It has been suggested that anti–citrullinated protein antibodies (ACPAs) play an important role in the pathogenesis of rheumatoid arthritis (RA). To exert their pathologic effects, ACPAs must recruit immune effector mechanisms such as activation of the complement system. Mouse models of RA have shown that, surprisingly, arthritogenic antibodies activate the alternative pathway of complement rather than the expected classical pathway. This study was undertaken to investigate whether human anti–cyclic citrullinated peptide (anti-CCP) antibodies activate the complement system in vitro and, if so, which pathways of complement activation are used.
We set up novel assays to analyze complement activation by anti-CCP antibodies, using cyclic citrullinated peptide–coated plates, specific buffers, and normal and complement-deficient sera as a source of complement.
Anti-CCP antibodies activated complement in a dose-dependent manner via the classical pathway of complement, and, surprisingly, via the alternative pathway of complement. The lectin pathway was not activated by anti-CCP antibodies. Complement activation proceeded in vitro up to the formation of the membrane attack complex, indicating that all activation steps, including the release of C5a, took place.
Our findings indicate that anti-CCP antibodies activate the complement system in vitro via the classical and alternative pathways but not via the lectin pathway. These findings are relevant for the design of interventions aimed at inhibition of complement-mediated damage in RA.
Rheumatoid arthritis (RA) is one of the most common autoimmune diseases, affecting ∼1% of the population (1, 2). In recent years novel treatment modalities have become available that have proven to be successful in relieving the symptoms of the disease. Nonetheless, insight into the mechanisms involved in RA pathogenesis is still limited. Anti–citrullinated protein antibodies (ACPAs) are a hallmark of RA (3). The occurrence of ACPAs, often measured using an anti–cyclic citrullinated peptide (anti-CCP) test, can occur years before the onset of disease (4, 5) and provides prognostic information (6). Studies in mice have shown that ACPAs can exacerbate established arthritis (7, 8), a finding that mimics the human situation, in which ACPA-positive patients have a more severe disease course than ACPA-negative patients (6, 9). Taken together, these observations and recent reports (10) that some of the well-known genetic risk factors for RA are predominantly associated with ACPA-positive RA suggest that anticitrulline immunity may play an important role in the pathogenesis of the disease. To elucidate the biologic mechanisms underlying RA, and possibly identify targets for intervention, it is therefore important to gain more insight into the development of ACPAs and the way in which they may contribute to disease pathogenesis.
Antibodies can activate effector mechanisms of the immune system largely via 2 pathways, namely, by binding to cellular Fc receptors and by activation of the complement system. One recent study has shown that ACPA can trigger Fc receptors (11). No information is currently available on the ability of ACPA to activate the complement system.
The complement system is an integral part of the innate immune defense, but it is also involved in the induction of the adaptive immune response and in the removal of waste, including immune complexes and dead cells (12–14). The complement system can be activated via 3 pathways: the classical pathway, the lectin pathway, and the alternative pathway (15). Each pathway is initiated by a specific recognition molecule. The classical pathway is initiated by C1q, the lectin pathway is initiated by mannose-binding lectin (MBL) or ficolins, and the alternative pathway is initiated by C3(H2O). Initiation of complement activation via each of these pathways results in the formation of C3 convertases that cleave C3 to produce biologically active complement fragments that result in opsonization, chemotaxis, and cytolysis.
Complement is present in synovial fluid and is partly produced locally and partly originates from the circulation (16). Complement levels in the synovial fluid of RA patients can be lower than those in controls because of local consumption (17). Complement activation products such as sC5b–9, Bb, C3a, and C1 inhibitor–C1s and C1q–C4 complexes have been shown to be increased in synovial fluid (18–22), and complement depositions in synovial tissue from RA patients can be visualized by immunohistochemistry (23). Recent genetic association studies have indicated that complement (C5 in one previous study ) may be involved in the pathogenesis of human RA. Collectively, the presence of activated complement fragments in synovial fluid, consumption of complement in synovial fluid, and the deposition of complement fragments in the synovium indicate that complement activation does occur in the synovial space and provide evidence of local complement activation in RA.
