Macrophages Induce the Inflammatory Response in the Pulmonary Arthus Reaction through Gα_i2 Activation That Controls C5aR and Fc Receptor Cooperation¹

The mechanisms by which the immune system controls effector responses to pathogenic Abs and immune complexes (IC)⁶ are of central importance for understanding how immunologic diseases develop. The initial response to IC challenge is local inflammation, which provokes perivascular edema, hemorrhage, vascular permeability, and leukocyte extravasation. Based on initial work in the classical model of IC inflammation, the Arthus reaction, it has long been understood that formation of IC propagates tissue damage primarily by complement activation via the classical pathway. This view, however, has been changed during the last years by the advent of gene knockout technology in mice. It is now clear that, dependent on the property of the target tissue, the alternative complement pathway also plays a crucial role (1), and that late complement components, C5a and C5aR, are essential effectors in the majority of complement-mediated inflammatory reactions (2, 3, 4). IgG FcR (FcγR) are the other well-established key effectors that regulate inflammatory responses by the balance of activating FcγRIII and inhibitory FcγRIIB signals (5). Activating FcγRI and FcγRIII are expressed in association with dimers of the signal-transducing FcRγ chain, which contains an ITAM sequence in its cytoplasmic tail (6). FcγRIII and FcγRIIB bind IC with comparable affinity and specificity, and coupling of FcγRIII with FcγRIIB, which has an ITIM motif in the cytoplasmic domain, results in tyrosine phosphorylation of FcγRIIB ITIM and subsequent inhibition of the FcγRIII ITAM-triggered activation signal (7). In vivo, FcγRIIB-deficient mice exhibit enhanced susceptibility in models of inflammatory autoimmune disease (8), whereas FcγRIII-deficient mice show diminished disease phenotypes (4, 9, 10). Recently published data for FcγRI-deficient mice re-emphasize the requirement of FcγRI for immune defenses against bacterial infection and, in concert with FcγRIII, for the development of autoimmune hemolytic anemia (11).

Despite increased knowledge of how complement and FcγR each contribute to various immunologic diseases, the critical role of their possible interaction has only recently been addressed (12). In skin Arthus reactions, the presence of a C3b-independent, C5a-generating pathway can overrule impaired humoral complement activation in C3-deficient mice (13), whereas C5aR modulates FcγR in alveolitis and peritonitis (14, 15). These findings indicate that the convergence of complement and FcγR effector pathways at the level of C5a:C5aR is of central importance for disease pathogenesis. However, the underlying cellular mechanisms have not yet been fully defined. In this study we show, through the use of adoptive cell transfer systems in conjunction with genetic deletion and pharmacological inhibition studies, that macrophages are the primary effectors that induce the lung Arthus reaction by sequential activation steps and provide novel insights into the specific requirements of the receptor systems involved, C5aR and FcγRIIB/FcγRIII. Our findings provide direct evidence for a refined model of the inflammatory cascade with local G_i-dependent C5aR-FcγR cross talk as the key regulatory event.

The generation of FcγRI-, FcγRIII-, and Gα_i2-deficient mice and their phenotypic characterization has been described in detail (11, 16, 17, 18). Gα_i3-deficient mice have been generated as described by Jiang et al. (19), and detailed analysis of this strain will be published elsewhere (our manuscript in preparation). FcγRIIB- and C5aR-deficient mice were provided by T. Takai (Department of Experimental Immunology and CREST Program of Japan Science and Technology Agency, Institute of Development, Aging, and Cancer, Tohoku University, Sendai, Japan) (20) and U. Höpken (Department of Tumor Genetics and Immunogenetics, Berlin, Germany) (21). All gene-deleted mice were backcrossed to C57BL/6 mice for four to eight generations and maintained under dry-barrier conditions. CD45.1⁺ congenic C57BL/6 mice (B6.SJL-PtrcaPep3b/BoyJ) were purchased from The Jackson Laboratory. C57BL/6 control mice were obtained from Charles River. All these mice were used at 8–14 wk of age. Animal experiments were conducted in accordance with current laws in combination with the regulations of the local authorities.

