Cleavage of CD14 on Human Gingival Fibroblasts Cocultured with Activated Neutrophils Is Mediated by Human Leukocyte Elastase Resulting in Down-Regulation of Lipopolysaccharide-Induced IL-8 Production

Reagents

LPS of Escherichia coli O55:B5, HYBRI-MAX (serum-free and protein-free medium), α1-antitrypsin (α1-AT), PMSF, 1,10-phenanthroline, FMLP, and PMA were purchased from Sigma (St. Louis, MO). Bacillus cereus phosphatidylinositol-specific phospholipase C (PI-PLC), pepstatin (an asparatic protease inhibitor), and E-64 (N-[N-(l-3-trans-carboxirane-2-carbonyl)-l-leucyl]agmatine, a cysteine protease inhibitor) were purchased from Boehringer Mannheim (Indianapolis, IN). FCS was purchased from Flow Laboratories (McLean, VA). HLE, human neutrophil cathepsin G, and N-methoxysuccinyl-Ala-Ala-Pro-Val-chloromethyl ketone (HLE/CMK) were purchased from Calbiochem-Novabiochem (La Jolla, CA). Human leukocyte proteinase 3 (PR3) was purchased from HyTest (Turku, Finland). α-MEM and 0.25% trypsin/1 mM EDTA were from Life Technologies (Rockville, MD). Human rIL-1α and human rTNF-α were obtained from Dainippon Pharmaceutical (Osaka, Japan). Anti-HLA-ABC FITC (G46-2.6, mouse IgG1), anti-CD59 FITC (p282, mouse IgG2a), and anti-CD26 mAb (M-A261, mouse IgG1) were from PharMingen (San Diego, CA). Anti-CD14 mAb (MEM-18, mouse IgG1) was from Monosan (Uden, The Netherlands). Anti-CD14 mAb (MY4, mouse IgG2b), anti-CD157 mAb (RF3, mouse IgG1), and all isotype control mAbs were obtained from Immunotech (Marseille, France). All other reagents were obtained from Sigma, unless otherwise indicated.

Isolation of human PMNs

PMNs from heparinized (10 U/ml) peripheral venous blood were isolated by density-gradient centrifugation on Mono-Poly resolving medium (ICN Biomedical, Costa Mesa, CA) at 300 × g for 30 min at room temperature (18). The fraction containing PMNs was harvested and washed twice with PBS at 4°C. The viability of these cells was greater than 98%, as judged by trypan blue dye exclusion. The purity of PMNs was above 95% morphologically.

Human gingival fibroblast

HGF were prepared from the explants of normal gingiva from 8- to 25-year-old patients with informed consent, as reported previously (19). Explants were cut into pieces and cultured in 100-mm-diameter tissue culture dishes (Falcon; Becton Dickinson Labware, Lincoln Park, NJ) in α-MEM supplemented with 10% FCS with a medium change every 3 days for 10–15 days until confluent cell monolayers were formed. The cells were detached with 0.25% trypsin/1 mM EDTA, washed with PBS, and subcultured in plastic flasks (Corning Coster, Acton, MA). After three to four subcultures by trypsinization, homogeneous, slim, spindle-shaped cells grown in characteristic swirls were obtained. The cells were used as confluent monolayers at subculture levels 5 through 15.

Flow cytometry

HGF were collected by trypsinization, washed with PBS (pH 7.2), and used for staining. For HLA-ABC and CD59 staining, 10⁵ HGF were stained with FITC-conjugated mAbs or FITC-conjugated isotype-matched mouse IgG at 4°C for 30 min. For CD14, CD157, and CD26 staining, cells were stained with each mAb or each isotype-matched control IgG at 4°C for 30 min, followed by incubation with FITC-conjugated goat anti-mouse IgG (BioSource International, Camarillo, CA) at 4°C for an additional 30 min. Staining was analyzed on a FACScan (Becton Dickinson, Mountain View, CA). For staining fixed HGF, monolayers of HGF in 24-well multiplates were treated with 1% (w/v) paraformaldehyde for 20 min at room temperature. After washing three times with PBS, HGF were treated with HLE, as described below, and harvested by trypsinization and then stained with anti-CD14 mAb. In the coculture experiment of HGF with PMNs, the HGF population was gated on the basis of forward and right-angle scatters. Data were collected for 5000 events, which were stored in list mode and then analyzed with Lysis II software (Becton Dickinson). The arithmetic mean was used in the computation of the mean fluorescence intensity (MFI).

