FHR5 Binds to Laminins, Uses Separate C3b and Surface-Binding Sites, and Activates Complement on Malondialdehyde-Acetaldehyde Surfaces

The complement system is the central part of the innate immune system and controls many physiological reactions. The self-amplifying cascade is spontaneously activated and the counterbalanced action of activators and inhibitors directs the newly formed effector components to target surfaces. Regulation is required to protect intact self-cells and tissue surfaces from toxic activation products. In line with this, activators such as factor H related-protein 5 (CFHR5) and properdin as well as inhibitors like factor H and C4 binding protein are essential to control complement action in time and space (1–4).

FHR5 is a 65 kDa human plasma protein with a serum concentration of 3–6 μg/ml and consists of nine short consensus repeats (SCRs) (5, 6). This complement activator binds to necrotic human endothelial cells, activated mesangial cells, and the extracellular matrix (ECM) (7, 8). The N-terminal SCRs1-2 contain a dimerization motif and bind properdin (7, 9). FHR5 shares sequence homology with factor H, the central inhibitor of the alternative pathway. Factor H binds C3b, dissociates preformed C3-convertases, and assists the serine protease Factor I in degradation of C3b (4, 5, 10). The N-terminal SCRs1-4 of factor H contain the regulatory region, and the C-terminal end contains the surface-binding region, and the heparin and C3b interaction sites (11–16).

The factor H/CFHR gene cluster is a hotspot for genetic variations and is associated with diseases (17–23). Variations in the CFHR5 gene are linked to kidney diseases, i.e., C3 glomerulopathy (C3G), dense deposit disease (DDD), IgA nephropathy, and atypical hemolytic uremic syndrome (24–27). In CFHR glomerulonephritis, disease-associated genomic rearrangements concerning the CFHR5 gene result in the expression of FHR1::FHR5, FHR2::FHR5, FHR5::FHR2 hybrid proteins, or FHR5 proteins with duplicated segments (17, 24, 28–30). Two related C3G patients (#635 and #638) with a 25 kbp chromosomal deletion express an oligomer-forming FHR2_1–2-FHR5 hybrid protein, which overactivates complement (7, 24).

There is evidence that complement activation is linked to lipids, which can be targeted by oxidation (31, 32). Phospholipids in particular are susceptible to oxidation induced by cellular stress. The oxidation leads to the formation of highly reactive degradation products, which can modify self-molecules. As a consequence, structural neoepitopes referred to as oxidation-specific epitopes are generated. Oxidation-specific epitopes have been characterized as a novel group of damage-associated molecular patterns and include malondialdehyde (MDA)-modified amino groups. MAA epitopes can be detected on apoptotic cells, microvesicles, or oxidized low-density lipoproteins (33–38). Malondialdehyde-acetaldehyde (MAA) represents an advanced MDA-lysine adduct that is stable and able to induce a potent immune response (31, 39). Recently, it was shown that factor H binds to MDA/MAA epitopes and bound factor H inhibits complement progression and decreases the proinflammatory effects of MDA/MAA epitopes (31).

In this study, we identify the ECM-component laminin and MAA epitopes as novel FHR5 ligands. To map ligand-binding regions of FHR5, we generated four fragments (FHR5_1–2, FHR5_3–4, FHR5_5–7, and FHR5_8–9). The binding sites for laminin-521, MAA epitopes, heparin, and necrotic cells were mapped to the middle region (SCRs5-7). However, surface-bound FHR5 contacts C3b with the C-terminal end (SCRs8-9). FHR5 activates complement on MAA surfaces and competes with factor H for MAA epitope binding. The disease-associated FHR2_1–2-FHR5 hybrid protein enhances complement activation on the MAA surface.

The constructs of FHR5 and fragment FHR5_1–2 (expression vector PcDNA4/To/myc-His B; Invitrogen, Karlsruhe, Germany) were generated as described previously (7). The C-terminal fragment FHR5_8–9 was generated by deleting exons I-VII in FHR5 construct with primers FHR5_del1-7-F/FHR5 _del1-7-R. The middle region FHR5_3–7 was generated by deleting exons I–II and exons VIII–IX. To generate FHR5_3–4 or FHR5_5–7, exons V–VII or exons III–IV of FHR5 were deleted in the FHR5_3–7 construct. Recombinant FHR5 and FHR5 fragments were expressed in HEK-293 cells (ACC 305; DSMZ, Braunschweig, Germany) by transient transfection with polyethylenimine (jetPEI; Polyplus, Illkirch, France) and purified by affinity chromatography as described previously (7). Silver staining was performed using a standard protocol. FHR2 was expressed in a Pichia pastoris expression system as described previously (40).

