INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

LIPOPOLYSACCHARIDE (LPS) is a potent secretagogue for a variety of cytokines from resident and inflammatory cells. Chemical isolation of lipid A (water-soluble form, triethylammonium salt) confirmed that lipid A was the active domain responsible for the induction of all known pathophysiological LPS effects (4, 5, 29). The results of a great number of biological experiments show that for full expression of typical in vivo manifestations such as fever and hypotension, the simultaneous presence of a bisphosphorylated -(1,6)-linked D-glucohexosamine disaccharide carrying six acyl groups [four molecules of (R)-3-hydroxy fatty acid (C-10 to C-16) and two secondary fatty acids in the form of two (R)-3-acyloxyacyl groups] in a defined structural arrangement (as in Escherichia coli) is a prerequisite (20). However, lipid A structures from some nonenterobacterial LPSs have been found to differ from Hemophilus and enterobacterial species in several parameters. For example, in lipid A from Chromobacterium violaceum (7, 30), Neisseria meningitidis (15), Pseudomonas aeruginosa (16), and Bacteroides fragilis (28), the nature, number, chain length, and location of fatty acids are different. In contrast to the asymmetric acylation pattern of fatty acids over the glucosamine (GlcN) disaccharide in lipid A of Hemophilus and E. coli, a symmetric distribution is present in C. violaceum and N. meningitidis lipid A (7, 30). The 3-hydroxy fatty acid chain lengths are smaller in lipid A of C. violaceum, N. meningitidis, and P. aeruginosa than in Hemophilus and enterobacterial lipid A. The major species of P. aeruginosa lipid A contains only five fatty acids because the primary acyl residue at position 3 of the reducing GlcN residue is lacking (16). Thus LPS of P. aeruginosa is significantly less toxic than enterobacterial LPS, and the presence of only five residues in lipid A may account for the low toxicity (14, 16).

The interaction of LPS with cells of the mononuclear system is perhaps the central event that triggers systemic reactions, resulting in endotoxic effects (3). On one hand, LPS activates mononuclear phagocytes both in vivo and in vitro to enhance functional capacity. Thus both the cytotoxic activity and phagocytic capacity of macrophages are significantly enhanced on exposure to biologically active LPS (21, 22). On the other hand, LPS induces the production of inflammatory mediators and immunoregulatory cytokines including prostaglandins, leukotrienes, platelet-activating factor (PAF), superoxide anion, hydrogen peroxide, nitric oxide, and interleukins as well as of tumor necrosis factor (TNF)- (27). These secondary, hormonelike mediators possess potent intrinsic bioactivity and contribute to the overall manifestation of endotoxic effects (27). In this sense, the LPS molecule is not toxic itself. As recognized by Thomas (26), it is the response of the host organism to LPS that makes LPS "poisonous." This seems to be true in other cells including endothelial cells, fibroblasts, and epithelial cells.

Although cytotoxicity of LPS from P. aeruginosa, i.e., Limulus amebocyte lysate (LAL) activity, is less potent than that from E. coli 0127:B8, P. aeruginosa infection induces a sustained prominent inflammation in the lung, including cystic fibrosis, diffuse panbronchiolitis, and chronic inflammatory lung diseases. In the present study, experiments were carried out to clarify the endotoxic effects of various LPSs, i.e., the release of neutrophil chemotactic activity (NCA) from type II alveolar epithelial-like cells, A549 cells, bronchial epithelial cell line BEAS-2B cells, and lung fibroblasts.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Culture and identification of type II alveolar epithelial cells, human fetal lung fibroblasts, and BEAS-2B cells. Because of the difficulty in obtaining primary human type II epithelial cells of sufficient purity, A549 cells (passage 75; American Type Culture Collection, Manassas, VA), a pulmonary type II epithelial cell line derived from an individual with alveolar cell carcinoma, were used (17). These cells retained many of the characteristics of normal type II epithelial cells such as surfactant production, cytoplasmic multilamellar inclusion bodies, and cuboidal appearance and had been extensively used to assess type II pneumocyte effector cell functions (2, 13, 24, 25). A549 cells were grown as monolayers on 35-mm-diameter tissue culture dishes as previously reported (12, 13). After the cells reached confluence, the cells were used for the experiment.

