. Our studies provide snapshots of the gating process in a K⁺ channel and highlight the allosteric pathway coupling the activation gate to the selectivity filter. Our studies also indicate a role for ion occupancy at the S2 site in the selectivity filter in inactivation in KcsA.

Results

The W67F substitution alters gating in the KcsA channel.

The W67F substitution in the KcsA channel is reported to severely perturb inactivation [25]. To confirm, we recombinantly expressed and purified the W67F channel, reconstituted it into lipid vesicles and used liposome-patch clamping to measure channel activity. Macroscopic currents of the KcsA channel are elicited by a pH jump from 8.0 to 3.0 on the cytoplasmic side [26]. On a pH jump, the wild type (WT) channels quickly activate as indicated by a rapid increase in current and then inactivate as indicated by the decay in current (Fig. 1C, D). Macroscopic currents for the W67F channel, following activation, showed a very slow decay in current indicating substantially reduced inactivation. We also observed a faster rate of activation for the W67F channels compared to the WT (Fig. 1E, F). The rate of activation is determined by an exponential fit to the increasing phase of the current following a pH jump [14]. Our measured rate of activation for the W67F channel was ~3–4 fold faster than the WT. These studies indicate that the W67F substitution alters gating in the KcsA channel and that the substitution affects both activation and inactivation.

A pre-open structure of the KcsA-W67F channel.

Towards understanding the altered gating properties, we determined the structure of the KcsA W67F channel. We initially determined the structure of the KcsA W67F channel under conditions that favor the closed activation gate. The KcsA W67F channel was well expressed and showed a similar size exclusion profile to the WT channel (Supplementary Fig.1A). Structures of the KcsA channel are determined in complex with a Fab [4]. For the W67F channel, we observed that the addition of Fab caused a dissociation of the channel tetramer (Supplementary Fig. 1A). We therefore explored strategies to stabilize the KcsA W67F tetramer and determined that the addition of the channel blocker TBA stabilizes the W67F channel tetramer on Fab binding. TBA binds to the central cavity of the channel and structural studies on the WT KcsA channel show that TBA does not perturb the structure of the channel or alter the ion occupancy at the selectivity filter sites (Supplementary Fig. 1B) [27].

We obtained crystals of the KcsA W67F-Fab in the presence of TBA that diffracted to 2.6 Å. The crystals were in the P4 space group with the asymmetric unit consisting of two copies of the channel subunit +Fab (referred to as Molecule I and II). The structure was determined by molecular replacement using a WT KcsA subunit+ Fab (pdb: 1k4c) as the search model (Table 1). The structures of molecule I and II are very similar (Supplementary Fig. 2). Molecule I is used for our analysis.

Table 1:

Data collection and refinement statistics

	W67F+TBA+Fab Pre-open	W67F+Fab Open	WT+Fab Open
Space group	P4	I4	I 4
a (Å)	154.32	155.25	134.8
b (Å)	154.32	155.25	134.8
c (Å)	74.24	74.42	72.15
Data collection
X-ray source	APS 23ID-B	APS 23ID-D	ALS 4.2.2
Wavelength (Å)	1.0	1.0	1.0
Resolution range (Å) (Highest res. shell)	48.80–2.64 (2.73–2.64)	47.60–2.72 (2.85–2.72)	47.66–3.20 (3.51–3.2)
Collected reflections	375568 (30171)	162627 (23991)	80448(18122)
Total /Unique	54531 (4451)	21473 (3166)	10831 (2566)
Completeness (%)	99.9 (100)	100 (100)	100 (100)
Multiplicity	6.9 (6.8)	6.8 (6.8)	7.4 (7.1)
I/σ(I)	9.1 (1.0)	8.1 (1.0)	8.8 (1.0)
CC ½	99.8 (40.3)	99.4 (32.2)	99.7 (37.8)
Refinement statistics
R -work (%)	0.23	0.22	0.24
R-free (%)	0.26	0.25	0.29
Clash score	4.79	3.43	8.78
Ramachandran plotb
Most favored (%)	95.34	95	96
Disallowed (%)	0.48	0.2	0.96
Rmsd bonds (Å)	0.02	0.03	0.002
Rmsd angles (deg)	0.51	0.50	0.57
PDB code	7M2H	7M2I	7M2J

Numbers in parentheses are statistics for the last resolution shell.

^bPerformed in MolProbity.

Rmsd, root-mean-square deviation.

