key: cord-0270519-bd9fh00n
authors: Braunger, Katharina; Ahn, Jiyoon; Jore, Matthijs M.; Johnson, Steven; Tang, Terence; Pedersen, Dennis V.; Andersen, Gregers R.; Lea, Susan M.
title: Targeting properdin - Structure and function of a novel family of tick-derived complement inhibitors
date: 2021-04-04
journal: bioRxiv
DOI: 10.1101/2021.04.02.438250
sha: 1f2e315d9e7cbed40e7dec27e0447b4981c61c21
doc_id: 270519
cord_uid: bd9fh00n

Activation of the serum-resident complement system begins a cascade that leads to activation of membrane-resident complement receptors on immune cells, thus coordinating serum and cellular immune responses. Whilst many molecules act to control inappropriate activation, Properdin is the only known positive regulator of the human complement system. By stabilising the alternative pathway C3 convertase it promotes complement self-amplification and persistent activation boosting the magnitude of the serum complement response by all triggers. We have identified a novel family of alternative pathway complement inhibitors, hereafter termed CirpA. Functional and structural characterisation reveals that CirpA family directly bind to properdin, inhibiting its ability to promote complement activation, and leading to potent inhibition of the complement response in a species specific manner. For the first time this study provides a full functional and structural characterization of a properdin inhibitor, opening avenues for future therapeutic approaches.

The complement system plays pivotal roles in immunity and cellular homeostasis 1, 2, 3 . The 35 tightly regulated proteolytic cascade can be activated through the Classical pathway (CP), Lectin pathway (LP) and Alternative pathway (AP), all of which ultimately lead to the formation of a surface-bound AP C3 convertase C3bBb. This then cleaves C3 into C3a and C3b, inducing a self-amplification loop, and leading to opsonisation with C3b and release of the anaphylatoxin C3a. Exceeding a certain density of surface bound C3b triggers a shift in substrate preference 40 towards C5 4 . The resulting production of the anaphylatoxin C5a, and the first component of the membrane attack complex (MAC) C5b, initiate the terminal steps of the cascade.

A plethora of host proteins are known to attenuate the response, as excessive, inappropriate, or prolonged complement activation can be highly detrimental to host tissue 5, 6 . A striking example is constituted by the disproportionate immune response seen in many severe COVID- 5 19 cases of the ongoing pandemic 7, 8, 9, 10 . In contrast, only a single positive regulator for has been identified in the complement system: Properdin, also known as factor P (FP) 11, 12, 13 .

FP natively forms oligomers which promote assembly of the C3bB proconvertase and stablise the AP C3 convertase, thereby extending its half-life 5-to 10-fold and inhibiting its degradation by factor I 12, 14, 15 . Besides this, there is an ongoing controversy about FP's potential role as a 10 pattern recognition molecule, in which it is suggested to bind directly to activating surfaces, thereby recruiting C3 and inducing complement activation 16, 17 .

FP is a highly flexible protein consisting of an N-terminal TGFb binding (TB)-domain followed by six thrombospondin type I repeats (TSR)-domains. The convertase binding region lies in a vertex formed between two FP molecules, interacting head-to-tail. High resolution structures 15

of pseudo-monomeric FP in isolation and bound to the C-terminal domain of C3b, as well as a moderate resolution structure of a FP-bound AP C3 convertase, have greatly improved our understanding of FP multimerization and convertase stabilization 18, 19, 20 .

FP inhibition leading to complement downregulation has the potential to ameliorate conditions caused by the overactivation of complement on host tissue in a way which is preferable over 20

inhibition at other stages. By acting at the level of the AP C3 convertases, FP-inhibition leads to a reduction in both C3a and C5a, as well as reduced MAC formation. FP-inhibition may also preserve non-amplified complement activation via CP and LP, allowing for some level of continued clearance of apoptotic cells. Additionally, FP concentration in serum (5-25 µg/mL) is much lower than for example C3 (1.2 mg/mL) or C5 (75 µg/mL), such that a lower dose of an 25 inhibitory agent might be sufficient to demonstrate therapeutic effects 21 .

Tick-saliva presents a vast source for pharmaco-active proteins, targeting a wide array of immune mechanisms 22 . Their therapeutic potential can be illustrated with the example of Nomacopan (Coversin, Akari therapeutics), a recombinant variant of the lipocalin family protein OmCI (Ornithodoros moubata Complement Inhibitor), which is currently under investigation in 30

several clinical trials at various stages. Its mechanism of action is binary, with tight binding of C5 inhibiting the terminal pathway of complement 23, 24 , and sequestration of the proinflammatory eicosanoid leukotriene B4 (LTB4) within an internal binding cavity 25 providing additional anti-inflammatory function.

