Accelerated enzyme engineering by machine-learning guided cell-free expression.

Area of Science:

• Biochemistry
• Synthetic Biology
• Machine Learning

Background:

• Enzyme engineering faces challenges in generating large sequence-function datasets for predictive design.
• Optimizing enzymes for specific chemical reactions requires efficient exploration of protein sequence space.

Purpose of the Study:

• To develop a machine learning (ML)-guided, cell-free platform for rapid enzyme engineering.
• To engineer amide synthetases for improved activity in synthesizing small molecule pharmaceuticals.

Main Methods:

• Integrated cell-free DNA assembly, gene expression, and functional assays.
• Evaluated 1217 enzyme variants across 10,953 reactions to map fitness landscapes.
• Developed augmented ridge regression ML models for predictive design.

Main Results:

• Engineered amide synthetases showed 1.6- to 42-fold improved activity for 9 target small molecule pharmaceuticals.
• The ML models accurately predicted enzyme variants with enhanced catalytic function.
• Demonstrated parallel optimization of enzymes for distinct chemical reactions.

Conclusions:

• The ML-guided, cell-free framework accelerates enzyme engineering.
• Enables iterative exploration of protein sequence space for specialized biocatalyst development.
• Promises to enhance the creation of custom enzymes for diverse applications.