INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

It is widely acknowledged that the eosinophil is one of the main effector cells in asthma. Eosinophils from peripheral blood do not only infiltrate into the central airways, but also infiltrate into peripheral lung tissue of asthmatic patients (1). During this infiltrating process eosinophils become activated, which can be deduced from positive staining for antibody EG2 in bronchoalveolar lavage (BAL) and airway wall biopsies (2, 4) as well as increased CD11b expression and decreased L-selectin expression in BAL (5). Cells that may be involved in activation of eosinophils include endothelial cells, epithelial cells, and fibroblasts, the latter being the prime cell in lung connective tissue. Owen and coworkers were the first to suggest that fibroblasts may contribute to eosinophil activation (leukotriene C₄ [LTC₄] production, capacity to kill Schistosoma mansoni larvae) and may increase eosinophil viability, using 3T3 fibroblasts and supplementary granulocyte/macrophage colony-stimulating factor (GM-CSF) (8, 9). Similar results were found when eosinophils were cultured with interleukin-5 (IL-5) and 3T3 fibroblasts (10). Other studies have confirmed the findings on eosinophil survival using human pulmonary fibroblasts without addition of supplementary cytokines (11, 12), suggesting a role for fibroblast-derived cytokines. Studies with lung fibroblast-conditioned medium showed GM-CSF to be a factor that increases eosinophil viability (11).

In addition to the soluble factors produced by fibroblasts, adherence of eosinophils to fibronectin, an extracellular matrix (ECM) protein produced by fibroblasts, may also contribute to the observed increased survival and activation of eosinophils (13). Furthermore, fibroblasts are able to express both intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) (17, 18). It has been shown that soluble ICAM-1 augments superoxide generation and eosinophil cationic protein (ECP) release by eosinophils (19, 20) and that very late antigen-4 (VLA-4)-mediated eosinophil adhesion to VCAM-1 stimulates superoxide generation (21). Recently, Yuan and colleagues (22) described a c-kit receptor on eosinophils to which fibroblast-derived stem cell factor (SCF) can bind in both soluble and membrane-bound form. This interaction leads to increased adhesion to fibronectin and VCAM-1 via activation of VLA-4 on eosinophils and may also be involved in activation of the eosinophils (22).

In this study we investigated activation of eosinophils in vitro by human lung fibroblasts, assessing eosinophil CD11b and L-selectin expression. Secondly, we investigated whether eosinophil activation is due to adhesive interactions between the two cell types or to cytokines secreted by fibroblasts.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subjects

Ten nonallergic volunteers were included for the isolation of peripheral blood eosinophils. Inclusion criteria were as follows: 18 to 40 yr of age; normal lung function; no airway obstruction; a total plasma immunoglobulin-E (IgE) < 100 IU/ml; no specific IgE against cat, dog, house dust mite, tree and grass pollen. All participants gave their informed consent and the study was approved by the medical ethics committee of the University Hospital Groningen.

Overall Study Design

Eosinophils were isolated from peripheral blood of nonallergic volunteers and cultured in vitro with confluent monolayers of human lung fibroblasts or fibroblast-conditioned medium. Fibroblasts were cultured with or without IL-1. Control eosinophils were cultured in medium alone. Neutralizing antibodies to GM-CSF and IL-8 were added to elucidate the contribution to eosinophil activation of these cytokines as produced by fibroblasts. After 4 h and 24 h of culture, we assessed eosinophil viability by propidium iodide (PI) exclusion and their activation by the expression of two adhesion molecules, CD11b and L-selectin, using flow cytometric analysis.