Coculture of HGF monolayer with PMNs

Confluent monolayers of HGF (approximately 10⁵ cells) in collagen I (rat tail tendon)-coated 24-well multiplates (Becton Dickinson Labware) were cocultured with the indicated number of purified PMNs in 500 μl of FCS-free α-MEM with or without 100 ng/ml PMA for 60 min at 37°C. Cells were collected by trypsinization, and expression of CD14 by HGF was analyzed by flow cytometry. To harvest the cell-free supernatants from PMNs, the indicated number of purified PMNs in a microtube containing 500 μl of FCS-free α-MEM were stimulated with 100 ng/ml PMA for 60 min at 37°C or primed with 100 ng/ml TNF-α for 10 min, followed by stimulation with 1 μM FMLP for 60 min at 37°C. After incubation of PMNs, supernatants were harvested by centrifugation, and kept at −20°C until use. For fixing PMNs, PMNs were stimulated with PMA similar to cell-free supernatant and fixed for 3 min at 4°C in PBS containing 3% (w/v) paraformaldehyde and 0.25% (v/v) glutaraldehyde (pH 7.4). After washing twice in HBSS, the indicated number of fixed PMNs were resuspended with 500 μl of FCS-free α-MEM. For protease inhibitor treatment, the indicated concentration of each inhibitor was preincubated with cell-free supernatant (500 μl) from 3 × 10⁶ PMNs or fixed PMNs (3 × 10⁶/500 μl) for 10 min at 37°C before addition onto monolayer cells. Percent inhibition for the supernatant and the fixed cells was calculated as ((MFI in the presence of supernatant and inhibitors) − (MFI in the presence of supernatant))/((MFI in the absence of supernatant) − (MFI in the presence of supernatant)) × 100 (%) and [(MFI in the presence of fixed PMNs and inhibitors) − (MFI in the presence of fixed PMNs)]/[(MFI in the absence of fixed PMNs) − (MFI in the presence of fixed PMNs)] × 100 (%), respectively.

HLE treatment

Monolayers of HGF in collagen I-coated 24-well multiplates (the well contained 300 μl of serum-free media, HYBRI-MAX) were treated with the indicated concentration of HLE at 37°C for the indicated times. For HLE inhibitor (α1-AT and human serum) treatments, 20 μg/ml HLE was preincubated with 100 μg/ml α1-AT or 10% (v/v) human serum for 15 min at 37°C before addition onto monolayer cells.

Reexpression of CD14 following proteolytic cleavage

Confluent monolayers of HGF in wells of 24-well collagen I-coated plates were treated with or without 20 μg/ml HLE in HYBRI-MAX for 30 min at 37°C. After gently washing monolayers with prewarmed HYBRI-MAX, 1 ml of prewarmed HYBRI-MAX with 0.5% FCS was added to the wells and incubated at 37°C. At the indicated times, cells were analyzed for the expression of CD14 by flow cytometry.

Preparation of cell membranes and treatment with enzymes

HGF were collected by trypsinization, and cells were suspended in hypotonic buffer (10 mM Tris-HCl, pH 7.4, and 1 mM MgCl₂) and incubated on ice for 30 min. Cells were then homogenized in a Dounce homogenizer by 15 strokes, and sucrose was added to a final concentration of 0.25 M. The homogenate was centrifuged at 500 × g for 5 min twice to remove nuclei. Supernatants were centrifuged at 15,000 × g for 30 min, and membrane pellets were suspended in PBS and stored at −20°C until use.

