Identification in human airways smooth muscle cells of the prostanoid receptor and signalling pathway through which PGE₂ inhibits the release of GM-CSF

Abstract

1. The prostanoid receptor(s) on human airways smooth muscle (HASM) cells that mediates the inhibitory effect of PGE₂ on interleukin (IL)-1β-induced granulocyte/macrophage colony-stimulating factor (GM-CSF) release has been classified.

2. IL-1β evoked the release of GM-CSF from HASM cells, which was suppressed by PGE₂, 16,16-dimethyl PGE₂ (nonselective), misoprostol (EP₂/EP₃-selective), ONO-AE1-259 and butaprost (both EP₂-selective) with pIC₅₀ values of 8.61, 7.13, 5.64, 8.79 and 5.43, respectively. EP-receptor agonists that have selectivity for the EP₁- (17-phenyl-ω-trinor PGE₂) and EP₃-receptor (sulprostone) subtypes as well as cicaprost (IP-selective), PGD₂, PGF_2α and U-46619 (TP-selective) were poorly active or inactive at concentrations up to 10 μM.

3. AH 6809, a drug that can be used to selectively block EP₂-receptors in HASM cells, antagonised the inhibitory effect of PGE₂, 16,16-dimethyl PGE₂ and ONO-AE1-259 with apparent pA₂ values of 5.85, 6.09 and 6.1 respectively. In contrast, the EP₄-receptor antagonists, AH 23848B and L-161,982, failed to displace to the right the concentration–response curves that described the inhibition of GM-CSF release evoked by PGE₂ and ONO-AE1-259.

4. Inhibition of GM-CSF release by PGE₂ and 8-Br-cAMP was abolished in cells infected with an adenovirus vector encoding an inhibitor protein of cAMP-dependent protein kinase (PKA) but not by H-89, a purported small molecule inhibitor of PKA.

5. We conclude that prostanoid receptors of the EP₂-subtype mediate the inhibitory effect of PGE₂ on GM-CSF release from HASM cells by recruiting a PKA-dependent pathway. In addition, the data illustrate that caution should be exercised when using H-89 in studies designed to assess the role of PKA in biological processes.

Introduction

Colony-stimulating factors (CSFs) such as granulocyte/macrophage colony-stimulating factor (GM-CSF) are responsible for the proliferation and differentiation of cells in the bone marrow (Metcalf, 1985; 1986). These cytokines also modulate the function of mature leukocytes, including eosinophils and neutrophils, promoting their activation and survival (Lopez et al., 1986; Giembycz & Lindsay, 1999). It is believed that by increasing the longevity of proinflammatory cells in tissues, CSFs may perpetuate airway inflammatory diseases such as asthma and chronic obstructive pulmonary disease (COPD) where eosinophils and neutrophils play a pathogenic role. It is now appreciated that human airways smooth muscle (HASM) cells have a substantial synthetic capacity and can contribute to inflammatory processes through the generation of a plethora of mediators including cytokines, chemokines and bioactive lipids (Johnson & Knox, 1997). Of relevance to the present study is the finding that HASM cells generate GM-CSF (Saunders et al., 1997; Lazzeri et al., 2001) that can be suppressed by endogenously synthesised or exogenously applied prostaglandin (PG) E₂ (Clarke et al., 2001; Lazzeri et al., 2001). The possibility exists, therefore, that agonism of specific prostaglandin receptors on HASM cells could provide an effective means of suppressing CSF production in diseases such as asthma and COPD and reduce, respectively, the eosinophil and neutrophil burden in the airways.

Five main classes of G-protein-coupled receptor for the naturally occurring prostanoid agonists have been defined and given the prefix DP, EP, FP, IP and TP (Coleman et al., 1994b; Breyer et al., 2001; Tsuboi et al., 2002) where the first letter refers to the natural ligand most selective for that receptor. Molecular biological techniques subsequently confirmed this pharmacological classification with the cloning and expression of cDNAs for representatives of the five prostanoid receptors in a number of species including humans (Coleman et al., 1994b; Breyer et al., 2001; Tsuboi et al., 2002). The finding that PGE₂ potently inhibits the release of GM-CSF from HASM cells in response to IL (interleukin)-1β (Clarke et al., 2001; Lazzeri et al., 2001) suggests that this response is mediated by one or more prostanoid receptors of the EP-subtype. Currently, pharmacological evidence and primary sequence information of partial and full-length cDNA clones indicates the presence of at least four EP-receptor variants denoted EP₁, EP₂, EP₃ and EP₄ (Coleman et al., 1994b; Narumiya et al., 1999; Breyer et al., 2001). These receptors can couple to several effector molecules (Breyer et al., 2001), thereby mediating a diverse array of biological responses (Narumiya et al., 1999; Kobayashi & Narumiya, 2002; Tsuboi et al., 2002) that can now be reasonably well defined with agonists and antagonists that can discriminate between EP-receptor subtypes.

