How complement pathways generate C3b, anaphylatoxins and C5b–9; why endothelial injury is tissue-specific and what positive, negative and retracted clinical findings show.
Complement is often drawn as a sequence of letters and arrows ending in a hole in a membrane. That diagram explains only part of its work. The cascade labels surfaces for phagocytosis, helps remove immune complexes and cellular debris, directs inflammatory cells, and links innate recognition to adaptive immunity. Injury develops when amplification extends beyond the surface and time window in which the response is useful. The same blood C3 concentration can therefore accompany very different complement activity in a renal glomerulus, a vessel wall, or the eye.
The liver produces most circulating components, but endothelial, renal and immune cells can also produce individual components locally. The plasma cascade, reactions on cell surfaces and intracellular processes cannot automatically be combined into a single mechanism. Three outputs are particularly important clinically: opsonization by C3 fragments, signaling by the anaphylatoxins C3a/C5a, and formation of the terminal C5b–9 complex. Each has distinct receptors, regulators and treatment implications.
The classical pathway starts with the C1 complex, containing recognition protein C1q and serine proteases C1r/C1s. The spatial arrangement of surface-bound IgM or IgG allows activation; a free antibody in solution is not equivalent to that surface. C1q can also interact with certain damaged-cell structures and innate recognition molecules, so “classical” does not mean exclusively antibody-dependent. C1s cleaves C4 and C2, allowing assembly of surface C3 convertase C4b2a. This article uses the widespread historical nomenclature in which the large catalytic C2 fragment is C2a; another coherent naming system calls it C2b. The changed designation does not identify a different enzyme.
The lectin pathway recognizes suitable arrangements of carbohydrate and acetylated groups through mannose-binding lectin, MBL, ficolins and collectins. Their associated MASPs are serine proteases, not zinc-dependent metalloproteases. Calcium matters for the organization of certain recognition complexes and interactions, but does not change the catalytic enzyme class. MASP-1 contributes to initiation and MASP-2 activation; MASP-2 cleaves C4 and C2. Studies of recognition-complex juxtaposition and MASP kinetics show that activation depends on spatial organization on a ligand-bearing surface, rather than the incidental presence of a lectin in plasma.
MASP-3 links this system to maturation of alternative-pathway factor D. In experiments with resting human blood, active MASP-3 mediated conversion of pro-factor D to the mature protease. This connects pathways through preparation of an enzyme; it does not mean that every alternative activation event needs a new lectin trigger. Nor can findings in resting blood be transferred without qualification to a clotting sample or severe systemic inflammation.
The alternative pathway is sustained by background changes in C3 and is especially important as an amplification loop for an established reaction. Hydrolysis of the internal thioester produces C3(H2O), retaining the peptide backbone and C3a domain: this is different from proteolytic production of C3a and C3b. C3(H2O) binds factor B; factor D cleaves bound B into Ba and Bb, generating fluid-phase convertase C3(H2O)Bb. Alternative-convertase interactions depend on Mg2+. Primary biochemical measurements of hydrolysis rates varied with the medium and other proteins; they cannot establish a universal rule that “1–2% of all C3 in the body turns over every hour.”
The protein structure shields the reactive thioester in native C3. Cleavage by a C3 convertase releases C3a, while the remaining C3b changes conformation. Structural studies by Janssen and colleagues showed a major displacement of the thioester-containing domain and exposure of interaction sites for factor B and regulators. The short-lived activated state permits covalent attachment to hydroxyl or amino groups on a neighboring surface. Competing hydrolysis by water limits the distance over which attachment can occur; already hydrolyzed fluid-phase C3b is not an intact new covalent label.
Surface-bound C3b recruits factor B. Factor D cleaves B within the appropriate complex, and Bb remains as the catalytic subunit of surface C3 convertase C3bBb. C3b is the platform, not the protease. Properdin stabilizes the convertase and participates in surface recognition under some conditions; it should not be presented as an obligatory universal initiating sensor. New C3b molecules create additional platforms, so classical or lectin initiation can acquire substantial alternative amplification. Its contribution depends on the activator, sample and measured output; a fixed percentage for the whole organism has not been established.
An increased density of C3b enables complexes with C5-convertase activity. They are commonly written C4b2a3b and C3bBbC3b, but these formulas abbreviate the spatial organization of enzyme and substrate on a surface. C5 cleavage yields soluble signaling peptide C5a and C5b, which starts terminal-complex assembly. C5 and C3 inhibition therefore differ fundamentally: the former leaves C3b formation and opsonization intact, whereas the latter acts earlier and more broadly.