Eosinophil Isolation

Eosinophils were isolated from peripheral blood anticoagulated with ethylenediamine tetraacetic acid (EDTA) by negative selection with anti-CD16-magnetic beads using a method modified from Hansel and coworkers (23). Briefly, blood was diluted 1:1 with 0.9% NaCl solution and centrifuged (20 min, 1,000 × g) over isotonic Percoll (Pharmacia LKB, Uppsala, Sweden) with a density of 1.082 g/ml to separate granulocytes from the rest of the leukocytes. Granulocytes were collected from the bottom of the tube and were treated twice with cold ammonium chloride (155 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM Na₂-EDTA) to lyse contaminating erythrocytes. Cells were resuspended in 2% fetal calf serum (FCS; GIBCO BRL, Paisley, UK) in phosphate-buffered saline (PBS/2%) in a concentration of 10⁸/ml, and 100 µl of immunomagnetic beads (Miltenyi Biotec, Bergisch Gladbach, Germany) per milliliter cell suspension was added. For immunomagnetic separation, a MACS type CS column (Miltenyi Biotec) was used. Eluted fractions contained 98 ± 3% eosinophils with a viability of 98 ± 1%. Eosinophils were resuspended in RPMI complete medium: RPMI-1640 (GIBCO BRL) supplemented with 2 mM L-glutamine (BioWhittaker, Verviers, Belgium), 125 µg/ml streptomycin (Radium-farma- Fisiopharma, Milan, Italy), 125 U/ml penicillin (Yamanouchi Pharma, Leiderdorp, The Netherlands), and 10% FCS.

Lung Fibroblasts

Lung fibroblasts of a healthy individual (CCD8-Lu; American Type Culture Collection, Rockville, MD) were phenotypically characterized. Immunohistochemical staining showed that these cells were positive for vimentin and fibronectin and negative for keratin and desmin; 0-1% of the cells were alpha smooth muscle actin positive, indicating myofibroblasts. Thus, no contaminating epithelial, endothelial, or smooth muscle cells were present in the fibroblast cultures.

Lung fibroblasts were cultured in Ham's F12 medium (BioWhittaker) supplemented with 2 mM L-glutamine, streptomycin (125 µg/ml), penicillin (125 U/ml), and 10% FCS, hereafter referred to as Ham's complete medium. For the experiments, fibroblasts in passage 12 were used.

Culture Conditions

Fibroblasts grown in 24-well plates (Costar, Cambridge, MA) reached confluence and were either stimulated with 10 U/ml IL-1 (Boehringer Mannheim, Mannheim, Germany) in Ham's complete medium, or remained in Ham's complete medium alone for 4 h. The wells were subsequently washed twice with RPMI-1640, and 1.5 × 10  ⁵ isolated eosinophils in 1 ml RPMI complete medium were added and cocultured for 4 or 24 h.

For experiments with fibroblast-conditioned medium, confluent fibroblasts were cultured on 1 ml RPMI complete medium after the aforementioned stimulation and washing steps. After 4 or 24 h of culture, the conditioned medium was harvested and 1.5 × 10⁵ pelleted isolated eosinophils were resuspended in the fibroblast-conditioned medium and cultured for 4 h and 24 h, respectively.

Polyclonal anti-GM-CSF and polyclonal anti-IL-8 (R&D Systems, Abingdon, UK) in a concentration of 40 µg/ml were applied to neutralize GM-CSF and IL-8 in the cocultures and conditioned media before the eosinophils were added. These concentrations were able to inhibit at least 1,000 pg/ml GM-CSF and at least 40 ng/ml IL-8 according to the manufacturer's product information.

For insert experiments, fibroblasts were cultured to confluence in 12-well plates (Costar). After the aforementioned stimulation and washing procedures, 0.4-µm pore inserts (Costar) were applied to the wells and 3.0 × 10⁵ eosinophils were added on top of the inserts. The total volume of these cultures was 2 ml. In this way, direct contact between the two cell types was prevented, but diffusion of soluble factors excreted by fibroblasts and eosinophils was allowed. Additionally, it allowed us to determine possible cytokine degradation over time in fibroblast-conditioned medium.

Flow Cytometric Analysis of Adhesion Molecule Expression and Survival

After culture, eosinophils were harvested and the wells were washed once with 1 ml RPMI-1640. Harvested eosinophils were divided into four portions, centrifuged (5', 590 × g) and washed in 2 ml FACS buffer (PBS, 2% FCS, 0.01% wt/vol NaN₃). Samples for detection of viability were resuspended in 150 µl FACS buffer and stored on ice until analyzed. CD11b, L-selectin, and IgG2a (negative isotype control) labeling was performed with the three remaining portions of eosinophils.

