Protection of host cells by complement regulators

Summary

The complement cascade is an ancient immune-surveillance system that not only provides protection from pathogen invasion but has also evolved to participate in physiological processes to maintain tissue homeostasis. The alternative pathway (AP) of complement activation is the evolutionarily oldest part of this innate immune cascade. It is unique in that it is continuously activated at a low level and arbitrarily probes foreign, modified-self, and also unaltered self-structures. This indiscriminate activation necessitates the presence of preformed regulators on autologous surfaces to spare self-cells from the undirected nature of AP activation. Although the other two canonical complement activation routes, the classical and lectin pathways, initiate the cascade more specifically through pattern recognition, their activity still needs to be tightly controlled to avoid excessive reactivity. It is the perpetual duty of complement regulators to protect the self from damage inflicted by inadequate complement activation. Here, we review the role of complement regulators as preformed mediators of defense, explain their common and specialized functions, and discuss selected cases in which alterations in complement regulators lead to disease. Finally, rational engineering approaches using natural complement inhibitors as potential therapeutics are highlighted.

1 | THE COMPLEMENT SYSTEM AT A GLANCE

In the late 19th century, the complement system was described as a heat-labile component of serum that “complements” antibodies in killing bacteria.¹ Since then, our collective knowledge concerning this innate immune cascade has increased dramatically. Some 50 soluble or cell membrane-tethered complement proteins comprise the cascade, including effector molecules, receptors and regulators.^2,3 The three main complement activation pathways, i.e. the classical, lectin, and alternative pathways (CP, LP and AP, respectively), merge into a common terminal pathway (TP) that mediates major effector functions. Activation of the CP and LP begins with the recognition of pathogens or danger signals by various pattern recognition molecules (PRM), which are oligomeric protein complexes of a bouquet-like structure. These recognition processes, which usually require multivalent interactions, trigger conformational changes within the PRM complexes, leading to sequential proteolytic activations of zymogens that are consecutively activated in a cascade-like fashion (Fig. 1).

FIGURE 1.

Sketch of the complement cascade. The complement cascade can be triggered by one of the three canonical activation pathways, the lectin (LP), the classical (CP), or the alternative pathway (AP). The central step in the three cascades is the activation of C3 into the anaphylatoxin C3a and the opsonin C3b through bimolecular C3 convertases (the proximal complement cascade). In the presence of complement regulators with cofactor activity, the plasma protease Factor I proteolytically inactivates C3b molecules into the late-stage opsonins iC3b and C3dg. If C3b inactivation does not proceed at sufficient speed, the AP-amplification loop provides a positive feedback mechanism. C3b molecules assemble more C3 convertases, followed by the production of more C3b molecules, thus amplifying the initial trigger many-fold. Newly produced C3b molecules can associate with the bimolecular C3 convertases of the proximal pathway to form the tri-molecular C5 convertases, marking the transition to the terminal (lytic) pathway. C5 convertases cleave, and thus activate, C5 into the anaphylatoxin C5a and C5b. Together with other complement components in serum, C5b builds membrane attack complexes. C5a is a potent anaphylatoxin and sends inflammatory signals to many different cells. Several other regulators efficiently control the cascade at different steps, including the proteins that cause the decay of the convertases. For simplicity, only cofactor activity is shown in this scheme

Traditionally, only immune complexes or foreign carbohydrate moieties, respectively, were thought to activate the CP or LP. However, their PRM can also directly recognize danger moieties on modified host surfaces such as apoptotic cells or damaged tissues. Danger recognition initiates several proteolytic activation events cumulating in the formation of C3 convertases of the CP and LP. These convertases perform the central task common to all activation pathways, i.e. the proteolytic activation of the abundant plasma protein C3 into its effector fragments C3a and C3b. Liberated C3a acts as anaphylatoxin that exerts diverse signaling functions through the C3a receptor (C3aR). In the bigger split product, C3b, a reactive thioester becomes exposed upon convertase cleavage, enabling covalent attachment of this protein to nearby surfaces. Surface-attached C3b and its proteolytically processed products iC3b and C3dg remain surface-tethered via the thioester and act as opsonins that mark the C3b/iC3b/C3dg-decorated surfaces for removal by phagocytosis.

In the absence of adequate regulation, CP/LP-produced C3b molecules can self-amplify through a positive feedback loop of the AP; in a process that is commonly termed the “AP amplification loop,” factor B (FB) binds to deposited C3b and becomes activated by factor D (FD) to form the AP C3 convertase complex (C3bBb), which cleaves additional C3 to C3b. Of note, AP C3 convertases may be further stabilized by the positive AP regulator properdin and thus promote complement activation.⁴

Accumulation of C3b to high densities favors the proteolytic activation of C5, which marks the initiation of the TP.^5,6 This activation is carried out by C5 convertases that process C5 into the potent anaphylatoxin C5a and the bigger split product C5b, which initiates the assembly of the membrane attack complex (MAC). The MAC can damage cells via insertion into their membranes, potentially leading to the formation of lytic pores. Thus, TP activation has the greatest inflammatory potential of the entire complement cascade.

Not merely a feedback loop for the amplification of CP/LP-produced C3b, the AP also constitutes a distinct initiation pathway. The essence of the arbitrary nature of the AP and its activation lies in the structural organization of the central, thioester-containing molecule C3, which functions by analogy to early immune strategies observed in Cnidaria species such as sea anemones and corals.⁷ Native C3 assumes a conformation that shields its internal thioester from activation and subsequent reactions with proteins or carbohydrates on target surfaces⁸; however, as part of the so-called “tick-over” mechanism, spontaneous hydrolysis of the thioester in C3 occurs continuously in a small fraction of C3 molecules. This process produces C3(H₂O) molecules that structurally and functionally resemble the proteolytic C3 activation product C3b.⁹ In comparison to C3, C3(H₂O) has undergone conformational rearrangements to expose its thioester, which, within a very short timeframe, can react with nucleophiles on nearby surfaces (Fig. 2A). Failure of individual C3(H₂O) molecules to attach to surfaces leads to the presence of soluble C3(H₂O) (or C3b) molecules that circulate in the fluid phase.^10,11 It is generally thought that such conformation-driven auto-activation of C3 occurs spontaneous and arbitrarily, although contact-triggered C3 activation has been described for some surfaces.¹¹

FIGURE 2.

Activation and regulation of C3. (A) Activation of C3. The top left sketch depicts C3, highlighting the anaphylatoxin domain C3a (ANA) in red and the thioester domain (TED) in black, with the thioester (yellow) being shielded. Enzymatic cleavage of the C3a anaphylatoxin domain by the C3 convertases of the CP/LP or AP enables a fast conformational transition of TED and facilitates the reactiveness of the thioester. C3 autoactivation via the tick-over mechanism also leads to conformational changes and exposure of the TED, but the presence of the anaphylatoxin domain is thought to sterically hinder this conformational transition, providing one potential explanation for the fact that only a small proportion of the C3 molecules auto-activates. (B) Regulation of complement convertases by decay-accelerating factors. The decay accelerator binds the convertase constituents and displaces the enzymatic component of the convertase Bb or C2a. Bb or C2a cannot re-associate with C3b to form new convertases, because they are missing the Ba or C2b portion, which had been cleaved off during the assembly of the convertases from the respective zymogen precursors Factor B and C2 by other activating proteases of the complement cascade. (C) Cofactors are essential to allowing recruitment of the regulatory protease Factor I to C3b and C4b. Proteolytic inactivation by Factor I only occurs in presence of cofactors. The C3b and C4b split products can no longer assemble convertases

Foreign and host surfaces are considered similarly susceptible to constant probing by the AP through the deposition of C3(H₂O) and/or C3b. It is therefore essential to have powerful control mechanisms in place to discriminate between various threat levels and shape the complement response accordingly. In the presence of danger patterns and the absence of efficient regulation, complement marks affected particles with opsonins for phagocytic removal, assembles the cell-damaging MAC, and alarms the body through the release of the inflammatory and chemoattractant mediators C3a and C5a, which recruit phagocytes to the site of activation.¹² Induction of the same effector responses in host tissue would impose a serious threat to the host and disrupt body homeostasis, and therefore these responses need to be actively prevented.

