FHR-1 Binds to C-Reactive Protein and Enhances Rather than Inhibits Complement Activation

The human factor H (FH) protein family includes the major alternative pathway regulator FH, its splice variant FH-like protein 1 (FHL-1), and five FH-related proteins (FHRs) (1). All members of this protein family are composed exclusively of complement control protein (CCP) domains (also known as short consensus repeats). Although the function of FH is well characterized, the role of the FH-related proteins is incompletely understood (1, 2).

Despite the limited information on FHR protein functions, genetic studies linked the FHR proteins to various diseases (reviewed in Ref. 1). The five CFHR genes have arisen through segmental duplications (3, 4), and this makes the genomic region encoding the FH and the five FHRs prone to misalignments and genomic recombination events. These may result in hybrid proteins involving FH and FHR-1 or FHR-3, gene conversion events, for example, between the exons coding for the C termini of FH and FHR-1, gene deletions affecting CFHR1, CFHR3, and CFHR4, as well as in duplication of exons (reviewed in Ref. 1). The CFH::CFHR1 and CFHR1::CFH hybrid genes are associated with atypical hemolytic uremic syndrome (aHUS) (5–7), the deletion of the CFHR1 gene predisposes to the autoimmune form of aHUS (8, 9), and a mutant FHR-1 protein with duplicated CCPs 1 and 2 was described as pathogenic in a patient with C3 glomerulopathy (10). These data indicate a role of FHR-1 in the regulation or modulation of complement activation.

FHR-1 is the most abundant glycoprotein among the FHRs, with an estimated plasma concentration of 40–100 μg/ml (11, 12). It is composed of five CCPs that are homologous to CCPs 6 and 7 and CCPs 18–20 of FH (13). Two FHR-1 glycoforms, FHR-1α (37 kDa) and FHR-1β (43 kDa), are detected in plasma, representing differentially glycosylated proteins (11). In addition, an acidic (FHR-1*A) and a basic (FHR-1*B) isoform of the protein exist, which differ in sequence in 3 aa (14). The CCPs 3, 4, and 5 of FHR-1 have a high degree of amino acid sequence identity (95, 100, and 97% in FHR-1*A and 100, 100, and 97% in FHR-1*B, respectively) to the three most C-terminal domains of FH (Fig. 1). These FH domains are responsible for host surface recognition and contain an important C3b binding site (15–19).

The homology between FHR-1 and FH suggests related functions and, indeed, FHR-1 binds to C3b (12, 20), although FH, having three binding sites for C3b, shows more pronounced binding. No cofactor activity and convertase decay accelerating activity were observed for FHR-1 at the C3b and C3 convertase level in line with the lack of homologous domains in FHR-1 to the complement regulatory domains of FH (11, 12). FHR-1 was described as an inhibitor of C5 convertase and formation of the terminal pathway membrane attack complex (12), but other groups could not confirm this activity (20, 21). CCP1 and CCP2 of FHR-1 display only weak similarity to the respective CCPs of FH (42 and 34% amino acid identity to CCP6 and CCP7 of FH, respectively), but are almost identical to CCPs 1 and 2 of FHR-2 and FHR-5. These domains contain a unique dimerization motif that allows for FHR-1, FHR-2, and FHR-5 to exist as dimeric species in vivo. Moreover, dimerization is thought to increase avidity for ligands and enhance the ability of FHR-1, FHR-2, and FHR-5 to compete with FH for C3b binding in vitro (21).

The FH C terminus contains binding site, in addition to C3b, for heparin and mediates cell surface binding such as attachment to endothelial cells, which was also described for FHR-1 (12, 20). Moreover, these FH domains contain binding sites for the pentraxins pentraxin 3 (PTX3) and C-reactive protein (CRP) (22–24). FHR-1 binding to PTX3 was described (23), but its interaction with CRP has not yet been analyzed.

CRP is a 115-kDa acute-phase protein with a disc-like form composed of five identical subunits, which are assembled noncovalently (25). The structure is stabilized by Ca²⁺ in human plasma, but CRP dissociates into its monomeric subunits in the absence of calcium, at low pH or at increased temperature (25–27). CRP levels in the plasma of healthy adults are low, with a median concentration of 0.8 μg/ml, but they increase as much as a 1000-fold during acute-phase reaction (28). Several ligands have been identified for CRP, including microorganisms, phosphorylcholine, complement components (e.g., C1q, FH, C4b-binding protein, ficolins), and proteins of the extracellular matrix (ECM), but the relationship between ligand binding and function is a matter of debate in most cases (25, 29). Surface-bound CRP has been shown to inhibit the alternative pathway, possibly through the interaction with FH (30, 31). The FH and FHL-1 402H variants, which are strongly associated with increased risk to develop age-related macular degeneration (AMD), were shown to bind CRP less strongly compared with the 402Y variants (32, 33). This interaction, among others, is thought to be important in the pathogenesis of AMD. The common CFHR3-CFHR1 gene deletion, causing FHR-1 deficiency, is protective in AMD (34).