Mouse models of RA have shown, surprisingly, that arthritogenic antibodies are fully dependent on the alternative pathway of complement, rather than the expected classical pathway. C3−/−, FB−/−, C5−/−, and C5a receptor−/− mice are protected against induction of arthritis, whereas C1q−/− and MBL−/− mice are fully susceptible (25, 26), indicating that in mouse models the alternative pathway is necessary and sufficient for disease induction (25, 27, 28).
Now that ACPAs are increasingly recognized as major players in RA, it is of relevance to know if and how these antibodies activate complement. This information could be beneficial for the understanding of the pathogenic process of human RA and is important for the optimal targeting of anti-complement therapeutics. Therefore, in the present study we addressed the questions of whether anti-CCP antibodies can activate the complement system and, if so, via which pathway(s) this activation occurs. Our results indicated that anti-CCP antibodies activate the complement system via the classical pathway, and, interestingly, also directly via the alternative pathway, but not via the lectin pathway.
PATIENTS AND METHODS
Patients were selected from the Leiden Early Arthritis Clinic, an inception cohort of patients with recent-onset arthritis (of <2 years' duration) that was initiated at the Department of Rheumatology of the Leiden University Medical Center in 1993. Baseline serum samples from 60 patients with anti-CCP-2–positive arthritis (>25 AU/ml in the CCP-2 assay [Euro-Diagnostica, Arnhem, The Netherlands]) were analyzed in this study. Patient characteristics are summarized in Table 1. The collection and use of patient samples was approved by the local medical ethics committee in compliance with the Helsinki declaration.
Complement-active sera and anti-CCP–positive standard.
As a source of complement, pooled normal human serum (NHS) from healthy controls was divided into aliquots and stored at −80°C until used. As additional sera, we used C1q-depleted serum, which was generated as previously described (29), and MBL-deficient serum from an MBL-deficient healthy donor. Sera from 4 RA patients with high anti-CCP titers were pooled and used as an anti-CCP–positive standard serum.
Buffers and reagents.
Human IgM was purified as previously described (29). Mannan (from Saccharomyces cerevisiae; M7504) and lipopolysaccharide (LPS) (from Salmonella typhose; L6386) were obtained from Sigma (St. Louis, MO). Rabbit anti-C1q, horseradish peroxidase (HRP)–labeled rabbit anti-C3c, rabbit anti–C5b–9, and HRP-labeled goat anti-rabbit were obtained from Dako (Carpinteria, CA). As additional antibodies, we used digoxigenin (DIG)–conjugated anti-C4d monoclonal antibody, anti-MBL monoclonal antibody 3E7 (HBT, Uden, The Netherlands), and rabbit anti-properdin (generated in house ). HRP-labeled anti-DIG was obtained from Roche (Indianapolis, IN).
The primary and secondary antibodies were diluted in phosphate buffered saline (PBS) containing 0.05% Tween 20 and 1% bovine serum albumin (PBST–BSA). Patient sera were diluted in PBST–BSA containing 10 mM EDTA (pH 7.5). The sera used as a complement source were diluted in either gelatin veronal buffer containing Ca2+ and Mg2+ (GVB++) (veronal buffered saline [VBS] containing 0.5 mM MgCl2, 2 mM CaCl2, 0.05% Tween 20, and 0.1% gelatin [pH7.5]) or Mg-EGTA (VBS containing 10 mM EGTA, 5 mM MgCl2, 0.05% Tween 20, and 0.1% gelatin [pH 7.5]).