The following Abs were used: anti-CD45.1 (clone A20), anti-CD45.2 (clone 104), anti-FcγRII/III (clone 2.4G2; all from BD Pharmingen), anti-F4/80 (Serotec), anti-FcγRIIB/Ly17.2 (clone K9.361) (22), and anti-C5aR (clone 1/36) (23). F4/80, CD45.1, and CD45.2 Ags on bronchoalveolar lavage (BAL)-alveolar macrophage (AM) cells (2 × 10⁴) were analyzed by flow cytometry using a FACScan flow cytometer (BD Biosciences). Expression of FcγRIIB was analyzed by staining with the anti-FcγRIIB mAb Ly17.2 conjugated to FITC. To analyze FcγRIII expression, BAL-AM cells were first treated with saturating doses of unlabeled anti-FcγRIIB Ly17.2 mAb, followed by PE-2.4G2 anti-FcγRII/III mAb staining, as previously described (14, 22).

Mice were anesthetized with ketamine and xylazine, the trachea was cannulated, and 150 μg of protein G-purified, rabbit anti-OVA IgG Ab (Sigma-Aldrich) was applied. Immediately thereafter, 20 mg/kg OVA Ag was given i.v.; Ab control animals received PBS instead of OVA Ag. In AM depletion experiments, mice intratracheally (i.t.) received liposome-encapsulated dichloromethylene diphosphonate (Cl₂MDP; at a concentration of 5 mg/ml; Lipo-clod) as previously described (Cl₂MDP was a gift of Roche Diagnostics) (24). Initial studies indicated a single dose of 100 μl and the application route used to be most effective for selective depletion of AM within 1–3 days, lasting for at least 5–6 days with renewing AM first detectable on day 7 and full repopulation after 18 days. Intratracheal delivery of PBS-containing liposomes (Lipo-PBS) did not show any depletion effects and served as the control. In additional blocking experiments, mice received either 1/36 mAb directed against C5aR to block C5aR activity, pertussis toxin (PTX) to inhibit G_i function, or soluble FcR fragments (sFcR), consisting of the unglycosylated extracellular part of human FcγRIIb (25), to inhibit cellular activation of FcγR. Doses of 20 μg of 1/36 mAb, 800 ng of PTX, and 50 μg of sFcR were given i.t. before the application of anti-OVA IgG. Mice were killed at various time points (ranging from 2–24 h) after initiation of the Arthus reaction, and lung tissues obtained after lavage were assayed for interstitial polymorphonuclear leukocyte (PMN) accumulation by myeloperoxidase (MPO) activity as previously described (12).

Pulmonary vasculature was gently flushed with PBS with a catheter positioned in the root pulmonary artery. Lungs were lavaged with PBS (1 ml, five times at 4°C) after cannulation of the trachea. The volume of the collected BAL fluid (BALF) was measured in each sample, and total cell count was assessed with a hemocytometer (Neubauer Zählkammer). The amount of erythrocytes represented the degree of hemorrhage. For quantitation of PMN accumulation, differential cell counts were performed on cytospins (10 min at 55 × g) stained with May-Grünwald/Giemsa using 300 μl of BALF. The concentrations of MIP-2, TNF-α, and IL-1β in BALF were measured in duplicate in appropriately diluted samples with respective MIP-2-, TNF-α-, and IL-1β-specific ELISA kits (R&D Systems) according to the manufacturer’s instructions. For FcγR and C5aR expression analysis, total RNA was prepared from BAL-AM cells. FcγR and C5aR mRNA levels normalized to tubulin were quantitated by TaqMan real-time RT-PCR using published FcγR/C5aR-specific primers and probes (14, 26).

