Effects of a Polarizable Force Field on Membrane Dynamics: Surface Viscosity, Lipid Diffusion, and Peptide Induced Pore Formation

The effects of the newly developed CHARMM polarizable lipid force field (FF), Drude2023, on selected lipid dynamical properties are compared with the additive CHARMM36 (C36), and an extension of C36, termed C36/LJ-PME, which includes long-range Lennard-Jones (LJ) interactions. Polarizability and long-range LJ interactions increase the membrane surface viscosity, which decreases the translational diffusion constants. Simulated diffusion constants for dipalmitoylphosphatidylcholine (DPPC) and dioleoylphosphatidylcholine (DOPC) extrapolated to infinite system size agree well with experiment for Drude2023, but overestimate experiment by 60% (on average) and a factor of 2.5 for C36/LJ-PME and C36, respectively. The relaxation time of lipid wobble is described about equally well by C36/LJ-PME and Drude2023, as consistent with the hexadecane viscosity for the FF, and both are more accurate than C36. Hence, physical improvements in the FF, which slowed down these dynamic processes led to better agreement with experiment. Lastly, bilayers containing 10 influenza fusion peptides and high fractions of lysolipids (which are known to accelerate pore formation) were simulated with C36 and Drude2023. Pore formation rates were comparable for the two FF for the bilayers with 80% and 90% lysolipid. However, while no pores formed in 24 μs (including a single 20 μs trajectory) in the 70% lysolipid with C36, 4 of 15 replicates formed pores in less than 1 μs with Drude. While the pathway to poration is qualitatively similar for the additive and polarizable FF for the systems studied, Drude2023 should be considered for quantitative studies of pore formation, and, in some cases, will accelerate the process.