Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Proteinases, which degrade basement membrane components, appear to play an important role in tumor cell invasion and metastasis (1); tumor cells elaborate proteinases which make holes in the basement membrane, disrupting the barrier and allowing cellular penetration. Inflammatory cells also transmigrate through the basement membranes; by analogy, they may also use degrading proteinases to migrate to the site of inflammation. Recent in vitro studies demonstrated that lymphocytes use matrix metalloproteinase (MMP)-2 and MMP-9 in migration through reconstituted basement membranes (2, 3). Neutrophils also use MMP-9 and elastase to migrate across basement membranes (4). Eosinophils specifically accumulate at sites of allergic inflammation and likely play important roles in the pathophysiology of diseases, such as asthma and allergic rhinitis (5). In contrast to the extensive data on the roles of adhesion molecules and cytokines in eosinophil binding to and migrating through endothelium in allergic inflammation (for review see Bochner and Schleimer [6]), the mechanisms by which eosinophils penetrate and traverse the basement membrane have not been investigated on a molecular level.

We have previously studied the migration of eosinophils through basement membrane components using Matrigel^®-coated chemotaxis chambers (7). In that report, we showed that both platelet-activating factor (PAF) and an eosinophil active cytokine are necessary for the basement membrane transmigration of eosinophils. In contrast, prior studies showed that in vitro transmigration of eosinophils in Boyden chemotaxis chambers and endothelial cell constructs was provoked by either chemoattractant or cytokine stimulation (8, 9). These variations may be caused by the different step(s) needed for basement membrane transmigration, such as attachment and degradation of basement membrane components. In addition, eosinophils in the inflammatory tissues of patients with eosinophilia have been shown to express MMP-9, a protease which can specifically degrade matrix proteins (10, 11), suggesting a potential role for this enzyme in migration of eosinophils.

Based on this information, we proposed that protease activity is required for eosinophil migration through basement membranes. We tested this hypothesis using a pore membrane coated with Matrigel^®, a gel containing basement membrane components (12), as a model of basement membrane. First, using scanning and transmission electron microscopy, we examined the degradation of the Matrigel^® layer during eosinophil transmigration. Second, using inhibitors of various types of proteinases, we tested whether proteinase activity is important for eosinophil transmigration. Finally, using gelatin zymography, immunoblotting, and an MMP-9 assay, we identified the proteinases involved in eosinophil transmigration and examined the conditions required for their activation by eosinophils.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Reagents

Human recombinant interleukin (IL)-5 was generously provided by Schering Corporation (Kenilworth, NJ). Actinomycin D, 1-10-phenanthroline, 3,4-dichloroisocouma-rin (3,4-DCI), trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane (E-64), leupeptin, chymostatin, elastatinal, and phosphoramidon were purchased from Sigma Chemical Co. (St. Louis, MO). PAF was purchased from Biomol (Plymouth Meeting, PA). The monoclonal antibody (mAb) to MMP-9 (6-6B) (13) was purchased from Oncogene Science (Uniondale, NY). Batimastat, a selective inhibitor of MMP (14), was generously provided by British Bio-Technology (Oxford, UK). The fluorogenic substrate, (7-methoxycoumarin-4-yl)acetyl-L-Prolyl-L-Leucyl-Glycyl- L-Leucyl-[N³-(2,4-dinitrophenyl)-L-2,3,-diaminopropionyl]- L-Alanyl-L-Arginine amide, was purchased from Peninsula Laboratories, Inc. (Belmont, CA).

Purification and ⁵¹Chromium (Cr)-labeling of Eosinophils

Eosinophils were isolated from peripheral blood specimens obtained from normal volunteers by density centrifugation with 1.085 g/ml of Percoll (Sigma) and by CD16-negative selection with a magnetic activated cell sorter (Miltenyi Biotec, Sunnyvale, CA), as previously described (15). The purity of the eosinophil preparations used in this study was 99.4 ± 0.1% (mean ± SEM, n = 41). Eosinophils were labeled with ⁵¹Cr by incubating them with 250 µCi of sodium chromate (Amersham, Arlington Heights, IL) for 1 h at 37°C in 750 µl of Pipes (piperazine-N,N'-bis[2-ethane sulfonic acid]) buffer containing 1% heat- inactivated calf serum (16). Before use, cells were washed 3 times and resuspended in RPMI 1640 medium containing 25 mM Hepes at 1 × 10⁶ cells/ml.

