Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Migration of leukocytes through vascular endothelium involves a multistage process in which the cell first becomes captured by the endothelium as it flows through the postcapillary venule. This event is mediated largely by P- and E-selectin on the endothelium and L-selectin on the leukocyte binding to counter-structures that include mucin-like receptors in which O-linked sugars are expressed on a serine threonine-rich polypeptide backbone (1). The 4 integrins 47 and 41 have also been shown to capture cells under flow conditions (2). Once captured, the cell becomes loosely tethered to the endothelial cell surface and rolls along it. It may then become detached or, alternatively, arrest, flatten, and transmigrate. The arrest phase is mediated by 1 and 2 integrins binding to endothelial- expressed counter-receptors that are members of the immunoglobulin family and include intercellular adhesion molecule (ICAM)-1, ICAM-2, and vascular cell adhesion molecule (VCAM)-1 (3). An activation step is required for the integrins to become functional. The activation signal is thought to be supplied by chemoattractants, expressed on the endothelial surface, bound either to specific receptors or the proteoglycan-rich endothelial glycocalyx (4). Each step in the migration cascade is required for successful migration to occur. As a result, there is potential for considerable diversity in the combination of signals that could direct migration and this may be responsible, in part, for the varied pattern of leukocyte accumulation seen in different inflammatory processes (5). The evidence for this model of migration is extensive and derived from several sources. These include in vitro observations of patterns of leukocyte binding to cultured endothelial cells and purified adhesion proteins under flow conditions (8, 9), direct observation of leukocyte traffic in animal models using intravital microscopy (10), and studies using monoclonal antibodies (mAbs) or gene-deleted mice to disable one or more adhesion receptors in the context of an inflammatory stimulus (11, 12).

Although the involvement of selectins and integrins in the migration process appears secure, there is still some uncertainty about the nature of the activation signal. Most work on the activation step has been undertaken with neutrophils in which the 2 "leukocyte" integrins are the predominant, if not the only, integrin receptors mediating adhesion to endothelium. The 2 integrins on neutrophils require an activation signal to become functional, and this is effectively supplied by low molecular-weight chemoattractants such as formylmethionyl leucylphenylalanine and platelet-activating factor (PAF) as well as the chemokine interleukin (IL)-8 (reviewed in 13). Consistent with this observation, integrin-mediated neutrophil adhesion is inhibited by pertussis toxin (PT), which inactivates the  subunit of the Gi protein coupled to seven-transmembrane chemoattractant receptors (14). However, less work has been undertaken on the activation step in other leukocytes. In lymphocytes there is evidence that nonchemoattractant mediators such as the cytokine hepatocyte growth factor could mediate the activation step (15).

There is increasing evidence, for example, in the field of lymphocyte homing, that selective expression and function of adhesion receptors can control patterns of leukocyte migration (16, 17). Another area in which this idea has received considerable attention in recent years is the mechanism by which the selective tissue accumulation of eosinophils occurs in diseases such as asthma (18). Eosinophils are end-stage myeloid cells, related ontogenically to basophils, which in humans are normally very few in number in both blood and tissue (19). There is considerable circumstantial evidence pointing toward eosinophil-derived mediators being responsible for the pathologic abnormalities associated with asthma and related allergic diseases (20). Eosinophils are present in increased numbers in the airways of asthmatics, despite only a small increase in the numbers of neutrophils, suggesting a relatively selective pathway of migration. This may be due to different patterns of adhesion receptor usage compared with neutrophils, specific chemoattractants, or prolonged survival of eosinophils under the influence of locally generated specific growth factors such as IL-5. Human peripheral blood eosinophils, unlike neutrophils, constitutively express very late antigen (VLA)-4 (21), and antibodies against this receptor have been effective at inhibiting eosinophil migration into the lung in animal models of asthma (22). Particular interest has been generated recently by the identification of a number of C-C chemokines that are chemotaxins for eosinophils but not neutrophils (23). Evidence is emerging that several of these chemokines are expressed in eosinophilic inflammation (24). Most eosinophil-active chemokines signal through the C-C chemokine receptor (CCR)-3, which is expressed predominantly by eosinophils and is therefore an ideal target for selective inhibition of eosinophil migration (27).