2 | CONTROLLING COMPLEMENT: THE IMPORTANCE OF COMPLEMENT REGULATORS

Healthy host tissues counteract complement-mediated damage by relying on an arsenal of efficient regulators that are either membrane-anchored or circulate as soluble proteins. The soluble components can be equipped with a host surface-homing mechanism that biases complement regulation toward self-structures over foreign surfaces. The constant arbitrary tick-over activation within the complement cascade marks a potential hazard to any surface. Preformed complement regulators on self-cells ensure that, even after initial opsonization, host surfaces are not affected by the actions of the AP amplification loop. This escape of self-surfaces from the consequences of the indiscriminate AP constitutes a phylogenetically ancient immune mechanism that was established several hundred million years ago.⁷

One of the major tasks of complement regulators is to rapidly inactivate any C3b molecules that have arbitrarily bound to unaltered host cells, before these C3b molecules can feed into the AP amplification loop. On altered host surfaces such as apoptotic cells, pattern recognition via the CP/LP leads to limited C3b deposition as a result of the presence of a reduced set of complement regulators that control convertase activity and help process deposited C3b molecules to iC3b and C3dg. These “late-stage” opsonins are only inactive with respect to the further formation of C3 convertases, and they still have important biological roles to play by promoting phagocytosis through binding to complement receptors on phagocytes and mediation of adaptive immune responses. Thus, thanks to the moderating action of complement regulators, iC3b/C3dg-opsonized apoptotic bodies are efficiently engulfed in a non-inflammatory manner by phagocytes, without large-scale proinflammatory TP activation.

Thus, the preformed regulators of complement activation bear the burden of protecting healthy self-structures from continuous undirected probing by the complement system. Strict regulation stops C3 deposition on normal host tissue through arbitrary activation, while on altered self-structures, limited complement activation is permitted, involving a residual regulation that results in moderate C3 opsonization and silent phagocytic clearance without a transition to the inflammatory TP. In contrast, pathogenic intruders are ideally met with a powerful and unregulated response of the complement cascade characterized by large-scale C3 opsonization, anaphylatoxin release, and MAC formation. Yet, contrary to these ideal scenarios, and as discussed in more detail later, pathogens have developed a diverse range of strategies to evade complement attack,¹³ and dysregulation of complement can contribute to autoimmune and inflammatory diseases.¹⁴ A profound understanding of the regulators and mechanisms involved in the control of complement activity is therefore essential if we are to elucidate disease processes and devise novel therapeutic strategies.

3 | THE RE GULATORS

Complement regulation needs to control all the major checkpoints of activation, amplification, and resulting effector functions and to cover both fluid-phase and surface activation. This need is reflected in the number and diversity of complement regulators, which encompass various circulating and membrane-bound proteins. Complement is considered a surface-directed effector pathway, and its regulation on cells and tissues therefore has received particular attention. However, poorly controlled fluid-phase activation can lead to exuberant C3b production by circulating C3 convertases, resulting, in the worst case, in C3 depletion and TP activation. Activation of the TP can cause bystander damage when soluble MAC complexes “spill over” to (host) surfaces. Soluble regulators are therefore critically needed to prevent such consumptive activation of complement components, to protect host cells from bystander activation. However, complement control by soluble regulators is not limited to the fluid phase, and some specialized fluid-phase regulators are equipped with targeting mechanisms that facilitate binding to self-surface markers on host cells. This mechanism generally reinforces complement control by membrane-anchored regulators on host tissue, but it is considered especially important for non-cellular host surfaces such as proteoglycan-rich basement membranes on fenestrated endothelium.

In contrast to the fluid-phase regulators, the effects of the membrane-anchored regulators are typically restricted in terms of their respective expression and tissue distribution. Although membrane-detached ectodomains of certain complement receptors/regulators can be detected in the circulation, their concentrations are very low, and their contribution to fluid-phase complement control may be marginal. An interesting exception to the strict surface vs fluid-phase distinction is the presence of CR1 (CD35) on erythrocytes: The vast number of erythrocytes in circulation can distribute complement regulation throughout the body, rather than being confined to a local tissue.

Because of the central role of the C3 and C5 convertases in driving complement activation, amplification and effector generation, convertase function and the resulting opsonization are particularly well controlled and involve a specialized set of regulators. In this review, we have divided the discussion on complement regulators into those non-convertase- directed mechanisms that target the initiation and effector steps of the cascade (Table 1) and those mechanisms that regulate the activities of the central C3 and C5 convertases (Table 2).

TABLE 1.

Non-convertase-directed regulators of complement activation

Regulator	Function	Regulated pathway	Main regulatory compartment
C1 inhibitor (C1-INH)	Inactivates C1r and C1s, MASP-1 and MASP-2	CP/LP	Fluid
sMAP	Binding to MBL, competition with MASPs	LP	Fluid
MAP-1	Binds to MBL/ficolins	LP	Fluid
Vitronectin (S protein)	Binds to C5b-7/8/9	TP (MAC formation)	Fluid
Clusterin	Binds to C5b-7/8/9	TP (MAC formation)	Fluid
CD59	Binds to C8 and C9	TP (MAC formation)	Surface

TABLE 2.

Convertase-directed regulators of complement activation

Regulator	Regulatory activity		Regulated pathway	Main regulatory compartment
	Decay	Cofactor
CR1	DAA	CA			CP/LP & AP	Surface
DAF	DAA	–	CP/LP & AP	Surface
MCP	–	CA	CP/LP & AP	Surface
C4BP	DAA	CA	CP/LP	Fluid/surface
Factor H	DAA	CA	AP	Fluid/surface
FHL-1	DAA	CA	AP	Fluid/surface
Factor I	Protease for degradation of C3b or C4b in presence of a cofactor		CP/LP & AP	Fluid (on surface only in conjunction with cofactor)

4 | NON-CONVERTASE-DIRECTED REGULATION

Although the convertases act as the “motor” of complement activation, their actions are typically initiated by the sensing activities of the CP/LP, and they produce effectors that may potentially harm not only the intended target but also heathy cells. Therefore, both CP/LP initiation and the TP effector arm need to be carefully controlled.

4.1 | Regulation of initiation pathways

At least three proteins regulate the initiation of the cascade by acting on the pattern recognition complexes of the CP and/or LP, which typically consist of a PRM (C1q, MBL, ficolins, collectins) and associated proteases (C1r/C1s and MASPs, respectively). All regulators acting at this step are soluble proteins with distinct mechanisms and specificities.

C1 inhibitor (C1-INH) is a secreted, heavily glycosylated single-chain plasma protein that impairs complement initiation via both the CP and LP. It is important to note that whereas the initiation of the LP is also controlled by other regulators (see below), C1-INH is the only known regulator that inhibits CP initiation.¹⁵ C1-INH is organized into a C-terminal serine protease inhibitor (serpin) domain and an N-terminal portion of unclear function and is mainly synthesized in the liver.¹⁵ As a member of the serpin family, C1-INH is a suicide inhibitor that presents to the protease a scissile peptide bond that matches its substrate specificity; after being cleaved by the respective protease, C1-INH remains irreversibly bound, thus permanently inhibiting the protease.¹⁶ Like other serpins, the inhibitory activity of C1-INH toward at least some of its targets is enhanced by negatively charged glycosaminoglycans (GAGs).¹⁷ C1-INH interacts with and inhibits the catalytic centers of activated C1r and C1s within the C1 complex of the CP, or interacts with MASP proteases within the MBL/ficolin/collectin complexes of the LP. In all cases, C1-INH limits the consumption of C2 and C4 after CP or LP activation.

However, the activity of C1-INH is not only restricted to complement proteases but also includes other important targets such as proteases of the fibrinolytic, clotting (e.g. activated factor XII, anti-thrombin III), and kinin pathways. Of note, C1-INH is the primary physiological inhibitor of the plasma kallikrein-kinin system.¹⁸ The importance of C1-INH for kallikrein regulation is exemplified by the clinical condition hereditary angioedema (HAE), which is caused by an absence or shortage of C1-INH function. C1-INH deficiencies are among the more common deficiencies observed in complement proteins,¹⁹ with the predominant type I deficiencies being caused by sequence alterations that affect proper secretion, and type II deficiencies being characterized by mutations that yield proteins with dysfunctional inhibitor activity. HAE is characterized by recurrent edema that affects the skin and mucosa; whereas edemas in other body parts are usually less serious, laryngeal edema can be life-threatening [reviewed in (20)].

