Factor H (FH) is a key alternative pathway regulator that controls complement activation both in the fluid phase and on specific cell surfaces, thus allowing the innate immune response to discriminate between self and foreign pathogens. However, the interrelationships between FH and a group of closely related molecules, designated the FH-related (FHR) proteins, are currently not well understood. Whereas some studies have suggested that human FHR proteins possess complement regulatory abilities, recent studies have shown that FHR proteins are potent deregulators. Furthermore, the roles of the FHR proteins have not been explored in any in vivo models of inflammatory disease. In this study, we report the cloning and expression of recombinant mouse FH and three FHR proteins (FHR proteins A–C). Results from functional assays show that FHR-A and FHR-B proteins antagonize the protective function of FH in sheep erythrocyte hemolytic assays and increase cell-surface C3b deposition on a mouse kidney proximal tubular cell line (TEC) and a human retinal pigment epithelial cell line (ARPE-19). We also report apparent KD values for the binding interaction of mouse C3d with mouse FH (3.85 μM), FHR-A (136 nM), FHR-B (546 nM), and FHR-C (1.04 μM), which directly correlate with results from functional assays. Collectively, our work suggests that similar to their human counterparts, a subset of mouse FHR proteins have an important modulatory role in complement activation. Further work is warranted to define the in vivo context-dependent roles of these proteins and determine whether FHR proteins are suitable therapeutic targets for the treatment of complement-driven diseases.
Understanding regulatory mechanisms by which the alternative pathway (AP) controls spontaneous activation of complement in the fluid phase and the amplification of complement on specific surfaces has important implications for treating complement-driven inflammatory disease. Unlike the classical or lectin pathways, the AP does not require any specific molecular recognition for its initiation but is activated by hydrolysis of C3 to C3(H2O) in the fluid phase, which results in activation and production of C3b through the activities of factor D, factor B, and properdin. This phenomenon, known as C3 tick-over, occurs spontaneously and allows for the rapid initiation and amplification of complement. Given that this pathway is responsible for ≥80% of the final downstream effect of initial specific activation of the classical and lectin pathways, precise control of the AP and its amplification loop is required (1).
Factor H (FH) is a major soluble complement regulator that is essential for controlling AP activation in the fluid phase and on cell surfaces. Several human diseases are associated with mutations and autoantibodies that alter either FH function, or the activities of the closely related FH-related (FHR) proteins. For example, mutations or polymorphisms in the CFH and CFHR gene family have been linked to the renal diseases atypical hemolytic uremic syndrome and IgA nephropathy as well as diseases that have glomerular pathologies, including dense deposit disease and FHR5 nephropathy, which are encompassed under the C3 glomerulopathy umbrella (2–4). Other autoimmune diseases associated with alterations within the CFH and CFHR gene family include systemic lupus erythematosus and age-related macular degeneration (AMD) (5–7).
Given the recent expansion in research toward understanding the human FHR protein family member disease associations, one question that remains unanswered is whether the mouse FHR (mFHR) proteins are functional orthologs to their human counterparts. The mouse FH (mFH) gene consists of 22 exons that share 63% homology with human CFH and encode a protein composed of 20 short consensus repeat (SCR) domains (8). Unlike its human counterpart, the mFH gene does not have a FHL-1 variant, although it does contain an unspliced exon (exon 9) that could encode an SCR domain with a stop codon. Similar to their human FHR counterparts, a total of five mFHR genes have been identified, and evidence for four mFHR proteins have been inferred from mRNA transcripts isolated from a mouse liver cDNA library; however, direct comparison with the human CFHR gene family is not straightforward (9, 10). These predicted mouse proteins also exhibit high sequence identity with important ligand-binding and self-surface recognition domains of mFH. After the initial characterization of the four classes of mFHR transcripts by Vik et al. (9), this gene family was not thoroughly examined until Hellwage et al. (10) published data on mFHR protein expression and binding partners. However, little subsequent work has been published that characterizes these genes, evaluates the functional roles of the proteins they encode, or examines the concentrations and functions of these proteins in vivo.
One potential limitation to studying the mouse FHR gene family lies with understanding the published nomenclature as it relates to that of the human FHR gene family. Although the nomenclature for the mouse FHR genes (labeling them alphabetically as A, B, C based on their position from mFH) was proposed more than a decade ago, a current search of genome browsers lists these genes under various aliases, including CFHR4/2, CHFR2, FHR-1, or FHR-3 (10). For example, the mouse gene referred to as CFHR-1 is located at the mFHR-E gene position, which is farthest from the mouse FH gene. This same gene was originally classified by Vik et al. (9) as clone 13G1 and is a “class D” mFHR transcript. Additionally, the mouse genes are predicted to have higher sequence identity to both full-length FH and to one another compared with the human FHR genes. Therefore, given our relative lack of knowledge about the potential functions of the mFHR proteins, we elected to investigate three constructs that were discussed in the two original publications on mFHRs. We generated an mFHR-A construct based on the original prototype sequence provided by Vik et al. (9), and the sequences of mFHR-B and mFHR-C constructs were generated based on work by Hellwage et al. (10). Notably, all three of the murine constructs that were chosen for this study lack a dimerization domain found in human FHR-1, FHR-2, and FHR-5 (11). However, other sequences may mediate dimerization of the murine FHRs, and therefore we chose to focus our efforts on obtaining a better understanding of mFHR protein function.
