Alternative Endocytic Pathway for Immunoglobulin A Fc Receptors (CD89) Depends on the Lack of FcRγ Association and Protects against Degradation of Bound Ligand*

IgA is the most abundant immunoglobulin in mucosal areas but is only the second most common antibody isotype in serum because it is catabolized faster than IgG. IgA exists in monomeric and polymeric forms that function through receptors expressed on effector cells. Here, we show that IgA Fc receptor(s) (FcαR) are expressed with or without the γ chain on monocytes and neutrophils. γ-less FcαR represent a significant fraction of surface FcαR molecules even on cells overexpressing the γ chain. The FcαR-γ2 association is up-regulated by phorbol esters and interferon-γ. To characterize γ-less FcαR functionally, we generated mast cell transfectants expressing wild-type human FcαR or a receptor with a point mutation (Arg → Leu at position 209) which was unable to associate with the γ chain. Mutant γ-less FcαR bound monomeric and polymeric human IgA1 or IgA2 but failed to induce exocytosis after receptor clustering. The two types of transfectant showed similar kinetics of FcαR-mediated endocytosis; however, the endocytosis pathways of the two types of receptor differed. Whereas mutant FcαR were localized mainly in early endosomes, those containing FcαR-γ2 were found in endo-lysosomal compartments. Mutant γ-less FcαR recycled the internalized IgA toward the cell surface and protected against IgA degradation. Cells expressing the two forms of FcαR, associated or unassociated with γ chains, may thus have differential functions either by degrading IgA antibody complexes or by recycling serum IgA.

In humans, IgA is found in the systemic and mucosal compartments; it is the second most common antibody class in blood and the major immunoglobulin at mucosal surfaces (1,2). More IgA is produced daily than all of the other immunoglobulin classes together (3). In serum, IgA is mainly monomeric and has a half-life around five times shorter than that of IgG because of its fast catabolism (2,4). Although the implications of secretory IgA in host defenses are well established (2), much less is known about the antibody-mediated functions of serum IgA in human blood. Serum IgA has been considered an antiinflammatory isotype capable of inhibiting several functions mediated by other isotypes including inhibition of IgG phagocytosis, bactericidal activity, oxidative burst, and cytokine release (5)(6)(7)(8)(9)(10). The molecular basis of these inhibitory functions is poorly understood; however, IgA-immune complexes can trigger effector cells after aggregation of IgA Fc receptor(s) (Fc␣R, 1 CD89), resulting in various immune effector functions such as phagocytosis, oxidative burst, and cytokine release (11)(12)(13).
Fc␣R are expressed on myeloid cells as heterogeneously glycosylated type I transmembrane proteins that can bind both IgA1 and IgA2 isotypes at the boundary between the C␣2 and C␣3 domains (14 -18). Polymeric IgA binds more efficiently to Fc␣R than does monomeric IgA (19,20). Fc␣R exist as at least two isoforms (a.1 and a.2) differing by a deletion in their extracellular domains and expressed alternatively on monocytes and alveolar macrophages (21). Several other splice variants, the corresponding native proteins of which have not been identified, have also been reported (21)(22)(23)(24)(25). Fc␣R are associated with the disulfide-linked FcR ␥ chain homodimer (26 -28). This interaction is resistant to treatment with Nonidet P-40 detergent, which contrasts with the dissociation of ␥ chains from Fc⑀RI or Fc␥RI in certain detergents (26,29,30). This strong interaction can be explained by the presence of two oppositely charged residues (Arg ϩ /Asp Ϫ ) in the transmembrane domain of the Fc␣R and ␥ chain, respectively (28). The ␥ chain contains a common immunoreceptor tyrosine-based activation motif in its cytoplasmic tail. Recently, it has been shown that signaling through Fc␣R-␥ 2 involves several tyrosine kinases including lyn, syk, and Btk (31,32). Recruitment and phosphorylation of syk and Btk were modulated by stimulation with interferon-␥ (IFN-␥) and/or phorbol esters, indicating that activation of tyrosine kinases through Fc␣R depends on the priming state of the cell (32).
FcR without signaling motifs in their cytoplasmic tails are associated with specialized subunits, such as ␥ or ␤ chains, and depend on their specific retention motifs to be fully expressed on the cell surface (33). In the absence of the ␥ chain they are degraded rapidly in the endoplasmic reticulum as in the case of Fc␥RIII (34). One remarkable feature of Fc␣R is that these receptors can be expressed fully at the surface of COS cells after transfection, without the signaling ␥ subunits (16). Despite the role of the ␥ chain in downstream Fc␣R signaling (28), we wondered whether the Fc␣R could exist and function as receptors when unassociated with the ␥ chain (␥-less Fc␣R) on myeloid cells. We have identified significant amounts of ␥-less Fc␣R in several cell types, including monocytes, neutrophils, and transfected cells overexpressing the ␥ chain. ␥-less Fc␣R and Fc␣R-␥2 are expressed on the same cells, and this constitutes the basis for differential endocytosis pathways of IgA, in which ␥-less receptors recycle IgA toward the cell surface whereas Fc␣R-␥2 undergo endo-lysosomal sorting for IgA degradation.
Cells-The human monocytic cell line U937 was maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mM glutamine, 100 IU/ml penicillin, and 100 mg/ml streptomycin. Polymorphonuclear cells and mononuclear cells were isolated from whole blood by Ficoll-Hypaque (Amersham Pharmacia Biotech) gradient centrifugation. Granulocytes were purified from red cell pellets by dextran sedimentation as described previously (15). Enriched monocyte populations (60 -80% pure) were obtained by subjecting mononuclear cells to rosette formation with 2-aminoethylisothiouronium bromide-treated SRBC, and nonrosetting cells were submitted to plastic adherence as described in Ref. 19 (40) were transfected with human Fc␣R and/or human ␥ chain and were maintained in DMEM (Life Technologies, Inc.) supplemented with 10% fetal calf serum, 2 mM glutamine, 1.5 g/ml puromycin (Sigma) and/or 1.0 mg/ml of G418 (Life Technologies, Inc.).