Recombinant FHR5 or FHR5 fragments (50 nM in Dulbecco's PBS [DPBS]) were coated on the surface of MaxiSorp microtiter plates (Nunc, Langenselbold, Germany). Gelatin (Merck, Darmstadt Germany) and BSA (Sigma-Aldrich, Taufkirchen, Germany) were immobilized for background measurement. After blocking (2% BSA in DPBS), C3b (50 nM in DPBS; CompTech, TX) was added to each well and the plate was incubated at 37°C for 1 h. After intensive washing, bound C3b was quantified by C3d polyclonal antiserum (Dako, Glostrup, Denmark) in combination with the corresponding secondary antiserum (Dako). Tetramethylbenzidine (TMB) Plus2 substrate (Kem-En-Tec Diagnostics, Taastrup, Denmark) was used to visualize binding. OD values were measured at 450 nm (Tecan Safire2; Tecan, Männedorf, Switzerland).

Formalin fixed, paraffin-embedded kidney biopsies of two patients with C3G [one DDD and one with membranoproliferative glomerular nephritis (MPGN) pattern] and two normal biopsies of transplanted kidneys at the time of transplantation (0 h biopsies) as normal control were immunohistochemically stained for FHR5. Tissue sections (1 μm) were deparaffinized, rehydrated through graded ethanol solutions, and pretreated with protease at 37°C for 30 min before blocking with normal horse serum (S-2000; Vector Industries). Sections were incubated with FHR5 mAbs (1:1000; R&D, Wiesbaden, Germany) overnight at 4°C. For detection using ZytochemPlus/POLAP100 (Zytomed Systems, Berlin, Germany), the slides were incubated with polymer 1, rinsed in PBS, and incubated with polymer 2. After washing in PBS, slides were stained in new fuchsin naphthol As-Bi phosphate substrate mixture (30 min) and in hemalaun (Mayer) for nuclear counterstaining (1 min).

Heparin-binding microtiter plates (BD, Heidelberg, Germany) were coated with heparin-albumin (50 μg/ml in DPBS; Sigma-Aldrich). After blocking (2% BSA in DPBS), recombinant FHR5, FHR5 fragments, or BSA (100 nM in DPBS) were added and incubated at 37°C for 1 h. Afterwards, bound FHR5 and FHR5 fragments were quantitated by c-myc mAb (clone 9E10; R&D) in combination with the corresponding secondary antiserum (Dako). TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan).

Laminins (BioLamina AB, Sundbyberg, Sweden) or gelatin (10 μg/ml in DPBS; Merck) were immobilized onto the surface of a MaxiSorp microtiter plate (Nunc). After washing and blocking (2% BSA in PBS-T), recombinant FHR5 and the FHR5 fragments (50 nM in DPBS) were added and incubated at 37°C for 1 h. Binding was detected by c-myc mAb (clone 9E10; R&D) in combination with the corresponding secondary antiserum (Dako). TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan).

HUVECs (CRL-1730; American Type Culture Collection, Manassas, VA) were cultured in DMEM medium (Lonza, Wuppertal, Germany) with 10% FBS (Sigma-Aldrich), 25 μg/ml gentamicin-sulfate, and 2 mM ultraglutamin (Lonza). Necrosis was induced by heating the cells (65°C) for 1 h. Necrotic cells were incubated with recombinant FHR5, FHR5 fragments, or BSA (500 nM in DPBS) at 30°C for 40 min. After washing, bound FHR5 proteins were detected using c-myc mAb (clone 9E10; R&D) followed by Alexa 647 mouse-antiserum (Thermo Fisher Scientific, Darmstadt, Germany). To distinguish between intact and necrotic HUVECs, cells were additionally stained with Annexin V-FITC (Thermo Fisher Scientific) and Viability Dye 780 (BioLegend, San Diego, CA). Binding was analyzed by flow cytometry (BD LSR II Flow Cytometer) and evaluated by FlowJo software. Dead cells were gated according to positive Annexin V and Viability Dye 780 staining.

