Recombinant human FHR1, FHR2, FHR3, FHR4 and FHR5 produced by Novoprotein (Shanghai, China) were purchased from Gentaur (Kampenhout, Belgium). Polyclonal goat anti-human FHR5 was purchased from R&D Systems (Wiesbaden, Germany). Purified human factor H (FH), C3, C3b, iC3b, C3c, C3d, factor I (FI), properdin (FP), goat anti-human FH and goat anti-human FB Ab were obtained from Merck (Budapest, Hungary). The anti-human FH mAb A254 and the anti-properdin mAb A235, and normal human serum (NHS) were from Quidel (obtained via Biomedica, Budapest, Hungary). Rabbit anti-human C3d, and HRP-conjugated goat anti-mouse Ig, rabbit anti-goat Ig and swine anti-rabbit Ig were obtained from Dako (Hamburg, Germany) and the HRP-labeled goat anti-human C3 Ab was from MP Biomedicals (Solon, OH). Human laminin, aggrecan, biglycan, decorin, collagen IV, fibronectin and vitronectin were purchased from Sigma-Aldrich (Budapest, Hungary).

The ECM proteins fibromodulin, osteoadherin, and PRELP were expressed recombinantly in HEK293 cells and purified using affinity chromatography due to the presence of His-tag (41).

Human properdin was purified as described in (42). Properdin oligomeric forms (dimers, P2; trimers, P3; and tetramers, P4) were isolated from pure properdin by size exclusion chromatography, as previously described (42). Briefly, pure properdin (5 mg) was loaded onto a Phenomenex BioSep-Sec-S4000 column (600 x 7.8 mm) with a guard column (75 x 7.8 mm) and eluted at a 0.5 ml/min flow rate in PBS. Purified, physiological forms of properdin were stored at 4°C and used within 2 week of separation (42, 43).

Recombinant human FHR5 fragments comprising CCPs 1-4, CCPs 3-7, and CCPs 8-9 were amplified by polymerase chain reaction using codon-optimized human FHR5 DNA template and specific primers ( Table 1), and cloned into the pBSV-8His Baculo-virus expression vector (44). The proteins were expressed in Spodoptera frugiperda (Sf9) cells and purified by nickel-affinity chromatography. Purified proteins were analyzed by Western blot using 10% SDS-PAGE under non-reducing conditions and by silver staining.

ECM proteins (0.5 mg/ml), gelatin (0.5 mg/ml) and MaxGel (0.8 mg/ml) were printed onto nitrocellulose-covered slides in triplicates. Air-dried slides were washed three times with PBS containing 0.05% Tween 20 and blocked with 4% BSA. In binding assays, the immobilized ECM proteins were incubated with FHR1 (5 µg/ml; 25 µg/ml; 50 µg/ml) or FHR5 (0.5 µg/ml; 5 µg/ml; 20 µg/ml), then detected with goat anti-human FH or goat anti-human FHR5 followed by Alexa-647 labeled anti-goat IgG.

To measure competition, printed proteins were incubated with 25 µg/ml FH with or without FHR1 (5 µg/ml; 25 µg/ml; 50 µg/ml) or FHR5 (0.5 µg/ml; 5 µg/ml; 20 µg/ml). Bound FH was detected using monoclonal mouse anti-FH Ab (A254) that does not cross-react with FHR1 and FHR5, and Alexa-546-conjugated goat anti-mouse IgG. After scanning, fluorescence intensities were calculated as median of the triplicates and background was subtracted.

Complement activation on protein microarray was analyzed by incubating immobilized ECM components with 20% NHS in the presence or absence of recombinant FHR1 (50 µg/ml) or FHR5 (20 µg/ml) diluted in DPBS containing Mg²⁺ and Ca²⁺ (Lonza). C3 deposition was detected with Alexa-555 labelled anti-C3 F(ab’)_2, and the sC5b-9 was detected with biotinylated monoclonal anti-sC5b-9 Ab and streptavidin-Alexa-488.

To analyze the binding of C3b, iC3b, C3c and C3d to FHR5 fragments, 5 µg/ml FHR5, FHR5 CCPs 1-4, CCPs 3-7, CCPs 8-9, and human serum albumin (HSA) as a negative control were immobilized on microtiter plate. After blocking with 4% BSA, 20 µg/ml C3b, iC3b, C3c and 10 µg/ml C3d were added to the corresponding wells. Bound proteins were detected with HRP-conjugated anti-human C3 or polyclonal rabbit anti-human C3d followed by HRP-conjugated swine anti-rabbit Ig. TMB High Sensitivity substrate solution (BioLegend) was used to visualize binding, and the absorbance was measured at 450 nm.

