C3 glomerulopathy–associated CFHR1 mutation alters FHR oligomerization and complement regulation

A familial case of C3G associated with an internal duplication of CFHR1.

We present the case of a 12-year-old boy (GN29), his mother (GN29M), and his maternal grandfather (GN29GF); see Methods for detailed clinical presentation and development of disease. Biopsies in all showed C3G (Figure 1). Ultrastructural changes were heterogeneous: in GN29, deposits were present within the GBM and mesangium; in GN29M, accumulation of electron-dense material was found in subendothelial and subepithelial GBM locations (Figure 1).

We were able to collect samples from the 2 affected members alive in the pedigree, GN29 and GN29M, and from 2 healthy relatives, GN29B and GN29U (Figure 2A). Complement analysis in the proband and his mother illustrated persistent decreased levels of C3 compared with their healthy relatives (Figure 2B), which suggests significant activation of complement. Notably, C3 levels in these patients, although decreased, were significantly higher than those usually found in DDD patients. Assays to identify C3Nef or anti-FH autoantibodies were negative. Levels of membrane cofactor protein (MCP; also known as CD46) on the surface of peripheral blood lymphocytes were normal in both patients. Levels of factor I (FI), factor B (FB), and FH were also normal.

Routine Western blot analysis to characterize anomalies in the FHRs, performed as previously described (16), identified an anomalous FHR band of approximately 70 kDa in the proband and his mother that was absent in their healthy relatives (Figure 2C). Proteomic analysis of material from this band revealed that it exclusively contained sequences corresponding to FHR1 (Supplemental Figure 2). Multiplex ligation-dependent probe amplification (MLPA) analysis identified 2 rearrangements in the CFHR1-CFHR5 region segregating in this pedigree, the polymorphic deletion of the CFHR3 and CFHR1 genes (Δ_CFHR3-CFHR1) and a partial duplication of the CFHR1 gene, which was only carried by the affected individuals (Supplemental Figure 3A) and likely explains the anomalous FHR1 band found by Western blot in these individuals. To further characterize this CFHR1 genomic rearrangement, genomic DNA from the proband was analyzed in a custom-made high-density comparative genomic hybridization (CGH) 15k array (see Methods), which was designed to identify copy number variations within the regulators of complement activation (RCA) gene cluster. These analyses revealed an internal duplication of exons 2–5 in CFHR1 that predicted a mutant FHR1 protein (Figure 2D), consistent with the anomalous FHR1 band detected in the proband and his affected mother. Long PCR amplification analyses using primers to exons 5 and 2 of CFHR1 resulted in the generation of DNA fragments that confirmed the fusion of intron 5 to intron 1 (Supplemental Figure 3B). As a whole, these data demonstrate that in pedigree GN29, the disease segregates with an anomalous mutant FHR1 protein resulting from an internal duplication of the CFHR1 gene. This protein had 9 SCRs, with the first 4 duplicated in tandem (Figure 2E). Thus, the mutant FHR1 protein was expressed at concentrations similar to those of the native FHR1 protein and could be easily detected in the plasma of the mutation carriers.

The mutant FHR1 protein resembles the mutant FHR5 originally described by Gale et al. in Cypriots (14), and the FHR2::FHR5 hybrid protein described by Chen et al. (11), in the sense that each of these 3 mutant proteins is characterized by the duplication of SCR1 and SCR2. In all cases, these aberrant proteins were associated with C3G, either DDD or C3-GN. Interestingly, the disease showed full penetrance in the carriers of these 3 mutations. It also segregated as a dominant trait, which suggests that, in contrast to the protective effect of the Δ_CFHR3-CFHR1 allele, duplication of the N-terminal region of these FHRs results in a gain of function.

Native FHR1 interacts with FHR2 and FHR5, but not with FH, FHR3, or FHR4A/B.

To understand the functionality of the N-terminal SCR1 and SCR2 domains of FHR1 and to determine the consequences of the duplication of this region in the FHR1 protein, we purified native FHR1 from FH-depleted human EDTA plasma using, as a first step, heparin chromatography with a 50- to 350-mM NaCl gradient. Although FHR1 was fully retained on the heparin column, this chromatographic fractionation did not result in a single elution peak containing FHR1; instead, FHR1 eluted in 4 distinct peaks at different NaCl concentrations. Immunodetection of FHR2 and FHR5 in the eluted fractions using the mouse mAb MBC125 (which recognizes an epitope common to FHR1, FHR2, and FHR5) showed that these 2 proteins, which were also fully retained in the heparin column, coeluted with FHR1 in the NaCl gradient (Figure 3). FHR2 coeluted with FHR1 in 2 distinct peaks: in peak 1, FHR1 and FHR2 appeared equimolar (FHR1-FHR2), whereas in peak 2, FHR1 was approximately 3 times more abundant than FHR2 (FHR1₃-FHR2). In peak 3, FHR1 eluted alone, and in peak 4, both FHR1 and FHR5 were present (FHR1-FHR5) (Figure 3). Identical results were obtained with FH-depleted plasma from individuals with each of the 3 CFHR1*A/B genotypes (Figure 3 and ref. 17), which indicates that the peculiar FHR1 elution pattern was unrelated to the CFHR1 common polymorphism. Since no evidence of FHR1 posttranslational modifications was obtained to justify this complex FHR1 elution pattern, we concluded that it was a consequence of interaction of FHR1 with other FHRs. Moreover, because the analysis of FH-depleted plasma from individuals lacking FHR1 demonstrated that in the absence of FHR1, FHR2 was not retained in the heparin column (Figure 3), we postulated that FHR1 interacts with FHR2 and FHR5. Using this method, we could not see any evidence of an interaction between FHR2 and FHR5: we could not detect retained FHR2 when FHR1 was absent, and we did not observe coelution of FHR2 with FHR5 when FHR1 was present.

To determine whether FHR1 also interacts with FH, FHR3, and FHR4A/B, a sample of EDTA plasma from a CFHR1*A/B heterozygote was passed through a MBC125 affinity column, and the presence of FH and all FHRs was analyzed in the flow-through and retained fractions by Western blot using a panel of specific Abs. Since FH, FHR3, and FHR4A/B proteins were not retained, we concluded that they do not interact with FHR1, FHR2, and FHR5, which were retained on the column (Supplemental Figure 4).

These experiments also showed that, when eluted from the MBC125 column at pH 2.5 and renatured in PBS, FHR1, FHR2, and FHR5 demonstrated elution profiles from heparin columns identical to those obtained using FH-depleted plasma (Supplemental Figure 5). These data suggest either that the interactions between FHR1 and FHR2 (or FHR5) are very strong and are maintained through acid elution, or that formation of the different FHR complexes is a reversible process. In practical terms, this observation facilitated the development of a simple 2-step protocol for the purification of the different FHR1 complexes: affinity chromatographic purification from whole EDTA plasma using MBC125 mAb, and subsequent fractionation of the FHR1 complexes by heparin chromatography.

FHR1 assembles into homo- and hetero-oligomeric complexes.

Purified FHRs corresponding to the heparin chromatography elution peaks 1–4 (Figure 3), as well as FHR2 and FHR5 purified from the plasma of FHR1-deficient individuals, were analyzed by gel filtration using a Superdex 200 PC 3.2/30 column previously calibrated using 3 SCR-containing proteins of different sizes: FH (20 SCRs), C4bpαΔCT (8 SCRs), and soluble DAF (sDAF; 4 SCRs). 350 mM NaCl was used in these experiments to avoid interaction of the FHRs with the carbohydrate components in the column matrix. Interestingly, the different purified FHRs eluted as a single peak with an apparent size that was consistent with the size of a FHR dimer (Supplemental Figure 6), suggestive of the formation of homo- and heterodimers. The exception was FHRs in peak 4 (Figure 3), which eluted as 2 overlapping peaks. Evidence of FHR1-FHR2 heterodimers was also obtained in immunoprecipitation experiments using mAb MBI-18, which specifically recognizes FHR2 (Supplemental Figure 7).

Purified proteins corresponding to the FHR heparin chromatography elution peaks 1–4 were also analyzed in 4%–16% blue native–PAGE (BN-PAGE), which provided formal proof of the existence of homo- and hetero-oligomers of the FHRs (Figure 4). FH and sDAF were included in these native gels as molecular weight markers, and the FHRs were visualized by silver staining and Western blot analysis (Figure 4A). Bands from the native gels corresponding to the different complexes were excised and further analyzed in 12% SDS-PAGE gels (Figure 4B). As a whole, these experiments were consistent with the existence of homo- and heterodimers of the different FHRs. Notably, in samples containing FHR1, they also identified higher-order oligomers, which we interpreted as tetramers. For example, when purified FHRs included in peak 1 (fractions A11 to B11; Supplemental Figure 5) were run in the native gels, 2 FHR complexes were obtained (Figure 4A), both of which contained FHR1 and FHR2 (Figure 4B, lanes 7 and 8). Interestingly, the apparent sizes of the FHR oligomers containing FHR1 and FHR2 were intermediate between those obtained from peaks containing only FHR1 or only FHR2, consistent with the size of the FHR1-FHR2 hetero-oligomers (Figure 4).

A tetrameric structure could explain the differences between peak 1 and peak 2 in the FHR heparin chromatography elution profiles (Figure 3). In fact, our native gels showed that peak 2 contained both FHR1-FHR1 homodimers and FHR1-FHR2 heterodimers, which suggests that the FHRs within this peak assemble into a tetramer formed by a FHR1-FHR1 homodimer and a FHR1-FHR2 heterodimer. In contrast, peak 1 only contained FHR1-FHR2 heterodimers, which suggests that it corresponds to a tetramer resulting from the interaction of 2 FHR1-FHR2 dimers. In agreement with this interpretation, when the purified proteins in peak 2 were denatured in 1M NaCl, dialyzed, and rechromatographed in the heparin column, they again displayed peaks 1–3 (Supplemental Figure 8).

Because we found only interactions among FHR1, FHR2, and FHR5, and these FHRs have almost identical N-terminal regions distinct from other FHRs, this region is the most likely candidate to mediate FHR oligomerization. It is unclear why FHR2 and FHR5 only formed homo-oligomers in the absence of FHR1.

The FHR1 mutant protein associated with C3G in the GN29 pedigree assembles into unusually large multimeric complexes.

Consistent with the idea that the N-terminal region of FHR1 mediates homo- and hetero-oligomerization, analysis of the FHRs eluted from heparin chromatography from the FH-depleted plasma of GN29M showed an abnormal profile, with the mutant FHR1 coeluting with FHR2 through most of the NaCl gradient (Figure 5A). Moreover, gel filtration chromatographic analysis of the FHRs affinity purified from the EDTA plasma of GN29M using MBC125 revealed 2 major peaks containing the mutant FHR1 protein, one of them with a size considerably exceeding that of the normal FHRs analyzed under the same conditions (Figure 5B). We argued that these apparently large size structures were suitable for structural analysis by EM and set up experiments with purified native and mutant FHR1 complexes. Despite intensive effort, native FHR complexes could only be detected at high concentrations as an intersection of thin filaments corresponding to nearby molecules that were very close together. Unfortunately, these native FHR complexes were not visible upon dilution to the concentration required to separate individual molecules (data not shown). Similar behavior was observed for FH, which we interpreted as a limitation of the EM method. Thin filaments like those formed by FH and FHR1 were at the size limit for observation using EM; thus, they were only detected when several molecules were very close together or part of a larger complex. In contrast, we were able to visualize protein complexes in a preparation enriched for the large-size mutant FHR1 complexes (Figure 6). Importantly, these structures were detected at concentrations when individual molecules were separate, which suggests that FHR1 mutant complexes are larger than native FHR1 proteins. Mutant FHR1 complexes appeared as multimeric, bundle-like structures, each containing several intersecting thin filaments. Interestingly, despite mutant FHR1 complexes displaying heterogeneous shapes with multiple arms in different orientations, their sizes were rather homogeneous, which indicates that the dimensions of the complexes found were within a discrete range, and very large aggregates were not detected. As a whole, the EM data support that mutant FHR1 is assembled into a larger multimeric complex.

GN29 mutant FHR1 shows increased interaction with surface-bound C3b, iC3b, and C3dg.

To explore the functional consequences of the effect of multimerization on the mutant FHR1, we performed surface plasmon resonance (Biacore) experiments. Because we could not demonstrate complement inhibitory capabilities for FHR1 in preliminary analyses (Supplemental Figures 9–11), we therefore focused on the interaction of FHR1 with surface-bound C3 fragments. Our results showed that when C3b was immobilized on a CM5 Biacore chip through the thioester group, mutant FHR1 bound more efficiently to C3b than the native FHR1 protein. The results illustrated fast binding and dissociation kinetics for the native FHR1 protein, whereas binding of the mutant FHR1 showed much slower dissociation (Figure 7A). As expected, these results support the hypothesis that multimerization of FHR1 in the mutant protein increases avidity for its ligands, leading to much enhanced binding to C3b compared with native FHR1. Similar results were obtained with iC3b and C3dg. Binding of native and mutant FHR1 was comparable to that of FH on the same surfaces: FH bound well to the C3b surface, illustrating marked avidity effects as would be expected on the “clustered C3b” surface. Conversion to iC3b almost eliminated binding of FH, whereas native and mutant FHR1 bound almost as well to iC3b as to C3b (Figure 7B).

GN29 mutant FHR1 shows enhanced competition with FH.

As a consequence of binding more efficiently to its ligands, the mutant FHR1 protein may compete better with FH. To investigate this possibility, we tested native and mutant FHR1 in a variation of the FH-dependent hemolytic assay previously developed in our laboratory (12). In this modified assay, a human serum sample was incubated with guinea pig erythrocytes (instead of the sheep erythrocytes used in the original assay) in the presence of 5 mM MgCl₂ and 8 mM EGTA. Guinea pig erythrocytes were efficiently lysed in this assay, but lysis could be prevented by the addition of extra FH (Figure 8, inset). Importantly, FHR1 competed with FH in a dose-dependent manner, influencing control of the AP on the surface of the guinea pig erythrocytes. As shown in Figure 8, the mutant FHR1 protein competed with FH more efficiently than did native FHR1. These data indicate that multimerization as a consequence of duplication of the N-terminal region enhances competition with FH for certain surfaces, resulting in increased activation of the complement AP.

Patients.

GN29 is a 12-year-old boy who, coincident with an episode of fever associated with an upper respiratory infection and purulent conjunctivitis by Haemophilus influenzae, presented with macroscopic hematuria at 15 months of age. 1 month earlier, he had a similar episode that resolved spontaneously in 3 days. At that time, moderate fever was also present, as well as the macroscopic hematuria and subnephrotic-range proteinuria. Blood pressure and renal function were normal. Plasma C3 levels were slightly below the normal range. A third episode, similar to the previous ones, occurred at age 20 months, precipitating a nephrology review. Despite preservation of renal function and normotension, he had persistent microhematuria of variable intensity that becomes macroscopic during intercurrent fever. Proteinuria remained in the subnephrotic range with no peripheral edema. However, in 2009, he developed edema and nephrotic-range proteinuria that was steroid responsive, coincident with a drop in plasma C3 levels (Figure 2B). Current medication is angiotensin converting enzyme inhibitor (ACEI) therapy, and proteinuria is now in the subnephrotic range. Notably, his C3 levels in plasma have always been below the normal range. A renal biopsy performed in 2008 revealed glomerular hypertrophy with diffuse capillary and mesangial thickening and hypercellularity (Figure 1A). The GBM was thick due to the linear accumulation of PAS-positive material that was also found locally in the mesangium (Figure 1, B and C). Immunofluorescence studies revealed very intense granular C3 staining along thickened capillary loops (Figure 1D) and within the mesangium, where focal IgM staining was also present (Figure 1E). Staining for IgG was negative (Figure 1F). Transmission EM demonstrated the presence of osmiophilic deposits in the GBM and mesangium (Figure 1G).

GN29M, mother of GN29, had microhematuria, trace proteinuria, arterial hypertension, and decreased plasma C3 levels in a routine screening in 1999, prior to the in vitro fecundation during her first and only twin pregnancy. She delivered by Caesarean section because of preeclampsia. Since 2002, her hypertension has been controlled with ACEI treatment and diet. Her renal function is preserved (0.79 mg/dl creatinine, corresponding to a creatinine clearance rate of 140 ml/min). She maintains a persistent variable proteinuria (trace to 50 mg/dl), moderate microhematuria (>50 erythrocytes/field), and chronic hypocomplementemia, with levels of plasma C3 persistently below the normal range. A renal biopsy performed in 2009 showed a C3G with intense C3 deposits similar to those observed in her son, negative IgG and IgM staining, and the presence of subendothelial and subepithelial GBM and mesangium osmiophilic deposits (Figure 1, H–O).

GN29GF, maternal grandfather of GN29, developed chronic renal insufficiency at 51 years of age. Renal biopsy diagnosis was membranoproliferative glomerulonephritis type II (now DDD). 6 years later, he presented with acute myocardial infarction and multivessel coronary artery disease. In 2001, he reached end-stage renal disease. He received a renal transplant, but died in 2007 from acute myocardial infarction.

Mutation screening and genotyping.

Genomic DNA was obtained from peripheral blood mononuclear cells using Puregene Blood Core kit B (QIAGEN) according to manufacturer’s instructions. Each exon of the CFH, MCP, CFI, C3, and CFB genes was amplified from genomic DNA using specific primers derived from the 5′ and 3′ intronic sequences, as described previously (2, 22, 28–30). Automatic sequencing was performed in an ABI 3730 sequencer using a dye terminator cycle sequencing kit (Applied Biosystems).

MLPA analysis and high-density CGH arrays.

Copy number variations and genomic rearrangements in the CFH-CFHR1-CFHR5 region were assessed by MLPA analysis and custom-designed high-density 8 × 15k oligonucleotide CGH arrays spanning the RCA gene cluster region in human chromosome 1q32 (median resolution, 110 bp; AMADID 040193; Agilent Technologies). MLPA analysis was done with genomic DNA with the P236 A1 ARMD mix 1 (MRC-Holland). For the CGH 8 × 15k microarrays, samples were hybridized with sex-matched normal human DNA (Promega Corp.) according to the manufacturer’s protocols (31). Microarray data were extracted and visualized using Feature Extraction Software (version 10.7) and Genomic Workbench Standard Edition (version 7.0; both Agilent Corp). Copy number–altered regions were detected using ADM-2 (set as 5) statistic provided by DNA Analytics, with a minimum number of 5 consecutive probes. Genomic build hg19 was used for the experiment. All work involved in the hybridization of the CGH 8 × 15k microarrays and data extraction was outsourced to Nimgenetics. Microarray data have been deposited in GEO (accession no. GPL16749).

Proteins and Abs.

C3b was prepared as previously described from human EDTA plasma (32). sDAF was a gift from S. Lea (Oxford University, Oxford, United Kingdom). C4bpαΔCT was prepared as previously described (33). FH from human plasma was purified by immunoaffinity chromatography. Briefly, filtered plasma was loaded into an immunoaffinity column coupled to an in-house anti-human FH mouse mAb (mAb214). Bound protein was eluted at low pH, and fractions containing the protein were pooled and loaded into a Superose-6 column equilibrated with 10 mM HEPES (pH 7.6) and 100 mM NaCl. MBC125 is a mouse IgG2a mAb raised against a recombinant protein containing the first 3 SCRs of FHR1. This Ab recognizes an epitope present in FHR1, FHR2, and FHR5 and thus allows for simultaneous quantitative detection of the 3 proteins in Western blot analyses. Other Abs were in-house mouse mAb MBI-18 against FHR2 (gift from B.P. Morgan, Cardiff University School of Medicine, Cardiff, United Kingdom); in-house rabbit polyclonal Ab against FHR3 and FHR4 and rabbit polyclonal Ab against FHR4 (gifts from P.F. Zipfel and M. Jozsi, Hans Knöll Institute for Natural Products Research, Jena, Germany); and OX24, a mouse mAb to FH originally made by R. Sim (University of Oxford, Oxford, United Kingdom).

BN-PAGE.

Samples were analyzed by BN-PAGE. Briefly, the different FHRs purified from the heparin columns were dialyzed overnight against 20 mM HEPES, pH 7.0, and 150 mM NaCl. Samples were mixed with loading buffer (50 mM BisTris, pH 7.2; 50 mM NaCl; 10% [v/v] glycerol; 0.001% [v/v] Ponceau S; 0.25% [w/v] Coomassie G-250) and loaded on precast BN-PAGE gels (Novex BisTris; Invitrogen). Electrophoresis was performed at 4°C according to the manufacturer’s protocol, and proteins were identified by silver staining and Western blot. The composition of each band detected by BN-PAGE was analyzed by 12% SDS-PAGE followed by Western blot using the MBC125 mAb. Silver staining and Western blots were carried out according to standard protocols.

Biosensor analysis/binding affinity assays using SPR.

All analyses were carried out on a Biacore T100 (GE Healthcare). A nidus of C3b (150 RU) was amine-coupled to a CM5 (carboxymethylated dextran) sensor chip as instructed by the manufacturer (NHS/EDC coupling kit; GE Healthcare). Further C3b was deposited via the internal C3 thioester group by forming convertase on amine-coupled C3b by flowing 50 μg/ml FB and 1 μg/ml factor D (FD) using the same buffer supplemented with 1 mM MgCl₂, followed by C3 as substrate for the convertase (34). 1,240 RU nascent C3b was covalently bound to the chip surface. Coupling in this manner results in “clustered” C3b on the surface as would be present on a target. To generate iC3b, the C3b surface was treated with successive cycles of FH (10 μg/ml) and FI (10 μg/ml) until convertase could no longer be formed. To generate C3dg, the iC3b surface was treated with successive cycles of soluble CR1 (50 μg/ml; gift from T Cell Sciences) and FI (10 μg/ml) until no further loss of mass (C3c) was evident. To assess binding to C3b, iC3b, or C3dg, samples were injected using the “High Performance” command, in HBS/P (10 mM HEPES, pH 7.4; 150 mM NaCl; 0.01% surfactant-P20) flowed at 20 μl/min and analyzed at 25°C. All data were double-referenced (data from control cell and blank injection subtracted). The chip surface was regenerated between injections using 10 mM sodium acetate, pH 4.0, and 1M NaCl. Data were evaluated using Biaevaluation software (version 4.1; GE Healthcare).

FH-dependent hemolytic assay.

The capacity of FHR1 and mutant FHR1 to compete the activity of FH to control AP on cellular surfaces was assessed with a hemolytic assay using guinea pig erythrocytes. In brief, guinea pig erythrocytes (TCS Biosciences) in AP buffer (0.5% packed cell volume) —consisting of veronal buffer saline (2.5 mM barbital; 1.5 mM sodium barbital; 144 mM NaCl, pH 7.4) with 5 mM MgCl₂, 8 mM EGTA, and 0.1% gelatin — were incubated with sufficient amounts of 10% serum in AP buffer and increasing amounts of FH for 1 hour at 37°C. After centrifugation, supernatants were read at 414 nm. Titration of FH before each experiment determined conditions for the competition assay. The minimal concentration of FH giving 30% lysis was mixed with different amounts of FHR1 proteins to measure their capacity to compete with FH (starting concentration, 150 nM). Serum without added FH was taken as 100% lysis, and erythrocytes diluted in AP buffer were used as blank for spontaneous lysis.

EM analyses.

Mutant FHR1 preparations were subjected to a final purification step using size-exclusion Superdex 200 chromatography (GE Healthcare) in 20 mM Tris-HCl, pH 7.5, and 350 mM NaCl before the preparation of EM grids. A selected fraction was adsorbed onto glow discharged carbon-coated grids and negatively stained with 2% (m/v) uranyl formate. Grids were visualized in a JEOL 1230 transmission electron microscope operated at 100 kV, and micrographs were recorded on a 4k × 4k TVIPS CMOS detector at a nominal magnification of ×50,000.

Statistics.

Statistical analysis of the FH-dependent hemolytic assay data in Figure 8 was done by 2-tailed t test (independent samples). The analysis was performed using SPSS software (version 21; SPSS). Results are expressed as a mean of 3 independent experiments. A P value of 0.05 or less was considered significant.

Study approval.

The studies described herein received IRB approval (Comision de Bioetica, Consejo Superior de Investigaciones Científicas). Patients and their relatives gave their informed consent.