A pre-existing hydrophobic collapse in the unfolded state of an ultrafast folding protein.

Insights into the conformational passage of a polypeptide chain across its free energy landscape have come from the judicious combination of experimental studies and computer simulations. Even though some unfolded and partially folded proteins are now known to possess biological function or to be involved in aggregation phenomena associated with disease states, experimentally derived atomic-level information on these structures remains sparse as a result of conformational heterogeneity and dynamics. Here we present a technique that can provide such information. Using a 'Trp-cage' miniprotein known as TC5b (ref. 5), we report photochemically induced dynamic nuclear polarization NMR pulse-labelling experiments that involve rapid in situ protein refolding. These experiments allow dipolar cross-relaxation with hyperpolarized aromatic side chain nuclei in the unfolded state to be identified and quantified in the resulting folded-state spectrum. We find that there is residual structure due to hydrophobic collapse in the unfolded state of this small protein, with strong inter-residue contacts between side chains that are relatively distant from one another in the native state. Prior structuring, even with the formation of non-native rather than native contacts, may be a feature associated with fast folding events in proteins.