The rheumatoid arthritis synovial fluid citrullinome reveals novel citrullinated epitopes in apolipoprotein E, myeloid nuclear differentiation antigen, and β-actin

Authors

  • Joyce J. B. C. van Beers,

    1. Nijmegen Center for Molecular Life Sciences and Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, The Netherlands
    Search for more papers by this author
  • Carla M. Schwarte,

    1. Nijmegen Center for Molecular Life Sciences and Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, The Netherlands
    Search for more papers by this author
  • Judith Stammen-Vogelzangs,

    1. Nijmegen Center for Molecular Life Sciences and Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, The Netherlands
    Search for more papers by this author
  • Els Oosterink,

    1. Nijmegen Center for Molecular Life Sciences and Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, The Netherlands
    Search for more papers by this author
  • Borut Božič,

    1. University Medical Centre Ljubljana, Immunology Laboratory, Ljubljana, Slovenia
    Search for more papers by this author
  • Ger J. M. Pruijn

    Corresponding author
    1. Nijmegen Center for Molecular Life Sciences and Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, The Netherlands
    • 271 Department of Biomolecular Chemistry, PO Box 9101, NL-6500 HB Nijmegen, The Netherlands
    Search for more papers by this author

Abstract

Objective

To generate a catalog of citrullinated proteins that are present in the synovia of patients with rheumatoid arthritis (RA) and to elucidate their relevance for the anti–citrullinated protein antibody response in RA.

Methods

Polypeptides isolated from the synovial fluid of patients with RA were identified by mass spectrometry. Three proteins (apolipoprotein E [Apo E], myeloid nuclear differentiation antigen [MNDA], and β-actin) were studied in more detail, using immunoprecipitation and Western blotting. The presence of autoantibodies to synthetic peptides derived from these proteins in sera from patients with RA, sera from patients with other diseases, and sera from healthy control subjects was studied by enzyme-linked immunosorbent assay (ELISA).

Results

RA synovial fluid samples displayed several distinct patterns of citrullinated proteins. Using mass spectrometry, (fragments of) 192 proteins were identified, including 53 citrullinated proteins, some of which contained multiple citrullinated residues. In addition to previously reported citrullinated proteins in RA synovia (e.g., vimentin and fibrinogen), a series of novel citrullinated proteins, including Apo E, MNDA, β-actin, and cyclophilin A, was identified. Immunoprecipitation experiments confirmed the citrullination of Apo E and MNDA. ELISAs demonstrated the presence of autoreactive citrullinated epitopes in Apo E, MNDA, and β-actin.

Conclusion

Synovial fluid samples from the inflamed joints of patients with RA contain many citrullinated proteins. Citrullinated Apo E, MNDA, and β-actin are novel antigens identified in RA synovial fluid, and only a limited number of their citrullinated epitopes are targeted by the immune system in RA.

Rheumatoid arthritis (RA) is a common autoimmune disease that is characterized by chronic inflammation of the synovial joints. The cause of RA remains unknown. Several factors, however, have been proposed to play a role in the pathogenesis of RA, including environmental (e.g., smoking), genetic (e.g., HLA–DRB1 shared-epitope alleles), and hormonal (female sex) factors (1, 2). A specific feature of the immune response in RA is the presence of anti–citrullinated protein antibodies (ACPAs) in patient sera. These antibodies can appear several years prior to disease onset, and increasing evidence suggests that they play a role in the pathophysiology of RA (3–6).

Because the initial studies to detect ACPAs in patient sera were performed with citrullinated molecules that do not occur in the inflamed joints of patients (e.g., citrullinated filaggrin and citrullinated synthetic peptides from random peptide libraries), identification of inflammation-associated citrullinated proteins severely lagged behind ACPA characterization. Several candidate citrullinated autoantigens have already been described (e.g., fibrin[ogen], vimentin, fibronectin, α-enolase, and others), but their roles in the pathogenesis of RA remain unclear (7–10). The presence of ACPAs reactive with the majority of the identified citrullinated antigens in RA sera has been described previously (11). Although citrullinated proteins have been reported to appear in most if not all inflamed tissue, the formation of citrullinated protein–specific immune complexes will be largely restricted to RA as a result of the disease-specific production of ACPAs. In view of the putative pathophysiologic role of such complexes, it is important to obtain a comprehensive view of the citrullinated proteins present in the inflamed joints of patients with RA and their recognition by ACPAs.