Eosinophils were incubated for 30 min at 4° C with 10 µl anti-Leu15 (CD11b), anti-Leu8 (L-selectin) (Becton Dickinson, San Jose, CA), or IgG2a (CLB, Amsterdam, The Netherlands) as well as 10 µl AB serum (20% in FACS buffer) to avoid nonspecific binding. Samples were washed twice with cold FACS buffer and incubated for 15 min at 4° C in the dark with 4 µl goat antimouse-fluorescein isothiocyanate (GAM-FITC) (Becton Dickinson) and 46 µl 5% AB serum. Cells were washed once with FACS buffer and resuspended in 135 µl FACS buffer. Samples were kept on ice until FACS analysis and prior to analysis 15 µl of 10 µg/ml PI (Sigma Chemical Co., St. Louis, MO) was added to allow discrimination between viable and dead cells. Only viable cells were assayed for adhesion molecule expression, by excluding the PI-positive cells.

Fluorescence was measured on a FACStar (Becton Dickinson) after calibration with SPHERO Rainbow fluorescent particles (PharMingen, Dan Diego, CA). Data analysis was performed with WinList for Win 32 (Verity Software House, Inc., Topsham, ME). Mean fluorescent intensity (MFI) by control antibody was subtracted from MFI for CD11b and L-selectin expression, yielding specific mean fluorescence (SMF) values. SMF of eosinophils cultured in RPMI complete medium alone at 4 h and 24 h, respectively, was used as a standard (100%). Adhesion molecule expression after coculture was expressed as the percentage with respect to the standard value.

Cytokine Levels in Cultures

Cytokine levels in the various conditions were measured both to ensure 100% neutralization by the added antibodies and to establish possible correlations with the activated state of eosinophils. All cell-free media were stored at 80° C for later measurements of cytokine concentrations. This was done using a GM-CSF (R&D Systems) and an IL-8 (CLB) enzyme-linked immunosorbent assay (ELISA) kit.

Statistical Analysis

Results are presented as median cytokine concentrations, median percentages of adhesion molecule expression or viability, and minimal and maximal values between brackets. Results were tested for differences using the one-tailed Wilcoxon signed rank test for related samples. Differences were considered significant for p < 0.05. Correlations were tested with Spearman's correlation test.

The first set of experiments comprised six experiments with six donors, the second set of experiments studying insert conditions comprised four experiments with four donors.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Activation of Eosinophils in Coculture with Lung Fibroblasts

Eosinophils were activated by fibroblasts, as indicated by increased expression of CD11b and decreased expression of L-selectin compared with eosinophils cultured alone. Eosinophils cocultured with IL-1-stimulated fibroblasts expressed more CD11b and less L-selectin than did eosinophils cocultured with unstimulated fibroblasts at 4 and 24 h (p < 0.05) (Figures 1a and 1b, significant differences not indicated). Specifically, eosinophils cocultured with fibroblasts stimulated with IL-1 expressed more CD11b (154 [139-213]% at 4 h, 210 [156- 292]% at 24 h, p < 0.05; Figure 1a, and less L-selectin (59 [55- 86]% at 4 h, 35.5 [31-72]% at 24 h, p < 0.05; Figure 2a) than did eosinophils cultured in medium alone (control = 100%) at both 4 h and 24 h. However, when eosinophils were cocultured with unstimulated fibroblasts, there was no significant change in CD11b and L-selectin expression (Figures 1b and 2b), except for increased CD11b expression at 24 h (130 [111- 187]%). A time-dependent decrease in MFI was observed, reflecting absolute levels of CD11b and L-selectin expression on cultured eosinophils compared with freshly isolated eosinophils. In some activating conditions there was no actual increase in absolute CD11b levels, but a preserved CD11b level compared with a decreased level in cultured control eosinophils. In all activating conditions there was a decrease in absolute L-selectin levels compared with freshly isolated eosinophils.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Eosinophil CD11b expression after 4 and 24 h of culture in (a) IL-1-stimulated human lung fibroblast conditions and (b) unstimulated human lung fibroblast conditions. Results are expressed as median percentages (min-max) of CD11b expression (n = 6) with respect to control eosinophils cultured alone (E0, 100% expression). EF0 = coculture with resting fibroblasts; EFIL-1 = coculture with IL-1-stimulated fibroblasts; ECM0 = culture in resting fibroblast-conditioned medium; ECMIL-1 = culture in IL-1-stimulated fibroblast-conditioned medium.

