Complement Factor B Mutations in Atypical Hemolytic Uremic Syndrome—Disease-Relevant Or Benign?

BASIC RESEARCH www.jasn.org

Complement Factor B Mutations in Atypical Hemolytic Uremic Syndrome—Disease-Relevant or Benign?

†‡ † † ‡ Maria Chiara Marinozzi,* Laura Vergoz,* § Tania Rybkine,* § Stephanie Ngo, | † † † Serena Bettoni, Anastas Pashov,¶ Mathieu Cayla,* § Fanny Tabarin,* § Mathieu Jablonski,* § † †† † Christophe Hue,* § Richard J. Smith,** Marina Noris, Lise Halbwachs-Mecarelli,* § | ‡ † Roberta Donadelli, Veronique Fremeaux-Bacchi,* and Lubka T. Roumenina* §

*Institut National de la Santé et de la Recherche Médicale UMRS 1138, Cordeliers Research Center, Complement and Diseases Team, Paris, France; †Université Paris Descartes Sorbonne Paris-Cité, Paris, France; ‡Assistance Publique— Hôpitaux de Paris, Service d’Immunologie Biologique, Hôpital Européen Georges Pompidou, Paris, France; §Université Pierre et Marie Curie (Paris-6), Paris, France; |IRCCS—Istituto di Ricerche Farmacologiche Mario Negri, Clinical Research Center for Rare Diseases Aldo e Cele Daccò, Ranica, Bergamo, Italy; ¶Molecular Medicine, Stephan Angelov Institute of Microbiology, Bulgarian Academy of Sciences, Soﬁa, Bulgaria; **Molecular Otolaryngology and Renal Research Laboratories and Rare Renal Disease Clinic, Departments of Pediatrics and Internal Medicine, Carver College of Medicine, University of Iowa, Iowa City, Iowa; and ††Laboratory of Immunology and Genetics of Transplantation and Rare Diseases, Mario Negri Institute for Pharmacological Research, Ranica, Bergamo, Italy

ABSTRACT Atypical hemolytic uremic syndrome (aHUS) is a genetic ultrarare renal disease associated with overactivation of the alternative pathway of complement. Four gain-of-function mutations that form a hyperactive or deregulated C3 convertase have been identified in Factor B (FB) ligand binding sites. Here, we studied the functional consequences of 10 FB genetic changes recently identified from different aHUS cohorts. Using several tests for alternative C3 and C5 convertase formation and regulation, we identified two gain-of-function and potentially disease-relevant mutations that formed either an overactive convertase (M433I) or a convertase resistant to decay by FH (K298Q). One mutation (R178Q) produced a partially cleaved protein with no ligand binding or functional activity. Seven genetic changes led to near-normal or only slightly reduced ligand binding and functional activity compared with the most common polymorphism at position 7, R7. Notably, none of the algorithms used to predict the disease relevance of FB mutations agreed completely with the experimental data, suggesting that in silico approaches should be undertaken with caution. These data, combined with previously published results, suggest that 9 of 15 FB genetic changes identified in patients with aHUS are unrelated to disease pathogenesis. This study highlights that functional assessment of identified nucleotide changes in FB is mandatory to confirm disease association.

J Am Soc Nephrol 25: 2053–2065, 2014. doi: 10.1681/ASN.2013070796

Atypicalhemolyticuremicsyndrome(aHUS)isa aHUS, giving good results in patients with and rare renal thrombotic microangiopathy disease without known complement genetic abnormality.4,5 characterized by mechanical hemolysis, platelet consumption, and acute kidney failure.1 aHUS is Received July 28, 2013. Accepted January 17, 2014. associated with genetic abnormalities in the pro- M.C.M. and L.V. contributed equally to this work. teins of the alternative complement pathway2 as – 3 Published online ahead of print. Publication date available at well as anti Factor H autoantibodies. The muta- www.jasn.org. tions may induce either loss of function of key complement regulators (such as Factor H [FH], Correspondence: Dr. Lubka T. Roumenina, Cordeliers Research Center, Institut National de la Santé et de la Recherche Médicale Factor I, or membrane cofactor protein [CD46]) UMRS 1138, Complement and Diseases Team, 15 rue de l’Ecole or overactivity of the C3 convertase components de Medecine, entrance E, ﬂ. 3, 75006 Paris, France. Email: lubka. C3 and Factor B (FB). Anti-C5 blocking antibody [email protected] Eculizumab has revolutionized the management of Copyright © 2014 by the American Society of Nephrology

J Am Soc Nephrol 25: 2053–2065, 2014 ISSN : 1046-6673/2509-2053 2053 BASIC RESEARCH www.jasn.org

Nevertheless, knowledge of the underlining genetic defect and its substrates (C3 and C5). The structures of most of these com- functional consequences is of importance for patients’ manage- plexes are known,22–24 and functional tests studying C3b ment, especially when the decision for transplantation should be binding and the activity of the C3 and C5 convertases are made.5–7 Routine screening of the aHUS susceptibility genes led available. to the identification of a large number of mutations and rare Four aHUS FB mutations resulted in the formation of either variants. In-depth functional analyses are required to predict hyperactive C3 convertase or a convertase unsusceptible to if the genetic change is related to the disease manifestation or complement regulation mechanisms.17,19 In contrast, two FB just a fortuitous association. The majority of tested mutations polymorphisms at position 7 (R7Q and R7W) were found to had, indeed, a functional defect.2 Nevertheless, five mutations be hypoactive compared with the most frequent FB variant found in aHUS patients—onemutationinFB,8 one mutation R7.25 The functional activity of FB, between R7 and R7Q var- in Factor I,9 two mutations in FH,10 and one mutation in iants, determines the arbitrary normal range.25 decay accelerating factor (CD55)11—were reported to lack a The objective of this study was to characterize the functional functional defect, and the link with complement dysregula- consequences of 10 newly identified FB genetic changes (Table tion for them remains unclear.12 Experimental verification of 1) and understand whether and how they are related to aHUS. functional consequences of a mutation is a time-consuming We sought to compare the in silico algorithms to predict the process. Therefore, there is a need to develop reliable in silico outcome of a given genetic change, searching for one that algorithms, validated with experimental data, to predict rap- shows a good agreement with experimental data. idly whether the discovered genetic change is pathogenic or not. FB mutations in aHUS are rare, with a frequency of 0%–4% RESULTS of the patients from different cohorts.13–16 To date, 15 FB genetic changes have been published8,13,15,17–21 (Table 1), FB Genetic Changes in the Normal Population but functional data are reported for only 5 of them.8,17,19 FB In the 1000 Genomes database,26 32 polymorphisms of FB forms the alternative pathway C3 and C5 convertases inter- have been reported (Supplemental Figure 1). Among them, acting with two partners (C3b and Factor D) and cleaves two three have been identified in aHUS patients (I217L

Table 1. Description of the identiﬁed FB mutations in aHUS patients Amino Acid No. of Presence of Other Variant at Cohort C3 Levels Source Change Patients Genetic Change Position 7 Mutations (previous studies) D254G France 2, family Low No RR/RW 19 F261L Spain/United 7 (family)+1 Low/NA 1, ADAMTS13, V832M rare NA/RR 17,21 States variant (United States) K298E Spain 1, de novo Low No NA 17 K325N France 1, de novo Low No RW 19 L408S Sweden/Italy 1+1 Low/low 1, FI p.G261D (Sweden) RR/RR 8,thisstudy Mutations (this study) R113W Italy 1 Low No RR 18 S141P United States 1 NA 1, FI p. Y369S RR 15 R178Q United States 1 Low No RR 15 K298Q United States/ 1+1 NA/low 1, MCP p. A304V, rare RR/NA 15,45 United Kingdom variant (United States) P344L France 1 Normal TM P501L, rare variant RQ 13 V430I France 1 Normal FI pY459S, rare variant RW 13,38,46 M433I United States 1 NA FH p.T956M, rare variant RR 15 Polymorphisms I217L France/United 1+1 Normal/NA No/1 FH p.Q950H, rare RR/RR 13,15 States variant K508R United States 3 Low, normal, NA 1, no NA, RR, RR 15,20 1, FI p. H138R 1, del CFHR3-CFHR1 E541A United States 1 NA 1, ADAMTS13 R7W variant NA 21 RR, homozygous for arginine (R); RW, heterozygous for arginine and tryptophan; NA, not available; FI, Factor I; MCP, membrane cofactor protein; TM, thrombomodulin; RQ, heterozygous for arginine and glutamine.