In this study we identified a novel family of complement inhibitors from the hard tick 35 Rhipicephalus pulchellus, hereafter termed CirpA (Complement inhibitor from R. pulchellus of the alternative pathway). CirpA1 is shown to target human properdin, leading to potent AP inhibition. Functional analysis of six CirpA homologs demonstrates a highly species dependent activity profile.

In addition, we present crystal structures of four CirpA homologs, revealing a conserved 40 lipocalin fold, and the crystal structure of the properdin-CirpA1 complex.

Our work represents the first comprehensive functional and structural characterisation of a properdin inhibitor. The in-depth structural and functional analysis gives insights into the mechanism of properdin binding and reveals a remarkable variety in lipocalin inhibition mechanisms. 45

Fractionation of R. pulchellus salivary gland extract (SGE) had previously led to identification of the CirpT family of complement inhibitors 26 . Strikingly, the first chromatographic step indicated a distinct second population with inhibitory activity specific for the alternative pathway 5 of complement activation (Extended Data fig 1a) . Using an analogous procedure, we further purified and enriched fractions with inhibitory activity, demonstrated by reduced MAC deposition in a standard complement alternative pathway activation assay using human serum (Fig 1a, b ; Extended Data Fig. 1a,b) .

The proteins in the active fractions were analysed by electrospray ionization-MS/MS following 10 trypsin digest. Analysis against two published cDNA databases 26, 27 resulted in a list of nine proteins containing a predicted signal peptide (Extended Data Fig. 1c ). The candidates were expressed recombinantly in Drosophila melanogaster S2 cells and the culture supernatants tested for complement inhibitory activity. Only one protein, subsequently termed CirpA1, was shown to specifically inhibit the alternative pathway of complement activation (Fig. 1c,  15 Extended Data Fig. 1d ). CirpA1 significantly inhibited the formation of both C3a and C5a via the alternative pathway (figure 1d, e) suggesting it acts at the point of C3 cleavage.

To pinpoint the target of CirpA1, we performed a pull-down assay from human serum utilizing immobilised CirpA1. Western Blot analysis resulted in a single positive hit: Properdin, also known as complement factor P (FP, Figure 1f ). FP is the only known positive regulator of the 20 human complement system. In the alternative pathway, FP binds to the C-terminal domain of C3b in the C3 convertase (C3 C345c), thereby increasing convertase half-life approximately ten-fold. To gain a better understanding how CirpA1 binding to FP leads to inhibition of the alternative pathway of complement activation we investigated how it interferes with FP function. We generated biotinylated C3b from purified C3 to couple it to streptavidin magnetic 25 beads in a physiological orientation 4 . We incubated the C3b coupled beads in human serum, supplemented with EDTA and EGTA to block complement activity, in presence or absence of CirpA1. Subsequent western blot analysis clearly shows that CirpA1 abolishes FP-C3b binding. Furthermore, CirpA1 is able to act in a competitive manner, releasing pre-bound FP (Fig 1g) . 30 Figure 1 Identification and functional characterisation of CirpA1 from Rhipicephalus pulchellus salivary glands. (a) Experimental procedure leading to the identification of CirpA1 (b) Fractionation of salivary gland extract hydrophobic interaction chromatography (top) used to identify fractions with inhibitory activity against the alternative pathway in a AP haemolysis assay (bottom). (c) Inhibition of complement pathways by CirpA1 measured in WIESLAB assays using human serum. CirpA1 specifically inhibited 5 the alternative pathway. Values were normalised for no-serum samples (0% MAC formation) and no-additive samples (100% MAC formation). Error bars, s.e.m. (n = 3). Curve fitting was carried out in GraphPad Prism using a dose response inhibition (variable slope) model. AP IC50 = 13 nM. (d-e) C3a (d) and C5a (e) levels in supernatants of the alternative pathway Wieslab assay. CirpA1 inhibited C3a and C5a formation through the alternative pathway; -ve = Ra-HBP2; +ve = EDTA. All proteins were added at a final concentration of 1 μM. Values were baseline-corrected for buffer-only samples. Error bars, s.d. (n = 3 Wieslab samples). ns (not 10 significant); ****P < 0.001 by unpaired two-tailed t-test, with no-additive sample as the reference. (f) CirpA1 binds FP, anti-FP blot using CirpA1 coated beads from serum compared to +ve and -ve controls (g) Anti-FP Western Blot after serum incubation of C3bcoupled beads in presence or absence of CirpA1 shows CirpA1 is able to prevent binding to C3b (E1) and displace the majority of pre-bound FP (E2, FT2); -ve = no added serum; +ve = no inhibitor.

