Fcgamma-receptors induce Mac-1 (CD11b/CD18) mobilization and accumulation in the phagocytic cup for optimal phagocytosis.

Functional interactions between Fcgamma-receptors (FcgammaR) and the beta2 integrin Mac-1 (CD11b/CD18) have been described, but the molecular basis of this relationship remains unclear. Although the glycosylphosphatidylinositol-linked receptor FcgammaRIIIB of human neutrophils is constitutively associated with Mac-1, we found no evidence for direct physical association between Mac-1 and the FcgammaR of mouse macrophages, which are transmembrane proteins. Nevertheless, Mac-1 accumulated in the phagocytic cup following engagement of FcgammaR by IgG-opsonized particles. Blocking the CD18 chains of beta2 integrins by using specific antibodies reduced Mac-1 accumulation in the cup. These antibodies or the addition of the recombinant CD11b I-domain inhibited the ingestion of IgG-opsonized particles. FcgammaR cross-linking stimulated cell adhesion to surfaces coated with Mac-1 ligands and in addition enabled macrophages to bind C3bi-opsonized particles, indicating that FcgammaR-derived signals induce activation of Mac-1. Measurements of fluorescence recovery after photobleaching revealed that whereas most (>80%) of Mac-1 is immobile in resting cells, stimulation of FcgammaR markedly increases the mobile fraction of the integrin. Activation of Mac-1 by FcgammaR required the activity of Src family tyrosine kinases, phosphatidylinositol 3-kinase and phospholipase C, with the release of diacylglycerol and stimulation of protein kinase C. Because elevated cytosolic Ca2+ was not required, we suggest that novel protein kinase C isoforms are involved in Mac-1 activation. These results suggest that FcgammaR stimulation promotes Mac-1 clustering into high avidity complexes in phagocytic cups by releasing the integrin from cytoskeletal constraints and enhancing its lateral diffusion. FcgammaR can enhance host defense by activating Mac-1 (and possibly other integrins), having a synergistic effect on pathogen engulfment and promoting the adherence of phagocytes at sites of infection.

Mac-1, a heterodimeric receptor primarily expressed in neutrophils and monocytes/macrophages, is composed of a specific ␣ chain (CD11b) and the ␤2 chain (CD18) which is common to the other members of the ␤ 2 integrin family (1). As is the case for other integrins, Mac-1 (also known as CD11b/CD18, ␣ M ␤ 2 , Mo-1, or CR3) activation is required for efficient binding to several ligands such as intercellular adhesion molecule 1, C3bi, or fibrinogen. Activation of the cells by specific agonists induces the receptor to undergo conformational changes, mobilization, and clustering by a process known as inside-out signaling (2,3).
A variety of studies have demonstrated that Mac-1 participates in a number of important aspects of the innate immune response, including phagocyte adhesion, migration, and engulfment of complement-opsonized particles (1,4). Such functions are generally triggered by direct binding of ligands like intercellular adhesion molecule 1 and C3 complement fragment C3bi to Mac-1. In addition, Mac-1 interacts with and appears to serve as a signaling partner for glycosylphosphatidylinositollinked receptors such as urokinase-type plasminogen activator receptor and CD14 (5).
Whereas Mac-1 can directly recognize components of the microbial wall (6,7), phagocytosis via this receptor is most efficient when the target particles are coated with complement fragment C3bi, a process known as opsonization. Opsonic phagocytosis can also be mediated by Fc␥-receptors (Fc␥R), 1 which recognize the constant region of IgG bound to antigenic particles (8). Unlike the integrins, however, Fc␥R are capable of binding and responding to their ligands without priming, i.e. stimulation of the cells by other agonists. In vitro studies using single opsonins have demonstrated that engagement of either Fc␥R or Mac-1 suffices to initiate phagocytosis. However, interactions between the two systems have been suggested by various observations. Neutrophils from leukocyte adhesion deficiency syndrome patients, which have mutations in the gene encoding CD18 (9), have a reduced ability to ingest IgG-opsonized red blood cells (IgG-RBC) (10). Moreover, antibodies that block Mac-1 function depressed IgG-mediated phagocytosis without impairing the binding of Fc␥R to its ligands (11). Also, Fc␥R-mediated adhesion of neutrophils to immobilized immune complexes was inhibited by antibodies targeting either CD18 or CD11b (12,13), and Mac-1-deficient mice displayed impaired immune complex-mediated recruitment of neutro-phils onto glomerular basement membrane in vivo or spreading on immune complexes in vitro (14). These studies and that of Coxon et al. (15) suggest that Mac-1 is required for sustained leukocyte adhesion following Fc␥R engagement.
The manner whereby Fc␥R and Mac-1 receptors interact remains largely obscure. One type of Fc␥R, the human Fc␥RIIIB, has been reported to exist in physical association with Mac-1. A carbohydrate-mediated interaction is thought to exist between these receptors, which undergo co-capping in human neutrophils (16,17). However, association with Fc␥RIIIB is unlikely to account for all the observations reported above. The expression of Fc␥RIIIB is restricted to human neutrophils (18,19), yet cooperation between Mac-1 and Fc␥R has been described in cell types that lack this receptor isoform. Moreover, it is unclear how clustering and signaling would be initiated by the complex of quiescent Mac-1 and Fc␥RIIIB, which is a glycosylphosphatidylinositol-linked and therefore signaling-incompetent form of Fc␥R.