2054 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 2053–2065, 2014 www.jasn.org BASIC RESEARCH rs144812066, K508R rs149101394, and E541A rs80318145) produced partially cleaved, no binding to C3b was detected and designated as mutations in the original reports, because by these techniques. they were not found in the control groups. Their frequencies in the normal population are 0.001, 0.002, and 0.006, respec- C3 Convertase Formation and Dissociation tively (Table 2, Supplemental Figure 1). The previously de- Five different tests were used to monitor the formation and/or scribed mutation K298E17 was also reported in this database the dissociation of alternative pathway C3 and C5 convertases as an extremely rare variant without reported frequency (Figures 3 and 4). (rs121909748). The formation of (C3bBb-Mg2+) C3 convertase was assessed in real time by SPR (Figure 3) and the combination of Structure Description ELISA and Western blot27 (Figure 4A). The formation of the All identified mutations were mapped on the structures of C3/C5 convertase on cell surface (Figure 4, B–E) and the dis- FB,23 C3bBD,22 and C3bBb24 (Figure 1, Tables 2 and 3). Three sociation of the C3 convertase on cell surface by FH (Figure 4F) mutations (R113W, S141P, and R178Q) are in the Ba comple- were measured by hemolytic tests. The positive control in ment control protein domains. These residues are released these tests was the gain-of-function mutation D254G,17,19,28 with the Ba fragment after FB cleavage and absent in the active which indeed, induced stronger formation of C3 convertase convertase C3bBb. The rare variant I217L is located in the by SPR (Figures 3A and 4A) and enhanced erythrocytes lysis linker between Ba and Bb fragments near the R234–K235 (Figure 4B) compared with the WT and formed a convertase scissile bond. resistant to dissociation by FH (Figure 4F). Also, K298E, Five mutations are in the von Willebrand factor type A known to form a convertase resistant to decay by FH, was in- domain, which is responsible for the binding to C3b. V430Iand cluded as an additional control.17 The resistance to dissociation 1 M433I are close to the Mg2 adhesion site. One mutation by FH of this mutation was confirmed by the hemolytic assay. (K298Q) affects the same residue as the previously published The SPR test showed only very mild perturbation, potentially mutation K298E.17 P344L is far from any known binding site because of a weaker sensitivity of this assay. As expected, the and located in a flexible proline/tryptophane-rich loop. R7Q polymorphism was hypoactive in terms of formation of The last two variants (K508R and E541A) are in the serine the C3 and C3/C5 convertases (Figures 3A and 4, A and B), and protease domain of FB—on the same surface but not close to the C3 convertase was decayed normally (Figures 3A and 4F). the catalytic triad Asp551, His501, and Ser674. No functional activity was detected for the supernatant from mock-transformed cells. Expression of Recombinant FB The mutations could be classified in five groups according to All recombinant proteins (except one) were generated and their functional behavior (Table 2): (1) gain of formation of expressed successfully, with the expected molecular weight and the C3 convertase by SPR and ELISA/Western blot assays concentration similar to the wild-type (WT) (Supplemental (S141P, I217L, D254G, M433I, and K508R); (2) gain of for- Figure 2). Obtained functional results are summarized in Ta- mation of the C3/C5 convertase by hemolytic test (D254G and ble 2. R178Q was secreted as a partially cleaved protein by both M433I); (3) resistance of the convertase to decay by FH human embryonic kidney 293T and Chinese hamster ovary (D254G, K298E, and K298Q); (4) lack of activity (R178Q); cells (Supplemental Figure 2A). and (5) activity within the normal range (R113W, P344L, V430I, and E541A). Binding to C3b The interaction between recombinant FB and its ligand C3b Complement Activation on Endothelial Cells was assessed by ELISA and in real time by surface plasmon FACS analyses allowed estimation of the deposition of C3 resonance (SPR). The gain-of-function mutation D254G fragments on endothelial cells (human umbilical cord vein served as a positive control for increased binding. The binding endothelial cells [HUVECs]) and, thus, the alternative pathway of recombinant R7Q and R7W to C3b was about 40%–50% activation in FB-depleted serum reconstituted with recom- weaker than the WTas measured by both techniques (Figure 2). binant FB (Figure 5, Table 2). Reduced C3 deposition was The binding variation between WT R7 and R7Q (lowest bind- detected for R7Q and R7W compared with WT (R7), thus ing allotype) with C3b was considered as the normal range. The allowing for establishment of a normal range. ELISA analysis did not show significant increase of the binding On resting HUVECs, only the positive control D254G of any of the newly characterized mutations. The SPR interac- showed increased C3 deposition, which was similar to the one tion results (Figure 2) showed that only M433I led to a mild observed when a blocking anti-FH antibody was added to increase (1.38-fold increase) of C3b binding affinity (far below serum (Figure 4A). On activated and heme-exposed cells, in- the5.4-foldincreaseobservedwiththepositivecontrol creased deposition was measured for the blocking anti-FH D254G). S141P showed faster association but also faster disso- antibody, D254G, as well as K298E, K298Q, and M433I (Fig- ciation of the formed complex. The remaining mutations were ure 5, B and C). E541A showed increased C3 deposition on within the normal range as SPR curve profiles (Figure 2, B–E) activated HUVECs but not on heme-exposed HUVECs (Fig- and binding affinity (Figure 2F). For R178Q, which was ure 5, B and C). R178Q had no functional activity in these tests

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2055 2056 AI RESEARCH BASIC ora fteAeia oit fNephrology of Society American the of Journal www.jasn.org

Table 2. Summary of the functional consequences of studied FB mutations based on this study and previous reports SPR and Surface C3 Surface C3/C5 Amino Acid FB FB Integrity C3 Binding C3 Binding In-Plate C3 Convertase Resting Activated Heme-Exposed Convertase Phenotype Change Domain Fragment on WB (ELISA) (SPR) Convertase Decay by FH HUVECs HUVECs HUVEC (Hemolytic) Assembly (Hemolytic) R113W CCP1 Ba Intact ,WT ,WT NA ,WT NA ,WT ,WT ,WT Benign S141P CCP2 Ba Intact =WT .WT .WT ,WT NA ,WT ,WT =WT Partial R178Q CCP3 Ba Cleaved No binding No binding NA No NA ,WT ,WT ,WT Unclear I217L Linker Ba Intact =WT =WT .WT ,WT =WT ,WT ,WT ,WT Partial K298Q VWA Bb Intact =WT =WT NA =WT ,WT =WT .WT .WT Complete P344L VWA Bb Intact =WT ,WT =WT ,WT =WT ,WT =WT ,WT Benign V430I VWA Bb Intact ,WT ,WT NA ,WT =WT ,WT ,WT ,WT Benign M433I VWA Bb Intact =WT .WT NA .WT =WT =WT .WT .WT Complete K508R SP Bb Intact ,WT =WT .WT =WT =WT ,WT ,WT ,WT Partial E541A SP Bb Intact =WT =WT NA =WT =WT =WT .WT =WT Partial D254Ga VWA Bb Intact .WT .WT .WT .WT ,WT .WT .WT .WT Complete K298E VWA Bb Intact =WT =WT NA =WT ,WT =WT .WT .WT Complete F261Lb VWA Bb Intact NA .WT NA NA NA NA NA NA Complete K298Eb VWA Bb Intact NA =WT NA NA NA NA NA NA Complete K325Na VWA Bb Intact .WT .WT NA .WT ,WT .WT NA NA Complete L408Sc VWA Bb Intact ,WT ,WT NA =WT =WT =WT ,WT NA Benign The data for the SPR C3 convertase decay showed differences compared with WT only for D254G. Results are from this study only unless otherwise noted. WB, Western Blot; CCP, complement control protein; VWA, von Willebrand factor type A domain; SP, serine protease domain. aResults are from both this study and ref. 19. mScNephrol Soc Am J bResults are from ref. 17. cResults are from ref. 8. 25: 2053 – 05 2014 2065, www.jasn.org BASIC RESEARCH

