Human Peripheral Blood Eosinophils Express a Functional c-kit Receptor for Stem Cell Factor that Stimulates Very Late Antigen 4 (VLA-4)–mediated Cell Adhesion to Fibronectin and Vascular Cell Adhesion Molecule 1 (VCAM-1)

rhSCF (catalogue no. 1833-01, lot no. B6326; Genzyme, Cambridge, MA), rh eotaxin (Endogen, Inc., Boston, MA), 2′,7′-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM) (Molecular Probes, Eugene, OR), human serum albumin (HSA) (Sigma Chem. Co., St. Louis, MO), purified anti-human c-kit YB5.B8 (PharMingen, San Diego, CA), purified anti-human c-kit (95C3) (Coulter, Miami, FL), purified anti-human integrin β₁ (CD29) (4B4) (Coulter), purified anti-human integrin α⁴ (CD49d) (A4-PUJ1) (Upstate Biotechnology, Lake Placid, NY), purified anti-human CD3 (HIT3a) (PharMingen), mouse laminin and the 40-kD chymotryptic fragment of human fibronectin (FN40) (GIBCO BRL, Gaithersburg, MD) were purchased. The preparation of purified VCAM-(mouse Cκ) fusion protein (VCAM-1-κ) was previously reported (52). Anti-human integrin α⁴ (CD49d) (B5G10, ascites) (53), anti-human integrin α⁶ (CD49f) (450-33D, ascites) (54), anti-human integrin β₁ (CD29) (A-1A5, ascites) (55), purified anti-human integrin α⁴β₇ (Act-1) (56), and the negative control antibody P3 (hybridoma culture supernatant) (57) were provided by Dr. M.E. Hemler (Dana-Farber Cancer Institute, Harvard Medical School). Purified anti-human c-kit mAb SR-1 (ascites) (58) was a gift from Dr. V. Broudy (University of Washington, Seattle, WA).

Blood was collected into sterile, heparinized syringes from the peripheral veins of nonatopic and atopic volunteer donors who gave informed consent. After the erythrocytes were sedimented with dextran for 45 min at 37°C, the granulocyte fraction was obtained by centrifugation through a cushion of Ficoll–Hypaque (1.77 g/ml; Pharmacia, Uppsala, Sweden) of the buffy coat at 350 g for 30 min. After the hypotonic lysis of residual erythrocytes, eosinophils were separated from neutrophils by negative immunomagnetic selection with a magnetic cell separation (MACS) column (Miltenyi Biotec, Sunnyvale, CA) (59). In brief, the erythrocyte-depleted granulocyte pellet was incubated for 45 min at 4°C with anti-CD16 mAb bound to immuno-magnetic beads. When the mixture was applied to a steel wire column in a strong magnetic field, the CD16⁺ neutrophils were retained, whereas the CD16⁻ eosinophils were highly purified in the fraction that flowed through the column. Contaminating T lymphocytes and monocytes were further removed by incubating the CD16⁻ fraction for another 15 min at 4°C with saturating concentrations of anti-CD3 and anti-CD14 magnetic beads, respectively (Miltenyi). Cytocentrifugation slides of the eosinophils stained with Wright's and Giemsa stains showed that the purity of the isolated eosinophils was greater than 95% in all experiments.

Cytofluorographic analyses of surface epitopes expressed by human peripheral blood eosinophils with or without rhSCF stimulation were performed by established procedures (52). Freshly isolated human peripheral blood eosinophils were resuspended in RPMI 1640 medium containing 10% FCS at a concentration of 10⁶ cells/ml and were divided into two identical fractions. rhSCF was added to one fraction to a final concentration of 100 ng/ml (5 nM). Alternatively, rh eotaxin was used at a final concentration of 24 nM and both fractions were incubated for 15 min at 37°C. The cells were harvested and washed once with cold PBS containing 0.5% HSA and 0.02% sodium azide (FACS buffer). Samples of 5 × 10⁵ cells were then incubated for 1 h on ice with primary antibodies (purified mAbs at a final concentration of 10 μg/ml or ascites at a final dilution of 1:500 or P3 culture supernatant at 1:4 dilution). The cells were washed once with FACS buffer and incubated in the dark for 1 h on ice with fluorescein isothiocyanate–conjugated goat anti–mouse IgG (GIBCO BRL) at a final dilution of 1:100. The cells were washed again with the FACS buffer, resuspended in 0.25 ml of PBS, and analyzed on a FACSort^® machine (Becton Dickinson, Oxnard, CA). For c-kit expression (Fig. 1), the results are presented as overlaid histograms and the fold increase of mean fluorescence intensity (MFI). The fold increase of c-kit expression was calculated by dividing the MFI units of SR-1 staining by the MFI units of P3 control mAb staining in each donor.

