Coarse-Grained Molecular Dynamics Simulations of Membrane

Aug 16, 2016 - It is well established that trehalose (TRH) affects the physical properties of lipid bilayers and stabilizes biological membranes. We p...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/JPCB

Coarse-Grained Molecular Dynamics Simulations of Membrane− Trehalose Interactions Jon Kapla, Baltzar Stevensson, and Arnold Maliniak* Department of Materials and Environmental Chemistry, Arrhenius Laboratory, Stockholm University, 106 91 Stockholm, Sweden ABSTRACT: It is well established that trehalose (TRH) affects the physical properties of lipid bilayers and stabilizes biological membranes. We present molecular dynamics (MD) computer simulations to investigate the interactions between lipid membranes formed by 1,2-dimyristoyl-sn-glycero-3phosphocholine (DMPC) and TRH. Both atomistic and coarse-grained (CG) interaction models were employed, and the coarse graining of DMPC leads to a reduction in the acyl chain length corresponding to a 1,2-dilauroyl-sn-glycero-3phosphocholine lipid (DLPC). Several modifications of the Martini interaction model, used for CG simulations, were implemented, resulting in different potentials of mean force (PMFs) for DMPC bilayer−TRH interactions. These PMFs were subsequently used in a simple two-site analytical model for the description of sugar binding at the membrane interface. In contrast to that in atomistic MD simulations, the binding in the CG model was not in agreement with the two-site model. Our interpretation is that the interaction balance, involving water, TRH, and lipids, in the CG systems needs further tuning of the force-field parameters. The area per lipid is only weakly affected by TRH concentration, whereas the compressibility modulus related to the fluctuations of the membrane increases with an increase in TRH content. In agreement with experimental findings, the bending modulus is not affected by the inclusion of TRH. The important aspects of lipid bilayer interactions with biomolecules are membrane curvature generation and sensing. In the present investigation, membrane curvature is generated by artificial buckling of the bilayer in one dimension. It turns out that TRH prefers the regions with the highest curvature, which enables the most favorable situation for lipid−sugar interactions.



INTRODUCTION Nature exhibits different strategies for the protection of organisms against environmental stress. Small sugars frequently act as regulators that stabilize biomembrane structures. The disaccharide trehalose (TRH) is known for its involvement in membrane stabilization processes in many living organisms,1 ranging from plants and fungi2,3 to bacteria and invertebrates.4−6 The research in this area in the last 30 years mainly focused on the cryoprotective and anhydrobiotic properties of TRH; in industry, TRH is typically used as a protective agent to preserve food and biomaterials at room temperature as well as under freeze-drying conditions. There are several views on the membrane−sugar interactions involved in biological processes. The research field can be summarized into two partially opposing views:7 (1) the direct interaction model, in which the sugars are in direct contact with the membrane surface, thus replacing water molecules8,9 and (2) the exclusion model, in which the sugars are kept out of the surface region, maintaining the hydration level of the bilayer. It is concluded that both models can be valid at the same time and that the exclusion of sugars takes precedence over the direct interaction in a concentration region where the lipid bilayer surface becomes saturated with sugar molecules. These findings were in fact supported by our atomistic molecular dynamics (MD) simulations of TRH−bilayer systems.10 From analyses of the trajectories, we found indications of strong attraction of TRH molecules to the bilayer surface at low to © 2016 American Chemical Society

moderate amounts of sugar. If scrutinized further, the direct interactions of TRH with the lipid bilayer can be divided into three main hypotheses: (a) sugar molecules substitute water in the headgroup region of the lipids, (b) sugar molecules interact with both the lipid headgroups and water molecules, keeping the hydration shell of the bilayer intact, and (c) vitrification of sugar molecules on the top of the bilayer. The recent development indicates that all three of these mechanisms may not be mutually exclusive.7,10,11 In addition, it has been suggested, based on neutron diffraction data,12 that TRH does not strongly associate with membranes and that the direct interactions are not particularly important for the protective properties of this sugar. The models and hypotheses used for the description of membrane−sugar interactions are based on a wealth of experimental and theoretical results. In particular, the changes in the physical properties of the lipid membranes related to these interactions have been investigated using scattering methods,13−16 thermochemical approaches,7,17−19 magnetic resonance techniques,11,20−22 and atomic force microscopy,23 just to mention a few. Investigation of membrane−sugar interactions using computational methods in general and MD simulations in particular is an increasingly active field of Received: June 30, 2016 Revised: August 10, 2016 Published: August 16, 2016 9621

DOI: 10.1021/acs.jpcb.6b06566 J. Phys. Chem. B 2016, 120, 9621−9631

Article

The Journal of Physical Chemistry B

Figure 1. Molecules used in the simulations: (a) TRH, (b) CG representation of TRH, (c) DMPC, and (d) CG representation of DLPC. The dashed arrows in (c) and (d) indicate the long axis of the lipid used for splay calculations.

