Complement component C3 - The “Swiss Army Knife” of innate immunity and host defense

1.1 | C3-mediated opsonization as central element in the elimination of threats (V体育官网入口)

Once they encounter a pathogen or other foreign cell, the various PRMs of the complement system detect and bind to distinct surface patterns and induce the cascade via several routes. 3,4,6 Although the classical pathway (CP) is induced by the C1 complex, which primarily binds to antibody complexes and also to other pathogen-or damage-associated molecular patterns (PAMPs, DAMPs), the various PRMs of the lectin pathway (LP) recognize microbial carbohydrates and similar markers (Figure 1A). In all cases, PRM-associated proteases become activated and cleave components C2 and C4 to form a reactive but unstable protease complex on the surface of the targeted cells; this CP/LP C3 convertase binds to and cleaves the abundant plasma protein C3 and induces a transformation that determines the fate of the cell. The convertase-mediated cleavage of C3 releases the small protein C3a and induces a conformational change in the remaining C3b fragment that exposes a highly reactive but short-lived thioester VSports. In close proximity to the initiating surface, the newly formed C3b can covalently bind to hydroxyl or amino groups, thereby opsonizing the cell. Importantly, formation of C3b enables the binding of the protease Factor B (FB), and the resulting pro-convertase (i. e. C3bB) is quickly transformed by Factor D (FD) into an active C3 convertase (i. e. C3bBb) that by itself can cleave more C3 into C3b, thereby creating an amplification loop for C3b deposition (Figure 1C). This mechanism is commonly, yet somewhat misleadingly referred to as the “alternative pathway” (AP) of complement, especially as this route can sometimes contribute up to 80% of the overall response, even after initiation by the CP or LP. 7,8.

Although not pathogen-specific, CP-and LP-mediated initiation still confers some selectivity, as these routes are dependent on the presence of certain microbial surface markers or, as in the case of antibody-mediated CP activation, receive assistance from adaptive immunity. However, complement also employs a simple yet elegant “tick-over” mechanism involving C3 that allows the system to increase baseline activity and probe cells in a less distinctive manner (Figure 1B): Although C3 predominantly circulates in its native form, a small fraction is constantly undergoing spontaneous hydrolysis to result in a distinct conformer termed C3(H2O). Without changing its composition, C3 thereby modifies its reactivity as a result of structural changes that generate a molecule with a functional spectrum similar to that of C3b. As such, C3(H2O) can bind FB and form AP C3 convertases that turn C3 into C3b and initiate the AP. Importantly, physical adsorption of C3 on various surfaces, such as microbial cells containing lipopolysaccharides (LPS) and also blood-gas interfaces or activated platelets, can induce a similar conformational activation in C3 (Figure 1B), thereby enhancing and directing the tick-over in the case of less-defined threats. 9 Finally, C3 may be directly cleaved by non-convertase proteases such as coagulation enzymes (e. g. thrombin, plasmin) or tissue kallikreins10,11; although these activities are typically low in comparison with the primary substrates, it is possible that this “extrinsic” pathway (Figure 1A) gains relevance under specific disease conditions or in local microenvironments VSports app下载.

Independent of the route of initiation, opsonization with C3b marks the starting point for the generation of a series of potent effectors with the aim of facilitating the clearance of the tagged particles (Figure 1D). In this context, C3b not only amplifies the complement response via convertase formation but also acts as an effector itself that mediated functions in innate and adaptive immunity. 12 Via binding to complement receptor 1 (CR1; CD35) on immune cells, it confers immune adherence that enables the shuttling of opsonized cells to the spleen and liver. Tissue-resident macrophages such as Kupffer cells in the liver also express the complement receptor of the immunoglobulin family (CRIg) that has been shown to bind C3b and induce phagocytosis. As briefly discussed later, binding of C3b to CD46 has an impact on adaptive immunity by influencing T-cell responses. The interaction of C3b with CR1/CD35 and CD46, both of which are members of the regulators of complement activation family (RCA; see below), mediates the degradation of this primary opsonin to iC3b (after removal of the C3f peptide) and/or C3dg plus C3c (Figure 1A). Although C3c is released into circulation, the late-stage opsonins iC3b and C3dg remain bound to the surface and exert potent signaling functions (Figure 1A, D). iC3b, in particular, is a highly versatile effector that not only maintains binding activity for CRIg but also gains affinity for the phagocytic integrin receptors CR3 (CD11b/CD18) and CR4 (CD11c/CD18), thereby largely enhancing complement-mediated phagocytosis V体育官网. Intriguingly, there is increasing evidence that the powerful iC3b-CR3 mechanisms are not only employed for microbial clearance but also for immunoediting during tissue development; recent studies have shown that complement is engaged during synaptic pruning with potential implications for the development of schizophrenia. 13,14 Finally, the conversion to iC3b also exposes a binding site for CR2 (CD21); this receptor is expressed on B cells (and follicular dendritic cells) and forms part of the B-cell co-receptor complex. The interaction with CD21 lowers the activation threshold of B cells and influences the generation of memory cells. Although the binding to CD21 is maintained in the end-stage opsonin C3dg, the interaction with CR4 appears to be lost. Interestingly, recent studies suggest that the interaction profile of C3dg with the related receptors CR3 and CR4 is distinct. C3dg still contains binding sites for CR3 and, at least in the case of densely C3dg-opsonized erythrocytes, the C3dg-CR3 interaction can mediate phagocytosis; the relevance of this mechanism in the context of different (patho)physiological scenarios needs to be further explored. In any case, the deposition and conversion of C3b provides a dynamic platform for tiered effector functions ranging from immune adherence and transport (C3b-CR1) to phagocytosis (C3b/iC3b-CRIg, iC3b/C3dg-CR3, iC3b-CR4) and adaptive immune modulation (C3b-CD46, iC3b/C3d-CD21).