For immunohistochemical analysis of the surface c-kit expression, peripheral blood buffy coats were prepared after dextran sedimentation of erythrocytes. Two fractions of the buffy coat, each containing 10⁶ white blood cells, were incubated on ice for 1 h with SR-1 (1:500 diluted ascites) and control P3 antibody (1:4 diluted hybridoma culture supernatant) in a total volume of 100 μl of FACS buffer, respectively. Cells were washed once with 3 ml of FACS buffer, and cytocentrifugation slides were prepared, with 4 × 10⁴ cells per slide. The slides were fixed in 4% paraformaldehyde (Polysciences, Inc., Warrington, PA) in PBS for 10 min at room temperature. After washing with PBS for 3–4 times, the slides were blocked with 2% chicken egg albumin (Sigma) for 30 min at room temperature, and were incubated with 1:10 diluted goat anti–mouse IgG(Fc)-conjugated gold particles (average size 10 nm) (Amersham International, Buckinghamshire, England) for 1 h at room temperature. The slides were washed first with PBS 2–3 times and then with distilled water 2–3 times. A silver enhancement procedure was carried out for 8 min at room temperature following the manufacturer's manual (AuroProbe™; Amersham). The slides were then counterstained with hematoxylin and eosin, mounted with Permount^® (Fisher Scientific, Pittsburgh, PA), and analyzed with a Leica microscope (Model Dialux 20). Results were photographically recorded using Kodak Royal Gold film ASA25.

Cells were attached in stasis to FN40, VCAM-1, and laminin as described (52). FN40 and laminin at concentrations of 3 μg/ml, 10 μg/ml, and 30 μg/ml were coated onto non–tissue culture grade 96-well microtiter plates (Nunc-Immuno Plate) in 0.1 M NaHCO₃ (pH 8.3) (100 μl/well) for 16 h at 4°C. The plates were washed twice with PBS (150 μl/well), and the nonspecific binding sites were blocked by incubation with 5% HSA in PBS (100 μl/well) for 45 min at 37°C. After two washes with PBS (150 μl/well), the plates were ready to be used. For the VCAM-1 assay, plates were precoated with goat anti–mouse κ (GIBCO BRL) at a 1:1,000 final dilution in 0.1 M NaHCO₃ (100 μl/well) for 16 h at 4°C. After two washes with PBS, the plates were coated with VCAM-1-κ at concentrations of 0.1, 0.5, and 1.0 μg/ml in 0.1 M NaHCO₃ (100 μl/well) for 2 h at 4°C. The plates were washed with PBS and blocked with HSA as mentioned above. The freshly isolated eosinophils were incubated with BCECF-AM (5 μg/ml) in RPMI 1640 containing 10 mM Hepes and 0.1% HSA (RPMI–HSA) for 30 min at 37°C, washed twice with PBS, and resuspended in RPMI–HSA at 4 × 10⁶ cells/ml. Samples (50 μl) of the cell suspension containing 2 × 10⁵ cells were added to each well of the ligand-coated plate; each well had been preloaded with either 50 μl of RPMI–HSA or 50 μl of RPMI–HSA containing defined concentrations of rhSCF (generally 10 nM) or defined concentrations of rh eotaxin (generally 48 nM). The plates were incubated for 15 min at 37°C, and the fluorescence of total input cells was quantitated by a fluorescence analyzer (Idexx Laboratories, Westbrook, ME). Unbound cells were removed by washing the plates with RPMI/10 mM Hepes 4–5 times at 150 μl/well until the cells in the control wells (5% HSA alone) were less than 5% of their total input. Cells remaining attached to the plate were quantitated after every wash with the same fluorescence analyzer. The percentage of specific binding to immobilized integrin ligands was calculated as the fluorescence of the cells detected after the final wash divided by the fluorescence of the total starting cells × 100, and assay results are presented as means ± SD of three independent experiments, each performed in triplicate.

