Using protein backbone mutagenesis to dissect the link between ion occupancy and C-type inactivation in K⁺ channels

Significance

C-type inactivation is a gating process that takes place at the selectivity filter of K⁺ channels. C-type inactivation is important in regulating cellular excitability. A defining characteristic of C-type inactivation is a dependence on the permeant ion, but the underlying mechanism is not known. We use protein backbone mutagenesis to alter ion binding at specific sites in the selectivity filter and determine the effect on inactivation. We show that C-type inactivation is linked to ion occupancy at a specific site in the selectivity filter. This study underscores the utility of unnatural mutagenesis for investigating the mechanisms of channel function. Furthermore, permeant ions modulate function in many channel families; therefore, the approaches used in this study are generally applicable.

Abstract

K⁺ channels distinguish K⁺ from Na⁺ in the selectivity filter, which consists of four ion-binding sites (S1–S4, extracellular to intracellular) that are built mainly using the carbonyl oxygens from the protein backbone. In addition to ionic discrimination, the selectivity filter regulates the flow of ions across the membrane in a gating process referred to as C-type inactivation. A characteristic of C-type inactivation is a dependence on the permeant ion, but the mechanism by which permeant ions modulate C-type inactivation is not known. To investigate, we used amide-to-ester substitutions in the protein backbone of the selectivity filter to alter ion binding at specific sites and determined the effects on inactivation. The amide-to-ester substitutions in the protein backbone were introduced using protein semisynthesis or in vivo nonsense suppression approaches. We show that an ester substitution at the S1 site in the KcsA channel does not affect inactivation whereas ester substitutions at the S2 and S3 sites dramatically reduce inactivation. We determined the structure of the KcsA S2 ester mutant and found that the ester substitution eliminates K⁺ binding at the S2 site. We also show that an ester substitution at the S2 site in the K_vAP channel has a similar effect of slowing inactivation. Our results link C-type inactivation to ion occupancy at the S2 site. Furthermore, they suggest that the differences in inactivation of K⁺ channels in K⁺ compared with Rb⁺ are due to different ion occupancies at the S2 site.

Potassium channels are a ubiquitous family of integral membrane proteins that facilitate the selective conduction of K⁺ ions across cellular membranes (1). K⁺ selectivity is achieved by a structural element in the K⁺ channel pore called the selectivity filter (2). The selectivity filter consists of four sequential ion-binding sites (labeled S1–S4, from the outside to inside) that are built using protein backbone carbonyl oxygen atoms and the threonine side chain from the protein sequence T-V-G-Y-G (Fig. 1A) (4, 5).

Fig. 1.

Ester substitutions in the selectivity filter of the KcsA channel. (A) Close-up view of the selectivity filter of the wild-type KcsA channel [Protein Data Bank (PDB): 1K4C]. Two opposite subunits are shown in stick representation, and the K⁺ ions bound are shown as purple spheres. The amide bonds (1′–4′) and the ion-binding sites in the selectivity filter (S1–S4) are labeled. (B) Macroscopic currents for the KcsA channels were elicited at +100 mV by a rapid change in pH from 7.5 to 3.0. (C) Single-channel currents recorded at steady-state conditions at pH 3.0. The pH 3.0 solution is 10 mM succinate, 200 mM KCl, and the pH 7.5 solution is 10 mM Hepes–KOH, 200 mM KCl. The data for KcsA^WT are from ref. 3.

In addition to selective conduction of K⁺, the selectivity filter acts as a gate to regulate the flow of ions through the pore (6–8). During this gating process, conformational changes at the selectivity filter convert it from a conductive to a nonconductive state. In voltage-gated K⁺ (K_v) channels, this gating process is referred to as C-type inactivation (9). C-type inactivation is a physiologically important process as it plays a direct role in regulating neuronal firing and in pacing cardiac action potentials (8). The KcsA K⁺ channel from Streptomyces lividans undergoes an inactivation process that is functionally similar to C-type inactivation in a eukaryotic K_v channel (10–13). As the KcsA channel is easily amenable to structural studies, it has become an important model system for understanding the structure of the selectivity filter in the C-type–inactivated state and the forces that drive inactivation (14, 15).

