Selectivity filter modalities and rapid inactivation of the hERG1 channel

Abstract

The human ether-á-go-go-related gene (hERG1) channel conducts small outward K⁺ currents that are critical for cardiomyocyte membrane repolarization. The gain-of-function mutation N629D at the outer mouth of the selectivity filter (SF) disrupts inactivation and K⁺-selective transport in hERG1, leading to arrhythmogenic phenotypes associated with long-QT syndrome. Here, we combined computational electrophysiology with Markov state model analysis to investigate how SF-level gating modalities control selective cation transport in wild-type (WT) and mutant (N629D) hERG1 variants. Starting from the recently reported cryogenic electron microscopy (cryo-EM) open-state channel structure, multiple microseconds-long molecular-dynamics (MD) trajectories were generated using different cation configurations at the filter, voltages, electrolyte concentrations, and force-field parameters. Most of the K⁺ permeation events observed in hERG1-WT simulations occurred at microsecond timescales, influenced by the spontaneous dehydration/rehydration dynamics at the filter. The SF region displayed conductive, constricted, occluded, and dilated states, in qualitative agreement with the well-documented flickering conductance of hERG1. In line with mutagenesis studies, these gating modalities resulted from dynamic interaction networks involving residues from the SF, outer-mouth vestibule, P-helices, and S5-P segments. We found that N629D mutation significantly stabilizes the SF in a state that is permeable to both K⁺ and Na⁺, which is reminiscent of the SF in the nonselective bacterial NaK channel. Increasing the external K⁺ concentration induced "WT-like" SF dynamics in N629D, in qualitative agreement with the recovery of flickering currents in experiments. Overall, our findings provide an understanding of the molecular mechanisms controlling selective transport in K⁺ channels with a nonconventional SF sequence.

Keywords: human ether-á-go-go channel; ion channels; long-QT syndrome; molecular dynamics.

Fig. 1.

Pore domain (PD) of the hERG1 channel. (A) Structural representation. The S5–P segments, P-helices, SF, outer-mouth vestibule, and S6 helices are highlighted in pink, silver, green, orange, and violet colors, respectively. Only two monomers are shown for clarity. The S5 helices of the PD were omitted for the same reason. The residues from the SF (₆₂₄SVGFG₆₂₈), N629 from the vestibule, and Q664 from the cytoplasmic gate are highlighted in licorice representation. The N, C, and O atoms are colored in blue, cyan, and red colors, respectively. The labels S0–S_cav indicate potassium or water SF binding sites. The pink shaded area indicates the region behind the filter for one of the monomers. (B) Radius profile along the pore domain. (C–E) Initial ion configurations at the SF used in our MD simulations. The symbols K, W, and 0 indicate potassium ions, water, and vacancy, respectively, at the SF binding sites S0–S4. A third ion was placed in S_cav binding site for the configuration shown in C. This configuration was also used for placing Na⁺ ions.

Fig. 2.

Potassium cation permeation events observed for hERG1 variants. (A) WT and (B) N629D. For each channel variant, the simulations were performed using the initial ion configuration [WKWKW], [K⁺] = 150 mM, V = 750 mV, and the NBFIX correction. Permeant cations are represented by different colors for each system. The binding sites at the SF along the pore axis are indicated by dashed lines. The acronym ES indicates the extracellular side of the channel. The black arrow indicates the direction of the applied electric field. The Inset in A zooms into the permeation events during the first 50 ns of the simulation.

Fig. 3.

Two-state HMM for the SF of hERG1-WT. The model was built using trajectories started from different ion configurations (Fig. 1 C–E), [KCl] = 150 mM, V = 750 mV, and the NBFIX correction for ∼10 µs of combined MD simulations (SI Appendix, Table S1). For each metastable state (states 1 and 2), we show representative configurations of the backbone for both pairs of opposite chains (Left and Middle), where the N, C, and O atoms are colored in blue, cyan, and red, respectively. The solid double-headed arrow indicates the average distance between backbone Cα atoms at the level of residues G626 and G628. We also show the dynamics of F627 side chains (Right). The cryo-EM configuration of the SF backbone (blue tubes) and F627 side chains (green licorice) are shown as well. Representative configurations explored by F627 side chains along the MD trajectories are shown as gray lines. Only 20 to 50 frames are superposed for clarity. Transition kinetics between the two states is evaluated using the mean first-passage time (MFPT) metric.

Fig. 4.

Hydrogen bond networks and hydration in the vicinity of the SF of hERG1 variants. (A–C) WT and (D–F) N629D. For each channel variant, the analysis was focused on the simulation performed using the [WKWKW] configuration, [K⁺] = 150 mM, V = 750 mV, and the NBFIX correction (∼4 µs long). A, B, D, and E show side views of the SF for each pair of opposite chains. C and F show the extracellular vestibule and SF for each hERG1 variant. The same coloring scheme as in Fig. 1 is used to highlight the structural motifs.

Fig. 5.

Two-state HMM for the SF of hERG1-N629D. The model was built using trajectories started from different ion configurations (Fig. 1 C–E), [KCl] = 150 mM, V = 750 mV, and the NBFIX correction for ∼10 µs of combined MD simulations (SI Appendix, Table S1). For each metastable state (states 1 and 2), we show representative configurations of the backbone for both pairs of opposite chains (Left and Middle) and for F627 side chains (Right). We used the same coloring scheme as in Fig. 3. Only 20 to 50 frames are superposed for clarity. Transition kinetics between the two states is evaluated using the mean first-passage time (MFPT) metric.

Fig. 6.

Average electrostatic potential (2D) xy maps at z position of residue 629 for hERG1-N629D. The simulations were performed using [K⁺] = 150 (A) and 500 (B) mM. The maps are shown perpendicular to the channel axis (cytoplasmic view). The segments of the S6 helices (blue cartoon) and the SF (gray tubes) are shown for spatial reference. Only one of the four D629 residues is shown for clarity, using ball-and-stick representation (C and O atoms are colored in cyan and red, respectively). Both simulations were performed using the [WKWKW] initial SF configuration, V = 750 mV, and the NBFIX correction. We obtained a map similar to B for the simulation performed using [K⁺] = 500 mM and V = 500 mV (not shown).

Fig. 7.

Reorganization of the interaction network at the SF of hERG1-N629D. Changes in the external [K⁺] modulate the repulsion among residues (−)D629 in the outer-mouth vestibule, favoring either the NaK-like configuration at low [K⁺] (A) or WT-like configuration at high [K⁺] (B). See Discussion for details.