Electron density corresponding to the pore helix and the selectivity filter is shown (Fig. 2A, B). Superposition of the W67F channel to the WT did not reveal any major structural changes in the pore helix and the selectivity filter regions (Fig. 2C). The structural comparison surprisingly showed changes in TM1 and TM2 helices with an outward-anti clockwise movement in the lower halves of the TM2 (12°) and the TM1 (10°) helices (Fig. 2D). Evaluation of the root mean square differences in the backbones of the TM1 and the TM2 helices of the W67Fmutant and the WT indicated that the movement of the helices in the W67F mutant took place around hinge points at residues 40–41 in TM1 and 99–100 in TM2 (Supplementary Fig. 3). This movement of the TM1 and TM2 helices in the W67F channel was also accompanied by changes in the side chain conformations along these helices. A major change in the side chain conformation was observed at F103 along with smaller changes at T101, L105 and T112 (Fig. 3A). F103 has been proposed to be a major node in the allosteric pathway coupling the activation gate to the selectivity filter [28–30]. Substitution of F103 with Ala results in a non-inactivation phenotype [28]. The conformation of the F103 side chain in the W67F structure is similar to the conformation of this sidechain in KcsA channel structures with the activation gate open (Fig. 3B) [28, 29].

Figure 2. Structure of the KcsA W67F channel in the pre-open state.

A) Electron density of the selectivity filter of KcsA W67F channel. The 2F_o-F_c electron density map contoured at 2.2 σ is shown with residues 71–80 as sticks, and the K⁺ ions in the selectivity filter are shown as purple spheres. B) Close-up view of the electron density corresponding to key residues in the pore helix and the selectivity filter. 2F_o-F_c electron density map contoured at 2.2 σ is shown. C) Superposition of the selectivity filter and pore helix residues 67–80 of KcsA W67F channel (blue) and the WT (green, pdb: 1k4c). D) Superposition of the KcsA WT channel in the closed state (1k4c) and the W67F channel in the pre-open state is shown. Two opposite subunits shown with the direction of movement of the inner and the outer helices between the closed and the pre-open state indicated by arrows.

Figure 3. Comparison of the closed and pre-open states of KcsA.

A) A comparison of the closed (WT) and the pre-open (W67F) structures shows side chain conformational changes along the inner helix. Residues showing significant conformational changes are shown in van der Waals representation. B) The flip of the F103 side chain with the movement of the inner helix is shown by superposition of the KcsA closed (WT), pre-open (W67F) and the open structures (E71A, pdb: 5vk6). C) HOLE profiles of the closed (WT, pdb: 1k4c, green) and the pre-open (W67F, blue) KcsA channel shows different constriction points for ion permeation, with the residues T107, A111 forming the constriction in the closed state and F103 forming the constriction in the pre-open state.

In spite of these changes, the structure of the W67F channel still corresponds to a closed pore as indicated by analysis using the HOLE program (Fig. 3C) [31]. This analysis does indicate different constriction points in the W67F and the WT channels. Due to the rotation of the side chain, the F103 residue forms the major constriction point in the W67F channel while the residues at the bundle crossing (T107, A111) act to restrict ion movement through the pore of the WT KcsA channel in the closed state.

The changes seen in the W67F mutant both in terms of the splaying of the TM helices as well as the changes in the F103 residue (and other side chains along TM2) indicates a partial movement of the helices towards the open state. The process of channel opening is proposed to involve transition through multiple closed states before a concerted conformational change leading to the open state [14]. We propose that the structure of the W67F mutant represents one of the intermediates along this pathway and therefore, corresponds to a “pre-open” state.

Effect of TBA and Fab on the KcsA W67F channel structure.

TBA was used to stabilize the W67F channel -Fab complex for structure determination, which raises a question on whether the changes seen are due to the presence of TBA. In the WT channel, no changes are observed on a comparison of the structures without and with TBA present (Supplementary Fig. 1B) [27]. We have previously observed that the Y82A KcsA channel, similar to the W67F mutant, dissociates partially on Fab binding (Supplementary Fig. 4A). The Y82A substitution dramatically enhances C-type inactivation.[15, 32] We tested the effect of TBA and observed that TBA was capable of stabilizing the Y82A tetramer on the binding of Fab (Supplementary Fig. 4B). We determined the structure of the Y82A channel bound to Fab in the presence of TBA (Supplementary Fig. 4C, Supplementary Table 1). The structure of the Y82A mutant channel was very similar to the structure of the WT KcsA channel. Importantly, the changes observed in the W67F channel such as the altered positioning of the inner helix or the conformational changes in residues along the inner helix were not observed in the Y82A+TBA structure (Supplementary Fig. 4D). These results therefore argue against the presence of TBA as the major reason for the structural changes observed in the W67F pre-open structure, though minor effects of TBA on the W67F channel structure cannot be ruled out.

