Design and Development of a Special-Purpose in silico Solvent for Modeling Lipid-Protein Interaction in a Highly Mobile Membrane Mimetic

Joshua Vermaas, University of Illinois at Urbana-Champaign

Despite the growing knowledge of the role of lipid molecules in regulation of membrane proteins, describing their detailed interaction with proteins has proven extremely challenging experimentally. While molecular dynamics simulations offer a powerful alternative, the slow dynamics of lipid molecules have significantly diminished the effectiveness of the method in this context. This problem was addressed to a large degree by a novel membrane representation, termed the Highly Mobile Membrane Mimetic (HMMM) developed in our lab. In HMMM, lateral diffusion of lipid molecules has been significantly enhanced by replacing the long acyl chains in the membrane interior with an organic solvent. While our original solvent choice, 1,1-dichloroethane (DCLE), accurately captures the dynamics of peripheral membrane proteins, extending the application of the model to transmembrane systems requires an optimized liquid solvent that better reflects the properties of the membrane interior while still enhancing lateral lipid diffusion. Finding conventional organic solvent models unsatisfactory, we have designed monoatomic and diatomic liquid Lennard-Jones in silico solvents based on methylene groups. Through an extensive search of parameter space, our solvent model matches the density and compressibility of the membrane interior. These novel in silico solvent models improve the agreement between the insertion free energy of amino acids for conventional and HMMM membranes, while retaining the high lateral diffusion essential to accelerated lipid-protein sampling. This is a critical step in extending the application of this fast membrane model to phenomena involving integral membrane proteins. We have also tested the stability of five integral membrane proteins with different folds within an HMMM membrane, including the new solvents, and all proteins remained stable during the simulations. Given the enhanced lipid diffusion in the HMMM model, these simulations have allowed us to provide a detailed description of distinct lipid-protein interactions.

Abstract Author(s): Josh Vermaas and Emad Tajkhorshid