Complement in disease: a defence system turning offensive

Danger sensing and complement activation

A key function of complement is the sensing of potential insults, which is achieved through the detection of pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs) by specialized recognition molecules (FIG. 1). C1q predominantly binds to IgG and IgM when arranged in antigen–antibody complexes¹² but also detects other molecules such as lipopolysaccharides and phosphatidylserine^4,13. In contrast, mannose-binding lectin (MBL) mainly recognizes carbohydrate patches that are rich in D-mannose and L-fucose. Although MBL has long been considered the prime initiator of the lectin pathway (BOX 1), it is now clear that other recognition molecules such as ficolins and collectin-11 are also involved in sensing carbohydrates and/or acetylated moieties on microbial and apoptotic cells⁴. In the circulation, the recognition molecules are associated with serine proteases: C1q associates with C1r and C1s, and MBL, ficolins, and collectin-11 form complexes with MBL-associated serine proteases (MASPs). Binding of these complexes to target surfaces leads to activation of the associated serine proteases, which then cleave the plasma proteins C2 and C4. Activation of C4 produces the opsonin C4b, which is deposited on the target surface via a thioester moiety. The cleavage fragment of C2, in turn, binds to C4b to generate a C3 convertase complex (termed either C4b2b or C4b2a in the literature), which activates the abundant complement component C3. Convertase-mediated cleavage of C3 liberates the anaphylatoxin C3a and generates the versatile opsonin C3b that, like C4b, covalently binds to activating surfaces. The high plasma concentration of C3 (~1 mg/ml), which is relatively inert in its native form, is a critical factor in the rapid reactivity of the complement system. Conversely, the ultra-short reaction time of the activated thioester moiety of C3b and C4b restricts opsonization to the immediate site of activation and reduces bystander opsonization of other cells⁴.

The amplification step that follows initial opsonization often determines the fate of the target cell (FIG. 1). Surface-deposited C3b binds the protease factor B (FB), which is subsequently cleaved by factor D (FD) to form another C3 convertase, C3bBb. Importantly, this convertase is able to activate additional C3 to form C3b and produce more C3 convertases, thereby creating a self-amplification loop that leads to efficient opsonization of unprotected cells and surfaces. Under most circumstances, C3-mediated amplification by the ‘alternative pathway’ (BOX 1) contributes most of the overall complement response^14,15. The increasing density of C3b opsonins on the target surface leads to the formation of convertases that preferentially cleave complement component C5. Upon activation, with release of the anaphylatoxin C5a, the resulting C5b associates with complement components C6 and C7, inserts into membranes and forms a membrane-attack complex (MAC) through binding to C8 and multiple copies of C9. Of note, although C4b can generate a C3 convertase, the presence of C3b is required for C5 convertase formation. Furthermore, C3 is continuously hydrolysed at a low rate, producing a C3b-like conformational state that can form alternative pathway-initiating C3 convertases with FB and FD. This ‘tick-over mechanism’ provides a state of complement alertness in the circulation but can also be actively induced when C3 is adsorbed on certain interfaces (for example, on platelet surfaces or biomaterials), and changes its conformation¹⁶.

Effector functions of complement

Activation and amplification of complement creates a variety of potent effectors (FIG. 1). The MAC with its lytic pore is probably the best-recognized of these effectors. However, not all cells are equally susceptible to MAC, and swift MAC-induced lysis mainly occurs in aged erythrocytes and certain Gram-negative bacteria. On nucleated cells, MAC formation is tightly regulated, assembled MAC complexes are actively removed from the membrane, and the lytic effects of MAC are counteracted by ion pumps¹⁷. These protective effects can, however, be overwhelmed — a principle that is used to achieve complement-dependent cytotoxicity in cancer therapy¹⁸. Of note, cell lysis is not the only effector function of MAC; even when formed at sublytic levels, MAC can induce damage or activation of host cells and act as a proinflammatory mediator on various cells ranging from neutrophils to platelets¹⁷.

The dominant effector in signalling danger and in the induction of immune-modulatory responses by complement is typically the other C5-derived activation product, C5a^8,19. This anaphylatoxin is a potent chemoattractant, recruiting neutrophils and other cells to the activation site and priming them by binding to C5a receptor 1 (C5aR1, also known as CD88). The role of a second C5a receptor, C5aR2 (also known as C5L2), is not fully resolved and might involve decoy or modulatory functions¹⁹. C5a also exerts a variety of other effects on various cell types, including contraction of smooth muscle cells, and is an important orchestrator of immuno–inflammatory crosstalk (see below). Although C3a and C5a were once thought to have analogous functions because of their structural similarities, the two anaphylatoxins are now considered to have distinct activity spectra. Compared to the proinflammatory C5a, the activity of C3a, which signals through the C3a receptor (C3aR), seems to be more diverse and context-specific. For example, C3a does not attract neutrophils but might even counteract the neutrophil-attractant effects of C5a, and is reported to have direct antimicrobial activity^20,21. As such, C3a must be regarded as a unique inflammatory modulator, not just a less-potent homologue of C5a²⁰.

