Computational design of metalloproteases

Area of Science:

• Biochemistry
• Enzyme Engineering
• Computational Biology

Background:

• Designing enzymes to cleave stable amide bonds is challenging due to substrate stability, unactivated leaving groups, and substrate flexibility.
• Previous de novo enzyme design efforts focused on more easily hydrolyzed ester bonds.

Purpose of the Study:

• To report the de novo design of zinc proteases capable of cleaving amide bonds.
• To utilize a fine-tuned AI model, RoseTTAFold Diffusion 2 for Molecular Interfaces, for enzyme and protein-protein interaction design.

Main Methods:

• Employing RoseTTAFold Diffusion 2 for Molecular Interfaces for de novo design of zinc proteases.
• Testing 135 designed enzymes in a single round to assess activity and cleavage specificity.

Main Results:

• 36% of the designed enzymes exhibited activity, cleaving amide bonds at the intended site.
• The most effective designed enzyme demonstrated a catalytic enhancement exceeding 10^8-fold for peptide bond hydrolysis.

Conclusions:

• De novo enzyme design has advanced to successfully target challenging amide bond hydrolysis.
• These engineered metallohydrolases hold potential for therapeutic and bioremediation applications.