With a key step towards faster, more personalized cancer treatments, scientists have unveiled a powerful AI-driven method to create custom-designed immunotherapy in just a few weeks. Instead of relying on a slow and complicated task to find the match of the innate immune system, this new approach uses artificial intelligence to create accurate proteins that can direct immune cells directly to cancer targets.
This method works by designing small synthetic proteins (coll minibinders) that stick to specific molecules in cancer cells. These molecules are known as the peptide-major histocompatibility complex (PMHC) and present protein fragments from within cancer cells on its surface.
Typically, T cells in the immune system use T cell receptors to recognize these fragments. However, it can take months or years to find and test the appropriate T-cell receptors from a patient or donor. Thanks to advanced generative models and simulations, scientists can now create these mini-binders completely computer-made, effectively test them, and create practical prototypes in the lab within just 4-6 weeks.
Faster ways to train your immune system
In a study published in Journal Science, researchers from the Institute of Technology Denmark (DTU) and researchers from the Scripps Research Institute described how AI platforms were used to target the well-known cancer protein called NY-ESO-1. This protein is found in many types of tumors and is known to activate the immune system.
Researchers trained the AI to design a mini-bind that can adhere firmly to the NY-ESO-1 PMHC structure. Once the protein was created it was inserted into the lab immune cells. These modified cells, called Impac-T cells, showed strong ability to recognize and kill cancer cells carrying NY-ESO-1 markers.
“The researchers are also working to help people understand the importance of their efforts,” said Kristoffer Haurum Johansen, a postdoctoral researcher at DTU and research co-author.
Related Stories
Fighting cancer with a digital blueprint
AI systems do not work with known targets alone. Scientists tested again this time using different PMHC targets from patients with metastatic melanoma. This target, called rvtdesilsy/hla-a*01:01, was not previously structurally mapped. Without known structures, the platform successfully generated minibinders that match new cancer proteins, opening the door for designing treatments for previously unstandardized cancers.
This is a major advantage in precision medicine. Instead of relying on limited data available or inaccessible immune cells, scientists can now use digital models to create effective treatments for targets specific to each patient's cancer. “We essentially create new eyes in the immune system,” says Timothy P. Jenkins, an associate professor at DTU and a senior author of the study. “Our platform designs molecular keys to target cancer cells using AI platforms, and they do so at an incredible speed.”
Smart screening for improved safety
One of the most challenging parts of the development of new immunotherapy is to avoid attacking healthy cells. Some cancer markers resemble proteins found in normal tissues. If treatments accidentally combine with these, they can cause serious side effects.
To prevent this, the researchers added a “virtual safety check” to the process. Computer simulations were used to test each minibinder against a wide range of PMHCs found in healthy cells. This allowed us to rule out potentially harmful designs before lab testing began.
“The accuracy of cancer treatments is extremely important,” said Sine Reker Hadrup, a professor at DTU and a co-author of the study. “By predicting and eliminating cross-reactions already in the design phase, we were able to reduce the risks associated with the designed proteins and increase the likelihood of designing safe and effective treatments.” This predictive step is key to making these treatments safe for people one day. By eliminating dangerous binders early, researchers can focus their resources only on the most promising and safe molecules.
From the bench to the bedside
The results of the lab are promising, but scientists still need to move carefully. Jenkins estimates it will take about five years for the first human clinical trial to begin. As the trial progresses, treatment may resemble existing methods used for certain hematological cancers.
In these treatments, doctors collect the patient's blood, extract the immune cells, and modify them in the lab. The modified cells are then returned to the patient's body, where they hunt and kill cancer cells.
With new AI methods, the process is faster and more personalized. Instead of waiting to find the right immune cells or matching matching receptors, doctors can use digital tools to design the perfect fit. This could be a game-changer for patients with solid tumors where current immunotherapy success is limited.
By adjusting treatment to each patient's tumor marker, doctors may be able to quickly treat cancers that were previously untreated. Furthermore, AI methods allow for easy exploration of rare or individual mutations that traditional therapies have overlooked.
A new era of precision oncology
The ability to design effective cancer target proteins from scratch has long been the goal of immunotherapy. This breakthrough shows that it is possible to move from digital blueprints to occupational immunotherapy in just a few weeks.
The team's work also demonstrates that these synthetic minibinders can function in the same way as natural receptors in directing T cells to target. When tested in the lab, engineered immune cells were as fatal to cancer as those created using natural methods, but were faster and safer.
Continuing improvements make this AI-powered approach more accessible, more accurate and much faster than ever. The future of immunotherapy is no longer just about understanding biology, but also about designing solutions with digital tools.
