Introduction
Rheumatoid arthritis (RA) is a progressive autoimmune disease characterized by chronic inflammation of the peripheral joints. It is a complex multifactorial pathology, in which genetic and environmental factors, like smoking, can play an important role in the onset of disease and the progression of the joint damage [
1,
2]. The presence of immune complexes (IC) in serum and synovial fluid (SF) of RA patients is likely to contribute to the pathogenesis of the disease and to articular damage, since they are responsible for the activation of complement, the stimulation of phagocytes through their Fc receptor and the release of chemotactic factors, cytokines, metalloproteinases and reactive oxygen intermediates [
3‐
6]. The formation of IC as such is not specifically related to autoimmune pathologies as it is a natural process, completing an immune response in the body. The antigen-antibody complexes are usually effectively removed by phagocytosis. However, it is known that an impaired clearance of these complexes can elicit or sustain an inflammatory response [
7,
8].
The pathological nature of IC has been suggested by several groups based on
in vitro studies. The effect of the SF IC from juvenile RA patients on healthy PBMCs was studied by Jarvis
et al. They found that especially the high molecular weight IC, separated by size exclusion chromatography from the other immunoglobulins and low molecular weight IC, were responsible for inducing a spectrum of pro-inflammatory cytokines, such as TNFα, IL-1β, IL6, IL8 and granulocyte-macrophage colony-stimulating factor (GM-CSF) [
9]. A comparison between IC from SF of RA patients, serum of RA patients and serum of healthy persons was made by Schuerwegh
et al. They demonstrated that IC isolated from RA serum and RA SF, in contrast to IC from healthy persons, had an effect on chondrocyte growth, NO production and apoptosis, thereby contributing directly to cartilage destruction in RA [
10]. Mathsson
et al. showed that polyethylene glycol (PEG) precipitated IC from RA SF induced the production of the pro-inflammatory cytokine TNFα in peripheral blood mononuclear cell (PBMC) cultures from healthy donors. When IC from RA serum or healthy serum were used, no elevated levels in TNFα could be seen [
11]. These reports show the relevance of IC in the joint destruction and the pathogenesis of RA.
The best known IC in RA is the rheumatoid factor (RF) bound to its antigen, the Fc domain of IgG. The RF, which is mainly IgM [
12], is used in diagnostic tests for RA and has a sensitivity of 78.6% and a specificity of 80.8% [
13]. The RF factor is also found in other diseases such as systemic sclerosis (20 to 30%) [
14] and occasionally in healthy persons (1.3 to 4%) [
5]. Besides the RF, immunoglobulins and complement factors, other components can also be present in IC from serum of RA patients. Indeed, recently, it has been shown that fibrinogen and citrulline-containing fibrinogen are present in the IC of RA patients [
15]. Because of the pathogenic nature of IC in RA, it is important to identify the antigens in these complexes. After identification of these antigens, a better understanding of the immunological process in the affected joints can be achieved.
Since anti-citrullinated protein/peptide antibodies (ACPA) are very specific for RA (specificity of 98%, sensitivity 68%) [
16] and high amounts of citrullinated proteins, like fibrinogen, have been detected in the joint of RA patients, it is likely that some antigens in IC of RA patients are citrullinated.
The isolation of IC and subsequent identification of the antigens is therefore of great importance in the understanding of RA. The isolation of IC from biological matrices has been tackled by many different techniques such as PEG precipitation [
10,
11,
17], C1q ELISA [
15] and immunoprecipitation [
18]. PEG precipitation is broadly used for the isolation of IC but the IC-fraction also contains a considerable amount of non immune complex (IC)-related proteins, such as albumine, haptoglobin and α1-antitrypsin [
17]. C1q ELISA will isolate IC that are bound to the C1q component of the complement and this method is gaining interest because of the high throughput possibilities. However, to capture IC by C1q ELISA, C1q must be present and accessible in the IC. IC from serum and SF can be isolated with a high purity by means of immunoprecipitation with proteinG, but it has the disadvantage of isolating the free immunoglobulins as well. For the identification of the antigens in IC, a sensitive method like mass spectrometry and immunodetection is necessary because of their low abundance.
