Fig 1.
Resolution of sevoflurane sites at the homotetrameric Kv1.2 structures C and O.
(A) Atomistic systems containing Kv1.2 structures (cyan) embedded in a fully-hydrated lipid bilayer (gray) were MD simulated to produce molecular ensembles considered for flexible docking calculations. (B) Docking poses for singly-occupied sites. Shown is the ensemble-average channel structures C and O, along with the set of centroid configurations of sevoflurane (points) determined from docking. Centroid configurations of sevoflurane were clustered as a function of their location on the channel structures, that is within the voltage-sensor (green), at the S4S5 linker (yellow), at the S4Pore (dark blue) and S6P-helix (light blue) interfaces, at the central cavity (violet) and extracellular face (pink). Each of these clusters was treated as an interaction site j for sevoflurane with volume δVj. (C) Following another round of docking calculations starting from structures in (B), solutions for doubly occupied sites were resolved by determining if volumes δVj could accommodate the centroid positions of two docked ligands at once. For better annotation, helices S4, S4S5 linker, S6 (**) and P-helix (*), as well as the pore and VSD domains are indicated in structures in (B) and (C). (D) Per site number of docking solutions for single (cyan) and double (gray) ligand occupancy. (E) Representative molecular structure resolved from docking. Voltage-sensor domains in two opposing channel subunits are not depicted for clarity in (B) and (C) lateral views.
Fig 2.
C and O state-dependent binding probabilities for different concentrations of sevoflurane at the reservoir.
(A) Marginal probabilities ρX(nj) of site j, for nj = 0 (gray), nj = 1 (blue) and nj = 2(cyan). Marginals at the extracellular face of the channel are negligible for every structure/concentration. (B) Probabilities ρX(n) for macrostates mapping an ensemble of accessible states
in which n ligands bind the receptor regardless their specific distributions over the binding sites. Here, ρX(nj) and ρX(n) were computed by coarse-graining over state probabilities in S4 Fig (cf. Computational Methods for details). Average number ⟨nX⟩ of bound ligands as a function of the reservoir concentration is indicated in (B).
Fig 3.
C and O position-dependent binding probabilities for diluted concentrations of sevoflurane in the bulk.
(A) Shown is the ensemble average structure of the channel (white) along with the density of sevoflurane (orange and cyan) in each of the binding sites (isovalues of 9x10-5 Å-3). Densities pertaining to sites S4S5 linker, S6P-helix and S4Pore are indicated with yellow symbols. As described in Computational Methods, the determination of
involved reweighing the marginal probability ρX(nj) at the binding site j by the local equilibrium density of sevoflurane ρX(R|nj). The marginal ρX(nj) was computed by coarse-graining over state probabilities in S2 Fig whereas, ρX(R|nj) was calculated from the centroid distributions of docking solutions shown in Fig 1B and 1C. (B) Projection of
along the transmembrane direction z of the system,
.
Fig 4.
Sevoflurane binding effects on C-O equilibrium.
(A) Kv1.2 open probabilities for different sevoflurane concentrations in solution. Ligand-free and ligand-bound ρO(V) curves were respectively computed from eqs (5) and (9) by taking into consideration parameters, Vm = −21.9mV and ΔQ = 3.85eo, for best two-state Boltzmann fit of measured data for Kv1.2 free of ligands [18]. A reference experimental curve (blue) is shown for sevoflurane at 1 mM concentration, with best two-state Boltzmann parameters Vm = −25.1mV and ΔQ = 4.00eo [18]. Relative to the ligand-free channel, 1 mM of sevoflurane shifts the open probability of the channel leftward by ΔVm ≈ −3mV. The inset explicitly shows both experimental (blue) and calculated (red) Vm shifts at 1mM of sevoflurane in solution. In black, ligand-free Vm is also shown for reference. For sevoflurane concentrations of 1, 10 and 100 mM, their respective open probabilities Vm are -22.3, -24.7 and -51.1 mV. Representative statistical errors for ρO(V) at 100mM sevoflurane concentration (error bars) were calculated by Monte Carlo bootstrap error analysis of the statistical uncertainty of independent FEP estimates considered in the calculations. (B) Decomposition analysis at 100mM ligand concentration. Shown is the FEP sampling overlap versus ratio values for C-O partition functions at the individual binding sites j. Per-site ratio values can be equal, smaller or larger than unity meaning respectively that sevoflurane binding is not conformational dependent, stabilizes the open structure or stabilizes the closed structure. Binding sites located nearby flexible protein regions for which the root-mean-square deviation (RMSD) between channel structures is larger than 4.0 Å are highlighted in red (cf. Computational Methods and S2 Fig for details). (C) Decomposition analysis of ρO(V) curves in terms of partition ratio values showing in (B). (D) Same decomposition analysis in terms of an aggregate per-site contribution across channels subunits. At 100mM, binding of sevoflurane at the S4S5 linker and S4Pore interface significantly stabilizes the open structure of the channel which contrasts the mild stabilization of the closed structure due to ligand binding at the S6P-helix interface.