Bone marrow cells (containing 64–68% PMN) from C57BL/6 and C5aR-null mice were suspended at 7.5 × 10⁶ cells/ml RPMI 1640 medium and 0.5% BSA. One hundred microliters of the bone marrow cell suspension (treated, or not, with anti-C5aR 1/36 mAb and PTX) was placed into the insert of a Transwell chemotaxis chamber, and the bottom well was filled with 600 μl of RPMI 1640/0.5% BSA, the same medium supplemented with 5 × 10⁻⁹ M rhC5a, appropriately diluted MH-S AM cell culture supernatants obtained at different times after IC activation, or BALF diluted 1/2 in RPMI 1640/1% BSA. BALF were obtained from Lipo-clod-treated and untreated C57BL/6 mice at 2 and 4 h after OVA:anti-OVA IC inflammation. BALF from mice receiving Ab, but not OVA Ag, served as controls. Inserts were combined to the lower chambers and incubated at 37°C and 6% CO₂ for 2 h. After the incubation, 50 μl of 70 mM EDTA solution was added to the lower chambers to release adherent cells from the lower surface of the membrane and from the bottom of the well. Plates were further incubated 30 min at 4°C, inserts were removed, and the transmigrated neutrophils were vigorously suspended and counted with a FACSCalibur for 1 min at 60 μl/min with gating on forward and side scatter. Migration of PMN from the insert to the bottom well was quantitated as a percentage of the total PMN loaded into the upper chamber.

Mouse MH-S AM cell line cells (27) were maintained in RPMI 1640 medium containing 10% FCS and supplements. In functional assays, 10⁶ MH-S cells were incubated overnight in six-well culture plates containing 1% FCS/RPMI 1640 medium and activated with 100 μg/ml heat-aggregated IgG IC (mouse IgG1) or 5 × 10⁻⁹ M rhC5a (Sigma-Aldrich). In some experiments MH-S cells were also incubated with anti-C5aR mAb 1/36 (100 μg/ml) or PTX (500 ng/ml; for 2 h; Calbiochem) to block C5aR-G_i activity. After various time periods (0–16 h), appropriate dilutions of culture supernatants from untreated and stimulated MH-S cells were examined for production of TNF-α and MIP-2 by ELISA and C5a-dependent chemotactic activity by Transwell migration assays. Total RNA was prepared and analyzed for FcγR mRNA synthesis by TaqMan RT-PCR.

For detection and identification of PTX-sensitive G_i proteins [³²P]ADP-ribosylation in the presence of PTX and immunoblot analysis of cell membranes were performed as previously described (28). In brief, mouse embryonic fibroblasts (MEFs) from wild-type (WT), Gα_i2^(−/−) and Gα_i3^(−/−) embryos (our manuscript in preparation), MH-S cells, and Jurkat T cells were harvested and disrupted by three freeze-thaw cycles. Cell membranes were isolated and subjected to [³²P]ADP ribosylation in the presence of PTX. Cell membranes (100 μg/lane) and PTX-treated membranes (40 μg/lane) were subjected to urea-containing SDS-PAGE, and G_i proteins were either detected by staining with a Gα_common Ab (AS 8) (29) and visualized using the ECL chemiluminescence system (Amersham Biosciences) and a Diana chemiluminescence-Imager (Raytest) or subjected to phosphorimaging using a FLA 5000 Fuji-Imager (Raytest), respectively.

Selective AM depletion in CD45.1⁺ congenic C57BL/6 recipient mice was performed by i.t. administration of Lipo-clod for 3 days. Reconstitution of such depleted CD45.1⁺ congenic mice with CD45.2⁺ AM was performed by i.t. injection of 5 × 10⁵ CD45.2⁺ BALF-AM cells from WT C57BL/6 donors (pretreated, or not, with either 20 μg/ml anti-C5aR 1/36 mAb or 500 ng/ml PTX for 2 h) or CD45.2⁺ B6 mice deficient in FcγRI, FcγRIIB, FcγRIII, or C5aR. The effectiveness of depletion of CD45.1⁺ AM and subsequent adoptive transfer of the different CD45.2⁺ AM cells was followed in individual mice by FACS analysis using anti-F4/80, anti-CD45.1, and anti-CD45.2 Abs. After allowing 24 h for reconstitution, the lung Arthus reaction was induced and assessed as described above.

AM depletion/reconstitution in CD45.2 C5aR^−/− and FcγRIII^−/− mice was performed by i.t. application of Lipo-clod for 3 days and subsequent transfer of CD45.1 AM, followed in individual mice by FACS analysis using F4/80, CD45.1, and CD45.2 staining. One day after reconstitution, the lung Arthus reaction was induced and assessed as described above.