For antibody-blocking experiments, cells were resuspended in cold RPMI–HSA at a concentration of 4 × 10⁶ cells/ml and divided into four identical lots. Purified mAbs were added to each lot to a final concentration of 10 μg/ml. The mAbs used in this study were anti-CD3 (negative control), anti-α⁴ (A4-PUJ1), anti-α⁴β₇ (Act-1), and anti-β₁ (4B4). Cells were incubated with the mAbs for 10 min on ice and added to each well of a FN40 (30 μg/ml) coated plate. During the 15-min incubation at 37°C, the mAb concentration in each well was reduced to 5 μg/ml by preloading the wells with RPMI–HSA with or without rhSCF or rh eotaxin. The remainder of the assay procedure was as described above.

The statistical significance of differences between sample means for each set of cells was based on comparison as determined by the Student's t test for matched pairs. Results are presented as mean ± SD.

To determine whether freshly isolated, peripheral human blood eosinophils expressed the c-kit receptor, the purified cells from eight separate donors (four nonatopic and four atopic) were analyzed by cytofluorography for the expression of c-kit. Three mouse mAbs against three independent epitopes of the human c-kit (58, 60) gave virtually identical expression for two of the donors (Fig. 1,A), with mean log fluorescence intensities of 2.0- and 3.3-fold over control, respectively, with mAb SR-1. The surface c-kit expression was subsequently confirmed in six additional donors with one of the three mAbs, SR-1 (Fig. 1 B). In every case, the c-kit expression was readily detectable, with a mean log fluorescence intensity of 2.0 to 3.6-fold greater than the negative control (IgG control mAb P3) (mean 2.8 ± 0.6 fold, n = 8).

To confirm c-kit expression by peripheral blood eosinophils and determine its potential expression by other circulating leukocytes, peripheral blood buffy coats from two separate donors (1 atopic and 1 nonatopic) were incubated with either SR-1 or control P3 antibody, and were subjected to immunohistochemical analysis using secondary antibody-conjugated gold particles and a silver enhancement procedure, followed by counterstaining with hematoxylin and eosin. As indicated by the counterstaining and shown for the nonatopic donor (Fig. 2), 100% of the eosinophils in the buffy coats were positive for c-kit, and eosinophils were the only cells displaying a signal for c-kit receptor. Identical results were obtained for the atopic donor (data not shown). A similar positive signal was also detected on freshly isolated human peripheral blood eosinophils after MACS column purification (data not shown).

To determine the functional integrity of the expressed c-kit receptor on eosinophils, the ability of rhSCF to augment their adhesion to ligands selective for integrin α⁴ and α⁶ was evaluated in a static adhesion assay. The 15-min time course for the adhesion assay was selected because a kinetic study showed that augmented adhesion peaked at 15 min, persisted for 30 min, and decreased to baseline by 45 min (data not shown). Adhesion to FN40 increased at all three inputs of 3, 10, and 30 μg/ml of immobilized FN40. In the absence of rhSCF, the adhesion of FN40 was limited (specific binding of 4.3 ± 3.5% at 30 μg/ml FN40). Concomitant stimulation with rhSCF (5 nM) augmented adhesion to 3, 10, and 30 μg/ml FN40 by 7.7 ± 1.4-fold, 5.3 ± 3.3-fold, and 5.4 ± 0.2-fold, respectively, compared with baseline (Fig. 3). Adhesion to rVCAM-1 also increased in relation to the input of ligand. In the absence of rhSCF, rVCAM-1 supported greater adhesion than FN40 (specific binding of 9.3 ± 9% and 22.5 ± 10% at 0.5 μg/ml and 1.0 μg/ml of rVCAM-1, respectively). Stimulation with rhSCF augmented adhesion to 0.1, 0.5, and 1.0 μg/ml rVCAM-1 by 12.7 ± 9.2-fold, 3.8 ± 2.5-fold, and 1.7 ± 0.6-fold, respectively, compared with baseline (Fig. 3). Higher baseline adhesion mediated by VCAM-1 compared with fibronectin has been previously reported (22). Cell adhesion to laminin also increased in a dose-dependent fashion with respect to ligand input but did not increase further with stimulation by rhSCF (Fig. 3). Treatment with rhSCF did not change the surface expression of integrin α⁴, α⁶, β₁, α⁴β₇, or c-kit receptor as evaluated by cytofluorographic analysis (Fig. 4), indicating that the increases in adhesion to the α⁴ ligands were not due to increased receptor expression.

