Potassium channel selectivity filter dynamics revealed by single-molecule FRET

Abstract

Potassium (K) channels exhibit exquisite selectivity for conduction of K⁺ ions over other cations, particularly Na⁺. High-resolution structures reveal an archetypal selectivity filter (SF) conformation in which dehydrated K⁺ ions, but not Na⁺ ions, are perfectly coordinated. Using single-molecule FRET (smFRET), we show that the SF-forming loop (SF-loop) in KirBac1.1 transitions between constrained and dilated conformations as a function of ion concentration. The constrained conformation, essential for selective K⁺ permeability, is stabilized by K⁺ but not Na⁺ ions. Mutations that render channels nonselective result in dilated and dynamically unstable conformations, independent of the permeant ion. Further, while wild-type KirBac1.1 channels are K⁺ selective in physiological conditions, Na⁺ permeates in the absence of K⁺. Moreover, whereas K⁺ gradients preferentially support ⁸⁶Rb⁺ fluxes, Na⁺ gradients preferentially support ²²Na⁺ fluxes. This suggests differential ion selectivity in constrained versus dilated states, potentially providing a structural basis for this anomalous mole fraction effect.

Fig. 1.. The SF-loop conformation is dependent on ion occupancies

a. Representative smFRET trajectories (above) and associated FRET amplitude histograms (below) at T120C sites under different ionic conditions as indicated. Fluorescence intensities of the donor Alexa Fluor 555 and acceptor Alexa Fluor 647 are colored cyan and red, respectively, FRET is colored blue. The FRET levels 0.2, 0.45 and 0.8 are marked by dashed lines. b. FRET contour maps and histograms for AF555/647 fluorophores labeled at diagonal T120C sites in the SF-loop, at increasing [K⁺] from 0–5 mM, on a 150 mM Na⁺ background, as indicated (n= 173, 333, 345, 422 and 386 traces). c. Fractional amplitudes from 3 state fits to FRET contour maps in b, plus Hill equation fits of [K⁺] dependence of fractional occupancy of high (F0.8, red, K_1/2= 1.34 mM, nH= 0.5) and low (F0.2, blue, K_1/2= 0.69 mM, nH= 0.5) states. The amplitudes of medium (F0.45, green) FRET populations were not fitted. d. FRET contour maps and histograms for AF555/647 fluorophores labeled at the diagonal A45C sites in the SF-loop, in 150 mM [K⁺] (n= 396 traces) or 150 mM Na⁺ (n= 524 traces), as indicated. e. FRET contour map and histogram for AF555/647 fluorophores labeled at the diagonal T120C sites in the SF-loop in asymmetric 145 mM [K⁺] on the cytoplasmic side, while the extracellular side of the channel faces 5 mM [K⁺] (n= 335 traces).

Fig. 2.. Permeant ion-dependent kinetics of SF-loop conformational dynamics

a. Concatenated smFRET trajectories (blue) in 150 mM KCl (n=363 traces) and 150 mM NaCl (n= 173 traces), together with raw idealizations (brown) and subsequent cut-off idealization into low (0.2), middle (0.45) and high (0.8) FRET states (green). b. Mean lifetime in low (L), medium (M) and high (H) FRET states as a function of [K⁺], from cut-off idealization into 3 states for concatenated trajectories as in a (and Supplementary Fig. 2). c. Transition rate constants between L, M and H (ML is rate constant for transition from M to L, etc) as a function of [K⁺].

Fig. 3.. Quaternary ammonium ions and impermeant NMDG stabilize low FRET states

a. FRET contour maps and histograms for AF555/647 fluorophores labeled at diagonal T120C sites of KirBac1.1 SF-loop with different concentrations of TEA⁺ (n= 420, 185, 177 and 214 traces). b. Fractional amplitudes from 3 state fits to FRET contour maps in a. Hill equation fits of dependence of fractional occupancy of high (F0.8, red) and low (F0.2, blue) states. The amplitudes of medium (F0.45, green) FRET populations were not fitted. c. FRET contour maps and histograms for AF555/647 fluorophores labeled at the diagonal T120C sites in the SF-loop, with increasing [K⁺] from 1–5 mM, on a 150 mM NMDG⁺ background, as indicated (n= 234, 212, 252, 191 and 240 traces). d. Fractional amplitudes from 3 state fits to FRET contour maps in b. Hill equation fits of [K⁺] dependence of fractional occupancy of high (F0.8, red, K_1/2= 0.12 mM, nH= 2.0) and low (F0.2, blue, K_1/2= 0.22 mM, nH= 2.0) states. Comparable fits for 150 mM Na⁺ background (Fig. 1e) are shown as dashed lines. The amplitudes of medium (F0.45, green) FRET populations were not fitted.

Fig. 4.. Non-selective KirBac1.1 mutants exhibit dynamic conformations that are not stabilized by either K or Na.

a. Ion-selectivity of KirBac1.1 WT and mutants labeled at diagonal T120C sites with Alexa Fluor 555 and 647 fluorophores. WT channels show flux inhibition by permeant K⁺, Rb⁺ and Cs⁺ ions but not by impermeant Na⁺ or Li⁺, whereas the mutant channels are inhibited by all ions. b. Representative smFRET trajectories at T120C sites for KirBac1.1 non-selective mutations, in 150 mM KCl or 150 mM NaCl, as indicated. Fluorescence intensities of the donor Alexa Fluor 555 and acceptor Alexa Fluor 647 are colored cyan and red, respectively, FRET is colored blue. c. FRET contour maps and histograms at diagonal T120C labeling sites of KirBac1.1 non-selective mutants (from left to right, n= 524, 485, 220, 238, 263 and 254 traces) in either 150 mM KCl (left) or 150 NaCl (right).

Fig. 5.. Permeant ion induced conductance in KirBac1.

a, b. Time course of ²²Na⁺ (A) or ⁸⁶Rb⁺ (B) uptake into liposomes (POPE:POPG, 75: 25%), with 450 mM internal and 0.05 mM external [K⁺] (blue), or with 450 mM internal and 0.05 mM external [Na⁺] (green). Liposomes were reconstituted with no protein (open symbols), or with KirBac1.1 (2.5 μg/mg lipid, filled symbols), n=3 in each condition. c. Cartoon representation of proposed model for ion-dependent conformational transitions in KirBac1.1. The selectivity filter exists predominantly in a constrained high FRET state (H), an intermediate medium FRET state (M) or a dilated low FRET state (L). Na ions in the pore favor transitions away from the H state, particularly the H to M and M to L transitions. K⁺ ions in the pore favor transitions to the H state, in which the SF generates the ‘canonical’ K⁺ selective ion binding sites.