Constructs, Expression Vectors, and Transfection-Human Fc␣R containing the R209L mutation was constructed by amplifying base pairs 591-891 of a previously described Fc␣R a.1 cDNA (21). The sense primer included the BanII restriction site at position 601 and introduced the R209L mutation (at base pairs 688 -690, with CTG replacing CGC). The amplified fragment was ligated to the remainder of the NH 2 -terminal cDNA via the BanII restriction site. The construct was checked by sequencing as described in Ref. 21. The Fc␣R a.1, Fc␣R (R209L), and human ␥ chain (41) were subcloned into pSR␣Neo (kindly provided by Dr. J. Di Santo, INSERM U429) a modified version of the pcDL-SR␣ promoter-based expression vector (42). RBL-2H3 cells were first transfected with 30 g of DNA by electroporation at 250 V and 1,050 microfarads using an Easyjet ϩ apparatus (Eurogenetec, Seraing, Belgium), then grown under 1 mg/ml G418 selection; resistant clones were selected for Fc␣R expression by means of flow cytometry. One Fc␣R-expressing clone was chosen and cotransfected with pSR␣Neohuman ␥-chain (30 g) and pSR␣-Puro (4 g) (43).
Cell Iodination, Immunoprecipitation, and Immunoblotting-Cell surface iodination with Na 125 I (1 mCi; Amersham Pharmacia Biotech) was carried out by the lactoperoxidase method (44). For immunoprecipitation of Fc␣R, cells (10 7 /ml) were lysed for 30 min at 4°C in PBS containing 1% digitonin (Aldrich), 0.02% sodium azide, 1% aprotinin, 1 mM diisopropylfluorophosphate, 5 mM iodoacetamide, and 1 mM phenylmethylsulfonyl fluoride. After centrifugation at 14,000 ϫ g for 30 min to remove insoluble materials, cleared lysates were immunodepleted of Fc␥R by using human IgG, 32.2 and IV.3 mAb, and precipitated with test mAb as described previously (35). Bound materials were treated or not treated with N-glycanase (Genzyme), and samples were subsequently prepared for SDS-PAGE (45). For immunoblotting, immunoprecipitated proteins were separated by SDS-PAGE and transferred electrophoretically to a nitrocellulose Hybond-C (Amersham Pharmacia Biotech) filter for 18 -20 h (46). The blots were incubated in blocking buffer composed of 25 mM Tris-HCl, pH 7.4, 137 mM NaCl, 2.7 mM KCl (TBS) containing 3% bovine serum albumin and 0.1% Tween 20 and then incubated with anti-␥ (1:500) for 2 h at room temperature. Horseradish peroxidase-conjugated goat anti-rabbit IgG was used a secondary Ab. Filters were developed using the Enhanced Chemiluminescence detection system (ECL; Amersham Pharmacia Biotech).
Coimmunoprecipitation of Receptor-bound 125 I-Anti-Fc␣R mAb-This was carried out as described in Ref. 47. Briefly, F(abЈ) 2 fragments of A77 mAb or mouse IgE were labeled with Na 125 I using the IODO-GEN method (48). Cells (5 ϫ 10 6 ) were incubated with 125 I-labeled test mAb (25-35 g/ml) for 1 h, washed in PBS and 0.1% NaN 3 , and then lysed in 0.5 ml of 1% digitonin buffer containing protease inhibitors as described above. After centrifugation, lysates were divided into two aliquots for 2-h incubations with either 20 g of 4D8 anti-␥ chain mAb or 50 g of RAM Ig Ab coupled to Sepharose 4B. These amounts of antibodies had been identified as saturating concentrations for precipitation of labeled Ab complexes. The percentage of specifically precipitated counts was calculated for each Ab after subtraction of nonspecific counts obtained using either irrelevant IgG1-coupled beads or lysates that had been preincubated with a 100-fold excess of unlabeled anti-Fc␣R mAb (nonspecific counts were always Ͻ3%). For RBL transfectants, 3 l of polyclonal rabbit anti-␥ chain antiserum plus protein A-coupled beads were used to coprecipitate both rat and human ␥ chains (49). Normal rabbit serum was used to determine background precipitation.
FIG. 1. Lack of colocalization between ␥ chain and some Fc␣R within intracellular vesicles after short term endocytosis. PMA-treated U937 cells preincubated with human IgG (10 mg/ml) to block Fc␥R were incubated with anti-Fc␣R A77 F(abЈ) 2 fragments plus GAM coupled to FITC on ice as described under "Experimental Procedures." After Fc␣R staining, cells were incubated at 37°C for 3 min, fixed, permeabilized, and stained with anti-␥ chain polyclonal Ab plus GAR coupled to Texas Red. Cells were observed under a confocal microscope and optically sectioned at 1.5-m intervals. A representative medial section of the horizontal slices is shown. No staining was observed when FITC-labeled secondary Ab or irrelevant IgG1 was used.
Immunofluorescence and Flow Cytometry-RBL transfectants (1 ϫ 10 6 ) were stained with 10 l of biotinylated A77 F(abЈ) 2 or irrelevant IgG1 F(abЈ) 2 fragments (at 0.1 mg/ml) for 30 min at 4°C followed by 10 l of 1/50 diluted streptavidin PE (Southern Biotechnology Associates) as developing reagent. For IgA binding, cells preincubated with human IgG (10 mg/ml) to block Fc␥R were incubated with 10 l of biotinylated purified IgA (0.5 mg/ml) for 1 h at 4°C followed by streptavidin PE. For two-color immunofluorescence analysis, viable U937 cells (2 ϫ 10 6 ), preincubated with an excess of human IgG (10 mg/ml) to block Fc␥R, were stained directly with 10 l of PE-labeled A59 anti-Fc␣R mAb (0.1 mg/ml) or with an irrelevant PE-labeled IgG1 control for 30 min at 4°C. After washing, cells were fixed with PBS containing 1% paraformaldehyde, permeabilized with PBS containing digitonin (10 g/ml) for 5 min at 4°C, and stained with anti-␥ chain rabbit antiserum (10 l at 1:100 dilution) or a control rabbit serum for 30 min at 4°C in PBS containing 0.05% Tween 20. After washes, cells were incubated with 10 l of FITC-labeled goat anti-rabbit antibodies (25 g/ml; human-and mouseadsorbed, purchased from Southern Biotechnology Associates) for 30 min at 4°C and analyzed by flow cytometry using a FACScalibur apparatus (Becton Dickinson). In some experiments, cytoplasmic molecules were evaluated on cells cytospun onto glass slides, fixed, and permeabilized for 20 min at Ϫ20°C with 95% ethanol and 5% acetic acid solution, washed, and incubated for 20 min with 4D8, A59, or control IgG1 mAb (0.05 mg/ml). FITC-labeled anti-mouse Ig Ab (0.05 mg/ml) was added as a developing reagent and mounted on coverslips.