View larger version (11K): [in this window] [in a new window]   Figure 2.   Eosinophil L-selectin expression after 4 and 24 h of culture in (a) IL-1-stimulated human lung fibroblast conditions and (b) unstimulated human lung fibroblast conditions. Results are expressed as median percentages (min-max) of L-selectin expression (n = 6) with respect to control eosinophils cultured alone (E0, 100% expression). For definition of abbreviations, see text and Figure 1.

Activation of Eosinophils by Fibroblast-conditioned Medium

When eosinophils were cultured in medium conditioned by fibroblasts stimulated with IL-1, they expressed more CD11b (155 [145-187]% at 4 h, 166 [131-210]% at 24 h, p < 0.05) and less L-selectin (57 [46-61]% at 4 h, 57 [27-72]% at 24 h, p < 0.05) at both time points than control eosinophils. After 4 h, the change in CD11b expression was as great as that seen in eosinophils in the corresponding (stimulated) coculture (Figure 1a), but the decrease in L-selectin expression was slightly greater (p < 0.05) (Figure 2a). After 24 h, eosinophils cultured in IL-1-stimulated fibroblast-conditioned medium expressed significantly less CD11b than did eosinophils in the corresponding coculture, though they expressed equivalent amounts of L-selectin. Eosinophils cultured in unstimulated fibroblast-conditioned medium did not change in CD11b expression after 4 h, neither in L-selectin expression after 4 and 24 h (Figures 1b and 2b), but an increase in CD11b expression was seen on eosinophils after 24 h. This increase in CD11b expression was smaller compared with the expression on eosinophils in the corresponding coculture (Figure 1b).

Figures 3a and 3b show that addition of anti-GM-CSF, but not anti-IL-8, to medium conditioned by IL-1-stimulated fibroblasts inhibited increased CD11b expression (155 [145- 187]% at 4 h, 166 [131-210]% at 24 h, p < 0.05) and decreased L-selectin expression (57 [46-61]% at 4 h, 57 [27-72]% at 24 h, p < 0.05) on eosinophils caused by IL-1-stimulated fibroblast-conditioned medium after 4 and 24 h (Figures 3a and 3b). The expression of CD11b (Figure 3a) and L-selectin (Figure 3b) on eosinophils after neutralization of GM-CSF was not different from the expression of control eosinophils (100%), except for the expression of L-selectin after 24 h, which was still decreased (84 [43-93]%, p < 0.05) (Figure 3b). In contrast, neither CD11b expression nor L-selectin shedding of eosinophils in coculture conditions was inhibited by anti-GM-CSF or anti-IL-8 (not shown). In the presence of anti-IL-8, eosinophils expressed significantly less L-selectin in conditioned medium of IL-1-stimulated fibroblasts at 24 h (Figure 3b). Eosinophils also expressed less L-selectin and more CD11b in coculture with unstimulated fibroblasts at 24 h and in coculture with stimulated fibroblasts at 4 h when anti-IL-8 was added (p < 0.05) (not shown). Adhesion molecule expressions of control eosinophils were not affected either by anti-IL-8 or anti-GM-CSF, except for a lower L-selectin expression induced by anti-GM-CSF after 4 h (p < 0.05) (not shown).

View larger version (13K): [in this window] [in a new window]   Figure 3.   The effect of anti-GM-CSF and anti-IL-8 in cultures with conditioned media of IL-1-stimulated fibroblasts (ECMIL-1) on eosinophil (a) CD11b and (b) L-selectin expression. Results are expressed as median percentages (min-max) (n = 6) of adhesion molecule expression with respect to control eosinophils cultured alone (E0, 100% expression).

