CG simulations of multiple TMEM16 structures capture lipid density in the TM4/TM6 pathway of scrambling competent members.

(A) Snapshot and POPC headgroup density (right) from atomistic simulations of Ca2+-bound nhTMEM16 (PDB ID 4WIS) previously published in ref. (36). Only the lipid headgroup choline (blue) and phosphate (red) beads are shown for clarity. Density (brown isosurface) is averaged from both subunits across 8 independent simulations totaling ∼2 μs (B) Snapshots from CG simulations of open Ca2+-bound nhTMEM16 (PDB ID 4WIS, green), afTMEM16 (PDB ID 7RXG, violet), TMEM16K (PDB ID 5OC9, orange), TMEM16F F518H (PDB ID 8B8J, blue), TMEM16K (orange), and TMEM16A (red). (C) Snapshots with lipid headgroup densities near simulated open (6QP6*) and closed (PDB ID 6QP6) TMEM16F. (D) Snapshot of simulated ion conductive TMEM16A (5OYB*). For each CG snapshot, again only the lipid headgroup choline (blue) and phosphate (red) beads are shown for clarity. Each density (brown isosurface) is averaged over both chains except TMEM16K and TMEM16A where only a single chain is used due to the structure’s asymmetry.

Number of scrambling events in and out of the canonical groove pathway.

Scrambling events where the lipid headgroup transitions between leaflets within 4.7 Å of the DOPC maximum density pathway. All other events were considered “out-of-the-groove”. For the full list of simulations and scrambling rates see source data file.

Simulated lipid scrambling differentiates closed/open conformations.

(A) Accumulated scrambling events during CGMD simulations of experimental and simulated (sim) structures of nhTMEM16 (green), afTMEM16 (violet), TMEM16K (gold), TMEM16F (blue), and TMEM16A (red). Inset values are the average rate and its standard deviation. Plots corresponding to structures described as “open” in their original publications (PDB IDs 4WIS (19), 6QM6 (61), 7RXG (59), 5OC9 (12), 8B8J (60), and 6QP6* (55)) are shaded. (B) Snapshots of the open nhTMEM16 simulation (PDB ID 4WIS) showing a single scrambling event over time. The tail (yellow) of the scrambling lipid is explicitly shown, while all other lipids only show the phosphate (red)/choline (blue) headgroup.

Lipid scrambling rates correlate with groove openness and membrane thinning.

(A) The minimal membrane thickness at the groove plotted against the median width of the groove measured based on the minimal distance between any two residues on TM4 and TM6 of the groove with the most scrambling events. The lower and upper error bars represent the 25% (Q1) and 75% (Q3) quartiles, respectively. Each data point is colored by the scrambling rate through that same groove. Dashed lines define minimal TM4/TM6 distance and membrane thickness requirements for robust scrambling (shaded grey quadrant). (B) Simulation time traces of the TM4-TM6 minimal distance at the most scrambling-competent groove of 4WIS, 5OC9, 8B8J, and 5OYB* (top to bottom). The dashed line indicates the 6 Å threshold we defined for scrambling-competent groove opening. Black dots indicate time points at which a scrambling event is completed. The solid curve is a recursively exponentially weighted moving average with smoothing factor 0.1, while the transparent curve is the raw distance values. (C) Density isosurfaces for DOPC headgroup beads (yellow) and average membrane surface calculated from the glycerol beads (blue) for representative nhTMEM16, TMEM16K, TMEM16F, TMEM16A, and afTMEM16 simulations. Panels are ordered left to right by decreasing scrambling rate. Cartoon beads and arrows in each image indicate the closest points between the inner and outer leaflet of the average surface.

Lipid scrambling events and lipid-protein residue contact in the dimer interface and canonical TM4/TM6 groove.