(Supplemental Table 1) (a=0.6576). It shows only moderate agreement between the methods. Calculating Cronbach’s a for all combinations of the four methods shows that Mutation Taster’s disagreement with the rest is the greatest. Omitting it brings the criterion (a=0.8053) to a level indicating high consistency. Using PolyPhen, only 4 of 15 genetic changes were predicted to be damaging, contrary to Mutation Taster and Sorting Intolerant from Tolerant (SIFT), which predicted 8 of 14 and 10 of 15 changes as damaging, respectively. Align- Grantham variation (GV)/Grantham deviation (GD) re- turned 6 of 15 damaging scores. Although the GV/GD, SIFT, and interaction site involvement algorithms correlated well with the functional assay results, PolyPhen and Mutation Taster showed insignificant or no correlation (Table 4). Com- bining these results with the analysis of agreement led to the conclusion that Mutation Taster does not contribute substan- tial information. However, the other four methods seem to complement each other. To make use of this fact, a consensus score was defined as the number of methods scoring positive for each mutation (score=0–4). The performance of this score Figure 1. Visualization of the complex between C3b and FB and and the optimal cutoff for classifying the mutations were es- the positions of studied mutations. The three FB domains are timated using receiver operating characteristic curve analysis represented as follows: complement control protein domains in (Supplemental Figure 3). The area under the curve was 0.92 red, the von Willebrand factor type A domain in cyan, and the (95% confidence interval, 0.66 to 1.00; P,0.001). Optimal serine protease domain in green, with the corresponding new classification was achieved with a cutoff of consensus score.1 mutations in magenta and previously characterized mutations in fi blue. C3b is in gray. The approximate position of the common (sensitivity=85.7%; speci city=87.5%). This yielded one false polymorphism at position 7 (not present in the crystal structure) is positive case and one false negative case. Interestingly, the represented in orange by the first crystallized amino acid at po- consensus score outperformed the other methods and equaled sition 10. The residues forming the catalytic triad are depicted the binding site involvement (Table 4). with green spheres. Similar results were obtained for 17 mutations in the C-terminal 19 and 20 domains of FH, for which the functional defect was well characterized31–33 (data summarized in ref. 2) (Figure 5, A–C). The remaining mutations showed either nor- (Supplemental Table 2). mal or reduced C3 deposition but were always in the normal range. Frequency of R32Q and R32W in aHUS Patients and Terminal complement C5b-9 deposition can be measured Healthy Donors on apoptonecrotic HUVECs when incubated with normal The position 7-based alleles were studied for aHUS patients of sera.19,29 Therefore, apoptonecrotic HUVECs were used as a the French aHUS cohort and French healthy donors. Healthy model surface to study the activity of the alternative pathway donors presented the allele frequency expected for the Cau- C5 convertase in the presence of FB mutations. Because the Ba casian population (Table 5). When the aHUS patients allele fragment does not participate in the C5 convertase activity, frequencies were compared with healthy donor frequencies, only the mutations within the Bb part as well as I217L were there was no significant difference. tested (Figure 5D). The two known gain-of-function mutations (D254G and K298E) as well as K298Q and M433I resulted in a significantly stronger C5b-9 deposition, whereas the DISCUSSION remaining tested mutations showed levels similar to the WT. Identification of a genetic abnormality in aHUS patients is of Prediction for the Damaging Probability of FB importance for selection of a therapeutic regimen and trans- Mutations plantation decisions. The mutations in FH, C3, or FB, for Four different algorithms were used to calculate the probability which a clear functional consequence hasbeen identified,affect that each mutation will induce alterations of the protein the formation and/or the regulation of the C3 convertase. structure and function (Table 3). The relation of the mutated These defects translate to overactivity of the terminal com- position to an interaction site was used as an additional pre- plement pathway. As a result, the presence of FH mutations in dictor. The consistency of predictions by those five methods patients’ sera induces a spontaneous lysis of sheep erythro- was measured using the Cronbach’s a-coefficient30 cytes,34,35 and the presence of FB or C3 mutations causes

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2057 2058 AI RESEARCH BASIC ora fteAeia oit fNephrology of Society American the of Journal

Table 3. In silico analyses for the FB genetic changes and their correlations with patients’ C3 levels and observed functional defect www.jasn.org Numbering Presence in Other Variant Consensus Score Mutation Without 1000 Genomes of the Same Residue PolyPhen HumDiv SIFT GV/GD Interaction (PolyPhen+ C3 Consumption Functional Taster the Signal Database; Found in 1000 Score Score Score Site SIFT+GV/GD+ in Patients Defect Score Peptide Frequency Genomes; Frequency Interaction Site) Complete phenotype D254G D254E; 0, 000 Benign Deleterious 65 Disease-causing Yes 1 Yes Yes F261L Probably Deleterious 15 Disease-causing Yes 1 Yes Yes damaging K298E Yes; — Benign Deleterious 15 Polymorphism Yes 1 Yes Yes K298Q Benign Deleterious 15 Polymorphism Yes 1 Yes Yes K325N Probably Deleterious 65 Polymorphism Yes 1 Yes Yes damaging M433I Benign Deleterious 0 Disease-causing No 0 NA Yes R178Q Probably Deleterious 0 Disease-causing No 1 Yes Yes damaging Incomplete phenotype S141P Benign Tolerated 0 Polymorphism No 0 NA Partial I217L Yes; 0, 001 Benign Tolerated 0 Disease-causing No 0 No Partial K508R Yes; 0, 002 Benign Tolerated 0 Polymorphism No 0 No Partial E541A Yes; 0, 006 Benign Tolerated 0 Polymorphism No 0 NA Partial Within the normal range (benign) L408S Possibly Deleterious 65 Disease-causing No 1 Yes No damaging

mScNephrol Soc Am J R113W R113Q; 0, 001 Benign Deleterious 0 Disease-causing No 0 Yes No P344L P344A; — Benign Deleterious 0 Polymorphism No 0 No No V430I V430V; 0, 018 Benign Tolerated 0 Disease-causing No 0 No No In the GV/GD column, 0 indicates no predicted defect, and a number$15 predicts functional defect. In the semantics of Mutation Taster, polymorphism means mutation without predicted functional signiﬁcance. In the consensus score, one indicates that two or more algorithms suggested a functional defect, and zero indicates that less than two algorithms predicted affected functions. 25: 2053 – 05 2014 2065, www.jasn.org BASIC RESEARCH

to ESRD in the majority of the cases.13,15,17–19,21 Therefore, the presence of a FB mutation is considered as a bad prog- nosis for the disease. C3 levels are frequently below the normal range in these patients, suggesting active complement consumption in the plasma caused by C3 convertase overactivity. The aim of thiswork was to understand if this phenotype is shared between all identified FB genetic abnormalities and find out whether other FB mutations could be causative factors for aHUS. We used 10 different binding and functional tests for the formation of the C3 and C3/C5 convertases and the resistance of the C3 convertase to dissociation by FH to classify 10 previously uncharacterized aHUS-associated genetic changes as potentially disease-causing or benign. We confirmed the previous reports that D254G and K298E are gain-of-function mutations forming overactive C3 convertase and/or convertase resistant to decay by FH.16,18 A similar phenotype is shared by two newly characterized mutations (M433I and K298Q). K298Q (removal of positive charge) had a milder effect compared with K298E (exchange of a positive charge with a negative charge). This result suggests that the positive charge at this position is of critical importance for the competition be- Figure 2. Binding of FB to its ligand C3b. (A) As tested by ELISA in the presence of tween FH and FB, which is in agreement Mg2+. The values are reported as fold of the WT, and the normal range (between WT with the structural data.36 In fact, all gain- and the R7Q polymorphism) is depicted with dashed and continuous lines (n=3–6). of-function mutations appeared to be di- – 2+ (B E) As tested by SPR in the presence of Mg . The binding of the mutant FB to C3b rectly located within the C3b binding site immobilized to a biosensor chip (ProteOn XPR36) was measured. (B) WT was com- (D254G and K325N) or buried beneath it pared with the normal polymorphisms W7 and Q7 and the known gain-of-function (F261L) or located in the area important mutation D254G. (C–E) The mutations tested in this study are compared with the WT. One representative experiment of three performed is shown. (F) The binding affinity for the electrostatic repulsion between FB and FH. In agreement with the experimen- (KA) represented as fold difference compared with WT (1.0). Of note, all mutations are introduced on a WT FB with R at position 7. *P,0.05; **P,0.001, Mann–Whitney tally determined functional consequences, nonparametric test calculated with GraphPad Prism 5. CCP, complement control C3 levels were low in all patients with these protein; SP, serine protease; vWF, von Willebrand factor. mutations for which data were available, indicating alternative pathway consump- enhanced C5b-9 deposition of apoptonecrotic endothelial tion in vivo. Therefore, low C3 levels in the patient together cells.19,29 The pathogenic role of C5 convertase overactivation with a mutation localized in an important interaction site are a also explains the therapeutic efficacy of Eculizumab in aHUS. strong indication that the mutation may be a gain-of-function. Eculizumab leaves the overactive C3 convertase untouched Nevertheless, for three patients from this study (R113W, but blocks the C5 cleavage.5 Therefore, an FB mutation should R178Q, and K508R) and two unrelated patients with a pre- be defined as gain of function when it is able to form an over- viously published mutation (L408S), the low C3 levels could active C3/C5 convertase or a C3 convertase resistant to decay not be explained by a gain of function of the respective FB by the regulators that is strong enough to translate the func- mutation. tional defect to the level of C5 cleavage on a cell surface. Gain- The mutation R178Q showed complete lack of functional of-function FB mutations were reported in 12 patients from activity and was produced as a cleaved recombinant protein. five families and are associated with a severe outcome, leading Unfortunately, no plasma samples were available from this

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2059 BASIC RESEARCH www.jasn.org

