0x668143c4…385fsent to0x30cccc3d…0f29·#11,865,625·view on Etherscan
B-cells recognize non-self only, and T-cells recognize self only
The "needle in a haystack" problem in the adaptive immune system is solved if T-cell receptors recognize self, and not the other way around. This also has a nice symmetry, one half of the adaptive immune system recognizes non-self, and the other half only self. Contrast.
Antigen presentation to T-cells with MHC I and II is a young science. It has not been explored that long. Younger than antibodies, since those were observable more easy. And since antibodies clearly bind to non-self, there would be a priori prejudice to assume T-cell receptors did as well.
It is known that self-antigen are fewer than non-self. This is self-evident. What is not as intuitive is that the self-antigen that MHC can present, might only be a fraction of all self-antigen peptide fragments that can be produced from the human body. And T-cells that learn to recognize self, with one self-antigen per T-cell receptor, would only have to recognize the self-antigen that MHC can present.
The upper bound for self-antigen that can be recognized by T-cell receptors, can be estimated quite easily. If MHC on average presents 10 amino acid long peptide fragments, there is 20^10 = 10^13 possible combinations. 20 being the number of amino acids. Out of these 10^13 combinations, MHC is only able to bind a fraction.
To then estimate how much of that fraction is self-antigen, the possible combinations of 10 amino acid long sequences that the human genome can encode can be delineated with the size of the human genome, 3*10^9 base pairs, maximally encoding 10^9 codons, that encode an amino acid each. If the fragments can be sliced from any point in the genome, then theoretically, any 10 amino acids from any point in the genome could be unique (since 10 amino acid long sequences can generate 10^13 combinations, a 10*10^13 = 10^14 long string of base pairs, whereas human genome is just 10^9. ) 10^9 combinations is a fraction of 0.0001 of 10^13 combinations, 10^9/10^13 = 10^-4.
The T-cell receptor diversity to recognize self becomes 10^13*fraction_MHC_can_present*10^-4 = 10^9*fraction_MHC_can_present.
During MHC presentation, the antigen would need to be verified to every single self-antigen MHC can bind, but, this might not be such a large population of T-cells.