Artificial Intelligence-Driven de Novo Design of Robust Enzymes to Enhance Their Performance

The booming artificial intelligence (AI) technology provides an opportunity to precisely carry out de novo design of enzymes and create new biocatalysts with significantly enhanced performance. In the past decade, successful de novo enzyme design cases, although they yielded modest improvements that fell short of targets, have shown that this ambitious goal is achievable, especially as AI now enables high-accuracy, from-scratch prediction of enzyme structures. Analyzing the structural features of current de novo enzymes highlights the need for greater design precision to create tailored, high-performance biocatalysts on demand. Herein, the main achievements, latest research progress, and numerous emerging innovation opportunities of de novo enzyme design in the context of AI are summarized and discussed. Building on previous in-depth research on the catalytic mechanisms of enzymes, de novo enzyme design is achieved by modeling the most critical transition state in the catalytic reaction. Currently, the dominant approach is an inside-out design strategy. AI-driven de novo enzyme design methods have great potential to enhance model accuracy and now emerge as promising approaches. It is hopeful to overcome the bottleneck of tailoring industrial enzymes, obtain robust and efficient biocatalysts, and thus meet greener and more economical development.