Four genetic changes (S141P, I217L, K508R, and K541A) had an incomplete phenotype. One or more performed tests showed increased binding or function, but the results were inconsistent between the different approaches. Moreover, neither of these mutations resulted in concomitant increase of the C3 fragments deposition on activated and heme-exposed endothelial cells and enhanced C5b-9 deposition on apoptonecrotic cells. Therefore, they do not fulfill the criteria set to consider a genetic change as a gain of function. Among the carriers of these genetic changes for whom data are available, only one patient, with K508R, had low C3 levels. Another patient with K508R had normal C3 levels, suggesting that this genetic change is not necessarily responsible for the C3 consumption. There- fore, the functional consequence of these genetic changes, if any, is mild and less likely to predispose directly to the aHUS pathogenesis. In addition, in two of these patients, additionalFactorImutationswereidentified, which could be the disease-predisposing factor rather than the genetic change in FB. When an identified genetic change in a cohort is not found in the corresponding healthy controls, 1000 Genomes and other databases for genetic variants should be consulted to exclude that the identified genetic change is just a rare variant un- Figure 3. Formation and dissociation of the C3 convertase by FH studied in real time by SPR. Recombinant mutant FB and plasma-purified Factor D (FD) were mixed in an related to the disease, which seemed to be Mg2+-containing buffer and injected over a C3b-coated biosensor chip. After 540 the case of I217L, K508R, and E541 de- seconds of spontaneous dissociation, FH was injected to dissociate the convertase. scribed here. The incomplete phenotype The binding of FH to immobilized C3b was measured in parallel and subtracted from observed for these polymorphisms suggests the signal obtained in the channels with the convertase. In every run, four mutants that they may represent risk factors affect- were compared with the WT, and the sixth channel served as a reference for the FH ing the aHUS penetrance and severity but binding. (A) The formation and dissociation of the convertase formed with the positive cannot be considered as disease-causing control D254G and the R7Q polymorphism were compared with the WT. (B–E) The mutations. Studies with large aHUS cohorts remaining FB mutations were tested. and ethnically matched healthy donors are needed to establish if the frequency of these patient to test if it happens in vivo or is an in vitro artifact. It polymorphisms is increased in the patients compared with could be hypothesized that intrinsic propensity for cleavage healthy controls. could facilitate the in vivo transition between C3 proconvertase The fact that a mild reduction of C3b binding and FB C3bB and active convertase C3bBb, which would explain the C3 functional activity was detected for certain mutations is in- consumption found in the patient. triguing. The two normal variants, Q7 and W7, are reported25 Three genetic changes (R113W, P344L, and V430I) had and confirmed here to be hypoactive compared with R7. either normal or slightly reduced C3b binding and convertase Therefore, we sought to investigate if FB hypoactivity is as- formation; the formedconvertase wasdecayed normally byFH, sociated with aHUS. No difference was found in the frequency and the C3 fragment deposition on endothelial cells was in the of R7, Q7, and W7 in the French cohort of adult aHUS pa- normal range. This benign phenotype could be because of their tients and French healthy volunteers, suggesting that the FB localization distant from known binding sites and explains hypoactivity is not associated with this disease. Moreover, why, in two of three cases, they were found in patients with these polymorphisms did not differ among aHUS patients normal C3 levels. and healthy donors in a study of 501 complement-related

2060 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 2053–2065, 2014 www.jasn.org BASIC RESEARCH

the hypoactive variants is small and not at the level that we would expect to cause aHUS on its own. Based on our experimental data, it could be concluded that eight newly characterized FB genetic changes are most likely benign variants rather than disease-associated mutations. Interestingly, in 10 of 28 patients, the FB genetic changes were associated with rare variants or mutations without identified functional consequences in the other tested complement genes or ADAMTS13 (a disintegrin and metal- loproteinase with a thrombospondin type 1 motif, member 13). At the opposite of our series, most of the aHUS patients in a recently described subgroup with combined mutations have at least one func- tionally relevant genetic change.38 This information is critical for analyses of the phenotype of patients and prediction of the survival or the outcome of a particular therapeutic regimen. Functional studies are frequently un- available to inform physicians of the pos- sible relation of a newly identified genetic change with the disease. Therefore, bioin- formatics tool(s) could be useful to get a prediction for the mutations role and guide the treatment decision. Unfortunately, the four prediction algorithms tested here agreed for only about one third of the mutations and gave discordant data for the rest. Performed statistical analyses suggested that none of these algorithms could be used confidently as a stand-alone approach Figure 4. Functional activity of FB mutations. (A) C3 convertase formation assay based because of incomplete agreement with the on ELISA/Western blot. C3 convertases C3bBb were formed in C3b-coated microtiter experimental data. The prediction effi- wells by incubation with FB and FD in the presence of Mg2+. C3 convertase formation was evaluated by the visualization of the Bb band. The intensity of the Bb band was ciency was improved when the worse- quantified, and the results are presented as percentage compared with the WT (n=2). performing method (Mutation Taster) (B–E) Hemolytic test for the functional activity of the cell surface-bound C3/C5 con- was excluded, and a consensus score was vertase formed by recombinant FB proteins. C3b-covered sheep erythrocytes are calculated for the remaining three methods incubated with increasing concentrations of FB and a fixed FD concentration. The (PolyPhen, SIFT, and Align-GV/GD). Add- number of lytic sites per cell (Z) was revealed by the addition of a source of terminal ing to this score the information onwhether complement components (n=2–5). (F) Hemolytic test for the FH-induced dissociation the genetic change falls within a known of the cell surface-bound C3 convertase (C3bBb-Ni) formed by the WT or mutant FB functional site (i.e., the area of the molecule on the surface of C3b-covered sheep erythrocytes. The results are presented as important for ligands binding or func- percentages of the lysis obtained in the absence of FH (Z approximately 2; i.e.,after tional activity) substantially improves the spontaneous dissociation of the formed C3 convertase; n=2–4). prediction. Nevertheless, the bioinformat- single-nucleotide polymorphisms in 47 complement genes.37 ics evaluation of the potential functional effect of a mutation Of note, R7 is the most frequent polymorphism at this posi- should be taken with caution, because even the improved score tion, most of the FB mutations in the patients are found on R7 proposed here gave one false positive and one false negative background, and the recombinant WT used in this study car- result for 15 analyzed FB mutations. Similar results were ob- ries R at position 7. The observed magnitude of function tained for the mutations in the C terminus of FH, well known change of R113W, P344L, L408S, and V430I compared with for their functional defect and association with aHUS.2,31–33

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2061 BASIC RESEARCH www.jasn.org

Figure 5. Complement activation on endothelial cell surface. C3 deposition on (A) resting, (B) heme-exposed, or (C) TNFa/IFNg- activated HUVECs incubated with FB-depleted serum, reconstituted with the recombinant FB, and measured by ﬂow cytometry (n=3– 6). The C3 deposition from the WT was taken as one, and the values obtained for the remaining mutations were expressed as a fold difference compared with the WT. The supernatant from mock-transformed cells (white bar) is used as the negative control, and D254G and WT+FH blocking antibody Ox24 (hatched bars) are used as positive controls. (D) C5b-9 formation on apoptonecrotic HUVECs. After incubation with FB-depleted serum reconstituted with recombinant FB, the cells were probed for C5b-9 deposition by ﬂow cytometry (n=3). The C5b-9 formation from the WT was taken as one and compared with the tested mutations. *P,0.05; **P,0.001, Mann–Whitney nonparametric test calculated with GraphPad Prism 5.

In conclusion, although some FB mutations induce com- necessary in all patients with identiﬁed FB mutations to plement overactivation and predispose to aHUS, others have a identify other abnormalities affecting the disease occurrence, normal functional activity and hence, are benign and unlikely especially in the presence of low C3 levels. The status of the to be related to the disease. In-depth screening of all suscep- genetic change identiﬁed in aHUS needs to be carefully tibility genes as well as for the presence of anti-FH antibodies is analyzed to prove the pathogenic role in the disease. We

2062 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 2053–2065, 2014 www.jasn.org BASIC RESEARCH

