Tumor Immunology Meeting 29-05-2017
How T cell cross-reactivity helps the immune system learn self-nonself discrimination
Inge Wortel
Department of Tumor Immunology,
Radboudumc
The immune system's T cell repertoire has to:
Diversity
Tolerance
1 Sewell (2012). Nature Reviews Immunology. 2 Yu et al (2015). Immunity. Images adapted from The Immune System, 3rd ed.(2009), Garland Science.
The T cell repertoire needs to learn which epitopes are "self" based on only a few examples in the thymus
Generalization:
Infer common features of a group of objects from
a few examples
Group of objects : paintings by Rembrandt
A description would be tedious! $\rightarrow$ Give examples:
Now find the third Rembrandt...
or
Can the T cell repertoire infer whether an epitope is "self" or "foreign" because of generalization during negative selection?
Only if:
1. Self epitopes have some common features
2. T cells can see those similarities
Cross-reactivity: T cells recognize multiple (related) epitopes
Three ingredients needed:
1 | Epitope | |
2 | (CD8+) TCR | |
3 | Matching rule |
Three ingredients needed:
1 | Epitope | peptide that binds MHC-I (9-mer) |
2 | (CD8+) TCR | |
3 | Matching rule |
Three ingredients needed:
1 | Epitope | peptide that binds MHC-I (9-mer) |
2 | (CD8+) TCR | "implicit" motif & matching length $k$ |
3 | Matching rule |
Three ingredients needed:
1 | Epitope | peptide that binds MHC-I (9-mer) |
2 | (CD8+) TCR | "implicit" motif & matching length $k$ |
3 | Matching rule | at least $k$ matches in a row |
Which peptides can bind to a given TCR?
$k$ | # binders | frequency |
6 | 1 | 1 in 64,000,000 |
5 | 39 | 1 in 650,000 |
4 | 1,160 | 1 in 55,000 |
3 | 30,800 | 1 in 2,000 |
2 | 766,879 | 1 in 80 |
1 | 16,954,119 | 1 in 4 |
lower $k$ | $\rightarrow$ less specific |
$\rightarrow$ more cross-reactive |
Fewer T cells survive if:
Precursor frequency: specific cells per million T cells
Higher tolerance if: | - More self "seen" |
- "Smart" choice of training set | |
- Cross-reactivity (lower $k$) |
Also foreign epitopes that "look like self" are deleted!
$\rightarrow$ overgeneralization
When T cells "see" all self epitopes during selection:
$\rightarrow$ With too high cross-reactivity ($k$ = 3), no cells survive that can detect foreign epitopes.
Generalization in the immune system:
If T cells do not see the whole "self": "unseen self" might still be recognized!
$\rightarrow$ but maybe with a lower precursor frequency?
Learning score: | 0 = no difference between "unseen self" and "foreign" |
1 = perfect discrimination |
Computed "smart" set is enriched in rare AAs, depleted of common AAs
$\rightarrow$ Could enrichment of rare AAs help self-nonself discrimination?
Bias for peptides with rare AAs:
$\rightarrow$ enough to improve self-nonself discrimination?
Generalization in the immune system:
Self-nonself discrimination cannot be perfect!
Test robustness:
How important in vivo?