Therefore, the aim of this study was to evaluate FHR-1 as a potential ligand for CRP and analyze the functional relevance of this interaction. We also aimed to investigate whether FHR-1, similar to FHR-4 and FHR-5 (35, 36), has a role in complement activation by supporting formation of the C3bBb convertase.

Recombinant human FHR-1*A (referred to as FHR-1 in this article), FHL-1, FHR-4A, and FHR-4B were generated using the pBSV-8His Baculovirus expression vector (37), expressed in Spodoptera frugiperda (Sf9) cells, and purified by nickel-affinity chromatography as described previously (23, 38, 39). Recombinant human FHR-5, PTX3, and biotinylated goat anti-human PTX3 Ab were obtained from R&D Systems (Wiesbaden, Germany). Recombinant human FHR-2-gst fusion protein was purchased from Abnova (Taipei, Taiwan), and FHR-3-gst fusion protein was purchased from Proteintech Europe (Manchester, U.K.). The C-terminal fragments of FH and the FHR proteins were generated as described previously (21). FHR-1*A and FHR-1*B were isolated from human plasma as described previously (23). Recombinant FH19-20 fragment with 14 different single amino acid substitutions was produced in yeast cells (40).

Purified human FH, C3, C3b, factor B (FB), factor D (FD), properdin (factor P [FP]) factor I (FI), C1q, C1, recombinant human CRP, goat anti-human FH Ab, goat anti-human FB Ab, goat anti-human C1q Ab, and goat anti-human C4 Ab were obtained from Merck (Budapest, Hungary). Modified monomeric form of CRP (mCRP) was generated from recombinant CRP by urea/EDTA chelation treatment as described previously (24). The anti-FH mAbs A254 and A255 and the anti-FP mAb A235 were from Quidel (Biomedica, Budapest, Hungary). The anti-FH mAb C18 (41) was purchased from Alexis Biochemicals (Lörrach, Germany). The goat anti-CRP Ab and the anti-mCRP mAb (clone CRP-8) were from Sigma-Aldrich (Budapest, Hungary). The anti–pentameric form of CRP (pCRP) mAb was purchased from Antibodies-online (Aachen, Germany). HRP-conjugated goat anti-human C3 was from MP Biomedicals (Solon, OH). HRP-conjugated swine anti-rabbit Igs, rabbit anti-goat Igs, and goat anti-mouse Igs were from Dako (Hamburg, Germany).

Normal human serum was collected from healthy individuals and pooled, and sera from patients with sepsis or aHUS were obtained according to protocols approved by the Institutional Review Board of Semmelweis University of Budapest and the National Ethical Committee (Scientific and Research Ethics Committee of the Medical Research Council). Written approvals for the diagnostic tests and research analysis were given by the patients or their parents. IgG fractions of aHUS patients with FH autoantibodies and of healthy individuals were isolated from serum or plasma using Protein G columns as described previously (23). FH-depleted human serum was purchased from Complement Technology (Tyler, TX).

To analyze binding of native FHR-1 from human serum to CRP, we coated microplate wells with 10 μg/ml CRP or gelatin in Dulbecco’s PBS (DPBS; Lonza, Cologne, Germany). Wells were incubated with 50% v/v normal human serum (NHS) in DPBS (with Ca²⁺ and Mg²⁺) for 1 h at 37°C. After washing, the bound proteins were eluted with nonreducing SDS sample buffer, separated on 10% SDS-PAGE, and transferred to nitrocellulose membrane. The blot was incubated with a polyclonal goat anti-FH Ab and the corresponding secondary Ab, and developed with an ECL detection kit (Merck).

To measure binding of CRP to FHRs at various pH, we immobilized 5 μg/ml FH, FHR-1, FHR-5, and human serum albumin (HSA) in microtiter plate wells and, after blocking with 4% BSA (w/v) in DPBS, added 2.5% v/v NHS or sepsis serum, or 5 μg/ml recombinant CRP in DPBS (with Ca²⁺ and Mg²⁺) at pH 7.4, 6.5, and 5.5. CRP was detected with the goat anti-human CRP Ab that recognizes both CRP forms (42) and HRP-conjugated rabbit anti-goat Ig. TMB PLUS substrate (Kem-En-Tec Diagnostics, Taastrup, Denmark) was used to visualize binding, and the absorbance was measured at 450 nm.

Interaction of FHRs with pCRP was measured in TBS (10 mM Tris, 140 mM NaCl, 2 mM CaCl₂, 1 mM MgCl₂ [pH 7.4]). FH, FHR proteins, and HSA as control protein were immobilized at 65 nM in microplate wells and, after blocking with 3% BSA (w/v) in TBS, incubated with up to 50 μg/ml pCRP. CRP binding was detected with the goat anti-human CRP Ab.

