Membrane curvature sensing and symmetry breaking of the M2 proton channel from Influenza A
- Cardiovascular Research Institute, Department of Pharmaceutical Chemistry, University of California, San Francisco, United States
- Graduate Group in Biophysics, University of California, San Francisco, United States
Figures
Figure 1
M2 channels and influenza egress.
(A) Structural organization of the nascent viral bud. M2 first accumulates in the host plasma membrane (PM), which is approximately flat and therefore has zero Gaussian curvature. M2 then diffuses to the neck of a nascent viral bud, which has negative Gaussian curvature characterized by principal curvatures, κ1 and κ2, of opposite sign. Finally, a small minority of the initial M2 population migrates into the outward-budding, convex cap of the mature virion, which is otherwise enriched in viral spike proteins neuraminidase (NA) and hemagglutinin (HA). Labels e and c highlight the extracellular face of the PM and the cytosolic face. (B) A sample of published M2 structures (indicated by PDB accession codes) grouped by reported symmetry (top row: C4 symmetry vs. bottom row: C2 symmetry) and relative position of their amphipathic helix (AH) domains: 2L0J on the far left turns sharply and lies perpendicular to the transmembrane (TM) domain, with the AH domain becoming less associated with the TM domain until 6OUG on the far right shows the AH continuing parallel to the TM helices.
Figure 2
Unrestrained simulations reveal dynamic amphipathic helix (AH) domains.
(A) RMSDs of Ca backbones over time. Yellow indicates the transmembrane (TM) domain only, red the AH domain only, and blue the full protein. Snapshots from unrestrained molecular dynamics (MD) are numbered along the X-axis with the starting structure overlaid in light gray, showing AH dynamics. (B) Double electron–electron resonance (DEER) data provided by the Howard lab (green) compared to distance estimates from simulation between labeled AH domain residues (blue), including the first 5 ns alone (yellow). Insets in upper and lower panels show the I51 (top) and F55 (bottom) residue on 2L0J (view from cytoplasm), with adjacent and diagonal distances represented by two black lines.
-
Figure 2—source data 1
Raw double electron–electron resonance (DEER) data from Figure 7 of Kim et al., 2015 (green curves in panel B).
- https://cdn.elifesciences.org/articles/81571/elife-81571-fig2-data1-v2.xlsx
Figure 3
Influence of lipid packing on bilayer tension and shape.
Comparison of tension and structural features of simulated membranes with the initial packing (left, panels A, C) versus the rebalanced leaflets (right, panels B, D). (A) Lateral pressure profile of the initial restrained 2L0J simulation, with bilayer midplane centered at z = 0. Upper leaflet in z > 0 range; lower leaflet in z < 0 range. Leaflet tensions and thicknesses at box edge (defined by calculating the average thickness excluding a 60 Å × 60 Å square cutout centered around the protein) at right. (B) Same as panel A for the rebalanced restrained 2L0J simulation. (C) Hydrophobic surfaces (top row) and leaflet thicknesses (bottom row) for the initial simulation. White cutout areas with black edges represent the footprint of the protein from the last frame of the simulation. (D) Same as panel C for the rebalanced simulation. Note that protein footprints in C and D differ due to motion of protein side chains.
Figure 4
M2 membrane deformation patterns from simulation.
(A) Structures for restrained-protein simulations. Extracellular loops in green, transmembrane (TM) domain in yellow, amphipathic helix (AH) domain in blue (polar/charged) and white (hydrophobic). Left: the fourfold solid-state NMR structure (PDBID 2L0J); center: parallel AH domain model (see main text for details); right: twofold solid-state NMR structure (PDBID 2N70). (B) Upper leaflet (extracellular face, e) and lower leaflet (cytosolic face, c) mean hydrophobic surfaces computed from molecular dynamics (MD; purple). Polar and charged side chains shown in blue, hydrophobic side chains in white. Structure of the parallel AH domain construct used in simulations, built from PDBID 2N70.
-
Figure 4—source data 1
Structure of the parallel amphipathic helix (pAH) model.
- https://cdn.elifesciences.org/articles/81571/elife-81571-fig4-data1-v2.zip
Figure 5
Surface height and thickness heatmaps for three different channel configurations.
Compressive deflections are negative/shown in bright colors. Expansive deflections are positive/dark colors. The same scale is used for the upper and lower leaflets to highlight the greater amount of deflection in the upper as compared to the lower leaflet. Arrows correspond to regions of high (red) and low (blue) compression noted in the text.