To identify potential biologically relevant homologs, the CirpA1 sequence was used to query the expressed sequence tag database (NCBI) as well as in-house R. appendiculatus 28 and R. pulchellus sialomes. This search revealed five homologs across different tick species (Fig 2a, Extended Data fig. 2b ) with varying sequence identity (hereafter named CirpA2-A6, pairwise sequence identity to CirpA1 43%-82% over ~200 residues). To compare their activity all CirpA 5 homologs were expressed in D. melanogaster S2 cells and assessed regarding their ability to inhibit alternative pathway complement mediated haemolysis via sera from multiple mammalian species.

We tested inhibitory activity against human, monkey, rat, and guinea pig serum (Fig. 2b) . CirpA1 showed comparable activity in human and monkey sera (IC50= 1.41x10 -8 and 1.38x10 -10 8 respectively) but only weak effects in rat serum and was inactive in guinea pig. Interestingly, for CirpA2, which shares 82% of amino acids with CirpA1 and only differs in the C-terminal 57 residues, the inhibitory potential is lost in human and strongly reduced in monkey. In contrast, CirpA6, the homolog with the lowest sequence identity to CirpA1 (43%), shows almost identical inhibition behaviour in human (IC50= 1.40x10 -8 ), considerable activity in monkey and weak 15

inhibition in rat. The homologs CirpA3-5 were inactive in the species tested in this study ( Fig  2b) .

To understand the distinct inhibition profiles, we set out to determine the structure of the 20 different CirpA homologs. To that end we overexpressed the proteins in Escherichia coli and purified them via refolding from inclusion bodies. Using this strategy, we were able to purify CirpA1, A3, A4 and A5 and determine their structure using X-ray crystallography to a resolution of 2.0 Å, 2.1 Å, 1.8 Å and 1.9 Å respectively. (Fig 2 c , d, Table 1 ).

The structure of CirpA1 was solved by molecular replacement using the tick-derived 25 complement inhibitor OmCI (pdb-id: 3ZUI) as an initial search model. The refined model of CirpA1 was used to determine the structures of CirpA3-5. All CirpA structures share a lipocalin fold with a short N-terminal a-helix followed by an eight-stranded beta-barrel and a longer Cterminal a-helix. Overall, the structures appear very similar, with the biggest conformational differences in the tilt angle of the C-terminal helix H2 as well as variations in loops between 30 beta strands 4 and 5 (L4-5) and between strands 7 and 8 (L7-8). 

Regions of high conformational variability between homologs indicated by dashed ellipsoids.

We next aimed to get more detailed insights into the mechanism of function of CirpA1 to potentially understand differences in inhibitory potential between the homologs. We first investigated whether FP was still able to form multimers upon binding of CirpA1. Size exclusion analysis of purified FP in presence of CirpA1 resulted in clear shifts towards higher molecular 5

weight for all FP populations with no noticeable changes in population ratio (fig 3a) . Hence, we concluded that CirpA1 binds all multimeric forms of FP without affecting multimer distribution.

Next, we set out to determine the structure of the CirpA1-FP. We chose to work with a minimal FP construct that had previously been used to determine the crystal structure of pseudomonomeric, minimal functional unit of FP (hereafter referred to as FPD2,3) 18, 19 . It consists of 10 a head-fragment with the TB domain as well as TSR1 and a tail fragment of TSR 4-6. Coexpression of both fragments in HEK293F cells results in formation of a vertex mimicking FP's physiological multimerization behaviour and is able to bind the C3 convertase 18 . Addition of CirpA1 to purified FPD2,3 results in a distinct shift of molecular weight (fig 3b) , as measured by SEC-MALS, indicating formation of a complex at 1:1 stoichiometry. 15

We were able to obtain crystals from the FPD2,3-CirpA1 complex and determine its structure to 3.4 Å by X-ray crystallography. The structure was solved by molecular replacement, using models of CirpA1 (this study) and FPD2,3 (pdb-id: 6S08) as search models. In the crystal lattice each FPD2,3 molecule interacts with two inhibitor molecules (fig 3c) . Knowing from our SEC-MALS analysis that each FPD2,3 binds only one CirpA1 in solution, we designed two CirpA1 20 mutants, introducing arginine residues at distinct positions in each interface (E170R/V173R and Q148R) which based on our structure would likely disrupt the corresponding interaction with FP. Using an alternative pathway haemolysis assay, we were able to show that CirpA1 E170RV173R displays similar properties as the wild-type. In contrast, the CirpA1 inhibitory effect is lost in the Q148R mutant indicating that interface 2 is the functionally relevant one. Analysis of the interaction surface, which is burying 864 Å 2 , reveals complementary charged patches, suggesting that binding is mediated by electrostatic interactions (Fig 4a) . 10 Superposition with structures of isolated FPD2,3 as well as FPD2,3 in complex with the Cterminal domain of C3b shows subtle conformational changes in the binding region (Fig. 4b) .