Traces of all scrambling lipids in a TMEM16F (PDB ID 8B8J) simulation (center top). Lipid scrambling from the inner to outer leaflet are illustrated as cyan traces and from the outer to inner leaflet as yellow traces. A cartoon depiction of two individual inward scrambling events along the TM4/TM6 groove (orange tail with red/blue headgroup) and the dimer interface (yellow tail with red/blue headgroup) with multiple snapshots over time (center bottom). Only headgroup, first and second tail beads are shown for clarity. Protein backbone colored by mean lipid headgroup interaction (dwell) time at the TMEM16F dimer interface (left) and TM4/TM6 groove (right).

Comparison between membrane deformations in cryo-EM nanodiscs and CG MD.

Front and side views of cryo-EM maps (left) and the final frames of our CG MD simulations (right). nhTMEM16: cryo-EM images (PDB IDs 6QM9 and 6QM4) were adapted from 2019, Kalienkova et al, published by eLife (61). The CG MD structures are PDB IDs 4WIS (with Ca2+) and 6QM4 (without Ca2+). afTMEM16: cryo-EM image (PDB ID 7RXG) was adapted from 2022, Falzone et al, published by Springer Nature (59). The CG MD structures are PDB IDs 7RXG (with Ca2+) and 7RXB (without Ca2+). TMEM16K: cryo-EM images (PDB ID 5OC9) were adapted from 2019, Bushell et al, published by Springer Nature (12). The CG MD structures are PDB IDs 5OC9 (with Ca2+) and 6R7X (without Ca2+). TMEM16F: cryo-EM images (PDB IDs 6QPC and 8B8J) were adapted from 2022, Arndt et al, published by Springer Nature (60). The CG MD structures are also PDB IDs 6QPC (WT with Ca2+) and 8B8J (F518H with Ca2+). Black arrows indicate a dip in lipid density near the groove entrance. All CG MD snapshots were rendered with PyMOL 2.5.0 (104), after selecting all lipid beads within 12 Å of the protein and matching the coloring to the colors used in the original cryo-EM images.

CG simulations of multiple TMEM16 structures with closed grooves lack lipid density in the TM4/TM6 pathway.

Snapshots from CG MD simulations of nhTMEM16 (PDB IDs 6QM4 (closed) and 4WIS (open), green), afTMEM16 (PDB IDs 7RXB (closed) and 7RXG (open), violet), TMEM16K (PDB IDs 6R7X (closed) and 5OC9 (open), gold), and TMEM16F (PDB IDs 6QPB (closed) and 6QP6* (open), blue) with phosphatidylcholine (PC) lipid headgroup density (yellow) and nearby lipids (yellow). Residues forming the closest distance between TM4 and TM6 (colored by residue type: basic (red), acidic (blue), and polar (green)) and lipids near the groove also shown. Each density is averaged over both chains.

Measuring lipid angles to detect scrambling events.

(A) For every time frame, for every lipid in the system, we defined a vector between the choline bead (NC3) and the two tail beads (C4A, C4B; dashed arrows) and calculated the angle θ between the average of those two vectors (solid arrow) with the z axis. (B) A schematic representation of a typical time trace for a lipid that scrambles from the upper membrane leaflet (θ ≈ 150°) to the lower membrane leaflet (θ ≈ 30°). A scrambling event is only counted when θ passes the threshold at the opposite leaflet with respect to its original location (35° for the lower, 145° for the upper).

Position traces of scrambling lipids in fungal TMEM16 simulations.

(A) Lipid traces for Ca2+-bound open nhTMEM16 (PDB ID 4WIS). (B) Lipid traces for Ca2+-bound open nhTMEM16 (PDB ID 6QM6). (C) Lipid traces for Ca2+ -bound open afTMEM16 (PDB ID 7RXG). Lipid traces are generated by fitting raw lipid headgroup center of mass positions to a smooth spline curve.

Position traces of scrambling lipids in the open TMEM16K simulation.