In this study we sought to define whether the transmembrane forms of Fc␥R, such as those expressed in murine cells (which lack Fc␥RIIIB), undergo functional interactions with Mac-1. To this end we studied the possible role of Mac-1 in the phagocytosis of IgG-RBC in unprimed RAW264.7 mouse macrophages. In particular, we explored whether direct physical interactions exist between Mac-1 and Fc␥R in these cells, and we considered the possibility that Fc␥R engagement may lead to inside-out activation of Mac-1.

EXPERIMENTAL PROCEDURES
Reagents-Dulbecco's modified Eagle's medium and fetal calf serum (FCS) were from Wisent Inc. BSA (IgG-free, low endotoxin), sheep RBC, and rabbit anti-sheep RBC IgG were obtained from ICN Biomedicals, and anti-sheep RBC IgM was from Accurate Chemicals. cDNA encoding the ␣ M I-domain fused to GST was kindly provided by Dr. E. F. Plow of the Cleveland Clinic Foundation (Cleveland, OH). 2 The purification of the recombinant proteins was performed as described (20). Hamster anti-mouse CD18 and rat anti-mouse CD11b-producing hybridoma, 2E6 and M1/70, respectively, were purchased from the American Type Culture Collection (ATCC), and purified rat IgG2b and rat anti-mouse CD16/CD32 (Fc␥RIII/II) were from Pharmingen. Rat anti-mouse CD11a mAb (M17/5.2) and anti-talin Ab were kind gifts from Drs. H. Ostergaard (Edmonton, Canada) and A. Kupfer (Denver, CO), respectively. F(abЈ) 2 fragments of anti-CD18 or anti-CD11a were prepared from the IgG fraction purified from hybridoma supernatants by using immobilized pepsin, according to the manufacturer's instructions (Pierce). FITC-conjugated anti-CD11b Ab (M1/70) and IgG2b isotype control and anti-hamster IgG were from Serotec and Caltag, respectively. Hamster IgG F(abЈ) 2 and Cy5-, Cy3-, or tritc-conjugated secondary Ab were from Jackson ImmunoResearch Laboratories. PP1, GF-109203X, and calphostin C were from Biomol. BAPTA-AM and Alexa 488-or rhodamine-conjugated phalloidin were from Molecular Probes. U73122 and 1,2-dioctanoyl-sn-glycerol (DiC 8 ) were from Calbiochem. All other reagents were from Sigma.
Aggregated human IgG (agg-IgG) was prepared by heating 10 mg/ml human IgG to 63°C for 20 min followed by spinning at 14,000 rpm for 10 min. The freshly prepared supernatant was used at 250 g/ml final concentration.
Cell Culture and Transfections-RAW264.7 macrophages were cultured in Dulbecco's modified Eagle's medium with 10% heat-inactivated FCS and transferred onto acid-washed and poly-L-lysine (1 g/ml)coated glass coverslips 1 day before the assays. For immobilized immune complex binding assays, RAW264.7 cells grown on glass coverslips were lifted by treating with PBS containing 2 mM EDTA for 15 min at 4°C. Cells were washed with PBS and allowed to recover at room temperature for 5 h in Hepes-buffered RPMI 1640 (HPMI). To study Fc receptor mobility, cells on poly-L-lysine-coated coverslips were transiently transfected with the Fc␥RI-␥-␥-GFP cDNA (21), a gift from Dr. A. D. Schreiber (Philadelphia), using FuGENE 6 (Roche Applied Science), according to the manufacturer's instructions, and used within 24 h of transfection.
Phagocytosis and Binding Determinations-Sheep RBC were opsonized with IgG as described (22). C3bi opsonization was performed by first incubating RBC with sub-agglutinating concentrations of IgM (1:10) in PBS with 0.5 mM CaCl 2 and MgCl 2 for 1 h at room temperature. Excess IgM was then washed off, and RBC were incubated with C5-deficient serum (1:6) for 20 min at 37°C with frequent mixing. C3bi-RBC were then washed and used immediately for binding assays.
To induce IgG-mediated phagocytosis, RAW264.7 cells were exposed to IgG-opsonized RBC (ϳ5 RBC/macrophage) in HPMI with 1% heatinactivated FCS at 37°C for 30 min. Cells were subjected to hypoosmolar shock by adding H 2 O for 20 s followed by three washes with PBS prior to fixation with 4% paraformaldehyde overnight or with methanol at Ϫ20°C for 10 min. For IgG-RBC binding assays, cells were exposed to IgG-opsonized RBC under the same conditions as for phagocytosis except that incubation was for 5 min at 4°C followed by 15 min at room temperature. Cells were washed extensively with PBS prior to fixation with methanol.
For C3bi-RBC binding assays, RAW264.7 cells were serum-starved for 2 h and then incubated with or without 100 nM PMA or 100 M of Blocking Experiments-Cells were washed and incubated for 10 min at room temperature with HPMI containing 5% FCS to block nonspecific binding sites. The cells were next treated for 15 min at room temperature with or without (control) 5 g/ml anti-CD18 (2E6), anti-CD11a (M17/5.2), or non-immune hamster IgG (all in F(abЈ) 2 fragments) with 1% FCS. Next, IgG opsonized RBC were added to the incubation mixture, and binding or phagocytosis experiments were conducted in the presence of the blocking Ab, as described above. To detect CD18 staining, the secondary Ab (TRITC anti-hamster IgG) was preabsorbed with IgG-RBC to eliminate cross-reactivity with opsonized RBC, by incubation for 1 h. The accumulation of CD18 in the cup versus membrane under normal and blocking conditions was determined from confocal images by determining the pixel density in the cup and in inactive plasma membrane regions by using NIH/Scion Image software.
