Braun, M.; Tripp, A.; Chakatok, M.; Kaltenbrunner, S.; Totaro, M. G.; Stoll, D.; Bijelic, A.; Elaily, W.; Hoch, S. Y. Y.; Aleotti, M.; Hall, M.; Oberdorfer, G.

Custom designed enzymes can further enhance the use of biocatalysts in industrial biotransformations, thereby helping to tackle biotechnological challenges of the 21st century. We present rotamer inverted fragment finder - diffusion (Riff-Diff) a hybrid machine learning and atomistic modeling strategy for scaffolding catalytic arrays in de novo protein backbones with custom substrate pockets. We used Riff-Diff to scaffold a catalytic tetrad capable of efficiently catalyzing the retro-aldol reaction. Functional designs exhibit a high fold diversity, with substrate pockets similar to natural enzymes. Some of the thus generated de novo enzymes show activities rivaling those optimized by in-vitro evolution. The design strategy can, in principle, be applied to any catalytically competent amino acid constellation. These findings are paving the way to address factors for the practical applicability of de novo protein catalysts in industrial processes and shed light on fundamental principles of enzyme catalysis.