(A) Lipid traces for Ca2+-bound open TMEM16K (PDB ID 5OC9). (B) Cartoon representation of aligned subunits of Ca2+-bound TMEM16K (PDB ID 5OC9). Lipid traces are generated by fitting raw lipid headgroup center of mass positions to a smooth spline curve.

Position traces of scrambling lipids in TMEM16F simulations.

(A) Lipid traces for Ca2+ -bound open TMEM16F F518H mutant (PDB ID 8B8J). (B) Lipid traces for Ca2+ -bound simulated open TMEM16F (6QP6*) (C) Cartoon representation of aligned subunits of Ca2+-bound simulated open TMEM16F (6QP6*). (D) Lipid traces for Ca2+ -bound simulated closed TMEM16F (PDB ID 8TAG). Lipid traces are generated by fitting raw lipid headgroup center of mass positions to a smooth spline curve.

Position traces of scrambling lipids in TMEM16A simulations.

(A) Lipid traces for Ca2+ -bound simulated conductive TMEM16A (5OYB*). (B) Lipid traces for Ca2+ -bound simulated open TMEM16A (7ZK3*6). (C) Lipid traces for Ca2+ -bound simulated open TMEM16A (7ZK3*10). (D) Lipid traces for Ca2+ -bound simulated open TMEM16A (7ZK3*8). (E) Snapshots from simulation of 5OYB* during lipid scrambling event (trace in A). Lipid traces are generated by fitting raw lipid headgroup center of mass positions to a smooth spline curve.

Membrane deformations for simulated nhTMEM16 structures.

Left column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Middle column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Right column: the sum of the upper and lower leaflet deformations, representing the bilayer thickness along z. In all plots, grey areas indicate grid points with lipid occupancy <2%. The black outline is the projected surface of the upper (z>0) or lower (z<0) portion of the protein dimer.

Membrane deformations for simulated afTMEM16 structures.

Left column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Middle column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Right column: the sum of the upper and lower leaflet deformations, representing the bilayer thickness along z. In all plots, grey areas indicate grid points with lipid occupancy <2%. The black outline is the projected surface of the upper (z>0) or lower (z<0) portion of the protein dimer.

Membrane deformations for simulated TMEM16K structures.

Left column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Middle column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Right column: the sum of the upper and lower leaflet deformations, representing the bilayer thickness along z. In all plots, grey areas indicate grid points with lipid occupancy <2%. The black outline is the projected surface of the upper (z>0) or lower (z<0) portion of the protein dimer.

Membrane deformations for simulated TMEM16F structures.

Left column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Middle column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Right column: the sum of the upper and lower leaflet deformations, representing the bilayer thickness along z. In all plots, grey areas indicate grid points with lipid occupancy <2%. The black outline is the projected surface of the upper (z>0) or lower (z<0) portion of the protein dimer.

Membrane deformations for simulated TMEM16A structures.

Left column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Middle column: xy-map of the distance along the z-axis from the bilayer midplane to the ensemble averaged positions of the glycerol linker (GL1 and GL2 beads). Right column: the sum of the upper and lower leaflet deformations, representing the bilayer thickness along z. In all plots, grey areas indicate grid points with lipid occupancy <2%. The black outline is the projected surface of the upper (z>0) or lower (z<0) portion of the protein dimer.

TM4 moves away from starting structure coordinates in open states.

(A) Overlayed CG representations of experimentally determined or simulated starting structures (grey) and snapshots from CG simulations. nhTMEM16 4WIS (green), afTMEM16 7RXG (violet), TMEM16K 5OC9 (orange), TMEM16F 8B8J (blue), and TMEM16A 5OYB* (red) with minimum TM4-6 distances for the starting CG structure (grey) and mean values with standard deviation from simulations. (B) Violin plots of minimum distances between TM4 and TM6 in a single groove with median value (cyan square) and 25-75% quartiles (cyan bars). (C) TMEM16F TM4/TM6 groove constriction point at residues F518H (on TM4) and W619 (on TM6). (D) TMEM16K TM4/TM6 groove constriction point at Y366 (on TM4).