␤-Hexosaminidase Assay-This was based on a method described previously (43). Briefly, transfectants were plated at 5 ϫ 10 4 cells in 100 l of complete DMEM in the absence of G418 and sensitized with anti-dinitrophenyl IgE Abs (1/200) or F(abЈ) 2 fragments of A77 mAb (0.01 mg/ml) for 4 h at 37°C. Cells were washed in Hanks' balanced saline solution containing 1% fetal calf serum and resuspended in the same buffer containing 100 ng/ml dinitrophenyl-human serum albumin (Sigma) or F(abЈ) 2 fragment RAM (40 g/ml), respectively. To determine spontaneous release, cells were incubated in the absence of Ag (for Fc⑀RI stimulation) or with irrelevant IgG1 F(abЈ) 2 fragments (for Fc␣R stimulation). Maximal release was determined with 100 nM PMA plus 1 M ionomycin as stimulant. After incubation for 1.5 h, hexosaminidase

FIG. 2. Identification of ␥-less Fc␣R on human myeloid cells.
2 ϫ 10 7 blood neutrophils (panel A) and PMA-activated U937 cells (panel B) were surface labeled with Na 125 I, and the membrane proteins were solubilized using a 1% digitonin lysis buffer as described under "Experimental Procedures." Lysates were divided into three aliquots and incubated with irrelevant IgG1 (lanes 1 and 4), anti-Fc␣R mAbs (lanes 2 and 5; A, A59 and B, A77 F(abЈ) 2 ), or anti-␥ Abs (lanes 3 and 6; A, 4D8 and B, rabbit antiserum) plus RAM Ig Ab (panel A) or protein G (panel B) coupled to Sepharose 4B beads. Eight immunoadsorptions were performed with an excess of monoclonal (panel A) or polyclonal anti-␥ chain (panel B) and followed by immunoprecipitations with test Abs. In panel A, immunoprecipitates were digested or not digested by N-glycanase, as indicated (N-gly) and analyzed by 10% SDS-PAGE (2MEϩ) and autoradiography. In panel B, immunoprecipitated 125 Isurface proteins were separated by 12.5% SDS-PAGE (2MEϪ), transferred onto a nitrocellulose membrane, and analyzed by autoradiography (top) and immunoblotting (bottom) using anti-␥ chain polyclonal Ab and horseradish peroxidase-conjugated anti-rabbit Ig Ab plus ECL.

FIG. 3.
A. ␥-less Fc␣R represents the major fraction of Fc␣R molecules on the cell surface. 5 ϫ 10 6 of U937 cells, monocytes, and neutrophils (empty, hatched, and filled bars, respectively) were incubated with 125 I-labeled A77 anti-Fc␣R F(abЈ) 2 fragments (25-35 g/ml) for 1 h at 4°C, washed, and solubilized in 1% digitonin lysis buffer. The lysates were divided into two aliquots and incubated for 2 h at 4°C with either RAM Ig Ab or 4D8 mAb anti-␥ chain coupled to Sepharose 4B. After washes, precipitated counts versus total counts were determined. Nonspecific precipitated counts were obtained using either irrelevant IgG1-coupled beads or lysates that had been preincubated with a 100fold excess of unlabeled anti-Fc␣R. Bars (mean Ϯ S.D. of at least three experiments performed in triplicate) show the calculated percentage of specifically precipitated 125 I-labeled A77 bound to cell surface receptors. Panel B, time-dependent stability of Fc␣R-␥ association in the mild detergent, digitonin. PMA-activated U937 cells were incubated with 125 I-labeled A77 anti-Fc␣R F(abЈ) 2 fragments for 1 h at 4°C, washed, lysed (time 0) for 15 min on ice in 1% digitonin lysis buffer, and centrifuged for 15 min at 14,000 ϫ g. Lysates were then incubated at 4°C for different time periods and immunoprecipitated by 4D8 anti-␥ chain mAb coupled to beads for 2 h at 4°C and analyzed as described in panel A.
secretion was analyzed in test supernatants by adding p-nitrophenyl N-acetyl-␤-D-glucosamine (1.3 mg/ml Sigma). The total cellular content of ␤-hexosaminidase was determined by lysis of adherent cells in 0.5% Triton X-100. Absorbance was determined at 410 nm in a microplate reader.
Internalization and Recycling Assays-This was performed as described elsewhere (50). Briefly, 1 ϫ 10 6 cells were incubated with 1 g of 125 I-F(abЈ) 2 fragments of A77 anti-Fc␣R mAb for 1 h at 4°C. After extensive washing, 10 l of F(abЈ) 2 fragments of rabbit anti-mouse antibodies (1 mg/ml) was added for 30 min. Excess antibody was removed, and endocytosis was induced by incubating cells at 37°C in RPMI 1640, 25 mM HEPES, 5% fetal calf serum for the times indicated. The reaction was stopped by placing the cells on ice. Any residual antibodies on the surface were removed by acid stripping (PBS, pH 2.5, at 4°C for 5 min). This acid treatment routinely removes 85-90% of surface-bound anti-Fc␣R F(abЈ) 2 . After pelleting, cell-associated counts were detected in a gamma counter. In recycling experiments the cells were incubated with 125 I-polymeric IgA1 or 125 I-Fab fragments of A77 alone for 1 h on ice, washed, and then either treated or not treated for 20 min with 0.6 mM primaquine (Sigma) before incubation at 37°C. Nonspecific counts were obtained by preincubating cells with a 100-fold excess of nonlabeled mAb or IgA. Data are expressed as percentages of total initial cell-associated counts and presented as the means Ϯ S.D. of at least three separate experiments.