For blocking experiments using the recombinant proteins GST-␣ M I domain or GST, IgG-RBC were preincubated for 5 min with or without 10 M recombinant protein in HPMI containing 0.2% BSA, and this suspension was then added to cells adherent to poly-L-lysine-coated coverslips. Phagocytosis was quantified as described above.
Immunostaining and Flow Cytometry-For internalization experiments cells were preincubated with 1 g/ml F(abЈ) 2 fragments of anti-CD18 (2E6), of non-immune hamster IgG, or of FITC-conjugated anti-CD11b Ab (M1/70), as described above, prior to washing and incubation with or without agg-IgG for 30 min at 37 or 4°C, as specified. For surface staining at 4°C, 0.1% azide was added to the incubation buffers. Cells were washed in PBS, fixed, and incubated with secondary Ab to detect human IgG and CD18.
For flow cytometric analysis cells were scraped from tissue culture flasks, washed, and incubated with agg-IgG in Dulbecco's modified Eagle's medium for 30 min at 37°C (or, in the case of controls, at 4°C in the presence of azide) and washed twice with ice-cold Hanks' buffered saline solution (HBSS) containing 2.5% FCS and 0.1% azide prior to blocking with 5% FCS for 10 min and staining at 4°C for 30 min with FITC-labeled anti-human-IgG, 2E6/FITC anti-hamster, FITC anti-CD11b, or the corresponding control Ab. Live cells excluding propidium iodide were analyzed for surface staining of human IgG or integrin chains with FACSCalibur and CellQuest software (BD Biosciences). To determine the number of integrin molecules per cell, the phycoerythrin (PE) fluorescence quantitation kit (Quantibrite TM PE, BD Biosciences) was used according to the manufacturer's instructions with saturating amounts of PE-anti-CD18 and -anti-CD11b mAb (Caltag Laboratories). The amounts of PE/Ab were determined by generating a standard curve with unconjugated R-PE (Molecular Probes), and the number of Ab molecules per cell were calculated.
To assess affinity changes of ␤ 2 integrins, cells were incubated with an activation epitope-specific mAb (mAb24, a gift from Dr. N. Hogg, London, UK). Cells were treated at 37°C for 15 min using mAb24 or mouse IgG1 (isotype control) in HBSS with 10 mM Hepes and 1% FCS with or without agg-IgG. Parallel incubations were performed in Ca 2ϩ ,Mg 2ϩ -free HBSS, containing 5 mM MgCl 2 , 1 mM EGTA, and 1% FCS. Alexa 488-conjugated F(abЈ) 2 fragments of anti-mouse IgG were used as secondary Ab.
Fluorescence Recovery after Photobleaching (FRAP)-The mobility of integrins and Fc receptors was estimated using FRAP. Cells layered on poly-L-lysine were either transfected with Fc␥RI-␥-␥-GFP cDNA or exposed to 1 g/ml FITC anti-CD11b Ab for 10 min at room temperature. The cells were washed, and the dorsal surface of flat lamellae was imaged immediately. Spots of ϳ2 m in diameter were photobleached using the full power of the 488 nm laser line of the Zeiss LSM 510 confocal microscope, resulting in a 70 -90% reduction of the fluorescence intensity. A similar spot from the area that remained unbleached was selected to serve as a control. Sequential images were acquired after photobleaching with a decreased laser intensity to minimize further photobleaching for a period up to 120 s. The data were exported and analyzed in Microcal Origin 6.0 software. Starting values were set as 100% recovery, and photobleached values were normalized to the unbleached regions at all time points to correct for any bleaching incurred during measurement or for changes in focal plane. The percent recovery (mobile fraction) of the measured population of labeled CD11b Ab or Fc␥R-GFP was determined as the ratio of the final fluorescence to the pre-bleach fluorescence. The half-time to recovery (t1 ⁄2 ) was also quantified. Images of the cells were taken before and after photobleaching and compared to verify that no gross morphological changes or drifting had occurred in the regions of interest.
Immobilized Immune Complex Binding Assays-BSA and BSA/anti-BSA (immune complex)-coated coverslips were prepared as described (14). Briefly, acid-washed coverslips were coated with poly-L-lysine (100 g/ml) and treated with 2.5% glutaraldehyde for 15 min. Coverslips were washed and coated with BSA (1 mg/ml) for 30 min and then blocked with 0.1 M glycine for 2 h. To create immobilized BSA/anti-BSA immune complexes, BSA-coated coverslips were incubated with 40 g of rabbit anti-BSA IgG in PBS for 1 h. RAW264.7 cells suspended in HPMI were plated onto BSA or BSA/ anti-BSA-coated coverslips and incubated for 8 min at 37°C. Cells were fixed with 4% paraformaldehyde and stained with phalloidin and/or primary Ab to integrins or talin, followed by secondary Ab.
