Why don't immune cells attack the body's tissues?
Researchers from Cold Spring Harbor Laboratory have unveiled a new mechanism explaining how T cells in the thymus, located above the heart, learn to avoid attacking the body’s own tissues. They achieve this by employing a process akin to generalization in artificial intelligence (AI) model systems. Each T cell encounters only a small subset of the body’s self-protein fragments, yet it must learn not to attack the rest. This process is known as negative selection, in which T cells that bind strongly to self-protein fragments are eliminated, thereby preventing the immune system from attacking healthy tissues.
Assistant Professor Hanna Meyer explained that the open question in immunology was how T cells could learn to avoid “friendly fire” if they were required to test against every peptide in the body—a task that could take a long time. The team hypothesized that the answer lies in the process of generalization. Associate Professor Sachit Navlakha noted that generalization is familiar in AI, where a model can learn from limited examples and then apply that learning to new cases, such as a dog-recognition model that does not need to see every image of a dog on the internet.
• The abundance of self-peptides in the thymus matches their abundance in tissues throughout the body, meaning that self-peptides common in the body tend to be common in the thymus as well, which is a fundamental prerequisite for the success of generalization.
• The ability of T cell receptors to recognize several similar peptides (cross-reactivity) provides the immune system with a shortcut. A T cell does not need to encounter every peptide it might later attack; if it recognizes a similar self-peptide during training, the thymus can eliminate it before it becomes dangerous.
The researchers formulated a model explaining how the thymus can function as a training environment and the body as a testing environment, helping to explain why this process fails in certain cases, leading to autoimmune diseases.
The researchers believe that this new model opens avenues for a deeper understanding of autoimmune diseases and may pave the way for developing new treatments targeting the correction of defects in the immune training process, potentially revolutionizing the treatment of these chronic diseases.
