B. Paul Morgan


Complement comprises a set of proteins present in plasma and other biological fluids, and on cell membranes, that together play key homeostatic roles in combating infection and disposal of waste. Complement is a central pillar of innate immunity, a ready-to-go, fast-response system that efficiently targets pathogens and toxic waste.

The discovery of complement dates back to the late 19th century when a number of pioneering biologists, including Josef Fodor, George Nuttal, and Hans Buchner, were exploring the bactericidal activities of plasma and serum.1 They showed that fresh serum contained an activity that could efficiently kill some types of bacteria, and that this bactericidal activity decayed as the serum aged ex vivo and was rapidly lost when the serum was exposed to heat. To describe this heat-labile activity, Buchner coined the enigmatic term “alexin,” from the Greek and roughly translated as “without a name.” Jules Bordet extended these serum bactericidal studies using cholera bacilli and showed that sera from choleraimmune individuals efficiently killed the organism, whereas sera from nonimmune individuals did not; heating the immune serum caused loss of this bactericidal effect, suggesting it was related to Buchner’s alexin.2 Bordet then did a clever experiment: first incubating cholera bacilli with heat-treated immune serum (which alone did nothing), then adding fresh nonimmune serum; the bacilli were killed, demonstrating that killing of cholera bacilli required two serum components, a heat-stable component present only in immune serum which he termed “sensitiser,” and a heat-labile component present only in fresh serum, alexin. Around the same time, Paul Ehrlich was exploring how immune serum caused hemolysis of animal erythrocytes; he also found that these same two components were required. The heat-stable component present only in immune serum he termed “amboceptor” (and later immune body or antibody), while the heat labile alexin, he called “complement,” to indicate that it merely complemented the inherent hemolytic effect of amboceptor.3

Over the first few years of the 20th century, an intense debate continued regarding the nature of complement and its relationship to antibody, with Bordet and Ehrlich as the main protagonists. The “complement fixation test,” developed by Bordet and his coworkers around 1900 as a means of testing whether an individual possessed antibodies against a particular bacterium (ie, was immune), relied on the fact that complement was consumed when antibody bound its target and demonstrated conclusively that complement was a distinct activity in serum.4 This test, the mainstay of immune diagnostics for a century, is still used for some pathogens today. The famous Glasgow physician and pathologist Sir Robert Muir, writing in 1906, described complement as “that labile substance of normal serum that is taken up by the combination of an antigen and its antibody,”5 a definition it would be hard to improve on today.

Over the next 20 or so years, a number of scientists used the serum fractionation techniques that were state of the art at the time to investigate the “substance” called complement. Euglobulin precipitation (by dialysing serum against water) revealed that neither the re-dissolved euglobulin precipitate nor the dialyzed serum supernatant alone possessed complement hemolytic activity; however, when re-combined, complement activity was restored, demonstrating the need for at least two components, termed C′1 and C′2. The precipitable euglobulin C′1 component was inactivated by heating to 56°C, while the soluble C′2 component was heat-stable. By the mid-1920s, other manipulations of serum, including adsorption on yeast particles, incubation with ammonia or treatment with cobra venom had shown that there were at least four separable components necessary for complement activity, termed C′1, C′2, C′3, and C′4.6

The physicochemical nature of complement also attracted much interest and debate until 1941, when a landmark paper from Louis Pillemer and his colleagues showed clearly that each of the components C′1 to C′4 was protein in nature.7 Recognition of the protein nature of the complement components fueled a frenzy of protein chemistry activity that, by the mid-1960s, had further refined the components C′1 to C′4, in particular demonstrating that C′3 actually contained six separate proteins; the nine complement component proteins were therefore termed C′1, C′2, C′4, C′3a, C′3b, C′3c, C′3d, C′3e, and C′3f. The euglobulin, C′1, was also shown to be a complex of three different proteins. In 1968, a Committee on Complement Nomenclature met under the auspices of the World Health Organization to simplify and standardize, resulting in the modern terminology, in order of reaction, C1, C4, C2, C3, C5, C6, C7, C8, and C9.8

Up to the 1950s, complement research was focussed on the antibody-dependent activity recognised by Bordet and Ehrlich. Then, in 1954, Pillemer made a startling discovery: serum contained a protein that he called properdin (from the latin, perdere, which means to destroy) that could trigger complement attack on pathogens without the need for antibody.9 This finding was hailed as a “magic bullet” against infection and was so newsworthy that it was featured in Time magazine under the banner “Medicine: Death to Germs.”10
Others strongly denied the existence of the extraordinary properdin system, accusing Pillemer of carelessness or worse; perhaps in part due to this criticism, Pillemer committed suicide in 1958, leaving others in the 1960s to confirm and extend his findings. In 1971, Hans Muller-Eberhard and Manfred Mayer independently provided definitive proof of the existence of the properdin system, which they called the alternate pathway (now alternative) to distinguish from the original antibody-dependent classical pathway of activation.11,12 Identification of complement proteins unique to this pathway soon followed.

Up until the end of the 1950s, complement proteins were considered, with no good evidence, to be minor plasma components, present in trace amounts and hence difficult to work with. This misconception was laid bare in 1960 with the demonstration by Hans Muller-Eberhard and colleagues that the β1c-globulin band visible on serum electrophoretograms, was in fact C3, a major component accounting for 1% to 2% of total plasma proteins.13 Other complement proteins were soon shown to be relatively abundant in plasma, opening the door to their purification and functional characterization.

The next major leap forward was the recognition of the enzymatic nature of complement. Irwin Lepow and colleagues in the late 1950s had demonstrated that C′1 was associated with enzymatic (esterase) activity. By the mid-1960s, they had achieved a remarkable understanding of this first step of complement activation14; they showed that the enzyme comprised three distinct protein subunits (C′1q, C′1r, C′1s), that formation of the C′1 complex required calcium ions, that its substrates were C′4 and C′2 in that order, and that the enzyme was controlled by a plasma C′1 esterase inhibitor. It was soon recognized that a puzzling disease, hereditary angioedema, long associated with complement activation, was caused by a deficiency of this inhibitor,15 launching the field of complement therapeutics.

