Incorporation and Validation of Flexibility Modelling Tools for De Novo Enzyme Design

Abstract

Computational de novo enzyme design is a rapidly evolving field, involving the bottom-up design of an active enzyme for an important chemical reaction, starting from an inactive protein scaffold. Many methods and pipelines have been in development for decades, and it has been shown that designing enzymes from a multitude of input templates (ensemble design) rather than a single one (static design) usually leads to better results. Methods for generating ensembles are not perfect and are prone to error as well as biases depending on the specific method used. In this work, backrub’s (BR) efficacy as an alternative ensemble generation method for the development of enzymes is explored and utilized to recapitulate pre-existing enzymes with known catalytic efficiencies approaching that of wild-type enzymes, as well as backrub serialized with other methods. These methods are compared to pre-existing ones that do not utilize backrub, and advantages as well as disadvantages are discussed and explained. Further on, backrub ensembles are used to design a new retro-aldolase, TyRA95.0, whose design is then computationally characterized. To conclude, recommendations are given to the next generation of enzyme engineers given what was learned from Backrub’s recapitulations and TyRA95.0 design, and new metrics for improved ensemble analysis are discussed.