To measure mCRP binding to FHR proteins, we immobilized FHR proteins and their C-terminal fragments in microtiter plate wells at 5 μg/ml concentration, followed by blocking with 3% BSA (w/v) in DPBS and incubation with 10 μg/ml mCRP. Binding of mCRP was detected as described earlier.

In some assays, 5 μg/ml CRP was immobilized in microplate wells in DPBS, which results in the decay of most bound CRP into the mCRP form (42). After blocking, FHR-1, FHR-5, and FH were added in serial dilutions for 1 h at 20°C, and their binding was detected using the goat anti-FH Ab.

To compare binding of mCRP with FHR-1*A and FHR-1*B, the two FHR-1 variants were immobilized at 4 μg/ml in DPBS in microplate wells, and binding of increasing amounts of mCRP was measured as described earlier.

To analyze PTX3 binding by FH19-20 mutants, we coated microtiter plates with 4 μg/ml each of the wild-type or mutant FH19-20 fragments, diluted them in TBS in 25 μl, overnight at 4°C. The wells were washed after each step with TBS containing 0.05% Tween 20 (v/v). After blocking with 4% BSA (w/v) in TBS for 1 h at 20°C, 5 μg/ml PTX3 was added in TBS for 1 h at 37°C. Bound PTX3 was detected with a biotinylated goat anti-PTX3 Ab followed by HRP-conjugated streptavidin.

To measure C1q binding to FHR-1–bound mCRP, we coated microplate wells with 4 μg/ml recombinant FHR-1 and, as control, with gelatin, then incubated with 5 μg/ml mCRP for 1 h at 20°C. After washing with DPBS containing 0.05% Tween 20 (v/v), C1q was added in increasing concentrations for 1 h at 20°C, and C1q binding was detected using anti-C1q and the corresponding secondary Ab. Binding of C1 was similarly measured by incubating the wells with 10 μg/ml C1 instead of C1q.

In inhibition assays, 5 μg/ml FHR-1 was immobilized in microplate wells. The mAbs C18 and A255 were added in 10 μg/ml in DPBS, and IgG fractions derived from sera of aHUS patients containing FH autoantibodies or sera of healthy individuals were added in 500 μg/ml, for 15 min at 20°C, followed by the addition of mCRP in 5 μg/ml final concentration. The binding of mCRP was detected as described earlier.

To measure competition between PTX3 and mCRP, we incubated immobilized FHR-1 (5 μg/ml) with 5 μg/ml PTX3 in the absence or presence of serial dilutions of mCRP in microtiter plate wells. Bound PTX3 was detected with a biotinylated goat anti-human PTX3 Ab.

For FH15-20/FHR-1 inhibition assays, 10 μg/ml recombinant FH15-20 fragment was immobilized and incubated with 5 μg/ml mCRP in the absence or presence of 1 μM FHR-1. Bound mCRP was measured with a polyclonal goat anti-human CRP Ab and the corresponding secondary Ab. Competition of FHR-1 with FHL-1 was similarly measured.

To measure the effect of FHR-1 on the assembly and activity of the C3bBb alternative pathway C3 convertase, we immobilized C3b at 5 μg/ml in microplate wells. FH (10 μg/ml), FHR-1 (50 μg/ml), FHR-5 (10 μg/ml), or BSA (50 μg/ml) was added together with FB, FD, and properdin to generate C3bBb convertase as described previously (36). The formed C3bBb was detected using polyclonal anti-FB Ab. The convertase activity was measured by adding 10 μg/ml purified C3 for 1 h at 37°C and quantifying the generated C3a by a C3a ELISA kit (Quidel). Formation of the C3bBb alternative pathway C3 convertase on surface-bound FHR-1 was measured as described previously (36).

To investigate decay accelerating activity of FHR-1, we built up the C3bBb convertase as described previously (36); then FH (2.5 μg/ml), FHR-1 (10 μg/ml), FHR-5 (10 μg/ml), or FH plus FHR-1 or FH plus FHR-5 was added for 30 min at 20°C in DPBS. The remaining convertase was detected with a polyclonal anti-FB Ab.

Cell-derived ECM was prepared by culturing the human retinal pigmented epithelial cell line ARPE-19 (ATCC, LGC Standards, Wesel, Germany) in 96-well tissue culture plates, coated with 0.2% (w/v) gelatin, in DMEM:F12 medium (Lonza) supplemented with 10% FCS, penicillin (100 U/ml), streptomycin (100 μg/ml), and Amphotericin B (250 ng/ml), in a cell incubator with 5% CO₂ (v/v) at 37°C. Cells were removed by incubation with DPBS containing 20 mM EDTA; removal was controlled visually by microscopy. To measure the formation of the C3bBb C3 convertase on cell-free ECM, we blocked the washed ECM with 4% (w/v) BSA in DPBS containing 0.05% Tween 20 (v/v); then the wells were sequentially incubated with 5 μg/ml mCRP, 50 μg/ml FHR-1, and 50 μg/ml C3b (all in DPBS, 1 h at 20°C, with washing steps). Purified FB (2 μg/ml), FD (0.2 μg/ml), and FP (4 μg/ml) were added in convertase buffer (containing 2 mM Ni²⁺, 4% [w/v] BSA, and 0.05% [v/v] Tween 20) at 37°C for 30 min to generate the C3bBb convertase as described previously (36). The formed C3bBb was detected with anti-FB Ab as described earlier.