In this study, the presence of citrullinated proteins and fragments thereof in the synovial fluid (SF) of patients with RA was investigated by Western blotting and mass spectrometry (MS). This approach yielded an inventory of citrullinated proteins (citrullinome) in the inflamed joints of patients with RA. The importance of the identified citrullinated residues for the recognition of 3 of these proteins by patient sera was assessed by enzyme-linked immunosorbent assay (ELISA) with synthetic peptides.

PATIENTS AND METHODS

Patient SF and sera.

Synovial fluid samples were obtained by joint puncture (arthrocentesis), using needles with a diameter of 1.6–2.2 mm. A sample was immediately centrifuged at 2,500g for 10 minutes at 4°C. The supernatant and pellet were stored separately in sterile containers at −80°C. All procedures were finished in <2 hours. The pellet fraction was resuspended in EGTA lysis buffer (50 mM Tris HCl, pH 7.4, 100 mM KCl, 1 mM dithioerythritol, 0.1% Nonidet P40 [NP40], 10 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and Complete Protease Inhibitor Cocktail [Roche]). Supernatant fractions were diluted with 4 volumes of EGTA lysis buffer. After sonification, sodium dodecyl sulfate (SDS) was added (final concentration 2%), and the fractions were heated and centrifuged at 12,000g. Supernatants were used for further analysis. All samples were obtained from patients for diagnostic and therapeutic purposes. No sample was obtained for research only, and all patients signed an informed consent that the remainder of biologic samples obtained for diagnostic purposes may be used anonymously for research. The study was approved by the local Medical Ethics Board of the University Medical Center Ljubljana.

Sera were collected from patients with RA (n = 80), patients with systemic lupus erythematosus (SLE; n = 32), and patients with primary Sjögren's syndrome (SS; n = 32) at the Department of Rheumatology of Radboud University Nijmegen Medical Centre and at St. Maartenskliniek Nijmegen. Sera from patients with multiple sclerosis (n = 32) were collected at the Multiple Sclerosis Center Nijmegen. Sera from patients with type 1 diabetes mellitus (n = 32) were collected at the Department of Internal Medicine of Radboud University Nijmegen Medical Centre. Sera from healthy individuals (n = 24) were purchased from the Sanquin blood bank (Nijmegen). Sera were stored at −80°C until used. The need for ethics approval to use these sera for autoantibody studies was waived by the local ethics committee; patient consent was obtained.

Western blotting.

Synovial fluid samples (10 μg total protein per sample) were separated by SDS–polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes. The blots were incubated with blocking buffer (5% nonfat dry milk and 0.1% NP40 in phosphate buffered saline) for 1 hour and incubated with the antibody of interest (anti–myeloid nuclear differentiation antigen [anti-MNDA], sc-6051 [Santa Cruz Biotechnology]; anti–apolipoprotein E [anti–Apo E], EP1374Y [Epitomics]; anti–β-actin, A5441 [Sigma], or anti–modified citrulline [AMC] antibody [Upstate]) in blocking buffer for 1–3 hours. After washing with blocking buffer, proteins were detected with an IR Dye 800–labeled secondary antibody (Molecular Probes), followed by infrared imaging (Li-Cor Odyssey infrared imaging system). For the detection of citrullinated proteins, blots were chemically modified as described previously (12), prior to incubation with the AMC antibody.

Mass spectrometry.

Two SF samples were depleted of albumin as described by Colantonio and coworkers (13), separated by SDS-PAGE, and stained with colloidal Coomassie brilliant blue. Gel slices were incubated with 20 μl trypsin solution (15 ng/μl trypsin in 25 mM NH4HCO3 and 5 mM n-octylglucoside). Peptides were extracted by adding 50% acetonitrile, 0.5% trifluoroacetic acid, and 5 mM n-octylglucoside, followed by sonification. The protein digests obtained in this manner were used for nano–liquid chromatography tandem MS (nano-LC-MS/MS) with an LTQ Fourier Transform Ion Cyclotron Resonance Mass spectrometer. Data were converted by BioWorks Sequest (Thermo Electron Corporation) into a peak list, which allowed peptide identification with the Mascot search program (version 2.1.03; Matrix Science) and the NCBInr database (www.ncbi.nlm.nih.gov) using Homo sapiens taxonomy. Mass deviations for precursor ions were set at 20 ppm, deviations for the mass of fragment ions were set at 0.8 dalton, and the maximal number of missed cleavage sites was set at 2. Technical details of the LC-MS/MS analysis and database search are available at the Arthritis & Rheumatism Web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Immunoprecipitation.

