Systemic lupus erythematosus (SLE) is the archetype of a systemic autoimmune disease. It is characterized by a dysregulated immune system, resulting in the generation of autoantibodies to numerous self-antigens and a broad spectrum of clinical manifestations. The exact cellular and molecular mechanisms leading to the disease remain incompletely understood (1) but may be elucidated by exploring the characteristics of self-antigens. One of these self-antigens is C1q, the first component of the classical complement activation pathway. Approximately 20-50% of unselected SLE patients have autoantibodies against C1q (anti-C1q) (2).

Positivity for anti-C1q is predictive for flares of lupus nephritis (LN) and anti-C1q levels correlate with overall disease activity (3). Additional lines of evidence suggest that these antibodies are directly involved in tissue injury (4): C1q deposition is a typical finding in severe LN and anti-C1q could be extracted from glomerular basement membrane fragments (5). Furthermore, C1q is a highly functional molecule (6) and experimental data support the assumption, that binding of anti-C1q alter those functions (7–10). However, the definite pathogenic role of the polyclonal anti-C1q remains to be determined and may strongly depend on the antibody binding site.

C1q is composed of 18 polypeptide chains (6 A-, 6 B- and 6 C-chains), that form six triple helices assembling to a structure that resembles a bouquet of tulips. Each chain has a short N-terminal region, followed by an ~81 residue-long collagen-like region (CLR) forming the stalk of the molecule and a ~135 residue-long C-terminal globular head region (gC1q) (11). The globular heads are mostly responsible for the recognition of target structures, e.g. Fc parts of bound immunoglobulins (12), surface proteins of pathogens and apoptotic cells (13). Upon binding of C1q, the CLR mediates immune effector mechanisms, including complement activation and enhancement of phagocytosis through interaction with cell surface receptors (14, 15).

Anti-C1q are polyclonal and primarily recognize neoepitopes on the CLR of C1q (16, 17) and to a lower extend also on gC1q (18, 19). These epitopes are cryptic, only exposed when C1q is in its bound form (20) and certainly located in different structures. However, so far little is known about the precise C1q epitopes (21, 22). As SLE patient-derived monoclonal anti-C1q Fabs recognize different C1q polypeptide chains in Western blot assay (22), they were used in a previous microarray-based peptide scan to identify peptide sequences recognized by anti-C1q (23). By this approach, Vanhecke et al. described a major linear epitope being located on the N-terminal C1q A-chain covering the arginine rich part of the chain, the so-called ‘A08’. Interestingly, this region is also known as a major binding site for non-immunoglobulin molecules (24), and could even be an early epitope allowing cross reactivity of antibodies that primarily target EBNA-1 of Epstein Bar Virus (EBV) due to sequence homology (25). Epitope spreading might then lead to a more diverse antibody repertoire against the whole C1q molecule. As C1q has more functional subunits than just ‘A08’, e.g. the globular heads, the lysins in the C-terminal CLR that mediate the interaction of C1q with the C1s₂C1r₂ tetramer (26) and widely unknown regions, which are responsible for the interaction with C1q receptors (15), antibodies targeting these structures might have different functional consequences and thus mediate different disease manifestations.

The aim of this study was to explore epitopes of C1q and determine whether epitope-specific antibodies against C1q can be linked to specific disease manifestations.

Since the study by Vanhecke et al., was limited to the use of monoclonal antibodies, which do not mirror the polyclonal character of anti-C1q in patients, we used 22 SLE patient sera to determine the epitope landscape of C1q. In addition, we applied an advanced epitope mapping method based on densely overlapping linear as well as cyclized peptides, to increase sensitivity, since many epitopes rely on protein folding, which can hardly be detected with standard microarrays based on linear peptides alone (27). Subsequently we investigated the conspicuous peptide sequences by analyzing a large cohort of well-defined SLE patients provided by the Swiss Systemic Lupus Erythematosus Cohort Study (SSCS).

