G7 · The Scientist Gallery · Attribution Series

Victora & Nussenzweig

B cells mutate their own receptor genes a million times faster than background and are killed unless it helped. The selection is not on affinity directly — it is on how much antigen you can show a T cell that has only so much help to give.

Part I · Rockefeller, 2012

Directed evolution, inside a body

Another title correction

The stub this page replaces was titled Victor Nussenzweig — the malaria immunologist at NYU, who with Ruth Nussenzweig developed the circumsporozoite work behind the RTS,S vaccine. The Book VII index card cites Victora & Nussenzweig, 2012, review in Cell: that is Gabriel Victora and Michel Nussenzweig at Rockefeller, on germinal centres. Michel is Victor and Ruth's son. The index card was right; the stub's title named the wrong Nussenzweig. CITED

A germinal centre is a structure that appears in a lymph node days after an infection, runs for a few weeks, and dissolves. Inside it, B cells deliberately mutate the genes encoding their own antigen receptor, at a rate about a million times the background — roughly $10^{-3}$ per base pair per division — and are then killed unless the mutation helped.

It is directed evolution, run inside a body, on a timescale of days. The output is antibody whose affinity has risen by four to five orders of magnitude.
Part II · The Indirect Criterion

Help is what is scarce

The question Victora and Nussenzweig settled is how the selection step works, and the answer is not the obvious one.

The obvious mechanism would be direct: a B cell whose receptor binds antigen better survives because it binds antigen better. What the two-photon imaging and the photoactivatable-GFP experiments showed is that the competition is indirect. A B cell in the light zone captures antigen from follicular dendritic cells, processes it, and presents it to a T follicular helper cell. T-cell help is the limiting resource, and it is allocated in proportion to how much antigen a B cell can present.

What that buys

Affinity is converted into a presented quantity, and the selection acts on that. The cells are not compared against a fixed standard; they are compared against each other, for a supply of help that does not grow. It is a relative criterion, and it moves as the population improves.

Cells that win help return to the dark zone, divide, mutate again, and come back. The cyclic re-entry model, proposed decades earlier, was confirmed by photoactivating cells in one zone and finding them in the other. CITED

Part III · The Arithmetic

Five orders of magnitude, a few rounds

The arithmetic of the whole process is a compounding, and it is tight.

what each round has to deliver

106 → 1011 L/mol is 105-fold, so
 5 rounds  →  1.000 dex/round  (10.00×)
 6 rounds  →  0.833 dex/round  (6.81×)
 8 rounds  →  0.625 dex/round  (4.22×)
10 rounds  →  0.500 dex/round  (3.16×)  COMPUTED

A toy model — mutate log-affinity by a Gaussian of $0.30$ dex, keep the top ten per cent, expand, repeat — reaches $5.89$ dex in ten rounds. That is $7.8\times10^{5}$-fold, against an observed $10^{5}$.

The model overshoots, and that is the finding

A toy selection scheme with no biology in it beats the real germinal centre by most of an order of magnitude. So something in the real system is holding it back — and the candidates are all interesting: affinity has a ceiling set by diffusion, most mutations destroy the receptor outright rather than degrading it gently, the selection is far less than top-ten-per-cent stringent, and the antigen supply is itself being consumed.

This page does not claim to know which. It claims that the gap exists and is the right thing to ask about. OPEN

An earlier draft of this chapter went further and claimed an optimal mutation rate — too little and nothing improves, too much and the receptors are destroyed. The model written to show it was monotone: every increase in mutation size improved the outcome, up to sizes that are biologically absurd. The claim was dropped rather than the model tuned until it produced one. The trade-off is real in the literature; it is not demonstrated here. OPEN

Part IV · What This Page Will Not Claim

Two conjectures, marked

The index card for this chapter proposes that each selection round is a step of the corpus's n-bonacci ladder, converging on $\tau = 2$. This page does not assert that. The germinal centre's per-round gain is set by mutation size and selection stringency, both measurable, and nothing measured here produces a recurrence. Recording the conjecture as a conjecture is the most this chapter can honestly do with it. OPEN

What it can assert is the shape, and the shape is the one this gallery keeps finding. A blind generator, a constraint that removes almost everything, and iteration — and what comes out is not designed but is also not arbitrary, because at every round the constraint left only a few directions open.

The thymus does this once, destructively, to build a repertoire that will not attack you. The germinal centre does it repeatedly, constructively, to build one antibody that will. Same operator, opposite sign, and the two organs are the only places in the body where a cell's own genome is deliberately damaged as part of normal function.
Part V · Placement

Where this is taught

And where it is taught:

TextWhere
Janeway’s Immunobiology, Murphy & WeaverGerminal centre dynamics and affinity maturation — the intravital imaging and the cyclic re-entry model of B cells between light and dark zones
Kuby Immunology, Punt et al.B-cell activation, differentiation and memory; germinal centres, Ch. 7 and 10 depending on edition
Annual Review of ImmunologyVictora & Nussenzweig, “Germinal centers”, vol. 30, 2012 — the review this chapter is built on
Reading the primary sources

“Victora-Nussenzweig” names two laboratories, not a book. The 2012 Annual Review of Immunology article is the consolidation (PMID 22224772); the 2010 Cell paper is the experiment. Rockefeller deposits its authors’ work in its Digital Commons and the primary papers are indexed in PubMed Central, which is the route in if a library portal is not to hand.

The 2010 Cell paper is eighteen months older than the review, and the review is in the textbook chapter. Cyclic re-entry had been a model since the 1990s and an argument for as long; photoactivating a cell in one zone and finding it in the other is what ended the argument, and the speed of the passage into teaching is the measure of how decisive the experiment was.

Place in the Series

Where this sits on the operator map

OperatorIn this chapterIn dm³
Csomatic hypermutation — a blind generator at $10^{-3}$/bp/divisioncompression: variation, produced without direction
Kantigen captured, processed, presented — affinity converted to a quantitythe quantity driven toward the selection threshold
FT follicular helper cells, finite — the cut is relative and movesthe fold CITED
Ureturn to the dark zone, divide, mutate, come backcyclic re-entry — the branch re-entering the chain

Verification

Every number on this page is produced by book7/ch-victora-nussenzweig-verify.py. It records in its own closing block what it establishes and what it does not.

References

G. D. Victora and M. C. Nussenzweig, “Germinal centers”, Annu. Rev. Immunol. 30, 2012, 429–457. · doi 10.1146/annurev-immunol-020711-075032 · PMID 22224772.
G. D. Victora et al., “Germinal center dynamics revealed by multiphoton microscopy with a photoactivatable fluorescent reporter”, Cell 143, 2010, 592–605.
A. D. Gitlin, Z. Shulman and M. C. Nussenzweig, “Clonal selection in the germinal centre by regulated proliferation and hypermutation”, Nature 509, 2014.
C. D. C. Allen, T. Okada and J. G. Cyster, “Germinal-center organization and cellular dynamics”, Immunity 27, 2007.
K. Murphy and C. Weaver, Janeway’s Immunobiology, 10th ed., Garland — germinal centre dynamics and affinity maturation.
J. Punt et al., Kuby Immunology, 8th ed., Macmillan — B-cell activation, differentiation and memory.
M. Meyer-Hermann et al., “A theory of germinal center B cell selection, division, and exit”, Cell Reports 2, 2012 — for the modelling this page's toy scheme is a crude shadow of.

← Ludger Klein Back to G7 Index Élie Metchnikoff →