Direct observation of Bin/amphiphysin/Rvs (BAR) domain-induced membrane curvature by means of molecular dynamics simulations

Abstract

The process of membrane curvature generation by BAR (Bin/amphiphysin/Rvs) domains is thought to involve the plastering of the negatively charged cell membrane to the positively charged concave surface of the BAR domain. Recent work [Peter, B. J., et al. (2004) Science, 303,495-499; Masuda, M., et al. (2006) EMBO J. 25, 2889-2897; and Gallop, J. L., et al. (2006) EMBO J. 25, 2898-2910] has demonstrated the importance of the charged, crescent-shaped surface and the N-terminal amphipathic helices (present in N-BAR domains) for generating membrane curvature. These experiments suggest that curvature is generated by the synergistic action of the N-terminal helices embedding in the lipid bilayer and the charged crescent-shaped dimer acting to "scaffold" membrane curvature. Here, we present atomistic molecular dynamics simulations that directly show membrane binding to the concave face of N-BAR domains, resulting in the generation of local membrane curvature that matches the curvature presented by the BAR domain. These simulations provide direct molecular-scale evidence that BAR domains create curvature by acting as a scaffold, forcing the membrane to locally adopt the intrinsic shape of the BAR domain. We find that BAR domains bind strongly through the maximum curvature surface and, additionally, at an orientation that presents a lesser degree of curvature, thus enabling N-BAR domains to induce a range of local curvatures. Finally, we find that the N-terminal region may play a role in biasing the orientations of N-BAR domains on the membrane surface to those that favor binding to the concave face and subsequent membrane bending.

Fig. 1.

Time evolution of membrane curvature with N-BAR domain binding and orientation. (a) Membrane curvature development for NBR1 (solid line), NBR2 (dashed line), and plain lipid bilayer (dotted line) in the region occupied by the BAR domain. Curvature is calculated from the discretized membrane shape (see Fig. 3b) at each time point. (b) Angle between the BAR domain concave binding surface and the membrane surface for NBR1 (solid line) and NBR2 (dashed line). The angle θ_BAR is the angle between the y coordinate axis (which is parallel to the membrane surface) and the BAR domain principal axis that is (initially) most nearly parallel to the y coordinate axis (see Fig. 2). (c) Number of lipid head group oxygen atoms within 4.2 Å of an Arg or Lys nitrogen (minimum association time of 1 ns; 50-ps sampling interval) for NBR1 (solid line) and NBR2 (dashed line).

Fig. 2.

Simulation snapshots of N-BAR domains inducing local membrane curvature. (a) Snapshot of simulation NBR1 at t = 10 ns. (b) Snapshot of simulation NBR1 at t = 27 ns. (c) Snapshot of simulation NBR2 at t = 10 ns. (d) Snapshot of simulation NBR2 at t = 27 ns. In simulation NBR1, the N-BAR domain maintains its concave surface facing the membrane and induces a higher degree of curvature. In simulation NBR2, the N-BAR domain tilts along the y coordinate direction (into the plane of the image) and presents a less concave surface to the membrane. In both cases, the membrane bends to match the curvature of the N-BAR domain surface facing the membrane. Charged phosphatidylserine head groups are shown in purple, and polar phosphatidylcholine head groups are green. (Solvent is not shown.)

Fig. 3.

Membrane and BAR domain curvatures and geometries. (a) Average curvature along the length of the membrane for NBR1 (solid line), NBR2 (dashed line), and the plain lipid bilayer (dotted line) calculated from the average membrane shape shown in b. The membrane shape is calculated by projecting the center of mass (COM) of the upper leaflet lipid head groups onto the x–z plane and averaging the COM in bins (n = 25) along the x axis. (c) Average curvature along the length of the BAR domain in simulation NBR1 calculated from the COM discretization (n = 8) of the BAR domain shape that is superimposed over the average atom positions in d. The number of points for the BAR domain discretization was chosen to be commensurate with the length scale of the membrane discretization. All quantities were averaged from 20 to 27 ns. Error bars represent the standard deviation calculated from 1-ns block averages.

Fig. 4.

Effect of the N-terminal region on binding orientation. Shown is the angle (θ_BAR) between the concave binding surface and the surface of the membrane for BAR domains and N-BAR domains on neutral (DOPC) or charged (DOPC/DOPS) membranes. N-terminal regions are not embedded in the membrane. The presence of the N-terminal region tends to restrict the angle of association of the BAR domain with the membrane.