The W67F KcsA channel tetramer dissociates on binding Fab. A close comparison of the channel-Fab interface in the WT and the W67F channel did not reveal any major structural changes in this interface (Supplementary Fig. 5A). We see a shift in the F103 side chain in the W67F structure compared to the WT (Supplementary Fig. 5B). A F103A substitution has previously been shown to stabilize the E71H and the Y82A channels on Fab binding.[15, 28] We investigated the effect of an Ala substitution at F103 and observed that the W67F/F103A channel, unlike the W67F channel, was stable on the binding of Fab (Supplementary Fig. 5C). This result indicates that destabilization of the W67F KcsA tetramer on Fab binding is linked to the changes at F103.

Changes in pH activation in KcsA W67F.

We investigated if the structural changes observed at the activation gate in the W67F mutant affect the pH dependence of channel opening. Towards this goal, we used EPR spectroscopy to compare the opening of the activation gate in the W67F mutant to the WT channel [8, 28]. For these experiments, we substituted G116 with Cys and introduced a spin label by modifying the Cys residue with (1-oxyl-2,2,5,5-tetramethylpyrrolidin3-yl) methyl methanethiosulfonate (Fig. 4A). The spin label at G116C has previously been used to report on the opening of the activation gate [8, 28]. When the activation gate is closed, the spin labels on the channel subunits are in close proximity, which result in a broadening of the EPR spectra by dipolar relaxation. Opening of the activation gate on lowering the pH increases the distance between the spin labels and thereby decreases the spectral broadening. We measured the EPR spectra at different pH values and monitored the change in the spectra from the change in the amplitude of the central resonance line (Fig. 4B). Fitting the change in amplitude with pH using a Hill equation gave pK_a values of 4.18 ± 0.13 (mean ± SD, n= 3, Hill coefficient 1.6 ± 0.4) for the WT channel and 4.51 ± 0.08 (1.3 ± 0.1) for the W67F mutant (Fig. 4C). The higher pK_a value observed for activation gate opening in the W67F mutant indicates an easier opening of the activation gate in the W67F mutant compared to the WT and is consistent with the faster activation observed in the W67F channel.

Figure 4. pH activation in WT and W67F KcsA channel probed by spin labelling and CW-EPR spectroscopy.

A) Cartoon representation of the KcsA channel in the closed (pdb: 1k4c) and the open states (pdb: 5vk6). Two opposite subunits shown. G116C, the site of modification by the spin label is shown as a yellow sphere. B) CW-EPR spectra for the WT and the W67F channel over a pH range of 3–7. C) The amplitude of the normalized CW-EPR spectrum’s central resonance line (normalized by the number of spin) was plotted versus the proton concentration [H⁺], and the Hill equation was fitted to the data. The pK_a for activation for the WT and the W67F channel were 4.18 ± 0.13 and 4.51 ± 0.08 (± SD, n = 3) respectively. Error bars correspond to the SD for n = 3 or the range of values at the pH 5.0 data point (n = 2).

Structure of the KcsA W67F channel in the open state.

Next, we determined the structure of the W67F channel with the activation gate open. Cuello et al have previously reported the use of a disulfide crosslink to trap the activation gate open for the structure determination of this state [15]. Using this approach, we obtained crystals (I4 form) of the locked open W67F mutant that diffracted X-rays to 2.72 Å and determined the structure (Fig. 5A, Table 1). The electron density map corresponding to the pore helix and the selectivity filter region of the W67F mutant channel is shown (Fig. 5B, C). The W67F mutant shows the selectivity filter in the conductive state, as anticipated based on the functional studies that show substantially reduced inactivation in this mutant. For comparison, we also determined a structure of the WT KcsA channel trapped with the activation gate open. The structure of the WT channel was at a lower resolution (3.2 Å) compared to the W67F mutant but clearly shows a constricted conformation for the selectivity filter (Fig. 5D, Table 1).

Figure 5. Structure of the KcsA W67F channel in the open state.

