We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Deep learning-guided discovery and engineering of binding peptides for accelerated enzymatic degradation of polyethylene terephthalate
Summary
Researchers used a deep learning model trained on genomic data from Ideonella sakaiensis to discover and engineer PET-binding peptides, then fused the optimized peptides with a PETase mutant to boost PET hydrolysis by up to 24.8-fold, with molecular dynamics simulations explaining how closer substrate docking drives the catalytic improvement.
Enzymatic polyethylene terephthalate (PET) degradation holds promise for environmental restoration. However, limited substrate catalytic capacity hinders its application in addressing PET plastic contamination. To enhance enzyme-substrate interaction, effective anchoring strategies are essential. This study presents a novel deep learning approach to identify and engineer high-performance PET-binding peptides from genomic data. Utilizing this approach, we discovered promising PET-binding peptides from the Ideonella sakaiensis genome and fused them with an optimized Ideonella sakaiensis PETase mutant to enhance PET hydrolysis. Remarkably, the Efficient Attention-Based Model for Computational Protein Design-optimized fusion proteins achieved a 2.0- to 24.8-fold increase in PET hydrolysis compared with the enzyme without an anchor. Importantly, we elucidated the mechanism by which the binding peptide domain enhances the catalytic activity of the enzyme against PET substrates, supported by comprehensive molecular dynamics simulations. This work establishes a robust deep learning framework for biocatalyst design and provides potent enzymatic solutions to address global PET plastic pollution.