To measure binding in a reverse setting, 20 µg/ml C3b was immobilized and incubated with 20 µg/ml FHR5, the FHR5 fragments CCPs 1-4, CCPs 3-7 and CCPs 8-9 diluted in DPBS containing Mg²⁺ and Ca²⁺ (Lonza). HSA was used as a negative control protein. Bound proteins were detected with polyclonal goat anti-FHR5 Ab and HRP-conjugated rabbit anti-goat Ig antibody.

To further analyze the binding of C3 and C3(H₂O) to FHR5, 10 µg/ml FHR5, CCPs 1-4, CCPs 3-7, CCPs 8-9 and as a negative control alpha-1-antitrypsin were immobilized in microplate wells, and incubated with 5 µg/ml C3, C3(H₂O), and C3b, which was used as a positive control. C3(H₂O) form was generated from C3 with ten freeze/thaw cycles. After washing steps, bound proteins were detected with HRP-conjugated anti-human C3 antibody.

To measure properdin binding, FHR proteins and C3b were immobilized at 5 µg/ml in DPBS containing Mg²⁺ and Ca²⁺ at 4°C overnight. After washing with DPBS containing 0.05% Tween-20, free binding sites were blocked by incubation with 4% BSA dissolved in DPBS containing 0.05% Tween-20 at 20°C for 1 hr, then 20 µg/ml of the various properdin forms diluted in DPBS containing Mg²⁺ and Ca²⁺ was added at 20°C for 1 hr. After washing, properdin binding was detected using the anti-properdin mAb A235 (1:1000) and the corresponding secondary Ab (1:1000).

To analyze FHR1 binding to ECM proteins, 5 µg/ml laminin, fibromodulin, osteoadherin, PRELP and HSA were coated and incubated with FHR1 (5 µg/ml or 10 µg/ml). Bound FHR1 was detected with goat anti-human FH and the corresponding secondary Ab. Inhibition of FHR1 binding to ECM proteins was measured by incubating the immobilized ECM proteins with 5 µg/ml FHR1 and simultaneously added mAb C18 (10 µg/ml), which recognizes CCP5 of FHR1, or mAb A255 (10 µg/ml), which does not recognize FHR1 and was used as a control. Binding of FHR1 was detected as described above.

For FHR5, ECM proteins were immobilized (10 µg/ml) and incubated with FHR5 and its fragments (CCPs 1-4, CCPs 3-7, CCPs 8-9) in equimolar concentrations (100 nM or 200 nM). After washing, bound proteins were detected as described above.

To measure the competition between FHR5 and FH for the binding to ECM proteins in microplate format, immobilized ECM proteins (10 µg/ml) were incubated with 50 µg/ml FH in the presence or absence of 20 µg/ml FHR5. HSA was used as a negative control. Bound FH was detected with monoclonal anti-human FH (A254) and HRP-conjugated goat anti-mouse Ig antibody.

To measure complement activation and C3 convertase formation on ECM-bound FHR1/FHR5, microplate wells were coated with the ECM proteins (5 µg/ml or 10 µg/ml) and, after blocking with 4% BSA, incubated with 10% NHS diluted in 5 mM Mg²⁺-EGTA in the absence or presence of FHR5 (10 µg/ml) or FHR1(10 µg/ml; 20 µg/ml) for 30 min at 37°C. Deposition of C3 fragments and factor B (FB) was detected using HRP-conjugated anti-human C3, and goat anti-human FB followed by HRP-conjugated rabbit anti-goat Ig, respectively.

To assess the functional consequence of competition between FHR5 and FH, laminin was immobilized at 10 µg/ml and, after blocking with 4% BSA, 100 µg/ml FH was added in the absence or presence of 20 µg/ml FHR5. After washing, the wells were incubated with 140 nM C3b and 300 nM FI diluted in DPBS containing Mg²⁺ and Ca²⁺ for 1 hour at 37°C. Supernatants were collected and subjected to 7.5% SDS-PAGE and Western blotting under reducing conditions. C3 fragments were detected with a HRP-conjugated anti-human C3 Ab and an ECL detection kit (Merck).

Statistical analysis was performed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, California). A p value <0.05 was considered statistically significant.

FH was previously shown to bind to several components of the extracellular matrix (ECM); moreover, FH represents the main alternative complement pathway inhibitor on ECM (31, 32), although its splice variant, factor H-like protein 1 (FHL-1) was reported as the major regulator in Bruch’s membrane (45). We hypothesized that, due to their sequence similarities with FH ( Figure 1 ), FHR1 and FHR5 could interact with certain ECM components, which is also supported by recent data (7, 10, 18).

