18. Take the plate away and nobody can see · Prof. Kwon's TCR-HLA and CD4/CD8, First Steps

From lesson 1 to here, what a T cell sees has always been fragment + plate.

Which means there are two ways to escape: change the fragment, or remove the plate.

Change the fragment

Lose one mutation and one fragment becomes invisible. The others stay exactly as they were.

Remove the plate

Every fragment that was riding on that plate becomes invisible. One move, done.

Tumours use the second one often. And the second one has a shortcut.

No exceptions. All 174,982 of them.

Python Runs in your browser

# What is class I always paired with
import thla

d = thla.vdj("I")
print("partner chain on class I rows")
for a, n in d["mhc_b"].value_counts().items():
    print(f"   {a:<10}{n:>10,}")
print()
n = int((d["mhc_b"] == "B2M").sum())
print(f"of {len(d):,} class I rows, paired with B2M : {n:,}   ({100*n/len(d):.1f}%)")

Why 100%

Class I cannot stand up on its own chain. A small protein called β2-microglobulin has to prop it from below before it will fold; it has to fold before it can load a fragment; and it has to load before it reaches the surface.

HLA-A · B · C · E · F · G — all use the same single B2M

And B2M is one gene on chromosome 15. Not six copies like HLA — a single kind.

So break B2M and every class I molecule leaves the surface at once. From a tumour's point of view there is no cheaper move. B2M mutations turn up repeatedly in melanomas and lung cancers that resist checkpoint inhibitors.

Different architecture, different weak spot. There is a reason tumours dodge CD8 first.

Python Runs in your browser

# Does class II have the same weakness
import thla

d2 = thla.vdj("II")
v = d2["mhc_b"].value_counts()
print(f"distinct partner chains on class II : {len(v)}")
print()
for a, n in v.head(8).items():
    print(f"   {a:<20}{n:>6,}")
print()
print("In class II the beta chain is itself several genes - DR, DQ, DP.")
print("Lose one and the rest remain. Not a single point of failure.")

Switch parts off one at a time. What goes off with them is the point of this lesson.

Loading…

Loss does not arrive in only one form

💥

B2M mutation

All class I disappears. Not reversible.

✂️

HLA LOH

Only the paternal or the maternal set is lost, whole. The remaining plates are perfectly fine — half the surface goes blind.

🚚

TAP · proteasome

The route that carries fragments to the groove is cut. The plates are there, and empty.

🔇

Expression silencing

The gene is intact but switched off. Give interferon-γ and it comes back.

The gap between the fourth one and the rest is large in the clinic.

Where this lesson meets the earlier ones

Fact from this lesson	Which lesson it came from	What it overturns
Every class I uses one B2M	Lesson 2 · groove architecture	six sets of plates hang off one gene
CD8 grabs α3	Lesson 9 · the coreceptor loop	no class I means no platform to grab either
HLA-E is nearly monomorphic	Lesson 17 · non-classical plates	it becomes an escape route a tumour can choose
Alloreactivity reads the plate	Lesson 15 · alloreactivity	with no plate, transferred T cells are no use either

That table is the real point of this lesson. A plate is not background; it is a part, and parts break.

🤔 Check yourself

Why does one B2M mutation remove all class I from a tumour?

🤔 Check yourself

How do you tell a mutation from expression silencing when a tumour shows no class I?

🤔 Check yourself

Why is a tumour cell that has cleared away all class I still at risk?