To assess possible inhibition of the terminal pathway by FHR-1, we induced complement activation by adding 20 μg/ml zymosan to 30% NHS (v/v) in DPBS in the absence or presence of 1 μM FH, FHR-1, or FH plus FHR-1 in a final volume of 20 μl. After incubation for 30 min at 37°C, terminal pathway activation was detected by measuring the generated C5b-9 complexes with a C5b-9 ELISA kit (Quidel).

To assess functional consequence of FHR-1/FH competition, we immobilized 10 μg/ml CRP in microtiter plate wells and, after blocking and washing, added 20% (v/v) NHS or FH-depleted serum in the absence or presence of 1 μM FHR-1 or 300 nM FHR-5, used as a control protein. After incubation for 30 min at 37°C, C3 deposition was detected with a HRP-labeled polyclonal anti-C3 Ab.

To measure complement activation and C3 convertase formation on FHR-1, we coated microtiter plate wells with 5 μg/ml FHR-1, FHR-4B, and HSA. After blocking with 4% BSA (w/v) in DPBS, wells were incubated with 10% NHS with or without 5 mM Mg²⁺-EGTA or 5 mM EDTA for 30 min at 37°C. Deposition of C3b, FB, and FP was detected using the corresponding primary and secondary Abs as described previously (35).

To measure classical pathway activation, we coated microplate wells with 4 μg/ml recombinant FHR-1 and, as control, with gelatin, and then incubated with 5 μg/ml mCRP for 1 h at 20°C. After washing with DPBS containing 0.05% Tween 20 (v/v), 1% (v/v) NHS in DPBS containing Mg²⁺ and Ca²⁺ (Lonza) or in DPBS containing 20 mM EDTA was added for 30 min at 37°C. C4-fragment deposition was detected using polyclonal anti-C4 Ab. To measure C4 deposition on ARPE-19 cell-derived ECM, we sequentially incubated the washed cell-free ECM with 20 μg/ml recombinant FHR-1 and 5 μg/ml mCRP, then exposed it to 1% human serum. C4 fragments were detected as described earlier.

HUVECs were cultured in EGM-2 medium (both from Lonza) supplemented according to the manufacturer’s instructions. Confluent cell layer was removed by incubation with Trypsin-EDTA solution (Lonza). Necrosis of HUVEC was induced by heat treatment (65°C, 30 min). Necrotic HUVECs (5 × 10⁵ cells per sample) were left untreated or were preincubated with 2.5 μg/ml mCRP, then incubated with 25 μg/ml recombinant FHR-1 in DPBS containing Mg²⁺ and Ca²⁺ (Lonza). After washing, cells were exposed to 5% NHS in buffer containing 5 mM Mg²⁺-EGTA. The C3bBb convertase was detected by serial incubation with anti-FB and Alexa 488–labeled rabbit anti-goat Ig Ab (Invitrogen, Thermo Fisher Scientific, Waltham, MA). Necrotic cells were gated based on positive staining for propidium iodide. Cells were measured using a DB FACSCalibur flow cytometer (BD Biosciences, Heidelberg, Germany) and CellQuest Pro software, and data were analyzed using FCS Express 3.0 (BD Biosciences).

Ligand binding sites on FH19-20 (Protein Data Bank accession code 2G7I) (43) were visualized using PyMOL (http://www.pymol.org).

Statistical analysis was performed using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA). A p value <0.05 was considered statistically significant.

First, we investigated whether binding of native FHR-1 (Fig. 1) to CRP can be detected from human serum. CRP was immobilized in microplate wells, which were then incubated with serum. The bound proteins were eluted and analyzed by SDS-PAGE and Western blotting using polyclonal anti-FH, which detects both FH and FHR-1. This experiment revealed prominent binding of FH and both FHR-1 glycoforms to CRP (Fig. 2A). This assay, however, does not exclude indirect binding, for example, via C3-fragments, which may explain the background binding of FH to the gelatin-coated control well.

The interaction of FHR-1 with CRP was also studied by ELISA. Previously, CRP was shown to bind to several of its ligands in a pH-dependent manner, showing increased binding at slightly acidic pH, which may be observed at inflammatory sites (44). We used serum from a sepsis patient to analyze the binding of native CRP to immobilized FHR-1 at various pH. No CRP binding was detected at pH 7.4, whereas significant binding of CRP to FHR-1 was detected at pH 6.5 and pH 5.5. Similar binding was observed in the case of the FHR-5 protein, whereas no CRP binding to FH was observed under these conditions (Fig. 2B). To analyze direct protein interactions, we also performed these experiments using recombinant native pCRP and obtained similar results (Fig. 2C).

