Structure and simulation of the Kv2.1 channel in the activated state.

(A) Structure of Kv2.1 determined by single-particle cryo-electron microscopy in a lipid nanodisc (17), viewed from the cell intracellular side and along the perpendicular to the membrane. The channel is an assembly of four distinct subunits (in colors); the transmembrane ion pore is formed at the center of the assembly, and the voltage sensors are found in the periphery, in a domain-swapped configuration. A K+ ion is shown in magenta, inside the selectivity filter. (B) Simulation system used in this study, comprising the channel (yellow), a phospholipid bilayer (gray), and a 300 mM KCl buffer (magenta, green). The figure depicts the final configuration of the 25-µs MD trajectory described in Figure 2. The total number of atoms is 201,954, some of which are omitted in the figure, for clarity. (C) Close-up of the pore domain and the selectivity filter. Note two protein subunits are omitted, for clarity. The configuration represented is that shown in panel (B); in this configuration, K+ ions are found in sites S1, S3 and S4 within the selectivity filter, while sites S0, S2 and Scav are transiently vacant.

Mechanism of K+ permeation through the Kv2.1 channel.

(A) Time traces of the position of K+ ions along the central axis of the channel as they reach and permeate the selectivity filter towards the extracellular side (black, red and blue). Ions that do not reach the filter are not shown for clarity (those that reach the filter but return to the cytoplasmic side before permeating are shown in orange and cyan). The approximate location of each of the K+ binding sites (S0 through S4, and Scav) along the pore is indicated alongside the plot. (B) Same as (A), for a fragment of the trajectory that illustrates the knock-on mechanism that initiates and completes each of the observed permeation events. (C) Density for K+ ions along the channel axis, relative to the bulk concentration value of 300 mM. Density peaks correspond to each of the K+ binding sites within and adjacent to the selectivity filter. For reference, gray horizontal lines indicate the average position of selected protein atoms: in the selectivity filter these are, from top to bottom, the backbone carbonyl oxygens of residues Y376, G375, V374 and T373, and the sidechain hydroxyl oxygen of T373; and in the cytoplasmic gate, the alpha-carbon of residue P408. All ions in the simulation system contribute to this profile, but only while they reside in a cylindrical volume of diameter equal to 12 Å, centered in and parallel to the channel axis, extending across the whole system.

Binding of TEA and RY785 to Kv2.1 channel and impact on K+ permeability.

(A) Time trace of the position of TEA (N atom) in the first 500 ns of the simulation, showing its spontaneous binding to the cavity between the selectivity filter and the cytoplasmic gate (marked by T373 and P406, respectively, whose position is indicated with gray traces). (B) Time trace of the position of TEA for the rest of the 5-µs simulation, alongside those for K+ ions within the selectivity filter, shown as in Figure 2. The location of each of the K+ binding sites therein is indicated alongside the plot. (C) Snapshot of the final snapshot of the simulation, with TEA bound to Scav. Only two of the four channel subunits are shown, for clarity. Residues P406 and I405 are marked in green. (D-F) Same as (A-C), for the simulation of RY785 binding, indicating separately the positions of the central N atom and of the S atom in the distal 5-membered ring.

Bound RY785 does not occlude the K+ pathway, but TEA does.

(A) Histograms of the position of TEA and RY785 as projected on the plane of the membrane, i.e. perpendicular to the pore, with the origin on its central axis. (B) Snapshot of the simulation of Kv2.1 and TEA described in Fig. 3, with a density map for TEA (blue mesh) calculated for all simulation snapshots. The map is viewed from the membrane plane, on the left, as well as from the cytoplasmic entrance of the pore, on the right. (C) Same as (B), for the simulation of Kv2.1 and RY785.

RY785 binds to the interior wall in the Kv2.1 cavity.

(A) Close-up of the snapshot shown in Fig. 4C, highlighting the most frequently observed protein contacts for RY785; these are on helix S6, which lines the cavity. Only two adjacent protein subunits are shown, for clarity. Note V409, P406 and Ile405 are at the narrowest point of the permeation pathway (aside from the selectivity filter). For reference, the inset shows an alignment of the sequence of this region of S6 (residues 396 to 409) with its equivalent in the KcsA K+ channel (residues 98 to 111), whose closed-state structure is known (25). (B) Statistical analysis of the interactions depicted in panel (A). A contact was defined as an instance wherein a C atom in the abovementioned hydrophobic sidechains was within 4.5 Å of one of the C/S atoms in RY785; a given snapshot might thus show multiple contacts with each side-chain, sometimes in two different subunits. The plot in the figure quantifies the percentage of simulation snapshots in which a given number of contacts were observed for each side-chain; e.g. in 75% of the snapshots RY785 contacts Val409 and Ile405 in one subunit, while Pro406, Ile401 and Val398 are contacted simultaneously in two subunits.

Bound TEA precludes K+ from accessing S4 binding site, but bound RY785 allows it.

(A) Five independent simulations of Kv2.1 bound to TEA wherein a knock-on event was artificially induced in the selectivity filter at t = 100 ns, to create a K+ vacancy in site S4. K+ did not reload this site in the subsequent 100 ns, except when TEA spontaneously dissociated prior to the knock-on event (right). (B) Same as (A), for Kv2.1 bound to RY785. In all simulations, a K+ ion reloads the S4 site within ∼10 ns of the induced knock-on event, corroborating RY785 is not an open-state blocker of Kv2.1.

Hypothetical mode of Kv2-channel inhibition by RY785.

R785 does not directly block K+ flow; instead, it reshapes the conformational free-energy landscape of the pore domain to stabilize an occluded, partially closed state that does not require S4 to reach the down state in full. The model is an inference based on the simulation data reported in this article and a previous electrophysiological study by Sack and co-workers (13). Note this is a simplified diagram of the functional cycle of the channel, which includes non-conductive or transiently inactivated states aside from the fully deactivated form.