In CirpA1, FP binding leads to structural changes in the three C-terminal beta strands, as well as specific rearrangements of individual amino acids contributing to complex stability (Fig 4c,  Suppl. Fig 2) . One example is Q148 which forms a hydrogen bond to FP Q338 (Fig 3f and Fig  15  4c ). Superposition with structures of the CirpA3-5 shows that CirpA3 and CirpA4 could rearrange analogously to form this interaction, while in CirpA5 Q148 is replaced with a glycine residue (Fig 4d) . Interestingly, sequence comparison of FP between the species tested for CirpA activity shows that Q/E at position 338 is unique to primates. Our structure suggests that substitution with aspartate (rat) or lysine (guinea pig) would likely weaken the interaction, 20

contributing to CirpA1 activity loss in these species (Fig 4e) .

Another CirpA1 residue undergoing a conformational rearrangement upon FP binding is Y122 which forms a hydrophobic pocket around properdin P377 together with CirpA1 residues F122, Y131 and F150. This pocket is incomplete in all three CirpA1 homolog structures.

Together with the fact that the CirpA1 regions forming the FP binding interface overlap with the regions of highest conformational variability between the CirpA homolog structures (Fig 4f) , 5

these changes might provide an explanation for lack of human FP-inhibition activity of CirpA3-A5 in our analysis. In a broader context of complement activation, CirpA1 is not the first known tick lipocalin mediating potent inhibition. The complement inhibitor OmCI binds the C5 in close vicinity to its 20 C345C domain and prevents cleavage of C5 by the C5 complement convertases, thereby preventing release of anaphylatoxin C5a and formation of the terminal membrane attack complex (MAC) (Fig 5a) . In contrast, CirpA1 acts at an upstream step, by binding to FP and interfering with assembly and stabilisation of the C3 convertase. Remarkably, despite their high structural similarity, OmCI and CirpA1 bind their targets in very different regions of the lipocalin 25 surface (Fig 5b-d) . In addition, OmCI displays a second mechanism of function by binding the proinflammatory eicosanoid leukotriene B4 (LTB4) within an internal binding cavity. The corresponding pocket in CirpA1 is blocked by a cluster of charged residues, precluding binding of a hydrophobic ligand in this position (Fig 5 e,f) . 

We have identified a novel complement inhibitor from the hard tick R. pulchellus. Our functional and structural analyses show that it targets FP, the only known positive regulator of the human 15 complement system. Sequence analyses identified five additional members of the CirpA family, with pairwise sequence identities to CirpA1 between 43-82%. Their anti-complement activity does not seem to correlate with the degree of sequence conservation and is specific to serum of certain species. To understand the differences in inhibition profile, we determined the structures of CirpA1, CirpA3, CirpA4, CirpA5 and CirpA1-FPD2,3. The structures revealed a 20

lipocalin-like fold for all CirpA homologs. We identified three regions with high conformational variability between homologs. All these regions are part of the binding interface between CirpA1 and FP, and hence the conformational differences might render other homologs incompatible with FP binding.

The natural hosts of the three tick species producing the characterised CirpA homologs are 25 mostly cattle and other larger domestic animals. In light of that, it seems likely that the proteins have evolved to target other species, which were not tested in this study. Lipocalins are known to be functionally diverse 29, 30 . In addition to binding small, often hydrophobic molecules in designated pockets in their beta-barrel, some have additional functionality by engaging in protein-protein interactions. Therefore, the lack of inhibition of AP mediated haemolysis 30 observed for CirpA2-5 might also be due to the fact that their activity lies within a different aspect of the complement cascade, or even in another biological pathway.

The functional versatility of lipocalin tick inhibitors is further illustrated by our comparison of CirpA1 and OmCI. OmCI binds leukotriene B4 and forms a complex with complement C5, thereby inhibiting the terminal steps of the complement response. Evaluating the structures of these inhibitors in isolation and in complex with FP or C5 shows that both their internal pocket as well as their protein binding regions are very distinct from each other. Nevertheless, both 5

lipocalins cause a potent block for the corresponding complement pathways.

The CirpA1-FPD2,3 structure also sheds light on how CirpA1 interferes with properdin function. High resolution structures of isolated FPD2,3 and the C3b-c345c-FPD2,3 complex, together with a moderate resolution structure of convertase-bound pseudo monomeric FP, have established that FP interacts with the convertase mainly through contacts with the C-terminal 10

C345c domain of C3b 18, 19, 20 . It binds the convertase via TSR5 and TSR6 near the factor B binding site. It is plausible that binding is strengthened through direct contacts between FP and Bb, but resolution limitations did not allow to unambiguously pinpoint interacting residues. Two distinct loops in FP were shown to play a crucial role in binding to C3b: residues 328-333 (336 in 18 ) (TSR5-stirrup or thumb) and residues 419-426 (TSR6-stirrup or index finger). They  15 embrace the C-terminal helix of C3b and might be involved in stabilising the Mg2+ coordination between C3b and Bb.