In this report a broad range proteome approach, by means of mass spectrometry, is used in order to find new antigens in IC. Because of the low abundance of the antigens and the excess of immunoglobulins, it is possible that not all antigens will be detected by this approach, especially antigens that have a molecular weight that correspond to those of immunoglobulins. Therefore, a second, very sensitive method such as immunodetection on 2D-PAGE, was chosen to confirm the results of the broad range proteome approach and to investigate whether known antigens in RA (e.g. fibrinogenβ (Fibβ), fibrinogenγ (Fibγ), fibronectin and vimentin) are present in these complexes. Besides the high sensitivity of immunodetection, Western blot makes it also possible to visualize different isoforms of a certain protein.
Since not only the identification of the antigens, but also their citrullination status was of interest, the choice of antibodies for immunodetection was based on previous reports on citrullinated proteins either in serum, SF or synovial tissue of RA patients. The comparison between citrullinated proteins in serum and SF was already reported by Takizawa
et al. [
19]. In their study, soluble antigens were studied in RA serum, RA SF and osteo-arthritis (OA) SF. They could only identify citrullinated fibrinogen in RA SF. However, two years later, also citrullinated fibronectin and citrullinated vimentin were found as soluble antigens in RA SF and synovial tissue [
20‐
22]. Citrullinated fibronectin was also detected in RA SF and synovial exosomes [
23,
24]. Additionally, the presence of citrullinated Fibβ and Fibγ in RA synovium has been reported by Matsuo
et al. [
25]. Based on these findings, immunodetection was performed with anti-Fibβ, anti-Fibγ, anti-fibronectin and anti-vimentin antibodies on 2D-PAGE with IC, followed by anti-modified citrulline (AMC) detection.
The citrullination of the antigens perfectly fits the model for the development and chronic nature of RA proposed by van Venrooij and Pruijn. They divided the process of autoimmunity in RA into five steps: an innocent inflammation in combination with massive apoptosis or impaired clearance can lead to the elevation of cytosolic Ca
2+ concentrations (1) followed by the activation of peptidylarginine deiminase (PAD) and the citrullination of proteins (2). When citrullinated antigens are presented to T cells, the production of ACPAs is triggered (3). Immune complexes can be formed if the antigens react with the auto-antibodies (4). These IC stimulate inflammatory processes (5) and cause a vicious circle of inflammation resulting in joint destruction for years [
26].
This study describes the isolation and characterization of antigens residing in IC of RA patients. We found that circulating IC in the serum of RA patients and healthy controls contain Fibβ and fibronectin, both in a non-citrullinated form. In IC, isolated from RA SF, on the other hand, Fibrinogenβ, Fibrinogenγ, fibronectin and vimentin were identified. More importantly, vimentin and a minor portion of Fibβ were found to be citrullinated in the isolated complexes from RA SF. However, these citrullinated antigens were only found in IC from SF of ACPA+ RA patients, while no citrullinated antigens were found in IC from SF of ACPA- RA patients or SpA patients.
Materials and methods
Patients and controls
Serum and synovial fluid were collected from patients fulfilling the American College of Rheumatology criteria for RA [
27] and European Spondyloarthropathy Study Group criteria for SpA [
28] (for patient information see Table
1). Sera from healthy donors were used as controls. Informed consent was obtained from patients and healthy controls and the study was approved by the local ethics committee. Detailed information on the identity of the samples used in each experiment is provided in Table
1. RF titers were determined with the Waaler Rose test and ACPA titers were measured with anti-CCP-EliA (Phadia, Freiburg, Germany).