Mouse monoclonal anti–Apo E antibody, goat polyclonal anti-MNDA antibody, and mouse monoclonal anti–β-actin were coupled to either protein A– agarose beads (Kem-En-Tec) or protein G–agarose beads (Thermo Scientific). Antibodies were coupled to beads in IPP 500 (500 mM NaCl, 10 mM Tris HCl, pH 8.0, 0.1% Tween 20, and 0.1% NP40) by incubation for 1 hour at room temperature. Beads were washed once with IPP 500 and twice with IPP 150 (150 mM NaCl, 10 mM Tris HCl, pH 8.0, 0.1% Tween 20, and 0.1% NP40). Subsequently, beads were incubated with SF, 5-fold diluted in IPP 150, for 2 hours at 4°C. Finally, beads were washed 3 times with IPP 150, after which the immunoprecipitated proteins were analyzed by SDS-PAGE and Western blotting.

Enzyme-linked immunosorbent assay.

Citrullinated peptides were synthesized based on the identified citrullination sites of Apo E, MNDA, and β-actin. These peptides (which were not purified by high-performance liquid chromatography) were used in the ELISA. Additional information is available at the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. Cutoff values were determined based on the values obtained for healthy control sera, as described previously (14).

Statistical analysis.

Statistical analyses were performed using an unpaired 2-tailed t-test. P values less than 0.05 were considered significant. Linear discriminant analysis was performed using DTREG software, version 10.6.0 (www.dtreg.com).

RESULTS

Citrullinated proteins in RA SF samples.

RA SF samples were separated by centrifugation into soluble and insoluble fractions. The presence of citrullinated proteins in both fractions was analyzed by Western blotting using AMC antibodies, after chemical modification of peptidylcitrullines on the blot. As shown in Figure 1, many citrullinated proteins were detected in RA SF samples. Moreover, different RA SF samples often showed different citrullination patterns. Some SF samples contained a broad range of citrullinated proteins (e.g., SF pellet fraction 10 [SFP10] and SFP11), whereas other SFs contained only a limited number of citrullinated proteins (e.g., SFP12 and SFP13). In addition, intermediate citrullination patterns were observed (e.g., SF supernatant fraction 2 [SFS2] and SFS7). The SFs containing the insoluble SF fractions (SFP) tended to show the most extreme citrullination patterns (either a broad range of citrullinated proteins or only few low molecular weight citrullinated proteins), while the soluble SF fractions (SFS) displayed more intermediate citrullination patterns. Some samples did not show detectable levels of citrullinated proteins (e.g., SFP4), even though these SF samples contained similar protein levels (Figure 1A).

Figure 1.

Citrullinated proteins in rheumatoid arthritis (RA) synovial fluid (SF). SF samples from patients with RA were fractionated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and analyzed by ponceau S staining (A) and immunoblotting using anti–modified citrulline antibodies (B) after citrulline modification on the blot. As a positive control, citrullinated histones (H; 200 ng and 1 μg) were analyzed in parallel. SFP = SF pellet fraction; SFS = SF supernatant fraction. The positions of molecular weight markers (kd) are indicated on the left.

Global identification of proteins and citrullinated proteins in RA SF.

After Western blot analysis was performed to screen RA SF for the presence of citrullinated proteins, SF samples from 2 patients (SFS9 [soluble fraction] and SFP38 [insoluble fraction]) were selected based on the presence of multiple citrullinated proteins. The depletion of albumin resulted in 2 fractions (p1 and p2) per sample (13). The polypeptides in these samples were separated by SDS-PAGE, after which each lane was dissected into 18 slices (Figure 2). After digestion with trypsin, the resulting peptides were eluted and analyzed by LC-MS/MS to determine the identity of the proteins and the positions of citrullinated residues (citrullination was set as one of the variable modifications to detect). The fragmentation pattern of each peptide was manually investigated to confirm the presence of peptidylcitrulline (1 dalton difference compared with peptidylarginine) and to exclude the possibility that the mass difference was attributable to deamidation in the same tryptic peptide.

Figure 2.