In context of the properdin bound C3 convertase, binding of CirpA1 results in a steric clash between CirpA1 and the protease component Bb. This direct competition might contribute to the CirpA1 inhibitory effect in vivo. However, it cannot be the only component explaining 20

CirpA1-mediated inhibition as our work shows that CirpA1 prevents FP binding to C3b in the absence of FB, and promotes dissociation of a preformed C3b-FP complex. Therefore, the CirpA1 inhibitory mechanism is likely to interfere directly with the FP-C3b interaction. Notably, the TSR5-stirrup is directly involved in CirpA1 binding. The TSR6-stirrup lies in immediate vicinity to another the CirpA interacting region and appears to be disordered in the CirpA1-25 FPD2,3 structure. While it is similarly flexible in isolated FP D2,3 it assumes a stable conformation upon C3b binding. We therefore rationalise that binding of CirpA1 impacts the conformational freedom of the stirrup loops thereby destabilising the interaction with C3b and possibly preventing conformational changes required for binding to the convertase.

Combining together the results from our structural and functional analyses, we are proposing 30 a model for CirpA1-mediated complement inhibition that mimics the decay acceleration models of convertase regulation (Fig. 5g) . In the absence of the inhibitor, the C3 convertase is bound and stabilised by multimeric FP. In presence of CirpA, the inhibitor binds free FP at the TSR5/6 junction, thereby blocking FP from binding the C3 convertase. Furthermore, CirpA1 is capable of binding to convertase-engaged FP molecules, triggering FP displacement. 35

In summary, our study represents the first all-inclusive characterisation of a human properdin inhibitor covering isolation from its natural source, target identification, functional as well as structural characterisation. In addition, our structure of inhibitor-bound properdin sheds light on the mechanisms involved in properdin-convertase binding, thereby marking an important contribution to understanding the human complement response. 40

R. pulchellus ticks were reared, and 250 salivary glands from 6-day fed female R. pulchellus were dissected according to Tan et al. 27 . The gland protein extract was topped up with 25 mM 45

Na2HPO4/NaH2PO4, pH 7.0 to 10 mL. The sample was then fractionated by sequential anion exchange, hydrophobic interaction chromatography and size exclusion chromatography (SEC). At each stage, eluted fractions and flow-through from the chromatographic columns were assayed for complement inhibitory activity, and the active fractions were further fractionated. First, protein extract was fractionated by anion exchange chromatography using a MonoQ 5/50 GL column (GE Healthcare), washed with 10 column volumes (CV) 25 mM Na2HPO4/NaH2PO4, pH 7.0, and eluted by a 0 to 0.5 M NaCl gradient over 30 CV in 500-μL 5

fractions. Active fractions were combined, mixed with an equal volume of 3.4 M (NH4)2SO4, pH 7.0, centrifuged (22,000 x g, 10 min), and the supernatant topped up to 5 mL with 1.7 M (NH4)2SO4, 100 mM Na2HPO4/NaH2PO4, pH 7.0. The sample was loaded onto a 1-mL HiTrap Butyl HP column (GE Healthcare) and washed with 5 CV of 1.7 M (NH4)2SO4, 100 mM Na2HPO4/NaH2PO4, pH 7.0. Elution was carried out by a 1.7 to 0.0 M (NH4)2SO4 gradient 10 over 15 CV in 1-mL fractions. The active fraction was topped up to 220 μL with PBS, separated by a Superdex S75 10/300 column (GE), and collected in 250 μL fractions. All fractions were buffer exchanged to PBS and concentrated.

Identified protein fractions with complement-inhibitory abilities were digested by Trypsin and 

Codon-optimized GeneArt strings were cloned cloned into pExpreS2-2 vector with the insect BiP signal sequence followed by an N-terminal 6-His tag. Transfections into Drosophila 30 melanogaster S2 suspension cells were carried out following the manufacturer's instructions (Expres2ion Biotechnologies).

CirpA1 was codon-optimised, cloned into pETM-14 using the NcoI and NotI restriction enzymes and transformed into E. coli BL21 (DE3) (New England Biolabs). CirpA3, CirpA4 and 35

CirpA5 were codon optimised, cloned into pET15b vector using the restriction enzymes NcoI and XhoI and transformed into E. coli BL21 (DE3) (New England Biolabs). Protein expression was carried out in LB broth (with 50 μg/mL kanamycin). Cells were induced with 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) for 16h at 20 C, resulting in expression in inclusion bodies. 40

The supernatant was harvested and filtered. Expressed proteins were subsequently purified using a cOmplete His-Tag Purification column (1mL column, Roche) and SEC (S200, 16/60, GE Healthcare) in PBS.

