Induction of macrophage secretion of tumor necrosis factor α through Fcγ receptor IIa engagement by rheumatoid arthritis–specific autoantibodies to citrullinated proteins complexed with fibrinogen


  • Cyril Clavel,

    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, Institut Fédératif de Recherche 30 (IFR30) Toulouse, France
    2. Laboratory of Cell Biology and Cytology, Institut Fédératif de Biologie, CHU de Toulouse, IFR30, Toulouse, France
    Search for more papers by this author
  • Leonor Nogueira,

    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, Institut Fédératif de Recherche 30 (IFR30) Toulouse, France
    2. Laboratory of Cell Biology and Cytology, Institut Fédératif de Biologie, CHU de Toulouse, IFR30, Toulouse, France
    Search for more papers by this author
  • Laetitia Laurent,

    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, IFR30, Toulouse, France
    Search for more papers by this author
  • Cristina Iobagiu,

    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, Institut Fédératif de Recherche 30 (IFR30) Toulouse, France
    2. Laboratory of Cell Biology and Cytology, Institut Fédératif de Biologie, CHU de Toulouse, IFR30, Toulouse, France
    Search for more papers by this author
  • Christian Vincent,

    1. Laboratory of Cell Biology and Cytology, Institut Fédératif de Biologie, CHU de Toulouse, IFR30, Toulouse, France
    Search for more papers by this author
  • Mireille Sebbag,

    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, IFR30, Toulouse, France
    Search for more papers by this author
    • Drs. Sebbag and Serre contributed equally to this work.

  • Guy Serre

    Corresponding author
    1. Unité Mixte de Recherche 5165 CNRS–Université Toulouse III, Institut Fédératif de Recherche 30 (IFR30) Toulouse, France
    2. Laboratory of Cell Biology and Cytology, Institut Fédératif de Biologie, CHU de Toulouse, IFR30, Toulouse, France
    • Unité “Différenciation Épidermique et Auto-immunité Rhumatoïde,” UMR 5165 CNRS–Université Paul Sabatier, Hôpital Purpan, Place du Dr Baylac, TSA 40031, 31059 Toulouse Cedex 9, France
    Search for more papers by this author
    • Drs. Sebbag and Serre contributed equally to this work.



Macrophage-derived tumor necrosis factor α (TNFα) is a dominant mediator of synovitis in rheumatoid arthritis (RA). This study was undertaken to assess whether and how immune complexes (ICs) formed by the interaction of disease-specific autoantibodies to citrullinated proteins (ACPAs) with their main synovial target antigen, citrullinated fibrin, contribute to TNFα production by macrophages.


An in vitro human model was developed in which monocyte-derived macrophages were stimulated with ACPA-containing ICs that were generated by capturing ACPAs from RA sera on immobilized citrullinated fibrinogen. Cellular activation was evaluated by TNFα assay in culture supernatants. Selective blockade of IC interactions with the 3 classes of Fcγ receptors (FcγR) was used to assess the contribution of each receptor to macrophage activation. In addition, 2 citrullinated fibrin–derived peptides bearing major ACPA epitopes were tested for their capacity to inhibit formation of macrophage-activating ACPA-containing ICs.


ACPA-containing ICs induced a dose-dependent TNFα secretion by macrophages from 14 of 20 healthy donors. The macrophage response was systematically higher than that of the paired monocyte precursors. TNFα secretion was not reduced by blockade of FcγRI or FcγRIII, but was strongly repressed when interaction of ICs with FcγRII was prevented. The 2 citrullinated peptides significantly inhibited ACPA reactivity to citrullinated fibrinogen and, when tested together, almost completely abolished formation of macrophage-activating ICs, thereby diminishing the secreted TNFα levels.


Our model demonstrates the inflammatory potential of ACPA-containing ICs via engagement of FcγRIIa at the surface of macrophages, strongly supporting their pathophysiologic involvement. Continuing dissection of these molecular pathways could open the way to new therapeutic approaches in patients with RA.

IgG autoantibodies to citrullinated (deiminated) proteins (ACPAs) are highly specific for rheumatoid arthritis (RA), being present in the serum of 75% of patients with RA (1). In fact, ACPAs seem to play a key role in the pathophysiology of RA, since these autoantibodies are detectable years before arthritis becomes clinically observable (2), have repeatedly been found to be associated with the most severe and erosive forms of the disease (for review, see ref. 1), and are produced by plasma cells of the rheumatoid synovium (3), wherein they concentrate and most likely interact with citrullinated proteins, of which citrullinated fibrin constitutes a prominent target (4).

Recent findings in mice have reinforced the hypothesis of the pathophysiologic importance of ACPAs. First, mice with collagen-induced arthritis have been shown to develop IgG antibodies that are specific for citrullinated peptides and that clearly enhance the severity and incidence of arthritis (5). Second, intraperitoneal injections of serum IgG from RA patients, but not from healthy individuals, induced inflammation and histologic lesions in the ankles of mice deficient in the inhibitory receptor for the Fc fragment of IgG, Fcγ receptor IIb (FcγRIIb) (6). In this mouse model, at least some of the RA-associated IgG antibodies exerting a pathogenic role probably correspond to ACPAs. Finally, observations of prolonged clinical improvement in patients with RA following anti-CD20 antibody therapy demonstrate the likely pathogenic involvement of B cells in the human disease (7).

