Complement Regulator FHR-3 Is Elevated either Locally or Systemically in a Selection of Autoimmune Diseases

The human complement factor H-related protein-3 (FHR-3) is a soluble regulator of the complement system. Homozygous cfhr3/1 deletion is a genetic risk factor for the autoimmune form of atypical hemolytic-uremic syndrome (aHUS), while also found to be protective in age-related macular degeneration (AMD). The precise function of FHR-3 remains to be fully characterized. We generated four mouse monoclonal antibodies (mAbs) for FHR-3 (RETC) without cross-reactivity to the complement factor H (FH)-family. These antibodies detected FHR-3 from human serum with a mean concentration of 1 μg/mL. FHR-3 levels in patients were significantly increased in sera from systemic lupus erythematosus, rheumatoid arthritis, and polymyalgia rheumatica but remained almost unchanged in samples from AMD or aHUS patients. Moreover, by immunostaining of an aged human donor retina, we discovered a local FHR-3 production by microglia/macrophages. The mAb RETC-2 modulated FHR-3 binding to C3b but not the binding of FHR-3 to heparin. Interestingly, FHR-3 competed with FH for binding C3b and the mAb RETC-2 reduced the interaction of FHR-3 and C3b, resulting in increased FH binding. Our results unveil a previously unknown systemic involvement of FHR-3 in rheumatoid diseases and a putative local role of FHR-3 mediated by microglia/macrophages in the damaged retina. We conclude that the local FHR-3/FH equilibrium in AMD is a potential therapeutic target, which can be modulated by our specific mAb RETC-2.

in rheumatoid patient samples. Furthermore, we identified local production of FHR-3 by microglia/macrophages in an aged donor retina with RPE atrophy -the latter being a typical hallmark of dry AMD. Additionally, we demonstrate that FHR-3 mAb RETC-2 reduced binding of FHR-3 to C3b reinforcing the local binding of the FHR-3 competing complement inhibitor FH. Thus, FHR-3-targeting therapeutics may offer an innovative strategy for local immune therapies for AMD and other complement-related diseases.

MaTerials anD MeThODs human Material, animals, and ethical statements
Collection of human blood and eye samples were approved by the local ethics committees (serum: University of Regensburg and Friedrich Schiller University Jena, Germany; eyes: University of Bern, Switzerland) and were obtained in accordance with the Declaration of Helsinki. Complement-depleted human sera were purchased from Complement Technology (Tyler, TX, USA). All human serum samples were stored at −80°C. Balb/c and C57/BL6 mice were obtained from Charles River Laboratories (Wilmington, MA, USA). Mice experiments were strictly performed according to the guidelines of replacement, refinement, and reduction of animals in research (20) and approved by the committee on the ethics of animal experiments of the regional agency for animal health Regierung der Oberpfalz, Veterinärwesen (TVA 54-2532.4-05/13).
The goat polyclonal antihuman FH antibody was obtained from Quidel (cat. A312, San Diego, CA, USA), the mouse mAb anti-FH was a kind gift of S. Berra (Università degli Studi di Milano, Milano, Italy) (21), and mouse mAb anti-FHR-3.1 and anti-FHR-3.4 were described previously (6). Rabbit anti-Iba1 polyclonal antibody was ordered from Wako Chemicals (cat. 019-19741, Neuss, Germany), and mouse anti-Glutamine Synthetase mAb, clone GS-6, was from Calbiochem/Merck Millipore (cat. MAB302, Darmstadt, Germany). StrepMAB-classic (cat. 2-1507-001) and StrepMAB-classic conjugated to horseradish peroxidase Our results unveil a previously unknown systemic involvement of FHR-3 in rheumatoid diseases and a putative local role of FHR-3 mediated by microglia/macrophages in the damaged retina. We conclude that the local FHR-3/FH equilibrium in AMD is a potential therapeutic target, which can be modulated by our specific mAb RETC-2.
The immunogenic peptide within SCR5 (Charité Universitätsmedizin Berlin, Institute of Biochemistry, Laboratory Proteolytic Systems, Berlin, Germany) was conjugated either to BSA for mouse immunization or to keyhole limpet hemocyanin for ELISA screening of positives clones.

