Plastic，It was hailed as one of the greatest inventions of the last century because of its lightness, durability and low cost. Today, however, the rampant "white pollution" has become a well-known environmental problem.

The problem has been exacerbated in recent decades by the dramatic rise in disposable plastic consumer goods and the prevalence of a "throwaway" culture. Plastic is found in every corner of the globe. They are "far away" -- as far as the uninhabited ice sheets of Antarctica, as deep as the deepest Mariana Trench on Earth, as high as Mount Everest, the roof of the world; Here we are again: microplastics in the environment have already seeped into our food, air and water.

There is growing evidence that plastics pose a major ecological challenge. Previous studies have shown that nearly 700 Marine species, more than 50 freshwater species and a wide range of land animals have swallowed plastic. Recently, scientists have also found traces of plastic in the meconium and blood of human babies. It is clear that plastic pollution has affected many aspects of human well-being.

Previously, a study published in the journal Science showed that，Over the next two decades, without extensive intervention, more than 1.3 billion tons of plastic waste will flow into the world's oceans and lands.

A broad intervention strategy is therefore urgently needed. In fact, recycling is the most effective way to reduce plastic waste. Yet globally, less than 10% of plastic is recycled.

Now, in a study published in Nature，A team of researchers from theUniversity of Texas at Austin used machine learning algorithms to design a powerful and extremely active hydrolase that can break down plastics in 24 hours that would normally take centuries to degrade.The research provides a solution for large-scale plastic recycling and reuse.

In this study，The researchers focused on solving cluster pairs Ethylene glycol phthalate (PET) contamination problem. PET Is an important polymer widely found in the packaging of most consumer goods, commonly used in mineral water bottles, beverage bottles, fruit and salad packaging and the appearance of various household appliances, among others. PET It now accounts for 12% of the world's total waste. Enzymes capable of degrading PET were reported as early as 2005 and have been preliminarily demonstrated by 19 different PET hydrolytic enzymes (PHE). However, most of these enzymes show significant hydrolytic activity only at high reaction temperatures and highly processed substrates. Most other PET hydrolases are less active at moderate temperatures and neutral pH. This greatly limits in situ/microbial degradation solutions for PET waste. This limit is critical because 40% of plastic waste will bypass recycling systems and flow into the environment.

To do this, the researchers used a machine learning model to predict a mutated version of a natural PET degrading enzyme (PETase) in an attempt to improve the thermal stability and activity of the enzyme.

Through engineering and testing, a mutant of the natural enzyme was created, which the researchers call a FAST-PETase (functional, active, stable, and tolerant PETase).

FAST-PETase contains five mutations compared to wild-type PETase and has higher PET degradation activity compared to wild-type and engineered alternatives. FAST-PETase can dehydrate PET plastics at 30-50℃.

The researchers then demonstrated the enzyme's effectiveness by testing 51 different post-consumer plastic containers, five different polyester fibers, and fabrics and water bottles made entirely of PET with FAST-PETase. In some cases, PET plastic can completely break down into monomers within 24 hours.

In addition, FAST-PETase can produce about 2.8 grams of colorless PET from 3 grams of colored PET, which can be synthesized into food-grade packaging material, a process that breaks through the challenge of recycling mixed-color PET products.

According to the researchers, FAST-PETase is able to truly close the recycling loop.

Taken together, these results demonstrate the application of structure-based machine learning to transform hydrolytic enzymes at moderate temperatures into a wide range of biocatalysts for recycling plastics economy. The research will help solve one of the world's most pressing environmental problems.