Autoantibodies against C1q strongly correlate with the occurrence of severe nephritis in patients with systemic lupus erythematosus (SLE). We undertook this study to determine whether identification of the C1q epitope(s) recognized by these autoantibodies might lead to a better diagnostic assay and help elucidate the putative role of C1q and anti-C1q in SLE.
SLE patient–derived anti-C1q Fab were used in a microarray-based peptide scan to identify the peptide sequence recognized by anti-C1q. Anti-C1q Fab binding to the target peptide was further analyzed using real-time interaction measurements (surface plasmon resonance) and peptide-based enzyme-linked immunosorbent assays (ELISAs).
A peptide scan of the collagen-like region of C1q identified 2 regions, 1 on the A chain and 1 on the B chain, that were the targets of the anti-C1q Fab. Binding was confirmed by surface plasmon resonance and showed nanomolar affinity. The A chain–derived peptide could specifically be detected in a peptide-based ELISA by SLE patient sera. Competition experiments suggested that this peptide represented one of the major linear epitopes of C1q that is the target of anti-C1q in SLE. Serum antibodies from most SLE patients but not from healthy individuals specifically bound to this epitope. Binding to the peptide correlated with binding of the same sera to native C1q but was found to be more sensitive for the detection of lupus nephritis.
We identified a major linear epitope of C1q that is the target of anti-C1q in SLE. The ELISA using this peptide was more specific and more sensitive than a conventional anti-C1q assay for the detection of active nephritis in SLE patients.
Systemic lupus erythematosus (SLE) is a systemic autoimmune disease that is characterized by the occurrence of autoantibodies to a number of self antigens, resulting in a broad spectrum of clinical and immunologic manifestations. One of these self antigens is C1q, the first molecule of the classical complement pathway; ∼20–50% of unselected SLE patients have antibodies directed against C1q (anti-C1q).
The importance of C1q in the pathogenesis of SLE is underscored by the observation that loss-of-function mutations in C1q, although very rare, are sufficient to cause SLE in >90% of affected individuals (1–3). This makes deficiency in C1q the strongest known susceptibility factor for SLE. In addition, low levels of C1q and other complement factors of the classical pathway are associated with active disease and the appearance of renal involvement in non–C1q-deficient SLE patients. Reduced serum C1q levels in SLE patients due to increased consumption or neutralization are possibly linked to the presence of anti-C1q autoantibodies (4, 5). Active nephritis in lupus patients has been found to be strongly associated with high levels of anti-C1q autoantibodies (6–15). However, not all patients with high levels of anti-C1q develop lupus nephritis, suggesting that anti-C1q antibodies are necessary but not sufficient for the development of proliferative lupus nephritis. Some studies only found a variable degree of association between anti-C1q levels and the occurrence of lupus nephritis (16). In part, this lack of association might be due either to too low sensitivity of the used assays or to false-positive test results caused by the binding of immune complexes to the globular heads of the immobilized C1q (17).
The C1q molecule is a multimeric glycoprotein composed of 18 polypeptide chains of 3 different types: C1q-A, C1q-B, and C1q-C. Each chain has a short N-terminal region followed by a collagen-like region (CLR) and a C-terminal globular head domain (18). To investigate the role of C1q and/or anti-C1q in the pathogenesis of SLE, our group previously isolated anti-C1q Fab from a phage display library, generated from bone marrow cells of an SLE patient (19).
Using these anti-C1q Fab, we now identify a C1q neoepitope as the main linear target of SLE anti-C1q antibodies. Serum antibodies from SLE patients but not from healthy individuals were found to bind specifically to this epitope, and the binding correlated with binding to native C1q.
In contrast to conventional anti-C1q enzyme-linked immunosorbent assays (ELISAs) where immune complexes in sera could obscure specific signals by binding to the C1q globular heads, a peptide ELISA based on the identified C1q neoepitope was found to be more specific and more sensitive for the detection of anti-C1q antibodies in SLE patient sera. Finally, the knowledge of a major C1q epitope might help to determine the potential role of C1q and/or anti-C1q in the pathogenesis of SLE.
MATERIALS AND METHODS
Serum antibodies and Fab.
Anti-C1q Fab from an SLE patient were generated as described (19). Control Fab were polyclonal Fab prepared from human IgG from healthy donors (GenWay Biotech).
