Predicting Transmembrane (TM) Domain Dimer Structures using Martini 3

Determination of the structure and dynamics of transmembrane (TM) regions of single-transmembrane receptors is key to understanding their mechanism of signal transduction across the plasma membrane. Although many studies have been performed on isolated soluble extra- and intracellular receptor domains in aqueous solutions, limited knowledge exists on the lipid embedded TM domain. In this study, we examine the assembly of configurations of receptor TM region dimers using the Martini 3 force field for coarse-grain (CG) molecular dynamics simulations. This recently published version of Martini has new bead types and sizes, which allows more accurate predictions of molecular interactions compared to the previous versions. At first glance our results with Martini 3 simulations show only a reasonable agreement with ab initio predictions using PREDDIMER (for TM domains only), AlphaFold2 Multimer and with available NMR derived structures for TM helix dimers. Specifically, 6 of 11 CG TM structures are similar to the NMR structures (within < 3.5 [A] mainchain RMSD) compared to 10 of 11 and 9 of 11 using PREDDIMER and Alphafold, respectively (7 structures of the latter are within 1.5 [A]) Surprisingly, AlphaFold2 predictions are more comparable with NMR structures when the database of 2001 (mainly composed of soluble proteins) instead of 2020 PDB structures are used. While there are some differences in the conditions used, the CG simulations primarily reveal that alternate configurations of the TM dimers that are sampled, which readily interconvert with a predominant population. The implications of these findings for our understanding of the signalling mechanism of TM receptors are discussed, including opportunities for the development of new pharmaceuticals, some of which are peptide based.