LetterThe following article is Open access

DPD simulation to reproduce lipid membrane microdomains based on fragment molecular orbital calculations

, , , , , and

Published 11 June 2024 © 2024 The Author(s). Published on behalf of The Japan Society of Applied Physics by IOP Publishing Ltd
, , Citation Hideo Doi et al 2024 Appl. Phys. Express 17 055001DOI 10.35848/1882-0786/ad4955

1882-0786/17/5/055001

Abstract

Lipid domains play a critical role in signal transduction and transport across cell membranes. The formation of domains in "HLC" ternary lipid bilayers composed of high transition temperature (high-Tm) lipids, low-Tm lipids, and cholesterol (Chol) has been extensively studied as a raft-like system. Recently, experiments were performed to control the formation of submicron domains in LLC lipid bilayers containing low-Tm phosphatidylethanolamine (PE), low-Tm phosphatidylcholine (PC), and Chol by manipulating the presence or absence of Chol. The formation of microdomains in this LLC mixture was replicated by dissipative particle dynamics simulation. The results show that domain formation can be replicated.

Export citation and abstractBibTeXRIS

Content from this work may be used under the terms of the Creative Commons Attribution 4.0 license. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

The cell membrane is a complex structure composed of lipids and membrane proteins. Its basic structure is the lipid bilayer. 1,2) Lipids with diverse physicochemical properties regulate physical properties such as molecular packing, fluidity, curvature, and hydration. 35) The organization of lipid molecules affects the dynamics and physical properties of bilayers. The physical properties and dynamics described in this context are responsible for essential cellular functions, such as membrane transport and intracellular signal transduction. 69) The cell membrane contains multicomponent lipids that are heterogeneously distributed due to the preferential distribution of certain lipid molecules and membrane proteins. The formation of lipid domains has been extensively studied to understand the response mechanisms of cell membranes and related pathologies. 10) Lipid domains are regions composed of specific lipids. The most common representative is the "raft," which is composed of sphingolipids and phospholipids with saturated acyl chains and cholesterol. 1116)

The use of artificial lipid bilayers has been instrumental in studying the fundamental properties and biological processes of cell membranes. 1125) Previous studies 17) have referred to ternary mixtures containing high transition temperature (high-Tm ), low Tm (low-Tm ), and cholesterol as "HLC" mixtures, which serve as models for lipid rafts. Various microscopic and spectroscopic techniques have been used to study HLC mixtures extensively. 1121) At room temperature, lipids can be classified into two phases: the gel phase and liquid crystalline phase. The gel phase contains high-Tm lipids, such as sphingomyelin and saturated phosphatidylcholine, while the liquid crystalline phase includes low-Tm lipids. When high-Tm and low-Tm lipids are mixed, they separate into gel and liquid crystalline phases with a wide range of lipid fractions. Incorporation of cholesterol leads to liquid-liquid phase separation, which occurs in the temperature range between the melting temperatures (Tm ) of high-Tm and low-Tm lipids. This results in the formation of two liquid phase domains: one containing high-Tm lipids and a cholesterol-rich phase, and the other containing low-Tm lipids. Studies of high- and low-Tm lipid mixtures have provided insight into the molecular principles of raft formation.

One significant difference between artificial lipid bilayers of HLC mixtures and cell membranes is the composition of their lipids. HLC bilayers are primarily composed of lipids with saturated and monounsaturated acyl chains, whereas biological membranes contain a significant proportion of polyunsaturated lipids. The influence of polyunsaturated lipids on lipid bilayers may play an important role in the biological and physiological functions of cell membranes. Several experimental 2631) and simulation studies 30,3235) have investigated the relationship between polyunsaturated lipids and cholesterol. These studies suggest that polyunsaturated lipids differ from monounsaturated and saturated lipids in that their acyl chains are essentially disordered. Wassall and Stillwell proposed the formation of non-rafted domains enriched in polyunsaturated lipids due to the low miscibility between cholesterol and phospholipids with docosahexaenoyl groups. 28,30) However, experimental studies characterizing the local compositional organization and elucidating the effects of polyunsaturated lipids on domain formation remain largely unexplored.

Goh and Tero recently reported the formation of a domain in a ternary lipid bilayer composed of low-Tm phosphatidylethanolamine (PE), low-Tm phosphatidylcholine (PC), and cholesterol (Chol). 36,37) This ternary lipid mixture is referred to as a "LLC" mixture and serves as a specific fusion site for proteoliposomes. 38) The PE+PC+Chol ternary bilayer can retain the activity of several ion channels. 39) The two low-Tm lipids mix uniformly to form a single lipid bilayer without any phase separation. 37) The formation of submicron domains at a temperature above the Tm of the two Low-Tm lipids in the presence of about 30 mol% Chol is a novel phenomenon that has not been reported previously. The research topic was to investigate the composition and mechanism of non-raft domain formation. The study aimed to investigate how the effect of the number and position of double bonds in PE affect domain formation in LLC mixed bilayers of polyunsaturated PE, monounsaturated POPC, and Chol. 36,37) It was found that the degree of unsaturation of polyunsaturated PE and the sn-position distribution of the double bonds play a crucial role in domain formation. Experimental evidence showed that separation of polyunsaturated PE from the bilayer composed of monounsaturated PC and Chol resulted in the formation of non-rafted domains.