Although there have been a number of studies detailing the expression of adhesion receptors and inflammatory mediators in allergic diseases such asthma (28), there is relatively little functional data in the context of clinical disease. In an attempt to define the events relevant to eosinophil adhesion in chronic airway inflammation, we have in recent years adapted the Stamper-Woodruff frozen-section assay (FSA), which has been used extensively to study lymphocyte homing (31), to study eosinophil and neutrophil binding to nasal polyp endothelium (NPE) as a model of eosinophilic airway inflammation. We have previously shown that eosinophils bind effectively and specifically to nasal polyp blood vessels and that this adhesion is almost completely inhibited by mAbs against P-selectin and P-selectin glycoprotein ligand (PSGL)-1 (32). We have also demonstrated that eosinophils bind more avidly than do neutrophils to both NPE and P-selectin under flow and express a structurally and functionally distinct isoform of PSGL-1 (33). This observation was recently supported by Kitayama and colleagues (34), who also found that eosinophils bind with greater avidity to P-selectin than do neutrophils under flow; although, using similar techniques, Patel and McEver found no difference in binding to P-selectin between eosinophils and neutrophils (35).

The purpose of the present study was to determine whether eosinophil and neutrophil binding in our assay was integrin- and activation-dependent consistent with the multistep model for adhesion described previously. We have found that for both cell types, successful adhesion required functionally competent 2 (and in the case of eosinophils, 1) integrins as well as an activation step. We found that, whereas neutrophil adhesion was mediated by PAF and IL-8 through PT-sensitive, G protein-linked receptors, eosinophil adhesion was inhibited neither by PT, a PAF antagonist, nor mAbs against CCR-3, IL-3, IL-5, or granulocyte macrophage colony-stimulating factor (GM-CSF).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

mAbs

The following blocking anti-integrin mAbs were used: anti-CD18 clones 7E4, AZN-L18, AZN-L27, CLB-LFA-1 (Sixth Human Leukocyte Differentiation Antigen [HLDA] Workshop, Kobe, Japan; 1996), MHM23 (Dako Ltd., High Wycombe, UK), and L130 (Becton Dickinson Ltd., Oxford, UK); anti-CD11a clones CD11a-5E4 (Sixth HLDA Workshop and a generous gift of Dr. W. Knapp, University of Vienna, Vienna, Austria), AZN-L20, AZN-L21 (Sixth HLDA Workshop), 38 (Cymbus Bioscience Ltd., Southampton, UK), and SPV-L7 (Bradsure Biologicals Ltd., Loughborough, UK); anti-CD11b clones 44 (Cymbus Bioscience Ltd.) and 2LPM19c (Dako Ltd.); anti-CD11c clones 3.9 (Cymbus Bioscience Ltd.) and KB90 (Dako Ltd.); anti-CD29 clone P4C10 (Life Technologies Ltd., Paisley, Scotland); anti-CD49d clone HP2/1 and anti-CD49f clone 4F10 (Serotec Ltd., Oxford, UK); anti-ICAM-2, clone BT-1 (Serotec); anti-47, Act-1 (a generous gift of LeukoSite, Inc., Cambridge, MA); and anti-VCAM-1, clone 4B9 (kindly donated by Roy Lobb of Biogen, Cambridge, MA).

The anti-IL-8 mAb (A5.12.14) and the anti-IL-8 receptor A and B mAbs (clones 9H1.5.1 and 10H2.12.1, respectively) were generous gifts from Genentech, Inc., San Francisco, CA. The anti-eotaxin receptor clone 7B11 (27) was kindly donated by LeukoSite, Inc. Anti-IL-3 and anti-GM-CSF mAbs were from Genzyme (Kent, UK). The anti-IL-5 mAb (clone mAb7) was a generous gift from Amanda Proudfoot of Glaxo Wellcome, Geneva, Switzerland.

