CD66b Regulates Adhesion and Activation of Human Eosinophils¹

The eosinophil is recognized as a proinflammatory granulocyte implicated in various allergic diseases, such as bronchial asthma and atopic dermatitis, and in immunity to parasites (1). Current evidence suggests that the pathophysiology of these diseases and conditions is closely related to proinflammatory mediators, such as eosinophil major basic protein, eosinophil cationic protein and reactive oxygen species, released when eosinophils are exposed to appropriate stimuli in the tissues (1). Several in vitro studies suggest that eosinophils express receptors for IgA, IgG, cytokines, chemokines, and complement components, and the effector functions of eosinophils are induced by various ligands for these receptors (2, 3, 4). However, the triggers and molecular mechanisms of eosinophil activation, in particular those involved in in vivo settings, are not fully understood.

Eosinophils express several Ig superfamily cell surface molecules, such as leukocyte Ig-like receptor/Ig-like transcript (LIR/ILT),³ LIR-1/ILT-2, LIR-2/ILT-4, LIR-3/ILT-5, LIR-7/ILT-1 (5), sialic acid-binding Ig-like lectins (6), IRp60 (7), CD52 (8), and CD48 (9). Ligation of these molecules with specific mAb either enhances or inhibits eosinophil activation. For example, cross-linking of CD48 induces activation and degranulation of eosinophils (9). In contrast, cross-linking of CD52, CD300a, and Siglec-8 inhibited superoxide production, migration and survival of eosinophils (6, 7, 8). Thus, complex networks of signals from both activating and inhibitory cell surface molecules likely regulate eosinophil effector functions, including the release of proinflammatory mediators.

CD66b (CEACAM8, CGM6, NCA-95) is a single chain GPI-anchored glycoprotein and is a member of the Ig superfamily, more specifically the human carcinoembryonic Ag (CEA) family (10). CEAs are widely used as colorectal carcinoma tumor markers but are expressed in normal tissues and during fetal development (10). Indeed, CD66b is exclusively expressed on human granulocytes and is recognized as a granulocyte “activation marker” (11, 12). However, the functions of this molecule are largely unknown in eosinophils (13, 14). Under normal conditions, neutrophils have minimal expression of CD66b, but incubation in vitro with inflammatory agonists, such as fMLP, rapidly increases this expression. Furthermore, expression of CD66b in eosinophils is enhanced in vivo in patients with a helminth infection, specifically with Schistosoma mansoni. (15). The eosinophil expression of CD66b is also enhanced in vitro by incubation with activated lymphocytes (16) or by brief incubation with IL-5 (unpublished observations). Interestingly, no homologue for CD66b has been identified in the mouse, suggesting that there may be strong selection pressure (e.g., exposure to microorganisms or parasites) during the evolution of the CEA family molecules. Thus, we hypothesized that CD66b may be critically involved in the activation of human eosinophils.

As specific membrane microdomains that are enriched in sterols and sphingolipids, lipid rafts play several important roles in immune cells. For example, they contribute to BCR signaling and Ag uptake (17, 18), to ligand-mediated signaling from T cell receptors (19, 20) and from high affinity IgE receptors (21), and to MHC class II-mediated Ag presentation (22, 23). GPI-anchored proteins are often localized in these lipid rafts, where they possibly play roles in regulating molecular interactions, signal transduction, and cell activation (24, 25, 26, 27). To the best of our knowledge, the presence and function of lipid rafts have not been previously reported in eosinophils.

We now show that perturbation of CD66b by mAb or by a natural ligand, galectin-3, results in strong cellular adhesion and degranulation of human eosinophils. Furthermore, lipid rafts and β₂ integrins are likely involved in eosinophil activation induced by CD66b. Our observations suggest that CD66b may be an important positive regulator of eosinophil effector functions during inflammation, host immunity, and allergic diseases.