Table 4. Contingency table signiﬁcance and correlation classify the mutations as deleterious. All calculations were performed between different algorithms and experimental data using MedCalc v12 software. Fisher’s Exact Correlation Algorithms Test, P Value Rho P Value Recombinant FB Production and Characterization The plasmid-containing recombinant FB construct and the muta- PolyPhen NS 0.343 0.21 genesis protocol were described previously.19 Proteins were expressed SIFT 0.03 0.661 0.01 GV/GD 0.04 0.600 0.02 in human embryonic kidney 293T cells after transfection with Gene Mutation Taster NS 0.071 1.0 Juice lipotransfection reagent (Calbiochem). Serum-free superna- Functional site 0.01 0.756 0.001 tant, containing recombinant proteins, was harvested after 3 days Consensus 0.01 0.732 0.002 of culture. A detailed description is given in Supplemental Material. NS, not signiﬁcant. The integrity of recombinant WT and FB mutants was tested by Western blot (precasted gels and rapid transfer system iBlot by Invitrogen). Proteins in the culture supernatants were separated by Table 5. Frequencies of FB variants R7, Q7, and W7 in SDS-PAGE, transferred to a nitrocellulose membrane, and probed French aHUS patients and normal controls with an in-house biotinylated polyclonal sheep anti-human FB Genotype RR WW QQ RQ RW QW antibody (Abcam, Inc.) followed by streptavidin–horseradish peroxy- Frequency, % n (%) n (%) n (%) n (%) n (%) n (%) dase (Amersham) and ECL substrate (GE Healthcare). Normal 64 (57) 1.1 (1) 0 (0) 16.9 (15) 12.4 (11) 5.6 (5) The FB content was assessed by sandwich ELISA using immobi- donors (n=89) lized polyclonal sheep anti-human FB antibody (Abcam, Inc.) for Patients 51.5 (86) 1.8 (3) 3.6 (6) 13.8 (23) 25.1 (42) 4.2 (7) capturing and biotinylated sheepanti-humanFBfollowedby (n=167) streptavidin–horseradish peroxydase (Dako) for detection as de- Allele Frequency, % R Q W scribed previously.19 Plasma-derived FB (Comptech, Tylor, TX) Normal donors (n=178) 78.7 11.2 10.1 served as a standard. Patients (n=334) 71 12.3 16.5 R, arginine; W, tryptophane; Q, glutamine. C3b Binding Characterization The ELISA approach for C3b–FB interactions was performed as previously described.19 suggest that rare variants in which frequency is not increased in The interaction of WT and mutant FB with C3 was also analyzed the patients cohorts and mutations without functional con- using SPR technology with ProteOn XPR36 equipment (Bio-Rad). sequences should not be included in the patient management. C3b was coupled to the GLC biosensor chip using standard amide- coupling technology. The recombinant WTand mutant FB were used as an analyte at concentrations 650, 325, 162.5, 81.25, and 40.625 pM CONCISE METHODS and injected at 50 ml/min in Mg2+-containing Hepes buffer (10 mM

Hepes [pH 7.4], 50 mM NaCl, and 10 mM MgCl2) over the C3b- In Silico Analyses bearing surfaces and an empty activated/deactivated flowcell as a con- PyMol (http://www.pymol.org/) and Chimera software39 (http:// trol. Data were analyzed using ProteOn Manager software, and the www.cgl.ucsf.edu/chimera) were used for structural data analyses. data from the blank flowcell were subtracted. Kinetic parameters were The probability that a genetic change induces an alteration of the pro- calculated by fitting the obtained sensorgrams into two state interac- tein structure and function was calculated with the following software: tion models as described.25 PolyPhen (http://genetics.bwh.harvard.edu/),40 Marinozzi_et_al_final. doc, Align-GV/GD (http://agvgd.iarc.fr/),41 Mutation Taster (http:// FB Functional Assays www.mutationtaster.org/),42 and SIFT (http://sift.jcvi.org/).43 The FB hemolytic activity and dissociation of the C3 convertase by FH were 1000 Genomes database26 was searched for the presence of each iden- tested as described previously.19 In addition, C3 convertase (C3bBb- tified FB genetic change to confirm the mutation status. Mg2+) formation assays, based on SPR17,25 or ELISA/Western blot, The pairwise contingency tables significance was determined using were performed to assess the convertase formation by different ap- Fisher’s exact test (P.0.05), and the correlation of the methods was proaches.27 A detailed description is given in Supplemental Material. estimated using the correlation coefficient. To test the internal consistency of the prediction by the battery of methods, the Cronbach’s Endothelial Cell Assay a-coefficient was calculated for all five methods together and all com- Resting, TNFa/IFNg-activated, heme-exposed, and apoptonecrotic binations of four methods to determine the contribution of each one HUVECs were prepared as described19,29,44 and incubated with 50 ml by omitting it. A consensus method was proposed based on the num- FB-depleted serum (CompTech) and 100 ml recombinant WTor mu- ber of methods predicting functional effects of each mutation. Its tant FB supernatants containing equal amounts of FB. Cells were performance was tested against the in vitro-determined functional labeled with anti-C3c (Quidel, San Diego, CA), anti-C5b9 neoepitope effects using receiver operating characteristic curve analysis. Using (a gift from Paul Morgan, Cardiff, UK), mAbs, or a control mouse the Yuden J index, a cutoff was determined for the consensus score to IgG1 followed by phycoerythrin-labeled secondary antibody

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2063 BASIC RESEARCH www.jasn.org

(Beckman Coulter, Roissy, France) and analyzed by flow cytometry Remuzzi G, Richard T, Sberro-Soussan R, Severino B, Sheerin NS, on a Becton Dickinson Facscalibur (Mountain View, CA) using Cell- Trivelli A, Zimmerhackl LB, Goodship T, Loirat C: Terminal complement Quest and FACS express software. inhibitor eculizumab in atypical hemolytic-uremic syndrome. NEnglJ Med 368: 2169–2181, 2013 5. Zuber J, Fakhouri F, Roumenina LT, Loirat C, Frémeaux-Bacchi V; Genetic Analyses French Study Group for aHUS/C3G: Use of eculizumab for atypical Genomic DNA from each patient from the French adult aHUS cohort haemolytic uraemic syndrome and C3 glomerulopathies. Nat Rev or healthy donors was obtained from peripheral blood leukocytes. The Nephrol 8: 643–657, 2012 frequencies of R7, Q7, and W7 (rs12614 and rs641153) were assessed 6. Le Quintrec M, Zuber J, Moulin B, Kamar N, Jablonski M, Lionet A, Chatelet V, Mousson C, Mourad G, Bridoux F, Cassuto E, Loirat C, by a direct sequencing of exon 2 of FB as described before for Rondeau E, Delahousse M, Frémeaux-Bacchi V: Complement genes fi 19 mutations identi cation. Screened individuals signed an informed strongly predict recurrence and graft outcome in adult renal transplant consent, and the project was approved by the local ethical committee recipients with atypical hemolytic and uremic syndrome. Am J Trans- according to the Declaration of Helsinki. plant 13: 663–675, 2013 7. Zuber J, Le Quintrec M, Sberro-Soussan R, Loirat C, Fremeaux-Bacchi V, Legendre C: New insights into postrenal transplant hemolytic uremic syndrome. Nat Rev Nephrol 7: 23–35, 2011 ACKNOWLEDGMENTS 8. Békássy ZD, Kristoffersson AC, Cronqvist M, Roumenina LT, Rybkine T, Vergoz L, Hue C, Fremeaux-Bacchi V, Karpman D: Eculizumab in an Part of the cytometric analysis was done at the Centre d’Imagerie anephric patient with atypical haemolytic uraemic syndrome and ad- vanced vascular lesions. Nephrol Dial Transplant 28: 2899–2907, 2013 Cellulaire et de Cytomètrie, Centre de Recherche des Cordeliers 9. Nilsson SC, Karpman D, Vaziri-Sani F, Kristoffersson AC, Salomon R, ’ UMRS 1138 (Paris, France). The Centre d Imagerie Cellulaire et de Provot F, Fremeaux-Bacchi V, Trouw LA, Blom AM: A mutation in factor I Cytomètrie is a member of the Université Pierre et Marie Curie Flow that is associated with atypical hemolytic uremic syndrome does not Cytometry Network. affect the function of factor I in complement regulation. Mol Immunol – This work was supported by grants from the Agence Nationale de la 44: 1835 1844, 2007 10. Tortajada A, Pinto S, Martínez-Ara J, López-Trascasa M, Sánchez-Corral Recherche (Grant 2009-2012 09geno03101I; Genopath), Assistance P, de Córdoba SR: Complement factor H variants I890 and L1007 while Publique-Hôpitaux de Paris (Programme Hospitalier de Recherche commonly associated with atypical hemolytic uremic syndrome are Clinique; Grant AOM08198), European Union FP7 (Grant 2012- polymorphisms with no functional significance. Kidney Int 81: 56–63, 305608; EURenOmics), Association for Information and Research on 2012 Genetic Renal Diseases (AIRG) France, and Institut National de la 11. Kavanagh D, Burgess R, Spitzer D, Richards A, Diaz-Torres ML, Santé et de la Recherche Médicale. S.B. received a fellowship from Goodship JA, Hourcade DE, Atkinson JP, Goodship TH: The decay accelerating factor mutation I197V found in hemolytic uraemic syn- Fondazione Aiuti per la Ricerca sulle Malattie Rare (ARMR) (Italy), drome does not impair complement regulation. Mol Immunol 44: M.N. was supported by Italian Telethon Grant GGP09075, and L.T.R. 3162–3167, 2007 was a recipient of European Molecular Biology Organization Long- 12. Kavanagh D, Anderson HE: Interpretation of genetic variants of un- Term Fellowship ALTF 444-2007. certain significance in atypical hemolytic uremic syndrome. Kidney Int – This work was presented as an abstract at the XXIV International 81: 11 13, 2012 13. Fremeaux-Bacchi V, Fakhouri F, Garnier A, Bienaimé F, Dragon-Durey Complement Workshop in Crete, Greece, October 10–15, 2012. MA, Ngo S, Moulin B, Servais A, Provot F, Rostaing L, Burtey S, Niaudet P, Deschênes G, Lebranchu Y, Zuber J, Loirat C: Genetics and outcome of atypical hemolytic uremic syndrome: A nationwide French series DISCLOSURES comparing children and adults. Clin J Am Soc Nephrol 8: 554–562, None. 2013 14. Kavanagh D, Kemp EJ, Richards A, Burgess RM, Mayland E, Goodship JA, Goodship TH: Does complement factor B have a role in the pathogenesis of atypical HUS? Mol Immunol 43: 856–859, 2006 REFERENCES 15. Maga TK, Nishimura CJ, Weaver AE, Frees KL, Smith RJ: Mutations in alternative pathway complement proteins in American patients with 1. Noris M, Remuzzi G: Atypical hemolytic-uremic syndrome. NEnglJ atypical hemolytic uremic syndrome. Hum Mutat 31: E1445–E1460, Med 361: 1676–1687, 2009 2010 2. Roumenina LT, Loirat C, Dragon-Durey MA, Halbwachs-Mecarelli L, 16. Westra D, Volokhina E, van der Heijden E, Vos A, Huigen M, Jansen J, Sautes-Fridman C, Fremeaux-Bacchi V: Alternative complement path- van Kaauwen E, van der Velden T, van de Kar N, van den Heuvel L: way assessment in patients with atypical HUS. J Immunol Methods 365: Genetic disorders in complement (regulating) genes in patients with 8–26, 2011 atypical haemolytic uraemic syndrome (aHUS). Nephrol Dial Transplant 3. Dragon-Durey MA, Sethi SK, Bagga A, Blanc C, Blouin J, Ranchin B, 25: 2195–2202, 2010 André JL, Takagi N, Cheong HI, Hari P, Le Quintrec M, Niaudet P, Loirat 17. Goicoechea de Jorge E, Harris CL, Esparza-Gordillo J, Carreras L, C, Fridman WH, Frémeaux-Bacchi V: Clinical features of anti-factor H Arranz EA, Garrido CA, López-Trascasa M, Sánchez-Corral P, Morgan autoantibody-associated hemolytic uremic syndrome. JAmSoc BP, Rodríguez de Córdoba S: Gain-of-function mutations in comple- Nephrol 21: 2180–2187, 2010 ment factor B are associated with atypical hemolytic uremic syndrome. 4. Legendre CM, Licht C, Muus P, Greenbaum LA, Babu S, Bedrosian C, Proc Natl Acad Sci U S A 104: 240–245, 2007 Bingham C, Cohen DJ, Delmas Y, Douglas K, Eitner F, Feldkamp T, 18. Noris M, Caprioli J, Bresin E, Mossali C, Pianetti G, Gamba S, Daina E, Fouque D, Furman RR, Gaber O, Herthelius M, Hourmant M, Karpman Fenili C, Castelletti F, Sorosina A, Piras R, Donadelli R, Maranta R, van D, Lebranchu Y, Mariat C, Menne J, Moulin B, Nürnberger J, Ogawa M, der Meer I, Conway EM, Zipfel PF, Goodship TH, Remuzzi G: Relative

