Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The influx of neutrophils into the airway is a common feature of several inflammatory diseases, including acute bronchitis, adult respiratory distress syndrome, and chronic obstructive pulmonary disease (1, 2). Increased numbers of neutrophils have also been recovered from the airways of patients with severe asthma, and of subjects undergoing acute asthma exacerbations (3, 4).

To be recovered in bronchoalveolar lavage fluid during such diseases, neutrophils must have migrated from the circulation to the airway lumen. To accomplish this, neutrophils must first adhere to, and migrate through, the vascular endothelium into the interstitial tissue in response to a chemotactic gradient. Cells must then undergo a similar sequence of events to cross the airway epithelium and enter the airway lumen. In the past decade, many of the events involved in endothelial adherence and transmigration have been delineated. It is known that the adherence of the endothelium can be increased upon exposure to a range of stimuli including interleukin (IL)-1, tumor necrosis factor (TNF)-, and endotoxin (5). In response to a spe-cific chemoattractant gradient, neutrophils initially adhere weakly to the endothelium in a process referred to as tethering or rolling. This weak adherence is dependent largely upon the class of adhesion molecules known as selectins (6). Subsequent firm adhesion of neutrophils to the endothelium occurs and is due primarily to the interaction of ₂-integrins on the neutrophil surface with intercellular adhesion molecule (ICAM)-1 on the endothelium (7). Finally, the neutrophil crawls toward intercellular junctions between adjacent endothelial cells and migrates through these junctions in a process that uses a combination of molecules, including platelet-endothelial adhesion molecule-1 (CD31) and ₂-integrins (8).

Although the mechanisms involved in transendothelial migration of neutrophils are well established, much less is known regarding the migration of neutrophils across the epithelial barrier. Presumably, the chemotactic gradient required to induce transepithelial migration must be generated either by other cells already present in the airway lumen, or else by the epithelial cell itself. It is clear that the epithelial cell is a rich source of chemoattractant molecules, including chemokines (9). In particular, IL-8, a chemokine that is a potent chemoattractant for, and activator of, neutrophils, can be produced in large amounts by epithelial cells in response to a variety of stimuli, including proinflammatory cytokines, bacteria, fungi, and respiratory viruses (10). It seems likely that transepithelial migration of neutrophils will show several unique characteristics. Not only does the spectrum of adhesion molecules expressed by epithelial and endothelial cells show marked differences (14, 15), but neutrophils must cross the epithelial barrier in a basolateral-to-luminal direction, in contrast to the apical-to-basolateral direction operative in transendothelial migration.

Many of the initial studies of the mechanisms of neutrophil transepithelial migration used cells grown on top of permeable supports and examined migration in the apical-to-basolateral direction (16). Given the relatively polarized distribution of some epithelial cell adhesion molecules, this may not adequately reflect the in vivo situation (14). Indeed, recent studies, using either a human tumor cell line with epithelial cell characteristics (19), or a type II alveolar carcinoma cell line (20), have demonstrated that neutrophil migration is more efficient in the basolateral-to-luminal direction. In the present study, we sought to develop a model of neutrophil transepithelial migration using a transformed normal human bronchial epithelial cell line (16HBE) grown at an air-liquid interface. This cell line has been shown to be able to form polarized monolayers with intact tight junctions and, when grown at an air-liquid interface, can develop cilia (21). We evaluated whether neutrophil transepithelial migration in this model is dependent upon the electrical resistance of epithelial monolayers. We also examined the selectivity of chemoattractant stimuli in inducing neutrophil transmigration, and the effects of cytokine stimulation of the epithelial monolayer on neutrophil transmigration. The role of selected adhesion molecules in the transmigration process was also examined.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials and Reagents