Because they contribute considerably to inflammation and joint destruction in both the acute and chronic phases of RA (8), synovial macrophages also play a pivotal role in the disease. In particular, macrophages constitute the major articular source of tumor necrosis factor α (TNFα) (9), a key proinflammatory cytokine in RA (10). Although contact with antigen- or cytokine-activated T cells has been proposed to play a role in TNFα production (11–13), the factors involved have not been completely elucidated. Numerous reports have described the arthritogenic potential of FcγR-mediated cell responses, notably through studies of the induction of experimental arthritis in mice deficient in different types of FcγR (14).

In the present study we wanted to verify that, in the rheumatoid synovial tissue, the immune complexes (ICs) formed following interaction of ACPAs with citrullinated fibrin deposits could stimulate secretion of TNFα by macrophages via crosslinkage with stimulatory FcγR. Macrophages were generated by differentiation of highly purified human monocytes, and immobilized ACPA-containing ICs were reconstituted in vitro. Monocytes and macrophages were compared for their expression of the 3 classes of FcγR, FcγRI, FcγRII, and FcγRIII, and for their TNFα-specific response to the surrogate IC heat-aggregated gamma globulins (HAGGs), used in an immobilized form (iHAGG), and to ACPA-containing ICs. In addition, blockade of the interaction of ICs with the various FcγR was used to identify which were involved in TNFα production by macrophages. Finally, we evaluated the capacity of citrullinated fibrin–derived peptides bearing major ACPA epitopes to inhibit formation of ACPA-containing ICs and the subsequent effect on the induced macrophage activation.


Citrullinated human fibrinogen.

Plasminogen-depleted human fibrinogen (95% pure; Calbiochem-VWR, Fontenay-sous-Bois, France) was further purified and then citrullinated, as previously described (15), to obtain citrullinated fibrinogen (designated C-fibrinogen). Noncitrullinated fibrinogen incubated in citrullination buffer alone (designated NC-fibrinogen) was used as the control. Buffer exchange to phosphate buffered saline (PBS; pH 7.4) was performed until complete reassembly of the 6 constitutive chains was achieved; this was monitored by sodium dodecyl sulfate–polyacrylamide gel electrophoresis under nonreducing conditions. Aliquots were lyophilized and kept at −80°C until used.

Citrullinated fibrin–derived peptides.

Two citrullinated human fibrin–derived synthetic peptides bearing major ACPA epitopes (16), α36-50cit38,42 (of amino acid sequence GPXVVEXHQSACKDS, in which X designates a citrullyl residue), β60-74cit60,72,74 (XPAPPPISGGGYXAX), and their totally noncitrullinated counterparts (with arginyl instead of citrullyl residues), α36-50 and β60-74, respectively, were purchased from NeoMPS (Strasbourg, France). All peptides were at least 90% pure. The β60-74cit60,72,74 peptide was synthesized with a C-terminal carboxamide instead of a free carboxylic group.

ACPA+ and ACPA− IgG.

ACPA+ IgG were prepared from a pool of 38 RA serum samples containing high levels of ACPAs, as detected both by indirect immunofluorescence on rat esophagus cryosections (17) and by enzyme-linked immunosorbent assay (ELISA) on C-fibrinogen (18, 19). ACPA− IgG were prepared from a pool of 20 serum samples obtained from patients with non-RA rheumatic diseases; the ACPA− IgG were totally nonreactive in the above-mentioned tests. All of the IgG were purified by affinity chromatography on a protein G column (5 ml HiTrap protein G; GE Healthcare, Saclay, France), as instructed by the manufacturer. The eluted IgG fractions were pooled and dialyzed against Dulbecco's PBS (D-PBS; Gibco, Cergy Pontoise, France) containing 0.5M NaCl, without calcium or magnesium. The IgG concentrations were estimated by measuring the optical density (OD) at 280 nm. Aliquots were stored at −80°C.

Monocyte purification and macrophage differentiation.

All reagents used for monocyte or macrophage preparation and culture were shown to contain <0.5 units/ml of endotoxin, as measured using a Limulus amebocyte lysate assay (QCL-1000; Cambrex Bioscience, Verviers, Belgium). Blood buffy coats from healthy adult donors (Établissement Français du Sang Pyrénées-Méditerranée, Toulouse, France) were diluted 1:2 in a D-PBS buffer containing 0.6% sodium citrate and 0.1% bovine serum albumin (BSA) (low endotoxin grade; Sigma-Aldrich, Saint Quentin Fallavier, France), referred to as PBNaCit, and layered onto Ficoll (1,077 gm/ml Biocoll; Biochrom, Berlin, Germany).

After centrifugation at 350g, the layer containing mononuclear cells was extracted and washed in PBNaCit, followed by loading onto a column of magnetic beads coated with anti-CD14 antibodies (CD14 MicroBeads; Miltenyi Biotec, Paris, France) to purify monocytes, according to the manufacturer's instructions. Monocyte purity, determined by flow cytometry analysis of CD14 expression (see below), was regularly higher than 95%. The monocytes were either directly stimulated or allowed to differentiate into macrophages over 7 days of culture (at 106 cells/ml) in macrophage–serum-free medium (SFM) (Gibco), supplemented with 50 units/ml penicillin, 50 μg/ml streptomycin, 10% fetal calf serum (Biowest, Nuaillé, France), and 100 ng/ml macrophage colony-stimulating factor (M-CSF; PeproTech, Levallois-Perret, France), in perfluoroalkoxy polymer culture inserts (Calbiochem-VWR).

