How Reliable are Molecular Dynamics Simulations of Membrane Active Antimicrobial Peptides

Abstract

Membrane-active antimicrobial peptides (AMPs) are challenging to study experimentally, but relatively easy to investigate using molecular dynamics (MD) computer simulations. For this reason, a large number of MD studies of AMPs have been reported over the recent years. Yet relatively little effort has focused on the validity of such simulations. Are these results reliable, and do they agree with what is known experimentally? And how much meaningful information can be obtained? To answer these questions, we demonstrate here some of the requirements and limitations of running MD simulations for several common AMPs: PGLa, melittin, maculatin and BP100. The two most important findings are: (a) Simulation results depend strongly on force field parameters, making experimental verification of the simulations obligatory, and (b) slow orientational and conformational fluctuations mean that much longer sampling timescales (multi-μs) are needed if quantitative agreement between simulation averages and experimental data is to be achieved.