The following materials were purchased: Dulbecco's minimal essential medium (DMEM), L-glutamine, and penicillin/streptomycin/fungizone from Biofluids, Inc. (Rockville, MD); fetal bovine serum (FBS) from GIBCO BRL (Grand Island, NY); piperazine- N,N'-bis(2-ethane sulfonic acid) (Pipes), formyl-methionine-leucine-phenylalanine (FMLP), and Type VII rat-tail collagen from Sigma Chemical Co. (St. Louis, MO); Hanks' balanced salt solution from Collaborative Research (Bedford, MA); Millicell 12 mm (1.13 cm²) polycarbonate membranes with 3-µm pores from Millipore (Bedford, MA); percoll from Pharmacia (Uppsala, Sweden); IL-8 and eotaxin from Peprotech (Rocky Hill, NJ); regulated on activation, normal T cell expressed and secreted (RANTES) from R&D (Minneapolis, MN); platelet-activating factor (PAF)- C₁₆ from Bachem (Torrance, CA); and leukotriene (LT) B₄ from Biomol (Plymouth Meeting, PA). Antibody to IL-8 was generated by immunization of rabbits with recombinant human IL-8. For inhibition experiments, a purified immunoglobulin (Ig) G fraction was used. Equal concentrations of preimmune IgG were used as a control.

PAG buffer contained 25 mM Pipes, 100 mM NaCl, 5 mM KCl, 0.1% D-glucose, and 0.003% human serum albumin, pH 7.4. PAGCM was PAG containing 1 mM MgCl₂ and 1 mM CaCl₂.

Neutrophil Purification

Neutrophils were isolated from ethylenediaminetetraacetic acid- anticoagulated blood from normal volunteers by percoll density gradient centrifugation and hypotonic lysis of erythrocytes. Cell count and viability were assessed using erythrocin B dye exclusion. The viability of neutrophils was always greater than 95%. Neutrophils were then labeled with ⁵¹Cr as described previously (5) and suspended in PAGCM at a concentration of 1.25 × 10⁶/ml.

Epithelial Cell Culture

The simian virus 40-transformed, immortalized 16HBE bronchial epithelial cell line (21) was a generous gift of Dr. D. Gruenert (University of California, San Francisco, CA). Cells were routinely cultured to confluence on collagen-coated T-75 flasks (Costar, Cambridge, MA) in 100% humidity and 5% CO₂ at 37°C. They were grown in DMEM containing 10% FBS, 2 mM L-glutamine, penicillin (100 U/ml), streptomycin (100 U/ml), and fungizone (250 ng/ml). The cells were used between passages 12 and 35.

For experiments, cells were grown on Millicell inserts (3-µm pore size) in 12-well plates. The inserts were placed inside the wells in the usual manner to create a well within a well. Then 4.5 ml of medium was added to each of the 12 wells. This completely submerged the polycarbonate insert. Using a glass pipette tip, the insert was then inverted, ensuring that no air bubbles were trapped under the insert, and exactly 1.7 ml was removed from each well. Each insert was carefully moved to the center of its well to avoid an overlapping meniscus at the edge of the well. Cells were suspended at 5 × 10⁶ cells/ml and each inverted insert was gently seeded with two sequential aliquots of 50 µl (a total of 0.5 × 10⁶ cells/insert). Plates were placed in an incubator at 37°C and 5% CO₂ and cells were grown on the micropore membrane with air above and medium below. Inserts were used for experiments typically at 6 to 8 d, when they were returned to their original orientation for measurement of electrical resistance and for transmigration experiments. A diagrammatic representation of this culture system is shown in Figure 1.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Culture model for the study of neutrophil migration across an epithelial layer in the physiologically relevant basolateral-to-luminal direction.

Measurement of Resistance

Transepithelial resistance was measured using an EVOM epithelial volt-ohmmeter (World Precision Instruments, Sarasota, FL). Measurements were made after standardizing the resistance in 10- and 100-mM solutions of KCl.

The inserts were removed from the 12-well plates and put into individual wells of 24-well plates, with the cell monolayer now on the lower face of the insert. Medium (0.6 ml) was put into each well outside of the insert, and a further aliquot (0.4 ml) of medium was put into the inner compartment formed by each of the inserts. These volumes created an identical fluid level in the inner and outer wells (and no hydrostatic pressure gradient).

The electrodes (STX-2; World Precision Instruments) had a chopstick design with one tip 2.5 mm longer than the other. This allowed the longer electrode to be inserted into one of the wells of the 24-well plate (outer well) and the shorter electrode inside the millicell insert (the inner well), without touching the epithelial monolayers. The electrodes were then separated by the membrane and the epithelial monolayer.

The relationship between electrical resistance and confluence was examined by light microscopy after staining membranes of differing resistances using a Wright's stain.

