Scaling up precision cancer treatment has become more feasible with the development of an AI platform that customizes protein components and enhances patients' immune cells for fighting cancer.
The innovative method, published in Science, represents a breakthrough as it proves computers can design proteins to redirect immune cells toward cancer cells using pMHC molecules.
ScienceThis advancement drastically reduces the timeline for discovering effective cancer treatments from years to just weeks.
"Essentially, we're providing new vision for the immune system. Current individual cancer treatment methods rely on identifying T-cell receptors in a patient's or donor’s immune system that can be applied therapeutically. This is an extremely time-intensive and challenging process. Our platform leverages an AI to design molecular keys targeting cancer cells at incredible speed, so a new lead molecule can be developed within 4-6 weeks," says Timothy P. Jenkins, Associate Professor at the Technical University of Denmark (DTU) and last author of the study.
The AI platform, a result of collaboration between DTU and the Scripps Research Institute in the USA, aims to overcome a significant hurdle in cancer immunotherapy by illustrating how scientists can produce targeted treatments for tumor cells while sparing healthy tissue.
Typically, T cells naturally recognize cancer cells through specific protein fragments called peptides, displayed on cell surfaces by molecules referred to as pMHCs. Utilizing this knowledge for therapeutic purposes is slow and challenging due to the variability in individual T-cell receptors.
In their study, researchers tested the AI platform's strength against a well-known cancer marker, NY-ESO-1, found across various cancers. They successfully designed a minibinder that tightly bonded with NY-ESO-1 pMHC molecules. When inserted into T cells, this protein created a unique product called IMPAC-T cells, which effectively directed T cells to kill cancer cells in lab experiments.
"It was thrilling to witness these computer-generated minibinders working efficiently in the laboratory," says Kristoffer Haurum Johansen, postdoc and co-author of the study at DTU.
The researchers applied their pipeline to design binders for a melanoma patient's target cancer type as well, generating effective binders for this specific target. This demonstrated that their method could be tailored for personalized immunotherapy against various cancers.
A significant innovation was creating a virtual safety check. Using AI, the team screened designed minibinders in relation to pMHC molecules found on healthy cells, filtering out those that might cause harmful side effects before experimentation.
"Precision is paramount in cancer treatment. By anticipating and eliminating cross-reactions during design, we minimize risks associated with these proteins while boosting the likelihood of creating a safe and effective therapy," says Sine Reker Hadrup, Professor at DTU and co-author of the study.
Jenkins anticipates that it could take up to five years before this new method is ready for initial human clinical trials. When ready, treatment will resemble current procedures using genetically modified T cells (CAR-T cells), which are used to treat lymphoma and leukemia.
The process involves drawing a patient's blood at the hospital, similar to a standard blood test. Immune cells from this blood sample are then extracted and modified in the lab to carry AI-designed minibinders. These enhanced immune cells are re-administered to the patient, acting like targeted missiles that precisely attack cancer cells.