Insert Experiments

In this set of experiments (n = 4), CD11b and L-selectin expression of eosinophils cultured under insert conditions were compared with expressions of eosinophils cocultured with fibroblasts, and cultured in conditioned medium. Eosinophils cocultured with IL-1-stimulated fibroblasts expressed significantly more CD11b (188 [168-249]% at 4 h, 246 [199-272]% at 24 h) than did eosinophils in the corresponding insert condition, i.e., 151 (139- 184)% at 4 h and 193 (171-234)% at 24 h, and conditioned medium, i.e., 182 (139-220)% at 4 h and 173 (169-214)% at 24 h (Figure 4a). CD11b expressions of eosinophils in the insert conditions were not different from the expressions in the corresponding conditioned media. Eosinophils cocultured with unstimulated fibroblasts for 4 h expressed the same levels of CD11b (134 [129-171]%) as eosinophils in corresponding insert condition (114 [96-133]%), but more than in corresponding conditioned medium (107 [93-137]%, p < 0.05) (Figure 4b). After 24 h, CD11b expression in coculture (195 [184-281]%) was higher than in insert condition (182 [133-240]%, p < 0.05) and conditioned medium (120 [114-199]%, p < 0.05), and higher in insert condition than in conditioned medium (p < 0.05).

View larger version (14K): [in this window] [in a new window]   Figure 4.   Eosinophil CD11b expression after 4 and 24 h of culture in (a) IL-1-stimulated human lung fibroblast conditions and (b) unstimulated human lung fibroblast conditions. Results are expressed as median percentages (min-max) (n = 4) of CD11b expression with respect to control eosinophils cultured alone (E0 = 100% expression). EF0 = coculture with resting fibroblasts; EFT0 = culture in insert in the presence of resting fibroblasts; ECM0 = culture in resting fibroblast-conditioned medium; EFIL-1 = coculture with IL-1-stimulated fibroblasts; EFTIL-1 = culture in insert in the presence of IL-1-stimulated fibroblasts; ECMIL-1 = culture in IL-1-stimulated fibroblast-conditioned medium.

L-selectin expression of eosinophils was not different in corresponding insert, coculture conditions, and conditioned medium (not shown). One exception was that in IL-1-stimulated conditions after 4 h, eosinophils showed significantly less L-selectin expression in conditioned medium than in the corresponding insert condition. In fact, no decreased L-selectin expression at all was found on eosinophils in the presence of stimulated fibroblasts under insert condition after 4 h compared with control (not shown).

Two Populations of Eosinophils with Regard to L-selectin Shedding

We observed two different populations of eosinophils with regard to L-selectin shedding after culture: eosinophils that had lost all L-selectin molecules from the surface (L) and eosinophils still expressing L-selectin (L⁺), but at lower levels than freshly isolated eosinophils (Figure 5). Nonisolated eosinophils in peripheral blood and freshly isolated eosinophils showed only one eosinophil population that expressed L-selectin (L⁺⁺). Under activating conditions (cocultures with IL-1-stimulated fibroblasts and conditioned medium of IL-1-stimulated fibroblasts) the population of eosinophils without L-selectin expression (L) tended to be larger than in control conditions at 4 h, reaching significance at 24 h (p < 0.05). Addition of anti-GM-CSF to conditioned medium of IL-1-stimulated fibroblasts inhibited the increase in the L-selectin negative population (L).

View larger version (13K): [in this window] [in a new window]   Figure 5.   Different levels of L-selectin expression on (top panel ) freshly isolated eosinophils (L++) and (bottom panel ) eosinophils cultured with IL-1-stimulated fibroblast-conditioned medium for 4 h (L and L+).

Eosinophil Viability

Percentages of viable eosinophils were significantly higher in coculture with unstimulated fibroblasts (90 [89-94]% at 4 h, 86 [64- 91]% at 24 h) and IL-1-stimulated fibroblasts (95 [88-97]% at 4 h, 80 [60-86]% at 24 h) (Figure 6) than in control conditions with only eosinophils present (84 [81-90]% at 4 h, 48 [34-64]% at 24 h). In all conditioned medium conditions, percentages of viable eosinophils were also higher (p < 0.05). There were no significant differences in viability of eosinophils between IL-1-stimulated versus unstimulated or coculture versus conditioned medium, except that viability was higher in coculture with unstimulated fibroblasts than in corresponding conditioned medium at 24 h (Figure 6). Anti-GM-CSF did not decrease eosinophil viability under most conditions, but reduced eosinophil viability in 24-h conditioned medium of IL-1-stimulated fibroblasts (71 [58-78]% in the presence of anti-GM-CSF versus 75 [59- 88]% without anti-GM-CSF, p < 0.05) (not shown).