Determination of antibody Variable regions
RNA of hybridoma cell lines was isolated using RNeasy Mini Kit (Qiagen, Hamburg, Germany) and subsequent synthesis of cDNA using the Quantitect reverse Transcription Kit (Qiagen, Hamburg, Germany). PCR for amplification of IgG variable regions (heavy and light chain) was performed using the mouse IgG Library Primer set (Progen, Heidelberg, Germany), according to the manufacturer's protocol and DNA-fragments were ligated into pGEM-T Easy plasmid (Promega, Mannheim, Germany) for E. coli XL-1 Blue competent cells (Agilent Technologies, Boeblingen, Germany) transformation. Following plasmid isolation (Plasmid Midi Kit, Qiagen, Hamburg, Germany), DNA-sequencing was performed by GeneArt (Thermo Fisher Scientific, Braunschweig, Germany) with pGEM-T specific primer M13 (Promega, Mannheim, Germany).
gel and non-gel lc-Ms/Ms analyses Samples were separated by SDS-PAGE and stained with Coomassie. Afterward, all visible bands were excised and subjected to in-gel-digestion as published previously (27). Resulting peptides were used for nano-LC-MS/MS analysis on a TripleTOF 5600+ mass spectrometer (Sciex, Darmstadt, Germany) as published previously (24). Database searches were accomplished using the ProteinPilot 4.5 software using the Uniprot-database (version 05/2014).
Samples for non-gel LC-MS/MS were diluted in 50 mM ammoniumcarbonate, then reduced with 100 mM dithiothreitol for 30 min at 60°C, and alkylated with 300 mM iodoacetamide at room temperature. Trypsin-mediated proteolysis was performed using a modified Filter-Aided Sample Preparation protocol as published previously (28). Resulting peptides were used for analysis on a LTQ-Orbitrap XL (Thermo Fisher Scientific, Braunschweig, Germany) connected with an Ultimate 3000 nano-HPLC system (Thermo Fisher Scientific, Braunschweig, Germany) as described previously (29). Peptides were identified and quantified using the Progenesis QI software (Non-linear, Waters) and the Mascot search algorithm with the Ensembl Human public database (29,30).

software and statistical analyses
Immunogenic and unique peptides of SCR5 were determined by AbDesigner (36). Sequence analyses of mAb RETC-2 and mAb RETC-3 variable regions were performed with VBASE2 (37) and Rosetta online (38) server. The visualization of antibody structures and complementarity-determining regions (CDR) was realized with Chimera (39). Data were statistically analyzed using GraphPad Prism 5 (GraphPad Software, San Diego, CA, USA). resUlTs novel Mouse mabs specifically interact with human Fhr-3 We generated four different mouse hybridoma cell lines against native human FHR-3 using a peptide immunization strategy. The isolated and purified four mAbs RETC-2, RETC-3, RETC-5,  and RETC-7 were specific for the C-terminal, fifth SCR (SCR5) domain of FHR-3. These antibodies were of different mouse isotypes. MAb RETC-3 and mAb RETC-7 were mouse IgG1 antibodies. MAb RETC-2 and mAb RETC-5 were mouse IgG2b isotypes ( Table 1). κ-light chains complemented the FHR-3 binding structures of the variable heavy chain regions. Sequence analysis and in silico simulations revealed the specific, variable paratope structure (Fab), including the CDRs of mAb RETC-2 and mAb RETC-3 ( Figure S1 in Supplementary Material).