Flow cytometry.

All mouse monoclonal antibodies (mAb) used in flow cytometry were from Beckman Coulter (Roissy, France). These included fluorescein isothiocyanate–labeled anti-FcγRI (clone 22) and anti-CD14 (clone MY4), phycoerythrin (PE)–labeled anti-FcγRII (clone 2E1) and anti-CD206 (clone 3.29B1.10), and PE/Cy5-labeled anti-FcγRIII (clone 3G8). Staining was performed by incubating cells (1–3 × 105 per tube) for 30 minutes at 4°C with 10 μl of each of the antibodies diluted in D-PBS. For each antibody, the mean fluorescence intensity (MFI) and the percentage of positive cells were determined with a Coulter Epics XL4C flow cytometer (Beckman Coulter). Nonspecific staining was controlled with mAb of the same isotype and labeled with the same fluorescent dye.

Monocyte and macrophage stimulation.

Flat-bottom 96-well plates (Nunclon Delta; Nunc-VWR, Roskilde, Denmark) were coated overnight at 4°C with C-fibrinogen (10 μg/ml in D-PBS, 50 μl/well) followed by blocking with D-PBS containing 2% BSA (PBS–BSA) for 1 hour at 4°C, and then washing with D-PBS containing 0.1% (volume/volume) Tween 20 (D-PBS–Tween). ICs were then generated by 2-hour incubation at 4°C with 100 μl of the ACPA+ IgG solution (0.2– 6.4 mg/ml) diluted in PBS-BSA containing 2M NaCl (PBS–BSA–2M). Controls were C-fibrinogen–coated wells incubated with either PBS–BSA–2M or ACPA− IgG solution, or NC-fibrinogen–coated wells incubated with either PBS–BSA–2M or ACPA+ IgG solution.

After washing the wells in D-PBS–Tween and in D-PBS, monocytes or macrophages were added in macrophage-SFM (50,000 cells/well, 2 wells per condition). Cells cultured in medium alone or with lipopolysaccharide (LPS) (0.5 μg/ml, from Escherichia coli strain O55/B5; Sigma-Aldrich) were used as negative and positive controls, respectively. In addition, cultures were performed on wells passively coated with HAGGs (50 μl/well at 20 μg/ml in D-PBS), prepared by heating human IgG (Sigma-Aldrich) for 20 minutes at 63°C and removing insoluble aggregates by centrifugation at 12,000g. After a 24-hour incubation at 37°C with 5% CO2, supernatants were harvested and immediately stored at −30°C until TNFα concentrations were determined by ELISA (OptEIA Human TNF ELISA; BD Biosciences, Pont-de-Claix, France), as recommended by the manufacturer.

Fcγ receptor blockade.

Macrophages were incubated for 45 minutes on ice with F(ab′)2 fragments of mouse mAb to FcγRI (mAb 10.1), FcγRII (mAb 7.30), or FcγRIII (mAb 3G8) (all from Höelzel Diagnostika, Köln, Germany), used at a ratio of 2, 4, or 8 μg of F(ab′)2 fragments per 200,000 cells. After washing in D-PBS, the macrophages were stimulated with immobilized ACPA-containing ICs, as described above.

Inhibition experiments with citrullinated fibrin–derived peptides.

Increasing amounts of the peptides α36-50cit38,42 and β60-74cit60,72,74, used alone or in combination (individual concentrations from 0.006 mg/ml to 1.560 mg/ml), were preincubated for 2 hours at 4°C with the ACPA+ IgG before being tested by ELISA on C-fibrinogen, as previously described (18). ACPA+ IgG were assayed at a concentration (16 μg/ml) that allowed an OD of ∼1.0 to be obtained in the absence of competing peptide. Similarly, in the assay of macrophage stimulation by ICs, ACPA+ IgG (diluted at 1.13 mg/ml in PBS–BSA–2M) were preincubated for 2 hours at 4°C with the peptides α36-50cit38,42 and/or β60-74cit60,72,74 (individual concentrations from 0.6 mg/ml to 0.8 mg/ml) before being added to the C-fibrinogen–coated wells and incubated with macrophages as described above. Controls included preincubations of ACPA+ IgG with C-fibrinogen or NC-fibrinogen, or with the noncitrullinated peptides α36-50 and/or β60-74 at the relevant matched concentrations.

Statistical analysis.

Data analyses were performed using Statistica for Windows (StatSoft, Tulsa, OK). Differences in the percentages of positive cells, in the MFIs, or in the TNFα concentrations between monocytes and macrophages were evaluated by Wilcoxon's tests. Correlations between TNFα levels induced by ACPA-containing ICs and those induced by iHAGG, and correlations between each of these levels and the percentage of FcγR-positive cells or corresponding MFIs were assessed by Spearman's rank correlation coefficients. P values less than or equal to 0.05 were considered significant.


