Computational Design of Protein Structures and Interfaces

Brian Kuhlman
University of North Carolina

Designing novel protein structures and interactions is challenging because it requires the identification of low energy conformations and sequences. Our approach for solving this problem has been to iterate between sequence design and structure refinement using the molecular modeling program Rosetta. This talk will focus on three recent design efforts that make use of this strategy: the redesign of a protein core to have zero sequence identity with the starting protein, the design of a novel binding conformation for a protein-binding peptide, and the de novo design of symmetric homodimers. In all three cases we have used biophysical characterization and x-ray crystallography to validate the designs. In most cases, there is close agreement between the design models and the experimental structures. A comparison of failed and successful designs shows that while the design and refinement of hydrophobic-mediated interactions are often successful, the design of novel hydrogen bond networks is still very challenging and often fails.

Presentation (PowerPoint File)

Back to Workshop I: Design of Drugs and Chemicals that Influence Biology