Transepithelial Migration Assay

After measurement of transepithelial resistance, the monolayers were removed from medium and inverted onto absorbent paper. They were then gently immersed in PAG to remove any remaining medium. Each monolayer was numbered so that its resistance was documented. Each experimental point (spontaneous and stimulated) was performed in triplicate.

The chemoattractant was made up at the desired concentration in PAGCM. The wells of a 24-well plate had 0.6 ml of chemoattractant solution added. Each insert was filled with 0.4 ml of ⁵¹Cr-labeled neutrophil suspension (5 × 10⁵ cells). The fluid levels in the inner and outer compartments were identical, avoiding a hydrostatic gradient. The 24-well plates were then incubated at 37°C.

After the appropriate length of time, cells that had migrated to the outer well were lysed using 2 M NH₄OH. The level of radioactivity from each sample was measured in a  counter, and the percent migration was calculated as follows: (counts per min [cpm] of migrated cells)/(cpm of total cells added to upper well) × 100. Specific migration was defined as: (% migration in presence of stimulus)  (% migration in absence of stimulus).

Cytokine Activation of the Epithelium

We (22) and others (15) have previously shown that incubation of epithelial cells with the combination of TNF- and interferon (IFN)- markedly enhances adhesion molecule expression. For the current studies, epithelial monolayers were immersed in medium containing TNF- (10 U/ml) and IFN- (30 U/ml) for 18 h. Resistance was measured after this treatment. The human recombinant TNF- and IFN- used each had a specific activity of 1 × 10⁷ U/mg and were obtained from Genzyme Diagnostics (Cambridge, MA).

Adhesion Molecule Inhibition Studies

The role of adhesion molecules on neutrophils in transepithelial migration was investigated by incubating [⁵¹Cr]-labeled neutrophils in a volume of 30 µl of PAGCM (3.5 × 10⁶ cells/ml) with appropriate blocking monoclonal antibodies (mAbs) to selected leukocyte adhesion molecules, or with a class-matched control antibody, at 4°C for 30 min. Antibodies were used at 10 times the concentration that, when incubated with neutrophils, gave a maximal fluorescence signal by flow cytometry. Cells were subsequently diluted with PAGCM to 1.25 × 10⁶ cells/ml and added to the epithelial layers as in previous experiments.

For investigations of the role of ICAM-1 on the epithelium, a F(ab')₂ fragment of a blocking antibody to ICAM-1 was used instead of an intact antibody to avoid interactions with Fc receptors on the neutrophil surface. Monolayers of 16HBE cells were placed in the orientation used for transmigration studies and incubated with or without mAb in PAGCM for 30 min at 37°C. Antibody was placed in both the inner and outer wells at volumes of 0.2 and 0.3 ml, respectively, and was aspirated before use of the monolayers in the transmigration assay. Antibody was used at 10 times the concentration that, when incubated with epithelial cells, gave a maximal fluorescence signal by flow cytometry.

All mAbs to leukocyte adhesion molecules used in our studies have previously been shown to block leukocyte adhesion to appropriate counterligands, either on endothelial cells or purified proteins immobilized on plates. mAbs against leukocyte integrins used were: H52, an IgG₁ antibody to CD18; MHM24, an IgG₁ antibody to CD11a; HC2, an IgG₁ antibody to CD11b (all obtained from Dr. J. Hildreth, Johns Hopkins University, Baltimore, MD); BU15, and IgG₁ antibody to CD11c (Immunotech, Westbrook, ME); and 240I, a recently developed blocking antibody to CD11d that has been shown to block CD11d-mediated binding of eosinophils to immobilized vascular cell adhesion molecule (VCAM)-1 (23). These antibodies were generously provided by Dr. Bruce Bochner (Johns Hopkins University). For incubation with epithelial monolayers, MEM-111, a blocking IgG_2a antibody to ICAM-1, and TÜ149, an IgG_2a antibody to HLA-1, were purchased from Caltag Laboratories (San Francisco, CA).

