FAST-PETase: The AI-Designed Enzyme That Breaks Down PET Plastic

Facebook ↗WhatsApp ↗
FAST-PETase Broke Down PET Samples in Laboratory Recycling Research

In April 2022, chemical engineers and molecular biologists at the University of Texas at Austin reported an enzyme they called FAST-PETase. In laboratory tests it took apart polyethylene terephthalate (PET), the clear plastic of drinks bottles, fruit punnets and polyester fabric, in a matter of hours to days rather than the decades or centuries the same material can persist in the environment. The work appeared in the journal Nature, and the short answer to the question most readers ask is this: yes, the enzyme genuinely breaks PET down, but only PET, and so far under controlled conditions rather than in a landfill or the sea.

That distinction matters, because the headlines that followed often spoke of an enzyme that "eats plastic" in general. Here is what the team actually built, how artificial intelligence helped, and what still has to happen before it changes what goes in your recycling bin.

What the Texas team created

The starting point was PETase, a natural enzyme first described in 2016 by Japanese researchers who found the bacterium Ideonella sakaiensis living on plastic debris at a bottle-recycling site. That bacterium uses PETase to snip the long PET chains into smaller pieces it can feed on. The natural version, however, is slow and loses its shape at modest temperatures, which limits any industrial use.

The Austin group, led by chemical engineer Hal Alper with first author Hongyuan Lu and colleagues including Andrew Ellington and Nathaniel Lynd, set out to make a sturdier, quicker variant. Their finished enzyme carries five changes to the natural protein, and the name is an acronym for "functional, active, stable and tolerant". The paper was published on 27 April 2022.

The university's announcement described PET as making up about 12 percent of global waste. That figure comes from the press release rather than from a waste audit cited in the paper, so it is best treated as a rough indication of scale, not a measured share of every country's rubbish.

How machine learning picked the mutations

Choosing which amino acids to change in a protein is a needle-in-a-haystack problem: there are far too many combinations to test one by one. The researchers used a machine-learning model called MutCompute, trained on many thousands of known protein structures, to predict positions where the natural enzyme's structure looked "unusual" for its surroundings and where a substitution might make it more stable.

The model's suggestions were combined with mutations from earlier engineering work, and the team tested the resulting variants in the lab. Ellington, whose group helped develop the model, framed the project as a meeting of fields, saying it brought together "synthetic biology to chemical engineering to artificial intelligence." The point is not that a computer invented an enzyme alone, but that prediction narrowed the search so that lab experiments could focus on the most promising candidates.

What the laboratory tests showed

According to the university, the team tried FAST-PETase on 51 different post-consumer PET items, including food containers and thermoformed packaging, as well as polyester fibres, fabric and whole drinks bottles. Some samples were largely broken down within 24 hours, and the enzyme worked at temperatures below 50 °C. That is cool by industrial standards, and it is one reason the group stressed lower energy use.

Crucially, the reaction is depolymerisation. The enzyme cuts PET back into its two building blocks, terephthalic acid and ethylene glycol, instead of grinding a bottle into ever-smaller fragments. Early viral posts claimed the enzyme simply breaks plastic "into smaller and smaller pieces", which would describe microplastics, the very problem recycling tries to avoid. The researchers also showed they could rebuild new PET from the recovered monomers, closing the loop in the lab.

Alper argued that this kind of enzymatic process could give companies far beyond the waste industry the chance to take a lead in recycling their own products, and he pointed to working at ambient temperatures as an advantage for possible clean-up uses in future.

Why PET is hard to recycle today

PET is one of the most widely recycled plastics, yet most of it still escapes the loop. Mechanical recycling shreds, washes and remelts bottles, but each round degrades the polymer a little, and coloured, dirty or mixed items are difficult to handle. Food trays, multilayer packaging and polyester clothing often end up landfilled or burned. Chemical recycling can recover pure monomers, but conventional routes tend to need high temperatures or harsh solvents.

Enzymes offer a gentler chemical route. FAST-PETase was not the first: the French company Carbios has worked on an engineered leaf-branch compost cutinase that operates at higher temperatures, described in Nature in 2020, and other groups have published their own variants. The Texas result added a fast enzyme working in a milder temperature range to a growing toolkit.

Common misunderstandings

"It eats all plastic"

No. PETase-type enzymes act on PET. Polyethylene bags, polypropylene lids, polystyrene foam and PVC have different chemistry and are untouched. Separate research looks at insects and microbes that may attack polyethylene, but that is a different story.

"It can be sprayed on landfills or the ocean"

Not as things stand. The tests were done in controlled reactors with prepared samples. Releasing engineered enzymes into the open environment raises practical and regulatory questions that the paper did not address, and Alper's comments about environmental clean-up were framed as a future possibility.

"Plastic waste is solved"

Even a perfect enzyme depends on collection, sorting and cleaning. If PET never reaches a facility, no enzyme can recycle it. Cutting single-use packaging and improving collection remain the most dependable steps.

From test tube to recycling plant

When the paper came out, the university said the team planned to work on scaling up production of the enzyme and on applying it at industrial and environmental sites. Scale-up is where many promising processes stall: enzymes must be produced cheaply, survive long runs, and cope with labels, glue, dyes and food residue. The economics also have to compete with virgin PET made from oil and gas, which is often cheap.

For readers, the practical takeaway is simple. Keep putting clean PET bottles and trays in the right recycling stream, choose refillable options where you can, and treat enzyme recycling as a promising technology under development rather than a reason to stop worrying about plastic. The FAST-PETase study is a solid piece of science; its real-world value will depend on the years of engineering that follow.

Sources: University of Texas at Austin News: Plastic-eating enzyme could eliminate billions of tons of landfill waste (27 April 2022); Lu et al., Machine learning-aided engineering of hydrolases for PET depolymerization, Nature (2022); Wikipedia: PETase; Wikipedia: Carbios; video via YouTube.

COMMENTS