Measurement of IgA Proteolysis after Internalization-RBL transfectants were plated in the absence of G418 at 0.5 ϫ 10 6 cells/ml in 24-well Costar tissue culture plates. 24 h later, the cells were incubated with biotinylated, 125 I-labeled dimeric IgA1 (1 g/well) in 0.2 ml of 0.1% bovine serum albumin, DMEM at 4°C for 1 h. The medium was removed after 1 h, and the cells were washed three times at 4°C. Cells were then incubated with streptavidin-labeled PE (10 g/ml) at 4°C for 15 min to induce receptor aggregation. After washings, cells were cultured in DMEM containing 0.1% bovine serum albumin and 100 g/ml unlabeled IgA1 for the times indicated. After incubation, the medium was removed, proteins were precipitated in 10% trichloroacetic acid, and acid-soluble and acid-insoluble radioactivities were counted in a gamma counter as described in Ref. 51.
Endocytosis Procedure by Confocal Microscopy-Adherent cells on glass slides were incubated at 4°C for 30 min with 100 l of 0.1 mg/ml A77 anti-Fc␣R F(abЈ) 2 fragments in PBS and 0.2% bovine serum albumin. After washings, cells were incubated with 100 l of 0.04 mg/ml RAM or FITC-coupled GAM for 30 min on ice. When indicated, cells carrying unlabeled antibodies were incubated further with 0.005 mg/ml GAR coupled to FITC to amplify aggregation. To visualize transferrin receptor-recycling vesicles, cells were cultured in serum-free DMEM for 30 min to deplete endogenous transferrin and incubated on ice with 100 nM human transferrin coupled to Cy3 (kindly provided by Dr. A. Benmerah, CJF-97-10, Necker Institute) together with anti-Fc␣R mAb as above. The slides were either warmed to 37°C for various times or kept at 4°C. Cells were washed, fixed in 3% paraformaldehyde for 10 min, and quenched twice in PBS containing 1 M glycine. For intracellular ␥ chain staining, cells were permeabilized with 0.05% saponin (Sigma) and stained with 3.5 g/ml purified anti-␥ chain polyclonal Ab plus 3.5 g/ml GAR coupled to Texas Red. To visualize the plasma membranes, cells were stained after endocytosis for 5 min at 4°C with 10 g/ml of wheat germ agglutinin coupled to Texas Red (38). After washing, the slides were mounted in 10% Moviol, 25% glycerol, Tris-HCl (100 mM, pH 8.5). Confocal laser scanning microscopy was carried out with a TCS4D confocal microscope based on a DM microscope interfaced with an argon/krypton laser. Simultaneous double fluorescence acquisitions were made with the 488 nm and 568 nm laser lines to excite FITC and Texas Red dyes using a 100 ϫ oil-immersion Plan Apo objective (numerical aperture 1.4). The fluorescence was selected using appropriate double-fluorescence dichroic mirror and band-pass filters (52).

Identification of ␥-less Fc␣R on Human Myeloid Cells-Be-
cause Fc␣R can be fully expressed at the surface of COS cells after transfection without the signaling ␥ subunits (16), we investigated whether all Fc␣R expressed on PMA-treated U937 cells were associated with the ␥ chain homodimer by means of confocal microscopy. Because ␥-less and ␥-associated Fc␣R could not initially be distinguished on the cell surface, we performed short term endocytosis of Fc␣R⅐anti-Fc␣R mAb complexes (FITC-labeled) to examine whether all internalized receptors colocalized with the ␥ chain (Texas Red-labeled) in the vesicles. As shown in Fig. 1, two types of intracellular vesicles were detected in single cells, in which Fc␣R was either colocalized (yellow) or not colocalized (green) with the ␥ chain. These results strongly suggest the existence of ␥-less and ␥-associated  (32) were cotransfected with ␥ chain and selected on the basis of the reactivity with anti-human ␥ chain 4D8 mAb. 10 7 cells were then solubilized in 0.5% Nonidet P-40 lysis buffer and immunoprecipitated using anti-human (hu) ␥ chain mAb (4D8)-coupled Sepharose 4B beads. Precipitated proteins were separated by 12.5% SDS-PAGE under nonreducing conditions and analyzed by immunoblotting using rabbit anti-␥ chain polyclonal Ab (poly) and horseradish peroxidase-conjugated anti-rabbit Ig Ab. One out of five clones was chosen (␣␥). PMAtreated U937 cells were used as positive control. Panel B, Fc␣R expression. 5 ϫ 10 5 nontransfected (NT), human Fc␣R-transfected (␣) and human Fc␣R/␥ transfected (␣␥) cells were stained with biotinylated A77 anti-Fc␣R mAb (solid lines) or biotinylated irrelevant IgG1 (dotted line, for ␣␥) and with streptavidin PE followed by FACS analysis. Panel C, partial association of Fc␣R with the ␥ chain on transfectants. Because human Fc␣R molecules were previously found to be associated with rat ␥ chain on transfectants (32), experiments were performed using a polyclonal Ab that recognizes both rat and human ␥ chains. 5 ϫ 10 6 cells were incubated with 125 I-labeled A77 anti-Fc␣R F(abЈ) 2 fragments and digitonin solubilized. The lysates were divided into two aliquots and incubated for 2 h at 4°C with either RAM Ig or polyclonal anti-␥ chain plus protein A-coupled beads, and the percentage of precipitated surface receptors was determined as described in Fig. 3A. Maximum percentages of precipitable 125 I-Ab-receptor complexes using anti-mouse Ig Abs exceeded 80%.

Fc␣R in the same cells.