The objective of the present study was to characterise the prostanoid receptor(s) through which PGE₂ inhibits GM-CSF release from IL-1β-stimulated HASM cells and probe the molecular basis of this effect. To this end, naturally occurring and synthetic prostanoid agonists and antagonists that have selectivity for the EP-receptor subtypes were used. In addition, the role of the cAMP/cAMP-dependent protein kinase (PKA) cascade was assessed using an adenovirus (Ad5) vector encoding the α-isoform of PKA inhibitor protein (PKIα), which is potent, extremely selective for PKA (Olsen & Uhler, 1991; Scarpetta & Uhler, 1993; Collins & Uhler, 1997) and devoid of the problems associated with many small molecule protein kinase inhibitors (Engh et al., 1996; Davies et al., 2000).

Methods

Isolation of HASM cells

Tracheal rings from either lung or heart and lung transplantation donors (7 female, 17 male, age range: 17–57 years; median age 36.5 years) 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 penicillin (100 U m1⁻¹), streptomycin (100 μg ml⁻¹) and amphotericin B (2.5 μg ml⁻¹). The smooth muscle layer was dissected free of adherent connective tissue and cartilage, and the epithelium was removed using a rounded scalpel blade. The smooth muscle was incubated (30 min; 37°C; 5% CO₂/air) in HBSS containing BSA (10 mg ml⁻¹), collagenase (type XI, 1 mg ml⁻¹) and elastase (type I, 3.3 U ml⁻¹). 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 described above with the elastase concentration increased to 15 U/ml. Dissociated cells were centrifuged (100 × g, 5 min, 4°C) and resuspended in Dulbecco's-modified Eagle's medium (DMEM) containing heat-inactivated foetal calf serum (FCS; 10% v v⁻¹), sodium pyruvate (1 mM), L-glutamate (2 mM), nonessential amino acids (1 ×) and the antimicrobial agents detailed above.

Primary culture of HASM cells

The HASM cell suspension was placed in a tissue culture flask (75 cm²) containing 6 ml supplemented DMEM and allowed to adhere (∼12 h) at 37°C in 5% CO₂/air. The culture medium was replaced after 4–5 days (12 ml) and thereafter every 3–4 days. When the cells reached confluence (∼10–14 days) and demonstrated a typical ‘hill and valley' appearance and positive immunostaining for α-actin (routinely >95%, Belvisi et al., 1997), they were seeded onto either 96-well plates (Costar UK Ltd., High Wycombe) at an initial density of 2000 cells per well or six well plates (Costar) at an initial density of 20,000 cell per well for cytokine release and Western blotting experiments, respectively. At subconfluence, the cells were growth arrested by being placed in DMEM containing apotransferin (5 μg ml⁻¹), insulin (1 μM), ascorbate (100 μM) and BSA (0.1% w v⁻¹) for 24 h. The medium was replaced with DMEM containing 3% FCSv v⁻¹ and drugs or appropriate vehicles as described below.

Infection of HASM cells with Ad5.CMV.PKIα

In some experiments, subconfluent, growth-arrested HASM cells were infected (MOI=100) with an E1⁻/E3⁻ replication-deficient Ad5 vector (Ad5.CMV.PKIα) containing a 251 bp DNA fragment encoding the complete amino-acid sequence of PKIα (Day et al., 1989; Meja et al., 2004), downstream of the constitutively active CMV immediately early promoter (Gomez-Foix et al., 1992; Lum et al., 1999). After 48 h, cells were processed for GM-CSF release and Western blotting as described below. To control for possible biological effects of the virus per se, the vector, Ad5.CMV.Null, expressing no transgene, was used in parallel. Preliminary experiments using immunofluorescence microscopy established that >95% of cells expressed PKIα 48 h after infection. This was determined by enumerating the number of HASM cells expressing the transgene as a percentage of total number of cells that counterstained with the nuclear marker, 4′,6-diamidino-2-phenylindole. Ad5.CMV.PKIα and Ad5.CMV.Null at an MOI of 100 had no effect on HASM cell viability 48 h after infection (data not shown).

Measurement of GM-CSF

HASM cells (naïve and virus-treated) were pretreated for 30 min with indomethacin (10 μM) and, where indicated, receptor antagonist before being exposed for a further 5 min or 30 min to prostanoid agonists or 8-Br-cAMP, respectively. IL-1β (100 pg ml⁻¹) was then added and the cells were incubated at 37°C in a thermostatically controlled incubator under a 5% CO₂ atmosphere. At 24 h, the amount of GM-CSF released into the culture supernatant was quantified by a sandwich ELISA (human DuoSet® development system, R&D Systems Europe, Abingdon) according to the manufacturer's instructions. The detection limit of this assay is 7.8 pg ml⁻¹.

