20. What we can say, and what we cannot · Prof. Kwon's TCR-HLA and CD4/CD8, First Steps

We covered one thing. The place where a TCR and an HLA meet. Not how the receptor gets built, not how the fragment gets into the groove.

Here is what came out of that one place.

Click a lesson. You get the one thing it decided, and what would have been overturned had it been decided differently.

Loading…

Twenty lessons in three sentences

👀

It sees the two together

A TCR sees neither the fragment nor the plate but the single face the two make. Which is why the same fragment on a different plate is a different thing.

📍

The polymorphism is in the groove only

What differs between people sits on the face a TCR touches, and the platform a coreceptor grabs barely moves. That is why one CD8 works on every class I there is.

⏱️

Not strength but time

It has to bind weakly and stay long enough for the signal to complete. The weakness was not an unfinished job — it was the design.

Come back to this cell whenever you lose track of where a number came from.

Python Runs in your browser

# Pull out every number this course used, in one place
import thla

f = thla.facts()
print("data")
print(f"   VDJdb human rows          {f['n_vdj']:>10,}")
print(f"   class I / II              {f['n_class']['I']:>10,} / {f['n_class']['II']:,}")
print(f"   distinct epitopes I / II  {f['n_epitope']['I']:>10,} / {f['n_epitope']['II']:,}")
print(f"   alleles appearing         {f['n_allele_seen']['I']:>10} / {f['n_allele_seen']['II']}")
print()
print("plates")
print(f"   HLA-A alleles             {f['hla_n']['A']:>10,}")
print(f"   HLA-E alleles             {f['hla_n']['E']:>10,}")
e = thla.domain_entropy("A")
print(f"   HLA-A alpha1 entropy      {e.loc['α1','mean']:>10.4f}")
print(f"   HLA-A alpha3 entropy      {e.loc['α3','mean']:>10.4f}")
print()
print("cross-reactivity")
print(f"   distinct CDR3beta         {f['cross']['n_cdr3b']:>10,}")
print(f"   binding two or more       {f['cross']['multi']:>10,}")

What we can say, and what we cannot

We can say

· which domain HLA polymorphism sits in
· how little the loop a coreceptor grabs varies
· how many rows of each plate and fragment this table holds
· how many fragments a given CDR3β spans
· roughly what share of people a set of plates covers, from cited frequencies

We cannot say

· how tightly any given TCR actually binds (we never measured K_D)
· the true size of cross-reactivity (we only know what was tested)
· which fragments would be immunogenic (we predicted nothing)
· whether a therapy will work in a given patient
· what any of it looks like (we used no structural coordinates)

The right-hand box is longer. That is normal.

Where to go next

What you want more of	Where it lives
How the receptor gets built — V(D)J recombination and the repertoire	"T Cell Receptors and the Repertoire, First Steps"
How a fragment reaches the groove — proteasome, TAP, neoantigens	"Antigen Presentation and Neoantigens, First Steps"
What switches on afterwards — PD-1, CTLA-4, checkpoint drugs	"Immune Checkpoints and Immunotherapy, First Steps"
The life and subsets of a T cell — naive, memory, exhausted	"T Cells, First Steps"
How a tumour gets away — the whole map of escape	"Tumour Immune Evasion, First Steps"

All five branch off the same point as this course. Whichever way you go, you will keep using the words fragment and plate.

🤔 Check yourself

Which of these did this course not measure?

🤔 Check yourself

Why does it matter that the loop a coreceptor grabs barely varies?

🤔 Check yourself

What warning did this course return to for twenty lessons?