2064 Journal of the American Society of Nephrology J Am Soc Nephrol 25: 2053–2065, 2014 www.jasn.org BASIC RESEARCH

role of genetic complement abnormalities in sporadic and familial 33. Manuelian T, Hellwage J, Meri S, Caprioli J, Noris M, Heinen S, Jozsi M, aHUS and their impact on clinical phenotype. Clin J Am Soc Nephrol 5: Neumann HP, Remuzzi G, Zipfel PF: Mutations in factor H reduce 1844–1859, 2010 binding affinity to C3b and heparin and surface attachment to endo- 19. Roumenina LT, Jablonski M, Hue C, Blouin J, Dimitrov JD, Dragon- thelial cells in hemolytic uremic syndrome. JClinInvest111: 1181– Durey MA, Cayla M, Fridman WH, Macher MA, Ribes D, Moulonguet L, 1190, 2003 Rostaing L, Satchell SC, Mathieson PW, Sautes-Fridman C, Loirat C, 34. Roumenina LT, Roquigny R, Blanc C, Poulain N, Ngo S, Dragon-Durey Regnier CH, Halbwachs-Mecarelli L, Fremeaux-Bacchi V: Hyperfunc- MA, Frémeaux-Bacchi V: Functional evaluation of factor H genetic and tional C3 convertase leads to complement deposition on endothelial acquired abnormalities: Application for atypical hemolytic uremic cells and contributes to atypical hemolytic uremic syndrome. Blood syndrome (aHUS). Methods Mol Biol 1100: 237–247, 2014 114: 2837–2845, 2009 35. Sánchez-Corral P, González-Rubio C, Rodríguez de Córdoba S, López- 20. Tawadrous H, Maga T, Sharma J, Kupferman J, Smith RJ, Schoeneman Trascasa M: Functional analysis in serum from atypical Hemolytic Ure- M: A novel mutation in the complement factor B gene (CFB) and mic Syndrome patients reveals impaired protection of host cells asso- atypical hemolytic uremic syndrome. Pediatr Nephrol 25: 947–951, ciated with mutations in factor H. Mol Immunol 41: 81–84, 2004 2010 36. Wu J, Wu YQ, Ricklin D, Janssen BJ, Lambris JD, Gros P: Structure of 21. Feng S, Eyler SJ, Zhang Y, Maga T, Nester CM, Kroll MH, Smith RJ, complement fragment C3b-factor H and implications for host pro- Afshar-Kharghan V: Partial ADAMTS13 deficiency in atypical hemolytic tection by complement regulators. Nat Immunol 10: 728–733, 2009 uremic syndrome. Blood 122: 1487–1493, 2013 37. Ermini L, Goodship TH, Strain L, Weale ME, Sacks SH, Cordell HJ, 22. Forneris F, Ricklin D, Wu J, Tzekou A, Wallace RS, Lambris JD, Gros P: Fremeaux-Bacchi V, Sheerin NS: Common genetic variants in com- Structures of C3b in complex with factors B and D give insight into plement genes other than CFH, CD46 and the CFHRs are not associ- complement convertase formation. Science 330: 1816–1820, 2010 ated with aHUS. Mol Immunol 49: 640–648, 2012 23. Milder FJ, Gomes L, Schouten A, Janssen BJ, Huizinga EG, Romijn RA, 38. Bresin E, Rurali E, Caprioli J, Sanchez-Corral P, Fremeaux-Bacchi V, Hemrika W, Roos A, Daha MR, Gros P: Factor B structure provides in- Rodriguez de Cordoba S, Pinto S, Goodship TH, Alberti M, Ribes D, sights into activation of the central protease of the complement system. Valoti E, Remuzzi G, Noris M; European Working Party on Complement Nat Struct Mol Biol 14: 224–228, 2007 Genetics in Renal Diseases: Combined complement gene mutations in 24. Rooijakkers SH, Wu J, Ruyken M, van Domselaar R, Planken KL, Tzekou atypical hemolytic uremic syndrome influence clinical phenotype. JAm A, Ricklin D, Lambris JD, Janssen BJ, van Strijp JA, Gros P: Structural Soc Nephrol 24: 475–486, 2013 and functional implications of the alternative complement pathway C3 39.PettersenEF,GoddardTD,HuangCC,CouchGS,GreenblattDM, convertase stabilized by a staphylococcal inhibitor. Nat Immunol 10: Meng EC, Ferrin TE: UCSF Chimera—a visualization system for ex- 721–727, 2009 ploratory research and analysis. JComputChem25: 1605–1612, 2004 25. Montes T, Tortajada A, Morgan BP, Rodríguez de Córdoba S, Harris CL: 40. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork Functional basis of protection against age-related macular de- P, Kondrashov AS, Sunyaev SR: A method and server for predicting generation conferred by a common polymorphism in complement damaging missense mutations. Nat Methods 7: 248–249, 2010 factor B. Proc Natl Acad Sci U S A 106: 4366–4371, 2009 41. TavtigianSV,DeffenbaughAM,YinL,JudkinsT,SchollT,SamollowPB, 26. Clarke L, Zheng-Bradley X, Smith R, Kulesha E, Xiao C, Toneva I, de Silva D, Zharkikh A, Thomas A: Comprehensive statistical study of Vaughan B, Preuss D, Leinonen R, Shumway M, Sherry S, Flicek P; 1000 452 BRCA1 missense substitutions with classification of eight recurrent Genomes Project Consortium: The 1000 Genomes Project: Data substitutions as neutral. J Med Genet 43: 295–305, 2006 management and community access. Nat Methods 9: 459–462, 2012 42. Schwarz JM, Rödelsperger C, Schuelke M, Seelow D: MutationTaster 27. Bettoni S, Donadelli R, Roumenina L, Fremeaux-Bacchi V, Noris M: A evaluates disease-causing potential of sequence alterations. Nat user-friendly method to study the effect of FH and impact of FB variants Methods 7: 575–576, 2010 on C3bBb convertase and C3bB proconvertase formation. Im- 43. Kumar P, Henikoff S, Ng PC: Predicting the effects of coding non- munobiology 217: 1034–1046, 2012 synonymous variants on protein function using the SIFT algorithm. Nat 28. Hourcade DE, Mitchell LM, Oglesby TJ: Mutations of the type A do- Protoc 4: 1073–1081, 2009 main of complement factor B that promote high-affinity C3b-binding. J 44. Frimat M, Tabarin F, Dimitrov JD, Poitou C, Halbwachs-Mecarelli L, Immunol 162: 2906–2911, 1999 Fremeaux-Bacchi V, Roumenina LT: Complement activation by heme 29. Roumenina LT, Frimat M, Miller EC, Provot F, Dragon-Durey MA, as a secondary hit for atypical hemolytic uremic syndrome. Blood 122: Bordereau P, Bigot S, Hue C, Satchell SC, Mathieson PW, Mousson C, 282–292, 2013 Noel C, Sautes-Fridman C, Halbwachs-Mecarelli L, Atkinson JP, Lionet 45.GilbertRD,FowlerDJ,AngusE,HardySA,StanleyL,GoodshipTH: A, Fremeaux-Bacchi V: A prevalent C3 mutation in aHUS patients Eculizumab therapy for atypical haemolytic uraemic syndrome due to a causesadirectC3convertasegainoffunction.Blood 119: 4182–4191, gain-of-function mutation of complement factor B. Pediatr Nephrol 28: 2012 1315–1318, 2013 30. Cronbach LJ, Warrington WG: Time-limit tests: Estimating their re- 46. Bienaime F, Dragon-Durey MA, Regnier CH, Nilsson SC, Kwan WH, Blouin liability and degree of speeding. Psychometrika 16: 167–188, 1951 J, Jablonski M, Renault N, Rameix-Welti MA, Loirat C, Sautés-Fridman C, 31. Ferreira VP, Herbert AP, Cortés C, McKee KA, Blaum BS, Esswein ST, Villoutreix BO, Blom AM, Fremeaux-Bacchi V: Mutations in components of Uhrín D, Barlow PN, Pangburn MK, Kavanagh D: The binding of factor H complement influence the outcome of Factor I-associated atypical he- to a complex of physiological polyanions and C3b on cells is impaired in molyticuremicsyndrome.Kidney Int 77: 339–349, 2010 atypical hemolytic uremic syndrome. J Immunol 182: 7009–7018, 2009 32. Lehtinen MJ, Rops AL, Isenman DE, van der Vlag J, Jokiranta TS: Mu- tations of factor H impair regulation of surface-bound C3b by three mechanisms in atypical hemolytic uremic syndrome. J Biol Chem 284: This article contains supplemental material online at http://jasn.asnjournals. 15650–15658, 2009 org/lookup/suppl/doi:10.1681/ASN.2013070796/-/DCSupplemental.