Western blot analysis

HASM cells were treated with 3% FCS for 24 h. The medium was removed and the cells washed with HBSS, lysed and proteins extracted in 20 mM Tris HCl – pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.1% (v v⁻¹) Nonidet P-40, 0.05% (w v⁻¹) sodium deoxycholate, 0.025% (w v⁻¹) SDS and 0.1% (v v⁻¹) Triton X-100 supplemented with PMSF (0.1 mg ml⁻¹), leupeptin (10 μg ml⁻¹) and aprotinin (25 μg ml⁻¹). Insoluble protein was removed by centrifugation (10,000 × g; 3 min) and aliquots of the resulting supernatant were diluted 1 : 4 in Laemmeli buffer (62.5 mM Tris-HCl – pH 6.8, 10% (v v⁻¹) glycerol, 1% (w v⁻¹) SDS, 1% (v v⁻¹) β-mercaptoethanol, 0.01% (v v⁻¹) bromophenol blue) and boiled for 5 min. Denatured proteins (25 μg) were separated by SDS–PAGE using a 4–20% gradient gel (BioRad; Hemel Hempstead) and transferred to Hybond enhanced chemiluminescence (ECL) membranes (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire) in Tris buffer (Tris base 50 mM – pH 8.3, glycine 192 mM, 20% v v⁻¹ methanol). The nitrocellulose was incubated overnight in Tris-buffered saline plus Tween-20 (TBS-T) (25 mM Tris-base – pH 7.4, 150 mM NaCl, 0.1% v v⁻¹ Tween 20) containing 5% (w v⁻¹) non-fat dry milk. The filters were incubated for 1 h at room temperature in TBS-T containing 5% non-fat dry milk and an anti-human PKIα, pCREB or β-actin polyclonal antibody (diluted 1 : 500, 1 : 1000 and 1 : 2000, respectively) as indicated. Membranes were washed with TBS-T and incubated with horseradish peroxide-conjugated sheep, anti-rabbit IgG (diluted 1 : 4000) in TBS-T/5% nonfat dry milk for 1 h at room temperature. The nitrocellulose was washed again in TBS-T and developed using ECL™ western blotting detection reagents on Kodak X-OMAT-S film.

Cell viability

At the end of each experiment, cell viability was determined colorimetrically by measuring the reduction of the tetrazolium salt, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide (MTT), to formazan by mitochondrial dehydrogenases (Mosmann, 1983; Hussain et al., 1993).

Drugs, antibodies and analytical reagents

IL-1β was from R&D systems (Abingdon, Oxon, U.K.), DMEM and HBSS were from Flow Laboratories (Rickmansworth, Hertfordshire, U.K.) and nonessential amino acids were purchased from Life Technologies (Paisley, U.K.). Indomethacin, FCS, MTT, PGD₂, PGE₂, PGF_2α, U-46619 (9,11-dideoxy-9α,11α-methanoepoxy prostaglandin F_2α) and H-89 were from Sigma-Aldrich (Poole, Dorset, U.K.). 16,16-Dimethyl PGE₂, 17-phenyl-ω-trinor PGE₂, AH 6809 (6-isopropoxy-9-oxoxanthine-2-carboxylic acid), SQ 29,548 (1S-[1α,2α(Z),3α,4α]]-7-[3-[[2-[(phenylamino) carbonyl]hydrazino]methyl]-7-oxa bicyclo[2.2.1]hept-2-y1]-5-heptenoic acid), AH 23848B ([1α(Z),2β,5α]-(±)-7-[5-[[(1,1′-biphenyl]-4-ylmethoxy)-3-hydroxy-2-(1-piperidinyl) cyclopentyl]-4-heptanoic acid), sulprostone, misoprostol methyl ester and butaprost methyl ester were obtained from Cayman Chemicals (Ann Arbor, MI, U.S.A.). All other synthetic prostanoid reagents were donated by the following: cicaprost (Schering AG, Berlin, Germany); ONO-AE1-259 ((16S)-9-deoxy-9β-chloro-15-deoxy-16-hydroxy-17,17-trimethylene-19,20-didehydro-PGE₂ – sodium salt; Ono Pharmaceuticals, Osaka, Japan) and L-161,982 ([4′-[3-butyl-5-oxo-1-(2-trifluoromethyl-phenyl)-1,5-dihydro-[1,2,4]triazol-4-ylmethyl]-biphenyl-2-sulphonic acid (3-methyl-thiophene-2-carbonyl)-amide]; Merck Frosst, Montreal, Canada). Goat anti-human PKIα (code sc#1944) and goat anti-human β-actin (code sc# 1615) were from Autogen Bioclear (Calne, Wiltshire, U.K.). pCREB (code 9191S) was purchased from New England Biolabs (Hitchin, Hertfordshire, U.K.).

Data and statistical analyses

Data points, and values in the text and figure legends, represent the mean±s.e.m. of ‘n' independent determinations using tissue from different donors. The concentration–response curves were analysed by least-squares, nonlinear iterative regression with the ‘PRISM' curve fitting program (GraphPad software, San Diego, U.S.A.) and pEC₅₀ and pIC₅₀ values were subsequently interpolated from curves of best-fit. Equieffective molar concentration ratios (e.c.r.) were calculated using the formula: IC₅₀ PGE analogue/IC₅₀ PGE₂.