Differential expression of Fcγ receptors and CD206 between monocytes and monocyte-derived macrophages.

Expression of the 3 FcγR and of the mannose receptor CD206 by monocytes and monocyte-derived macrophages was analyzed by flow cytometry (Table 1). As expected (20), nearly all of the monocytes expressed FcγRI and FcγRII (median 94.4% and 97.1% positive cells, respectively), whereas only a small proportion expressed FcγRIII and CD206 (median 14.7% and 3.1% positive cells, respectively).

Table 1. Expression of Fcγ receptors (FcγR) and CD206 on monocytes and macrophages*
 No. of donors% positive cellsMFI
  • *

    Values are the median (range) percent positive cells and median (range) mean fluorescence intensity (MFI), evaluated by flow cytometry in paired samples.

  • P < 0.05 versus monocytes, by Wilcoxon's test.

 FcγRI1794.4 (15.5–98.6)6.6 (1.6–19.2)
 FcγRII1697.1 (82.5–99.2)21.5 (0.7–102)
 FcγRIII1914.7 (0–57.2)12.6 (0–57.5)
 CD20663.1 (0.8–5.1)5.4 (2.9–10.4)
 FcγRI1726.0 (2.9–84.4)3.7 (0–16.2)
 FcγRII1698.9 (78–100)24.7 (5.5–57.6)
 FcγRIII1986.1 (63.6–98.1)19.8 (7–60.4)
 CD206684.2 (60.4–94.2)16.2 (5.3–32.8)

After 7-day culture in the presence of M-CSF, cells were stained with May-Grünwald-Giemsa for microscopic examination. The results showed that the cells had adopted the typical morphologic features of macrophages (results not shown). Moreover, after 7 days in culture, a significant induction of CD206 expression (median 84.2% positive cells; P = 0.028 versus monocytes) and of FcγRIII expression (median 86.1% positive cells; P = 0.00013 versus monocytes) was noted in a very high proportion of the differentiated macrophages (Table 1). This was also accompanied by a small increase in the proportion of FcγRII-positive cells (median 98.9% positive cells; P = 0.023 versus monocytes) and a significant reduction in both the percentage of FcγRI-positive cells (median 26% positive cells; P = 0.0014) and its corresponding MFI (median 6.6 in monocytes versus 3.7 in macrophages; P = 0.0042).

Increased sensitivity of macrophages, compared with monocytes, to activation induced by iHAGG.

To determine whether differentiation into macrophages influenced the sensitivity to ICs, the TNFα secretion induced by iHAGG was assessed in paired monocytes and macrophages from 6 donors. Figure 1 shows that the sensitivity of both monocytes and macrophages to iHAGG varied between donors, with the induced TNFα concentrations ranging from ∼40 pg/ml to ∼150 pg/ml in monocytes and from ∼160 pg/ml to ∼1,300 pg/ml in macrophages. Nevertheless, macrophages systematically secreted higher amounts of TNFα, at a mean ∼6-fold higher level than that secreted by the paired monocytes. Accordingly, the TNFα values obtained with macrophages were found to be significantly higher than those obtained with monocytes (P = 0.028).

Figure 1.

Activation of paired monocytes and macrophages by immobilized heat-aggregated gamma globulins (iHAGG). Monocytes and macrophages obtained from 6 blood donors (pairs a–f) were cultured on iHAGG-coated culture wells. Response values are the difference in the mean secreted concentrations of tumor necrosis factor α (TNFα) after 24 hours of culture between iHAGG-coated and uncoated wells, with the latter always representing a very low percentage (median 0%, range 0–7.4%) of the levels of TNFα induced by iHAGG.

Induction of TNFα secretion from macrophages by ACPA-containing ICs.

The TNFα responses to ACPA-containing ICs and to iHAGG were evaluated in the macrophages from 20 donors. Figure 2A shows a representative response to increasing amounts of ACPA-containing ICs that were reconstituted by incubating increasing doses of ACPA+ IgG on C-fibrinogen–coated wells. The ICs induced a dose-dependent TNFα secretion, with a bell-shaped distribution, reaching maximal levels after incubation with an ACPA+ IgG concentration of 1.6 mg/ml (thus considered the optimal concentration of ACPA+ IgG for that donor). Most of the induced secretion of TNFα could be attributed to a specific effect of ICs, since, in control conditions (ACPA− IgG on C-fibrinogen–coated wells and ACPA+ IgG on NC-fibrinogen–coated wells), the median levels of TNFα never exceeded 42 pg/ml (Figure 2A).

Figure 2.

Activation of macrophages by anti–citrullinated protein autoantibody (ACPA)–containing immune complexes (ICs), and analysis of the subsequent TNFα response. A, The macrophages from 1 blood donor were stimulated with increasing amounts of ACPA-containing ICs, which were reconstituted by incubating citrullinated fibrinogen (C-FBG)–coated culture wells with ACPA+ IgG at increasing concentrations. Under control conditions, macrophages were cultured on C-FBG–coated wells incubated with increasing concentrations of ACPA− IgG (C-FBG + ACPA IgG) or on noncitrullinated fibrinogen (NC-FBG)–coated wells incubated with increasing concentrations of ACPA+ IgG (NC-FBG + ACPA+ IgG). Results are the median and range of duplicate measurements of TNFα secretion at various IgG concentrations. B, In macrophages from the same donor, the maximal concentration of TNFα obtained with the ACPA-containing ICs (C-FBG + ACPA+ IgG at the optimal concentration) is compared with that obtained in culture medium alone, in the presence of lipopolysaccharide (LPS), in the presence of iHAGG, or in the presence of immobilized C-FBG or NC-FBG not incubated with any IgG solution. Results are the median and range of duplicate measurements. See Figure 1 for other definitions.