We hypothesized that mFHRs, similar to their human counterparts, regulate complement activation by antagonizing FH function. To test this hypothesis, a comprehensive characterization of the mFHR proteins was undertaken. Hemolytic assays were used to assess the function of recombinant proteins. To characterize the interaction between mFHRs and C3b/C3d fragments, a combination of ELISA and surface plasmon resonance (SPR) experiments were performed. To elucidate the functional homology between a conserved C3d binding site on the human and mouse FHR proteins, mutant mFHRs were generated and their activity was assessed. Using in vitro experiments with two cell lines, the role of mFHRs on complement regulation on different cell surfaces was evaluated. In summary, these analyses provide a better understanding of a subset of mFHR orthologs and provide a potential means to modulate FHR function in an informative manner in vivo and better understand how manipulation of their activities can potentially ameliorate the human diseases to which these proteins contribute in a pathophysiologically important manner.
Materials and Methods
DNA and protein sequences for the human FH, mFH, and mFHR families and C3b/C3d proteins were analyzed using the Bioinformatics Resource Portal (ExPASy) of the Swiss Institute of Bioinformatics (https://www.expasy.org), the Ensembl database, which is a joint project between the European Bioinformatics Institute and the Wellcome Trust Sanger Institute (http://www.ensembl.org), and the UCSC Genome Browser hosted by the University of California, Santa Cruz (https://genome.ucsc.edu).
mFH and FHR protein expression
DNA sequences of full-length mFH and mFHR constructs, including mFHR-A [based on the sequence for class A clone 3A4/5G4 by Vik et al. (9)], mFHR-B, mFHR-C, and mFH19–20 were generated commercially (GeneArt by Life Technologies). The unique UniProt identifiers for each construct include: mFH (UniProt: P06909), mFHR-A (UniProt: Q61407), mFHR-B (UniProt: Q4LDF6), mFHR-C (UniProt: Q0KHD3), and mFH19–20 (UniProt: P06909 residues 1112–1234). Each DNA sequence was codon-optimized (Homo sapiens) and engineered into a Gateway pDONR221 entry vector (Life Technologies) following previously published construct design and experimental conditions (12). For protein expression, FreeStyle 293-F cells (Invitrogen) were obtained and subcultured in FreeStyle 293 expression medium every 3 d. Transfections were performed using FreeStyle MAX reagent (Life Technologies) according to the manufacturer’s instructions. Cell supernatant was harvested 5–7 d posttransfection using centrifugation at 8000 rpm for 15 min in sterile 250-ml tubes.
Mutant mFHR constructs
To create mutant mFHR constructs, a QuickChange Lightning site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) was used according to the manufacturer’s directions. Generation of an N340A/D342A mFHR-A mutant was created with the following primers that were codon-optimized for mammalian cell expression: 5′-CCT ATC GAC GCC GGG GCT ATC ACC AGC CTG-3′ (forward) and 3′-GGA TAG CTG CGG CCC CGA TAG TGG TCG GAC-5′ (reverse) to create the sequence PIDAGAITSL. The mutant N220A/D222A mFHR-B construct was created with primers forward 5′-CCC ATC GAC GCT GGG GCT ATC ACC AGC CTG-3′ (forward) and 3′-GGG TAG CTG CGA CCG CGA TAG TGG TCG GAC-5′ (reverse) to create the sequence PIDAGAITSL.
FH and FHR protein purification
Purification of the FH and FHR proteins was performed using immobilized metal affinity chromatography. Harvested supernatant was diluted with 5× buffer (0.1 M sodium phosphate [pH 7.8], 0.1 M imidazole, and 2.5 M NaCl). Samples were applied to either a 5-ml HisTrap HP column (GE Healthcare Life Sciences) or a column containing HisPur Ni-NTA resin (Thermo Fisher Scientific) as per the manufacturer’s instructions. After applying supernatant to the Ni-NTA column, bound His-tagged proteins were eluted using a linear imidazole gradient from 20 mM and 0.5 M imidazole. Most proteins were completely eluted from the resin using 0.2 M imidazole; however, 1- to 2-ml fractions were collected and analyzed by monitoring the OD ratio at 260/280 nm. Protein purity was assessed by SDS-PAGE. Fractions containing purified protein were pooled and concentrated at room temperature using a Vivaspin 20 spin concentrator (Millipore). Proteins were buffer exchanged into Dulbecco’s PBS (DPBS; 2.7 mM KCl, 1.5 mM KH2PO4, 138 mM NaCl, 8.1 mM Na2PO4 [pH 7.4]). A polishing step was performed by applying protein to either S200 or S300 HiPrep 16/60 Sephacryl size exclusion columns (GE Healthcare) on an ÄKTAPure HPLC system (GE Healthcare) that had been equilibrated with DPBS. Peak fractions were eluted in 1-ml volumes and analyzed by SDS-PAGE. Protein concentrations were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific) by entering molecular mass values and molar extinction coefficients for each protein. Purified proteins were flash frozen in 50- to 100-μl aliquots using liquid nitrogen and stored at −80°C. SDS-PAGE analysis was used to analyze protein purity for both glycosylated and deglycosylated proteins. To deglycosylate proteins, a Remove-iT PNGase F and a magnetic chitin bead protocol (New England Biolabs) was used according to the manufacturer’s instructions.