C3b and its products do more than amplify the cascade. Factor I converts C3b to iC3b with suitable cofactors; iC3b no longer assembles an active alternative convertase but retains opsonic activity. CR1 binds C3b/C4b and participates in immune-complex processing, whereas CR3 and CR4 recognize iC3b and assist phagocytes in capturing labeled objects. Subsequent C3dg/C3d generation maintains a link to adaptive immunity through CR2. Proteolytic shutdown of a convertase therefore does not instantly remove every immunological consequence of the label.
Factor H recognizes a combination of C3b and host-surface features, including sialylated glycans and glycosaminoglycans. It competes with factor B, accelerates alternative-convertase decay and acts as a factor I cofactor. Structural studies of H complexes with C3b and sialylated sugars explain why changes in recognition domains may impair protection of a particular surface despite the protein remaining present in plasma. H concentration alone is not enough: it must bind where C3b has appeared.
Factor I is a serine protease with restricted activity unless the appropriate substrate–cofactor complex is present. Cell-surface CD46 supports C3b/C4b cleavage as a cofactor; CD55 accelerates convertase decay; CD59 limits terminal assembly and C9 incorporation into the membrane complex. CD55 and CD59 have GPI anchors, whereas CD46 is transmembrane. C1 inhibitor, a serpin, acts earlier on active C1r/C1s and certain MASPs; C4-binding protein regulates C4b-containing complexes. These mechanisms are not interchangeable, and an increase in one marker does not guarantee preservation of all other levels.
Convertase decay and proteolysis of its platform are distinct reactions. CD55 can accelerate dissociation of the catalytic subunit, whereas factor I with a cofactor irreversibly changes C3b. Terminal regulators act after C5b formation. Host injury can consequently originate from defective surface recognition, inadequate cofactor activity, convertase persistence or loss of membrane protection, even when the eventual laboratory profiles partly overlap.
Human C3a and C5a contain 77 and 74 amino-acid residues, respectively. Disulfide bonds stabilize their compact structures, and the C-terminal region is important for receptor activation. However, C3a and C5a are not interchangeable chemoattractants. C5a is particularly important for neutrophil and monocyte recruitment and activation; C3a responses depend more strongly on cell type, state and receptor repertoire. An EC50 or affinity measured in one cellular system cannot be declared a universal threshold in human tissues.
Plasma carboxypeptidase N and membrane carboxypeptidase M remove the C-terminal arginine to form desArg derivatives. These are zinc-containing carboxypeptidases, unlike the serine proteases of the convertases. C3a-desArg substantially loses canonical C3aR activity; C5a-desArg retains context-dependent residual activity and receptor interactions. Processing should not be reduced to “both peptides are completely switched off.” Similarly, disputed metabolic effects of C3a-desArg should not be described as an unquestioned universal axis through C5aR2.
C3aR and C5aR1 are seven-transmembrane receptors. Important C5aR1 mechanisms include Gi coupling, release of signaling Gβγ subunits, and activation of PI3Kγ and PLCβ. PI3K generates membrane PIP3, helping organize a migrating phagocyte's leading edge; PLCβ cleaves PIP2 into IP3 and DAG. IP3 promotes Ca2+ release from the endoplasmic reticulum, DAG participates in PKC activation, and Rac/Cdc42 connect these signals to actin remodeling. This is a regulated network of parallel processes, rather than a single linear “calcium–movement” sequence.
C5aR2, formerly C5L2, differs from C5aR1: it lacks typical canonical G-protein coupling but can recruit β-arrestins and alter signaling responses. Primary studies show why “passive decoy” is inadequate. Nevertheless, the direction of its effect depends on the cell and inflammatory context; C5aR2 cannot be assumed always anti-inflammatory or always injury-promoting. Receptor presence, ligand binding and the clinical effect of blockade are three separate evidence levels.
Chemotaxis requires a spatial gradient and a cell able to resolve it. Local C5a generation helps direct a phagocyte to a focus, whereas prolonged systemic stimulation can accompany desensitization, receptor internalization and impaired directional movement. Neutrophil dysfunction in sepsis is multifactorial, involving receptor signaling, metabolism and other mediators. The disappearance of surface C5aR1 does not explain all immune paralysis, and a C5a concentration does not predict an inevitable outcome.