Fractionation of 2 RA SF samples for citrullinome analysis. Two SF samples (SFS9 and SFP38) from patients with RA were selected for (citrullinated) protein identification by liquid chromatography mass spectrometry (LC-MS) analysis. These samples were depleted for albumin, resulting in 2 pellet fractions (p1 and p2) and a supernatant fraction containing most of the albumin. All 4 samples (pellet fractions for both patients) were fractionated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and stained with colloidal Coomassie brilliant blue (cCBB) or analyzed with anti–modified citrulline (AMC) antibodies on Western blots. The cCBB-stained gel was divided into 18 slices, which were in-gel digested with trypsin, after which the resulting peptides were analyzed by LC-MS. The presence of citrullinated proteins (or fragments thereof) in the gel slices is indicated on the right of each panel. The positions of molecular weight markers (kd) are indicated on the left of each panel. Apo E = apolipoprotein E; MNDA = myeloid nuclear differentiation antigen (see Figure 1 for other definitions).

A total of 192 proteins with Mascot scores of ≥30 were identified in the SF samples, irrespective of their citrullination; 40 and 45 of these proteins were detected solely in the material from the soluble fraction (SFS9) and the insoluble fraction (SFP38), respectively. The identified proteins could be categorized into 14 groups according to their function. These groups included proteins that play a role in the immune response (n = 40), glycolysis (n = 9), signal transduction (n = 4), lipid metabolism (n = 8), blood coagulation (n = 13), cell mobility/structure/integrity (n = 35), cell death (n = 4), protein folding (n = 6), protein synthesis (n = 7), vesicle formation/protein transport (n = 6), molecular transport (n = 16), protease inhibition (n = 6), and cell adhesion (n = 4), and proteins that could be categorized according to other functions (n = 34). Additional information is available in Supplementary Table 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. Peptide sequences for many of these proteins were observed not only in material from the gel slice corresponding to the known molecular weight for the full-length protein but also in ≥1 slices corresponding to lower molecular weights. This strongly suggested that many proteins are subjected to proteolytic cleavage events in the SF. A comparison of the proteins detected in both patient samples demonstrated that 107 proteins (56%) were present in both samples, in spite of the fact that the starting material differed with regard to solubility.

Although the majority of the peptides identified by LC-MS/MS did not contain citrullinated residues, citrullinated residues were detected in peptides from 53 proteins (28% of all proteins identified); 35 proteins were detected solely in the material from SFS9, and 12 were detected only in the material from SFP38 (Table 1); additional information is available in Supplementary Table 2, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. The citrullinated proteins could be divided into 4 major groups, based on their function (Supplementary Table 1). The majority of the citrullinated proteins were involved in the immune response (n = 14); cell mobility, structure, or cell integrity (n = 12); lipid transport (n = 5); or the coagulation cascade (n = 4). The remaining citrullinated proteins were divided randomly over the other functional groups.

Table 1. Citrullinated polypeptides detected in rheumatoid arthritis synovium*
Citrullinated proteinUniProtKBSFS9SFP38MW × 10−3
  • *

    SFS9 = synovial fluid supernatant fraction 9; SFP38 = synovial fluid pellet fraction 38.

  • Derived from the positions in the sodium dodecyl sulfate–polyacrylamide gels containing SF material.

  • The protein was present in the sample, but no citrullinated residues were detected.

40S ribosomal protein S16P62249+∼14
ADP-ribosylation factor 1P84077+∼14
Alpha-1-antitrypsin (serpin A1)P01009+∼43
Alpha-2-macroglobulinP01023+>94−43
Antithrombin-IIIP01008+∼40
Apolipoprotein A-IP02647+∼20
Apolipoprotein A-IVP06727+∼40
Apolipoprotein B-100P04114+>94
Apolipoprotein EP02649++43−30
Apolipoprotein L-IO14791+∼35
Beta-actinP60709++43−<14
Cathepsin GP08311+20−<14
Clusterin (apolipoprotein J)P10909+∼35
Complement C3P01024+∼75
Complement component C4-AP0C0L4+∼30
Complement factor BP00751+∼43
Complement factor HP08603+>94−43
Complement factor H-related protein 1Q03591+∼20
CoroninP31146+∼43
Cyclophilin A (peptidylprolyl isomerase A)P62937++∼14
Fibrinogen, alpha-chainP02671++67−<14
Fibrinogen, beta-chainP02675+<14
FibronectinP02751++>94−67
Galectin-2P05162+<14
Gamma-actinP63261+∼43
GelsolinP06396+∼43
Glia maturation factor gammaO60234+∼20
Glyceraldehyde-3-phosphate dehydrogenaseP04406+43−20
Hexokinase 3P52790+∼80
Histone cluster H1.3P16402+<14
Histone H2A.JQ9BTM1+<14
Histone H2B type 2-EQ16778+<14
Histone H3.3P84243+<14
Ig mu chain C regionP01871+67−43
Immunoglobulin lambda locusQ6PJF2+∼25
Inter-alpha-trypsin inhibitor heavy chain H2P19823+>94−50
Inter-alpha-trypsin inhibitor heavy chain H4Q14624+>94−40
Keratin, type I cytoskeletal 9P35527+∼67
Keratin, type II cytoskeletal 1P04264+∼80
Kininogen 1 light chainP01042-2+∼67
Myeloid cell nuclear differentiation antigenP41218++43−14
MyeloperoxidaseP05164+∼43
Myosin 9P35579+∼80
PeriostinQ15063+∼70
Probable E3 ubiquitin-protein ligase MYCBP2O75592+∼94
Proteoglycan 4Q92954+∼14
Profilin 1P07737+∼14
Prothrombin (coagulation factor II)P00734+∼43
Pyruvate kinaseP14618+∼43
Rho GDP-dissociation inhibitor 2P52566+∼20
Serum amyloid A4P35542+∼14
VimentinP08670+43−30
VinculinP18206+∼94