Although some aspects remain elusive, C3b is also essential for the induction of the terminal pathway of complement effector generation (Figure 1A, E). Increasing the density of C3b during complement activation and amplification leads to a substrate specificity shift in the generated convertases from C3 to its paralog C5. Although the CP/LP C3 convertases are composed of cleavage-generated C4 and C2 fragments, an association with C3b is necessary for shifting to C5 convertase activity. Once formed, C5 convertases cleave C5 into C5b, thereby releasing the small C5a protein. In contrast to C3b and C4b, C5b does not covalently deposit via a thioester, but rather associates with C6 and C7 to form a complex that can insert into membranes. The binding of C8 and multiple copies of C9 forms the membrane attack complex (MAC) that acts as a lytic pore to damage or kill susceptible cells and organisms.

1.2 | Spreading the danger signal: The role of C3a

Although MAC-induced lysis and opsonin-mediated phagocytosis already represent efficient means of microbial clearance, the translation of local complement activation into cellular signals and the coordination of downstream responses add another critical layer to complement’s defensive actions. Alongside the immunomodulatory activities of opsonins, which act in direct contact with target and effector cells (see above), the remote actions of the anaphylatoxins C3a and C5a are highly important for shaping the immune response (Figure 1A). C5a, in particular, is a powerful chemoattractant that signals via two C5a receptors (C5aR1/CD88, C5aR2/C5L2) and exerts activities that range from immune cell activation and priming to smooth muscle contraction and beyond. In addition, C5a-mediated signaling is engaged in a plethora of crosstalk activities with other defense and homeostatic pathways.

C3a has long been considered a less potent cousin of C5a, with a similar activity spectrum but a 10-to 100-fold weaker activity. However, newer studies have suggested that C3a, which signals through the G protein-coupled C3a receptor (C3aR), shows a distinct and context-specific profile that sometimes even counteracts C5a-induced activities.¹⁵ In fact, C3a appears to assist in some unique functions, with our group and others having demonstrated its involvement in such diverse processes as tissue development and regeneration (e.g. in the retina, liver),^16–19 homing of hematopoietic stem cells,^20,21 and migration of neural crest cells,^22,23 among others. Although C3a has also been suggested to signal via other receptors, these activities have remained relatively unexplored (e.g. the binding of C3a to the receptor of advanced glycation end-products²⁴) or remain controversial, as in the case of the proposed metabolic effects of C3a and its degradation product C3a-desArg via C5aR2 stimulation.^15,25 Interestingly, C3a has also been reported to act as a preformed mediator of defense outside its role in complement, as it may act as an antimicrobial peptide that binds to and destabilizes bacterial membranes.²⁶ Although many aspects of its function remain to be fully explored, C3a appears to hold a few more functional cards than initially thought.

1.3 | Tuning C3 activity through interaction with modulators and regulators

The molecular pattern on a cell surface largely defines the binding of PRMs and/or adhesion of C3; complement is therefore intrinsically less likely to be engaged on host cells than on foreign particles. However, given the potentially deleterious actions of an activated complement system on or near host cells, circulating and membrane-bound complement regulators offer an essential layer of protection and confer the necessary selectivity. Although soluble C1 inhibitor and non-proteolytic splice products of MASPs help control the CP and/or LP, membrane-bound CD59 prevents the formation of MAC, and carboxypeptidases tame/modulate anaphylatoxin activity.²⁷ Given their central role in modulating complement activity, however, the majority of complement regulators directly target C3 and its fragments and complexes.^27,28 These inhibitory proteins typically belong to the regulators of complement activation (RCA) family, which includes the membrane-bound CR1 (CD35), decay acceleration factor (DAF; CD55) and membrane cofactor protein (MCP; CD46), and the circulating regulators C4b-binding protein (C4BP), Factor H (FH), and FH-like protein 1 (FHL-1).²⁷

With one exception (see below), these regulators mainly act via binding to C3b.^28,29 This interaction can provide decay acceleration activity (DAA) toward the AP C3 convertase as a result of the steric displacement of the Bb fragment from the C3bBb complex. Binding of some RCAs to C3b also provides a binding site for the protease Factor I (FI), which cleaves C3b into iC3b and/or C3dg (termed cofactor activity [CA]). Among membrane-bound regulators, CD55 only provides DAA and CD46 only CA, whereas CR1 combines both activities and is the only RCA to enable the final degradation to C3dg. Although C4BP has been reported to bind C3b and act as a cofactor, albeit at high molar excess, C4BP is generally considered a CP/LP-selective regulator.^30,31 In contrast, FH and its shorter splice product FHL-1 both show strong selectivity for the AP and provide DAA and CA upon binding to C3b. FH is a particularly interesting regulator because it combines its inhibitory activities with pattern recognition capabilities that allow this plasma protein to recognize self-surface patterns (e.g. glycosaminoglycans or sialic acids that are usually absent from microbial cells), thereby assisting regulation on host cells.²⁷

Complement-mediated attack on microbial intruders can therefore be seen as a result of both the presence of foreign signatures and an absence of non-self patterns.³² This delicate interplay between promoting and preventing signatures can be exploited by modulators to fine-tune complement activity and create a directed and measured response to specific targets.⁴ Such modulators typically contain pattern recognition regions and binding sites for complement components, activators, and/or regulators. For example, although properdin is best known for its ability to stabilize the C3bBb convertase, the protein has also been reported to act as an initiator that binds markers such as LPS and recruits soluble C3b (as a starter molecule) to the target surface (Figure 1B); however, the relevance of this bridging function is still debated.^33–35 Although the molecular details remain to be elucidated, there is increasing evidence for a recruitment of C3b by P-selectin on platelets and endothelial cells that could induce or exacerbate complement activation.³⁶ FH-related proteins (FHRs) contain domains homologous to the C-terminal region of FH, which binds both C3b/iC3b/ C3dg and self-surface patterns. As a consequence, they can compete with FH for host surface binding and de-regulate complement activity on certain surfaces such as apoptotic cells.^37,38 Finally, members of the pentraxin family are known as context-specific modulators of complement activity but seem to mediate their activity via binding to PRMs and regulators rather than by directly interacting with C3 fragments.³⁹

