SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Objective

Antibodies directed toward citrullinated proteins (e.g., anti–cyclic citrullinated peptide antibodies) are highly specific for rheumatoid arthritis (RA) and are produced locally at the site of inflammation. Although the presence of citrullinated proteins in rheumatoid synovium has been described in the literature, it is uncertain whether their presence is specific for RA. The present study was undertaken to investigate this.

Methods

The local production of the anti–citrullinated protein antibodies was investigated by comparing the concentration of the antibodies (corrected for the total amount of IgG present) in paired samples of serum and synovial fluid from RA patients. The presence of citrullinated proteins in the synovial tissue was investigated by immunohistochemical analysis of synovial tissue from RA patients and from patients with other arthropathies, using a variety of specific antibodies to citrullinated proteins.

Results

In RA patients, anti–citrullinated protein antibodies constituted a 1.4-fold higher proportion of IgG in synovial fluid compared with serum, which is indicative of a local production of the antibodies. Immunohistochemical staining of citrullinated proteins was observed in the lining layer, the sublining layer, and in extravascular fibrin deposits in inflamed synovial tissue from RA as well as non-RA patients.

Conclusion

The presence of citrullinated proteins in the inflamed synovium is not specific for RA, but rather, it may be an inflammation-associated phenomenon. The high specificity of the anti–citrullinated protein antibodies is, therefore, most likely the result of an abnormal humoral response to these proteins.

The presence of antibodies directed toward citrullinated proteins in the serum of patients with rheumatoid arthritis (RA) has been described in great detail (for review, see ref.1). These antibodies (e.g., antiperinuclear factor antibodies [2], so-called antikeratin antibodies [3], antifilaggrin antibodies [4, 5], anti-Sa antibodies [6, 7], and anti–cyclic citrullinated peptide [anti-CCP] antibodies [8, 9]) can be detected in up to 80% of RA patients with 98% specificity and are directed against proteins containing the amino acid citrulline. Citrulline is a nonstandard amino acid; it is not incorporated into proteins during translation. It can be generated, however, by posttranslational modification of arginine residues by peptidylarginine deiminase (PAD; EC 3.5.3.15) enzymes (for review, see ref.10).

The antibodies appear to be produced locally at the site of inflammation (i.e., the synovium), since they constitute a higher proportion (7.5-fold) of IgG in synovial tissue than in paired serum samples (11). They are thought to be produced by local plasma cells, since significant amounts of the antibodies have been detected in culture supernatants of synovial tissue fragments obtained from antifilaggrin antibody–seropositive RA patients (11). Furthermore, anti-CCP–producing B cells have been detected in the synovial fluid (SF) of RA patients. B cells isolated from the SF or bone marrow of anti-CCP–positive RA patients, but not anti-CCP-negative patients, were found to actively produce IgM anti-CCP antibodies without stimulation, whereas B cells from the peripheral blood required stimulation in order to produce such antibodies (12). The local presence of anti-CCP–secreting cells in inflamed joints provides evidence of an antigen-driven maturation of CCP-specific B cells at the site of inflammation in RA and, thus, suggests the presence of citrullinated synovial proteins in the synovium. It has previously been suggested that the presence of citrullinated proteins in synovial tissue might be specific for RA (13). Alternatively, it is conceivable that the humoral response to citrullinated proteins, rather than the presence of these proteins, is specific for RA (14).

Therefore, the aim of this study was to investigate whether citrullinated proteins are detectable in the synovium of RA patients and disease controls with the use of various antibodies for immunohistochemical analysis. We used recombinant antibodies directed toward citrullinated proteins selected from patient-derived phage display libraries (RA3 and A2-2) (15), as well as IgG antibodies purified with CCP-1 peptide (8) obtained from the serum of RA patients. For comparison with previous studies (13, 16), antibodies to chemically modified citrulline (anti-MC antibodies) (17–19) and commercially available anti–L-citrulline antibodies (20, 21) were used.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients.

