Understanding and Designing Protein Beta-Sheets

Our goal is a quantitative algorithm for protein design which is not limited to particular protein folds. In this endeavor there have been previous successes designing protein cores, where van der Waals packing, and the tendency of hydrophobic amino acids to avoid contact with solvent, are the dominant forces. On the surfaces of proteins, efforts at α-helix surface design have also been successful, where hydrogen bonding and α-helix propensities are additionally important. However, there are no algorithmically designed stable β-sheet surfaces. One of the energy terms expected to be important for β-sheet surface design is β-sheet propensity. No concise theory explaining the amino acids' differing β-sheet propensities has previously been developed. In this thesis, I examine the underlying physical-chemical basis for β-sheet propensities, and show that they are caused primarily by van der Waals interactions between the side chains and the local backbone. I then consider an additional energy term, a penalty for the exposure of hydrophobic surface area. This is not a thermodynamic term, but rather one that can be justified through "negative design," in which alternative badly folded ground state structures are disfavored. I show experimentally that this term improves the algorithm's predictive ability, and determine its strength in the context of our previously published energy expression. In order to do this, I developed a two body approximation for buried and exposed surface area calculation which very closely reproduces the true surface areas. Finally, I develop a general method for calculation of the optimal energy expression for protein design, from theoretical lattice model studies, and apply it to real proteins. In particular the method is applicable to β-sheet surfaces. The β-sheet surfaces of two real proteins are thus redesigned and made experimentally. The culmination is a protein of greater stability than the naturally occurring protein. This is the first time greater stability has been achieved solely through mutations to the β-sheet surface, and marks a major step towards an ability to completely design de nova arbitrary proteins of arbitrary size. Successful protein design will lead to many practical applications, from new catalysts for industrial processes, to improved stability for existing medicines, to completely novel enzymes.