As such, C3 and its fragments serve multiple purposes across the entire complement cascade: (i) by acting as an initiator of complement activation through tick-over of C3 and passive adsorption to surfaces (i.e. C3(H₂O)) and potential surface-capturing of C3b by modulators such as properdin or P-selectin; (ii) by driving the amplification loop as substrate (C3) and C3 convertase component (C3b); (iii) by functioning as direct effector proteins to mediate phagocytosis and/or immune stimulation/modulation (C3a, C3b, iC3b, C3dg); and (iv) by enabling the generation of terminal pathway effectors (C5a, MAC) as essential component of the C5 convertases (C3b). In order to comprehend how nature has designed and refined such a potent and versatile defense molecule, it is worth taking a journey from evolutionary to biochemical and deeply molecular aspects of C3.

3.1 | The anatomy of C3: One scaffold that determines many functions

The structure of C3 at 3.3-Å resolution was reported in 2005 and revealed an intricately shaped protein composed of 13 domains (Figure 4A–C).⁵⁸ The α-and β-chain each contain six individual domains (Figure 4D); surprisingly, and quite unusually, a thirteenth domain is shared between the two chains. The majority of the β-chain was found to consist of a “key ring”-shaped core formed by five domains with a hitherto-undescribed fold, which were termed MG domains to identify them as a common structural element of the α2M protein family (see above). The five complete MG domains (MG1–5) are followed by a partial MG domain (MG6^β) and a hydrophobic linker domain (LNK) that loops through the core ring. An anaphylatoxin (ANA) domain, which is prone to be released as the C3a fragment, defines the start of the α-chain. This domain is followed by the second part of the MG6 domain (MG6α) to complete the key ring and attach the remainder of the α-chain on top of the MG core. Following another MG domain (MG7), a split CUB domain (CUB^g, CUB^f) frames and holds the globular thioester-containing domain (TED) and leads to a final MG domain (MG8). A short anchor region finally connects the C-terminal complement domain (CTC; previously termed C345C^29,58). The unusual presence of two split domains (MG6, CUB) in C3 indicates that functionally critical elements (i.e. the proC3 processing site, C3a, and CUB-TED) were added by gene insertion during evolution and assumed increasing importance in host defense.⁵⁸

The arrangement of the 13 domains in native C3 largely explains its relatively inert nature when compared to its cleavage fragments. The orientation of the TED is of particular importance, as this domain contains the reactive thioester moiety that mediates covalent binding to targeted surfaces. In native C3, the thioester bond (Cys-988/Gln-991) is intact and tucked away in a hydrophobic pocket at the MG8-TED interface; although shielded from solvent contact, the thioester region still resides close to the protein surface.⁵⁸ Upon activation, Gln-991 forms a reactive acylimidazol intermediate with His-1104 that is further stabilized by Glu-1106; approximation of these residues is hindered in native C3, thereby providing another mechanism that prevents ready conversion and explains the slow hydrolysis rate (approximately 0.2%–0.4%/h) in this state.^58,81 Moreover, with the exception of the dimerization site on the MG ring that mediates the binding of C3 to convertases (see below) and the scissile loop of ANA/C3a, all known ligand-binding sites in C3b/iC3b/C3dg are buried or sterically hindered by the arrangement of the α-chain in native C3.

"VSports" 3.2 | Activation in variations: From tick-over to opsonization

Despite sharing the same domain organization with native C3, the hydrolyzed form C3(H₂O) has remained enigmatic, largely because of the absence of a high-resolution structure. Functional studies have long suggested that C3(H₂O) adapts a C3b-like structure, but one that still contains the ANA domain.^81,82 Initial structural evidence was provided by studies with monoclonal antibodies that revealed comparable neoepitopes after conversion of C3 to either C3(H₂O) or C3b.^78,83 Hydrogen-deuterium exchange mass spectrometry (HDX-MS) identified several areas with changes in solvent accessibility upon hydrolysis of C3, particularly in the α-chain and its TED domain.⁸⁴ Detailed mechanistic insights were offered by electron microscopy (EM) studies suggesting that native C3 represents a “kinetically trapped” conformational state, in which the bulky ANA domain acts as a “door stop” to prevent the formation of the reactive conformer.⁸⁵ In an unfavorable high-energy (and therefore rare) event, potentially enabled by unfolding of MG6, ANA may slip through the key ring of the MG core, leading to an irreversible conformational change with an extended TED-CUB and exposed thioester. Such displacement of the ANA domain in C3(H₂O), with rearrangement of CUB and TED alongside MG7 and MG8, has been supported by quantitative cross-linking mass spectrometry (QCLMS).⁸⁶ Although EM, QCLMS, and independent small-angle X-ray scattering studies indicate an arrangement of the TED-CUB-MG core interface that is slightly different in C3(H₂O) and C3b,^85–87 the relevance and functional implications of the difference are not entirely clear. Binding studies have indicated distinct yet sometimes conflicting differences in the interaction profiles of the two proteins toward FB and FH,^88,89 but additional insight from structural and/or functional assays will be necessary to reveal the full picture of this important complement initiator.