Mouse C3d expression and purification
A construct encoding wild-type mouse C3d in a pCR TOPO vector was obtained from a previous laboratory stock. This construct was sequenced with M13 forward and reverse primers to confirm that the DNA sequence could be translated to match the UniProt identifier P01027 (residues 1002–1303). The thioester-containing domain (TED) of C3b is contained within this sequence. Four residues in this sequence (Cys-Gly-Glu-Gln) make up a 15-member thiolactone ring that facilitates the covalent interaction between C3b and cell surfaces through residue Q1013. Primers were used to mutate C1010A to prevent dimerization and reactivity of C3d by mutating the sequence from PAGCGEQN to PAGAGEQN as previously described (13). Primers (forward, 5′-CCC GCA GGC GCT GGG GAA CAG AAC-3′, reverse, 3′-GTT CTG TTC CCC AGC GCC TGC GGG-5′) and the QuickChange Lightning site-directed mutagenesis kit were used (Agilent Technologies) followed by transformation of XL-10 Gold ultracompetent cells (Stratagene). Subcloning of the mC3D construct into a pGEX-2T GST expression vector (GE Healthcare Life Sciences) was performed. Dual digestion of both vectors was performed using BamHI-HF and EcoR1-HF restriction enzymes per the manufacturer’s specifications (New England Biolabs) followed by gel purification and fragment ligation using T4 DNA ligase. Both pGEX forward and reverse primers were used to confirm the sequence of the mC3d-pGEX-2T GST expression construct prior to transformation and expression of protein in BL21 (DE3) cells.
Circular dichroism spectroscopy
Recombinant mC3d was analyzed using circular dichroism (CD) spectroscopy. A Jasco J-815 spectropolarimeter (Jasco, Easton, MD) with a Jasco DP-500/PS2 system was used in this experiment. A Lauda model RMS circulating water bath (LAUDA-Brinkman, Delran, NJ) was used to maintain temperature control of the optical cell. Results are expressed as mean molar ellipticity [θ] (deg·cm2·dmol−1) and were calculated using the following equation: [θ] = [θobs × MRW/10lc], where θobs is the observed ellipticity in millidegrees, MRW is the mean residue weight of the peptide (peptide molecular mass/number of residues), l is the optical path length of the cell in centimeters, and c is the peptide concentration in milligrams per milliliter. A solution of 1 mg/ml mC3d in PBS was used in this experiment. Variable wavelength measurements of peptide solutions were scanned at 5°C from 195 to 250 nm in 0.5-nm increments at a scan rate of 50 nm/min.
Differential scanning fluorimetry
The thermal stability profiles of mFH, mFHR, and C3d recombinant proteins were investigated by combining 75 μg/ml protein in PBS (catalog no. 21-040-CV; Corning) and dye in a total reaction volume of 30 μl. A fresh 300× stock of SYPRO Orange (5000×) was diluted in fresh DMSO (Sigma-Aldrich) and a final concentration of 10× SYPRO Orange per reaction was used. This specific dye has excitation/emission wavelength profiles of 490 and 575 nm. Experiments were performed in a 96-well PCR plate (Bio-Rad Laboratories) that was sealed using optical-quality sealing tape (Bio-Rad Laboratories). An iQ5 real-time PCR detection instrument heated samples at 1°C/min from 20 to 95°C with a fluorescence reading recorded every 0.2°C. The melting temperature (Tm) for each of the proteins was determined by fitting data to a Boltzmann model, and data were normalized in GraphPad to make comparisons between different proteins (14).
ELISA with mouse C3d
To examine the interaction between mC3d and mFH/mFHR proteins, ELISA assays were used. Clear flat-bottom polystyrene 96-well microplates (Corning) were coated overnight at 4°C with 50 μl per well of recombinant mC3d at a concentration of 10 μg/ml in 50 mM sodium bicarbonate buffer (pH 8.8). Plates were washed three times in wash buffer (PBS plus 0.05% Tween 20) before blocking with 100 μl per well PBS plus 1% BSA. Plates were again washed prior to addition of mFH or mFHR proteins. Following incubation with mFH/mFHR proteins for 1 h at room temperature, plates were washed before addition of 1:1000 anti-6X His tag Ab (catalog no. ab1187; Abcam) in 100 μl PBS plus 1% BSA for 1 h at room temperature. Finally, plates were washed before developing with ABTS. Following a 30-min RT incubation, OD values were determined at wavelength of 405 nm.
Surface plasmon resonance
A Biacore T200 instrument (GE Healthcare Life Sciences) was used to calculate apparent KD values for mFH and mFHR proteins and mC3d. A series S sensor CM5 chip (GE Healthcare Life Sciences) was chemically coupled with 2500 response units of mC3d in 10 mM acetate buffer (pH 4) using standard amine coupling techniques. To assess binding with mC3d, mFH and mFHR wild-type and mutant proteins were serially diluted in HBS-EP+ buffer (10 mM HEPES [pH 7.4], 150 mM NaCl, 3 mM EDTA, and 0.005% [v/v] surfactant P20) and injected at a flow rate of 15 μl/ml for ∼80 s while dissociation of mC3d:mFH/mFHR complexes was monitored to 600 s after injection. The running buffer was HBS-EP+. Guided by the Biacore evaluation software, a two-state reaction model was determined to provide the best fit and therefore this model was used to estimate the KD values presented in this study.
Erythrocyte hemolytic assays
The functional activities of mFH and mFHR proteins were analyzed using erythrocyte hemolytic assays that have been previously described (15). Sheep erythrocytes (Complement Technology) were obtained and resuspended in Mg-EGTA buffer (0.1 M MgCl2, 0.1 M EGTA [pH 7.3]). To each 24-μl reaction, ∼0.7 to 1 × 106 cells (in Mg-EGTA buffer) were added to the bottom of a PCR tube. Either mFH or mFHR proteins were added to the cells followed by normal mouse serum (NMS) from C57BL/6 mice (40% final concentration). Reactions were incubated at 37°C for 30 min and quenched using 200 μl gelatin veronal buffer containing EDTA. Following centrifugation, 180 μl was read at OD412. As a negative control, the same reaction mixture was prepared using cells and NMS in PBS without recombinant protein. The percentage of hemolysis was determined by subtracting the OD412 of the background hemolysis control (PBS plus NMS) and dividing by the maximum hemolysis of cells in sterile water.
