AI-Designed Proteins May Offer Protection Against Lethal Snake Venom

By utilizing AI, scientists have engineered proteins that counteract the toxins from cobras and other venomous snakes. This proof-of-concept approach could eventually lead to a new treatment for snakebites. In laboratory tests, the tailored proteins saved the lives of mice exposed to a lethal dose of toxins, as reported by researchers on January 15 in Nature.

“These proteins are truly effective,” says Michael Hust, an antibody researcher at the Technical University of Braunschweig in Germany, who was not involved in the study. “The mice are surviving. This is exactly what we aim for.”

This research builds on the groundbreaking work that earned three scientists the 2024 Nobel Prize in Chemistry. In 2022, medical biotechnologist Timothy Jenkins came across a preprint from David Baker’s lab at the University of Washington, one of the Nobel laureates. The preprint described AI-designed proteins that adhere strongly to specific molecules.

This sparked an idea: could AI design a protein that binds to and neutralizes snake venom toxins?

Jenkins’ Quest for New Snakebite Therapies

Jenkins, from the Technical University of Denmark, had spent years trying to create new therapies for snakebites, which claim around 100,000 lives globally each year. Venomous snake bites can inject a wide range of toxins, including three-finger toxins, which paralyze muscles, stop hearts, and disrupt breathing. While antivenoms exist, Jenkins notes that the technology is outdated. “There’s not much profit in it, so innovation has been limited,” he says.

Currently, antivenom production involves milking venom from snakes, a dangerous process, and injecting small amounts of venom into horses or other animals to harvest antibodies. These antibodies, when administered to snakebite victims, neutralize the toxins. However, manufacturing antivenoms is expensive and slow, prompting scientists to explore alternative methods. One promising approach involves screening a large collection of lab-made antibodies to identify those that target specific toxins.

With AI, scientists can now design toxin-targeting proteins more quickly and affordably. Jenkins and Baker teamed up to create custom proteins using an AI model called RFdiffusion, a free protein-design tool similar to image-generating AIs. Rather than creating images, RFdiffusion generates protein designs that precisely target molecules scientists wish to neutralize.

Teaching the Model to Build Proteins from Known Structures and Amino Acid Sequences

Baker’s team had previously trained the AI model using all known protein structures and their amino acid sequences—the molecular building blocks that fold into a protein’s 3D shape. They then computationally deconstructed these shapes, teaching the model how to assemble a complete protein from its parts, much like learning to build a car engine by disassembling one.

Baker and Jenkins used the AI to design proteins that would bind to venom toxins, and then they synthesized these proteins in the lab. Like a magnetic cover that prevents a key from fitting into a lock, the custom proteins stopped the toxin from attaching to cells.

The team injected 20 mice with the proteins 15 minutes after or simultaneously with a lethal dose of cobra venom. All the mice survived.

“We were very, very excited about this,” Jenkins says, calling it a clear demonstration of the proteins’ effectiveness. The next step for the team is to develop these proteins into a product for human testing. Researchers will need to ensure that the custom proteins are safe and do not bind unexpectedly to human tissues, according to Hust.

Jenkins acknowledges that this study is just the first step toward neutralizing venom’s harmful effects. “It was very much about proving that this incredibly new technology works,” he says.

Read the original article on: Science News

Read more: Scientists Recommend Eating More Python