Researchers use AI to design enzymes that cut specific proteins

In short: Researchers used AI design tools to build new enzymes that can cut chosen proteins, and tested them on disease-related targets.

What happened

Proteins are long chains of building blocks called amino acids. The links between them, called peptide bonds, are so stable they could last for hundreds of years on their own. In living things, special enzymes called proteases act like scissors and can cut these bonds quickly.

Researchers, mainly from David Baker’s lab at the University of Washington, used AI-based protein design tools to create new proteases. Instead of copying a full natural enzyme, they kept the small “cutting” part and redesigned most of the rest. They tried two types, a zinc protease (it uses a zinc atom to help the cut) and a cysteine protease (it uses one specific amino acid as the cutter).

The team used software to draft new protein shapes and then pick the amino acids that would fold into those shapes. They aimed the designs at specific target sequences, meaning the enzyme should cut at a chosen spot. Some designs worked right away, and others improved after the researchers made many small changes, similar to trying different parts to make a tool fit better.

As a demonstration, the researchers designed zinc proteases that could cut proteins linked to neurological disorders, including Alzheimer’s and frontotemporal dementia. They reported cutting the intended targets in cells, but they also noted that turning this into a treatment would be difficult.

Why it matters

If scientists can reliably design protein “scissors” that only cut one target, it could eventually help research on diseases caused by harmful proteins. But real-world use is not close, because delivering these enzymes into the right cells and avoiding immune reactions is still a major challenge.

Source: Arstechnica