To investigate whether the ␥-less receptor population could also be detected in detergent extracts, we performed immunodepletion experiments using digitonin-solubilized cells because the Fc␣R-␥2 interaction is resistant to digitonin treatment (26). As shown in Fig. 2, immunoprecipitation of surfaceiodinated Fc␣R from blood neutrophils and PMA-treated U937 cells resulted in the appearance of the expected broad band of 55-75 kDa. Precipitation with an anti-␥ chain antibody gave rise to similarly sized species. Treatment of anti-␥ chain mAbassociated glycoproteins with N-glycanase resulted in 32-and 36-kDa bands that comigrated with those observed with the anti-Fc␣R mAb ( Fig. 2A). After extensive immunodepletion with anti-␥ chain mAb, anti-Fc␣R mAb-reactive molecules could still be precipitated ( Fig. 2A). Similar results were obtained with PMA-treated U937 cells using a polyclonal anti-␥ chain Ab (Fig. 2B), in which the same treatment eliminated most of the ␥ chain molecules (Fig. 2B, bottom). Conversely, immunoadsorptions with an anti-Fc␣R mAb (A59) completely eliminated 55-75 kDa-4D8 mAb reactive proteins (not shown). These results reveal two forms of Fc␣R, ␥-less and ␥-associated that are expressed on the surface of U937 cells and blood neutrophils. Because of the extensive immunoadsorption it was, however, impossible to quantify the two forms of Fc␣R.
To estimate the amounts of ␥-less Fc␣R we used a coimmunoprecipitation assay validated previously for Fc⑀RI (47). Cells were loaded with 125 I-labeled anti-Fc␣R mAb F(abЈ) 2 fragments. Labeled receptor-Ab complexes were solubilized in the presence of 1% digitonin and precipitated with either anti-␥ chain (4D8)-or anti-mouse Ig Ab. As shown in Fig. 3A, whereas total precipitable amounts of 125 I-Ab⅐Fc␣R complexes using anti-mouse Ig Abs exceeded 70% (total Fc␣R), those precipitated with the anti-FcR␥ mAb were significantly lower on U937 cells, monocytes, and neutrophils (16 Ϯ 5%; 21 Ϯ 3%, and 29 Ϯ 5%, respectively). The results were almost similar when 125 Ilabeled Fab fragments of anti-Fc␣R were used instead of F(abЈ) 2 fragments in two comparative experiments (17 Ϯ 0.5% versus 23.5 Ϯ 1.5% in monocytes and 26 Ϯ 3% versus 32 Ϯ 1% in neutrophils, respectively). Similar results were also obtained using a rabbit anti-␥ chain antiserum (data not shown). We then analyzed the time-dependent dissociation of the ␥ chain from the Fc␣R in digitonin lysates. Fc␣R-␥ complexes were FIG. 5. Fc␣R and ␥ chain are mostly expressed in the same cells. Viable U937 cells preincubated with an excess of human IgG to block Fc␥R were stained directly with PE-labeled A59 anti-Fc␣R mAb or with an irrelevant PE-labeled IgG1 control. After washes, digitonin-permeabilized cells were stained with anti-␥ chain rabbit antiserum or a control rabbit serum and with FITC-labeled goat antirabbit antibodies as a developing reagent, as described under "Experimental Procedures." Two-color immunofluorescence analysis was then carried out by flow cytometry. The values inside the boxes represent the percentage of cells. a Untreated or treated U937 cells (5 ϫ 10 6 ) were loaded with 125 Ilabeled A77 anti-Fc␣R mAb F(abЈ) 2 (25-35 g/ml) for 1 h at 4°C. After solubilization in 1% digitonin lysis buffer, counts specifically precipitated with 4D8 anti-␥ chain mAb or anti-mouse IgG Ab were determined as described under "Experimental Procedures." For total Fc␣R, the expression index (mean Ϯ S.E. of at least three experiments performed in triplicate) was calculated independently for each experiment by the ratio of cpm of treated cells to cpm of untreated cells. Total precipitated Fc␣R binding sites in untreated cells was 6,102 Ϯ 788 (determined from seven different experiments). For Fc␣R-␥2 complexes, the expression index of the percentage of ␥ chain association with Fc␣R was calculated independently for each experiment and expressed as the ratio of treated to untreated cells.
b Values indicate modulation of total Fc␣R-␥2 complexes. c Statistically significant differences (p Ͻ 0.05, Mann-Whitney test).
very stable over a 24-h period, ruling out the possibility that partial association of Fc␣R with the ␥ chain was the result of their instability in detergents (Fig. 3B). Partial Association of Fc␣R with the ␥ Chain Is Not Dependent on the Amounts of Expressed ␥ Chain-To examine the effect of the amounts of expressed ␥ chain, we transfected a previously established Fc␣R ϩ RBL transfectant (32) with the human ␥ chain. One clone (␣␥) expressing large amounts of human ␥ chain was selected (Fig. 4, A and B). Coimmunopre-cipitation experiments, using a polyclonal Ab that recognizes both rat and human ␥ chains, showed that in these transfectants the fraction of ␥-associated receptors did not increase compared with cells transfected with Fc␣R only (Fig. 4C). As a control we coprecipitated the Fc⑀RI using the same anti-␥ Ab. In agreement with previous observations (47), a major fraction of Fc⑀RI (Ͼ50%) was coprecipitated in these transfectants (data not shown). This demonstrated that the amount of anti-␥ Ab used in the assay was not limiting. These results also indicated that association of the ␥ chain with Fc␣R did not depend on the amount of expressed ␥ chain.
Fc␣R and ␥ Chain Are Coexpressed in a Single Cell Population-The experiments shown in Fig. 1 suggested that both ␥-less and ␥-associated Fc␣R are expressed on the same cells.
