Multi-scale coarse graining and molecular dynamics simulations of vesicle formation

Abstract

The basic mechanisms of living cells have attracted an increased interest in recent years, triggered perhaps by the prospect of creating life in the laboratory through synthetic biology. This once-futuristic idea has become realistic through scientific progress in the past decade. One of the main challenges in current synthetic biology is the design of a so-called cell, a cell with only those parts which are essential for its survival. An important feature of such a minimal cell is its casing, as a boundary with the environment. Amphiphilic, self-aggregating molecules such as phospholipids and fatty acids may form a membrane to serve as such a casing, in the shape of spontaneously formed spherical vesicles. To understand the self-aggregation behavior of these membrane components, it is important to understand their molecular properties and the intermolecular interactions. However, not all of these properties can be observed experimentally, which has lead to the use of in silico methods, such as molecular modeling, in which molecular systems are represented by computational models. A frequently used technique to model the dynamic behavior of a system is molecular dynamics, where successive configurations are generated by integrating Newton’s laws of motion. Molecular dynamics may be used with detailed atomistic models, but also with so-called coarse grained models, where groups of atoms are represented by single coarse grained particles. Coarse grained models are often employed to study systems with longer time and length scales, for which the calculation time in an atomistic representation is unfeasibly long. However, in coarse grained modeling it can be challenging to find suitable model parameters. Therefore, in our work we develop a new coarse graining method. Furthermore, we employ (coarse grained) molecular dynamics to investigate the properties of lipid membranes. In the first part of this thesis, we introduce a novel method to coarse grain an atomistic simulation, the CUMULUS coarse graining method. Combined with the iterative Boltzmann inversion procedure, this coarse graining method can be employed to derive coarse grained force fields in a multi-scale approach, in order to reproduce structural properties from simulations at the atomistic level. This approach is applied on systems containing pure water, sodium chloride solutions, and water–octanol mixtures. Also, the CUMULUS method is tested on two systems containing larger molecules, poly-norbornene in chloroform and a poly(propylene imine) dendrimer in water. Importantly, the obtained force fields are found to be transferable to systems of different composition. Having developed this method, we investigate the role of protonation in the behavior of oleic acid vesicles. First, in an atomistic simulation of a periodic oleic acid membrane, a strong hydrogen bonding network between protonated and deprotonated species of oleic acid is observed. Next, a coarse grained oleic acid model is constructed with the CUMULUS coarse graining method and the iterative Boltzmann inversion procedure. In this model, two different headgroup types are used to represent the protonated and deprotonated species of the molecules. Although this model is parametrized successfully, the phase behavior of the coarse grained oleic acid membranes does not match that of experiments. The second part of this thesis is aimed at understanding the behavior of lipid vesicles and their encapsulation of biomacromolecules. We employ an existing coarse grained model of lipids and water to perform simulations of the spontaneous transition of a flat to a spherical vesicle. Our investigations reveal that this transition follows a molecular pathway we denominate bilayer bulging. Our analysis indicates that the inward forces exerted by the solvent on the edge of the membrane are the main driving force behind this bulging pathway. Upon addition of water-soluble proteins to these simulations, it is found that variation of the nonbonded interaction between the proteins and the membrane surface significantly affects the encapsulation efficiency. When the protein–membrane interaction is neutral, the encapsulated protein concentration is below that of the solution. Increasing this interaction results in an increase of the encapsulation efficiency to values above those of the solution, with a linear relationship between the encapsulation efficiency and the strength of adhesion of the proteins to the membrane surface. Furthermore, our simulations indicate that the protein encapsulation efficiency does not depend on the size of the proteins nor on the speed of vesicle formation. In this work we have extended the toolbox of coarse grained molecular modeling and designed new simulations to study lipid membranes. Furthermore, we have investigated the properties of lipid membranes at the molecular level and gained insight into the formation of vesicles and their encapsulation of proteins. Our findings help to construct the required theoretical framework to understand the behavior of lipid membranes in their application as cell membranes for artificial cells.