Researchers have used AI-based design tools to create new enzymes that can break down specific proteins, demonstrating their effectiveness by cutting proteins associated with neurological diseases. The development short-circuits the natural evolutionary process that typically diversifies protein-digesting enzymes over long periods.
Proteins consist of chains of amino acids connected by peptide bonds. These bonds are inherently stable, with an estimated half-life of several hundred years, which explains why most proteins do not persist indefinitely in Earth's environment. Organisms have evolved mechanisms to catalyze the breakdown of these bonds, resulting in proteases that generally degrade proteins quickly once they enter the environment.
The new approach involves taking the catalytic core from two types of protease and replacing the rest of the protein structure. This allows the resulting enzymes to be tuned to cut specific target proteins in half. By using these AI-designed tools, the researchers bypassed the need for natural diversification, creating specialized enzymes that perform targeted digestion rather than general breakdown.
The chemical foundation of these processes lies in the peptide bond itself, formed when an amino group links to an acid group, releasing a water molecule. Amino acids possess both an amino and an acid group, enabling them to form bonds on two sides and build long chains. The side chains of these amino acids provide distinctive chemistry that gives the protein backbone its specific properties.