Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cytokines are a large group of mediators that play a critical role in determining the nature and duration of the inflammatory response. They play a key role in the pathophysiologic changes in airway diseases such as chronic asthma or chronic obstructive pulmonary disease and are being increasingly recognized as important therapeutic targets (1). Most recently the proinflammatory cytokine granulocyte macrophage-colony stimulating factor (GM-CSF), which stimulates the maturation, activation, and survival of a number of inflammatory cells, has been recognized as important in the pathology of airway and allergic disease.

GM-CSF is produced by several airway cells, including macrophages, eosinophils, T lymphocytes, fibroblasts, endothelial cells, epithelial cells, and airway smooth muscle cells. Indeed, due to its documented effects on eosinophils (4), GM-CSF has been classified as an important cytokine in asthma.

Our group and others have recently suggested that in addition to its well-characterized contractile function, airway smooth muscle is actively involved in the inflammatory response (5). Indeed, human airway smooth muscle (HASM) releases a number of inflammatory mediators, including GM-CSF (7) and prostaglandins (PGs) (6) after stimulation with inflammatory cytokines. PG release by these cells is predominantly regulated by the inducible form of cyclooxygenase (COX)-2 under inflammatory conditions (6, 10). COX-2 is thought to regulate inflammatory events in humans, particularly those associated with arthritis (11), cardiovascular disease (12), and gastrointestinal disorders (11). In other settings, COX-2 induction in HASM cells regulates cellular function, including proliferative responses (15) and "desensitization" of adenylyl cyclase- mediated responses (16). However, the potential influence of COX-2 activity on GM-CSF production by HASM has not been addressed. This aspect of cytokine signaling is important because selective inhibitors of COX-2 are currently in clinical trials for the treatment of rheumatoid or osteoarthritis as well as for the prevention of some forms of cancer.

Thus, the purpose of this study is to investigate the possible role of COX products in GM-CSF release by HASM cells. We have previously shown that the predominant product of HASM cells is PGE₂ (15). PGE₂ acts on at least four prostanoid receptors (EP_[1-4]) to cause its biologic effects (17). However, in many biologic settings, the responses of PGE₂ are mediated by activation of adenylyl cyclase with a subsequent increase in cyclic adenosine monophosphate (cAMP) formation. Thus, in order to further understand how GM-CSF is formed by HASM cells, we have also assessed the effects of other cAMP modulating pathways on cytokine production by these cells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of HASM Cells

As described previously (6, 18), tracheal rings, from either heart or heart and lung transplantation donors (two females, five males, 27-45 yr of age), were dissected under sterile conditions in Hanks' balanced salt solution (HBSS) (in mM: NaCl 136.8, KCl 5.4, MgSO₄ 0.8, Na₂HPO₄ 7H₂O 0.4, CaCl₂ · 2H₂O 1.3, NaHCO₃ 4.2, and glucose 5.6) supplemented with the antibiotics penicillin (100 U ml¹) and streptomycin (100 µg ml¹), and the antifungal amphotericin B (2.5 µg ml¹). The smooth muscle layer was dissected free of adherent connective tissue and cartilage, and the epithelial layer was removed using a rounded scalpel blade. The smooth muscle section was then incubated for 30 min at 37° C in 5% CO₂/air in HBSS containing 10 mg ml¹ bovine serum albumin (BSA) and the enzymes collagenase (type XI, 1 mg ml¹) and elastase (type I, 3.3 U ml¹). After the removal of any remaining connective tissue, the smooth muscle was chopped finely and incubated for a further 150 min in the enzyme solution outlined previously with the elastase content increased to 15 U ml¹. Cells were centrifuged (100 × g, 5 min) at 4°C and resuspended in Dulbecco's modified Eagle's medium (DMEM) containing heat-inactivated fetal calf serum (FCS) (10% vol/vol), sodium pyruvate (1 mM), L-glutamate (2 mM), nonessential amino acids (1×) and antimicrobial agents as previously described.

Primary Culture of HASM Cells

Cells were placed in a tissue culture flask (75 cm²) with 6 ml of supplemented DMEM and incubated at 37°C in 5% CO₂/air. The cells adhered after approximately 12 h and the culture medium was replaced after 4 to 5 d (12 ml) and subsequently every 3 to 4 d. After approximately 10 to 14 d, the cells reach confluence and are identified by their typical "hill and valley" appearance and positive immunostaining for -actin. Cells were passaged into 2 × 75 cm² flasks and grown on 96-well plates at a seeding density of 2,000 cells/well. With this approach, cells could be maintained in culture over several passages (usually four to nine). At subconfluence, the cells were growth-arrested for 24 h, being placed in a serum-free medium containing the supplements outlined previously and BSA (0.1%). Then cells were treated for 24 h with test drugs containing 3% FCS, in the presence or absence of a mixture of the cytokines interleukin (IL)-1 and tumor necrosis factor (TNF)-, each at 10 ng ml¹. The COX inhibitors were added 10 min before the addition of the cytokine mixture at the concentrations stated in the figure legends.

