Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Polymorphonuclear leukocytes (PMN) and eosinophils (Eos) are important cellular participants in a variety of acute and chronic inflammatory reactions in airway tissue. These related but differentially regulated types of cell are often densely distributed in airway tissue in response to viral or bacterial infections, during adult respiratory distress syndrome and in allergic disease, and can be found in significant numbers in the lumen of the airway in these diseases (1). The presence of these cells represents an important component of the host response to infection; however, their persistent presence and activation are thought to be largely responsible for the pathophysiology of several debilitating diseases including cystic fibrosis, chronic bronchitis, and asthma (7). Adhesive interactions are thought to be important for both the localization and activation of leukocytes. Adhesion-dependent signaling has been reported to enhance both oxygen radical formation and degranulation responses in PMN and Eos (10).

Accumulation of PMN and Eos in the airway depends first on transmigration across the vascular endothelium, a stepwise process requiring concerted action of selectins, which initiate contact under flow conditions; integrins, which firmly tether leukocytes to the endothelium; and integrin- cell-adhesion molecule (CAM) interactions involved in the passage of leukocytes between endothelial cells (16). Once PMN and Eos are beyond the endothelial barrier, chemotactic signals frequently lead to the accumulation of these cells near mucosal epithelial cells as well as in the lumen of the airway. Airway epithelial cells have been shown to be an abundant source of chemokines for both PMN and Eos under appropriate stimulatory conditions (9, 21, 22). These cells also express intercellular adhesion molecule (ICAM)-1, an important adhesive ligand for both PMN and Eos (23, 24). Airway epithelial cells may therefore be important not only for retention and activation of PMN and Eos, but also in the passage of these cells across epithelial cells and into the airway itself. The activation state of epithelial cells may therefore play an important role in dictating the final destination of emigrating leukocytes.

Although the molecular interactions between vascular endothelial cells and leukocytes have been extensively described, the nature of the interactions between PMN or Eos and epithelial cells are less clear. Eos express a number of different adhesion molecules that participate in their interactions with endothelial cells. L-Selectin and ligands for P-selectin promote rolling of Eos along the endothelial cell surface under shear forces resulting from normal blood flow (25, 26). Recent studies suggest that very late antigen-4 (VLA-4; ₄₁ integrin) may also participate in the rolling process (27, 28). In the absence of shear forces resulting from flow dynamics, adhesion is primarily mediated through integrins. Firm adhesion and transcellular passage appear to involve both VLA-4 and ₂ integrins, depending in part on the activation state of the endothelial cell (29). The principal known couterligands for ₂ integrins and VLA-4 are ICAM-1 and vascular cell adhesion molecule (VCAM)-1, respectively (18, 29). In addition to expressing ₁ and ₂ integrins, Eos also express ₄₇, a counterligand for mucosal addressin cell-adhesion molecule (MadCAM) (33, 34). Integrin molecules, particularly ₁ integrins expressed by Eos and other leukocytes, also frequently mediate cell-matrix adhesion (10, 35).

PMN also express L-selectin and ligands for E- and P-selectin, as well as ₂ integrins (18, 20, 26, 29). PMN may also express ₅₁ and ₆₁ integrins (35, 36). Although resting human PMN have been shown to express ₁ integrins, they appear not to express ₄ integrin (VLA-4), and are therefore incapable of interacting with VCAM-1 (25, 29, 30, 35). This differential expression of VLA-4 has been postulated to account for the selective recruitment of Eos in asthma and allergic diseases. Tissue generation of Eo-specific cytokines serves as an alternative mechanism by which preferential localization of Eos may occur. In particular, interleukin (IL)-3, IL-4, and IL-5 have been closely associated with allergic diseases (37). IL-4 is known to selectively upregulate VCAM-1 on endothelial cells (32, 40, 41). IL-3 and IL-5 have been shown to induce both the differentiation and the activation of Eos (38, 42). In addition, IL-8, regulated on activation, normal T-cell expressed and secreted (RANTES), and granulocyte-macrophage colony-stimulating factor (GM-CSF), all of which have chemotactic activities, are produced by inflamed pulmonary epithelium, and may act as an endogenous source of leukocyte-specific chemoattractants (9, 48). Given the importance of adhesion in the activation and localization of PMN and Eos to airway tissues, we investigated the relative contributions of ₁, ₂, and ₇ integrin molecules on Eos and PMN to these cells' adhesion to primary human airway epithelial cells. We also characterized the inflammatory signals and the counterligands on epithelial cells that regulate these adhesive interactions.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Reagents