Erythrocyte hemolytic competition assays were performed to evaluate the competition between mFH and the mFHRs. For these reactions, 2 μM recombinant mFH was added to 1 × 106 sheep erythrocytes in a 24-μl reaction volume with 20% NMS. mFH was added to the cells 5 min prior to adding the 20% NMS and either 1 μM mFHR-A or 4 μM mFHR-B to the samples. Samples were then incubated at 37°C for 30 min and quenched using 200 μl of gelatin veronal buffer containing EDTA. Following centrifugation, 180 μl was read at OD412. The data represent mean ± SEM from at least three separate experiments with three replicates per sample.
Murine renal tubular epithelial cells (TECs) were cultured in DMEM media supplemented with 10% FBS, 0.1 g/ml penicillin/streptomycin, and 1% nonessential amino acids (Thermo Fisher Scientific). Cells were grown to 80–90% confluence prior to analysis. The cells were treated with Accutase and then washed using PBS. Cells were incubated in 10% NMS in the presence or absence of increasing amounts of recombinant mFHR proteins for 30 min. Surface-bound C3b was determined by incubating cells with a 1:150 dilution of FITC-labeled goat anti-mouse C3 Ab (no. 55500; Cappel) in the dark on ice for 60 min. Cells were next washed twice and resuspended in FACS buffer (1× PBS with 1% BSA). Samples were run using a FACSCalibur flow cytometer (BD Biosciences), and results were analyzed using FlowJo software. To measure C3b deposition on retinal pigment epithelial cells (ARPE-19), experiments were performed according to previously published methods (16). Briefly, ARPE-19 cells were cultured in DMEM F12 (Invitrogen) with 10% FBS and 1× penicillin/streptomycin. Stressed cells were incubated with 1 mM H2O2 for 1 h in fresh media whereas unstressed cells were incubated in fresh media. Cells were treated with Accutase (Innovative Cell Technologies), washed, and resuspended in DPBS. Next, cells were incubated with mFHR proteins for 15 min prior to addition of 10% NMS and further incubation at 37°C for 35 min. Cell surface C3b deposition was examined by gating cells based on their forward scatter height and side scatter properties. At least 60,000 cells were analyzed and the mean fluorescence intensity was plotted.
Generation and assessment of recombinant FH and FHR proteins
The positions of the mFH and mFHR gene family were analyzed using various genome browsers prior to designing recombinant DNA constructs (Fig. 1A). To analyze the function of mFH and mFHR proteins, expression constructs for mFH, mFHR-A, mFHR-B, mFHR-C, and mFH19–20 were synthetically generated (Fig. 1B). The sequence for mFHR-A was based on the prototype transcript sequence for the class A clones (3A4/5G4) characterized by Vik et al. (9), whereas the mFH19–20 construct consisted of the last two C-terminal SCR domains of full-length mFH (15, 16). Constructs were engineered with N-terminal His6 tags rather than C-terminal tags, as the C-terminal domains of mFH and FHR proteins contain putative C3b/C3d and GAG binding sites. Recombinant proteins were expressed by transient transfection of mammalian 293-F cells. This system was chosen to produce proteins with glycosylation patterns closer to native protein expression patterns, which may have implications for proper protein folding and function. Additionally, we sought to produce proteins that did not require deglycosylation to produce a uniform protein population and that could be used in functional studies and in vivo disease models. Lastly, 293-F cell expression provided a serum-free method for expressing and purifying mFHR proteins, while preventing cross-reactivity and/or contamination from complement components and other proteins that are present in serum-based expression media.
Recombinant proteins were purified from supernatant using a combination of immobilized metal affinity chromatography and gel filtration methods. Analytical SDS-PAGE was used to examine protein purity under reducing conditions (Fig. 2A, 2B). Although bands for glycosylated mFH, mFHR-A, and mFHR-C are observed near their predicted molecular weights, mFHR-B appears as a band at ∼55 kDa. Therefore, deglycosylation of the proteins using PNGase F was performed. Following deglycosylation of the proteins, PNGase F was removed from each sample before performing SDS-PAGE. When deglycosylated, all proteins run at their predicted molecular masses: mFH, ∼140 kDa; mFHR-A, ∼55 kDa; mFHR-B, ∼35–40 kDa; and mFHR-C, ∼110 kDa. In a separate experiment, PNGase F, which has a molecular mass of ∼36 kDa, was not removed after deglycosylation of mFHR-A and mFHR-B and can be seen as a band running beneath both mFHR-A and mFHR-B (Fig. 2B). Differential scanning fluorimetry (DSF) was used to assess the quality of the folded proteins and analyze their stability in PBS. This technique proved to be an effective method for analyzing the quality of batches of protein produced from different transfections. DSF also allowed for a way to examine protein stability in different buffer conditions. Normalized results from DSF show that mFH and mFHR proteins exhibit denaturation curves with Tm of ∼60–65°C in PBS, suggesting that the recombinant proteins are properly folded (Fig. 2C).
mFHR-A and mFHR-B are potent antagonists of mFH on sheep erythrocytes
Previous research has shown that the C-terminal end of human FH plays a critical role in the regulation of complement on cell surfaces (15). The C-terminal SCR domains of FH (SCRs 19–20) can discriminate between host and nonhost surfaces. In sheep erythrocyte hemolytic assays, lysis of normal sheep RBCs occurs in the presence of the recombinant FH19–20 inhibitor whereas fluid phase complement activation remains unaffected. Both FH and the FH19–20 inhibitor have ∼7-fold higher affinity for surfaces coated with C3b and high amounts of sialic acid (15). Additionally, competition assays showed that FH19–20 inhibits FH-mediated control of AP-driven complement lysis on host-like polyanion-bearing cells.