Synovial biopsy samples were obtained from an actively inflamed joint of 23 patients with RA (22) and 31 control patients. Disease duration in the RA patients was <1 year, as measured from the first clinical signs of arthritis, irrespective of which joint was initially affected. Clinical inflammation was defined as both joint swelling and pain at the time of physical examination. The control group consisted of 8 patients with inflammatory osteoarthritis (OA) (23), 5 patients with reactive arthritis (ReA), 14 patients with other inflammatory joint diseases (e.g., undifferentiated arthritis, gout, and calcium pyrophosphate dihydrate crystal deposition disease), and 4 patients with noninflammatory joint diseases (trauma or joint disease of unknown origin).

Laboratory assessments of all study subjects included measurement of serum levels of C-reactive protein (CRP) and rheumatoid factor (RF). Radiographs of the hands and feet were obtained for diagnostic purposes. Synovial fluid was obtained at the time of biopsy.

All patients gave their informed consent. The study protocol was approved by the Committee of Medical Ethics of the Academic Medical Center, University of Amsterdam.

Antibodies.

Anti-MC antibodies (rabbit antibodies directed against chemically modified citrulline) were a kind gift from Dr. Tatsuo Senshu (Yokohama City University, Yokohama, Japan) and have been described previously (17, 18). Rabbit antibodies directed against L-citrulline were obtained from Biogenesis (Poole, UK).

Recombinant patient-derived antibodies (RA3 and A2-2) have been described elsewhere (15). Briefly, a phage library of random combinations of heavy-chain and light-chain variable fragments was constructed from messenger RNA (mRNA) that had been isolated from bone marrow samples from anti-CCP–seropositive RA patients. Phages bearing recombinant antibodies specific for CCP peptides were selected by several rounds of panning. Anti-CCP–specific antibody clones were recloned into a pUC119 plasmid vector containing a VSV-G tag (derived from vesicular stomatitis virus G protein), a His6 tag, and a mouse Cκ-domain tag to facilitate detection and enhance stability of the recombinant antibodies. Bacteria (Escherichia coli) containing the antibody constructs were grown in 500-ml cultures, and antibody expression was induced by the addition of IPTG. Antibodies were purified from periplasmic fractions by nickel–nitrilotriacetic acid (Qiagen, Crawley, UK) chromatography, followed by dialysis against phosphate buffered saline (PBS).

In addition, anti-CCP antibodies were purified from a pool of strong anti-CCP–positive RA sera. Serum was diluted 10 times in PBS500 (PBS containing 500 mM NaCl and 0.05% Nonidet P40) and coupled to a protein G–Sepharose column (Amersham Pharmacia Biotech, Little Chalfont, UK). After extensive washing with PBS500, bound IgG molecules were eluted with elution buffer (100 mM glycine HCl [pH 2.5], 500 mM NaCl, 0.05% Nonidet P40). The collected fractions were neutralized with 1M Tris (pH ∼11) and dialyzed against PBS500.

Fractions containing IgG, as determined by enzyme-linked immunosorbent assay (ELISA), were further purified on a CCP-1 peptide column. Biotinylated CCP-1 peptide (cfc1-cyc2 [8]) was kindly synthesized by Dr. Jan-Wouter Drijfhout (Leiden University Medical Center, Leiden, The Netherlands) and coupled in PBS to a HiTrap streptavidin–Sepharose HP (Amersham Pharmacia Biotech) column. Purified IgG fractions were applied to the column. After extensive washing with PBS500, bound antibodies were eluted by pH shock with elution buffer, immediately neutralized with 1M Tris (pH ∼11), and dialyzed against PBS. Fractions containing anti-CCP antibodies, as determined by ELISA, were pooled and concentrated by centrifugation over a Centricon column (10-kd cutoff; Millipore, Watford, UK).

Because human antibodies cannot be used directly for immunohistochemistry on human tissue sections (a secondary anti-human IgG antibody will also detect IgG that is present in the tissue), the antibodies were directly conjugated to horseradish peroxidase (HRP). Labeling of CCP-purified antibodies with HRP (a kind gift from Dr. Will Roeffen, St. Radboud University Medical Center, Nijmegen, The Netherlands) was performed using the periodate method, with a molar input HRP:IgG ratio of 4 (24). After dialysis against PBS, the labeled antibodies were supplemented with thimerosal (0.01%), glycerol (50%), and fetal calf serum (1%), and then stored at −20°C.

Measurement of antibodies to citrullinated proteins.

