Accelerating the discovery of industrial PET-degrading enzymes: Evolving paradigms from bioprospecting to computational discovery.

Polyethylene terephthalate (PET) waste remains a major environmental and resource challenge, and enzymatic depolymerization offers a promising route for recovering terephthalic acid and ethylene glycol under aqueous conditions for closed-loop PET recycling. However, the practical deployment of PET-degrading enzymes is constrained by substrate crystallinity and morphology, limited thermal robustness near the glass transition temperature, biocatalyst manufacturability and reusability, product inhibition and pH-mediated activity loss, and the heterogeneity and contamination of post-consumer feedstocks. This review summarizes the evolution of PET hydrolase discovery from phenotype-driven bioprospecting to sequence-based mining, structure-guided searches, and AI-assisted prioritization. These approaches are compared in terms of throughput, novelty, false-positive risk, substrate relevance, and their capacity to identify enzymes with industrial fitness. We further discuss the importance of hierarchical validation, in which rapid screening is followed by activity measurements on PET substrates with defined crystallinity and process relevance. Finally, standardized substrate characterization, kinetic reporting, and open data sharing are highlighted as prerequisites for reliable comparison across studies and the development of AI-ready datasets. Such integration should help distinguish enzymes that are merely active under simplified screening conditions from candidates with measurable, comparable, and application-relevant potential for plastic circularity.