Figure 6
Leaflet thicknesses and position of the bilayer midplane.
Upper and lower leaflet thicknesses are plotted as time-averaged surface values in the top and middle rows, with lighter shades of blue denoting regions of leaflet thinning. The height of the bilayer midplane is plotted in the bottom row. For each system, the midplane height is calculated in reference to the average position of the four L36 residues of the tetramer. These residues sit at exactly the midplane height (z = 0) in the 2L0J simulation. Deflections above this height are shown in purple, deflections below are shown in orange.
Figure 7
Lipid tilt around different M2 conformations.
The top row shows representative all-atom snapshots extracted from the protein-restrained, equilibrium simulations of 2L0J, parallel amphipathic helix (AH) domain model, and 2N70 (2, 3, and 4 in Table 1, respectively). Black stars in each snapshot highlight purple lipids discussed in the main text. Two-dimensional tilt surfaces computed for each simulation in the top row separated out by upper (middle row) and lower (bottom row) leaflets. The mean lipid tilt at a given position in the X–Y plane with respect to the Z-axis (membrane normal) is reported in degrees and color coded according to the scale on the right. Cholesterol is excluded from the tilt calculations. Molecular images of the proteins are oriented as in Figure 4, with the viewpoint origin in the lower right quadrant of the tilt surfaces shown in the middle and bottom rows. Full size images are included as source data.
-
Figure 7—source data 1
Full size image of lipids around 2L0J.
- https://cdn.elifesciences.org/articles/81571/elife-81571-fig7-data1-v2.pdf
-
Figure 7—source data 2
Full size image of lipids around the parallel amphipathic helix (AH) domain model.
- https://cdn.elifesciences.org/articles/81571/elife-81571-fig7-data2-v2.pdf
-
Figure 7—source data 3
Full size image of lipids around 2N70.
- https://cdn.elifesciences.org/articles/81571/elife-81571-fig7-data3-v2.pdf
Figure 8
Mean deflection and tilt as a function of distance from the protein center.
Mean deflection (A) and tilt (B) as a function of radial distance from the protein center for restrained simulations. Points are colored by local density, such that bright spots reflect highly sampled values of deflection/tilt versus distance. At a fixed value of distance, a vertical scan highlights the range of mean deflections/tilts sampled and the nature of the distribution of mean deflections/tilts seen at that distance.
Figure 9
Boundary conditions extracted from molecular dynamics (MD) simulations.
(A) Parallel amphipathic helix (AH) domain M2 protein in a lipid bilayer. The molecular surface is shown with hydrophobic residues in white and charged and polar residues in blue. The mean hydrophobic core of membrane, with equilibrium thickness Lc, is between the purple surfaces of the upper/extracellular leaflet e and lower/cytosolic leaflet c. The mid-plane surface between the upper and lower leaflets is CM. The insets show the boundary conditions for the upper leaflet extracted at the lipid-excluded surface of the protein: (top inset) the vertical displacement from equilibrium of the upper leaflet u+ and (bottom inset) the slope normal to the protein. (B) Membrane upper and lower bounds of the mean hydrophobic core at the protein. Solid blue: the upper and lower mean leaflet locations; dashed red: fourth-order Fourier approximation. Dotted black: equilibrium leaflet positions. The boundary parameter is a measure of the distance around the boundary between the lipid-excluded surface of the protein and the membrane leaflet. (C) Solid and dotted blue: slopes of the membrane normal to the boundary. Dashed red: fourth-order Fourier slope approximations.
Figure 10
Continuum model membrane deformations compared to molecular dynamics (MD) surfaces.
Left column: MD upper and lower leaflet mean positions. Right column: continuum model minimum energy upper and lower leaflet surfaces for a flat membrane (200 Å by 200 Å membrane with zero displacement and slope on the outer boundary – entire patch not shown), with inner boundary conditions at the protein taken from the MD surfaces in the left column. (A) 2L0J. (B) Parallel AH domain model (pAH). (C) 2N70.
Figure 11
C2 symmetry broken conformations are stabilized in membranes with negative Gaussian curvature.
