What has an immune system seen? The answer to this question could be incredibly valuable for patients with cancer, autoimmune disorders and infectious diseases. Knowing what patient’s immune cells respond to could pave the way for better treatment. Personalized vaccines, cancer immunotherapy response prediction, new therapies…
foto: First authors Marius Messemaker (right) and Bjørn Kwee
Daunting task
Understanding the code controlling how T cells recognize antigens is one of the most fundamental challenges in immunology. Our T cells carry an enormous diversity of receptors, and each disease presents its own set of antigens. Working out how the pool of T cells in a patient recognizes the antigens specific to their disease is, at best, a daunting task.
So far, computer models, such as Alphafold3, have not shown to be a great help. “People thought the models weren’t good enough”, says group leader Ton Schumacher. “We wondered: could the lack of performance also be due to the quality of the data?”
In-house technology
And so his group came up with a huge experiment. They created thousands of T cell receptors with their in-house technology and tested them for their reactivity to known viral antigens. “Half of the public data turned out to be incorrect”, says PhD Marius Messemaker, referring to the commonly used reference database VDJdb, which contains data from around 200 labs around the world. “So a large part of the data used to judge existing prediction models was unreliable.”
Unseen antigens
“The good news is: using our validated data, we showed that AlphaFold3 distinguishes T cell receptors that recognize these viral antigens from those that don’t about three times out of four”, says PhD Bjørn Kwee. “In addition, we found that the model could shortlist TCRs that recognize patient-specific cancer antigens in a melanoma patient, without ever having seen those antigens during model training.”
“For the first time, we’re hopeful about the feasibility of in silico prediction of T cell receptor reactivity based on genetic sequence information alone”, says Schumacher. “We still have major steps to make, but it’s an important step towards personalized therapies.”
è Read the publication in Immunity
Research at the NKI is financially supported by KWF Dutch Cancer Society and the AVL Foundation. This project was supported by Cancer Grand Challenges.