Mouse anti-human CD66b mAb (clones 80H30 and GM-2H6, both IgG1 isotype) were obtained from Serotec and Abcam, respectively. Mouse anti-human CD66a mAb (clones 29H2 and GM8G5, both IgG1 isotype; clone 283340, IgG2b isotype) were purchased from Novocastra Laboratories, Alexis Biochemicals, and R&D Systems, respectively. Control mouse IgG1, polyclonal Ab for the Src kinase family (rabbit anti-Hck, Fgr, Lck, Lyn), control rabbit IgG, anti-CD11b mAb (clone 2LPM19C, IgG1) and peroxidase-conjugated rabbit anti-mouse IgG Abs were obtained from Santa Cruz Biotechnology. F(ab′)₂ of anti-human CD66b mAb (80H30) and control mouse IgG1 were generated by using an Ab F(ab′)₂ fragmentation kit (Pierce Biotechnology). FITC-conjugated anti-CD11b mAb (clone ICRF44, IgG1) was from Ancell Corporation, and FITC-conjugated control mouse IgG1 and anti-CD11a (clone G43-25B, IgG2a) were purchased from BD Biosciences. Control mouse IgG2b was from R&D Systems. FITC- and peroxidase-conjugated cholera toxin B subunit, peroxidase-conjugated goat anti-rabbit IgG, human serum albumin, methyl β-cyclodextrin (MβCD), F(ab′)₂ of goat anti-mouse IgG (Fc specific), FITC-conjugated F(ab′)₂ of sheep anti-mouse IgG, and myelin basic protein (MBP, exogenous substrate of Src kinase family in vitro kinase assay) were obtained from Sigma-Aldrich. Human rgalectin-3 was purchased from RDI; human rIL-5 and phosphatidylinositol-specific phospholipase C (PI-PLC) were obtained from Calbiochem. A Src kinase-specific inhibitor (PP1) was purchased from A. G. Scientific. Anti-phosphotyrosine mAb (4G10) was obtained from Upstate Biotechnology.

Human eosinophils were isolated from 15 healthy volunteers using Percoll density gradient centrifugation and the MACS system for negative selection with anti-CD16 microbeads (Miltenyi Biotec) as described earlier (28). The purity of eosinophils was >96%, and cells were used immediately. The Mayo Clinic Rochester Institutional Review Board approved the protocol to obtain blood from volunteers; all provided informed consent.

For CD66b detection, eosinophils (10 × 10⁶ cells) were lysed in lysis buffer (25 mM HEPES (pH 7.4), 1% Nonidet P-40, 125 mM NaCl, 100 μM Na₃VO₄,1 mM PMSF, and protease inhibitor mixture); the lysate was centrifuged at 15,000 × g for 20 min at 4°C; the supernatant was incubated with anti-CD66b mAb (80H30, 3 μg/ml) or control mouse IgG1 overnight at 4°C; protein A-Sepharose (40 μl) was added; and the mixture was incubated for 3 h at 4°C. After several washings with lysis buffer, a 5-fold excess of sample buffer was added, and the mixture was boiled at 95°C for 10 min. After centrifugation at 15,000 × g, supernatants were subjected to 10% SDS-PAGE. Proteins in the gel were transferred onto nitrocellulose paper; after blocking with 5% skim milk in PBST (0.2% Tween 20 and PBS) for 1 h at room temperature, the paper was incubated with the appropriate primary and secondary Abs, and the ECL system was used to detect CD66b.

To detect activated Hck and other Src-family molecules, eosinophils (5 × 10⁶ cells/sample) were incubated with 10 μg/ml anti-CD66b mAb (80H30) on ice for 30 min. Then the primary Ab was cross-linked with 50 μg/ml F(ab′)₂ of goat anti-mouse IgG for intervals between 0 and 30 min at 37°C. Reactions were stopped with cold PBS; cells were centrifuged at 1,000 × g for 5 min at 4°C and immediately lysed with lysis buffer for 1 h on ice. Lysate was centrifuged at 15,000 × g for 15 min at 4°C. Supernatants were incubated with anti-Hck polyclonal Ab (3 μg/ml) overnight at 4°C. Precipitation with protein A-Sepharose (40 μl) and Western blot analysis were performed, as described above, except 5% BSA was used instead of skim milk, and the appropriate primary and secondary Abs were used in the ECL system to detect phosphorylated Hck.