(A) Transfer free energy ΔΔG(K) for moving M2 from a flat region to a region with Gaussian curvature K = ±1/Rc2, where Rc is the radius of curvature (top X-axis). Solid lines at negative K show ΔΔG for three M2 models in saddles. Solid (dashed) lines at positive Gaussian curvature show ΔΔG in concave (convex) membranes. Labeled points B, C, and D on the plot indicate the energy for the membranes shown in panels B–D. For all shapes, K1 = ±K2. (B) Parallel amphipathic helix (AH) domain model in a saddle. Inset: rotated close-up view of the lower leaflet distortion around the AH domain. (C) 2L0J in a concave spherical membrane. (D) 2L0J in a convex spherical membrane.
Appendix 1—figure 1
Leaflet surfaces (left and middle columns) and bilayer thicknesses (right column) for the original restrained 2L0J simulation and restrained Repack simulations 1–4.
The upper leaflet has 200 lipids, and the lower leaflet lipid count is indicated in parentheses.
Appendix 1—figure 2
Leaflet tilt surfaces for the original restrained 2L0J simulation and restrained Repack simulations 1–4.
The upper leaflet has 200 lipids, and the lower leaflet lipid count is indicated in parentheses.
Appendix 1—figure 3
Leaflet thicknesses for the original restrained 2L0J simulation and restrained Repack simulations 1–4.
The upper leaflet has 200 lipids, and the lower leaflet lipid count is indicated in parentheses.
Appendix 1—figure 4
Amphipathic helix (AH) domain mobility in Repack 4, unrestrained 2L0J simulations.
(A) Trajectory RMSDs for the AH domain. Independent replicates are color coded. (B) Trajectory RMSDs for the transmembrane (TM) domain. (C) I51–I51 distance distributions from unrestrained simulations. Dashed curves are color coded per trajectory. The aggregate distance distribution across all replicates is solid blue. DEER data from the Howard lab in solid green (see MT Figure 2B for source data). Vertical dashed lines represent the adjacent and diagonal distances for I51 in 2L0J. (D) F55–F55 distance distributions with the same color coding as panel C.
Appendix 1—figure 5
Boundary conditions extracted from molecular dynamics (MD) simulations.
(A, B) Membrane–protein boundary information for the fourfold 2L0J structure. (C, D) Membrane–protein boundary information for the twofold 2N70 structure. Upper and lower bounds of the mean hydrophobic membrane core (defined by the surface separating the lipid headgroups from the acyl chain) along the membrane–protein contact curve for 2L0J (A) and 2N70 (C). Solid blue: mean MD membrane hydrophobic boundaries. Dashed red: fourth-order Fourier series approximation of the MD boundary. Slopes of the mean MD hydrophobic surfaces at the protein in the direction normal to the boundary for 2L0J (B) and 2N70 (D). Blue solid (dotted): upper (lower) leaflet slopes. Dashed red: fourth-order Fourier series approximation of the MD slopes. The boundary parameter is the length around the boundary, normalized to .
Appendix 1—figure 6
Influence of cholesterol enrichment on membrane curvature sensing by different M2 channel conformations.
Membrane parameters from Table 2 were used to calculate the membrane bending energy for moving from a flat membrane region to regions of differing curvature. (A) A 0% cholesterol membrane with 28.5 Å core thickness. (B) A 50% cholesterol membrane with a 35-Å core thickness. The 30% cholesterol case is shown in Figure 11A. Positive Gaussian curvatures correspond to concave (solid curves) or convex (dashed curves) spherical caps, while negative Gaussian curvature values correspond to saddles.
Appendix 1—figure 7
Membrane shapes which approximate an ideal spherical cap or saddle.
(A) Dash–dotted red: circle of radius R = 200 Å. Dashed blue: quadratic approximation of circle, valid for . Solid black: the m = 0 analytic biharmonic solution finite at the origin behaves quadratically for . (B) Dashed blue: the ideal saddle is quadratic along the X and Y axes with radii of curvature and , respectively. Solid black: the m = 2 analytic biharmonic solution diverges from quadratic when Å. The numeric solutions from the membrane solver are not shown because they are indistinguishable from the analytic solutions at this scale.