One of the hallmarks of C-type inactivation is a dependence on the permeant ion (6, 7). The rate of C-type inactivation decreases when the K⁺ concentration is increased or when the permeant ion is changed from K⁺ to Rb⁺ (16, 17). Crystallographic studies on K⁺ channels have shown that a change in the permeant ion or its concentration results in changes in the ion occupancy at the binding sites in the selectivity filter (18, 19). For example, K⁺ and Rb⁺ at similar concentrations show different occupancies at the ion-binding sites, and the channel exhibits different rates of inactivation in K⁺ compared with Rb⁺ (3, 16, 20), which suggests a link between ion occupancy at the selectivity filter and inactivation (21, 22). The influence of permeant ions on inactivation has been proposed to arise from a “foot in the door”-like effect in which ion binding at a specific site prevents inactivation, similar to the presence of a foot in the doorway that prevents a door from closing (16, 23). The binding site responsible for the foot in the door effect is suspected to be at the extracellular side of the channel, but the exact location of the binding site, whether in the selectivity filter or at the extracellular mouth of the filter, is not known (6).

In this study, we investigate this link between ion binding at the selectivity filter and inactivation. The approach that we use is to alter ion binding at the selectivity filter sites and to determine the effect on inactivation. The S1–S3 ion-binding sites in the selectivity filter are constructed by backbone carbonyl oxygens. Therefore, conventional site-directed mutagenesis does not allow us to alter these sites. Instead, we use chemical synthesis and nonsense suppression approaches to introduce amide-to-ester substitutions in the protein backbone to perturb ion binding to specific sites in the selectivity filter (24, 25).

Amide-to-ester substitutions have previously been used to engineer the protein backbone for studies on protein stability and folding (26). Ester bonds are isosteric to amide bonds but have altered hydrogen-bonding properties and reduced electronegativity at the carbonyl oxygen (27). This reduction in the electronegativity of the carbonyl oxygen, by roughly one-half compared with an amide bond, perturbs ion binding to the selectivity filter. Amide-to-ester substitutions have previously been reported in the selectivity filters of the Kir2.1 and the KcsA K⁺ channels (28, 29). In the Kir2.1 channel, an ester substitution for the 3′ amide bond (see Fig. 1A for nomenclature) was found to reduce channel conductance and to produce distinct subconductance levels. In the KcsA channel, an ester substitution for the 1′ amide bond was found to reduce channel conductance, and a crystal structure of the ester mutant showed that the ester substitution decreased ion occupancy at the S1 site. Neither of these studies examined the effect of the ester substitutions on inactivation.

Here we substitute the 1′, 2′, and 3′ amide bonds in the selectivity filter of the KcsA K⁺ channel with esters and investigate the effect on inactivation. We determine the crystal structure of the 2′ ester mutant of the KcsA channel to examine the effect of the ester substitution on the structure and ion occupancy of the selectivity filter. We also investigate the effect of an ester substitution at the 2′ amide bond in the selectivity filter on inactivation in the voltage-gated K⁺ channel, K_vAP. Our results show that the S1 and S2 sites in the selectivity filter do not act as the foot in the door sites to prevent inactivation. Unexpectedly, we find that a lack of ion binding at the S2 site reduces inactivation.

Results

Ester Substitutions in the Selectivity Filter of the KcsA Channel.

We used the previously described semisynthesis of the KcsA channel to introduce amide-to-ester substitutions into the selectivity filter (29, 30). The semisynthesis provides a truncated but functional form of the KcsA channel. Briefly, native chemical ligation is used to assemble the KcsA polypeptide from a recombinant thioester peptide (residues 1–69) and a synthetic peptide (residues 70–123, with a N-terminal Cys) containing the desired ester linkage. The ligation product is folded to the native tetrameric state using lipid vesicles. Using this strategy, we were able to generate KcsA channels with ester substitutions for the 1′ (G79-ester), 2′ (Y78-ester), and the 3′ (G77-ester) amide bonds in the selectivity filter (Fig. 1A). We were unable to generate an ester substitution for the 4′ amide bond due to difficulties in incorporation of the T75-V76 ester linkage during peptide synthesis. We purified the semisynthetic KcsA ester mutants and reconstituted them into lipid vesicles for recording channel activity. As a control for our measurements, we generated a semisynthetic KcsA channel with the wild-type (WT) selectivity filter sequence (KcsA^WT).