C3a and C5a were purified from yeast-activated human serum as described previously (51). Endotoxin contamination was determined to be less than 10 pg/µg protein through the Limulus amoebocyte lysate assay (BioWhittaker, Walkersville, MD). Antibodies to the adhesion molecules ICAM-1 (clone P2A4), VCAM-1 (clone 51-10C9), and ₇ integrin (clone FIB27, a rat antimurine ₇ that cross-reacts with human ₇) were purchased from PharMingen, (San Diego, CA). Blocking antibody to _M integrin (CD11b; clone 2LPM19c) was purchased from Dako Corporation (Carpinteria, CA), and blocking antibody to _x integrin (CD11c; clone 3.9) was obtained from Serotec (Raleigh, NC). The antibody IB4 (52), directed against CD18, the common subunit of ₂ integrins, was provided by Dr. Karl Arfors (Experimental Medicine Inc., Princeton, NJ). The antibodies 33B6 (anti-₁) (53) R3.1 (anti-CD11a; _L integrin) (19), and the Fab fragment of R6.5 (anti-ICAM-1; CD54) (19) were a generous gift of Dr. Scott Simon of the Baylor University College of Medicine, Houston, TX. The activating anti-₁ integrin antibody TS2/16 was purchased from Endogen (Cambridge, MA). The cytokines TNF- and IFN- were purchased from PeproTech (Rocky Hill, NJ). All culture media were purchased from BioWhittaker. Dextran 70 was supplied by Baxter Diagnostics, Inc. (McGaw Park, IL). Ficoll-Paque Plus was obtained from Pharmacia Biotech (Piscataway, NJ). Goat antimouse IgG-coated magnetic beads and a magnetic separation unit were purchased from Advanced Magnetics, Inc. (Cambridge, MA). Unconjugated mouse monoclonal antihuman antibodies against CD16 were purchased from Biosource International (Camarillo, CA).

Cell Culture

Human bronchial epithelial cells (HBECs) were isolated from human large bronchi and trachea obtained from unusable organ-transplant specimens according to previously described methods (54). Briefly, after washing with Earle's balanced salt solution (EBSS) containing penicillin/streptomycin and 25 mM 4-(2-hydroxyethyl)-1-piperazine-N'-2-ethanesulfonic acid (Hepes), the epithelial layer of the airways was stripped away with a scalpel and incubated with 0.1% Pronase (Sigma Chemical Co., St. Louis, MO) for 30 min at 37°C. The cells were dispersed and filtered through sterile 60-mm Nitex mesh (Cellector; E-C Apparatus Corp., Holbrook, NY). After centrifugation, the cells were suspended in bronchial tracheal epithelial growth medium (BEGM) and cultured in standard tissue-culture treated flasks in a humidified atmosphere at 37°C and 5% CO₂. Medium was exchanged every 2 d until cells reached confluence between 6 and 10 d after plating. Cells were passaged by treatment with trypsin-ethylenediaminetetraacetic acid (EDTA) solution and reseeding at a density of 1:4. Cells were used in the third to ninth passage. Homogeneity and purity of the cultures were confirmed by cytokeratin staining.

PMN Preparation

PMN were isolated from the peripheral blood of healthy human donors according to a modification of standard leukocyte isolation procedures. Blood was drawn into syringes containing dextran (8 ml/60-ml syringe) and EDTA to achieve a final concentration of 10 mM EDTA. After sedimentation of blood specimens for 1 h, leukocyte-rich plasma was recovered into a 250-ml conical tube and centrifuged at 250 × g for 10 min. The pellet was diluted in 24 ml of EBSS, and 6-ml aliquots were carefully layered over 7 ml of Ficoll-Paque in 15-ml conical centrifuge tubes. After centrifugation at 450 × g for 12 min, the top interface, containing the mononuclear cell layer, was removed and discarded. The PMN-containing fraction appeared as a diffuse cell suspension above the cell pellet. The upper third of the suspension was removed, washed, and resuspended in EBSS. Purity was determined by staining in a solution of 0.1% eosin Y in 10% acetone to enumerate the total number of leukocytes and the percentage of Eos. PMN were typically  98% pure.

Preparation of Eos

Eos were isolated from gradients of the same density used to isolate PMN. After removal of the mononuclear cell- and PMN-containing fractions above the cell pellet, the remaining red cells in the pellet were lysed for 10 s with cold water. The tonicity was restored with 10× phosphate-buffered saline and the remaining cells were washed with EBSS and resuspended (10⁷/ml) in RPMI-1640 medium containing 10% fetal calf serum, 10 mM Hepes, and 1% penicillin/streptomycin. Eos were further purified by negative selection of remaining PMN as previously described (55). The PMN-specific mouse anti-CD16 antibody was added (1 µl per 3 million PMN) and incubated at room temperature (RT) for 15 min. Magnetic beads (50 particles per PMN) were added to this mixture and incubated at RT for 30 min with occasional mixing. The beads, together with the adherent PMN, were then removed by magnetic separation. Remaining Eos were collected, washed once, resuspended in EBSS, and counted as described for PMN. Eos were  99% pure as assessed through eosin staining.