This study aimed at identifying and exploring the clinical relevance of epitope specific autoantibodies against complement C1q (anti-C1q) in patients with SLE. Considering the multiple functions of C1q, the role of C1q in SLE as well as the striking association of anti-C1q with active LN and SLE disease activity, a better understanding of epitopes of C1q could aid understanding the pathogenic mechanisms of SLE as well as improving diagnostic procedures. By using SLE patient sera in an advanced epitope mapping method, we first identified three epitopes on the C1q A- and three on the B-chain. In subsequent exploration of clinical relevance of these epitopes in a large cohort of patients by peptide-ELISA, two of the investigated peptides were significantly better recognized by serum IgG of SLE patients than of healthy controls. In addition, positivity for four of the investigated peptide-specific antibodies showed associations with selected SLE disease manifestations. These primarily explorative analyses might point to distinct functional properties of the measured peptide-specific antibodies.

The most obvious association with SLE was found for IgG antibodies targeting an epitope on the N-terminal C1q A-chain. The corresponding peptides were named ‘A09’ and ‘A15’ respectively, based on the position of their first AA on the CLR of C1q. Anti-A09 correlated with fever, arthritis, thrombocytopenia, and the occurrence of anti-dsDNA antibodies, while patients with anti-A15 had an increased probability of having active disease, arthritis, leukopenia, low complement and antiphospholipid antibodies. Regarding the occurrence of proteinuria, which is a typical finding in lupus nephritis, anti-C1q showed significantly better diagnostic performance than anti-A09 as well as anti-A15. This finding apparently is not in line with previous studies which showed that anti-A15-ELISA is more specific and more sensitive than a conventional anti-C1q assay for the detection of active lupus nephritis in SLE patients (10) (23). These differences in observation might be explained by the differences in patient selection and number. Both previous studies examined exclusively (10) or predominantly (23) lupus patients with renal biopsy-proven lupus nephritis, whereas most patients analyzed in the present study had long lasting, stable disease resembling an unselected clinical outpatient cohort of patients with closely monitored disease. Furthermore the sample sizes of the preceding studies were substantially smaller, than in the present study, namely n=61 (23) and n= 210 (10) versus n= 378 in our study presented here.

Nevertheless the present study is in line with the study from Vanhecke et al. showing that anti-A15 is better in discriminating asymptomatic donor sera from SLE patient sera than anti-C1q. In the present study, this discrimination was in the same range as the reported diagnostic performance of anti-dsDNA antibodies (40). Hence, anti-A-15 might serve as a diagnostic marker for SLE. However, to determine the real discriminatory power, it will be of importance to also investigate anti-A09 and -A15 in other inflammatory rheumatic diseases, and to perform a direct comparison with anti-dsDNA antibodies.

Furthermore, anti-C1q showed a weak correlation with anti-A09 or anti-A15 antibodies, respectively as also observed by Vanhecke et al. (23). In line with these findings, Wu et al. recently described that anti-A15 antibodies derived from 10 lupus nephritis patients bound to A15 but not to intact C1q (41). Regarding the potential functional consequences of anti-A09 and -A15, it should be noted that both peptides include a major binding site of C1q for non-immunoglobulin molecules (24). With regard to the interaction of A09 compared to A15 with binding partners other than anti-C1q, we hypothesized that the arginines being present in A09 as well will lead to very similar interactions as observed for A15. However, the differences in correlation with clinical parameters between anti-A09 and ant-A15 suggest that either the antibodies have a different potential to interfere with the known interactions of C1q with the described non-immunological molecules and/or receptors, or may point to differences in interaction between the two sites themselves with these binding partners.