Statistical analysis was performed using the SPSS V. 9.0 statistical package. To analyze differences in mean values between groups, a two-sided unpaired Student’s t test was used.

Nonredundant and codominant roles for C5a/C5aR components of complement and the regulatory FcγR system have been implicated in Arthus reactions triggered by IC of foreign Ags depending on the mouse strain and tissue site of inflammation (30). Among the distinct effector cells involved, mast cells are essential in models of vasculitis and arthritis (31, 32). In contrast, they are less important in alveolitis (13). Because AM are major cells in alveoli, we hypothesized that AM may represent the critical effector cells in lung pathology. In C57BL/6 mice, induction of the lung Arthus reaction by rabbit anti-OVA IgG and OVA Ag results in a robust hemorrhagic response with rapid secretion of cytokines and chemokines, followed by accumulation of PMN in lung tissue, reaching maximal levels of PMN transmigration into alveoli at 8–24 h (33). To assess the role of AM effector cells, we evaluated the effect of their depletion in kinetic studies (at 2–8 h) in mice receiving Cl₂MDP-containing liposomes (Lipo-clod) i.t. (24). This treatment resulted in AM depletion within 3 days compared with control mice receiving Lipo-PBS (Fig. 1,a). IC-induced signs of inflammation specified by PMN accumulation in lung tissue, PMN influx into alveoli, and hemorrhage were all significantly lower in Lipo-clod-treated mice than in Lipo-PBS control animals (Fig. 1,b). Moreover, analysis of BALF from AM-depleted mice revealed markedly reduced levels of MIP-2, TNF-α, and IL-1β (Fig. 1 c).

In line with the kinetic data of macrophage depletion, the roles of FcγRIII and C5aR have recently been shown to correlate with mast cell-independent secretion of such mediators in the lung Arthus response (13, 33). IC-induced responses of enhanced vs diminished hemorrhage and alveolar PMN infiltration were consistently observed at 24 h in FcγRIIB compared with FcγRIII and C5aR mutant mice (14, 33, 34) (data not shown), whereas lack of the high affinity FcγRI had no substantial influence on the inflammatory reaction (WT vs FcγRI knockout: PMN, 20.54 ± 3.92 × 10⁵ vs 17.96 ± 3.83 × 10⁵; n = 8–16; p > 0.6; RBC, 114.59 ± 21.27 × 10⁶ vs 107.20 ± 44.60 × 10⁶; n = 8–16; p > 0.7). Local inhibition of FcγR by sFcR or of C5aR by the neutralizing anti-C5aR Ab 1/36 as well as AM depletion each resulted in markedly suppressed alveolar PMN influx and hemorrhage at 24 h in IC-challenged C57BL/6 mice (Fig. 2,a). BAL-AM cells displayed changes in the mRNA levels of FcγR, but not C5aR at 2 h after IC challenge (Fig. 2,b). These early IC-induced FcγR alterations are C5aR-dependent, because anti-C5aR 1/36 mAb-treated mice showed neither induction of FcγRIII αγ-chains nor suppression of FcγRIIB (Fig. 2,b). Surface FcγR expression was up-regulated by IC in the case of FcγRIII and down-regulated in the case of FcγRIIB, with this opposite regulation being sensitive to C5aR blockade (Fig. 2 c). Collectively, these results establish macrophages as primary effectors that trigger the lung Arthus reaction and, consistent with our previous data in C5aR-deficient mice (14), validate the dichotomous effects of C5aR on inhibitory FcγRIIB and activating FcγRIII in the induction of the inflammatory cascade.

Several studies of the role of complement have indicated a greater importance of C5a than C3 in models of arthritis and different settings of the Arthus reaction (4, 21), which may be explained by a C3 bypass mechanism in the generation of C5a at sites of inflammation (13). Analysis of chemotactic activities of BALF from IC-challenged mice, measured on C5aR⁺ compared with C5aR-null PMN, showed (within 2 h) C5aR-dependent PMN chemotaxis that was completely diminished in Lipo-clod-treated mice, which suggests the requirement for AM cells in early C5a production in the lung (Fig. 2,d). AM-depleted mice displayed reduced, but detectable, C5aR-dependent chemotaxis 4 h after IC challenge (Fig. 2 d), indicating activation of plasma complement as the expected complementary C5a generation pathway. However, such humoral production of C5a was not sufficient to trigger substantial inflammation in the absence of AM.