Furthermore, FHR5 binding to the cell surface was analyzed by confocal microscopy (Zeiss LSM 710 confocal microscope and ZEN2009 software; Jena, Germany). To this end, necrotic cells were incubated with FHR5 (20 μg/ml in DPBS) for 30 min. After washing, cells were stained with c-myc mAb (clone 9E10; R&D) in combination with Alexa 488 mouse-antiserum (Thermo Fisher Scientific). The nucleus and the cell membrane were visualized by DAPI (Sigma-Aldrich) or wheat germ agglutinin (WGA) TexasRed (Thermo Fisher Scientific) respectively.

HUVECs were cultured as described above. Then 1 d before the experiment, culture medium was changed to serum-free medium. Cells were harvested and necrosis was induced by heating the cells (65°C) for 20 min or 1 h. Intact cells were kept on ice. Afterwards, intact and necrotic HUVECs were stained with the natural monoclonal IgM Ab NA-17 (41). The IgM isotype control (IC) (clone MM-30; BioLegend) was used to detect background staining. To distinguish between intact and necrotic HUVECs, cells were additionally stained with Viability Dye 780 (BioLegend). Binding was analyzed by flow cytometry and evaluated with FlowJo software. Intact cells were gated according to negative Viability Dye 780 staining. Necrotic cells were gated according to positive Viability Dye 780 staining.

MAA epitopes were visualized by microscopy using the monoclonal NA-17 Ab as described above. The nucleus and the cell membrane were visualized by DAPI (Sigma-Aldrich) or WGA TexasRed (Thermo Fisher Scientific) respectively.

MAA-BSA (42) (10 μg/ml in DPBS) was coated ono the surface of MaxiSorp microtiter plates (Nunc). For control, sham-treated BSA (sh-BSA, 10 μg/ml in DPBS) was immobilized. After washing and blocking (2% BSA in DPBS), recombinant FHR5 and FHR5 fragments (50 nM in DPBS) were added, and the plate was incubated at 37°C for 1 h. Following washing, bound FHR5 and FHR5 fragments were quantitated with c-myc mAb (clone 9E10; R&D) in combination with the corresponding secondary antiserum (Dako). FHR2 binding was tested under the same conditions using the αFHR1 polyclonal antiserum (generated in house) for detection. TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan). Dose-dependent FHR5 binding was performed in two directions. On the one hand, increasing concentrations of MAA-BSA or sh-BSA (2.5, 5, 7.5, 10 μg/ml in DPBS) were coated and FHR5 was added in a constant concentration (3 μg/ml). On the other hand, a constant concentration of MAA-BSA or sh-BSA (10 μg/ml) was immobilized followed by the incubation of increasing FHR5 concentrations (2.5, 5, 7.5, 10 μg/ml in DPBS). The remaining steps were performed as described above.

The binding affinity of FHR5 and MAA-BSA was measured by biolayer interferometry using the BLItz system (Pall ForteBio, Portsmouth, U.K.). His-tagged FHR5 (20 μg/ml in DPBS) was loaded onto Ni-NTA biosensors (VWR, Darmstadt, Germany) for 240 s. After blocking (0.05% gelatin in DPBS, 120 s) and washing (DPBS, 45 s), MAA-BSA was added in different concentrations (72, 36, 18, 9, 4.5 μM), and association was measured (240 s). Dissociation was performed in DPBS and followed for 240 s. Sh-BSA was used as control and the K_D was calculated using BLItz analysis software.

Additionally, the FHR5-MAA interaction was analyzed by Surface Plasmon Resonance using the Biacore 3000 instrument (Biacore, GE Healthcare, Freiburg, Germany). Briefly, MAA-BSA was immobilized to the flow cells of a CM5 sensor chip (Biacore) via standard amine coupling. FHR5 or FHR5 fragments (100 nM in PBS) were flowed onto the MAA-coupled chip or onto a blank control chip at 5 μl per min (25°C).

MAA-BSA or sh-BSA (10 μg/ml) was immobilized onto the surface of a MaxiSorp microtiter plate (Nunc). After washing and blocking (2% BSA in DPBS), normal human serum (NHS) or serum from a C3G patient (20% in DPBS) was added to the coated wells. The plate was incubated overnight at 4°C. After intensive washing, proteins were eluted from the surface with reducing sample buffer (Carl Roth, Karlsruhe, Germany) and separated by SDS-PAGE. Thereafter, Western blot analysis was performed. Bound FHR5 was visualized by FHR5 polyclonal antiserum (AF3845; R&D) in combination with the corresponding secondary antiserum (Dako). The membrane was developed with the Fusion Fx system. Contrast was enhanced by FusionCapt Advance FX7 (Vilber Lourmat, Eberhardzell, Germany).