Anaphylatoxin-mediated attraction and priming of immune cells are important prerequisites for phagocytosis, which contributes to the removal of microbial threats and cellular debris through antibody and complement-assisted mechanisms. Although C1q can contribute to phagocytosis by binding to its receptors gC1qR and cC1qR (also known as calreticulin)²², complement-mediated phagocytosis is mainly driven by the interactions of opsonins on target surfaces with complement receptors on phagocytes. Complement receptor 1 (CR1, also known as CD35), which is ubiquitously expressed on erythrocytes and many nucleated cells, binds both C3b and C4b. This adhesion receptor can facilitate Fcγ-receptor-mediated uptake of antibody-tagged cells, and erythrocyte-bound CR1 is instrumental in transporting opsonized particles to the spleen and liver⁵. Importantly, CR1 acts as a regulator of complement activation by mediating the degradation of C3b to iC3b and C3dg, thereby shifting the signalling profile of the opsonins. Although the term iC3b had been introduced to imply an ‘inactive’ form in terms of convertase formation, iC3b is in fact among the most versatile effectors. It is a major ligand for the integrin receptors CR3 (also known as CD11b/CD18) and CR4 (also known as CD11c/CD18), which are found on most phagocytic cells. Moreover, C3b and iC3b bind to the complement receptor of the immunoglobulin family (CRIg), a phagocytic receptor on tissue-resident macrophages such as Kupffer cells²³. Finally, iC3b and C3dg modulate adaptive responses by binding to CR2 (also known as CD21) on B-cells and follicular dendritic cells (as discussed below). Thus, C3-derived opsonins fulfil various immune-modulatory functions, and their density and dynamic conversion on distinct cell surfaces shape the effector profile of the complement system.

Regulating the complement response

Modulators of complement activity in the circulation and on the surface of cells are essential to prevent complement from attacking healthy host cells and to adjust the response of the complement system to microbial and homeostatic insults (FIG. 1). Regulation of the complement system starts at the initiation steps through the control of recognition molecules and their associated proteases. The circulating C1 esterase inhibitor (C1-INH) inactivates C1r, C1s, and MASP-1/2 but also acts on non-complement serine proteases^4,24. Splice products of MASPs (MAP-1 and sMAP), which lack the protease domains of MASPs, compete with MASPs for binding to MBL, ficolins, and collectins²⁵.

The assembly of C3 and C5 convertases is particularly well-controlled by proteins from the ‘regulators of complement activation’ (RCA) family. These modular proteins, which are composed of complement control protein (CCP) domains, prevent the formation of convertase complexes and destabilize existing ones. They also act as cofactors for the serine protease factor I (FI) to enable the degradation of C3b and/or C4b. Among RCA members, the membrane-bound CR1 is particularly potent, since it acts on all C3 and C5 convertases and mediates full cleavage of both C4b and C3b. The other membrane-bound RCA family members confer either cofactor activity (membrane cofactor protein; also known as CD46) or convertase decay activity (decay accelerating factor; also known as CD55). The soluble RCAs C4b-binding protein (C4BP) and factor H (FH) control complement activation in the circulation and assist in modulating the response on host cell surfaces. C4BP acts on C4b-containing convertases and mediates the degradation of C4b. FH is a 20-domain protein, the regulatory activity of which is mediated by its binding to hydrolysed C3 and C3b, thereby controlling both tick-over activation and the amplification loop. The N-terminus of FH provides convertase decay activity for C3bBb and enables FI-mediated degradation of C3b to iC3b. FH also has surface-recognition capabilities that discriminate between healthy and apoptotic host cells and microbial particles. The C-terminus of FH harbours an additional binding site for C3-derived opsonins and areas that recognize polyanionic patches found in glycosaminoglycans and sialic acids on host surfaces. FH-like protein 1 (FHL-1), a splice product that consists of the first seven FH domains, has the same regulatory capacity as FH but lacks some recognition capabilities; its smaller size, however, might allow better tissue penetration with potential implications for local convertase regulation²⁶.

By tightly controlling opsonization and convertase formation, RCAs impair C3 and C5 activation and largely prevent effector generation on host cells. However, given the potentially devastating effects of C5-derived effectors, additional mechanisms exist to keep their activity in check. The membrane-bound protein CD59 binds to active forms of C8 and C9 and prevents MAC formation. In addition, clusterin and vitronectin act as MAC regulators and scavenge partially formed MAC in solution. Finally, many host cells swiftly remove C5b–9 complexes via exocytosis or endocytosis¹⁷.