A) Superposition of the KcsA W67F channel in the closed state (blue) and the open state (yellow) is shown. Two opposite subunits are shown. B) Electron density of the selectivity filter of KcsA W67F channel in the open state. The 2F_o-F_c electron density map contoured at 2.5 σ is shown with residues 71–80 as sticks, and the K⁺ ions in the selectivity filter are shown as purple spheres. C) Close-up view of the electron density corresponding to key residues in the pore helix and the selectivity filter. 2F_o-F_c electron density map contoured at 2.5 σ is shown. D) The constricted selectivity filter observed in the KcsA WT channel in the open state. 2F_o-F_c electron density map for the selectivity filter contoured at 1.8 σ is shown with residues 71–80 as sticks, and the K⁺ ions in the selectivity filter are shown as purple spheres. E) Superposition of the selectivity filter and pore helix (residues 67–80) for the KcsA W67F channel in the pre-open (cyan) and the open state (yellow). Only a single subunit is shown. F) Change in the rotameric state and the hydrogen bond interactions of E71 in the KcsA W67F channel in the pre-open and the open state. The water molecule present is indicated by a red sphere.

A comparison of the selectivity filter and the surrounding regions in the W67F channel in the open and the pre-open state shows two characteristic differences: a shift of the E71 side chain to a distinct rotameric state and a loss of ion occupancy at the S2 site (Fig. 5B, E). The rotameric switch of the E71 side chain results from the breaking of an H-bond between the side chain carboxyl group and the 2’ amide bond of the selectivity filter (Fig. 5F). This rotamer of the E71 side chain is also observed in KcsA structures with the selectivity filter in the constricted state as seen in the structure of the WT KcsA channel with the activation gate open. The constricted selectivity filter is also observed in structures of the KcsA channel determined at low K⁺ concentrations and in KcsA mutants with specific amino acid substitutions in and around the selectivity filter [4, 15, 33, 34]. In this conformation, the S2 and the S3 ion binding sites are disrupted which renders the channel non-conductive. The observation that the rotameric switch of the E71 side chain is associated with the constricted selectivity filter suggests that the rotameric shift in the E71 side chain on the opening of the activation gate imposes a strain on the selectivity filter, which destabilizes the conductive state and drives the filter into the constricted state.

Another characteristic feature of the W67F open structure is the loss of ion occupancy at the S2 site (Fig. 5B, 6). Consistent with the changes at the S2 site, we observed a lower single channel conductance for the W67F mutant compared to the WT (Supplementary Fig. 6). We also investigated if the loss of ionic occupancy at the S2 site influences ionic selectivity. We probed ionic selectivity using a K⁺ flux assay and observed no changes in ionic selectivity in the W67F mutant compared to the WT channel (Supplementary Fig. 7).

Figure 6. Ion distribution in the selectivity filter.

K⁺ binding to the selectivity filters in the KcsA W67F- open, W67F pre-open, Y78-ester (pdb: 4msw) and G77-ester (pdb: 5ebw) is shown. All datasets were scaled to 2.7 Å and the F_o-F_c omit maps contoured at 4 σ is shown.

A loss of ion occupancy at the S2 site was previously observed for the Y78- and G77-ester mutants of the KcsA channels that carry an amide to ester substitution in the protein backbone of the selectivity filter (Fig. 6) [20, 35]. The Y78- and G77-ester mutant channels also show substantially reduced inactivation as observed for the KcsA W67F channel. These observations provide a strong link between ion occupancy at the S2 site and the ability of the selectivity filter to inactivate. It has been proposed based on molecular dynamic simulations that the lack of inactivation in the Y78- and G77-ester mutants may be due to a disruption of the interaction of water molecules with the protein backbone of the selectivity filter during inactivation [36]. This is proposed to arise from the lack of a H-bond between the protein backbone and water molecules due to the replacement of the amide N-H in the WT channel by an oxygen atom in the ester mutants. There are no changes in the protein backbone of the selectivity filter in the W67F channel, which indicates that the lack of ion occupancy at the S2 site underlies reduced inactivation in this case and for the Y78- and G77-ester mutants.

The lack of ion occupancy at the S2 site explains why the rotameric shift of the E71 side chain on the opening of the activation gate does not drive the filter into the inactivated and the constricted or the deep-inactivated state. We suggest that the selectivity filter in the W67F open structure, which shows a rotameric shift of the E71 side chain but a conductive state of the selectivity filter, corresponds to the structure of the selectivity filter in a “pre-inactivated” state. These studies indicate that the E71 side chain and the S2 site in the selectivity filter are two key nodes in the allosteric mechanism coupling the activation gate to the selectivity filter.