The effects of various LPSs from E. coli serotypes and P. aeruginosa were assessed on other lung cell types: BEAS-2B cells (a kind gift from Dr. Curtis Harris, National Institutes of Health, Bethesda, MD) and human fetal lung fibroblasts (HLFs) from an established commercially available cell line (human lung, diploid, passage 27; American Type Culture Collection). BEAS-2B cells and HLFs were cultured on 35-mm-diameter tissue culture dishes. After 4-6 days in culture, the cells reached confluence and were then used for the experiments.

Exposure of A549 cells, BEAS-2B cells, and HLFs to LPSs. Medium was removed from the cells by washing twice with serum-free Ham's F-12 medium, and the cells were incubated with Ham's F-12 medium without fetal calf serum (FCS) in the presence and absence of a variety of E. coli LPSs [serotypes 0127:B8 from both Difco (D; Detroit, MI) and Sigma (S; St. Louis, MO), 0111:B4 (S), 055:B5 (S), and 026:B6 (S)] and P. aeruginosa LPS [serotype 10 (S)] at concentrations of 0, 0.1, 1.0, 10, and 100 µg/ml and cultured for 12, 24, 48, 72, and 96 h at 37°C in a humidified 5% CO₂ atmosphere. In some experiments, A549 cells were stimulated with LPSs in the presence of 10% FCS. LPSs from Sigma were prepared by extraction into phenol with the method of Westphal and Luderitz (29). However, the content of lipid A was not determined. LPS from Difco was prepared similarly, and the content of lipid A was 9.7%. LPSs obtained from Sigma were tested for biological activity with LAL (Table 1). Each LPS did not cause injury (no deformity of cell shape or detachment from the tissue culture dish and viability of >95% of cells by trypan blue dye exclusion) to A549 cells, BEAS-2B cells, and HLFs after 72 h of incubation at the maximal concentration. The culture supernatant fluids were harvested and stored at 80°C until assayed. At least seven separate A549 cell supernatant fluids were harvested from cultures for each experimental condition.

Measurement of NCA. Polymorphonuclear leukocytes were purified from heparinized normal human blood with the method of Boyum (1). The amount of NCA in the supernatant fluids was quantified in 48-well microchemotaxis chamber (Neuro Probe, Cabin John, MD) as previously described (6). Neutrophils that had completely migrated through the filter were counted in five random high-power fields (HPFs; ×1,000) from each duplicate well by light microscopy. Chemotactic response was defined as the mean number of migrated cells per HPF. Ham's F-12 medium without FCS was incubated identically with A549 cells, and the supernatant fluids harvested were used to determine background neutrophil migration. Formyl-methionyl-leucyl-phenylalanine (fMLP; 10⁸ M in Ham's F-12 medium; Sigma) and normal human serum that was complement activated by incubation with E. coli LPS 0127:B8 (D) and diluted 10-fold with Ham's F-12 medium were used as positive controls.

To determine whether the migration was due to movement along a concentration gradient (chemotaxis) or stimulation of random migration (chemokinesis), a checkerboard analysis (32) was performed with A549 cell supernatant fluids harvested at 72 h in response to 100 µg/ml of E. coli LPS serotype 0127:B8 (D). To do this, various dilutions of A549 cell supernatant fluids (1:1, 1:3, 1:9, and 1:27) with target cells were placed below and above the membrane of the microchemotaxis chamber to make a variety of concentration gradients across the membrane.

Partial characterization of NCA. Partial characterization of NCA was performed with supernatant fluids harvested after 72 h of incubation at a concentration of 100 µg/ml of E. coli LPS 0127:B8 (D). Sensitivity to proteases was tested by incubating the supernatant fluids with trypsin (100 µg/ml; Sigma) for 30 min at 37°C followed by the addition of a 1.5 M excess of soybean trypsin inhibitor (Sigma) to terminate the proteolytic activity, and then the chemotactic activity was evaluated. The lipid solubility was evaluated by mixing the supernatant fluids twice with ethyl acetate, decanting the lipid phase after each extraction, evaporating the ethyl acetate to dryness, and resuspending the extracted material in the Ham's F-12 medium used for cell culture before the chemotaxis assay. Both extracted and extractant materials were evaluated for chemotactic activity. Heat sensitivity was determined by maintaining a supernatant fluid at 98°C for 15 min.

