As computationally designed enzymes usually possess low initial catalytic efficiency, directed evolution is used for optimization, which is done in iterative cycles of gene diversification by mutagenesis and screening for improved catalysts. Laboratory evolution thus relies on high-throughput screening techniques that must be applicable for enzymes with low starting activity. A first work package of this project was hence to establish that microfluidic-based fluorescence-activated droplet sorting (FADS), a ultrahigh-throughput method used to screen for (retro-)aldolase activity, is suitable for low initial catalytic activity. We could successfully demonstrate this by evolving a previously designed de novo (retro-)aldolase (RA95), which is based on a natural protein scaffold. The results showed that alternative evolutionary trajectories leading to enzymes with opposite enantioselectivity can be explored efficiently when starting straight from the computational design. This part of the project has been published (Ref. 1) and provides the basis for the evolution of aldolases implemented in de novo protein scaffolds, which is in progress.
Another important aspect of the project was to gain a mechanistic understanding of changes that occur during the directed evolution of artificial aldolases. Here, the evolutionary history of a previously optimized (retro-)aldolase (RA95) was analyzed using single turnover enzyme kinetics, isotope exchange and kinetic isotope effect measurements in combination with structural analysis. The results have been published (Ref. 2) and provide important feedback for enzyme design. Specifically, a shift in the rate-determining step was identified and the importance of a catalytic tetrad in the enzyme’s active site that emerged during evolution was emphasized. As a consequence, this catalytic motif will be introduced in a new generation of de novo aldolases.
The first generation of (retro-)aldolase designs in the actual de novo scaffolds showed severe problems with solubility and stability and were generally not sufficiently active to start the optimization by directed evolution. A second generation of designs with modifications in both the active sites and the scaffolds themselves shows promising initial results and is currently being optimized.
In addition to the aldolase approach, de novo protein scaffolds were equipped with binding sites for metal cofactors, which provide a broad range of non-natural activities. A crystal structure of a metal-binding de novo protein has been determined and provides the basis for our future work. This part of the project is still at a preliminary stage, but will be continued.
In summary, the overall goal of this study, generating an efficient enzyme from a naïve de novo protein scaffold, has not yet been fully achieved, but work in progress. The results obtained in the period of the fellowship represent important milestones and were disseminated in two peer-reviewed publications and contributions two several international conferences.
[1] *Obexer, R., *Pott, M., *Zeymer, C., Griffiths, A.D. and Hilvert, D. (2016) „Efficient Laboratory Evolution of Computationally Designed Enzymes with Low Starting Activities Using Fluorescence-Activated Droplet Sorting”
Protein Engineering, Design & Selection 29, 355-366
[2] Zeymer, C., Zschoche, R., and Hilvert, D. (2017) „Optimization of Enzyme Mechanism along the Evolutionary Trajectory of a Computationally Designed (Retro-)Aldolase”
Journal of the American Chemical Society 139, 12541-12549