Therapeutic measures based on C1-INH largely rely on reconstituting the inhibitor using plasma-purified or recombinant preparations, and several C1-INH-based drugs have reached the clinic.²¹ Although C1-INH preparations are currently only indicated for HAE treatment, their complement-directed activities make them potential options for complement-mediated or multifactorial conditions. For example, C1-INH has shown promising results in a model of antibody-mediated transplant rejection²² and is being evaluated in clinical trials for this and other indications.²¹

In addition to C1-INH, the initiation of the LP is also controlled by other proteins, including sMAP, MAP-1, and potentially anti-thrombin. sMAP (or Map19) is a 20-kDa, non-enzymatic splice variant of the MASP2 gene that mainly consists of the first CUB and the EGF-like domain found in MASP-2.²³ Thus, in comparison to the bigger MASP2 protein, sMAP lacks one CUB and two complement control protein (CCP) domains as well as the C-terminal serum protease domain. sMAP binds to MBL (and at least some ficolins²⁴) and is thought to compete with the enzymatically active MASP proteins for association with these PRM; however, the exact role of sMAP is still debated.^23,25

MAP-1 (or Map44) is a 45-kDa serum protein that binds to MBL and ficolins to inhibit the deposition of C4 by preventing the initiation of the LP.²⁶ The MAP-1 gene product derives from alternative splicing of the MASP1 gene. Compared to MASP-1, the splice product lacks one CCP domain and the complete serine protease domain but still contains the first CUB, the EGF-like and the second CUB domain, which are crucial for the binding of MASP proteins to MBL and ficolins. ²⁷ Not only is MAP-1 expressed by hepatocytes and found in serum, but high local expression of MAP-1 has also been detected in myocardial and skeletal muscle tissue, pointing to a potentially specific role of MAP-1 in these tissues.²⁶

4.2 | Modulation of anaphylatoxin effector functions

Anaphylatoxins, and in particular C5a, are highly potent inflammatory mediators and chemoattractants whose spatiotemporal activity needs to be controlled. In addition to rapid systemic clearance of these effectors, carboxypeptidase has a direct regulatory role. Carboxypeptidases quickly tame anaphylatoxin activity by cleaving the C-terminal arginine residue of C3a and C5a, generating their desarginated (“desArg”) forms. Although the desArg forms were traditionally considered “inactive” or at least of markedly reduced potency, recent data indicate that C5a-desArg maintains significant levels of cell signaling activity.²⁸ Furthermore, C5a-desArg binds to a second C5a receptor (C5aR2, C5L2) with high affinity. Although the exact role of these events remains to be fully explored, it appears that desArg anaphylatoxins are not completely inactivated but may instead exhibit distinct biological functions when compared to the non-regulated forms.

4.3 | Regulation of the terminal pathway

The initiation of the TP is marked by the splitting of C5 into C5a and C5b. C5b associates with components C6 to C8 and multiple molecules of C9 to form C5b-9 complexes that efficiently insert into membranes. Whereas C5b-7 complexes can already enter and disturb membrane lipid bilayers, the insertion is inefficient and not very stable without the binding of C8 and C9.²⁹ The recruitment of C8 and C9 to C5b-7 complexes therefore marks a critical step in efficient MAC formation and a critical point-of-action for TP regulators.

Vitronectin (or S protein; not to be confused with protein S) is a secreted glycoprotein of approximately 75 kDa that is mainly produced in the liver and found in the circulation at a concentration of approximately 400 μg/mL.^30–32 In addition, vitronectin is detected in many tissues. Recent data suggest that this glycoprotein, which contains an RGD integrin motif, acts as an adhesive protein in the extracellular space and interstitium to facilitate mammalian tissue repair and remodeling, particularly after trauma [reviewed in (33)]. In terms of complement regulation, vitronectin binds to nascent C5b-7, C5b-8, and C5b-9 complexes, blocks their incorporation into cell membranes, and prevents the polymerization of C9.³² Vitronectin-bound complexes of the TP components C5b to C9 are commonly referred to as “SC5b-9.” Finally, vitronectin is also thought to help solubilize SC5b-9 complexes by binding to hydrophobic patches in C5b-9 and preventing protein aggregation.

The inhibitory mechanism of the amphiphilic protein clusterin (apolipoprotein J; SP-40, 40) resembles that of vitronectin. Clusterin scavenges nascent C5b-7, C5b-C8, and C5b-9 complexes in the fluid phase before they can come into contact with a lipid bilayer, and thereby it inhibits the membranolytic insertion of soluble TP complexes.³⁴ Clusterin can also become incorporated into SC5b-9 complexes, suggesting some orthogonality in the regulatory mechanisms of these two fluid-phase TP inhibitors. Clusterin is a approximately 70–80 kDa glycoprotein, consisting of two disulfide-linked polypeptide chains (α and β), that is secreted into plasma. Aside from this circulating form of clusterin, which has cytoprotective effects, other less abundant, non-secreted forms of clusterin are believed to have chaperone-like functions.³⁵ Specific isoforms of and sequence variations in clusterin have been linked to the development of Alzheimer’s disease,³⁶ and modulation of apoptosis has been associated with certain cancers [reviewed in (37)].

On most host–cell surfaces, the major regulator of TP activity is the glycosylphosphatidylinositol (GPI)-anchored protein CD59 (protectin). Human CD59 is a 20-kDa glycosylated protein that shows exceptionally broad distribution across many human cells and tissues. As a result of enzymatic cleavage of the GPI by phospholipase C, a minor fraction of the polypeptide can also be found in soluble form (e.g. in blood plasma or urine). CD59 inhibits the final sequence of MAC assembly on cell membranes by binding to C8 and C9 within C5b-8 or C5b-9 complexes and inhibiting further recruitment and polymerization of C9 into a cytolytic pore. CD59 deficiency can occur in isolation as the result of a homozygous mutation in the encoding gene, leading to a global absence of functional CD59 from all cell surfaces.^38,39 In PNH patients, somatic mutations in hematopoietic stem cells affect enzymes that are essential for producing GPI anchors (most commonly observed in the PIG A gene). Mutated enzymes lead to a relative or absolute absence of GPI-anchored proteins from the progeny of the affected stem cells, including the complement regulators CD59 and CD55 (DAF; see below).^40,41 While both conditions are characterized by chronic hemolysis, isolated CD59 deficiency also shows a pronounced neurological phenotype. Severe or fatal thromboembolic events in PNH are caused by a generalized procoagulant state that is thought to be fostered by complement activation and hemolysis.^39,41,42

5 | CONVERTASE-DIRECTED RE GULATORS

A key feature of the complement cascade is that it allows for rapid activation and subsequent amplification via the positive feedback loop of the AP. Although the catalytic functions of the C3 and C5 convertases are important for this imminent reactivity, it is paramount to keep the convertases under strict regulatory pressure to ensure that activation and amplification only occur in the right biological context, i.e. on pathogens or altered self. Indeed, no other part of the cascade is controlled by as many regulators as convertase-driven C3 and C5 activation. Once the convertases are formed, these complexes typically decay within minutes and cannot reassemble. Although this decay provides an intrinsic regulatory mechanism, large amounts of C4b/C3b molecules can still be produced within the time frame of the natural convertase decay, necessitating that the convertases are tightly controlled by convertase-targeted regulators to supplement their intrinsic decay. Extending this concept, one further regulatory strategy is to accelerate the natural slow decay of the convertase complexes [termed decay acceleration activity (DAA)] and interfere with new assembly (Fig. 2B). However, since surface-bound C4b or C3b still retains the potential to rebuild fresh convertases, a second regulatory strategy aims at removing these scaffolds by facilitating their proteolytic degradation [termed “cofactor activity” (CA)]. By binding to C3b and/or C4b, some regulators act as cofactors by enabling the serum protease Factor I (FI) to cleave these “early-stage” opsonins into split products (iC3b/C3dg, iC4b/C4d) that can no longer from convertases (Fig. 2B), thus bringing complement activation/amplification to a halt. Importantly, some surface-attached, inactivated (or “late-stage”) opsonins such as iC3b and C3dg still have biological function, since they can be recognized by complement receptors to promote phagocytosis. Thus, FI serves two purposes: to downregulate complement activity by preventing ongoing convertase formation⁴³ and to transform surface-attached early-stage opsonins into late-stage ones to facilitate efficient phagocytic uptake.

FI, a 63-kDa glycoprotein built from two polypeptide chains that is primarily synthesized by the liver, is a highly specific serine protease requiring the presence of a suitable cofactor that mediates substrate binding. It is thought to circulate in a proteolytically inactive form that is maintained by the non-catalytic heavy chain, which allosterically modulates the enzymatic activity of the serine protease domain.⁴⁴ Only in a ternary complex with a cofactor and the substrate (C3b or C4b) is the allosteric inhibition released to enable proteolytic inactivation. The few cases of FI deficiency that have been reported are typically associated with the occurrence of recurrent infections.⁴⁵ The absence of FI leads to uncontrolled activation of the AP amplification loop and results in consumptive loss of C3 and, consequently, an impaired complement cascade.

The proteins acting as cofactors for FI and/or accelerators of convertase decay belong to the regulators of complement activation (RCA) family. With the exception of FI, all convertase-directed complement regulators belong to this family, which thus far consists of six homologous members. RCA proteins are encoded on a gene cluster on chromosome 1q32⁴⁶ and consist mainly of CCP domains, which have a distinct beta-sandwich structure that is stabilized by disulfide bonds [(47, 48) and reviewed in (49)]. Aside from their common domain structure, the regulatory CCP domains of all six RCA proteins have recently been found to share a common binding mode for C3b.⁵⁰ The members of the RCA family are complement receptor 1 (CR1, CD35), membrane cofactor protein (MCP, CD46), decay-accelerating factor (DAF, CD55), C4b-binding protein (C4BP), and Factor H (FH) and its alternative splice product Factor H-like 1 (FHL-1) (Table 2 and Fig. 3). Down-regulation of complement activation has to take place selectively on healthy self-surfaces and in the fluid phase (to prevent consumption), but complement activity must be allowed to proceed against microorganisms and other unwanted particles. Given their distinct profiles concerning localization, CA/DAA activity, and target specificity (Table 2), the RCA proteins act in concert to provide the required selectivity.