Around the same time, Muller-Eberhard and colleagues showed that activation of C′3 was also enzymatic in nature. They showed that a cell-bound complex of C′1-activated C′4 and C′2 could cause activation of many molecules of C3 and their deposition on the cell surface in an active form.16 It took another 15 years before the mechanism by which activated C′4 and C′3 attached to membranes was identified, when Brian Tack’s group showed that both these molecules possess a labile, buried thioester group, exposed on activation, that covalently attached the proteins to surfaces.17,18

One problem with starting a history is knowing when to stop. The brief history above is incomplete and ends 40 years ago, and much that is noteworthy has occurred since then! However, that more recent history will form part of later descriptions, so I will draw a line here. Before moving on, I will remark that the use of “apostrophes” in complement nomenclature was quietly dropped in the late 1970s, so henceforth they will not be featured.


In the 21st century, the term complement encompasses some 35 plasma and membrane proteins. The pathway components (Table 36.1) are outnumbered by regulatory proteins (Table 36.2) that limit activation in plasma and on self cells, and receptors (Table 36.3) that bind complement proteins to trigger a range of cellular responses. The complement proteins interact with one another to provide an effective and efficient antimicrobial defense system, along with a growing list of other roles, for example, in immune complex handling and priming for adaptive immune responses. Critical features include activation by diverse triggers, enzymatic amplification at multiple steps, and rigid control to prevent damage to self. Complement activation can be initiated in a variety of ways to target different pathogens and toxic agents. The literature describes three distinct activation pathways, although in reality, these pathways are closely interlinked. Antibodydependent activation, described by Ehrlich and others at the beginning of the 20th century, was for many years the only known pathway, now termed the classical pathway (CP). The re-discovery by Pillemer in the 1950s of an antibodyindependent, pathogen-triggered pathway for complement activation laid the foundations for the alternative pathway (AP). A third activation pathway, antibody independent and triggered by pathogen-specific sugars, was described in the 1980s; at first given a variety of names to reflect the activating sugars, it is now universally known as the lectin pathway (LP). These three activation systems share common components but also have pathway-specific ones.

The Classical Pathway

Antibody-dependent triggering of the CP begins with immunoglobulin (Ig)G or IgM antibody bound to its target antigen either on a pathogen or host cell membrane, or on an immune complex. IgM, a large, pentameric molecule, is the most efficient activator of complement; a single IgM molecule bound to antigen is, in theory, sufficient to generate a nidus for complement activation. Free IgM in plasma does not activate complement; however, binding to antigen through its five antigen binding domains generates major structural change affecting the whole molecule, causing it to transition from a planar to a staple conformation. These events expose complement binding sites in the constant (Fc) regions of each of the five subunits that initiate the process of complement activation. In contrast, multiple IgG molecules, bound close together on the target, termed an immune array, are needed to trigger complement activation. Again, conformational changes occur, exposing complement initiation sites in the Fc regions of the molecule. Not all IgG subclasses possess complement binding sites; IgG1 and IgG3 are strong complement activators, whereas IgG2 is a weak activator and IgG4 does not activate complement at all.

The first component of the CP is C1, a large multimeric protein comprising one copy of C1q and two copies each of C1r and C1s, held together noncovalently in a calcium-dependent complex. C1q, the recognition unit of the C1 complex, is itself multimeric, made up of six subunits, each comprising a collagenous stalk and a carboxy-terminal globular head; the six subunits are tightly associated along the collagenous stalks but separate in the head regions, giving the classical electron microscopy image of a “bunch
of tulips” (Fig. 36.1). To add further complexity to this, the most complex of complement proteins, each of the six C1q subunits is itself assembled from a trimer of homologous chains termed C1q-A, C1q-B, and C1q-C, intertwined in a triple helix in the stalk but distinct in the globular head. C1r and C1s, the proteolytic units of the C1 complex, are homologous serine proteases that associate with one-another in a calcium-dependent linear tetramer complex in the order C1s-C1r-C1r-C1s. When bound in the C1 complex, the tetramer folds upon itself in a figure-eight conformation and sits between the globular heads, held in place by ionic bonds between acidic residues in C1r/C1s and a single basic residue in each of the six C1q stalks (see Fig. 36.1).19

TABLE 36.1 Component Proteins of the Complement Pathway




Plasma Level

Classical pathway


460 kDa collectin, six subunits each of three 25 kDa chains

Binds immobilized IgG/IgM to initiate the CP

150 mg/L


85 kDa single chain

In C1 complex, activates C1s

50 mg/L


85 kDa single chain

In C1 complex, cleaves C4/C2

50 mg/L


S-S bonded heterotrimer, α, 97 kDa, β, 75 kDa, γ, 33 kDa

C4b fragment target-bound via thioester is the receptor for C2

500 mg/L


100 kDa single chain

C2a fragment bound to C4b cleaves/activates C3

25 mg/L

Alternative pathway

Factor B

110 kDa single chain, C2 homologue

Bb fragment bound to C3b cleaves and activates C3

200 mg/L

Factor D

25 kDa single chain protease

Cleaves factor B to activate

5 mg/L


Oligomers of 53 kDa chain

Stabilizes the C3bBb complex

20 mg/L

Lectin pathway


200-600 kDa collectin, two to six subunits each comprising three 32 kDa chains

Binds mannan sugars on pathogens to initiate the LP

0-5 mg/L (broad normal range)

Ficolin-1 (M-ficolin)

440 kDa lectin, 12 subunits each 36 kDa

Binds carbohydrate epitopes on pathogens to initiate the LP

0.05 mg/L

Ficolin-2 (L-ficolin)