Immunofluorescence and Confocal Microscopy-Following treatment, RAW264.7 cells were washed with PBS and fixed with 4% paraformaldehyde in PBS overnight at 4°C or for 20 min at room temperature. Immunostaining was performed by permeabilization with 0.1% Triton X-100 in PBS containing 100 mM glycine for 20 min before blocking for 1 h with 5% FCS in PBS. Staining with anti-integrins Ab and phalloidin was for 1 h at room temperature in PBS containing 1% FCS. Following washing, samples were incubated for 1 h at room temperature with appropriate secondary antibodies. Samples were analyzed by differential interference contrast and fluorescence confocal microscopy using a Zeiss LSM 510 microscope with a 100ϫ oil immersion objective. FITC, Cy5, and Cy3 channels were examined using the conventional laser excitation lines and filter sets.
Statistical Analysis-All data were expressed as mean values Ϯ S.E. Student's two-tailed t tests were performed to assess the significance of differences using InStat software (GraphPad, San Diego).

CD18 Is Required for Optimal IgG-mediated Phagocytosis of RBC-
We initially confirmed the expression and quantified the density of CD18 and CD11b on the surface of RAW264.7 cells by flow cytometry (see "Experimental Procedures"). An average of ϳ3 ϫ 10 5 CD18 molecules/cell of CD18 and 1.6 ϫ 10 5 CD11b molecules/cell was estimated, which is similar to values reported for monocyte-derived macrophages (23). These data indicate that in RAW264.7 cells nearly half of the CD18 chains were associated with CD11b chains to form the Mac-1 integrin receptor.
To assess whether a functional interaction exists between ␤ 2 integrins and transmembrane Fc␥R, we examined the effect of blocking monoclonal antibodies (mAb) against CD18 (24) on the phagocytosis of IgG-RBC by RAW264.7 cells. Fig. 1 shows that pretreatment with F(abЈ) 2 fragments of the blocking mAb 2E6 impaired phagocytosis significantly. The phagocytic index dropped from 287 Ϯ 31 in the controls (n ϭ 9) to 132 Ϯ 4 in antibody-treated cells (n ϭ 4; p Յ 0.001). This effect was specific, because a comparable concentration of non-immune F(abЈ) 2 fragments produced no significant inhibition of phagocytosis (Fig. 1). Inhibition of phagocytosis by anti-CD18 was not due to interference with the binding of the opsonins to the Fc␥R, because measurements of the number of IgG-RBC associated with the phagocytes revealed no difference between control and antibody-treated cells (Fig. 1, solid bars).
Although we had no available F(abЈ) 2 fragments of blocking mAb against the ␣ chain of Mac-1, we speculated that this receptor may have been responsible, at least in part, for the CD18-dependent component of phagocytosis, based on its abundance and on the data of Graham et al. (11), who found that Mac-1 is the major ␤ 2 integrin contributing to Fc-mediated phagocytosis in primary human monocytes and neutrophils. A unique domain of about 200 amino acids of the extracellular moiety of CD11b, known as the I-domain, has been implicated in the binding of Mac-1 to several ligands (20). To verify whether Mac-1 is required for optimal phagocytosis of IgG-RBC, we tested the effect of a recombinant GST fusion protein of the I-domain of CD11b (GST-␣ M I-domain) on phagocytosis. Parallel control experiments were performed by using the GST protein alone. Fig. 1 shows that in the presence of the GST-␣ M I-domain fusion protein, the phagocytic index was reduced by 55% (to 128 Ϯ 27) as compared with untreated controls, whereas preincubation in the presence of GST alone had no effect on phagocytosis. These results confirm that CD11b/CD18 is implicated in Fc␥R-induced phagocytosis, at least in part through the I-domain of CD11b.
In addition to CD11b/CD18, we found that RAW264.7 cells express also CD11a/CD18 (LFA-1). We therefore examined the possible role of this ␤ 2 integrin during IgG-induced phagocytosis. F(abЈ) 2 fragments of blocking mAb against the ␣ chain of LFA-1 did not alter phagocytosis of IgG-RBC (Fig. 1). These results imply that not all ␤ 2 integrin family members participate in IgG-mediated phagocytosis. We therefore focused on CD18 and CD11b during the rest of this study.
Accumulation of CD18 and CD11b at the Phagocytic Cup-To investigate the relationship between Fc␥R and Mac-1, we first studied the distribution of CD18 and CD11b during phagocytosis of IgG-RBC. Fig. 2, A-D, shows that both CD18 and CD11b accumulate in phagocytic cups, identified by the FIG. 1. IgG-mediated phagocytosis, but not binding of IgG-RBC, is dependent on CD18 and CD11b. Red cell phagocytosis (white bars) or binding (black bars) assays were initiated by the addition of IgG-RBC to RAW264.7 cells pretreated without (Cnt, control; GST, ␣ M I-domain;) or with 5 g/ml F(abЈ) 2 fragments of non-immune (NI) hamster IgG, F(abЈ) 2 fragments of mAb against mouse CD18 (2E6), or mAb against mouse CD11a (M17/5.2). In experiments with the recombinant proteins, IgG-RBC were preincubated with 10 M ␣ M Idomain-GST or GST prior to phagocytosis. Preincubation without the recombinant proteins did not affect phagocytosis (not shown). The reaction was allowed to proceed for either 5 min at 4°C followed by 15 min at room temperature to measure binding or for 30 min at 37°C to assess phagocytosis. Binding index or phagocytic index, defined as the number of IgG-RBC bound or ingested, respectively, per 100 cells was determined from 200 to 400 cells counted per condition in 3-4 replicate experiments. **, p Յ 0.001; *, p Ͻ 0.05. presence of adherent IgG-RBC and, more importantly, by the enrichment in F-actin. Fig. 2E displays a typical case of CD18 accumulation in the cup and Fig. 2F illustrates the line-scanning method used to quantify the enrichment in CD18 and CD11b at the cup, in comparison to its density in the contralateral membrane. In 45 determinations, the density of CD18 at the cup consistently exceeded the levels found in other regions of the plasmalemma (62 Ϯ 2% over the overall plasmalemma level). These results are consistent with the phagosomal localization of Mac-1 reported previously (25) in lipopolysaccharideprimed murine peritoneal macrophages during phagocytosis of IgG-zymosan.