The presence of antibodies to citrullinated proteins in serum and synovial fluid was investigated by the Rapscan RA kit (CCP-1 test; Euro-Diagnostica, Arnhem, The Netherlands) according to the manufacturer's instructions. To determine IgM or IgA anti-CCP titers, isoclass-specific secondary antibodies (Dako, High Wycombe, UK) were used instead of the supplied (IgG-specific) secondary antibody. Total IgG, IgM, and IgA concentrations in serum and SF were determined by nephelometry on a Cobas Fara-2 centrifugal analyzer (Roche, Almere, The Netherlands).

Synovial biopsy.

Small-bore arthroscopy (2.7-mm arthroscope; Storz, Tuttlingen, Germany) was performed under local anesthesia, as previously described in detail (25). Synovial biopsy samples were obtained from the entire joint using a 2-mm grasping forceps (Storz). At least 6 tissue samples per patient were collected and snap-frozen en bloc in Tissue-Tek OCT (Miles, Elkhart, IN). Frozen blocks were stored in liquid nitrogen until they were sectioned for staining. Five-micrometer sections were cut in a cryostat and mounted on glass slides (Star Frost adhesive slides, Knittelgläser, Braunschweig, Germany). Slides were stored at −70°C until they were used.

Immunohistochemistry.

Recombinant patient-derived antibodies (RA3 and A2-2).

Sections were fixed in acetone, and endogenous peroxidase activity was inhibited using 0.1% sodium azide and 0.3% hydrogen peroxide in PBS. Primary antibodies were incubated for 60 minutes, followed by incubation with HRP-conjugated goat anti-mouse Ig (Dako) for 30 minutes. Enhancement of the signal was performed with biotinylated tyramide (Perkin Elmer, Warrington, UK) for 15 minutes and HRP-conjugated streptavidin (Dako). HRP activity was detected using hydrogen peroxide as substrate and aminoethylcarbazole as dye. Slides were counterstained with Mayer's hematoxylin and mounted in Kaiser's glycerol gelatin (Merck, Darmstadt, Germany).

CCP-purified HRP-conjugated human antibodies.

Immunohistochemistry was performed as described for the RA3 and A2-2 antibodies, except that incubation with HRP-conjugated goat anti-mouse secondary antibody was omitted.

Anti-MC antibodies.

Prior to incubation with the primary antibody, sections were treated for 3 hours at 37°C in modification solution consisting of 2 parts solution A (0.025% [weight/volume] FeCl3, 4.6M H2SO4, and 3.0M H3PO4), 1 part solution B (1% diacetyl monoxime, 0.5% antipyrine, and 1M acetic acid), and 1 part H2O. Control sections were incubated in a mixture of 2 parts solution A and 2 parts H2O. After extensive washing with PBS, slides were subsequently incubated with 3% H2O2 in methanol (15 minutes) and then with 5% normal swine serum (Dako X0901) with 1% bovine serum albumin (BSA) in PBS (30 minutes). Samples were incubated overnight at room temperature with 0.125 μg/ml of anti-MC antibodies in PBS plus 1% BSA. After washing, the sections were incubated for 30 minutes at room temperature with biotinylated swine anti-rabbit secondary antibody (Dako E0431) diluted 1:200 in PBS plus 1% BSA. Finally, the slides were incubated for 30 minutes with Vectastain ABC reagent (Vector, Burlingame, CA) and then developed using hydrogen peroxide as substrate and diaminobenzidine (Sigma, St. Louis, MO) as dye.

Rabbit anti–L-citrulline antibodies.

Immunohistochemistry for rabbit anti–L-citrulline antibodies (Biogenesis antibodies) was performed as described previously (13, 26).

Microscopic analysis.

After immunohistochemical staining, coded sections were analyzed in a random order by one observer (TJMS). Qualitative scores (positive/negative) were determined for each section.

Statistical analysis.

Wilcoxon's signed rank test was used to compare the relative concentrations of anti-CCP antibodies in paired samples of serum and SF. Spearman's rank correlation test was used to calculate the extent of association between serum and SF concentrations of anti-CCP antibodies. Fisher's exact test was used to investigate differences in the frequency of staining between RA patients and controls. The independent-samples t-test was used to investigate correlations between the anti-CCP titers or CRP levels and the frequency of staining.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Clinical features.

