Accelerated Enzyme Design: A Smart Combination Approach Leveraging AlphaFold2 for Ultra-Low Throughput Screening

TrRosetta has proven its efficacy as a reliable starting point for the design of stable and functionally diverse enzymes. Similarly, AlphaFold2 has emerged as a dependable tool for enzyme design. Building upon these advancements, we introduce a novel approach, aptly named CSFp, which represents a strategic combination of SMD, FuncLib, and pSUFER, aiming to enhance enzyme activity through computationally focused ultra-low throughput screening. This approach leverages a structure predictor, AlphaFold2, and eliminates the requirement for experimental structural data. The CSFp pipeline comprises a multi-step process, including: (1) structure prediction using AlphaFold2-based Pointsite for pocket prediction and molecular docking, (2) identification of key residues within (un)binding tunnels using SMD, (3) generation of small yet intelligent libraries of mutants by FuncLib, (4) filtering of results based on predicted relative reactivity with MD simulations, (5) tagging of suboptimal residues during protein folding by pSUFER, and (6) experimental screening of combinatorial mutations to prevent superior mutants from escaping the local optimal space. The efficacy of this strategy was evaluated using a challenging case of biocatalytic relevance: the design of D-lactonohydrolase (D-Lac) with enhanced catalytic activity and enantioselectivity. Our results demonstrate the potential of CSFp as a powerful tool for accelerating enzyme design, facilitating the development of novel biocatalysts with desirable properties.