Tables
Table 1
List of simulations.
| ID | Label | Gromacs | Length (ns) | PDBID | Bilayer # lipids | Restraints |
|---|---|---|---|---|---|---|
| 1 | Unrestrained 1 | 2020.6 | 2829 | 2L0J | 200 upper; 150 lower | No |
| 2 | 2L0J 1 | 2020.6 | 3808 | 2L0J | 200 upper; 150 lower | Yes |
| 3 | pAH 1 | 2020.6 | 3760 | Parallel AH | 200 upper; 150 lower | Yes |
| 4 | 2N70 1 | 2020.6 | 1727 | 2N70 | 200 upper; 150 lower | Yes |
| 5 | 2L0J Repack 1 | 2020.6 | 2740 | 2L0J | 200 upper; 158 lower | Yes |
| 6 | 2L0J Repack 2 | 2020.6 | 2500 | 2L0J | 200 upper; 166 lower | Yes |
| 7 | 2L0J Repack 3 | 2020.6 | 2890 | 2L0J | 200 upper; 174 lower | Yes |
| 8 | 2L0J Repack 4 – Final | 2020.6 | 5000 | 2L0J | 200 upper; 183 lower | Yes |
| 9 | Unrestrained 2 | 2020.6 | 2650 | 2L0J | 200 upper; 183 lower | No |
| 10 | Unrestrained 3 | 2020.6 | 3000 | 2L0J | 200 upper; 183 lower | No |
| 11 | Unrestrained 4 | 2020.6 | 3000 | 2L0J | 200 upper; 183 lower | No |
| 12 | Unrestrained 5 | 2020.6 | 3000 | 2L0J | 200 upper; 183 lower | No |
| 13 | pAH Repack Final | 2020.6 | 3000 | Parallel AH | 200 upper; 183 lower | Yes |
| 14 | 2N70 Repack Final | 2020.6 | 3000 | 2N70 | 200 upper; 183 lower | Yes |
Table 2
Default elastic membrane material properties.
| Parameters | % cholesterol | Values | Reference |
|---|---|---|---|
| Membrane thickness (LC) | 0 | 28.5 Å | Argudo et al., 2017 |
| 30 | 35 Å | This manuscript | |
| 50 | 37 Å | Scaling from Ferreira et al., 2013 | |
| Surface tension (α) | 0 | 3.0 × 10−13 N/Å | Latorraca et al., 2014 |
| 30 | 3.0 × 10−13 N/Å | “ ” | |
| 50 | 3.0 × 10−13 N/Å | “ ” | |
| Bending modulus (KC) | 0 | 29 kBT | Average value described in Methods |
| 30 | 65 kBT | Scaling from Henriksen et al., 2006 | |
| 50 | 68 kBT | Scaling from Pan et al., 2009 | |
| Gaussian modulus (KG) | 0 | −26 kBT | Relation from Hu et al., 2012 |
| 30 | −56 kBT | “ ” | |
| 50 | −61 kBT | “ ” | |
| Areal compression modulus (Ka) | 0 | 2.13 × 10−11 N/Å | Henriksen et al., 2006 |
| 30 | 3.55 × 10−11 N/Å | “ ” | |
| 50 | 3.73 × 10−11 N/Å | Scaling described in Methods |
Appendix 1—table 1
Leaflet properties for restrained 2L0J simulations with varying lipid numbers.
| Simulation | Area per lipid (Å2) | Thickness at box edges (Å) | Tension (mN/m) | |||
|---|---|---|---|---|---|---|
| (lower lipid #) | Upper leaflet | Lower leaflet | Upper leaflet | Lower leaflet | Upper leaflet | Lower leaflet |
| Original (150)* | 42.5 | 49.9 | 18.5 | 16.3 | −10.8 | 24.6 |
| Repack 1 (158) | 43.8 | 49.0 | 18.2 | 15.8 | −13.1 | 20.2 |
| Repack 2 (166) | 44.4 | 47.4 | 17.9 | 16.4 | −11.8 | 13.2 |
| Repack 3 (174) | 45.2 | 46.1 | 17.8 | 17.0 | −9.6 | 7.6 |
| Repack 4 (183)† | 46.4 | 45.1 | 17.4 | 17.2 | −3.8 | 8.0 |
-
*
The original restrained 2L0J simulation used a semi-isotropic Berendsen barostat for pressure coupling. All repacked simulations, and extension simulations for Gromacs-LS calculations, use a Parrinello–Rahman barostat. Other simulation parameters were identical for all simulations.
-
†
Lipid ratio used for all subsequent 2L0J analysis.
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Membrane curvature sensing and symmetry breaking of the M2 proton channel from Influenza A
eLife 13:e81571.
https://doi.org/10.7554/eLife.81571