Human serum/plasma was obtained from 61 SLE patients who fulfilled at least 4 of the American College of Rheumatology revised classification criteria (20) and from 72 healthy blood donors. The majority of patients (36 of 61 [59%]) had biopsy-confirmed active renal disease at the time of sampling, and another 13 patients (21%) had had lupus nephritis confirmed by renal biopsy in the past. Most patients with “active nephritis” had biopsy-confirmed class III and/or class IV nephritis. Four patients had class II nephritis and 2 patients had class V nephritis. The study was approved by the Ethics Committee of the University Hospital Basel and fulfilled the ethical guidelines of the most recent Declaration of Helsinki.
Anti-C1q Fab that were previously shown to bind to the CLR of the A and/or B chain of C1q (19) were now used to screen an overlapping peptide library representing CLR sequences of C1q-A and C1q-B, consisting of 95 and 97 amino acids, respectively. Peptide synthesis and Fab binding were performed by AbD Serotec using PepStar microarrays. Briefly, 13-mer peptides overlapping by 11 amino acid residues, resulting in 42 C1q-A plus 43 C1q-B peptides, were printed onto glass slides in triplicates. These microarray slides were incubated with anti-C1q Fab followed by a fluorescence-labeled secondary antibody. The Fab clones A4 and A14 were each analyzed on separate microarrays. The means of the triplicate fluorescence signals are expressed as relative signal intensities.
Biotinylated and nonbiotinylated peptides with >95% purity were synthesized by GenScript. Peptides A08 (GRPGRRGRPGLKG) and B78 (PGKVGPKGPMGPK) are derived from the C1q-A chain and the C1q-B chain, respectively. Peptide A08-C (GAPGKDGYDGLPG), which was derived from the C1q-C chain, is located at the N-terminal region homologous to peptide A08 and was used as a negative control peptide.
Measurement of binding of anti-C1q Fab to immobilized peptides by surface plasmon resonance.
The interaction of 2 anti-C1q Fab clones, A4 and A14, with 3 peptides (A08, B78, and an irrelevant control peptide) was analyzed using surface plasmon resonance technology with BIAcore 2000 equipment. Biotinylated peptides were bound to a sensor chip SA (GE Healthcare) with preimmobilized streptavidin by injecting 50 μl of a 0.5 mg/ml peptide solution into 3 different flow cells, and an equivalent level of immobilization was obtained (870 response units [RU] for A08, 886 RU for B78, and 910 RU for the negative control peptide). Purified Fab were used as analytes at concentrations of 170 nM, 85 nM, 42.5 nM, 21.3 nM, and 10.6 nM and were flowed at 10 μl/minute in phosphate buffered saline (PBS) (Gibco). The surface was regenerated with 0.5M NaCl, 0.05M NaOH. Data were analyzed using BIAevaluation software (GE Healthcare), and the data from the negative control peptide flow cell were subtracted. Kinetic parameters were calculated by fitting the obtained sensorgrams into 1:1 interaction with the drifting baseline algorithm, which gave the lowest chi-square value.
Peptide and anti-C1q ELISAs.
Purified C1q (Complement Technology) or NeutrAvidin (Pierce Biotechnology) proteins were coated overnight at 4°C at 5 μg/ml in carbonate buffer (0.1M sodium carbonate, pH 9.6) in MaxiSorp microplates (Nalge Nunc International). After washing, NeutrAvidin plates were incubated with biotinylated peptides at 5 μg/ml in PBS for 2 hours at room temperature. Nonbound peptides were removed by washing, serum samples were added, and plates were incubated for 1 hour at room temperature with shaking. The optimal serum dilutions were found to be 1:50 for the anti-C1q ELISA and 1:800 for the peptide ELISA. Washed C1q-coated plates were incubated with serum samples diluted in high-salt buffer (0.05% PBS–Tween 20 supplemented with 1M NaCl), while sera were diluted in 0.05% PBS–Tween 20 for the peptide ELISA.
