Full Sequence Design of an Alpha-Helical Protein and Investigation of the Importance of Helix Dipole and Capping Effects in Helical Protein Design

Our goal is an objective, quantitative design algorithm based on the physical chemical forces which determine protein structure and stability. To this end, we have developed a cyclical protein design strategy which utilizes theory, computation, and experimentation using a variety of protein systems. We address the inverse folding problem using a protein design algorithm which objectively predicts protein sequences which are compatible with a given fold. Our protein design methodology was developed using a variety of proteins, and therefore should be generalizable to many folds and motifs. To test the generalizability and expand the size of proteins we have designed, engrailed homeodomain (enh), a 51-residue helix-turn-helix motif, was used as a target motif. A series of design calculations and experiments on the thirty surface positions of enh were performed to probe the importance of the helix dipole and capping effects in protein design. Rules for which types of residues were allowed at the helix termini were introduced systematically, resulting in progressively more stable proteins. The first design in the series, which had no considerations for the helix dipole or capping effects, was shown to have the same thermal stability as wild-type enh and the protein with the most stringent rules has a Tm of 75 °C, 32° higher than wild-type and the first design. Therefore, helix dipole and capping effects have a large impact on our ability to design stable proteins. The ten core residues of enh were included in the design calculation. The resulting protein, a 29-fold mutant of wild-type, has a Tm of 81 °C. The full sequence design of enh was computed stepwise. The eleven boundary residues were designed in the context of the surface-core design. The resulting protein, a 39-fold mutant of wild-type enh, has a melting temperature of 114 °C and is 4.7 kcal/mol more stable than wild-type. The structure of the boundary-surface-core design was solved by NMR techniques and found to be in excellent agreement with the target structure. The top 10 structure have a backbone root-mean-square standard deviation of 0.45 Å and the root-mean-square standard deviation between the model structure and experimental backbones is 1.25 Å. The side chain selection algorithm was also extended to the design of peptides to bind tightly to MHC class I proteins. A circular dichroism spectrometry assay was developed to determine the peptide dissociation constants. Three designed peptides were bound more tightly to the MHC class I molecule H-2Kd than known peptides. In addition, an investigation of the removal of disulfide bonds from toxin folds is discussed.