Cells were harvested by centrifugation and lysed by homogenisation. The lysate was spun at 20,000 rpm, 4 C for 30 min and the cell pellet fraction was re-suspended in 40 ml of Buffer A (50 mM Tris, 150 mM NaCl, pH 8, 0.05% (v/v) Tween 20) using a hand-held homogeniser. After incubation on a rotary wheel at 4 oC for 1 h, the re-suspended pellet was centrifuged at 20,000 rpm at 4 C for 30 min. This wash step was repeated once more. The cell pellet was 5 then resuspended in 40 ml of Buffer B (8 M urea, 1 mM EDTA, 100 mM Tris, 25 mM DTT, pH 8) for solubilisation of inclusion bodies, and incubated on a rotary wheel at 4 C for 2 h. Following centrifugation, the supernatant fraction was retained and filtered. For refolding, the supernatant was added dropwise to Buffer C (1 mM cysteine, 2 mM cystine, 20 mM ethanolamine, 1 mM EDTA, pH 11) with stirring. The protein-containing refolding buffer was left overnight at 4 C, 10 and then concentrated down to ~40 ml using a 10 kDa Vivaflow 200 filtration device (Sartorius). For CirpA1, the His-Tag was removed following concentration via cleavage with 3C protease whilst dialysing against 2 l of Buffer D (50 mM Tris, 150 mM NaCl, pH 8) overnight at 4 C. Uncleaved material was removed by reverse nickel purification. All inhibitors were purified by SEC (S75, 26/60, GE Healthcare) in Buffer D. CirpA3, CirpA4 and CirpA5 were further purified 15 by ion exchange chromatography (Mono Q 5/50 column, GE Healthcare). Homogeneous protein populations eluted in the peak (CirpA3, CirpA4) and flow-through (CirpA5) fractions and were dialysed against 50 mM Tris, 20 mM NaCl, pH 8.5, overnight at 4 C.

An affinity column was generated with CirpA1 using the Pierce NHS-Activated Agarose Slurry 20 (Thermo Scientific) following the manufacturer's instructions. In brief, CirpA1 was mixed with the slurry on a rotary wheel overnight at 4 C. After coupling, the remaining active sites were blocked with 1 M ethanolamine. The slurry was then packed into an empty cartridge to generate a CirpA1 'column'. Normal human serum was acquired from the John Radcliffe Hospital, Oxford. After loading the serum onto the CirpA1 column and washing with PBS, properdin was 25 eluted with 0.2 M glycine-HCl, pH 3. The pH of the elution fraction was adjusted by adding 50 μl of neutralisation buffer (1 M Tris, pH 9) per 1 ml of eluate. Elution fractions were further purified by SEC (S200 10/30 GE Healthcare) in PBS, pH 7.4. The presence of properdin in the elution fraction was confirmed by SDS-PAGE and Western blot.

Monomeric properdin lacking TSR domains 2 and 3 was prepared as previously described 18 .

Purified FPcdel2,3 was mixed with refolded CirpA1 at a 1:1.5 ratio and incubated for 5 min at RT. The complex was separated from excess CirpA1 by SEC (S200 16/60 GE Healthcare) in 35 20 mM Tris/HCl pH 7.5, 100 mM NaCl. A 1:1 stoichiometry was confirmed using SEC-MALS (S200, 10/300 GL GE Healthcare).

Complement inhibition ELISAs were performed using a Wieslab complement system screen (Euro Diagnostica) following the manufacturer's instructions, with sample added prior to serum. 40

Assays for the classical and lectin pathways were performed with sheep red blood cells (TCS Biosciences) sensitized with anti-sheep red blood cell stroma antibody (cat. no. S1389, Sigma-Aldrich), alternative pathway assay was performed with rabbit red blood cells (TCS Biosciences). Human serum dilution of 1:18, 1:101 and 1:101 were used for the AP, CP and LP, respectively. Alternative pathway haemolysis inhibition assays were carried out with rabbit 45 red blood cells (TCS Biosciences) as described previously 28 . Serial dilutions of tick inhibitors purified from insect cells were used to assay interspecies activity whereas wt and mutant CirpA1 purified from E.coli was used to distinguish Properdin binding sites. Briefly, fifty microliters of cells (2 × 10^8 cells/mL) were incubated in an equal volume of diluted serum (1 h, 37 °C, shaking), supplemented with 2 μL of purified inhibitor or control. Cells were pelleted and haemolysis was quantified at A405 nm of supernatant. Cells with serum only were used for normalization (100% activity). Final serum dilutions used was as follows: 1:5 (human), 1:6 5

(monkey), 1:3 (rat) and 1:5 (guinea pig). Human serum from healthy volunteers was prepared as previously described 28 ; Macaca fascicularis serum was a kind gift from John Davis and Elena di Daniel; rat and guinea pig serum were from Complement Technology.

