[Paper Review] Efficient Simulation of Tunable Lipid Assemblies Across Scales and Resolutions

John M. A. Grime, Jesper J. Madsen|arXiv (Cornell University)|Oct 11, 2019

Lipid Membrane Structure and Behavior49 references18 citations

TL;DR

This paper presents a computationally efficient, particle-based coarse-grained model for simulating tunable lipid assemblies across multiple scales and resolutions. Using soft, short-ranged pairwise potentials and a modular bead representation (2–5 beads per lipid, plus a quasi-monolayer 1.5-bead model), the model accurately forms micelles, bilayers, and vesicles with tunable phase behavior, demonstrating improved numerical stability and dynamics over existing models while capturing key membrane physics such as bending and phase transitions.

ABSTRACT

We present a minimal model for simulating dynamics of assorted lipid assemblies in a computationally efficient manner. Our model is particle-based and consists of coarse-grained beads put together on a modular platform to give generic molecular lipids with tunable properties. The interaction between coarse-grained beads is governed by soft, short-ranged potentials that account for the renormalized hydrophobic effect in implicit solvent. The model faithfully forms micelles, tubular micelles, or bilayers (periodically infinite, bicelles, or vesicles) depending on the packing ratio of the lipid molecules. Importantly, for the self-assembled bilayer membranes it is straightforward to realize gel and fluid phases, the latter of which is most often of primary interest. We show that the emerging physics over a wide range of scales and resolutions is (with some restrictions) universal. The model, when compared to other popular lipid models, demonstrates improvements in the form of increased numerical stability and boosted dynamics. Possible strategies for customizing the models (e.g., adding chemical specificity) are briefly discussed. An implementation is available for the LAMMPS molecular dynamics simulator [Plimpton. J. Comp. Phys. 117, 1-19. (1995)] including illustrative input examples from the simulations we present.

Motivation & Objective

• To develop a computationally efficient, minimalistic model for simulating diverse lipid assemblies across scales and resolutions.
• To improve numerical stability and dynamics compared to existing coarse-grained lipid models, particularly those using hard-core or Lennard-Jones potentials.
• To enable tunable membrane properties such as fluid and gel phases, packing ratios, and bending behavior through adjustable interaction parameters.
• To provide a modular, extensible framework for simulating membrane-associated biological processes with customizable chemical specificity.
• To demonstrate the model's universality in reproducing large-scale morphological transitions and membrane physics.

Proposed method

• The model uses a modular particle-based framework with 2–5 beads per lipid, representing head, interface, and tail regions with distinct interaction parameters.
• Non-bonded interactions are governed by a soft, short-ranged pair potential: U_pair(r) = - (A/a) sin(r·a) for r ≤ r₀ and - (B/b) sin(π/2 + (r - r_c)·c) for r₀ < r < r_c, with smooth decay to zero at r_c.
• Intramolecular forces are modeled via harmonic bond (K_b(r - r₀)²) and angle (K_θ(θ - θ₀)²) potentials to maintain lipid chain rigidity.
• The hydrophobic effect is implicitly renormalized into cohesive interactions between interface and tail beads, while head beads have purely repulsive interactions.
• The model is implemented in LAMMPS and validated through simulations of micelles, bilayers, vesicles, and in silico pulling experiments.
• Vesicle packing is performed using a constant volume per lipid (VPL) strategy to avoid instability and reduce need for flip-flop relaxation.

Experimental results

Research questions

• RQ1 Can a minimal, coarse-grained lipid model with soft, short-ranged potentials self-assemble into diverse morphologies such as micelles, bilayers, and vesicles?
• RQ2 How does the model’s dynamics and numerical stability compare to established models like SDK or Lennard-Jones-based approaches?
• RQ3 To what extent can the model reproduce key membrane properties such as phase behavior (fluid vs. gel), bending, and morphological transitions?
• RQ4 Can the model’s tunability be leveraged to simulate complex biological processes at reduced computational cost?
• RQ5 How does the choice of lipid packing strategy (constant APL vs. VPL) affect vesicle stability and equilibration?

Key findings

• The model successfully forms micelles, tubular micelles, and bilayers (including periodic, bicellar, and vesicular forms) based on lipid packing ratio, with self-assembly occurring without external constraints.
• The fluid bilayer phase is stable and exhibits realistic dynamics, with the ability to tune the phase behavior by adjusting lipid flexibility and interaction parameters.
• The quasi-monolayer model (1.5 beads per lipid) demonstrates high computational efficiency and enables in silico pulling experiments on vesicles, revealing membrane bending and mechanical properties.
• The model shows improved numerical stability and boosted dynamics compared to models using $1/r^{12}$ or Lennard-Jones potentials, particularly at highly coarse-grained resolutions.
• Constant volume per lipid (VPL) packing strategy produces more stable vesicles than constant area per lipid (APL), minimizing the need for flip-flop relaxation and reducing structural artifacts.
• The soft, short-ranged potential ensures smooth force decay without truncation or shifting, enhancing stability and enabling long-time simulations.