Western blotting

Cell membranes from confluent HGF grown in 25-cm² flasks were suspended in 100 μl of HBSS containing 20 μg/ml HLE or 5 U/ml PI-PLC and treated for 60 min at 37°C. Membrane pellets and supernatants were separated by centrifugation at 15,000 × g for 30 min at 4°C, and supernatants were dried by vacuum centrifugation. Membrane pellets and dried supernatants were solubilized with 20 μl of Laemmli sample buffer (10% glycerol, 1% SDS, 0.0025% bromphenol blue, and 50 mM Tris-HCl, pH 6.8) at 100°C for 5 min. Samples were separated by SDS-PAGE (10%). Proteins were transferred to a polyvinylidene difluoride membrane (ATTO, Tokyo, Japan) using a semidry transblot system (ATTO). The blot was blocked for 2 h with 3% BSA/PBS, followed by incubation with 2 μg/ml sheep anti-human CD14 polyclonal Ab (Genzyme/Techne, Minneapolis, MN) in 3% BSA/PBS for 2 h at room temperature. The blot was washed four times with PBS and then incubated for 2 h with HRP-conjugated affinity-purified donkey anti-sheep IgG (Jackson ImmunoResearch, West Grove, PA) at 1:2000 in 3% BSA/PBS. After washing, CD14 was visualized with diaminobenzidine. The M_r of the proteins was estimated by comparison with the position of the standard (Bio-Rad Laboratories, Hercules, CA).

Detection of IL-8 by ELISA

A total of 5 × 10⁴ HGF was cultured in α-MEM with 10% FCS in wells of 24-well collagen I-coated plates until confluent, and an additional 4-day culture was allowed. After washing with PBS three times, confluent monolayers of HGF were treated with 20 μg/ml HLE in HYBRI-MAX for 30 min at 37°C. HLE-treated monolayer cells were gently washed twice with prewarmed HYBRI-MAX, followed by addition of test stimulants in 500 μl of HYBRI-MAX without FCS for 4 h. For the blocking experiment with mAb, confluent monolayers of HGF were pretreated with 5 μg/ml dialyzed anti-CD14 mAb (MY4) or isotype control mouse IgG2b at 37°C for 30 min before addition of stimulants. After stimulation, the supernatants were collected and the level of IL-8 in the supernatants was determined with a human IL-8 ELISA kit (Endogen, Woburn, MA). IL-8 assays were performed exactly as instructed by the ELISA manufacturer. The concentration of IL-8 in the supernatants was determined using the Softmax data analysis program (Molecular Devices, Menlo Park, CA). Each sample was assayed in triplicate.

Statistical analysis

All experiments in this study were performed at least three times to test the reproducibility of the results, and representative findings are shown. In some experiments, experimental values are given as means ± SDs. The statistical significance of differences between two means was evaluated by Student’s unpaired t test.

Reduction of CD14 expressed on HGF by activated PMNs

Confluent monolayers of HGF were cocultured with purified PMNs in the presence or absence of PMA, and CD14 expression was evaluated by flow cytometry using the MEM-18 mAb. When HGF were cocultured with 3 × 10⁵ PMNs in the presence of 100 ng/ml PMA for 1 h, the MFI of CD14 was significantly (p < 0.01) reduced (MFI = 52% of control), and 3 × 10⁶ of activated PMNs caused up to 95% reduction of the CD14 expression (Fig. 1⇓A). However, unactivated PMNs lacked the activity at any number of PMNs. Next, we examined whether the reduction of CD14 expression required secreted molecules from PMNs or cell-to-cell contact between HGF and PMNs. When the HGF monolayer was incubated with the culture supernatant derived from the indicated number of PMA-activated PMNs, CD14 expression on HGF was significantly (p < 0.01) reduced, and the supernatant from 3 × 10⁶ of activated PMNs reduced the expression to approximately 75% of the control value (Fig. 1⇓B). When activated PMNs were fixed with 3% paraformaldehyde, as described previously (20), to prevent continuous secretion of molecules from PMNs during coculture with the HGF monolayer, the CD14 expression on HGF was also significantly (p < 0.01) reduced and 3 × 10⁶ of fixed PMNs reduced to approximately 70% of the control (Fig. 1⇓C). In addition to PMA stimulation, we examined whether a physiological stimulus also had a simillar effect. Fig. 1⇓D shows that when PMNs were stimulated with 1 μM FMLP for 60 min after priming with 100 ng/ml TNF-α for 10 min, the supernatant from those PMNs showed CD14 cleavage activity, to a degree approximately 80% that of the PMA supernatant, in a cell number-dependent fashion. These findings indicate that the reduction of CD14 expression on HGF was caused by molecules that were both released and expressed by activated PMNs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Coculture of PMNs and HGF results in reduction of CD14 expression on HGF. A, A confluent monolayer of HGF was cultured with the indicated number of PMNs in the presence or absence of 100 ng/ml PMA for 1 h. B, A confluent monolayer of HGF was cultured for 1 h with cell-free supernatant from the indicated number of PMNs stimulated with 100 ng/ml PMA for 1 h. C, A confluent monolayer of HGF was cultured for 1 h with the indicated number of 3% paraformaldehyde-fixed PMNs, which had been stimulated with 100 ng/ml PMA for 1 h. D, A confluent monolayer of HGF was cultured for 1 h with cell-free supernatant from the indicated number of PMNs stimulated with 1 μM FMLP for 1 h after priming with 100 ng/ml TNF-α for 10 min or stimulated with 100 ng/ml PMA for 1 h. After harvesting by trypsinization, HGF were stained with MEM-18 or isotype-matched mAb, and CD14 expression on HGF was analyzed by flow cytometry. Representative findings of three independent experiments are expressed as the mean ± SD of the MFI (% of control) of triplicate assay. Statistical significance is shown (∗, p < 0.01 vs samples without PMNs).