To investigate the effects that mFHR proteins exhibit on FH protection of host-like surfaces, hemolytic assays using normal sheep RBCs were performed. As a positive control for deregulation of FH function, mFH19–20 was designed and expressed using methods described. Normal sheep erythrocytes were incubated for 30 min at 37°C with complement-sufficient wild-type mouse serum (40%) and increasing concentrations of mFHR-A, mFHR-B, mFHR-C, mFH, and mFH19–20 (Fig. 3). In agreement with previous studies we found that mFH19–20 activates complement by inhibiting FH protection of sheep RBCs (15, 17). Interestingly, both mFHR-A and mFHR-B appear to be more potent antagonists of mFH in this assay (Fig. 3A). Thus, complete hemolysis was observed with addition of ∼4 μM mFHR-A, whereas 2-fold higher concentrations of mFHR-B were required to induce just >50% lysis of erythrocytes. In contrast, 5 μM mFH19–20 inhibited mFH cell-surface protection of the erythrocytes by only 25%. Interestingly, mFHR-C did not antagonize mFH function on these cell surfaces. Even in the presence of high concentrations of mFHR-C (20 μM), no erythrocyte hemolysis was observed (Fig. 3B). Furthermore, competition assays were performed using mFH to demonstrate that incubation of sheep RBCs with mFH prior to addition of mFHR-A or mFHR-B prevents hemolysis of cells (Fig. 3C).
mFH and mFHR proteins bind mouse C3d, and mutant FHR proteins are inactive
To explore the molecular mechanism underlying the different hemolytic activities observed with the mFHR proteins, we next investigated how mFH and mFHRs engage the complement component, C3d, an analog of the thioester-containing domain in C3b. Complement regulation by FH requires binding to C3b/C3d, and we hypothesized that differential binding by the mFHRs to C3b/C3d may modulate AP activity. Previous work with mFHR-B produced in Pichia pastoris provided the first evidence that mFHRs interact with human C3b (10). However, differences between human and mouse C3b exist; therefore, recombinant mouse C3d was produced. Human C3d produced in our laboratory was also available in these studies for comparison. Recombinant production of human and mouse C3d using Escherichia coli was performed followed by purification of the proteins using a GST column and size exclusion chromatography. SDS-PAGE separation showed a product at ∼35 kDa, which is the predicted molecular mass for mC3d (Supplemental Fig. 1A). Analysis of the mC3d secondary structure was performed using CD. The CD spectrum revealed an α helical confirmation for mC3d, a result that is in agreement with the x-ray crystallography structure of human C3d (Supplemental Fig. 1B) (18). Additionally, thermal shift assays for mC3d and hC3d were performed, and the apparent Tm for mC3d (>40°C) was estimated to be slightly higher than that for hC3d (<40°C) (Supplemental Fig. 1C).
To investigate mC3d binding to mFH and mFHRs, ELISA assays were performed. Microtiter plates were coated with C3d, and binding of mFH, mFHR-A, and mFHR-B proteins was observed (Fig. 4B). Cross-species reactivity between human C3d and mFH and mFHR proteins was also observed (Supplemental Fig. 2). To further explore this interaction, mutant constructs were made. Four key residues within human FH SCR 19 and one residue within SCR 20 have side chains that form intermolecular hydrogen bonds with C3d. All five of these key residues are conserved within the mFH and mFHR family (Fig. 4A). Several other residues that are also important for the human FH19–20:C3d interface are indicated. To test this interaction, mutations were made to residues corresponding to human FH residues N1117 (N1117A) and D1119 (D1119A) to produce mutant mFHR-A and mutant mFHR-B proteins. No binding was observed between mC3d and mutant mFHR-A or mutant mFHR-B by ELISA (Fig. 4B).
Binding affinities of mFHR proteins and mC3d directly correlate with their functional activities
Earlier studies have analyzed the interaction between mFH and C3b using SPR. The binding of full-length mFH and truncated versions of mFH, including mFH SCRs 1–5 and mFH SCRs 18–20, to both mouse and human C3b has been demonstrated (19). However, the apparent KD values for this interaction have not been determined, and the interaction between mFH (and the mFHRs) and mC3d has not been evaluated.
Using previously described conditions for SPR experiments with mFH and C3b, binding of mFH and mFHR proteins to mC3d was investigated. We used standard amine coupling of ∼2500 response units of mC3d onto a CM5 sensor chip. Serial dilutions of mFH, mFHR-A, mFHR-B, mFHR-C, and mFH19–20 proteins were made, and sensorgrams generated from the different dilutions were analyzed using BIA evaluation software. Data were fitted using a two-state reaction model, and apparent KD values for each of the proteins were calculated and compared.
Apparent binding affinities varied among the different proteins. The highest binding affinities were observed for mFHR-A (KD of 136 nM) and mFHR-B (KD of 546 nM). These affinities were ∼8-fold and ∼2-fold greater than the apparent binding affinity of mFHR-C at KD of 1.04 μM. The weakest binding affinities were observed for mFH (KD of 3.85 μM) and the recombinant mFH19–20 inhibitor (KD of 22 μM) (Fig. 5A, 5D). No binding interaction was observed for mutant mFHR-A and mFHR-B proteins (containing the N1117A and D1119A pair mutation). Whereas binding of mFH to mouse C3b has been previously analyzed (19), affinity values were not calculated. However, two different dissociation constants have been calculated using SPR for the interaction between human C3d and FH19–20. Kajander et al. (20) calculated a KD value of 0.18 μM for the C3d:FH19–20 interaction whereas Morgan et al. (13) determined a KD value between 6.2 and 8.2 μM (13, 20).
