Replicating PET Hydrolytic Activity by Positioning Active Sites with Smaller Synthetic Protein Scaffolds

Area of Science:

• Biotechnology
• Enzyme Engineering
• Computational Biology

Background:

• Evolutionary constraints limit natural enzyme diversity, hindering enzyme discovery and engineering.
• Artificial intelligence advances in protein design enable de novo enzyme creation beyond traditional methods.

Purpose of the Study:

• To computationally design novel polyethylene terephthalate hydrolases (PET hydrolases) with enhanced properties.
• To leverage AI and molecular computations to create enzymes not found in nature.

Main Methods:

• Utilized a computational strategy involving functional motif extraction from a template enzyme (leaf-branch compost cutinase, LCC).
• Employed deep learning algorithms and molecular computations for sequence regeneration, screening, and refinement.
• Experimental validation of designed enzymes for PET hydrolytic activity.

Main Results:

• Successfully created designer PET hydrolases with significant PET hydrolytic activity.
• Designer enzymes were at least 30% shorter than the LCC template.
• RsPETase1 demonstrated robust expressibility, comparable catalytic efficiency (kcat/Km) to LCC, and thermostability (Tm = 56°C) with only 34% sequence similarity.

Conclusions:

• Enzyme diversity can be expanded by computationally designing protein scaffolds around functional motifs.
• This approach generates opportunities to acquire highly active and robust enzymes absent in nature.
• The study demonstrates a powerful strategy for de novo enzyme engineering.