We set out to clarify which form of CRP interacts with FHR-1. Binding of the native pentameric pCRP to all human FHR proteins was measured by ELISA in Ca²⁺-containing buffer at physiological pH. To this end, FH family proteins were immobilized in microtiter plate wells, and binding of pCRP added in increasing concentrations up to 50 μg/ml (∼435 nM) was measured using polyclonal CRP-specific Ab. pCRP bound strongly to FHR-4, as expected (42), and much more weakly to FHR-3 (Fig. 3A). There was marginal binding to FHR-5 at the highest investigated pCRP concentration, whereas no pCRP binding to FH, FHR-1, and FHR-2 was observed at this pH. However, mCRP that was generated from pCRP by urea-chelation treatment showed strong binding to FHR-1, FH, and FHR-5 (Fig. 3B).

Previously, an mCRP binding site in CCPs 19 and 20 of FH was identified (24). Therefore, we studied whether the homologous CCPs 4 and 5 of FHR-1 also harbor a binding site for mCRP. To this end, the homologous C-terminal two domains of all five FHR proteins were studied for mCRP binding in ELISA. CCPs 4 and 5 of FHR-1 bound mCRP similar to CCPs 19 and 20 of FH. None of the other homologous FHR C-terminal domains bound mCRP (Fig. 3B).

The two FHR-1 allelic variants were also analyzed for mCRP binding using FHR-1*A and FHR-1*B purified from plasma of homozygous carriers of each allele. The two variants bound mCRP equally well (Fig. 3C).

Binding of FHR-1 to surface-bound CRP was measured by ELISA. To this end, CRP was immobilized in microplate wells, resulting in the generation of mCRP (42), and serial dilutions of FHR-1, FH, and FHR-5 were added. Binding of the three FH family proteins was detected with polyclonal anti-FH. Although the Ab shows less reactivity with FHR-5, because of the lower sequence identity to FH, than between FH and FHR-1, a stronger signal for FHR-5 was obtained compared with that for FHR-1 (Fig. 3D).

To further confirm the C-terminal mCRP binding site in FHR-1, we used the mAb C18, which binds in CCP20 of FH and CCP5 of FHR-1 (40, 41). This mAb strongly inhibited mCRP binding to FHR-1 (∼70% inhibition under the experimental conditions), whereas a control mAb that binds in the middle portion of FH did not interfere with mCRP binding (Fig. 4A).

Because most FH autoantibodies that are associated with the autoimmune form of aHUS also bind within these FH domains and cross-react with FHR-1 (20, 40, 45), we tested three IgG preparations isolated from FH autoantibody-positive aHUS patients. Of the three IgGs, two inhibited mCRP binding to FHR-1 (Fig. 4B), by ∼70 and ∼50%, respectively, whereas all three inhibited C3b binding (data not shown).

Some of the aHUS-associated C-terminal FH mutations were shown to impair the binding of FH to CRP, indicating the role of FH residues 1183, 1197, 1210, and 1215 in CRP binding (24). Because the same FHR-1 and FH domains were shown to include a binding site for the CRP homolog pentraxin PTX3 (22, 23, 35), we analyzed the binding of PTX3 to 14 different recombinant FH CCPs 19 and 20 fragments containing single amino acid substitutions. This approach, similar to results obtained previously using peptide array (23), identified residues 1182–1186 and 1203–1215 relevant in PTX3 binding, thus at least partially overlapping with the mCRP binding site (Fig. 4C). In line with this, the two pentraxins mCRP and PTX3 partially competed for FHR-1 binding; mCRP caused an ∼40% reduction in PTX3 binding at the tested highest concentration (Fig. 4D). The pentraxin binding sites, as well as the binding sites for C3b and sialic acid, are shown on a surface representation of the FH19-20 structure (Fig. 4E).

Recent evidence supports a role for some of the FHR proteins as competitive inhibitors of FH on certain ligands, such as C3b, pentraxins, and the ECM (10, 21, 35). Therefore, we investigated whether FHR-1 and FH compete for CRP binding because of the homologous mCRP binding sites in FHR-1 CCPs 4 and 5 and FH CCPs 19 and 20. To this end, the C-terminal FH fragment CCPs 15–20 were immobilized on microplate wells, and binding of mCRP to these domains in the presence of FHR-1 was measured. FHR-1 at 1 μM concentration significantly inhibited mCRP binding to FH CCPs 15–20; FHR-1 caused an ∼25% inhibition in binding as compared with the control protein HSA (Fig. 5A). FHR-1 did not compete with FHL-1 for mCRP (data not shown).

