Computational enzyme design by catalytic motif scaffolding

Area of Science:

• Biocatalysis
• Protein Engineering
• Computational Chemistry

Background:

• Enzymes are valuable biocatalysts due to their selectivity and efficiency.
• Current de novo enzyme design methods are inefficient, yielding low-activity enzymes requiring extensive optimization.
• Industrial applications of custom enzymes are limited by the cost and complexity of current design processes.

Purpose of the Study:

• To develop a novel computational strategy for designing de novo protein catalysts.
• To demonstrate the general applicability of the new method for diverse chemical transformations.
• To create custom enzymes with high catalytic activity and stereoselectivity.

Main Methods:

• Developed rotamer inverted fragment finder-diffusion (Riff-Diff), a hybrid machine learning and atomistic modeling approach.
• Applied Riff-Diff to design enzymes for the retro-aldol and Morita-Baylis-Hillman reactions.
• Utilized high-resolution structural analysis to validate active site precision.

Main Results:

• Designed novel enzymes for two distinct chemical reactions with high catalytic activity and stereoselectivity.
• Achieved catalytic performance comparable to enzymes optimized through in vitro evolution.
• Structural analysis confirmed near-atomic precision in active site design for multiple enzyme constructs.

Conclusions:

• Riff-Diff is a powerful strategy for designing de novo protein catalysts with tailored functions.
• The method significantly advances the potential for practical applications of custom enzymes in synthesis.
• This work provides fundamental insights into protein design principles and enzyme catalysis.