Transition states for protein folding using molecular dynamics and experimental restraints

The mechanism through which a given sequence of amino acids finds its way to a global free energy minimum cannot yet be predicted by theory or numerical simulation. Much of the information available on the protein folding mechanism derives from the so-called values. These are believed to probe the structure of the rate limiting step, or transition state, for the folding of two-state proteins. In recent years experimental values have been widely used to benchmark the results of simulations, mostly of unfolding, which have been achieved using detailed sequence-dependent molecular models. A few years ago a novel technique was proposed which uses values as restraints so that only conformations which are transition-state-like are sampled in the simulation. This technique, albeit grounded on several approximations and assumptions, has provided an unprecedented structural representation of the transition state for folding. Here we explore various issues concerning the generation of ensembles of structures representing the transition state. One important result is that by allowing a large tolerance on the experimental restraints the information contained in the latter is lost; this suggests that an experimental error on the values which is too large might affect the results of restrained simulations and the picture provided by them.