Predictive folding of a β-hairpin protein in an all-atom free-energy model

Using an all-atom free-energy model we reproducibly and predictively fold a monomeric stable tryptophane-zipper (pdb-code 1LE1) from unfolded starting conformations to experimental accuracy. We construct the folding free-energy landscape under physiological conditions with a fraction of the cost of methods that simulate the folding pathway. We identify a metastable helical ensemble of conformations in the folding funnel and discuss its impact on the folding scenario. A comparison of the energetic contributions of the competing ensembles rationalizes the stabilization of the native ensemble and illustrates opportunities for peptide design.