Anaphylatoxin activity is mainly regulated by carboxypeptidases that rapidly convert C3a and C5a to their desarginated forms (des-Arg), which profoundly reduces their effector functions. Residual and sometimes distinct activities of C3a-desArg and C5a-desArg when compared to their parent anaphylatoxins have been reported, although the physiological consequences of these activities remain unclear^19,27. Nevertheless, these findings indicate that conversion steps do not lead to complete inactivation of complement effectors, but rather to a shift in their activity and/or signalling profiles.

Whereas the above-mentioned regulators tame the complement response to protect host cells, the plasma protein properdin (also known as factor P) acts as a positive modulator. Properdin binds to and stabilizes the C3 convertase of the amplification loop, enhancing the overall complement response. Although properdin is considered to be an initiating recognition molecule because of its ability to bind to surface structures such as lipopolysaccharide and heparin and its ability to provide an interaction platform for C3b or hydrolysed C3, the implications of this function are still debated^28,29. Properdin had long been thought to be the only positive regulator of complement activation; however, several molecules that tailor complement activity on specific cell surfaces have now been described, giving rise to the concept of context-specific complement modulation (BOX 2).

Although complement is mainly regarded as an intra-vascular system, increasing evidence supports an important role for complement components that are produced locally by specific cells or tissues³⁰. Indeed, complement components are not only synthesized by the liver but by many other cells. For example, adipocytes are the primary source of FD production, and platelets and neutrophils store several complement proteins in their granules. Differential activation of complement in particular tissues is an important but not yet fully appreciated phenomenon that might help the immune system to distinguish between beneficial and pathogenic triggers.

Crosstalk with coagulation systems

Cooperation between the complement and coagulation systems has many potential implications for health and disease³¹. Both cascades are driven by tiered, serine protease-mediated activation steps, and examples of each pathway activating the other have been reported. For instance, coagulation enzymes such as thrombin and kallikrein can activate C3 and/or C5, whereas certain MASPs can cleave fibrinogen, prothrombin and factor XIII, among others. However, these cross-activation activities are typically low when compared to their normal routes of activation, and the (patho) physiological implications of this crosstalk remain to be explored. Independent of their mode of generation, anaphylatoxins influence coagulation through direct effects on platelets, neutrophils and endothelial cells or by stimulating pro-coagulant cytokines. For example, C5aR1 can induce expression of tissue factor (TF), thereby triggering the extrinsic coagulation pathway³². Conversely, thrombin-activatable fibrinolysis inhibitor (TAFI) is a carboxypeptidase (carboxypeptidase B2), the active form of which desarginates anaphylatoxins and tames their effects. von Willebrand factor (vWF) has a complex modulatory role in complement–coagulation interplay. Ultra-large multimeric forms of vWF, as observed after tissue injury, can provide a binding platform for C3b to trigger complement activation. Concurrently, vWF interacts with FH and enhances its cofactor activity for the FI-mediated degradation of C3b. Conversely, FH seems to interfere with the hydrolysis of vWF multimers by ADAMTS13. This bidirectional modulation has been proposed to strengthen platelet aggregation while keeping complement activation in control³³.

The connection between complement and platelet activation has garnered considerable interest in the past few decades, but remains obscure³⁴. Platelets seem largely unaffected by complement when in their quiescent form; upon activation, however, platelets engage with complement through a complex interplay. For example, chondroitin sulfate exposed on the platelet surface is recognized by both the activator C1q and the regulators FH and C4BP. Exposed gC1qR can trigger additional C1q binding to platelets, and P-selectin, acting as a ligand for C3b, can initiate convertase formation on the surface of activated platelets. A 2015 study demonstrated that C3 can adhere to activated platelets whereupon it is transformed to a hydrolysed-like state that is capable of forming convertases and binding to CR3 (REF. 35). The release of complement components such as FD from the α-granules of activated platelets might fuel convertase-mediated complement turnover. Platelets themselves are well-protected from complement damage through the expression and engagement of complement regulators^31,34; however, the generation of complement effectors could be important in enhancing platelet activation. Soluble MAC (that is, sC5b–9) can trigger the secretion of the platelets’ α-granules, and C1q induces the expression of P-selectin, among other effects³⁴. Anaphylatoxins have also been described as platelet activators. Although C3aR and C5aR1 are barely detectable on resting platelets, both are strongly upregulated upon platelet activation. The physiological effects of complement–platelet crosstalk on the defensive actions of complement or on platelet clearance remains debated, but current observations suggest that tempered complement activation on activated platelets by effector-mediated stimulation serves to sustain or amplify platelet activation³¹.

