A Coarse-Grained Model For Protein Backbone Dynamics

When one wants to simulate the folding of proteins or the dynamics of intrinsically disordered proteins, atomistic simulations very soon become prohibitvely expensive. This is even more so when multiple unfolded proteins are considered as for example in the case of amyloid formation. For these scenarios coarse-grained models are required. Here we present a newly developed hierarchic coarse-grained model which builds upon Langevin Dynamics. Using this very efficient solvent-free simulation technique allows for more freedom in the modelling than the united-atom-approaches with their still spherical super-atoms. Here we show how we base the model of the protein backbone on non-spherical building blocks with off-center bonds. By construction then the allowed regions in the Ramachandran angle space are reproduced. Non-local steric interactions, electrostatics, and hydrogen bonds form a second layer and shift the secondary structure propensities according to the residue types. We also demonstrate that poly-peptides form alpha-helical or beta-strand structures according to their amino-acid composition even when we start from random initial configurations. This model can then be applied to efficiently simulate the folding and association of multiple unfolded proteins like A-beta or alpha-synuclein.