420 kDa lectin, 12 subunits each 35 kDa

Binds carbohydrate epitopes on pathogens to initiate the LP

5 mg/L

Ficolin-3 (H-ficolin)

590 kDa lectin, 18 subunits each 3 kDa

Binds carbohydrate epitopes on pathogens to initiate the LP

5 mg/L


90 kDa single chain

Uncertain; in mouse, activates pro-fD to active fD

5 mg/L


74 kDa single chain

In complex with MBL or ficolin, cleaves C4/C2

0.4 mg/L


94 kDa single chain

Uncertain, perhaps as MASP-2

4 mg/L


19 kDa single chain

Suggested MASP-2 inhibitor

0.2 mg/L


44 kDa single chain

Suggested MASP-2 inhibitor

1.4 mg/L



S-S bonded heterodimer, α, 110 kDa, β, 75 kDa

Central component in all pathways, C3b major opsonin

1200 mg/L


S-S bonded heterodimer, α, 115 kDa, β, 75 kDa

Binds C6 to initiate TP, C5a major effector molecule

75 mg/L

Factor I

S-S linked heterodimer, heavy (50 kDa) and light (38 kDa) chains

Serine proteasel cleaves C3b/C4b in presence of cofactor

30 mg/L

Terminal pathway


110 kDa single chain

Binds C5b, C5b6 receives C7

50 mg/L


100 kDa single chain

Binds C5b6, C5b-7 attaches to target cell

90 mg/L


α (64 kDa) and γ(22 kDa) chains S-S linked, β chain (65 kDa) noncovalently associated

Binds C5b-7, C5b-8 receives C9

60 mg/L


70 kDa single chain

Binds C5b-8 to form MAC

60 mg/L

CP, classical pathway; fD, factor D; Ig, immunoglobulin; LP, lectin pathway; MAC, membrane attack complex; MASP, mannan-binding lectin-associated serine protease; MBL, mannan-binding lectin; TP, terminal pathway.

For C1 activation to occur, at least two of the six head domains must be engaged simultaneously, explaining both the efficiency of activation by the multivalent IgM and the

need for a critical surface density of IgG, the immune array, for activation; this latter limitation reduces the risk of inappropriate activation of complement on host tissues. Binding of the C1q globular heads to antibody Fc is itself a complex series of events; an exposed calcium ion in the head mediates the initial binding and induces further binding events that cause rotation of the head domain. These conformational changes stress the C1s-C1r-C1r-C1s tetramer tightly gripped between the C1q stalks, thereby triggering the autoactivation of C1r via cleavage at a single site to yield a two-chain disulphide-bonded active protease. Activated C1r then cleaves its homologous substrate, C1s, at a single site to generate the active C1s protease.*,19

TABLE 36.2 Regulatory Proteins of the Complement Pathway




Plasma Level/Cell Expression

Fluid phase

C1 inhibitor

Single chain, 100 KDa, heavily glycosylated

Serine protease inhibitor binds and inactivates C1r, C1s, MASP-2, others

150 mg/L

Factor H

Single chain, 150 kDa, 20 SCRs

AP convertase decay accelerator and fI cofactor

300 mg/L

Factor H-like 1

Single chain, 42 kDa, seven SCRs

As for fH

10 mg/L

C4b binding protein

550 kDa oligomer comprising seven α chains (eight SCRs) and one β chain (three SCRs)*

CP convertase decay accelerator and fI cofactor

200 mg/L

Carboxypeptidase N

280 kDa dimer of heterodimers of 83 kDa and 55 kDa

Inactivates C3a/C5a by removing C-terminal Arg

30 mg/L

S protein/vitronectin

84 kDa single chain

Binds C5b67 in fluid phase to inhibit MAC formation

250 mg/L


70 kDa heterodimer of 35 kDa chains

Binds C5b67 in fluid phase to inhibit MAC formation

150 mg/L



220 kDa single chain, 30 SCRs, TM

CP/AP convertase decay accelerator and fI cofactor

RBC, WBC, renal, others


70 kDa single chain, four SCRs, GPI

CP/AP convertase decay accelerator

Broadly distributed


60 kDa, single chain, four SCRs, TM

CP/AP fI cofactor

Broad, absent from RBCs


20 kDa globular protein, four S-S bonds, heavy glycosylation, GPI

Binds C5b-8 on membrane to inhibit MAC formation

Broad, all blood cells, etc.

AP, alternative pathway; Arg, arginine; CP, classical pathway; DAF, decay accelerating factor; fH, factor H; fI, factor I; GPI, glycosyl phosphoinositol; LHR, long homologous repeat; MAC, membrane attack complex; MASP-2, mannan-binding lectin-associated serine protease-2; RBC, red blood cell; SCR, short consensus repeat; TM, transmembrane; WBC, white blood cell.

*Common isoform; other oligomers of C4bp, α7β0, and α6β1, are also found in plasma.

Common isoform; forms comprising 37 SCRs (gain of LHR) and, rarely, 23 SCRs (loss of LHR) occur.

The fH-related proteins 1 to 5 are omitted for simplicity and because their functions are unconfirmed.

TABLE 36.3 Receptors for Products of Complement Activation



Ligand; Function

Cell Expression

C3/C4/C5 fragment receptors

CR1 (CD35)

TM, single chain, 30 SCRs

C3b/C4b; binds immune complexes; B-cell activation

RBC, WBC, FDC, renal, others


TM, single chain, 15 or 16 SCRs

iC3b/C3dg/C3d; sensitizes B cells for response to antigen

B cells, some T cells, FDC

CR3 (CD11b/CD18)

Integrin heterodimer, α 160 kDa, β 95 kDa, TM proteins

iC3b; phagocytic receptor, leukocyte migration

Myeloid cells, NK cells

CR4 (CD11c/CD18)

Integrin heterodimer, α 150 kDa, β 95 kDa, TM proteins

iC3b; phagocytic receptor, leukocyte migration

Myeloid cells, T cells, NK cells


45 kDa single chain Ig super-family member

C3b/iC3b; phagocytic receptor

Tissue macrophages

C3a receptor (C3aR)

54 kDa heptaspan G protein-coupled receptor

C3a; activates cell responses

WBC, brain and renal cells, etc.

