Researchers Use AI To Create PET-Eating Enzymes
by Rudy Sanchez on 05/09/2022 | 2 Minute Read
Methods using wild and engineered enzymes quickly degrade plastics such as polyethylene terephthalate (PET) in a process called depolymerization, which breaks the polymers into monomers. These monomers can then get pieced together into virgin PET, unlike recycling which turns collected PET into rPET, with a finite number of cycles before the resulting material is unsuitable for use.
As promising as it sounds, current methods require precise conditions within narrow temperatures and pH levels. Additionally, the plastic material often requires pretreatment. Scientists at the University of Texas, Austin, have developed an enzyme using artificial intelligence (AI) that breaks down 51 types of PET in a broader range of temperature and acidity levels in addition to untreated plastic.
The new enzyme, which researchers call FAST-PETase, is an acronym for “functional, active, stable, and tolerant PETase.” Their work was published in the journal Nature, which details how scientists used AI to identify optimal mutations, finding a balance between robustness and activity. The authors posited that highly engineered enzymes lack the optimization that comes from evolution. Using a neutral, structure-based, deep learning neural network, the team identified four mutations that they tested in 29 different combinations, eventually landing on FAST-PETase as the best performing, using a series of tests at different temperatures and pH.
Researchers see FAST-PETase as a commercially viable closed-looped solution to PET thanks to its high activity rate at ambient temperatures and its ability to break down even tinted PET. While the enzyme can only break down one type of plastic, the researchers could potentially develop other plastic-eating enzymes. That could be a big win for all of the single-use polymer-packed products we have currently sitting in landfills and all of the PET plastics we produce ad nauseum.