Mutant mFHR-B exhibits loss of function in hemolytic assays
To assess the function of the mutant mFHR-B protein compared with wild-type protein, a sheep erythrocyte hemolytic assay was performed using previously described methods and conditions. Results show that mutation of two out of five major putative binding residues (corresponding to human FH residues N1117A and D1119A) that form intermolecular hydrogen bonds with C3d sufficiently abrogates the ability of mFHR-B to inhibit mFH resulting in diminished hemolysis of erythrocytes (Fig. 6). At a concentration of 2 μM mFHR-B, ∼20% hemolysis of sheep erythrocytes is observed, whereas at 8 μM mFHR-B, >60% lysis occurs. We observe negligible hemolysis at mutant mFHR-B concentrations of 2, 4, and 8 μM. These results are in good agreement with our hypothesis that the AP is modulated by mFHR proteins through their direct engagement and competition with mFH for binding to C3d and the TED domain.
mFHRs activate complement on different cell surfaces
We also explored whether mFHR proteins can impair mFH function on different cell surfaces. Murine renal TECs have been shown to express the membrane-bound regulator Crry on their basolateral surface whereas FH regulates complement activation on their apical surface (21). TECs are targets of complement-mediated damage in vivo due to dysregulation and excessive activation of the AP (22). Complement activation in vitro occurs during the incubation of cells with 10% wild-type mouse serum and can be monitored by evaluating C3b deposition on the surface of TECs using flow cytometry. In a separate study, a FH binding partner, annexin 2, impaired complement regulation by FH and increased complement activation on TEC cells. The effect of annexin 2 on C3b deposition was reversed by the addition of recombinant FH, but not addition of the inhibitor FH19–20 (17). In this study, we probed the ability of murine FHR proteins to act as antagonists of FH function on TEC cells by incubating the cells with mFHR-A or mFHR-B proteins in the presence of 10% NMS. As anticipated, both mFHR-A and mFHR-B act as potent antagonists of mFH, increasing complement activation on TEC surfaces as indicated by the increase in C3b deposition (Fig. 7).
Given the association with FHR proteins and AMD, the ability of the murine FHRs to enhance complement activation on a human retinal pigment epithelial cell line (ARPE-19) was explored. Research by Thurman et al. (16) has described a mechanism by which immortalized ARPE-19 cells regulate complement activation on their surfaces. Using similar assays as described in the TEC experiments, they report that treatment of ARPE-19 cells with H2O2 to induce oxidative cell stress resulted in decreased surface expression of CD55 and CD59 and impaired FH cell-surface control. In this study, we used flow cytometry to examine the degree of C3b deposition on both unstressed and stressed ARPE-19 cells incubated with either mFHR-A or mFHR-B in the presence of 10% NMS. Addition of mFHR-A or mFHR-B increased complement activation on both types of cells. The ability of mFHR-A and mFHR-B to increase complement activation on healthy cells (unstressed) further supports the idea that mFHR-A and mFHR-B are potent antagonists of mFH on different nucleated cell surfaces (Fig. 8).
mFH and mFHR Ab production
Armenian hamsters were obtained from Cytogen hamsters (Cytogen Research and Development, West Roxbury, MA). Hamsters were prebled to obtain a stock of serum prior to immunization. On day 0, hamsters were immunized by s.c. injection of 100 μg of protein in CFA. On day 14, hamsters were boosted with 50 μg of mFH/mFHR protein in IFA. On day 21, hamsters were bled and serum was tested by ELISA. Once a reasonable Ab titer was determined or after day 35, hamsters were given an i.p. injection with 50 μg of protein in PBS. On day 38 or later, the spleen was harvested and fused with mouse myeloma SP2/0 cells, and hybridomas were screened using recombinant mFH and mFHR proteins per Kappler–Marrack protocols (https://www.nationaljewish.org/research-science/programs-depts/biomedical-research/labs/kappler-marrack-research-lab/protocols).
The complement system is a major mediator of immune surveillance and homeostasis. At present, there is an increased focus on understanding the regulation of the different pathways, especially the AP, which is the primary pathway by which the system damages cells and tissues. The spontaneous hydrolysis of C3 leading to generation of C3b must be well controlled to prevent inappropriate and potentially rapid AP complement activation. FH is a major soluble complement regulator that is essential for controlling AP activation in plasma and on cell surfaces through the engagement of polyanionic surface markers, such as sialic acid, and C3 fragments, such as C3b and C3d. Mutations or deletions in FH, particularly within the C-terminal region, are associated with a number of different autoimmune and inflammatory illnesses. Moreover, pathogens and altered-host cells, such as cancer cells, exploit the protective function of FH by recruiting FH to their surfaces as part of an immune evasion/modulation strategy (23–25).
Although a significant amount is known about the structure and function of FH, the functional roles of a group of structurally similar molecules, designated the FHR proteins, are less understood. In this study, we present a more extensive characterization of the mFHR-B and mFHR-C proteins in addition to evaluating a third putative FHR protein, mFHR-A, whose sequence is derived from the original characterization of class A mFHR transcripts. Following expression and purification of proteins using a 293-F mammalian expression system, functional assays were performed to understand how mFHR proteins modulate complement activation by competing with mFH for C3b/C3d on self-surfaces.