Statistics

All data are presented as means ± standard error of the mean. Time- and concentration-dependence of migration was assessed using one-way repeated measures analysis of variance. Student's two-tailed t test was used to calculate the statistical significance of differences between groups.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The relationship between the resistance of 16HBE monolayers and confluence was studied using light microscopy of stained membranes of different resistances. Monolayers with resistances significantly < 250  rarely covered the whole of the membrane. By contrast, cell cultures with resistances of > 250  completely covered the membrane. After 6 d, the resistance of monolayers was usually in the range of 250 to 700 . With further incubation, however, resistances continued to increase, reaching values up to 2,000 . There were no further obvious visual differences when monolayers of higher resistance were stained and examined.

To examine the relationship between resistance and transmigration, labeled neutrophils were added to the basolateral sides of monolayers of differing resistances and the migration of neutrophils in the absence of any chemotactic stimulus, or in the presence of 100 ng/ml of IL-8, in a 2-h period was measured. Data are expressed in two ways. In the first set of experiments, percent migration to IL-8 was plotted against resistance (Figure 2). As can be seen, an exponential curve was seen demonstrating a negative relationship between resistance and migration. Spearman rank correlation analysis indicated that resistance was negatively correlated with percent migration (rho = 0.73, P = 0.0001). In additional experiments, we also compared migration to buffer and to IL-8 in membranes with different ranges of resistances. As shown in Figure 3, subconfluent membranes (< 250 ) permitted high spontaneous migration of neutrophils (7.4 ± 1%), and extremely high migration of neutrophils in response to IL-8. By contrast, confluent monolayers (resistance 250 to 700 ) permitted low spontaneous migration (2 ± 0.5%) but showed a 6-fold enhancement of neutrophil migration to IL-8 (12.4 ± 1.3%). At levels of resistance above 700 , transmigration was markedly reduced. Given that it became more difficult to measure migration accurately using membranes with resistances > 700 , and that such high resistances are not observed in vivo or with cultured monolayers of primary human epithelial cells, all subsequent experiments were performed using inserts with resistances between 250 and 700 .

View larger version (16K): [in this window] [in a new window]   Figure 2.   Relationship between transepithelial migration of neutrophils in response to IL-8 and the electrical resistance of epithelial monolayers. The electrical resistance of polycarbonate membranes (n = 93) containing subconfluent (< 250  resistance) or confluent monolayers of 16HBE cells were recorded, and the migration, in a 2-h period, of neutrophils across these monolayers in response to 100 ng/ml of IL-8 was determined.

View larger version (29K): [in this window] [in a new window]   Figure 3.   The dependence of spontaneous and IL-8-induced neutrophil transepithelial cell migration on electrical resistance of the monolayers. Migration was determined over a 2-h incubation period. For spontaneous migration, buffer was placed in the lower chamber on the apical side of the epithelial monolayer. Migration to IL-8 (100 ng/ml) was determined using membranes with similar resistances. Membranes were used when monolayers were subconfluent, having a resistance < 250  (n = 24); when resistances were in the range of 250 to 700  (n = 42); and when resistances were > 700  (n = 36).

Neutrophil transmigration to 100 ng/ml of IL-8 occurred in a time-dependent manner (P < 0.05) over a 3-h period (Figure 4). The greatest rate of migration occurred within the first 2 h, perhaps reflecting gradual equilibration of the chemotactic gradient. Given that migration at 2 h was approximately 80% of that seen at 3 h, we chose to use 2 h as our routine incubation time. Migration to IL-8 at 2 h was dose-dependent (Figure 5) over a concentration range from 30 to 300 ng/ml. Moreover, this dose-response curve was shifted significantly (P < 0.05) to the left when the epithelial monolayer was preincubated with a combination of TNF- and IFN-. Migration to IL-8 was dependent upon the creation of a chemotactic gradient, because no specific transmigration was observed when IL-8 was preincubated with a specific blocking antibody (specific migration above buffer control = 5.4 ± 1.6% in the presence of preimmune IgG and 0.3 ± 0.4 in the presence of anti-IL-8; n = 3). Migration was also inhibited when equal concentrations of IL-8 were added to both sides of the epithelial monolayers (specific migration above buffer control = 4.6 ± 0.6% with IL-8 only below the monolayer and 0.2 ± 0.4 with IL-8 on both sides of the monolayer, n = 4).

View larger version (12K): [in this window] [in a new window]   Figure 4.   Time course of transepithelial migration of neutrophils across epithelial monolayers in response to buffer and to 100 ng/ml of IL-8. All monolayers used had resistances in the range of 250 to 700 . Four complete time courses were performed, with each time point for spontaneous and IL-8- induced migration being performed in triplicate in each experiment.