Table 1
Rheumatoid factor and CCP values of RA patients
RA1 | results 1&2 | RA | 1,280 | 2,839 |
RA2 | results 1&2 | RA | 320 | 265 |
RA3 | results 1&2 | RA | 0 | > 1,600 |
SF1 | results 3 | RA | 6 | 0 |
SF2 | results 1&2&3 | RA | 80 | 1 |
SF3 | results 3 | RA | 0 | 1 |
SF4 | results 3 | RA | 0 | 1,6 |
SF5 | results 3 | RA | 0 | 2 |
SF6 | results 3 | RA | 5 | 2 |
SF7 | results 3 | RA | 0 | 3 |
SF8 | results 3 | RA | 40 | 4 |
SF9 | results 3 | RA | 0 | 5 |
SF10 | results 3 | RA | 0 | 5 |
SF11 | results 3 | RA | 80 | 7 |
SF12 | results 3 | RA | 160 | 10 |
SF13 | results 1&2&3 | RA | 351 | 107 |
SF14 | results 3 | RA | 183 | 340 |
SF15 | results 1&2&3 | RA | 1,280 | 533 |
SF16 | results 3 | RA | 1,280 | 608 |
SF17 | results 3 | RA | 1,280 | 710 |
SF18 | results 3 | RA | 2,560 | 740 |
SF19 | results 3 | RA | 10,240 | 767 |
SF20 | results 1 | RA | 1,280 | 1,141 |
SF21 | results 3 | RA | 640 | 1,294 |
SF22 | results 3 | RA | 0 | > 1,600 |
SF23 | results 3 | RA | 160 | > 1,600 |
SF24 | results 3 | RA | 0 | > 1,600 |
SF25 | results 3 | RA | 640 | 1,775 |
SF26 | results 2 | RA | 227 | ND |
SF27 | results 3 | SpA | 0 | 0 |
SF28 | results 3 | SpA | ND | 1,6 |
SF29 | results 3 | SpA | 0 | 2 |
SF30 | results 3 | SpA | 0 | 2 |
SF31 | results 3 | SpA | 0 | 3 |
SF32 | results 3 | SpA | 0 | 4 |
SF33 | results 3 | SpA | 0 | 11 |
SF34 | results 3 | SpA | 0 | ND |
SF35 | results 3 | SpA | ND | ND |
SF36 | results 3 | SpA | 0 | ND |
SF37 | results 3 | SpA | 0 | ND |
SF38 | results 3 | SpA | 0 | ND |
Immunoprecipitation with Immobilized Protein G
IC were further purified by affinity immunoprecipitation with Immobilized Protein G (Pierce, Rockford, IL, USA). 400 μL beads were washed twice with 500 μL phosphate buffered saline (PBS). A total of 50 μL serum or SF was mixed with 450 μL PBS and added to the beads. Sample and beads were placed on a rocker for four hours at 4°C. The beads with the bound IC were washed five times in 500 μL PBS. The pellet of protein G beads was resuspended in reducing Laemmli sample buffer for five minutes at 95°C. After centrifugation (5 minutes, 460 g) the supernatant was stored at -20°C.
Protein concentrations were determined by Coomassie (Bradford) Protein Assay (Pierce, Rockford, IL, USA) and 2 D Quant (GE Healthcare, Uppsala, Sweden).
One-dimensional gel electrophoresis (1D-PAGE)
Protein samples were dissolved in Laemmli buffer (50 mM TrisHCl, pH 6.8, 2% SDS, 10% glycerol, bromophenol blue) with 5% β-mercapto-ethanol and incubated at 95°C for five minutes. The samples were loaded on a 10% TrisHCl polyacrylamidegel (Biorad, Hercules, CA, USA) and electrophoresis was performed by applying 150 V for 30 minutes, followed by 200 V for one hour.
Two-dimensional gel electrophoresis (2D-PAGE)
For 2D-PAGE, protein samples were first precipitated overnight in acetone at -20°C. After centrifugation at 18,000 g for 10 minutes the samples were air dried. A total of 100 μg was resuspended in 200 μL rehydration buffer (7 M Urea, 2 M Thiourea, 2% CHAPS, 0.2% carrier ampholytes, 100 mM DTT, bromophenol blue). The sample was introduced passively in an IPG strip (11 cm, pH 4 to 7) (Biorad) as previously described [
29]. Iso-electric focussing was performed in a Protean IEF Chamber (Biorad) according to the following program: 100 V, 30 minutes, linear voltage slope - 250 V, 30 minutes, linear - 500 V, one hour, linear - 1,000 V, one hour, linear - 8,000 V, four hours, rapid - 8,000 V, 35,000 V hours, rapid - 500 V, 20 h, rapid. Subsequently the strips were equilibrated in equilibration buffer (50 mM TrisHCl, pH 8.8, 6 M Urea, 20% glycerol, 2% SDS) containing 1.5% DTT for 15 minutes, followed by 4% IAA in equilibration buffer for 15 minutes.
Gel electrophoresis was carried out on a 10% TrisHCl PAGE using 150 V for 30 minutes, followed by 200 V for one hour.
Western blot
After a 15-minute equilibration of the gels and the nitrocellulose membranes (Biorad) in CAPS (pH 11), electrophoretic transfer of proteins was performed by tank blotting in a Trans Blot Cell (Biorad), with CAPS (pH = 11), at 50 V for three hours. Successful transfer of proteins was checked by means of Ponceau S staining.