AM cells are constitutively active in converting extracellular C5 into C5a (35), but whether AM directly link IC formation to C5a production in an autonomous manner remains unknown. In this report, kinetic studies using supernatants from HA-IgG1-stimulated MH-S AM cells demonstrated the presence of mediators with chemotactic activity on PMN from FcγRIII^−/− and WT C57BL/6 mice. This activity was greatly ineffective on PMN from C5aR^−/[minus) mice (Fig. 3,a), the residual migration of C5aR^−/− PMN was found to be MIP-2-dependent (data not shown). These findings indicate that activated AM generate chemotactically active C5a independent of humoral complement activation. In line with this, MH-S AM not only displayed regulated FcγR mRNA expression after a 2- to 4-h period of incubation with rhC5a (5 × 10⁻⁹ M), but also equally responded to HA-IgG1 exposure, showing the same marked reduction in FcγRIIB mRNA and enhanced expression of FcγRIII (Fig. 3,b). We also analyzed whether this HA-IgG1-induced increase in the ratio of FcγRIII to the inhibitory FcγRIIB depends on C5aR signaling and is required for activation of AM. Stimulation of FcγR by IgG1 aggregates, but not of C5aR by rhC5a alone (at 5 × 10⁻⁹ M; data not shown), resulted in increased protein levels of MIP-2 and TNF-α in supernatants of AM cell cultures (Fig. 3,c). Importantly, however, IgG1-induced MIP-2 and TNF-α secretion was completely inhibited by incubation with 100 μg/ml anti-C5aR 1/36 mAb (Fig. 3,c), which was not observed with an isotype control Ab of irrelevant specificity (data not shown). Moreover, treatment of cells with 500 ng/ml PTX for 2 h to functionally inactivate C5aR-associated G_i-type G proteins caused a similar impaired IgG1 activation, as documented by strongly reduced synthesis of MIP-2 and TNF-α (Fig. 3,c). These results together with the observation that treatment with either 1/36 or PTX resulted in blockade of the inverse regulation of FcγRIIB and FcγRIII (Fig. 3 d) suggest a tightly regulated IC responsiveness-encoding pathway in AM that involves autonomous C5a production and activation, and G_i family member-dependent signaling of C5aR, FcγR modulation toward the FcγRIII activation phenotype, and, in turn, chemokine and cytokine production.

Because IC-triggered FcγR activation on AM appears to be C5a:C5aR-G_i-dependent in vitro, we assessed the significance of PTX-sensitive G_i protein activity in pulmonary IC inflammation in vivo. IC-challenged C57BL/6 mice that received PTX showed diminished PMN and RBC levels in alveoli (Fig. 4,a), which indicated that one or more PTX-sensitive G proteins are required for disease induction. A recent analysis demonstrated only Gα_i2 and Gα_i3 in lung tissue homogenates (36). Gα_i2 and Gα_i3 are widely expressed and, with respect to lung, MH-S AM cells coexpress G_i proteins with a stoichiometric excess of Gα_i2 over Gα_i3, as shown by autoradiograms from PTX-catalyzed [³²P]ADP ribosylation and immunoblot analysis (Fig. 4,c). Thus, we considered Gα_i2 to be particularly important for C5aR-dependent effects in immune inflammation. To address this issue, G_i-knockout mice deficient in either Gα_i2 or Gα_i3 were analyzed in the Arthus reaction. Gα_i2/3 double mutants were not included in these experiments due to their embryonic lethality. As illustrated in Fig. 4 d, Gα_i3^−/− mice (n = 6; p > 0.7 compared with C57BL/6 mice) developed normal signs of inflammation at 24 h, whereas Gα_i2^−/− mice exhibited a strong defect in hemorrhage and alveolar PMN influx that was even more pronounced in C5aR mutant mice (Gα_i2^−/− vs C5aR^−/−: PMN, 6.21 ± 2.09 × 10⁵ vs 1.25 ± 0.35 × 10⁵; n = 7–8; p = 0.03; RBC, 50.79 ± 16.46 × 10⁶ vs 11.37 ± 0.29 × 10⁶; n = 7–8; p = 0.03). These data suggest that C5aR-associated G_i2 is the functionally dominant G protein in this process. with some compensatory role of G_i3, especially when G_i2 is absent.

