B-Cell Depletion: Why Anti-CD20 Spares the Antibody

You can remove almost every B cell from an animal and barely move its antibody titre. That is not a failed experiment — it is the expected result, and it follows from one fact: CD20 is switched off as a B cell becomes a plasma cell. Anti-CD20 clears the recirculating pool and leaves the long-lived plasma cell, which has no CD20 to bind, sitting in its bone-marrow niche secreting the pathogenic antibody exactly as before. Depleting B cells and depleting antibody are two different experiments, and conflating them is the most common mistake in this field.

B-cell depletion and the BAFF–APRIL axis, target through to readout. Open the interactive version to click any protein for its role and the matching validated reagent.

The target you choose decides which cells you keep

B-lineage surface proteins switch on and off along development, and a depleting antibody can only see what is expressed at the moment it arrives. B220 (CD45R) covers essentially the whole mouse B compartment and gives the broadest depletion. CD19 appears earlier than CD20 and — critically — is retained on plasmablasts, so anti-CD19 reaches a stage anti-CD20 does not. CD22 marks the mature cell and is the inhibitory co-receptor of the BCR. At the far end, CD38, BCMA and CD138 mark plasmablasts and plasma cells.

Laid out that way, the choice becomes a question rather than a preference. If the goal is to remove the cells that will become antibody factories, a CD20 or CD19 agent is right. If the goal is to remove existing antibody, you need something that sees the plasma cell — and that means the CD38 or BCMA end of the row, with the different toxicity profile that comes with it.

Depletion changes the survival economy

BAFF and APRIL are made by myeloid and stromal cells at a rate that does not fall when B cells disappear. Remove most of the compartment and the survivors face a large excess of ligand. Free BAFF rises after depletion, and the cells that made it through are rescued.

Mechanically, BAFF-R is dominant on naive and memory cells. Engagement drives degradation of TRAF3, which normally holds NIK (MAP3K14) in a state of constant destruction. Release NIK and the non-canonical NF-κB pathway switches on; the output is BCL-2 and MCL-1, and therefore survival. TACI and BCMA read APRIL as well as BAFF, and BCMA in particular is what keeps the long-lived plasma cell alive in its niche.

This is why B-cell return after depletion is not simple repopulation, and why combining depletion with BAFF blockade behaves differently from either alone: the two interventions are acting on the same limited resource from opposite ends. It is also why the timing of a second dose matters more than dose size in many models.

Three killing mechanisms, three different failure modes

A bound depleting antibody can do three things, and they saturate for different reasons. It can engage CD16 on NK1.1⁺ cells for antibody-dependent cellular cytotoxicity. It can drive phagocytosis by CD11b⁺ macrophages. And it can fix C1q, running C3 and the C5b-9 membrane attack complex.

ADCC runs out when effector cells are exhausted or when their Fc receptors are already occupied by circulating IgG — which is why depletion is less efficient in a hypergammaglobulinaemic animal. Phagocytosis is held back by CD47 on the target cell signalling into SIRPα on the macrophage; that brake is why CD47 blockade deepens depletion without touching the depleting antibody at all. Complement is consumed locally and depletes, so CDC plateaus in tissue even while it looks efficient in a plate assay.

In most in vivo settings macrophage phagocytosis in liver and spleen turns out to be the dominant mechanism, not ADCC and not complement — which is the opposite of the intuition most people carry over from cell-based assays. That single discrepancy is worth testing directly rather than assuming, and it is cheap to test.

Type I and type II anti-CD20 are not interchangeable

Two anti-CD20 antibodies binding overlapping epitopes can depend on entirely different mechanisms. Type I agents (the rituximab class) redistribute CD20 into lipid rafts, which clusters Fc regions and makes them exceptionally good at fixing C1q. Type II agents (the obinutuzumab class) do not redistribute CD20, fix complement poorly, and instead drive strong homotypic adhesion and direct, non-apoptotic cell death, with more efficient phagocytosis.