Both of the α⁴ integrins expressed on eosinophils, VLA-4 (α⁴β₁) and α⁴β₇, serve as receptors for FN40 and VCAM-1 (19). Antibody-blocking experiments were performed in the assay for adhesion of eosinophils to 30 μg/ml FN40 with or without rhSCF stimulation (5 nM). Although the fold increase in response to rhSCF was similar at each concentration of FN40 (see Fig. 3), the greatest absolute increment in adhesion occurred at 30 μg/ml FN40. mAbs against α⁴ (A4-PUJ1) and β₁ (4B4) completely abolished eosinophil adherence to FN40, whereas the anti-α⁴β₇ mAb (Act-1) had no effect and was comparable to the control mAb of irrelevant specificity (anti-CD3) (Fig. 5). Similarly, cell adhesion to rVCAM-1 in the absence or presence of rhSCF (5 nM) was also completely blocked by anti-α⁴ and anti-β₁ mAbs (data not shown). Therefore, rhSCF–c-kit ligation specifically augmented the adhesion of VLA-4 to its ligands, fibronectin, and VCAM-1.

The effect of rhSCF at concentrations ranging from 12.5 ng/ml (0.625 nM) to 200 ng/ml (10 nM) on eosinophil adhesion to 30 μg/ml FN40 was studied. Enhancement of VLA-4 binding to FN40 by rhSCF was significant at 2.5, 5, and 10 nM as compared with the unstimulated replicate eosinophils (P <0.05) (Fig. 6). Each fourfold increment in rhSCF concentration produced a statistically significant gain in binding (P <0.05) except for the final increment between 2.5 and 10 nM, suggesting that a plateau was reached.

Eotaxin, a selective eosinophil chemoattractant (61, 62), belongs to the CC chemokine family. Other members of this family are known to be stimuli for eosinophil adhesion via integrin α⁴ (63). Preliminary dose-dependence experiments indicated that the effect of 24 nM rh eotaxin was similar to the effect of 5 nM rhSCF for augmenting eosinophil adhesion to 30 μg/ml FN40. Thus, the effects of these concentrations on eosinophil adhesion were compared over three concentrations of FN40 (Fig. 7). Eotaxin at 24 nM increased eosinophil binding to FN40 to the same extent as 100 ng/ml (5 nM) rhSCF at each static input of FN40 ligand. Eotaxin did not change the surface expression of c-kit, α⁴, α⁶, β₁, and α⁴β₇ by cytofluorographic analysis (data not shown), and the augmented adhesion of these cells to FN40 produced by eotaxin was completely blocked by mAbs against α⁴ and β₁ (Fig. 8).

The dose-related effect of rh eotaxin on human eosinophil adhesion to FN40 (30 μg/ml) was analyzed in the absence and presence of 1.25 nM rhSCF. The approximate EC₅₀ for rh eotaxin-augmented adhesion occurred at the lowest concentration studied, 3 nM, and the plateau was reached at 12 nM (20.8 ± 13-fold, n = 3). The effect of the concomitant presence of rhSCF at slightly less than its EC₅₀ (see Fig. 6) was generally somewhat additive and did not extend the maximum reached with the plateau doses of rh eotaxin alone (Fig. 9).

The finding that mature human peripheral blood eosinophils express a functional c-kit receptor for SCF reveals that a stromal ligand can activate effector cells conventionally linked to allergic inflammation. Although originally appreciated for its central role in the hematopoiesis of all lineages, SCF is elaborated in the peripheral microenvironment by diverse cell types, including endothelial cells and fibroblasts (34–37). In addition to the functional implications for eosinophil- and mast cell–directed inflammation, the study also uncovers a potential limitation to the use of c-kit detection as a marker of mature mast cells.