Isolated eosinophils were washed with PAB (PBS, 3% BSA and 0.1% sodium azide) buffer twice and incubated with mouse anti-human CD66b mAb (80H30), anti-human CD66a mAb (283340), control mouse IgG1, or control mouse IgG2b for 45 min on ice. After washing with PAB, cells were incubated with FITC-conjugated F(ab′)₂ of sheep anti-mouse IgG for 30 min on ice. Cells were washed twice with PAB, fixed with 1% formaldehyde for 20 min and analyzed using a FACScan (BD Biosciences), with Becton Dickinson lysis II software. In some experiments, eosinophils (5 × 10⁶ cells) were pretreated with PI-PLC (1 U) for 1 h at 37°C to remove the GPI-anchored protein(s) from the cell surface (29). Cells were fixed with 1% formaldehyde for 20 min and washed with PAB before incubation with anti-CD66b mAb.

Superoxide generation was measured by superoxide dismutase-inhibitable reduction of cytochrome c, as described earlier (2). In brief, for the experiments with immobilized CD66a mAb and CD66b mAb, plates were first coated with 50 μg/ml F(ab′)₂ of goat anti-mouse IgG. After washing, the plates were incubated with anti-CD66a mAb (29H2 and GM8G5, 1 μg/ml), anti-CD66b mAb (80H30 and GM-2H6, 0.05 to 1 μg/ml) or control mouse IgG1 in 1% HSA (Sigma-Aldrich) in PBS for 2 h at 37°C and washed twice with 0.9% NaCl. Eosinophils (10⁵ cells/well in HBSS with 10 mM HEPES and 0.01% gelatin) were incubated in anti-CD66a, anti-CD66b or isotype control Ab-coated plates, and superoxide production was measured (2). For the experiments with F(ab′) ₂ of mAb, plates were directly coated with 1 μg/ml F(ab′) ₂ of anti-CD66b mAb (80H30) or control mouse IgG1, and blocked with 1% HSA. In certain experiments, we used IL-5 (10 ng/ml), PMA (1 ng/ml) and galectin-3 (10 μg/ml) to stimulate eosinophils. To quantify eosinophil degranulation, the concentrations of EDN in the cell-free supernatants were measured by specific RIA, as described earlier (2). To examine the effect of cholesterol depletion on eosinophil activation via CD66b, eosinophils were incubated with the cholesterol depletion reagent, MβCD (10 mM), for 1 h at 37°C, 5% CO₂ before exposure to each stimulus. To examine whether the Src kinase family is involved in eosinophil activation, cells were incubated with a Src kinase specific inhibitor (PP1, 10 μM) for 30 min at 37°C, 5% CO₂. To remove GPI-anchored CD66b from the cell surface, purified eosinophils were incubated with PI-PLC (1U) for 1 h at 37°C, 5% CO₂. After each preincubation, cell viability, using trypan blue exclusion, was >99%.

To investigate CD66b cross-linking effects on morphological changes, polystyrene beads or tissue culture plates were coated with F(ab′)₂ of goat anti-mouse IgG (50 μg/ml) at 4°C overnight, washed with 0.9% NaCl, and incubated with anti-CD66b mAb (80H3) or control mouse IgG1 for 1 h at 37°C. Freshly isolated eosinophils were incubated with Ab-coated beads or plates for 30 min or 3 h at 37°C, respectively. After fixation with 3.7% formaldehyde for 20 min at 4°C, the suspensions were washed twice with PBS. Images were taken with a Zeiss LSM 510 confocal laser scanning microscope (Carl Zeiss).