Molecular-sieve column chromatography of the chemotactic activity. To determine the approximate molecular mass of the released NCA in the supernatant fluids harvested after 72 h of incubation with 100 µg/ml of E. coli LPS 0127:B8 (D), we performed molecular-sieve column chromatography with Sephadex G-100 (50 × 1.25 cm; Pharmacia, Piscataway, NJ). At a flow rate of 6 ml/h, the A549 cell culture supernatant fluid was eluted with PBS, and every fraction after the void volume was evaluated for NCA in duplicate. The molecular-mass markers were bovine serum albumin (66 kDa), cytochrome c (12.3 kDa), and quinacrine (450 Da).

Effects of metabolic inhibitors on NCA release. The effects of the nonspecific lipoxygenase inhibitors nordihydroguaiaretic acid (NDGA; 100 µM; Sigma) and diethylcarbamazine (DEC; 1 mM; Sigma) and the 5-lipoxygenase inhibitor AA-861 (100 µM; Takeda Pharmaceutical, Tokyo, Japan) were evaluated in response to 100 µg/ml of E. coli LPS 0127:B8 (D) after 72 h of incubation. To further examine the involvement of protein synthesis in the release of chemotactic activity, we added cycloheximide (10 µg/ml; Sigma) to inhibit protein synthesis. At these concentrations, NDGA, DEC, and AA-861 inhibited the release of leukotriene (LT) B₄ in the cell cultures in response to E. coli 0127:B8 (D) (10, 11) and did not cause cytotoxicity to A549 cells after 72 h of incubation.

Effects of polymyxin B on the release of NCA from A549 cells, BEAS-2B cells, and HLFs in response to LPS. The remarkably high concentrations of LPSs (10-100 µg/ml) required to cause substantial effects were 100-1,000 times greater than those required for monocyte/macrophage effects. Although this might reflect biological differences in receptors and signaling, it raised the possibility that there might be contaminating molecules for the observed effects; therefore, we used polymyxin B (10 µg/ml) to abrogate the effects of LPSs on the release of NCA from A549 cells, BEAS-2B cells, and HLFs.

Effects of LTB₄- and PAF-receptor antagonists on NCA. The LTB₄-receptor antagonist ONO-4057 (ONO Pharmaceutical, Tokyo, Japan) and the PAF-receptor antagonist TCV-409 (Takeda) at a concentration of 10⁵ M were used to evaluate the responsible NCA in the crude supernatant fluids and in the lowest-molecular-mass marker separated by column chromatography.

Measurement of LTB₄ and PAF in the supernatant fluids. The concentration of LTB₄ in the supernatant fluids was measured by RIA as previously described (12). PAF in the supernatant fluids was quantified with the scintillation proximity assay system (12). This assay system combines the use of a high specific activity, tritiated PAF tracer with an antibody specific for PAF, and a PAF standard similar to that used in the method for measuring LTB₄.

Effects of polyclonal antibodies to interleukin-8 and granulocyte colony-stimulating factor. Neutralizing antibodies to interleukin (IL)-8 and granulocyte colony-stimulating factor (G-CSF) were purchased from Genzyme (Cambridge, MA). Anti-IL-8 and anti-G-CSF antibodies were added to the A549 cell supernatant fluids that were harvested 72 h after exposure to a concentration of 100 µg/ml of E. coli LPS 0127:B8 (D) without serum at the concentrations recommended for inhibiting these cytokines, and the resulting preparations were incubated for 30 min at 37°C. The A549 cell samples containing the antibodies were then used for chemotactic assay. The antibodies to IL-8 and G-CSF were tested by blockade of the chemotactic response of neutrophils to each human recombinant cytokine. These antibodies did not influence the chemotactic response to endotoxin-activated serum (data not shown). To assess the nonspecific effect of IgG, nonimmune IgG (Genzyme) was added to the supernatant fluids, incubated for 30 min at 37°C, and used for chemotactic assay. The nonimmune IgG did not affect the neutrophil chemotaxis in response to LPS-stimulated A549 cell supernatant or endotoxin-activated serum (data not shown).

Measurement of IL-8, G-CSF, and IL-6 in the supernatant fluids. The concentrations of IL-8, G-CSF, and IL-6 in A549 cell supernatant fluids were measured with enzyme-linked immunosorbent assay according to the manufacturer's directions. Assay kits for IL-8 and IL-6 were purchased from R&D systems (Minneapolis, MN), and the minimum detectable concentrations of IL-8 and IL-6 with these assays were 0.156 pg/ml and 10.0 pg/ml, respectively. G-CSF (chemiluminescence enzyme immunoassay) kit was obtained from CLAW Pharmaceutical (Tokyo, Japan), and the minimum detectable concentration of G-CSF was 1.0 pg/ml.