Six citrullinated polypeptides, Apo E, β-actin, cyclophilin A (CYP-A), fibrinogen α-chain, fibronectin, and MNDA, were detected in both patients (Figure 2 and Table 1). The majority of the citrullinated polypeptides corresponded to fragments of the total protein, based on their positions in the gel slices. Some of the citrullinated proteins have previously been identified in RA, such as fibrinogen (9, 10), vimentin (15), and fibronectin (16). The majority, however, have not been previously described to occur in a citrullinated form in the inflamed joints of patients with RA. There was some overlap between the data obtained with material from both patients (Table 1 and Supplementary Table 2). Citrullinated fibronectin and fibrinogen were detected in both RA SF samples. Citrullinated vimentin and histones were detected in only 1 of the 2 samples analyzed. For 4 other proteins (Apo E, MNDA, β-actin, and CYP-A), citrullinated peptides were detected in both patient samples (Figure 3); additional information is available in Supplementary Table 3, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. For all of the citrullinated peptides derived from Apo E, MNDA, β-actin, and CYP-A, the unmodified forms were also detected, with the exception of the unmodified counterpart of the peptide containing a citrulline at position 204 of MNDA.

Figure 3.

Citrullinated residues in apolipoprotein E (A), myeloid nuclear differentiation antigen (B), β-actin (C), and cyclophilin A (D) in rheumatoid arthritis synovial fluid (RA SF). Shaded areas represent sequences that were covered in both RA SF samples (SF supernatant fraction 9 [SFS9] and SF pellet fraction 38 [SFP38]). Citrullinated residues are indicated with a black-boxed R.

Citrullinated Apo E, MNDA, and β-actin in multiple RA SF samples.

Because only 2 RA SF samples were used to identify citrullinated proteins, we next investigated the presence of Apo E, MNDA, and β-actin in multiple RA samples, using Western blotting with specific antibodies. The results showed that Apo E, MNDA, and β-actin were readily detected in the SF samples of virtually all patients (Figure 4). To elucidate the citrullination state of these proteins, they were immunoprecipitated using antibodies to each of these proteins, followed by Western blotting with the AMC antibody (after chemical modification of the citrulline residues on the blot). Citrullinated Apo E was detected in 4 of 10 SF samples, and MNDA was observed in 3 of 10 SF samples (Figures 4D and E). Control Western blots demonstrated that Apo E and MNDA were precipitated from each of the samples used. In contrast, the anti–β-actin antibody failed to precipitate the protein, which means that no conclusions can be drawn regarding the prevalence of citrullinated β-actin in RA SF.

Figure 4.

Detection of citrullinated apolipoprotein E (Apo E), myeloid nuclear differentiation antigen (MNDA), and β-actin in multiple RA SF samples. The presence of Apo E (A), MNDA (B), and β-actin (C) in RA SF samples was analyzed by Western blotting, using specific antibodies to each of these proteins. To analyze their citrullination state, Apo E (D) and MNDA (E) were immunoprecipitated from 10 RA SF samples and subjected to Western blot analysis with anti–modified citrulline antibodies. As a positive control (+), citrullinated soybean trypsin inhibitor was analyzed in parallel. See Figure 3 for other definitions.

Autoreactivity of citrullinated Apo E, MNDA, and β-actin.

