Overview of CFTR channel structure.

(A) Cartoon representations of the 3D structure of CFTR (PDB: 6MSM). Different parts of the channel are coloured according to the domain (left) or the half channel (HC; right) to which they belong. The dashed lines indicate the membrane region. (B) Side view of the channel lumen. Cl- ions (cyan spheres) from multiple simulation snapshots are overlaid to TM helices (gray cartoon), outlining the space accessible to ions inside the pore. Intracellular Cl- ions cross a lateral intracellular (IC) entrance (z∼-40 Å) into the inner vestibule which leads to the pore bottleneck (z∼0 Å) separating the inner vestibule from its outer or extracellular (EC) counterpart. Note the absence of Cl- from the bottleneck region in this “near-open” state of the channel.

Conformational ensemble of TM helices.

(A) Spatial distribution of TM helices in the plane (x,y) of the outer lipid leaflet. Conformational transitions result in well resolved bimodal distributions of TM helices 1 and 11. The extracellular ends of TM helices 3-4 and 9-10 overlap in this plane. (B) In contrast to the outer leaflet, the spatial distribution of TM helices in the plane of the inner membrane leaflet (x,y) is conserved throughout the simulations. (C) 2D histograms of outer leaflet positions of pore-lining helices (TM1, 6, 8, 11, and 12) and TM2 in the squished conformation of the pore (blue) versus expanded pentagonal conformation (red). Helices in the blue-shaded region belong to half-channel 1, whereas those in the unshaded region belong to half-channel 2. (D) Top (extracellular) view of representative structures of the squished (blue) and pentagonal (red) conformations superimposed. Cl- ions (green spheres) are overlaid from multiple simulation snapshots to indicate the location of the channel pore.

Conservation of overall structure of ATP-bound CFTR in MD simulations.

Left: Distributions of RMSDs from the average of the MD structural ensemble. Average values are indicated in the legend. Right: Distribution of the distance between the centres of mass of the two NBDs (blue). The distances in experimental structures of ATP-bound (PDB: 6MSM; red dashed line) and ATP-free (PDB: 5UAK; green dashed line) states are shown for comparison.

Principal component analysis of the CFTR conformational ensemble in the TM region.

(A) Distribution of TM conformations projected onto the plane of the first two principal components reveals 4 conformational states labelled α-δ. The cross indicates the location of the initial cryo-EM structure (PDB: 6MSM). Contour levels are evenly spaced values of probability density. (B) Spatial distribution of pore-lining helices (TM1, 6, 8, 11, 12) in the plane of the outer leaflet (x,y) for conformational states α-δ. Contour levels are evenly spaced values of probability density. Concentric circles are centered on the pore axis and spaced by 4 Å. (C-D) Porcupine representation (red) of eigenvectors (C) PC1 and (D) PC2 overlaid on the CFTR structure (cyan) to illustrate the collective movements in each dimension. The largest collective motion, PC1, captures the concerted movement of TM1, 2, and 11 from near-open states β and γ to the pentagonal pore conformation, α (see also Fig. 2C).

Time evolution of the fractional population of conformational states α-δ indicating their relative stability on the timescale of the MD simulations.

Supplementary information for PCA analysis.

Left: fractional variance explained by each component. Right: cumulative fractional variance of principal components.

Structural features of the pore bottleneck.

(A) Cartoon representation of the bottleneck region in putative open state α. Pore-lining helices are shown in cyan except for TM6 (tan) and TM11 (magenta). The blue surface encapsulating the interior of the open pore was computed with HOLE2 (Smart et al., 1996). (B) Distributions of the minimum radius (Rmin) in the pore bottleneck. Pore radii are calculated with HOLE2. Average values are indicated in the legend. (C) Distribution of the distances of pore-lining helices (TM1, 6, 8, 11, and 12) to the centre of the pore for conformational states α-δ. State α is unique for showing a unimodal distribution, consistent with the symmetrical arrangement of pore-lining helices.

Symmetry of TM helix conformations in the pore bottleneck.

Distributions of the distances of the five pore-lining helices to the centre of the pore in conformational states α-δ. In contrast to states β-δ, the strong overlap between narrow TM helix distributions in state α reflects their nearly symmetrical arrangement.

Detailed structural features of the pore bottleneck.

(A-B) Distributions of distance between the Cα of residues R334 and T338 onTM6, and that of T1122 and S1118 on TM11 in the bottleneck region for conformational states α-δ. (C-D) Distributions of solvent accessible surface area (SASA) of bottleneck-lining residues T1115 and S1118 from TM11.

Cl- permeation pathways through the pore bottleneck.