Effect of protease inhibitors on reduction of CD14 by activated PMNs

Since activated PMNs release various proteases and induce the cell surface-bound serine proteases (20, 21), we next examined the effect of various protease inhibitors on the reduction of CD14 expression by cell-free supernatant from PMA-stimulated PMNs. A naturally occurring serine protease inhibitor, α1-AT, and a synthetic serine protease inhibitor, PMSF, showed a marked inhibitory effect (83.6% and 79.9%, respectively) on the reduction of CD14 caused by cell-free supernatant (Table I⇓). Furthermore, HLE/CMK, a specific inhibitor of HLE, exhibited 62.1% inhibition of CD14 reduction caused by the supernatants. However, the inhibitory effect was scarcely observed in 1,10-phenanthroline (a metalloprotease inhibitor) and E-64 (a cysteine protease inhibitor), and only slight inhibition (15.9%) was observed in pepstatin (an asparatic protease inhibitor). These findings indicated that among the various proteases released from activated PMNs, serine proteases, in particular HLE, participated in CD14 cleavage caused by the cell-free supernatant.

Since serine proteases including HLE are expressed on the PMN cell surface upon activation (20), we next examined whether cell surface HLE as well as released HLE participated in the CD14 reduction by coculturing the HGF monolayer with 3% paraformaldehyde-fixed PMNs in the presence or absence of various protease inhibitors. PMSF and HLE/CMK only partially inhibited the CD14 reduction at 14.8% and 11.1%, respectively. Moreover, α1-AT, 1,10-phenanthroline, E-64, and pepstatin showed almost no inhibitory effect. These findings indicated that although cell surface serine proteases actually participated in the reduction of CD14 expression, the degree was very low. This observation might be due to: 1) the protease inhibitors were ineffective at cell-to-cell contact proteolysis, or 2) some protease(s) that was/were not inhibited by the protease inhibitors used in this study might be involved in the reduction. No inhibitory effect of α1-AT was supported by the study of Owen et al. (20), in which cell surface HLE was resistant to inhibition by naturally occurring protease inhibitors. The CD14-reducing activity induced by TNF-α and FMLP treatment of PMNs shown in Fig. 1⇑D was also inhibited by HLE/CMK at approximately 90% (data not shown). Therefore, the findings shown in Fig. 1⇑ and Table I⇑ indicate that the reduction of CD14 on HGF caused by activated PMNs was performed mainly by serine proteases, probably HLE in the supernatant and, although the degree was low, also partially performed by cell surface HLE.

