Efficient laboratory evolution of computationally designed enzymes with low starting activities using fluorescence-activated droplet sorting

De novo biocatalysts with non-natural functionality are accessible by computational enzyme design. The catalytic activities obtained for the initial designs are usually low, but can be optimized significantly by directed evolution. Nevertheless, rate accelerations approaching the level of natural enzymes can only be achieved over many rounds of tedious and time-consuming laboratory evolution. In this work, we show that microfluidic-based screening using fluorescence-activated droplet sorting (FADS) is ideally suited for efficient optimization of designed enzymes with low starting activity, essentially straight out of the computer. We chose the designed retro-aldolase RA95.0, which had been previously evolved by conventional microtiter plate screening, as an example and reoptimized it using the microfluidic-based assay. Our results show that FADS is sufficiently sensitive to detect enzyme activities as low as k_cat/K_m = 0.5 M^-1s^-1. The ultra-high throughput of this system makes screening of large mutant libraries possible in which clusters of up to five residues are randomized simultaneously. Thus, combinations of beneficial mutations can be identified directly, leading to large jumps in catalytic activity of up to 80-fold within a single round of evolution. By exploring several evolutionary trajectories in parallel, we identify alternative active site arrangements that exhibit comparably enhanced efficiency but opposite enantioselectivity.