Neutrophil activation and priming can facilitate degranulation and the respiratory burst. NOX2 transfers electrons from NADPH through FAD and heme centers to oxygen: O2 becomes superoxide and subsequently H2O2. Myeloperoxidase uses H2O2, chloride and a proton to produce hypochlorous acid: H2O2 + Cl− + H+ → HOCl + H2O. These reactions are useful in the appropriate antimicrobial compartment, but extracellular release of proteases and oxidants can damage endothelium and matrix. Neutrophil extracellular trap formation depends on the stimulus and conditions; C5a does not automatically activate one universal NETosis program.
Anaphylatoxins can contribute to degranulation of responsive mast-cell and basophil populations. Release of stored histamine must be distinguished from new synthesis of lipid mediators, including leukotrienes and prostaglandins. C3aR/C5aR1 expression and responses differ among cell populations; the term “anaphylatoxin” does not make all anaphylaxis complement-dependent. The relationship between receptor signals and clinical syndromes is explored further in the article on the mast-cell activation mediator cascade.
Increased vascular permeability can involve mediators from other cells and direct endothelial effects. Shivshankar and colleagues' study in human umbilical-vein endothelial cells identified regulated C3a/C5a receptor expression, altered migration and effects on interaction with lymphoid cells. It does not demonstrate an identical response in every vascular bed. In another primary study, Sartain and colleagues found lower alternative activation and more pronounced regulatory properties in brain microvascular endothelial cells than in glomerular cells. This difference helps explain tissue-selective injury, but a cellular model does not measure an individual's disease risk.
Inflamed endothelium changes leukocyte adhesion, barrier contacts and the balance between procoagulant and anticoagulant signals. Complement interacts with cytokines, platelets and coagulation, while heme released during intravascular hemolysis can further sustain activation. The result is a mutually amplifying loop, rather than a strictly one-way chain. Interactions with inflammasomes are also model-dependent; NLRP3 and IL-1β production describe another inflammatory regulatory level that cannot be equated with C3 cleavage itself.
After C5 cleavage, unstable C5b binds C6, followed by C7, C8 and C9 molecules. Sequential assembly enables membrane insertion of MACPF-domain proteins and pore formation. Cryo-electron microscopy reveals a complex asymmetric architecture; a teaching diagram showing identical cylinders in a circle is schematic, not an exact atomic model. Neither C9 number nor complex geometry should be assigned a single fixed value for every biological membrane.
Surface susceptibility matters more than C5b–9 presence alone. Terminal complement is especially important against certain Gram-negative bacteria; a thick wall and other surface features may prevent effective membrane attack. Nucleated cells can remove or internalize damaged membrane regions and tolerate sublytic exposure. Such exposure alters ion fluxes and signaling pathways and may sustain inflammation without immediate osmotic lysis. An erythrocyte lacking regulatory protection and an endothelial cell with repair mechanisms consequently respond differently.
Soluble sC5b–9 associates with proteins limiting membrane insertion, including vitronectin and clusterin. An increase indicates terminal activation in the relevant sample, rather than directly counting pores in a kidney or vessel wall. Membrane and soluble terminal complexes are not synonyms. Likewise, an absence of massive lysis does not exclude clinically meaningful complement-dependent injury.
In paroxysmal nocturnal hemoglobinuria, PNH, a somatic PIGA defect in a hematopoietic clone impairs GPI-anchor synthesis. Loss of CD55/CD59 makes cells particularly vulnerable to complement. This is an acquired clonal disease, rather than an ordinary dietary deficiency or simply “high C3.” C5 inhibition reduces intravascular hemolysis, but persistent C3-dependent erythrocyte uptake by phagocytes can sustain extravascular anemia. Bone-marrow status and other causes also affect hemoglobin; full normalization is therefore not guaranteed by one selected target.
In complement-mediated thrombotic microangiopathy, traditionally discussed in the setting of atypical hemolytic uremic syndrome, aHUS, endothelial injury involves impaired cascade control. CFH, CFI, CD46, C3 or CFB variants and factor H autoantibodies may contribute, but a genetic factor is not always sufficient without additional triggers. Thrombotic thrombocytopenic purpura with severe ADAMTS13 deficiency, infection-associated HUS and secondary microangiopathies require differential assessment. Not every microangiopathy establishes a need for the same complement therapy, and normal C3 does not exclude local dysregulation.
C3 glomerulopathy is defined by morphology together with predominant C3 deposition, not a single low C3 concentration. Nephritic factors may stabilize convertases, and regulation may be impaired genetically or acquired; convertase-activity studies show why a functional mechanism should be distinguished from antibody presence alone. In age-related macular degeneration, CFH variants are associated with risk, but genotype is not a ready-made individual prognosis. During ischemia–reperfusion, damaged surfaces can activate complement recognition and amplification; a biologically plausible target nevertheless does not guarantee successful pharmacological intervention.