Fc␥R and CD11b Chains Do Not Interact Constitutively-Because Fc␥R are known to accumulate at the phagocytic cup, the concomitant accumulation of Mac-1 suggests that the two types of receptors may interact physically. To define whether the two receptor types are constitutively associated, we analyzed whether internalization of Fc␥R induced by soluble IgG complexes drives also the internalization of Mac-1. When added at 4°C, agg-IgG binds to the surface of RAW264.7 cells, which can be readily verified by immunofluorescence (Fig. 3, A and G) or flow cytometry (cf. Fig. 3G, inset) by using an antihuman secondary antibody. At 37°C, the receptor clustering triggered by agg-IgG induces endocytosis of the Fc␥R, which can be visualized by immunofluorescence (Fig. 3, D and J) and quantified by flow cytometry using antibodies to human IgG (cf. Fig. 3J, inset) or Fc␥R (not shown; see Ref. 26 for example). Note that whereas a sizable fraction of Fc␥R becomes internalized under these conditions, the density of Mac-1 at the cell surface remained unaltered. This was apparent by microscopy (cf. Fig. 3, B and H with E and K, respectively) but more accurately established by flow cytometry (cf. Fig. 3, C and I with F and L, respectively). Similar results were obtained when internalization was induced by cross-linking Fc␥R by using sequentially mouse IgG and anti-mouse IgG (not shown).
These data strongly suggest that Mac-1 and Fc␥R are not constitutively associated. Nevertheless, it is conceivable that constitutively associated receptors may undergo dissociation upon endocytosis and that Mac-1 recycles rapidly to the surface. We therefore used an alternative approach to evaluate the interaction of the receptors, which did not require receptor cross-linking. This was accomplished by comparing the lateral mobility of Mac-1 and of Fc␥R by using fluorescence recovery after photobleaching (FRAP). Receptors on the dorsal membrane, near the edge of the cells, were chosen for these measurements to minimize the curvature and thereby ensure that the area under study was within the confocal plane. As depicted in Fig. 4, A-F, and quantified in Fig. 4, G and H, the vast majority of Fc␥R was mobile in the plane of the membrane. Conversely, only a small fraction (19%) of Mac-1 was mobile, and the rate of lateral displacement of these mobile receptors was modestly and yet significantly slower than that of Fc␥R. These results are in agreement with earlier observations: the mobility of Mac-1 had been reported to be limited by its interaction with the cytoskeleton (27), whereas Fc␥R was found earlier not to be restrained by cytoskeletal association (28). Together with the results of Fig. 3, these data indicate that, unlike Fc␥RIIIB, the transmembrane Fc␥R and Mac-1 are not associated to a significant degree.
Ligand Binding by Mac-1 Contributes to Its Phagosomal Accumulation-To investigate whether the ligand binding ability of Mac-1 is required for its concentration at the cup, cells were pretreated with the F(abЈ 2 ) fragments of 2E6, a CD18 blocking mAb prior to incubation with IgG-RBC. Control experiments incubating cells with 2E6 in the absence of IgG-RBC, followed by staining with phalloidin, ensured that the antibodies did not directly affect the cytoskeletal architecture. No changes in F-actin were observed by confocal imaging under these conditions (not shown), discarding the possibility of a direct interference by these antibodies.
In the presence of IgG-RBC, the accumulation of CD18 at the cup was reduced to only 31 Ϯ 0.7% over the plasmalemma level (n ϭ 66). When compared with cells that were not pretreated (62%, see above), the significant drop (p Ͻ 0.0001) in CD18 and likely Mac-1 concentration at the phagocytic cup by mAb 2E6, which inhibits their ligand binding capacity (24), suggests that accumulation is due in part to binding of ligands present in the region of particle engulfment. Such ligands could be located on the opsonized RBC or could be part of the exofacial aspect of cellular components that accumulate at the sites of phagocytosis.
CD11b/CD18 Is Activated through Fc␥R-induced Signals-In the resting, unstimulated state, unbound integrins are anchored to cytoskeletal elements and therefore immobile in the plane of the membrane. The finding that Mac-1 accumulates at the cup by lateral displacement and interaction with ligands suggests that activation of at least a fraction of the integrin must have occurred upon engagement of Fc␥R. To test this hypothesis we examined whether Fc␥R-derived signals modulate the avidity and/or affinity of Mac-1 for its ligands. To evaluate the activation of Mac-1, we quantified its ability to (a) induce the formation of focal contacts and (b) bind C3bi.
It was shown previously that murine neutrophils adhere to surfaces coated with immune complexes by a mechanism involving Mac-1 (14). We confirmed these observations. When plated on surfaces coated with only BSA, unstimulated RAW264.7 cells do not form substantive contacts as detected by diffuse integrin and F-actin staining (Fig. 5A). When plated on immune complexes formed by adding anti-BSA antibodies to BSA-coated surfaces (Fig. 5B), the cells displayed prominent F-actin-rich pericellular rings and dots that likely represent focal contacts, because they stained positively for Mac-1 (Fig.  5B, inset) as well as talin (not shown).