To further exclude the possibility that partial association was caused by heterogeneity in ␥ chain expression on a given cell population, we carried out two-color FACS analysis of cell surface Fc␣R and intracellular ␥ chain. Fig. 5 shows the expression of both Fc␣R and ␥ chain on U937 cells, before and after IFN-␥ or PMA treatment. The majority of cells were Fc␣R ϩ /␥ ϩ . Even though the expression level of both molecules was heterogeneous, the absence of two independent contour plots ruled out the existence of a subpopulation expressing only one of these proteins. The coexpression of Fc␣R and ␥ chain was also found in neutrophils as determined by cytoplasmic stainings (not shown). (15,17,38,53,54), we investigated whether they also affected expression of Fc␣R-␥2 complexes as determined by coimmunoprecipitation. U937 cells were cultured with IL-1␤, IFN-␥, GM-CSF, PMA, or ionomycin for 18 h. Table I shows that Fc␣R expression on the cell surface was enhanced significantly by PMA or GM-CSF, whereas ionomycin diminished receptor expression by about half. Interestingly, despite the lack of effect on Fc␣R surface expression, IFN-␥ promoted a significant increase (about 1.5-fold) in ␥ chain association with Fc␣R. PMA also significantly favored ␥ chain association with Fc␣R (about 1.8-fold). Treatment with IL-1␤, GM-CSF, or ionomycin had no significant effect on ␥ chain association with Fc␣R.

Modulation of the ␥ Chain Association with Fc␣R on U937 Cells by Phorbol Esters and IFN-␥-Because a variety of agents have been described to modulate surface expression of Fc␣R
␥-less Fc␣R Binds IgA but Does Not Induce Exocytosis-We first established stable transfectants expressing either a wildtype or a mutant (R209L) Fc␣R by using the Fc␣R-negative rat mast cell line RBL-2H3. As shown in Fig. 6A, the selected transfectants expressed similar levels of Fc␣R. In contrast to the wild-type, the mutant receptor did not associate with en-FIG. 6. Characterization of Fc␣R transfectants. Panel A, expression of wild-type (WT) or mutant (R209L) Fc␣R on RBL transfectants. 5 ϫ 10 5 cells were stained with biotinylated A77 anti-Fc␣R mAb (solid lines) or with biotinylated irrelevant IgG1 (dotted lines) and with streptavidin PE as described in Fig. 4. Panel B, absence of ␥ chain association with R209L Fc␣R.Nontransfected (NT) RBL cells and transfectants expressing wild-type or R209L Fc␣R (10 7 ) were solubilized in 1% digitonin lysis buffer and immunoprecipitated using A77 anti-Fc␣R F(abЈ) 2 fragments coupled to Sepharose 4B beads. Precipitated proteins were separated by 12.5% SDS-PAGE in nonreducing conditions and analyzed by immunoblotting using rabbit anti-␥chain polyclonal Ab, horseradish peroxidase-conjugated anti-rabbit Ig Ab, and ECL as described in the legend of Fig. 2. Similar results were obtained with two other clones. a Cells (0.5 ϫ 10 6 ) preincubated with 100 g of human IgG to block Fc␥R were stained with 5 g of biotinylated IgA plus streptavidin PE. Biotinylated anti-Fc␣R F(abЈ) 2 of A77 mAb was used to allow approximate estimation of receptor numbers on transfected cells. Biotinylated human monomeric IgG was a negative control. IgA2 was composed of 80% monomers and 20% polymers. dogenous ␥ chains of RBL cells (Fig. 6B). Both types of receptor specifically bound monomeric and polymeric IgA1 or IgA2 molecules, as this binding was inhibited by My43 anti-Fc␣R mAb (Table II). However, ␥-less Fc␣R bound more IgA than wildtype Fc␣R, despite their similar levels of Fc␣R expression (evaluated using mAb A77). The capacity of wild-type and mutant (R209L) receptors to mediate downstream events was examined by measuring the capacity of cells to degranulate in response to receptor stimulation. As a control the releasing capability of each individual transfectant was tested by stimulating cells through Fc⑀RI. Maximal release was obtained with PMA and Ca 2ϩ ionophore. As shown in Fig. 7A, activation through mutant receptors did not lead to significant release of the granular enzyme, ␤-hexosaminidase, whereas the response to stimulation through wild-type Fc␣R was comparable to that induced by Fc⑀RI. ␥-less Fc␣R⅐IgA Complexes Are Rapidly Endocytosed and Recycled to the Cell Surface-We next examined endocytosis as a second function for both types of Fc␣R. As shown in Fig. 7B, wild-type and mutant receptors internalized immune com-plexes at similar rates and amounts, indicating a potential endocytic function of ␥-less Fc␣R molecules. Analysis of endocytosis by means of confocal microscopy revealed numerous intracellular vesicles containing Fc␣R in transfectants expressing ␥-less or ␥-associated Fc␣R (Fig. 8). However, close inspection revealed a marked difference in the localization of intracellular endocytic vesicles between the mutant and wild-type Fc␣R transfectants. The internalized mutant (R209L) Fc␣R was localized very close to the periphery, whereas wild-type receptors were also found in vesicles deeper inside the cell. Colocalization experiments revealed that Fc␣R was partially found within recycling vesicles that stained positively for transferrin receptors in both types of transfectants (Fig. 9). Recycling was also suggested by flow cytometry experiments in which high amounts of receptor complexes were still detectable on the cell surface even after endocytosis for 90 min (Fig. 10A). R209L-Fc␣R mutant transfectants had significantly more anti-Fc␣R mAb-receptor complexes on the cell surface than wildtype transfectants (exceeding 50% of initial fluorescence intensity values), suggesting a preferential role of ␥-less receptors in recycling. To demonstrate receptor recycling, we took advantage of the recycling inhibitor primaquine. This drug blocks endocytic recycling vesicles from reaching the cell surface, thus accumulating the internalized ligand inside the cell, as described for recycling of internalized monomeric IgG by Fc␥RI (50). 125 I-Polymeric IgA was bound to transfected Fc␣R molecules and allowed to internalize for various periods in the presence or absence of primaquine. As shown in Fig. 10B, primaquine-induced accumulation of internalized 125 I-polymeric IgA was significantly higher in R209L-Fc␣R ϩ mutant transfectants than in cells expressing wild-type receptors, indicating that ␥-less Fc␣R recycles IgA toward the cell surface.