To test this hypothesis, first we used protein microarray technique. Various ECM proteins were printed onto nitrocellulose-covered slides in triplicates. The slides were incubated with increasing concentrations of FHR1 ( Figure 2A ) and FHR5 ( Figure 2B ), and their binding was detected using polyclonal antibodies to FH and FHR5, respectively. Both FHRs bound to several ECM proteins, including laminin, osteoadherin, PRELP and vitronectin, in a dose-dependent manner. Binding of FHR1 and FHR5 to laminin, osteoadherin and PRELP was confirmed by ELISA; in addition, prominent binding to fibromodulin was also detected in ELISA ( Figure 3 ). This difference between the two types of assays might be due to the different surfaces and measurement methods.

To determine the binding sites for the analyzed ECM proteins within FHR5, recombinant FHR5 fragments comprising CCPs 1-4, CCPs 3-7, and CCPs 8-9 were generated, expressed in insect cells and purified ( Figure 1 , Table 1 and Supplementary Figure 1 ). Binding of these FHR5 fragments to the immobilized ECM proteins was determined by ELISA using polyclonal anti-FHR5. We found that CCPs 3-7 of FHR5 bound to laminin, fibromodulin, osteoadherin and PRELP, indicating that FHR5 binds to these ECM proteins through its middle part, whereas the other tested FHR5 fragments did not bind ( Figures 3A, B ). FHR1 also bound to immobilized laminin, fibromodulin, osteoadherin and PRELP in ELISA ( Figure 3C ). Presumably, FHR1 binds through its most C-terminal domain to these ECM proteins, in contrast to FHR5, since when FHR1 was preincubated with the monoclonal antibody C18, which recognizes the CCP5 domain of FHR1, FHR1 binding to laminin, fibromodulin, osteoadherin and PRELP was inhibited. In contrast, the mAb A255, which binds to FH but not to FHR1, did not affect FHR1 binding in this assay ( Figure 3D ). Since laminin binding has not been described for FH before, we tested laminin binding to the FH fragments comprising the N-terminal regulatory domains CCPs 1-4, the middle part CCPs 8-14 (which contains domains related to FHR5) and the C-terminal CCPs 15-20 that contains domains homologous to the FHR1 C-terminus. Laminin bound only to CCPs 15-20, supporting a C-terminal binding site in both FH and FHR1 ( Supplementary Figure 2 ).

To further analyze the interaction of FHR1 and FHR5 with ECM proteins and their effect on the functional activity of FH, we tested whether FHR1 and FHR5 compete with FH for binding to ECM proteins. In the protein microarray setup, the immobilized ECM proteins were incubated with FH together with increasing amounts of FHR1 and FHR5. Binding of FH was detected with an anti-FH mAb that does not recognize FHR1 and FHR5. Both FHR1 and FHR5 inhibited the binding of FH to osteoadherin, PRELP, fibromodulin, laminin, vitronectin and collagen IV in a dose-dependent manner, as well as to MaxGel that was used as a control ( Figures 4A, B ). Competition between FHR5 and FH was confirmed by ELISA where laminin, osteoadherin, fibromodulin and PRELP were immobilized in microplate wells and incubated with 50 µg/ml FH in the absence or presence of 20 µg/ml FHR5. FHR5 strongly inhibited FH binding to these ECM proteins ( Figure 4C ).

To study whether the competitive inhibition of FH binding by FHR5 impairs the complement regulatory activity by removing FH, a surface cofactor assay was performed. Laminin was immobilized in microplate wells and incubated with FH together with or without FHR5. After removing the unbound proteins by extensive washing, FI and C3b were added to the wells to allow for cleavage of C3b. After incubation, the supernatants were collected, the proteins were separated by SDS-PAGE and the C3b cleavage products were visualized by Western blot. FHR5 competitively inhibited the cofactor activity of FH on laminin coated surfaces as less C3b α’-chain was cleaved in the presence of FHR5 ( Figure 4D ).