Reagents

The PAF antagonist WEB 2086 was a generous gift from Dr. Peggy Ganong (Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT). Control mouse myeloma proteins MOPC (immunoglobulin [Ig]G1, IgG2a, and IgG2b, mixed in equal proportions), PT, and sodium azide were purchased from Sigma Chemical Co. (Poole, Dorset, UK). PAF was supplied by Bachem Ltd. (Saffron Walden, UK). Recombinant human (rh) GM-CSF and rh eotaxin were obtained from R&D Systems Europe Ltd. (Abingdon, Oxford, UK).

Preparation of Eosinophils and Neutrophils

Eosinophils were isolated from 100 ml blood from normal volunteers who showed only mild or no atopic symptoms (median starting count, 0.47 × 10⁶ eosinophils/ml of blood) and who were not taking medication at the time of venesection. Purification was as described previously (32) by density gradient centrifugation and negative immunomagnetic selection using the magnetic-activated cell separation (MACS) system. Neutrophils were isolated from nonatopic volunteers in a similar manner to eosinophils, omitting the final immunomagnetic selection step. Eosinophils and neutrophils were isolated from different donors. Purity and viability of both cell types was > 95%.

FSA

Nasal polyps were obtained after routine surgery in the Ear, Nose, and Throat Department (Leicester Royal Infirmary, Leicester, UK). They were received within 30 min of removal, washed briefly in phosphate-buffered saline (PBS), dissected, and snap-frozen in liquid N₂. All tissue samples were stored in vapor-phase liquid N₂ until required. The FSA was carried out as described previously (32). Briefly, isolated leukocytes were resuspended at 5 × 10⁶/ml in Medium 199 (M199) containing 25 mM N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid and L-glutamine (Life Technologies Ltd.) with 2% fetal bovine serum (FBS), treated where appropriate, and layered over 8-µm nasal polyp sections (NPS). mAbs against cell-expressed molecules were incubated with the leukocytes for 15 min at room temperature (RT) before being added to the NPS. mAbs against endothelial-expressed molecules were diluted in M199 and incubated on the NPS for 30 min at RT, then the solution was tipped off, leukocytes were added, and the FSA continued. All mAbs were used at saturating concentrations. Slides were rotated at 70 rpm (setting 5) on a "Belly Dancer" shaker (Scotlab, Lanarkshire, Scotland) for 30 min at RT. Unbound cells were tipped off and slides were fixed in 2% glutaraldehyde and then washed in PBS. May-Grunwald-Giemsa staining was used to visualize bound cells. Leukocyte adhesion was assessed on blinded slides by counting the number of blood vessels that bound two or more cells from 100 vessels located randomly within each section and calculated as the percentage of blood vessels that bound cells. Leukocytes added experimentally were easily distinguished from leukocytes present in nasal polyp tissue, as they could be seen to be on different planes of the tissue. Each condition was carried out on at least triplicate sections. For the majority of experiments, different leukocyte donors and polyps were used. In total, 37 eosinophil donors, 32 neutrophil donors, and polyps from 34 subjects were used. Data have generally been presented as percentage binding. However, only a limited number of conditions can be undertaken on each experiment. The inherent variability of the assay has led us to express some of the data as percent inhibition, for reasons of clarity.