KcsA is a pH-gated channel (10, 31). With a jump in the intracellular pH from 7.5 to 3.0, the KcsA^WT channels rapidly activated and then slowly inactivated (Fig. 1B). During inactivation, the current showed an initial rapid decrease followed by a gradual decrease to a value that was 38 ± 11% (n = 12) of the peak current at the end of the pH pulse (Fig. S1, Table S1). A single exponential fit to the initial phase (10 s) gave a time constant of 2.76 ± 1.04 s (S1 Materials and Methods). For the G79-ester mutant, we observed inactivation with an initial time constant of 2.94 ± 0.85 s (n = 5) followed by a gradual decrease to a value of 32 ± 11% of the peak current at the end of the pH pulse. These values are very similar to KcsA^WT (Student t test, P > 0.05), which indicates that the G79-ester substitution does not affect inactivation. The Y78-ester and the G77-ester mutants, in contrast, showed very little inactivation (Fig. 1B). For these ester mutants, the currents following activation decayed very slowly, and the current observed at the end of the pH pulse was 76 ± 11% (n = 5) for the Y78-ester mutant and 73 ± 8% (n = 5) for the G77-ester mutant, significantly higher than observed for KcsA^WT (Student t test, P < 0.05).

The inactivation properties of the ester mutants observed using macroscopic currents was also evident in recordings showing single-channel activity (Fig. 1C, Fig. S2). At a steady-state pH 3.0 on the intracellular side, KcsA^WT has a low open probability (P_o = 0.17 ± 0.04, n = 4) as it resides mainly in the inactivated state (10). The G79-ester shows a reduced single-channel conductance as previously reported (29). The G79-ester mutant also has a low P_o (0.14 ± 0.05, n = 3), similar to the WT channel and consistent with the G79-ester substitution not affecting inactivation. The Y78- and G77-ester mutants also show a lower single-channel conductance compared with the KcsA^WT but have a much higher P_o (Y78-ester: P_o = 0.51 ± 0.20, n = 3; G77-ester: P_o = 0.37 ± 0.09, n = 3). The higher P_o observed for the Y78- and the G77-ester mutants is in keeping with the reduced inactivation observed for these mutants.

We also incorporated the Y78-ester substitution into the KcsA channel by using the in vivo nonsense suppression approach (Fig. S3). We used an orthogonal tRNA/tRNA synthetase pair identified by the Schultz group, which incorporates (S)-3-(4-hydroxyphenyl)-2-hydroxypropionic acid (HPLA) at an amber stop codon, thereby substituting the amide bond with an ester (32, 33). This approach allowed incorporation of the Y78-ester substitution into the full-length KcsA channel. Inactivation in the Y78-ester mutant obtained using the in vivo suppression approach was considerably reduced compared with the WT KcsA channel, similar to the effect observed with the semisynthetic Y78-ester KcsA channel.

The electrophysiological measurements on the KcsA ester mutants therefore show that all of the ester substitutions in the selectivity filter have a similar effect of reducing the single-channel conductance but different effects on inactivation. The Y78- and G77-ester substitutions substantially reduce inactivation whereas the G79-ester substitution does not affect inactivation.

Structure of the Y78-Ester KcsA Channel.

To evaluate the effect of the ester substitution on the structure of the selectivity filter and ion occupancy, we determined the crystal structure of the Y78-ester mutant. We were not able to determine the structure of the G77-ester as the yields sufficient for the structure determination could not be obtained. The structure of the KcsA G79-ester mutant has been previously reported (29).

The Y78-ester KcsA channel was crystallized as a complex with an antibody Fab fragment in the presence of 300 mM KCl (5). The structure was solved by molecular replacement and refined to 2.1-Å resolution (Table S2). The electron density map for the selectivity filter of the Y78-ester is shown in Fig. 2A. The Y78-ester linkage was clearly resolved and is transplanar as seen in the F_o-F_c electron density omit map (Fig. 2B). Superposition of the selectivity filter of the Y78-ester and the WT KcsA channel is shown in Fig. 2C. Overall, these structures are very similar with a root-mean-square deviation of 0.214 Å for the selectivity filter (residues 75–79). One obvious difference is that the Y78-ester is lacking a K⁺ ion in the S2 site (Fig. 2A). Examination of the area surrounding the selectivity filter in the Y78-ester mutant reveals a change in the conformation of the E71 residue. In the WT channel, the E71 side chain forms a hydrogen bond (H-bond) with the amide hydrogen of Y78 and a carboxyl–carboxylate H-bond with the side chain of D80 (Fig. 2D) (5). This interaction between E71 and D80 side chains has been proposed to be one of the key determinants for inactivation in the KcsA channel (13). In the Y78-ester mutant, the amide-to-ester substitution disrupts the H-bond between E71 and the protein backbone, and the E71 residue is in a different rotameric state (Fig. 2E). The H-bond between E71 and D80 is present but the H-bond distance (2.55 Å) and the Cα-Cα distance (9.55 Å) of these residues is shorter in the Y78-ester mutant than the WT channel (2.63 and 10.00 Å, respectively). A shortening is also observed in the selectivity filter of the Y78-ester mutant with a distance of 8.88 Å between the S1 and S4 sites (as measured between the center of the ions at the S1 and the S4 sites) compared with 9.95 Å for the WT channel.