(A) Side and (B) top views of the bottleneck region in the putative open state α with overlaid Cl- ions (coloured spheres) showing three different Cl- permeation routes. Blue: 1-6 pathway between TM1 and 6; red: 1-12 pathway between TM1 and 12; green: intermediate pathway. (C) Pore conformations in which Cl- permeation occurs are indicated as scatter points on the (PC1,PC2) plane, coloured by permeation pathway. (D) Breakdown of Cl- permeation pathways followed in each conformational state (N: number of permeation events) for (top) the full voltage range (-500 < U < 500 mV) and (bottom) the physiological voltage range (-120 < U < 120 mV). The number of times each permeation pathway was followed is indicated on top of the bar graphs.

Projection of the two earlier trajectories in which Cl- permeation events occurred in our previous study (Zeng et al., 2023) onto the (PC1,PC2) plane. (Left) traces of the two trajectories coloured by timestep. (Right) The conformations sampled in the two simulations indicated as red scatter points. Conformations in which Cl- translocations occurred are marked in black “x”. The snapshots from these two trajectories used to initiate the extended trajectories reported in this work are marked in cyan “x”.

Solvation of Cl- by protein and water inside the pore of CFTR.

(A) Cl- binding sites inside the inner vestibule. Regions of high Cl- density (blue mesh) indicate the binding sites where Cl- ions preferentially reside. Protein sidechains coming into direct contact with Cl- ions in its first solvation shell are shown; blue labels highlight cationic residues among them. (B-C) Solvation environment of Cl- ions that underwent translocation. The bottleneck region is indicated as the gray shaded region (-5 < z < 5 Å). (B) The average numbers of polar groups from protein sidechains (black) and of water molecules (red) in the first solvation shell of Cl- ions (nsolv) along the channel axis are shown together with their sum (green). The average number of non-polar protein residues in the first solvation shell of Cl- ions is indicated in purple. (C) Axial distribution of protein residues making significant sidechain contributions to the solvation of Cl- ions. F337 is also shown. (D-E) Distribution of the number of water molecules (NH2O) in the first solvation shell of Cl- ions in the bottleneck region of (D) the 1-6 pathway (blue) and of (E) in the central pathway (green) compared to bulk water (gray).

Solvation of Cl- ions by permeation pathway in the pore bottleneck.

Statistics of solvation by water and protein of Cl- ions that followed (A-B) the 1-6 pathway and (C-D) the central pathway (combined 1-12 and intermediate pathways) in the pore bottleneck. The grey shaded region indicates the pore bottleneck (-5 < z < 5 Å). For details, see the caption of Figure 6.

Current-voltage (I-V) relationship for each of the four conformational states.

(A) Entire voltage range sampled and (B) physiological and experimentally accessible range. Zero-intercept linear regression fits of data points in the positive (0 < U < +120 mV) and in the negative (-120 < U < 0 mV) TM voltages are shown for state α as red straight lines with conductance (slope) indicated. The gray shaded regions are bound by the experimental single channel conductance range of 6-10 pS (Sheppard and Welsh, 1999). Negative and positive currents indicate Cl- efflux to and influx from the extracellular space, respectively.

Effect of R334 conformation on Cl- permeation. (A-B)

Snapshots of the extracellular mouth of the pore with R334 in the (A) up and (B) down states. TM helices 6 (tan), 11 (magenta) and 12 (yellow) are shown together with loop ECL-6. In the down conformation, R334 can form cation-π and π-π interactions with F337 and an ion pair with Cl- in the pore bottleneck. (C-D) Iso-surfaces reveal an increase in the average Cl- ion density in the bottleneck region in state α from (C) the up state to (D) the down state of R334. (E) The current-voltage relationship shows that, in putative open state α, the down conformation of R334 (red) increases Cl- conductance 5-fold in the physiological voltage range (inset) compared to the up conformation (blue).

The sidechain conformation of R134 does not influence Cl- permeation in CFTR. All analyses are restricted to the α state. (A) The axial position of the centre of charge (Cζ atom) of the R134 sidechain samples two main states “up” and “down”. (B-C) Estimates of Cl- conductance computed from these two conformational states do not differ significantly.

Distribution of distance (d) between centre of charge (Cζ atom) of R334 and centre of the phenyl ring of F337 in conformational states R334-up (blue) and R334-down (red).

Effect of salt-bridge interactions between R334 and ECL-6 loop residues on Cl- permeation.