The study group with active RA whose samples were used for immunohistochemical analysis consisted of 17 women and 6 men. The mean (±SD) age of this group was 63 ± 12 years (range 41–80 years), and the mean (±SD) duration of disease was 5.2 ± 2.7 months (range 0.5–12 months) at the time of synovial biopsy. Their mean serum level of CRP was 49 ± 33 mg/liter (range 11–118). Erosions were present in 17 of the 23 patients. Twenty patients were seropositive for IgM-RF. Fifteen of 21 RA patients (71%) were seropositive for anti-CCP; anti-CCP status was not available for 2 patients. Most of the RA patients were being treated with nonsteroidal antiinflammatory drugs; 2 of them were also receiving disease-modifying antirheumatic drugs at the time of arthroscopy. None of the RA patients were being treated with corticosteroids.

The control patient group consisted of 12 women and 19 men. The mean age of this group was 51 ± 19 years (range 22–84 years), and the mean duration of disease was 8.9 ± 13.7 months (range 0–60 months). The mean serum CRP level was 27 ± 32 mg/liter (range 3–125). No CRP data were available for 4 of these patients. None of the control patients had erosions. One control patient was seropositive for IgM-RF. RF status was not available for 3 of the control patients. None of the disease controls who were tested were positive for anti-CCP; anti-CCP status was not available for 2 patients.

Anti-CCP antibodies constitute a higher proportion of Ig in synovial fluid than in serum.

Concentrations of total IgG, IgM, and IgA as well as Ig class–specific anti–CCP-1 antibodies were measured in paired serum and SF samples from 21 RA patients. Fourteen of these patients were positive for IgG anti–CCP-1 antibodies, both in serum and SF. Relative concentrations of the antibodies (corrected for the total amount of IgG present) were, on average, 1.4 times higher in SF than in serum (P < 0.05 by Wilcoxon's signed rank test) (Table 1 and Figure 1). There was a significant correlation between the concentration of anti-CCP antibodies in serum and SF (Spearman's rank correlation rs = 0.90, P < 0.0001). The relative concentration of IgM and IgA anti-CCP antibodies was higher in SF than in serum, although for IgA, the difference was not statistically significant because of the small sample size (n = 7). Our results thus support the notion that anti-CCP antibodies are produced locally in the synovium of RA patients. Consequently, citrullinated proteins should most likely be present in inflamed synovial tissue.

Table 1. Comparison of the relative concentrations of Ig class–specific anti-CCP antibodies in paired samples of serum and synovial fluid obtained from patients with rheumatoid arthritis*
Ig classSerum, mean ± SDSynovial fluid, mean ± SDRatio, synovial fluid to serumNo. of paired samples
  • *

    Concentrations are expressed as class-specific anti–cyclic citrullinated peptide 1 (anti–CCP-1) units/total mg of class-specific Ig.

  • P < 0.05 by Wilcoxon's signed rank test.

IgG109.1 ± 73.8159.1 ± 125.21.414
IgM372.6 ± 169.3605.1 ± 702.31.618
IgA406.6 ± 610.7811.1 ± 1130.92.87
thumbnail image

Figure 1. Comparison of relative IgG anti–cyclic citrullinated protein (anti-CCP) antibody concentrations (expressed in anti–CCP-1 units/mg of total IgG) in paired samples of serum and synovial fluid from patients with rheumatoid arthritis (n = 14). Data points above the diagonal line represent patients in whom the anti-CCP concentration in synovial fluid was higher than that in serum. Anti-CCP antibodies constituted, on average, a 1.4-fold higher proportion of IgG in synovial fluid than in serum.

Download figure to PowerPoint

Presence of citrullinated proteins in the synovium of RA and control patients.

To investigate the presence of citrullinated proteins in the synovial tissue, immunohistochemistry with various anti–citrullinated protein antibodies was performed. After staining, coded sections were analyzed in random order. Qualitative scores were determined independently for 4 different morphologic areas of the tissue sections: the intimal lining layer, synovial sublining, fibrin deposits, and endothelial tissue. Staining of citrullinated proteins was observed in samples from the RA patients as well as in those from disease controls (Table 2).