Fig. 2.

Structure of the selectivity filter of KcsA Y78-ester. (A) Stereoview of the electron density of the selectivity filter of KcsA Y78-ester. The 2F_o-F_c electron density map contoured at 2.0 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 ester bond between G77 and Y78. The F_o-F_c omit map (residues 77–78 omitted) is contoured at 3.0 σ. (C) Superposition of residues 71–80 of KcsA Y78-ester (red) and the WT (blue) (PDB: 1K4C). The hydrogen bond interactions of E71 in the WT (D) and Y78-ester KcsA (E). 2F_o-F_c electron density contoured at 2.0 σ for E71, D80, and the water molecule is shown. The Y78 side chain is not shown for clarity.

Effect of the Y78-Ester Substitution on the Ion Distribution in the Selectivity Filter.

To evaluate ion occupancy in the selectivity filter of the Y78-ester mutant, we scaled the data for the Y78-ester mutant to the WT KcsA channel and plotted one-dimensional electron density maps sampled along the selectivity filter axis as previously described (18, 19). In the WT channel, four peaks of roughly equal density that correspond to roughly equal occupancy of K⁺ ions at the four binding sites are observed (Fig. 3A). In the Y78-ester mutant, there is no significant electron density detected at S2, which indicates a lack of ion binding at the S2 site (Fig. 3B). Furthermore, the electron density peaks at the S1, S3, and S4 sites in the Y78-ester mutant are roughly similar to the WT, which indicates that the ion occupancy at these sites is similar to the WT. The lack of ion binding at the S2 site is therefore not accompanied by an increase in occupancy at the other sites. The total ion occupancy in the selectivity filter of the Y78-ester mutant is therefore lower at 1.7 compared with a value of 2.1 for the WT KcsA channel.

Fig. 3.

Ion distribution in the selectivity filter of WT KcsA and the ester mutants. K⁺ binding to the selectivity filters of WT KcsA (A) (PDB: 1K4C), Y78-ester (B), G79-ester (C) (PDB: 2H8P), and Rb⁺ binding to WT KcsA (D) (PDB: 1R3I). F_o-F_c electron density omit maps (KcsA residues 75–79, K⁺ ions, and lipid omitted) along the central axis of the selectivity filter, contoured at 3.0 σ, are shown. Below each structure, the one-dimensional plot of the electron density sampled along the central axis of the selectivity filter is shown.

The effect of the ester substitution in reducing ion occupancy was also observed in the previously reported G79-ester mutant (29). In this case, the ester substitution at the 1′ amide bond results in a decrease in ion occupancy at the S1 site without changing the ion occupancy at the other three sites (Fig. 3C). Similar to the Y78-ester, the total ion occupancy in the G79-ester channel is lower than the WT (1.4 vs. 2.1).

The crystal structure of the KcsA channel shows the presence of two ion-binding sites above the selectivity filter that have been referred to as the S0 and the S-1 sites (5). The ion occupancies of S0 and S-1 sites in the Y78-ester were similar to the WT channel, indicating that the effect on inactivation was not due to changes in ion occupancy at these sites (Fig. S4).

The ion distribution profile for Rb⁺ in the WT KcsA channel shows reduced Rb⁺ occupancy at the S2 site in the selectivity filter (Fig. 3D) (19). Inactivation in the WT KcsA channel, similar to other K⁺ channels, is reduced in Rb⁺ (20), consistent with a decrease in ion occupancy at the S2 site being linked to reduced inactivation.

Effect of an Ester Substitution at S2 Site on Slow Inactivation in the K_vAP Channel.