Measurement of PGE₂ by Radioimmunoassay and of GM-CSF by ELISA

The COX metabolite PGE₂ was measured by radioimmunoassay and the cytokine GM-CSF by use of a specific sandwich enzyme-linked immunosorbent assay (ELISA) (6, 19). Cells were grown in 96-well plates and treated with cytokines in various combinations in the presence of 3% FCS.

Cell Viability

Cell viability was measured at the end of each experiment by the ability of respiring cells to convert 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide to formazan as described previously (6). None of the treatments used affected viability measured in this way.

Materials

Radiochemicals were obtained from Amersham International (Bucks, UK). IL-1, TNF-, and interferon- were purchased from R&D Systems Europe Ltd. (Abingdon, Oxfordshire, UK). L-745,337 was supplied by Merck Frosst (Montreal, Canada). Rolipram was supplied from Schering Aktiengesellschaft (Berlin, Germany) and SB 207499 from SmithKline Beecham Pharmaceuticals (King of Prussia, PA). OptEIA human GM-CSF ELISA set was purchased from PharMingen (San Diego, CA). Amphotericin B, nonessential amino acids, and sodium pyruvate were purchased from Life Technologies Ltd. (Paisley, UK). All other materials were purchased from Sigma Chemical Company (Poole, UK).

Statistical Analysis

Results are shown as the mean ± standard error of n experiments; cells from at least three separate patients were usually used for each protocol. Where appropriate, data were analyzed by Kruskal-Wallis nonparametric analysis of variance followed by Dunn's test for multiple comparisons, or in the case of "normalized" data by one-sample t test. All treatments were compared with control values and P < 0.05 was considered to be significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Release of PGE₂ and GM-CSF by HASM Cells

Under control culture conditions, HASM cells released low or undetectable levels of PGE₂ and GM-CSF. However, when cells were treated with a combination of IL-1 and TNF- (each at 10 ng/ml) for 24 h, they released relatively high levels of PGE₂ and GM-CSF (Figure 1A).

View larger version (17K): [in this window] [in a new window]   Figure 1.   (A) Release of GM-CSF (open columns) and PGE2 (solid columns) from HASM cells under basal or cytokine stimulation (cytmix: IL-1 and TNF- each at 10 ng ml1). Results are shown as the mean ± standard error of the mean (SEM) of nine determinations from three patients. Significant differences between basal and cytokine-treated cells were calculated by unpaired t test. **P < 0.01; ***P < 0.001. (B) The effect of indomethacin (108 to 105 M) on GM-CSF (open squares) and PGE2 (solid squares) release from HASM cells treated with cytokines. Results are shown as the mean ± SEM of nine determinations from three patients. Significant differences between control and indomethacin-treated cells were calculated by Kruskal-Wallis nonparametric ANOVA followed by Dunn's test for multiple comparisons. *P < 0.05; **P < 0.01.

Effects of Indomethacin and Other NSAIDs on PGE₂ and GM-CSF Release by HASM Cells

The mixed COX-1/COX-2 inhibitor indomethacin caused a concentration-dependent inhibition of PGE₂ release by HASM cells stimulated with cytokines (Figure 1B). By contrast, indomethacin induced a concentration-dependent increase in GM-CSF release by the same cells (Figure 1B). Indomethacin had no effect on the levels of PGE₂ or GM-CSF released by cells cultured without cytokines (n = 9; data not shown).

In separate experiments, a number of structurally different nonsteroidal anti-inflammatory drugs (NSAIDs), including aspirin, diclofenac, meloxicam, nimesulide, and L-745337, similarly inhibited PGE₂ production by cytokine-treated cells (data not shown). Similarly, to observations made with indomethacin, at concentrations where PGE₂ production was blocked, all NSAIDs tested caused an increase in GM-CSF release by cytokine-treated cells (Figure 2). However, there was a certain "rank ordering" of the efficacy of the NSAIDs to increase GM-CSF release. This was indomethacin > aspirin > diclofenac > meloxicam > L-745337  nimesulide (Figure 2).