Transmigration Assay

The migration of eosinophils through basement membrane components was investigated using BioCoat Matrigel^® Invasion Chambers (Becton Dickinson, Bedford, MA), as described previously (7). To optimally induce migration of eosinophils, 500 µl of RPMI 1640 containing 0.1 µM PAF was added to the lower chamber in 24-well plates, and 10 ng/ml of IL-5 was added to both upper and lower chambers. Subsequently, 200 µl of ⁵¹Cr-labeled eosinophils (1 × 10⁶ cells/ml) was added to the upper chambers, and the plates were incubated for 4 h at 37°C. After the cells remaining in the upper chambers were removed by aspirating and wiping with cotton swabs, the transmigrated cells were harvested by washing the bottom side of the chamber membrane and the lower chamber with 500 µl of ice-cold Pipes solution containing 1% heat-inactivated calf serum and 5 mM EDTA. ⁵¹Cr radioactivity in the harvested solutions was counted using a gamma counter. We calculated percent migration of eosinophils as the ratio of ⁵¹Cr activity in the lower chamber divided by the total ⁵¹Cr activity initially added to the upper chamber. To study the role of proteinases in eosinophil migration, inhibitors of proteinases or neutralizing antibody to MMP-9 (6-6B) were added to both upper and lower chambers 10 min before the addition of IL-5 and PAF and left in the chambers during the 4-h incubation. At the concentrations used, these proteinase inhibitors did not inhibit superoxide production by eosinophils stimulated by 0.1 µM PAF (17) (data not shown). We also studied the effect of actinomycin D solutions (0.1 and 1 µg/ml) on Matrigel^® transmigration by adding it to both chambers. After transmigration, cell-free supernatants from the upper chambers were collected and analyzed for gelatinolytic activity by gelatin zymography, as described below.

The migration of eosinophils through a bare membrane was examined using 24-well Falcon^® cell culture inserts (Becton Dickinson), which have the same pore size (8 µm) and pore density and are identical in material composition to Matrigel^® Invasion Chambers, but are not coated with Matrigel^®. ⁵¹Cr-labeled eosinophils were suspended in RPMI 1640 medium containing 25 mM Hepes and 0.1% human serum albumin. Eosinophil suspensions, 200 µl, were added to the upper chambers, and 500 µl of medium with or without 10 nM PAF was added to the lower chamber. Chambers were incubated at 37°C and 5% CO₂ for 45 min. The numbers of transmigrated cells into the lower chamber were determined by the same procedure as described above. Previous experiments (7) showed that when a bare membrane was used as a barrier, the dose-response curve of eosinophil migration response to PAF was bell-shaped, and 10 nM of PAF was optimal. To study the effects of inhibitors of proteinases, the inhibitors were added to both upper and lower chambers 10 min before the addition of PAF and left in the chambers during the 45-min incubation.

Electron Microscopy

The morphologic changes of the Matrigel^® layer and the transmigrating eosinophils were observed by scanning electron microscopy and transmission electron microscopy (TEM). Unlabeled eosinophils were incubated for 30 min, 1 h, 2 h, and 4 h in the Matrigel^® Invasion Chamber with or without IL-5 and with or without PAF. Medium in both the upper and the lower chambers was replaced with Trump's fixative (18), and the chambers with attached eosinophils were fixed overnight at 4°C. For TEM, membranes were removed from the chamber, en bloc stained with 2% uranyl acetate, dehydrated in progressive concentrations of ethanol and 100% propylene oxide, and embedded in Spurr's resin (19). Thin sections were cut on an ultratome, placed on 200-mesh copper grids, stained with lead citrate, and viewed on a JEOL 1200EX transmission electron microscope. For scanning electron microscopy, membranes were dehydrated in progressive concentrations of ethanol, dried by the critical point method (CO₂), mounted on stubs, sputter-coated with gold, and observed in a JEOL 6400 scanning electron microscope.

Gelatin Zymography

To characterize the gelatinolytic activity of cell-free supernatants from the eosinophil migration chambers and from eosinophil conditioned media (CM), samples were analyzed by gelatin zymography. Cell-free supernatants were obtained from the upper Matrigel^® chambers after eosinophil incubation with IL-5 and PAF for 2 h. To obtain eosinophil CM, eosinophils (5 × 10⁶ cells/ml) were suspended in RPMI 1640 medium supplemented with 25 mM Hepes and incubated in 96-well tissue culture plates with or without IL-5 and with or without PAF for 4 h at 37°C. Gelatinolytic activities in these supernatants were determined by zymography, as previously described with minor modifications (20). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (7.5% or 12% polyacrylamide) was performed by using gels copolymerized with 0.3 mg/ml gelatin (Sigma). The samples were diluted in nonreducing sample buffer and electrophoresed for 3 h at a constant current of 20 mA at 0°C. The gels were washed twice with 50 mM Tris-HCl containing 2.5% Triton X-100, 1 µM ZnCl₂, and 10 mM CaCl₂ for 15 min to remove SDS; rinsed with 50 mM Tris-HCl, 1 µM ZnCl₂, and 10 mM CaCl₂; and then incubated overnight at 37°C in incubation buffer (50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 1 µM ZnCl₂, 10 mM CaCl₂, 0.02% NaN₃). In some experiments, the gels were incubated in the presence or absence of various proteinase inhibitors. After staining with 0.25% Coomassie blue and destaining, gelatinolytic activities were detected as a transparent spot in the remaining blue-stained gel. Gels were dried and scanned by ScanJet 4C (Hewlett Packard, Palo Alto, CA), and the black-white image was inverted and the density of bands was measured using NIH Image, version 1.59.