Functional activity of the 3 complement activation routes was analyzed by enzyme-linked immunosorbent assay using IgM as the ligand for the classical pathway, LPS as the ligand for the alternative pathway, and mannan as the ligand for the lectin pathway (31). MaxiSorp plates (Nunc, Naperville, IL) were coated with human IgM at 2 μg/ml, with LPS at 10 μg/ml, or with mannan at 100 μg/ml in coating buffer (100 mM Na2CO3/NaHCO3 [pH 9.6]) overnight at 4°C. After each step, plates were washed 3 times with PBS containing 0.05% Tween 20. Residual binding sites were blocked with PBS containing 1% BSA for 1 hour at 37°C. After washing, plates were incubated with NHS dilutes at 2% in either GVB++ or PBST–BSA–EDTA and incubated for 1 hour at 37°C. After washing, bound C3 was detected with HRP-labeled rabbit anti-human C3c. After final washing, the enzyme activity of HRP was detected by incubating the plate with ABTS, according to the recommendations of the manufacturer (Sigma). Absorbance at 415 nm was determined using a microplate biokinetics reader (BioTek Instruments, Winooski, VT). The MBL binding assay was essentially the same as the C3 activation assay on mannan, but using anti-MBL antibodies for detection (29).
Complement activation on CCP-2 plates.
To analyze the complement-activating potential of anti-CCP antibodies, we used CCP-2 plates (Euro-Diagnostica) that had been blocked with PBS–1% BSA for 1 hour at 37°C. After washing, plates were incubated with patient sera diluted 1:50 in PBST–BSA–EDTA. Following incubation for 1 hour at 37°C, plates were washed and incubated for 1 hour at 37°C with complement-active NHS as a source of complement, used at either 1% or 10% diluted in either GVB++ or Mg-EGTA. After washing, activation of complement was determined by detection of plate-bound complement fragments with specific antibodies diluted in PBST–BSA for 1 hour at 37°C. Bound antibody was detected using matched secondary HRP-labeled reagent diluted in PBST–BSA for 1 hour at 37°C. Following final washing steps, plates were developed using the substrate ABTS, and absorbance was read at 415 nm. A schematic overview of this assay is shown in Figure 1D.
Correlations were calculated using Pearson's correlation coefficient, and the significance of differences between groups was calculated by Student's t-test using GraphPad Prism software (GraphPad Software, San Diego, CA). P values less than 0.05 were considered significant.
Establishment of an assay to detect and compare complement activation by anti-CCP antibodies in vitro.
We developed a novel assay to specifically analyze the complement-activating potential of anti-CCP antibodies. This 2-step assay first allows binding of anti-CCP antibodies to CCP-2 plates without activating complement in the patient serum. The complement-activating capacity is subsequently assessed by addition of a fixed source of exogenous complement. We prevented complement activation in the anti-CCP–positive samples by using a dilution buffer containing EDTA (32).We performed a series of assays to verify that the use of a buffer containing EDTA would indeed prevent complement activation in anti-CCP–positive sera but would not hamper the binding of anti-CCP antibodies to the CCP-2 plate.
First, we used a plate-bound classical pathway activation assay to test the effect of diluting NHS in either PBST–BSA–EDTA, a buffer that does not allow complement activation, or GVB++, a buffer that allows complement activation via all 3 pathways. We observed that PBST–BSA–EDTA prevented the complement activation that occurred when NHS was diluted in GVB++ (Figure 1A).
Next, we analyzed whether PBST–BSA–EDTA prevented complement activation via each of the 3 pathways. Plate-bound assays were used to specifically test the activation of the classical, alternative, and lectin pathways, using plate-bound IgM, LPS, and mannan, respectively. We observed activation of the classical, alternative, and lectin pathways, all 3 of which were blocked when NHS was diluted in PBST–BSA–EDTA (Figure 1B). These data show that the NHS used as a source of complement was activated via all 3 complement pathways and that all of these pathways were blocked by PBST–BSA–EDTA.
Finally, we tested whether dilution of the anti-CCP–positive sera in PBST–BSA–EDTA would have a negative impact on the binding of anti-CCP antibodies to the CCP-2 plate. We observed no effect of PBST–BSA–EDTA on the binding of anti-CCP antibodies to CCP-2 plates (Figure 1C). The sequential steps used to analyze complement-activating potential are summarized in Figure 1D.