The practical consequence is that a complement-deficient animal, or one in which local C3 has been consumed, will show a much larger drop in efficacy with a type I agent than with a type II one. If you are comparing depletion across strains or across disease states, the antibody format is a variable, not a constant.

The differentiation programme underneath

Antigen through the BCR engages SYK and then BTK — the node BTK inhibitors act on — while CD22 restrains the same signal. CD4⁺ T follicular helper cells supply IL-21, and IL-6 from stroma completes the push towards plasma-cell differentiation.

Inside the cell the order is fixed and worth memorising. IRF4 at low concentration sustains the B-cell fate; at high concentration it forces commitment and induces Blimp-1. Blimp-1 shuts the B-cell programme down and licenses XBP1, which builds the secretory apparatus. Everything downstream of Blimp-1 is committed. That commitment is exactly why plasma cells are hard to remove: they are no longer running the programme your depleting antibody was designed against.

Blocking the signal rather than deleting the cell is the alternative strategy, and it targets a different point. A BTK inhibitor leaves the compartment intact and silences it. A depleting antibody removes the compartment and leaves the plasma cells. Neither does both.

What the readout is actually telling you

IgM falls quickly after depletion because it comes largely from short-lived cells that are continuously replaced from the pool you just removed. IgG does not, because long-lived plasma cells keep making it for months. A serological panel that shows IgM down and IgG flat is not a partial success — it is the signature of a complete, correctly working CD20 depletion.

For an autoimmune model the endpoint that matters is usually the specific autoantibody rather than the total. Anti-dsDNA IgG in a lupus model behaves like total IgG, not like IgM, and a depletion strategy that does not move it has not addressed the pathology however impressive the flow plots look.

Four ways this gets measured wrong

Counting circulating B cells only. Blood clears first and clears most completely. Spleen, lymph node, peritoneum and the tissue of interest all clear less efficiently, and marginal-zone and B-1 cells are notoriously resistant. A clean blood plot is compatible with substantial tissue persistence.

Staining with the depleting clone. If the depleting antibody is still bound, the detection antibody for the same epitope cannot bind, and a partially depleted animal reads as fully depleted. Use a different clone, a different epitope, or a lineage marker the treatment does not target.

Assuming the mechanism. ADCC, phagocytosis and complement are separable in vivo but only if you separate them. Removing one effector arm at a time is the only way to know which one your depletion depends on, and it is the control set most often skipped.

Reading antigen loss as depletion. Anti-CD20 can strip CD20–antibody complexes from the surface by trogocytosis, leaving a live, CD20-negative B cell that no longer stains and is no longer a target. It looks identical to depletion on a plot and is the opposite of it.

Where the surviving cells come from

Repopulation after depletion is not a re-run of the original repertoire. Two sources contribute, and they behave differently. Transitional cells emerging from the bone marrow bring a naive, largely unselected repertoire; residual memory cells that escaped depletion in tissue bring back exactly the specificities you were trying to remove. Which source dominates decides whether depletion is curative or merely a pause.

This is where the survival axis and the depletion axis meet. A high free-BAFF environment after depletion preferentially rescues cells that would normally have been deleted at the transitional checkpoint, because BAFF availability is that checkpoint. The clinically observed pattern of autoimmunity that returns worse after B-cell reconstitution has this as its mechanistic candidate, and it is testable in a model by measuring free BAFF against the returning repertoire rather than against the cell count.

Which reagent answers which question

This map has an unusually deep In Vivo layer, and the useful way to think about it is as two separate control sets: one that varies the depth of depletion, and one that varies the mechanism. Most published models run the first and skip the second.

A note on reagent specificity

The NIK / MAP3K14 antibody linked from this map is validated for rat reactivity only. Anti-CD47 and anti-BAFF-R are biosimilar-format research antibodies rather than functional-grade In Vivo reagents, and are linked here because they are the right tools for detection rather than for blockade. All three are flagged in the tooltips on the interactive version.

For research use only. Not for use in diagnostic or therapeutic procedures.