Molecular dynamics simulations suggest possible activation and deactivation pathways in the hERG channel

Abstract

The elusive activation/deactivation mechanism of hERG is investigated, a voltage-gated potassium channel involved in severe inherited and drug-induced cardiac channelopathies, including the Long QT Syndrome. Firstly, the available structural data are integrated by providing a homology model for the closed state of the channel. Secondly, molecular dynamics combined with a network analysis revealed two distinct pathways coupling the voltage sensor domain with the pore domain. Interestingly, some LQTS-related mutations known to impair the activation/deactivation mechanism are distributed along the identified pathways, which thus suggests a microscopic interpretation of their role. Split channels simulations clarify a surprising feature of this channel, which is still able to gate when a cut is introduced between the voltage sensor domain and the neighboring helix S5. In summary, the presented results suggest possible activation/deactivation mechanisms of non-domain-swapped potassium channels that may aid in biomedical applications.

Fig. 1. Side view of hERG channel.

A single subunit (a) and the whole protein (b) are colored by domain: in blue, the voltage sensor domain ranging from helix S1 to loop L45, including the positively charged helix S4; in red, the pore domain, composed of helix S5, P-Loop, selectivity filter, and helix S6; in gray, the carboxy-terminal domain with the C-Linker and the cyclic nucleotide-binding homology domain.

Fig. 2. Contact analysis and activation/deactivation hypotheses.

a Contact map of the first subunit of the system with Q_g = 8e during the transition from O to C state. The black dots represent the contacts in the initial O conformation not broken during the simulation while the colored symbols indicate the contacts formed or broken at different times (t_f and t_b, respectively) during TMD. b Illustration of the two activation/deactivation hypotheses displaying the relevant conserved interactions in the C state.

Fig. 3. Activation/deactivation pathways of the closed system with Q_g = 8e.

Paths identified in subunits I (a), II (b), III (c), and IV (d). Panel e schematizes the main families of activation/deactivation pathways identified in all systems. Arrows describe the preferred routes of motion propagation: blue arrows refer to S4→L45→S6 route; red arrows refer to the L45→S6 route; green arrows refer to S4→S1→S5→S6. Black dots correspond to residues on the path with CI > 0.15 (see Supplementary Table 1); yellow circles refer to the pathological mutation R531Q/W known to alter the gating of the channel inducing the LQTS. The mutations W410S, Y420C, and T421M impair both the trafficking and the gating. Average minimal path lengths $⟨d_{\min }⟩$ are also reported.

Fig. 4. Communication pathways in split channels.

Split channels disconnected before D540 (a) and after G546 (b): dotted circles are centered on the source and sink regions used in the network analysis; these locations correspond to helices S4 and S6; red arrows schematize the S4→L45→S6 and S4→L45→S5→S6 paths while green arrows refer to S4→S1→S5→S6 path. Average minimal path lengths $⟨d_{\min }⟩$ are reported in the G546-split system (b); no pathways are found for D540 (a).

Fig. 5. Pseudo-potential of mean force for hERG with Q_g = 8e as a function of geometrical descriptors, relevant for the O→C transition.

The geometrical observables are: displacement and rotation of helices S4 and bending of helices S6 (a); rotation of the cytosolic domains with respect to the transmembrane ones (b). Pseudo-PMF maps as follows: S4 rotation vs S4 displacement (c); S6 bending vs S4 displacement (d); S6 bending vs S4 rotation (e); PD-CTD rotation vs S6 bending (f); PD-CTD rotation vs S4 displacement (g); PD-CTD rotation vs S4 rotation (h). Dashed lines are a guide to the eye. Bin widths are as follows: S4 displacement: 1.0 Å; S4 rotation: 2°; S6 bending: 1°; PD-CTD rot: 1°. Free energies were expressed in kcal/mol.