View larger version (10K): [in this window] [in a new window]   Figure 6.   Eosinophil survival after 4 and 24 h of culture expressed as median percentage (min-max) of vital cells excluding PI (n = 6). *p < 0.05 versus E0, #p = 0.05 versus EF0 24 h (Wilcoxon signed rank test, one-tailed).

Cytokine Levels in Cultures

In order to establish correlations between eosinophil activation and cytokine concentrations and to check whether the added quantity of neutralizing antibody was enough to neutralize all GM-CSF and IL-8 produced under all conditions, we determined cytokine concentrations. Eosinophils cultured in medium alone produced small quantities of GM-CSF and IL-8. Fibroblasts stimulated with IL-1 produced about 20 times more GM-CSF and IL-8 than did unstimulated fibroblasts. Table 1 shows that, at 24 h, cytokine concentrations in fibroblast-conditioned medium with or without eosinophils cultured in it after 24 h were lower than in cocultures of eosinophils and fibroblasts, reaching significance in most cases.

The GM-CSF concentration in eosinophil cocultures at 24 h (76 [30-155] pg/ml unstimulated, 963 [504-1,191] pg/ml IL-1-stimulated) was significantly higher than in corresponding fibroblast culture without eosinophils (42 [30-50] pg/ml, 681 [589-999] pg/ml, respectively). We found a positive correlation (p < 0.001) between CD11b expression and GM-CSF concentrations (r = 0.87) as well as IL-8 concentrations (r = 0.88) with total paired data. There was a negative correlation between L-selectin expression and GM-CSF levels (r = 0.82) as well as IL-8 levels (r = 0.80). No correlation was found between eosinophil viability and cytokine concentrations.

Total GM-CSF was neutralized under all conditions, because concentrations were below 1,000 pg/ml, except in a coculture with IL-1-stimulated fibroblasts of one donor (1,191 pg/ml). IL-8 concentrations in the stimulated fibroblast conditions after 24 h were too high to ensure 100% neutralization by anti-IL-8, because they exceeded 40 ng/ml (Table 1).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results show that normal human lung fibroblasts are able to activate isolated peripheral blood eosinophils in vitro as demonstrated by increased CD11b expression and decreased L-selectin expression of eosinophils.

We found a larger increase in CD11b expression in coculture than in corresponding conditioned medium conditions after 24 h of culture. This difference may partly be due to degradation of cytokines after 24 h. Conditioned medium in which eosinophils have been cultured (ECM0, ECMIL-1) is in fact 24 h older than the medium from cocultures (EF0, EFIL-1) and fibroblasts alone (F0, FIL-1) and shows lower GM-CSF concentrations (Table 1). Further experiments, in unstimulated conditions after 24 h, revealed a higher CD11b expression on eosinophils in coculture than in corresponding insert condition. In IL-1-stimulated conditions at 4 and 24 h, we found a smaller increase in CD11b expression in the insert condition and conditioned medium than in the corresponding coculture. No differences in GM-CSF levels of cocultures and corresponding insert conditions were found (not shown). These results together imply that direct contact between fibroblasts and eosinophils is required to increase CD11b expression of eosinophils. This is not necessarily a direct activation through adhesion molecules, but can be a secondary effect. For instance, we found higher GM-CSF concentrations in cocultures than in fibroblast cultures at 24 h of unstimulated and IL-1-stimulated fibroblasts, suggesting either that eosinophils have started to become cytokine-producing cells, or that fibroblasts have become activated by eosinophils and thus release more cytokines. It has been previously described that fibroblasts can become activated by eosinophils (24, 25). The activation of eosinophils by IL-1-stimulated fibroblast-conditioned medium appears to be mediated almost entirely by GM-CSF, because neutralizing anti-GM-CSF antibodies prevented an increase in CD11b expression.

