Probabilistic computational protein design: Advances in methodology and the incorporation of non-biological molecular components

Computational protein design method has been developed to understand the underling physicochemical principles that determine the uniqueness of protein conformation often endowing biological function. Many designed sequences have been successful in revealing the energetic ingredients that govern the global and local details of folding; such successes are due in part to the advances in searching methods and scoring functions. Given the intrinsic complexity of proteins, myriads of subtle interactions govern the folding and approximations must be used for efficient computation. Sensitivity to the scoring energy function remains to be examined for discrimination of native-like ensembles of proteins. In this thesis, an entropy-based protein design strategy providing deterministic site-specific amino acid probabilities has been explored upon the variation of molecular mechanical potential energy functions. Secondly, the flexibility and extensibility of the method have been demonstrated in engineering de novo function in the human H-ferritin for noble metal cluster growth. The method has been used to de novo design of a tetra-alpha-helical bundle for anesthetic halothane binding, where the calculation is expedited by a symmetry approximation and expanded to accommodate non-natural amino acids and/or guest-molecules, thereby simultaneously determining the amino acid sequences and the bound ligand conformations. Finally, the probabilistic strategy has been applied to understand physicochemical properties in natural systems and to facilitate experimental studies. The folding of a Zn-binding domain has been explored by computationally designed substitution with the natural fluorescent probe Trp mutation. The probability profiles in the allosteric enzyme Adelnylate kinase has been used to design sequences that may have preferential stabilities for the open and the closed conformations of the enzyme in an attempt to understand the role of allostery in enzyme catalysis. As the last example, methods for designing sequences with targeted pI values have been introduced and applied to the partial redesign of cytochrome P450 2C9.