C5a receptor (CD88)

45 kDa heptaspan G protein-coupled receptor

C5a > C5adesArg; activates cell responses

Broadly expressed


37 kDa heptaspan; G protein coupling uncertain

C5a/C5adesArg/C3a/C3adesArg; activates cell responses?

Broadly expressed

Receptors for C1q

CR1 (CD35)

See above

C1q; uncertain

See above


125 kDa single chain TM

C1q/MBL/SP-A; phagocytic receptor

Myeloid cells, endothelium


Tetramer 33 kDa subunits

C1q; phagocytic receptor

WBC, platelets

α2β1 integrin

Integrin, heterodimer

C1q, collagens, laminins, decorin, etc.; cell activation

Broadly expressed

CR1, complement receptor 1; CR2, complement receptor 2; CR3, complement receptor 3; CR4, complement receptor 4; CRIg, complement receptor of the immunoglobulin superfamily; FDC, follicular dendritic cell; Ig, immunoglobulin; MBL, mannan-binding lectin; NK, natural killer; RBC, red blood cell; SCR, short consensus repeat; TM, transmembrane; WBC, white blood cell.

FIG. 36.1. Structure of C1. A: C1q comprises six units tightly associated through the collagenous tail, separating in the neck region to spread the globular heads out from the tail. Each unit is itself made up of three closely apposed subunits, C1q-A, -B, and -C. C1r2C1s2 is assembled in a linear complex in the order C1r-C1s-C1s-C1r. B: In the C1 complex, the C1r2C1s2 complex winds into a figure-of-eight conformation and binds among the globular heads in a calcium-dependent complex.

The C1s serine protease has two substrates, C4 and C2, the next two proteins in the classical pathway sequence. The illogical ordering here reflects the fact that complement components were named chronologically, according to the order of their discovery (C′1 > C′2 > C′3 > C′4), rather than according to their position in the reaction. C4 is a relatively abundant protein, present in plasma at around 0.5 g/l. It is a large (210 kDa), disulphide-bonded heterotrimer. Activated C1s captures C4 from the fluid phase, perhaps in part through interaction of its short consensus repeat (SCR) domains with C4, then cleaves the C4 α chain at a single site near the amino terminus, releasing a 77 amino-acid fragment, C4a, and exposing in the cleaved α′ chain of the large fragment C4b a labile thioester group (Fig. 36.2A). Although most of the nascent C4b formed will decay in the fluid phase through hydrolysis of the thioester, a small proportion will bind reactive hydroxyl or amino groups on the activating surface, creating a cluster of covalently bound C4b around the initiating IgG/C1 complex. Immobilized C4b binds the next component in the sequence, C2, in a magnesium-dependent complex. C2 is a single-chain plasma protein of mass 100 kDa and plasma concentration around 25 mg/L; it is the most heat-labile of the complement proteins, destroyed by brief incubation of plasma at 56°C. C4b-bound C2 is cleaved by activated C1s in an adjacent IgG/C1 complex, releasing a 30 kDa fragment C2b, while the 70 kDa C2a fragment, an active serine protease, remains associated with C4b on the surface.,20,21

The magnesium-dependent C4b2a complex is the CP C3 convertase, the next activation enzyme in the sequence. C2a in the C4b2a complex is an active serine protease that cleaves C3, a two-chain 190 kDa protein, homologous to C4 and the most abundant of the complement proteins at around 1 g/l in plasma.22 Cleavage releases a 77 amino-acid fragment, C3a, from the amino terminus of the α chain of C3, exposing in the large fragment, C3b, a labile thioester group essentially as described previously for C4b (Fig. 36.2B). Again, most of the C3b formed decays by thioester hydrolysis, but a small fraction covalently binds the activating surface, clustering around the site of activation. Some of the C3b formed will directly bind C4b2a through its thioester to form a trimolecular complex, C4b2a3b; this binding is not a random event but occurs at a single specific site in C4b, placing C3b in the correct orientation for succeeding steps of activation. The C4b2a3b complex contains a binding site for C5 involving interactions with both C4b and C3b in the complex. C5, another homologue of C4 and C3, is a 200 kDa, two-chain molecule present in plasma at about 100 mg/L; importantly, C5 lacks the critical thioester group and so cannot bind covalently to targets. Once bound to C4b2a3b, the CP C5 convertase, C5 is cleaved by C2a in the complex, releasing a 74 amino-acid fragment, C5a, from the α chain of C5 and leaving the large fragment, C5b, loosely attached to the convertase. Cleavage of C5 is the final enzymatic step in the CP (Fig. 36.3).

Two features of CP activation are critical to its roles. First, amplification at each of the enzymatic steps is critical for efficient activation; thus, a single active IgG/C1 complex will deposit an abundance of C4b in the vicinity of the initiating

IgG/C1, each C4b2a complex formed will in turn cause deposition of many copies of C3b on the surrounding membrane, and each C5 convertase will also cleave multiple C5 molecules. Second, the nascent thioester groups in C4 and C3 are critical components of complement activation; without these entities, complement activation on surfaces would be an impossibility and the system would not function. The thioester group is formed from interaction between a glutamine and a cysteine residue that, in the intact molecule, are buried in the protein structure (Fig. 36.4).17,18 When activated by the convertase, a major conformational change occurs that exposes the internal thioester bond in C3b and C4b, making it very unstable and highly susceptible to attack by nucleophiles such as hydroxyl groups (-OH) and amine groups (-NH2) in membrane proteins and carbohydrates, creating a covalent bond that locks C3b and C4b onto the surface.23 The exposed thioester is highly labile and rapidly inactivated by hydrolysis, restricting binding of C3b and C4b to the immediate vicinity of the activating enzyme.