GM-CSF has been shown before to increase CD11b expression on eosinophils when cultured for 24 h (26, 27). Increased CD11b expression of eosinophils due to contacts with cells has also been described for IL-1-stimulated umbilical vein endothelial cells (HUVEC) (28). Transmigration through stimulated HUVEC monolayers and contact with stimulated HUVEC membrane fragments caused a larger increase in CD11b expression than IL-1-stimulated, HUVEC-conditioned medium (28). Apparently, the most important mechanism to increase CD11b on eosinophils differs between endothelial cells and fibroblasts. From our results we conclude that CD11b upregulation is mainly mediated by GM-CSF, but influence of adhesive contacts with IL-1-stimulated fibroblasts after 24 h of coculture may be involved as well.

The decrease in L-selectin expression of eosinophils was virtually the same in cocultures with fibroblasts, in corresponding conditioned medium, and in insert conditions. This implies that decreased L-selectin expression results exclusively from soluble factors and that direct cell contacts are not a prerequisite. Under IL-1-stimulated insert conditions after 4 h, L-selectin expression on eosinophils was not decreased at all compared with control eosinophils. This lack of decrease is probably an effect caused by insufficient diffusion of culture medium through the insert filter after 4 h, which can also be responsible for the smaller CD11b expression observed under the same conditions (Figure 4b). Neutralizing antibodies to GM-CSF were able to prevent a decrease in L-selectin expression induced by IL-1-stimulated fibroblast-conditioned medium on eosinophils almost entirely, indicating a major role for GM-CSF in this process. It has previously been described that incubation with GM-CSF alone can cause decreased L-selectin expression on granulocytes (29). Mengelers and colleagues, however, were unable to show decreased L-selectin expression on eosinophils cultured with comparable GM-CSF concentrations for 4 h (30). In the latter study, eosinophils were purified from peripheral blood using 10 nM fMLP stimulation and subsequent centrifugation over a discontinuous Percoll gradient. This difference in eosinophil isolation procedure might be the reason for the dichotomy with our results. We conclude that in our study decreased expression of L-selectin on eosinophils in coculture with fibroblasts is only mediated by GM-CSF.

In contrast to the effect of anti-GM-CSF, we did not detect an inhibitory effect of anti-IL-8 on eosinophil activation, despite the fact that IL-8 was produced in large quantities by IL-1-stimulated lung fibroblasts. Furthermore, CD11b correlated positively with IL-8 concentrations, whereas the correlation was negative for L-selectin. After 4 h of culture in conditioned medium of IL-1-stimulated fibroblasts (levels below 40 ng/ml), anti-IL-8 did not have an inhibitory effect. Although the concentration of anti-IL-8 could not entirely inhibit high IL-8 levels (> 40 ng/ml) as found in conditioned medium of IL-1-stimulated fibroblasts after 24 h (Table 1), we do not regard this to be the reason for the inability to inhibit eosinophil activation. According to the manufacturer's product information, a partial inhibition of ± 70% would still be expected. It has previously been described that (primed) eosinophils can respond to IL-8 when used as a chemotactic factor (31, 32), but other indicators of activation have not been studied so far. Finally, we did not find evidence for an activating effect of fibroblast-derived IL-8 on eosinophil activation.

Anti-GM-CSF-induced inhibition of eosinophil activation was not found under coculture conditions. A simple explanation for this observation is not evident, but it may be an effect of poor accessibility of the anti-GM-CSF to eosinophils that rest on GM-CSF-producing fibroblasts. In contrast, in conditioned medium all GM-CSF can be neutralized before reaching the eosinophils. A second explanation is that adhesive interactions (CD11b/ICAM-1, VLA-4/VCAM-1, VLA-4/fibronectin, or c-kit/ SCF) could become important mediators for eosinophil activation when GM-CSF is neutralized under coculture conditions. Surprisingly, addition of anti-IL-8 increased CD11b expression and decreased L-selectin on eosinophils in some coculture conditions, indicating eosinophil activation. An explanation for this finding is speculative, but anti-IL-8 may nonspecifically activate fibroblasts or may activate fibroblasts via a feedback mechanism by scavenging fibroblast-produced IL-8. Overall, we did not find indications that fibroblast-produced IL-8 plays a role in eosinophil activation.