In recent years, there has been a growing need to develop highly functional materials in the fields of materials science and drug discovery. To achieve this, researchers commonly use dissipative particle dynamics (DPD) and coarse-grained molecular dynamics simulations. In particular, DPD simulations are used to predict mesostructures, simulate lipid membranes, 4046) polymer electrolyte membranes, 47) drug delivery systems, 48) and to design material properties. The simulations involve the analysis and design of each molecule and particle, which requires the use of χ parameters to describe their interactions. Typically, these parameters are derived from experimental data or classical molecular mechanics (MM). 49,50) Alternatively, the parameters can be calculated using FCEWS, 51,52) a workflow system that uses the fragment molecular orbital (FMO) calculations 53,54) to reliably perform non-empirical DPD simulations 42,47,5557) (referred to as FMO-DPD). FCEWS could thus provide robust support for DPD simulations of unknown phenomena.

The importance of studying phase separation phenomena by simulation is widely recognized. However, simulations present several challenges. For example, the accurate description of phase separation in simulations depends on several factors, including initial conditions and interaction parameters. In addition, the behavior of biomolecules such as polyunsaturated lipids is challenging due to their complexity, diverse interactions, and limited experimental data. These factors complicate efforts to fully understand phase separation phenomena. Therefore, performing DPD simulations after parameter determination using FCEWS is a useful approach.

The present study aims to simulate microdomain formation induced by polyunsaturated lipids using a series of DPD simulations (with a set of FMO-derived parameters). The non-empirical nature of the method provides an advantage in simulating delicate behaviors that depend on the difference of lipid species. The investigation focuses on five types of polyunsaturated phospholipids, POPC, and Chol, as shown in Fig. 1. The main points of interest are the degree of unsaturation and the arrangement of the double bond to the two carbon chains. The five types of polyunsaturated phospholipids are referred to as 0+2PE to 6+6PE based on their degree of unsaturation.

Fig. 1. Refer to the following caption and surrounding text.

Fig. 1. The molecular structures of five polyunsaturated phospholipids, POPC, and cholesterol were the subject of DPD simulations. The area inside the blue line represents 1DPD bead treated as molecular fragments.

Standard image High-resolution image

Goh and Tero observed lipid bilayers composed of polyunsaturated phospholipids, POPC, and Chol using fluorescence microscopy, atomic force microscopy, and force-distance curve measurements. They found that polyunsaturated PE causes the formation of domains, where the domains are rich in PE composition. 36) The distribution of double bonds into two chains is more likely to form domains in the order of 6+6PE, 3+3PE, 0+6PE, 1+1PE, and 0+2PE, as shown in Fig. 1. In addition, Goh and Tero found differences in domain formation between 3+3PE and 6+6PE, and between 0+6PE and 3+3PE. The main objective of the present study is to replicate this trend in domain formation of polyunsaturated fatty acids with and without cholesterol molecules using DPD simulation analysis.

The DPD model was created using segments of polyunsaturated phospholipids, POPC, and Chol as shown in Fig. 2. The χ parameters representing the interaction of these DPD models were calculated and set using the FCEWS system. 51,52) The calculation of the χ parameters was performed at the standard FMO-MP2 level with the c.c.-pVDZ 58) basis functions, where the ABINIT-MP program 59,60) was used. The fragmentation details of each molecule, the DPD model, and the χ parameter are described in the Supporting Information (SI). The DPD simulations used approximately 100 000 particles with a dimensionless cell size of 32.2 (22.9 nm) and a DPD time step of 0.025. A total of 100 000 steps were performed, corresponding to 140 ns. The temperature was set to 1 (300 K). The simulation was also performed with the membrane as the initial structure. The composition of the membranes was 82% water, 14% POPC, and 4% PE (membrane components: POPC:PE = 78:22) without cholesterol, and 85% water, 11.4% POPC, 3.6% PE and 2% cholesterol (POPC:PE:Chol = 67:21:12) with cholesterol. To maintain membrane tension, the ratio of membrane particles to total particles was adjusted to approximately 17%. The membrane structure was created using J-OCTA, 61) and DPD simulations were performed using COGNAC. 62)

Fig. 2. Refer to the following caption and surrounding text.