In comparing the maximal concentrations of TNFα secreted in other control conditions, we found that, as shown in Figure 2B, negligible amounts of TNFα were secreted when immobilized C-fibrinogen or NC-fibrinogen was incubated without IgG. Furthermore, iHAGG induced a higher TNFα secretion than did ICs (mean ∼2.5-fold higher levels). Finally, as expected, LPS induced the highest levels of TNFα (median ∼800 pg/ml). Similar results with regard to the TNFα responses to ACPA-containing ICs were obtained with macrophages from 14 of the 20 tested donors, with the distribution of TNFα secretion following a bell-shaped curve and reaching maximal concentrations of more than 75 pg/ml (Table 2). In the other 6 donors, low to negligible TNFα secretion (lower than 75 pg/ml) was observed, regardless of the dose of IC used (Table 2).

Table 2. Macrophage activation induced by ACPA-containing ICs or by iHAGG*
DonorΔ[TNFα]ACPA-containing ICsΔ[TNFα]iHAGGRatio
  • *

    Values are the difference between the mean tumor necrosis factor α (TNFα) concentrations (in pg/ml) obtained after culture on C-fibrinogen–coated wells incubated with anti–citrullinated protein autoantibody–positive (ACPA+) IgG at the optimal concentration (range 0.4–3.2 mg/ml) compared with culture on C-fibrinogen–coated wells incubated with ACPA− IgG at the same concentration, or the difference in mean TNFα secretion in response to immobilized heat-aggregated gamma globulins (iHAGG) compared with that in uncoated wells. ICs = immune complexes.

  • Calculated as Δ[TNFα]ACPA-containing ICs/Δ[TNFα]iHAGG.


Despite the large interindividual variability, comparison of the TNFα concentrations obtained with ICs and those obtained with iHAGG in the cultures of macrophages from all 20 donors (Table 2) showed that the response to iHAGG was systematically higher (a mean 3-fold higher) than that to ICs. The 2 responses were significantly correlated (r = 0.45, P = 0.05). However, no correlations were observed between either of these responses and the percentage of positive cells or the MFI for any of the 3 FcγR.

Increased sensitivity of macrophages, compared with monocytes, to activation induced by ACPA-containing ICs.

Because the responses to iHAGG and to ACPA-containing ICs were correlated, we predicted that activation by ACPA-containing ICs would be more readily achieved in macrophages than in monocytes. Monocytes from 6 of the 20 donors were stimulated with ACPA-containing ICs, and their levels of TNFα secretion were compared with those of the paired macrophages (as indicated in Table 2). Indeed, with only one exception, the ICs induced higher amounts of TNFα in macrophages than in the paired monocytes (mean ∼4.5-fold higher levels; P = 0.044) (additional detailed data are available from the corresponding author upon request).

Dependency of IC-induced TNFα secretion on IgG–FcγRII interactions.

To evaluate involvement of the FcγR in the TNFα response of macrophages to ACPA-containing ICs, the macrophages were preincubated with increasing amounts of F(ab′)2 antibody fragments to each of the 3 FcγR; these antibody fragments are known to be able to block the interaction of the receptors with IgG. Several doses of ICs were tested. Preincubation with F(ab′)2 fragments of antibodies to FcγRI or FcγRIII did not significantly inhibit the TNFα secretion by macrophages (as observed in macrophages from 3 donors [results not shown]). In contrast, TNFα secretion was inhibited (up to 80%), in a dose-dependent manner, when interaction with FcγRII was prevented using F(ab′)2 fragments of antibodies to this receptor (as observed in macrophages from 2 donors, with 3 doses of ICs for each). In fact, the mean percentage inhibitions obtained with the 2 highest doses of anti-FcγRII F(ab′)2 fragments were statistically significant (Figure 3).

Figure 3.

Effects of Fcγ receptor II (FcγRII) blockade on secretion of tumor necrosis factor α (TNFα) by macrophages activated by anti–citrullinated protein autoantibody (ACPA)–containing immune complexes (ICs). Macrophages were preincubated with F(ab′)2 fragments of anti-FcγRII monoclonal antibody 7.30 at the indicated increasing doses, and then stimulated with ACPA-containing ICs. The percentage inhibition of TNFα secretion was calculated in relation to the TNFα secreted when macrophages were preincubated without F(ab′)2 fragments. Results are the mean and SD percentage inhibition corresponding to the 6 sets of data obtained from 2 blood donors with 3 different doses of ACPA-containing ICs per donor. After verifying that the obtained amounts of inhibition did not significantly vary between donors or between different doses of ACPA-containing ICs (by Kruskal-Wallis analysis of variance [ANOVA]), but that the inhibition was globally influenced by the FcγRII antagonist concentration (by Friedman's ANOVA), Wilcoxon's tests were performed to evaluate the significance of the inhibition obtained at the different antagonist concentrations. ∗ = P < 0.05 versus the lowest concentration.