Eosinophils (2 × 10⁷ cells) were lysed in the TNE buffer (10 mM Tris-HCl, NaCl 150 mM, and EDTA 5 mM) with 1% Triton X-100, and the lysate was passed 5 times through 25 gauge needles. The lysates were mixed with equal volumes of 85% sucrose in TNE buffer with protease inhibitor mixtures and transferred to a centrifuge tube (30). Samples were overlaid with 6 ml of 35% sucrose and 4 ml of 3.5% sucrose in TNE buffer. After centrifugation for 16 h at 288,000 × g in a Beckman Coulter SW41Ti rotor, fractions (1 ml each) were collected from the top of the gradients. Each fraction was precipitated with 20% TCA and washed twice with cold acetone.

Eosinophils (10⁷ cells) were stimulated with immobilized anti-CD66b mAb or control Ab for 10 or 30 min. at 37°C. Then, cells were lysed in 25 mM HEPES (pH 7.4), 1.5% Nonidet P-40, 125 mM NaCl, 100 μM Na₃VO₄,1 mM PMSF and protease inhibitor mixture for 30 min on ice. After centrifugation at 15,000 × g for 20 min, the supernatants were incubated with anti-Hck, -Fgr, -lyn and -lck Abs (3 μg/ml) for 3 h at 4°C, followed by incubation with protein A-Sepharose overnight at 4°C. After the immunoprecipitated complexes were washed with lysis buffer and the kinase reaction buffer (100 mM Tris-HCl (pH 7.2), 125 mM MgCl₂, 25 mM MnCl₂, 2 mM EGTA, 100 μM Na₃VO₄, 2 mM DTT), the complexes were suspended in 30 μl of kinase reaction buffer with 10 μCi [³²P]ATP (PerkinElmer Life Science) and substrate MBP (10 μg/sample) and finally incubated for 20 min at 30°C. Reactions were stopped with 5× sample buffer, and samples were subjected to SDS-PAGE. Gels were dried for 1 h at 80°C and then exposed to the bio-MS film (Kodak) for 30 min at −80°C and developed.

Eosinophils (5–6 × 10⁶ cells) were activated by incubation with anti-CD66b mAb (80H30) or control mouse IgG1 (10 μg/ml) and F(ab′)₂ of goat anti-mouse IgG (50 μg/ml) for 15 min at 37°C. Activation was stopped by adding cold PBS. After washing with PBS, cells were lysed with lysis buffer (0.5% Nonidet P-40, 20 mM Tris-Cl (pH 7.4), 150 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂, 10% glycerol, and a protease inhibitor mixture) for 30 min on ice. Lysates were precleared with protein A/G-Sepharose and incubated with anti-CD66b mAb or control Ab (2 μg/ml) overnight at 4°C. Subsequently, protein A/G-Sepharose was added and incubated for 2 h at 4°C. After washing (3×) with lysis buffer, proteins were eluted by adding sample buffer and boiling for 5 min at 95°C. CD11b was detected by Western blot analysis with anti-CD11b mAb (clone 2LPM19C; Santa Cruz Biotechnology).

The cellular localization of CD11a and CD11b molecules after stimulation with anti-CD66a or CD66b mAb were examined by a capping experiment. Eosinophils (1 × 10⁶ cells) were activated by incubation with 10 μg/ml anti-CD66a mAb (GM8G5) or anti-CD66b mAb (80H30), and F(ab′)₂ of goat anti-mouse IgG (50 μg/ml) for 30 min at 37°C. Eosinophils were fixed with 3.7% formaldehyde for 20 min at room temperature, washed with 1% PAB and spread by using cytospin. After staining with FITC-conjugated anti-CD11b mAb (1/200 dilution) or FITC-conjugated anti-CD11a (1/100 dilution) in PAB for 40 min at room temperature and washing (3×) with 1% PAB, anti-fade reagent was added and images were taken with a LSM 510 confocal laser scanning microscope.