Detection of citrullinated proteins
The presence of citrullinated proteins on the nitrocellulose blots was detected using the anti-modified citrulline (AMC) detection kit (Upstate, Charlottesville, VA, USA) according to the manufacturer's protocol. Each AMC detection was accompanied with a positive control, as indicated in the manufacturer's protocol.
Protein identification
Visualization of proteins in the gels was performed using Sypro Ruby Protein Gel staining (Invitrogen, Carlsbad, CA, USA) for at least three hours after a 30-minute fixation in a 10% MeOH, 7% acetic acid solution. After staining, the gel was washed twice with a 10% MeOH, 7% acetic acid solution. Proteins of interest were excised from the gel and digested with modified sequence grade porcine trypsin (Promega, Madison, WI, USA) as described earlier [
30]. Proteins were analyzed and identified by LC-MSMS, using a Q-TOF Ultima Mass spectrometer (Waters, Milford, MA, USA) combined with ESI source. The data were processed using Mascot Distiller and searched against the Swissprot human database, using the in-house mascot daemon searching algorithm. Identification was considered positive with a
P-value < 0.05.
Immunodetection
Before immunodetection, each blot was blocked for one hour in 0.3% Tween-20 in PBS. Vimentin was detected with the mouse anti-human vimentin antibody (clone V9, Sigma, St. Louis, MI, USA) at a concentration of 1/400 in 0.3% Tween-20 in PBS. After overnight incubation, vimentine was detected with HRP labelled goat anti-mouse IgG followed by ECL detection. Detection of Fibβ, Fibγ and fibronectin was performed using respectively rabbit anti-human Fibβ, rabbit anti-human Fibγ and rabbit anti-human fibronectin. Anti-rabbit HRP labelled antibody was used as a secondary antibody. ECL detection was carried out by means of Supersignal West Dura Extended Duration Substrate (Pierce).
Following each immunodetection, the blot was stripped for 30 minutes at 50°C with stripping buffer (2% SDS, 0.1 M β-mercapto-ethanol, 0.05 M Tris pH 6.8) and washed three times with 0.3% Tween20 in PBS. To check the stripping efficiency, the blot was re-incubated with secondary antibody and detected with ECL. Afterwards the blot was stripped for another 15 minutes, before incubation with a new primary antibody. Additionally, the sequence of antibodies used for immunodetection varied throughout the different experiments in order to exclude false positive results. Protein patterns were scanned and digitized using the VersaDoc Imaging System (Biorad).
Discussion
One of the purpose of this study was to identify the antigens in IC from serum of RA patients. However, the joint is the primary target in the pathology of RA and high amounts of antibodies [
32,
33], citrullinated proteins [
19‐
21], and IC [
10,
34] are found in RA synovial fluid. Moreover, Wipke
et al. showed that IC in the joint are necessary to initiate inflammation [
8]. Investigation of the IC in SF could thus be even more informative than analysing serum or plasma.
In order to isolate IC, immunoprecipitation was performed on serum and SF. However, a major drawback of this technique is the co-purification of high amounts of free immunoglobulins. Previous experiments had already shown that this contaminating factor caused interference during further identification of the antigen by mass spectrometry (data not shown). To separate immunoglobulins from other proteins, 1D-PAGE in combination with in gel digestion and mass spectrometry was performed as a broad range proteome approach in order to detect new potential antigen in IC. Despite the high abundance of immunoglobulins in the purified IC, Fibβ was detected in the IC of RA SF as well as in the flow-through. At the MW of Fibβ, a clear positive AMC staining on Western blot was detected in the isolated IC fraction, while no citrullinated proteins were detected in the flow-through. This indicates that Fibβ could be present in a citrullinated form in IC in SF of RA patients. In RA serum, however, no citrullinated proteins were detected in the IC fraction. This was in contrast to the findings of Zhao
et al. [
15]. They reported the presence of citrullinated Fibβ in IC from plasma of CCP+ RA patients and confirmed their presence in synovial tissue. The discrepancy in results might be explained by the difference in isolation method and the fact that they worked with plasma instead of serum. Indeed because serum does not contain any clotting factors, some of the fibrinogen, citrullinated or native, free or in immune complexes, might have already been removed during clotting. Additionally, patient variability or disease status variation can not be excluded as contributing factors.
By means of this broad range approach, Fibβ, possibly citrullinated, was found as an antigen in SF from RA patients. However, it should be noted that due to the low abundance of the antigens and the excess of immunoglobulins, some antigens will not be detected in this approach, especially the antigens that co-migrate with immunoglobulins on 1 D SDS PAGE.