Coordination with other immune pathways

Complement activation products, particularly C3a and C5a, can modulate the activation state of most immune cell types, including neutrophils, monocytes, macrophages, and dendritic cells^19,36,37 (FIG. 2). For example, a positive feedback mechanism between complement and neutrophil activation has been described in vitro: C5a can activate neutrophils to secrete properdin that, in turn, activates complement on released neutrophil extracellular traps to generate more C5a. This perpetuation of neutrophil-derived inflammation signals might enhance defence mechanisms but also have pathological consequences in diseases such as antineutrophil cytoplasmic autoantibody (ANCA)-associated vasculitis (see below)^38,39. By contrast, C3a-mediated signalling demonstrated a protective effect in a mouse model of ischaemia–reperfusion injury (IRI), constraining neutrophil mobilization in response to injury⁴⁰. Complement has also been linked to the development and recruitment of myeloid-derived suppressor cells (MDSCs), which are increasingly recognized as important players in the immune response to cancer or transplants. For example, the induction of MDSC development by hepatic stellate cells in cultured murine dendritic cells was strongly dependent on C3, and addition of exogenous iC3b to dendritic cell cultures influenced MDSC differentiation⁴¹. In the tumour microenvironment, complement activation leads to the recruitment and/or activation of MDSCs in a C5a-dependent manner, as shown in a syngeneic cervical cancer model⁴².

Overwhelming and inappropriate activation

Although much of the interest in complement-associated diseases has been driven by the identification of mutations in complement components by GWAS, in many clinical conditions complement is fully functional. However, as seen in conditions related to systemic inflammatory response syndrome (SIRS), too much of a good thing can ‘turn bad’. SIRS occurs when defence systems are suddenly presented with an overwhelming amount of PAMPs or DAMPs. As an early warning system and immune mediator, the induction of an intense complement response fuels hyperinflammatory events, which can be more devastating than the initial trigger. This fateful reaction has been described in sepsis but also after severe trauma, burns and in other SIRS-associated conditions. Numerous studies have identified complement as a key contributor to hyperinflammation in sepsis, with C5a in particular acting as an adverse effector through functions ranging from the release of cytokines and reactive oxygen species to consumptive coagulopathy, which contribute directly to lethality^81,82. Genetic ablation or pharmacological blockade of C5a signalling or C3 activation improves outcomes in animal models of sepsis^81,83,84. Moreover, in vitro and in vivo studies have provided evidence of synergistic effects of blocking complement and TLRs simultaneously^85,86, underscoring the crosstalk that occurs between these systems in pathological events. In patients with severe trauma, the presentation of DAMPs after tissue injury can trigger complement activation, which induces SIRS and contributes to complications such as multi-organ dysfunction and/or death^87–89. Given their simultaneous activation at the site of injury, crosstalk between complement, coagulation pathways and TLRs might be particularly relevant in this setting⁸⁹. Interestingly, findings from animal studies suggest that complement might also influence the repair process after trauma by inducing osteoclastogenesis and contributing to wound healing^89,90. A 2015 study of spinal cord injury in mice showed that early inhibition of C5aR1 during the acute phase of injury improved clinical outcomes, whereas extended blockade during the chronic phase of injury did not, indicating dual roles for complement in inflammation and repair⁹¹.

The presence of autoantibodies in the circulation and on self-surfaces can also trigger inflammation and tissue damage through inappropriate complement activation. For example, immune complexes present in systemic lupus erythematosus (SLE) cause consumption of complement components in the circulation and deposition of complement activation products in the kidney (lupus nephritis) and other organs⁹². Similar mechanisms are involved in antiphospholipid antibody syndrome, myasthenia gravis, and many other autoimmune diseases^3,92. In patients with ANCA-associated vasculitis, neutrophil priming by C5a or other stimuli leads to the expression of proteins that are recognized by autoantibodies. This recognition further activates complement, immune cell recruitment, and neutrophils, causing a vicious cycle that induces inflammation and tissue damage to blood vessels, with the kidney being among the most affected organs⁹³. Finally, the interaction of circulating natural antibodies with neoepitopes exposed during tissue damage can trigger antibody-mediated complement activation and contribute to conditions such as IRI⁹⁴. Conversely, autoantibodies against complement components can directly modulate the complement response. Some examples, such as anti-FH antibodies that impair the surface-recognition properties of FH in atypical haemolytic uraemic syndrome (aHUS), or C3/C4 nephritic factors that stabilize assembled convertases and contribute to C3 glomerulopathies, are discussed in further detail below^59,95.

Ineffective removal of debris

Another mechanism by which complement is inadvertently activated is through the failed removal of debris (FIG. 3), which is thought to contribute to the progression of age-related diseases. Ageing, lifestyle factors, and oxidative stress can lead to the formation of deposits and/or plaques that are sensed by PRRs, and the involvement of complement in this process is complex. During early stages of disease onset, the complement system has been attributed important roles in the clearance of debris and prevention of disease progression. Once debris begins to accumulate and cannot be efficiently cleared, however, perpetual activation of the complement cascade induces inflammation and tissue damage that drives degenerative processes. In models of Alzheimer disease, C1q and C3 exert neuroprotective effects via opsonic clearance of amyloid-β aggregates, whereas C1q-mediated complement activation by accumulating amyloid-β fibrils increasingly generate C5-derived effectors that maintain a focal inflammatory state, thus exacerbating disease progression^96,97. Similar complement-related mechanisms have been described in other neurological and neurodegenerative diseases, as well as in atherosclerosis^96–100. Another prominent example of a disease of ageing that is mediated by complement is age-related macular degeneration (AMD), a chronic degenerative eye disease, in which complement activation contributes to the geographic atrophy and choroidal neovascularization that causes loss of vision¹⁰¹. Although the aetiology of AMD is complex, the presence of complement components in lipoprotein deposits (known as drusen) often observed in the early stages of AMD suggests that insufficient clearance and accumulation of debris leading to activation of complement is an important factor^2,101.