The diminished inflammatory response exhibited by mice treated with PTX or lacking C5aR (Fig. 4) may be explained by impaired C5aR-G_i-dependent modulation of FcγR that is found after C5aR blockade in vivo (Fig. 2) and during AM stimulation in vitro (Fig. 3). In addition, the other more traditional function of C5a as PMN-specific chemoattractant was abolished by preincubation with either PTX or anti-C5aR 1/36 mAb in vitro (Fig. 4,b). Thus, it is not clear from these experiments which one of the two or the combination of C5a:C5aR-G_i-mediated activities is central for disease pathogenesis. To tackle this question, we performed several adoptive cell transfer experiments using CD45.1/2 congenic mice. As an initial strategy, AM cells were first depleted for 3 days with Lipo-clod in congenic CD45.1⁺ C57BL/6 recipient mice, which then were reconstituted with BALF-AM cells from CD45.2⁺ C57BL/6 donor mice, followed by IC challenge on day 4 and evaluation on day 5 (Fig. 5,a). The allotypic CD45 marker served as a tool to follow the course of AM depletion and reconstitution (Fig. 5,b). IC-challenged CD45.1⁺ and CD45.2⁺ C57BL/6 mice showed the expected identical response of alveolar PMN migration and hemorrhage, which was inhibited by AM depletion, but was fully restored through CD45.2⁺ AM engraftment in AM-depleted CD45.1⁺ hosts (Fig. 5,c). PTX and anti-C5aR 1/36 mAb treatments of CD45.2⁺ AM cells before transfer resulted in no disease recovery (Fig. 5 c). Thus, ex vivo-performed functional inactivation of G_i proteins in AM was sufficient for complete disease prevention. This suggests that the C5aR-G_i-triggered pathway in AM is of much greater importance than the C5aR-G_i-dependent chemotactic response of PMN, which remains intact in this adoptive transfer model.

In the next set of experiments, AM cells from FcγR and C5aR knockout mice were used for reconstitution. Recovery of lung pathology through CD45.2⁺ AM in CD45.1⁺ C57BL/6 recipient mice was equally achieved by transfer of FcγRI-deficient AM and was strongly enhanced or completely abolished after reconstitution with CD45.2⁺ AM specifically lacking FcγRIIB, or FcγRIII and C5aR, respectively (Fig. 6,b). Flow cytometric detection of F4/80, CD45.1, and CD45.2 Ags by FACS analysis verified the successful repopulation of CD45.2⁺, F4/80⁺ AM from the various mouse mutants throughout all experiments (Fig. 6,a). Thus, AM with impaired G_i function (see Fig. 5) or lacking C5aR or FcγRIII (Fig. 6) do not longer possess disease-inducing capacity. The results define an absolute requirement for the macrophage C5a:C5aR-G_i-FcγRIIB/FcγRIII axis in the lung Arthus reaction.