Statistics. In experiments in which multiple measurements were made, differences between groups were tested for significance with one-way analysis of variance, with Fisher's multiple range test applied to data at specific time and dose points. In experiments in which single measurements were made, the differences between groups were tested for significance with Student's paired t-test. In all cases, a P value < 0.05 was considered significant. Data are expressed as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Release of NCA from A549 cells, BEAS-2B cells, and HLF monolayers. LPSs stimulated the release of NCA from A549 cells in a time- and dose-dependent manner. However, there was a difference among E. coli serotypes and species [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 026:B6 (S); P < 0.01 between each LPS; Fig. 1, Table 2]. In contrast, LPSs derived from E. coli 0127:B8 (S) and 0111:B4 (S) did not stimulate A549 cells to release NCA. HLFs and BEAS-2B cells responded similar to A549 cells to various kinds of LPSs (Table 3). The release of NCA from HLFs and BEAS-2B cells was in the order P. aeruginosa > E. coli 0127:B8 (D) > E. coli 026:B6 (S) (P < 0.01 between each LPS; data not shown). LPS derived from E. coli 0111:B4 (S) did not stimulate HLFs and BEAS-2B cells. The lowest concentration of LPSs to stimulate A549 cells was 0.1 µg/ml. Increasing the concentration of LPSs up to 100 µg/ml progressively increased the release of NCA. The release of NCA began 12 h after exposure to LPSs, and the released activity was cumulative even after 72 h (Fig. 1B). The chemotactic activities to fMLP and activated serum were 120.4 ± 8.7 and 170.6 ± 15.4 neutrophils/HPF, respectively.

View larger version (20K): [in this window] [in a new window]   Fig. 1.   A: dose-dependent release of neutrophil chemotactic activity in response to Pseudomonas aeruginosa lipopolysaccharide (LPS) serotype 10 (), Escherichia coli LPS 0127:B8 from Difco (), and E. coli LPS 026:B6 from Sigma () from A 549 cell monolayers after 72 h of incubation. Values are means ± SE; n = 8 monolayers. There were significant differences between each group, P < 0.01. * P < 0.01 compared with medium alone. B: time-related release of neutrophil chemotactic activity in response to 100 µg/ml of P. aeruginosa LPS serotype 10 (), E. coli LPS 0127:B8 from Difco (), and E. coli LPS 026:B6 from Sigma () from A 549 cell monolayers. Values are means ± SE; n = 8 monolayers. There were significant differences between each group, P < 0.01. * P < 0.01 compared with supernatant fluids without incubation.

Checkerboard analysis revealed that A549 cell supernatant fluids to which E. coli LPS 0127:B8 (D) was added induced neutrophil migration in the presence of a gradient across the membrane in a concentration-dependent manner, but weak migration of a small number of neutrophils without a gradient (Table 4). Thus the passage of neutrophils was consistent with chemotactic rather than chemokinetic migration.

Inhibition of the release of NCA from A549 cells, HLFs, and BEAS-2B cells by polymyxin B. Polymyxin B inhibited the release of NCA from A549 cells, HLFs, and BEAS-2B cells in response to 100 µg/ml of LPSs derived from P. aeruginosa, E. coli 0127:B8 (D), E. coli 055:B5 (S), and E. coli 026:B6 (S) almost completely (Table 3). Polymyxin B per se did not affect the neutrophil chemotactic response to fMLP and activated serum.

Partial characterization of the released NCA. The NCA released from A549 cells was heterogeneous in its character. NCA was sensitive to heat, extractable into ethyl acetate, and partly digested by trypsin (Fig. 2). Incubation of A549 cells with cycloheximide inhibited the release of NCA (Fig. 3). The nonspecific lipoxygenase inhibitors NDGA and DEC and the 5-lipoxygenase inhibitor AA-861 attenuated the release of NCA (P < 0.01; Fig. 3). NDGA, DEC, and AA-861 did not have any effects on fMLP- and activated serum-induced neutrophil chemotaxis (data not shown).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Partial characterization of neutrophil chemotactic activity obtained from A549 cell supernatant fluids harvested after 72 h of incubation in response to 100 µg/ml of E. coli LPS 0127:B8 from Difco. EA; ethyl acetate; F12, Ham's F-12 medium. Values are means ± SE; n = 6 monolayers. * P < 0.01 compared with crude supernatant fluids.