All the error bars represent SEM. Two-sided differences between two samples were analyzed with the Student t test or Mann-Whitney U test. Values of p < 0.05 were considered significant.

CD66b is a GPI-anchored glycoprotein that is exclusively expressed on human granulocytes (14). Both flow cytometry analyses (Fig. 1,A) and Western blot (Fig. 1,B) show that CD66b is expressed abundantly on human eosinophils isolated from healthy individuals. CD66a, another member of the human CEA family, is also expressed abundantly on eosinophils (Fig. 1,A). Eosinophil CD66b is highly glycosylated, as demonstrated by its migration around a 100 kDa molecular mass range (Fig. 1,B), when compared with the 31 kDa predicted polypeptide molecular mass (31). To investigate functions of CD66b in eosinophils, we first examined the effects of CD66b cross-linking by specific mAb on superoxide generation. Eosinophils were incubated with anti-CD66b mAb immobilized onto tissue culture plates through goat anti-mouse IgG Fc-specific Ab. Eosinophils incubated with immobilized anti-CD66b mAb actively generated superoxide in a time-dependent manner; in contrast, eosinophils incubated with immobilized anti-CD66a mAb or medium alone did not generate superoxide (Fig. 2,A). The concentration-response study shows that eosinophils were activated by as little as 0.05 μg/ml anti-CD66b mAb, but they did not respond to 1 μg/ml anti-CD66a mAb or control mouse IgG1 Ab (Fig. 2,B). Other clones of anti-CD66a mAb, including 29H2 (mouse IgG1) and 283340 (mouse IgG2b), did not generate superoxide (data not shown). In contrast, two clones of anti-CD66b mAb, namely 80H30 and GM-2H6, robustly induced superoxide generation (Fig. 2,C). Furthermore, F(ab′)₂ of anti-CD66b Ab, which had been immobilized directly onto the plates, potently activated eosinophils (Fig. 2 D). Thus, ligation of eosinophil surface CD66b, but not CD66a, by Ab strongly activates eosinophils and induces superoxide generation in an Fc receptor-independent manner.

We examined whether other effector functions of eosinophils are induced by CD66b cross-linking, and we examined functions induced by an authentic stimulus for eosinophils, IL-5, and by a pharmacologic stimulus, PMA. Eosinophil superoxide production induced by cross-linking CD66b was comparable to that induced by IL-5 (10 ng/ml) and ∼58% of that induced by PMA (1 ng/ml) (Fig. 3,A). As measured by release of a granule protein EDN into supernatants, CD66b cross-linking also induced eosinophil degranulation, (Fig. 3,B); the magnitude of eosinophil degranulation induced by CD66b cross-linking was comparable to that induced by IL-5 and ∼65% of that induced by PMA. Morphologically, eosinophils adhered and spread onto tissue culture plates coated with anti-CD66b mAb (Fig. 3,C). A photomicrograph of polystyrene beads coated with anti-CD66b mAb clearly depicts eosinophils spreading over the spherical bead surface (Fig. 3 C). No degranulation or cellular adhesion was observed with eosinophils incubated with immobilized mIgG1 Ab or anti-CD66a mAb. Thus, cross-linking of CD66b triggers various effector functions in eosinophils, including adhesion, superoxide production and degranulation.

We next examined whether galectin-3, a physiologic ligand for both CD66b and CD66a (32), recapitulates the anti-CD66b mAb results. Galectin-3 is a member of a family of β-galactoside-binding animal lectins; it is produced by macrophages and mast cells during inflammation (33, 34) and is implicated in eosinophilic airway inflammation in mice (35). At 10 μg/ml galectin-3 significantly induced superoxide production and degranulation of eosinophils (p < 0.05, n = 3) (Fig. 4, A and B). These eosinophil responses to galectin-3 were roughly similar to those induced by cross-linking CD66b with mAb (1 μg/ml). Cross-linking of CD66a by anti-CD66a mAb (29H2, 1 μg/ml) did not induce superoxide production or degranulation (Fig. 4, A and B).