Whether PMN can transmigrate into alveoli is probably dependent on their response to the chemotactic factors expressed in lung tissue. The failure of PMN to migrate to the alveolar space in AM-depleted mice could be attributed to a reduced expression of C5a (Fig. 2,d) and CXCR2 ligands (MIP-2; Fig. 1,c). The reported IC-tethering function of FcγRIII also suggests a role of neutrophil FcγRIII for PMN migration (37). However, the finding that defects in C5aR-G_i-FcγRIII on local AM cells were severe enough to cause severely impaired recruitment of otherwise C5aR- and FcγRIII-positive PMN (Figs. 5 and 6) suggests a minor role for C5a:C5aR and IC:FcγRIII signaling on PMN in the Arthus reaction. To further confirm this idea, AM cells from CD45.1 mice were transplanted in WT and C5aR- or FcγRIII-deficient CD45.2 mice (Fig. 7,a). This approach was used to selectively restore the macrophage C5a:C5aR-G_i-FcγRIII axis, but leaving C5aR/FcγRIII on PMN impaired. Transfer of C5aR/FcγRIII^+/+ AM rescued the diminished Arthus reaction in both FcγRIII^−/− and C5aR^−/− mice, as shown by a normalized hemorrhagic response (Fig. 7,b). This rescue effect was also observed for PMN recruitment with normal numbers of migrated FcγRIII^−/− and C5aR^−/− PMN detectable in BALF after C5aR/FcγRIII^+/+ AM reconstitution (Fig. 7,b). Consistent with the FcγRIII^−/− and C5aR^−/− AM transfer studies in C5aR/FcγRIII^+/+ mice (Fig. 6), the scenario of a C5a:C5aR-G_i-FcγR-influenced inflammatory lung microenvironment that induces the migration of PMN to alveoli thus appears dependent on the local macrophage, rather than the PMN side.

The results show that the inflammatory response in the Arthus reaction is initiated by macrophages that use a G_i-dependent signal for the communication of two key players of the immune system, the C5aR and the paired FcγRIIB/FcγRIII. This conclusion is based in part on the observation that AM-depleted mice resisted alveolar IC inflammation and that restoration of susceptibility after AM transfer required the functional presence of C5aR and FcγRIII (but not of FcγRI) on these effector cells. Our finding that FcγRIIB-deficient AM induce a substantially enhanced inflammatory phenotype in this transfer model supports previous work (34), which indicated that the balance between stimulatory FcγRIII and inhibitory FcγRIIB and their coordinate expression is critical in the development of inflammatory lung IC disease. Taking into consideration that the inhibitory FcγRIIB is more abundantly expressed on AM than FcγRIII (14), it is important to strengthen that for an efficient IC activation of macrophages, a change in the FcγRIIB to FcγRIII expression ratio is required. Agonists such as LPS and IFN-γ trigger inverse regulation of FcγR in vitro (26), whereas in IC disease the C5a anaphylatoxin plays a central role (14). Local inhibition of lung C5aR abrogates inflammation by preventing C5a-dependent induction of macrophage FcγRIII and suppression of FcγRIIB. Our adoptive transfer studies demonstrate that functional loss of G_i-coupled C5aR on AM causes unresponsiveness to IC in the lung Arthus reaction without compensation by the other known C5aR-G_i-dependent pathway, that is, C5a-triggered chemotaxis for migration of PMN to sites of acute inflammation.

Earlier work suggested the coupling of C5aR to PTX-sensitive G_i class proteins and Gα₁₅, a PTX-insensitive G_q class α subunit (38). Genetic inactivation of Gα₁₅ in mice revealed functional redundancy of C5a-stimulated G_q class signaling (39), indicating that C5aR may function through both G_q proteins, i.e., Gα₁₅, and one or more of the three distinct G_i1/2/3 isoforms. Our in vitro results show that IC-activated MH-S AM cells have the capacity for plasma complement-independent generation of bioactive C5a, and that a PTX-sensitive C5aR signal is critically required for the modulation of FcγR and IC:FcγR-induced MIP-2 and TNF-α production. Thus, G_i family members play a central role in the cellular C5aR-FcγR cross talk. Furthermore, C5aR- and Gα_i2-deficient, but not Gα_i3-deficient, mice, display a diminished IC response in vivo. These data suggest that the AM effector cell response is largely determined by initial C5aR-dependent G_i2 activation connected to G_i-mediated changes in FcγR expression levels shifting the balance of FcγRIIB/FcγRIII toward FcγRIII activation in immune inflammation.

Although playing a minor role in the Arthus response of the lung (13), mast cells are essential effector cells in hypersensitivity type III (Arthus-like) reactions at several other tissue sites. In skin and synovium, for example, they function as a cellular link between autoantibodies and inflammatory disease of autoimmune vasculitis and arthritis (31, 32). C5aR and FcγRIII are known activators of mast cell function (40, 41) as well as critical effectors in the mast cell-dependent K/B×N model of autoimmune arthritis induced by articular formation of glucose-6-phosphate isomerase/anti-glucose-6-phosphate isomerase IC (4). Thus, future work will address the question of whether C5aR/FcγR communication is also a major cellular event on mast cells, as has been shown in this study for macrophages.

