Decoding Plastic Waste: Novel AI-Guided Engineering of PETase Mutants for Enhanced PET Plastic Bioremediation

Overview

Polyethylene terephthalate (PET) represents one of the most prominent forms of synthetic plastic pollution, accounting for over 50 million tons of annual waste worldwide. Although the discovery of natural PET-degrading enzymes like PETase has provided a foundational base for biological recycling through bioremediation, the native enzyme exhibits limitations including low thermal tolerance, poor activity against high-crystallinity PET, and narrow pH ranges. To overcome these challenges, I developed PETase-M14, a computationally engineered PET hydrolase derived from Ideonella sakaiensis PETase, enhanced via an AI-guided design framework. My engineering strategy utilized a transformer-based neural network (StructuRiNet) to predict mutation fitness, integrated with a probabilistic regression model (BayesRegression) that estimates PET hydrolytic efficiency using structural, thermodynamic, and biochemical descriptors. Through iterative silico selection, I introduced 14 synergistic mutations designed to enhance thermal stability, folding energetics, and PET substrate binding. Experimental characterization of PETase-M14 revealed a substantial increase in degradation activity over wild-type PETase on untreated amorphous PET films, alongside a significant improvement in thermal stability (+19–22°C Tm). The enzyme retained >90% of peak activity across a broad pH range (6.0–8.5) and temperature window (30–55°C). PETase-M14 also efficiently degrades postconsumer plastics, achieving up to 92.8% mass reduction within 120 hours. This work demonstrates the power of AI-assisted enzyme engineering to accelerate bioremediation optimization and provides a practical route toward enzymatic PET recycling under environmentally realistic conditions.

Competition history

• ISEF 2026 Biochemistry · Entry BCHM005

Resources

• ISEF abstract page ↗

Related projects

Closest projects by meaning, across every fair and year in the corpus.