View larger version (19K): [in this window] [in a new window]   Fig. 3.   Inhibition of release of neutrophil chemotactic activity by diethylcarbamazine (DEC), nordihydroguaiaretic acid (NDGA), AA-861, and cycloheximide in response to E. coli LPS 0127:B8 from Difco. Values are means ± SE; n = 6 monolayers. * P < 0.01 compared with stimulus alone.

Partial purification of NCA. Molecular-sieve column chromatography with Sephadex G-100 revealed that NCA was heterogeneous in size, with estimated molecular masses at 19 kDa, 8 kDa, and 400 Da (Fig. 4). NCAs of 19- and 8-kDa molecular masses were the predominant activity.

View larger version (11K): [in this window] [in a new window]   Fig. 4.   Molecular-sieve column chromatographic findings of released neutrophil chemotactic activity from A549 cell monolayer in response to 100 µg/ml of E. coli LPS 0127:B8 from Difco after 72 h of incubation. Nos. and arrows on top, molecular mass and position, respectively, of chemotactic peaks of indicated markers. Data are representative of 1 of 4 experiments.

Inhibition of NCA by LTB₄-receptor antagonists. The NCA in crude samples and the lowest-molecular-mass peak activity separated by column chromatography were inhibited by the addition of the LTB₄-receptor antagonist ONO-4057 by 25 and 60%, respectively (P < 0.01; Fig. 5, Table 5). In contrast, the effects of the PAF-receptor antagonist was not significant. The LTB₄-receptor antagonist inhibited NCA in the supernatant fluids released from A549 cells by LPSs derived from P. aeruginosa and E. coli 026:B6 (S), 0127:B8 (S), and 055:B5 (S) (data not shown). Each receptor antagonist at a concentration of 10⁵ M completely inhibited the neutrophil migration in response to a 10⁷ M concentration of LTB₄ and PAF but showed no inhibitory effects on fMLP- or endotoxin-activated serum-induced neutrophil and monocyte chemotaxis (data not shown).

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Inhibition of neutrophil chemotactic activity in A549 cell supernatant fluids harvested at 72 h in response to 100 µg/ml of E. coli LPS 0127:B8 with anti-interleukin (IL)-8 (Anti-IL-8) and anti-granulocyte colony-stimulating factor (Anti-G-CSF) antibodies and leukotriene (LT) B4-receptor antagonist. Values are means ± SE; n = 6 monolayers. * P < 0.01 compared with untreated supernatant fluids.

Concentration of LTB₄ and PAF in the supernatant fluids. The measurement of LTB₄ in supernatant fluids with RIA revealed that A549 cells released a significant quantity of LTB₄ in the baseline culture condition. However, the addition of E. coli LPS 0127:B8 (D) at a concentration of 100 µg/ml for 72 h did not induce LTB₄ release from A549 cells [68.9 ± 15.4 (LPS treated) vs. 55.4 ± 13.4 pg/ml (control); P = 0.10]. PAF was not detected in the baseline or LPS-stimulated supernatant fluids (<40 pg/ml).

Inhibition of NCA by polyclonal antibodies to IL-8 and G-CSF. The anti-IL-8 antibody significantly blocked the chemotactic response to neutrophils. The inhibition of total NCA by anti-IL-8 antibody was 40% (Fig. 5). The antibody to IL-8 inhibited 8-kDa chemotactic activity separated by molecular-sieve column chromatography by 70%. Anti-G-CSF antibody inhibited NCA in the supernatant fluids by 30% (Fig. 5) and 18-kDa chemotactic activity by 70%. These antibodies inhibited NCA released from A549 cells by LPSs derived from P. aeruginosa and E. coli 026:B6 (S), 0127:B8 (S), and 055:B5 (S) (data not shown). The anti-IL-8 and anti-G-CSF antibodies and the LTB₄-receptor antagonist together inhibited the total NCA up to 80% (Fig. 5).

Concentrations of IL-8, G-CSF, and IL-6 in the supernatant fluids. Measurement of IL-8, G-CSF, and IL-6 after 24 h of incubation revealed that LPSs at a concentration of 100 µg/ml significantly stimulated the release of IL-8 [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 055:B5 (S) > E. coli 026:B6 (S) = E. coli 0127:B8 (S)], G-CSF [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 026:B6 (S) = E. coli 0127:B8 (S) > E. coli 055:B5 (S)], and IL-6 [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 0127:B8 (S) > E. coli 055:B5 (S) > E. coli 026:B6 (S); Table 2]. E. coli LPS 0111:B4 (S) did not stimulate the release of IL-8, G-CSF, and IL-6 (Table 2).