For pull-down assay, 0.1 mg/mL of purified protein was immobilized on Pierce NHS-activated 10 magnetic beads (Thermo Fisher) following the manufacturers' instructions. The beads were incubated with 10 mM EDTA and 50 μL serum (21 °C, 30 min). The beads were washed 3 times with 1 mL PBS + 0.05%Tween20, once with 100 μL PBS, and boiled in 50 μL SDS-PAGE loading buffer. Elutions were analysed by SDS-PAGE and semi-dry Western blotting.

Biotinylated C3b (bC3b) was prepared as previously described 4 . 0.1 mg/mL bC3b in PBS was immobilised on MagStrep "type3" XT Beads (iba) using 250 uL resin per mL. To prevent complement activation and thereby uncouple the effects of properdin binding to C3b vs. Properdin binding to fully assembled convertases, human serum was pre-treated with 10 mM EDTA and 10 mM EGTA. Pre-inhibited serum was prepared by adding 10 uM CirpA1 to the 20 inactivated serum and incubating 30 min at 25C. To investigate Properdin binding to C3b in the presence or absence of CirpA1 25 uL bC3b bound beads were incubated for 2.5h at 25 C in 100 uL of inactivated serum or 100 uL pre-inhibited serum. 25 uL beads without bC3b were incubated with inactivated serum as a negative control. The beads were washed 5 times with 100 uL PBS. To study competitive binding of CirpA1, 25 uL of beads that had not previously 25

contained CirpA1 were subsequently incubated for 30 min with 100 uL 10uM CirpA1 in PBS at 25C. The FT was collected, and the beads were washed 5 times with 100 uL PBS. Beads from all reactions were resuspended in twice the bead volume of SDS sample buffer and heated to 95 C for 10 min to elute all bound protein immediately after the last wash. The elutions from all reactions as well as the flow-through from the competition sample were analysed by SDS-30 PAGE followed by semi-dry Western Blotting against Properdin.

For blotting the SDS/PAGE-separated proteins were transferred to a PVDF membrane (Amersham Hybond P0.2 PVDF, 55 GE) by semiwet transfer (Bio-Rad) and blocked for 1 h with PBS/2% milk. 35

Primary antibody (goat α-FP: 1:2,000, Complement Technology). Secondary antibody (donkey α-goat HRP, Promega, 1:10,000). For His-tagged proteins, the Penta-His HRP Conjugate Kit (Qiagen) was used following manufacturer's instructions.

Blots were developed using ECL Western Blotting Substrate (Promega) and imaged using Amersham Hyperfilm ECL (GE Healthcare) or using a ChemiDoc XRS+ imaging system 40 (Biorad).

For SEC-MALS, 100 μL of protein sample at 1 mg/mL was injected onto an S200 10/300 column (GE Healthcare) equilibrated in PBS. Light scattering and refractive index were measured using a Dawn Heleos-II light scattering detector and an Optilab-TrEX refractive 45 index monitor. Analysis was carried out using ASTRA 6.1.1.17 software assuming a dn/dc value of 0.186 mL/g.

Refolded CirpA1 in 50 mM Tris, 150 mM NaCl, pH 8 was concentrated to 17.8 mg/mL. The protein was mixed with an equal volume of mother liquor containing 0.2 M imidazole malate, 5 pH 6, 30% (w/v) PEG4000, and crystallized in 300-nL drops by a vapor-diffusion method at 21 °C. Crystals were cryoprotected in mother liquor supplemented with 20% (v/v) PEG400 and flash-frozen in liquid N2. Data were collected on beamline I04 at the Diamond Light Source (Harwell, United Kingdom), wavelength: 0.9795 Å, as specified in Table 1 The structure of Cirp-A5 was built and refined through cycles of automated model building by Buccaneer 37 and refinement by REFMAC5 38 . Subsequently, the model was subjected to multiple rounds of manual rebuilding in Coot 39 and refinement in REFMAC5 38 or Phenix 40 .