Chemotaxis

Eosinophils were resuspended at 5 × 10⁶/ml in M199/2% FBS as before. They were incubated with either WEB 2086 (10⁵ M, 15 min, RT), anti-CCR-3, clone 7B11 (10 µg/ml, 15 min, RT), or PT (100 ng/ml, 2 h, 37°C) before being used in the assay. All mAbs/reagents were diluted in M199/2% FBS. Solutions of PAF (10⁶ M) in ethanol were evaporated over nitrogen and redissolved in the same buffer. Eotaxin was also prepared by dilution in this buffer at concentrations of 1 to 100 nM. PAF and eotaxin as positive controls for chemotaxis, and buffer as a negative control, were added to the lower wells of a 48-well microchemotaxis chamber (Polyfiltronics Ltd., Middlesex, UK) and covered with an 8-µm nitrocellulose membrane (Sartorius Instruments Ltd., Belmont, Surrey, UK). For each experiment, at least three replicates of each condition were performed. Appropriately treated cells were added to the upper chamber and allowed to migrate for 90 min at 37°C. The chambers were placed in a humidified box to minimize evaporation. The membrane was removed after incubation, washed, fixed, and stained with hematoxylin to allow visualization of migrated cells. The membrane was then mounted and left overnight at 4°C before counting. Results were expressed as the number of cells migrating to the underside of the filter per 10 high-power fields.

Immunohistochemistry

NPS, 6 µm, were air-dried, fixed in 100% acetone (10 min at RT), and then immunostained using the streptavidin- biotin, alkaline phosphatase method, as recommended by the manufacturer (Dako Ltd.). A New Fuchsin substrate kit (Biomen Ltd., Berkshire, UK) was used to visualize immunocomplexes, and sections were counterstained with Mayer's hematoxylin (Sigma Chemical Co.).

Statistics

Statistical comparisons were undertaken using a standard Student's t test; differences with P < 0.05 were considered significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

As we have previously shown, eosinophils and neutrophils bound selectively to blood vessels with little background binding to stromal tissues. Ethyleneglycol-bis-(-aminoethyl ether)-N,N'-tetraacetic acid (EGTA) almost completely abrogated adhesion for both cell types (Figure 1). Again, as previously reported, eosinophils bound in greater numbers to NPE than did neutrophils, with approximately twice as many blood vessels supporting eosinophil adhesion.

View larger version (104K): [in this window] [in a new window]   Figure 1.   Eosinophils and neutrophils bind specifically to NPE. The FSA was carried out as shown in MATERIALS AND METHODS. Representative example of the pattern of eosinophil and neutrophil binding with (C and D) and without (A and B) EGTA. A and C: eosinophils; B and D: neutrophils. Arrows denote cells bound.

We have undertaken a detailed examination of the role of CD18 integrins in eosinophil and neutrophil binding using a large panel of mAbs available either commercially or through the Sixth HLDA Workshop. A total of 15 blocking mAbs were investigated. Figures 2 and 3 show results for selected antibodies that are representative of our findings. Using a panel of six mAbs directed against CD18, we observed consistent inhibition of both eosinophil (range, 35 to 52%) and neutrophil (range, 48 to 78%) adhesion (Figure 2 and data not shown). Consistent patterns of inhibition were also observed with mAbs against the individual CD11 integrin chains. In the case of eosinophils, inhibition was seen with antibodies against all three integrins, although for anti-CD11c this did not reach significance (Figure 3 and data not shown). In contrast, neutrophil adhesion was not inhibited by anti-CD11a mAbs.

View larger version (25K): [in this window] [in a new window]   Figure 2.   Eosinophil and neutrophil adhesion to NPE is CD18-dependent. Eosinophils and neutrophils from different donors were purified (see MATERIALS AND METHODS) and incubated for 15 min at RT with either anti-CD18 mAb clone MHM23, control antibodies, or EGTA-treated assay medium before use in the FSA, also described in MATERIALS AND METHODS. Filled columns, eosinophils, n = 5; hatched columns, neutrophils, n = 6. Values shown indicate means ± standard error of the mean (SEM). *P < 0.05, **P < 0.01.

View larger version (58K): [in this window] [in a new window]   Figure 3.   Contribution of individual CD11 integrins to eosinophil and neutrophil binding to NPE. Assay conditions were as described in Figure 2. Antibodies used were CD11a (AZN-L21); CD11b (2LPM-19c); and CD11c (3.9). Filled columns, eosinophils, n = 3-7; hatched columns, neutrophils, n = 3-5. Values represent the mean percentage inhibition of cell adhesion compared with control antibodies ± SEM. *P < 0.05.