Adhesion Studies

HBECs were grown to confluence in standard 24-well culture plates in complete medium. Twenty-four hours before the adhesion assay, complete medium (BEGM) was replaced with fresh medium without epidermal growth factor and bovine pituitary extract. In some experiments, cells were treated with either interferon (IFN)- or tumor necrosis factor (TNF)- 24 h before the adhesion assay. Immediately before addition of leukocytes, medium was removed from the HBEC cultures and the cells were washed with 0.5 ml EBSS. Purified PMN (0.5 × 10⁶/well) or Eos (0.1 × 10⁶/well) were added in a final volume of 0.25 ml EBSS. The plates were returned to a 37°C incubator for 15 min to reequilibrate the temperature and allow the leukocytes to contact the confluent cells. Selected stimuli were added and the wells and the plates were returned to the incubator for 15 min. Nonadherent cells were then removed by shaking the plates at 200 rpm for 15 s, decanting the supernatant, washing with warm EBSS, and decanting. Remaining adherent cells were quantitated by measuring the peroxidase content of the wells and comparing these values with the peroxidase levels in serially diluted samples of PMN or Eos as subsequently outlined. After washing away of the nonadherent cells, 0.25 ml of 0.5% hexadecyl-trimethyl ammonium bromide was added to each well to extract cellular peroxidase. A 50-µl aliquot from each well was added to a 96-well assay plate, and 150 µl of substrate (o-dianisidine, 1 mg/ml in 0.1 M phosphate buffer containing 0.001% H₂O₂, pH 6.5) was added. Absorbance was measured at 450 nm. The number of adherent cells was determined by comparing peroxidase activity with that of serial dilutions of a standard number of purified Eos or PMN. In preliminary experiments it was found that HBECs did not contain significant endogenous peroxidase, and that negligible peroxidase release from Eos or PMN occurred during the adhesion assay.

Reverse Transcription-Polymerase Chain Reaction

Polymerase chain reaction (PCR) primer sets for ICAM-1, VCAM-1, and -actin were purchased from Stratagene (San Diego, CA). Total RNA was isolated from 1-5 × 10⁶ cells using TRIzol reagent (GIBCO BRL, Gaithersburg, MD) according to the manufacturer's instructions. Total RNA was reverse transcribed with Superscript II reverse transcriptase (200 U/assay) and oligo deoxythymidine_12-18 (GIBCO BRL) at a 1 mM final concentration according to the manufacturer's protocol. Reverse transcription (RT)- PCR was performed in a 50-µl final volume containing: 5 µl 10× Taq reaction buffer, 26.4 µl double distilled H₂O, 6 µl of 25 mM MgCl₂, 0.4 µl 25 mM deoxynucleotide triphosphates, 0.2 µl Taq polymerase (5 U/ml) (Perkin-Elmer, Foster City, CA), 2 µl (1 mM final concentration) of the 5' sense and 3' antisense primers, and 10 µl of 1:10 diluted DNA. After an initial denaturation at 94°C for 5 min, and a 5-min annealing at 60°C, samples were run for 30-35 cycles under the following conditions: 1 min denaturation at 94°C, 1-min annealing at 60°C, and 2-min extension at 72°C. Following amplification, PCR products were separated on a 3% agarose gel and visualized by ethidium bromide staining.

Flow Cytometry

For measurement of ₂ integrin on PMN and Eos, mixed granulocyte populations were stained with either phycoerythrin (PE)-conjugated CD9 (an Eo-specific marker) or PE-conjugated CD16 (a PMN-specific marker) along with fluorescein isothiocyanate (FITC)-conjugated IB4 (anti-₂ integrin). The mixed granulocytes were stimulated for 15 min at 37°C with various stimuli as indicated in the text. Cells were then immediately placed on ice and incubated with antibodies (5 µg/ml in 0.1 ml) in EBSS containing 10% fetal bovine serum, 0.05% NaN₃, and 0.1% bovine gamma globulins for 45 min. Cells were washed twice with cold medium and analyzed on either a FACScan or FACSort instrument (Becton Dickinson, Mountain View, CA) using CellQuest software (Becton Dickinson). Cells were gated according to either CD9 or CD16 expression. A minimum of 2,500 Eos (CD9⁺/CD16) or 5,000 PMN (CD9/CD16⁺) were collected and analyzed.