In contrast to its functional cousin, C3b is the best-characterized C3 fragment, with several high-resolution crystal structures describing this key opsonin in isolation or in complex with various ligands.^29,90–95 With the notable exception of one study, which has meanwhile been retracted,^96,97 all C3b-containing structures agree on the domain arrangement and have provided a rationale to explain the difference between the essentially inert C3 and the highly reactive C3b. The 4-Å crystal structure of C3b in isolation, published in 2006,⁹¹ showed that the β-chain of C3b remains largely unchanged when compared to C3, thereby identifying the key ring built by domains MG1–6 as a stable core element. In contrast, the α-chain in the activated form has undergone profound conformational changes, with the CUB in an extended position, the TED located at the other side of the MG core with the thioester approximately 95 Å from its original location, and the MG7–MG8 domains rearranged. The analogy of a “puppeteer” composed of a body (MG1–6, LNK), shoulder (MG7–8), neck (anchor), head (CTC), and arm (CUB), who holds a puppet (TED) was previously coined; in C3, the puppet is held up to the shoulder, whereas it is dropped in C3b, leading to an extended arm and a twisted shoulder (Figure 4E).⁹¹ This dropping of TED is considered critical for bringing the short-lived acylimidazole intermediate of the reactive thioester into close proximity with the target surface; a single electropositive patch around the thioester moiety may help orient the reactive C3b on negatively charged surfaces such as bacterial cells.⁹¹

Of equal importance is that the fact that the transformation of C3 to C3b reveals several previously cryptic binding sites that define its versatile involvement in activation, regulation, and signaling. For example, the repositioned CUB is involved in binding both the convertase precursor FB and most regulators of the RCA family, including CR1/CD35, CD46, CD55, and FH.^29,90,92 In the case of CRIg, the binding site for this receptor is located in the key ring area (MG3–6, LNK) that is present in both C3 and C3b; here, the specificity for C3b is determined by rearrangements of the MG3 and LNK domains upon activation.⁹⁴

3.3 | Fueling the powerhouse of complement: The inner workings of the AP C3 convertase (VSports app下载)

In many physiological and pathological situations, the C3 convertase of the AP is arguably the most powerful component of complement activation; the assembly, spatiotemporal activity, and regulation of this complex often determines the fate of a probed cell. In the absence of true specificity, it is expected that distinct safety mechanisms have evolved to prevent uncontrolled activity of the convertase. Surprisingly, and in contrast to most other steps in the cascade systems of host defense, the two enzymes involved in AP C3 convertase formation, i.e. FB and FD, circulate as mature enzymes rather than in zymogen form. Moreover, the same protein scaffold acts as both substrate (i.e. C3) and part of the enzyme complex (i.e. C3b). The fascinating story how this unique constellation is translated into an efficient yet controlled mechanism of complement activation and amplification was only fully reconstructed in recent years with the help of several structural studies by our group and others (Figure 5).

All the individual pieces of this puzzle had been contributed earlier via the crystal structures of C3 and C3b,^58,91 FB and its Bb fragment,^98,99 and FD.¹⁰⁰ However, the primary obstacle to putting the pieces together and achieving a structural image of the AP C3 convertase was the fact that the C3bBb complex irreversibly dissociates within a few minutes, thereby preventing successful crystallization. An attempt to solve the structure of the pro-convertase as an initial step involved the use of CVF, a snake-derived C3b/C3c that forms more stable convertases (see below), and a gain-of-function double mutant of FB (D254G/N260D¹⁰¹) to increase the stability of the complex.¹⁰² The resulting crystal structure at 2.2-Å resolution revealed that CVF and presumably, by analogy, C3b mainly interact with FB at its head and shoulder region via domains MG2, MG6, MG7, CUB, and CTC. Most of the contacts are provided by the Ba segment of FB, whereas the interface with the Bb segment is restricted to the CTC domain of CVF. In agreement with the long-standing notion that convertase formation is magnesium-dependent, the distorted metal-ion-dependent adhesion site (MIDAS) of FB was thought to adopt a high-affinity state in the CVF-bound form that chelated a Mg²⁺ ion together with the C-terminus of CVF. In contrast, and quite surprisingly, the rest of the FB structure, and in particular the catalytic region, shows little to no change when compared to free FB, thereby leaving critical questions concerning convertase formation unresolved.

An important breakthrough in this endeavor came with the use of the bacterial complement inhibitor SCIN (see below), which had previously been shown to trap the AP C3 convertase in a kinetically stable yet inactive state.^103,104 Indeed, in vitro convertase formation in the presence of SCIN resulted in stable complexes that could be crystallized at 3.9-Å resolution.⁹³ The crystal structure showed a dimeric and symmetric complex formed by two C3b, Bb, and SCIN molecules each (i.e. C3b₂Bb₂SCIN₂). When extracted from this structure, the two C3bBb complexes were highly comparable and showed that, like the CVF-FB structure, Bb was mainly bound to the CTC domain of C3b, suggesting that the high-affinity MIDAS state is maintained. As a result of this interaction, Bb appears to “dangle” from the top of C3b in the final AP C3 convertase. The dimeric nature of the SCIN-stabilized convertase complex also revealed critical information about potential substrate binding and processing. Intriguingly, the data showed that the two copies of C3b were dimerizing at a region that is retained in both C3 and C3b, pointing to a potential contact site between the C3 substrate and convertase-forming C3b. Indeed, replacement of one C3b copy in the dimeric complex with native C3 resulted in a model in which the protease domain in Bb is directed toward the ANA domain of C3. Although the catalytic site and scissile loop are still 30 Å apart, it was speculated that the presumed conformational flexibility of the CTC domain would allow Bb to swing closer to the C3 cleavage site in the uninhibited substrate-convertase complex and allow activation of C3.⁹³