We noted that the percent of inhibition of cell adhesion by anti-CD18 was significantly greater for neutrophils (67.7%) than for eosinophils (43.9%) (P < 0.025 for mAb MHM23). This raised the possibility of other integrins being involved in eosinophil adhesion. We therefore examined the effects of a mAb directed against CD29. Anti-CD29 gave consistent inhibition of eosinophil adhesion (23.6%, P < 0.0005), although this was considerably less than for anti-CD18. Inhibition with a combination of anti-CD18 and anti-CD29 mAb was significantly greater than anti-CD18 alone (87.1%, P < 0.025), and brought binding almost down to levels observed with EGTA (Figure 4). No effect of the anti-CD29 mAb was observed on neutrophil adhesion. To investigate which of the 1 integrins were involved in eosinophil adhesion, we used mAbs against 4, 6, and VCAM-1. Anti-4 and anti-VCAM-1 caused inhibition of adhesion that was similar in degree to that observed with anti-CD29 (Figure 5). No effect was observed with the anti-6 mAb.

View larger version (28K): [in this window] [in a new window]   Figure 4.   Eosinophil, but not neutrophil, adhesion to NPE is mediated by 1 integrins. Assay conditions were as described in Figure 2. mAbs used were CD18 (2, clone MHM23) and CD29 (1, clone P4C10). Filled columns, eosinophils, n = 4-7; hatched columns, neutrophils, n = 3. Data shown indicate mean values ± SEM. *P < 0.025, **P < 0.005, ***P < 0.0005.

View larger version (46K): [in this window] [in a new window]   Figure 5.   Eosinophil adhesion to NPE uses CD49d and VCAM-1 but not CD49f. Assay conditions were as described in MATERIALS AND METHODS. Antibodies used were: CD18 (2, clone MHM23), CD49d (4, clone HP2/1), CD49f (6, clone 4F10), and VCAM-1 (clone 4B9). Filled columns, eosinophils, n = 2-8. Values represent the mean percentage inhibition of cell adhesion compared with control antibodies ± SEM. *P < 0.01, ***P < 0.0005.

Integrin function is dependent on cell activation. We therefore investigated whether either activation before adding the cells to the tissue or inhibition of activation during the assay had any effect. Incubation for 30 min at 37°C with GM-CSF (10⁹ M) had no effect on either eosinophil or neutrophil adhesion. In contrast, treatment of both cell types with azide (0.65%; 0.1 M) inhibited adhesion by approximately 50% (Figure 6). We investigated the possibility that cell activation was occurring through PT-sensitive serpentine receptors. PT completely inhibited neutrophil adhesion but had no effect on eosinophil adhesion (Figure 7). Consistent with this observation, a well- established PAF antagonist (WEB 2086) and blocking mAbs against IL-8RA and IL-8RB each inhibited neutrophil adhesion (Figure 8) with no effect on eosinophil adhesion (Figure 9). An anti-IL-8 mAb also inhibited neutrophil adhesion by a mean of 46% (mouse IgG control, 19.61 ± 1.22; % blood vessel versus anti-IL-8 mAb, 10.6 ± 1.16%; n = 3, P < 0.05). Combining WEB 2086 with the two anti- IL-8R antibodies had no additional effect for either cell type (data not shown). In addition, a blocking mAb against CCR-3 failed to inhibit eosinophil adhesion (Figure 9) despite specifically inhibiting chemotaxis to eotaxin (data not shown). Similarly, PT inhibited eosinophil chemotaxis to both PAF and eotaxin, and WEB 2086 inhibited eosinophil chemotaxis to PAF (data not shown). Finally, blocking mAbs against the eosinophil growth factors IL-5, IL-3, and GM-CSF had no effect on eosinophil adhesion in our assay (Figure 10).