In the initial experiments, the surface expression of the c-kit receptor was established by cytofluorographic analysis of freshly isolated human peripheral blood eosinophils. The expression was similar among the cells of all eight donors tested (Fig. 1), irrespective of the presence of donor atopy by history. To exclude an unsuspected specificity, three mAbs directed against separate epitopes of the human c-kit were shown to yield nearly identical fluorescence signals. The positive surface c-kit expression was also demonstrated by immunohistochemical analysis of peripheral blood buffy coats (Fig. 2). Eosinophils were uniformly positive for c-kit expression and were the only cell type expressing c-kit in these preparations. The presence of c-kit was then confirmed by the functional, dose-related response of the eosinophils to SCF signal in static adhesion assays (Fig. 6). An earlier study did not detect a c-kit signal on peripheral blood eosinophils, and functional assays with SCF were not performed (64). However, this same study did demonstrate low level c-kit expression by human peripheral blood basophils and a priming effect of SCF on their IgE-dependent release of histamine (64). Because eosinophils and basophils are closely related and arise from a common progenitor (65), the expression of c-kit and their activation via SCF can now be added to the list of their shared characteristics. Moreover, a triad of hematopoietic allergic effector cells, mast cells, basophils, and eosinophils, would appear to share the expression of c-kit.

rhSCF–c-kit stimulated eosinophil adhesion to FN40 and VCAM-1, but not to laminin (Fig. 3), implying a response via an α⁴ integrin. Antibody-blocking studies revealed that the adhesion of eosinophils to FN40 (Fig. 5) and VCAM-1 (data not shown) was mediated by VLA-4 without the involvement of α⁴β₇. Because the surface integrin receptor numbers were unchanged after treatment of the cells with rhSCF, including α⁴, α⁶, β₁, and α⁴β₇ (Fig. 4), the enhanced adhesiveness of VLA-4 induced by rhSCF was likely due to increased avidity of VLA-4 to its ligands. This observation is supported by an earlier study with the transformed human cell line MO7E (66). The α⁴β₇ integrin did not contribute to the fibronectin and VCAM-1 binding, suggesting that the avidity change of VLA-4 by SCF may be conveyed through either the unique β subunit or the combined effect on the α and β heterodimer. The fact that rhSCF did not increase VLA-6 (α⁶β₁)–mediated adhesion to laminin favors the latter hypothesis. Thus, the SCF–c-kit engagement on human peripheral blood eosinophils provides an inside–out signal to increase transiently the avidity of an integrin, VLA-4, for its cell surface and matrix ligands.

Recently, Weber and colleagues (63) demonstrated that the CC chemokines RANTES and MCP-3, as well as the anaphylatoxin C5a, rapidly increased α⁴ integrin avidity on eosinophils and augmented their adhesion to VCAM-1 and fibronectin with a peak at 15 min after stimulation. This result prompted us to examine the effect of another CC chemokine, eotaxin, on eosinophil adhesive function. The EC₅₀ values of 1.25–2.5 nM for rhSCF (Fig. 6) and of 3 nM for rh eotaxin (Fig. 9) were similar with 30 μg/ml of FN40. Similarly, the magnitudes of eosinophil binding to the inputs of 3, 10, and 30 μg/ml of FN40 were similar at the plateau concentrations of rhSCF and of rh eotaxin (Fig. 7). The cell adhesion response to each stimulus (rhSCF and rh eotaxin) for the ligand FN40 depended entirely on VLA-4, as shown by mAb inhibition (Figs. 5 and 8). Stimulation of the peripheral blood eosinophils with less than the EC₅₀ for the stromal cytokine, rhSCF, and incremental amounts of the chemokine, rh eotaxin, appeared additive, and the combination neither increased the maximal cellular response nor revealed a negative counterregulatory action for the common target, VLA-4.