View larger version (16K): [in this window] [in a new window]   Figure 5.   Effects of cytokine pretreatment of epithelial monolayers on transepithelial migration of neutrophils. In each of six experiments, epithelial monolayers were preincubated with or without a combination of TNF- and IFN- as described in MATERIALS AND METHODS. Migration to buffer or to 100 ng/ml of IL-8 was then monitored over a period of 2 h.

Transepithelial migration of neutrophils was stimulus-specific. In addition to IL-8, pronounced migration was observed in response to either LTB₄ or FMLP (10⁷ M), whereas PAF (10⁷ M) was an ineffective stimulus for neutrophil transmigration. Moreover, no transmigration was observed with the C-C chemokines eotaxin or RANTES at a final concentration of 100 ng/ml (Figure 6).

View larger version (46K): [in this window] [in a new window]   Figure 6.   Specificity of chemotactic stimuli in inducing transepithelial migration of neutrophils. Data are expressed as specific migration, as defined in MATERIALS AND METHODS. The number of experiments (n) with each stimulus is shown above the appropriate histogram column.

Preincubation of the epithelial monolayers with antibody to ICAM-1 reduced transepithelial migration by an average of 70% (Figure 7a). Although analysis of the raw data did not indicate statistical significance by t test, this was due to the variability in baseline migration between the experiments. In each of three experiments, however, migration was reduced in the presence of anti-ICAM-1 (P < 0.05 for paired comparison of normalized data). Experiments also revealed that transmigration was dependent, at least in part, on neutrophil expression of ₂-integrins (Figure 7b). Preincubation of neutrophils with antibody to CD18, the common  chain of these integrins, reduced transepithelial migration by approximately 50% (P < 0.05 versus control). To determine which individual ₂-integrins may serve as counterligands for adhesion, we examined the effects of antibodies to individual -chains of these molecules (Figure 8). Blockade of either CD11a or CD11b inhibited transmigration by approximately 35% in each case. More modest, but statistically significant (P < 0.05 versus control), inhibition was also noted with antibody to CD11c. By contrast, antibody to CD11d did not inhibit transepithelial migration of neutrophils.

View larger version (17K): [in this window] [in a new window]   Figure 7.   Dependence of transepithelial migration of neutrophils on ICAM-1 and CD18. (a) In each of three experiments, epithelial cell monolayers were incubated with F(ab')2 fragments of either anti-ICAM-1 or class-matched control antibody as described in MATERIALS AND METHODS. Specific migration (above buffer control) to IL-8 was then determined over a 2-h incubation period. (b) In each of six experiments, neutrophils were preincubated with monoclonal anti-CD18 or a class-matched control as described in MATERIALS AND METHODS. Specific migration to IL-8 was then determined over a 2-h incubation period.

View larger version (50K): [in this window] [in a new window]   Figure 8.   Effects of specific antibodies to individual -chains of 2-integrins on transepithelial migration of neutrophils. In each of three matched experiments, neutrophils were incubated with antibodies to CD11a, CD11b, CD11c, CD11d, or a class-matched control. Specific migration (above buffer control) to IL-8 was then determined over a 2-h incubation period.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The complex mechanisms by which neutrophils exit the circulation to reach sites of inflammation have been extensively studied in recent years. In several disease states, however, neutrophils that have traversed the endothelial barrier subsequently migrate across the epithelium to reach the airway lumen. Initial studies of the mechanisms regulating transepithelial migration used monolayers of various epithelial cell lines grown on permeable supports but monitored neutrophil migration in the nonphysiologic apical-to-basolateral direction (16). More recent studies using epithelial carcinoma cell lines, however, have demonstrated that migration occurs primarily in the physiologic basolateral-to-luminal direction (19, 20). This finding is not unexpected, given that many macromolecules, including some adhesion molecules, are distributed in a polarized fashion to apical versus basolateral epithelial membranes (14, 24).