J Am Soc Nephrol 25: 2053–2065, 2014 aHUS Complement Factor B Mutations 2065 Complement factor B mutations in atypical hemolytic uremic syndrome – disease-

relevant or benign?

Supplemental Materials and Methods

In silico analysis

The crystal structures of FB, the C3 convertase C3bBb and the C3bBD triple complex are available in Protein Data Bank (PDB ID 2OK5; 2WIN and 2XWB). The residues affected by the mutations were visualized using PyMol and Chimera softwares. If a FB residue was in a distance of 1 Å or less from a residue from the partner molecule, it was considered to belong to the binding site. Protein numbering throughout this study is according to the sequence of the mature protein, without signal peptide. Wherever appropriate, the numbering according to the protein with leader peptide (starting p., having additional 25 residues) will be given for reference.

Recombinant FB production

Site-directed mutagenesis

The mutations were introduced in a FB gene containing plasmid by site-directed mutagenesis using the QuikChange® II XL Site-Directed Mutagenesis kit (Stratagene, La Jolla, CA), according to the manufacturer’s instructions. E. coli XL-Gold bacteria were transformed in order to amplify the quantity of plasmid, extracted thanks to a midi-prep technique (Qiagen).

The constructs were completely sequenced to confirm that no additional mutations had been introduced. The transient expression of recombinant FB proteins was conducted in transitory transfected HEK-293T cells cultured 3 days in DMEM+glutaMAX TM -I 4,5g/l D-Glucose + Pyruvate medium (Gibco, Paisley, GB) without foetal calf serum. A production test was performed also in CHO-K1 cells.

Recombinant FB characterization

The integrity of recombinant wild type (WT) and mutants FB was tested by Western blot

(ready made gels and rapid transfer system iBlot by Invitrogen). The proteins in the culture supernatants were separated by SDS-PAGE, transferred to a nitrocellulose membrane and probed with in house biotinylated polyclonal sheep anti-human FB antibody (Abcam), followed by Streptavidin-HRP (Amersham) and ECL (GE Healthcare).

The FB content was assessed by sandwich ELISA, using immobilized polyclonal sheep anti- human FB antibody (Abcam) for capturing and biotinylated sheep anti-human FB, followed by streptavidin-Horseradish Peroxydase (HRP) (Dako) for detection, as described previously

[9]. Plasma derived FB (Comptech, Tylor, TX) served as a standard.

C3b binding characterization

Enzyme-linked immunosorbent assay (ELISA)

Microtiter wells were coated with purified human C3b at 10 µg/ml in PBS (pH 7,4), for 1 hour at 37 °C. The residual binding sites were blocked by addition of BSA at 2% in PBS, during 1 hour at 37°C, and washed three times with Hepes buffer (10mM Hepes pH 7.4,

25mM NaCl, 10mM MgCl 2 and 0,05% Tween 20). Then, serial dilutions of the supernatants containing the recombinant FB were made in the same Hepes buffer containing 4% BSA and incubated in the plate during 1 hour at 37°C. Three washings were done with the same buffer and then the wells were incubated with a biotinylated anti-FB antibody (dilution 1:1000) for 1 hour at 37°C. After three more washings, the wells were incubated 1 hour with Streptavidin-

HRP (dilution 1:3000) at 37°C. Finally, the interactions were revealed using TMB substrate and the reaction was stopped by 2M H 2SO 4. The OD 450nm was measured.

Surface plasmon resonance (SPR)

The interaction of wild type and mutant FB with C3 was analyzed using surface plasmon resonance technology with ProteOn XPR36 equipment (BioRad). C3b was coupled to the

GLC biosensor chip, using standard amide-coupling technology, according to the manufacturer’s instructions. The recombinant wild type and mutant FB were used as an analyte at concentrations 650, 325, 162.5, 81.25 and 40.625 pM and were injected at 50

µl/min in Mg 2+ - containing HEPES buffer (10mM Hepes pH 7.4, 50mM NaCl, 10mM

MgCl 2,) over the C3b containing surfaces and an empty, activated/deactivated flowcell as a control. Data were analyzed using ProteOn Manager software and the data from the blank flowcell were subtracted. Kinetic parameters were calculated by fitting the obtained sensorgrams into two state interaction model.

FB functional assays

FB hemolytic activity

Briefly, samples made from 100 µl 1x10 8 C3b-covered sheep erythrocytes/ml, 40ng of purified human factor D (Comptech) and recombinant Factor B proteins (serial dilutions starting from

10 µg/ml, recombinant WT or mutant forms) in DGVB containing CaCl 2 and MgCl 2were incubated at 30°C for 30 min. The negative control did not contain factor B. The alternative pathway C3 convertases sites were developed with 300 µl of a 1/40 dilution of rat serum in

GVB-EDTA buffer at 37°C for 45 min. Hemolysis was detected at OD 414nm. Hemolytic activity levels for the factor B mutants were expressed as Z values, which is equivalent to the average number of lytic sites per cell.

C3-convertase (C3bBb-Mg 2+ ) formation assay based on ELISA/western blot

One mutation per domain – S141P for the CCP domains, I217L for the linker, P344L for the vWF domain and K508R for the SP domain were selected for assessment of the convertase formation by a different assay.

