Because receptors are built at random, a good share of them recognise your own body. The immune system handles this in two stages: delete the worst during training, then police the ones that slipped through.
Central tolerance: the thymus
Young T cells move from the bone marrow to the thymus and go through two tests.
| Test | Question | If it fails |
|---|---|---|
| Positive selection | Can this receptor recognise your own MHC at all? | Dies from neglect. A useless receptor gets no survival signal, and this is the fate of most cells. |
| Negative selection | Does it stick too strongly to your own proteins? | Told to die, because it's dangerous. |
Only a few percent of the T cells that enter come out alive. The survivors have a narrow window: able to see MHC, but not reacting strongly to self. Some cells that grip one of the body's own proteins firmly, just short of the line for deletion, are kept and turned into regulatory T cells, which come up below.
There's a remarkable trick that makes this work. The thymus would normally have no reason to make insulin or the proteins found in the eye. A gene called AIRE makes certain thymus cells produce small amounts of thousands of proteins from all over the body, so T cells can be tested against them. People born with a broken AIRE gene get a severe autoimmune disease attacking many organs.
B cells go through something similar in the bone marrow. A self-reactive B cell gets a chance to swap its receptor for a new one, and it dies if that also fails. Later on, B cells usually depend on T cell help, so a self-reactive B cell without a matching helper rarely gets far.
Peripheral tolerance: everywhere else
The thymus can't show everything, and low-affinity self-reactive cells get through. Everyone has them in their blood. What matters is that they're kept under control.
- Anergy. A T cell that sees its target without a danger signal is silenced rather than activated.
- Regulatory T cells. A specialised population, marked by the protein FOXP3, actively suppresses other cells nearby.
- Checkpoints. CTLA-4 and PD-1 are brakes on the T cell itself. CTLA-4 acts early on, in the lymph node, while PD-1 acts later, in the tissues.
- Deletion. Repeated strong stimulation can push a cell into dying.
- Hiding. Some tissues, like parts of the eye, are largely out of reach of immune cells.
Regulatory T cells and the 2025 Nobel Prize
Shimon Sakaguchi spent the 1980s and 90s arguing for a dedicated class of suppressor cells when the idea was out of fashion. Removing them from mice caused autoimmune disease in many organs, and restoring them prevented it.
Mary Brunkow and Fred Ramsdell then traced a mouse strain with fatal autoimmunity to a broken gene, Foxp3, and showed that mutations in the human version cause IPEX, a severe autoimmune disease of boys. Sakaguchi connected the two: FOXP3 is the master switch for regulatory T cells. The three shared the 2025 Nobel Prize in Physiology or Medicine.
Why this matters
Every autoimmune disease is, in some form, a failure of this system. Cancer therapy takes advantage of it in reverse: the checkpoint drugs deliberately release the brakes so T cells will attack tumours, and the side effects are autoimmune.
Most self-reactive T cells die in the thymus, and those that survive are held back by regulatory T cells and built-in brakes. Having some self-reactive cells is normal. Losing control of them is the disease.