Anti-CCP antibodies activate complement via the classical and alternative pathways but do not activate the lectin pathway.
To analyze whether anti-CCP antibodies activate complement, we conducted a novel assay using anti-CCP antibodies from a pool of anti-CCP–positive sera. Complement activation was detected by measuring plate-bound C3. We observed strong dose-dependent complement activation by anti-CCP antibodies (Figures 2A and B). Since C3 can be activated by all 3 pathways, we next set out to determine the relative contribution of each of these pathways.
To discriminate classical and lectin pathway activation from alternative pathway activation we analyzed 3 parameters. First, we compared deposition of C3 with deposition of C4. Second, we used a buffer that allows alternative pathway activation but blocks classical pathway and lectin pathway activation (32). Third, we analyzed complement activation at several concentrations of NHS, since the alternative pathway can only be activated effectively with 10% NHS, whereas the classical and lectin pathways are activated with much lower concentrations of NHS (32).
Incubation of plates that were loaded with a fixed concentration of anti-CCP antibodies from an anti-CCP–positive standard with NHS diluted in GVB++ as a complement source resulted in strong dose-dependent deposition of both C4 and C3 (Figures 2A and B). Since C4 is used by both the classical and lectin pathways (Figure 3A), these data indicate that either the classical pathway, the lectin pathway, or both were activated. Since C3 can be activated by all 3 pathways, the data could indicate that C3 was activated via the classical pathway or lectin pathway and potentially also directly via the alternative pathway. The use of NHS diluted in Mg-EGTA, a buffer that allows complement activation only via the alternative pathway, resulted in deposition of C3 but not C4 (Figures 2A and B). These results therefore indicate that these antibodies directly activate the alternative pathway.
To determine whether the classical or lectin pathway was responsible for C4 deposition, both C1q-depleted serum and MBL-deficient serum were used in the same assay. C1q-depleted serum diluted in GVB++ did not show any C4 deposition (Figure 2A), indicating that all C4 deposition was dependent on C1q and that MBL or ficolins present in the serum did not contribute to C4 deposition. The C1q-deficient serum did, as expected, still activate the alternative pathway, as evidenced by the observed C3 deposition (Figure 2B). The MBL-deficient serum, also diluted in GVB++, showed normal C4 deposition and C3 deposition that was not different from that observed with NHS, indicating that MBL and the lectin pathway are not involved in complement activation by anti-CCP antibodies.
Under conditions in which C3 deposition was observed to take place via the alternative pathway only (NHS in Mg-EGTA or C1q-depleted serum in GVB++), we observed C3 deposition with 10% serum only, and not with 1% serum, confirming true alternative pathway activity. Taken together, these data indicate that anti-CCP antibodies can recruit both the classical and alternative pathways, but not the lectin pathway, of complement activation.
Detailed analysis of the complement-activating potential of anti-CCP antibodies from patients with early arthritis.
We next analyzed the complement-activating potential of anti-CCP antibodies in 60 patients with anti-CCP–positive arthritis. We observed that anti-CCP antibodies from all of these patients supported activation of C3, C4, and binding of C1q (Figure 3B), similar to the observations with the anti-CCP–positive standard. No binding of MBL to anti-CCP antibodies was observed (Figure 3B). To verify that the lack of MBL binding was not due to technical limitations, we incubated increasing concentrations of NHS in GVB++ on CCP-2 plates loaded with anti-CCP antibodies and on plates coated with mannan, a well-established ligand for MBL. Strong dose-dependent binding of MBL to the mannan-coated plate, but no binding to the plate loaded with anti-CCP antibodies, was observed (Figure 3C). This indicates that MBL did not bind to anti-CCP antibodies from these patients, and further confirms and extends the observation that anti-CCP antibodies do not use the lectin pathway to activate complement.