Table 2. Immunohistochemical detection of citrullinated proteins in synovial tissue obtained from RA patients and control patients*
Antibody, study groupNo. of patientsLining layerSublining layerFibrin depositsEndothelial tissue
  • *

    The control group consisted of 8 patients with inflammatory osteoarthritis, 5 with reactive arthritis, 14 with other inflammatory joint diseases, and 4 with noninflammatory joint diseases. Values are the number (%) of patients. RA = rheumatoid arthritis; CCP-1 = cyclic citrullinated peptide 1.

  • P < 0.05 versus control patients, by Fisher's exact test.

RA3     
 RA patients158 (53)6 (40)13 (87)7 (47)
 Control patients155 (33)5 (33)9 (60)3 (20)
A2-2     
 RA patients2312 (52)21 (91)15 (65)1 (4)
 Control patients314 (13)22 (71)5 (16)3 (10)
CCP-1     
 RA patients236 (26)21 (91)6 (26)4 (17)
 Control patients3114 (45)23 (74)6 (19)2 (6)

Staining with RA3 antibodies (Table 2 and Figure 2) was seen in both RA and non-RA synovial tissue and was localized in the intimal lining layer, endothelium, and fibrin deposits. Extracellular staining corresponding to fibrin deposits was observed in the majority of the sections examined. Staining was also observed in the cytoplasm of lining cells and in the cytoplasm of endothelial cells. Control staining with irrelevant recombinant antibody SB12 (directed toward the Pfs48/45 protein of the malaria parasite Plasmodium falciparum [27]), which contained the same tags as the RA3 antibody, showed negative results.

thumbnail image

Figure 2. Immunohistochemical staining of citrullinated proteins in synovial tissue using citrulline-specific RA3 antibody. A, Rheumatoid arthritis (RA) synovial tissue stained with RA3 antibody. B, Control staining of RA synovium with irrelevant recombinant antibody SB12, which contains the same tags as the RA3 antibody. C, Non-RA synovial tissue stained with RA3 antibody. D, Control staining of non-RA synovium with irrelevant recombinant antibody SB12.

Download figure to PowerPoint

A2-2 antibody staining (Table 2 and Figure 3) was mainly observed in the cytoplasm of the sublining cells, both in RA and in non-RA synovial tissue. The intimal lining layer and the fibrin deposits stained with A2-2 antibodies in many of the RA samples and in only a few of the non-RA samples (P < 0.05 by Fisher's exact test). Staining of endothelial cells was rare. Control staining with irrelevant recombinant antibody SB12 (27), which contained the same tags as the A2-2 antibody, showed negative results.

thumbnail image

Figure 3. Immunohistochemical staining of citrullinated proteins in synovial tissue using citrulline-specific A2-2 antibody. A, Rheumatoid arthritis (RA) synovial tissue stained with A2-2 antibody. B, Control staining of RA synovium with irrelevant recombinant antibody SB12, which contains the same tags as the A2-2 antibody. C, Non-RA synovial tissue stained with A2-2 antibody. D, Control staining of non-RA synovium with irrelevant recombinant antibody SB12.

Download figure to PowerPoint

CCP-purified HRP-conjugated human antibodies (Table 2 and Figure 4) showed a scattered staining pattern, with intense staining in both RA and non-RA tissues, particularly in the synovial sublining and, to a lesser extent, in the intimal lining layer. In both groups, endothelium was rarely stained. Positive staining on granulocytes could also be detected by these antibodies. Fibrin deposits were stained in only a few of the samples in both groups. Control stainings with IgG antibodies derived from RA sera and purified with the noncitrullinated peptide cf0-cyc2 (8) showed negative results.

thumbnail image

Figure 4. Immunohistochemical staining of citrullinated proteins in synovial tissue using cyclic citrullinated peptide 1 (CCP-1)–purified antibodies. A, Rheumatoid arthritis (RA) synovial tissue stained with CCP-1–purified antibodies (Abs). B, Control staining of RA synovium with IgG antibodies derived from RA sera and purified with the noncitrullinated peptide cf0-cyc2. C, Non-RA synovial tissue stained with CCP-1–purified antibodies. D, Control staining of non-RA synovium with IgG antibodies derived from RA sera and purified with the noncitrullinated peptide cf0-cyc2.

Download figure to PowerPoint

A subanalysis was performed to examine correlations between the levels of anti-CCP antibodies or CRP and staining with any of the 3 antibodies in any of the 4 subcompartments. In RA patients, there was no correlation between the anti-CCP titers and the staining of citrullinated proteins or between the CRP levels and the staining of citrullinated proteins.

