A broadly applicable artificial selection system for biomolecule evolution

Abstract

Biocatalysis offers an attractive alternative to traditional chemical catalysis. However, it is often found that an enzyme with the optimal properties for a specific application is not available within the natural repertoire of enzymes. It is then desirable to obtain an improved variant by altering the sequence of a known enzyme, in a process known as protein engineering. Directed evolution is one of the most powerful tools for protein engineering. In directed evolution, the process of natural evolution is mimicked in the laboratory at a much shorter timescale and selecting for properties that make the enzyme (or any other type of biomolecule) more suitable for an application of human interest. The main bottleneck of directed evolution is the identification of the desired variants amongst a majority of variants without the sought altered or improved property. Selection approaches link the desired activity to an increased survival rate or improved growth. While in principle such methodologies allow for ultra high-throughput analysis of libraries, most selection techniques have a limited scope, and can only be applied to a relatively reduced set of biomolecules or properties. This thesis presents the most broadly-applicable artificial selection system for the evolution of biomolecules ever reported. The selection platform is based on an engineered E. coli strain with impaired regeneration of NAD+, causing a conditional growth defect during anaerobic fermentation. By directly or indirectly linking the activity of the biomolecules of interest to the oxidation of NADH, cells can be rescued from this growth defect. The efficacy of such selection system has been demonstrated by using it to select alcohol dehydrogenase, imine reductase and nitroreductase variants with altered or enhanced catalytic properties, as well as an isopropanol-producing metabolic pathway with optimised regulatory elements leading to a maximised yield of isopropanol. These results confirm the wide scope of the developed selection system, which can replace conventional screening currently used in many cases of direct relevance for industrial processes. Increasing the throughput of the variant search process by many orders of magnitude will lead to the discovery of novel biomolecules and accelerate the implementation of biocatalysis.