Immunoblot Analysis of CM

Immunoblotting was performed to examine the MMP-9 activity in eosinophil CM. After gel electrophoresis of CM samples at nonreducing conditions in 7.5% polyacrylamide gel, the proteins were transferred to a polyvinylidene difluoride microporous membrane (Immobilon^TM PVDF; Millipore, Bedford, MA). The blots were blocked for 3 h with 5% bovine serum albumin (BSA) and 0.5% Tween 20 in Tris-buffered saline (TBS). After several washes with TBS containing 0.5% Tween 20, blots were incubated for 2 h at room temperature with 1 µg/ml MMP-9 mAb, and 1 h with peroxidase-conjugated rabbit antimouse IgG (0.5 mg/ml) (DAKO, Carpinteria, CA) in TBS containing 2% BSA and 0.5% Tween 20. After 3 washes, blots were incubated for 1 min with ECL^TM solution (Amersham), and exposed to X-ray film.

Fluorometric Assay of MMP Activity

A substrate degradation assay was used to measure MMP activity in eosinophil CM, as previously described (21). Briefly, 100 µl of CM and 400 µl of a fluorogenic substrate for MMP, (7-methoxycoumarin-4-yl)acetyl-L-Prolyl-L-Leucyl-Glycyl-L-Leucyl-[N³-(2,4-dinitrophenyl)-L-2,3-diaminopropionyl]-L-Alanyl-L-Arginine amide (final concentration 2 µM), in Hanks' balanced salt solution (HBSS) containing 1 µM ZnCl₂, 10 mM CaCl₂, and 10 mM Hepes were mixed and incubated for various times up to 120 min at 37°C. After the incubation, fluorescence intensities (_ex 328 nm, _em 393 nm) were measured by a Shimadzu RF-1501 fluorophotometer (Kyoto, Japan). Fluorescence intensities increased linearly with incubations between 5 and 120 min. Therefore, the differences in fluorescence intensities between 60 min and 120 min of incubation were used to monitor MMP enzymatic activities. To confirm the specificity of this assay for MMP, CM samples were analyzed with or without 1 µM of an MMP inhibitor, batimastat.

Statistical Analysis

Significant differences were determined using analysis of variance (ANOVA) with post hoc analysis of Fisher's protected least significant difference or paired t test using StatView 4.0 (Abacus Concepts, Berkeley, CA) for Macintosh.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Optimal Conditions for Eosinophil Migration through Matrigel^®

We have previously shown that both PAF and an eosinophil active cytokine are necessary for the basement membrane transmigration of eosinophils (7). As shown in Figure 1, IL-5 at 10 ng/ml alone in both upper and lower chambers did not induce eosinophil transmigration, nor did PAF at 0.1 µM alone in the lower chamber. In contrast, when 10 ng/ml of IL-5 was added to both chambers, PAF at 0.1 µM strongly induced transmigration of eosinophils. Other conditions of IL-5 alone or PAF alone were also tested (e.g., PAF alone in both upper and lower chambers); however, no significant eosinophil migration was observed in these conditions (7). These findings suggest that both PAF and IL-5 are required for optimal migration of eosinophils through Matrigel^®.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Effects of IL-5 and PAF on eosinophil transmigration through Matrigel®. IL-5 (10 ng/ml) or medium alone was added to both upper and lower chambers, and PAF (0.1 µM) or medium alone was added to the lower chamber. 51Cr-labeled eosinophils were added to the upper chamber, and the 51Cr activity of transmigrated eosinophils in the lower chamber 4 h after incubation at 37°C was measured. Data are presented as mean percent eosinophil migration ± SEM of 3 separate experiments. *Significant difference (P < 0.05) from eosinophil migration in the absence of both IL-5 and PAF. n = none.

Electron Microscopy of Matrigel^®-coated Membrane

To examine the morphology of eosinophils and the Matrigel^®-coated membrane during eosinophil transmigration, we examined the upper and the lower surfaces of the membrane by scanning electron microscopy. In the absence of IL-5, PAF, or both IL-5 and PAF, the pores in the polyethylene terephthalate membrane were completely filled with Matrigel,^® and the upper surface of the Matrigel^®-coated membrane was smooth; no cells transmigrated even after 4 h of incubation (Figure 2A). In the presence of IL-5 (10 ng/ml) and PAF (0.1 µM), apparent transmigration of eosinophils was observed by 30 min after the initiation of incubation (Figure 2B). Small holes had formed under the eosinophils, and cells appeared to be entering the holes on the upper side of the membrane (Figure 2B). In some areas, more marked erosion of the Matrigel^®-coated membrane had occurred and more cells had accumulated at this time point (Figure 2C). Erosion of the Matrigel^®-coated membrane increased with time, and after a 4-h incubation in the presence of IL-5 and PAF, the erosion sites appeared strikingly disrupted (Figure 2D). In Figures 3A and 3B, showing the lower side of the membrane, cells in the presence of IL-5 and PAF with expanded pseudopods are migrating through the hole at 1 h. On the lower side of the membrane, transmigrated cells increased up to 2 h (Figure 3C), and at 4 h, almost all cells attached to the membrane appeared morphologically disrupted (Figure 3D). IL-5 and PAF in the absence of eosinophils did not cause any morphologic changes in the Matrigel^®-coated membrane (not shown). Figures 2A and 2D show eosinophils in the absence (Figure 2A) or presence (Figure 2D) of both IL-5 and PAF and demonstrate marked differences in the morphology of the Matrigel^®-coated membrane, suggesting important roles for IL-5 and PAF in the eosinophil activation and subsequent disruption of the Matrigel^®. The electron microscopy was performed 3 times with eosinophils from different donors. The observations were consistent among the experiments. In addition, these results are also compatible with the quantitative analysis of eosinophil transmigration (Figure 1), showing that transmigration of eosinophils occurred only in the presence of IL-5 and PAF. We also tested whether an eosinophil lysate affected the Matrigel^®-coated membrane, but there was no change in the surface of the membrane (not shown).

