Final Report Summary - ABDESIGN (Computational design of novel protein function in antibodies)
The design of large and complex functional proteins, such as antibodies and enzymes, is a major challenge in protein engineering due to the astronomically large sequence and conformation space of such proteins. As part of the AbDesign Starting Grant, the Fleishman lab established a new strategy, called evolution-guided atomistic design, for designing such proteins. This strategy uses information from structural-bioinformatics analyses to limit Rosetta atomistic design calculations to regions of the sequence-structure space that are likely to encode stable, well-folded and functional proteins, thus enabling the design even of large and complex proteins. Foremost, we demonstrated that this new strategy leads to the first design of atomically accurate antibody variable domains and enzymes. Some of these enzymes exhibited as many as 150 mutations from any natural protein, yet showed high stability and catalytic efficiency similar to those observed in enzymes from thermophilic organisms and orders of magnitude more than observed in previously designed enzymes. We furthermore developed a general method, called PROSS, for designing stable proteins, including therapeutic enzymes and a malaria vaccine immunogen, dramatically reducing costs of production and protein stability without impairing biological activity. Previous stability design methods relied on iterating structural analysis, genetic screening, and crystallography, which can now be circumvented by an essentially one-shot design strategy. Very recently, we extended this strategy and developed a method called FuncLib to design variants of enzymes with vastly improved catalytic efficiency, including therapeutically relevant hydrolases of highly poisonous nerve agents, such as Russian VX and soman. In the past, achieving highly efficient enzymes and antibodies was only possible by relying on laborious and intensive rounds of in vitro evolution. The methods developed through AbDesign are therefore a paradigm shift enabling efficient, one-shot optimisation of large and complex proteins through completely rational and automated methods. We therefore developed PROSS and FuncLib into simple web accessible servers that are free to use to all academics. The PROSS server, for instance, has been used by more than 500 labs from around the world since its launch two years ago. Since most drugs under development are based on proteins and additionally many industrial processes rely on enzymes and binding proteins, these new methods may have a broad impact, circumventing frustrating impasses to the application of proteins in research, biotechnology and medicine. Based on the methods developed through the AbDesign project, our lab is now developing new methods to design completely new antibodies that target disease causing agents.