Simulation Assisted Improvement of Plastic Degradation Enzyme PETase based Machine Learning Tools

Polyethylene terephthalate (PET) plastic is one of the most widely used plastic primarily due to its flexibility, endurance, and low cost. However, the plastics one-time use nature and long degradation time have led to massive waste accumulation, damaging our ecosystem, health, and biodiversity. While previous degradation methods are ineffective due to their high cost and low efficiency, the discovery of two enzymes PETase and MHETase in the bacteria Ideonella sakaiensis to degrade PET and mono(2-hydroxyethyl), a reaction intermediate in PET degradation, respectively, sparked the idea of a sustainable approach to degradation. Ever since, many approaches, including directed evolution, rational protein engineering, and computational redesign strategies, have optimized PETase in terms of its thermostability, catalytic activity, and more. This study proposes the incorporation of newly developed machine learning-based computational tools, including MutCompute, AlphaFold, and DiffDock, into a holistic protein engineering process to predict optimal PETase mutations. Here, in-silico experiments using machine learning tools as well as molecular dynamics simulation and interactions analysis screened for large amounts of PETase mutants in a time and cost-saving manner. Degradation assay coupled with mass analysis and high-performance liquid chromatography techniques then experimentally characterized PETase and its chosen mutants; thus, further screening found the most viable PETase mutant. Using various strategies, the project directly tackles one of the major global issues sustainability by bio-recycling PET. The research also aims to pave the way for introducing a new, imitable process for the more effective and resource-efficient engineering of all proteins.