FIGURE 3.

Convertase-directed regulators of the RCA cluster. The functions and CCP domain structure of the regulators of complement activation are illustrated. CP/LP regulatory functions are depicted on the left, and AP functions on the right side. Arrows indicate the respective regulatory activities. The RCA proteins are comprised mostly of CCP domains, which are depicted as green ovals. The connecting region in C4BP that links all eight chains is shown in yellow. For DAF, the GPI-anchor is indicated. For CR1 and MCP, the transmembrane and cytoplasmic regions are shown

5.1 | Membrane-bound convertase regulators

Since membrane-bound regulators are embedded in the cell membrane, their regulatory functions are restricted to the host surfaces they protect. Of the three RCA members that are tethered to the plasma membrane, CR1 and MCP are embedded via transmembrane domains, whereas DAF is fixed to the outside membrane leaflet via a GPI anchor.

CR1 is an integral membrane glycoprotein composed of an N-terminal ectodomain, a transmembrane region, and a C-terminal cytoplasmic tail (Fig. 4A).⁵¹ Membrane-liberated CR1 molecules consisting of the ectodomain alone (soluble CR1; sCR1) are found as soluble proteins in the plasma, but only in low concentration. CR1 is widely expressed on human cells, including all blood cells except platelets, natural killer cells, and most T cells.^52,53 CR1 numbers range from 100 to 1000 on erythrocytes to the low ten thousands on leukocytes. The copy numbers and size/domain organization of expressed CR1 may depend on genetic variations in the CR1 gene [reviewed in (54)]. For example, CR1 density on erythrocytes is influenced by a “quantitative” (or number/expression) polymorphism,^55,56 and selected amino acid changes in CR1 define the Knob’s blood group antigens.

FIGURE 4.

Natural regulators as templates for engineered complement inhibitors. (A) Structural features of CR1 and the CR1-inspired, targeted complement inhibitor Mirococept. CR1 comprises an ectodomain area of 30 CCP domains (circles), followed by a transmembrane region (TM) and a cytoplasmatic tail (CYTO). The domain organization of CR1 into four long homologous repeats (LHR) is shown. The first three CCP domains of LHRs A-C all bind C3b but have different convertase-directed functions. Regulatory site 1 contains decay acceleration activity (DAA). The two copies of regulatory site 2 are virtually identical in amino acid sequence and harbor cofactor activity (CA). In Mirococept, a positively charged peptide (which aids the localization of the engineered inhibitor to the negative phosphate groups of the plasma membrane) links the functional site 1 of CR1 to a fatty acid introduced for membrane localization. (B) The soluble regulator FH is exclusively built from 20 CCP domains. Regions with specialized regulatory or architectural function are indicated. The splice variant FHL-1 consists of CCP domains 1–7 and has a unique C-terminal amino acid stretch of four amino acids added by splicing. In FHΔ10–15, domains 10–15 are deleted, linking CCP 9 directly to CCP 16. The optimized miniFH version uses 12 glycines to connect the four N-terminal domains to CCPs19–20. The number of glycines in the linker were chosen to allow both C3b recognition patches in miniFH to bind simultaneously to C3b

A size polymorphism in the ectodomain, which has arisen through duplication or deletion of highly homologous repeating units termed long homologous repeats (LHR), results in four known polypeptide sizes⁵⁷ containing three to six LHR. The most abundantly observed CR1 size variant is built from 30 CCP domains, which are organized in four LHR (A-D) of seven CCPs, followed by two additional CCP domains at the C-terminus. In LHR A-C, the regulatory functions are located in the three most amino-terminal CCPs, whereas LHR D does not exhibit regulatory activity.^58,59 Regulatory site 1 encompasses CCPs1–3 in LHR A and harbors DAA for both the CP/LP and AP.⁶⁰ There are two copies of regulatory site 2 within LHR B (CCPs8–10) and LHR C (CCPs15–17) that are virtually identical in sequence and function; these sites bind C3b and C4b more efficiently than does site 1, and they have CA for FI-mediated cleavage of these opsonins.⁶¹ The three functional sites are thought to cooperate when CR1 binds to surface patches bearing clusters of C3b and C4b molecules.^61,62 In addition to efficiently controlling complement activation, the transport of C3b-and C4b-opsonized immune complexes to the spleen and liver is another major function of CR1 on red blood cells (termed “immune adherence”).⁶³

The number polymorphism of CR1 has pathological implications, for example, in the case of PNH. Low CR1 numbers on erythrocytes have been associated with a less favorable outcome in PNH patients, ^64,65 which is not surprising, given that PNH erythrocytes only have two remaining RCA left to control the AP (i.e. CR1 and FH). It is probably because of the redundancy of convertase-directed regulators that the absence of two GPI-anchored complement regulators from PNH erythrocytes does not produce a more severe disease phenotype. Aside from a continuous low level hemolysis (probably resulting from the lysis of old erythrocytes), PNH erythrocytes are remarkably stable under steady-state conditions in the absence of complement-activating triggers (e.g. infection, trauma, or surgery). With regard to infectious disease, quantitative CR1 polymorphism and blood group variants are thought to contribute to malaria pathology [reviewed in (54)], but no functional consequences of the blood group variants on complement regulation have been detected in malaria.⁶¹

DAF is composed of four CCP modules that harbor DAA activities, followed by a serine/threonine-rich stalk domain that links the regulator to the GPI anchor on the membrane. With one N-glycosylation site and several O-glycosylation sites in the stalk, DAF has a molecular size of approximately 70 kDa. DAF is expressed on almost all peripheral blood cells and on tissues such as endothelial and epithelial cells. DAF deficiency is known from the Inab (null) phenotype of the Cromer blood group, but those with this phenotype do not, per se, exhibit increased hemolysis,⁶⁶ since their erythrocytes have normal numbers of CD59 that protect them from complement-mediated lysis under steady-state conditions. However, erythrocytes of the Inab phenotype do show moderately increased C3 fixation and/or hemolysis under challenging conditions, underlining the role of DAF in protecting host cells from complement attack.^67,68 Aside from complement regulation, DAF appears to also serve as an anti-adhesive surface glycoprotein that modulates the rate of neutrophil transmigration across mucosal epithelia.⁶⁹

MCP (CD 46) is a transmembrane protein that is widely expressed on most cell surfaces, with the notable exception of erythrocytes. Like DAF, MCP is comprised of four regulatory CCP modules and a heavily O-glycosylated serine/threonine/proline-rich stretch, yet it differs in also possessing a transmembrane region, followed by an intracellular domain. MCP protects cells by acting as a cofactor for the FI-mediated degradation of both C3b and C4b.⁷⁰ In addition to its complement-regulatory functions, MCP is increasingly recognized as a participant in several other functions, including cell signaling, metabolism, and development. Over the past two decades, MCP/CD46 has also emerged as a link between innate and adaptive immune responses, especially with regard to T-cell biology [reviewed in (71)]. For example, stimulation of CD46 on CD4⁺ T cells by C3b, C4b, or non-complement-related ligands has been implicated in the induction of a protective Th1 phenotype and the subsequent switch to a self-regulatory phenotype, which contracts the initiating immune response to restore homeostasis.

5.2 | Soluble convertase regulators

The RCA family of proteins includes three soluble regulators, C4BP, FH, and FHL-1. These plasma proteins have the important function of controlling complement turnover in the fluid phase to prevent exuberant production of inflammatory complement mediators and depletion of complement proteins. However, all soluble regulators can also be specifically recruited to self-surfaces via polyanionic host markers^{72– 74} to reinforce complement regulation by membrane-anchored regulators on host cells and tissues.

C4BP is a heterogeneous oligomeric glycoprotein of about 570 kDa that is mainly synthesized in the liver. Various isoforms of C4BP have been described, with the major form being a complex of seven identical α-chains and one β-chain that are organized in a spider-like arrangement through connecting regions in their C-termini. ^75,76 The seven α-chains each contain eight CCP modules, and the single β-chain is composed of three CCP modules [reviewed in (77)]. C4BP specifically regulates the CP, with both DAA and CA activity residing within the first three N-terminal CCP domains of the α-chains. Potentially, C4BP can associate with GAGs on cell surfaces, but simultaneous binding to GAGs and C4b is unlikely because their binding sites overlap.⁷⁸ Aside from complement regulation, C4BP has a role in regulating the activity of the anticoagulant protein S, providing a link between the clotting and complement cascades and a potential mechanism for targeting apoptotic cells.^79,80 It is noteworthy that the majority of circulating C4BP exists as a complex with protein S as the result of a strong interaction mediated by the C4BP β-chain.