In control patients, however, there appeared to be a general trend toward higher CRP levels in those with positive staining. Patients who had positive staining of fibrin deposits with RA3 antibody had significantly higher CRP values (P < 0.05) than did those without positive staining. Likewise, significantly higher CRP values were observed in patients with positive A2-2 antibody staining of the sublining layer as well as with positive CCP-purified antibody staining of the intimal lining layer. In all other cases, the P values were not significant (see Table 2).

The subanalysis examining correlations between the anti-CCP titers and the frequency of staining could not be performed in the control patients because all of them were negative for anti-CCP antibodies.

There appeared to be a trend toward a higher frequency of positive staining in RA patients compared with control patients, although statistical significance was reached only for A2-2 staining of the intimal lining layer and the fibrin deposits. In all other cases, the differences were not statistically significant. The presence of citrullinated proteins in the synovium was clearly not specific for RA.

Anti–modified citrulline antibodies.

Staining with anti-MC antibodies (Figure 5) was observed in the majority of RA tissues. Non-RA tissues also stained with anti-MC antibodies, but not as intensely as the RA tissues. Extravascular fibrin deposits and the sublining layer showed positive staining. In RA patients, the intimal lining layer was also stained. Control stainings, where diacetyl monoxime and antipyrine were left out during chemical modification, were negative.

thumbnail image

Figure 5. Immunohistochemical staining of citrullinated proteins in synovial tissue using antibodies to chemically modified citrulline (anti-MC). A, Rheumatoid arthritis (RA) synovial tissue stained with anti-MC antibody (AMC). B, Control staining of RA synovium (diacetyl monoxime and antipyrine were left out during chemical modification). C, Non-RA synovial tissue stained with anti-MC antibody. D, Control staining of non-RA synovium (diacetyl monoxime and antipyrine were left out during chemical modification).

Download figure to PowerPoint

Anti–L-citrulline antibodies.

The majority of RA tissues and a few non-RA tissues showed positive staining with the commercially available rabbit anti–L-citrulline antibody (Figure 6), a finding that is consistent with our previous observations (26). In general, only a few cells in the synovial sublining that had clear plasma cell morphology were positive. In addition, the irrelevant antibody control (rabbit anti–fluorescein isothiocyanate [anti-FITC]) showed similar staining patterns, indicating nonspecific binding of rabbit immunoglobulins to plasma cells. The intimal lining layer, fibrin deposits, and endothelium were not stained by this antibody.

thumbnail image

Figure 6. Immunohistochemical staining of citrullinated proteins in synovial tissue using anti–L-citrulline antibody (obtained from Biogenesis). A, Rheumatoid arthritis (RA) synovial tissue stained with anti–L-citrulline antibody. B, Control staining of RA synovium with an irrelevant antibody (rabbit anti–fluorescein isothiocyanate). C, Non-RA synovial tissue stained with anti–L-citrulline antibody. D, Control staining of non-RA synovium with an irrelevant antibody.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Autoantibodies directed toward citrullinated proteins are highly specific for RA and are present very early in the disease. The association of high titers of the antibodies with an erosive disease outcome suggests a possible role in the pathophysiology of the disease (1). If such a functional relationship exists, the antibodies are expected to be present at sites of inflammation. We therefore sought to determine whether these antibodies are present in higher concentrations in SF than in serum. By comparing paired samples of serum and SF, we observed that anti-CCP antibodies constituted a higher proportion of IgG in SF than in serum. In addition, there was a clear correlation between anti-CCP levels in serum and SF. Taken together, these findings indicate that the anti-CCP antibodies are produced locally in the inflamed synovial compartment. Another study, however, reported equal titers of anti–citrullinated protein antibodies (antifilaggrin autoantibodies) in serum and SF (11). In that study, the antibody titers were expressed in units per milliliter. Because total IgG concentrations are significantly higher in serum that in SF, antibody concentrations should preferentially be expressed in units per milligram of IgG. Using those units of measure, one can see that in that study as well, anti–citrullinated protein antibodies constituted a higher proportion of IgG in SF than in serum (11). Data from SF samples obtained from our non-RA patients are not included in this study, because previous analyses of SF samples from more than 100 non-RA patients (consisting mainly of OA patients) did not yield any anti-CCP–positive samples (Vossenaar ER, et al: unpublished observations).