MAA-BSA was immobilized onto the flow cells of a CM5 sensor chip (Biacore). Fragment FHR5_5–7 (200 nM in 75 mM PBS) was preincubated with heparin (300 μg/ml in 75 mM PBS) and flowed on a MAA-BSA–coupled chip. In parallel, binding of FHR5_5–7 without heparin preincubation was tested.

Apoptosis was induced on Jurkat T cells by irradiating with UVC 100 mJ per cm. After irradiation, cells were kept in RPMI 1640 (Life Technologies, Invitrogen, Carlsbad, CA) without FBS and with penicillin-streptomycin for 16 h at 37°C and 5% CO₂ in a humidified environment. Input was taken out and kept on ice until required, whereas the rest of the cells were incubated with 40 μg/ml of IC IgM (clone MM-30; BioLegend) or LR04 Ab (mouse natural IgM against MDA epitopes) (43) for 30 min at 4°C. Everything was washed in PBS and subsequently incubated with anti-mouse IgM microbeads (Militenyi Biotec, Bergisch Gladbach, Germany). Flow-through containing MDA^high (IC-n) and MDA^low (LR04-n) fractions was collected using an MS MACS column (Miltenyi Biotec) according to the manufacturer’s protocol. Efficiency of fractionation was confirmed using flow cytometry after staining an aliquot of cells with 10 μg/ml of IC or LR04 Abs detected with anti-IgM APC (clone II/41; Thermo Fisher Scientific, eBioscience, San Diego, CA) and extent of apoptosis was monitored with Annexin V PE Apoptosis Detection Kit (eBioscience). Incubation of IC-n and Lr04-n cellular fractions with FHR5 was performed using 1% BSA in DPBS. Bound FHR5 was detected using FACS Calibur (BD).

To test whether FHR5 competes with factor H for MAA binding, MAA-BSA or sh-BSA (10 μg/ml) was immobilized onto the surface of a MaxiSorp microtiter plate (Nunc). After washing and blocking (2% BSA in DPBS), FHR5 or BSA was added to factor H (250 nM in DPBS; CompTech) in increasing concentrations (5–15 μg/ml). Thereafter, the mixture was added to the coated wells. Factor H binding was detected by M16 mAb (11) in combination with the corresponding secondary antiserum (Dako). Competition of FHR5 and factor H was also analyzed with heat-inactivated serum. To this end, heat-inactivated human serum (hiHS) (20% in DPBS) was added to MAA-BSA– or sh-BSA–coated wells either alone or in combination with recombinant FHR5 or FHR2 (20 μg/ml in DPBS). Factor H binding was detected with M16 mAb in combination with the corresponding secondary antiserum (Dako). TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan).

MAA-BSA or sh-BSA (10 μg/ml in DPBS) was immobilized onto the surface of a MaxiSorp microtiter plate (Nunc). After washing and blocking (2% BSA in DPBS), FHR5 (20 μg/ml in DPBS) was added and factor H (50 μg/ml in DPBS) was applied. After incubation and intensive washing, a mixture of C3b (140 nM) and factor I (220 nM) was added to each well. The plate was incubated for 1 h at 37°C. Next, the supernatants were collected, separated by SDS-PAGE, and Western blot analysis was performed. C3 cleavage was visualized by C3 polyclonal antiserum (A213; CompTech) in combination with the appropriate secondary antiserum (Dako). The Western blot was developed with the Fusion Fx system. Contrast was enhanced by FusionCapt Advance FX7. The band intensity was determined with ImageJ (44).