To rule out the potential nonspecific effects of galectin-3, we examined the effects of PI-PLC, an enzyme that specifically removes GPI-anchored proteins from the cell surface. Incubation with PI-PLC decreased the number of CD66b on the eosinophils’ surface (Fig. 5,A). Furthermore, pretreatment with PI-PLC decreased eosinophil superoxide production induced by cross-linking CD66b mAb or by soluble galectin-3 (Fig. 5 B, p < 0.01, n = 3). In contrast, superoxide production induced by PMA was not affected by PI-PLC treatment. Thus, not only anti-CD66b cross-linking mAb, but also a natural ligand for CD66b induces eosinophil activation.

To understand better the molecular mechanisms for eosinophil activation induced by CD66b, we hypothesized that lipid rafts may be involved. Because there were no reports for lipid rafts in eosinophils, we first investigated whether CD66b is localized to lipid rafts. Unstimulated eosinophils (i.e., incubated with control mouse IgG) and activated cells (i.e., incubated with anti-CD66b mAb) were lysed and a sucrose gradient with ultracentrifugation was used to separate the membrane fractions (30, 36, 37). In unstimulated eosinophils, CD66b was detectable in both lipid rafts and in nonraft fractions; CD66b cofractionated with an authentic lipid raft marker GM1 (30) and with Lyn, which has been mainly detected in the lipid raft fractions (Fig. 6,A). In stimulated eosinophils, the numbers of CD66b molecules in the lipid raft fractions dramatically increased (Fig. 6 A), suggesting active recruitment of CD66b from nonraft compartments to raft compartments during activation.

We also investigated the roles of these lipid rafts in eosinophil activation induced by CD66b cross-linking. Because cholesterol depletion specifically disrupts lipid rafts without affecting other membrane structures, we used a cholesterol depletion reagent, MβCD, which has been used for other cell types (30, 38, 39, 40). By trypan blue exclusion, the MβCD pretreatment did not affect eosinophil viability (data not shown). When cholesterol was depleted, eosinophil adhesion to the anti-CD66b mAb-coated beads was markedly inhibited (Fig. 6,B, inset). Furthermore, MβCD pretreatment nearly abolished eosinophil superoxide production induced by cross-linking CD66b, but showed no effects on superoxide production induced by PMA (Fig. 6 B). Thus, CD66b is localized to lipid rafts, and their integrity is critical for the functions of CD66b.

We investigated downstream events of cellular activation induced by CD66b cross-linking. In lymphocytic cells, some Src kinase family proteins localize in lipid rafts where they play a key role in cell activation mediated by GPI-anchored surface molecules (26, 41, 42, 43, 44). Eosinophils express four types of Src kinases, including Hck, Fgr, Lyn, and Lck (45, 46). Therefore, one or more Src kinases may be involved in eosinophil activation through CD66b. Eosinophils were activated by cross-linking CD66b, and changes in the phosphorylation levels of Hck, Fgr, Lyn, and Lck were analyzed by immunoprecipitation and by immunoblot with anti-phosphotyrosine mAb (47). Increased phosphorylation of Hck was observed within 1 min after CD66b cross-linking; it reached a peak at 5 min and decreased gradually thereafter (Fig. 7,A). These results were reproducible among three different eosinophils donors and showed a 193 ± 30% increase in Hck phosphorylation at 30 min (mean ± SEM, p < 0.05, n = 3). An in vitro kinase assay using an exogenous substrate, MBP, showed that the enzymatic activity of Hck also increased within 10 min of CD66b cross-linking (Fig. 7,B). Similar experiments were repeated with Fgr, Lyn, and Lck, but we did not detect increased tyrosine phosphorylation or increased kinase activity of these molecules after CD66b cross-linking (data not shown). To examine the role of Hck activation in eosinophil effector functions, eosinophils were pretreated with the Src kinase specific inhibitor, PP1. Both superoxide generation and EDN release induced by CD66b cross-linking were abolished by PP1 treatment, but superoxide generation and EDN release induced by PMA were not affected by PP1 (Fig. 7, C and D). Thus, Hck likely plays a key role in eosinophil activation mediated by CD66b cross-linking.