Antagonist affinity was calculated using the equation pK_B=log (CR-1) – log [B] (Schild, 1949), where CR is the concentration ratio calculated from the EC₅₀ of agonist in the presence of the antagonist divided by the EC₅₀ of the agonist alone, K_B is the equilibrium dissociation constant and [B] is the concentration of antagonist. In the experiments described herein, the term apparent pA₂ is substituted for pK_B as antagonists were used at one concentration only, which precludes assumptions being made about the nature of the antagonism.

Where appropriate, data were analysed statistically using Student's paired t-test or by one-way ANOVA/Newman–Keuls multiple comparison test. The null hypothesis was rejected when P<0.05.

Results

We have reported previously that IL-1β promotes a time- and concentration-dependent release of GM-CSF from HASM cells with a t_1/2 and EC₅₀ of >18 h and 16 pg ml⁻¹, respectively (Clarke et al., 2001). In the experiments described herein, IL-1β was used at 100 pg ml⁻¹ (∼EC₉₀) and GM-CSF was measured in the culture supernatant 24 h after addition of the stimulus. None of the compounds or their vehicles used in these experiments affected cell viability as determined by the reduction of MTT to formazan. None of the vehicles used had any significant effect on GM-CSF release (data not shown).

Effect of naturally occurring prostaglandins, cicaprost and U-46619 on GM-CSF release

PGD₂, PGE₂, PGF_2α, U-46619 and cicaprost had no effect on GM-CSF release from indomethacin (10 μM; 30 min)-pretreated HASM cells. However, the GM-CSF elaborated in response to IL-1β was suppressed by PGE₂ in a concentration-dependent manner with a maximal effect and pIC₅₀ of 79% and 8.61, respectively (Figure 1, Table 1 ). Cicaprost was also active but the concentration–response curves that described the suppression of GM-CSF were complex and better described by a two-site (mean r²=0.956) rather than single sigmoidal function (mean r²=0.765; Figure 1). Analysing the data in this way yielded pIC₅₀ values for the high- and low-affinity components of 9.45, which accounted for 34.1±5.5% of the effect, and 4.69 (estimate), respectively. PGD₂, PGF_2α and U-46619 had no significant effect on GM-CSF release (Figure 1, Table 1).

Figure 1.

Effect of prostanoids on IL-1β-induced GM-CSF release. Adherent HASM cells were pretreated for 5 min with varying concentrations of PGD₂, PGE₂, PGF_2α, cicaprost and U-46619 before being exposed to IL-1β (100 pg ml⁻¹). Cells were maintained at 37°C in a thermostatically controlled incubator under a 5% CO₂ atmosphere and the amount of GM-CSF released into the culture supernatant was quantified at 24 h by a sandwich ELISA. Each data point represents the mean±s.e.m. of four to 13 determinations (see Table 1) using tissue from different donors. Indomethacin (10 μM) was present through the experiment. Note that the data for PGD₂ and PGF_2α are superimposed. See Methods for further details.

Table 1.

Potency of prostanoid and EP/IP-selective agonists at suppressing GM-CSF generation

Prostanoid	n	Receptor selectivitya	Inhibition of GM-CSF release pIC50	e.c.r. PGE2=1
PGE2	13	EP1≈EP2 ≈EP3≈EP4	8.61±0.43 (78.6±6.2)	1
ONO-AE1-259	8	EP2>>>EP1=EP3=EP4	8.79±0.26 (8.79±0.3)	0.7
16,16-dimethyl PGE2	7	EP2⩾EP3=EP1>EP4	7.13±0.37 (83.0±9.2)	30
Misoprostol	5	EP2=EP3>EP1>EP4	5.64±0.25 (64.7±6.9)	935
Butaprost	4	EP2>>EP1>EP3>EP4	5.43±0.26 (56.1±10.9)	1518
Cicaprostb	11	IP>EP4	9.45±0.26	0.14
			>5	>4081
PGF2α	4	FP	>5 (12.5±17.9%)	>4081
PGD2	4	DP	>5 (13.4±13.6%)	>4081
U-46619	9	TP	>5 (−22.1±11.2%)	>4081
17-phenyl-ω-trinor PGE2	4	EP1>EP3>EP2>EP4	>5 (9.7±11.8%)	>4081
Sulprostone	4	EP3>EP1>>EP2>EP4	>5 (−22.7±8.6%)	>4081

Values in parentheses show the percentage inhibition of GM-CSF release evoked by 10 μM prostanoid agonist. Equieffective concentration ratios >1 and <1 indicate that the agonist in question is less and more potent than PGE₂, respectively.

^aSelectivity derived from studies in isolated cells and tissues.

^bConcentration–response curve described by a two site sigmoidal function.