The assay was performed as reported. Microtiter wells were coated with 3 ug/mL C3b

(ComplTech) in PBS by overnight incubation at 4°C, blocked with 1% BSA for 1h at 37°C, and washed with a phosphate buffer supplemented with 10mM MgCl 2. C3bBb complexes were formed by incubating C3b coated wells for 12 min at 25°C with FB (ComplTech; 1000 ng/mL) and FD (ComplTech; 5 ng/mL) both diluted in a phosphate buffer supplemented with

0.5% BSA and 10mM MgCl 2. After wash, the protein complexes were detached from microtiter wells with EDTA 10 mM and SDS 1%, subjected to 10% SDS-PAGE, and transferred by electroblotting to PVDF membrane (Amersham). Proteins were detected with rabbit anti-human CFB antibody (Atlas; 1:500) followed by HRP anti-rabbit antibody (Vector

Lab.; 1:30000) and the ECL system (Amersham). C3 convertase formation was evaluated by the visualization of the Bb band (60 KDa) and the intensity of the band was quantified by

ImageJ. Sample with WT FB was used as positive control and results expressed relative to it:

% of C3bBb formation = 100 X (Bb band with variant FB sample – background)/(Bb band with WT FB sample – background).

Dissociation of the C3 convertase by FH

C3bBbconvertases were assembled on sheep erythocytes using recombinant WT or mutant factor B (1

(without plasma).

Endothelial cells assay

HUVEC were isolated from human umbilical cord veins and cultured in M199 medium

(Gibco, Paisley, GB) with 20% fetal calf serum, supplemented with endothelial cells growth supplement and heparin. Third passage resting cells were grown to confluence into 24-well plates. When activated cells were used, the wells were treated overnight with TNF α/IFN γ.

Alternatively, the cells were exposed to 100 µM heme, as described (27). The overnight detached HUVEC were collected and shown to be late apopto-necrotic cells (Annexin V and propidium iodide positive). After washing with PBS, adherent cells were incubated with 50 µl

FB-depleted serum, (CompTech, Tylor, TX), and with 100µl recombinant WT or mutant FB supernatants, containing equal amount of FB. Used lots of FB-depleted serum did not contain detectable residual FB, as measured by ELISA but had normal C3 levels. Supernatant from

HEK293T cells transfected with the vector alone (SN0) was used as a negative control.

Blocking anti-FH antibody Ox24 was added in the FB depleted serum, reconstituted with the

WT FB as a positive control for complement dysregulation. After 30-minute incubation at

37°C, the supernatant was discarded and the wells were washed 3-fold with PBS. Adherent cells were detached by incubation with PBS-5mM EDTA for 30 minutes. The resting, activated and apopto-necrotic cells were labeled with anti-C3c, a monoclonal antibody

(Quidel, San Diego, CA, USA), anti-C5b9 neoepitope antibody (kind gift form Prof. Paul

Morgan, Cardiff, UK) or a control mouse IgG1, followed by phycoerythrin (PE)-labeled secondary antibody (Beckman Coulter, Roissy, France). Cells were analyzed by flow cytometry on a Becton Dickinson Facscalibur or LSRII, using FCS express software.

Genetic analysis

Genomic DNA from each patient from the French adults aHUS cohort or healthy donor was obtained from peripheral-blood leukocytes. The frequency of R7, Q7 and W7 (rs12614 and rs641153) was assessed by a direct sequencing of exon 2 of FB. Screened individuals signed an informed consent and the project was approved by the local ethical committee. Supplemental Table 1. Cochran’s alpha coefficient was used to test the internal

consistency of the available algorithms for prediction of functional effects of mutations

using FB.

Cronbach's alpha with raw variables

Cronbach's alpha 0.6576

95% lower confidence limit 0.3589

Effect of dropping variables

Variable dropped Alpha Change

BS 0.4444 -0.2132 GV/GD 0.3944 -0.2633 MT 0.8053 0.1477 PolyPhen 0.6603 0.002607 Sift 0.5693 -0.08836

Supplemental Table 2. In silico analysis for the FH mutations in SCR19-20.

mut FH mut FH position PolyPhen Polyphen SIFT Mut taster GV/GD Prediction c.3356 1.000 Probably G1119D Asp119Gly A>G Damaging Deleterious Disease causing 65 c.3469 1.000 Probably W1157R Trp1157Arg T>A/C Damaging Deleterious Disease causing 65 c.3514 -- Damaging E1172X Glu1172Stop G>T Deleterious Disease causing na c.3546 0.826 Possibly R1182S Arg1182Ser G>C Damaging Deleterious Polymorphism 35 c.3548 0.998 Probably W1183L Trp1183Leu G>T Damaging Deleterious Polymorphism 55 c.3551 0.000 Benign T1184R Thr1184Arg C>A Tolerated Polymorphism 0 c.3566 0.994 Probably L1189R Leu1189Arg T>G Damaging Deleterious Polymorphism 65 c.3565 0.278 Benign L1189F Leu1189Phe C>T Tolerated Polymorphism 0 c.3572 0.151 Benign S1191L Ser1191Leu C>T Deleterious Polymorphism 65 c.3590 0.962 Possibly / V1197A Vla1197Ala T>C Probably D. Deleterious Disease causing 25 c.3581 0.405 Benign G1194D Gly1194Asp G>A Tolerated Polymorphism 0 c.3593 0.724 Possibly E1198A Glu1198Ala A>C Damaging Deleterious Polymorphism 65 c.3628 0.024 Benign R1210C Arg1210Cys C>T Tolerated Disease causing 15 c.3611 0.997 Probably G1204E Gly1240Glu G>A Damaging Deleterious Polymorphism 65 c.3643 1.000 Probably R1215G Arg1215Gly C>G Damaging Deleterious Polymorphism 65 c.3644 0.994 Possibly / R1215Q Arg1215Gln G>A Probably D. Deleterious Polymorphism 35 c.3676 1.000 Probably P1226S Pro1126Ser C>T Damaging Deleterious Disease causing 65

(A)

CCP CCP CCP Von Willebrand – type A Serine Protease domain 1 2 3 domain R7 W7 Q7

F261 K325N

D254 K298E K298Q (B)

CCP CCP CCP Von Willebrand – type A Serine Protease domain 1 2 3 domain

R113W R178 P344 V430I S141 M433I K508R I217L E541A

227G,S,D 2.52G,E 254D,E 301E,D 340M,I 371G,R 264A,V 309W,C 344P,A 390R,H 266K,N 323K,M 484E,K 354R,H 298K,E 332K,T,E 355T,P 626D,E 444I,M 540K,E 704K,R 8P,A 93S,A 167S,G 552V,I 644 V,L 49R,H 113R,Q 207E,K (C)

CCP CCP CCP Von Willebrand – type A Serine Protease domain 1 2 3 domain

Supplemental Figure 1 : Schematic representation of the Factor B gene. (A) Frequent polymorphisms in the healthy population, (B) FB mutations and rare polymorphisms (light blue box, italic) found in aHUS patients, (C) ultra rare polymorphisms found in the healthy controls (1000 genomes database).

(A) (A1)

MW P344L I217L V430I D254G K508R S141P R113W R7Q R7W SN0 WT 98

WT K298E K298Q M433I E541A 98

MW (A2) WT R178Q 98

(B) Recombinant FB production 200

150

100 % WT of 50

0 I I G P Q L 0 WT 7Q 4 3W 7L 8E 8 4 1A R R7W SN0 1 433 11 I2 29 V43 M D25 R S141R178Q K K29 P34 K508RE54

Supplemental Figure 2: Integrity of FB mutated recombinant proteins and production levels. A) Western blot. Supernatants from each cell culture were diluted 1/10 and migrated on 10% PAGE gels. The presence and integrity of the recombinant FB was detected by an anti-FB polyclonal antibody. B) Production level of the recombinant FB mutations, tested by sandwich ELISA, using plasma derived FB as a standard. The values of the WT in each production were taken as 100% and the corresponding levels of the mutations expressed at the same time were calculated.

Consensus 100

80 Sensitivity: 85.7 Specificity: 87.5 Criterion : >1 60

40 Sensitivity

0 0 20 40 60 80 100 100-Specificity

Supplemental Figure 3. ROC curve analysis of performance of the consensus score. The number of methods (among Polyphen, Sift, GDGV and functional site involvement) that predict a given mutation as deleterious was used as a consensus score. Mutation taster was omitted due to a disagreement with the rest of the methods (Cronbach’s alpha analysis) and poor correlation with the functional assays. The optimal cut off was determined by determining the Yuden index J and the corresponding criterion value (marked by the inset). The 95% Confidence Intervals are shown with a dashed line.