Mouse Ig interacts with their variable Fab-part with the corresponding target epitope (Figure 1), and their constant Fc-part is recognized by Fc-receptors or complement components (41,42). We used a combination of gel-based ( Figure 2B) and non-gel More than 30 different proteins were immunoprecipitated from non-reduced NHS using RETC-2 and isotype control (not shown). Labeled fractions after Coomassie staining were identified by mass spectrometry analyses and contained FHR-3. Unlabeled protein bands contained other serum proteins (e.g., serum albumin, apolipoproteins, Igs). (c) RETC-2 specifically precipitated FHR-3 (green star) in comparison to an isotype control mAb. Additional peptide fragments, characterized with non-gel LC-MS/MS, were complement proteins.  LC-MS/MS ( Figure 2C) protein identifications to dissect the interactions into Fab-part (anti-FHR-3 specific) and Fc-part (general mouse Ig isotype specific) mechanisms. Therefore, we compared the precipitated protein patterns from native human serum using either anti-FHR-3 mAb RETC-2 or the corresponding isotype Ig (Figures 2B,C; Table S1 in Supplementary Material). Fab-specific FHR-3 counts were exclusively identified in immunoprecipitations using the novel anti-FHR-3 mAb at different protein sizes (Figures 2B,C), but not with the isotype controls ( Figure 2C; Table S1 in Supplementary Material). In contrast, FH and FHR-5 bound to all mouse IgG2b whereas FHR-1, FHR-2, or FHR-4 did not interact with the IgGs (Table S1 in supplementary material). Given the fact that no other FH-family members were consistently detected in combination with FHR-3, the data suggested that FHR-3 circulates in vivo as a monomer or as a homo-multimer.
The constant, non-specific Fc-antibody parts of both the anti-FHR-3 mAb and the isotype control mAb precipitated additional serum proteins (Table S1 in supplementary material). Separation of mAb RETC-2 precipitated proteins by SDS-PAGE with subsequent Coomassie staining revealed protein bands with 20 different mobilities ( Figure 2B) and 10 bands with isotype control (data not shown). On average, 30 different proteins were identified as Fc-and/or immunoprecipitation-beads associated interaction partners for both the FHR-3 specific and the corresponding isotype control antibodies. These proteins included abundant proteins, such as keratin, apolipoproteins, Igs, and serum albumin. Our analysis focused on complement-associated proteins. More than 20 human complement proteins of all three complement activation pathways bound to the murine Fc-antibody parts IgG1 or IgG2b, irrespective of the antibodies specificity (Fab-part) ( Figure 2C; Table S1 in Supplementary  Material). The human complement proteins C3 and C5 were detected with the highest spectral counts as binding partners for all mouse Fc-parts ( Figure 2C). MAbs RETC-2, RETC-3 as well as the corresponding isotype controls precipitated also proteins of the terminal pathway (C6-C9) and important soluble regulators (clusterin, vitronectin, properdin) ( Table S1 in Supplementary Material). In addition, characteristic components of the classical pathway (C1, C2, C4), the lectin pathway (MASP1, ficolin-2), and the alternative pathway (CFB) were associated with mouse IgG2b but not with mouse IgG1 Fc-parts antibodies (Table S1 in Supplementary Material).
In conclusion, these results showed that the specific FHR-3antibodies detected FHR-3 from human serum. However, we confirmed a general interaction of human complement proteins with the mouse Fc-antibody part. We concluded that mAb RETC-2 will need to be humanized in the future for further functional interaction studies involving this mAb and human serum containing all complement proteins.

Quantified Fhr-3 levels Varied in rheumatic Diseases and sle
An immunoassay for FHR-3 quantification from human samples was established, using mAb RETC-2 antibody as capture antibody and biotin-labeled anti-FHR-3.4 mAb (which reacts with FH and FHR-3) as detection antibody (6) (Figures 3A,D).
We performed chromosomal DNA analysis of leukocytes from healthy blood donors to determine positive and negative controls for the immunoassays. The positive samples (P1, P2) resulted in one specific 1.5 kb-fragment for the N-terminal cfhr3-gene region (SCR1/2), and negative samples (N1, N2) were homozygous for the cfhr3 deletion ( Figure 3C). FHR-3 serum levels from healthy positive (P1, P2) and negative (N1-N3) controls were consistent with the respective genotype and were detected without any false-negative or false-positive signals in the immunoassay (Figures 3A,C). The ELISA and genotyping results (Figures 3A,C) correlated with the immunoblot detection, showing protein bands at 50-55 kDa for FHR-3 positive serum samples and no signals for the FHR-3-deficient sera (Figure 3B).
