web analytics

AI-designed proteins target toxins in deadly snake venom

AI-designed proteins target toxins in deadly snake venom

Artificial intelligence could take the bite out of some snake venoms.

Using AI, scientists have designed proteins that say not so fassst to toxins wielded by cobras and some other snakes. In lab tests, these new proteins saved mice from a lethal dose of those snake toxins.

Researchers shared these results January 15 in Nature. Their approach could one day offer much-needed new treatments for snakebites.

The new proteins “are really doing their job,” says Michael Hust. He was not involved in the work. But he does study antibodies at the Technical University of Braunschweig in Germany. Antibodies are proteins that help a body fight off toxins and other invaders.

With AI-designed proteins, “mice are surviving,” Hust says. And this, he notes, “is what we all want.”

A need for new antivenoms

Worldwide, snakebites kill some 100,000 people each year. Timothy Jenkins has spent years trying to develop new treatments for such deadly injuries. Jenkins works on new medical technologies at the Technical University of Denmark in Lyngby.

Through bites, venomous snakes can deliver a blizzard of toxic chemicals. These are known as toxins. Among the most dangerous are molecules called three-finger toxins. These can paralyze muscles, stilling the heart — and someone’s ability to breathe.

Antivenoms exist to counter them. But the methods used to make them are outdated, Jenkins says. Currently, they’re created by first extracting snakes’ venom. This is “like handling a live hand grenade,” Jenkins says. Then a small dose of that extracted venom gets injected into a horse or other large animal. Its body responds by making toxin-fighting antibodies. People can then harvest those antibodies to use as an antivenom.

Do you have a science question? We can help!

Submit your question here, and we might answer it an upcoming issue of Science News Explores

When given to a snakebite victim, these antibodies will bind to the venom’s toxins and shut them down. But making antivenom this way is costly. It’s also time-consuming. So scientists have been searching for new ways to make antivenom therapies.

Back in 2022, Jenkins came across some research by David Baker. He’s a biochemist at the University of Washington School of Medicine in Seattle. Baker’s work described AI-designed proteins that stick like superglue to non-venom molecules.

That sparked an idea for Jenkins. Could AI think up a design that clamps onto — and takes down — snake toxins, too?

Designed by AI, made by scientists

Jenkins and Baker paired up to do this.

They used a generative AI model called RFdiffusion. This free protein-design tool works sort of like the AIs that generate images. But instead of whipping up a picture of the pope in a puffer jacket, RFdiffusion designs proteins that match a target molecule.

Explainer: What is generative AI?

In the past, Baker’s team had trained their model on all known protein structures. They also had taught the model the strings of amino acids that fold into the 3-D shapes that those proteins make. Then, the researchers asked the model to take those shapes apart. That taught the model how to put together a complete protein from its building blocks — like learning how to build a car engine by taking it apart.

Finally, Baker and Jenkins asked the AI to design proteins that would glom on to the toxins in snake venom. Afterward, they took the designs and made those proteins in the lab. In tests, the proteins kept toxins from latching onto the cells they would ordinarily kill.

In the lab, human cells exposed to toxins from black-necked spitting-cobra venom typically die within about 90 minutes (left). Dying cells appear to light up like a pink Christmas tree. But treating the cells with AI-designed anti-toxin proteins five minutes after toxin exposure helped save them (right).   University of Washington Institute for Protein Design

To see how the proteins would work in mice given a lethal dose of cobra toxins, they treated 20 of those rodents with the new AI-designed proteins. Some mice got the proteins at the same time as the toxins. Others got the proteins 15 minutes later.

Every mouse survived.

“We were very, very excited about this,” Jenkins says. The next step will be to use these proteins to create a product that can be tested in people. Scientists will need to ensure the custom proteins are safe. For instance, they must not bind unexpectedly in human tissues, Hust says.

Jenkins agrees. His team’s new study “was very much just proving that this extremely new technology works.”

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.