Discussion

Here, we report structural and functional studies on the W67F mutant of the KcsA channel. We confirm that the W67F substitution perturbs inactivation in KcsA. We find faster activation and changes in the pH dependence of activation in the W67F channel compared to the WT. We report structures of the W67F channel with the activation gate in the pre-open and the open states. With the activation gate open, we find a loss of ion occupancy in the S2 site in the selectivity filter which confirms a role for ion occupancy at the S2 site in determining the propensity for the selectivity filter to inactivate.

Structures of the KcsA channel have previously been reported with the activation gate at various stages of opening. Channel opening in the KcsA channel takes place through a counterclockwise motion of the TM1 and TM2 helices [8, 37]. The pre-open W67F structure provides an intermediate point in the counter clockwise trajectory of the TM1 and the TM2 helices between the closed and the open states. The pre-open structure of the W67F channel shows a flip of the F103 side chain compared to the WT closed channel while the open structures show similar positioning of the F103 side chain (Supplementary Fig. 8). This indicates that changes in F103 side chain takes place in the very initial stages of the opening of the activation gate. The studies confirm that the F103 residue is a key node in the allosteric coupling in the KcsA channel. Further, the changes in the activation gate in the W67F structure were observed as a result of a substitution (W67F) in the vicinity of the selectivity filter. We also measured an increase in the rate of activation and a change in the pH dependence of opening in the W67F channel. These results emphasize the allosteric coupling between the activation gate and the residues surrounding the selectivity filter.

The W67F structure with the activation gate open suggests that the E71side chain and the S2 site are two other nodes in the allosteric coupling of the activation gate to the selectivity filter. A disruption at either of these nodes affects channel inactivation. In the E71A mutant, for example, the opening of the activation gate does not strain the selectivity filter due to the substitution of the E71 side chain and the E71A mutant therefore affects inactivation [15, 32]. Similarly, mutations that effect ion occupancy at the S2 site such as the W67F, Y78- and G77-ester substitutions affect inactivation due to perturbation of ion occupancy at the S2 site.

It is presently not obvious from the structure how the W67F substitution effects ion occupancy at the S2 site. The effect on ion occupancy at the S2 site is observed when the activation gate is open while the S2 site is occupied when the activation gate is closed. These results are broadly consistent with NMR and thermal denaturation studies which show that the opening of the activation gate changes ion binding affinity at the selectivity filter.[30, 38] However, measurements of ion binding affinity to the KcsA channel using Isothermal titration calorimetry (ITC) did not show a change in affinity for K⁺ with the opening of the activation gate. [39] It should be noted that the NMR and ITC measurements do not report on the ion binding affinity of specific sites in the selectivity filter. While it is not known as to why ion occupancy at the S2 site is required for filter inactivation in the KcsA channel, a similar effect has been reported in other channels. A Y199 ester substitution in the selectivity filter the K_vAP channel, which is expected to perturb ion occupancy at the S2 site, results in reduced inactivation [35]. In the MthK channel, it has been proposed that the lower ion occupancy at the S2 site is the basis for the difference in the structural transitions with K⁺ concentrations for the structurally similar selectivity filters in the MthK and the KcsA channels [40]. These studies further support an important role for ion occupancy at the S2 site for the K⁺ channel selectivity filter to transition into the inactivated state.

Our studies highlight three nodes, F103, E71 and the S2 site in the selectivity filter in the allosteric pathway coupling the activation gate to the selectivity filter. Other studies have identified residues T74, T75 and I100 as also participants in this allosteric pathway [29, 30, 33]. Combining the results of this investigation with the previous studies on the gating in the KcsA channel suggest the following pathway. The opening of the activation gate causes a flip in the F103 residue. This flip in the F103 side chain exerts a force on the lower half of the selectivity filter/ pore helix. This transmission involves I100 in the inner helix and T74 and T75 at the base of the pore helix/selectivity filter. The force imposed on the pore helix due to the opening of the activation gate causes a rotameric shift in the E71 side chain. The rotameric shift in the E71 side chain strains the selectivity filter thereby rendering the selectivity filter in the open state to be a metastable conformation that rapidly transitions to the inactivated and subsequently the deep-inactivated or the constricted state. The strain on the filter due to the rotameric shift of the E71 side chain can be relieved by the loss of ion occupancy specifically at the S2 site. Alternately the strain on the selectivity filter can also be relieved by perturbations in the interactions of the selectivity filter with the surrounding residues. The structures of the intermediates that we have determined in this study support the temporal order with the switch of the F103 side chain taking place in the initial stages of the opening of the activation gate while the rotameric switch of the E71 side chain takes place in the later stages.