Effect of EP-selective prostanoid agonists on GM-CSF release

To gain information on the possible role of an EP-receptor subtype in mediating the inhibitory effect of PGE₂ on GM-CSF release, a variety of PGE analogues, which differ in their selectivity for the EP₁-, EP₂-, EP₃- and EP₄-receptor subtypes, were examined (Figure 2, Table 1). By themselves none of the ligands had any effect on GM-CSF release. However, in IL-1β-stimulated HASM cells, 16,16-dimethyl PGE₂, ONO-AE1-259, misoprostol and butaprost suppressed the release of GM-CSF in a concentration-dependent manner, with a maximum inhibition of 83, 68, 65 and 56%, respectively, at the highest concentration (10 μM) of drug examined (Figure 2). The rank order of agonist potency was ONO-AE1-259 >16,16-dimethyl PGE₂ > misoprostol > butaprost. 17-Phenyl-ω-trinor PGE₂ suppressed GM-CSF output at concentrations above 1 μM whereas sulprostone was inactive at all concentrations examined (Figure 2, Table 1).

Figure 2.

Effect of EP-selective prostanoid agonists on IL-1β-induced GM-CSF release. Adherent HASM cells were pretreated for 5 min with varying concentrations of six synthetic PGE analogues before being exposed to IL-1β (100 pg ml⁻¹). Cells were maintained at 37°C in a thermostatically controlled incubator under a 5% CO₂ atmosphere and the amount of GM-CSF released into the culture supernatant was quantified at 24 h by a sandwich ELISA. Each data point represents the mean±s.e.m. of four to 13 determinations (see Table 1) using tissue from different donors. Indomethacin (10 μM) was present through the experiment. See Methods for further details.

Effect of an EP₂-receptor antagonist on the inhibition of GM-CSF release evoked by PGE₂, 16,16-dimethyl PGE₂ and ONO-AE1-259

AH 6809 is an antagonist at the human EP₁-, EP₂- and DP-receptor subtypes (Coleman et al., 1987; Keery & Lumley, 1988). As 17-phenyl-ω-trinor PGE₂ (EP₁-selective) and PGD₂ were inactive in this system (Table 1, Figure 2), AH 6809 was used to determine if EP₂ receptors mediated the inhibitory effects of PGE₂, 16,16-dimethyl PGE₂ and ONO-AE1-259 on GM-CSF release. Pretreatment of HASM cells with AH 6809 (10 μM) produced a rightwards shift of the concentration–response curve that described the inhibition of GM-CSF release by PGE₂, 16,16-dimethyl PGE₂ and ONO-AE1-259 from which apparent pA₂ values of 5.85±0.31, 6.09±0.22 and 6.1±0.7 were derived, respectively (Figure 3). These affinity estimates were not statistically significant from one another (P>0.05; one-way ANOVA).

Figure 3.

Effect of AH 6809 on the inhibition of IL-1β-induced GM-CSF evoked by PGE₂, 16,16-dimethyl PGE₂ and ONO-AE1-259. Adherent HASM cells were pretreated (30 min) concurrently with indomethacin and AH 6809 (both 10 μM) and then for a further 5 min with PGE₂ (a), 16,16-dimethyl (b) or ONO-AE1-259 (c). IL-1β (100 pg ml⁻¹) was then added to the cells and the GM-CSF released into the culture medium was quantified at 24 h by a sandwich ELISA. Each data point represents the mean±s.e.m. of three to five determinations using tissue from different donors.

Effect of EP₄-receptor antagonists on the inhibition of GM-CSF release evoked by PGE₂ and ONO-AE1-259

To evaluate the role of EP₄-receptors in mediating suppression of GM-CSF release, the antagonists AH 23848B (Coleman et al., 1994a) and L-161,982 (Machwate et al., 2001) were employed using PGE₂ and ONO-AE1-259 as agonists. In the presence of AH 23848B or L-161,982, at concentrations ∼10 and ∼100 times higher than their affinity at EP₄-receptors, respectively, the inhibitory effect of PGE₂ and ONO-AE1-259 was not significantly altered in terms of IC₅₀ or maximal inhibitory response (Figure 4).

Figure 4.

Effect of AH 23848B and L-161,982 on the inhibition of IL-1β-induced GM-CSF release evoked by PGE₂ and ONO-AE1-259. Adherent HASM cells were pretreated (30 min) concurrently with indomethacin (10 μM) and either AH 23848B (30 μM) or L-161,982 (2 μM) and then for a further 5 min with PGE₂ (a and b) or ONO-AE1-259 (c and d). IL-1β (100 pg ml⁻¹) was then added to the cells and the GM-CSF released in to the culture medium was quantified at 24 h by a sandwich ELISA. Each data point represents the mean±s.e.m. of three to five determinations using tissue from different donors.

Effect of 8-Br-cAMP on IL-1β-induced GM-CSF release

Pretreatment of HASM cells with 8-Br-cAMP inhibited the release of GM-CSF from IL-1β (100 pg/ml)-stimulated HASM cells with a pIC₅₀ of 3.76±0.14 (n=3). At the highest concentration of drug tested (1 mM), the elaboration of GM-CSF was suppressed by 96.2±3.1% (n=3).