Table 1. Summary of the Simulated Systemsa label

TRH

lipids

H2O

water model

AA-0 MP-0 MP-TRHL P3-TRHL P2-TRHL PP-TRHL 00-TRHL MP-TRHH P3-TRHH PP-TRHH 00-TRHH AA-PMF MA-PMF MP-PMF P3-PMF AA-MET MP-MET P2-MET

0 0 720 720 720 720 720 1440 1440 1440 1440 2 2 2 2 2 2 2

1152 1152 1152 1152 1152 1152 1152 1152 1152 1152 1152 128 128 128 128 0 0 0

90 000 18 000 31 500 31 500 31 500 31 500 31 500 31 500 31 500 31 500 31 500 5000 1593 1593 1593 2000 500 500

TiP3P PW PW PW PW PW PW PW PW PW PW TiP3P W PW PW TiP3P PW PW

comments Slipids + GLYCAM Martini; polarized water model P4 → P3 P2 → SP2 P4 → P3 and P2 → SP2 TRH−TRH interactions turned off P4 → P3 P4 → P3 and P2 → SP2 TRH−TRH interactions turned off umbrella sampling umbrella sampling umbrella sampling umbrella sampling, P4 → P3 metadynamics metadynamics metadynamics, P2 → SP2

PX → PY indicates change of bead from PX to PY, the superscripts, L = low (720) and H = high (1440), refer to the content of TRH molecules. The models denoted W and PW are the default water model in the Martini force field and the polarized water model, respectively. The simulations labeled MA and MP correspond to the default Martini force field and that where polarized water model is used, respectively. a

parameters within the scope of the force field, we compare bilayer-specific properties and free energies of the binding of TRH to the lipid bilayer to tune the CG results toward the atomistic findings. In particular, we calculate the PMF, which reflects the free energy of interaction between a single TRH molecule and the bilayer. Eighteen systems were simulated, differing in the level of interaction details, simulation parameters, and molecular composition. The reference atomistic simulation consisted of a bilayer with 1152 1,2-dimyristoylsn-glycero-3-phosphocholine (DMPC) lipids, solvated using 90 000 TiP3P water molecules (corresponding to 78 water molecules per lipid). The CG simulations consisted of a lipid with shorter alkyl chains (12 instead of 14 carbon atoms), 1,2dilauroyl-sn-glycero-3-phosphocholine (DLPC) lipid, solvated using water beads described by the standard and polarizable water models in Martini. The CG lipid−TRH systems were simulated with two contents (720 and 1440) of TRH

research.10,22,24−31 Common to most recent publications are the relatively small systems simulated (typically 128 lipids), which may induce finite-size effects that influence the reliability of the estimated properties such as diffusion32,33 and area fluctuations. 34 In addition, the TRH order parameters calculated from trajectories generated in our recent atomistic simulations22 were 3 orders of magnitude higher than the corresponding parameters derived from nuclear magnetic resonance (NMR) experiments. This indicates that the short time scale of the simulations and the number of molecules are insufficient to obtain proper averaging of molecular motions. Furthermore, the exchange of TRH molecules between the membrane surface and water bulk is not adequately sampled. In the present work, we aim to overcome these shortcomings by using larger system sizes and extend the time scale using coarse-grained (CG) MD simulations with parameters from the extensively used Martini force field.35 Using different sets of 9622

DOI: 10.1021/acs.jpcb.6b06566 J. Phys. Chem. B 2016, 120, 9621−9631

Article

The Journal of Physical Chemistry B molecules. In addition, we simulated smaller systems for the calculation of free energies for TRH−TRH interactions. The molecules and their descriptive nomenclature are shown in Figure 1, and the simulation details are collected in Table 1. Finally, we consider curvature sensing and formation in lipid bilayers, which is a rapidly developing research area in biophysics and membrane physical chemistry. In particular, proteins,36,37 peptides,38,39 nanoparticles,40 and also carbohydrates41−43 interact differently with the biomembrane depending on the local curvature. The effect of the curvature on the molecular interactions can, in principle, be monitored by considering aggregates with different geometrical shapes, such as spherical vesicles or cylinders. However, here we create three bilayers with different degrees of buckling generated by squeezing the membrane in one dimension. These bilayers interact with one TRH molecule, and we monitor the probability distribution for finding the TRH molecule in regions with different local curvatures.



RESULTS AND DISCUSSION We start this section by considering the complex molecular interactions in the ternary systems consisting of water, TRH, and lipids. In particular, we determine the PMF for TRH− membrane interactions subsequently used in a simple two-site model, which, in turn, describes the adsorption of sugar molecules on the bilayer surface. Furthermore, we calculate the free energy for TRH−TRH interactions in an isotropic aqueous solution. The physical properties of lipid bilayers are characterized by calculating the area per lipid and area compressibility modulus together with splay and bend moduli, which describe various deformation processes. We close this section by considering one TRH molecule interacting with bilayers exhibiting different curvatures, modeled by artificial squeezing (buckling) of the membrane. TRH−Bilayer Interaction. In Figure 2a, the PMF determined from the umbrella sampling simulations is displayed. The center of masses of the two TRH molecules (one on each side of the bilayer) are constrained to a position along the bilayer normal using an umbrella-type potential, limiting the translational freedom of the molecules to a window in the xy plane. Clearly, there are substantial differences between the PMFs calculated using different interaction models. The MA-PMF system, in which the standard water model in Martini is employed, shows a small minimum at 1.7 nm with the interaction energy of