With a refined model of the assembled convertase in hand, the question concerning the formation of the complex was reinvestigated. Under improved conditions, including Ni²⁺ instead of Mg²⁺ ions and again involving the FB gain-of-function mutant, it was possible to obtain a structure of the C3bB pro-convertase at 4-Å resolution.⁹⁰ Surprisingly, the domain arrangement of C3b-bound FB differed significantly from that observed in free FB and in CVF-FB.^90,99,102 Although the contact interface of the Ba segment and of Bb with the CTC domain of C3b remained largely the same, the serine protease (SP) domain of FB underwent a large rotation (84°), thereby forming additional contacts with the MG2 and CUB domains of C3b. It was speculated that the arrangement observed in CVF-FB resembles a closed “loading state” of FB, which dynamically transitions to an open “activation state” that would allow the binding of FD.^90,102,105 Indeed, co-crystallization of a functionally inactive mutant of FD with the proconvertase resulted in a C3bBD* complex, confirming that FD specifically binds to the open form of C3b-bound FB.⁹⁰ This interaction brings the catalytic site of FD into close proximity with the scissile loop of FB to allow its cleavage and the release of Ba. Importantly, FD interacts with C3bB via a hitherto unidentified exosite distant from the catalytic site. Moreover, FD possesses a self-inhibitory loop that prevents ready access to the catalytic center in circulation but is rearranged in the bound form to induce an active conformation only upon binding.

Taken together, these structural and additional binding studies provide fascinating insights into an intricate molecular mechanism tailored for instant, yet controlled, tagging of targeted particles (Figure 5).^90,93 Both C3 and FB are present in rather high plasma concentrations (1.2 and 0.2 mg/mL, respectively) to allow instant availability but do not react with each other in circulation. Similarly, the lower concentration of FD (0.002 mg/mL), its self-inhibitory loop, and the missing binding site in the closed form of unbound FB prevent a solution reaction between these two partners. The initial generation of an active form of C3, i.e. C3b or C3(H₂O), by any means on a target surface provides a binding platform for FB as the result of the exposure of new binding areas, primarily on the CUB domain. A fast initial interaction of the Ba segment of FB with this region and Mg²⁺-specific binding of the MIDAS to the CTC domain of C3b enable a quick but specific “loading” of the protease component to form the pro-convertase C3bB. Once bound, dynamic switching of the SP domain to an open conformation of FB allows the binding of FD; only then is the self-inhibitory loop of FD allosterically removed to give access to the catalytic center. Binding and cleavage of the scissile loop in FB release Ba and produce the final C3bBb convertase. Although Ba initially restricts C3’s access to the dimerization site in the pro-convertase C3b, its dissociation now allows the binding of the C3 substrate to the convertase. The presumed flexibility of the anchor and CTC interface likely enable the Bb part to swing closer to the scissile loop to cleave the ANA domain; the feasibility of such a movement has been supported by the arrangement of CVF in complex of C5.¹⁰⁶ Release of C3a removes the “door stop” that kinetically traps C3 and initiates the profound conformational change to C3b that quickly brings the newly formed, short-lived thioester intermediate into contact with the nearby surface and exposes another FB binding site to fuel amplification (Figure 5A).

This tiered interaction of three components, each necessitating a conformational transition, already presents an intricate means of intrinsic control with respect to convertase formation.⁹⁰ The low stability of the assembled convertase, with a half-life of approximately 3 minutes at 37°C, adds another layer of control. Binding studies have shown that the Ba segment is essential for pro-convertase formation because it provides rapid, Mg²⁺-independent loading and brings the MIDAS into contact with the C-terminus of C3b. After the release of Ba, the Bb-C3b continuously dissociates, but Bb cannot rebind in the absence of the Ba segment; even in the presence of Mg²⁺, no significant binding of Bb to C3b can be observed (Figure 5B).⁹³ Thereby, convertase activity relies on a constant supply of FB and the generation of new C3bBb complexes to maintain or even amplify the opsonic response. Moreover, this mechanism provides a basis for the ability of complement regulators to extrinsically control complement activation on host surfaces and confer selectivity between self and non-self (Figure 5B).

3.4 | A tale of two activities: A common binding mode defines regulatory functions

The N-terminal four domains of the soluble AP regulator FH exert both decay acceleration and CA. Crystallization of the C3b-FH[1–4] complex at 2.7-Å resolution therefore provided an important starting point for studying the molecular basis of these two major regulatory mechanisms.⁹² The structure revealed that FH[1–4] occupies a large, continuous contact area that covers an entire flank of C3b; despite this extended binding site, the interaction of FH[1–4] with C3b results in only a rather weak affinity of approximately 10 μM. The N-terminal portion of FH[1–4] binds at the top of C3b (the MG7 domain and α′-NT region), partially occupying the area where FB and Bb bind. Indeed, superposition of C3b-FH[1–4] with the C3bBb complex from the SCIN-stabilized convertase structure reveals a profound steric clash between FH and Bb that can be exacerbated by the N-linked glycan in Bb and electrostatic repulsion between the two proteins. It is therefore reasonable to propose that the DAA of FH is achieved by a competition for the Bb binding site on C3b, leading to a displacement of Bb and accelerated decay of the AP C3 convertase (Figure 5B). Cofactor activity, on the other hand, can be explained by the observation that C3b-bound FH[1–4] forms a new binding platform for FI that is mainly formed by CCP domains 1–3 of FH and the CUB domain of C3b.⁹² Subsequent docking studies have confirmed and extended this model and have shown that the binding of FI to this joint C3b-FH site releases allosteric inhibitory restraints in FI and brings the catalytic center of FI close to the scissile loop in the CUB domain of C3b to enable the first cleavage (between Arg-1281 and Ser-1282) to iC3b₁.¹⁰⁷

