Integrating rational design and machine learning to engineer a novel plastic depolymerase to break down PBAT-based...

Biodegradable polybutylene adipate terephthalate (PBAT) represents a promising alternative to conventional polyolefins and has gained widespread adoption in packaging, agricultural mulch films, and diverse industrial applications. Despite its biodegradable classification, PBAT's intrinsically low biodegradation efficiency in natural environments caused persistent ecological contamination. A critical challenge in addressing this problem remains the development of efficient PBAT-degrading enzymes with broad thermal adaptability. In this study, we identified a new PET hydrolase (ThPETase) from Thermobifida halotolerans via consensus sequence analysis and subsequently engineered it through structure- and surface charge-guided design, integrating the machine learning model EvolvePro. The optimized variant, DCTPC_4M_S122H (ThPETase_N48D/D204C/L213T/S214P/D253C/S68K/D105R /G194Q/D232K/S122H) showed a remarkable 33.9°C increase in melting temperature (T_m) compared to the wild-type enzyme, and demonstrated an enhanced hydrolytic activity against PBAT powders across a broad temperature range (37-70°C), achieving up to a 91-fold improvement in PBAT degradation efficiency at 37°C. Notably, it efficiently degraded PBAT-based mulch films within 20 h at 37°C, 8 h at 50°C, and merely 6 h at 60°C, significantly outperforming previously reported enzymes. The enzymatic degradation yielded high-purity terephthalic acid (≥99 %), which can be directly reclaimed as a feedstock for PET and PBAT synthesis. This enables a closed-loop recycling system for sustainable plastic waste management.