LPSs at a concentration of 10 µg/ml after 72 h of incubation without serum caused a small increase in the release of IL-8 [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 026:B6 (S) > E. coli 0127:B8 (S) = E. coli 055:B5 (S) = E. coli 0111:B4 (S); Fig. 6], G-CSF [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 026:B6 (S) = E. coli 0127:B8 (S) > E. coli 055:B5 (S); Fig. 7], and IL-6 [P. aeruginosa > E. coli 0127:B8 (D) > E. coli 0127:B8 (S) > E. coli 026:B6 (S) = E. coli 055:B5 (S) = E. coli 0111:B4 (S); Fig. 8]. E. coli LPS 0111:B4 (S) did not stimulate the release of IL-8 and G-CSF. Although the release of IL-8, G-CSF, and IL-6 was remarkably augmented by the addition of 10% FCS (Figs. 6-8), a dependency on serotypes and species was observed.

View larger version (16K): [in this window] [in a new window]   Fig. 6.   Release of IL-8 from A549 cell monolayers in response to 10 µg/ml of LPSs derived from E. coli and P. aeruginosa (Pseudo) after 72 h of incubation in absence (solid bars) and presence (open bars) of serum. CONT, control; D; Difco; S; Sigma. Values are means ± SE; n = 9 monolayers. * P < 0.01 compared with control.

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Release of G-CSF from A549 cell monolayers in response to 10 µg/ml of LPSs derived from E. coli and P. aeruginosa after 72 h of incubation in absence (solid bars) and presence (open bars) of serum. Values are means ± SE; n = 9 monolayers. ND, not detected. * P < 0.01 compared with control.

View larger version (15K): [in this window] [in a new window]   Fig. 8.   Release of IL-6 from A549 cell monolayers in response to 10 µg/ml of LPSs derived from E. coli and P. aeruginosa after 72 h of incubation in absence (solid bars) and presence (open bars) of serum. Values are means ± SE; n = 9 monolayers. * P < 0.01 compared with control.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The cytotoxicity of LPS derived from P. aeruginosa is less potent than that from E. coli 0127:B8. However, P. aeruginosa infection induces a prominent sustained inflammation in the lung, including cystic fibrosis, diffuse panbronchiolitis, and chronic inflammatory lung diseases. LPS from P. aeruginosa stimulated airway epithelial cells and fibroblasts to release NCA, i.e., IL-8 and G-CSF, more potently than E. coli LPSs. The high potential for P. aeruginosa LPS to stimulate lung cells may, at least in part, explain the prominent sustained lung inflammation observed with P. aeruginosa infection.

The present study demonstrated that E. coli LPS 0111:B4 did not stimulate lung cells very potently, but P. aeruginosa LPS and E. coli LPSs 0127:B8, 055:B5, and 026:B6 stimulated the lung cells. Thus our results confirm the previous studies of Standiford et al. (24) and Pugin et al. (19) that E. coli LPS 0111:B4 did not stimulate lung cells without serum (LPS-binding protein) and extend the results that other different forms of LPSs may stimulate A549 cells, BEAS-2B bronchial epithelial cells, and human lung fibroblasts even without serum. LPSs from different E. coli serotypes and P. aeruginosa stimulated the release of NCA, i.e., IL-8 and G-CSF, from A549 cells. The presence of serum augmented the release of IL-8 and G-CSF. However, a difference among LPSs from E. coli and P. aeruginosa was observed. Thus the response of lung cells to LPSs may be differently regulated depending on the E. coli serotype and species involved.

The potential for LPS from P. aeruginosa to stimulate lung cells was most prominent. The differing stimulatory potential among LPSs from E. coli serotypes to release NCA was also evident. The release of IL-8 or G-CSF was regulated by LPSs depending on serotype and species. Because NCA consisted predominantly of IL-8 and G-CSF, the concentrations among NCA, IL-8, and G-CSF were correlated closely with each other. These data suggest that the differing stimulatory potentials of NCA among LPSs may be related to differences in the release of IL-8 and G-CSF.