TRIUMPH was a double-blind randomized placebo-controlled trial in 87 patients with PNH. Intravenous eculizumab was given at 600 mg weekly for four weeks, then 900 mg one week later and 900 mg every two weeks through week 26. The primary outcomes were hemoglobin stabilization and transfused red-cell units. Stabilization without transfusion occurred in 21 of 43 active-group participants, 49%, versus 0 of 44; median transfused units were 0 versus 10. Median LDH area under the time curve was 85.8% lower. This is an intravascular-hemolysis marker result, not an “85.8% mortality reduction”; the trial does not justify attributing an exact percentage of prevented thromboses to it.
PEGASUS was an open-label randomized phase 3 trial in 80 adults with PNH whose hemoglobin remained below 10.5 g/dL despite eculizumab. Following four weeks of combined run-in treatment, 41 participants switched to subcutaneous pegcetacoplan, 1080 mg twice weekly, and 39 continued intravenous eculizumab on their existing regimen. The primary outcome was hemoglobin change at week 16. The adjusted between-group difference was 3.84 g/dL in favor of pegcetacoplan, not each patient's mean increase. Transfusion avoidance was 85% versus 15%. However, noninferiority for LDH change was not demonstrated; the open-label design, selected residual-anemia population and short comparative period limit generalization. Injection-site reactions and breakthrough hemolysis also remain part of the benefit–risk assessment.
The negative APEX-AMI result is particularly relevant to the idea of “suppressing complement after any injury.” In this double-blind placebo-controlled trial, 5745 patients with ST-elevation myocardial infarction undergoing primary coronary intervention received pexelizumab or placebo. The anti-C5 agent was administered as a 2 mg/kg bolus before intervention, followed by 0.05 mg/kg/hour for 24 hours. Primary 30-day mortality was 4.06% versus 3.92%; hazard ratio 1.04, 95% confidence interval 0.80–1.35, p=0.78. A mortality benefit was not established. This does not disprove complement's role in ischemic injury, but it refutes an automatic progression from that mechanism to effective treatment in this population and regimen.
Avacopan blocks C5aR1 rather than C5 cleavage, potentially modifying anaphylatoxin signaling while preserving C5b–9 formation. Clinical evidence must nevertheless be assessed separately from the target's appeal. ADVOCATE randomized 331 patients with ANCA-associated vasculitis to oral avacopan 30 mg twice daily or a prednisone taper; all also received rituximab or cyclophosphamide followed by standard treatment. The primary outcomes were remission at week 26 and sustained remission at week 52. On June 29, 2026, NEJM retracted the publication: primary-endpoint assessments in nine patients had been readjudicated after database lock and unblinding without disclosure in the article. Its previously reported percentages cannot be presented as reliable efficacy confirmation.
On April 27, 2026, FDA's CDER proposed withdrawing Tavneos approval, stating that efficacy for the approved use had not been validly demonstrated and the application contained material untrue statements. FDA also reported serious drug-induced liver injury, including vanishing bile duct syndrome and fatal cases. A proposed withdrawal and a final regulatory decision are distinct events; this announcement alone does not establish completion of the procedure in every country. It also explicitly states that the drug does not eliminate glucocorticoid use. A universal promise of “steroid-free remission” is therefore scientifically and clinically incorrect here.
Choosing C3, C5, C5aR1, factor B or factor D blockade changes the protective functions left intact. Serum models after meningococcal vaccination show that consequences depend on the target, antibody response and studied organism; they do not guarantee infection safety in practice. Human factor D deficiency is also associated with serious infections, so preservation of part of the classical pathway does not imply absence of risk. Infection prevention and monitoring depend on the specific drug and current clinical requirements, rather than a general notion of inhibitor “selectivity.”
C3 and C4 measure component concentrations, not the entire functional system. Low values may reflect consumption, synthesis or deficiency, while normal values can coexist with local activation. CH50 assesses functional integrity of the classical and common terminal cascade; AH50/AP50 assesses the alternative and common terminal cascade. Isolated low CH50 suggests consideration of early classical components; isolated low AH50, alternative components; low values in both direct attention to shared components, consumption or sample quality. This guides further investigation, rather than independently making a diagnosis.