View larger version (157K): [in this window] [in a new window]   Figure 2.   Scanning electron photomicrographs of the upper side of the Matrigel®-coated membrane incubated with eosinophils. (A) Four-hour incubation without PAF or IL-5. The cells appear spherical, and the Matrigel®-coated membrane surface is smooth. (B) and (C) Thirty minutes after incubation with 0.1 µM PAF and 10 ng/ml IL-5. Both a hole under a cell (B) and erosion of the membrane (C) are observed. (D) Four-hour incubation with 0.1 µM PAF and 10 ng/ml IL-5. Erosion appears more extensive in area and more cells have accumulated at this site. Bar = 2 µm.

View larger version (144K): [in this window] [in a new window]   Figure 3.   Scanning electron photomicrographs of the lower side of the Matrigel®-coated membrane incubated with eosinophils. (A) and (B) One hour after incubation with 0.1 µM PAF and 10 ng/ml IL-5. Eosinophils with pseudopods are present in holes in the Matrigel® (A) and change their shape when they adhere to the lower side of the membrane (B). (C) Two hours after incubation with PAF and IL-5. A cell with many pseudopods is seen next to a hole. (D) Four hours after incubation with PAF and IL-5. A ruptured cell adherent to the lower side of the membrane is shown. Bar = 2 µm.

The structure of the membrane was also examined by TEM; the upper side of the membrane was coated with Matrigel^® about 3 µm thick, and the pores in the membrane were filled or covered with Matrigel^®. Figure 4 shows the results after 2 h of eosinophil transmigration in the presence of both IL-5 and PAF. Eosinophils were attached to the upper side of the membrane with pseudopods, the thickness of the Matrigel^® layer near the attached eosinophil was decreased, and the Matrigel^® appeared to be eroded away from the eosinophil (Figure 4A; compared with 2C and 2D). Figure 4B shows an eosinophil with small pseudopods in a pore with a Matrigel^®-free space around the cell.

View larger version (118K): [in this window] [in a new window]   Figure 4.   Transmission electron photomicrograph of eosinophils transmigrating Matrigel®-coated membrane. (A) and (B) show cross-sections of the membrane 2 h after incubation of eosinophils with PAF and IL-5. (A) An eosinophil is attached to the upper surface of the membrane. The cell is bound to Matrigel® with its pseudopods. The layer of Matrigel® under the cell is thinner compared to the area distant from the cell. (B) An eosinophil within a pore in the membrane. The top of the figure is the upper chamber. Matrigel®-free space was observed around the cell, especially the cell periphery that is closest to the lower chamber (arrows). Bar = 2 µm.

Effect of Protease Inhibitors on Matrigel^® Transmigration of Eosinophils

To investigate the role of eosinophil-derived proteases in migration of eosinophils through Matrigel^®-coated membranes, we examined the effects of various protease inhibitors on eosinophil transmigration induced by PAF and IL-5. As shown in Figure 5, transmigration of eosinophils induced by 0.1 µM PAF and 10 ng/ml IL-5 was inhibited by 1,10-phenanthroline (a metalloproteinase inhibitor) and by 3,4-DCI (a serine proteinase inhibitor), but not by E-64 (a cysteine proteinase inhibitor). To further identify the serine proteinases involved in Matrigel^® transmigration of eosinophils, we used different types of serine proteinase inhibitors, including chymostatin, leupeptin, elastatinal, and eglin C (60-63), and found that chymostatin partially inhibited eosinophil transmigration, but others did not. This result suggests that a chymotrypsin-like serine protease is involved in eosinophil migration through Matrigel^®.

