Computational design of metallohydrolases.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology and Bioinformatics
• Synthetic Biology

Background:

• De novo enzyme design aims to create proteins with optimized active sites for specific chemical reactions.
• Existing generative AI methods like RFdiffusion require precise specification of catalytic residue positions and backbone coordinates, limiting design flexibility.
• There is a need for more efficient and flexible computational tools for designing novel enzymes with high catalytic activity.

Purpose of the Study:

• To introduce RFdiffusion2, an improved generative AI method for de novo enzyme design.
• To design novel zinc metallohydrolases using RFdiffusion2, starting from quantum chemistry-derived active site geometries.
• To demonstrate the capability of RFdiffusion2 to generate highly active and accurate enzyme designs.

Main Methods:

• Utilized RFdiffusion2, a generative AI tool that removes the need for specifying catalytic residue positions and backbone coordinates.
• Designed zinc metallohydrolases by integrating quantum chemistry-derived active site geometries with the RFdiffusion2 algorithm.
• Experimentally tested 192 de novo enzyme designs generated through two rounds of RFdiffusion2.

Main Results:

• The most active enzyme designed exhibited a catalytic efficiency (kcat/KM) of 16,000 M^-1s^-1, significantly exceeding previously designed metallohydrolases.
• A second round of design yielded three additional highly active enzymes, with kcat/KM up to 53,000 M^-1s^-1 and kcat up to 1.5 s^-1.
• The crystal structure of the most active design closely matched its computational model, validating the accuracy of the RFdiffusion2 method.

Conclusions:

• RFdiffusion2 enables the direct computational generation of highly active enzymes without experimental optimization.
• The designed enzymes feature preorganized active sites that facilitate substrate binding and catalysis.
• This advancement paves the way for a new generation of potent, computer-designed biocatalysts.