Analysis of adhesion-molecule expression on epithelial cells was done with mouse monoclonal antibodies to ICAM-1 and VCAM-1. Briefly, confluent cells were detached with EDTA/trypsin solution, placed on ice, centrifuged, and resuspended at 0.5 × 10⁶ cells/ml in labeling buffer containing 5 µg/ml of anti-adhesion-molecule antibody. Cells were thereafter treated and analyzed as described in the preceding section.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Regulation of CAM Expression on HBECs

The two best-characterized static adhesion molecules for leukocytes are ICAM-1 and VCAM-1. ICAM-1 has been shown to be essential for PMN adhesion to endothelial cells, whereas Eos have been shown to interact both with ICAM-1 and VCAM-1 (25, 29, 30, 32). We therefore investigated the expression and regulation of ICAM-1 and VCAM-1 on HBECs by several inflammatory cytokines. ICAM-1 was expressed constitutively by HBECs and was upregulated by treatment with IL-1, TNF-, or IFN- (Figure 1). The greatest increases in ICAM-1 expression resulted from treatment with IFN-, and optimal upregulation required a 24-h pretreatment with any of these stimuli (see Figure 3 for IFN-; data not shown for TNF- or IL-1). IL-4 had no effect on ICAM-1 expression on HBECs. VCAM-1 could not be detected on HBECs (Figure 1). VCAM-1 was also not inducible with any of the cytokines tested, including IL-4, which selectively induces VCAM-1 expression on endothelial cells (32, 40, 41). Furthermore, although both ICAM-1 and -actin mRNA were readily detectable with RT-PCR, no VCAM-1 messenger RNA (mRNA) was detectable in either stimulated or unstimulated HBECs (data not shown).

View larger version (32K): [in this window] [in a new window]   Figure 1.   Expression of ICAM-1 and VCAM-1 by HBECs. HBECs were grown to confluence in six-well tissue-culture plates. Individual wells were then stimulated with IL-4 (10 ng/ml), IL-1 (10 ng/ml), TNF- (25 ng/ml), or IFN- (50 ng/ml), or were left unstimulated. Cells were harvested 24 h later and stained for ICAM-1 (CD54) or with control mouse IgG (negative control) for flow cytometry. The closed histograms represent staining with specific antibody; the mean fluorescence intensity for a minimum of 5,000 cells is shown for each treatment group. Staining with the control mouse IgG is shown by the open histograms. ICAM-1 was expressed on unstimulated HBECs, and its expression was increased after treatment with IL-1, TNF-, or IFN-. VCAM-1 was not expressed by HBECs under any of the treatment conditions. Results are representative of a minimum of three experiments.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Transcription-dependent regulation of ICAM-1. HBECs were cultured and stimulated with IFN- (50 ng/ml) in the presence or absence of actinomycin D (500 ng/ml) for 24 h. Cells were then harvested and analyzed for ICAM-1 expression by flow cytometry. Relative expression of ICAM-1 was normalized to initial values for three separate experiments and expressed as mean fluorescence intensity (arbitrary units) ± SEM.

To determine whether the differences in efficacy between TNF- and IFN- were a function of stimulus concentration, we conducted a dose-response study. The results shown in Figure 2 demonstrate that although TNF- and IFN- were essentially equipotent on a molar basis for induction of ICAM-1, the maximal response attainable (efficacy) with TNF- was substantially weaker than that with IFN-. The calculated EC₅₀ for TNF- was 3.7 ng/ml, as compared with 2.2 ng/ml for IFN-. In contrast, TNF- led to a maximal increase in ICAM-1 expression of 4.5-fold, as compared with a 21-fold increase in response to IFN-. Therefore, in contrast to its relative effect on endothelial cells, IFN- appears to much more effectively induce ICAM-1 expression on epithelial cells than do either IL-1 or TNF-.

View larger version (14K): [in this window] [in a new window]   Figure 2.   Dose-dependence of ICAM-1 expression in response to TNF- and IFN-. HBECs were cultured and stimulated with the indicated concentrations of either TNF- (closed squares) or IFN- (closed triangles) as described in Figure 1. Expression of ICAM-1 was determined with flow cytometry at 24 h after stimulation, and was expressed as a percent increase over basal levels. Responses to IFN- are indicated on the y-axis and those for TNF- on the right y-axis. Although each cytokine showed comparable EC50 values, IFN- was much more effective in increasing expression of ICAM-1. Data are representative of two separate experiments.

The kinetics of upregulation of ICAM-1 expression suggested a dependence on de novo synthesis of this adhesion molecule. To confirm that IFN- upregulation of ICAM-1 occurred at a transcriptional level, we stimulated cells with IFN- in the presence of actinomycin D, an inhibitor of mRNA synthesis. Under these conditions, upregulation of ICAM-1 was inhibited by nearly 90% (Figure 3), indicating that IFN- regulates ICAM-1 gene expression in airway epithelial cells.