The experiment in Fig. 5C provides further evidence that Fc␥R stimulation activates integrins and, in addition, demonstrates that such stimulation does not require direct interaction between these molecules. In these experiments cells were plated on BSA, and Fc␥R were stimulated with soluble agg-IgG. As above, stimulation induced the formation of multiple focal contacts, even though the adherent surface lacked Fc␥R ligands and exposed only BSA, a known ligand of Mac-1 (29). Soluble agg-IgG itself does not induce focal contact formations, as judged by CD11b/CD18 staining in Fig. 3. Note that addition of soluble agg-IgG leads to Fc␥R internalization (e.g. Fig. 3), whereas Mac-1 clustered on the adherent BSA-coated surface, implying an indirect relationship between the receptors. Activated Mac-1 can also mediate the binding and internalization of C3bi-coated particles. We therefore measured the binding of C3bi-opsonized RBC to RAW264.7 cells to assess the state of activation of Mac-1. As shown in Fig. 5D, binding of C3bi-coated particles to unstimulated macrophages was marginal. As reported earlier (30), inside-out priming of Mac-1 by pretreatment with phorbol esters unmasked the ability of Mac-1 to bind C3bi-opsonized RBC. Remarkably, stimulation of cells with agg-IgG also enabled the cells to bind C3bi-coated particles (Fig. 5D). We ascertained that binding of C3bi-RBC by agg-IgG-stimulated cells occurred via Mac-1, because it was inhibited by preincubating the cells with 1 g/ml of the mAb M1/70, a Mac-1-specific blocking mAb (31,32). In view of the fact that agg-IgG does not induce up-regulation of Mac-1 expression in RAW264.7 cells (see Fig. 3), these results demonstrate that stimulation of Fc␥R suffices for inside-out activation of Mac-1.
Mechanism of Mac-1 Activation by Fc␥R-Integrin activation can result from increased affinity or avidity for its ligands, or from a combination of both (33). We initially probed whether the affinity of Mac-1 was increased upon Fc␥R engagement. This was tested by measuring the appearance of an epitope on the ␣ chains of human ␤ 2 integrins, including Mac-1, that is recognized by mAb-24 only after the integrin undergoes the conformational change associated with activation (34). Because the conformation-sensitive mAb does not recognize murine integrins, we used the human monocytic U937 line for these experiments. As reported (35), flow cytometric analysis revealed a marked increase in the availability of the epitope recognized by mAb-24 on the surface of U937 cells in the presence of high Mg 2ϩ /EGTA (ratio of MFI of treated over control ϭ 5.2 Ϯ 0.6, see Fig. 6, A and B), indicative of increased affinity. In contrast, no significant change in mAb-24 binding was observed in cells stimulated with agg-IgG (ratio of MFI of treated over control ϭ 1.05 Ϯ 0.05, see Fig. 6, A and C). These results suggest that changes in the affinity of CD11b are not the primary mechanism underlying Mac-1 activation following engagement of Fc␥R.
Even when affinity is not changed, binding to targets can occur through clustering of integrins, which increases the avidity for ligand. Avidity increase in ␤ 2 integrins can result from enhanced mobility of the receptors along the plasma membrane after their release from the cytoskeletal constraints, which allows receptor clustering (3,36). We therefore inquired whether the mobility of Mac-1 is increased by stimulation of Fc␥R. The mobility of CD11b chains was measured in resting and stimulated cells using FRAP, as described above, and the results are presented in Table I. In unstimulated cells, only a small fraction of CD11b (19%) was found to be mobile during the course of our experiments. As a positive control, we confirmed earlier observations (37) that activation of PKC by phorbol esters released a sizable fraction of the integrin from its constraints, increasing the mobile fraction to nearly 60%. More importantly, we also found that stimulation of the cells with agg-IgG produced a marked enhancement in the fraction of mobile CD11b to 48% (Table I). We therefore concluded that by releasing the cytoskeletal constraints that limit the mobility of Mac-1 (3,36), Fc␥R stimulation promotes the formation of high avidity interactions between the integrin and its ligands, which are clustered on the surface of phagocytic particles.
Fc␥R-induced Signals Leading to Activation of CD11b/ CD18 -Because Fc␥R were found earlier not to associate directly with Mac-1, the increased mobility of the latter must be transduced by signaling intermediates. To define the nature of the transduction pathway, we measured the effects of a variety of agonists and inhibitors with well defined specificity on the ability of Fc␥R to promote binding of C3bi-opsonized RBC to RAW264.7 cells. The results of these studies are summarized in Fig. 7. Consistent with previous reports (38), stimulation with exogenous diacylglycerol (DiC 8 ) sufficed to enable the cells to bind C3bi-opsonized particles. The extent of binding was similar to that elicited by Fc␥R cross-linking, suggesting that they may share a common pathway, namely activation of PKC. This notion was tested by using two inhibitors, calphostin C and GF-109203X, that block PKC by distinct mechanisms. As illustrated in Fig. 7, both inhibitors precluded the activation of Mac-1 by Fc␥R, confirming a role for PKC in the process.