Using this specific immunoassay, we aimed to determine FHR-3 levels in different standardized serum samples, using complement-deficient sera exemplarily (Figure 3A). Tested serum samples varied in their FHR-3 levels between 0.63 μg/mL for C3-depleted (C3 dpl ) and 1.69 μg/mL for C5 dpl sera ( Figure 3A). FHR-3 was not detected in FH dpl , IgG dpl , or normal mouse serum (Figures 3A,B). This is explained by heparin-based immunodepletion of FH from the serum that also removed heparin-affine FHR-3.
Repetitive FHR-3 quantification using positive control and complement-depleted sera revealed an interassay variation of the sandwich ELISA of 12-23% ( Figure 3A).
We then investigated whether FHR-3 levels were systemically changed in complement-associated diseases (Figures 3D and 4; Table S2 in Supplementary Material). Serum samples from 21 healthy, young (mean age 26) volunteers showed a FHR-3 concentration of 0.41-2.49 μg/mL with a mean concentration of 1.06 + 0.53 μg/mL ( Figure 3D; Table S2 in Supplementary Material), whereas the patient samples showed a FHR-3 concentration range of 0.31-12.29 μg/mL. On average, FHR-3  concentrations of patients with aHUS (mean 1.6 μg/mL), choroidal neovascularization (CNV, mean 1.83 μg/mL), systemic sclerosis (SSc, mean 1.92 μg/mL), and connective tissue diseases (CTD, mean 1.76 μg/mL) were slightly increased ( Figure 3D; Table S2 in Supplementary Material). However, serum samples from non-steroid-treated SLE, rheumatoid arthritis (RA), and polymyalgia rheumatica (PR) patients were associated with potentially biological relevant, significantly increased FHR-3 serum levels (Figures 3D and 4; Table S2 in Supplementary Material). In our small cohort, the FHR-3 concentrations were threefold to fourfold higher in non-steroid-treated SLE (mean 4.14 μg/mL), RA (mean 3.12 μg/mL), and PR (mean 3.37 μg/mL) patients in comparison to controls (Figures 3D and  4B; Table S2 in Supplementary Material). Steroid treatment of SLE patients decreased the FHR-3 amounts in serum by 48% ( Figure 4B). A correlation of FHR-3 levels and clinical parameters revealed a putative relation of FHR-3 levels and lupus nephritis diagnose or blood sedimentation rate (BSR) in SLE patients, respectively (Figures 4A,C). Increased C-reactive protein (CRP) levels in RA patients corresponded to higher FHR-3 concentrations (Table S3 in Supplementary Material). FHR-3-deficient serum samples were mainly found in aHUS (38%), non-steroid treated SLE (17%), and in control (14%) cohorts (Table S2 in Supplementary Material). FHR-3 protein analyses in the aHUS group were confirmed by genetic analysis of all samples using multiplex ligation-dependent probe amplification (data not shown). In summary, these experiments showed that FHR-3 levels in serum were altered in complement-associated inflammatory diseases. This suggested that systemic FHR-3 concentrations were associated with complement activity and inflammation.