Fig. 2. Schematic of the molecular structure and DPD model of unsaturated fatty acids. (a) Molecular structure, (b) DPD model of 0+2PE, (c) molecular structure, (d) DPD model of cholesterol.

Standard image High-resolution image

First, we performed DPD simulations in a system of polyunsaturated phospholipids and POPC without Chol to confirm the absence of domain formation. The simulation results for five types of polyunsaturated PEs and POPC molecules, with PEs in purple and POPC in light blue, are shown in Fig. 3. All polyunsaturated phospholipids (PEs) remained stable as bilayers and were mixed with POPC molecules without forming a phase-separated or aggregated domain structure. The number of PE clusters (blue area) counted in Figs. 3(a)–3(e) was approximately 80. This lack of domain formation is in agreement with the experimental observations. 36)

Fig. 3. Refer to the following caption and surrounding text.

Fig. 3. Snapshots of DPD simulation with POPC and PE blends. (a) 0+2PE, (b) 1+1PE, (c) 0+6PE, (d) 3+3PE and (e) 6+6PE are used as PE. PE is shown in purple and POPC in light blue.

Standard image High-resolution image

Next, the simulations shown in Fig. 4 were aimed at determining the presence of domains in polyunsaturated phospholipids, POPC, and Chol. 36) The results showed that no domain formation was observed in the case of 0+2PE and 1+1PE. However, very small domains appeared to be formed in the case of 0+6PE. Experiments also showed that 0+6PE caused more domain formation than 0+2PE and 1+1PE, and the area fraction of the submicron domain was 5%–10% of 3+3PE. Simulation confirmed that microdomains are present in both 3+3PE and 6+6PE, with larger domain formation observed in 6+6PE. Figure 4 shows the number of PE clusters: (a) 91, (b) 94, (c) 48, (d) 32, and (e) 7, respectively. This result is consistent with the experimental trend that 6+6PE forms microdomains several times larger in area than 3+3PE.

Fig. 4. Refer to the following caption and surrounding text.

Fig. 4. Snapshot of a DPD simulation with POPC, PE and cholesterol mixtures. (a) 0+2PE, (b) 1+1PE, (c) 0+6PE, (d) 3+3PE, and (e) 6+6PE are used as PE. PE is shown in purple, POPC in light blue, and cholesterol in yellow.

Standard image High-resolution image

To explain these phenomena in terms of free energy, we compare the χ parameter, which is a dimensionless free energy parameter. The χ parameter between particles of propane, butane, and cholesterol is about −3, while the χ parameter between particles of 2-butene, 1-butene, and propene, which are unsaturated carbon chains, is about 5, 8, and 8, respectively. These particles repel cholesterol. The main difference between the carbon chain interaction of Chol-1+1PE and Chol-POPC, also between Chol-0+2PE and Chol-POPC is the presence of a double bond, and the interaction energy and domain formation do not differ significantly. The results correspond to the previous AFM observation that shows no domain formation in the 1+1PE+POPC+Chol or 0+2PE+POPC+Chol bilayer. 36) On the other hand, Chol-6+6PE has 12 particles with double bonds, which promotes phase separation. Comparing 0+6PE and 3+3PE, there is little difference in the repulsive interaction of their double bond parts. However, there is a difference in the attractive interaction. 0+6PE is more attractive because existence of one saturated carbon chain drastically increase the affinity for Chol. 29) As a result, phase separation seems to be enhanced in 3+3PE, even with the same number of double bonds. These results confirm that polyunsaturated PE causes domain formation. The domains have a PE-rich composition, and double bonds are more likely to form domains when they are split into double strands.

In the present study, five types of polyunsaturated phospholipids and cholesterol molecules were analyzed using FMO-DPD simulations from two perspectives: the LL mixture without Chol and the LLC mixture. The simulation results for LL and LLC mixtures of five different PEs with different degrees of unsaturation are in agreement with experimental observations. No domains were observed in the LL mixtures regardless of the PE species. In the LLC mixtures, no domain formation was observed in the 0+2PE and 1+1PE mixtures, while microdomains appeared in the 0+6PE, 3+3PE and 6+6PE mixtures. The size of the domains increased in the order of 0+6PE, 3+3PE and 6+6PE, which is consistent with the experimental trends. The dependence of domain formation on the degree of unsaturation of PE was also discussed in terms of χ parameters.

FMO-DPD has been used in the past in lipid systems to obtain results in agreement with experiments, 42,56) and again good agreement was obtained. The drawback of this method is the cost of parameter determination by FMO calculations, but this is becoming more efficient with the introduction of machine learning. 63) Our intention is to further demonstrate the effectiveness of this method by applying it to different lipid systems.

Acknowledgments

This work was supported by Rikkyo SFR. All calculations with FCEWS were performed on the supercomputer MASAMUNE-IMR (Project Nos. 20S0024 and 2012SC0008).

undefined