FIG. 36.2. Activation and Inactivation of C4 and C3. A: C4 is activated by cleavage of a 77 amino acid fragment, C4a, from the amino terminus of the α-chain, catalyzed by C1s in the classical pathway (CP) or MASP-2 in the lectin pathway (LP). Activation exposes the thioester in the α′-chain of C4b, enabling attachment to surfaces. The plasma enzyme factor I (fI) cleaves and inactivates C4b in the presence of appropriate cofactors, releasing the large, multichain C4c and leaving the C4d fragment attached to the surface. B: C3 is activated by cleavage of a 78 amino acid fragment, C3a, from the amino terminus of the α-chain, catalyzed by the C3 convertase of the CP/LP or alternative pathway. Activation exposes the thioester in the α′-chain of C3b, enabling attachment to surfaces. The plasma enzyme fI cleaves and inactivates C3b in the presence of appropriate cofactors, releasing a small fragment, C3f, from the α′-chain, and leaving the iC3b fragment attached to the surface. Further cleavage events, not shown here, occur in the presence of CR1 as cofactor, releasing the large, multichain C3c and leaving the C3dg fragment attached to the surface. Finally, C3dg is further degraded by plasma proteases, releasing the small C3g peptide and leaving C3d attached to the target.

FIG. 36.3. Classical Pathway Activation. C1s in the activated C1 complex bound to immobilized antibody cleaves C4. C4b binds the surface through its thioester and acts as a receptor for C2, which is then cleaved by C1s to yield the C4b2a complex (classical pathway [CP] C3 convertase). C4b2a cleaves C3, depositing C3b on adjacent membrane. C3b binding in the C4b2a complex acts as a receptor for C5, presenting it for cleavage by C2a in the complex. Cleavage of C5 to form C5b is the last enzymatic step in the CP and initiates the terminal pathway.

FIG. 36.4. Thioester Group Structure and Activation. A: The buried thioester in C3 and C4 is formed from the residues -Cys-Gly-Glu-Gln-. The bond is formed from a condensation reaction involving the Cys thiol group and the Glu carboxyl group in this four amino acid sequence. B: Hydroxyl groups on target proteins or sugars “attack” the exposed thioester in C3b or C4b in a reaction that is catalyzed by an adjacent histidine residue in the complement protein. The reaction creates a covalent bond, rigidly attaching C3b or C4b to the target.

The Alternative Pathway

The concept of an AP of complement activation grew out of the recognition that some pathogens activated complement without the need for antibody. The controversial history was briefly described previously, but today the existence of the AP is not in doubt, although whether it should be considered a separate pathway remains an area of debate. The AP functions in two interlinked ways, first as an efficient amplification loop to drive further complement activation whatever the initiating trigger, and second as an always-on pathogen sensor ready to attack foreign surfaces.24,25 The trigger for AP activation is the presence of “activated” C3; this can be “classically activated” nascent C3b on membranes or in plasma, or C3 that has been hydrolyzed in plasma, C3(H2O). Spontaneous hydrolysis of C3 occurs under physiological conditions at approximately 5% of total C3 per day, driving a constant, low-grade “tickover” activation of the AP in plasma. Factor B (fB), a 110 kDa single-chain protein and structural homologue of C2, binds to form a magnesium-dependent complex, C3bB or C3(H2O) B; in its bound state, fB is cleaved by a plasma serine protease, factor D (fD), releasing the 50kDa Ba fragment and leaving the serine protease fragment, Bb, in the complex to form the AP C3 convertase. The enzyme fD is a small (25 kDa) serine protease present in tiny amounts (1 mg/L) in plasma as an active enzyme with just one substrate, fB in complex with C3b/C3(H2O). Bb in the C3bBb or C3(H2O)Bb convertase cleaves C3 exactly as occurs in the CP, releasing C3a and generating C3b that can bind to targets and/or bind more fB to continue the AP amplification cycle—a positive feedback amplification loop (Fig. 36.5). C3bBb on targets will catalyze the deposition of many C3b molecules as described for the CP convertase, and C3b binding to a specific site on C3b in the convertase will create a trimolecular C5 convertase in which C5 can be cleaved by the adjacent C2b, events that are analogous to those described in the CP. I have yet to mention the protein for which the AP was first named: properdin. This complicated molecule, an oligomer comprising two, three, or four copies of a 53Da single-chain protein, binds and stabilizes the AP convertases, reducing their inherent tendency to fall apart (“decay”) and increasing markedly their capacity to perpetuate activation. Recently, a second role for properdin, binding to bacterial surfaces and forming a platform for AP convertase assembly, has been emphasised.26

The recognition that the AP is always on, in a constant tick-over state, explained how complement is so efficiently activated on pathogens. Tick-over ensures that all plasmaexposed surfaces are continuously showered with C3b; on self cells, inhibitory mechanisms described in the following sections ensure that no amplification occurs. However, on pathogens, lacking such protection, AP amplification kicks in, rapidly coating the surface with C3b. It is important
to note that the AP is inexorably linked to the CP in that C3b generated through the latter will feed into the former to amplify activation. It therefore does not matter whether the initial C3b is generated by the CP or AP (or, indeed, the LP that follows), the AP amplification loop will amplify the response, particularly when occurring on the intrinsically activating surfaces of pathogens.

FIG. 36.5. Alternative Pathway Activation. The initiating activated C3 may be C3(H2O) formed by spontaneous hydrolysis in the fluid phase, or C3b, formed from classical pathway/lectin pathway activation in the fluid or surface phase. Activated C3 binds factor B, rendering it susceptible to cleavage/activation by factor D. The resulting enzyme, C3(H2O)Bb or C3bBb, cleaves more C3b, creating a positive feedback amplification loop in the fluid phase and depositing more C3b on the target. C3b binding into a target-bound C3bBb enzyme acts as a receptor for C5, presenting it for cleavage by Bb in the complex. Cleavage of C5 to form C5b is the last enzymatic step in the alternative pathway and initiates the terminal pathway.

