Complement: The Brake Matters More Than the Trigger

Complement is firing on your own cells right now. Not in response to anything — C3 hydrolyses spontaneously in plasma at a low, constant rate, and every hydrolysed molecule is a potential convertase looking for a surface. The alternative pathway has no off switch and no trigger to wait for. What keeps your erythrocytes and endothelium intact is not that complement fails to land on them; it is that Factor H, CD55 and CD59 take it off again faster than it accumulates. Read that sentence twice, because almost every complement disease in humans is a failure of the second layer rather than an over-active first one.

The complement cascade: three initiation routes, one amplification hub, and the regulator layer that decides self from non-self. Open the interactive version to click any protein for its role and the matching validated reagent.

Three ways in, and only one of them needs an immune system

The classical route is the one named after antibody, and it is the fussiest. C1q has six globular heads and needs several of them engaged at once to activate, which is why a single IgG molecule on a surface does nothing and a densely clustered patch does a great deal. A single IgM pentamer, already carrying the geometry, is enough on its own and is roughly a thousand-fold more efficient per molecule. This is a recognition-density problem, not an affinity problem, and it explains why epitope copy number on a target cell often predicts complement-dependent cytotoxicity better than the antibody's K_D does. Once C1q is engaged, C1r autoactivates and cleaves C1s, and C1s is the protease that does the actual work on C4 and C2. C1q also binds apoptotic cells and immune complexes directly with no antibody involved at all, which is why C1q deficiency causes lupus rather than immunodeficiency: the classical pathway is a waste-disposal system before it is a weapon.

The lectin route reads sugar instead of antibody. MBL recognises the terminal mannose and N-acetylglucosamine patterns displayed by microbial surfaces and not by mature vertebrate glycoproteins, and it activates MASP-1 and MASP-2. MASP-2 then does exactly what C1s does — same two substrates, same convertase — and MASP-1 amplifies MASP-2 as well as feeding the alternative pathway directly. Two entirely different sensors converging on one enzymatic step is a recurring design in innate immunity, and it means a C4/C2 readout cannot by itself tell you which route fired.

The alternative route needs nothing. Tick-over C3 lands wherever it lands; Factor D — which circulates already active and at vanishingly low concentration, making it rate-limiting — cleaves surface-bound Factor B to build C3bBb; and properdin stabilises the result. Properdin is worth pausing on: it is the only positive regulator in the entire system, extending the convertase half-life roughly ten-fold. Everything else that binds a convertase takes it apart.

C3 is the hub, and amplification is the whole point

Both convertases converge on C3, and here the system stops behaving like a cascade and starts behaving like an explosion. C3b deposited on a surface can seed more alternative convertase on that same surface, so a handful of initiating events becomes thousands of C3b molecules within minutes. That exponential quality is the reason complement is fast enough to matter against a dividing bacterium, and it is also the reason the regulators have to be constitutive rather than induced. A brake that needed transcription would arrive several thousand C3b molecules too late.

Two outputs come off the hub. C3a diffuses away as an anaphylatoxin and signals through C3aR1 on mast cells, basophils and myeloid cells. iC3b — C3b that Factor I has already cut — stays put. It can no longer form a convertase, which sounds like inactivation and is why it is named that way, but it is read avidly by CR3 and is the dominant opsonin in most tissues. Calling iC3b "inactivated" is a historical accident that has misled a lot of people into ignoring it.

Why total C3 is the wrong thing to measure

Plasma C3 sits at around 1 mg/mL. Consuming even a biologically dramatic amount of it barely dents that number, so a total-C3 ELISA in an activated animal will often come back reassuringly normal. The fragments are what move: C3a, C5a, iC3b, C4d and C5b-9 all rise by orders of magnitude because they are near-absent at baseline. Measure the fragment, not the zymogen. The one caveat is pre-analytical: complement activates in the tube, so EDTA plasma collected on ice and spun promptly is not a nicety here, it is the difference between a real result and an artefact.

The regulators are the actual story in human disease

Run down the list of complement diseases and the pattern is almost monotonous. C1-inhibitor restrains C1r, C1s and the MASPs; lose it and you get hereditary angioedema. Factor H discriminates self from non-self by binding host sialic acid and glycosaminoglycans and accelerating decay of the alternative convertase specifically where those markers are; mutations in its C-terminal recognition domains cause atypical haemolytic uraemic syndrome, and the common Y402H variant is one of the strongest genetic risks for age-related macular degeneration. Factor I does the proteolytic cutting but is useless alone — it needs a cofactor, either soluble (Factor H, C4BP) or membrane-bound (CD46, CR1).