In all patients analyzed, we also observed C3 deposition when NHS was incubated in Mg-EGTA, indicating that anti-CCP antibodies have the capacity to directly activate the alternative pathway. We analyzed additional markers of complement, and observed that properdin, which is the only positive regulator of complement and contributes to the activity of the alternative pathway, bound to CCP antibodies (Figure 3B), further confirming the ability of anti-CCP antibodies to activate the alternative pathway in all patients analyzed. We also detected deposition of C9, a marker of activation of the final components of the complement cascade up to the formation of the membrane attack complex (Figure 3A). These results indicate that anti-CCP antibodies can activate the entire complement cascade and also indicate that C3a and C5a must be generated (Figure 3A).
Relationship of the extent of complement activation to anti-CCP antibody titer.
For this study we selected patients who were IgG anti-CCP–positive and for whom the anti-CCP isotype usage was known. We correlated the IgG anti-CCP titer with activation of the different pathways by analyzing the correlation with C1q for the classical pathway, with C4 for the classical and lectin pathways, with C3 for all pathways, and with C3 from NHS in Mg-EGTA for the alternative pathway only. We observed a strong positive correlation between IgG anti-CCP levels and C3, C3 deposition via the alternative pathway only, C1q, and C4 (Figures 4A–D). These data indicate that the extent of complement activation is correlated with anti-CCP antibody levels.
Approximately 58% of the IgG anti-CCP antibody–positive sera examined in this study were also positive for IgA anti-CCP antibodies, and 54% were also positive for IgM anti-CCP antibodies (Table 1). Since both IgA and IgM have been implicated in the activation of the lectin pathway (33, 34), we also analyzed in more detail whether there were indications of activation of the lectin pathway in IgA or IgM anti-CCP antibody–positive patients. Despite the fact that we observed stronger complement activation, as measured by deposition of C3, in samples from patients who were also positive for IgM anti-CCP (P < 0.0001) or IgA anti-CCP (P < 0.0002) compared with samples from patients who were positive for IgG anti-CCP only (Figures 5A and B), we did not observe any MBL binding in these IgA anti-CCP or IgM anti-CCP antibody–positive patients (Figures 5A and B). Taken together, these data further confirm that anti-CCP antibodies activate the classical and alternative pathways but not the lectin pathway of complement, even in patients who are also positive for IgA and IgM anti-CCP antibodies.
Anti-CCP antibodies have gained considerable interest in the field of arthritis, and it has been suggested that they play a role in disease progression. The exact events leading to the formation of these antibodies are thus far unknown, as are the mechanisms by which they may contribute to tissue damage and inflammation. However, anti-CCP antibodies have been shown to aggravate collagen-induced arthritis in mice, which suggests that anti-CCP can be pathogenic (7, 8).
In this study, we showed that anti-CCP can activate the human complement system via the classical and alternative pathways but not via the lectin pathway. Recently, Clavel et al (11) demonstrated that in vitro, anti-CCP antibodies interact with Fc receptors and activate immune cells. Taken together, these observations indicate that anti-CCP antibodies can trigger the 2 most prominent immune effector mechanisms used by antibodies.
Although one could expect that anti-CCP antibodies can activate the complement system, not all antibody isotypes do so to the same extent, and antibodies from different patients may use different pathways for activation. IgM, IgG3, and IgG1 are the most potent complement-activating isotypes, followed by IgG2 and IgA, whereas IgG4 and IgE do not activate complement. The anti-CCP antibody response involves all IgG subclasses as well as IgM and IgA (35, 36). The traditional view that the antibodies IgM and multivalent IgG can activate the complement system only directly via the classical pathway has recently been challenged by observations that differentially glycosylated forms of IgG (37), IgM (33), and IgA (38, 39) can also activate the alternative pathway and/or lectin pathway of complement (33, 34, 37–40). In our experiments we did not observe any binding of MBL to immunoglobulins, not even in the patients who were positive for IgA or IgM, which are isotypes that have the ability to mediate such interaction. Moreover, genetic association studies (41) have also indicated that MBL is unlikely to play a major role in RA. Although ficolins could also activate the lectin pathway, we did not observe C4 deposition in C1q-depleted serum, which excludes a role for MBL- and ficolin-mediated complement activation. A possible limitation of our study is that we have analyzed anti-CCP antibodies present in serum for their ability to activate the complement system. These antibodies could be different from the antibodies present in the inflamed joint, since the latter could have different characteristics, such as a different glycosylation profile, a property which is known to influence the lectin pathway.
