Molecular simulation of discoidal HDL lipoprotein particles

Abstract

High-density lipoprotein (HDL) has been determined to play an essential role as an atheroprotective in atherosclerosis – the thickening and hardening of arteries. HDL particles are important delivery vehicles in the reverse cholesterol transport (RCT) pathway wherein excess cholesterol is removed from peripheral tissues and transported to the liver for redistribution or removal from the body. The particles undergo a sequential development from being lipid-free comprising only the protein Apolipoprotein A-I (ApoA-I), to discoidal HDL, and finally forming spherical HDL. ApoA-I appears to stabilise the particles as well as influencing the activity of the enzymes ABCA and LCAT, and recognition of the HDL particles by the scavenger receptors in the liver. Apolipoprotein Zaragoza (ApoA-I Z) and Apolipoprotein Milano (ApoA-I M) are natural mutations of Apolipoprotein A-I. Carriers of these mutations have a lower level of HDL-cholesterol and yet a low risk of cardiovascular disease. Discoidal HDL (dHDL) particles take the form of a lipid disk stabilised by two ApoA-I molecules surrounding the edges. Here we have investigated the structure and physical characterisation of dHDL using molecular dynamics simulations. Being soft matter, the structure of dHDL and its self-assembly is still not entirely resolved. We have employed molecular dynamics simulation (using both atomistic and coarse-grained models), including enhanced sampling methods (temperature replica exchange, Hamiltonian replica exchange, Jarzynski’s non-equilibrium, metadynamic, and umbrella sampling approach) to explore the interactions of ApoA-I proteins in isolation, self-assembly of dHDL particles, and the free energy surface for the chain-chain interaction of ApoA-I proteins within the dHDL complex. Simulations have been carried out on the wild type and the ApoA-I mutants ApoA-I Z and ApoA-I M. The simulations of ApoA-I monomers and dimers in water solutions indicate that wild-type ApoA-I is more stable and more rapid in changing conformation than ApoA-I Z and ApoA-I M mutants. With respect to the self-assembly of dHDL, the standard MD simulations do not converge to equilibrium as the emergent structure becomes kinetically locked. The use of thermal replica exchange is also ineffective, being inefficient for large systems. Hamiltonian exchange wherein the ApoA-I protein chains are gradually transformed to generate a soft-core potential was observed to be more effective and indeed generated the double-belt structure with the experimentally known helix 5/2 registration. The free energy surface for the ApoA-I – ApoA-I pair interaction/registry in dHDL was inaccessible by Jarzynsky’s non-equilibrium approach and metadynamics but could be characterised using umbrella sampling. The surface suggests that H5/2 is the most stable form in both ApoA-1 and ApoA-I Z in dHDL.