Twelve peptide sets, each comprising a citrulline-containing peptide and the corresponding arginine-containing peptide, were synthesized based on the identified citrullination sites in Apo E (2 peptide sets), MNDA (3 peptide sets), and β-actin (7 peptide sets). Some of these peptides contained >1 citrullinated residue, and all peptides contained a C-terminal biotin group. The reactivity of RA sera with these peptides was assessed by ELISA. First, the reactivity of established RA sera with these peptides was analyzed, and the results showed that a subset of these peptides were recognized in a citrulline-dependent manner by a variable number of RA sera. Additional information is available in Supplementary Figure 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. The most frequently recognized peptides, one each for Apo E, MNDA, and β-actin (the citrullinated versions of Apo E197–207, MNDA121–135, and β-actin190–216) were selected for further studies. Using a larger number of patient sera, the citrulline-dependent reactivity of RA sera with these peptides was confirmed (Figures 5A–C). When cutoff values were determined using sera from healthy individuals, 27%, 16%, and 27% of the RA sera analyzed were shown to be reactive with citrullinated Apo E197–207, MNDA121–135, and β-actin190–216, respectively.

Figure 5.

Recognition of citrullinated apolipoprotein E (Apo E), myeloid nuclear differentiation antigen (MNDA), and β-actin epitopes by rheumatoid arthritis (RA) sera. A–C, Presence of autoantibodies against different citrullinated epitopes of Apo E (A), MNDA (B), and β-actin (C) in RA sera and sera from healthy individuals, as determined by enzyme-linked immunosorbent assay (ELISA) using synthetic peptides as antigen. Both citrullinated (X) and arginine-containing (R) peptides were used. Only data for the most frequently recognized peptides are shown. D–F, Reactivity toward citrullinated Apo E (D), citrullinated MNDA (E), and citrullinated β-actin (F) in additional sera from patients with RA, patients with primary Sjögren's syndrome (pSS), patients with multiple sclerosis (MS), patients with systemic lupus erythematosus (SLE), and patients with type 1 diabetes mellitus (T1D) (32 sera for each group), as determined by ELISA. Symbols represent individual data points; broken lines indicate the cutoff values. ∗ = P < 0.05; ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. OD450 = optical density at 450 nm; NS = not significant.

To shed more light on the specificity of the autoreactivity with these peptides, sera from patients with other autoimmune diseases (32 patients with multiple sclerosis, 32 patients with primary SS, 32 patients with SLE, and 32 patients with type 1 diabetes mellitus) were analyzed (Figures 5D–F). The results showed that 4% of the control sera (sera from patients with other autoimmune diseases; n = 126) showed reactivity toward citrullinated Apo E197–207 (Figure 5D), 3% showed reactivity toward citrullinated MNDA121–135 (Figure 5E), and 2% showed reactivity toward citrullinated β-actin190–216 (Figure 5F), leading to specificities of 96%, 97%, and 98%, respectively.

DISCUSSION

In the present study, a proteomic analysis of albumin-depleted SF samples obtained from the inflamed joints of 2 patients with RA led to the identification of 192 proteins. Because such SF samples are known to contain insoluble material, the soluble fraction of the sample from one patient and the insoluble fraction of the sample from the other patient were analyzed to obtain wide coverage of the synovial proteome. Of these 192 proteins, 107 (56%) were detected in material from both patients, in spite of the difference in solubility.

The molecular weight of the majority of the polypeptides identified, as detected with SDS-PAGE, was often lower than the predicted value based on the calculated mass of the corresponding full-length polypeptide. Moreover, several polypeptides with distinct molecular weights were identified in multiple gel slices. An example is the fibrinogen α-chain, which has a calculated mass of 69.8 kd but was detected in material from the gel corresponding to molecular weights of ∼70,000 to <14,000. This heterogeneity is most likely caused by proteolytic fragmentation. Indeed, it was previously demonstrated that the activity of several proteases is increased in RA synovium (17, 18). In a previous study, in which the proteome of 6 RA SF samples was analyzed (19), 24 distinct proteins were identified, 15 of which were also detected in the present study. These data indicate that as a result of the heterogeneity of the composition of SF among distinct patients, the currently available data do not yet represent a comprehensive catalog of proteins that can be found in RA SF.

The 192 proteins identified in the present study can be subdivided into 14 groups according to their annotated functions (immune response, glycolysis, signal transduction, lipid metabolism, blood coagulation, cell mobility/structure/integrity, cell death, protein folding, protein synthesis, vesicle formation/protein transport, molecular transport, protease inhibition, cell adhesion, and other functions). Immune response–related processes, e.g., T cell activation (20), the formation of immune complexes (15), and complement activation (21, 22), have been well documented in RA SF. Apolipoproteins, which are involved in lipid binding and transport, have also been previously detected in RA SF (23, 24). It has been shown that the concentration of certain apolipoproteins (Apo A1 and Apo B) is increased in RA SF, which suggests a role for these apolipoproteins in inflamed synovium (23).