The observation that properdin was present on anti-CCP antibodies in the in vitro model is interesting in view of recent reports describing properdin as a recognition molecule, like C1q and MBL, in addition to its role as a stabilizing factor of the alternative pathway C3 convertase (30, 42–44). Although our data clearly indicate the involvement of properdin and thus the alternative pathway, they do not discriminate between the role of properdin as a recognition molecule and its role as part of the stabilized alternative pathway C3 convertase.
In mouse models of arthritis clear evidence has been provided that the alternative pathway in particular, and not the classical pathway, is responsible for organ damage, based on experiments using complement-deficient animals (25, 27) or a specific alternative pathway inhibitor (26). The present study provides evidence that human anti-CCP antibodies activate both the classical pathway and the alternative pathway and thereby mimic the results obtained from in vivo models. Taken together, these data suggest that the classical and alternative pathways, but not the lectin pathway, are involved in the effects mediated by arthritis-exacerbating antibodies. Although it is difficult to dissect the relative contribution of the alternative versus the classical pathway, since the alternative pathway also functions as an amplification of the classical pathway (45), our data suggest that at high serum concentrations the alternative pathway may contribute more than the classical pathway. This suggestion is based on the observation that there is almost no difference in C3 deposition between NHS in GVB++ and C1q-deficient serum in GVB++. However, dedicated experiments are needed to resolve this issue.
The complement system plays a complex role in physiology. Binding of complement fragments to immune complexes can mediate a plethora of effects, such as prevention of immune complex precipitation or solubilization of existing immune complexes, but it can also result in tissue pathology when complement activation takes place on tissue-bound immune complexes (13, 46). Because of these various aspects, it is currently not possible to conclude from these data that complement activation on anti-CCP antibodies contributes to tissue damage or is beneficial since it mediates immune complex clearance. Intriguingly, however, it has been reported that the extent of classical pathway activation is associated with disease activity, as shown by the correlation between C1q–C4 complexes and the Disease Activity Score in RA (46), indicating that complement activation could be causally related to the tissue damage observed in RA.
A recent trial of a C5a receptor inhibitor in RA (48) did not show efficacy. As with any negative study, it is difficult to understand why the study failed to show beneficial effects. Obviously, this could be due to the fact that this type of intervention is not effective. However, it is striking that in that particular study only 29% of the placebo control group and almost 90% of the C5a receptor antagonist–treated group were positive for anti-CCP antibodies. Since it is known that anti-CCP status is strongly correlated with disease outcome, it is difficult to compare these 2 groups.
We analyzed the complement-activating potential of anti-CCP antibodies in a group of 60 patients and observed that anti-CCP antibodies from all patients activated the complement system. These observations suggest that the complement system could be involved in RA pathogenesis in all anti-CCP–positive patients. We used a constant pool of NHS as the source of complement rather than using sera from the CCP-positive patients to activate complement. The major advantage of this approach is that it allowed confidence that the data were not influenced by potential complement consumption or drug-induced effects on the ability of the complement system to be activated.
In conclusion, anti-CCP antibodies found in patients with arthritis can activate the human complement system via both the classical and the alternative pathways but not via the lectin pathway. These data indicate that anti-CCP antibodies are capable of activating a major immune effector mechanism via which they may contribute to tissue damage.
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. Trouw 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. Trouw, Huizinga, Toes.
Acquisition of data. Trouw, Haisma, Levarht.
Analysis and interpretation of data. Trouw, van der Woude, Ioan-Facsinay, Daha, Huizinga, Toes.