The Lectin Pathway

The LP, first described only in 1987,27,28 shares features with both the AP and CP. Like the AP, it provides antibody-independent “innate” immunity, activated by pathogens independent of antibody. Its similarities to the CP are legion; indeed, it only differs in the initiation step and is perhaps better considered as a different route to CP activation that bypasses the need for antibody.29 The C1 complex is replaced by a structurally similar multimolecular complex, comprising a C1q-like recognition unit, either mannan-binding lectin (MBL) or ficolin (a family of three proteins in man), and MBL-associated serine protease-2 (MASP-2) that provides the enzymatic activity. The recognition units bind carbohydrate epitopes, N-acetyl glucosamine for both ficolins and MBL, and mannose for MBL alone; these ligands are abundant in the cell walls of diverse pathogens, including bacteria, yeast, fungi, and viruses, making them targets for LP activation. Each MBL subunit comprises a homotrimer of 32 kDa chains, an amino-terminal collagen-like region responsible for trimerization in a triple helix, a short α-helical neck region, and a globular carbohydrate recognition domain. Subunits assemble into oligomers containing between two and six oligomers, the latter closely resembling the C1q hexamers. Indeed, C1q and MBL are members of the collectin family of proteins characterized by globular head regions with binding activities and long collagenous tail regions with diverse roles. Serum levels of MBL are highly variable in the population, from undetectable to 5 mg/L. Ficolins are novel lectins, structurally similar to C1q and MBL through the collagenous regions, but with head regions comprising fibrinogen-like domains.30 Three ficolins are described in man, termed ficolin-1, ficolin-2, and ficolin-3. Serum concentrations of all are low, ficolins-2 and -3 around 5 mg/L and ficolin-1 about 0.05 mg/L, although reported levels for each vary widely. MASP-2 is a member of a family of homologous lectin-binding proteins, structural homologues of C1r/C1s, three of which express protease activity (MASP-1, MASP-2, MASP-3) while the other two do not (MAp19, also called sMAP; MAp44, also called MAP-1). After a decade of debate, claim and counterclaim, it is now generally agreed that MASP-2 is the critical enzyme of the lectin pathway: a MASP-2 dimer associates with the MBL or ficolin oligomer in a calcium-dependent manner to generate a complex that is necessary and sufficient to create the activation enzyme.31 The biological roles of the other members of the MASP protein family remain obscure, although MASP-1 has recently been shown to be critical for AP activation in the mouse as it activates profD.32 MASP-2 complexed with the MBL or ficolin oligomer captures C4 from the fluid phase, likely via its SCR domains, then cleaves C4 at a single site, identical to that targeted by C1s, to release C4a and generate C4b that binds through its thioester to the surface adjacent the initiating enzyme. The rest of the sequence mirrors that of the CP: C2 is captured onto C4b and presented for cleavage by MASP-2
in an adjacent complex, and the resultant enzyme, C4b2a, continues activation through C3 and C5.

FIG. 36.6. Terminal Pathway. C5b, still attached to the convertase, binds C6, then C7. The trimolecular C5b67 complex is released to the fluid phase. A fraction of the complexes formed attach through hydrophobic interaction to the membrane. Membrane bound C5b67 recruits C8, then multiple copies of C9. C9 monomers unfold, insert into and through the membrane, and polymerize to form a transmembrane pore through which ions and water can freely flow. The inset shows an electron micrograph of a complement-lyzed cell; circular membrane attack complex lesions, a light protein rim surrounding a dark pore, are readily seen.

The Terminal Pathway

The terminal pathway (TP), sometimes referred to as the membrane attack pathway, is a final, common pathway for all activation routes. Cleavage of C5 is the last enzymatic step in the complement sequence and the final step of each activation pathway. The TP is a system almost unique in nature where five plasma proteins join to create an amphipathic membrane-inserted complex, the membrane attack complex (MAC), that creates a lytic pore in the membrane (Fig. 36.6).33 The TP begins with the binding of the next component in the sequence, C6, to C5b still in the grip of the C5 convertase. C6 is a 100 kDa single-chain protein present in plasma at around 50 mg/L. Conformational changes during formation of the C5b6 complex weaken the grip of the convertase and create a binding site for the next component, C7, a 95 kDa single-chain molecule, plasma concentration about 90 mg/L. C6 and C7 are homologous molecules and are genetically linked, with genes adjacent on chromosome 5p. Incorporation of C7 causes further loosening of grip, releasing the trimolecular C5b67 complex into the fluid phase. The newly released C5b67 complexes shower down onto the lipid membrane surrounding the convertase and bind firmly to the surface via a hydrophobic site in the complex, thereby creating a nidus for continued assembly of the MAC. This is an inefficient process; the large majority of C5b67 complexes formed are inactivated in the fluid phase before they can bind membranes. Spontaneous inactivation occurs rapidly even when the C5b67 complex is assembled from pure proteins; in plasma, several proteins act as C5b-7 inhibitors to further accelerate inactivation. Those C5b67 complexes that do bind membranes then recruit the next protein in the sequence, C8, a heterotrimeric molecule (α and β chains each approximately 61 kDa, γ chain, 22 kDa; α and γ covalently linked, β noncovalently associated) present in plasma at about 80 mg/L. Binding of C8 introduces additional hydrophobicity, causing the resultant C5b-8 complex to embed more firmly in the membrane. There is some evidence that the C5b-8 complex can cause membrane disruption and leakiness, but the major membrane disruption necessary to kill bacteria or other target cells requires the recruitment of multiple copies of the final component of the MAC, C9, a 70 kDa single-chain protein present in plasma at around 60 mg/L. The first globular C9 molecule binds C8 in the C5b-8 complex and undergoes major conformational rearrangement, unfolding to reveal a hydrophobic face that allows insertion into and through the membrane lipid bilayer. As additional C9 molecules are recruited, they in turn unfold and insert, aligning with the first C9, like barrel staves; with the recruitment of about 10 C9 molecules, the barrel is completed, creating a protein-lined channel through the membrane, the MAC (see Fig. 36.6). The C9 hydrophobic faces tightly lock the MAC in the membrane while the opposite, hydrophilic faces create a channel through which water and ions can flow, the MAC pore.** In electron micrographs of complement-lyzed targets, the MAC is readily visible as a
ring of about 10 nm internal diameter. The C5b-8 complex is displaced to the edge of the ring, resembling a pan handle.