Similarly, iHAGG-induced TNFα production by macrophages was inhibited (up to 69%) when interaction with FcγRII was prevented. In contrast, no inhibition was observed when interactions with FcγRI or FcγRIII were blocked (results not shown).

Strong inhibition of IC-induced macrophage activation by citrullinated fibrin–derived peptides.

We previously described 2 citrullinated fibrin–derived peptides, α36-50cit38,42 and β60-74cit60,72,74, which bear major epitopes that are recognized by ACPAs (16). To evaluate their capacity to inhibit the reactivity of a highly polyclonal mixture of ACPAs such as that contained in the ACPA+ IgG solution, this solution was preincubated with increasing amounts of the peptides before its reactivity was tested by ELISA on C-fibrinogen. Figure 4A shows that both α36-50cit38,42 and β60-74cit60,72,74 inhibited the reactivity of the ACPA+ IgG to C-fibrinogen, with inhibitions of up to ∼20% and up to ∼60%, respectively. Moreover, the inhibition increased to up to ∼70% when preincubation with both peptides was performed, a reduction in response approaching that obtained with the whole C-fibrinogen molecule (up to ∼80% inhibition). No inhibition was observed after preincubation with the noncitrullinated peptides α36-50 and/or β60-74 or with the NC-fibrinogen (Figure 4A).

Figure 4.

Effect of citrullinated fibrin–derived peptides on the enzyme-linked immunosorbent assay (ELISA) reactivity of anti–citrullinated protein autoantibody–positive (ACPA+) IgG, and on the capacity of ACPA+ IgG to form macrophage-activating immune complexes (ICs). A, Increasing amounts of the peptides α36-50cit38,42 and β60-74cit60,72,74 were preincubated, alone or in combination, with the ACPA+ IgG before being tested by ELISA on C-fibrinogen (C-FBG). Preincubations with the noncitrullinated peptides α36-50 and/or β60-74 with C-FBG or with noncitrullinated fibrinogen (NC-FBG) were performed as controls. The percentage inhibition of the reactivity in the presence of peptide(s), calculated in reference to that obtained without competitor(s), was plotted against the competitor concentrations. Note that inhibition percentages equaled 0 in the 4 control conditions (corresponding symbols therefore merge at 0). B, The ACPA+ IgG were preincubated with the peptides α36-50cit38,42 and β60-74cit60,72,74, alone or in combination (at the increasing concentrations indicated), and then used to generate ICs by incubation on C-FBG–coated culture wells. Preincubations with no competitor or with C-FBG at the indicated concentrations were also performed. Results are the median and range of duplicate measurements of tumor necrosis factor α (TNFα) concentrations secreted by macrophages in response to the ICs. The percentage inhibition of TNFα secretion in reference to that secreted without competitor(s) is indicated by the values above the bars.

Competition experiments with the fibrin-derived peptides were then performed before macrophage stimulation by ACPA-containing ICs. Figure 4B shows representative data from 1 of the 2 donors tested. In accordance with the ELISA results, a dose-dependent inhibition of TNFα secretion was observed when macrophages were stimulated with ICs reconstituted with ACPA+ IgG preincubated with α36-50cit38,42 or with β60-74cit60,72,74 (inhibitions up to ∼35% and up to ∼90%, respectively). Similar to the results described above, preincubation with the 2 peptides approached the inhibition efficiency of C-fibrinogen (up to ∼95% inhibition with the 2 peptides versus up to ∼100% inhibition with C-fibrinogen). No inhibition was observed after preincubation with noncitrullinated peptides or with NC-fibrinogen (results not shown).


We proposed that in the rheumatoid synovium, the disease-specific immunologic conflict between ACPAs and citrullinated fibrin plays an important role in the self-maintenance of tissue inflammation (15). Indeed, through activation of inflammatory effector mechanisms that lead to formation of new fibrin deposits and local expression of peptidylarginine deiminases that secondarily citrullinate these deposits, this conflict could sustain formation of the major ACPA target (15). One way to assess the relevance of our self-maintaining inflammation hypothesis was to confirm the capacity of ACPA-containing ICs to trigger TNFα secretion through engagement of FcγR on macrophages. Herein, by the use of an in vitro model, we were able to show that the contact of macrophages with such ICs can induce TNFα secretion. This reinforces our hypothesis and provides a strong additional argument for the pathogenic role of ACPAs in RA.

The purpose of this in vitro model was to mimic the pathophysiologic conditions of the human rheumatoid synovial tissue. The first step was to generate cells displaying characteristics of rheumatoid synovial tissue macrophages. In contrast to most prior studies, in which, in an appraisal of the inflammatory role played by antibodies in RA, the effect of ICs on myelomonocytic cells was examined using stimulation of blood-derived monocytes or mononuclear cells (21–24), but rarely macrophages (25, 26), we chose to generate macrophages from purified blood-derived monocytes and to compare the sensitivity of both cell types to IC-induced activation.