After washing, bound antibody to the peptides was detected by incubation for 1 hour at room temperature with an alkaline phosphatase (AP)–conjugated rabbit anti-human IgG antibody (diluted 1:5,000 in 0.05% PBS–Tween 20; Promega) for detection of IgG or with AP-conjugated anti-human Fab (diluted 1:1,000 in 0.05% PBS–Tween 20; Jackson ImmunoResearch) for detection of bound Fab. For the anti-C1q ELISA, the same secondary antibodies were used but diluted in high-salt buffer. After washing, color development with AP substrate (Sigma-Aldrich) was performed as recommended by the manufacturer. The reaction was stopped by addition of the same volume of NaOH, and color development was read at 405 nm. We used 0.05% PBS–Tween 20 as washing buffer for both the peptide and anti-C1q ELISAs. For the peptide ELISA, the signal obtained from incubating antibodies to an irrelevant biotinylated peptide was considered background, and this optical density (OD) value was subtracted from the specific signal. For the anti-C1q ELISA, background binding of the AP-conjugated secondary reagent to C1q was subtracted from the specific binding. The data are expressed either as OD values or as relative level of expression (percent signal compared to control). For the analysis of cohorts of healthy controls and SLE patients, we standardized the experiments by expressing the data in units relative to the OD values obtained from reference SLE sera (set as 1,000 units) that show high levels of binding in the peptide ELISA or in the anti-C1q ELISA and that were included in each experiment/plate.
The cutoff for a positive signal to evaluate the binding of patient sera was determined as the interquartile mean (IQM) of the values obtained with the control sera (normal human serum [NHS]) plus 3 times the SD. A trimmed mean like the IQM was used since it gives a much more robust estimation (an estimation not greatly affected by outliers) of the average than the arithmetic mean. Samples were tested in duplicate or triplicate within a single experiment, and experiments were performed twice or more. At the time of measurements, the examiner (DV) was blinded to the clinical state of the patients at the time of sampling.
Binding of sera to A08 or C1q was performed as described earlier except that sera were diluted in 0.05% PBS–Tween 20 for both peptide and anti-C1q ELISAs and were preincubated with soluble, nonbiotinylated peptides as indicated for 3 hours at room temperature prior to binding to A08 or C1q. An amount of peptide was used to reach a molar excess of at least 200-fold (peptide competition assay at www.abcam.com) the average amount of IgG present in NHS (10–15 mg/ml IgG). For ELISA experiments using denatured C1q, the stock solution (1 mg/ml) was heated for 30 minutes at 56°C before dilution and coating on ELISA plates.
Statistical analysis was conducted using GraphPad Prism software, version 4.03. Differences in antibody titers were analyzed by a 2-sided Mann-Whitney test and Spearman's rank correlation coefficient.
Identification of the peptide sequence of C1q that is recognized by SLE Fab.
Anti-C1q Fab generated from an SLE patient and selected for binding to immobilized C1q were shown previously to bind to the CLR of the C1q-A and/or C1q-B chains of C1q (19, 21). In the present study, a peptide scan was performed to detect the peptide sequence that is recognized on C1q-A and/ or C1q-B (Figure 1). Two regions, 1 on C1q-A and 1 on C1q-B, showed significantly higher binding (∼1,000-fold for the highest signal) with SLE Fab A4 as compared to the other peptide sequences. Fab clone A14 also bound the same C1q-A and C1q-B regions, although the level of binding was lower than that of Fab A4 (data not shown). The 13-mer peptide sequences that resulted in the highest binding by the Fab were observed to be A08 (eighth peptide of the scan, representing an A-chain region), corresponding to the GRPGRRGRPGLKG peptide sequence, and B78 (78th peptide of the scan, representing a B-chain region), corresponding to the PGKVGPKGPMGPK peptide sequence.
Confirmation of anti-C1q Fab binding to A08 and B78 peptides using surface plasmon resonance analysis.
Binding of anti-C1q Fab clones to the target peptides A08 and B78 was analyzed by surface plasmon resonance analysis. As shown in Figure 2, Fab A4 and A14, but not the irrelevant polyclonal control Fab, bound to immobilized A08 and B78 peptides. In contrast, the Fab did not bind to a control peptide under the same conditions. The association was relatively rapid for the 4 interactions tested (Ka = ∼1 × 104/msec), and the formed complexes were relatively stable (Kd = 1 × 10−4 − 1 × 10−3/second), giving a nanomolar affinity (KD = ∼1 × 10−8M). These affinities are comparable to the affinities of Fab binding to the C1q CLR determined in a previous study (KD of 5.16 × 10−8M and 8.35 × 10−8M for Fab A4 and Fab A14, respectively) (19). Both Fab bound also to the C1q CLR, but not to native C1q when these were immobilized on a CM5 chip (data not shown).
Development and validation of a peptide-based ELISA.