Misguided activation by foreign materials

Finally, complement can be inappropriately activated by the biomaterials and foreign cells that are increasingly used in modern medicine (FIG. 3). Polymers and metal surfaces present in haemodialysis filters, extracorporeal circuits, oxygenators, implants, and drug delivery devices can be recognized as foreign intruders by complement, which induces an inflammatory response that can affect the functionality of the device and patients’ quality-of-life^2,102. The mechanisms of complement activation by biomaterials are diverse and material-dependent but often involve the adsorption of plasma proteins such as albumin or immunoglobulins, which bind and activate recognition molecules or complement components such as C3 (REF. 102). Cell and organ transplants, in contrast, typically express antigenic epitopes that trigger complement activation via antibody-dependent routes⁶⁰. Foreign-body reactions in the context of kidney transplantation and haemodialysis are discussed in further detail below.

Missing and dysfunctional complement components

Whereas the above-described disease mechanisms are mainly attributed to excessive complement activation in the presence of an intact complement system, missing or dysfunctional components can induce or exacerbate an entirely new set of problems. Primary deficiencies have been reported for many key components of the complement system, but the frequencies and clinical impact of these deficiencies are diverse¹⁰³. Deficiency of early complement components is commonly associated with autoimmune and immune complex-mediated diseases, with more than 90% of patients with deficiencies in C1 proteins developing SLE-like disorders⁹². Impaired scavenging and clearance of apoptotic cells and immune complexes are thought to be major causal factors. Moreover, since complement participates in the elimination of self-reactive lymphocytes, deficiencies in complement components might favour the production of autoantibodies by affecting B-cell tolerance⁹². Interestingly, and despite a prominent role for C3 opsonization in ‘waste disposal’ (that is, clearance of debris and apoptotic cells), C3 deficiencies are not strongly associated with autoimmune disease. These paradoxical roles of C1q and C3 have been explained by the observation that, whereas C1q primarily facilitates phagocytic uptake, C3 opsonins also mediate intracellular trafficking of apoptotic cargoes and enhance endocytic processing of antigens that are required for generating potentially adverse T-cell responses⁹.

Interestingly, especially in view of the intricate effector roles of C5a and MAC, primary deficiencies in terminal components (C5 to C9) seem to have limited clinical consequences, most prominently an increased risk of meningococcal infection. Similarly, C3 deficiency is mostly associated with increased susceptibility to certain pyogenic infections, such as those caused by meningococci, pneumococci and haemophilus influenzae. Finally, deficiencies in complement regulators such as FH or CD46, though rare, are typically associated with kidney diseases¹⁰³. Paroxysmal nocturnal haemoglobinuria (PNH) is an ultra-rare but severe haematological disease in which somatic mutations in the genes responsible for glycosylphosphatidylinositol anchor synthesis lead to clonal populations of blood cells that lack several membrane proteins, including CD55 and CD59. As a consequence, PNH erythrocytes show increased susceptibility to complement-mediated haemolysis, for example, resulting from bystander activation during infection, with consequential anaemia and high thrombotic risk¹⁰⁴.

More common than deficiencies are genetic variations in complement components that define the functionality of the complement system and often represent a tipping point if and when clinical issues manifest. In general, gain-of-function mutations in activator components (for example, in C3) or loss-of-function mutations in regulators (for example, in FH) predispose individuals to disease susceptibility¹¹. The advent of GWAS in particular has led to a surge in newly identified mutations and polymorphisms that correlate with disease. In February 2016, the FH aHUS mutation database¹⁰⁵ listed 1,146 genetic alterations in C3, FH, FI, and CD46 alone, some 400 of which were disease-related (mostly aHUS but also AMD and other conditions). Despite this wealth of information, the functional consequences of distinct genetic variations are only beginning to be understood. For example, a strong correlation between the common Tyr402His variation in FH and risk of developing AMD was described in 2005 (REF. 106), yet the exact pathological process is not yet clear. Current data indicate that the Try402 and His402 variants of FH mainly differ in their binding to self-pattern and oxidative stress markers such as malondialdehyde¹⁰⁷.