Contact analysis of lipid headgroup high density sites identified from previous AA simulation.

(A-C) Charged residues and nearby POPC lipids near two high lipid phosphate density sites at the intracellular and extracellular entry of the open nhTMEM16 (PDB ID 4WIS) canonical groove (“SC”, “SE”) identified from previous AA simulation © 2016, Bethel & Grabe, published by PNAS (36). (C) Contact frequency with any lipid (grey bars) and only scrambling lipids (cyan bars). Dwell times with scrambling lipids shown as black points. (D) The nhTMEM16 CG backbone colored by total number of contact events with any lipid. Grey spheres indicate lipid headgroup positions in a single snapshot.

Position traces of scrambling lipids with total lipid headgroup density.

Lipid traces and lipid headgroup density (yellow) for Ca2+ -bound scrambling competent TMEM16s: nhTMEM16 (PDB ID 4WIS, green), afTMEM16 (PDB ID 7RXG, purple), TMEM16K (PDB ID 5OC9, orange), TMEM16F F518H (PDB ID 8B8J, blue) and simulated TMEM16F (6QP6*). Each image is viewed from the extracellular or cytosolic (TMEM16K) space. Lipid traces are generated by fitting raw lipid headgroup center of mass positions to a smooth spline curve.

Average duration of interaction between scrambling lipids and TM4/TM6 groove lining residues.

The canonical groove of each experimentally solved (A) and simulated (B) open scramblase structure is colored by the average duration of each interaction (dwell time) between scrambling lipids and groove lining residues. The distribution of average dwell times at individual residues is shown as a histogram below each structure.

Dwell time distribution and contact frequency for TM4/TM6 groove lining residue across homologs.

Contact frequency (cyan bar, left y-axis) and distribution of interaction dwell times (black scatter dots, right y-axis) between scrambling lipids and canonical groove lining residues with the 15 longest average interaction dwell times. Residues are sorted by the contact frequency. Frequency of contact with any lipid (scrambling and non-scrambling lipids taken together) is shown as grey bar. The red dashed-line rectangle indicates two previously identified residues near high lipid phosphate density in an all-atom (AA) simulation (36).

Lipids enter the dimer interface in atomistic and CG simulations of TMEM16F.

Left: a snapshot of a 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) lipid that entered the TMEM16F (PDB ID 6QP6) dimer interface from the outer leaflet during an all-atom (AA) simulation. Right: snapshots at the same timepoint of 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) lipids that entered the TMEM16F (PDB ID 6QM6) dimer interface during a CG simulation. Nearby side chains are colored by residue type: basic (red), acidic (blue), and polar (green).

Dimer interface hydrophobicity and lipid positions.

Images in each row were taken from CG MD simulations of 5 different TMEM16s. The first column depicts TM helices forming one half of the dimer interface (rest of the protein not shown) colored by residue type (small/hydrophobic: white, charge/polar: blue). The second column depicts snapshots of the same dimer interface helices with overlayed positions of water (light blue spheres) and lipid head groups (red and blue spheres) every 100 frames of the last 9 μs of each simulation. The last two columns are two views of the same snapshot showing the protein with its annulus of lipids (yellow). The dimer interface-forming helices are colored green (nhTMEM16), purple (afTMEM16), blue (TMEM16F), orange (TMEM16K), and red (TMEM16A).

Dwell time analysis for scrambling events observed at TMEM16F F518H mutant dimer interface.

The scrambling region is delineated by TM3, TM4, TM5, TM9, and TM10 both monomers. TM3 and part of TM5 are transparent for clarity. The backbone region of the dimer interface is colored by dwell time of the scrambling lipids. Sites with prolonged dwell time are circled in cyan (center left) and shown in zoomed-in images (center right). Distributions of dwell times at each site are shown as violin plots (far right).