FHR-1 was added to NHS (Fig. 5B) and FH-depleted serum (Fig. 5C) and incubated on CRP-coated wells to assess the functional effect of this competition. Even when exogenous FHR-1 was added at 2 μM concentration to 12.5% human serum, no significant increase in C3 deposition was detected. By contrast, FHR-5 at 300 nM significantly increased C3 deposition in both sera (Fig. 5B, 5C).

Because previous reports are controversial regarding FHR-1 as a complement inhibitor (12, 20, 21), we set out to clarify this issue. First the ability of FHR-1 to regulate the alternative pathway C3 convertase was measured. FHR-1 up to 1 μM did not inhibit the binding of FB to C3b and the formation and activity of the C3bBb convertase (Fig. 6A, 6B). FHR-1 also did not interfere with the ability of FH to regulate the C3 convertase even if FHR-1 was added in 16-fold molar excess to FH, whereas FHR-5 at 10-fold molar excess to FH resulted in ∼50% inhibition of FH decay accelerating activity (Fig. 6C).

The capacity of FHR-1 to inhibit the terminal pathway was assessed in serum. Activation of the alternative pathway was induced by addition of zymosan, and the formation of soluble C5b-9 complexes was measured by ELISA. In this assay, 1 μM (and even 2 μM) FHR-1 did not inhibit zymosan-induced C5b-9 generation, whereas addition of 1 μM FH resulted in ∼40% inhibition of C5b-9 formation (Fig. 7).

To analyze whether FHR-1–bound mCRP can still bind C1q, we immobilized FHR-1 in microplate wells and incubated it with or without mCRP. Then serial dilutions of purified C1q were added and C1q binding was measured. C1q bound to FHR-1 in a dose-dependent manner, and mCRP, particularly at lower C1q concentrations, significantly enhanced C1q binding (Fig. 8A). C1q showed background binding to gelatin, used as control, and mCRP did not affect this interaction. Similarly, mCRP bound to FHR-1 was able to bind the purified C1 complex (Fig. 8B).

We also assessed whether this interaction supports classical pathway activation. FHR-1 was immobilized in microplate wells and preincubated or not with mCRP before incubating with 1% NHS. The bound mCRP on immobilized FHR-1 was able to activate the classical pathway and caused significantly enhanced C4 deposition (Fig. 8C). We also measured classical pathway activation on cell-derived ECM. ARPE-19 cells were cultured in 96-well plates, and the ECM produced by these retinal pigmented epithelial cells was used in an ELISA setting. The washed, cell-free ECM was incubated with recombinant FHR-1 and mCRP as described earlier, followed by addition of 1% human serum. On this ECM surface, mCRP when bound to FHR-1 significantly increased classical pathway activation measured as increased deposition of C4-fragments (Fig. 8D).

FHRs lack the C3b and C3 convertase regulating activities of FH, but it was demonstrated that C3b binding to FHR-4 and FHR-5 can allow the formation of a fully active C3bBb convertase (35, 36). We therefore tested whether FHR-1 shares this ability. To this end, FHR-1 was immobilized on microplate wells, and formation of C3bBb in vitro was measured by sequential incubation with C3b and FB plus FD plus properdin. A significant amount of C3bBb was formed on FHR-1 in this assay, but this was significantly less than that formed on C3b (Fig. 9A). The smaller amount of C3bBb formed on FHR-1 in this particular assay is explained by the smaller amount of C3b bound to FHR-1 compared with the amount of C3b bound to the plate surface when directly immobilizing C3b. The FHR-1–bound convertase was active as demonstrated by the conversion of C3 to C3a (Fig. 9B), suggesting that, similar to FHR-4 and FHR-5, FHR-1 is able to support activation of the alternative pathway. FB and properdin did not directly bind to FHR-1 (Supplemental Fig. 1).

We also tested whether complement activation on FHR-1 occurs in serum. Wells coated with FHR-1, and as positive and negative controls, respectively, with FHR-4B and HSA, were incubated with NHS in buffer containing Mg²⁺/EGTA to allow activation of only the alternative pathway. The binding of C3 fragments, FB and properdin, to FHR-1 was detected, similar to FHR-4B, indicating the formation of the properdin-stabilized C3bBb alternative pathway convertase and complement activation in serum by FHR-1 (Fig. 9C).

To test whether mCRP has an influence on this ability of FHR-1 to activate complement, for example, because of partly overlapping binding site for C3b (see Fig. 4E), we first measured whether mCRP-bound FHR-1 can still bind C3b. To this end, mCRP was immobilized in microplate wells and incubated with FHR-1, followed by the addition of C3b. This experiment showed that mCRP-bound FHR-1 was still able to bind C3b, but required higher C3b concentration than without mCRP (Fig. 10A, Supplemental Fig. 1). In line with this, when ARPE-19 cell–derived ECM was used, mCRP on the ECM was able to bind FHR-1, and such bound FHR-1 could serve as a platform for the assembly of the C3bBb convertase (Fig. 10B).