Therefore, in a second approach, 2 D PAGE was combined with immunodetection to overcome the interference of the immunoglobulins as well as the low abundance of the antigens. By means of 2 D PAGE and immunodetection we wanted to confirm the previous results from the broad proteome approach and also analyse the presence of well-known antigens in RA, such as Fibβ, Fibγ, fibronectin and vimentin.
Fibβ and fibronectin were found in IC obtained from healthy serum, RA serum and RA SF, indicating that the presence of Fibβ and fibronectin in IC is not specific for RA or RA SF. Fibγ and vimentin could only be detected in the IC from RA SF (Figure
2). The extra isoforms of Fibβ (Figure
2a, box x) were also exclusively found in SF. Besides the presence/identification of the antigens in IC, the citrullination status of these antigens was analyzed. No citrullinated proteins were detected in IC of RA serum, while RA SF contained different citrullinated proteins, which confirmed previous 1D-PAGE analysis. The citrullinated proteins in IC from RA SF were identified as vimentin and Fibβ. The fact that vimentin was found as antigen during 2 D analysis and not during the broad range proteome approach with 1 D SDS PAGE, can be explained by the fact that vimentin has the same molecular weight as the heavy chain of immunoglobulins. The detection of citrullinated Fibβ on 2D-PAGE confirmed our results from the broad range approach. During later analysis we found that citrullinated Fibβ could only be detected in patient 13 (SF13) (Table
2), which was present in the pool for 1D-PAGE as well as the pool for 2D-PAGE. However, it should be noted that fibrinogen, citrullinated or not, could possibly be deposited in synovium tissue and thereby not, or in a lesser extend, be detected in synovial fluid.
Next, we analysed individual SF samples from CCP+ RA patients and CCP- RA patients. SpA patients were included as disease controls. Since fibronectin was present in healthy serum as well as RA serum and RA SF, we focussed on Fibβ, Fibγ and vimentin during further analysis. When individual samples of CCP+, CCP- and SpA synovial fluid were processed, we observed absence of citrullinated proteins in IC in the CCP- patients and SpA patients (n = 24) while half of the CCP+ patients contained citrullinated vimentin or Fibβ.
Citrullinated fibrinogen is known to be present in SF of RA patients [
25]. Our data showed that in IC of SF only Fibβ was citrullinated and that Fibγ was present exclusively in IC of SF, but in a non-citrullinated form. The Fibβ isoforms detected on the AMC blot of one patient, however, were only a minor fraction of the Fibβ isoforms detected with anti-Fibβ. The lower abundance of Fibβ could be due to residing fibrinogen in the synovial tissue.
A remarkable difference between our data and previous reports is the presence of citrullinated vimentin in IC of CCP+ RA SF. This antigen could not be detected in IC of healthy or RA serum. Citrullinated vimentin is known to be an important antigen in RA [
35] and is present in synovial fluid [
21], but its presence in IC has not been reported. Moreover, no citrullinated antigens were found in IC of SF from our control groups, consisting of CCP- RA patients and SpA patients.
Since CCP+ patients have a more destructive course of disease and because CCP+ patients contain citrullinated vimentin in their IC, we hypothesize that citrullinated vimentin plays an important pathophysiological role in the perpetuation of RA. Moreover, in contrast to Fibβ, vimentin is an intracellular antigen and therefore not expected in IC. Additionally, the presence of intracellular citrullinated proteins in the synovium is specific for RA, while extracellular citrullinated proteins lack this specificity [
36]. These RA specific synovial intracellular citrullinated proteins are also associated with significantly higher systemic and local ACPA levels and with local ACPA production in the joint [
36].
The locations where the authors' completed their education follow: KVS, PharmD, Ghent University, Belgium; KT, MSc, PhD, Ghent University Hospital, Belgium; MDC, PharmD, Ghent University, Belgium; DE, MD, PhD, Ghent University Hospital, Belgium; FDK, MD, PhD, Ghent University Hospital, Belgium; DD, PharmD, PhD, Ghent University, Belgium
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KVS performed most of the practical work and data-analysis and wrote the manuscript. KT assisted in performing ultracentrifugation, immunoprecipitation and Western blot analysis. She also helped to draft the manuscript. MDC gave practical assistance during the experiments. DE, FDK and DD helped in the design of the study and the critical analysis of the data. All authors read and approved the manuscript.