For our current studies, we used a transformed normal human bronchial epithelial cell line that forms polarized monolayers with intact tight junctions to develop a model to study further the mechanisms of transepithelial migration in the basolateral-to-luminal direction. By seeding cells on the lower face of culture well inserts that had been inverted, and feeding at a liquid-air interface until confluence was achieved, the inserts could then be returned to their proper orientation, allowing neutrophils added to the wells to migrate in the physiologic direction. Part of our goal was to be able to develop a system in which a simple measure of electrical resistance could be used to monitor the confluence and integrity of monolayers on each individual culture insert for use in migration studies. To accomplish this, we performed initial studies using membranes of differing resistances using a Wright's stain. We found that resistances of > 250  were associated with intact monolayers, assessed by staining, and resulted in markedly lower levels of "spontaneous" neutrophil transmigration compared with monolayers with resistances of < 250 . We also found a clear inverse relationship between electrical resistance and migration such that, at very high electrical resistances, little transmigration occurred. It has previously been suggested that the degree of transepithelial migration was largely independent of the resistance of the monolayer, but this was based upon a comparison of different cell lines with different inherent resistances (18).

Obviously, migration across different cell lines could also be affected by other parameters, including differential expression of adhesion molecules. To our knowledge, our current studies provide the first direct analysis of the relationship between resistance and neutrophil transmigration in a single airway epithelial cell population. Although surgical specimens of human bronchi show an epithelial resistance of only 100 /cm² (25), monolayers of primary bronchial epithelial cells show a mean resistance of approximately 450 /cm² (26). It has been suggested that this increased resistance in vitro is caused by geometric changes that lead to a decrease in the length of tight junctions per unit area of cell membrane (26). It is unclear why electrical resistances in the 16HBE cells continue to increase with extra time in culture, inasmuch as no obvious morphologic changes were noted by light microscopy, but there is precedent for some primary human epithelial cell monolayers to attain resistances up to 1,800 /cm² (26). Given our preliminary transmigration data, we chose to restrict our subsequent studies to monolayers with resistances between 250 and 700 , a range centered around the mean resistance for primary human cell monolayers. Using these guidelines, our current model provides a simple way to exclude individual inserts in which cells have not achieved confluence or else have achieved supraphysiologic resistances.

Using this model system, migration of human neutrophils across epithelial monolayers showed both time- and concentration-dependence in response to IL-8. This migration was dependent upon the existence of a chemotactic gradient, because when IL-8 was preincubated with blocking antibody, or when equal concentrations of IL-8 were added to both sides of the membrane, the level of migration observed was not different than that observed to buffer control. It was of interest to note that a mean 4-fold increase in neutrophil transmigration was seen to IL-8 in subconfluent monolayers compared with 6-fold in monolayers within our optimal range. This relatively robust response to IL-8 in subconfluent monolayers was surprising, given that IL-8 gradients should equilibrate more rapidly in such cultures. The response in subconfluent monolayers, however, also showed greater individual variability.

Other known chemoattractants for neutrophils, including FMLP and LTB₄, were also potent inducers of neutrophil transepithelial migration, whereas the C-C chemokines eotaxin and RANTES, which are mainly chemotactic for eosinophils, monocytes, and some T cells (27), did not cause neutrophil transmigration. Interestingly, although PAF is known to be a potent activator of neutrophils (28) and provocation with PAF can increase airway neutrophilia (29), this lipid was not a stimulus for neutrophil transepithelial migration in our model. These data are in good agreement, however, with those of Liu and coworkers, who also found PAF to be a poor stimulus for inducing neutrophil transepithelial migration (19).

Cytokines such as TNF- and IFN- are found in increased amounts in the airways during inflammatory diseases (30). Both of these cytokines are known to stimulate epithelial cell responses, including increasing surface expression of ICAM-1 (15, 22). Stimulation of epithelial monolayers with the combination of these cytokines in our model system significantly enhanced neutrophil transmigration to IL-8, clearly demonstrating that the epithelial monolayer is an active participant in the transmigration process.

