Abstract
Biocatalysis is an emerging and sustainable approach to depolymerize highly hydrophobic plastic polyesters such as poly-(lactic acid) (PLA), a bioplastic widely used in packaging and disposable items. Some enzymes, including lipases, cutinases, and proteases, have been described to hydrolyze PLA, but the activity strongly depends on stereochemistry and crystallinity. In this study, we explored the activity of the versatile lipase fromOphiostoma piceae (OPE) and three engineered variants (N81A, N94A, and N81/94A) on polylactic acid (PLA), comparing the experimental data with predictions from two computational methodologies, Thermal Titration Molecular Dynamics (a classical method) and the machine learning-guided XLPFE scoring function. Experimentally, mutant N81A showed the highest PLA hydrolytic activity, followed by WT OPE, with N94A and N81/94A being substantially less effective. This combined approach served to validate the reliability of these computational strategies for predicting enzyme interactions and highlights the importance of using long enough model substrates to guide future enzyme optimization.