Introducing specificity into protein design

Computational protein design is an attractive field of study because it provides a rigorous and verifiable test of protein science, and because it can generate useful molecules for science and medicine. A number of approximations and assumptions have been made that render the computational design of proteins feasible. These simplifications are at times incompatible with the goal of designing functional proteins. One remarkable aspect of protein function is the exquisite specificity with which proteins identify their molecular partners, ligands, and substrates. This dissertation expands the methodology of computational protein design to incorporate molecular specificity. The first contribution relates to the scoring functions used to rank protein structures. These functions typically neglect protein electrostatics because evaluation by conventional algorithms is computationally prohibitive. We have developed a rapid and accurate continuum electrostatics model that is suitable for design calculations. The second limitation of current protein design methodology is the focus on stabilizing the desired target state. We have developed a framework in which both stability for a target state and specificity against competitor states is possible. For the first time, design calculations can explicitly optimize the specificity of a protein, rather than just the stability. A third limitation of protein design methodology addressed by this dissertation is the requirement for an experimentally determined backbone before a design may begin. The design of proteins with arbitrary functions may require backbones not present in experimental databases. We have developed a parametric model for describing the backbones of (beta/alpha)8 barrel proteins. This protein fold has proven to be a versatile scaffold for enzymatic function. The parametric model allows for the simple specification and optimization of backbone coordinates for members of this fold. The extensions to computational protein design methodology developed in this dissertation bring the field closer to ultimate goal of understanding and engineering protein function.