All FHR proteins bind C3b, the main ligand of FH (3). FH has multiple recognition sites for C3b and other C3 fragments, but the C-terminal domains harbor the major binding site for surface-bound C3b. Because of the conserved FH C-terminal domains, FHRs are supposed to share this C-terminal C3b recognition site. We set out to determine the binding sites of various C3 fragments in the FHR5 protein, using recombinant FHR5 fragments comprising CCPs 1-4, CCPs 3-7 and CCPs 8-9 that cover the whole protein. FHR5 and its fragments were immobilized in microtiter plate wells and incubated with C3b, iC3b, C3c and C3d. FHR5 bound soluble C3b mainly via the C-terminal CCPs 8-9 ( Figures 5A, D ), in agreement with recent data (18). Similarly, both iC3b and C3d bound to the CCPs 8-9 fragment of FHR5, whereas C3c did not bind to any of the FHR5 fragments nor to the whole protein ( Figure 5A, B ). In reverse setting, where C3b was immobilized in microtiter plate wells and incubated with FHR5 fragments, CCPs 3-7 showed significant binding to surface-bound C3b ( Figure 5C ). Thus, FHR5 appears to have two C3b binding sites, one for surface-bound C3b and another for fluid-phase C3b, suggesting that FHR5 when bound to a surface either via deposited C3b or to other ligands exposed on the surface, such as bound pentraxins (11) or ECM proteins, it can potentially recruit C3b from fluid-phase and allow for alternative pathway activation by recruiting an active C3bBb convertase (7).

Since C3 fragments are generated upon complement activation, we also tested whether intact C3 and C3(H₂O), which is physiologically constantly generated in serum at a low rate, can bind to FHR5. Purified C3 was subjected to freeze-thaw cycles to generate C3(H₂O) ( Supplementary Figure 3 ). Both C3 and C3(H₂O) bound to the C-terminal fragment CCPs 8-9 of FHR5 ( Figure 5E ); however, it cannot be excluded that during the experiment C3 was ticking over and actually C3(H₂O) what was bound.

Previously, we showed that FHR5 can activate the alternative pathway by binding C3b and recruiting an active C3 convertase enzyme, as showed by the deposition of C3b, Bb and properdin on immobilized FHR5. However, FHR5 did not bind FB or properdin directly (7). Rudnick et al. on the other hand described direct binding of properdin to FHR5 and suggested that properdin when bound to FHR5 recruits C3b and thus initiates complement activation (18). To resolve this contradiction, and since properdin notoriously tends to aggregate and generate higher order oligomers when stored for longer time or freeze-thawed (43), we analyzed the binding of purified commercial properdin (P), as well as purified properdin dimers, trimers, and tetramers (P2, P3, and P4, respectively). There was no significant binding of the physiologically occurring properdin forms to FHR5, nor to the other FHRs tested in parallel; however, there was prominent binding to C3b, used as a positive control ( Figure 6A ). In addition, since FHR5 CCPs 1-2 were proposed to harbor a properdin binding site (19), we analyzed our recombinant FHR5 fragments and no properdin binding to any of these was found in ELISA ( Figure 6B ). Altogether, these results suggest that physiological serum properdin forms are unlikely to bind directly to FHR5 and other FHRs.

Competition between FH and FHR1 or FHR5 for binding to ECM is only one way how the regulation of complement activation can be compromised at such a surface. Previously, we demonstrated that both FHR1 and FHR5 allow formation of the alternative pathway C3 convertase when immobilized on microplate wells and rather support than inhibit complement activation. We tested whether FHR1 and FHR5 show this activity when bound to ECM proteins. To this end, ECM proteins were immobilized and incubated with NHS in the presence or absence of recombinant FHR1 or FHR5 in a buffer supplemented with 5 mM Mg²⁺-EGTA to allow only the activation of alternative pathway. Generation of the alternative pathway C3 convertase on the surface bound FHR proteins was confirmed by measuring deposition of C3b and factor B (FB). Both FHR1 and FHR5 significantly enhanced the amount of bound C3-fragments on the ECM proteins ( Figures 7A , 8A ). FHR5 also induced significant FB deposition, whereas FHR1 increased the amount of ECM-bound FB to a lesser extent and at higher serum concentration ( Figures 7B , 8B , and data not shown). These results indicate that both FHR1 and FHR5 support alternative pathway activation on ECM proteins.

In the case of FHR5, we further studied this using MaxGel and ECM produced in vitro by ARPE-19 cells. These two ECMs of different origin have different composition, laminin dominating in MaxGel and collagen IV in ARPE-derived ECM ( Supplementary Figure 4C ). When these ECMs were exposed to serum in the presence of FHR5, strongly increased deposition of all components of a C3bBbP properdin-stabilized alternative pathway C3 convertase was detected ( Supplementary Figures 4A, B ).

We used the ECM protein microarray to analyze the effect of FHR1 and FHR5 on terminal complement pathway activation by measuring the deposition of the C5b-9 complex. Immobilized ECM proteins were incubated with NHS with or without the addition of recombinant FHR1 or FHR5. In this assay, FHR5 increased the deposition of C5b-9, along with the deposition of C3 fragments, on osteoadherin, PRELP, fibromodulin and vitronectin ( Figure 9 ), while FHR1 had no such effect.