Further support for the notion that anti-CCP antibodies are produced locally in the synovial compartment comes from the observation that in RA patients, the antibody levels constitute a 7.5-fold higher proportion of IgG in synovial tissue than in serum (11). Moreover, plasma cells producing anti–citrullinated protein antibodies have been isolated from RA synovial tissue (11). Another study showed that peripheral blood B cells are capable of producing anti-CCP antibodies only upon stimulation with CD40 ligand and interleukin-10 or with anti-CD3–activated T cells (12). In contrast, SF B cells from anti-CCP–seropositive RA patients spontaneously produced IgM anti-CCP antibodies (12). Consistent with these data, we found that IgM anti-CCP antibodies constituted a higher proportion of IgM in RA SF compared with RA serum, suggesting that IgM anti-CCP–secreting B cells are present in inflamed joints. This is indicative of an antigen-driven maturation of CCP-specific B cells at sites of inflammation in RA (12). Therefore, it is likely that citrullinated proteins are present in the inflamed RA synovium.

The data presented here confirm and extend the findings of previous studies showing the presence of citrullinated proteins in RA synovial tissue (13, 16). We observed staining in the cytoplasm of various infiltrating cells as well as in extravascular deposits of fibrin. Western blotting previously confirmed that a variety of citrullinated proteins may be present, one of which was identified as citrullinated fibrin (16). Of importance, our study is the first to show that the presence of citrullinated proteins is not specific for rheumatoid synovial tissue, since these proteins can be detected in synovial tissue from patients with various inflammatory arthritides. This was confirmed using several antibodies that recognize different sets of citrulline-containing epitopes. It should be noted that the antibody response to citrullinated proteins in RA patients is very heterogeneous. Each patient's serum recognizes a different subset of citrullinated peptides (28). This indicates that various citrullinated epitopes exist on (several) synovial proteins.

The recombinant antibodies RA3 and A2-2 were selected from patient-derived phage display libraries for their reactivity with citrullinated peptides (15). In addition, HRP-conjugated human IgG antibodies were purified with CCP-1 peptide from a pool of CCP-seropositive RA sera. All 3 types of antibodies are reactive with most citrullinated peptides (and not with noncitrullinated control peptides) by ELISA and are positive for antiperinuclear factor and antikeratin antibodies by immunofluorescence. The antibodies stain the cornified layer of the epidermis, which contains citrullinated filaggrin and keratins (29). Furthermore, they recognize citrullinated filaggrin (by antifilaggrin antibody test) isolated from human epidermis. The antibodies also recognize in vitro–citrullinated fibrinogen (but not unmodified fibrinogen) on Western blots. It should be noted that RA3 antibody is strongly reactive with both the Aα-chain and the Bβ-chain of fibrinogen, whereas A2-2 is strongly reactive with the Aα-chain, but shows only moderate reactivity with the Bβ-chain (15) (the γ-chain of fibrinogen can not be citrullinated because it does not contain arginine residues). This may explain the observed differences in the frequency of staining of the fibrin deposits with both antibodies (Table 2).

For comparison with previous studies, we also used anti-MC antibodies, which are reactive with chemically modified citrulline (16), and anti–L-citrulline antibodies, which were obtained from Biogenesis (13), on a limited number of synovial tissue sections. Ideally, such antibodies recognize all citrullinated proteins, independently of the amino acids that flank the citrulline residues. The antibodies must also be nonreactive with free L-citrulline because this is also present in the synovium as a byproduct of nitric oxide synthesis. The anti-MC antibodies that were used in the study by Masson-Bessière and colleagues (16) as well as those used in a study investigating the expression of citrullinated proteins in animal models of arthritis (14) target chemically modified citrullines. In this method, which was developed by Senshu and coworkers (17–19), the citrulline side-chain is specifically modified by chemical treatment into complex structures that are so bulky that the influence of flanking amino acids on epitope recognition becomes negligible. Noncitrullinated proteins cannot be modified by the chemical treatment and are thus not recognized by these antibodies.

