All-atom and coarse-grained molecular dynamics simulations of a membrane protein stabilizing polymer

Abstract

Amphipathic polymers called amphipols (APols) have been developed as an alternative to detergents for stabilizing membrane proteins (MPs) in aqueous solutions. APols provide MPs with a particularly mild environment and, as a rule, keep them in a native functional state for longer periods than do detergents. Amphipol A8-35, a derivative of polyacrylate, is widely used and has been particularly well studied experimentally. In aqueous solutions, A8-35 molecules self-assemble into well-defined globular particles with a mass of ∼40 kDa and a R(g) of ∼2.4 nm. As a first step towards describing MP/A8-35 complexes by molecular dynamics (MD), we present three sets of simulations of the pure APol particle. First, we performed a series of all-atom MD (AAMD) simulations of the particle in solution, starting from an arbitrary initial configuration. Although AAMD simulations result in stable cohesive particles over a 45 ns simulation, the equilibration of the particle organization is limited. This motivated the use of coarse-grained MD (CGMD), allowing us to investigate processes on the microsecond time scale, including de novo particle assembly. We present a detailed description of the parametrization of the CGMD model from the AAMD simulations and a characterization of the resulting CGMD particles. Our third set of simulations utilizes reverse coarse-graining (rCG), through which we obtain all-atom coordinates from a CGMD simulation. This allows a higher-resolution characterization of a configuration determined by a long-timescale simulation. Excellent agreement is observed between MD models and experimental, small-angle neutron scattering data. The MD data provides new insight into the structure and dynamics of A8-35 particles, which is possibly relevant to the stabilizing effects of APols on MPs, as well as a starting point for modeling MP/A8-35 complexes.

Figure 1

A) Chemical Structure of the ungrafted, octylamine grafted, and isopropylamine grafted units, with the coarse-grained mapping. B) Grafting sequences simulated (white = ungrafted, black = octylamine grafted, gray = isopropylamine grafted).

Figure 2

Illustration of the multi-scale strategy employed in the present study. First, AAMD simulations are executed and analyzed. Second, the results from the AAMD simulations are used to parametrize a CGMD model. Third, coordinates from CGMD simulations are then used to start rCG (all-atom) simulations, allowing higher resolution characterization of equiilbrated particles.

Figure 3

Comparison of the bond distance distributions from the all-atom and coarse-grained models (AAMD = black, CGMD = gray).

Figure 4

Comparison of the angle and dihedral angle distributions from all-atom and coarse-grained models (AAMD = black, CGMD = gray).

Figure 5

Snapshots from CGMD simulation illustrating de novo particle assembly (blue = ungrafted, red = octylamine grafted, gray = isopropylamine grafted, water and ions removed for clarity).

Figure 6

Component radial density distributions describe the organization of the particle, contrasting the different simulation strategies.

Figure 7

The number of water molecules found within 0.5 nm of the amphipol center of mass (AAMD = black, rCG = gray).

Figure 8

Radii of gyration over time, describing structural relaxation and contrasting the different simulation strategies.

Figure 9

Particle shape is described by fitting the structure to an ellipsoid and calculating the semiaxial ratio (a/c), with 1 indicating a spherical particle and larger values indicating elongation.

Figure 10

SANS from experimental ensemble (gray triangles) [Gohon et al. 2006], AAMD simulations (solid), and rCG simulation (dotted). The vertical dotted lines indicate the Guinier region, in which the linear slope of these profiles can be related to the radius of gyration.

Figure 11

Mean squared displacement from amphipol simulations (AAMD and rCG) compared with POPC lipid bilayer and SDS detergent micelle, for peripheral atoms (A) and hydrocarbon chain terminals (B).

Figure 12

A) Histogram of radius of gyration from 97 CGMD simulations using randomly generated sequences. B) These simulations suggest a relationship between particle shape (quantified by the a/c semiaxial ratio) and sequence. Black = random sequences, Blue = averages, Red = well-characterized sequences (from left to right Homogeneous, Random, and Block).