CirpA1 was copurified with the FPdel2,3 by SEC (S200 16/60 GE Healthcare) in 20 mM Tris/HCl pH 7.5, 100 mM NaCl and concentrated to 10.5 mg/mL. The protein was mixed with an equal volume of mother liquor containing 0.15 M Potassium thiocyanate, 0.1 M Tris, pH 7.5, 18 % w/v PEG 5000 MME, and crystallized in 200-nL drops by a vapor-diffusion method at 21 °C. Crystals were cryoprotected in mother liquor supplemented with 20% (v/v) PEG400 and 5

flash-frozen in liquid N2. Data were collected on beamline I04-1 at the Diamond Light Source (Harwell, United Kingdom), wavelength: 0.915890 Å, as specified in Table 1 . The structure of the complex was solved by molecular replacement using MolRep 34 within CCP4 with the structures of CirpA1 (PDB ID code 7BD2, this study) and FPcdel2,3 (PDB ID code 6S08). The initial model was subjected to multiple rounds of manual rebuilding in Coot 39 and refinement 10

in Phenix 40 . The protein chemistry of the final models was validated using MolProbity 42 . The structures are characterized by the statistics shown in Table 1 . Interactions between CirpA1 and FP have been predicted by PDBePISA 43 . Protein structure figures were prepared using Pymol v2.4 (Schrödinger). Electrostatic potentials were calculated using the APBS program in Pymol 44 . 

Let's tie the knot: Marriage of complement and adaptive immunity in pathogen evasion, for better or worse

Complement system part Imolecular mechanisms of activation and regulation

Complement: A key system for immune surveillance and homeostasis

Molecular insights into the surface-specific arrangement of complement C5 convertase enzymes

15 Complement system part II: Role in immunity

The renaissance of complement therapeutics

Systemic complement activation is associated with respiratory failure in COVID-19 hospitalized patients

The complement system in COVID-19: friend and foe?

Complement as a target in COVID-19?

Complement and tissue factor-enriched neutrophil extracellular traps are 30 key drivers in COVID-19 immunothrombosis

Properdin: A multifaceted molecule involved in inflammation and diseases

Properdin: Binding to C3b and Stabilization of the C3b-dependent C3 convertase

The properdin system and immunity: I. Demonstration and isolation of a new serum protein, properdin, and its role in immune phenomena

A molecular concept of the properdin pathway

Properdin and nephritic 5 factor dependent C3 convertases: requirement of native C3 for enzyme formation and the function of bound C3b as properdin receptor

The properdin pathway: an "alternative activation pathway" or a "critical 10 amplification loop" for C3 and C5 activation?

Emerging Roles of a Pattern-Recognition Molecule

Structural Basis for Properdin Oligomerization and Convertase Stimulation in the Human Complement System

Insights Into Enhanced Complement Activation by Structures of Properdin and Its Complex With the C-Terminal Domain of C3b

Novel assays to distinguish between properdin-dependent and properdin-independent C3 nephritic factors provide insight into properdin-inhibiting therapy

The use of tick salivary proteins as novel 30 therapeutics

Complement Inhibitor of C5 Activation from the Soft Tick Ornithodoros moubata

The Structure of OMCI, a Novel Lipocalin Inhibitor of the Complement System

Bifunctional lipocalin ameliorates murine immune complex-induced acute lung injury

An inhibitor of complement C5 provides structural insights into activation

Sexual differences in the sialomes of the zebra tick, Rhipicephalus pulchellus

Structural basis for therapeutic inhibition of complement C5

The lipocalin protein family: Structure and function

Substrate prediction of Ixodes ricinus salivary lipocalins differentially expressed during Borrelia afzelii infection

CPFP: A central proteomics facilities pipeline

SignalP 4.0: Discriminating signal peptides from transmembrane regions

FFAS server: Novel features and 20 applications

Molecular replacement with MOLREP

Overview of the CCP4 suite and current developments

CHAINSAW: A program for mutating pdb files used as templates in molecular replacement

The Buccaneer software for automated model building. 1. Tracing protein chains

REFMAC5 for the refinement of macromolecular crystal structures

Features and development of Coot

Macromolecular structure determination using X-rays, neutrons and electrons: Recent developments in Phenix

PHENIX: A comprehensive Python-based system for macromolecular structure solution

MolProbity: All-atom structure validation for macromolecular 5 crystallography

Inference of Macromolecular Assemblies from Crystalline State

Improvements to the APBS biomolecular solvation software suite

We acknowledge the Diamond Light Source and the staff of beamlines I04 and I04-1 for access under proposals mx12346 and mx18069. We further acknowledge the European Synchrotron Radiation Facility and the staff of beamline MASSIF1. We thank M. Slovak (Institute of Zoology, 20Bratislava, Slovakia) for providing salivary glands; John Davis and Elena di Daniel (Oxford Drug Discovery Institute) for providing Macaca fascicularis serum; Simon Newstead (Biochemistry, Oxford) for assistance with Synchrotron data collection