␥-less Fc␣R Protects IgA from Degradation-As it has recently been shown that ␥ chains mediate endocytic trafficking to lysosomes and are important for ligand degradation and antigen presentation (55), we examined the ability of wild-type and ␥-less Fc␣R to sort for IgA degradation. Biotinylated and iodinated dimeric IgA was bound to Fc␣R on cells, followed by cross-linking using streptavidin PE to induce internalization. IgA proteolysis was monitored by determining the fraction of trichloroacetic acid-soluble radioactive counts in the supernatant between 30 and 120 min after endocytosis induction. As shown in Fig. 10C, no degradation of dimeric IgA1 was observed in R209L transfectants, whereas time-dependent IgA1 degradation was measured in wild-type Fc␣R containing ␥-associated receptors. No IgA degradation was detected in the absence of cross-linking in both types of transfectants (data not shown).

DISCUSSION
In this study, we report the existence of both ␥-associated and ␥-less surface Fc␣R on blood monocytes and neutrophils as well as on U937 cells. This is demonstrated by three different technical approaches, which included confocal microscopy after Fc␣R endocytosis, SDS-PAGE analysis of Fc␣R immunodepleted in ␥ chains, and finally by coimmunoprecipitation assay of Fc␣R using anti-␥ Ab. Results of this last assay suggest that ␥-less Fc␣R represent a significant fraction of surface Fc␣R molecules. The majority of cells express the two types of Fc␣R.
Although the positively charged arginine residue at position 209 of the Fc␣R transmembrane domain is critical for the interaction with the ␥ chain (Ref. 28 and our results), the mechanism underlying and regulating the partial association of Fc␣R with the ␥ chain is unknown. Genomic cloning has revealed a single gene encoding Fc␣R (56) anti-Fc␣R mAb F(abЈ) 2 followed by RAM F(abЈ) 2 or mouse IgE plus dinitrophenyl-human serum albumin, as described under "Experimental Procedures." Secreted ␤-hexosaminidase was analyzed in the supernatants. Maximal release was obtained after incubation of cells with PMA plus ionomycin. Similar results were obtained with two other clones. Panel B, kinetics of Fc␣R-mediated endocytosis are independent of their association with ␥ chains. RBL transfectants expressing either wild-type (closed circles) or R209L (open circles) Fc␣R were loaded with 125 I-A77 anti-Fc␣R mAb F(abЈ) 2 at 4°C for 1 h, washed, and incubated for 30 min with RAM F(abЈ) 2 fragments. Cells were warmed rapidly to 37°C for the indicated time periods followed by acid treatment at 4°C to remove cell surface-bound mAbs. Non-acid-releasable counts were determined and expressed as percentage of total initial cell-associated counts and presented as the mean Ϯ S.D. from at least three separate experiments. partial association of Fc␣R and the ␥ chain cannot be explained by the presence of a structurally different Fc␣R protein. This is further supported by experiments showing partial ␥ association in RBL cells transfected with the Fc␣R cDNA. It is also unlikely that intracellular amounts of ␥ chains are limiting for Fc␣R-␥2 expression because the amount of coprecipitated receptors with anti-␥ was unchanged even when human ␥ chains were overexpressed. Rather, our data suggest that ␥ chain association with Fc␣R may be a regulated process susceptible to modulation by a variety of agents. In this context, it is interesting to note that IFN-␥ favored ␥ chain association with Fc␣R independently of surface expression of the corresponding ␣ chains, whereas phorbol esters increased both the amount of total Fc␣R and the percentage of Fc␣R-␥2. GM-CSF enhanced total Fc␣R but not Fc␣R-␥2. Furthermore, recruitment of the tyrosine kinases syk and Btk after Fc␣R activation is modulated by these agents (32) and may be a consequence of amounts of Fc␣R-␥2 complexes. Our results do not rule out the presence of another uncharacterized chain that would compete with the ␥ chain for association with the arginine residue in the Fc␣R transmembrane domain. Taken together these results indicate that the formation of multimeric Fc␣R is independent of the amounts of ␥ chains expressed and can be regulated by environmental factors that could be of physiologic relevance at inflammatory sites.
To examine the functional role of ␥-less Fc␣R we established transfectants expressing ␥-less Fc␣R (R209L mutants) or both types of receptor using the mast cell line RBL-2H3. We found that mutant ␥-less Fc␣R bound IgA as efficiently, or even better, than wild-type receptors that contained Fc␣R-␥2. This seems to be different from Fc␥RI and III where coexpression with ␥ chain enhances ligand affinity (57). Our results confirm that ␥ chain association with Fc␣R is not essential for IgA binding (16,20). After cross-linking, cells expressing ␥-less Fc␣R failed to release the granular marker ␤-hexosaminidase after Fc␣R aggregation, suggesting that ␥ chains play a key role in Fc␣R-mediated signaling pathways leading to exocytosis. A role for IgA in eosinophil degranulation has been demonstrated previously (58). Our study also corroborates previous observations on B cell transfectants expressing R209L Fc␣R in which downstream signals such as Ca 2ϩ mobilization and IL-2 release were absent (28).
Although ␥-less Fc␣R were unable to mediate downstream signaling, we found that both wild-type and mutant ␥-less FIG. 8. Intracellular localization of internalized ␥-associated and ␥-less Fc␣R after 60-min endocytosis. Adherent RBL cells (wild-type (WT) and R209L mutant) were incubated successively for 30 min on ice with F(abЈ) 2 of the anti-Fc␣R A77 mAb, with F(abЈ) 2 fragments of RAM and with GAR coupled to FITC before incubation of the cells at 37°C for 60 min as described under "Experimental Procedures." Cells were finally incubated or not with wheat germ agglutinin coupled to Texas Red (WGA) to delimit the plasma membrane. Cells were observed under a confocal microscope and optically sectioned at 1.5-m intervals. A representative medial section of the horizontal slices is shown. No staining was observed when FITC-labeled secondary Ab or irrelevant IgG1 F(abЈ) 2 was used.

FIG. 9. Colocalization of Fc␣R and transferrin receptors (TfR) in intracellular vesicles after 15-min endocytosis.