(A-B) Snapshots of the extracellular mouth of the pore showing the dunked sidechain of R334 forming salt bridges with (A) E1124 and (B) E1126. (C-D) Iso-surfaces reveal a greater Cl- ion density in the extracellular mouth of the pore in open state α when (C) the glutamate sidechains are disengaged from the central pore, compared to (D) when they form a salt bridge with R334. (E-F) Top view of representative conformations of ECL-6 (cyan) in the (E) loop-outward and (F) loop-inward states. The helices lining the pentagonal pore are shown in gray. (G) The current-voltage relationship shows that the salt-bridge interactions decrease Cl- conductance in the R334-down, open state of CFTR, especially in the physiological voltage range (inset).

Analysis of charged sidechain arrangements in the extracellular mouth of the penta-helical pore (state α).

(A) Axial distribution of the centres of charge of R334 (Cζ atom), E1124 and E1126 (Cδ atoms). Ion pairing between R334 and either glutamate residue is precluded when glutamate sidechains are oriented outwards. In the R334 vs E1126 contour plot (bottom left panel), the boxed region highlights the “down-out” state with R334 down and E1126 out. In the E1124 vs E1124 contour plot (bottom middle panel), the boxed region highlights the “out-out” state with both glutamate residues out. (B) Scatter plot of apparent Cl- conductance (g) versus fractional population of R334 down without salt-bridge with ECL-6 glutamate residues [fdown-out-out, i.e. intersection of the two red boxes in (A)] at different TM voltages (indicated in mV in black). The circles are colour coded according to the sign of the voltage (yellow: positive; black: negative). The positive correlation shows that at negative voltages, the sidechain conformations of the ECL-6 loop glutamate residues and R334 modulate Cl- conductance. At positive voltages, the systematic under-sampling of the R334-down and glutamates-out conformation provides a plausible explanation for the apparent inward rectification of CFTR in the simulations.

Structural differences between the ATP-unbound, closed state and the open state of human CFTR in the gate region.

(A) Top and (B) side views of representative structures taken from MD simulations of ATP-free CFTR (left; black) and of the MD-open state of ATP-bound CFTR (right; red). To reach the open state, the outer segments of TM10 (cyan) and 11 (magenta) must separate to accommodate the intercalation of TM12 (yellow) between them. (C) 2D histograms of outer leaflet positions of pore-lining TM helices (1, 6, 8, 11, and 12) together with TM 2 and 10 from MD simulations of (black) ATP-free CFTR versus (red) MD-open state of CFTR. The largest movements of TM helices needed to reach the open state are indicated by arrows. Helices in the blue-shaded region belong to half-channel 1, whereas those in the unshaded region belong to half-channel 2. (D) Distribution of distances separating the centres of mass of the outer segments of TM10 and 11 in the ATP-free closed state (black) and in the four ATP-bound conformational states α-δ.

NBD1-NBD2 separation in the 20 simulations of ATP-free CFTR.

Time evolution of the distances separating the two halves of the catalytic site (orange) and of the degenerate site (blue). Dashed lines indicate the distances in ATP-bound, NBD-dimerized PDB structure 6MSM (red: catalytic site; blue: degenerate site). Background colors highlight the simulations in which (green) full dimerization; (caramel) partial dimerization; or (white) no dimerization occurred.

Conservation of TM helix arrangement at the extracellular end of the pore despite fast NBD dimerization in simulations of ATP-free, closed state CFTR. (Left)

Distribution of centre-of-mass distance between NBDs showing an NBD-separated population (d > 40 Å) and NBD-dimerized population (d < 40 Å). Distributions of (x,y) positions of TM helices at the extracellular end (middle) before and (right) after NBD-dimerization show that their relative arrangement is conserved. Blue crosses in the 2D contour plots indicate the position of TM helices in the ATP-free, unphosphorylated, and NBD-separated human CFTR (PDB: 5UAK) at the start of the simulations.

Analysis of changes in helical kinks underlying the transition between near-open and open states.

(A) Cartoon representation of CFTR highlighting residues located at helical kinks on TM 1 (blue), 2 (red), and 11 (magenta) where structural changes occur upon channel opening. The sidechains of these residues (Hα atoms for glycine) are shown. The three helices completing the pore bottleneck are shown in dark gray. All other parts of the protein are translucent. (B-G) Analysis of helical kink magnitude and orientation. (Left) Distribution of kink angle θ (kink magnitude) and wobble or swivel angle (kink orientation) are shown for each of the four conformational states at (B) G126 on TM2; (D) W1098 on TM11; and (F) G85 on TM1. (Right) Contour plots showing the correlation between kink or swivel angles and principal component 1 (PC1), which captures conformational transitions between MD-closed and MD-open states (see Figure 3). Note that the uncertainty in kink orientation is high for small kink angles θ.

Analysis of helical kink and orientation angles at (A-B) P99 on TM1 and (C-D) P140 on TM2.

See the caption of Fig. 11 for details.