The strategy of rational design to engineer enzymes is to predict the potential mutants based on the understanding of the relationships between protein structure and function, and subsequently introduce the mutations using the site-directed mutagenesis. Rational design methods are universal, relatively fast and have the potential to be developed into algorithms that can quantitatively predict the performance of the designed sequences. Compared to the protein stability, it was more challenging to design an enzyme with improved activity or selectivity, due to the complexity of enzyme molecular structure and inadequate understanding of the relationships between enzyme structures and functions. However, with the development of computational force, advanced algorithm and a deeper understanding of enzyme catalytic mechanisms, rational design could significantly simplify the process of engineering enzyme functions and the number of studies applying rational design strategy has been increasing. Here, we reviewed the recent advances of applying the rational design strategy to engineer enzyme functions including activity and enantioselectivity. Five strategies including multiple sequence alignment, strategy based on steric hindrance, strategy based on remodeling interaction network, strategy based on dynamics modification and computational protein design are discussed and the successful cases using these strategies are introduced.

Flexible sites are potential targets for engineering the stability of enzymes. Nevertheless, the success rate of the rigidifying flexible sites (RFS) strategy is still low due to a limited understanding of how to determine the best mutation candidates. In this study, two parallel strategies were applied to identify mutation candidates within the flexible loops of Escherichia coli transketolase (TK). The first was a “back to consensus mutations” approach, and the second was computational design based on ΔΔG calculations in Rosetta. Forty-nine single variants were generated and characterised experimentally. From these, three single-variants I189H, A282P, D143K were found to be more thermostable than wild-type TK. The combination of A282P with H192P, a variant constructed previously, resulted in the best all-round variant with a 3-fold improved half-life at 60 °C, 5-fold increased specific activity at 65 °C, 1.3-fold improved k_cat and a T_m increased by 5 °C above that of wild type. Based on a statistical analysis of the stability changes for all variants, the qualitative prediction accuracy of the Rosetta program reached 65.3%. Both of the two strategies investigated were useful in guiding mutation candidates to flexible loops, and had the potential to be used for other enzymes.

Significance: Rigidifying flexible sites is a powerful method to improve enzyme stability. However, if the highly flexible regions form the active site, modifying them risks losing activity due to the activity–stability trade-off. We hypothesized here that regions outside the active site whose dynamics were highly correlated to flexible active sites, would provide good targets for stabilizing mutations. To test this hypothesis, six variants were constructed in the 3M variant of Escherichia coli transketolase. The best variant had a 10.8-fold improved half-life at 55 °C, and increased the T _m and T _agg by 3 °C and 4.3 °C, respectively. The variants even increased the activity, by up to threefold. This study highlights how protein engineering strategies could be potentially improved by considering long-range dynamics.

Significance: Incomplete understanding of the mechanisms of epistasis between two or more substitutions in a protein can hamper protein-engineering strategies. With Escherichia coli transketolase as a model, we explore the epistatic interactions between a set of stabilizing mutations from across two different domains within the protein structure. Surprisingly, not all pairwise effects between distant mutations from the surface and core regions of different domains were additive. Additionally, the epistatic behaviors observed were dependent on the type of stability measured. We found single mutations that altered local protein dynamics, which then induced correlated changes in the dynamics of a second domain of the same monomer. This mechanism mediated epistasis between distant mutations.

Traditional protein engineering methods, such as directed evolution, while effective, are often slow and labor-intensive. Advances in machine learning and automated biofoundry present new opportunities for optimizing these processes. This study devises a protein language model-enabled automatic evolution platform, a closed-loop system for automated protein engineering within the Design-Build-Test-Learn cycle. The protein language model ESM-2 makes zero-shot prediction of 96 variants to initiate the cycle. The biofoundry constructs and evaluates these variants, and feeds the results back to a multi-layer perceptron to train a fitness predictor, which then makes prediction of second round of 96 variants with improved fitness. With the tRNA synthetase as a model enzyme, four-rounds of evolution carried out within 10 days lead to mutants with enzyme activity improved by up to 2.4-fold. Our system significantly enhances the speed and accuracy of protein evolution, driving faster advancements in protein engineering for industrial applications.