MAA-BSA or sh-BSA (5 μg/ml in DPBS) was immobilized onto the surface of a microtiter plate (Nunc). After washing and blocking (2% BSA in PBS-T), NHS (10%) was diluted in Mg-EGTA buffer (20 mM HEPES, 144 mM NaCl, 7 mM MgCl₂, 10 mM EGTA, pH 7.4). To investigate the influence of the FHR2_1–2-FHR5 hybrid protein, 5% patient serum was diluted with 5% complement active NHS in Mg-EGTA buffer. Serum was added to the wells and the plate was incubated for 40 min at 37°C. After additional washing steps, deposited C3 activation fragments were detected by C3d polyclonal antiserum (Dako) in combination with the corresponding secondary antiserum (Dako). TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan). To demonstrate the influence of FHR5, an additional step was included. FHR5 or BSA (200 nM in DPBS) was added to the MAA-BSA– or sh-BSA–coated wells and incubated for 1 h (room temperature). After washing, NHS (2% in Mg-EGTA buffer) was added and C3 fragment deposition was detected as described above.

MAA-BSA (10 μg/ml in DPBS) or laminin-521 (10 μg/ml in DPBS) was immobilized onto the surface of a MaxiSorp microtiter plate (Nunc). Heparin-albumin (50 μg/ml in DPBS) was coated onto a heparin-binding microtiter plate (BD). After washing and blocking (2% BSA in PBS-T), biotinylated FHR5 [biotinylated as described previously (45), 75 nM in DPBS] was mixed with increasing concentrations of FHR5_5–7 or FHR5_1–2 (0–75 nM) and then the mixture was added to the coated wells. After incubation for 1 h (37°C) and washing, bound FHR5 was quantitated by streptavidin-HRP (Amersham Life Science). TMB Plus2 substrate (Kem-En-Tec Diagnostics) was used to visualize binding. OD values were measured at 450 nm (Tecan).

Statistical analysis of the data was performed by t test for unpaired data using GraphPad Prism 5 for Windows (GraphPad Software, San Diego, CA).

To map binding sites within FHR5, fragments that span the complete protein (FHR5_1–2, FHR5_3–4, FHR5_5–7, and FHR5_8–9) were generated (Supplemental Fig. 1A). After purification, FHR5 and the fragments were separated by SDS-PAGE and visualized by silver staining. FHR5 appeared as a 72 kDa band. FHR5_1–2 was detected as 23 and 17 kDa, FHR5_3–4 as a 25 kDa, FHR5_5–7 as a 35 kDa, and FHR5_8–9 as a 25 kDa band (Supplemental Fig. 1B).

To localize the interaction site, C3b binding to immobilized FHR5 and FHR5 fragments was evaluated. C3b bound to full-length FHR5 (Fig. 1A) and to fragment FHR5_8–9, but not to FHR5_1–2, FHR5_3–4, and FHR5_5–7 (Fig. 1B). Thus, surface-bound FHR5 recruits C3b via the C-terminal SCRs8-9.

FHR5 binds to heparin with the middle region (6). Therefore, we used the novel fragments to fine-map the heparin-binding site. As expected, full-length FHR5 bound to immobilized heparin (Fig. 2A). Also, FHR5_5–7 bound, whereas FHR5_1–2, FHR5_3–4, and FHR5_8–9 did not (Fig. 2B). Thus, FHR5 binds to heparin via the SCRs5-7 and this heparin-binding region is separated from the C3b-binding region.

Immunohistochemically stained sections of kidney biopsy from a C3G patient showed strong granular FHR5 positivity in the glomerular mesangium, as well as along the thickened peripheral capillary walls (Fig. 2C). This pattern indicates FHR5 deposition corresponding to the electron dense deposits in the mesangium, in the subendothelial space (MPGN pattern) or within the glomerular basement membrane (GBM) seen in electron microscopy (data not shown). No relevant staining within the tubulointerstitium could be detected.

Colocalization of FHR5 and laminin was confirmed in vivo by proximity ligation assay using kidney biopsy with lupus nephritis class IV. Numerous FHR5 proteins were in close proximity to laminin chains α5, β2, and γ2 and mostly in projection with the peripheral GBM (Supplemental Fig. 2, panels I–III). In a normal, 0 h transplant kidney very few FHR5 molecules were identified in proximity with laminin α5, β2 or γ2 (Supplemental Fig. 2, panels IV–VI). Thus, in the GBM of a diseased kidney, FHR5 and laminins are in close proximity.

FHR5 binds to necrotic cells and to ECM components (7, 8). As laminins are constituents of the GBM and the mesangial matrix, we analyzed FHR5 binding to laminin isoforms. FHR5 bound to laminin-521 (Fig. 2D), the major component of the GBM and, with less intensity, to laminin-211, the major component of the mesangium (46, 47). As FHR5 binds to the GBM-specific laminin-521, we mapped the binding site for this isoform. Fragment FHR5_5–7 bound to immobilized laminin-521, whereas the other fragments did not (Fig. 2E). Thus, FHR5 contacts laminin-521 via the middle SCRs5-7.