Information about immune effector cells and their functions may impact our understanding of diseases and the design of therapeutic strategies. Different effector cells may exhibit different responses to pathogenic products. Moreover, different receptor systems communicate in inflammatory responses, as shown in this study for C5aR and the regulatory FcγRIIB/FcγRIII system. Based on the opposing FcγR functions, strategies that result in enhanced inhibitory FcγRIIB expression have been proposed as potential new therapeutic approaches for the treatment of autoimmune diseases (8). Therefore, because FcγR are targets for Ab-mediated effects, the characterization of factors and pathways that are capable of controlling inhibitory as well as activating FcγR is an important issue. Our results identify G_i-dependent C5aR signaling in the control of the expression and activation of FcγR on macrophages in vitro. Similar to the in vivo blockade of FcγR, genetic deletion of Gα_i2 and C5aR have beneficial effects in vivo. Targeting of C5aR with an anti-C5aR Ab prevents IC disease through the maintenance of macrophage FcγRIIB at high expression levels and FcγRIII at low expression levels. Previous work suggests different mechanisms in the therapeutic action of i.v. Ig (IVIG) in animal models of immune thrombocytopenic purpura and asthma (42, 43). In the immune thrombocytopenic purpura study, it is reported that administration of IVIG protects against disease in a FcγRIIB-dependent manner. After treatment with IVIG, macrophages show high levels of FcγRIIB, indicating that enhanced signaling of the inhibitory FcγRIIB might be essential for the protective effect of IVIG. In the asthma study, the protective function of IVIG was shown to depend on neutralization of C3a and C5a. In light of the apparent regulatory interactions between C5aR and FcγR, it will be important next to determine whether engagement of an as yet unidentified IVIG receptor (44) or the function of IVIG as a C5a scavenger (43) is responsible for high FcγRIIB levels and, thus, disease protection.

In conclusion, our results imply a revised multimolecular model of the inflammatory cascade (as summarized in Fig. 8). IC deposition represents the initial step that may provoke inflammation by classical and alternative pathways of complement activation, resulting in C3-dependent C5a production (1, 30). Remarkably, macrophages function as an alternative source of C5a, which helps to explain the varying importance of C3 and C5a in IC disease (4, 21, 45). The functional loss of C5aR on macrophages completely abolishes IC inflammation in otherwise C5a-, C5aR-, and FcγR-competent mice, whereas systemic defects of C3, although required for humoral C5a production, are partially tolerated (13). These data indicate the importance of local C5a-induced cellular mechanisms in disease pathogenesis and show that the traditional function of C5a as a PMN chemoattractant is less important than previously suspected. Downstream of AM- and plasma complement-derived C5a, macrophages trigger disease via multiple intracellular steps, one of which involves the C5aR-associated Gα_i2 subunit. It is the G_i-dependent part of the C5aR signaling machinery that is absolutely essential for a switch in the balance of the inhibitory and activating FcγRIIB/FcγRIII pair toward FcγRIII, the previously predicted main initiator of the inflammatory cascade. These findings thus confirm the relevance of the regulatory FcγR system as the primary target of IC-mediated effects and refine our understanding of the initiating mechanism by dissecting two different functions of C5a: chemoattraction of PMN and local control of FcγR, which is dependent on C5aR-associated G_i function with G_i2 as a novel molecular target for drug design in the treatment of inflammation and autoimmune diseases (46, 47). Finally, the effectiveness of ex vivo G_i inhibition by PTX in adoptive cell transfer studies establishes this system as a valuable tool for mutation-based identification of the relevant signaling molecules of the cellular C5aR-Gα_i2-FcγR axis in the context of immune inflammation in vivo.

The authors have no financial conflict of interest.

We thank P. Sondermann, T. Takai, and U. Höpken for providing sFcR-, FcγRIIB-, and C5aR-deficient mice, and A. Braun for critically reading the manuscript.