View larger version (35K): [in this window] [in a new window]   Figure 5.   Effect of proteinase inhibitors and mAb on Matrigel® transmigration of eosinophils stimulated by IL-5 and PAF. Inhibitors or MMP-9 mAb (10 µg/ml) were added to both the upper and lower chambers and preincubated with eosinophils for 10 min before the addition of stimuli. IL-5, 10 ng/ml, was added to both the upper and lower chambers, and 0.1 µM PAF was added to the lower chamber. Mouse IgG1 (10 µg/ml; Organon Teknika, Durham, NC) was used as a control for MMP-9 antibody. The results are normalized to percent of control, to which no protease inhibitors were added or to which control immunoglobulin, mouse IgG1, was added (for MMP-9 mAb). Transmigration of eosinophils in the controls (no inhibitors) was 8.17 ± 1.29 of total number of eosinophils. Data are presented as mean percent of controls ± SEM of 3 or more experiments performed in duplicate. *P < 0.05, **P < 0.01 from the control (white bar) of each group.

Because a metalloproteinase inhibitor (1,10-phenanthroline) inhibited transmigration effectively, we examined the effect of a specific inhibitor of MMP, batimastat (14). As shown in Figure 5, batimastat inhibited eosinophil transmigration in a concentration-dependent manner, and 100 nM batimastat inhibited eosinophil migration by 85%. The batimastat concentration causing 50% inhibition (IC₅₀) of eosinophil migration was approximately 3 nM, similar to a reported IC₅₀ value of batimastat for inhibition of MMP-9 (92 kD type IV collagenase, gelatinase B) in vitro (4 nM; Wang and colleagues [14]). Phosphoramidon, an inhibitor of some metalloproteinases, did not inhibit transmigration of eosinophils. To further investigate the role of MMP for eosinophil transmigration, we studied the effects of MMP-9 mAb. As shown in Figure 5, anti-MMP-9 significantly inhibited eosinophil migration through Matrigel^®. All in all, these findings suggest that MMP, and specifically MMP-9, plays an important role in the migration of eosinophils through a Matrigel^®-coated membrane. We also studied the effects of actinomycin D on eosinophil Matrigel^® transmigration in the presence of IL-5 and PAF. Actinomycin D did not inhibit transmigration of eosinophils during 4 h of incubation; percent eosinophil migration without actinomycin D, and with actinomycin D, 0.1 µg/ml or 1 µg/ml, were 7.2 ± 1.2, 6.7 ± 0.5, and 7.0 ± 2.8%, respectively.

To investigate the specificity of the effects of proteinase inhibitors, batimastat and 3,4-DCI were tested on eosinophil migration through a bare membrane. As shown in Figure 6, 10 nM of PAF induced approximately 27% eosinophil migration through a bare membrane. This eosinophil migration through a bare membrane was not affected by up to 100 nM of batimastat (less than 10% of inhibition) (Figure 6), although batimastat at 100 nM inhibited eosinophil migration through Matrigel^® by 85% (Figure 5). In contrast, a serine protease inhibitor, 3,4-DCI, at 10-µM partially inhibited eosinophil migration through a bare membrane. Thus a metalloproteinase inhibitor, batimastat, specifically inhibited eosinophil migration through Matrigel^®, but not through a bare membrane.

View larger version (19K): [in this window] [in a new window]   Figure 6.   Effect of proteinase inhibitors on eosinophil migration through a bare membrane. Inhibitors were added to both the upper and lower chambers and preincubated with eosinophils for 10 min before the addition of 10 nM PAF to the lower chamber. Data are presented as mean percent eosinophil migration ± SEM of 3 separate experiments. *Significant difference (P < 0.05) from PAF-induced eosinophil migration in the absence of inhibitor.

Analyses of Proteinases in the Supernatants from Eosinophil Transmigration Assays

To characterize the proteinases produced by eosinophils during transmigration, the supernatants from eosinophil transmigration chambers were analyzed by gelatin zymography. Eosinophil transmigration was performed in the presence of IL-5 and PAF for 2 h in the Matrigel^® migration chamber, as described in MATERIALS AND METHODS. Zymography of cell-free supernatants from the upper chambers using 7.5% polyacrylamide gels demonstrated distinct gelatinolytic activity at approximately 95 kD (Figure 7, lane 1). Less distinct gelatinolytic activities at 84, 120, and 131 kD were also detected. Similar gelatinolytic activities were observed in cell lysates of fresh eosinophils (lane 2), although 84 kD lytic activity was barely detectable. Zymography was also performed with 12% PAG; no gelatinolytic activities were detected at less than 84 kD. The gelatinolytic activities at 84, 95, 120, and 131 kD were abrogated when substrate gels were incubated with ethylenediaminetetraacetic acid (EDTA) (5 mM; lane 3) and batimastat (100 nM; lane 4), whereas inhibitors of a serine proteinase (3,4-DCI, 100 µM; lane 5) and elastase (elastinal, 100 µM; lane 6) had no effect. These findings suggest that the major gelatinolytic activity in cell-free supernatants from eosinophil transmigration assays is caused by a 95 kD protein with enzymatic characteristics typical of MMP, presumably MMP-9 (92 kD type IV collagenase, gelatinase B) (22).