Next, we investigated the effect of an ester substitution at the 2′ amide bond of the selectivity filter on C-type inactivation in a K_v channel. For this investigation, we used K_vAP, an archaebacterial K_v channel (34). The K_vAP channel is activated by depolarization, and on sustained depolarization shows inactivation that is similar to C-type inactivation in a eukaryotic K_v channel like Shaker (3, 34). Furthermore, the K_vAP channel is readily expressed in Escherichia coli, a requirement for the in vivo nonsense suppression approach. We replaced Y199 in the K_vAP selectivity filter (equivalent to Y78 in KcsA, Y445 in Shaker B) with HPLA using nonsense suppression, thereby replacing the 2′ amide bond in the selectivity filter with an ester (Fig. 4A). We incorporated the K_vAP Y199-ester mutant into planar lipid bilayers for measurement of channel activity. Single-channel activity for the Y199-ester is shown (Fig. 4B). The single-channel conductance of the Y199-ester mutant was reduced (18.0 ± 0.7 pS, n = 9 in 150 mM K⁺ at +100 mV) compared with the WT channel (136 ± 4.5 pS, n = 5), similar to the effect of the ester substitutions in the KcsA channel. Voltage gating of the K_vAP Y199-ester mutant was similar to the WT channel (Fig. S5).

Fig. 4.

Ester substitution in the selectivity filter of the K_vAP channel. (A) Structure of the selectivity filter of wild-type K_vAP (PDB: 1ORQ). Two diagonally opposite subunits of the tetrameric channel are shown in stick representation, and the K⁺ ions bound are shown as purple spheres. The amide bond replaced by an ester is indicated by an asterisk. (B) Representative single-channel traces of the WT and Y199-ester K_vAP channels recorded at +150 mV. (C) Representative macroscopic currents for WT and Y199-ester K_vAP channels in response to the voltage protocol shown above. Solid blue lines indicate the single (WT) or double (Y199-ester) exponential fit used to obtain the inactivation time constants. (D) Recovery from inactivation for WT (circles) and Y199-ester K_vAP (triangles) was measured using paired 10-s depolarization steps to +100 mV with increasing interpulse intervals at −100 mV. The peak current at the second pulse (I) divided by the peak current at the first pulse (I_max) is plotted as a function of the interpulse duration. Solid lines represent single exponential fits used to obtain the time constant for recovery from inactivation. Error bars represent SD for three or more independent measurements. All experiments in this figure were recorded in symmetrical solutions containing 150 mM KCl, 10 mM Hepes–KOH (pH 7.5).

To investigate inactivation in the Y199-ester mutant, we elicited macroscopic currents by depolarization to +100 mV from a resting potential of −100 mV. We observed that inactivation in the Y199-ester mutant was substantially slower compared with the WT channel (Fig. 4C). For WT K_vAP channels, the decay in current during inactivation is fit by a single exponential to give a time constant of 469 ± 69 ms (n = 4) and a steady-state value of 1.0 ± 0.2% of the peak current. For the Y199-ester, we observed that a double exponential was required to fit the current decay and gave time constants of 3,742 ± 631 ms (69%, n = 4) and 117 ± 23 ms (31%) with a steady-state value that was 6 ± 6% of the peak current. We are presently not certain of the origin of the minor fast component during inactivation in the Y199-ester. We investigated recovery from inactivation using a paired pulse protocol (Fig. S5) and observed that the recovery from inactivation for the WT and the K_vAP Y199-ester mutant was very similar with a time constant of recovery of 7.9 ± 2.5 s (n = 3) for the WT channel compared with 12.6 ± 3.9 s (n = 4) for the Y199-ester mutant at −100 mV (Fig. 4D and Fig. S5). Overall, we observed that the effect of the ester substitution at the 2′ position in the selectivity filter of the K_vAP channel mirrors the effects observed for the 2′ ester substitution in the KcsA channel.

Discussion

Although it is well established that ion binding to the selectivity filter of K⁺ channels modulates C-type inactivation, the underlying mechanism is not yet understood. To investigate, we used amide-to-ester substitutions to perturb ion binding to the selectivity filter of the KcsA channel and determined the effect on inactivation. We showed that an ester substitution at the 1′ position has no effect on inactivation whereas ester substitutions at the 2′ and the 3′ positions reduced inactivation. We used X-ray crystallography to show that the 2′ ester substitution in the KcsA channel eliminated K⁺ binding to the S2 site. We also showed that an ester substitution at the 2′ position in the selectivity filter of the K_vAP channel reduced inactivation, similarly to the KcsA channel.