FHR5 binds to late apoptotic and necrotic human cells (7). To localize the cell surface binding site, we tested FHR5 and the fragments for binding to necrotic cells. Full-length FHR5 [mean fluorescence intensity (MFI)_FHR5 41,272, MFI_control 4691, Fig. 3A], FHR5_5–7 (MFI_5–7 14,087), and to a lesser extent FHR5_8–9 (MFI_8–9 6251, Fig. 3B) bound to necrotic HUVECs. FHR5_1–2 and FHR5_3–4 did not bind. Thus, FHR5 binds to necrotic human endothelial cells mainly via the middle region (SCRs5-7). When binding was evaluated by confocal microscopy, FHR5 (green fluorescence) attached to specific regions of the necrotic surface (Fig. 3C). These regions lacked staining of the cell-surface marker WGA (red fluorescence) and represent sites where membrane integrity is lost.

As MDA/MAA epitopes are formed due to the oxidation of membrane phospholipids (38), we assumed that FHR5 binding sites at the necrotic cell surface correlate with the exposure of MDA/MAA epitopes. MDA/MAA epitopes were visualized on the surface of necrotic HUVECs by confocal microscopy with the specific mAb NA-17 (Fig. 4A) (41). NA-17 reactivity (green fluorescence) was prominent at damage sites that lacked WGA reactivity (red fluorescence). Therefore, FHR5 and MDA/MAA epitopes are localized at the same sites of necrotic cells. In contrast to necrotic cells, intact HUVECs showed low NA-17 reactivity (MFI 776, MFI_control 250). The amount of MAA/MDA epitopes was dependent on the time of necrosis induction (Fig. 4B, middle and right panel, Supplemental Fig. 3A). NA-17 staining was present after 20 min of necrosis induction (MFI 22,461, MFI_control 2530) and increased with time (60 min: MFI 29,376, MFI_control 1705).

As FHR5 binds to sites of necrotic HUVECs where MAA epitopes are localized, we hypothesized that FHR5 directly binds to MAA epitopes. Full-length FHR5 bound to immobilized MAA-BSA (Fig. 5A) in an ELISA-based assay and binding was dose dependent (Supplemental Fig. 3B, 3C). This interaction was specific as FHR5 did not bind to sh-BSA. The binding region for MAA epitopes was mapped to FHR5_5–7. None of the fragments FHR5_1–2, FHR5_3–4, FHR5_8–9, or FHR2 bound to MAA-BSA (Fig. 5B, Supplemental Fig. 3D). Binding of FHR5 and fragment FHR5_5–7 was further evaluated in real-time by surface plasmon resonance (Fig. 5C). MAA-BSA was coupled to a sensor chip and FHR5 or the fragments were added as analytes. FHR5 and again FHR5_5–7 bound to the MAA-BSA–coupled chip. Fragment FHR5_8–9 bound weakly whereas FHR5_1–2 and FHR5_3–4 did not bind. Thus, FHR5 contacts both MAA epitopes and heparin via the SCRs5-7. To investigate if the binding regions overlap, we tested whether heparin influences FHR_5–7 binding to MAA-BSA. Indeed, heparin preincubation reduced FHR5_5–7 binding to MAA-BSA (Fig. 5D).

The affinity of the FHR5::MAA interaction was determined by biolayer interferometry. His-tagged FHR5 was bound to an Ni-NTA sensor. Binding of MAA-BSA to sensor-attached FHR5 showed a strong association profile and rather stable complexes upon removal of the ligand (Fig. 5E). FHR5 binds to MAA epitopes with nanomolar affinity (K_D of 20.47 nM).

Next, we asked if serum-derived FHR5 and the C3G-associated hybrid protein FHR2_1–2-FHR5 also bind to MAA-BSA. Therefore, NHS or serum from patient #638 was added to MAA-BSA–coated wells. Following incubation, surface-bound proteins were eluted, separated by SDS-PAGE, and analyzed by Western blot. Serum-derived FHR5 and also the disease-associated FHR2_1–2-FHR5 hybrid protein bound to MAA-BSA (Fig. 5F, lane 1 and 3).