5.3 | FH and FHL-1

FH, a 155-kDa glycoprotein, is the major fluid-phase regulator of the AP and is solely composed of 20 CCP domains (Fig. 4B).⁸¹ An alternative splice version of the FH gene encodes the Factor H-like 1 protein (FHL-1), which consists of FH CCPs1–7 plus four additional residues at its C-terminus. FHL-1 occurs at a serum concentration of approximately 1 μM, which is about a third to half of the FH level.^27–29 Although the N-terminal four domains in FH (and FHL-1) contain all the complement-regulatory functions (i.e. CA and DAA), the C-terminal CCP20 domain of FH is especially important for host-surface recognition.⁸² Consistent with the lack of the 13 C-terminal FH domains in FHL-1, it is thought that FHL-1 mainly acts in the fluid phase, whereas FH controls the AP in both compartments, i.e. on host surfaces and in the fluid phase. Five FH-related proteins (FHR) complete the family of FH proteins [recently reviewed in (83)]. FHRs lack the N-terminal CCP domains of FH and FHL-1 that contain all the complement inhibitory functions; instead, they are mainly composed of CCPs sharing various degrees of identity with the FH domains that are specialized in host-surface recognition. Thus, rather than negatively regulating C3 activation, FHRs, which appear as homo-or heterodimers, are thought to compete with FH for binding to certain surfaces. This process has been termed FH de-regulation and may have implications for physiological processes and also for certain pathologies.⁸³

6 | DISEASES ASSOCIATED WITH ALTERATIONS OF COMPLEMENT REGULATION

Inappropriate complement activation and/or insufficient regulation is evident in many human pathologies and can be either causative for the disease or occur secondarily, exacerbating the disease outcome [reviewed in (3, 14, 84)]. Examples of complement acting as an exacerbating factor are ischemia-reperfusion injury or autoimmune hemolytic anemia, whereas diseases that are directly linked to complement dysregulation include the AP-mediated conditions paroxysmal nocturnal hemoglobinuria (PNH), atypical hemolytic uremic syndrome (aHUS), C3 glomerulopathy (C3G), and age-related macular degeneration (AMD). In contrast to CP/LP activation, which occurs in response to activating danger patterns (e.g. tissue damage resulting from ischemia or antibody binding), the AP exhibits perpetual background activation with a propensity for signal amplification and, thus, it necessitates especially strict control through preformed regulators. This includes all AP-directed regulators and those that deal with the consequences of exuberant AP regulation (i.e. TP inhibitors such as CD59). The complete absence of CD59 in isolated CD59 deficiencies or the combined DAF and CD59 deficiency in PNH that were discussed above exemplify this point.

The severe kidney condition aHUS is strongly linked to abnormal regulation of the AP. Many, mostly heterozygous, sequence variations in AP regulators and a few mutations in complement effectors are known to predispose to complement-mediated damage in the glomerulus. The complement-mediated damage then triggers the actual thrombotic microangiopathy, often with fatal consequences (Table 3). Despite this clear association with an impairment of complement regulators, in about 50% of aHUS cases, the genetic factors, if any, that predispose to the disease are still unknown. Another characteristic of the aHUS-linked mutations is that the penetrance of these sequence variations is incomplete.⁸⁵ Thus, the alterations in the affected complement proteins predispose individuals to the disease, rather than directly causing it. It is thought that the simultaneous occurrence of an environmental complement-triggering event (e.g. an infection, surgery or childbirth) and a mutational phenotype offset the imbalance in the cascade, and causing the disease to become manifest. Aside from single mutations in specific complement genes, digenic inheritance is rarely observed (e.g. mutations in FH and MCP, FH and FI, or MCP and FI). Thus, two mutations with a “mild” functional effect in different regulatory proteins can add up, an observation that is in agreement with the “complotype” concept observed in vitro.⁸⁶ The regulator with the highest frequency of aHUS-associated mutations is FH; it is striking that about half of those genetic changes are found in the two C-terminal domains CCP19 and 20, implying that the surface and/or C3b recognition functions of the FH C-terminus are especially important for complement control by FH.^87,88

TABLE 3.

Genetic changes in convertase-directed regulators and complement effector molecules that are associations with aHUS^a,^b,^c

Gene	Chromosomal Locus	Frequency in aHUS	Commonly observed genetic changesd	Main effect
Factor H	1q31.3	30%	Mutations	Impaired cell-surface regulation
		3–5%	CFH/CFHR1 hybrid allele
FHR3, FHR1	1q31.3	5–15%	Deletions involving CFHR3 and CFHR1	Impaired cell-surface regulation due to occurrence of anti-FH autoantibodies
FHR1, FHR4	1q31.3		Deletions involving CFHR1 and CFHR4
MCP (CD46)	1q32.2	12%	Mutations	Reduced expression
Factor I	4q25	5–10%	Mutations	Low enzymatic activity
Factor B	6p21.33	1–4%	Mutations	Abnormally stabilized C3 convertase
C3	19p13.3	5%	Mutations	Abnormally stabilized C3 convertase/resistance to inactivation by Factor I

^aFor about half of aHUS cases no predisposing genetic change is detectable.

^bSequence abnormalities in thrombomodulin and DGKE also predispose to aHUS and account for about 5% of cases, but are not listed because they are not convertase-directed regulators.

^cTable is adapted from REF 85.

^dMajority of mutations are heterozygous.

The AP-mediated kidney condition C3G is characterized by abnormal complement activity resulting from sequence variations in, or autoantibodies against, convertase-directed regulators or complement effectors.⁸⁵ These alterations lead either to insufficiently controlled C3 turnover in the fluid phase, C3b deposition on the glomerular basement membrane, or a combination of these, resulting in severe glomerular damage. As in aHUS, impairment of FH regulation appears to be the most frequently observed alteration in complement functionality in C3G.⁸⁹ However, contrary to aHUS, in which the mutations/autoantibodies usually affect the surface-regulatory functions located at the FH C-terminus, the C3G-associated FH mutations/autoantibodies commonly map toward the molecule’s N-terminal domains that exert DAA/CA. These functional impairments of FH again underscore the consequences resulting from impaired function of the preformed regulators.

In recent years, the progressive eye disease AMD has received particular attention as a clinical condition linked to complement dysregulation. While different causative factors are being discussed with regard to the pathogenesis of AMD, at least the AMD sub-entity geographic atrophy (GA) appears to be clearly driven by an insufficiently regulated AP [reviewed in (90)]. Of the approximately 20 chromosomal regions that have been identified as harboring variants associated with AMD, complement proteins are the most highly implicated. Among the complement genes linked to AMD development (CFH, C2/CFB, CFI, C3, and C9), the FH Y402H single nucleotide polymorphism (SNP) in CCP7 is the most strongly correlated with disease development (see below).^91,92

The numerous sequence alterations in complement regulatory proteins that are associated with a broad spectrum of diseases underline the critical importance of controlling complement activation and amplification for the maintenance of tissue homeostasis. FH-mediated regulation and surface protection appear to play a particularly important role in many complement-related diseases. The following section will therefore provide an updated overview of the biology of this versatile regulator.

7 | NOVEL INSIGHTS IN FH AND FHL-1 BIOLOGY

One of the fascinating features of FH is that it can act as both a fluid-phase regulator, inhibiting consumption of AP components in circulation, and as a surface regulator by binding to host cells. This versatility is largely determined by its structural organization, which is solely built from CCP domains that are organized into multiple functional units. The four N-terminal CCP domains in FH contain all the regulatory functionality of FH, while CCP7 and CCP20 contain the host-surface recognition patches that specifically recognize negatively charged sialic acid and GAG moieties on host surfaces (Fig. 4B).⁹³

The regulatory functions of FH encompass causing the decay of AP convertases as well as enabling the FI-mediated degradation of C3b to iC3b. Both tasks necessitate the binding of FH to its target, C3b, for which it employs two C3b-binding regions located in the four domains nearest the N-terminus (FH1-4) and the two domains nearest the C-terminus (FH19-20). ⁹³ Early delineation studies with recombinant FH segments, domain deletion constructs, short peptides, or FH fragments produced by limited tryptic digestion gave important insights into the biological functions of this indispensable AP regulator ^94–96 [reviewed in (81)]. The functional data from fragment approaches does not always reflect the function of the entire protein, indicating that the apparently simple domain arrangement of the 20 CCP domains in FH is more complex than initially anticipated. Both the recombinant domain stretches FH1-4 and FH19-20 bind C3b, but only the C-terminal two domains, FH19-20, also bind efficiently to the C3b degradation products iC3b, C3dg, and C3d [with C3d corresponding to the thioester domain (TED) in C3b]. This is remarkable, since full-length FH strongly binds C3b but fails to efficiently recognize either iC3b or C3dg. Therefore, within the FH molecule, the C-terminal domains must be hidden in a compact, rather than linear, overall conformation of the 20 CCP domains.

Early biophysical data were interpreted to suggest that the 20 FH domains are arranged linearly, resulting in a flexible, extended structure, and therefore the functional activities of single domains or domain stretches would simply “add up”.⁸¹ This view was amended by further biophysical studies, which indicated that FH assumes a folded-back, yet still flexible conformation.⁹⁷ It was suggested that the affinity and regulatory potency of FH is low for a single C3b but enhanced for C3b clusters, a situation that was explained by the supposedly flexible FH structure that allows the various C3b binding sites within FH to connect between adjacent, surface-attached C3b molecules (Fig. 5A). Polyanionic self-markers were proposed to induce multimerization of flexible FH molecules and further increase the avidity for C3b clusters on host surfaces.⁹⁸

FIGURE 5.