Recent studies have further refined our molecular understanding of RCA-mediated regulatory functions by comparing the co-crystal structures of C3b with CR1/CD35 (domains 15–17), MCP/CD46, DAF/ CD55, and the viral RCA mimic SPICE (see below).²⁹ The comparative analysis of these structures demonstrated that all these regulators invoke the same general binding mode as FH[1–4]. However, marked differences in the exact contact interface and the resulting binding strength of different domain interactions help explain their distinct regulatory profiles. For example, although the dual-activity regulator FH[1–4] engages all four N-terminal domains for C3b binding, the CA-selective MCP almost exclusively binds via its CCP3–4 domain at the lower half of C3b that is involved in forming the FI-binding site, and the DAA-conferring DAF predominantly occupies the areas of C3b in proximity to the FB site. Despite these differences, the common binding mode suggests that all regulators have evolved from a common origin and that MCP and DAF have enhanced their respective functions at the cost of dual activity.²⁹ In addition to improving our understanding of the molecular mechanisms of complement regulation, these C3b-RCA structures have already become and will continue to be essential for deciphering the functional consequences of the disease-related mutations and polymorphisms of RCA proteins.^4,29,92,108

3.5 | Access and restriction: Explaining the receptor specificity of C3-derived opsonins

One of the key features of complement-mediated opsonization is the unique ability to engage various immune receptors in a sequential, context-specific manner. In this regard, complement regulators not only impair the activation and amplification loop but also process the opsonins to forms that exhibit distinct signaling profiles. Interestingly, C3b contains all the receptor contact areas as preformed sites but cannot engage the entire spectrum of signaling tasks. The gain and loss of ligand binding during the breakdown from C3b to iC3b and C3dg is elegantly mediated by a series of remarkable transitions (Figure 6).

As discussed above, the activation of C3 to C3b leads to an extension of the CUB-TED interface that contains binding areas for members of the RCA family. Alongside their regulatory activities, some of the membrane-bound RCAs also act as adhesion and/or signaling receptors. For example, CR1 on erythrocytes is important for shuttling C3b-opsonized particles to the spleen and liver and is known to facilitate antibody-mediated phagocytosis via Fc-receptors.^6,109 The interaction of CD46 with C3b, on the other hand, can induce or modulate intracellular signaling events on various lymphocytes, including the induction of a productive Th1 phenotype in T cells.^110,111 Finally, the binding of CRIg to the key ring of C3b enables phagocytic clearance of opsonized particles by Kupffer cells and other tissue-resident phagocytes^94,112; of note, CRIg may also exert AP-specific regulatory functions by interfering with the binding of C3 to C3bBb and was recently reported to act as a direct PRM for Gram-positive bacteria.^112,113

The CA of certain RCAs initiates the degradation of C3b to iC3b by FI through the release of the C3f peptide from the CUB domain. There is currently no crystal structure of iC3b available, but EM studies have provided important but somewhat conflicting insights into the structural consequences of this transformation. Although some models predict a disordered CUB with a distant and flexible orientation between the C3d and C3c parts of the molecule,⁸⁵ other EM studies suggested a more compact conformation resembling the tucked-in TED in native C3.¹¹⁴ Recently performed HDX-MS experiments by our group comparing C3b and iC3b have found full solvent exposure across the entire CUB domain, suggesting complete unfolding and a loss of the interaction between TED and the MG core and supporting the initial EM results indicating a distant, flexible configuration (Ricklin and Lambris, unpublished observations). In this model, the removal of C3f and subsequent unfolding remove the RCA and FB binding sites on the previously ordered CUB domain and explain the loss of convertase formation and most RCA-related regulator/receptor functions in iC3b; however, the binding to CRIg remains intact.¹¹²

Conversely, an intact CUB sterically restricts the access of CR2/ CD21 to its binding site on the TED in C3b; unfolding of CUB eliminated this blockage and provided an explanation for the selectivity of CR2-mediated adaptive immune signaling for iC3b and C3dg. A similar situation likely applies to the integrin receptor CR3 (CD11b/CD18): Recent studies have revealed that at least one major binding area of the receptor is located on the TED.^115,116 Although distinct from the CR2 site, this CR3 contact region is also sterically hindered by CUB in C3b and only becomes accessible in iC3b. In contrast to CR2, however, additional binding sites on the C3c segment may be involved in the interaction with iC3b.¹¹⁵ Interestingly, the related CR4 (CD11c/ CD18) does not seem to bind the same site on TED but rather interacts primarily with the C3c part of iC3b, most likely at the MG3/MG4 interface.^115,117 As this area is preserved between C3b and iC3b, intracellular transformations would not account for the iC3b selectivity of CR4; the authors therefore speculated that the more flexible orientation of iC3b on the surface determines opsonin selectivity.¹¹⁷ Additional studies may be needed to fully explore the distinct binding modes of CR3 and CR4 toward complement opsonins.

The final transformation of iC3b to C3dg can only be mediated by CR1/CD35, but the molecular determinants of this enhanced CA are still being investigated. Although it has been shown that C3dg (40 kDa) can be further processed to C3d (35 kDa) after cleavage of the C3g peptide by various proteases in vitro, this transformation is currently not considered to be physiologically relevant under most circumstances.¹¹⁸ With the removal of the C3c segment, including its MG core, C3dg loses the ability to bind to CRIg. The interaction with CR2, however, and the associated B-cell stimulation function, are fully preserved in C3dg. In fact, the comparatively small size, high stability, and maintained B-cell activity make multimeric C3d a candidate for the development of novel vaccine adjuvants.^119,120 Although CR3 has long been considered to have strict opsonin specificity for iC3b, the above-mentioned molecular studies have confirmed the existence of a high-affinity binding site in the TED domain.¹¹⁵ Direct binding studies confirmed that the recombinant α_MI domain of CR3 interacts with iC3b, C3dg, and C3d but not with C3b.¹¹⁶ In vitro, we were further able to show that densely C3dg-opsonized erythrocytes from patients suffering from paroxysmal nocturnal hemoglobinuria (PNH) are recognized and phagocytosed in a CR3-dependent manner, suggesting a physiological role for the C3dg-CR3 interaction.¹¹⁶ Whether this route is dependent on C3dg density and how it compares kinetically with other uptake mechanisms in vivo remain to be further investigated.