Integrin adhesion molecules, such as α_Mβ₂ (Mac-1, CD11b/CD18) and α₄β₁ (VLA-4, CD49d/CD29) are likely important for eosinophil recruitment into inflammation sites in allergic diseases. These integrins, particularly β₂ integrin, can regulate eosinophil degranulation in response to stimulation with various physiologic secretagogues (48, 49). Furthermore, ligation of CD11b by mAb, without other exogenous stimuli, triggers eosinophil activation (50). Because CD66b cross-linking induced tight adhesion of eosinophils and their effector functions (Fig. 3), we investigated the potential interaction between CD66b and CD11b. Eosinophils incubated with anti-CD66b mAb and F(ab′)₂ of goat anti-mouse IgG showed distinct clustering and capping of CD11b (Fig. 8 A). In contrast, eosinophils incubated with anti-CD66a mAb and F(ab′)₂ of goat anti-mouse IgG showed that CD11b was distributed evenly over the cells’ surfaces. Anti-CD66b mAb did not induce clustering of another β₂ integrin molecule, CD11a, suggesting a specific interaction between CD66b and CD11b.

To investigate their interactions at a molecular level, both unstimulated (i.e., incubated with control mouse IgG) and activated (i.e., incubated with anti-CD66b mAb) eosinophils were lysed and immunoprecipitated with anti-CD66b mAb; the molecules associated with CD66b were depicted by Western blot analysis. In unstimulated eosinophils, both CD11b and CD66b were detectable in the immunoprecipitants with anti-CD66b mAb (Fig. 8 B). In stimulated eosinophils, the amount of CD11b molecules in the CD66b immunoprecipitants did not change, suggesting that CD11b is constitutively associated with CD66b without cellular activation.

Herein, we characterized the functions of CD66b in human eosinophils. The CD66b glycoprotein is highly expressed on the eosinophils’ surface, and cross-linking the CD66b with specific mAb or a ligand, galectin-3, induces tight cellular adhesion, superoxide generation, and degranulation. The magnitudes of the eosinophil effector functions induced by cross-linking the CD66b molecule were comparable to those induced by an authentic agonist, IL-5. Furthermore, independent of the eosinophil activation induced by CD66b cross-linking, CD66b constitutively localizes in lipid rafts. Cholesterol depletion and disruption of lipid rafts blocks eosinophil activation and adhesion mediated by CD66b. Cross-linking CD66b activates a Src kinase, Hck, and the cellular functions are completely abrogated with the Src kinase specific inhibitor, PP1. These observations suggest a model wherein lipid rafts, as a member structure, and Hck, as a signaling molecule, are essential for eosinophil activation mediated by CD66b cross-linking.

The CD66b glycoprotein is GPI-anchored to the cell membrane and is constitutively localized in lipid rafts; it does not have cytoplasmic domains and may be unable to interact directly with intracellular signaling molecules. However, Hck can also be localized in lipid rafts (51). Thus, CD66b could directly interact with Hck through lipid-lipid interactions between the GPI of CD66b and the palmitoylate of Hck. In addition, CD66b cross-linking could induce an accumulation of microdomain lipid rafts, resulting in the formation of macromolecules and inducing an accumulation and activation of signal-related molecules, including Hck. Finally, other transmembrane proteins might interact with CD66b and directly activate the intracellular signaling pathways. For example, integrins can function as adaptors in GPI-anchored protein-mediated cell activation (52). Our observations suggest that CD11b (i.e., α-chain of a β₂ integrin) is constitutively associated with CD66b and that they form a macromolecular complex after CD66b ligation (Fig. 8 A). Furthermore, cross-linking of CD11b, without any other exogenous stimulus, leads to tyrosine phosphorylation of several intracellular proteins and induces eosinophil degranulation (50). Although another CEA family member, CD66a, is highly expressed on the eosinophils’ surface, cross-linking of CD66a did not activate eosinophils. Unlike CD66b, CD66a has conventional transmembrane anchors as well as a 74-aa cytoplastic domain (10). The unique molecular features of CD66b, including the GPI anchor, localization to lipid rafts, and interactions with CD11b, may explain this molecule’s potent ability to trigger eosinophil functions. Thus, several nonmutually exclusive hypotheses are possible for CD66b-mediated mechanisms involved in the activation and effector functions of eosinophils. Future experiments that use molecular biological approaches, such as small interfering RNA, will be necessary to elucidate the precise mechanisms for CD66b-mediated eosinophil activation.