There are key points in the allosteric pathway that have to be further explored. These include determining how the interactions surrounding the selectivity filter influence the ion occupancy at the selectivity filter sites and why a change in ion occupancy at the S2 site stabilizes the conductive state of the selectivity filter. We anticipate that structure determination of additional intermediate states and studies using computational approaches will help deduce the mechanistic details of this allosteric pathway, which is critical for K⁺ channel function.

Materials and Methods

Protein expression, purification and crystallization.

The KcsA constructs used in this study were expressed from a pQE60 vector in E. coli XL-10 cells [3]. The W67F and other amino acid substitutions in the KcsA constructs were generated by Quickchange mutagenesis. The KcsA construct used for the determination of the structure of the open state carried a 1–20 deletion at the N-terminus and the H25Q, A28C, R117Q, E118C, E120Q, R121Q and R122Q amino acid substitutions [15]. The XL-10 transformants of KcsA W67F were grown in LB medium at 37 °C with 100 μg/mL ampicillin to an OD₆₀₀ of 1.0. Protein expression was induced by the addition of 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) and cells were grown for an additional 4 hours at 37 °C. XL-10 transformants of the open mutants of KcsA WT and the W67F channels were grown in LB medium with 0.2% (w/v) glucose to an OD₆₀₀ of 0.6. The cultures were then supplemented with 10 mM BaCl₂, protein expression was induced by the addition of 0.4 mM IPTG and further grown at 25 °C for 16–18 hours. Cells were harvested by centrifugation at 4000g at 4 °C and re-suspended in Tris-lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM KCl, 250 mM sucrose, 1 mM MgCl₂). Membranes were prepared, solubilized by decyl-β-d-maltoside (DM) and the KcsA channels were purified by Co²⁺-affinity chromatography as previously described [41]. The C-terminal 35 residues were removed by treatment with chymotrypsin at 1:15 ratio (chymotrypsin: protein) for 2 hours at room temperature. The chymotrypsinized channels were purified by size exclusion chromatography (SEC) in 50 mM Tris-HCl, 150 mM KCl, 0.25% DM, pH 7.5 (SEC buffer). For crystallization, the KcsA channels were complexed with a Fab fragment from KcsA IgG. The KcsA -Fab complexes were purified by SEC and concentrated to 10.0 OD units/ml for crystallization. Tetra butyl ammonium iodide (10 mM) was added to the SEC buffer used for the purification of the W67F and Y82A KcsA channels and the complexes with Fab. For the open structures, the Cys mutant channels following purification were already in the cross-linked state. An additional cross-linking strategy such as treatment with an oxidizing reagent was not required. Due to cross-linking of the subunits, the W67F open channels were stable as a tetramer and addition of TBA was not necessary. KCl was added to a final concentration of 300 mM prior to crystallization of the open mutant channels.

Crystallization and structure determination.

KcsA W67F+TBA-Fab and Y82A+TBA-Fab crystals were grown by mixing 1 μL of the protein with 1 μL of well solution (25–28% polyethyleneglycol (PEG) 400 (v/v), 50 mM magnesium acetate, 50 mM MES pH 6.0– 6.5)[4]. The W67F open mutant was crystallized using a well solution of 30–32% PEG400, 50 mM magnesium acetate, 50 mM MES pH 6.2. KcsA WT open mutant was crystallized using a well solution of 33% PEG 400, 100 mM MES pH 6.5–6.75, 100 mM calcium chloride. Crystals were cryo-protected by increasing the PEG400 concentration in the reservoir to 42% (v/v) and were flash frozen in liquid nitrogen.