Our use of anti-MC antibodies confirmed the findings obtained with the recombinant RA3 and A2-2 antibodies as well as with the purified human CCP antibody: the presence of citrullinated proteins is not specific for RA synovium, although staining tended to be more intense in synovium from RA patients than in that from the disease control patients. This tendency may be explained by the fact that the RA patients had higher levels of CRP (49 ± 33 mg/liter) than did the control group (27 ± 32 mg/liter). In control patients, there appeared to be a slight trend toward higher CRP levels in patients with positive staining. This finding supports the idea that citrullinated proteins are generated during inflammation; patients with more active inflammation will thus be more likely to produce citrullinated proteins.

Immunohistochemical staining with the anti–L-citrulline antibody from Biogenesis (13, 26) revealed a pattern that was quite different from that obtained with the other methods described above. We observed staining of a few isolated cells with plasma cell morphology, but no staining of the extracellular structures. We have previously suggested that this antibody may not be the antibody of choice for the detection of citrullinated proteins (26). These rabbit antibodies have been raised against poly-L-citrulline conjugated to keyhole limpet hemocyanin and were generated for the detection of nitric oxide synthase activity, which yields L-citrulline as a byproduct (20, 21). This is illustrated by the fact that the staining can be completely abolished by competition with free L-citrulline (13). The staining observed with these antibodies might be partly explained by nonspecific binding of rabbit IgG to (rheumatoid factor–positive) plasma cells, since we observed similar staining patterns using an irrelevant control antibody (rabbit anti-FITC) and demonstrated double staining with plasma cell markers (26).

Two possible explanations can be considered for the high specificity of the anti–citrullinated protein antibodies for RA. One explanation might be that there is an RA-specific expression of citrullinated antigens in the rheumatoid synovium that leads to an immune response. Alternatively, the presence of citrullinated proteins could be a common phenomenon in any inflamed synovial tissue, but RA patients might have an abnormal humoral response to these proteins.

The existence of genetic polymorphisms in the PAD4 gene (30) supports the first possibility (for a discussion of this concept, see refs.31 and32). A certain haplotype of PAD4 that is associated with susceptibility to RA encodes for an mRNA with increased stability. This could result in an increased production of PAD4 and, after activation of PAD, an increased citrullination of proteins. It has been shown that RA patients who were homozygous for the susceptible haplotype were significantly more often positive for anti–citrullinated protein antibodies than were patients who were heterozygous or patients who were homozygous for the nonsusceptible haplotype (30). The actual (enhanced) presence of citrullinated proteins was, unfortunately, not investigated in that study.

Our current observation that citrullination of synovial proteins is not specific for RA is supported by animal studies showing that various synovial proteins, including extravascular fibrin, are citrullinated during inflammation (14). Citrullinated proteins have been detected in a model of chronic inflammation (collagen-induced arthritis) as well as in a model of more acute joint inflammation (streptococcal cell wall–induced arthritis).

In the present study, the expression of citrullinated proteins in the non-RA group was not associated with circulating anti-CCP antibodies. Similarly, animal studies have revealed that arthritic mice do not produce anti–citrullinated protein antibodies despite the presence of citrullinated proteins (14, 33). Thus, the high specificity of the anti–citrullinated protein antibodies for RA appears to be the result of an abnormal humoral response, rather than representing disease-specific expression of citrullinated proteins. Indeed, it has recently been shown that citrullinated peptides can be bound much more efficiently by DR4 molecules than by corresponding noncitrullinated peptides (34). Thus, the genetic background associated with RA (35) might provide the possible basis of a citrulline-specific immune response in rheumatoid synovial tissue.

In conclusion, the highly specific presence of anti–citrullinated protein antibodies in RA cannot be explained by a specific expression of citrullinated proteins at sites of inflammation. It is more likely that RA patients exhibit an abnormal humoral response to citrullinated proteins that may be present in any form of (synovial) inflammation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The authors wish to thank Dr. Tatsuo Senshu (Tokyo, Japan) for kindly providing the antibodies directed toward chemically modified citrulline, Suzanne Nijenhuis, Kalok Cheung, and Ben de Jong (Nijmegen, The Netherlands) for valuable technical assistance, and Drs. Han Zendman and Reinout Raijmakers (Nijmegen, The Netherlands) for useful discussions and critical reading of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES