Regulation of human neutrophil Fcγ receptor IIa by C5a receptor promotes inflammatory arthritis in mice

Authors


Abstract

Objective

Rheumatoid arthritis culminates in joint destruction that, in mouse models of disease, is supported by innate immune molecules, including Fcγ receptors (FcγR) and complement. However, these findings may not be predictive of the outcome in humans, given the structural differences between murine and human activating FcγR on neutrophils, a prominent component of joint exudates. The aim of this study was to examine the role of human neutrophil FcγRIIa in the development of arthritis and probe the underlying mechanism by which FcγRIIa initiates disease.

Methods

K/BxN mouse serum transfer–induced arthritis was examined in mice expressing human FcγRIIa on neutrophils but lacking their own activating FcγR (γ-chain–deficient mice). The role of mast cells, complement (C3 and C5a), and CD18 integrins in FcγRIIa-initiated disease was examined using cell reconstitution approaches, inhibitors, and functional blocking antibodies, respectively. Crosstalk between the complement receptor C5aR and FcγRIIa on neutrophils was evaluated in vitro.

Results

The expression of human FcγRIIa on neutrophils was sufficient to restore susceptibility to K/BxN serum–induced neutrophil recruitment, synovitis, and bone destruction in γ-chain–deficient mice. Joint inflammation was robust and proceeded even in the absence of mast cells and vascular permeability, features shown to contribute to disease in wild-type mice. Neutrophil recruitment was dependent on the presence of a CD18 integrin, lymphocyte function–associated antigen 1, and C5aR. In addition, C5aR significantly enhanced FcγRIIa-mediated phagocytosis and oxidative burst in vitro.

Conclusion

Human and murine activating FcγR on neutrophils are not functionally equivalent, and in humans, they may play a primary role in arthritis. Crosstalk between neutrophil C5aR and FcγRIIa is essential for disease progression, thus highlighting a new aspect of complement during the effector phase of inflammatory arthritis.

Rheumatoid arthritis (RA) is an inflammatory disease of the joints with 1% prevalence in industrialized nations (1, 2). Neutrophils, lymphocytes, mast cells, macrophages, synovial tissue cells, and platelet microparticles present in the inflamed synovium have been implicated in the evolution of RA (3, 4). Circulating autoantibodies are present in a majority of patients with RA and joint tissue is frequently covered with immune complexes (ICs) (5, 6). Mechanisms of autoantibody-driven disease have begun to be elucidated in mouse models of arthritis, such as the K/BxN mouse model, which arises from the breakage of T cell tolerance followed by the generation of IgG autoantibodies to the glycolytic enzyme glucose-6-phosphate isomerase (GPI).

The K/BxN mouse model of arthritis shares certain clinical features with RA, including distal symmetric erosive polyarthritis and, at the tissue level, pannus that is erosive into bone and cartilage, prominent vascular hyperplasia, and neutrophil-rich synovial fluid. The finding that disease can be passively transferred with autoantibodies from these mice to normal recipients, together with the results obtained in knockout mouse models, support the view that although T and B cells are required for the initiation of disease, autoantibodies and innate immune mediators are the effector mechanisms that promote synovitis and joint destruction (7).

Fcγ receptors (FcγR) for complexed IgG and the complement system play essential roles in disease progression. Polymorphisms in FcγR are associated with RA (8, 9), and findings in mouse studies suggest that these receptors play a significant role in disease pathogenesis. For example, in mouse models of arthritis, the incidence and severity of disease are reduced in mice deficient for the γ-chain, which is required for the expression and signaling of activating FcγR, whereas a deficiency in the inhibitory receptor FcγRIIb exacerbates disease (10). Mast cells are among the FcγR-bearing cell types that contribute to articular inflammation, via release of mediators that induce vascular permeability, promote neutrophil recruitment, and activate resident synovial fibroblasts and macrophages (3).

Generation of complement components is a key event in the effector phase of the disease. Complement C5a levels are correlated with neutrophil accumulation in patients with arthritis (11) and in the K/BxN mouse model of arthritis, whereas mice lacking C5 or C3 are resistant to disease (12). C5a exerts its effects primarily by binding the C5a receptor (C5aR; CD88), which is expressed on a wide variety of hematopoietic cells, including neutrophils, mast cells, and macrophages (13).

Synergism between complement C3 receptors and FcγR has long been recognized to enhance adherence and ingestion of complement fragment C3b and IgG-opsonized targets (14, 15). However, the role of crosstalk between complement receptor C5aR and FcγR has been appreciated only recently. A hierarchical relationship between the 2 receptors has been suggested. C5a attracts FcγR-bearing leukocytes, and C5aR transcriptionally regulates expression of inhibitory and activating FcγR on macrophages to lower the threshold of IC activation (13). In other cases, complement and FcγR clearly play redundant roles (16), or one pathway dominates over the other (17).

Despite the significant strides made in mouse models, the results may not be predictive of the outcomes in humans, because the repertoire of activating FcγR on neutrophils that bind ICs differs significantly between the 2 species. Human neutrophils express 2 types of activating FcγR, both of which are uniquely human. Human FcγRIIa contains an immunoreceptor tyrosine–based activation motif (ITAM) and ligand-binding domain in a single polypeptide chain, and FcγRIIIb is glycophosphatidylinositol-linked. In contrast, murine neutrophils contain a transmembrane FcγRIII and a species-specific FcγRIV that complex with an ITAM-containing common γ-chain. Moreover, the contribution of FcγR alone to disease pathogenesis remains unclear, since deletion of the γ-chain in mice eliminates more than just the activating FcγR (18).

Results of previous studies suggest that expression of human FcγRIIa and FcγRIIIb on neutrophils of mice lacking their endogenous activating FcγR (i.e., γ-chain−/− mice) restores susceptibility to nephrotoxic serum nephritis and the reverse passive Arthus reaction (induced by soluble ICs) (19). In the present study, we explored the role of human FcγRIIa on neutrophils in a mouse model of inflammatory arthritis and probed the underlying mechanisms by which this receptor mediates disease.

MATERIALS AND METHODS

Mice.

Human FcγRIIa–expressing, γ-chain–deficient (hFcγRIIa+/γ−/−) mice and wild-type mice with CD18 silenced in neutrophils (miR-CD18) were generated as previously described (19, 20). Mice expressing human CD18 on neutrophils but lacking endogenous murine CD18 (hCD18+/mCD18−/−) were generated as follows. Human CD18 complementary DNA was subcloned into the Bgl II site of the pUC18-hMRP8 vector (21). A Hind III–Eco RI fragment was released and injected into zygotes from C57BL/6J mice. Transgenic mice were generated in the transgenic facility of Brigham and Women's Hospital. A mouse that was a high-expressing founder of hCD18 (hCD18+) was crossed with murine CD18–deficient (mCD18−/−) mice (22) on a C57BL/6J-F12 background and bred to be hemizygous for the hCD18 transgene and deficient in mCD18, as assessed by polymerase chain reaction of genomic DNA and flow cytometry analysis of cells.

All mice were maintained in a virus- and antibody-free facility at the New Research Building animal housing facility at Harvard Medical School. Mice used for each experiment were 8–10 weeks of age and sex matched. The Harvard Medical School Animal Care and Use Committee approved all procedures in this study.

Induction of arthritis by serum transfer, and pharmacologic treatments.

K/BxN mouse serum (25–150 μl) was administered intraperitoneally (IP) on days 0 and 2. Every second day up to day 14, the severity of arthritis was graded in all 4 paws (to yield the clinical arthritis score), and ankle joint thickness was measured on both sides using a micrometer (23, 24). The ankle joints were isolated to obtain tissue specimens, and the specimens were stained with either hematoxylin and eosin for histologic evaluation (25) or rat monoclonal antibody NIMP-R14 (Abcam) for neutrophil quantitation (26). The total neutrophil number in the closed articular cavity was counted and this value was divided by the total area of the articular cavity (in mm2) in each specimen, measured using ImageJ software (National Institutes of Health). For quantification of mast cell degranulation, mice were administered 100 μl of normal mouse serum or K/BxN mouse serum and were killed 18 hours later, and toluidine blue stain was applied to the tissue specimens (24).

For complement depletion, 12.5 units/mouse Cobra venom factor (CVF; Quidel) was injected IP on day −1, and 6.25 units CVF/mouse was injected on days 2, 5, and 10. A C5a receptor antagonist (A8Δ71–73) (27) was injected intravenously (IV) 20 minutes before and 2 hours after K/BxN serum injection, and then injected IV every 12 hours up to day 4. For blockade of lymphocyte function–associated antigen 1 (LFA-1), 100 μl of K/BxN serum was administered on days 0 and 2, and 75 μg of functional blocking rat anti-CD11a IgG2a antibody (clone M17/4; BD PharMingen) or rat IgG isotype control was injected IV on days 0, 2, and 4, as previously described (22).

Analysis of vascular permeability in mouse limbs.

Mouse heat-aggregated IgG (mHAGG) was generated essentially as previously described (28). One hundred microliters of monomeric mouse IgG (5 mg/ml), mHAGG, K/BxN serum, or normal mouse serum was diluted with 150 μl of 0.3% Evans blue and 50 μl phosphate buffered saline (PBS), and injected IV. All 4 limbs of each mouse were harvested and incubated with 600 μl dimethylformamide at room temperature for 4 days, and the extent of Evans blue staining was quantified in the supernatant by measuring the absorbance at 595 nm.

Engraftment of bone marrow–derived neutrophils (BMNs).

Mature neutrophils (>95% pure) were isolated from the bone marrow of femurs and tibias of hFcγRIIa+−/− mice or γ−/− mice, as described previously (29). BMNs (2 × 107) were administered IV into recipient W/Wv mice on days 0, 1, 2, and 3, and 200 μl K/BxN serum was administered on days 0 and 2.

In vitro neutrophil assays.

Ligand-binding assay.

BMNs (2 × 106) were stimulated with or without 100 nM leukotriene B4 (LTB4), 100 nM human C5a, and 100 μg/ml FMLP (Sigma) for 20 minutes at 37°C, and washed twice at 4°C. Cells were incubated with mHAGG at 1:100 dilution on ice for 1 hour, washed again, and stained with Alexa-conjugated donkey anti-mouse IgG (Invitrogen) and allophycocyanin (APC)–conjugated anti–Gr-1. The mean fluorescence intensity (MFI) in the Gr-1–positive cell population was evaluated by fluorescence-activated cell sorter analysis, to detect surface expression of mHAGG.

Neutrophil phagocytosis.

Sheep red blood cells (RBCs) (Lampire) were labeled with fluorescent dye PKH67, according to the manufacturer's instructions (Sigma). For analysis of IgG-opsonized RBCs, the RBCs were incubated with anti-sheep RBC IgG (Cedarlane) for 1 hour. For analysis of C3b-opsonized RBCs, the RBCs were incubated with anti-sheep RBC IgM (Cedarlane) for 1 hour, followed by incubation with C5-deficient serum (Sigma) for 1 hour. RBCs were coincubated with neutrophils at a 10:1 ratio in the presence or absence of C5a (5 μg/ml), phorbol myristate acetate (PMA; 100 ng/ml), or DMSO at 37°C for 30 minutes. External RBCs were lysed in ice-cold water, followed by the addition of trypan blue to quench fluorescence from the remaining external RBCs. Cells were visualized under a fluorescence microscope to assess the percentage of neutrophils with internalized RBCs. The normalized percentage of phagocytosis was calculated by subtracting the phagocytosis value of control RBCs from the phagocytosis value of IgG- or C3b-coated RBCs.

FcγR crosslinking–induced generation of reactive oxygen species (ROS).

BMNs (1 × 106) were suspended in PBS without Ca2+/Mg2+, and then incubated with 10 μg/ml mouse anti-hFcγRIIa (cloneIV.3; StemCell Technologies) on ice for 30 minutes. After washing in PBS twice, cells were incubated with or without 500 ng/ml hC5a for 30 minutes. Luminol (50 μM) in PBS with Ca2+/Mg2+ was added, followed immediately by the addition of goat anti-mouse F(ab′)2 (36 μg/ml; Jackson ImmunoResearch). ROS generation (expressed in relative light units) was continuously monitored over time using a 6-channel bioluminat LB-953 luminometer (Berthold).

Statistical analysis.

All data obtained in vivo are presented as the mean ± SEM. The data from in vitro experiments are presented as the mean ± SD. Statistical differences were analyzed with the unpaired t-test. P values less than 0.05 were considered significant.

RESULTS

Promotion of K/BxN mouse serum–induced arthritis by neutrophil-specific expression of hFcγRIIa.

K/BxN serum–induced arthritis was evaluated in hFcγRIIa+−/− mice, mice lacking the common γ-chain (γ−/− mice), and wild-type mice. Those γ−/− mice transgenically expressing the other human neutrophil FcγR, FcγRIIIb (19), were excluded from the analysis, as hFcγRIIIb does not recognize mouse IgG (results not shown). K/BxN serum at a dose range of 100–150 μl is the standard dosing level for induction of disease in wild-type mice. We used only 50 μl of K/BxN serum, because even this lower dose resulted in significant morbidity in hFcγRIIa+−/− mice, as indicated by our results herein. Similar to findings previously reported (12), γ−/− mice showed no evidence of disease, according to clinical and histologic parameters. In contrast, hFcγRIIa+−/− mice demonstrated disease activity that significantly exceeded that observed in wild-type mice (Figure 1A), despite having levels of FcγRIIa that were comparable to those of its murine counterpart, FcγRIII, on wild-type neutrophils (19).

Figure 1.

Induction of arthritis by K/BxN mouse serum in Fcγ receptor type IIa (FcγRIIa)–transgenic mice. A, Disease kinetics in wild-type (WT) and human FcγRIIa–expressing, γ-chain–deficient (hFcγRIIa+−/−) mice were assessed at various time points after induction of arthritis by intraperitoneal injection of the indicated doses of K/BxN serum on days 0 and 2. The total clinical score of arthritis in all 4 limbs and the change in ankle joint thickness in the hind limbs were evaluated. For comparison, the effects of 100 μl K/BxN serum were assessed in γ−/− mice. Bars show the mean ± SEM of 5 mice per group. B, Ankle joints of WT, γ−/−, and hFcγRIIa+−/− mice were examined histologically on day 14 after injection of 100 μl K/BxN serum, by staining tissue sections with hematoxylin and eosin (left) or rat monoclonal antibody NIMP-R14 for neutrophils (right). Representative samples are shown. Arrows indicate bone erosion. Bn = bone; Ca = cartilage; S = synovium. Bars = 200 μm. C, Histologic scores for inflammation and for bone and cartilage erosion were determined on days 4 and 14 after disease induction in joint sections stained with hematoxylin and eosin. Bars show the mean ± SEM of 10 joints from 5 mice per group. ∗ = P < 0.05 versus WT mice.

It is noteworthy that the inhibitory FcγRIIb is also expressed in hFcγRIIa+−/− animals and likely acts as a functional molecule, since administration of intravenous immunoglobulin, which mediates protection in wild-type mice through its action on FcγRIIb (30), was found to inhibit FcγRIIa-mediated inflammation in the K/BxN model (results not shown). Our histologic findings revealed that hFcγRIIa+−/− mice exhibited severe synovial hyperplasia and bone erosion (Figure 1B) with robust neutrophil infiltration in the synovium and bone, in contrast to that observed in γ−/− mice (Figure 1C). Thus, the expression of activating human FcγRIIa on neutrophils was sufficient to restore neutrophil influx and susceptibility to arthritis in this mouse model, in the absence of other activating FcγR.

Lack of association of hFcγRIIa-mediated arthritis with mast cell degranulation or vascular leakage.

Mast cell–deficient W/Wv and Sl/Sld mice fail to develop K/BxN serum–induced arthritis, and degranulated mast cells have been observed in the joint tissue of wild-type animals within hours of K/BxN serum transfer (24). As has been previously reported, our results showed that, after K/BxN serum transfer, the ankle specimens of wild-type mice exhibited significant mast cell degranulation, whereas mast cells remained intact in γ−/− mice, which do not succumb to disease. However, the ankle specimens of hFcγRIIa+/γ−/− mice also exhibited virtually no mast cell degranulation, despite the development of significant synovitis (Figure 2A).

Figure 2.

Contribution of mast cells to the development of K/BxN serum–induced arthritis in hFcγRIIa+−/− mice. A (left), Mast cell degranulation 18 hours after injection of 100 μl K/BxN serum in the ankle joints of γ−/−, WT, and hFcγRIIa+−/− mice was assessed by staining ankle joint sections with toluidine blue (upper panels and lower left) or staining serial sections from hFcγRIIa+−/− mice with hematoxylin and eosin (lower right) to document the inflammatory cell infiltration in the area. Representative specimens are shown (original magnification × 100). Arrows indicate mast cells; arrowheads indicate neutrophils. Msl = muscle. A (right), The mean ± SEM percentage of degranulated mast cells in the ankle joints of γ−/−, WT, and hFcγRIIa+−/− mice was determined at 18 hours after serum induction. B (left), Change in vascular permeability was determined by Evans blue staining of the limbs of WT and γ−/− mice to assess vascular leakage, 45 minutes after intravenous administration of 100 μl of mouse heat-aggregated IgG (mHAGG). Representative specimens are shown. B (right), The extent of Evans blue staining was determined in the ankle joints of γ−/−, WT, and hFcγRIIa+−/− mice, 45 minutes after administration of either mHAGG or 100 μl K/BxN serum. Bars show the mean ± SEM of 5 mice per group. Broken horizontal lines indicate the mean values in control WT animals treated with monomeric mouse IgG (left) or normal mouse serum (right) (n = 3 per group). ∗ = P < 0.005 versus γ−/− and hFcγRIIa+−/− mice. See Figure 1 for other definitions.

K/BxN serum induces joint-localized vascular permeability within 5 minutes of administration, a process that depends on mast cells, neutrophils, and FcγR in C57BL/6 mice (31). Heat-aggregated, nonspecific mouse IgG also induces vascular leakage, indicating that IgG ICs, rather than other components that are potentially present in K/BxN serum, are sufficient to promote permeability (31). Formation of ICs induced by either mHAGG or K/BxN serum did not result in vascular permeability in hFcγRIIa+/γ−/− mice, and indeed, permeability was similar to that observed in the γ−/− and wild-type mice treated with (control) monomeric IgG (Figure 2B). Permeability was also evaluated 18 hours after the introduction of K/BxN serum. At this later time point, vascular leakage, as revealed by Evans blue staining, was significantly increased in the hFcγRIIa+−/− mice (results not shown), which likely occurs secondary to the robust inflammation developing in these animals. In summary, in the presence of hFcγRIIa on neutrophils, disease proceeds through mechanisms that are independent of mast cell degranulation and is not associated with early changes in vascular permeability.

Development of FcγRIIa-mediated arthritis in the absence of mast cells.

To directly evaluate the contribution of mast cells to FcγRIIa-induced arthritis, BMNs isolated from hFcγRIIa+/γ−/− or γ−/− mice were injected IV into the W/Wv strain of mice for 4 consecutive days, and K/BxN serum was delivered into the reconstituted W/Wv mice on days 0 and 2. Absence of mast cells in the W/Wv mice was confirmed by immunostaining of ear skin (Figure 3A). W/Wv mice reconstituted with hFcγRIIa+/γ−/− BMNs exhibited significantly more disease activity than did mice injected with γ−/− BMNs (Figure 3B). Thus, in the context of human neutrophil FcγRIIa, mast cells are not essential for the development of arthritis.

Figure 3.

Expression of hFcγRIIa on neutrophils associated with induction of K/BxN mouse serum–induced arthritis in mast cell–deficient W/Wv mice. Purified bone marrow–derived neutrophils (BMNs) from hFcγRIIa+−/− or γ−/− mice were delivered intravenously for 4 consecutive days into mast cell–deficient W/Wv recipient mice. A, Ear skin specimens from WT and W/Wv mice were stained for mast cells (indicated in blue). Mast cells (arrow) were detected only in WT mice and not in W/Wv mice, as expected. B, Disease kinetics in W/Wv mice reconstituted with BMNs from γ−/− mice (γ−/− BMN→W/Wv) or BMNs from hFcγRIIa+−/− mice (hFcγRIIa+−/− BMN→W/Wv) were determined at various time points after arthritis induction, by assessing ankle joint swelling in the hind paws (representative images [top panels]; bars = 2 mm) as well as the total clinical score of arthritis in 4 limbs and change in ankle joint thickness in the hind limbs. Bars show the mean ± SEM of 4 mice per group per experiment. See Figure 1 for other definitions.

Required presence of neutrophil CD18 for the development of FcγRIIa-mediated RA.

Mice that lack all CD18 integrins, or mice in which these integrins are functionally blocked, develop minimal arthritis (22). The significance of CD18 integrins on neutrophils was evaluated by exploiting 2 complementary genetic approaches, one involving the silencing of CD18 selectively in neutrophils of wild-type mice (miR-CD18 mice) (20), and the other involving expression of human CD18 selectively in neutrophils of mice lacking endogenous CD18 (hCD18+/mCD18−/− mice) (details available from the corresponding author upon request). These 2 mouse models, together with wild-type mice and CD18-deficient mice, were subjected to K/BxN serum–induced arthritis. Neutrophil CD18 silencing impaired the development of arthritis and CD18-deficient mice were resistant to disease, whereas hCD18+/mCD18−/− mice developed arthritis that was comparable to that in wild-type mice (Figure 4A). Taken together, these results provide compelling evidence that neutrophil CD18 is critical in the effector phase of arthritis.

Figure 4.

Involvement of neutrophil CD18 in the pathogenesis of K/BxN mouse serum–induced arthritis. A, Arthritis was induced with 100 μl of K/BxN serum, and disease severity was evaluated in wild-type (WT) mice (n = 6), mice with CD18 silenced in neutrophils (miR-mCD18+; n = 6), CD18-deficient mice (mCD18−/−; n = 5), and CD18-deficient mice expressing human CD18 in neutrophils (hCD18+/mCD18−/−; n = 5). B, Human Fcγ receptor type IIa–expressing, γ-chain–deficient (hFcγRIIa+−/−) mice received 100 μl of K/BxN serum on days 0 and 2 and were treated with either anti–lymphocyte function–associated antigen 1 monoclonal antibody (anti–LFA-1 mAb) or isotype IgG control, administered by tail vein injection at the indicated time points (arrows). In A and B, the clinical score of arthritis in 4 limbs and the change in ankle joint thickness in the hind limbs were evaluated at various time points after disease induction. C, Effects of the anti–LFA-1 mAb or IgG control in hFcγRIIa+−/− mice were assessed by staining tissue sections with hematoxylin and eosin (H&E) or NIMP-R14. Representative samples are shown. Bn = bone; Ca = cartilage; S = synovium. Bars = 200 μm. D and E, Histologic scores for inflammation and bone and cartilage erosion were determined in tissue sections stained with H&E (D), and polymorphonuclear neutrophil (PMN) influx into the articular cavity was assessed in samples stained with NIMP-R14 (E) on day 6 after disease induction in ankle joints treated with either anti–LFA-1 mAb or IgG control. ∗ = P < 0.005. Bars in A, B, and D show the mean ± SEM.

Given the reliance of K/BxN serum–induced arthritis on neutrophil CD18, we assessed the contribution of CD18 specifically to the induction of arthritis in the hFcγRIIa-expressing mice. Studies have shown that a functional blocking antibody to LFA-1 blocks K/BxN mouse serum–induced arthritis in wild-type mice (22). Our results showed that this antibody completely suppressed the clinical and histologic features of arthritis in hFcγRIIa+−/− mice (Figures 4B, C, and D). Moreover, these effects correlated with minimal neutrophil accumulation in the mouse joints after antibody blockade (Figure 4E).

Essential role of complement in FcγRIIa-mediated RA.

Wild-type and hFcγRIIa+−/− mice were evaluated following the depletion of C3 by systemic administration of CVF (Figure 5A) or following the blockade of C5aR using a specific antagonist of the C5a receptors CD88 and C5L2 (27) (Figure 5B). Our results showed that arthritis did not develop in either mouse strain treated with CVF (Figure 5A). However, C5aR blockade attenuated arthritis only in hFcγRIIa+−/− mice, whereas the C5aR antagonist in wild-type mice failed to have any effect, and indeed there was a trend toward greater severity of clinical disease in wild-type mice, for reasons that are unclear (Figures 5B and C). Consistent with these findings, a significant reduction in neutrophil influx was observed only in the hFcγRIIa+−/− mice treated with the C5aR antagonist (Figure 5D). Thus, C5aR regulates hFcγRIIa-dependent neutrophil recruitment and joint injury.

Figure 5.

Role of complement C3 and C5a in FcγRIIa-mediated arthritis. WT and hFcγRIIa+−/− mice were administered K/BxN mouse serum on days 0 and 2 (using doses of 100 μl and 50 μl, respectively, to ensure similar severity of disease in both strains). A, Cobra venom factor (CVF) or phosphate buffered saline (PBS) was injected intravenously (IV) into WT and hFcγRIIa+−/− mice on the indicated days (arrows), and the total clinical score of arthritis in 4 limbs and the change in ankle joint thickness in the hind limbs were evaluated at various time points after disease induction. Bars show the mean ± SEM of 5 mice per group. B, C5a receptor antagonist (A8Δ71–73) or PBS was delivered IV into hFcγRIIa+−/− mice at the indicated time points (arrows), and the total clinical score of arthritis in 4 limbs and the change in ankle joint thickness in the hind limbs were evaluated up to 4 days after arthritis induction. Insets, Same treatments in WT mice. Bars show the mean ± SEM of 9–10 mice per group. ∗ = P < 0.05 versus CVF-treated or A8Δ71–73–treated group for all points up to 8 days and 4 days, respectively. C and D, Histologic scores of inflammation and bone and cartilage erosion on day 4 after disease induction (C) and quantitation of polymorphonuclear neutrophil (PMN) influx into the articular cavity (D) were assessed in A8Δ71–73–treated and PBS-treated WT and hFcγRIIa+−/− mice. Bars are the mean ± SEM of 5 mice per group. Consistent with the results in B, A8Δ71–73 blocked neutrophil influx only in hFcγRIIa+−/− mice and not in WT mice. ∗ = P < 0.05 versus PBS-treated group. See Figure 4 for other definitions.

Promotion of hFcγRIIa-dependent neutrophil cytotoxicity by C5a.

Activation of the recruited neutrophils via FcγR and the type 3 complement receptor (CR3) is essential for tissue damage in vivo (29, 32). We therefore conducted studies in vitro to explore the possibility that, in addition to effects on neutrophil recruitment, C5a regulates arthritis in hFcγRIIa-expressing mice by increasing FcγRIIa-dependent neutrophil cytotoxicity. The bacterial peptide FMLP increases FcγRIIa-dependent ligand binding (33). Human FcγRIIa+−/− neutrophils treated with C5a or other G protein–coupled receptor agonists, FMLP and LTB4, exhibited enhanced binding of ICs, although the MFI in treated cells was not significantly higher than that in untreated controls (P = 0.07) (Figure 6A).

Figure 6.

Promotion of FcγRIIa-mediated phagocytosis and generation of reactive oxygen species (ROS) by C5a. Bone marrow–derived neutrophils (BMNs) were harvested from hFcγRIIa+−/− and γ−/− mice, and binding of mouse heat-aggregated IgG (mHAGG) (A), phagocytosis of IgG- and C3b-opsonized fluorescently labeled red blood cells (RBCs) (B), and FcγRIIa crosslinking–induced generation of ROS (C) were analyzed. A, BMNs were incubated without or with agonists (100 nM human C5a, 100 nM leukotriene B4 [LTB4], or 100 μg/ml FMLP) and mHAGG, and bound IgG was detected by flow cytometry. Bars show the mean ± SEM mean fluorescence intensity (MFI) in 5 experiments. B, BMNs were incubated with C5a, phorbol myristate acetate (PMA), or negative controls (phosphate buffered saline [PBS] or DMSO) for 30 minutes, along with IgG- or C3b-coated fluorescently labeled sheep RBCs. Left, Representative green fluorescent images of phagocytosis by hFcγRIIa+−/− neutrophils (arrows) are shown. Bar = 50 μm. Right, The normalized percentage of phagocytosis under each condition was determined in γ−/− and hFcγRIIa+−/− BMNs. Bars show the mean ± SD results in 3 experiments. ∗ = P < 0.05. NS = not significant. C, BMNs were incubated with or without antibody to FcγRIIa (anti-RIIA) followed by PBS or C5a (+ C5a) for 30 minutes. ROS were detected following the addition of a secondary antibody against FcγRIIa (2° Ab). The generation of ROS, detected following FcγRIIa crosslinking, was significantly enhanced by C5a in hFcγRIIa+−/− samples, whereas minimal ROS generation was detected in control samples (hFcγRIIa+−/− neutrophils without FcγRIIa antibody [C5a + 2° Ab or 2° Ab]) and γ−/− neutrophils. Representative results from 1 of 4 experiments are shown. See Figure 4 for other definitions.

We also examined the phagocytic functions of FcγRIIa in the presence of C5a. Phagocytosis of IgG-opsonized RBCs was undetectable in hFcγRIIa-expressing neutrophils, whereas pretreatment of cells with C5a for 30 minutes induced significant phagocytosis (Figure 6B). The enhancement was specific for FcγR, since phagocytosis of complement C3b–opsonized RBCs, mediated by CR3 (32), was largely unaffected by C5a but was robust in the presence of PMA (Figure 6B), as has been previously described (32).

To determine whether C5a can regulate FcγR signaling downstream of receptor engagement, we evaluated whether C5a could enhance FcγRIIa-crosslinking–induced generation of ROS (Figure 6C). ROS was not detected in hFcγRIIa+−/− neutrophils pretreated with C5a for 30 minutes (Figure 6C), which was as expected, since ROS was found to return to baseline levels within 5 minutes of C5a treatment (results not shown). FcγRIIa crosslinking alone produced minimal production of ROS. In contrast, hFcγRIIa+−/− neutrophils pretreated with C5a prior to FcγRIIa crosslinking exhibited a significant and rapid enhancement of ROS production (Figure 6C). C5a, under the treatment conditions used in the above-described assays, did not increase surface FcγRIIa levels (results not shown). In summary, C5a is essential for FcγRIIa-dependent proinflammatory effector functions.

DISCUSSION

There is considerable evidence in humans and animal models that ICs and innate immune mediators promote inflammation in the joints and other tissues. However, the hierarchy of events and the downstream cellular and molecular effector mechanisms remain largely unknown. Furthermore, since significant divergence exists in the FcγR structure and function between mice and humans, the relevance of results obtained only with murine FcγR for human IC-driven diseases is debatable. Previous mechanistic studies in the K/BxN mouse serum–induced arthritis model have shown that antigen–antibody IC deposition in the articular cavity leads to activation of tissue-resident FcγR-expressing cells that then feed back to initiate recruitment of circulating leukocytes (34). Importantly, since FcγRIIa is found only in higher primates, the role of neutrophils activated by this pathway remains unexplored, both in this model and in human disease. Herein we demonstrate that expression of human FcγRIIa in neutrophils alone confers significant susceptibility to K/BxN mouse serum–induced arthritis, which, in fact, exceeds that observed in wild-type mice, which express their whole complement of endogenous FcγR.

We further show that FcγRIIa on neutrophils is acutely regulated by complement C5a. The crossregulation of C5aR and FcγRIIa provides an important link between the FcγR and the complement network, both of which connect upstream initiating events in RA to downstream effector responses (12). C5a also contributes to neutrophil recruitment, but it is the integrin LFA-1 that is critical in this respect, as its blockade ameliorates FcγRIIa-dependent neutrophil influx and the associated joint destruction. Thus, human FcγRIIa, C5aR, and LFA-1 represent an important regulatory complex on neutrophils that can coordinate neutrophil influx and cytotoxicity in IC-driven inflammation.

Engagement of human neutrophil FcγRIIa is well known to trigger a cascade of signaling events that lead to phagocytosis, ROS production, protease and leukotriene release, and cytokine production in vitro (1). It is expressed on neutrophils, as well as a number of leukocyte subsets, platelets, mast cells, Langerhans' cells, and dendritic cells (35, 36). In previous studies, transgene expression of FcγRIIa, primarily on platelets of wild-type mice, increased susceptibility to collagen-induced arthritis (37). Herein the neutrophil-selective expression of FcγRIIa in the absence of other FcγR demonstrates the potency of this molecular pathway in activating neutrophils in the context of inflammatory arthritis. We cannot formally rule out a role for the ∼20% of monocytes expressing FcγRIIa (19) in the development of arthritis. However, K/BxN mouse serum–induced arthritis proceeds normally in colony-stimulating factor 1–deficient (op/op) mice (30), which have a >90% reduction in monocytes and macrophages (38), and FcγRIIa is not detectable on macrophages of FcγRIIa+−/− mice (19). Therefore, the contribution of FcγRIIa on monocytes is likely not significant.

Interestingly, the inflammatory pathway triggered by FcγRIIa does not proceed via increases in vascular permeability, a function previously ascribed to both neutrophils and mast cells (31). Furthermore, the stimulation of neutrophils via FcγRIIa is not associated with mast cell degranulation and, indeed, proceeds in the absence of these cells. Thus, human and murine activating FcγR are not functionally equivalent, with human neutrophil FcγRIIa playing a primary and proximal role in the effector phase of arthritis. This might require a reevaluation of our conclusions regarding the pathogenesis of arthritis based on previous studies on murine receptors.

Our findings suggest that neutrophils were recruited through their own FcγRIIa. GPI expression has been observed on the surface of the synovial lining and on the endothelial cell surface (39). The latter observation suggests the possibility that FcγRIIa may directly interact with GPI–anti-GPI ICs within the joint vasculature in K/BxN serum–induced arthritis to promote neutrophil recruitment, similar to that shown, by intravital microscopy, to occur in the cremaster muscle following intravascular and tissue IC deposition (19).

Our analyses demonstrate a surprising requirement for the integrin LFA-1 in FcγRIIa-dependent contributions of neutrophils to arthritis. Although neutrophil recruitment in the joints of wild-type mice was previously reported to be dependent on neutrophil CD18 (22, 40), the requirement for LFA-1 in our studies was somewhat unexpected, as FcγRIIa can directly bind ICs both in vitro and in vivo (19). The absolute requirement for additional adhesion receptors was thus not apparent a priori. LFA-1 is known to co-cap with FcγR (41), and crosslinking of FcγR triggers proximity between the CD18 integrins and cortical microfilaments (42). We speculate that FcγRIIa engagement may lead to activation of LFA-1 and its subsequent interaction with its ligand, intercellular adhesion molecule 1, on the activated endothelium.