Data were collected either at ALS beamline 4.2.2 or at APS beamline 23ID-B. The X-ray data were integrated and scaled using XDS [42]. The structures were solved by molecular replacement using Phaser with the WT structure (PDB: 1K4C) as the search model [43]. For the closed KcsA W67F dataset, KcsA residues 67 and 74–79, lipids, water and ions were omitted from the search model while KcsA residues 82 and 74–79, lipids, water and ions were omitted from the search model used for the KcsA Y82 dataset. KcsA residues 22–34, 64–80, and 101–121, lipids, waters, and ions were omitted from the search model used for the KcsA W67F-open dataset. For the open KcsA WT dataset, KcsA residues 71 and 74–79, lipids, waters, and ions were omitted from the search model. The residues deleted from the model prior to molecular replacement were manually built using Coot [44]. The models were refined using automated refinement in Phenix and manual refinement using Coot [45]. X-ray data collection and refinement statistics are summarized in the Table 1. Pymol was used for preparing final figures [46] and HOLE program was used to analyze pore size of the channels [31].

Electrophysiology measurements.

The purified KcsA channels were reconstituted into soybean polar lipid vesicles at a 1:40–1:200 protein: lipid ratio as previously described (Matulef et al, 2016, 2018). Liposomes were dried down in a desiccator for 8–36 hours at 4 °C and then rehydrated in 200 mM KCl, 5 mM MOPS pH 7.0 for 2–16 hours at 12–22 °C. Giant unilamellar blisters were formed by adding the rehydrated liposomes into 200 mM KCl, 10 mM MOPS, 40 mM MgCl₂, 0.5 mM CaCl₂, pH 7.0. Inside-out patches were pulled from blisters using pipettes filled with 200 mM KCl, 0.5 mM CaCl₂, 10 mM succinate, pH 4. For recording channel activity, a RSC-160 rapid solution changer was used for rapid switching of the bath solution from 200 mM KCl, 10 mM MOPS pH 8.0 to 200 mM KCl, 10 mM succinate pH 3.0. On a pH jump the KcsA channels activate after a delay. As discussed by Chakrapani et. al., this delay is due to factors such as the position of the pipette with respect to the solution exchanger, the geometry of the pipette, the shape of the patch as well as the binding of H⁺ by other moieties such as lipids and buffer molecules. These factors govern the time it takes for H⁺ ions from the bulk solution to diffuse to the pH sensor residues on the channel.[14] This delay (~100–300 msec) was excluded from our analysis. Patches that took a longer time for perfusion were excluded from our analysis. The increase in current after the delay was fit by a single exponential to determine the time constant for activation.

EPR Spectroscopy.

KcsA channels with the G116C substitution were labeled with the spin probe (1-oxyl-2,2,5,5-tetramethylpyrrolidin-3-yl) methyl methanethiosulfonate (Toronto Research) at a 20: 1 probe-to-channel ratio for 30 min at room temperature followed by overnight incubation at 4 °C. The unreacted spin probe was then removed using size exclusion chromatography in 20 mM HEPES-KOH, 150 mM KCl, 0.25% DM. The spin-labeled protein was reconstituted into Soy polar lipid vesicles at a protein-to-lipid ratio of 1:500 as previously described [39]. Following reconstitution, the proteoliposomes were collected by ultracentrifugation at (200,000 g) for 45 minutes at 4 °C. Buffer exchange for changing pH was carried out using freeze thaw cycles and the vesicles were collected by ultracentrifugation before the spectra were collected.

Continuous-wave EPR spectra were collected at room temperature on a Bruker E500 X band spectrometer equipped with a dielectric resonator using the following parameters: microwave frequency of 9.79 GHz, modulation frequency of 100 kHz, incident power of 2 mW, and modulation amplitude of 1 G. EPR spectra were collected over a pH range of 3–7 in 15 mM citrate-phosphate and 150 mM KCl. The spectra were normalized by spin number. The amplitude of the central resonance line of the CW-EPR spectra vs. [H⁺] was plotted and fit using a Hill equation to determine the pK_a for activation.

Data Availability.

Atomic coordinates and the structure factors for the KcsA W67F+TBA pre-open, W67F pre-inactivated, WT-open and Y82A+TBA closed structures have been deposited in the PDB under accession codes 7M2H, 7M2I, 7M2J and 7RP0.

Highlights.

• Ion flux through a K⁺ channel is regulated by channel activation and inactivation.

• W67F in KcsA prevents inactivation and alters pH activation.

• Structures of KcsA W67F determined in the pre-open and the pre-inactivated state.

• Ion binding at second ion site in the selectivity filter required for inactivation.

• Insights into the allosteric pathway coupling activation and inactivation in KcsA.