Structural models of C3b binding by FH and miniFH. (A) The traditional FH model proposes that the CCP domains are flexibly linked to each other. This flexible linking should allow the binding of C3b and C3dg. Binding to adjacent C3b (or C3b and C3dg) molecules by one molecule of FH would be especially favored. (B) The updated FH model predicts that FH assumes a more rigid, conformationally closed structure. One molecule of FH binds to its two binding patches on one C3b molecule. When the FH N-terminus engages C3b, ideally with simultaneous recognition of anionic host markers on self-surfaces via CCPs19–20, a conformational rearrangement in FH “decrypts” the C-terminus and enables efficient binding and regulation of C3b. In C3dg (or iC3b), the FH1–4 binding site of C3b is absent (or degraded). In this situation, the compact center of FH directs the N-terminus to pack against the C-terminal domains 19–20 and thus shields these domains from binding to C3dg (or iC3b). (C) In miniFH, CCPs1–4 and 19–20 are flexibly linked by 12 glycines to allow simultaneous binding of both binding sites on C3b. Because of the flexible linker, CCPs19–20 are not shielded and thus are readily available for the targeting of C3dg (or iC3b)

How the FH splice product FHL-1, consisting of only the first seven N-terminal FH CCPs and a unique stretch of four amino acids, contributes to AP regulation was largely unaddressed. FHL-1 is found at approximately 1 μM in serum, which corresponds to 30–50% of the FH concentration in the blood.^99–101 Initial functional characterization of FHL-1, in an isolated system in which FHL-1 was allowed to exclusively probe DAA on sheep erythrocytes, indicated that FHL-1 exhibits only approximately 1% of the activity of FH in this system.¹⁰² Consequently, the role of FHL-1 in overall AP regulation was considered to be rather small. However, the 2005 discovery of a strong link between the Y402H SNP within CCP7 of the CFH gene and the development of AMD^103–106 has made the splice variant a focus of interest, since the Y402H change also occurs in FHL-1 and is not exclusive to FH.

8 | INSIGHTS INTO FHL-1 FUNCTION

Several studies have examined the functional consequences of the 402 SNP in the context of full-length FH, FHL-1, the isolated CCP7 domain, or partial FH fragments entailing CCP6-8. These studies have shown that the Y402H SNP results in modified binding to several ligands such as GAGs,^107–109 C-reactive protein,^110,111 zinc ions,¹¹² the extracellular matrix protein fibulin 3,¹¹³ and oxidation end-products, ^114,115 with many of these findings generating controversy among the different reports. Remarkably, these studies have revealed that the differential binding properties of the Y402H variant are pronounced when the mutation is studied in the context of isolated, recombinant fragments of FH. Within full-length FH, however, such differential binding is either undetectable, or only barely detectable,¹⁰⁷ suggesting that functional changes resulting from the 402 SNP in CCP 7 should be more pronounced in the splice variant FHL-1 than in FH. After all, FHL-1 is only equipped with one C3b (CCPs 1–4) and one polyanion-binding site (CCP7) and therefore relies to a greater extent on this area, whereas FH also contains such sites at its C-terminus, which may dilute the overall effect of the SNP. Importantly, genetic studies associating the 402 SNP with AMD cannot distinguish whether the functional consequences are derived from FH, its splice variant, or both, since FH and FHL-1 share the same gene. With the different reports indicating various molecular effects for the 402 SNP, there is no bona fide consensus as to why FH 402H predisposes individuals to develop AMD. A novel potential explanation for the FH-associated AMD pathophysiology is provided by a recent study that focused on the under-represented investigation of the role of FHL-1 in the eye.¹¹⁶ This report showed that FHL-1 is predominantly found in Bruch’s membrane, which is the extracellular matrix structure under the retina. In contrast, FH is mainly localized to the extracellular matrix of the choroid, the vascular layer of the eye. Part of the reason for this unusual distribution likely lies in the high local expression of FHL-1 at this site. However, with a molecular size of 49 kDa, FHL-1 is also expected to show better passive diffusion across Bruch’s membrane from the choroidal vasculature than is FH at 155 kDa. The AMD-associated Y402H SNP does not influence the diffusion of FHL-1 across Bruch’s membrane, but it strongly biases the binding to heparan sulfate (HS; a type of GAG) within Bruch’s membrane, with 402Y binding much more efficiently. This finding is especially meaningful in light of the previous observation that the HS content in Bruch’s membrane decreases with age.¹¹⁷ The implications are that the FHL-1 402H version is less efficient in localizing sufficient complement control to Bruch’s membrane.

That FHL-1 is the predominant AP regulator in Bruch’s membrane may explain the strong link between AMD and CCP7. With regard to localizing efficient AP-directed complement control by FH to the kidney, the host recognition sites in CCPs19–20 appear to be more important than the recognition site in CCP7.^118,119 The shifted balance between FH and FHL-1 in ocular tissue puts more pressure on FHL-1 to provide efficient complement control at the surface, including Bruch’s membrane, via the host-surface recognition patch in CCP7.

Although the contribution of the splice variant FHL-1 to AMD pathology has emerged as an increasingly critical factor, this finding does not imply that the function of FH is dispensable in the eye, since mutations at the FH C-terminus can be responsible for an early onset of the disease.¹²⁰ Moreover, outnumbering FH at Bruch’s membrane does not necessarily mean that the majority of AP control at this site is supplied by FHL-1, since the relative regulatory potencies of FH and FHL-1 have to be taken into account. Both FH and FHL-1 regulate the AP by binding to C3b, and thus causing convertase decay and acting as a cofactor for C3b degradation. Published data indicate that the N-and C-termini of FH cooperate in binding to C3b, since the affinity of FH for C3b (K_D approximately 0.5–1 μM) is significantly higher than that of the N-terminal (FH1–4, K_D approximately 10–14 μM) or C-terminal (FH19–20, K_D approximately 5 μM) sites assayed in isolation. ^93,121 Remarkably, the affinity of FHL-1 for C3b was recently determined to be 1.5 μM: just half that of FH.¹²² This result suggests that the N-terminal binding site in FH for C3b actually extends beyond domains 1–4 and includes one or more of the domains 5–7. Direct comparison of the overall regulatory capacity of FHL-1 to FH in lysis assays of PNH erythrocytes in serum showed equal activities for FH and its splice version when the proteins were added to normal human serum. Given that the C-terminal FH domains are critical for specifically conferring protection from the AP on host cells via sialic acid recognition, as on erythrocytes,^{73,74,123–125} the finding of comparable activity between FH and FHL-1 is surprising. However, together with the data on Burch’s membrane, where FHL-1 outnumbers FH, this similarity in activity implies that FHL-1 probably does account for the bulk of AP regulation within Bruch’s membrane, thereby further supporting the relevance of FHL-1 with regard to the AMD-associated Y402H polymorphism. However, further comparative studies in different host cells and tissues will have to be conducted to determine whether the parity in AP regulation between FH and FHL-1 is generally true. Another interesting point for further investigation is how FHL-1 and FH, which are present simultaneously in serum, may cooperate in protecting host cells from the AP. With FHL-1 being a more efficient regulator than previously anticipated, at least in certain environments, it is of particular interest to determine the function of the 13 C-terminal domains in FH that are absent from FHL-1.

9 | AN UPDATED STRUCTURAL MODEL OF FH

In terms of functional sites, the most obvious difference between FH and FHL-1 is that FHL-1 lacks the C-terminal domains, CCPs19–20. Thus, FHL-1 misses one important C3b binding site as well as the important C-terminal host recognition site, which is critically involved in protecting erythrocytes and the glomerular endothelium, among other tissues. That the C-terminal FH domains are important for complement inhibition can be deduced from the cluster of aHUS-associated mutants in CCPs19–20 and from several reports showing that a lack of CCPs19–20 in FH, or a competition with recombinant C-terminal domains, results in severe AP damage to host cells.^82,87,88 It is thought that FH efficiently regulates the AP in the fluid phase without much contribution from the C-terminus^82,126 [reviewed in (81)]. On host–cell surfaces, the simultaneous binding of FH to polyanionic surface markers while engaging C3b is thought to largely enhance regulatory activity. The absence of CCPs19–20 from FHL-1, however, does not diminish the molecule’s ability to protect erythrocytes in vitro,¹²² thereby raising the question of the exact contribution made by the FH C-terminus to AP control.