Fhr-3 is localized in Microglia/ Macrophages and Fh in Müller cells in the retina
Previous studies revealed that FHR-3 deficiency is protective of AMD but a risk factor for the development of aHUS (9, 10, 13). We detected no critical differences in systemic FHR-3 protein concentrations of healthy donors compared with AMD or aHUS patients ( Figure 3D). Therefore, a local effect of FHR-3 could be of relevance. We stained FHR-3 in a retina from a 92-year-old donor (Figure 5), which showed an increased number of invading macrophages (or activated microglia cells; Figures 5A,E) in comparison to a retina from a 64-year-old donor (Figures 5D,F). Surprisingly, we identified FHR-3 in activated microglia/macrophages in the central retina of the 92-year-old donor using mAb RETC-2 (Figure 5A, arrows; Figures 5B,C, magnified). FHR-3 was not detected in resting microglia/macrophage in the periphery of the younger retina ( Figure 5D). Fluorescent signals in the RPE (Figures 5A,D,F) were identified as unspecific autofluorescence ( Figure S2 in Supplementary Material). Interestingly, there is a patchy lag of autofluorescent RPE in the 92-year-old donor retina, indicative of local RPE atrophy. Invading macrophages and/or activated microglia were most prevalent in these tissue areas (Figures 5A,E). In contrast, the RPE appears to be largely intact in the 64-year-old control (Figures 5D,F).  In order to confirm the local retinal FHR-3 production, mRNA expression analysis from eye tissue was performed. We isolated mRNA from different RPE cells (primary human RPE cells, cultivated human RPE cells, and ARPE19 cells), as well as human liver cells as control, and tested cfhr3 together with cfh expression using qRT-PCR ( Figure S3 in Supplementary Material). Cfhr3 mRNA was only detected in liver cells but not in RPE-cells. However, cfh specific mRNA was determined in RPE and liver cells ( Figure S3 in Supplementary Material). Thus, different retinal cell types were separated from a human retina, and cfhr3 expression levels were compared to cfhr3-mRNA in ARPE19 cells using Taqman PCR ( Figure S3 in Supplementary Material). Cfhr3 expression levels were 2-to 140-fold higher in microglia/macrophages isolated from human retina compared to other retinal cell types ( Figure  S3 in Supplementary Material).
Factor H-related protein-3 is highly identical to FH and interacts with FH-binding partners. This suggested a similar binding pattern of FHR-3 and FH. Upon FH-staining of a human retina, Müller cells and photoreceptor cell segments were identified as FH-expressing cells (Figure 6). FH did not colocalize with FHR-3 in microglia/macrophages. The FH expression pattern in Müller cells appeared to be different in two investigated donor retinae. In the 92-year donor retina, the outer stem processes of Müller cells (Figures 6A-C), Müller cells enwrapping photoreceptor somata or the photoreceptor somata itself, and Müller cell microvilli reaching into the layer of photoreceptor segments were FH-positive (Figures 6D,E). In contrast, in the 64-yearold donor retina, FH labeling was more prominent in the inner stem processes (including perisynaptic side branches) of Müller glia spanning the inner plexiform layer (Figures 6D-F). This finding might be explained by the different degree of tissue damage as the disarranged retinal layers and activated microglia/macrophages in the 92-year-old retina were indicative of a more pronounced tissue inflammation with concomitant RPE loss and neurodegeneration compared to the 64-year-old donor (Figures 6A,D,F).
In summary, these stainings of the human retina and expression analyses demonstrated that FHR-3 expression might be locally regulated and that it is likely to depend on the activation level of microglia/macrophages. Our novel anti-FHR-3 mAb RETC-2 revealed that FH and FHR-3 were expressed in retinal cell types with entirely different functions.

anti-Fhr-3 mab reTc-2 reduced Molecular interaction of Fhr-3 with c3b but not with heparin
Having shown that FHR-3 is expressed by microglia/macrophages in the retina (Figure 5), we asked whether our novel anti-FHR-3 mAb RETC-2, directed against the SCR5 domain, interferes with the local binding characteristics of FHR-3. Therefore, we used competitive immunoassays to analyze the effect of mAb RETC-2 on FHR-3 interactions.
Factor H-related protein-3 binds in vivo to C3b on cell surfaces mainly via a simultaneous interaction with glycosaminoglycan chains. C3b interacted in vitro (without glycosaminoglycan chains) with recombinant FHR-3 and native FH (Figure 7A). The FHR-3-C3b interaction was not influenced by the addition of FH (Figure 7C). However, when anti-FHR-3 mAb RETC-2 was added to FHR-3 and C3b, the FHR-3-C3b binding was reduced by 32% (Figure 7C). This indicated either that in addition to SCR5 further domains are involved in the interaction of FHR-3 with C3b or that the binding strength of mAb RETC-2 was not sufficient to replace FHR-3 completely.