Adherent RBL cells (wild-type (WT) and R209L mutant) were first cultured in serum-free DMEM for 30 min to deplete endogenous transferrin and then successively incubated with anti-Fc␣R A77 mAb, RAM, and GAR coupled to FITC on ice as described in Fig. 8. After Fc␣R staining, cells were incubated with 100 nM human transferrin coupled to Cy3 before incubation of the cells at 37°C for 15 min. Cells were observed under a confocal microscope and optically sectioned at 1.5-m intervals. A representative medial section of the horizontal slices is shown. No staining was observed when FITC-labeled secondary Ab or irrelevant IgG1 F(abЈ) 2 was used.
Fc␣R were able to endocytose after receptor clustering. Mutant ␥-less Fc␣R was as efficient as wild-type receptors for internalization. Thus, endocytosis mediated by mutant ␥-less Fc␣R does not depend on the presence of tyrosine-based motifs in the cytoplasmic tail. Indeed, this has also been shown for other FcR lacking the ␥ chain as is the case of Fc␥RI and Fc␥RIIb2 that mediate endocytosis of immune complexes (59 -61). Our results point to major differences in endocytic pathways between these two forms of Fc␣R. In particular, internalized ␥-less Fc␣R were only localized close to the periphery, whereas internalized wildtype Fc␣R (containing ␥-less and ␥-associated receptors) underwent deeper compartmentalization, suggesting ␥ chain sorting for the endo-lysosomal pathway. Fc␣R endo-lysosomal compartmentalization has been demonstrated previously on blood monocytes by their colocalization with cathepsin D (38). Furthermore, a role for ␥ chains in mediating endocytic sorting to lysosomes that leads to antigen presentation has been demonstrated recently for Fc␥R (55). Therefore, we focused on the characterization of biological functions mediated by ␥-less Fc␣R. Intracellular vesicles containing ␥-less Fc␣R colocalized with those containing transferrin receptors, suggesting that they were involved in recycling of Fc␣R and its bound ligand. Further evidence for the recycling of IgA by mutant ␥-less Fc␣R was provided by the effects of primaquine, an inhibitor of receptor recycling. The significant increase in internalized polymeric IgA by transfectants expressing mutant ␥-less Fc␣R treated by primaquine is strongly indicative of Fc␣R-ligand recycling. Reflux of IgA toward the cell surface has been observed in blood monocytes from patients with alcoholic liver cirrhosis who have increased levels of serum IgA (38). Finally, our results indicate that mutant ␥-less Fc␣R are unable to sort for IgA degradation even when cells are cultured for 2 h with cross-linked dimeric IgA, whereas cells expressing ␥-associated Fc␣R degraded bound IgA.
Taken together, these findings point to the existence of myeloid cells expressing two types of Fc␣R with or without ␥ chains. They differ in the type of endocytic pathway used for the internalized ligand, which led us to propose the existence of an alternative mechanism that protects IgA from degradation. Because IgA bound to wild-type Fc␣R is not degraded without cross-linking, the use of each pathway may depend on the degree of aggregation of Fc␣R molecules. Cross-linking by IgAimmune complexes could thus increase numbers of Fc␣R-␥2 complexes delivering signals for downstream events such as degradation of IgA-immune complexes, processing, and antigen presentation. In agreement with this proposal, a previous study has shown that only large sized macromolecular IgA are efficiently and rapidly cleared from the circulation in humans, whereas clearance of smaller sized IgA polymers is considerably slower (62).
The protective role of ␥-less Fc␣R may be important in view of maintaining serum IgA concentrations that would certainly counterbalance the rapid catabolism of IgA through other receptors such as the hepatocyte asialoglycoprotein receptor, which interacts with IgA through their carbohydrates (63,64). Simultaneous expression of ␥-associated and ␥-less Fc␣R might thus increase cellular flexibility in carrying out alternative functions, independently, which either mediate IgA-antigen processing and presentation to major histocompatibility complex molecules or recycle the IgA monomer continuously to achieve serum IgA homeostasis. Wild type (open circles) and R209L mutant (closed circles) Fc␣R transfectants were stained with A77 anti-Fc␣R F(abЈ) 2 fragments and then with RAM F(abЈ) 2 fragments before incubation of cells at 37°C for the times indicated. Cells were then surface stained with PE-labeled F(abЈ) 2 fragments of goat anti-rabbit IgG Ab and analyzed by FACS. Results were calculated as follows 100 Ϫ [(100 ϫ (x of A77 mAb incubated at 37°C Ϫ x of negative control incubated at 37°C)/(x of A77 mAb incubated at 0°C Ϫ x of negative control incubated at 0°C) in which x is the computer mean fluorescence intensity value of each FACS profile. * p Ͻ 0.05 in Student's t test. Panel B, increased intracellular accumulation of ␥-less Fc␣R⅐IgA complexes in the presence of primaquine, a recycling inhibitor. Wild-type (open bars) and R209L mutated (closed bars) Fc␣R transfectants were loaded with 125 I-polymeric IgA1 at 4°C for 1 h and then warmed rapidly to 37°C in the presence or absence of 0.6 mM primaquine. Non-acid-releasable counts were determined at the time points indicated, calculated as a percentage of total initial cell-associated counts, and presented as the ratio of primaquine-treated to untreated cells. Results are expressed as the mean Ϯ S.D. of three separate experiments. * p Ͻ 0.03 in Student's t test. Panel C, degradation of IgA bound to Fc␣R. Cells expressing wild-type (open bars) or R209L mutant (closed bars) plated in triplicate for each time point were preincubated in serum-free medium for 30 min at 37°C followed by incubation with 125 I-labeled IgA for 1 h at 4°C with streptavidin PE. Internalization of aggregated receptors was induced by incubating the cells at 37°C. After the indicated times, trichloroacetic acid (TCA)-soluble radioactivity released into the medium was determined as described under "Experimental Procedures." Each experimental point is expressed as a percentage of the total radioactivity recovered. Results of three experiments are presented as the mean Ϯ S.D.