Several of these genetic alterations might have only a small effect on the function of the affected component but might have considerable effects on the complement system as a whole by exacerbating an imbalance in complement activation, particularly in chronic conditions such as AMD. Furthermore, an increasing prevalence and diversity of complement mutations and polymorphisms increase the chance that an individual will carry a distinct set of variants, or ‘complotype’, which defines the reactivity of the complement system and an individual’s susceptibility to complement-related disease¹¹. These concepts were underscored by studies showing that the combination of polymorphisms in C3, FB, and FH with weak individual consequences could result in a profound shift in complement activity¹⁰⁸. Perhaps most intriguingly, mutations and/or polymorphisms, dysfunctions, and deficiencies in complement proteins are expected to lead to a systemic imbalance that can affect all cells and tissues, yet clinical manifestations are primarily detected in particular organs, such as the eye and, even more prominently, the kidneys.

aHUS and other thrombotic microangiopathies

The strong connection between complement, coaglation pathways and endothelial cells and the consequences of imbalances therein are particularly obvious in thrombotic microangiopathies (TMAs). Although diseases with distinct aetiologies such as aHUS and thrombotic thrombocytopenic purpura (TTP) fall under the TMA spectrum, common features define their pathology, including microangiopathic haemolytic anaemia, thrombocytopenia, vascular damage with thrombosis, and organ injury^114,116,117. Essential insights into TMA disease pathogenesis have been acquired from studies of aHUS, a rare kidney disorder associated with imbalanced complement activation, which results in damage to the kidney endothelium and thrombus formation. Notably, aHUS is strongly associated with polymorphisms and mutations of activators (C3 and FB) and regulators (CD46, FH, and FI) of the ‘alternative pathway’ (REF. 118) (BOX 1). More than 300 aHUS-associated genetic alterations are listed in the FH aHUS mutation database to date, with mutational hotspots primarily in the C-terminus of FH and the C3d-domain of C3 (REF. 105). Variations in C3 and FB enhance convertase-mediated generation of C3b, whereas alterations in genes that encode regulators of complement disturb the control of convertase activity and C3b deposition on renal endothelium which, as discussed previously, strongly relies on fluid-phase regulators. Importantly, the devastating effects of imbalanced complement activation on the renal endothelium are primarily mediated by excessive opsonization and formation of C5b–9 on the cell surface, rather than uncontrolled activation of the fluid phase¹¹⁹. Alongside activation of endothelial cells and platelets by C5b–9, the anaphylatoxins C3a and C5a induce a local proinflammatory environment that promotes the formation of microthrombi in the renal micro-vasculature¹¹⁴. The release of haeme during haemolytic periods might act as secondary ‘hit’ by further inducing complement activation, since haeme has been shown in vitro to enhance C3 activation, decrease CD55 and CD46 expression on endothelial cells, and increase the expression of P-selectin as a ligand for C3b¹²⁰.

As a disease essentially conditioned by the excessive generation of C5-derived effectors, aHUS is successfully treated by the anti-C5 antibody eculizumab¹²¹. A longitudinal study in which patients with aHUS were treated for up to 1 year demonstrated that eculizumab normalized systemic levels of C5a and C5b–9 and markers of renal injury and endothelial damage; however, eculizumab decreased but did not normalize levels of the complement marker Ba (an FB fragment released during convertase formation) or soluble VCAM-1, a marker of endothelial cell activation¹²². Although this finding reflects ongoing complement activation despite blockage of C5, the implications for disease progression are not yet clear; therapeutic intervention at upstream levels, for example, by inhibiting MASP-2 are currently being evaluated¹²³.

Although the involvement of complement in aHUS pathogenesis is well-established, studies over the past few years suggest that complement has contributory roles in other TMA diseases^114,124. Typical HUS is induced by Shiga toxin-producing Escherichia coli (STEC), and increased complement activation has been linked to Shiga toxin-induced expression of the proposed C3b ligand P-selectin on endothelial cells¹²⁵. Beneficial effects of eculizumab in STEC-induced HUS were reported in connection with the STEC outbreak in Germany in 2011, but the effect of this treatment on HUS remains controversial^114,124,126. TTP is caused by decreased activity of ADAMTS13, a metalloproteinase that cleaves vWF. The influence of complement on TTP is not well defined, although the interaction between FH and vWF oligomers, endothelial damage, and microvascular thrombus formation might lead to secondary complement activation^33,114,124. Thus far, complement-targeted therapies have only been explored in one case study of TTP with favourable outcome^114,127. Of note, however, the clinical distinction between aHUS and TTP remains challenging in some cases, and individual patients may have clinical signs of both TTP and aHUS^128,129; more research is needed to elucidate the effect of complement on TMA beyond aHUS.

