Researchers have created an AI-driven platform that can arm immune cells with custom-built proteins—within just 4 to 6 weeks, this new approach could slash development timelines from years to mere weeks, potentially revolutionizing cancer immunotherapy.
Developed by teams at the Technical University of Denmark (DTU) and the Scripps Research Institute, the AI platform digitally engineers these protein components to lock onto peptides presented by cancer cells via pMHC molecules.The innovation, detailed in Science, enables scientists to design protein “minibinders” that guide T cells—our immune system’s frontline warriors—to attack specific cancer targets.
In lab tests, researchers used the AI to target NY-ESO-1, a cancer marker found in multiple tumor types. The team designed a minibinder that latched tightly onto NY-ESO-1 pMHCs. When inserted into T cells, the engineered cells—dubbed “IMPAC-T” cells—successfully hunted and killed cancer cells. The method was also validated on targets from a metastatic melanoma patient, proving its versatility across different cancers. To ensure safety, the researchers developed a “virtual safety check” using AI to predict and eliminate minibinders that might mistakenly attack healthy cells.
This kind of precision is crucial. It significantly reduces the risk of harmful side effects. Clinical trials could begin in about five years. The therapy, similar to existing CAR-T treatments, would involve collecting a patient’s blood, editing their immune cells with the AI-designed proteins, and reinfusing them to act as cancer-seeking missiles. For precision medicine, this is a leap toward faster, safer, and more scalable cancer care—powered by AI.
“We’re essentially giving the immune system new eyes,” says DTU Associate Professor Timothy P. Jenkins, senior author of the study. Traditional therapies rely on identifying T-cell receptors from patients, a tedious and imprecise process. “Our system skips that by designing molecular keys from scratch in just a few weeks,” he adds.