To further analyze the binding of anti-C1q antibodies to the peptides and to be able to analyze large numbers of different SLE sera, we developed a peptide-based ELISA. Biotinylated peptides A08 or B78 were bound via coated NeutrAvidin on ELISA plates. Anti-C1q Fab (clone A14) and a control SLE patient serum bound to the A08 peptide but not to NeutrAvidin only or to NeutrAvidin conjugated to a control biotinylated peptide. Under the same conditions, we could not detect any binding of the SLE antibodies to the B78 peptide (Figure 3A), even though binding to B78 had been observed in the peptide scan and surface plasmon resonance experiments.
Specificity of binding to A08 by patient polyclonal serum was confirmed by inhibiting this binding with increasing amounts of soluble A08 peptide. The same amounts of a control peptide had no effect on binding of the antibodies to immobilized A08 (Figure 3B). As shown in Figure 3C, the level of inhibition by soluble A08 peptides ranged from 20% to 90% for different SLE patient sera.
Binding of patient sera to C1q is partially inhibited by soluble A08.
The anti-C1q response in SLE patients is likely to be polyclonal or oligoclonal rather than monoclonal (22, 23). To evaluate the contribution of the identified epitopes to the binding of patient sera to C1q, we preincubated the sera with excess amounts of A08 or B78 peptides to block binding of the antibodies to polystyrene-bound C1q. Results obtained for sera from 3 patients, in which the binding to immobilized A08 could be differentially inhibited by soluble A08, are shown in Figure 4A. Binding of polyclonal nonpurified IgG to immobilized native C1q could be blocked by soluble A08, and this inhibition was even more pronounced for binding of the sera to more linear epitopes, exposed after heat denaturing of C1q at 56°C. No inhibition was seen with soluble B78 (data not shown). The same amounts of control peptides did not affect binding to C1q. As summarized in Figure 4B, inhibition of binding was observed for the majority (n = 33) of the 38 SLE sera tested. Depending on the sera, inhibition of binding to denatured C1q ranged from 10% to 90% (average 40%). These data suggest that A08 represents a major linear C1q epitope for SLE autoantibodies.
Evaluation of A08 peptide ELISA as a diagnostic tool to screen SLE patient sera.
Using the A08 peptide ELISA, we next wanted to evaluate the prevalence of autoantibodies to this C1q epitope in a cohort of SLE patients as compared to healthy donors. Serum antibodies from 61 patients with confirmed SLE and from 72 healthy donors were analyzed for their binding to immobilized C1q or C1q-derived A08 peptides. In contrast to antibodies in sera from healthy individuals (NHS), serum antibodies from most SLE patients specifically bound to the A08 epitope (Figure 5). When we compared both assays, the peptide-based ELISA allowed slightly better discrimination between SLE patients and controls than the C1q assay, suggesting that the peptide ELISA has a higher specificity.
For most sera, binding to A08 correlated with binding to native C1q since those that contained anti-C1q antibodies also bound the A08 peptide (Figure 6). Statistical analysis of the 2 assays showed a significant correlation (rs = 0.3344, P = 0.0084) with linear regression (P = 0.036). However, the level of binding was not comparable in both assays.
Some SLE sera (5%) showed binding to C1q but not to A08. This fraction was smaller than that in healthy donors (17%). Conversely, 21% of SLE sera showed significant binding to A08 but not to C1q compared to 11% of sera in the control group. Finally, 56% of SLE sera were positive for both anti-C1q and anti-A08 compared to 3% of control sera.
In total, 79% of the SLE patient sera scored positive when tested with the A08 ELISA compared to 62% when using the anti-C1q ELISA. In addition, the peptide ELISA was better in discriminating asymptomatic donor sera from patient sera (79% SLE sera versus 14% NHS) than the C1q-based ELISA (62% SLE sera versus 20% NHS).
Since sensitivity and specificity are cutoff dependent, we also generated receiver operating characteristic curves for the discrimination between SLE patients and normal donors. The area under the curve for the A08 peptide ELISA was better than for the conventional anti-C1q ELISA (0.86 versus 0.78; for each curve P < 0.0001).
Within the subgroup of patients with active nephritis at the time of sampling, more patients were found to be positive for anti-A08 than for anti-C1q (75% versus 67%). The majority of the 13 patients who scored anti-A08 positive but anti-C1q negative had active nephritis (46%) or had had nephritis previously confirmed by biopsy (30%).