Cell migration is a complex process requiring constant changes of the cell-substratum bonds with formation of new bonds (adhesion) at the leading edge of a cell and breakdown of old bonds (deadhesion) at the rear of the cell (67). This transient response is essential for adhesion to be followed by random or directed cell migration. Progressive movement within a tissue may require integrated activation of one integrin and resultant negative regulation of another integrin on the same cell as the ligand binding requirements change. For example, in Chinese hamster ovary cells, engagement of the transfected human α^IIbβ₃ with its specific ligand caused a transdominant inhibition of other integrins and resulted in suppressed adhesion of endogenous hamster α⁵β₁ to fibronectin, as well as transfected human α²β₁ to collagen (68). In another study, the activation of the transfected integrin α^IIbβ₃ in Chinese hamster ovary cells, detected by mAb PAC1 binding, was suppressed by the introduction of H-Ras and Raf-1 to these cells (69). This suppressive activity was correlated with activation of the ERK MAP kinase pathway and did not involve other dual specificity kinase sequences such as SEK–c-Jun or MKK–p38 MAP kinase (69). Thus, the activation of a particular integrin by cytokines such as SCF, which leads to an enhanced binding of that integrin to its ligand, may contribute to an integrated signal for coordinated cellular events such as cell migration.

Although found in small numbers in the circulation of healthy people, eosinophils mainly reside in the mucosal tissues in vivo, with an estimated tissue:circulation ratio of 200:1 (1). The mechanisms involving the tissue-specific localization of eosinophils are not commonly addressed, as most studies have focused on the incremental recruitment with immunologically elicited inflammation. SCF is a normal constitutive product of endothelial cells and fibroblasts and can be detected in the serum (36, 70). Therefore, it is a candidate to mediate a basal level interaction between circulating eosinophils and endothelial cells by facilitating interaction between VLA-4 and VCAM-1 at very low levels of VCAM-1 expression (71). Alternatively, exposure of eosinophils to SCF in the extracellular matrix may mediate their transient attachment to fibronectin, altering their migration speed and influencing their tissue localization. Indeed, Palecek and colleagues (72) have demonstrated that at low concentrations of ligand, cell migration speed increased as integrin binding affinity increased. Thus, SCF is a candidate for participating in the tissue distribution of eosinophils under normal physiologic conditions.

The interaction of VLA-4 and VCAM-1 plays a key role in lymphocyte and eosinophil adhesion and extravasation in both in vitro and in vivo models of allergic inflammation (18, 22–26). VCAM-1 expression is upregulated in the eosinophil-rich, inflamed airway tissue of individuals with asthma (73, 74). VLA-4 is expressed on eosinophils and lymphocytes, but not neutrophils; and the c-kit receptor expression is limited to eosinophils among these three cell types. Thus, under conditions of allergic inflammation, local SCF could contribute both by priming mast cells to elaborate IL-4 (46, 75) and TNF-α (75) with consequent upregulation of endothelial cell VCAM-1 expression (76), and by mediating a transiently increased avidity for that ligand by eosinophil VLA-4. This possibility is supported by a recent finding in mice of the association of allergen challenge-induced eosinophilic airways inflammation with increased levels of histamine and SCF in bronchoalveolar lavage fluid and of SCF in serum (70). In that study, the administration of antibody to SCF before allergen challenge markedly decreased histamine release and pulmonary eosinophil infiltration. The inflammatory response induced by allergen challenge also increases the production of the eosinophil-selective chemoattractant, eotaxin (61), which is a product of a number of cell types including epithelial cells, endothelial cells, fibroblasts, and even eosinophils (62). Therefore, increased concentrations of eotaxin and SCF could provide a coordinated signal for the selective recruitment of eosinophils, beginning with transiently enhanced eosinophil adhesion to endothelial cells within the vasculature through VLA-4–VCAM-1 interaction, followed by directed tissue movement through adhesion/deadhesion in a chemotactic gradient, and final eosinophil tissue retention through regulated integrin–matrix interaction. Such a matrix interaction may also prolong eosinophil survival through autocrine production of IL-3 and GM-CSF initiated by VLA-4–fibronectin interaction (28).

We have provided evidence that SCF is an agonist for eosinophil adhesion. These findings have potential relevance for physiologic eosinophil trafficking and for allergic inflammation, in which mast cell activation and eosinophil accumulation are both prominent features.

This work was supported by National Institutes of Health grants AI-22531, AI-31599, AI-01304, and HL-36110. Q. Yuan is the recipient of a research fellowship from the Charles A. King Trust and The Medical Foundation.