FIG. 36.7. Complement Regulation. Regulatory proteins are present both in plasma (dotted boxes) and on cell surfaces (solid boxes) to control complement activation at multiple stages in the activation and terminal pathways.


Complement is designed to efficiently target and destroy pathogens and other foreign cells; however, the nature of the system dictates that it also targets host cells and thus has the potential to cause tissue damage and disease—a double-edged sword. Damage to self is minimized by the presence of numerous regulatory proteins and control mechanisms. Complement regulatory proteins are present both on host cell membranes and in the fluid phase, collaborating to minimize activation and suppress amplification of complement. Each stage of each of the complement pathways is controlled through inhibition or accelerated decay of complement enzymes or by physical interference (see Table 36.2 and Fig. 36.7).

Control of Initiation of the Classical Pathway and Alternative Pathway

Activated C1 is regulated by a plasma serine protease inhibitor called C1 inhibitor (C1inh), a 100 kDa heavily glycosylated single-chain protein present in plasma at around 150 mg/L. C1inh binds C1r and C1s in activated C1 and forms a tight complex (C1inh-C1r2C1s2), simultaneously stripping them from the Ig-bound C1q. Like other serine protease inhibitors, C1inh undergoes an autocatalytic cleavage event during formation of the complex that stabilizes its binding to substrate; it thus behaves as a suicide inhibitor, inactivated during the process of inhibition. C1inh is the only plasma inhibitor for activated C1, but it has several other substrates, notably plasma kallikrein and the coagulation enzymes, factor XIa and factor XIIa.34 C1inh also inhibits the LP, binding to and removing MASP-2 from the activated MBL/ficolin-MASP complex to switch off activation.

Control of the CP/LP C3 Convertase

The CP/LP C3 convertase, C4b2a, is regulated by inhibitors present in the fluid phase and on membranes. C4b binding protein (C4bp) is a complex, multi-chain plasma protein present in plasma at about 200 mg/L. The predominant structure for the C4bp oligomer comprises seven copies of an eight-SCR α-chain and one copy of a three-SCR β-chain (α7β1), locked together through disulphide bonds in the short, carboxy-terminal oligomerization domains, yielding a spider-like eight-armed structure (Fig. 36.8).††,35 Oligomers missing the β-chain and containing seven (α7β0) or eight (α8β0) α-chains are found in various amounts in different individuals. Each of the α-chains in C4bp contains a binding site in its three amino-terminal SCRs that can bind C4b in the C4b2a complex. C4bp exerts two different inhibitory activities: first, it displaces C2a from C4b, termed “accelerated decay,” and second, it acts as a cofactor for enzymatic cleavage of C4b by the plasma enzyme factor I (fI), a two-chain (heavy, 50 kDa; light, 38 kDa), disulphide-bonded serine protease present at about 30 mg/L in plasma.‡‡ C4b is inactivated by cleavage at two sites in the α′-chain straddling the thioester, thereby releasing a large fragment, C4c, to the fluid phase and leaving the small C4d fragment attached to membrane (see Fig. 36.2A); neither C4c nor C4d are of particular biological
relevance. The β-chain of C4bp binds and inactivates Protein S, an important anticoagulation protein, one of many fascinating links between complement and coagulation.36

FIG. 36.8. Complement Regulators in the Regulators of Complement Activation Cluster. The complement regulators (and receptors) encoded in the regulators of complement activation cluster are all assembled from the same building block, the short consensus repeat (SCR) domain. The simplest, the fluid-phase regulator factor H, comprises just 20 SCRs arranged like beads on a string; C4bp is more complicated but is dominated by SCRs, the oligomer held together by short non-SCR joining sequences at their carboxy-termini. On the membrane, complement receptor 1 (CR1) is a very large molecule, comprising 30 SCRs in its most common isoform, with short transmembrane and cytoplasmic regions at its carboxy-terminus. The first 28 SCRs are arranged in four groups of seven called long homologous repeats, with repeating homology. Complement receptor 2 structurally resembles CR1 but comprises only seven SCRs in its most common isoform. Decay accelerating factor and membrane cofactor protein both comprise four amino terminal SCRs followed by a heavily carboxylated stalk and either a glycosyl phosphoinositol anchor (decay accelerating factor) or transmembrane and cytoplasmic regions (membrane cofactor protein).