These findings concerning FH’s structure can be reconciled by a structural model in which a compact center arranges the N-terminal half of FH so as to shield the C-terminal domains in a partially cryptic conformation (Fig. 5B). The occluded C-terminus may be released when the N-terminus engages C3b. It is anticipated that such conformational rearrangement into an “open” FH conformation upon C3b binding is facilitated by the simultaneous binding to polyanionic markers on host surfaces. This structural model also provides an explanation for the finding that recombinant FH CCPs19–20 in isolation can efficiently bind to TED in the C3b degradation products iC3b and C3dg,¹²⁷ but the same domains within FH cannot.^122,125,128 Only a compact and simultaneously rigid FH conformation can actively shield the C-terminal engagement of TED by occluding the TED binding site on CCPs19–20. Evidence for a compact FH conformation has already been described earlier.¹¹⁸ The updated model also predicts the domain arrangement in FH to be rigid, since a cryptic C-terminus, and the functional data supporting its relevance, could not be brought into agreement with a flexible FH molecule in which CCPs19–20 are easily accessible for TED engagement in iC3b or C3dg.

Thus, an updated FH model is proposed that features a compact and rigid rather than an elongated and flexible structure, bringing both termini in FH in close proximity to shield the C-terminus from readily binding to TED. This model is corroborated by data showing that FH domains 10–15 pack against each other and assume a compact conformation to position both FH termini in a close spatial arrangement. ^48,129 Structural data further support this model, since the available co-crystal structures of C3b:FH1–4¹³⁰ and C3d:FH19–20^131,132 reveal that both C3b binding sites of FH are located close together on the same side of C3b. Only a folded-back structure of FH would enable a single FH molecule to simultaneously bind both binding sites on C3b and to engage in recognition of self-surface polyanions. Recent data have confirmed that previously identified residues of CCP20, which were thought to be crucial for recognizing anionic host markers, are still accessible when CCPs19–20 engage TED.¹²³ Despite being a soluble plasma protein, and hence not tethered to plasma membranes, FH can selectively increase its residence time on self-surfaces through exploitation of polyanionic self-surface markers on host tissue.

The revised model also takes into consideration the implications for the stoichiometry of FH-mediated regulation. Although the available data increasingly suggest that a compact and rigid FH conformation favors an interaction with C3b in a 1:1 fashion, the alternative possibility of a complex in which one FH binds to two neighboring C3b molecules on a surface (1:2) cannot be completely excluded. Indirect support for the 1:1 interaction model comes from a report that describes the interaction of C3b with the recombinant FH deletion molecule FHΔ10–15 (Fig. 4B), which lacks the six central compactly folded domains 10–15. Deleting CCP10–15 is predicted to break the compact and rigid FH architecture and locate both FH termini far apart from each other, thereby favoring a 1:2 binding mode. Indeed, interaction analysis of FHΔ10–15 binding to C3b has confirmed that in the artificially extended structure of FHΔ10–15, a 1:2 interaction model with C3b is feasible. In conclusion, the data on FHΔ10–15 indirectly support the proposed compact, rigid structural model of FH predicting that FH preferably interacts with C3b in a 1:1 binding mode.¹²² With the new structural model of FH corroborated by an increasing amount of molecular data, questions arise about the functional benefit of such a complex and intricate binding mode of a seemingly simple regulator.

10 | THE UPDATED STRUCTURAL ROLE OF FH IN THE LIGHT OF EVOLUTION

Although some functional aspects of FH regulation still need to be revisited in the light of the new molecular insights, one can already speculate about why evolution would favor a mode in which a shielded C-terminus restricts the binding of FH to the C3b degradation fragments iC3b and C3dg:

1. Improved selectivity for danger is one potential explanation. Foreign cells and altered self-surfaces constantly activate complement and become opsonized with C3b. Although this arrangement ideally leads to rapid removal of danger, altered host cells and many pathogens have devised strategies to evade the innate immune response and regulate complement activity (see below), often leading to a massive accumulation of iC3b and C3dg on these regulated target surfaces. Efficient targeting of FH to iC3b/C3dg-covered “dangerous” structures under scrutiny by the complement system would interfere with further complement surveillance. Under normal circumstances, FH should instantly prevent any AP amplification on healthy host tissue but not accumulate on cells posing a potential insult. In evolutionary terms, targeting of FH to late-stage opsonins could generate an unjustified advantage in complement regulation on the attacked cells and disturb the indiscriminate tickover-driven surveillance mechanism of the AP. After all, accumulation of high iC3b and C3dg densities is not expected to occur on healthy host tissue, because an undisturbed set of preformed regulators of defense would limit C3b deposition and amplification early on.

2. A second, but related, reason could be linked to the concept of FH de-regulation by FHR proteins, which are thought to aid the AP’s immune surveillance by competing the regulator FH off surfaces with high densities of C3b, iC3b, or C3dg.^133,134 If FH were to have an open conformation that allowed its C-terminal CCPs19–20 to readily engage TED in C3b/iC3b/C3dg, FH would tend to resist being competed off by the multimeric FHR. However, with efficient FH engagement being restricted solely to C3b-coated surfaces, complement-probed surfaces covered in iC3b/C3dg do not achieve this increased measure of protection and may be scrutinized by another round of complement surveillance.

11 | INTERPLAY BETWEEN PATHOGENS AND COMPLEMENT REGULATORS

The principle mode of action of the ancient innate immune defense directed to discriminate between host and invader largely relies on a “missing self-protection,” i.e. the absence of regulators and self-makers on foreign cells that protects them from host defense systems, and in particular, from complement. However, the evolutionary pressure on invading microbes paved the way for the development of several strategies that allow to subvert complement attack to successfully invade the human host [reviewed in (13)]. Thanks to their potent and perfectly tailored inhibitory activity against the key steps of complement activation, exploiting natural host regulators such as RCA is among the most widely used strategies for complement evasion. Some viruses express proteins that mimic human RCA and efficiently control the convertases of the complement cascade. A well-known example is vaccinia virus complement control protein (VCP), a member of a homologous family that has evolved in several orthopox viruses as condensed forms of human RCAs; despite containing only four CCP domains, VCP binds C3b, C4b, and heparin and potently impairs complement activation.^135,136

Of the various complement evasion mechanisms identified thus far, the recruitment of soluble host regulators is the most commonly observed strategy and often involves FH/FHL-1, C4BP, and/or vitronectin binding. When recruiting host regulators to the microbial surface, pathogens employ strategies remarkably similar to those of their host cells, preferentially capturing a particular human regulator in an active conformation.¹²⁶ For instance, most pathogens bind FH through one of the molecule’s surface recognition patches around CCP7 or 20 [reviewed in (137)]. To recruit FH, bacteria often mimic host polyanions either by directly expressing sialic acid, an untypical carbohydrate for microbes, or by exposing bacterial proteins that mimic such moieties.^138,139 Viruses, fungi, and parasites have also been shown to recruit FH. A special case is the human malaria parasite Plasmodium falciparum, which expresses different molecules during diverse stages of its life cycle to recruit FH and FHL-1 for complement evasion.^140,141

Another common interaction between pathogens and complement proteins involves the hijacking of complement receptors/regulators of the RCA family as adhesion receptors for cell entry. Recent insights again come from malaria parasites, with P. falciparum merozoites exploiting CR1 as an entry receptor.^142,143 It is particularly interesting that the binding of the responsible malaria adhesin PfRh4 to the N-terminus of CR1 with nanomolar affinity^144,145 blocks the DAA of CR1.⁶² Whether this functional impairment of DAA is merely a side effect of merozoite adhesion or may actually enhance the invasion pathway through “voluntary opsonization,” a strategy previously reported for other pathogens, remains to be elucidated. However, taking into account the fact that merozoites recruit FH to impair complement, the concept of voluntary opsonization seems less likely, yet the decoration of merozoites with iC3b may serve to enhance its interaction with erythrocytes by binding to CR1. With regard to being an entry receptor, MCP is particularly often exploited by a large number of pathogens [reviewed in (146)], likely because it is widely expressed across almost all cell types, making it a worthwhile target.

12 | THE TRANSLATIONAL APPROACH: ENGINEERING OF NATURAL COMPLEMENT REGULATORS

As illustrated above, inappropriate regulation of complement activity is increasingly recognized as a causal or exacerbating factor in a number of human diseases [reviewed in (147)]. Advances in the understanding of disease mechanisms have been paralleled by rapidly expanding insights into physiological complement regulation and microbial complement evasion at a fine-coarse level. This progress has initiated a new surge in the development of therapeutic intervention strategies based on natural complement regulators.