On the membrane, CD55 accelerates decay of both convertases and CD59 blocks C9 insertion into the forming pore. Both are GPI-anchored, and that shared anchor is exactly why paroxysmal nocturnal haemoglobinuria — a somatic PIGA defect in GPI anchoring, not a complement gene defect at all — presents as complement-mediated haemolysis. The cascade in those patients is entirely normal. The brake is simply missing from one clone of cells.

What this means for your model

If you are trying to reproduce complement-driven pathology, the productive question is usually not "how do I activate complement harder" but "what is failing to switch it off". Regulator knockouts, blocking the cofactor rather than the protease, and surfaces that lack sialic acid all get you further than adding more antibody. And when a model does not phenocopy the human disease, check species differences in the regulator layer first — mouse Crry has no direct human equivalent and covers ground that CD46 and CD55 divide between them in people.

Three outputs, and only one of them is lysis

C5 cleavage starts the terminal pathway: C6, C7 inserting the complex into the bilayer, C8 initiating the pore, and up to eighteen copies of C9 polymerising into the barrel of C5b-9. The membrane attack complex gets the diagrams and the attention, and for most nucleated cells it is the least important arm of the system. Sublytic MAC is survivable — cells shed it by vesiculation and endocytosis, and the surviving cell is often left activated rather than dead. Lysis matters enormously for erythrocytes and for Neisseria, and rather less for the tissue you are usually studying. Terminal-component deficiency in humans presents almost exclusively as recurrent meningococcal disease, which tells you how narrow that arm's indispensable role really is.

The two arms that do more work are opsonisation and chemotaxis. iC3b on a surface is read by CR3 (CD11b/CD18) on neutrophils and macrophages and by CRIg on Kupffer cells, which strips C3b-opsonised particles straight out of portal blood. CR1 on erythrocytes does something different again: it ferries immune complexes to the liver and spleen for transfer to resident macrophages without phagocytosing them itself, which is why primates offload immune complexes onto red cells and rodents do not. Meanwhile C5a through C5aR1 is one of the most potent neutrophil chemoattractants known, with C5aR2 acting as a non-signalling decoy that tunes the response.

Where to block, if you are blocking

Anti-C5 — the eculizumab position — is the classic intervention, and its logic is worth understanding: it stops the MAC and stops C5a while leaving C3-mediated opsonisation completely intact. In PNH that is precisely the trade you want, and it is also why anti-C5 patients can develop extravascular haemolysis of C3b-coated cells that the drug does nothing about. Blocking further upstream at C3 removes opsonisation too, with the infection risk that implies. There is no neutral place to intervene in this cascade; every choice trades one arm for another.

Reading the pathway: a practical panel

Why there are only two functional-grade antibodies on this map

It is worth saying plainly rather than papering over. Of the thirty-eight proteins on the diagram, thirty-six link to a validated ELISA and only two link to an In Vivo functional-grade antibody. That ratio reflects the biology, not a gap in the catalogue. Complement is a cascade of soluble plasma proteins present at milligram-per-millilitre concentrations, turning over continuously and replenished by the liver. Antibody blockade of a soluble protein at that abundance is a pharmacology problem, not a bench reagent problem — it is why the successful complement drugs are dosed like drugs. You measure this pathway; you do not usually block it with an antibody at the bench.

Where blockade is the right experiment, it belongs on the cells that read the output. Anti-CD11b removes CR3-dependent phagocytosis of iC3b-opsonised targets, which separates "complement was deposited" from "complement did something". Anti-Ly6G depletes the neutrophils C5a recruits, which separates chemotaxis from tissue damage. Those two experiments answer most of the questions people reach for a complement blocker to answer, and both use reagents that behave predictably in vivo. Both are low-endotoxin, functional-grade formats — which matters more than usual here, since endotoxin contamination activates complement directly and will confound the very readout you are running.

The short version

Three ignition routes, one amplification loop, and a regulator layer doing the discriminating. The classical route needs clustered antibody, the lectin route needs microbial sugar, and the alternative route needs nothing at all and never stops. C3 turns any of those into thousands of molecules within minutes, which is why the brakes must be constitutive. Lysis is the least important of the three outputs for most cells; opsonisation and C5a-driven chemotaxis do more. And when you go looking for a phenotype, check the regulators before you blame the trigger — the cascade is usually fine.

For the antibody-dependent side of opsonisation and killing, the Fc receptor map covers ADCC, ADCP and CDC side by side; for the depletion setting where C1q, C3 and C5b-9 do part of the work, see B-cell depletion and the BAFF–APRIL axis.

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