In this study, we also identified 14 proteins that are involved in the coagulation cascade. One of these is fibrinogen, a protein previously found in RA synovium (9, 25, 26). Fibrinogen can be converted into fibrin and accumulates in fibrin deposits, which is a pathologic hallmark of RA. This association with the coagulation cascade is important, because it has been suggested that proteins of the coagulation cascade have proinflammatory properties and in this way contribute to the pathogenesis of RA (27). Coagulation may also be the cause of the relatively high overlap of proteins identified in the soluble and insoluble SF fractions; proteins may be trapped in fibrin deposits. Furthermore, many of the proteins identified in RA SF represent proteins that are known to reside in serum or plasma as well. This is in agreement with previous studies, in which a large overlap among the proteomes of SF and plasma from patients with RA was observed (19).

Fifty-three of the identified polypeptides appeared to be citrullinated on at least 1 position; 105 distinct citrullinated peptides were found. Also, it is noteworthy that this set will not represent a comprehensive collection of citrullinated polypeptides in RA SF, not only because the MS data only partially covered the polypeptide sequences (see Supplementary Table 2), but also because of the heterogeneity of citrullination patterns observed for different RA patients by Western blotting (Figure 1). The citrullination of several proteins, fibrinogen (25), fibronectin (16, 28), vimentin (15, 29), albumin, and β-actin (30), in the inflamed joints of patients with RA confirms previous observations, although citrullinated fibrinogen α-chain and β-actin have not yet been detected in RA SF but only in synovial tissue. Citrullinated histones have been detected only in HL-60 cells (31, 32).

In conclusion, our data revealed 53 citrullinated proteins, 6 of which (Apo E, β-actin, CYP-A, fibrinogen α-chain, fibronectin, and MNDA) were detected in both of the RA SF samples analyzed. Three of the newly identified citrullinated proteins (Apo E, MNDA, and β-actin) were found not only in both patients who were analyzed in detail but also frequently in the SF of other patients with RA. In view of the heterogeneity of the citrullination (as described above), it is likely that citrullinated proteins that were detected in only 1 of the 2 analyzed samples are also more frequently present in SF from other patients. A few other citrullinated polypeptides previously reported to be present in the inflamed joints of patients with RA, such as α-enolase (33), fibrinogen γ-chain (30), and collagen (type I and II) (34, 35), were not detected in the present study. Alpha-enolase and fibrinogen γ-chain, however, were detected in both RA SF samples analyzed (see Supplementary Table 1), and it should be taken into account that the MS data cover only part of these proteins. Therefore, the possibility exists that citrullinated residues may have escaped detection, for example as a result of relatively low citrullination efficiency or as a result of relatively poor ionization of the corresponding peptides.

An indication that Apo E might play a role in RA came from a recent study in which Apo E deficiency was shown to lead to exacerbated collagen-induced arthritis (CIA) in mice (36). It is tempting to speculate that the inflammation-associated citrullination of Apo E results in similar effects. A role for Apo E in the pathophysiology of arthritis is further supported by the observations made by Asquith and coworkers, who reported that Apo E–deficient mice are resistant to the development of CIA (37). However, the exacerbation of CIA observed in one study and the resistance to CIA observed in another study are clearly contradictory. Although the reason for these discrepancies is not yet clear, they may be related to the use of different CIA models; both the mouse strains used and the source of the collagen that was used to induce arthritis were different.

Previously, Matsuo and colleagues identified citrullinated mutant β-actin in RA synovial tissue. Mutant β-actin differs from wild-type β-actin at 2 amino acid positions (β-actin139V→M and β-actin295A→D) (30). Recently, Darrah et al identified β-actin as a target of peptidylarginine deiminase 2 in ionomycin-activated neutrophils (38). Those investigators mapped 6 residues (amino acid positions 176, 183, 196, 206, 312, and 372) that can be citrullinated, 5 of which (positions 183, 196, 206, 312, and 372) were shown to be citrullinated in RA SF in our study.