Stimulation of Mac-1 by DiC 8 and block of the Fc␥R effect by calphostin C, a diacylglycerol antagonist, suggest that the effects are mediated by conventional and/or novel isoforms of PKC. The source of endogenous diacylglycerol is likely to be the hydrolysis of phospholipids by PLC. Accordingly, we found that inhibition of PLC by U73122 virtually eliminated the activation of Mac-1 by Fc␥R (Fig. 7). The specificity of the PLC inhibitor was verified by addition of exogenous diacylglycerol, which bypassed the block induced by U73122. Although the phospholipase D-phosphatidic acid pathway may also contribute to the elevation of intracellular diacylglycerol levels in activated RAW264.7 cells (39), these data strongly suggest that PLC is a major contributor to diacylglycerol production during Fc␥R-induced Mac-1 activation. To distinguish between conventional and novel isoforms of PKC, we studied the effect of intracellular calcium chelation. Preloading the cells with BAPTA by incubation with the precursor acetoxymethyl ester, under conditions that effectively buffer thapsigargin-induced intracellular calcium changes (not shown), had little effect on the binding of C3bi-opsonized particles. Because conventional PKC isoforms are more sensitive to calcium omission than their novel counterparts, we believe that the latter played a predominant role in the stimulation of Mac-1 by Fc␥R.
Fc␥R-induced signaling depends on the activation of Src family and Syk tyrosine kinases, which in turn target several effector molecules including PLC␥ (40, 41) and phosphatidyl- The cells were then fixed, permeabilized, and stained for F-actin (main panels and upper insets) and CD11b (lower insets). Confocal images were acquired at the base of the cells. Sites of F-actin accumulation, including pericellular rings and focal contacts, are marked by arrows. Note sites of co-localization of CD11b with F-actin accumulation at focal contacts (arrows in insets). Bars ϭ 5 m. D, RAW264.7 cells were exposed to C3bi-opsonized RBC for 20 min in the absence (control) or presence of 100 nM PMA or 250 g/ml agg-IgG. The binding index (see legend to Fig. 1) was calculated from at least 100 cells from three experiments. *, p Ͻ 0.001. inositol 3-kinase (PI3K) (42). We tested the involvement of tyrosine kinases in Mac-1 activation by using the potent inhibitor PP1, which targets preferentially Src family kinases. As illustrated in Fig. 7, Fc␥R-induced Mac-1 activation was abolished in the presence of this inhibitor. The inhibitor had no effect when Mac-1 was activated directly by PMA (not shown), confirming the prediction that PKC-dependent Mac-1 activation is downstream of Src family kinase activation. Finally, we tested the role of PI3K, which was described to contribute to Mac-1-mediated spreading of neutrophils on immune complexes (12). The Fc␥R-induced binding of C3bi-RBC was found to be obliterated in cells pretreated with the PI3K inhibitor LY294002 (Fig. 7). This observation implies that PI3K is required either upstream or in addition to the activation of PLC. DISCUSSION It was recently reported that, in spreading human neutrophils, integrins can propagate their own activation by inducing an influx of calcium and the activation of calpain (43). The protease was proposed to mobilize the integrins by cleavage of their anchorage sites to the cytoskeleton. This mechanism does not seem to account for the ligation of C3bi-opsonized particles by Mac-1 in RAW264.7 cells, because chelation of intracellular calcium was without effect. Moreover, it is well established that the fraction of active Mac-1 in unstimulated macrophages is insufficient to elicit measurable responses and that a priming stimulus is required for their activation (44,45). As detailed in this paper, cross-linking of Fc␥R can provide such a stimulus.
Cooperation between Fc␥R and integrins had been suggested by several observations, including the reduced ability of leukocyte adhesion deficiency neutrophils to take up IgG-opsonized particles (10) and the observation that antibodies that block Mac-1 function depressed IgG-mediated phagocytosis without impairing the binding of Fc␥R to its ligands (11). In the case of human neutrophils, these effects could be attributed to the known direct association between Mac-1 and Fc␥RIIIB. However, the latter receptor is restricted to human neutrophils and cannot explain the cooperation observed between integrins and Fc␥R in mice (14). Two classes of phagocytic Fc␥R are expressed in mouse cells, Fc␥RI and Fc␥RIIIA, which are both transmembrane proteins. Because no mAbs are available against murine Fc␥RI, we assessed the contribution of Fc␥RIIIA in Fc␥R-induced Mac-1 activation by using a blocking mAb. The mAb used (anti-mouse CD16/CD32, 10 g/ml) recognizes also Fc␥RII, but the latter is not involved in the activation of phagocytosis in mice. The mAb inhibited the binding of C3bi-RBC induced by agg-IgG by about 57% but did not affect PMA-induced C3bi-RBC binding, as expected. This observation suggests that both Fc␥RIIIA and Fc␥RI are involved in this process.