View larger version (123K): [in this window] [in a new window]   Figure 7.   Gelatin zymographs of cell-free supernatants from eosinophil transmigration experiments. Supernatants from the upper chambers of eosinophil transmigration experiments were analyzed by gelatin zymography using SDS-PAGE with 7.5% PAG containing 0.3 mg/ml of gelatin under nonreducing conditions. Lanes 1, 3, 4, 5, and 6 show supernatants from eosinophil Matrigel® transmigration chambers after 2 h incubation, and lane 2 shows cell lysates from unstimulated eosinophils. The PAG matrix was treated with the following inhibitors before analyzing gelatinolytic activity: lane 1, no inhibitor; lane 3, 5 mM EDTA in the absence of CaCl2 and ZnCl2; lane 4, 100 nM batimastat; lane 5, 100 µM 3,4-DCI; lane 6, 100 µM elastinal.

Gelatin Zymography, Immunoblot, and MMP Enzymatic Activity of CM

To characterize the stimulus-dependent regulatory mechanisms of 95 kD MMP production by eosinophils, we studied eosinophil CM obtained as described in MATERIALS AND METHODS. Eosinophils were incubated with or without IL-5 and with or without PAF for 4 h at 37°C. As shown in Figure 8A, unstimulated eosinophils spontaneously released 95 kD proteolytic activity as revealed by gelatin zymography. The production of this 95 kD gelatinolytic activity was increased by stimulating cells with either IL-5 or PAF; 0.1 µM PAF was more potent than 10 ng/ml IL-5 in inducing secretion of this protein. Densitometric analyses of 4 experiments show that spontaneous release of 95 kD gelatinolytic activity was increased 55.1 ± 6.7%, 128.1 ± 13.1%, and 149.0 ± 11.7% by 10 ng/ml IL-5 alone, 0.1 µM PAF alone, and both IL-5 and PAF, respectively (mean ± SEM) (Figure 8B). Further, immunoblotting with MMP-9 mAb indicated that this 95 kD gelatinolytic activity was, in fact, MMP-9 (Figure 8C), and densitometric analysis of the bands showed that IL-5 alone, PAF alone, and both IL-5 and PAF increased the release of MMP-9 by 45.5%, 242.3%, and 290.9%, respectively, compared with the response without IL-5 and PAF (Figure 8D).

View larger version (35K): [in this window] [in a new window]   Figure 8.   Gelatin zymograph and immunoblot of eosinophil CM. Eosinophil CM specimens were obtained as described in MATERIALS AND METHODS by incubating eosinophils in the presence or absence of IL-5 or PAF for 4 h. CM were analyzed by gelatin zymography (A) and densitometric analysis of the gel (B) or by immunoblot using MMP-9 mAb (C) and densitometric analysis of the blot (D). Lane 1, eosinophils cultured with no IL-5 or PAF; lane 2, eosinophils cultured with 10 ng/ml IL-5 alone; lane 3, eosinophils cultured with 0.1 µM PAF alone; lane 4, eosinophils cultured with both IL-5 and PAF. (A) and (C) are representative of 4 experiments. (B) shows the mean density ± SEM from 4 independent experiments.

The enzymatic activity of MMP in eosinophil CM was quantitated by using a fluorogenic substrate for MMP, (7-methoxycoumarin-4-yl)acetyl-Pro-Leu-Gly-Leu-(3[2,4-dinitrophenyl]-L-2,3-diaminopropionyl)-Ala-Arg-NH₂ (21). The same eosinophil CM samples used in gelatin zymography (Figures 8A and 8B) and in the immunoblot analysis (Figures 8C and 8D) were used for the assay. As shown in Figure 9A, serial dilutions of pooled CM showed enzymatic activity in a concentration-dependent and linear manner; and batimastat, a selective MMP inhibitor, almost completely inhibited the activity, validating the specificity of the assay. As shown in Figure 9B, CM of eosinophils cultured with medium alone showed a small amount of MMP enzymatic activity. The CM of eosinophils cultured with IL-5 alone did not show increased enzymatic activity. Interestingly, CM of eosinophils cultured with PAF also showed no increases in enzymatic activity over baseline, although PAF enhanced the release of 92 kD MMP-9 from eosinophils (Figures 8B and 8D). In contrast, CM of eosinophils incubated with both IL-5 and PAF showed a 3-fold increase in MMP enzymatic activity (Figure 9B), suggesting that both IL-5 and PAF are required for optimal production of MMP enzymatic activity by eosinophils, and that neither one is sufficient for optimal stimulation of eosinophils for MMP activation.