Why does substituting an amide bond with an ester reduce inactivation? Here we focus on the 2′ ester substitution because we were able to determine the structure of the KcsA 2′ ester mutant. The 2′ ester substitution affects both the H-bond interactions of the amide bond and the ion occupancy at the S2 site. The reduced inactivation in the 2′ ester mutant can therefore be due to either of these effects. We rule out disruption of the H-bond as the cause for reduced inactivation in the 2′ ester mutant for the following reasons: (i) The 2′ amide bond in the selectivity filter of the WT KcsA channel forms an H-bond with E71, which is disrupted in the 2′ ester mutant, and the E71 side chain is in a different conformation compared with the WT. However, the E71–D80 interaction that is critical for inactivation is maintained. (ii) The crystal structure of the K_vAP channel does not indicate any H-bond interactions between the 2′ amide bond in the selectivity filter and the surrounding residues, but the 2′ ester mutant of the K_vAP channel shows reduced inactivation similar to the KcsA 2′ ester mutant. (iii) Inactivation of the WT KcsA channel is reduced in Rb⁺, but the H-bond interactions of the selectivity filter of the KcsA channel in K⁺ and Rb⁺ are identical (18, 19). The ion occupancy profile in K⁺ and Rb⁺, however, is different, with low Rb⁺ occupancy at the S2 site. Therefore, we propose that reduced inactivation in the 2′ ester mutant is not due to the disruption of the H-bond but rather due to the lack of ion occupancy at the S2 site.

In the 1′ (G79-ester) and the 2′ (Y78-ester) mutants, the total ion occupancy in the selectivity filter and the single-channel conductance (reflecting K⁺ residence times in the selectivity filter) are roughly similar, but inactivation is reduced only in the Y78-ester mutant. This indicates that inactivation is regulated by ion occupancy at specific sites in the selectivity filter rather than total ion occupancy or residence time in the selectivity filter. Reduced inactivation in Rb⁺ compared with K⁺ has been proposed to arise due to greater residence time of Rb⁺ in the selectivity filter (6). As the occupancy profile of Rb⁺ shows lower ion occupancy at the S2 site, similar to the occupancy profile of K⁺ in the 2′ ester mutant, we propose that reduced inactivation in Rb⁺ is due to lack of ion occupancy at the S2 site and not due to a greater residence time of Rb⁺ in the filter.

Structural and computational studies on the KcsA channel have previously suggested an involvement of the S2 site in inactivation (14, 35). Structures of a constitutively open mutant of the KcsA channel with varying degrees of opening at the cytoplasmic gate (the bundle crossing of the pore-lining helices) showed a reduction in the occupancy at the S2 site with the extent of opening of the cytoplasmic gate. This led the authors to propose that inactivation proceeds with a loss of ion binding at the S2 site (14). Computational studies have similarly suggested that inactivation is inhibited by ion binding to the S2 site (35). These studies therefore predict that a reduction in the ion occupancy at the S2 site would increase the rate of inactivation. We observe that a reduction in ion occupancy at the S2 site decreases the rate of inactivation, and so our results do not support the conclusions from these studies.

Our results suggest that the S2 site must be occupied for the channel to inactivate. Previous studies on inactivation in the KcsA channel have indicated that the H-bond between E71 and D80 is important as perturbation of this interaction results in reduced inactivation (13). Structures of the KcsA channel with substitutions at E71 (A, S, I, Q) that result in reduced inactivation do not show a correlation with ion occupancy at the S2 site (10, 13, 36). Similarly, the Y78-ester mutant does not perturb the H-bond between E71 and D80, which suggests that this H-bond interaction and ion occupancy at the S2 site have independent effects on inactivation. Further studies will be necessary to determine how these factors modulate the transition of the selectivity filter from the conductive to the inactivated state.

Materials and Methods

The ester mutants of the KcsA and K_vAP channels were generated either by semisynthesis or by the in vivo nonsense suppression approach (29, 30, 32, 33). Crystals of the KcsA Y78-ester were grown as a complex with a Fab fragment as described (5). Electrophysiological measurements were carried out using giant liposome patch clamp for the KcsA channels and planar lipid bilayers for the K_vAP channels as previously described (34, 37, 38). Data are presented as mean ± SD. Detailed descriptions of materials and methods are provided in SI Materials and Methods.