To confirm that FHR5 binds to the surface of dead cells via MAA epitopes, apoptotic Jurkat cells were immune depleted and two fractions, namely MAA^high (IC-n) and MAA^low (LR04-n), were collected. Afterwards, FHR5 was incubated together with the cells and binding was analyzed by flow cytometry. FHR5 bound with lower intensity to MAA^low (LR04-n) cells (Supplemental Fig. 3E, 3F). Thus, the amount of MAA epitopes influences the strength and intensity of FHR5 binding to apoptotic cell surfaces.

The activator FHR5 and the inhibitor factor H bind to MAA epitopes (31). Therefore, we hypothesized that both proteins compete for binding. FHR5, but not BSA, inhibited factor H binding to MAA-BSA and the effect was dose dependent (Fig. 6A). Consistently, FHR5 when added to hiHS reduced factor H binding to the MAA surface (Fig. 6B). When added to hiHS, FHR2 did not affect factor H binding. Interestingly, the oligomer-forming FHR2_1–2-FHR5 hybrid (hiHS patient #638) allowed significantly lower factor H binding.

MAA-bound factor H has cofactor activity and assists in factor I–mediated C3b degradation (31). As FHR5 competes with factor H for MAA epitope binding, we asked whether FHR5 affects the factor H–mediated cofactor activity. To this end, FHR5 was bound to MAA-BSA–coated wells, Factor H was added and following washing, Factor I and C3b were added. After incubation, the supernatants were separated by SDS-PAGE and the C3b cleavage products were visualized by Western blot. FHR5 reduced the cofactor activity of factor H on the MAA-BSA–coated surface (Fig. 6C, lane 2, and 3, 6D).

FHR5 binds to necrotic cells via MAA epitopes and competes with factor H for binding. As MAA-bound factor H inhibits complement, we asked whether FHR5 activates complement on this specific surface (31). Therefore, NHS was added to MAA-BSA–coated wells and complement activation was monitored by measuring C3 fragment deposition. Upon NHS challenge, increased C3 fragment deposition was detected. Interestingly, C3 fragment deposition on MAA-BSA was stronger when patient serum containing the FHR2_1–2-FHR5 hybrid protein was present (Fig. 7A).

To prove the complement activating role of MAA-bound FHR5, immobilized MAA-BSA was preincubated with FHR5 or BSA. Afterwards, complement-active NHS was added and C3 fragment deposition was quantified. Indeed, MAA-bound FHR5 enhanced the C3 fragment deposition (Fig. 7B). Thus, FHR5 has an activating effect on MAA epitope–exposing surfaces (Fig. 8).

FHR5 is a surface-acting complement activator (7). Mutations and copy-number variations in the CFHR5 gene are associated with kidney diseases including C3G and atypical hemolytic uremic syndrome (24, 25, 27–29). To identify surface ligands of FHR5 in the damaged glomerulus, we first asked whether FHR5 binds to glomerular ECM components and to modified lipids, specifically MAA epitopes that are exposed on necrotic cells. Indeed, FHR5 bound strongly to laminin-521, the major constituent of the GBM and also to mesangial laminin-211 (46, 47). FHR5 also bound to oxidation-specific MAA epitopes and to necrotic human endothelial cells at the site of severe membrane damage (48). Our findings show that FHR5 binds to glomerular matrix components, suggesting that FHR5 activates complement at damaged sites.

To characterize the domain structure of FHR5 and to map binding domains, four fragments which span the complete protein were generated. These fragments, each two or three SCRs in length, identified the middle region (SCRs5-7) of FHR5 as the binding sites for laminin-521, MAA epitopes, necrotic cells, and heparin. Recently, the binding site for heparin and C-reactive protein was localized to the middle domain. However, this study did not exclude the contribution of the SCRs3-4 (6). FHR5_5–7 when added inhibited binding of intact full length FHR5 to the MAA-BSA–, heparin-, and laminin-521–coated surface. The N-terminal deletion fragment FHR5_1–2 did not influence FHR5 binding to any of the matrices. Thus the binding region within the newly generated fragment retains its tertiary structure (Supplemental Fig. 4A–C). Interestingly, surface-bound FHR5 recruits C3b via a separate region, which is localized in the C-terminal SCRs8-9. Thus, FHR5 similar to the complement inhibitor FHL1 uses two distinct regions to contact heparin, cell surfaces, and C3b (49). However, this profile differs from that of FHR1 and factor H, which contact cell surfaces and C3b with their homologous C-terminal regions. The lower degree of homology of the C-terminal SCRs is in agreement with this difference. The SCR8 of FHR5 has 64% homology with the SCR4 of FHR1 and the SCR19 of factor H. Conversely, the SCR9 of FHR5 has 41% homology with the SCR5 of FHR1 and the SCR20 of factor H (5, 16). In summary, FHR5 has three major interaction sites (Fig. 8): 1) the N-terminal dimerization and properdin binding region (SCRs1-2) (7, 9); 2) the surface binding middle region (SCRs5-7), which contacts heparin, laminin-521, MAA epitopes, necrotic cell surfaces, and C-reactive protein (6); and 3) the C-terminal C3b binding region (SCRs8-9).