Previously, FHR-1 was shown to bind to necrotic HUVECs (46). Therefore, HUVECs were necrotized by heat treatment, and the necrotic cells were incubated with FHR-1 alone or with mCRP followed by FHR-1, and exposed to 5% human serum. FHR-1 increased complement activation, as measured by the presence of FB as part of the alternative pathway convertase on the cell surface, which was enhanced by preincubation with mCRP (Fig. 10C). These data indicate that FHR-1 can enhance complement activation in cooperation with mCRP on certain noncellular and cellular surfaces.

Recent studies have changed our understanding of the functions of the FHR proteins (1), demonstrating a role of these proteins, including FHR-1, in enhancing complement activation, mainly via competition with the alternative pathway inhibitor FH. Although it is well established that FHR-1 does not compete with FH on host cell surfaces under normal healthy conditions (reviewed in Ref. 1), very little is known about the FHR-1 ligands and the conditions under which competition between FHR-1 and FH may occur. Similarly, the possibility that FHR-1 has additional functions has not been sufficiently explored. CRP-complement cross-talk is implicated in the opsonophagocytic removal of dying cells and cellular debris (24, 47, 48). Because CRP–FH interaction was described to downregulate inflammation (24, 31) and enhance the silent phagocytosis of apoptotic particles (24), and particularly impaired interaction between CRP and FH was suggested to be involved in the pathogenesis of AMD (32, 33), we investigated the ability of FHR-1 to bind CRP.

In this study, we show that FHR-1, in its native form from serum and also in recombinant form, binds to CRP. As in the case of several of its ligand interactions, CRP binds to FHR-1 in a pH-dependent manner with increased binding at slightly acidic pH, a condition characteristic to inflammatory environments (Fig. 2). A more detailed analysis revealed that it is not the pCRP that binds FHR-1, but the modified mCRP form, which is sometimes termed monomeric or denatured CRP, although this does not necessarily mean formation of CRP monomers or the requirement of harsh conditions. mCRP may be generated at acidic pH, increased temperature, in low-Ca²⁺ conditions or by binding to membranes and surfaces including plastic ones in vitro (26, 27, 30, 42, 44, 49, 50). Remarkably, although CRP has been shown to bind FH also at lower pH (44), in our assays, CRP binding to FHR-1 and FHR-5 could readily be detected under conditions when FH binding was not observed (Fig. 2).

The CRP binding site in FHR-1 is located in the C terminus of the protein (Figs. 3, 4). This is not unexpected because CCPs 4 and 5 of FHR-1 are homologous with CCPs 19 and 20 of FH, which were previously shown to contain a major CRP binding site (24, 51). The FHR-1*A isoform is associated with AMD (52). We found no difference between the two FHR-1 isoforms in binding mCRP. The similar capacity of FHR-1*A and FHR-1*B to bind mCRP is likely explained by the fact that the CCPs 4 and 5 are identical in the two FHR-1 allelic variants.

FHR-1 indeed competed with the C-terminal FH fragment FH15-20 (Fig. 5), but not with FHL-1, which contains another mCRP binding site in CCP7. In serum, however, FHR-1, in contrast with FHR-5, caused only slightly, but not significantly, increased C3 deposition on CRP via competitive inhibition of FH (Fig. 5), likely because of its weaker binding to CRP compared with FH and FHR-5 (Fig. 3D). In addition, FH contains three CRP binding sites (24, 31, 51). For the FH–CRP interaction, an apparent K_D value of 4.2 μM was determined, for the FH fragment CCPs 6–8 (Y402 variant), a K_D value of 3.9 μM, and for CCPs 16–20, a K_D value of 15.3 μM (51). Thus, in FHR-1, the weaker CRP binding site of FH is conserved. Therefore, given the serum concentrations of FH and FHR-1, it is unlikely that under normal circumstances, even at enhanced CRP concentrations, functionally significant competition between the two FH-family proteins occurs that translates into enhanced complement activation. It cannot be excluded, however, that under certain conditions, such as local acidification, increase in FHR-1 concentration, or increased homo- and hetero-oligomerization between FHR-1 and FHR-5, complement deregulation may also take place on CRP. Such scenarios are feasible, because transcriptome analysis showed that the CFHR1 gene is highly expressed in human retinal pigment epithelium (53), altered FHR-1 protein levels were described under inflammatory and disease conditions (54, 55), and mutation in CFHR1 may result in higher-order FHR-1 homo-oligomer formation with increased capacity to outcompete FH (10). In addition, normal FH serum levels show significant variability in individuals, with reported ranges of 116–562 (56) and 124.4–402 μg/ml (57).