Among the patients with active biopsy-proven lupus nephritis at the time of sampling, we could not identify significant differences in terms of antibody positivity. Two patients with class IV nephritis and 1 patient with class II nephritis were single positive for anti-C1q. In comparison, 2 patients with class IV nephritis and another 2 patients with class II nephritis were single positive for anti-A08. In addition, both patients with class V nephritis were found to be anti-A08 positive but anti-C1q negative.
In the present study, we identified a major linear epitope of C1q that is targeted by anti-C1q antibodies from SLE patients. The epitope was identified using C1q-specific Fab derived from an SLE patient. In a previous study, we showed that these Fab were the result of an antigen-driven, affinity-matured immune response to a neoepitope of C1q (19).
In a microarray-based peptide scan, the anti-C1q Fab bound to 2 regions, 1 on the C1q A chain (represented by the A08 peptide) and 1 on the C1q B chain (represented by the B78 peptide). Binding of the Fab to A08 and B78 could be confirmed by real-time interaction measurements (surface plasmon resonance). Affinity of both A4 and A14 Fab toward A08 and B78 peptides was in the nanomolar range. However, using ELISA, only binding to A08 could be detected. It remains to be elucidated why biotinylated B78 peptides could be recognized by the antibodies when immobilized on BIAcore chips and with the peptide scan but not on ELISA plates. It is possible that differences in the biochemical properties of the adsorbing surface (negatively charged Dextran in BIAcore chips and hydrophobic polystyrene in ELISA) may result in either the maintenance or the loss of the conformation of the B78 peptide. Alternatively, since binding affinities of the anti-C1q Fab for B78 on BIAcore chips were lower than for A08, the affinity might be too low to detect B78 binding by ELISA. A similar differential reactivity in surface plasmon resonance experiments and ELISAs was also observed with anti-DNA monoclonal autoantibodies from SLE patients (24).
Interestingly, both A08 and B78 epitopes represent functional regions of complement C1q. Epitope A08 is almost identical to amino acid residues 14–26 of C1q-A that were shown by the group of Gewurz (25–27) to contribute to complement activation by DNA, C-reactive protein (CRP), and serum amyloid P component—all being implicated in the pathogenesis of SLE. Although CRP and DNA were later shown to bind to the globular domain of C1q (28–30), the role of the charged area of the C1q-A chain in these interactions remains poorly understood and needs further investigation. Moreover, the 14–26 peptide was shown to have a complement inhibitory effect in vivo (31), indicating that this area of the molecule participated in complement activation.
The B78 epitope, which unfortunately, could not be assessed in a peptide-based ELISA, is part of the binding regions of C1s and C1r on C1q (32). The importance of this region was recently demonstrated by the discovery of a single amino acid mutation in the B78 peptide region of C1q-B found in a lupus patient that affected the functionality but not the expression level of the mutant C1q (33). Accordingly, antibodies to this domain of C1q may create an acquired functional C1q deficiency similar to what has been suggested to result from antibodies against the C-terminal globular head domain (34). Studies are ongoing to elucidate the role of the binding of anti-C1q autoantibodies to the A08 epitope as well as to the B78 epitope in the pathogenesis of SLE.
Using the A08 peptide ELISA, we further demonstrated that serum antibodies from most SLE patients bound to the peptide and this binding correlated with the binding to C1q. However, comparing A08 peptide with native C1q as an antigen in ELISA revealed that the A08 assay had higher sensitivity and allowed better discrimination between low level–positive and true-negative sera. These differences are likely due to the fact that the large C1q molecule with a maximum of six A08 epitopes results in fewer epitopes per surface area compared to the immobilized peptides, reducing the sensitivity of the assay. In addition, the globular heads of C1q are well known to bind immune complexes nonspecifically. To minimize this nonspecific binding, classic C1q-based ELISAs are typically performed in high-salt buffers. However, high-salt conditions are not physiologic and could destroy potential epitopes or abolish the binding of antibodies to particular epitopes. These differences in properties between the peptide and anti-C1q ELISAs could explain why some of the SLE sera (21%) were found to bind to the A08 peptide of C1q but not to native C1q molecules. This detection difference underscores the additional diagnostic potential of the peptide ELISA.
