Deep learning-aided redesign of a hydrolase for near 100% PET depolymerization under industrially relevant conditions

Abstract Biotechnological plastic depolymerization and recycling have emerged as suitable options for addressing the plastic-waste pollution crisis in a circular plastic economy. Enzymatic degradation of poly(ethylene terephthalate) (PET), as the most typical representative, has evolved over the past two decades, with a major breakthrough achieved by using the LCC variant that permitted 90% conversion of PET on an industrial scale. Despite the achievements, the last 10% residual PET becomes nonbiodegradable due to physical aging, which has hampered its application in real industrial scenarios. In the present study, we addressed current challenges by employing a computational strategy that incorporates a protein language model and force-field-based algorithms to engineer a hydrolase from the bacterium HR29. The redesigned variant, TurboPETase, outperformed all the PET hydrolases reported thus far with regard to industrial application, enabling nearly 100% depolymerization of untreated PET containers, pretreated postconsumer PET bottles and their lower-grade products. The full degradation of pretreated PET at high industrially relevant scales (up to 300 g L-1) can be accomplished in as little as 10 h, with a maximum production rate of 77.3 gTPAeq L 1 h-1, demonstrating great potential for enzymatic PET recycling. Kinetic parameters derived from the inverse Michaelis‒Menten model and structural analysis suggest that the improved depolymerization performance may be attributed to a more flexible PET-binding groove that facilitates the targeting of more specific attack sites. Collectively, our results constitute a significant advance in the understanding and engineering of effective industrially applicable polyester hydrolases and provide guidance for further efforts on other mass-produced polymer types in this intriguing research field.