The binding site in FHR-1 CCPs 4 and 5 of CRP apparently overlaps with that of the related pentraxin PTX3 (Fig. 4) (35). This binding site was confirmed using the mAb C18, which binds in FHR-1 CCP5 and inhibits both mCRP and PTX3 binding to FHR-1, and by analysis of recombinant mutant proteins; in addition, aHUS-associated FH autoantibodies may also impair these interactions (Fig. 4) (23, 24). In line with these results, we observed partial competition between mCRP and PTX3 for binding to FHR-1. During acute-phase response, systemically strongly increased CRP levels may favor FHR-1 binding to CRP over PTX3, but when the amount of PTX3 is locally increased, PTX3 binding may prevail. Of note, PTX3 is expressed in human retinal pigment epithelial cells and is upregulated by IL-1β and TNF-α (58).

Thus, it seems that, because of its weaker binding to CRP and PTX3 compared with FH, normally FHR-1 does not significantly deregulate complement activation via competition with FH on these ligands. Deregulation may, however, occur when avidity of FH to these ligands is decreased (e.g., because of FH mutations or FH polymorphisms), or if the avidity of FHR-1 is increased (e.g., because of mutation influencing the oligomeric state of the protein, or when high levels of ligand deposition enable divalent binding by FHR-1). It is possible that there is competition between FHR-1 and FH on other modified or newly exposed ligands, such as malondialdehyde epitopes and proteins of the ECM (currently under investigation). For example, FHR-1 could likely pass through the Bruch’s membrane and may compete with FHL-1; even though FH has a higher concentration than FHL-1, because of its size it cannot pass the pores in the fenestrated endothelium in the Bruch’s membrane very efficiently (59). Under these conditions, interaction of FHR-1 with the ECM and CRP may enhance complement activation (Figs. 8, 10). The inverse association of the CFHR1 gene deletion with AMD could partly be explained by our data: in AMD, local inflammation and pH change may favor FHR-1–CRP interaction, but the lack of FHR-1 would leave more room for the anti-inflammatory complement regulators FH and FHL-1.

Because of the debated complement-inhibiting function of FHR-1, we examined its role in alternative pathway C3 convertase regulation and terminal pathway inhibition. In our assays, FHR-1 did not prevent assembly of the alternative pathway C3 convertase, did not accelerate the decay of the convertase, and did not competitively inhibit the convertase decay accelerating activity of FH (Fig. 6). Both FH and FHR-1 bind to C3b/C3d via their C termini, and a recent study confirmed that in FHR-1 the C-terminal C3b binding site of FH is essentially conserved (60). Previously, both weaker (a K_D value of 6.4 μM for the interaction of CCPs 3–5 of FHR-1 with C3b versus 2.6 μM for CCPs 18–20 of FH with C3b) (12, 61) and similarly strong (a K_D value of ∼4 μM for the interaction of FHR-1–like mutant FH19-20 with C3b versus ∼6 μM for FH19-20) (19, 62, 63) binding of FHR-1 to C3b as compared with FH were described; in our convertase assay, no significant competition was observed. FHR-1 could also not inhibit soluble C5b-9 generation in our zymosan-induced complement activation assay (Fig. 7). Likewise, in a recent study, we found no inhibition of serum C5b-9 generation caused by liposomes or Cremophor EL micelles (64). Thus, our results and the majority of literature data (10, 21, 64) do not support the previously suggested terminal pathway-inhibiting function of FHR-1.

Because C1q is one of the main complement ligands of CRP that can initiate classical pathway activation, we examined the influence of FHR-1 on this interaction. We found that FHR-1–bound mCRP could still bind C1q; the observed C1q binding to FHR-1 was significantly enhanced by mCRP (Fig. 8). C1q binding also occurred in the context of C1. Moreover, FHR-1–bound mCRP supported classical pathway activation, also on ECM produced in vitro by retinal pigmented epithelial cells (Fig. 8).

Importantly, FHR-1 by binding C3b allowed for the assembly of a functionally fully active alternative pathway C3 convertase and supported alternative pathway activation (Fig. 9). When FHR-1 was bound on mCRP, more C3b was required to form the C3bBb convertase on FHR-1 (Fig. 10A, 10B), likely because of the partially overlapping binding sites (Fig. 4E). This interaction, however, also supported alternative pathway activation on the surface of necrotic cells exposed to serum, indicating that mCRP and FHR-1 may cooperate in enhancing opsonization (Fig. 10C). Thus, we identify a new function of FHR-1 in the activation of complement, a role in striking contrast with that of the inhibitor FH, and similar to that described for FHR-4 and FHR-5 (35, 36).

In summary, CRP is identified and characterized as a novel ligand of FHR-1. Our results add to the growing body of evidence that, contrary to previous claims, FHR-1 is not an inhibitor of the terminal complement pathway. Instead, FHR-1 is shown to allow alternative pathway C3 convertase formation and alternative pathway activation. The FHR-1–mCRP interactions can enhance both classical and alternative pathway activation. Thus, FHR-1 promotes, rather than inhibits, complement activation.

The authors have no financial conflicts of interest.