View larger version (41K): [in this window] [in a new window]   Figure 6.   Eosinophil and neutrophil adhesion to NPE is activation-dependent. Eosinophils and neutrophils from different donors were purified and incubated as follows before use in the FSA (see MATERIALS AND METHODS): GM-CSF, 109 M, 30 min, 37°C; azide, 0.65% final concentration, 30 min, 37°C. Filled columns, eosinophils, n = 3-4; hatched columns, neutrophils, n = 4- 8. Values represent means ± SEM. *P < 0.01, **P < 0.0005. Incubation at 37°C resulted in a decrease in baseline eosinophil, but not neutrophil, adhesion. All eosinophil donors had < 0.4 × 106 eosinophils/ml blood.

View larger version (41K): [in this window] [in a new window]   Figure 7.   Neutrophils, but not eosinophils, signal via a PT-sensitive, seven-transmembrane linked receptor. Eosinophils and neutrophils from different donors were isolated as described in MATERIALS AND METHODS and incubated with PT (100 ng/ml) for 2 h at 37°C before use in the FSA (see MATERIALS AND METHODS). Filled columns, eosinophils, n = 4; hatched columns, neutrophils, n = 3. Values represent means ± SEM. *P < 0.01, **P < 0.005. Eosinophil donors were as described in Figure 6.

View larger version (28K): [in this window] [in a new window]   Figure 8.   Neutrophil adhesion to NPE is mediated by PAF and IL-8. Cell purification and FSA conditions were as described in MATERIALS AND METHODS. In addition, the PAF antagonist WEB 2086 was incubated with the cells at 105 M, 15 min, RT. mAbs used: IL-8RA (9111.5.1); IL-8RB (10H2.12.1). Hatched columns, neutrophils, n = 5-6. Values represent means ± SEM. *P < 0.025, **P < 0.005. Eosinophil donors were as described in Figure 6.

View larger version (63K): [in this window] [in a new window]   Figure 9.   Eosinophil adhesion to NPE is not mediated by PAF, IL-8, or eotaxin. mAbs used: IL-8RA (9111.5.1); IL-8RB (10H2.12.1); CCR-3 (anti-eotaxin receptor, 7B11). WEB 2086 is a PAF antagonist. Assay conditions were as described in MATERIALS AND METHODS and Figure 8. Filled columns, eosinophils, n = 3-5. Values represent means ± SEM. ***P < 0.0005. Eosinophil donors were as described in Figure 6.

View larger version (47K): [in this window] [in a new window]   Figure 10.   Eosinophil adhesion to NPE is not mediated by IL-3, IL-5, or GM-CSF. mAbs used were IL-3, GM-CSF (Genzyme), and IL-5 (clone mAb7). Assay conditions were as described in MATERIALS AND METHODS. Filled columns, eosinophils, n = 3. Values represent means ± SEM. Eosinophil donors were as described in Figure 6.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We have demonstrated that eosinophil and neutrophil adhesion to NPE is dependent on CD18 integrins. Eosinophil adhesion seemed to be largely mediated by CD11a and b, whereas neutrophil adhesion was inhibited by antibodies against CD11b and c. Previous studies using cytokine-stimulated human umbilical vein endothelial cells (HUVEC) have also demonstrated that eosinophil and neutrophil adhesion and transmigration are dependent on CD18 integrins (36). In these studies LFA-1 and Mac-1 appeared to be involved for both cell types. However, we found that whereas eosinophil adhesion was consistently inhibited by anti-LFA-1 mAbs, no inhibition was observed for neutrophils using the same antibodies. This may be related to the state of activation of these receptors or to the counter-structure being recognized on the endothelium.

We have not identified the counter-receptor on NPE recognized by the CD18 integrins. ICAM-1 is expressed on NPE, but we previously reported lack of inhibition using mAbs against ICAM-1 that inhibit both the Mac-1 and LFA-1 binding sites (39). We were also unable to inhibit adhesion with an anti-ICAM-2 mAb (data not shown). It seems possible, therefore, that the principal ligand for the 2 integrins expressed by inflamed airway endothelium remains to be determined.