C3 glomerulopathy

The term C3 glomerulopathy, introduced in 2010 as consensus nomenclature¹³⁰, refers to a group of renal pathologies characterized by detection of C3 opsonins in glomeruli, with minimal or no deposition of immunoglobulins. These forms of renal disease are typically caused by abnormal control of the alternative pathway of complement activation, in contrast to forms of immunoglobulin-associated membranoproliferative glomerulonephritis, which can be caused by autoimmune reactions or infection and are characterized by strong IgG or IgM staining alongside C3 and C1q. C3 glomerulopathies can be subdivided into dense deposit disease (DDD) and C3 glomerulonephritis on the basis of histological observations^95,130,131. C3 glomerulopathies develop as a consequence of imbalanced complement activation and lead to proteinuria and haematuria, with progression to renal failure within 10 years of diagnosis in approximately 50% of patients. The high mortality is a consequence of the lack of effective therapy and poor prognosis after kidney transplantation, with disease recurrence being observed in over 50% of all cases¹³².

In contrast to aHUS, in which endothelial cell damage and local deposition of complement fragments are considered driving factors, uncontrolled activation and consumption of complement in the fluid phase leading to bystander opsonization of ill-protected surfaces (such as the GBM) is considered the predominant mechanism in C3 glomerulopathy. The reasons for this hyperactivation of complement are diverse and can include autoimmune mechanisms, particularly the presence of C3/C4 nephritic factors that stabilize C3 convertases, and genetic alterations in C3, FH, FI and/or FHR proteins. Whereas C3 nephritic factor is associated with chronic activation of the amplification loop by stabilizing C3 convertases, a heterozygous C3 mutation associated with DDD was found to impair the FH-mediated degradation of C3b by FI¹³³. Similarly, genetic variations in FH and FI observed in patients with C3 glomerulopathy result in the loss of regulatory functions. In contrast to aHUS-related mutations and polymorphisms in FH, which typically affect the surface-recognizing C-terminus, those associated with C3 glomerulopathy generally occupy a focal area in the regulatory domains. Genetic alterations in the region encoding FHR proteins, including point mutations, polymorphisms, deletions, and rearrangements, have been identified in patients with C3 glomerulopathy. The pathologic mechanisms by which alterations in FHR proteins affect complement activation remain elusive, but likely involve deregulation and other modulatory effects on the complement system⁹⁵ (BOX 2). The increased complement turnover generates C5a that is recognized as a promoter of necrotic and inflammatory lesions in the glomeruli, which progress to renal fibrosis and impairment of kidney function¹²¹.

The vast diversity of pathologic mechanisms that underlie C3 glomerulopathies renders diagnosis and classification challenging, and has implications for the development of treatment strategies. Indeed, whereas anti-C5 therapy has developed into a gold standard for the treatment of aHUS, only select patients with C3 glomerulopathies will benefit from the same treatment^126,134. Given the central involvement of excessive amplification loop activity in most patients, strategies that target C3 and/or the convertases are currently under investigation.

Transplantation and IRI

Many kidney disorders will eventually progress to end-stage renal disease (ESRD) requiring haemodialysis and/or kidney transplantion²³. Although transplantation is the desirable treatment option for patients with ESRD, organ shortages and compatibility issues limit the number of transplantations performed worldwide. Complement has rapidly moved into the spotlight of transplantation medicine, with involvement in most complications that can arise during the procedure⁶⁰. In deceased organ donors, complement activation occurs shortly after brain death or cardiac arrest. Clinical studies have shown a direct correlation between the presence of complement activation fragments in the plasma of deceased donors and poorer post-transplantation outcome, suggesting that inhibition of complement and coagulation pathways in the donor could result in decreased tissue damage and better transplant outcomes¹³⁵.

In addition, the inevitable period of cold ischaemia during organ collection and storage induces hypoxia, inflammation, and tissue damage that contribute to rejection episodes and graft loss. During reperfusion, ischemia-induced DAMPs such as histones, heat-shock proteins, and biglycan activate TLRs and provide neoepitopes for natural antibodies, which subsequently activate complement⁵⁶. The increased expression of complement proteins, the loss of glycocalyx and complement regulators, and the damage and detachment of endothelial cells as a consequence of acute kidney injury can exacerbate complement activation on tubular epithelial cells, contributing to the production of proinflammatory cytokines⁶⁰. Interestingly, whereas mice that lack FB are protected from kidney IRI, Fb/Tlr2 double-knockout mice are not, suggesting counter-regulatory effects between complement and TLR2 during IRI¹³⁶. While complement-related tissue damage during IRI is mainly dependent on C5-derived effectors with recruitment and activation of neutrophils by C5a and cell activation and damage by C5b–9, the initial triggers of complement activation in kidney IRI remain elusive⁶⁰. In addition to antibody-mediated activation, components of the lectin pathway are increasingly identified as inducing factors. For example, a 2016 study showed that collectin-11 can recognize an abnormal L-fucose pattern on postischaemic renal tubule cells and activate complement via MASP-2¹³⁷. Independent of the mechanisms, acute kidney injury caused by complement and other inadvertently triggered defence systems during donor death and IRI affects the quality of the graft and largely contributes to the rejection risk in transplantation.