View larger version (18K): [in this window] [in a new window]   Figure 9.   Spectrophotofluorometric analysis of MMP enzymatic activity. Eosinophil CM were obtained as described in MATERIALS AND METHODS by incubating eosinophils in the presence or absence of IL-5 or PAF for 4 h. MMP enzymatic activities in the samples were analyzed by a fluorogenic substrate for MMP, (7-methoxycoumarin-4-yl)acetyl-Pro-Leu-Gly-Leu-(3-[2,4-dinitrophenyl]-L-2,3-diaminopropionyl)-Ala-Arg-NH2 as described in MATERIALS AND METHODS. (A) CM from eosinophils of 4 different donors stimulated by IL-5 and PAF were pooled. The MMP enzymatic activities in serial 3-fold dilutions of CM were measured without (open circle) or with an MMP inhibitor, batimastat (closed circle). (B) MMP enzymatic activities in CM of eosinophils incubated with or without PAF and with or without IL-5 were determined. The results are normalized to percent fluorescence intensity of control, in which eosinophils were incubated without stimuli, 13.7 ± 3.2 fluorescence units. The fluorescence intensity of culture medium alone (without eosinophils) was zero. Data are expressed as mean percent control ± SEM from 4 independent experiments. *Significant difference (P < 0.05) from the MMP activity of the eosinophil CM incubated without IL-5 or PAF.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

We have previously shown that both PAF and an eosinophil active cytokine, such as IL-5 or granulocyte macrophage colony-stimulating factor, are necessary for the basement membrane transmigration of eosinophils (7). In this study, we found that basement membrane transmigration of eosinophils is dependent upon activation of eosinophil-derived proteinases, especially MMP, that degrade unique components of basement membrane. First, using scanning electron microscopy and TEM, we observed degradation of the Matrigel^® layer during and after transmigration of eosinophils stimulated by IL-5 and PAF. Second, using various proteinase inhibitors and antibody against MMP-9 in a transmigration assay, we found that MMP-9, a proteinase implicated in lymphocyte basement membrane transmigration in vitro (2, 3), and serine proteinase(s) are involved in the migration of eosinophils through Matrigel^® membrane. In contrast, eosinophil migration through a bare membrane was not affected by an MMP inhibitor. Finally, we could detect MMP-9 gelatinolytic activity and immunoreactivity in the supernatants from eosinophil transmigration chambers and in eosinophil CM. A marked increase in MMP enzymatic activity was found in the supernatants of eosinophils cultured with both IL-5 and PAF, but not with IL-5 alone or PAF alone. All in all, these findings indicate that degradation of the basement membrane by MMP-9 plays an important role in basement membrane transmigration of human eosinophils, similar to lymphocytes (2, 3) and neutrophils (4).

Although MMP-9 is produced by various inflammatory cells, little is known about the stimulus-dependent regulatory mechanisms of release and activation of this enzyme. Recent reports suggest that MMP-9 activity is regulated by cytokines and chemotactic peptides. For example, in a lymphocyte cell line, secretion of 92 kD MMP-9 and type IV collagenolytic activity was enhanced by chemotactic peptide, vasoactive intestinal peptide, and cytokines, such as IL-2 and IL-4 (3). In our study, we also found that a combination of PAF and IL-5 effectively enhanced release and activation of MMP-9 in normal human eosinophils. Furthermore, using PAF alone as a stimulus, we observed that the release and the activation of MMP-9 was dissociated. As shown by zymography and immunoblotting in Figures 8B and 8D, PAF alone strongly enhanced release of 92 kD MMP-9 from eosinophils to levels comparable with those in the presence of both PAF and IL-5. On the other hand, PAF alone did not induce transmigration of eosinophils (Figure 1), and did not enhance MMP enzymatic activity in eosinophil CM (Figure 9). Furthermore, degradation of the Matrigel^® layer by eosinophils was observed only in the presence of both IL-5 and PAF and not in the presence of PAF alone (Figure 2). Therefore, we speculate that PAF is effective in inducing release of MMP-9 from eosinophils, but ineffective in inducing activation of the enzyme. The reason PAF fails to induce transmigration of eosinophils (Figure 1), in spite of its potent ability to enhance release of MMP-9, may be due to its inability to activate the released enzyme; additional signals, such as that provided by IL-5, may be required for optimal activation of MMP-9 enzymatic activity. By TEM, we found firm adhesion of eosinophils to basement membrane components with Matrigel^®-free spaces around the transmigrating surface of the eosinophils (Figure 4). Further, we previously reported that eosinophil cellular adhesion molecules, ₁ and ₂ integrins, are involved in eosinophil migration through Matrigel^® (7). Therefore, we suspect that MMP-9 is activated in the local microenvironment around transmigrating cells, such as on the surface of adherent eosinophils, in conjunction with exposure to soluble mediators, such as IL-5 and PAF.