6.1 | Taming the convertase

As the powerhouse of complement activation and amplification, the C3 convertases are an obvious promising choice to alleviate the detrimental effects of complement activation; this assertion is underscored by the fact that the majority of complement regulators, as well as many evasion proteins produced by human pathogens, interfere with the assembly, activity, or stability of the convertases (see above).²⁸ Importantly, these endogenous and exogenous convertase inhibitors already provide potent templates and starting points for the development of therapeutic convertase inhibitors. In the case of CP/LP C3 convertases, the clinically available C1-INH preparation (indicated for the treatment of hereditary angioedema) partially acts at this level by blocking serine proteases (i.e. C1r/C1s, MASPs) that are involved in cleaving C4 and C2, yet it also inhibits non-complement enzymes. Antibodies and molecules that specifically inhibit either C1s or members of the MASP family are therefore being developed.^161,190 The indications for such compounds are mainly focused on diseases with an established strong involvement of a distinct initiation pathway, as is the case for the CP in autoimmune hemolytic anemia for instance. For indications with a more complex initiation profile or predominant exacerbation via the amplification loop, targeting the AP C3 convertase might prove more rewarding.

As in the case of the above-mentioned approach for the CP/LP, inhibiting the serine proteases that drive the formation of the active convertase is a suitable approach.^190,192,194 Currently, the main focus appears to be on FD, which is the partner with the lowest abundance and which represents a bottleneck in convertase formation. For example, Genentech has developed an antibody fragment (lampalizumab) that binds to the exosite of FD and prevents its binding to the pro-convertase¹⁹⁵; in phase II clinical trials, intravitreal injection of lampalizumab has shown positive effects in a subset of patients suffering from a dry form of AMD (geographic atrophy), and Genentech has recently initiated phase III trials. Although lampalizumab appears to best suited for local application in the eye, at least two companies are working on small-molecule FD inhibitors that may be administered orally for systemic applications such as in PNH. Achillion recently started phase I trials with their lead compound ACH-4471, and a FD inhibitor from Novartis appears to be in preclinical development. Of note, Novartis is also working on small-molecule inhibitors of FB. It will be interesting to follow the clinical development of these low molecular weight inhibitors and learn more about their pharmacokinetic and pharmacodynamic behavior under clinical conditions. Other approaches to influence convertase assembly and stability are the use of antibodies against C3b, typically binding at the shoulder region and preventing binding of FB, or against properdin.^161,190,192

Even though the utilization of the natural convertase-regulating power of RCA proteins has long been considered, the direct use of physiological inhibitors such as FH or soluble CR1 has proved to be challenging, partly because of their comparatively large size (150 and 240 kDa, respectively). More recently, engineered smaller versions of human RCAs have entered the spotlight and shown some interesting benefits concerning tissue/cell targeting and efficacy. TT30, a fusion protein combining CCP domains 1–5 of FH with CCP1–4 of CR2, has played a pioneering role in this approach.¹⁹⁶ Although the FH part provides full AP regulatory activity, the CR2 segment mediates binding to iC3b and C3dg, thereby directing the regulator toward cells that are under complement attack. TT30 and related proteins have shown promising effects in various disease models, including PNH and ischemia-reperfusion injury^197,198; Alexion has evaluated TT30 in a phase I trial but has not announced further development plans.

Another approach to reducing the size of FH while conferring enhanced targeting properties has been taken by our group and others.^199–201 Analysis of the crystal structures of C3b-FH[1–4] and C3d-FH[19–20] have shown that the two functionally important termini are situated in close proximity and can be bridged to produce a mini-FH protein with 6 instead of 20 domains.¹⁹⁹ Interestingly, the recombinant construct not only maintains binding affinity for C3b and GAG but also shows higher activity in AP inhibition models than does FH. This mini-FH achieves stronger binding to the late-stage opsonins iC3b/C3dg, likely because of the better accessibility of the C3d binding site in FH[19–20] after the removal of the middle segment and because the connectivity between the domain areas plays an important role.²⁰² As a consequence, mini-FH analogs are ideally targeted toward host surfaces under complement distress. Despite missing the surface-recognition part of FH, the splice variant FHL-1 may also be considered as a therapeutic option (e.g. for local applications) because of its small size and presumably improved tissue penetrance.²⁰³ More recently, CRIg and the LP regulator MAP-1 have been explored as means of directing regulators to sites of ongoing complement activity.^204,205 The modular organization of most complement regulators has certainly opened fascinating possibilities for inhibitor design.

6.2 | Protecting the substrate: Direct inhibition of C3 by compstatin analogs

Convertase inhibitors can exhibit powerful activity but are typically pathway-specific and prevent C3 activation via either the CP, LP, or AP. In some situations, however, total blockade of C3-mediated opsonization and effector generation is desired. Such an effect can be achieved by peptides of the compstatin family, which bind to a functional hotspot on the C3 substrate and prevent its activation by any of the C3 convertases.^206,207 Originally discovered by our group in 1996 during phage-display peptide library screening,²⁰⁸ the compstatin family has since grown to include several distinct analogs with unique properties, two of which are currently in clinical development.^190,206