Reduction of CD14 on HGF by HLE treatment

Since the major serine proteases released by activated PMNs are HLE, cathepsin G, and PR3, we next examined the effect of HLE, cathepsin G, and PR3 on CD14 expressed by HGF. Fig. 2⇓A shows a representative FACS profile, indicating that marked reduction of CD14 expression on HGF treated with 20 μg/ml HLE in the serum-free condition for 1 h was observed compared with untreated cells. This reduction of CD14 on HGF reached 85% at 10 μg/ml (340 nM) of HLE, and was totally achieved at 20 μg/ml (680 nM) of HLE for 60-min treatment (Fig. 2⇓B). In contrast, cathepsin G, which is also inhibited by α1-AT (22) and PMSF (23), and PR3 exhibited much less activity for CD14 reduction compared with the same concentration (nM) of HLE. These findings indicated that HLE was the main serine protease from activated PMNs responsible for CD14 reduction in the coculture experiment, as shown in Table I⇑. The reduction of CD14 by HLE was inhibited almost completely by pretreatment with 100 μg/ml α1-AT and 10% (v/v) of human serum, which is known as a source of α1-AT (Fig. 2⇓C). The time kinetics experiment revealed that complete reduction of CD14 was observed after 60-min treatment, and that even 30-min treatment was adequate to almost completely reduce CD14 from the cell surface (Fig. 2⇓D).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Reduction of CD14 on HGF by HLE treatment. HGF monolayer cells were treated with the indicated concentrations of HLE for the indicated times at 37°C. After treatment, cells were harvested by trypsin-EDTA and stained with MEM-18 or matched isotype Ab and analyzed by flow cytometry. A, A representative FACS profile of CD14 expression on HGF after treatment with or without 20 μg/ml HLE for 1 h. B, HGF were treated with the indicated concentrations of HLE, cathepsin G, or PR3 for 1 h. A total of 680 nM of HLE, cathepsin G, and PR3 is equal to 20, 16, and 20 μg/ml, respectively. C, HGF monolayer cells were treated with 20 μg/ml HLE for 1 h in the presence or absence of 100 μg/ml α1-AT or 10% (v/v) human serum. D, HGF were treated with 20 μg/ml HLE for the indicated times. Representative findings of three independent experiments are expressed as the mean ± SD of the MFI (% of control) of triplicate assays.

We next compared the sensitivity of CD26, CD59, CD157, and MHC class I as well as CD14 expressed on HGF by HLE. Since CD26 is induced by IL-1α stimulation on HGF (19), the HGF monolayer was stimulated with IL-1α for 5 days before analysis of CD26 sensitivity. HLE treatment (20 μg/ml) of HGF for 1 h reduced the expression of CD14 completely, but only slight reduction of CD26, CD59, CD157, and MHC class I (at most 12%, 28%, 7%, and 24% reduction, respectively) was observed (Fig. 3⇓), indicating that HLE preferentially reduced CD14. Next, we examined whether CD14 could be reexpressed on the cell surface after reduction of CD14 by HLE. CD14 was gradually reexpressed in the normal culture condition, and approximately 75% of CD14 was recovered within 48 h (Fig. 4⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Preference reduction of CD14 on HGF by HLE treatment. HGF monolayer cells were treated with 20 μg/ml HLE for 1 h, followed by harvesting cells by trypsinization. CD14, CD26, CD59, CD157, and MHC class I expression on HGF were measured by flow cytometry. Concerning the expression of CD26, the HGF monolayer was stimulated with 10 ng/ml rIL-1α for 5 days in α-MEM supplemented with 10% FCS before HLE treatment. Representative findings of three independent experiments are expressed as the mean ± SD of the MFI (% of control) of triplicate assays.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Reexpression of CD14 on HGF after reduction by HLE treatment. HGF monolayer cells were treated with 20 μg/ml HLE for 30 min at 37°C. After washing three times, monolayer cells were recultured in serum-free medium (HYBRI-MAX) supplemented with 0.5% FCS for the indicated times. After harvesting by trypsinization, cells were stained with MEM-18 and analyzed by flow cytometry. The findings are expressed as the percentage of the MFI of control cells incubated for the same time without pretreatment of HLE. Representative findings of three independent experiments are expressed as the mean ± SD of the MFI (% of control) of triplicate assays.