In accordance with previous studies, we observed a competitive effect of FHR-3 and FH for C3b binding (5) (Figure 7B). A molecular ratio above 12 FHR-3 to 1 FH molecule significantly reduced the binding of FH to C3b (Figure 7B). MAb RETC-2 binding to FHR-3 ameliorated the FHR-3 interference with FH for C3b binding by 29% and resulted in an increased FH concentration bound to C3b (Figure 7D). We found stronger RETC-2 effects for the interaction of FH and FHR-3 with oxidative stress epitopes (manuscript in preparation).
Previous reports showed an in vitro interaction of recombinant FHR-3 with heparin ( Figure S4 in Supplementary Material) (3,43). In order to better understand FHR-3 interaction with cell surfaces, we investigated whether mAb RETC-2 interferes with FHR-3 binding to heparin, a model for highly sulfated glycosaminoglycan chains. We conclude that the FHR-3-heparin interaction was not influenced by anti-FHR-3 mAb RETC-2 ( Figure S4 in Supplementary Material). The data indicate that the heparin-binding region in FHR-3 is not located in the SCR5 domain and thus not in the C-terminus of FHR-3.
These results showed that FHR-3 competes with FH for C3b binding and that this can be modulated by specific mAbs to the SCR5 domain of FHR-3 without affecting the interaction with heparin.

DiscUssiOn
We here report of new, highly specific murine antibodies for a well-defined epitope of FHR-3, which we used for systemic as well as local detection and functional modulation of FHR-3. FHR-3 is a member of the FH protein family, which are important regulators of the complement system and associated with several diseases, such as AMD and aHUS (4). The specific detection of was reduced by FHR-3 in a dose-dependent manner. Statistic was performed using one-way ANOVA with multiple comparisons and FHR-3 (0 nM) as control (**p < 0.01). (c) MAb RETC-2, but not FH, reduced interaction of FHR-3 to immobilized C3b. Statistic was performed using one-way ANOVA with multiple comparisons and FHR-3 as control column (**p < 0.01). (D) FH binding to C3b was reduced by FHR-3. This inhibitory effect was reversed by the addition of mAb RETC-2. Data represent mean values +SD of three independent experiments. Statistics were performed using One-way ANOVA with multiple comparisons and FH as control column (****p < 0.0001) and two-tailed unpaired t-test (***p < 0.001). proteins depends mostly on distinct binding of antibodies to the target antigen. Due to the high sequence identities among the FH protein family, the majority of previously generated antibodies showed cross-reactivity with other FH-family members ( Figure  S5 in Supplementary Material) (6,44), with the exception of mAb HSL1 (19). HSL1 binds to SCR4-5 of FHR-3 and was used to determine binding of FHR-3 to pathogens. However, a distinct binding epitope and functional modulation of FHR-3 using mAb HSL1 have not been described.
In this study, the anti-FHR-3 mAbs were generated against a selected peptide sequence of the SCR5 domain. The paralogous SCR20 in FH is known for most of disease-associated missense mutations (45)(46)(47) and binding of autoantibodies (48). This C-terminal domain of FH/FHR proteins also facilitates the binding to C3d and glycocalyx in vivo (49) (Figure S5 in Supplementary Material). Therefore, it was a favorable target for function-modulating antibodies.
Our mAb RETC-2 specifically detected FHR-3, native and denatured FHR-3, but no other FHRs. SCR9 domain of FHR-4A showed the highest identity with the FHR-3 target domain used for antibody generation (93.8%). Although only two amino acids (aa R285 and aa R288) of the immunogen were different in the corresponding region of FHR-4A, mAb RETC-2 was highly specific for FHR-3 and did not interact with FHR-4A. We exclusively precipitated FHR-3 and no other attached FH-family member from human serum, which supports the hypothesis that FHR-3 circulates as a monomer or homo-multimer in human serum, in contrast to FHR-1, FHR-2, and FHR-5, which form beside homomultimers also heterodimers and share a dimerization motif (18).