Differential Responsiveness of PMN and Eos to Inflammatory Stimuli

Several cytokines, chemokines, and humoral mediators have been associated with airway inflammation. Among these, the anaphylatoxin C5a, the chemokine IL-8, and the cytokines IL-3, IL-5, and GM-CSF deserve special attention. IL-3 and IL-5 have been closely linked to allergic inflammation (37, 42). The importance of IL-8 and C5a have been demonstrated in a number of in vivo models of airway inflammation, and both IL-8 and GM-CSF are produced in substantial quantities by airway epithelial cells (3, 9, 56). We therefore examined the abilities of these various mediators to modulate expression of adhesion molecules on Eos and PMN, and their ability to induce leukocyte adhesion to airway epithelial cells. As shown in Figure 4, the various mediators showed differing specificities for responsiveness of PMN or Eos. Of the five mediators studied, GM-CSF and C5a were the only ones that enhanced expression of ₂ integrins on both Eos and PMN. IL-8 had a similar effect only on PMN, whereas IL-5 showed Eo-specific effects. Although Eos have previously been shown to express receptors for and to respond physiologically to IL-3, this cytokine had no effect on expression of ₂ integrins by Eos. Differences in the responsiveness of each cell type were independent of dose or of the stimulus used (i.e., C5a and GM-CSF had comparable EC₅₀ values for both Eos and PMN; the maximal response for either cell type was comparable for all effective stimuli). Differences in responsiveness were also independent of the time of exposure, and changes in ₂ integrin expression were stable for up to 2 h (data not shown).

View larger version (15K): [in this window] [in a new window]   Figure 4.   Induction of 2 integrin expression by inflammatory mediators. Human peripheral blood granulocytes were enriched for Eos (> 10% Eos) by dextran sedimentation and density-gradient centrifugation as described in MATERIALS AND METHODS. Mixed PMN and Eos were stimulated for 30 min at 37°C with C5a (100 ng/ml), IL-3 (25 ng/ml), IL-5 (25 ng/ml), IL-8 (80 ng/ml), GM-CSF (10 ng/ml), or medium alone (NS). Cells were then stained at 4°C for expression of 2 integrin, using FITC-conjugated anti-CD18 antibody IB4. PMN and Eos were differentiated upon analysis by staining with a PE-conjugated antibody to CD9. Data are the mean ± SEM of three experiments (*P < 0.05 versus unstimulated cells).

To extend the observations of upregulation of ₂ integrins outlined in Figure 4 to the physiologic response of cellular adhesion, we also studied the effects of each of the mediators named earlier on adhesion of PMN and Eos to HBECs. As shown in Figure 5, the effects of the mediators on granulocyte adhesion closely reflected their observed effects on adhesion-molecule expression. GM-CSF and C5a promoted adhesion of both Eos and PMN to airway epithelial cells, whereas IL-8 showed PMN specificity and IL-5 exhibited Eo specificity. IL-3 was without effect on either cell type, even under conditions in which epithelial cells were primed for adhesion by treatment with IFN-. Because of its occurrence with PMN as well as Eos, and its established importance in airway disease, the C5a-mediated adhesion of Eos to airway epithelial cells was studied in greater detail with regard to its effects and mechanisms.

View larger version (21K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 5.   Adhesion of Eos and PMN to HBECs. Purified human PMN (A) or Eos (B) were incubated either with resting HBECs or HBECs that were prestimulated for 20-24 h with IFN- (50 ng/ml) in the presence of the indicated stimuli (concentrations are as in Figure 4). Nonadherent cells were removed, and remaining adherent cells were determined by peroxidase activity. Data are the mean ± SEM of three experiments (*P < 0.05 versus unstimulated HBECs; P < 0.05 versus IFN--stimulated HBECs).