Reduction of CD14 on paraformaldehyde-fixed HGF by HLE treatment

These findings suggest that proteolytic cleavage of CD14 on the cell surface by HLE was the most plausible mechanism, although there remain two other possibilities that CD14 was internalized or shed by endogenous enzymes following HGF activation by HLE. To clarify this, HGF were fixed with 1% paraformaldehyde before HLE treatment. As shown in Fig. 5⇓, after fixation of HGF, HLE was still capable of effectively cleaving CD14 (p < 0.01). Considering that the shedding and the internalization of CD14 should require activating endogenous enzyme(s) or cell activation, this finding suggests that HLE cleaved CD14 on the cell surface proteolytically.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Effect of fixation on the reduction of CD14 on HGF by HLE treatment. HGF monolayer cells were fixed with 1% paraformaldehyde for 10 min at room temperature. After washing, untreated or fixed cells were incubated with 20 μg/ml HLE for 1 h at 37°C. Expression of CD14 on HGF was assessed by flow cytometry. Representative findings of three independent experiments are expressed as the mean ± SD of the MFI (% of control) of triplicate assays. Statistical significance is shown (∗, p < 0.01 vs respective control).

Immunoblot analysis of CD14 expressed on HGF

Another approach was to examine the proteolytic cleavage of CD14 by HLE. Purified cell membranes of HGF were treated with HLE, and both the cell membrane and extracellular supernatant were analyzed by Western blotting using anti-CD14 polyclonal Ab. As a positive control, PI-PLC, an enzyme that specifically removes GPI-anchored proteins from the cell surface (9), was used in this experiment. In untreated cell membranes, a strong 55-kDa CD14 band was detected, whereas this band disappeared after HLE treatment and partial reduction of the band was observed after PI-PLC treatment (Fig. 6⇓). This partial reduction of the CD14 band in the PI-PLC-treated cell membrane was consistent with flow cytometry in which CD14 expression was partially reduced (MFI, 75.7 (control) vs 28.9 (PI-PLC treatment)). In the extracellular supernatant, a strong 55-kDa band was detected in the PI-PLC-treated extracellular supernatant, whereas no bands were detected in HLE-treated extracellular supernatant or in controls. This finding and Fig. 5⇑ indicate that reduction of CD14 by HLE resulted from direct proteolysis, but not from shedding or internalization following cell activation, and could be degraded into multiple CD14 fragments that were not detected by polyclonal Ab for CD14.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Immunoblot analysis of CD14 expressed on and released by HGF after HLE treatment. Purified cell membrane of HGF was treated with 20 μg/ml HLE or 5 U/ml PI-PLC for 1 h at 37°C. After the cell membrane and the supernatant were separated by ultracentrifugation, both specimens were then solubilized with Laemli sample buffer, subjected to 10% SDS-PAGE, and transferred to polyvinylidene difluoride membrane. The blot was probed with an anti-CD14 polyclonal Ab. Molecular mass markers (kDa) are shown on the left. Findings are representative of two independent experiments.

HLE treatment of HGF resulted in the reduction of IL-8 production by HGF in response to LPS

The observation that HLE cleaved CD14 on HGF raises the question whether HLE inhibits LPS-induced IL-8 production by HGF, since it was reported that HGF produce IL-8 in response to LPS stimulation in a membrane CD14-dependent pathway (8). Preincubation of HGF with 20 μg/ml HLE for 30 min markedly suppressed IL-8 production from HGF upon stimulation with 10 ng/ml LPS (Fig. 7⇓A). Anti-CD14 mAb (MY4) inhibited LPS (10 ng/ml)-induced IL-8 production by HGF to background levels, but isotype-matched control Ab did not exhibit any inhibitory effect. We performed similar experiments with IL-1α as a CD14-independent stimulant. As shown in Fig. 7⇓B, HLE pretreatment with HGF did not significantly change IL-8 production triggered by IL-1α. MY4 and isotype-matched Ab were both unable to inhibit IL-8 production by HGF triggered with IL-1α. Thus, these findings indicate that HLE treatment of HGF specifically inhibited the CD14-dependent cell activation triggered by LPS.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

HLE treatment of HGF resulted in the reduction of IL-8 production by HGF in response to LPS. HGF monolayer cells were treated with 20 μg/ml HLE for 30 min at 37°C, or pretreated with MY4 (5 μg/ml) or isotype control IgG2b (5 μg/ml) before stimulation with 10 ng/ml LPS (A) or 10 ng/ml IL-1α (66) for 4 h at 37°C under serum-free conditions. The amount of IL-8 in the supernatants was analyzed by ELISA. Representative findings of three independent experiments are shown as the mean ± SD of triplicate assays. Difference was significant at p < 0.01 (∗).