Role of PKA in the action of PGE₂ on GM-CSF release

To determine the role of PKA in mediating the effect of PGE₂ on GM-CSF release from HASM cells H-89, a purported selective inhibitor of PKA (Chijiwa et al., 1990), and a virus vector, Ad5.CMV.PKIα (Lum et al., 1999; Meja et al., 2004), containing DNA encoding the complete amino-acid sequence of PKIα, an endogenous, potent and highly selective inhibitor of PKA (Glass et al., 1986; Olsen & Uhler, 1991) were employed. In the virus studies, western blotting was used to confirm expression of the PKIα transgene. In uninfected cells, PKIα was not detected in any experiment. However, 48 h after infection with Ad5.CMV.PKIα (MOI=100), a single peptide was labelled by the anti-PKIα antibody that migrated as a 12 kDa band on SDS polyacrylamide gels (Figure 5). In preliminary studies, the efficiency of transgene expression at 48 h was found to be >95% (see Methods for details).

Figure 5.

Expression of PKIα in HASM cells infected with Ad5.CMV.PKIα. Adherent cells were cultured until 50% confluent and then infected with Ad5.CMV.Null, Ad5.CMV.PKIα (MOI=100) or left untreated (naïve) for 48 h at 37°C. Cells were growth arrested in serum-free medium and processed by western blotting for PKIα expression (a) and the house-keeping protein, β-actin (b). Data are representative of three independent determinations using tissue from different donors. See Methods for further details. (Key: lanes 1 & 2, Naïve; lanes 3 & 4, Ad5.CMV.Null; lanes 5 & 6, Ad5.CMV.PKIα).

HASM cells exposed to IL-1β elaborated GM-CSF in an amount that was not significantly altered by H-89 (10 μM; 30 min) or following infection with Ad5.CMV.PKIα or Ad5.CMV.Null (MOI=100; 48 h) (Figure 6a,b). PGE₂ (10 μM) and 8-Br-cAMP (1 mM), which was used as a positive control, suppressed IL-1β-stimulated GM-CSF release by a mechanism that was prevented in cells expressing the PKIα transgene but not those infected with the empty virus or pretreated with H-89 (Figure 6a,b). Paradoxically, the phosphorylation by 8-Br-cAMP and PGE₂ of CREB, a well established substrate for PKA, was abolished in HASM cells infected with Ad5.CMV.PKIα and H-89 confirming that PKA was inhibited by both interventions (Figure 6c).

Figure 6.

Effect of H-89 and PKIα on 8-Br-cAMP- and PGE₂-induced GM-CSF release from and CREB phosphorylation in HASM cells. Adherent cells were pretreated with H-89 (10 μM; 30 min) or infected with Ad5.CMV.Null or Ad5.CMV.PKIα (MOI=100; 48 h). 8-Br-cAMP (1 mM) or PGE₂ (10 μM) were then added for 30 min. At this point, cells were either processed immediately for CREB phosphorylation by western blotting (a) or exposed to IL-1β (100 pg ml⁻¹) for 24 h to promote GM-CSF release, which was measured by ELISA (b and c). Each bar represents the mean±s.e.m. of four independent determinations using tissue from different donors. Indomethacin (10 μM) was present throughout the experiment. See Methods for further details. ^*P<0.05, significant inhibition of IL-1β-stimulated GM-CSF release. (Key: lanes 1 & 2, Naïve; lanes 3 & 4, Ad5.CMV.Null; lanes 5 & 6, Ad5.CMV.PKIα, lanes 7 & 8, H-89).

Discussion

We have reported previously that endogenously synthesised and exogenously applied PGE₂ can negatively regulate the elaboration of GM-CSF from HASM cells elicited by the proinflammatory cytokine, IL-1β (Clarke et al., 2001; Lazzeri et al., 2001). The experiments described in the present study were designed to further this work by characterising the prostanoid receptor(s) and second messenger pathway through which PGE₂ mediates this inhibitory effect. As PGE₂ was the most potent prostanoid at suppressing GM-CSF release from HASM cells, we reasoned that this might reflect the activation of an EP-receptor subtype. This conclusion was supported by the additional findings that cicaprost, an IP-receptor agonist, was a weak inhibitor of GM-CSF release and U-46619 (TP-agonist), PGD₂ and PGF_2α were inactive.

To identify the inhibitory EP-receptor(s) on HASM cells, a pharmacological analysis was performed by examining a panel of agonists and antagonists that have varying degrees of selectivity for the EP-receptor subtypes. The finding that 17-phenyl-ω-trinor PGE₂ was greater than three orders of magnitude less active than PGE₂ in this system provides persuasive evidence that the elaboration of GM-CSF from HASM cells is not mediated by receptors of the EP₁-subtype (Lawrence et al., 1992).