The mentioned core sequence was previously described to allow crossreactivity between antibodies directed against EBNA-1 of EBV and C1q (25). In addition, Wu et al. could show that BALB/C mice, which were immunized with the A15 peptide, developed anti-C1q antibodies. They concluded that A15 is important for development of anti- C1q antibodies, but epitope spreading might then lead to a more diverse antibody repertoire against the whole C1q molecule. In line with this finding, generation of anti-A09 and -A15 in SLE patients seem to be an early event in the course of the disease. Additionally, data from multivariate regression suggest that anti-A09 and -A15 have higher stability over time with lower dependency on disease activity ( Supplementary Figure 4 ). Taken together, these results support the hypothesis, that molecular mimicry is an early event in the pathogenesis of SLE, with the formation of anti-A15 antibodies being an intermediate step, but might also explain the weak correlation between anti-A15 and anti-C1q.

With regard to IgG antibodies against the other described peptide epitopes, no overall differences in antibody levels between SLE patients and control sera were observed. However, when judging on the significance of these antibodies, one has to keep in mind that quantitatively peptide-specific anti-C1q are only representing a small fraction of total anti-C1q, and they only occur in a subgroup of patients that is likely to be too small to have an impact on overall differences between unselected SLE patients and healthy controls. As the study hypothesis was that antibodies against distinct epitopes of the multifunctional C1q molecule are associated with a specific disease expression, we thus also explored the association of the peptide antibody positivity with the clinical presentation of SLE. Patients with anti-A86 were more likely to have vasculitis and leucopenia, and the presence of anti-B43 was associated with arthritis and anti-dsDNA antibodies.

However, because of the explorative character of these analyses, confirmatory studies in large cohorts of SLE patients covering a broad spectrum of SLE manifestations and taking interrelations into account will be required to define the definite role of the described peptide antibodies.

Lastly, with regard to anti-C1q levels we observed that anti-C1q are associated with a much wider range of clinical disease manifestations than previously described. So far, anti-C1q antibodies have mainly been studied in association with LN (42–44). However, while confirming this known association in the present study, we also observed a clear association with arthritis OR= 4.811 (2.722 - 8. 543), skin involvement OR= 2.646 (1.522 -4.613), vasculitis OR= 8.757 (2.093 -43.629) and serositis OR= 3.065 (1.175 - 7.514). Therefore, anti-C1q could be more broadly considered as marker of SLE disease activity. This observation could be due to the large number of SLE patients investigated in our study. To the best of our knowledge, to date our study is the largest ever on anti-C1q in SLE patients.

The main limitation of the present study is its retrospective observational character. In addition, in spite of the large number of investigated patients, the sample size was still too small to make a clear statistical statement for some of the investigated disease features and thus would require even larger cohorts. Moreover, longitudinal data on the described antibodies will be of importance in the future. Lastly, despite the extensive character of our epitope mapping, the expression of peptides only partially resembles the conformation of the corresponding peptide sequences as part of the complete C1q molecule, and the expression of peptides probably differed between their expression in the initial epitope mapping versus the ELISAs performed in the large SLE cohort. Furthermore, our methodologies were not able to detect and describe antibodies against epitopes involving two or more chains of C1q. Thus, the described peptide-specific antibodies are likely still representing only a fraction of total anti-C1q.

In conclusion, in this exploratory and largest study to date on anti-C1q in SLE patients we describe six candidate epitopes of anti-C1q and their clinical associations in SLE patients. Two N-terminal located A-chain epitopes, which provide good discrimination between SLE patients and healthy individuals, might serve as a biomarker of the disease. In addition, peptide-specific anti-C1q were found to be associated with specific disease manifestations, but their potential impact on clinical patient management and for the understanding of pathogenic mechanisms needs to be confirmed.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Swissethics (ethical committee of the Canton Vaud, Switzerland Ref. No. 2017-01434). The patients/participants provided their written informed consent to participate in this study.

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. JK 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. JK, JW, MT. Acquisition of data. JK, JL, AK-G, CC, UH-D, CR, and MT. Analysis and interpretation of data. JK, PR, SV, and MT.

This work was supported by a project grant of the Swiss National Science Foundation (SNSF) given to MT (320030_200423).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.