Adhesion of PMN and Eos to HBECs

To investigate potential mechanisms leading to leukocyte- epithelial interactions, we studied the adhesion of PMN and Eos to either resting HBECs or HBECs stimulated either with IFN- or TNF-. In addition to investigating HBEC activation, we investigated the effects of C5a, a potent stimulus for both PMN and Eos, on leukocyte-epithelial cell adhesion. The results are summarized in Figure 6. Despite the marked upregulation of ICAM-1 expression on HBECs by IFN-, PMN adhesion in the absence of additional stimuli, was increased by only 6.5% over baseline, and Eo adhesion was increased by only 7.4% over baseline, by pretreatment with IFN-. Similar increases were observed in response to pretreatment of HBECs with TNF-. C5a alone produced a marked increase in adhesion of both Eos and PMN to unstimulated HBECs (24% PMN adherent in response to C5a, versus 2.5% in the absence of stimuli; 20% Eos adherent in response to C5a, versus 1.9% under control conditions). Optimal adhesion of PMN and Eos resulted when HBECs were pretreated with either IFN- or TNF- and leukocytes were stimulated with C5a. To characterize the molecular basis for adhesion of PMN and Eos to HBECs, cells were pretreated with blocking antibodies to potential leukocyte and epithelial-cell adhesion molecules before stimulating adhesion. The effects of blocking antibodies to ₁, ₂, and ₇ integrins on Eo and PMN adhesion are summarized in Figure 7. PMN exhibited complete dependence on ₂ integrins for adhesion to HBECs, as shown by a more than 90% inhibition of adhesion by the anti-₂ antibody IB4 (Figure 7A). In contrast, the relative contribution of ₂ integrins to Eo adhesion depended on the nature of stimulation. C5a-stimulated Eo adhesion to IFN--primed HBECs was found to be almost fully blocked by IB4, suggesting an exclusive utilization of ₂ integrins (CD18) (Figure 7B). Under these conditions, antibodies to either ₁ or ₇ integrins were largely ineffective at inhibiting adhesion. However, C5a-stimulated adhesion to resting HBECs showed a codependence on both ₂ and ₁ integrins. Treatment of Eos with anti-₄ integrin antibodies suggested that the ₁ component of adhesion was primarily mediated through VLA-4 (₄₁), since treatment with either anti-₄ or anti-₁ antibody blocked adhesion to a similar degree.

View larger version (15K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Figure 6.   Combined effects of leukocyte and epithelial stimuli on PMN and Eo adherence. Isolated PMN (A) or Eos (B) were cultured with HBECs under the indicated conditions. Where indicated, HBECs were prestimulated for 20-24 h and washed before addition of leukocytes and C5a. Maximal adhesion for either cell type required activation of both leukocytes and epithelial cells. Data are the mean ± SEM of a minimum of six experiments. (*P < 0.05 versus unstimulated HBECs; P < 0.05 versus stimulated HBECs.)

View larger version (18K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 7.   Effect of blocking antibodies to leukocyte integrins on Eo and PMN adhesion. Purified human PMN (A) or Eos (B) were treated with antibodies to the indicated integrin subunits (10 µg/ml) for 15 min before addition to either resting HBECs or HBECs incubated for 20-24 h with IFN- (50 ng/ml). Adhesion in each group was stimulated by addition of 10 nM C5a. Data represent the percent suppression of adhesion by each antibody as compared with stimulated adhesion in the absence of antibody treatment. Data are the mean ± SEM of four to seven experiments (*P < 0.05).

To confirm that Eos may adhere to HBECs through ₁ integrins, we used a ₁ integrin-activating antibody, TS2/ 16 (60), as a proadhesive stimulus. This antibody has previously been shown to induce cellular adhesion through ₁ integrins. Treatment of Eos with this antibody increased adhesion to unstimulated HBECs from 6% to 35% (data not shown). If the Eos were preincubated with the neutralizing anti-₁ antibody before stimulation with TS2/16, adhesion was suppressed by 67%, further supporting a potential role for ₁ integrins in mediating adhesion of Eos to HBECs.

To further characterize the ligands involved in the ₂ integrin interactions, we used antibodies _L (CD11a; LFA-1), _M (CD11b; Mac-1), and _X (CD11c), as well as antibodies to ICAM-1. The results are summarized in Figure 8. Adhesion to IFN--activated HBECs (in which ₂ integrin utilization predominated) was most effectively blocked by antibodies to CD11b for both PMN and Eos. Both cell types showed a partial dependence on CD11a and no significant contribution from CD11c. Combination studies done with multiple antibodies confirmed that adhesion could be completely blocked by a combination of anti-CD11a and anti-CD11b antibodies.

View larger version (18K): [in this window] [in a new window]   View larger version (14K): [in this window] [in a new window]   Figure 8.   Effect of blocking antibodies to ICAM-1 and to  subunits of 2 integrins on Eo and PMN adhesion to HBECs. Adhesion of C5a-stimulated PMN (A) or Eos (B) to IFN--treated HBECs was measured in the presence of blocking antibodies to indicated adhesion molecules. Where indicated, antibodies to individual  subunits were combined to determine their additive effect. Data are expressed as the percent suppression of leukocyte adhesion in the absence of inhibitory antibodies. Results are the mean ± SEM of three to five experiments (*P < 0.05).