FHR5 is an important mediator of glomerular pathology and was initially identified in glomerular immune deposits of patients with kidney diseases (50, 51). Immunostaining of FHR5 showed primarily mesangial and capillary reactivity in biopsies from patients with focal glomerular sclerosis, membranous nephropathy, MPGN I, IgA nephropathy, postinfectious glomerulonephritis, and lupus nephritis. In these cases, FHR5 reactivity correlated with C3b and C5b-9 reactivity (50). In this study, we show that FHR5 binds to laminins, in particular to laminin-521, the major constituent of GBM. FHR5 binding to laminin-521 is in agreement with immunohistological localization of FHR5 in intact kidneys and kidneys from C3G patients (50). A recent proteomic approach with glomeruli of DDD patients found FHR5 and also laminin 5 and 2 chains in glomerular deposits (51). Additionally, staining of laminin α5, β2, and γ1 in the kidney is more prominent in disease scenarios (46).

FHR5 also bound with high intensity to MAA epitopes, which have been linked to several diseases including the retinal disease age-related macular degeneration and atherosclerosis (33). In addition, MDA modifications of laminins were proposed (52). Factor H binds to MAA/MDA epitopes on the surface of apoptotic and necrotic cells, assists in the inactivation of C3b, and contributes to tissue homeostasis by blocking the proinflammatory effects of modified cells. The age-related macular degeneration risk variant of factor H_H402 binds with lower intensity to MAA/MDA epitopes compared with the protective variant factor H_Y402. This difference in binding intensity correlates with the strength of complement inhibition and the inflammatory response (31). The interaction of both regulators, FHR5 and factor H, with MAA epitopes suggested binding competition. Indeed, the presence of FHR5 inhibited factor H binding to MAA epitopes and decreased the factor H mediated cofactor activity (Fig. 8). Despite the lower plasma levels [FHR5: 3–6 μg/ml (6), factor H: 156 μg/ml (53)] FHR5 has an activating effect on the MAA surface.

Copy-number variations in the CFHR5 gene can result in hybrid genes and the expression of FHR1::FHR5, FHR2::FHR5, FHR5::FHR2 or FHR5 variants (17, 24, 28–30). Importantly, these genetic changes occur in the presence of an intact factor H gene (24). Enhanced binding of the oligomer forming FHR2_1–2-FHR5 to MAA surfaces also suggests that other FHR5 hybrid proteins bind with higher intensity to oxidized lipids and enhance complement activation on such modified human surfaces. As FHR2 did not bind to MAA epitopes, the overactivating effect of the hybrid protein is interpreted by the potential to form oligomeric complexes (7).

In summary, we identified laminins and MAA epitopes as novel FHR5 ligands. Notably, FHR5 binds laminins and surfaces with the middle region (SCRs5-7) and contacts C3b with a separate region (SCRs8-9). Binding of FHR5 to laminins, in particular laminin-521, the major constituent of the GBM, provides a new link for FHR5 function in the kidney and for disease pathology. The activating effects of FHR5 on modified surfaces like MAA epitopes and competition with factor H provide a new explanation how FHR5 can facilitate silent noninflammatory removal of dead cells. The overactivating effect of hybrid proteins like FHR2_1–2-FHR5 on these modified surfaces provides a new link for disease pathogenesis.

We thank the patients and the family for participating in the study and for contributing valuable specimens. We thank Steffi Hälbich, Laura Broschat, and Julian Matthes (Department of Infection Biology, Leibniz Institute for Natural Product Research and Infection Biology, Jena, Germany) for excellent technical support.

The authors have no financial conflicts of interest.