In our culture conditions, differentiation was accompanied by a substantial decrease in the percentage of FcγRI-positive cells, a slight increase in the percentage of FcγRII-positive cells, and a dramatic increase in the percentage of FcγRIII-positive cells. Whereas the induction of FcγRIII during macrophage differentiation is documented (20, 27), we could not find any reports describing similar M-CSF modulation of FcγRI expression. Expression of all 3 types of FcγR by rheumatoid synovial tissue macrophages has been reported (25, 28), with one study reporting a lower expression of FcγRI compared with FcγRII and FcγRIII (28). This shows a good correspondence between these cells and the macrophages generated at the level of FcγR expression.

In addition, our strategy to use macrophages rather than monocytes was validated. Our results showed that, in comparison with monocytes, macrophages exhibited systematically higher TNFα responses to ICs.

The second step was to reconstitute ICs to resemble those formed by local interaction between ACPAs and citrullinated fibrin, which, in the RA synovium, is mostly present in insoluble amorphous deposits (4, 15). In vitro citrullinated human fibrinogen was chosen as the antigen because it is easier to handle than fibrin and harbors most, if not all, of its ACPA-targeted epitopes, potentially allowing formation of multivalent ICs able to mediate FcγR coaggregation. It was immobilized to imitate fibrin deposits. ICs were then formed by specific immunocapture of ACPAs from RA sera, available as ACPA+ IgG solutions, since the goal of this purification was the elimination of serum proteins interacting with ICs and macrophages, notably complement proteins. The use of ACPAs derived from RA sera allowed us to test the effect of a mixture of ACPA IgG, representing that found in vivo, i.e., corresponding to mainly IgG1 and IgG4 (18), and possibly harboring particular disease-related glycosylation patterns (29), all of which are characteristics that are potentially influential in the interaction with FcγR and, consequently, in the mediated effector functions (30–32).

Interestingly, our observation that iHAGG were more potent than reconstituted ACPA-containing ICs as an inducer of TNFα secretion by both monocytes and macrophages emphasizes that differences exist in the efficiency of differently constituted IgG aggregates (whether artificially induced by heat or formed by cognate interactions with multiepitopic antigens) in triggering effector functions, confirming the relevance of using “genuine” ICs. Although this was clearly beyond the scope of the present study, it would also have been interesting to compare the effect of ACPA-containing ICs with that of other ICs involving other RA-associated antibodies that might play a role in the pathophysiology of RA (e.g., rheumatoid factor), in order to evaluate their respective contributions to synovial inflammation. Similarly, following interaction of ACPAs with soluble (minor) antigenic targets present in the synovial tissue or fluid (33, 34), soluble ICs may also form. Thus, it would be interesting to assess their inflammatory potential.

To identify which of the 3 described classes of FcγR is involved in the IC-induced stimulation of macrophages, TNFα secretion was evaluated after selective blockade of receptor interaction with the IgG. F(ab′)2 fragments of mouse mAb to these receptors (clones 10.1, 7.30, and 3G8, specific for FcγRI, FcγRII, and FcγRIII, respectively), which recognize epitopes coinciding with the binding site for IgG and/or which have previously been used successfully as FcγR blockers, were utilized (35–37). Among members of the FcγRII class of receptors, the stimulatory FcγRIIa and inhibitory FcγRIIb types are present at the surface of macrophages (38), and both are recognized by the 7.30 mAb. The results obtained clearly indicate a largely predominant role for the engagement of FcγRIIa and no or limited involvement of FcγRI or FcγRIII in macrophage activation, since the net effect of F(ab′)2 fragments of the 7.30 mAb was a very high inhibition of TNFα secretion. Moreover, we did not observe any significant inhibition after blockade of FcγRI or FcγRIII. Whether and to what extent this activation is modulated by engagement of FcγRIIb remains to be determined.

Two recent studies from Rönnelid et al have also indicated a major role for FcγRIIa in the induction of TNFα by peripheral blood mononuclear cells (PBMCs), after stimulation of these cells with ICs connected to the pathophysiology of RA. In these studies, PBMCs were stimulated by ICs precipitated by polyethylene glycol from the serum or synovial fluid of RA patients (23) or by ICs formed by the interaction of surface-bound human type II collagen (CII) with RA sera exhibiting IgG reactivity to this antigen (24). Both stimuli were shown to induce TNFα secretion, which mainly originated from monocytes, as shown by depletion experiments. Moreover, the use of Fab fragments of the IV.3 mAb to FcγRII, a clone that preferentially binds to the FcγRIIa isoform, was shown to inhibit this secretion, whereas blockade of FcγRIII could not negatively influence its secretion.

In addition, supporting evidence of the role of FcγRIIa in human RA was also obtained after transfer of the human gene of this primate-restricted receptor in mice (39). Indeed, in comparison with nontransgenic controls, transgenic mice exhibited enhanced susceptibility to collagen-induced arthritis (CIA) or to arthritis induced by passive transfer of an anti-CII mAb. Furthermore, CIA in the FcγRIIa-transgenic animals was significantly reduced after treatment with F(ab′)2 fragments of an FcγRIIa-specific mAb. In addition, consistent with observations reported in the present study, HAGG-induced secretion of TNFα by peritoneal macrophages was shown to be predominantly mediated by the transgene-encoded FcγRIIa. Finally, genomic studies have predicted that the mouse FcγRIII gene is an ortholog of the human FcγRIIa gene (40). This suggests a functional equivalence identifying the human FcγRIIa as a major suspect in the pathophysiology of RA, in view of the fact that most studies of experimental arthritides in mice have pointed to a crucial, even if not exclusive, role for the stimulatory FcγRIIIa in joint inflammation (14).