Activation products such as C3 fragments and sC5b–9 answer a different question: whether a reaction has left a measurable trace in the sampled compartment. Tissue deposition, circulating products and convertase activity can diverge. C3 glomerulopathy and microangiopathy diagnosis therefore integrates clinical, morphological, hematologic and functional findings. A genetic variant needs classification and comparison with the phenotype; a variant of uncertain significance is not an automatic indication for lifelong inhibition.
Preanalytics matter particularly because complement may activate after blood collection. EDTA plasma is often used for activation products to chelate Ca2+/Mg2+ and reduce ongoing cascade activity; functional assays have the opposite aim of preserving reaction capacity and usually require appropriate serum. Studies by Mollnes and subsequent stability investigations show dependence on temperature, delay and anticoagulant. A universal rule that “all complement tests need EDTA” is incorrect. Handling and storage must match the laboratory's validated method, and an elevated result in a poorly prepared sample must not become a diagnosis.
Rosmarinic acid illustrates the distinction between an early paper's title and an established mechanism. A 1988 study attributed reduced hemolysis to C3-convertase inhibition. In Sahu, Rawal and Pangburn's targeted 1999 study with purified components, half-maximal inhibition of covalent C3b attachment occurred at 34 µM. Corresponding serum hemolytic-test values were 180 µM for the classical and 160 µM for the alternative pathway. Significant C3-convertase enzyme inhibition was not found even at 10,000 µM. Reduced final hemolysis therefore does not prove direct blockade of the convertase catalytic site: interference with C3b attachment can produce a similar assay output.
These conditions do not establish an oral human dose. A clinical recommendation would require evidence of free exposure to parent compound and metabolites, distribution into the relevant compartment, target engagement and improvement in a meaningful outcome. These biochemical studies do not supply that chain. Claims of “selective factor B blockade at 15 µM” and a ready-made rosmarinic-acid protocol to prevent aHUS or PNH consequently do not follow from them.
For curcumin, quercetin, luteolin, omega-3 and specialized pro-resolving mediators, possible effects on cytokines, lipid metabolism and cell membranes are not equivalent to clinically proven restoration of CD55/CD59 or selective complement control. A reduced inflammatory marker does not identify the altered cascade level or demonstrate preserved antimicrobial function. There is insufficient basis to turn a receptor hypothesis or experimental concentration into universal supplement regimens, promise thrombosis prevention or replace treatment of complement-mediated disease. Nutrition and general vascular-risk prevention matter, but address a different evidence question.
Complement cannot be assessed by a “less is better” principle. Functional loss impairs microbial and immune-complex clearance; uncontrolled reactions injure host tissues. The appropriate question includes the activation site, regulatory defect, specific effector output and clinical outcome that can actually be modified. Answers differ for PNH, endothelial microangiopathy and myocardial infarction, as both positive and negative trials show.
An evidence-based sequence is to establish the disease phenotype, distinguish component quantity from function and activation, identify the likely mechanism, and choose a target with demonstrated benefit in the relevant population. Then assess the remaining protective functions, infection risks and other treatment limitations. Neither low C3, high sC5b–9 nor an attractive molecular diagram replaces that sequence. Distinguishing mechanism from demonstrated outcome makes complement a useful clinical concept rather than a reason for universal immune suppression.
They converge on C3-convertase formation and subsequent effector outputs. The alternative pathway can also amplify a response initiated through the classical or lectin pathway.
No. C5a is particularly important for neutrophil and monocyte recruitment, while C3a responses depend more on cell and receptor context. Their desArg derivatives cannot both be treated as equally completely inactive.
No. Component concentration, cascade function and local activation answer different questions. Diagnosis needs clinical findings and additional laboratory evidence.
C5 blockade limits C5a and terminal-complex generation while preserving C3b formation. C3 blockade acts earlier and also affects opsonization, changing both benefit and remaining protective functions.
No. Positive TRIUMPH and PEGASUS findings concern selected PNH populations. In APEX-AMI, anti-C5 treatment did not reduce post-infarction mortality in the studied regimen.
NEJM retracted the publication in 2026 over reassessment of primary outcomes. Its former percentages cannot be presented as reliable efficacy confirmation; FDA also reported serious regulatory and liver-safety concerns.
No. Activation-product assays seek to reduce reactions after collection; functional tests need preserved cascade capacity. Sample type and handling depend on the laboratory method.
The presented evidence does not support that conclusion. A biochemical concentration or inflammatory-marker change does not establish clinical exposure, target engagement and improved outcomes. No universal supplement protocol follows.