Besides a previous estimation on FHR-3 levels (5), there has been only one other report using an immunoassay to measure FHR-3 levels in healthy individuals, reporting systemic FHR-3 levels of 0.7 μg/mL on average, and a major influence of copy number variation in cfhr3 on the variation of the FHR-3 levels within a population (6). According to the high specificity of mAb RETC-2, we confirmed the FHR-3 concentration in human serum probes derived from healthy individuals in a concentration ranged between 0.41 and 2.49 μg/mL (excluding cfhr3deficient donors). This concentration is lower than other soluble complement regulators in blood (e.g., 13-30 μg/mL properdin; 40-100 μg/mL clusterin) (24,50) and even 9-to 1800-fold below the reported systemic FH concentration (116-562 μg/mL) (51).
Previous studies on the deletion of cfhr3/cfhr1 genes revealed a protecting effect in AMD and IgAN but revealed a risk factor for aHUS and SLE (9,10,12,13,15). Serum FHR-3 concentrations in patients suffering from AMD or aHUS showed a similar FHR-3 concentration range as healthy controls, although there seemed to be a trend toward higher levels in this small cohort.
These results should be further investigated in larger patient cohorts. However, in probes from patients with the autoimmune diseases SLE, RA, and PR threefold to fourfold increased systemic FHR-3 levels were detected.
Systemic lupus erythematosus is a complex disease with heterogeneous sub-phenotypes, which are all characterized by the production of autoantibodies, complement dysregulation, and inflammatory tissue injury (52). In previous studies, reduced FH levels and FH deficiency have been associated with SLE (53,54). This indicates that the FH-homeostasis is highly relevant and that FH-competitors, such as FHR-3 (5), could affect disease progression. While we did find significant differences in FHR-3 levels within our SLE cohort, our results only include few patient samples and further studies are required to elucidate the role of FHR-3 in different clinical SLE phenotypes.
Additionally, we tested patients suffering from autoimmune diseases for FHR-3 levels, as we assumed elevated systemic FHR-3 levels could be involved in inflammation (16). Complement activation is important in RA as immune complexes fix complement, resulting in the release of chemo-attractants for macrophages (55). It is known that FH is an important inhibitor of this complement activation in joints (56). FHR-3 has not been found to be involved in RA pathogenesis so far, but mouse studies recently revealed a higher mRNA concentration of mouse CFHR-C in the spleen of RA mice compared to healthy mice (57). It remains unclear whether increased systemic FHR-3 concentrations could result in local competition of FH and FHR-3 in joints and result in complement dysregulation.
Polymyalgia rheumatica is associated with inflammation in proximal joints (58) and viral stimulation of the immune system, including autoantibodies and classical pathway activation, has been proposed as a possible pathomechanism (59,60). Almost 30 years ago, Smith et al. identified proteins of the FH-family in immune complexes of PR patients (61). According to the specificity of the used anti-FH antibody at that time, the detected proteins could correspond to FHR-2, FHR-3, FHR-4, FHR-5, or FH-like than to the described FH (1). An association of FHR-3 with PR has not been described before, but increased systemic FHR-3 levels support the hypothesis of complement dysregulation in PR.
Concluding that systemic FHR-3 levels were not exceedingly altered in AMD patients, we focused further on the local presence of FHR-3 to decipher the role of cfhr3/cfhr1 gene deletion in this disease. FHR-3 and members of the FH-family are primarily known as secreted plasma proteins, which are produced in liver cells (40). In addition, a myeloma cell line was also tested positive for cfhr3 transcripts (62). Here, we describe the local expression of cfhr3/FHR-3 on mRNA and protein levels in the retina by putatively activated microglia or invading macrophages. Active microglia/macrophages are a common characteristic in retinal degenerative diseases and are associated with the release of inflammatory complement components (63)(64)(65). Increased local FHR-3 concentration in the damaged retina could influence the FH/FHR-3 balance. However, FHR-3 did not colocalize with FH in the human retina, as the FH protein was found in Müller and photoreceptor cell segments. So far, FH was detected at the choroidal site of the Bruch's membrane and in RPE in the human retina (66). The previously reported FH-expressing RPE forms the blood-retina barrier and the Müller cells span the whole retina (66). Both cell types are known for complement protein expression, suggesting a responsibility for maintenance of the immune homeostasis in the eye (67)(68)(69).