Inhibition using the six anti-CD18 mAbs averaged 44% for eosinophils and 68% for neutrophils. We considered the possibility that this difference was due to the ability of eosinophils also to use 1 integrins and in particular VLA-4. Eosinophils can use VLA-4/VCAM-1 to bind to and migrate through venular endothelium, and there is a considerable body of evidence from animal models suggesting that VLA-4/VCAM-1 interactions may be important in controlling eosinophil accumulation in allergic disease (40). In this study, using an anti-CD29 antibody, we observed reliable though modest inhibition of eosinophil (but not neutrophil) adhesion. The lack of an effect of the anti-CD29 mAb on human peripheral blood neutrophils is consistent with their lack of expression of 1 integrins under physiologic conditions. Overall, similar degrees of inhibition to the CD29 mAb were observed with antibodies against 4 and VCAM-1, although these results were less consistent than those observed with anti-CD29, suggesting that the 1 component of eosinophil adhesion to NPE was through VLA-4/VCAM-1. No inhibition was seen with antibodies against 47 (not shown) or 6.

On both eosinophils and neutrophils, the CD18 integrins need to be activated through outside-in signaling to be able to bind to their counter-structures. We therefore explored the importance of cell activation in our assay. The requirement of an activation step for eosinophil adhesion was suggested by our previously reported observation that eosinophil adhesion was temperature-dependent (32). Optimal adhesion was observed at RT with a small fall in adhesion at 37°C and an approximately 50% reduction in adhesion at 4°C. To avoid any effects of in vivo priming, we used normal donors for our activation experiments for both eosinophils and neutrophils. In vitro priming with GM-CSF, although previously reported as upregulating CD18-dependent adhesion (43), had no effect on either eosinophil or neutrophil binding in our assay. Similarly, preactivation with the C-C chemokine regulated on activation, normal T cell expressed and secreted (RANTES) had no effect on adhesion of either cell type to NPE (data not shown). In contrast, the metabolic inhibitor azide significantly blocked adhesion for both cell types, offering further support for the idea that cell activation occurring during the assay is required for effective adhesion. The inhibitory action of azide led us to explore the effects of PT, which inhibits the function of the majority of known chemoattractant seven-transmembrane Gi-linked receptors. A striking difference was observed between eosinophils and neutrophils, with complete inhibition of neutrophil adhesion but no effect on eosinophil adhesion. The agents involved in neutrophil activation were further defined using the PAF antagonist WEB 2086 and anti-IL-8R mAbs. Neutrophil adhesion was inhibited by WEB 2086 and anti-IL-8RA and -8RB, consistent with previous studies in which both PAF and IL-8 were involved in neutrophil activation during adhesion to endothelium (44). Variability was observed in the amount of inhibition with the PAF antagonist and the anti-C-X-C receptor mAbs, suggesting that for a given polyp and blood donor the cells were predominantly being activated by either IL-8 or PAF. The inhibitory effect of the anti-IL-8 mAb confirmed the involvement of this cytokine, but does not exclude a role for other C-X-C chemokines. In addition, other chemoattractants such as leukotriene B₄ or C5a may be involved in neutrophil activation because PT was more effective at blocking adhesion than the combined effect of WEB 2086 and the anti-IL-8R antibodies. The polyp tissue is the likely source of PAF and IL-8, although we cannot be sure that PAF and IL-8 are being presented to the neutrophil expressed on the surface of the endothelium. It is possible that other cell types in the polyp, such as T cells, fibroblasts, endothelial cells, mast cells, epithelial cells, and resident eosinophils, are releasing chemotaxins into the assay medium during the experiment. IL-8 in particular has been shown to be generated in substantial quantities by nasal epithelium from patients with rhinitis (47). Indeed, using the anti-IL-8 mAb we found rather diffuse staining throughout the polyp tissue (not shown).