The phage display-derived compstatin was identified as a cyclic peptide of 13 amino acids that binds to human C3 (and C3c-containing fragments) and potently inhibits both AP and CP/LP-induced C3b deposition.²⁰⁸ Importantly, compstatin showed a narrow species specificity toward C3 from humans and non-human primates.²⁰⁹ From a molecular perspective, the cyclic nature of the peptide was found to be essential, and several key residues were identified.²⁰⁷ Based on this information, and in the absence of structural data, the first rounds of optimization using natural amino acids resulted in an analog (4W9A) that was acetylated at the N-terminus and included two amino acid substitutions (V4W and H9A), thereby gaining a 13-fold increase in affinity.^207,210 The use of non-natural amino acids, and in particular use of an indol N-methylated variant of Trp in position 4, improved the activity gain to more than 260-fold over the original compstatin.^207,211,212 Early on, some of these compstatin analogs showed promising efficacy data, for example in ex vivo models of kidney transplantation or biomaterial-induced complement activation.^213–215 The lead analog at that point (2006) was licensed to Potentia Pharmaceuticals; this analog performed successfully in phase I trials in AMD but did not reach the desired endpoints after administration of much lower doses in subsequent phase II trials. Meanwhile, the technology was transferred to Apellis Pharmaceuticals; although the unmodified compound (APL-1) is being considered for use in COPD, a PEGylated derivative (APL-2) has been clinically developed for local administration in AMD and for systemic application in PNH.^161,190

The lack of a crystal structure made it challenging to obtain a rational optimization of this promising lead compound. In addition, its molecular mechanism of action has long been unclear, as biochemical analyses have localized the compstatin binding area to a part of the β-chain that was considered distant from both the C3a cleavage site and the FB binding region.²¹⁶ These gaps were filled in 2007 with the publication of a crystal structure of C3c in complex with the compstatin analog 4W9A.²¹⁷ The structure revealed that compstatin binds to a shallow groove at the interface between MG4 and MG5 on the outside of the key ring of C3. An early hypothesis that this binding might interfere with the interaction of the C3 substrate and the convertase was subsequently confirmed by the structure of the SCIN-stabilized C3bBb complex (see above).^93,217 Superposition of these structures indeed showed that the compstatin binding site is located on the C3/C3b dimerization interface, which is presumed to be essential to allowing the binding and activation of C3. Although the binding of a second compstatin molecule to C3b of the convertase complex can enhance this effect in the case of the AP, the potent inhibition of CP/ LP-mediated complement activation suggests that blockage of the substrate is sufficient to achieve the inhibitory effect.

Binding site analysis and the use of tools from peptide and medicinal chemistry, including backbone N-methylation, have enabled the generation of next-generation analogs with a marked improvement in C3 binding affinity and inhibitory activity.^{163,206,211,218} The lead analog, Cp40, was the first analog with subnanomolar binding affinity for C3 (K_D=0.5 nM), marking an approximately 5000-fold improvement over the initial compstatin.¹⁶³ Structural analysis showed that N-methylation of Gly induces a bound-like conformation of Cp40 in solution, which may improve target binding. Moreover, the addition of a D-Tyr at the N-terminus extends the binding site and forms additional contacts with C3.¹⁶³ Despite the strong activity gain, Cp40 maintains its species specificity for human/primate C3. The tight target binding and the high plasma concentration of C3 also contribute to the unique pharmacokinetic properties of Cp40, which remains in the circulation much longer than do typical peptide drugs.^163,206 Studies in NHP have demonstrated that target-saturating drug concentrations can be maintained with repetitive subcutaneous injections of as little as 1 mg/kg Cp40 and administration intervals of 12–24 hours, or even less frequent intervals [(163, 206) and Ricklin & Lambris, unpublished observations].

Next-generation compstatin analogs, and in particular Cp40, have been used in a variety of disease models and have shown promising preclinical efficacy in PNH,²¹⁹ hemodialysis-induced complement activation,²²⁰ C3G,²²¹ sepsis-associated organ damage,^222,223 xenotransplantation,^224–226 hemorrhagic shock,²²⁷ malarial anemia,²²⁸ and periodontal disease,^185,229,230 among others. The last model is especially intriguing because periodontal disease is primarily caused by a microbial pathogen (i.e. Porphyromonas gingivalis), making complement inhibitors a rather counterintuitive choice. However, it has been shown that P gingivalis actively induces an inflammatory milieu to gain a competitive advantage, and in the process causing dysbiosis, inflammation, and bone loss; the anti-inflammatory effect of complement inhibition by Cp40 re-shifts the balance and markedly improves both immunological and clinical parameters.^229,230 This examples nicely illustrates that, as discussed above, complement inhibition needs to be considered in the specific context involved; in some indications, taming the deleterious effects of complement-mediated inflammation may be more important than controlling infection. Cp40 was licensed by Amyndas Pharmaceuticals and is currently being developed for the treatment of PNH, ABO-incompatible kidney transplantation, C3G, and periodontal disease.

"VSports" 6.3 | Moving ahead toward C3 inhibitors in the clinic

Although the development of C3-level complement inhibitors with clinical potential took substantially longer than expected, several highly promising candidates are now waiting in the wings. The overcoming of technical hurdles played an important role in this advancement. As demonstrated by Cp40, feasible treatment schemes for maintaining target-saturating levels may even be achieved for abundant targets such as C3. Although pharmacokinetic parameters may be even further improved by chemical modifications, for example by employing PEGylation or albumin-binding moieties,^219,231 the cost-benefit ratio needs to be carefully considered in each case. Another game-changer in the development of therapeutic C3-targeted drugs is the regained trust in complement inhibition in general, including the blockage of C3.^4,190,192 The clinical experience has been largely positive and, in the case of eculizumab, the potential risk of infection by meningococci can be addressed by vaccination. For C3-targeted inhibitors, such anti-infective strategies may need to be expanded to other encapsulated bacteria such as pneumococci. Importantly, the risk of C3 inhibition needs to be considered in the right context, as local, systemic, acute, and chronic treatment present highly distinct situations. The clinical trials that are being conducted at this point and in the near future will be critical for advancing our current mainly hypothetical questions about the benefits and risks of blocking C3 activation in humans to evidence-based discussions.