The local binding and competition of FHR-3 and FH on surfaces is an important mechanism for the regulation of immune homeostasis (5,19). We showed that 12 times more FHR-3 than FH molecules are needed for the displacement of FH on immobilized C3b in vitro. Systemic physiological conditions are associated with a FHR-3:FH ratio of 1:33 to 1:7400 (70). This relationship could be altered in pathophysiological processes. Reduced FH concentrations in stressed RPE and the detection of FHR-3 produced by macrophages in the damaged retina suggested a local shift of the FHR-3/FH equilibrium in retinal degeneration (71,72).
The interaction of mAb RETC-2 with FHR-3 partly influenced the competition of FHR-3 and FH for binding to C3b. As mAb RETC-2 binds in SCR5 of FHR-3, this suggests that SCR4 and SCR5 (19) of FHR-3 are involved in C3b binding ( Figure S5 in Supplementary Material). Indeed, the relevant amino acids in SCR19 of FH involved in C3b binding are conserved in SCR4 and SCR5 of FHR-3 (22, 73) ( Figure S5 in Supplementary Material). Our results are in agreement with previous studies, which showed that both C-terminal domains of FHR-3 (SCR 4-5) mediated the binding of FHR-3 to C3b, indicating that SCR5 is involved but not sufficient for C3b binding (3).
Factor H-related protein-3 interacts with cell surfaces not only via its C3b/C3d but also via its heparin-binding sites (3). MAb RETC-2 did not interfere with heparin binding to FHR-3, suggesting that the heparin-binding site is not located in SCR5 of FHR-3. These results are supported by the fact that the relevant amino acids for heparin binding identified in SCR20 of FH are not conserved in SCR5 of FHR-3 ( Figure S5 in Supplementary Material). Instead, a high amino acid identity is present between SCR2 of FHR-3 and SCR7 of FH, the latter including a heparinbinding site (1). Thus, FHR-3 probably binds with its N-terminal domain and not via the C-terminus to heparin ( Figure S5 in Supplementary Material).
To date, antibodies targeting the immune system offer promising therapies in different autoimmune diseases (74)(75)(76). Based on genetic association studies, FHR-3 seems to be a Janusfaced target: beneficial effects needed in aHUS and deleterious effects avoided in AMD. Inhibition of FHR-3 may therefore confer a therapeutic option in some autoimmune pathologies. Given the lack of effective therapy for the dry form of AMD and the reported local FHR-3 production in the damaged eye, FHR-3 inhibition by a humanized version of RETC-2 may be an effective therapeutic strategy (Figure 8). An unbalanced, local FHR-3/FH equilibrium could be the cause for local activation of microglia/macrophages in the human retina, as FH can not efficiently bind to the modified cell surfaces (Figure 8A). MAb RETC-2 interfered with FHR-3 binding to C3b fragments and enhanced FH binding. Our FHR-3 specific mAb may have a positive effect in the retina, helping to restore the local homeostasis by enhanced FH binding on cell surfaces to neutralize stress reactions (Figure 8B). Additional investigations of the cOnclUsiOn In summary, we generated specific antibodies against FHR-3, which detected endogenous FHR-3 in the retina and allowed to determine FHR-3 serum concentrations. Enhanced serum levels of FHR-3 found in sera from patients with autoimmune diseases, such as RA, SLE, and PR, underline the important role of FHR-3 in homeostasis. Our study confirms that FHR-3 competes with FH for C3b binding via SCR5 and binds to heparin via the N-terminal region. As the novel RETC-2 antibody inhibits competition of FHR-3 with FH and thus enhances local FH binding, our newly generated mAb RETC-2 may be a useful therapeutic target in specific autoimmune diseases.
aUThOr cOnTriBUTiOns NS and DP developed concept and designed the study. NS, AG, and DP designed experiments. NS, AG, JR, SH, VE, and DP performed experiments. NS, AG, BE, SH, VE, CS, and DP analyzed and discussed data. RP, TK, DW, BE, VE, PZ, and CS provided material. NS, AG, PZ, CS, and DP wrote the manuscript. RP, TK, and DW discussed and commented on the manuscript.