To minimize granulocyte adhesion through Fc interactions, we used either the Fab₂ fragment of the anti-ICAM-1 antibody R6.5, or an intact anti-ICAM-1 antibody coupled with a Fab fragment of antimouse-Fc to block exposed Fc of the primary anti-ICAM-1 antibody. Adhesion was blocked to the extent of approximately 50% by Fab₂ treatment directed against ICAM-1, again with no discernible difference between Eos and PMN. Under identical treatment conditions, PMN adhesion to human umbilical vein endothelial cells (HUVECs), which has previously been described as being primarily ICAM-1-dependent (19, 29), was also suppressed by approximately 50% by treatment with the anti-ICAM-1 Fab₂ fragment (data not shown). This suggests that the increase in ICAM-1 on epithelial cells in response to IFN- reflects ICAM-1 as an important, if not the principal, ligand for integrin-mediated leukocyte adhesion to HBECs.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We found that adhesion of PMN and Eos to cultured epithelial cells can be differentially induced by a number of mediators. IL-8 was found to be a PMN-specific stimulus for adhesion, whereas IL-5 showed Eo specificity (Figure 5). C5a and GM-CSF shared activities on both cell types, whereas IL-3, a cytokine that has been shown through several other parameters to activate Eos, failed to affect adhesion of either PMN or Eos to cultured epithelial cells. We also investigated several inflammatory cytokines for their potential proadhesive effects on epithelial cells (Figure 1). In accord with previous reports (23, 61), we found that epithelial cells expressed ICAM-1 but not VCAM-1, and that IFN- was the most effective stimulus of those we tested for inducing ICAM-1 expression. TNF- and IL-1 caused only moderate increases in ICAM-1 expression, whereas IL-4 was without activity on HBECs. This contrasts with the effects of these cytokines on endothelial cells, which express both ICAM-1 and VCAM-1, and which respond strongly to IL-1, TNF-, and IL-4, but are essentially unresponsive to IFN- (23, 62). Therefore, the signals that may promote leukocyte adhesion are not only different for PMN and Eos, but also for the endothelium of the vasculature and epithelial tissue cells in the airway.

Despite the prominent increases in expression of ICAM-1 by epithelial cells in response to IFN- and TNF-, activation of HBECs only weakly enhanced adhesion of unstimulated granulocytes. This suggested that additional signals, leading to the activation of leukocytes themselves, were required for significant adhesion. We therefore further characterized the mechanism of leukocyte adhesion to resting and stimulated epithelial cells under conditions of leukocyte activation by C5a (a common Eo/PMN stimulus). Although expression of ICAM-1 in response to TNF- was demonstrably less marked than it was in response to IFN-, TNF- was equally effective as a costimulatory signal for leukocyte adhesion. These results suggest that TNF- may contribute to leukocyte-epithelium adhesion through effects independent of ICAM-1, perhaps by the induction of paracrine signals that lead to the activation of leukocytes. Alternatively, TNF- may induce expression of ICAM-1 in an active state, whereas IFN- treatment induces a high level of expression of ICAM-1 in a latent or low-affinity form. This is only a conceptual hypothesis, however, since there is no current evidence that ICAM-1 can exist in different affinity states.

Under all treatment conditions, adhesion of PMN depended exclusively on ₂ integrins, primarily _M2 (Mac-1). Interestingly, adhesion-molecule utilization by Eos depended to a great degree on the activation state of the epithelium. Adhesion of Eos to unstimulated HBECs involved both ₄₁ (VLA-4) and ₂ integrins (again, principally Mac-1). The relative contribution of VLA-4, however, was markedly diminished when epithelial cells were prestimulated with IFN-. Blocking studies with Fab₂ fragments directed against ICAM-1 or with intact antibody to ICAM-1 coupled with Fab-anti-Fc antibody fragments confirmed that ICAM-1 represented the principal counterligand for ₂ integrins in these studies. Although treatment directed against ICAM-1 suppressed adhesion by only approximately 50% in these experiments, we were also unable to obtain complete suppression of PMN adhesion to HUVECs with these treatment protocols. The failure to more fully block adhesion probably results from the relatively low affinity of Fab fragments as compared with intact antibodies. Use of Fab fragments in these studies was necessary to prevent granulocyte binding to HBECs through Fc receptor interaction with the exposed Fc regions of the blocking antibodies. We cannot exclude, however, that interactions between ₂ integrins and epithelial ligands other than ICAM-1 may contribute to adhesion between granulocytes and HBECs. Our data suggesting that ICAM-1 is the principal counterligand for ₂ integrins on epithelial cells are in agreement with the findings in several previous studies (23, 63). Although other reports have suggested an ICAM-1-independent mechanism of adhesion between PMN and epithelial cells (66), no alternative ligand for ₂ integrins has yet been identified on epithelial cells.

