Researchers at MIT and Microsoft have used Artificial Intelligence to design molecular sensors that could detect early signs of cancer via a urine test. The sensors rely on short proteins, or peptides, that are engineered to be cut by specific proteases, which are enzymes known to be overactive in cancer cells. When these sensors are introduced into the body, they circulate and interact with cancer-linked proteases, creating a measurable signal in urine that could reveal the presence of disease at an early stage.
The method centers on nanoparticles coated with peptides that are designed by an Artificial Intelligence model to meet precise biochemical criteria. Nanoparticles coated with these proteins, called peptides, can give off a signal if they encounter cancer-linked proteases once introduced into circulation: The proteases will snip off the peptides, which then form reporter molecules that are excreted in urine. Earlier work in this area emerged from the lab of Sangeeta Bhatia, which previously used trial-and-error approaches to find peptides cleaved by specific proteases, but those efforts often produced ambiguous results. The new Artificial Intelligence model makes it possible to optimize peptides for sensitivity and specificity to particular proteases.
According to principal researcher Ava Amini, the ability to tune sensors to a protease that is key to a certain cancer can provide a highly sensitive and specific diagnostic signal. Bhatia’s lab is now collaborating with ARPA-H on an at-home kit that could potentially detect 30 types of early cancer, using these Artificial Intelligence designed peptides as the core sensing element. The same design framework could also allow these peptides to be incorporated into cancer therapeutics, extending the impact of the technology beyond diagnostics into treatment.