This is the first report showing two populations of eosinophils with different levels of L-selectin expression after activation. Previous studies have shown two L-selectin-expressing populations of neutrophils after 4 h of culture (33), and 1 h of lipopolysaccharide (LPS) treatment (34). These results were not observed after 10 to 30-min platelet-activating factor (PAF), fMLP, or IL-5 treatment (34) of neutrophils or eosinophils (34, 35). The development of L-selectin negative eosinophils (L) in our study is probably due to extended culture times or different modes of activation. This development seems to be regulated by the extent of eosinophil activation mediated by GM-CSF, because we found a decreased percentage of the L population when anti-GM-CSF was added to conditioned medium. It is plausible that the L population comprises eosinophils that are most activated. Another explanation for this phenomenon can be that the L population is senescent or early-phase apoptotic eosinophils (still excluding PI). In the circulation, L-selectin expression decreases on neutrophils as they age (33). When neutrophils become apoptotic after 18 to 24 h of culture, they show no expression of L-selectin, while their nonapoptotic counterparts continue to express L-selectin (36). However, the observation that anti-GM-CSF decreases the development of the L population in our study in the presence of conditioned medium after 4 h does not support this explanation. Anti-GM-CSF neutralizes GM-CSF and the latter inhibits apoptosis. Therefore, the former should have stimulated eosinophil apoptosis and consequent development of the L population. We conclude that eosinophils show two states of L-selectin shedding; in our study the L population is due to GM-CSF-mediated activation.

Other studies have shown that increased eosinophil viability by fibroblasts is partly mediated by fibroblast-derived GM-CSF. In our experiments, viability of eosinophils in coculture or conditioned medium from unstimulated or stimulated fibroblasts were all significantly increased compared with control eosinophils (Figure 6). We did not find a difference between unstimulated and IL-1-stimulated fibroblast conditions. This indicates that survival is dependent on a factor that is apparently not influenced by IL-1 stimulation of fibroblasts. No correlation between eosinophil viability and GM-CSF levels in different conditions has been found. However, a partial blocking of eosinophil viability by anti-GM-CSF was found in fibroblast-conditioned medium after 24 h. Unlike the activation of eosinophils, survival seems to be largely dependent on a soluble factor other than GM-CSF produced by fibroblasts. This result is supported by other studies. Zhang and coworkers (12) found that viability of eosinophils was only partly reduced when anti-GM-CSF was added to cocultures of bronchial myofibroblasts and eosinophils under insert conditions, and Vancheri and coworkers (11) found substantial but not total inhibition of increased eosinophil viability by anti-GM-CSF in fibroblast-conditioned medium. We suggest that soluble factors other than GM-CSF released by fibroblasts are the main factors responsible for increased eosinophil viability, although an attribution of direct cell contacts cannot be excluded. From preliminary experiments it became clear that coculture media do not contain IL-5 or RANTES under conditions used in our study.

In conclusion, our study shows that normal human lung fibroblasts are able to activate isolated peripheral eosinophils from nonatopic healthy individuals in vitro as demonstrated by increased CD11b expression and decreased L-selectin expression of eosinophils as well as an increased viability. GM-CSF produced by fibroblasts appears to be the most important factor in eosinophil activation. The knowledge that normal human lung fibroblasts are able to activate eosinophils may have important implications for the understanding of the pathogenesis of asthma. Infiltration of eosinophils in the lungs of asthmatic patients leads to eosinophil-fibroblast contact and interaction, in addition to interactions with endothelium and epithelium (28, 37). Fibroblasts thus may contribute to the activation of eosinophils as observed in pulmonary biopsies and BAL. Activating effects on eosinophils are more pronounced when fibroblasts have been in contact with a proinflammatory cytokine such as IL-1, a situation that is thought to reflect the inflamed state of asthmatic lungs.