View larger version (19K): [in this window] [in a new window]   Figure 2.   Effect of different NSAIDs on GM-CSF release by cytokine-treated HASM cells. Cells were treated with IL-1 and TNF-, each at 10 ng ml1 in the absence (vehicle; 0.01% dimethyl sulfoxide) and in the presence of different COX inhibitors (indomethacin, diclofenac, meloxicam, nimesulide, and L-745,337 each at 105 M; aspirin at 104 M) with increasing selectivity, COX-1 versus COX-2 (left to right). Drug-treated groups were compared with that of vehicle by the one-sample t test for normalized data. *P < 0.05; **P < 0.01; ***P < 0.001.

Effect of PGE₂ and Fenoterol on GM-CSF Release by HASM Cells

PGE₂ (10⁷ to 10⁵ M) had no effect on GM-CSF release by HASM cells stimulated with cytokines (10 µM concentration only shown; Figure 3A). By contrast, when endogenous PGE₂ production was blocked with indomethacin (10⁶ M), exogenous PGE₂ inhibited GM-CSF by HASM cells stimulated with the combination of IL-1 and TNF- (Figure 3B). In separate experiments, the ₂-agonist fenoterol (10⁷ to 10⁵ M), like PGE₂, did not affect GM-CSF release by HASM cells stimulated with cytokines (Figure 4A). However, again as was observed with PGE₂, fenoterol reduced GM-CSF release by cells treated with cytokines plus indomethacin (Figure 4B). The stable analogue of prostacyclin, cicaprost (10⁸ to 10⁵ M), did not modify GM-CSF release in HASM cells stimulated with cytokines either in the absence or presence of indomethacin (data not shown). Please note that the experiments depicted in components A and B of Figures 3 and 4 were not paired.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Effect of exogenously applied PGE2 (105 M) on GM-CSF release by HASM cells stimulated with cytokines (TNF- plus IL-1; 10 ng ml1; cytmix) in the absence (A) or presence (B) of indomethacin (106 M). Results are shown as the mean ± SEM of nine determinations from three patients. Responses of treated cells were compared with control by Kruskal-Wallis nonparametric ANOVA followed by Dunn's test for multiple comparisons. *P < 0.05; **P < 0.01; ***P < 0.001.

View larger version (13K): [in this window] [in a new window]   Figure 4.   Effect of fenoterol (1 × 105 M) on GM-CSF release by HASM stimulated with IL-1 and TNF- (10 ng ml1 for each; cytmix) in the absence (A) or presence (B) of indomethacin (106 M). Results are shown as the mean ± SEM of nine determinations from three patients. Responses of treated cells were compared with control by Kruskal-Wallis nonparametric ANOVA followed by Dunn's test for multiple comparisons.

Effect of the Phosphodiesterase Type 4 Inhibitors Rolipram and SB 207499 on GM-CSF Release by HASM Cells

The phosphodiesterase enzyme (PDE) type 4 selective inhibitor rolipram (10⁸ to 10⁵ M) significantly inhibited GM-CSF release generated by HASM cells stimulated with cytokines (Figure 5A). Similarly, another structurally unrelated PDE type 4 inhibitor, SB 207499, inhibited GM-CSF production in HASM cells exposed to cytokines (Figure 5B). Further experiments showed that the effects of rolipram (10⁵ M; 24.74 ± 7.53% of control) on GM-CSF release by cytokine-treated cells was not altered when cells were pretreated with indomethacin (19.16 ± 11.93% of control; n = 9).

View larger version (17K): [in this window] [in a new window]   Figure 5.   The effect of rolipram (108 to 105 M) (A) and SB 207499 (108 to 105 M) (B) on GM-CSF release from HASM cells exposed to a mixture of cytokines (cytmix: IL-1 and TNF- each at 10 ng ml1). Results are shown as the mean ± SEM of nine determinations from three patients. Significant differences between drug- and vehicle-treated cells were calculated by Kruskal-Wallis nonparametric ANOVA followed by Dunn's test for multiple comparisons. *P < 0.05; **P < 0.01.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

It is now clear from the work of our group and that of others that airway smooth muscle cells can be manipulated to express a secretory phenotype and thereby become active in inflammatory responses. Here we have confirmed our previous findings that HASM releases both GM-CSF and COX products when stimulated with the inflammatory cytokines IL-1 and TNF-. Because GM-CSF increases the survival of eosinophils and thereby mediates some of the aspects of allergic lung disease (20), it is thought that agents or pathways that modulate this cytokine may be of physiologic or therapeutic relevance for the understanding/treatment of patients with such conditions. Here we demonstrate the presence of a negative feedback mechanism by which COX activity limits GM-CSF production by HASM cells.

