A simple coarse-grained model for self-assembling silk-like protein fibers

Collagen-silk-collagen triblock polypeptides can self-assemble at low pH into nanometer thin fibers with a length in the order of micrometers. Previously we predicted, via all-atom simulations, the structure of the folded silk domain to be a beta-roll. In this work we develop a simple coarse-grained model of the silk domain to enable a numerical study of the fiber's properties and formation on a larger length and time scale. As an initial coarse-grained model for the fiber forming protein we chose the model of Brown et al., Proc. Natl. Acad Sci. U.S.A., 2003, 100, 10712-10717. We adapted this model, and optimized its parameters to reproduce the all-atom molecular dynamics simulation structural data. The unknown strength of the attraction between the beads representing the residues is optimized by computing the Potential of Mean Force for unfolding a strand of the beta-roll, using non-equilibrium steered MD simulations in combination with the Jarzynski relation. Using these optimized parameters we observed spontaneous folding of a short peptide. The coarse-grained beta-roll, as well as a much larger stack (a fiber) of beta-rolls, were found to be stable. Moreover, the predicted fiber persistence length is in agreement with experiment. The efficacy of the mapping of a coarse-grained system onto an all-atom simulation is discussed. The approach opens the way for large-scale simulations of fibers, based on molecular structure, and allows investigation of their nucleation, growth, cross-linking mechanism, network dynamics, and rheology.