P. aeruginosa LPS had the higher potential than the E. coli LPSs examined. However, LAL activity of P. aeruginosa was less than that of E. coli LPS 0127:B8. Because polymyxin B blocked the effects of LPSs, the possibility of contaminating molecules in LPSs to stimulate lung epithelial cells and fibroblasts was low. The cytotoxicity of P. aeruginosa LPS is reported to be less than that of E. coli by virtue of its specific lipid A structure (14, 16). Thus the cytotoxic potential of LPS, i.e., LAL activity, may not correlate with the stimulating potential of LPS to release NCA.

The primary culture of rat type II pneumocytes contains at least 3-5% alveolar macrophages. When we tried a primary rat type II pneumocyte culture, the contamination was 5-15%. It is probable that the effects of LPS to stimulate rat type II pneumocytes usually involves the activation of macrophages and subsequent release of IL-1 and TNF-, which are very potent secretagogues of airway cells. In that case, the primary rat type II cell culture was not suitable for the present experiment. However, primary bovine bronchial epithelial cells could be obtained with the method of Wu and Smith (31). The purity of bronchial epithelial cells was >98%. The bovine bronchial epithelial cells responded to E. coli LPS 0127:B8 (D) and released NCA and monocyte chemotactic activity, i.e., LTB₄ and TNF- without serum (8, 9). Although we did not examine the effect of every LPS on primary bovine bronchial epithelial cells, E. coli LPS 0127:B8 (D) stimulated primary bronchial epithelial cells. These data suggest that E. coli LPS has the stimulatory potential for primary airway epithelial cells.

The structure of lipid A in some nonenterobacterial LPSs has been found to differ from enterobacterial species in several parameters. The 3-hydroxy fatty acid chain lengths are smaller in lipid A of P. aeruginosa than in E. coli lipid A. The major species of P. aeruginosa lipid A contains only five fatty acids because the primary acyl residue at position 3 of the reducing GlcN residue is lacking (16). Thus LPS of P. aeruginosa is significantly less toxic than enterobacterial LPS, and the presence of only five residues in lipid A may account for the low toxicity (14, 16). Although the striking structural difference of lipid A between P. aeruginosa and E. coli is reported, there are no data available to explain the difference in functional and binding properties of LPSs to airway epithelial cells. In contrast, the structural pattern of lipid A among E. coli serotypes is quite similar. Although E. coli LPS 0111:B4 did not stimulate airway epithelial cells, E. coli LPS 0111:B4 stimulated renal cortical epithelia to release IL-8 (23), suggesting that the stimulation of airway epithelial cells by LPS might be regulated differently among epithelial cells. Although the precise structural mechanisms to stimulate airway epithelial cells by LPSs is uncertain, the functional and binding properties of LPSs to airway epithelial cells might determine the difference in LPS effects on airway epithelial cells.

The release of NCA in response to P. aeruginosa LPS was increased four- to fivefold compared with the constitutive release of NCA. This release of NCA was greater than that from 10⁶ alveolar macrophages/ml in response to IL-1, TNF-, and P. aeruginosa LPS (data not shown). Moreover, the release of NCA, IL-8, and G-CSF from A549 cells by P. aeruginosa LPS was almost equivalent to that from the same cells in response to 500 pg/ml of IL-1 or 1,000 U/ml of TNF- (data not shown). Thus the results of the present study suggest that P. aeruginosa LPS may contribute to the recruitment of neutrophils into the lung by stimulating lung epithelial cells and fibroblasts.

The concentration of LPSs required to stimulate A549 cells, BEAS-2B cells, and HLFs was greater than that required for monocyte/macrophage stimulation. Although the actual concentration of LPSs at the site of bacterial infection or colonization is uncertain, the LPS concentration in bronchoalveolar lavage fluids in patients with adult respiratory distress syndrome was 1-1,585 pg/ml (18). Because bronchoalveolar lavage fluid was diluted 50-100 times, the local concentration of LPSs at the site of bacterial infection or colonization would be far higher than that generally found in the epithelial lining fluid. Thus the high concentration of LPSs may be accessible at the site of the infection or colonization.

In conclusion, LPS derived from P. aeruginosa stimulated lung epithelial cells and fibroblasts to release NCA. LPSs derived from E. coli were less potent, and E. coli LPS 0111:B4 did not stimulate the release of NCA. Although the release of IL-8 and G-CSF was augmented by the addition of serum, the different stimulating potential among LPSs was observed. These data suggest that P. aeruginosa LPS may, at least in part, induce the sustained prominent lung inflammation observed at the site of P. aeruginosa infection.