Eosinophil adhesion was unaffected either by PT, the PAF antagonist, the anti-IL8R antibodies or an antibody against CCR-3 that is the major chemokine receptor so far identified on peripheral blood eosinophils. This was despite PT inhibiting chemotaxis to PAF and eotaxin, and WEB 2086 and the anti-CCR-3 mAb inhibiting chemotaxis to PAF and eotaxin, respectively. Nonetheless, the involvement of CD18 integrins, which require activation to be functional, together with the inhibitory effect of azide and reduced binding at 4°C, each suggest that an activation step was occurring.

PAF is an effective though nonspecific chemoattractant for eosinophils and has been shown to enhance CD18- dependent adhesion to HUVEC (48, 49). IL-8 has not been shown to be active on resting peripheral blood eosinophils, so it is not surprising that no effect was seen with the IL-8R antibodies (50). However, there is increasing literature demonstrating that several C-C chemokines, signaling primarily through the CCR-3 receptor, are effective and often highly specific eosinophil chemoattractants that are expressed in eosinophilic inflammation, including nasal polyps (51, 52). RANTES, C5a, and macrophage chemotactic protein-3, acting through a PT-sensitive signaling pathway, were able to enhance adhesion of eosinophils to purified VCAM-1 via VLA-4 and purified ICAM-1 via Mac-1 (53). In contrast, eotaxin (but not RANTES) was able to enhance adhesion to endothelial cells via a VLA-4-dependent pathway (54). It is therefore not clear why eosinophil activation was not mediated by nasal polyp- derived chemoattractants including chemokines. The role of C-C chemokines in disease may therefore be predominantly to direct eosinophil chemotaxis to the mucosal surface rather than to mediate the activation step in endothelial adhesion. It is possible that chemokines that can signal though the CCR-1 receptor are playing a part in eosinophil activation in our assay. However, we feel this is unlikely because CCR-1 almost certainly signals through a PT-sensitive receptor. Kitayama and colleagues recently demonstrated a modest reduction in adhesion of eosinophils to HUVEC under shear flow conditions with a mAb against the CCR-3 receptor (55). This is in contrast to what we found in our system using the same anti-CCR-3 mAb. However, the effect Kitayama and coworkers observed required a combination of tumor necrosis factor- and interferon-, whereas IL-4, a cytokine more relevant to allergic inflammation, did not result in C-C chemokine production by HUVEC. The recent identification of the CX₃C subfamily of chemokines, which can enhance adhesion through a non-PT-sensitive mechanism (56), means that it remains possible that a chemokine is triggering the activation step. Eosinophil growth factors, in particular IL-5, GM-CSF, and IL-3, are highly effective at enhancing CD18-dependent eosinophil adhesion (57). However, we were unable to detect any inhibition using antibodies against these cytokines in the FSA.

In summary, we have demonstrated that eosinophil and neutrophil adhesion to NPE in the ex vivo FSA conforms to the multistep paradigm for leukocyte adhesion that has been defined largely in animal models and in vitro adhesion assays. We used the assay to characterize the molecular basis for granulocyte adhesion in a model of chronic airway inflammation. We found that for neutrophil binding to NPE, P-selectin mediates the selectin step, CD11b-c the integrin step, and PAF and IL-8 (possibly with additional chemotaxins) the activation step. Using this model we have described potentially important differences between eosinophil and neutrophil adhesion pathways. These include expression by eosinophils of an isoform of PSGL-1 that binds P-selectin with greater avidity than neutrophils, the preferential use of CD11a by eosinophils, the additional use of VLA-4 by eosinophils but not neutrophils, and, perhaps most strikingly and unexpectedly, a difference in the nature of the activation signal between these two cell types. To this must be added another step: the likelihood that eosinophil-specific chemokines stimulate selective chemotaxis of eosinophils into the tissue. Each of these differences has potential implications for the treatment of eosinophilic diseases (such as asthma) by selective blockade of eosinophil migration. In addition, we have defined two major uncertainties in the adhesion cascade: the nature of the CD18 counter-structure on NPE recognized by eosinophils and neutrophils, and the identity of the signal-mediating eosinophil activation.