The role of complement in neutrophil FcγRIIa-driven responses is also noteworthy. In wild-type mice, the alternate pathway of complement activation drives K/BxN mouse serum–induced arthritis (12). Herein we show that C3 and C5a are required for FcγRIIa-induced disease pathogenesis. The effect of C5aR antagonism on neutrophil accumulation may have been expected, as C5a is a powerful neutrophil chemoattractant. The additional information provided by our studies is that the effect of C5a appears to occur cooperatively with FcγRIIa on neutrophils, since wild-type mice were resistant to C5aR blockade with the antagonist. The reason for the discrepancy between our results with the antagonist and previous studies that have shown that C5aR (CD88)–deficient mice are protected from K/BxN mouse serum–induced arthritis (12) is unclear. It is possible that antagonist penetration into the joint tissue is inadequate to fully block C5aR. In contrast, the C5a receptor antagonist also blocks the other C5aR, C5L2, which may act as a decoy receptor for C5a, as has been previously demonstrated in some other models (13).

It is important to understand how C5a regulates FcγRIIa-dependent arthritis. Previous studies suggest that the regulatory impact of C5a on FcγR on macrophages is mainly due to transcriptional down-regulation of the α-chain of FcγRIIb (13) or up-regulation of the γ-chain (43, 44), which is lacking in our system. Our studies in vitro suggest that C5a has a more immediate effect by promoting FcγRIIa-dependent phagocytosis and ROS generation in neutrophils. The absolute requirement for C5a in FcγRIIa-mediated phagocytosis of IgG-opsonized RBCs suggests that FcγRIIa is relatively inactive in resting cells, which we anticipate may have evolved as a mechanism to maintain circulating neutrophils in a quiescent state, much like leukocyte CD18 integrins that are activated to bind the ligand only upon stimulation with mediators of inflammation (45). Consistent with this notion, under physiologic flow conditions in vitro, FcγRIIa was insufficient to capture neutrophils on plate-immobilized ICs (46) or IgG-coated endothelial cells (47), but in the presence of a protein G–coupled receptor agonist, such as chemokines, FcγRIIa firmly adhered to ICs both in vitro (47, 48) and in vivo (19).

We speculate that the C5a pathway may be particularly important under conditions of low-avidity engagement of FcγR, where it may regulate the amplitude of the response. Interestingly, C5a appears to have some selectivity in its activity toward FcγR. The effect of C5a on complement C3/CR3–induced phagocytosis was marginal relative to its effects on FcγR-mediated uptake. There is increasing appreciation of the crossregulation between ITAM-associated FcγR and Toll-like receptors (TLRs) and cytokine receptors (18, 49). However, unlike the TLRs and cytokine receptors, which indirectly link to ITAM-containing receptors, C5aR and FcγR crosstalk in a direct manner. The observation that C5a enhanced ROS generation upon direct clustering of FcγRIIa at the cell surface suggests a role for C5a/ C5aR downstream of IgG binding. C5aR may relocalize FcγRIIa to lipid rafts enriched in signaling molecules (50, 51), and/or engagement of C5aR and the ITAM-containing FcγRIIa may lead to convergent MAPK signaling (52).

In conclusion, the results of our study provide insight, previously unappreciated, into the functional contributions from neutrophil FcγRIIa to joint inflammation in a humanized mouse model. Our studies focus attention on neutrophil FcγRIIa and, by extension, neutrophils as critical links between antibodies and immunologic injury. This could not have been predicted from previous studies, which were focused on the murine FcγR. Translationally, since FcγRIIa-bearing neutrophils comprise the predominant population in inflamed joint fluid in RA, with neutrophil influx estimated at more than a billion cells/day in a single joint, it is likely that these insights bear significant relevance for our understanding of disease physiology and suggest that neutrophil FcγRIIa may represent a viable therapeutic target for treatment of RA. Furthermore, there is growing evidence of the role of neutrophils and ICs in numerous other human diseases, such as glomerulonephritis, IC vasculitis, skin autoimmune diseases, and others (1). Thus, it is highly likely that our finding of an FcγRIIa:C5aR:LFA-1 receptor triad that functions to provide tight regulation and fine tuning of neutrophil responses to ICs has relevance in other autoimmune diseases that display evidence of IC pathophysiology.

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. Mayadas 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. Tsuboi, Lee, Mayadas.

Acquisition of data. Tsuboi, Ernandez, Li, Nishi, Mekala, Hazen.

Analysis and interpretation of data. Tsuboi, Ernandez, Li, Nishi, Cullere, Mekala, Köhl, Lee, Mayadas.

Acknowledgements

The hMRP8 promoter construct was kindly provided by Dr. E. Lagasse (University of Pittsburgh) and I. L. Weissman (Stanford University). We are grateful to Nick Calderone (Brigham and Women's Hospital) for excellent technical assistance and Asuka Shimizu for generously helping with the measurements obtained from histologic slides.

Ancillary