Early attempts to employ this concept used purified or recombinant versions of natural complement regulators. In 1984, complete DAF (including the GPI anchor) was extracted and purified from human erythrocytes and transferred to other cells to protect them from complement attack.¹⁴⁸ Subsequently, a recombinant soluble version of CR1 (aCR1; TP10, Avant Therapeutics), which lacks the transmembrane and cytoplasmic domains, was investigated in vivo as a means of suppressing post-ischemic myocardial inflammation and necrosis. ¹⁴⁹ Also, exogenous but highly potent RCA mimics derived from pathogens, such as VCP from vaccinia virus, have been investigated for their potential therapeutic effects,¹⁵⁰ but immunogenicity concerns common to exogenous proteins have imposed challenges to the chronic therapeutic use of these agents. Although consideration has been given to treating patients with concentrated doses of FH, after purifying FH from human plasma or expressing the protein recombinantly, these strategies have not advanced to the clinical stage.^151–153

Given their modular organization, members of the RCA family of complement regulators are particularly suitable for protein engineering, and such strategies have resulted in several drug candidates over the past few years. An example of an engineered complement inhibitor is Mirococept (APT070), a recombinant, membrane-targeted form of CR1. This inhibitor links the first three CCP domains of CR1, which exert DAA for both CP/LP and AP convertases, with a charged addressin peptide and a myristoyl moiety that allows incorporation into the plasma membrane.^77,154 Mirococept is currently in clinical trials for kidney transplantation after preperfusion of the donor organ with the drug. The majority of engineering approaches directed at improving the regulatory capacity of natural complement inhibitors for therapeutic purposes combine the N-terminal regulatory part of FH with an entity for targeting host surfaces.

13 | ENGINEERED FH-BASED REGULATORS

As the major AP-specific regulator that harbors both DAA and CA functionality, FH is an ideal candidate for engineering enhanced complement inhibitors.^{151–153,155} For example, it is well established that FH protects vulnerable PNH erythrocytes from complement-mediated hemolysis to some degree.¹²⁴ Recent in vitro studies investigated whether increasing the FH concertation in serum would enable complete protection of PNH erythrocytes from lysis. Although possible, this approach required a doubling of the concentration of FH to achieve full protection.^122,156 Given the rather high plasma concentration of FH (2–3 μM; 0.3–0.5 mg/ml),^99,100 the protein amounts needed for supplementation treatment would be very high. Engineered, targeted FH-based inhibitors that exceed the regulatory activity of FH could potentially alleviate this problem. In analogy to FH, these compounds would block the upstream process of C3b deposition and thus inhibit both pathophysiological mechanisms involved in erythrocyte destruction in PNH: the AP-driven C3 opsonophagocytosis that cumulates in extravascular hemolysis^157,158 and the TP-mediated process of MAC assembly that drives intravascular hemolysis.

The superposition of the C3d:FH19–20 and C3b:FH1–4 complexes onto TED revealed the close proximity of the FH1–4 C-terminus and the FH19–20 N-terminus (see above).¹³¹ This finding inspired the idea that FH CCPs1–4 could be directly fused with CCPs19–20 by a peptide linker, yielding the artificial fusion protein miniFH, with largely reduced size and increased production yield when compared to the parental FH (Fig. 4B). Biological evaluation of this engineered FH version showed that the two artificially connected FH termini enable miniFH to bind simultaneously to both FH binding sites on C3b.¹²⁵ According to the updated structural model of FH (see above) the direct connection of both FH termini, omitting the 14 central domains of FH, breaks the compact and rigid FH conformation and renders its C-terminus readily available in a non-shielded conformation. Thus, miniFH conserves not only the crucial regulatory and polyanion-binding functions located at the FH termini but also provides a novel feature by decrypting CCPs19–20 and rendering miniFH more accessible to iC3b/C3dg (Fig. 5C). This approach allows for improved targeting of the inhibitor to vulnerable host tissues that experience continuous complement activation, such as PNH erythrocytes,¹²⁵ and it is expected to increase the local regulator concentrations. Indeed, in clinically relevant ex vivo assays on PNH erythrocytes, miniFH indeed showed 10-fold higher potency in protecting PNH erythrocytes than did FH,^122,125,156 indicating an improved inhibitory capacity for the engineered regulator and indirectly supporting the hypothesis that a liberated FH C-terminus is able to potentiate complement regulatory function. This notion was further supported by reports that investigated alternative versions of miniFH^159,160 and a study that utilized a FH deletion construct that is missing the central domains 10–15 (i.e. FHΔ10–15).¹²² All these engineered versions exhibited de-shielded CCPs19–20 and outperformed the parental FH in protecting PNH erythrocytes by several-folds. ^122,156 Notably, the different linker lengths introduced between the two C3b binding sites of the engineered miniFH versions influenced the ability to protect PNH cells. The optimized peptide linker in miniFH showed the highest activity in PNH cell protection, illustrating that an optimal spatial orientation between functional domain segments is important for potentiating inhibitory activity on host cells.

To a certain degree, the regulatory mechanism of miniFH is comparable to the engineered FH-CR2 fusion protein that combines FH CCPs1–5 with CCPs1–4 of complement receptor 2 (CR2) [(161); reviewed in (162)]. The binding activity of CR2 for iC3b and C3dg gives the FH-CR2 fusion protein a targeting profile toward sites of ongoing complement turnover that is similar to that of miniFH. However, FH-CR2 is not equipped with the recognition capabilities of the FH C-terminus for polyanionic self-surface patterns and only contains a single binding region for C3b within the FH CCPs1–5. Direct comparison of these FH-based, engineered inhibitors in an ex vivo PNH protection assay showed high activity for both inhibitors, with miniFH being five times more potent than FH-CR2.¹⁵⁶ A potential reason for the higher activity of miniFH over FH-CR2 on PNH erythrocyte surfaces may be that miniFH recognizes host anionic markers via FH CCPs19–20. Moreover, the lower affinity of miniFH for iC3b/C3dg when compared to FH-CR2 may allow for faster transition rates from the inactivated opsonins to the activation product C3b, which needs to be efficiently controlled to prevent erythrocyte destruction. Both miniFH and FH-CR2 are promising candidates for the treatment of PNH, especially in patients with suboptimal response to eculizumab, and for treating other diseases.^125,163 A FH-CR2 construct developed by Taligen (TT30) has recently been investigated in phase 1 trials by Alexion, yet no further plans have been announced. A miniFH variant (AMY-201) is currently being considered for clinical development by Amyndas.

In conclusion, the design and characterization of engineered FH-based regulators has already resulted in interesting molecules with therapeutic potential that are undergoing further evaluation in clinically relevant disease models. Comparing various engineered constructs of the natural regulators FH and FHL-1 has also advanced our insights into the general requirements of convertase-driven complement regulation and suggests that the turnover rate of the regulators plays a role in determining anti-complement activity.

14 | FUTURE ASPECTS OF CONVERTASE-DIRECTED REGULATION

Convertase-directed regulators fulfill the vital task of tightly controlling complement activation and amplification processes early on in the cascade and keeping the critical balance between activation and regulation to maintain tissue homeostasis. Ideally, the diverse regulators work in concert to (selectively) allow a necessary minimum of complement activation for the cascade to perform its continuous immune-surveillance functions. The collaboration between regulators has to be dynamically adapted to cope with different threat situations. Aside from specific activation events, which are mainly controlled through the pattern recognition complexes of the CP/LP, it is the mix of available complement regulators that determines the functional consequences of complement activation. Pathogens devoid of regulators ideally face a full-scale response, including effector generation that leads to direct killing and/or phagocytic removal, whereas a milder response restricted to opsonization should be induced by dying host cells. Efficient removal of apoptotic cells largely depends on the FI-and regulator-mediated conversion of C3b to iC3b and subsequent uptake by macrophages via CR3 and CR4.¹⁶⁴ In contrast to phagocytosis of iC3b-opsonized microbes by CR3, which creates an overall inflammatory milieu,¹⁶⁵ the complement-dependent uptake of apoptotic bodies via the iC3b-CR3/CR4 axis promotes an anti-inflammatory signature with TGFβ release.^{164,166–168} Conversion of C3b to iC3b on apoptotic cells appears to depend on the recruitment of FH, which binds to several surface molecules on apoptotic bodies. ¹⁶⁹ Thus, the preformed regulators of defense not only negatively regulate the complement cascade to influence whether proximal or terminal signals are liberated but also produce the CR3-ligands iC3b and C3dg to direct efficient and anti-inflammatory uptake of turnover products to aid tissue homeostasis.

There is increasing evidence that iC3b molecules on self-particles promote an anti-inflammatory, potentially tolerogenic milieu through the recognition of iC3b by CR3.^170,171 Given that C3dg has recently also been considered a ligand of CR3, the same notion probably holds true for C3dg-tagged self-cells. ^158,172 It will be very exciting to see how future studies will address this emerging theory of iC3b/C3dg-mediated fostering of anti-inflammatory signals, especially in the context of complement-mediated diseases such as AMD.^173,174 Convertase-directed complement regulators are expected to play a major part in this process, since they are responsible for generating the late-stage opsonins iC3b and C3dg from the C3 activation product C3b.

The past few years have generated a wealth of new and fascinating insights into the molecular mechanisms of complement regulators, their involvement in disease processes, and their potential for therapeutic use. In view of emerging concepts such as competitive de-regulation, context-specific accessibility of binding/targeting sites, and regulator-mediated modulation of inflammatory responses, the coming decades in the research into preformed complement regulators promise to be equally exciting.