Two homologous cell surface complement inhibitors act in tandem to regulate the C4b2a complex. Decay accelerating factor (DAF; CD55) is a broadly distributed glycosyl phosphoinositol (GPI)-linked membrane protein made up of four SCR domains and a heavily glycosylated stalk at the carboxy terminus that links to the GPI anchor.37 As the name implies, DAF accelerates decay of the convertase; it binds membrane-associated C4b2a and displaces C2a, then releases because its affinity for C4b alone is low. Decay allows the second inhibitor, membrane cofactor protein (MCP; CD46), access to bind membrane-associated C4b. MCP, like DAF, is broadly distributed, though absent from erythrocytes, and comprises four SCR domains and a heavily glycosylated stalk, but differs in that it is a transmembrane protein with an intracellular domain that has important signalling roles, described in a later section.38 As its name implies, it is a cofactor for fI-mediated cleavage of C4b to C4c and C4d. This two-step regulation is important because decay alone leaves intact C4b on the surface, which can recruit more C2 and form a new convertase; in contrast, fI cleavage destroys C4b, thereby switching off activation. Another membrane protein, complement receptor 1 (CR1; CD35), can also regulate the C4b2a enzyme through decay and cofactor activities. CR1 is an enormous molecule expressed on erythrocytes, the majority of leukocyte subsets, dendritic cells, glomerular podocytes, and a few other cell types. CR1 comprises, in its most common isoform, 30 SCRs, with a transmembrane region and short cytoplasmic domain at the carboxy-terminus.39 The amino-terminal 28 SCRs are arranged in seven blocks, termed long homologous repeats (LHRs), in which the first SCR in each LHR is homologous to the first SCR in every other LHR, the second to every other second SCR, and so on. The functionally important parts of CR1 are the C3b-/C4b-binding sites contained in the first four SCRs of the first three LHRs; although differing in relative activities and ligand binding affinities, each of these sites can both decay the C4b2a convertase and bind C4b to catalyze its cleavage by fI. Because the functional sites are at the cell-distal end of a very large, elongated molecule, interaction with C4b or C4b2a on the same cell (intrinsic activity) is limited; however, interaction with C4b or C4b2a on adjacent surfaces (extrinsic activity), for example on circulating immune complexes, is favoured and is an important physiological role of erythrocyte CR1.40

Control of the Alternative Pathway C3 Convertase

The AP C3 convertase, C3bBb, is regulated in a very similar manner to the CP convertase. C4bp has only very low affinity for C3b or C3bBb, likely of no physiological relevance. Its role is taken by another plasma protein, factor H (fH), an elongated, single-chain protein made up entirely of 20 SCRs, present in plasma at about 300 mg/L.41 The complement regulatory activity of fH resides in SCRs 1 to 4; these four domains bind the C3bBb convertase, displace Bb from C3b (accelerated decay), and act as cofactor for fI to cleave C3b at two internal sites in the α′-chain, generating the large membrane-bound fragment, iC3b, and releasing a small peptide, C3f. iC3b cannot drive further complement activation but is an important ligand for complement receptors, as described in a later section. On membranes, the C3bBb convertase is regulated by the same proteins that control C4b2a. DAF binds C3bBb, displaces Bb, and is then released (decay acceleration); MCP binds C3b and catalyzes its cleavage by fI to yield iC3b and C3f (see Fig. 36.2B). CR1, through its C3b-/C4b-binding sites, both decays C3bBb and binds C3b to catalyze its cleavage by fI. Importantly, CR1 catalyzes a second fI cleavage event, an additional cut that releases the large fragment, C3c to the fluid phase, leaving a small piece of the α′ chain, C3dg, attached to the membrane through the thioester. Plasma proteases then chew C3dg down further to C3d, an important ligand for complement receptors.

Control of the Anaphylatoxins

The fragments C3a and C5a released during activation of complement are extremely potent inflammatory mediators, attracting phagocytes to the site of activation (chemotaxis),
and activating them to release their cargoes of enzymes, reactive oxygen species, and other proinflammatory molecules. These powerful effects are limited in time and space by the rapid inactivation of C3a and C5a in plasma and tissue fluids. The principal inactivator is carboxypeptidase N (CPN), a metallocarboxypeptidase made up of four noncovalently associated subunits, two identical 85 kDa regulatory subunits that stabilize the enzyme, and two identical 55 kDa catalytic subunits, present in plasma at 30 mg/L. CPN is a zinc-dependent enzyme that cleaves carboxy-terminal arginine or lysine residues from proteins or peptides; it removes the carboxy-terminal arginine (Arg; N) residue from both C3a and C5a.42 The residual peptides, C5adesArg and C3adesArg have, respectively, much reduced or absent proinflammatory activities because of reduced or absent capacity to bind the receptors for the parent molecules. Plasma levels of these peptides can be measured and provide a useful index of ongoing complement activation in man and animal models. Although C3adesArg and C5adesArg are inactivated with respect to proinflammatory activity, they retain or acquire other activities that may be of equal biological importance. CPN also removes the carboxy-terminal Arg from bradykinin, altering receptor binding and biological properties of this important mediator.

Control in the Terminal Pathway

Control at the C3 convertase stage is the crucial pinch-point of complement activation; nevertheless, later stages are also subject to control. C5 convertases of both activation pathways are subject to the attentions of the same regulators described for the equivalent C3 convertases, limiting their lifespan through decay acceleration and cofactor collaboration with fI. As further insurance against damage to self, the TP is also regulated at multiple stages. Once released from the convertase, the C5b67 complex must make the perilous journey to the membrane—an entirely random migration that must be accomplished before the membrane-binding site decays or is blocked, all of a fraction of a second. The vast majority of the C5b67 complexes formed never bind membrane; the site is hydrolyzed or the plasma scavenger proteins S-protein (also called vitronectin) and/or clusterin bind and block the site.43,44 Perhaps the most efficient fluidphase inhibitor of the TP is the next component in sequence, C8; if C8 binds C5b67 before it attaches to membrane, the binding site is lost. Because the process is so inefficient, large amounts of the waste product, the terminal complement complex (TCC) containing all the TP components, clusterin and S-protein, are found in plasma when complement activation occurs in vivo. TCC levels in plasma can be measured and provide an excellent index of ongoing complement activation. Despite the attention of these fluid-phase inhibitors, some C5b67 complexes will bind the membrane, usually on the same cell and close to the triggering convertase, but occasionally on adjacent, innocent bystander cells. C8 then binds, and with recruitment of multiple C9 molecules, the MAC forms. Membrane regulation of this process is provided by CD59

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

May 30, 2016 | Posted by in IMMUNOLOGY | Comments Off on Complement

Full access? Get Clinical Tree

Get Clinical Tree app for offline access