The unique characteristics of CD66b molecules, which are highly glycosylated, GPI-anchored, and localized in lipid rafts, may suggest two biological insights for innate immunity and molecular interactions. First, the highly glycosylated CD66b may be a receptor for lectin-type molecules generated by other inflammatory cells or be an innate receptor for the products of microorganisms. For example, activated macrophages and mast cells produce galectin-3 during inflammation (34, 53), and galectin-3 induces the production and release of proinflammatory mediators by eosinophils, likely through CD66b (Fig. 4). Galectin-3 may also decorate the cell surface carbohydrate molecules on microorganisms; this would allow eosinophils to recognize potential targets (54, 55). Furthermore, CEA family molecules directly recognize the carbohydrate cell surface structure of microorganisms (56). Second, CD66b may physically interact with other membrane molecules and form macromolecular complexes of transmembrane proteins and signaling molecules within the lipid rafts. These macromolecular complexes may then act as a nucleus to generate intracellular activation signals and to induce the effector functions of eosinophils. Indeed, the CEA family molecules, including CD66b, induce homophilic and heterophilic receptor aggregation and are considered to be part of a molecular receptor complex on the cell surface (10). In eosinophils, one of the components of these macromolecular complexes is likely CD11b (Fig. 8). In the future, the isolation and characterization of the molecular complexes containing CD66b on the surface of human eosinophils will be particularly important. These analyses should include studies of signal transduction pathways used by these receptor complexes and should search for additional natural ligands.

Despite an extensive search, the genes encoding the CD66b protein have not been identified in mice or rats (57). In contrast, CD66b homologues have been identified in two primate species, baboon and African green monkey (58). Thus, CEA family molecules are probably undergoing rapid evolution; there must be strong selection pressures on these molecules, such as exposures to microorganisms and parasites. Mouse eosinophils are relatively resistant to various activation stimuli (59) and eosinophil degranulation is rare in mouse models of eosinophilic inflammation (60). Could this lack of vigorous activation of mouse eosinophils compared with human eosinophils be due to the lack of CD66b? Thus, it will be important to generate mice transgenic for human CD66b, to examine the functions of eosinophils from these mice in vivo and in vitro, and to evaluate the roles of eosinophils more critically in immunity, host immunity and disease pathology.

In conclusion, the cross-linking of CD66b by specific Abs or a ligand, galectin-3, activates a Src family kinase, Hck, and triggers tight cellular adhesion and production and release of proinflammatory mediators in human eosinophils. Furthermore, CD66b is localized in lipid rafts and is physically associated with an adhesion molecule, CD11b; upon cross-linking, CD66b induces a marked clustering of CD11b molecules. CD66b may be a versatile positive regulator of eosinophil effector functions through recognition of their environmental ligands and/or physical interactions with other membrane molecules. Future studies will identify the importance of CD66b during inflammation, host immunity, and allergic diseases and will elucidate potential therapeutic approaches to target this molecule.

We thank Diane Squillace for technical assistance, Cheryl R. Adolphson for editorial assistance, and LuRaye S. Eischens for secretarial help.

The authors have no financial conflict of interest.