Another important finding is that macrophages were much more susceptible to IC-induced activation than were monocytes, probably reflecting multiple differentiation-induced modifications that globally mediate acquisition of enhanced competence for immune defense. However, as far as modulation of FcγR expression is concerned, the increased FcγRIII expression on macrophages is unlikely to constitute an important explanation, since blocking experiments ruled out a major involvement of this receptor in the macrophage response to ACPA-containing ICs or iHAGG. In contrast, we demonstrated the role of FcγRIIa in this response, although the percentage of FcγRII-positive cells was not drastically different between monocytes and macrophages. It is quite plausible that, during differentiation, the relative proportions of FcγRIIa and FcγRIIb are modified in favor of FcγRIIa, leading to a net stimulatory effect of FcγRII engagement in macrophages, whereas the inhibitory effect of FcγRIIb could be more prominent in monocytes.

The interindividual variability in the TNFα responses of macrophages to ACPA-containing ICs or to iHAGG could not be correlated with variations in the expression of any of the 3 classes of FcγR, particularly FcγRII. However, again, variations in the relative proportions of FcγRIIa and FcγRIIb may account for the interindividual variability. It could also be determined genetically by the presence of the H131R polymorphism of FcγRIIa, which modulates the affinity of this receptor to IgG2 (41) and, possibly, although this has not been documented, to differently glycosylated IgG, and/or by the presence of the I232T polymorphism of FcγRIIb, which has been shown to modulate the capacity of this receptor to down-modulate inflammatory responses of dendritic cells to ICs (42).

In competition experiments performed with ELISAs or macrophage stimulation assays, the lower inhibitory capacity of the peptide α36-50cit38,42 in comparison with that of the peptide β60-74cit60,72,74 was rather surprising, since both had been shown to be recognized by a high proportion of ACPA-positive RA sera (16) and both were strongly and equally recognized by the ACPA+ IgG solution (results not shown). Differences in the conformation adopted by the peptide α36-50cit38,42 depending on whether it is immobilized or in solution probably account for this inconsistency. Nevertheless, the capacity of both peptides to substantially inhibit the reactivity of a highly polyclonal mixture of ACPAs to C-fibrinogen confirms that both peptides represent major epitopes recognized by ACPAs on this antigen. Moreover, the essentially additive character of the inhibitory effect obtained when using both peptides confirms the previously observed rather limited overlap of the subgroups of ACPAs identified by each peptide (16) that, in conjunction with their major character, allows them to represent almost all of the reactivity of ACPAs to C-fibrinogen. Therefore, the peptides α36-50cit38,42 and β60-74cit60,72,74 represent highly interesting potential tools for future immunotherapeutic strategies.

Thus, in this study using a totally human in vitro model that is perfectly dedicated to the study of the interaction between macrophages and ACPA-containing ICs, we were able to demonstrate that this interaction can trigger the production of TNFα via engagement of FcγRIIa. Further use of this model should permit molecular dissection of the interaction and its functional consequences, and should enable more detailed explorations of how they could be modulated. In particular, the role of FcγRIIb in the down-modulation of the inflammatory effect should be elucidated.

It has been shown that the expression levels of FcγRII and FcγRIII are significantly increased in synovial tissue macrophages from rheumatoid joints as compared with joints with traumatic injury, and in monocyte-derived macrophages from RA-affected individuals as compared with healthy individuals, with the latter observation having functional consequences, since macrophages in patients with RA produce higher levels of TNFα in response to soluble HAGGs (25). Furthermore, monocytes from RA patients exhibit an altered balance between FcγRIIa and FcγRIIb, in favor of the stimulatory receptor, that correlates with enhanced TNFα production after stimulation with immobilized IgG (43). Comparison of the response of macrophages to ACPA-containing ICs between RA patients and healthy individuals is therefore also warranted. All of these studies will contribute to the accumulating evidence from ongoing research on the use of FcγR antagonists as a promising new therapeutic approach in patients with RA.


Dr. Serre 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 design. Clavel, Sebbag, Serre.

Acquisition of data. Clavel, Nogueira, Laurent, Iobagiu.

Analysis and interpretation of data. Clavel, Nogueira, Laurent, Sebbag, Serre.

Manuscript preparation. Clavel, Laurent, Sebbag, Serre.

Statistical analysis. Vincent.


We thank B. Pipy (Université Toulouse III) for helpful discussions, and B. Fournié, L. Zabraniecki, and O. Lemaire (Hôpital Purpan, Toulouse) and A. Cantagrel and A. Constantin (Hôpital Rangueil, Toulouse) for providing the patient data and sera. The skillful technical assistance of R. Llobera, M.-F. Isaïa, M.-P. Henry, H. Bagat, and K. Chassagne is gratefully acknowledged. We also thank F.-E. L'Faqihi-Olive (Plateau technique de Cytométrie, IFR30, Toulouse) for providing technical assistance in the flow cytometry analyses.