CFHR5 Genetic Variations and Serum Levels in Patients with Immune-Complex-Mediated Membranoproliferative Glomerulonephritis and C3- Glomerulopathy


 Background: Factor H-related-5 (FHR-5) is a member of the complement Factor H protein family. Due to the homology to Factor H, the main complement regulator of the alternative pathway, it may also be implicated in pathomechanism of kidney diseases where Factor H and alternative pathway dysregulation play a role. Here, we report the first observational study on CFHR5 variations along with serum FHR-5 levels in immune-complex-mediated membranoproliferative glomerulonephritis (IC-MPGN) and C3 glomerulopathy (C3G) patients together with the clinical, genetic, complement and follow-up data.Results: 120 patients with a histologically-proven diagnosis of IC-MPGN/C3G were enrolled in the study. FHR-5 serum levels were measured in ELISA, the CFHR5 gene was analyzed by Sanger-sequencing, and selected mutants were studied as recombinant proteins in ELISA and SPR.Eight relevant CFHR5 variations in 14 patients (11.7%) were observed, 5 of them identified as pathogenic for C3G. The FHR-5G278S and FHR-5R356H mutations altered the interaction of FHR-5 with C3b, when compared to the FHR-5WT. Serum FHR-5 levels were lower in patients compared to controls. Low serum FHR-5 concentration at presentation associated with better renal survival during the follow-up period, furthermore, it showed clear association with signs of hypocomplementemia and clinically meaningful clusters.Conclusions: Our observations support the hypothesis that FHR-5 protein and its genetic alterations play a role in the pathogenesis of IC-MPGN/C3G.

* group comparisons were made with Mann-Whitney U test between "total" and "controls" ** LPVs were detected in the following genes: CFH, CFI, CFB, C3, CD46, THBD. Previously identi ed and functionally characterized missense, as well as nonsense and splice site mutations were categorized as LPV. Novel missense variations were regarded as LPVs if their minor allele frequency was < 1% in the 1000Genomes Project phase 3 (http://browser.1000genomes.org/index.html) or in gnomAD (https://gnomad.broadinstitute.org/).    there was no difference between the various histology-based groups. The speci city of the FHR-5 ELISA was con rmed by western blot (WB) showing that the monoclonal anti-FHR-5 antibody used as capture antibody did not detect FH or any FHR other than FHR-5, and it recognizes FHR-5 WT , FHR-5 G278S , and FHR-5 R356H ( Fig. 1 and data not shown).
The detailed evaluation and interpretation of the CFHR5 variations are presented in Table 2A. The in silico predictions of the variations are presented in Table 2B.
Remarkably, the identi ed missense and frame-shift CFHR5 variations affect the short consensus repeat (SCR) domains 1-6 in the FHR-5 protein (Fig. 2). We consider 5 variations likely pathogenic, because of 1) variation was previously reported in C3G patients and/or is a recognized pathogenic factor of the disease (p.E163Kfs*10, p.E163Rfs*35) or based on literature data (24) showing segregation with the disease (in case of P46S, carried by HUN1502 and HUN2446) or because 2) results of the functional analysis (for G278S and R356H see below, observed in patients HUN225, HUN290, HUN769, HUN1190, HUN1325) (Table 2A).
In patient HUN593 carrying both a missense and a frame-shift CFHR5 variation (C208R, c.479_480insA), the serum level of FHR-5 was not markedly decreased (1.88 mg/L) contrary to what was expected, and we further investigated this sample on WB, where FHR-5 protein was detected in the expected position ( Fig. 3) with comparable intensity to that of wild-type FHR-5, con rming the observation done by ELISA (Fig. 2). The possible explanation is that the two identi ed CFHR5 variations are located on the same allele and thus the other allele is intact (unfortunately, we have not received samples from the parents, therefore we cannot con rm this hypothesis).
In patient HUN2446 three different CFHR5 variations were identi ed including one frame-shift and two missense variations (P46S, C208R, c.479_480insA). As it was expected, the serum level of FHR-5 was decreased (0.47 mg/L). The patient's mother carries two CFHR5 variations (C208R, c.479_480insA) whereas the father is heterozygous for CFHR5 P46S. Both of the parents are healthy (Fig. 4). Interestingly, the patient's mother did not have any signs of renal disease whereas another patient carrying the same two CFHR5 variations (HUN593) is diagnosed with IC-MPGN. HUN593 also had anti-Factor B antibody and is heterozygous for CFH Y402H that are additional risk factors for the disease.
FHR-5 serum levels were similarly low in patients, irrespective of CFHR5 variations, when compared to controls (Fig. 5). There was no remarkable association of the FHR-5 protein concentrations with the localization of the variations in various FHR-5 domains (Fig. 2).

Functional characterization of CFHR5 variations
To determine whether CFHR5 variations in uence FHR-5 functions, we measured serum FHR-5 binding to C3b. FHR-5 binding was signi cantly lower in patients carrying the G278S mutation compared with those expressing only wild-type FHR-5, carrying mutations in the dimerization domains SCR1-2 (P46S, V110A, K144N) or carrying the R356H mutation in the predicted ligand binding region SCR5-7 (Fig. 6A). To con rm that the reduced binding is due to the mutations and not to the differences in FHR-5 serum level,  Table 3, patients with CFHR5 variations are presented as a reference group. Seventeen out of the 92 patients (median: 1.53years; minmax: 0.05-6 years) progressed or stayed in ESRD ( Table 4). None of the patients with CFHR5 variation(s) progressed to ESRD during follow-up (Table 3 and Fig. 7). Fifteen patients with high FHR-5 levels (ESRD rate: 0.38 event/patient/year; median follow-up: 4.5 years, min-max: 0.05-6) progressed to ESRD whereas the same was observed only for 2 patients with low FHR-5 (ESRD rate: 0.04 event/patient/year; median follow-up: 1.98 years min-max: 0.11-6) concentrations. Patients with higher FHR-5 levels had the worst renal survival, when compared to patients with low FHR-5 concentrations (p = 0.034), or when the three groups were analyzed together (p = 0.016) (Fig. 7).

Discussion
In this study, we have examined the FHR-5 serum levels and the CFHR5 genetic variations in a large group of IC-MPGN/C3G patients and in healthy controls. Furthermore, we identi ed 5 different CFHR5 variations that are considered to have a likely pathogenic effect, either as 1) con rmed by our functional studies (in case of G278S and R356H), or 2) the variation was previously published in C3G patients and/or is a recognized pathogenic factor of C3G (p.E163Kfs*10, p.E163Rfs*35) or 3) based on literature data (24) showing segregation with the disease (in case of P46S).
To the best of our knowledge, this is the rst observational study describing FHR-5 levels together with the presence of likely pathogenic CFHR5 variations as well as clinical and laboratory data in a reasonably large group of IC-MPGN/C3G patients. In this cohort, 11.7% of the patients carry CFHR5 variations, representing 8 different missense/frame-shift variations.
Similarly to C3, C4, CP and AP activity, FHR-5 levels were lower in patients than in healthy controls.
Patients with low FHR-5 levels had superior renal survival compared to patients with higher FHR-5 levels, and this association was independent of CFHR5 variation carrier status. Interestingly, FHR-5 levels and CFHR5 variations showed clear association with clusters of the patients: patients with hypocomplementemia, low FHR-5 levels, presence of CFHR5 variations and good renal outcome fall into cluster 1, whereas patients in cluster 3 and 4 had higher levels of FHR-5 with worse renal outcome.
Genetic alteration of CFHR5 as a pathogenic factor was rst described by Gale et al (27), where they showed that internal duplication of exons 2-3 of CFHR5 leads to familial C3G termed CFHR5nephropathy. FHR-5 was reported to be colocalized in renal tissue together with other complement containing immune-deposits (19,37), although the pathogenic role of this protein was not fully understood and it was hypothesized that it may have a physiological role in complement activation in the kidney but large observational studies are missing. In a large American cohort (n = 104) only CFHR5 variations were examined without further comparison with serum levels and clinical data.  (39)) and the dimerization ability of FHR-5, but functional studies are needed to con rm this hypothesis. We also identi ed 2 different frame-shift mutations that are caused by the insertion of one or two adenine bases (E163Rfs*35; E163Kfs*10), affecting the SCR3 domain that has unknown function. Interestingly, in a previous case report, the E163Rfs*35 mutation occurred along with low FHR-5 concentration in one patient suffering from glomerulonephritis following streptococcal infection but not in unaffected carriers (34). In our cohort the FHR-5 levels of these three patients with frame-shift mutations were not decreased when analysed by ELISA and WB methods. This could be explained by the fact that these patients are heterozygous and the intact allele can produce the detected protein.
The remaining 3 missense variations (G278S, R356H and the novel C208R) affected the SCR4-6 domains that are partly responsible for the binding of C-reactive protein (21), heparin (SCR5-7(21, 40)), laminin (SCR5-7 (40)) and necrotic human endothelial cells (SCR5-7 (40)) based on previous studies (and the binding of pentraxin-3 to SCR5-7 was also hypothesized (23)). Of these, the C208R variation occurred together with the E163Kfs*35 mutation in one IC-MPGN patient (and it was also observed in two aHUS patients of our registry, unpublished results) and in one C3GN patient with the P46S variation also. Three patients carrying the G278S variation had variable levels of FHR-5 and both patients carrying the R356H variation had FHR-5 concentration below the median level observed in controls. We could not nd any additional etiological factors in the examined genes and complement autoantibodies except for CFHR5 variations (resulting in P46S/E163Rfs*35/C208R/G278S/R356H) in 7 patients. By analyzing recombinant proteins harboring these mutations, we found the FHR-5 G278S variant has decreased C3b binding ability (Fig. 3). Since FHR-5 was shown to compete with FH for ligand binding, among others for C3b, the reduced competition capacity of FHR-5 might explain our observation that patients carrying this variant tend to have better renal outcome (none of them progressed to ESRD during the median 1.36 yearlong follow-up period (min: 0.35 year; max: 5.12 years)), as a more appropriate regulation is achieved by FH. Nevertheless, C3b is only one of the common FH/FHR ligands, the function of FHR-5 and the effect of mutations should be further investigated. In an other perspective, interaction between FHR proteins and the extracellular matrix components can also modify the regulator activity of FHR proteins. The interactions between FHR-5 protein and the surface components such as glycosaminoglycans may also play a role in complement dysregulation (25). As C3G is considered as a polygenic disease, variations in CFHR5 are supposed to play a disease-modifying role in C3G predisposition and disease development. In recent years, two studies performing CFHR5 sequencing in aHUS patients (n = 54 and n = 65) were published which reported some novel CFHR5 variations, including two alterations coding for K144N and R356H that were observed in our patients as well (29,30). On the other hand, a large study was reported including 500 IgAN patients (31) carrying several CFHR5 variations but none of these mutations was observed in our patients. FHR-5 levels were higher in IgAN patients than in control subjects in several studies (32,33)  patients with lower FHR-5 level, than in patients with higher FHR-5 level. This observation is similar to the observation obtained for IgAN patients (32). We do not have a formal explanation for this phenomenon, but our hypothesis is that decreased level/function of deregulator FHR-5 protein allows FH to regulate complement activation more e ciently. If the delicate balance between FH and FHR-5 function could be shifted due to a damaging CFHR5 mutation, the pathogens (of which many have the capacity to interact with FH, (42)) may escape complement action as a result of decreased deregulator activity during disease development. In turn, later in disease course patients having lower deregulator activity, and hence, appropriate regulation by FH, may have a better renal survival. In addition, patients with lower baseline eGFR have higher FHR-5 serum levels which also support this hypothesis. Remarkably, in this cohort no ESRD developed during follow-up in patients with CFHR5 variations.
All these observations support the possibility that FHR-5, and thus carriage of CFHR5 variations, may play a role in the pathogenesis of IC-MPGN/C3G but these are not the only disease-causing factors. Our hypothesis is that the observed CFHR5 variations may be additional ne-tuning factors in disease pathogenesis and progression. In many cases autoantibodies against different complement components occur together with genetic variations or risk polymorphism, however, we have not identi ed a clear association between them in our previous studies (4,36). Recently, a large-scale whole-genome sequencing study did not nd a clear relationship between the identi ed rare variations and C3G, however, a strong association was identi ed between primary MPGN and a haplotype containing DQA1*05:01; DQB1*02:01 and DRB1*03:01. Of these, DQB1*02:01 and DRB1*03:01 are associated with different autoimmune diseases such as rheumatoid arthritis and membranous nephropathy. These genes are coding components of the MHCII molecule (found on the surface of antigen-presenting cells) which plays an important role in the adaptive immune response and in (auto)antibody production. These results raise the possibility that although genetic variations have a disease-modifying effect, it is the aberrant adaptive immune mechanism, thus autoimmunity that could be the key mechanism in the background of C3G (as shown by the high occurrence of autoantibodies) rather than the genetic abnormalities (43) We have further analyzed whether CFHR5 variations and serum FHR-5 concentrations are in connection with the recently described (18) and validated clinically meaningful clusters (4). IC-MPGN/C3G patients were clustered based on clinical, histological, complement and genetic data and clear association with disease pathogenesis and renal survival was observed supporting the relevance of the clusters, and our study found that FHR-5 levels were lower in cluster 1 along with higher prevalence of CFHR5 variations. Cluster 1 was also characterized by younger age of onset, higher complement activation with higher prevalence of complement autoantibodies and better renal survival. On the contrary, worst renal survival was observed in clusters 3 and 4 in our study, and patients in cluster 3 and 4 had higher FHR-5 levels.

Conclusions
In conclusion, our study is the rst to report observational data on CFHR5 variations together with functional characterization, and serum FHR-5 levels at disease presentation in a large group of IC-MPGN/C3G patients. We observed that 14 patients (11.7%) were carriers of 8 different CFHR5 variations, among them we considered 5 as pathogenic or likely pathogenic based on functional analysis or detailed evaluation and segregation data from the literature (Table 2A) Biopsy data were collected using standardized questionnaire forms from pathologists (n = 73) or extracted from the biopsy descriptions (n = 47).
Eighty-ve subjects formed the control group (68 adults, 17 children). All of them were referred for routine medical examination and none of them had any known disease at time of blood sampling.
Both the healthy subjects and the patients or their parents gave their written approval -after informed consent -for the analysis in accordance with the Declaration of Helsinki.

Molecular Genetic Analysis
In 111 patients the whole coding region of the gene encoding FHR-5 (CFHR5; OMIM# 608593) was screened by direct bidirectional DNA sequencing, as described in the case of further disease-associated genes (CFH, CFI, CD46, THBD, CFB, C3) that were sequenced also (44). No DNA samples were available from the remaining 9 patients. These genes were not sequenced in the included healthy subjects; instead of that the available public databases (1000Genomes Project, gnomAD) were used to acquire data about allele frequencies in more reliable and larger healthy populations. Primer sequences and PCR conditions are available upon request. Multiplex ligation-dependent probe ampli cation (MLPA) was performed to study the presence of copy-number alterations in the chromosomal regions of the CFHR1, CFHR2, CFHR3 and CFHR5 genes, with the SALSA MLPA probemix P236-A3 (MRC-Holland, Amsterdam, the Netherlands) following the manufacturer's instructions.
Polymorphic variations are numbered as + 1 from the A allele of the ATG translation initiation site. Those missense, nonsense or splice site mutations that were previously identi ed were categorized as likely pathogenic variations. In order to include every variation and not to miss information, novel variations were also included if they were found with a minor allele frequency of < 1% in 1000Genomes Project phase 3 (http://browser.1000genomes.org/index.html) or in gnomAD (https://gnomad.broadinstitute.org/). The possible functional effect of novel variations was predicted in silico using the following online prediction tools, such as PolyPhen (version2) (

Generation Of Recombinant Wild-type And Mutant Fhr-5 Variants
Coding sequence of wild-type CFHR5 (wtCFHR5) was codon-optimized for insect cell expression system (Integrated DNA Technologies, Inc.) and cloned into the pBSV-8His expression vector (50). Two mutants were ampli ed from the wtCFHR5-containing vector with mutagenic forward primers introducing the mutations FHR-5 G278S and FHR-5 R356H . Sequences and mutations were con rmed by sequencing.
Recombinant proteins were produced in Spodoptera frugiperda (Sf9) cells after co-transfection of the various CFHR5-containing expression vectors with linearized baculovirus DNA (Oxford Expression Technologies Ltd.) and puri ed from the supernatant by nickel a nity chromatography.

Measurement Of The Interaction Of Fhr-5 With Puri ed C3b
To compare the C3b binding ability of FHR-5 WT , FHR-5 G278S and FHR-5 R356H , microtiter plate wells Measurements were performed at 50 µl/min ow rate, association was followed for 120 s and the dissociation for 600 s. Data were processed and analyzed with ProteOnManager software. The curves were corrected by subtracting the non-speci c binding responses obtained from control, the ovalbumin captured channel. Binding curves were t to bivalent analyte model and the equilibrium dissociation constants were calculated from the directly estimated association and dissociation rate constants (K D = k d /k a ). The experiment was performed twice on separate GLC biosensor chips.

Determinations Of Complement Parameters
Samples (serum, EDTA-anticoagulated-and sodium-citrate-anticoagulated plasma) were taken from the antecubital vein, or from a central venous catheter. Cells and supernatants were separated by centrifugation after the sample was taken, and transferred to our laboratory. Separated aliquots were stored at − 70 °C until measurements.
AP activation was measured by a commercially available kit (Wieslab AP ELISA KITs, EuroDiagnostica, Malmö, Sweden), according to the manufacturer's instructions.
Total CP activity was measured by a home-made hemolytic titration test based on Mayer's method (51).
Radial immunodiffusion was performed to measure the antigenic concentrations of Factor I and Factor B, using speci c antibodies (52). Levels of FH, C1q and antibodies against FH, C1q,(52-54) C3 and Factor B were measured with in-house ELISA methods (4) whereas C3NeF and C4NeF titer was determined based on hemolytic method (4,55).
In order to split FHR-5 levels into high and low groups, receiver operating characteristic analysis was made. Kaplan-Meier analysis with log-rank test was performed to examine patients' renal survival.
For the statistical analysis IBM SPSS Statistics 20 and Graph Pad Prism 5 software were used. Availability of data and materials The dataset used and/or analyzed during the current study are available from the corresponding author on reasonable request.