EP₂- and EP₄-receptors can couple to adenylyl cyclase via Gs and enhance cAMP biosynthesis (Coleman et al., 1994a; Breyer et al., 2001). As 8-Br-cAMP was found to suppress GM-CSF release from HASM cells, it was considered likely that the EP₂- and/or EP₄-receptor might mediate the inhibitory effect of PGE₂ and related ligands in this system. Although agonism of certain splice variants of the human EP₃-receptor (e.g. EP_3B, EP_3C) can also couple positively to adenylyl cyclase (Namba et al., 1993), the failure of sulprostone to suppress GM-CSF release suggests that the EP₃-subtype does not regulate the expression of the CSF2 gene in HASM cells. As illustrated in Figure 3, PGE₂, 16,16-dimethyl PGE₂, butaprost, misoprostol and ONO-AE1-259 effectively reduced the elaboration of GM-CSF. Evidence that these ligands were acting through EP₂-receptors was derived primarily from two pieces of data. First, the agonist activity of butaprost and ONO-AE1-259 reported in the present study is considered diagnostic of EP₂-receptors (Boie et al., 1997; Kiriyama et al., 1997; Suzawa et al., 2000). Second, AH 6809 antagonised the inhibitory effect of PGE₂ and related ligands with an affinity consistent with EP₂-receptor-mediated events in other tissues including human myometrium (pA₂=5.85, (Hillock & Crankshaw, 1999) and the human recombinant EP₂-subtype expressed in Chinese hamster ovary (CHO) cells (pA₂=6.5 (Woodward et al., 1995). It is noteworthy that AH 6809 is also an antagonist at the EP₁- and DP-receptor (Coleman et al., 1987; Keery & Lumley, 1988; Woodward et al., 1995). However, as neither 17-phenyl-ω-trinor PGE₂ nor PGD₂ inhibited GM-CSF release, AH 6809 was used as a selective antagonist at the human EP₂-receptor. Despite antagonist affinity estimates implicating EP₂-receptors in mediating the inhibition of GM-CSF release, several observations were not entirely consistent with this conclusion. For example, a prediction based on data obtained with established EP₂-receptor-containing tissues such as the rabbit ear artery and cat trachea is that butaprost should be 10- to 100-times less potent than PGE₂ (Gardiner, 1986; Coleman et al., 1988; Nials et al., 1993; Lydford et al., 1996). However, in HASM cells, butaprost was considerably weaker than predicted with an e.c.r. of 1518 relative to PGE₂. Similarly, while misoprostol effectively reduced GM-CSF release, it was 935-times less potent than PGE₂ unlike its activity on the rabbit ear artery and the cat trachea where it is equieffective with the natural ligand (Coleman et al., 1988; Lydford et al., 1996). Consequently, the rank order of agonist potency for the suppression of GM-CSF output from HASM cells was markedly different from other EP₂-receptor-containing tissues. However, rather than invoke the involvement of multiple or a novel EP-receptor subtype(s), we suggest that a likely cause of these discrepancies is our use of the methyl esters of butaprost and misoprostol, which are considerably less potent (typically 10–300-times) than their respective free acids (Tsai et al., 1991; Abramovitz et al., 2000; Tani et al., 2001). For example, butaprost free acid has an affinity (K_i∼80 nM) for the human and murine recombinant EP₂-receptor subtype that is 39- and 32-fold higher than butaprost methyl ester (K_i∼2.9 μM) respectively (Abramovitz et al., 2000; Tani et al., 2001). A more dramatic difference is seen with misoprostol where the free acid is >300-times more potent (K_i=34 nM) than the methyl ester (K_i=10.3 μM) (Abramovitz et al., 2000). If this interpretation is correct, then the present study highlights a fundamental difference between HASM cells and other EP₂-receptor-expressing preparations where the methyl esters of misoprostol and butaprost have the expected e.c.r. relative to PGE₂. Mechanistically, this would imply that unlike many tissues HASM cells lack appreciable esterase activity and cannot convert esterified compounds such as misoprostol and butaprost to their free acids. Thus, these apparent anomalies not withstanding, the data obtained in the present study provide good evidence that PGE₂ suppresses GM-CSF release by interacting with prostanoid receptors of the EP₂-subtype. This conclusion is consistent with the expression by cultured HASM cells of mRNA and protein for the EP₂-receptor subtype (Clarke, D.L., Belvisi, M.G., Smith, S.J., Yacoub, M.H., Meja, K.K., Newton, R., Slater, D.M., Giembycz, M.A. – manuscript under review).

In other human tissues including synovial fibroblasts (Inoue et al., 2002), monocyte-derived macrophages (Takayama et al., 2002) and peripheral blood mononuclear cells (Takahashi et al., 2002), PGE₂ has been shown to regulate cytokine, chemokine and adhesion molecule expression by acting through EP₄-receptors. The failure of two EP₄-receptor antagonists, AH 23848B and L-161,982 (Coleman et al., 1994a; Machwate et al., 2001), to displace to the right the concentration-response curves that described the inhibition of GM-CSF release by PGE₂ and ONO-AE1-259 at a concentration ∼10 and 100 times greater than their affinity at EP₄-receptors, respectively, indicates that this is not the case in HASM cells. This conclusion is further supported by the finding that PGE₂ did not suppress the elaboration of GM-CSF in the subnanomolar concentration range, which is typical of many other EP₄-receptor-expressing preparations (Coleman et al., 1994b).