To assess the functional activity of recombinant mFH and mFHR proteins, different hemolytic assays were performed. Previous work has demonstrated that the last two SCR domains of FH (SCRs 19 and 20) are critical for FH-mediated protection of cell surfaces. To address whether the mFHR proteins deregulate mFH cell-surface protection, functional assays using mFH19–20 inhibitor, mFHR-A, mFHR-B, and mFHR-C were performed. In these hemolytic assays, normal sheep erythrocytes are added to complement sufficient wild-type mouse serum, but hemolysis of the cells is prevented by mFH that is present in the serum. Reduction of mFH from the surface of these cells results in AP activation and subsequent hemolysis. In this study, we were able to demonstrate that addition of mFHR-A and mFHR-B in the presence of NMS induced hemolysis of sheep erythrocytes. Both mFHR-A and mFHR-B act as potent antagonists of mFH on sheep erythrocyte cell surfaces. Although the mFH19–20 inhibitor and mFHR-B induced ∼50% hemolysis at ∼7 μM concentration, mFHR-A is more potent and induced 100% hemolysis at ∼2 μM. We also demonstrate that hemolysis can be prevented by addition of exogenous recombinant mFH to sheep erythrocytes prior to addition of mFHRs. Interestingly, mFHR-C does not appear to increase hemolysis of sheep erythrocytes but may exert effects on specific cell surfaces, which has been suggested by other studies of mFHRs in different mouse models of autoimmune disease (26). Taken together, the results from hemolytic assays show that mFHR proteins are able to perturb the overall function of the AP. Furthermore, our results suggest that mFHRs may have different biological functions and/or recognize different surface ligands.
Note that mFHR-A is a prototype for a protein encoded by the mFHR-A gene. An mRNA sequence isolated from a normal 5-mo-old mouse liver (BC026782) shares high homology to this prototype and can be found in the UCSC Genome Database; however, evidence for an mFHR-A protein does not currently exist. Most interestingly is that the mFHR-A prototype, which shares >95% identity to mFHR-B SCRs 5–7 and 19–20, is a more potent inhibitor of FH than mFH19–20. The mFHR-A prototype contains two additional SCR domains corresponding to mFH SCRs 8–9, which may enable the protein to more efficiently compete with FH for binding to C3b on cell surfaces. The additional SCR domains within the mFHR-A and mFHR-B proteins may help them to engage GAG surfaces better than mFH19–20, thus making them more potent antagonists.
Additionally, our results with both mFHR-A and mFHR-B agree with functional studies of the human FHR proteins. Hemolytic assays with guinea pig erythrocytes show that both FHR-1 and FHR-5 inhibit FH function by competitively blocking the interaction between FH and C3b (11). Structural studies have established that this interaction occurs through FH SCRs 19–20 and the reactive TED located on the C3d part of C3b. Five residues (Asn1117, Asp1119, Gln1139, Tyr1142, and Tyr1190) within SCRs 19–20 form intermolecular hydrogen bonds with C3d. Sequence alignment of human FH and the mFH family shows that these residues are conserved between the two species. Therefore, we predicted that mFH and mFHRs would bind C3d, and mutation of conserved residues within the binding interface would inactivate our proteins.
Apparent dissociation values for FHR-3 and FHR-4 were calculated as 87 and 260 nM, respectively (27). They also demonstrated that FHR-3 and FHR-4 have similar binding affinities with C3b and that the reverse analysis, with FHRs immobilized on a chip and C3d/C3b in fluid phase, produced similar results. One caveat is that the kinetics of the human proteins were analyzed using a 1:1 binding model whereas the mouse proteins were fitted using a two-state reaction model. Taking this into consideration, the binding kinetics of mFHR-A and mFHR-B could also be compared with SPR results from the binding interaction observed between C3b and mutant human FHR-51212–9 protein (11, 28).
This mutant, found in patients with C3 glomerulopathy, has a duplicated dimerization domain (from duplication of SCRs 1–2). Binding affinities for wild-type mFHR-5 and mFH to amine couple C3b were determined to be 1.7 and 6.6 μM. Although a binding affinity for the FHR-51212–9 mutant was not determined, functional analysis of this mutant using hemolytic assays also showed greater hemolysis than did wild-type controls. This result was attributed to the mutant’s increased ability to form higher order complexes and compete with FH for binding to C3b. Gel filtration chromatography (data not shown) was used to analyze the native mFHR-B, and results suggest that mFHR-B may be capable of forming higher order complexes. Given that the fit of both the mFHR-A and mFHR-B curves resembles the curve for FHR-51212–9, additional analysis of these proteins for the formation of higher order complexes than dimers is necessary.
To further analyze the functional homology between the human and mouse C3d binding site, mutation of the putative C3d binding site on mFHR-A and mFHR-B was performed. Results show that mutation of two residues (corresponding to human FH residues N1117 and Asp1119) was sufficient to disrupt mC3d binding of mFHR-A and mFHR-B in both ELISA and SPR assays. Additionally, mutant mFHR-B exhibited significant loss of function in a sheep hemolytic assay further, suggesting that mFHRs compete with mFH for the same target (C3b/C3d) on cell surfaces. The direct correlation between mFHR binding affinity for C3d and functional activity in hemolytic assays further establishes how these proteins modulate complement through competition with mFH for C3b/C3d on cell surfaces.