Most previous studies have failed to fully explore the potential roles of ₁, ₇, or different  integrin subunits on PMN and Eo adhesion to airway epithelium. There is general agreement that PMN adhesion is a predominantly ₂ integrin-dependent event. Studies by Stark and colleagues demonstrated a particularly strong _L (CD11a) component for PMN adhesion, whereas the same antibodies to _L that were used to show this component were ineffective in blocking Eo adhesion (69). In their studies, Stark and colleagues did not investigate the contribution of _M (CD11b). Our studies suggest a predominant dependence on _M, although some inhibition of adhesion was also demonstrated with antibodies to _L. In two previous studies of Eo adhesion to epithelial cell monolayers, antibody blockade of CD18 (₂ integrins) was found to be much more effective in suppressing adhesion to activated epithelial cells (80-90% suppression) than in blocking adhesion to unstimulated epithelial cells (50-60% suppression) (66, 69). This is in agreement with our current findings. Our investigation of ₁ and ₄ (VLA-4) integrin interactions explains this observation by demonstrating that Eo adhesion to resting epithelium occurs through both ₂ and ₁ (apparently VLA-4) interactions. In addition to associating with the ₁ subunit, ₄ may also associate with ₇ integrin and mediate adhesion through interaction with MadCAM. Although eosinophils express ₇ integrins, blocking antibodies to this subunit failed to inhibit eosinophil adhesion, suggesting that the role of ₄ integrins is restricted to their association with ₁ (VLA-4). The identity of the counterligand for VLA-4 on epithelial cells remains unknown, since epithelial cells do not express VCAM-1. Because VLA-4 is also a known ligand for matrix proteins such as fibronectin, eosinophil adhesion to HBECs may be mediated indirectly through extracellular matrix proteins produced by cultured epithelial cells (10, 35).

The present study was the first comprehensive comparison of Eo and PMN adhesion to nontransformed human epithelial cells. As described earlier, previous studies have investigated either Eo or PMN adhesion in a variety of systems. Many of these previous studies used either transformed epithelial cell lines or nonhuman cells (63, 69, 70), which may not accurately reflect leukocyte adhesion to normal human epithelium. Studies by Tosi and coworkers have demonstrated marked differences in adhesion of PMN to primary human epithelial cells and virally transformed human epithelial cells (67, 68). In addition, proadhesive stimuli have generally not been characterized, and nonphysiologic stimuli such as phorbol esters were often used in studies of PMN and Eos adhesion, which may not reflect signaling pathways of natural ligands (66, 69). We chose to study a number of immunologically relevant stimuli on the basis of their association with acute or chronic airway disease. C5a and IL-8 have been established as important chemotactic signals in airway disease (3, 56), and IL-8 is a prominent product of activated epithelial cells (9, 21, 49). Levels of IL-3 and IL-5 have been correlated with allergic and asthmatic airway disease, and have defined effects on Eo production, differentiation, and activation (37, 42). GM-CSF is also a prominent product of activated airway epithelium, and has been shown to enhance Eo survival (50, 71, 72).

Our findings that enhancement of PMN adhesion with C5a, IL-8, or GM-CSF, and of Eo adhesion with C5a, IL-5, or GM-CSF, further support the roles of these mediators in inflammatory disease. Our finding that leukocyte activation promotes adhesion is in general agreement with those in most previous studies of Eo and PMN adhesion to epithelial- and endothelial-cell cultures. However, these findings are in direct contrast to those in one previous study that showed no enhancement of PMN adhesion to stimulated primary human epithelial cell cultures by the formyl peptide formylmethionylleucylphenylalanine (64).

The levels of PMN and Eo adhesion to epithelial cell cultures in our study were comparable regardless of the activation state of the epithelial cells. These results suggest that restricted expression or utilization of adhesion molecules on epithelial cells cannot account for leukocyte specificity in different inflammatory conditions. However, there are alternative mechanisms through which epithelial cells may promote leukocyte-specific accumulation in airway tissue. Our results suggest that epithelial-cell production of IL-8 and GM-CSF may favor PMN adhesion and may therefore contribute to PMN accumulation in the airway lumen. Eo accumulation in the airway might also be promoted through HBEC-derived GM-CSF or RANTES (9, 48). We have recently found that IL-8 and RANTES may be inversely regulated and released preferentially by epithelial cells, depending on the stimulus employed (M. Jagels, unpublished observations). Epithelial cells may therefore contribute to cell-specific recruitment though differential release of Eo-specific or PMN-specific chemoattractants. By defining the proadhesive and chemotactic signals and adhesive interactions utilized by epithelial cells and leukocytes under pseudophysiologic conditions, we may begin to understand the molecular mechanisms underlying selective leukocyte recruitment, survival, and leukocyte-mediated injury of mucosal epithelium in inflammation.