Agonism of EP₂-receptors evokes responses that are thought to rely exclusively on the activation of the cAMP/PKA cascade (Tsuboi et al., 2002). However, in many biological systems, compelling evidence implicating this signalling pathway is difficult to gain as many compounds marketed as PKA inhibitors, such as H-89, are isoquinolinesulphonyl derivatives, which are remarkably nonselective (Davies et al., 2000), presumably because they block a conserved ATP-binding site found among many protein kinases (Engh et al., 1996). The reported limitations in H-89 prompted us to adopt an additional approach to establish the role of PKA in the inhibition, by PGE₂ and 8-Br-cAMP, of IL-1β-induced GM-CSF release. Thus, HASM cells were infected with an adenovirus vector encoding the complete amino-acid sequence of PKIα, one of three endogenous, potent (K_i∼50–100 pM) and highly selective inhibitors of PKA (Olsen & Uhler, 1991; Scarpetta & Uhler, 1993; Collins & Uhler, 1997; Meja et al., 2004). Indeed, PKIα, at a concentration ∼10⁶ times higher than its affinity for PKA, does not inhibit the highly homologous cGMP-dependent protein kinase (Glass et al., 1986) to which it is most closely related (Takio et al., 1984). As shown in Figure 6, CREB phosphorylation and the inhibition of GM-CSF release evoked by PGE₂ and 8-Br-cAMP was abolished in HASM cells expressing PKIα, indicating that the CSF2 gene is negatively regulated, directly or indirectly, by PKA. H-89 also prevented PGE₂- and 8-Br-cAMP-induced CREB phosphorylation under identical experimental conditions but, in contrast to PKIα-expressing cells, failed to block the reduction in GM-CSF output. These results agree with data published by Takayama et al. (2002), who found that H-89 abolished CREB phosphorylation in human monocyte-derived macrophages evoked by PGE₂ and 8-Br-cAMP but not the inhibition of MIP-1β release. Similarly, the ability of 8-Br-cAMP and β₂-adrenoceptor agonists to suppress the release of eotaxin from TNFα-stimulated HASM cells was not prevented by H-89 (Pang & Knox, 2001). Taken together, these results imply that although H-89 abolishes cAMP-dependent CREB phosphorylation, other signalling intermediates are affected that oppose the functional consequences associated with the inhibition of PKA. The identity of these additional intracellular targets in HASM cells is unknown and was not a subject of this investigation. However, isoquinoline sulphonamides can bind many proteins with high affinity including certain carboxy-terminal domain kinases, which have a profound influence on gene transcription (Dubois et al., 1994). Based on these data, we suggest that caution should be exercised when using H-89 in studies designed to assess the role of PKA in biological responses.

Cicaprost (Sturzebecher et al., 1986) also suppressed GM-CSF release from HASM cells although the concentration–response relationship was complex and better accommodated by a two-site sigmoidal function. The IC₅₀ of cicaprost for the high-affinity component, which accounted for ∼34% of the response, was comparable to that found on well characterised IP-receptor containing tissues including human platelets and vascular smooth muscle, suggesting that the suppression of GM-CSF output could also be mediated through this receptor subtype (Sturzebecher et al., 1986; Stanford et al., 2000). Although IP-receptor antagonists are currently unavailable, the knowledge that HASM cells express IP-receptors (Clarke, D.L., Belvisi, M.G., Smith, S.J., Yacoub, M.H., Meja, K.K., Newton, R., Slater, D.M., Giembycz, M.A. – manuscript under review) and that cicaprost is a potent and selective IP-receptor agonist (Abramovitz et al., 2000; Dong et al., 1986) with weak activity at the human EP₂-receptor subtype (Breyer et al., 2001) supports this conclusion. Moreover, a role for EP₄-receptors can be excluded on the basis that although cicaprost binds with nanomolar affinity to this subtype (Abramovitz et al., 2000), neither AH 23848B nor L-161,982 antagonised the inhibitory effect of PGE2, ONO-AE1-259 and 16,16-dimethyl-PGE2 on GM-CSF release. At high concentrations (>1 μM) of cicaprost, the second component of the response curve became evident (IC₅₀ ∼20 μM), which, based on the aforementioned discussion, may involve the EP₂-subtype.

In conclusion, PGE₂ inhibited GM-CSF release from IL-1β-stimulated HASM cells by acting through prostanoid receptors of the EP₂-subtype and this effect was mediated by PKA. Evidence is also provided that agonism of IP-receptors may also negatively regulate the expression of the CSF2 gene. Finally, H-89 did not reproduce the effect of PKIα, possibly due to its ability to inhibit other protein kinases (Davies et al., 2000), and should not be considered a selective inhibitor of PKA.

Abbreviations

CTD

carboxy-terminal domain

CHO

Chinese hamster ovary

COPD

chronic obstructive pulmonary disease

CSF

colony-stimulating factor

DMEM

Dulbecco's modified Eagle's medium

HBSS

Hanks' balanced salt solution

IL

interleukin