A literature review suggests that various MMP, including MMP-9, are released as pro-enzymes and are activated extracellularly by other proteinases (23); the 92 kD MMP-9 proenzyme is converted to the active form of 86 kD MMP-9 through cleavage of an N-terminal peptide. In our study, we found both 95 kD and 84 kD gelatinolytic activity in CM from eosinophils cultured in the presence of IL-5 and PAF; in contrast, only 95 kD gelatinolytic activity was found in lysates of fresh eosinophils (Figure 7). These findings suggest that 95 kD MMP-9 is preformed and stored within the cells. Once eosinophils are stimulated by IL-5 and PAF, MMP-9 is released and quickly converted to the 84 kD active form through the N-terminus cleavage. The promptness of eosinophil response to stimuli, as evidenced by degradation of Matrigel^® and commencement of transmigration as early as 30 min after exposure (Figure 2B), is compatible with this interpretation. Further support is provided by the failure of a transcription inhibitor, actinomycin D, to block transmigration of eosinophils. On the other hand, eosinophils resident in inflammatory tissues have been shown to transcribe the messenger ribonucleic acid for MMP-9 (9, 10), suggesting that eosinophils also have the capacity to synthesize MMP-9 de novo. Therefore, it seems likely that eosinophils use two sources of MMP-9: preformed MMP-9, which is released quickly upon cellular activation and used for transmigration; and de novo synthesized MMP-9, which may have importance in chronic inflammation.

Serine proteinases are also implicated in eosinophil trans-migration. The inhibitory effect of 3,4-DCI, a serine proteinase inhibitor, on the eosinophil transmigration was as potent as the metalloproteinase inhibitor, 1,10-phenanthroline (Figure 5). Further analyses suggest that this serine proteinase has chymotrypsin-like serine proteinase activity because an inhibitor of chymotrypsin, chymostatin, partially but significantly inhibited eosinophil transmigration (Figure 5). It is possible that serine proteases are involved in eosinophil migration responses in general because 3,4-DCI inhibited eosinophil migration through both Matrigel^® (Figure 5) and bare membrane (Figure 6) without affecting their superoxide production. Alternatively, chymotrypsin-like serine protease may be involved in eosinophil basement membrane transmigration by acting as an activator of MMP-9, modulating release of MMP-9 from cells, or directly degrading basement membrane components. First, serine proteinases, such as cathepsin G, trypsin, -chymotrypsin, and tissue kallikrein, are known to activate MMP-9 (24, 25). Among them, trypsin and kallikrein likely do not activate eosinophil MMP-9 because the inhibitor, leupeptin, did not block transmigration. Therefore, cathepsin G or -chymotrypsin may activate MMP-9 in eosinophils. Second, recently, a 28-kD chymotrypsin-like protease was identified within human eosinophils (26). The chymotrypsin inhibitor, chymostatin, significantly inhibited PAF-induced eosinophil degranulation, suggesting a role for this enzyme in the degranulation mechanism of eosinophils (26). Therefore, it is possible that a chymotrypsin-like protease is involved in release of MMP-9 from PAF-stimulated eosinophils. Finally, serine proteinases can degrade basement membrane components. Mast cell chymases, tryptase, neutrophil elastase, cathepsin G, and trypsin all degrade type IV collagen (27); however, there is no evidence to support the presence of these enzymes in eosinophils. Although eosinophils reportedly contain elastase (28), elastase likely does not play a key role in transmigration of eosinophils in our system because two types of elastase inhibitor, including elastinal and eglin C (60-63), did not inhibit transmigration of eosinophils. Furthermore, no gelatinolytic activity was detected in zymography at less than 84 kD (neutrophil elastase is approximately 28 kD [28]). In addition, elastase enzymatic activity has been found in neutrophils using a fluorogenic elastase substrate, methylsuccinylalanylalanylprolylvalyl-methylcoumarin amide (29), but not in eosinophils (data not shown). In the future, it will be interesting to identify this chymotrypsin-like serine proteinase in eosinophils, which likely has importance for transmigration and potentially for other effector functions of eosinophils.

In this study, we showed that proteinases, including MMP-9 and a chymotrypsin-like serine proteinase, are implicated in migration of eosinophils through basement membrane in vitro. The involvement of MMP-9 was further confirmed by gelatin zymography, immunoblot, and MMP enzymatic assay. These findings suggest that in inflammatory lesions, eosinophils migrate through basement membranes by releasing and activating MMP-9 and by degrading basement membrane barriers when they are activated by PAF and IL-5. Eosinophils are the predominant cells in tissues involved in bronchial asthma and other allergic diseases (30, 31). In these diseases, a large number of eosinophils transmigrate not only subendothelial but also subepithelial basement membranes. Ståhle-Bäckdahl and associates reported MMP-9 messenger ribonucleic acid expression in eosinophils infiltrating inflammatory sites (10, 11), suggesting production of MMP-9 by eosinophils within tissues. In addition, by zymography and enzymatic assay, we found both 92 kD MMP-9 proenzyme and 86 kD active form of MMP-9 in the bronchoalveolar lavage fluid specimens obtained from patients with asthma during asthma attacks (S. Okada and coworkers, manuscript in preparation). Therefore, MMP-9 and other proteinases released from eosinophils, which are transmigrating basement membranes and participating in inflammatory reactions, may contribute to destruction of tissue structures in disease typified by a histologic description of asthma as eosinophilic desquamative bronchiolitis (32). Furthermore, this tissue destruction may be followed by remodeling of airways with vascular engorgement and thickening of the subepithelial basement membrane (33, 34), changes found in chronic asthma. Control of the release and/or activation of MMP-9 in eosinophils may be another strategy for the treatment of allergic inflammation.