In view of our findings, we propose a possible mechanism to account for the activation of Mac-1, and possibly other integrins, following the engagement of Fc␥R. Briefly, we suggest that cross-linking of Fc␥R leads to the activation of Src family FIG. 6. Fc␥R stimulation does not induce the expression of an ␣ chain activation-specific epitope. U937 cells were incubated at 37°C for 15 min with mAb24 which recognizes an epitope exposed on the ␣ chains of human ␤ 2 integrins by an activation-induced conformational change. Incubation was in the absence (A) or presence of 250 g/ml agg-IgG (C). B illustrates cells incubated at 37°C in the presence of Mg 2ϩ /EGTA, conditions known to induce the conformational changes associated with activated integrins. Cells were analyzed by flow cytometry after staining with Alexa 488-conjugated F(abЈ) 2 fragments of anti-mouse IgG. Isotype-matched mAb (mouse IgG1) were used as controls (not shown).  kinases and stimulation of PLC␥, which have been independently established in various laboratories (for review see Ref. 42). The release of diacylglycerol from phosphoinositides would then activate PKC, which would in turn release the integrins from their cytoskeletal anchorage. Novel isoforms of PKC are likely to be prominent in this event, inasmuch as diacylglycerol, but not calcium elevation, was required for Mac-1 activation. Indeed, novel PKC isoforms have been attributed a prominent role in phagocytosis (46). PKC may alter cytoskeletal structure via phosphorylation of F-actin-bundling substrates like macrophage-enriched myristoylated alanine-rich C kinase substrate (MacMARCKs), which is involved in enhancing the mobility of ␤ 2 integrins (37). Consistent with this notion, myristoylated alanine-rich C kinase substrate is enriched in the phagosomes of PMA-treated peritoneal macrophages during phagocytosis of IgG beads (47). By having acquired lateral mobility as a result of cytoskeletal detachment, the integrins could then diffuse to sites of ligand accumulation, forming high avidity complexes. Whereas PKC or other mediators may also enhance the affinity of the integrins for their ligands, we failed to detect such an increase by using the conformation-sensitive mAb-24. A previous report revealed the emergence of a new epitope associated with Mac-1 activated by Fc␥R in human neutrophils. The mAb used, CBRM1/5, recognizes a conformational change in the I-domain of Mac-1 that reflects increased affinity to ligand (12). Because our data are consistent with agg-IgG-induced changes in the avidity rather than affinity of Mac-1, it is possible that Fc␥R-mediated activation of Mac-1 occurs through different mechanisms depending on the Fc␥R type and/or the extent of Fc␥R clustering (33). Also, unlike human neutrophils (12,13), RAW264.7 macrophages do not up-regulate Mac-1 expression in response to agg-IgG (see Fig.  3). Therefore, increased lateral mobility appears to be the primary source of enhanced Mac-1 activity. A schematic version of our model is presented in Fig. 8.
It was reported previously (12) that inhibition of PI3K activity prevents ␤ 2 integrin-mediated adhesion of neutrophils to immobilized immune complexes and the expression of the activation epitope of Mac-1 by immune complexes. In addition, LFA-1 clustering induced through chemokine receptors was found to be dependent on PI3K activity (33). Our studies also suggested that PI3K needs to be active for successful activation of Mac-1 by Fc␥R. It is possible that PI3K is necessary for the activation of PLC␥. This lipase can be effectively stimulated by Tek family kinases like Btk, which require 3Ј-phosphorylated phosphoinositides for activation (48). Alternatively, products of PI3K may cooperate with PKC in the activation of the integrins, possibly by affecting the function of GTPases of the Rho and/or Rap families (30,49,50).
In an earlier study using human monocytes, Brown et al. (51) proposed the existence of two distinct subpopulations of Mac-1. They suggested that ϳ60% of the surface Mac-1 is mobile and relocates to the ventral side of the cells as they adhere to C3bi-coated surfaces. The remaining 40% was proposed to be immobile, and this fraction was postulated to partake in phagocytosis. The much larger fraction of mobile Mac-1 found in monocytes (60%) compared with RAW264.7 cells (19%) may be characteristic of suspended cells or may reflect differences in the species or differentiation state of the cells. What was not implicit in the study by Brown et al. (51) is how the immobile Mac-1 contributed to the phagocytic process. To the extent that integrins are optimally activated upon cross-linking, lateral mobilization would be required for them to contribute effectively to particle engulfment. Our study provides a putative mechanism for the mobilization of this immobile fraction, not engaged by adherence to the substrate.
Proteins such as talin, ␣-actinin, and paxillin, which are typically recruited to the membrane by integrins, are found lining the phagocytic cup formed during ingestion of IgG-opsonized particles (47). This seemingly paradoxical observation can be explained if integrins are mobilized by active Fc␥R and diffuse to cluster at the cup. The nature of the ligands that attract integrins to cups formed by IgG-RBC is not clear. However, our results suggest that ligand binding involves the I-domain of the CD11b chain. Based on the cooperativity between the lectin and I-domains of integrins described previously (6,52), it is possible that Mac-1 ligands also include surface carbohydrates of the opsonized-RBC. Coexistence of Fc␥R and Mac-1 ligands is fully expected to occur on the surface of naturally opsonized particles. Thus, the initial engagement of Fc␥R would serve to mobilize the integrins, Resting macrophages contain freely mobile Fc␥R, whereas Mac-1 is largely bound and immobilized to the actin cytoskeletal matrix (A). Upon addition of IgG-coated particles, Fc␥R cluster and induce the activation of Src family tyrosine kinases, PI3K and PLC␥ (B). Localized diacylglycerol (DAG) production activates novel PKC (nPKC) that initiates cytoskeletal rearrangement leading to Mac-1 release from its cytoskeletal constraints (B). Mobile integrins then move to the sites of phagocytosis where they can interact with ligands, re-associate with the cytoskeleton, and contribute to the ingestion of particles (C). which would ligate appropriate targets on the particle surface, including complement fragments and carbohydrate moieties of the invading microorganisms (52). The combined effect of the two receptor types would thereby fortify the phagocytic response.