We have previously shown that HASM cells express COX-1 under control culture conditions and COX-2 after stimulation with cytokines (6, 15). Here we show that a range of NSAIDs, including the COX-2 selective inhibitor L745337, increases GM-CSF production when PGE₂ release is blocked. However, despite the same apparent level of COX reduction (i.e., 100%), there was some degree of differential effect displayed by the different NSAIDs on the amount of GM-CSF stimulated. In fact, they elevated GM-CSF production with the following rank order of efficacy: indomethacin > aspirin > diclofenac > meloxicam > L-745337  nimesulide. This observation was not expected because the ordering of drugs fits exactly with their selectivity as inhibitors of COX-1 over COX-2 (21). There are two possible explanations for these data: (1) different NSAIDs increase GM-CSF production by prostanoid- dependent and -independent pathways, and (2) in these cells, despite a clear induction of COX-2, COX-1 regulates the production of GM-CSF. Data presented in this study and others (15) show that COX-2 selective inhibitors block PGs released by cells to the outside surrounding medium. However, it is not clear whether there are residual levels of COX products released intracellularly that are not detected in our experiments. Thus, for our second explanation to be valid, there would have to be some mechanism for compartmentalization of products from the different forms of COX. Such a mechanism could take the form of intracellular lipid bodies, which have been described in some cells (22). Whether a similar phenomenon exists in airway smooth muscle remains to be established.

PGE₂ is the predominant prostanoid released by HASM cells treated with cytokines (15). Thus, in order to further understand the mechanisms of NSAID-induced GM-CSF production, we assessed the effects of PGE₂ on cytokine production by these cells. Indeed, we found that exogenous PGE₂ inhibited GM-CSF production. This observation adds weight to the mechanism of NSAID-induced GM-CSF production being due to inhibition of COX activity. However, we also found that inhibition of endogenous COX activity was required before any inhibitory effects of PGE₂ were observed. This suggests that PGE₂ production by the cells was in excess of that required to modulate GM-CSF production.

PGE₂ activates EP₂ and EP₄ (prostanoid) receptors on airway smooth muscle (23). These receptors are coupled via stimulatory G protein (Gs) to the membrane-bound enzyme adenylyl cyclase, activation of which leads to increased production of cAMP. Thus, it is likely that PGE₂ inhibits GM-CSF production by HASM cells by stimulating the production of cAMP. This is supported by two other protocols used in this study. First, the ₂-adrenoceptor agonist fenoterol, which shares a similar receptor-mediated signal transduction pathway with PGs (i.e., activation of adenylyl cyclase), also reduced GM-CSF production by these cells. Second, inhibition of the degradation of cAMP with inhibitors of type 4 PDE enzymes reduced GM-CSF production. In the case of fenoterol, like PGE₂, the presence of an NSAID was required for it to reduce GM-CSF. Because cytokines such as IL-1 do not alter the number or distribution of adrenoceptors (24), this observation can be explained by the heterologous desensitization of adenylyl cyclase by -agonists and PGE₂ in these cells (16). Most recently, the ability of TNF- to increase G_i-2 without altering G_s has been suggested as a possible mechanism of adenylate cyclase desensitization (25). Whether this process involves COX metabolites remains to be investigated. By contrast to results obtained with PGE₂, cicaprost, which preferentially activates prostacyclin receptors (IP), had no effect on GM-CSF production either in the presence or absence of NSAIDs. This observation is in direct contrast to other work from our group using human vascular smooth muscle cells (26). In vascular smooth muscle, IP, and not EP, ligands potently reduce GM-CSF production.

Several PDE isoenzymes (at least eleven families) have been identified in airway smooth muscle, although the proportions of these enzymes varies between species. In HASM cells, types 3 and 4, which degrade cAMP, and type 5, which breaks down cyclic guanosine monophosphate, are functionally important (27). As mentioned previously, we found that two structurally different type 4 inhibitors reduced GM-CSF production by cytokine-activated cells. These observations were most probably due to an increase in the level of cAMP in the HASM cells after PDE inhibition. Initially, we thought that PGE₂ production by these cells would account for the majority of the adenylyl cyclase activity. However, the inhibitory effects of both PDE inhibitors tested was unaffected by indomethacin. This suggests that there are either other mediators present that activate adenylyl cyclase or that there is a high "basal" activity of this enzyme.

In conclusion, when HASM cells are activated with cytokines, they release PGE₂, which acts as a braking mechanism to limit the coproduction of GM-CSF. In addition to PGE₂, other elevators of cAMP also reduce GM-CSF production by these cells. These observations have two implications for our understanding of airway disease. First, the NSAID-induced increase in GM-CSF production may explain why these drugs are not of therapeutic benefit in allergic airway disease. Moreover, this pathway may be particularly developed in the subgroup of asthmatic patients that react clinically to aspirin and related drugs. Second, -adrenergic agonists or PDE 4 inhibitors, which elevate cAMP, may owe some of their therapeutic benefits in the treatment of asthma to reduction in GM-CSF.