The prominent role of complement in cell-mediated and antibody-mediated transplant rejection has long been recognized, with increased levels of C5a and soluble C5b–9 commonly detected in the plasma and urine of transplant recipients during rejection episodes^60,138. Moreover, C4d-containing opsonins are among the most reliable biomarkers of rejection in kidney biopsy samples¹³⁹. Several mechanisms link complement to kidney transplant injury, and local complement production is of particular importance in this context. Tubular epithelial cells are the main target but also a major source of complement in the kidney undergoing rejection, yet graft-infiltrating cells also contribute to local complement production⁶⁰. In crossover transplantation studies in mice, allograft rejection was entirely dependent on complement production by the donor tissue; transplantation of kidneys from C3-knockout mice into wild-type mice was successful without any requirement for immunosuppression¹⁴⁰.

The role of complement in cellular rejection seems to be complex and some aspects remain controversial. The interaction between APCs and alloreactive T cells leads to the secretion of complement proteins and local generation of C3a and C5a; signalling through C3a and C5a receptors on APCs, and potentially T cells, amplifies the T_H1-cell response and graft rejection. Activation of C3aR and C5aR1 on APCs and CD4⁺ T cells sends proliferation, co-stimulatory, and survival signals for T cells; blockage of C3a and C5a-induced responses results in the endogenous production of TGF-β by APCs and T cells¹⁴¹. CD4⁺ T cells with a regulatory phenotype suppressed immune responses in a mouse model of graft-versus-host disease, providing proof-of-principle of the benefits of complement inhibition during transplantation^141,142.

As improvements in immunosuppression have decreased the clinical impact of cellular rejection, attention has increasingly turned to antibody-mediated rejection (ABMR), which remains associated with high rates of graft loss. Complement integrally participates in ABMR by causing direct tissue damage, regulating the production of donor-specific antibodies (DSAs) in B cells, potentiating the injury caused by antibodies, and fostering thrombo-inflammatory complications. In fact, the combination of serum DSAs and positive C4d staining in the allograft is used as a criterion to define ABMR in renal allografts. The binding of antibodies to the allograft strongly activates complement, resulting in the generation of C5a and C5b–9, which activate neutrophils and endothelial cells, among others. Although a rare occurrence as a result of improved crossmatching, complement-mediated damage is particularly pronounced in hyperacute rejection due to the presence of pre-formed antibodies. Of note, however, complement also contributes to the induction of DSAs as opsonization with iC3b and C3dg decreases the threshold of B-cell activation, as described above, and complement activation is required for the alloantigen-induced priming of B cells¹⁴³. Crosstalk between complement, coagulation, and FcR-mediated processes further potentiates the pro-coagulant and inflammatory responses, leading to endothelial damage, thrombosis, and graft loss¹⁴⁴.

An area that has gained interest is the potential link between complement inhibition and graft accommodation through the sustained protection of endothelial cells from DSAs and complement. In monkeys that had been pre-sensitized with skin grafts, depletion of complement induced long-term survival of a subsequently transplanted kidney¹⁴⁵. In this model, graft accommodation was associated with upregulation in the expression levels of complement regulatory proteins and anti-apoptotic genes. Many aspects of accommodation as an ‘acquired resistance’ to complement-mediated injury of endothelial cells remain elusive and might involve modulation of surface-regulation and cellular responses. Given the central involvement of complement in most aspects of transplant-related complications, and the potential prospect of allowing transplantation across ABO, HLA or species barriers, therapeutic complement inhibition is currently considered a promising strategy^60,146.

Haemodialysis-related complications

Patients awaiting kidney transplantation and other patients with impaired kidney function must undergo dialysis. While essential, this procedure exacerbates clinical complications such as muscle wasting, endothelial and immune cell dysfunction, malnutrition, and anaemia; it also increases the risk of cardiovascular disease, which represents a major cause of mortality^23,147.

During the contact of blood with the haemodialysis filters and extracorporeal circuits, the exposed biomaterials are recognized as foreign surfaces and activate complement. Although this effect was most obvious for older cellulose-based haemodialysis filters, modern synthetic filters with improved biocompatibility still induce substantial complement activation through all initiation pathways^3,147. The generation of C5a during haemodialysis-induced complement activation is associated with neutrophil activation and an increase in TF expression as a primary initiator of coagulation. Indeed, patients on haemodialysis have an elevated risk of thrombosis and exhibit increased expression of oxidative stress markers and proinflammatory cytokines — factors that are all strongly correlated with mortality in this population^147,148. The impact of haemodialysis-induced complement activation on overall inflammation, or even mortality, is difficult to assess given the complexity of the underlying diseases and treatments. Nevertheless, repetitive, chronic–acute inflammatory and complement-mediated triggers are likely to contribute to disease progression and/or affect quality of life. Further research into the biocompatibility of haemodialysis filters and potential treatment options might open new avenues for decreasing the burden of haemodialysis.