Stochastic simulation of structural properties of natively unfolded and denatured proteins.

A new simulation strategy based on a stochastic process has been developed and tested to study the structural properties of the unfolded state of proteins at the atomistic level. The procedure combines a generation algorithm to produce representative uncorrelated atomistic microstructures and an original relaxation method to minimize repulsive non-bonded interactions. Using this methodology, a set of 14 unfolded proteins, including seven natively unfolded proteins as well as seven "classical" proteins experimentally described in denaturation conditions, has been investigated. Comparisons between the calculated and available experimental values of several properties, at hydrodynamic and atomic level, used to describe the unfolded state, such as the radius of gyration, the maximum length, the hydrodynamic radius, the diffusion coefficient, the sedimentation coefficient, and the NMR chemical shifts, reflect a very good agreement. Furthermore, our results indicate that the relationship between the radius of gyration and the hydrodynamic radius deviates from the Zimm's theory of polymer dynamics for random coils, as was recently observed using single-molecule fluorescent methods. Simulations reveal that the interactions between atoms separated by three chemical bonds (1-4 interactions) play a crucial role in the generation process, suggesting that the unfolded state is essentially governed by bonding and short-range non-bonding interactions.