The participation of the epithelial monolayer in neutrophil transmigration is mediated, in part, by the adhesion molecule ICAM-1, inasmuch as preincubation with a specific blocking antibody to this molecule markedly reduced transmigration. These observations contrast with those of Liu and colleagues, who failed to demonstrate any inhibition of migration across a lung carcinoma cell line with epithelial characteristics when the monolayer was preincubated with an antibody to ICAM-1 (19). This may reflect our use of a normal bronchial epithelial cell population rather than a carcinoma cell population. It is feasible that the role of ICAM-1 in cell migration could depend upon both the stimulus and cell type under study. Thus, while both cell types express ICAM-1, the membrane distribution (e.g., apical versus basolateral) and role in adhesion may vary. Alternatively, the difference between our observations and those of Liu and colleagues (19) could be explained by different mAbs to ICAM-1, or differences in protocol. Neutrophils could adhere to epithelial cells treated with normal, intact anti-ICAM-1 via interactions with the Fc receptors expressed on their surface. Liu and colleagues (19) attempted to prevent such interactions by preincubating neutrophils with anti-FcRIIIb and anti-FcRIIIa, whereas we completely avoided the possibility of such interactions by using an F(ab')₂ fragment of anti-ICAM-1. Our data indicating a role of ICAM-1 in neutrophil transepithelial migration is supported by studies showing that neutrophil adherence to epithelial cells is at least partially mediated by ICAM-1 (31, 32), and by the demonstration that in vivo administration of either antisense oligonucleotides or mAbs to ICAM-1 inhibited neutrophil recruitment to the airways in a murine model of endotoxin-induced neutrophilia (33).

The neutrophil counterligands for ICAM-1 are members of the ₂-integrin family of adhesion molecules. When neutrophils were preincubated with the antibody against CD18, the common -chain of this family of adhesion molecules, transmigration was inhibited by approximately 50%. Antibodies to the individual -chains of the ₂-integrins yielded variable results. Marked inhibition of transmigration was seen with antibodies to CD11a and CD11b. Inhibition of transmigration by anti-CD11b is consistent with the data of Liu and associates (19), but these authors did not observe inhibition by CD11a. Again, this difference could be explained by variations in the epithelial cell preparations used. Our data are consistent, however, with studies supporting a role of CD11a in neutrophil adhesion to epithelial cells (32, 34). Although both CD11a/CD18 (also called lymphocyte function-associated antigen-1) and CD11b/CD-18 (also called Mac-1) appear to contribute to neutrophil transepithelial migration in our model, our data also provide the first evidence for a role for CD11c/CD18. We also tested, for the first time, the effect of the recently discovered ₂-integrin CD11d/CD18 on transepithelial migration of neutrophils. This molecule, more commonly referred to as d₂, is expressed on most types of human leukocytes, including neutrophils, and was originally shown to bind preferentially to ICAM-3 as a counterligand (35). More recently, d₂ has been shown also to be an alternative ligand for VCAM-1 (23). The lack of effect of antibody to d₂ in our model system would imply that neither ICAM-3 nor VCAM-1 plays a role in ₂-integrin-dependent neutrophil transepithelial migration.

It is of interest that inhibition produced by anti-ICAM-1 antibodies appears greater in our studies than those produced by antibodies to ₂-integrins, which appear to inhibit about 50% of migration. This may in part be technical, in that baseline migration (in the absence of antibody) was lower in the anti-ICAM-1 experiments, and it is often easier to markedly inhibit modest responses compared with robust responses. We must also consider alternative possibilities, however. These data could be taken to suggest that there are ICAM-1-dependent, but CD11/CD18-independent, adhesion pathways. Alternatively, CD11/CD18 expression on the surface of neutrophils may be increased during transmigration in membrane areas that are not accessible to the soluble antibody. Further studies will be necessary to resolve this issue.

In summary, we have developed a model in which 16HBE cells can be cultured in a fashion to study transepithelial migration of leukocytes in the physiologically relevant basolateral-to-luminal direction. The suitability of monolayers on each insert for experimentation can be easily determined by measuring electrical resistance. The migration of neutrophils across suitable monolayers is dependent upon the presence of a chemotactic gradient of a relevant neutrophil chemoattractant and is mediated, in part, by interactions between ICAM-1 on the epithelial cells and appropriate ₂-integrin counterligands on the neutrophil. It must be noted, however, that antibodies to neither ICAM-1 nor ₂-integrins completely prevented neutrophil transmigration, even though they were used at levels 10-fold higher than those needed to provide saturation of their respective ligands (as determined by flow cytometry), indicating that additional ligands on both cell types must also play a role. The model system described here should prove useful in permitting further studies aimed at identifying these additional ligands and at identifying the steps in the migration process in which specific adhesion molecules play a role.