One Ligand, Two Receptors, Opposite Jobs: TNFR1 vs TNFR2

Blocking TNF cures rheumatoid arthritis in some patients and triggers demyelination in others. That is not a dosing problem, and it is not a patient-selection problem. It is the direct consequence of one structural fact: TNF has two receptors that do opposite things. TNFR1 is on nearly every cell, carries a cytoplasmic death domain, and drives inflammation and apoptosis. TNFR2 is restricted to regulatory T cells, myeloid cells and neurons, has no death domain, and drives survival, suppression and repair. Every anti-TNF antibody in clinical use blocks both. Once you sort the whole superfamily by that one feature, nineteen ligands and twenty-nine receptors stop looking redundant and start looking like a design. Click any protein for the matching Assay Genie ELISA kit or In Vivo antibody.

The TNF receptor superfamily sorted by function rather than by name: the ligand layer, the death-domain receptors against the survival-only receptors, the costimulatory arm now being developed as agonists, the complex I signalling machinery, and the inflammatory output. Open the interactive version to click any protein for its role and the matching validated reagent.

The death domain is the dividing line

The superfamily is usually taught as a list, which is why it is usually forgotten. Sort the receptors by a single structural feature instead and the function falls out. Receptors with a cytoplasmic death domain — TNFR1, Fas and the TRAIL receptors — can recruit TRADD and caspase-8, and can therefore kill the cell. Receptors without one — TNFR2, CD40, 4-1BB, OX40, CD27, RANK, BAFF-R, GITR — can only recruit TRAF adaptors, and can therefore only signal survival, proliferation and NF-κB.

Nothing else about the family predicts function as cleanly. Not ligand identity, not tissue distribution, not nomenclature. So when a paper describes a TNFRSF member as pro-apoptotic or pro-survival, the domain is the thing to check before the mechanism is believed — and it is also the thing that tells you whether an agonist or an antagonist is the sensible intervention.

TNFR1 versus TNFR2 is the whole clinical story

TNFR1 mediates the fever, the acute-phase response through IL-6 and CRP, the IL-1β amplification loop, and the joint destruction. That is the arm anti-TNF is aimed at, and against that arm it works extremely well. TNFR2 is a different receptor with a different job. It is expressed most highly on Tregs, where it is required for their expansion and suppressive function — TNFR2-high Tregs are the most suppressive subset — and it drives IL-10 production. It is also expressed on oligodendrocyte precursors, where it drives remyelination.

Non-selective TNF blockade therefore removes a brake at the same moment it removes an accelerator. That is the accepted explanation for the most instructive failure in the field: TNF blockade helps rheumatoid arthritis and worsens multiple sclerosis. It is also the likely explanation for paradoxical psoriasis on anti-TNF therapy. And it is why TNFR2-selective agonists are now in development — the opposite intervention to the one that made the class famous, aimed at the same molecule.

The practical consequence for modelling: blocking the ligand, blocking TNFR1 selectively and blocking TNFR2 selectively are three different experiments with three different answers. If a study reports "TNF blockade" and means ligand neutralisation, it has not told you which receptor arm the phenotype depended on. Mouse-specific reagents exist for all three — anti-TNF, anti-CD120a and anti-CD120b — so the discrimination is available.

Signalling is a race between two complexes

TNF binding TNFR1 does not immediately kill anything. It first assembles complex I at the membrane: TRADD, TRAF2, RIPK1 and cIAP2. cIAP2 ubiquitinates RIPK1, which routes the signal to NF-κB and produces survival genes plus the inflammatory output. Only if that complex fails to form — or if the ubiquitination is stripped — does RIPK1 disengage from the membrane and assemble the cytosolic complex II with caspase-8.

The default outcome of TNF is therefore survival, and apoptosis is what happens when the survival arm is disabled. This is exactly why TNF alone kills very few cells in culture but TNF plus cycloheximide kills readily, and it is why IAP antagonists convert TNF from a growth signal into a death signal without touching TNF itself. cIAP2 is the switch. If you are trying to make TNF cytotoxic, that is the node to move.

The same architecture, aimed elsewhere

Bone. RANKL acting on RANK drives osteoclast differentiation, and osteoprotegerin is a soluble decoy receptor that soaks up RANKL before it arrives. Resorption is set by the RANKL:OPG ratio, not by RANKL alone, which is the single most common measurement error in bone work. Denosumab is essentially a synthetic version of the decoy — a drug that copies an endogenous regulator rather than inventing a new mechanism.

B cells. BAFF and APRIL are survival factors, and BAFF-R is the receptor that makes a transitional B cell viable. Belimumab blocks BAFF; the reason it is a slow drug clinically is that it removes a survival signal rather than killing anything.

Costimulation. CD40, 4-1BB, OX40, CD27 and GITR are the arm being developed as agonists, and they converge on CD25 and effector output. The hard problem here is mechanical: agonism requires receptor clustering, not merely occupancy. That makes agonist antibodies exquisitely sensitive to isotype, Fc γR engagement and epitope in a way blocking antibodies simply are not, and it is why anti-human CD40 clones behave so differently from one another. If you are screening an agonist, the Fc format is part of the pharmacology, not a formulation detail.

Reading the family at a glance

The In Vivo angle — and one assay warning

Fourteen of the 37 nodes on this map carry a functional-grade In Vivo antibody, which makes this the deepest blockable layer in the library after the T-cell engager map. Unusually, it covers both sides of the TNFR1/TNFR2 question rather than only the popular one: TNF itself, TNFR1 (CD120a) and TNFR2 (CD120b), plus RANKL, CD70, CD40 in mouse and human, 4-1BB, OX40, ICOS, CD25, GM-CSF, IL-10 and ICAM-1. That is enough to separate the two receptor arms in vivo, which is the experiment this map exists to make obvious.

One warning that matters more here than almost anywhere else: endotoxin. LPS induces TNF directly and potently. A blocking antibody carrying residual endotoxin therefore raises the very cytokine you are administering it to remove, and the result is a null or paradoxical experiment with no obvious cause. Low and ultra-low endotoxin grades are not a nicety on this pathway; they are the difference between a readable result and a wasted cohort. The measurable side is well covered too — human TNF-α, TRAIL, FasL, LIGHT, TWEAK, OPG, BAFF-R, IFN-γ and CRP are all quantitative analytes.

The one-line version

Sort the receptors by death domain and the family becomes predictable. TNFR1 inflames and kills, TNFR2 suppresses and repairs, every clinical anti-TNF blocks both, and the survival complex forms first — so apoptosis is what happens when survival fails, not what TNF does by default.

For the cytokine-neutralisation view of the same output layer, see the cytokine neutralisation network; for the costimulatory arm in detail, see T-cell co-stimulation and agonists; and for the BAFF–APRIL branch, see B-cell depletion and the BAFF–APRIL axis.

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