MNDA is expressed in the nuclei of granulocytes and monocytes. MNDA is a member of the p200 or HIN-200 protein family, which can be induced by interferon. Interferons play a central role in both the innate and adaptive immune responses (39). Other p200 protein family members have been implicated in SLE and systemic sclerosis (40). Finally, it has been reported that MNDA binds to nucleophosmin/B23, which also has been identified as a citrullinated protein (31).

A fourth novel citrullinated protein detected in RA SF is CYP-A, of which only 1 citrullinated peptide was identified in both patients analyzed. This is not the result of low sequence coverage, because coverage was 36% for SFS9 and 43% for SFP38 (see Supplementary Table 3). CYP-A is part of the enzyme family of peptidylprolyl-cis-trans-isomerases, which are involved in protein folding. It has been described previously by Yang and coworkers that CYP-A can up-regulate metalloproteinase 9, which contributes to cartilage and bone destruction in patients with RA (41). In agreement with our data, CYP-A has already been described to be present in RA SF. Moreover, it was suggested previously that CYP-A acts as a cytokine in inflammation (42).

This is the first study (the second in the case of β-actin) in which citrullinated Apo E, MNDA, and β-actin are shown to be targets of the immune system in RA. For each of these proteins, we observed immunodominant citrullinated epitopes, and peptides containing these epitopes, Apo E197−207, MNDA121−135, and β-actin190−216, were recognized by sera from 27%, 16%, and 27%, respectively, of patients with established RA. A linear discriminant analysis demonstrated that a combination of the immunodominant peptides of these 3 proteins appeared to be a better predictor of RA than the individual peptides alone (data not shown), but it should be stressed that this study was not aimed at raising the sensitivity of ACPA detection. Instead, these molecules may be valuable ACPA targets for the development of (multiplex) assays to study ACPA fine specificities (43–45). It remains to be established whether ACPA subspecificities targeting 1 or more of these epitopes are associated with clinical phenotypes.

It is also important to note that the reactivity observed with the synthetic citrullinated peptides may not fully reflect reactivity with the corresponding citrullinated antigens. Conformational aspects and inaccessibility in the complete proteins may reduce the importance of the epitopes identified. In addition, the selection of a single epitope, even when it is the most frequently targeted epitope, may allow the detection of only a subset of the antibodies directed to the corresponding citrullinated antigen. As a consequence, the frequencies of autoantibodies to Apo E, MNDA, and β-actin may be higher than those described above, and this may also explain the higher frequency of anticitrullinated β-actin antibodies in RA reported by Darrah and coworkers (38). Recognition of the antigens in the inflamed joints might be further affected by fragmentation of the antigens (as discussed above), which may also lead to the generation of neoepitopes. The presence of citrullinated proteins and ACPAs in the SF of ACPA-positive patients will lead to the formation of immune complexes. Previously, citrullinated vimentin was identified as a “prominent” citrullinated antigen in immune complexes from the SF of ACPA-positive patients with RA (15). These immune complexes may play an important role in pathophysiology, because they can activate immune cells to produce cytokines, which are the driving force in the chronicity of RA (46). The extent to which immune complexes containing citrullinated Apo E, MNDA, and β-actin are involved in such processes remains to be determined.

In conclusion, a total of 192 different proteins were identified in RA SF. Fifty-three of the identified polypeptides appeared to be citrullinated on at least 1 position, and 105 distinct citrullinated peptides were found. Our data revealed 3 novel citrullinated autoantigens (Apo E, MNDA, and β-actin) in RA SF that are targeted by a subset of ACPAs. The identified epitopes of native citrullinated proteins occurring in the inflamed joints of patients with RA will be helpful for the development of assays for ACPA fine-specificity profiling. Such profiles may allow the subclassification of ACPA-positive patients with RA with respect to disease progression and/or responsiveness to specific treatments, although it remains to be established whether ACPA fine specificities correlate with clinical phenotypes. The increased insight into the nature and extent of protein citrullination in the inflamed joints of patients with RA may also provide clues to the molecular mechanisms underlying joint damage. The biologic activity of many proteins will be affected by their citrullination, and it will be interesting to investigate to what extent this contributes to pannus formation and cartilage and bone damage.

AUTHOR CONTRIBUTIONS

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. Pruijn 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. Van Beers, Božič, Pruijn.

Acquisition of data. Van Beers, Schwarte, Stammen-Vogelzangs, Oosterink.

Analysis and interpretation of data. Van Beers, Pruijn.

Acknowledgements

We would like to thank Dr. Sander van Dooren and Annemarie van der Heijden for technical assistance and Dr. Reinout Raijmakers for his help with interpretation of the MS data.

Ancillary