Different patient sera showed different levels of inhibition of binding to A08 and C1q by soluble A08 peptides. This differential inhibition could be due to differences in IgG concentrations between sera or differences in affinity/avidity of binding to A08. Soluble monomeric peptides have lower avidity compared to multimeric immobilized peptides. Sera that have a low affinity for the peptide will only bind when the avidity is high enough (immobilized peptide) and will not bind monomeric peptides. Accordingly, for some sera, there was low or no inhibition with soluble A08 in the competition ELISA. Conversely, for those sera in which the peptide could inhibit up to 90% of binding, avidity is likely of less importance.
Interestingly, binding of patient sera to the complete C1q molecule could be partially inhibited by soluble A08 peptides. C1q is a large multimeric molecule, and autoimmune antibodies that can recognize the CLR, but sometimes also the globular heads, have been described. Thus, taking into account that the anti-C1q response in SLE patients is polyclonal (22, 23, 34), an inhibition of binding of patient sera to native C1q by soluble A08 peptides, which represent only a single epitope, can be considered significant. The polyclonal nature of the sera is also evident after heat denaturation of C1q, which mainly abolished conformational epitopes. For some sera, this denaturation resulted in higher binding to C1q, but for others, reduced binding. Binding to the remaining linear epitopes after heat denaturation could be blocked up to 90% by soluble A08 peptides. The polyclonal reactivity of patient sera to C1q could also explain why a number of sera bound to a similar extent to A08 (1 epitope) but showed different levels of binding to C1q (additive effect of multiple epitopes), resulting in a statistically relevant but weak linear regression when comparing the two assays. Together with the observation that serum antibodies from most SLE patients, but not from healthy individuals, specifically bind to the A08 epitope, we conclude that the A08 peptide represents a linear epitope that is a major target of anti-C1q patient sera.
Previous studies showed that anti-C1q antibodies could be detected in 20–50% of unselected SLE patients (with or without active lupus nephritis). This proportion increases to >90% in patients with active renal disease (13). The cohort used in the current study predominantly included patients who had had an episode of severe lupus nephritis in the past or who had active lupus nephritis at the time of sampling, as confirmed by renal biopsy. Accordingly, a relatively high fraction of patients were found to be anti-C1q positive. Interestingly, more patients were found to have antibodies to A08, suggesting a higher sensitivity of the assay. Within the subgroup with active lupus nephritis at the time of sampling, more patients were found to be positive for anti-A08 (75%) than for anti-C1q (67%), suggesting that the anti–A08 peptide ELISA is also more sensitive for active disease than the classic anti-C1q assay. The majority of the patients who were anti-A08 positive and anti-C1q negative had active nephritis or had had nephritis in the past. Taken together, the findings show that the peptide-based assay presented here provides more specific and more sensitive results than a conventional anti-C1q assay.
A similar observation was made for the diagnosis of rheumatoid arthritis, in which an assay for anti–cyclic citrullinated peptide antibodies was found to have superior specificity with similar sensitivity compared to the detection of rheumatoid factors in patient sera (35, 36). Analogously, the A08 peptide assay appears to represent an improved alternative assay for the detection of anti-C1q autoantibodies in SLE. However, additional studies will have to be performed using well-defined large cohorts of patients with long followup periods in order to determine the diagnostic value of the anti-A08 assay in SLE patients.
In conclusion, we identified a major linear target of SLE-associated anti-C1q antibodies, a marker of severe lupus nephritis. A peptide-based ELISA using this epitope was found to be more specific and sensitive for the detection of anti-C1q antibodies than a conventional anti-C1q assay. The identification of this anti-C1q epitope could help us to understand the pathogenic role of anti-C1q antibodies and eventually might even lead to the design of therapeutic synthetic peptides blocking anti-C1q binding.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Vanhecke had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Vanhecke, Wan, Schaller, Trendelenburg.
Acquisition of data. Vanhecke, Roumenina, Osthoff, Schaller, Trendelenburg.
Analysis and interpretation of data. Vanhecke, Roumenina, Wan, Osthoff, Schaller, Trendelenburg.
We thank Dr. Eliska Potlukova (Third Clinic of Medicine, Charles University, Prague, Czech Republic), Dr. Margarita Lopez-Trascasa (Department of Immunology, University Hospital La Paz, Madrid, Spain), and Dr. Solange Moll (Institutes of Pathology, University Hospitals of Geneva and Lausanne, Switzerland) for providing the serum samples from patients with active lupus nephritis. We thank the Blood Transfusion Center Beider Basel (Basel, Switzerland) for providing the serum samples from asymptomatic donors.