Additionally, the important role of other cell surface markers must also be addressed. The sheep hemolytic assay is dependent on FH recognition of C3b/C3d as well as sialic acid residues that are present on the cell surface. The ability of FH to engage C3b increases 10-fold in the presence of sialic acid residues, and the structural basis for FH recognition of sialic acid on cell surfaces has been established (25). A recent study has proposed that FH simultaneously engages sialic acid and C3b to regulate complement activation (29). Mutation of sialic acid binding residues on FH19–20 resulted in the loss of its ability to antagonize FH. This suggests that the behavior of FH, and perhaps the ability of FHRs to antagonize FH function, is dependent on both C3b/C3d and surface sialic acid cell surface content. Implicit in this theory is that cells that possess low amounts of surface sialic acids (such as rabbit erythrocytes) are not subject to this same mechanism.
Understanding the behavior of FHRs on different cell surfaces is critical for clarifying their associations with disease and explaining the different disease phenotypes. In this study, we also examined complement activation by mFHRs on two relevant cell lines. Our results showed increased C3b deposition on both murine tubular epithelial cells as well as normal human retinal pigment epithelial cells occurs when cells are incubated with NMS with addition of mFHR-A or mFHR-B. With both cell types, mFHR-A appears to generate greater complement activation than does mFHR-B, which supports our other binding and functional data. Additionally, complement activation by mFHRs on a human cell line and demonstration that mFHRs can bind human C3d provide evidence that the murine proteins can serve as orthologs of human FHRs in animal studies. Taken together, these data provide important support for further evaluation of the roles of mFHRs in vivo.
Whereas it is fairly straightforward to compare the sequence homology of C3b/C3d and GAG-binding sites between the different mFH and human FH protein families, it is much more challenging to make direct comparisons. That is, does mFHR-B function more like human FHR-3 and FHR-4 or does it behave like FHR-5? For direct comparisons such as this, better tools are required. Development of Abs that recognize specific mFHR proteins is important for elucidating the different biological roles that these proteins may have and is currently being investigated. We have generated hamster anti-mouse monoclonal and polyclonal anti-FH and anti-FHR Abs and observe bands that replicate the original data by Hellwage et al. (10). Although we attempted to generate mAbs against recombinant mFHR-A and mFHR-B, we were not able to identify Abs specific for mFHR-A and mFHR-B in wild-type or FH−/− mouse serum (Supplemental Fig. 3).
Finally, our results demonstrated that mFHR-C does not inhibit FH function in a sheep erythrocyte assay. These results suggest the possibility that mFHR-C has other important context-dependent biological functions. Compared to the mFHR-A and mFHR-B proteins, mFHR-C contains twice as many SCR domains (14 SCRs), but is 6 SCR domains shorter than mFH. Notably, mFHR-C (as well as all other FHR proteins) lacks sequence homology to FH SCRs 1–4, which are regulatory domains. Perhaps mFHR-C has a different role in cell recognition and complement regulation? One could speculate that mFHR-C may have context-dependent functional roles, such as in the kidney. Perhaps mFHR-C can destabilize C3 convertases on certain cell surfaces? Or maybe mFHR-C has fluid phase regulatory abilities and acts as a cofactor for FI? These results suggest that we must evaluate the mFHR proteins in a variety of different contexts to fully understand how they modulate complement. We must examine how FHRs regulate complement in fluid phase versus on different cell surfaces, taking into consideration that these proteins may function differently in the context of an autoimmune disease model versus a cancer or infectious disease animal model. Understanding the differences between mFHR-A/mFHR-B and mFHR-C may offer insight into how FHR proteins exert influence over complement regulation, which could allow for strategic treatment against different diseases.
In summary, these results demonstrate that murine FHR proteins modulate complement activation by antagonizing FH function on cell surfaces. Interestingly, we also show that neighboring genes, mFHR-B and mFHR-C, encode FHR proteins that have unique functions. Although mFHR-B is a potent inhibitor of mFH, mFHR-C does not appear to have inhibitory functions, which may be context-dependent. Similar to human FHR-1 and FHR-5, mFHR-A or mFHR-B compete with mFH for binding to C3b/C3d on cell surfaces and have similar C3d binding affinities to FHR-3 and FHR-4. We identified that the binding interaction between C3d and mFHR proteins is conserved between human and murine systems. Furthermore, we demonstrate that mFHR binding affinities for C3d directly correlate with their functional activity in hemolytic assays. These results will be helpful for guiding future work in different animal models of disease, such as atypical hemolytic uremic syndrome and AMD, which could potentially lead to the development of novel therapeutics.
We acknowledge Shaun Bevers for advice on CD experiments, which were carried out at the Biophysics Core Facility, which is part of the Structural Biology and Biochemistry Department at the University of Colorado Anschutz Medical Campus. We also thank Lori Sherman and Michelle Randolph for assistance with protein production.
This work was supported in part by National Institutes of Health Grant R01 AR-51749, an investigator grant from the Beckman Institute for Macular Research (to V.M.H.), University of Pennsylvania Health System Grant AI-18785 (to P.M.), the Victor W. Bolie and Earleen D. Bolie Graduate Scholarship Fund (to A.H.A.), and the University of Colorado Cancer Core Protein Production Shared Resource, which receives direct funding support from the National Cancer Institute through Cancer Center Support Grant P30CA046934.
The online version of this article contains supplemental material.
Abbreviations used in this article:
age-related macular degeneration
differential scanning fluorimetry
normal mouse serum
short consensus repeat
surface plasmon resonance
tubular epithelial cell
J.M.T. and V.M.H. receive royalties from Alexion Pharmaceuticals. K.J.M. receives funding from Gemini Therapeutics. The other authors have no financial conflicts of interest.