|Home | About | Journals | Submit | Contact Us | Français|
Rationale: A significant population of patients with severe asthma and chronic obstructive pulmonary disease is less responsive to β2-adrenoceptor agonists and corticosteroids, and there are possible safety issues concerning long-term use of these drugs. Inhaled prostaglandin E2 (PGE2) is antiinflammatory and a bronchodilator in patients with asthma, but it also causes cough.
Objectives: We aimed to identify the receptor involved in PGE2-induced sensory nerve activation and cough using a range of in vitro and in vivo techniques.
Methods: Depolarization of vagal sensory nerves (human, mouse, and guinea pig) was assessed as an indicator of sensory nerve acitivity. Cough was measured in a conscious guinea pig model.
Measurements and Main Results: Using an extensive range of pharmacological tools, we identified that the EP3 receptor mediates PGE2-induced depolarization of sensory nerves in human, mouse, and guinea pig. Further supporting evidence comes from data showing that responses to PGE2 are virtually abolished in isolated vagus nerves from EP3-deficient mice (Ptger3−/−). Finally, we demonstrated the role of the EP3 receptor in vivo using a selective EP3 antagonist to attenuate PGE2-induced cough.
Conclusions: Identification of the receptor mediating PGE2-induced cough represents a key step in developing a drug that is antiinflammatory and a bronchodilator but without unwanted side effects.
Prostaglandin E2 (PGE2) has been shown to be antiinflammatory and a bronchodilator in animal models and subjects with asthma. However PGE2 causes cough as a side effect and the receptor mediating this effect is currently unknown.
We have identified the receptor involved in PGE2-induced cough, which could lead to the development of a novel therapy for airway disease that is both a bronchodilator and antiinflammatory but devoid of tussive side effects.
Asthma and chronic obstructive pulmonary disease (COPD) represent major problems for world public health as chronic respiratory diseases are currently the fourth leading cause of death (1). Asthma is characterized by reversible airflow obstruction caused by bronchoconstriction, inflammation, and airway hyperresponsiveness. In contrast, the airflow obstruction in COPD is not fully reversible and often worsens progressively (2). Currently, bronchodilators, such as long-acting β2-adrenergic receptor agonists, are the mainstay treatment but often fail to relieve symptoms of COPD and severe asthma (3) and some possible concerns have been raised over long-term use (4). Inhaled corticosteroids successfully control airway inflammation in most patients with asthma; however, they are of negligible benefit in patients with severe asthma (5) and are often not effective in patients with COPD (6). Concerns exist over long-term side effects of high doses of inhaled corticosteroids (7). Collectively these factors represent a significant need for the development of new, safe therapies for airway inflammatory diseases.
Inhaled prostaglandin E2 (PGE2) is a bronchodilator in normal subjects (8–10) and in patients with asthma and chronic bronchitis (8, 11). In addition, PGE2 has antiinflammatory properties in patients with asthma (12, 13), providing an ideal dual therapy for the treatment of these diseases. Despite the benefits of inhaled PGE2, the development of prostanoid agonists for the treatment of airway inflammatory diseases has been hindered as prostanoids induce irritancy of the upper airway resulting in a reflex cough. Cough is initiated by stimulation of sensory afferent nerve endings in the airways. PGE2 excites airway afferent nerves (14, 15), which concurs with the transient cough seen in both normal subjects and patients with asthma during studies with inhaled PGE2 (8, 10, 13). However, it is not yet known how PGE2 causes this tussive response.
There are currently nine known prostanoid receptors. PGE2 has relatively low affinity for the FP, IP, TP, DP, and CRTh2 receptors (encoded by Ptgfr, Ptgir, Tbxa2r, Ptgdr, and Gpr44 genes) and acts predominantly via the EP receptors. The EP receptors have been subclassified into EP1, EP2, EP3, and EP4 (encoded by Ptger1–Ptger4 genes) (16–18) and distinct signaling pathways initiate diverse and opposing downstream effects in different tissues (17–19).
To date, no study has investigated the receptor responsible for the PGE2-induced cough. The results of this study could provide key proof of concept data demonstrating the selectivity profile required for any candidate compound to elicit antiinflammatory/bronchodilator effects without the sensory irritant side effect liability. To identify the tussive receptor, we used models of sensory nerve activation and cough. Guinea pigs, as the only rodents that possess a cough reflex, are the most appropriate species for investigating cough in vivo. Although mice do not cough per se, comparing sensory nerve activation in nerves from prostanoid receptor–deficient mice is fundamental to this study. Thus, we established that responses to PGE2 in guinea pig and mouse mimic those in human vagal sensory nerves. Using a pharmacological approach and prostanoid receptor–deficient mice, we provide substantial evidence that the EP3 receptor mediates PGE2-induced sensory nerve activation. Finally we provide in vivo proof of concept data demonstrating inhibition of PGE2-induced cough by a selective EP3 receptor antagonist.
Male C57BL/6 mice (18–20 g) and Male Dunkin-Hartley guinea pigs (250–350 g) were purchased from Harlan (Bicester, Oxon, UK). Breeding pairs of mice devoid of one of the following genes: Ptger1 (EP1), Ptger2 (EP2), Ptger3 (EP3), Ptgdr (DP), Ptgfr (FP), Ptgir (IP), or Tbxa2r (TP), had been backcrossed at least eight times onto the C57BL/6 background. Ptger4−/− mice do not survive on the C57BL/6 background due to patent ductus arteriosus (20), so they were backcrossed on a mixed background of 129/Ola X C57BL/6. Mice were kindly provided by Dr. Shuh Narumiya, Kyoto University, and breeding colonies maintained at Imperial College, London. Experiments were performed in accordance with the UK Home Office guidelines for animal welfare based on the Animals (Scientific Procedures) Act 1986.
Sensory nerve depolarization was measured as previously described (21–23). Concentrations of vehicle (0.1% ethanol) or PGE2 were applied to guinea pig, mouse, or human nerves in a random order for 2 minutes each, washing the tissue in between. No more than five stimulations were generated per section of nerve. Human vagus nerves were obtained from donor patients for heart or heart/lung transplants performed at The Royal Brompton or Harefield Hospital. Approval was obtained from the Royal Brompton and Harefield ethics committee after receiving the relevant consents from relatives.
A concentration of 10 μM PGE2 was selected from the concentration response and the effect of a range of antagonists was investigated in the guinea pig and mouse: 0.1% dimethyl sulfoxide (DMSO) vehicle, 1 μM GW848687X (EP1) (24), 0.2 μM L826266 (EP3) (25), 1 μM GW627368X (EP4) (26), 10 μM AL8810 (FP) (27), 10 μM AH6809 (EP1/2DP) (28), 1 μM SQ29548 (TP) (29), and 1 μM RO3244794 (IP) (30). Concentrations of antagonists were selected that were approximately 100-fold the pA2 value (defined as the negative logarithm of the molar concentration of an antagonist that would produce a twofold shift in the concentration–response curve for an agonist) as is common practice. Where available, agonists that act preferentially on one of the receptors (10 μM) were profiled in parallel to PGE2:PGF2α (FP), PGD2 (DP), iloprost (IP), and U46619 (TP). Agonists were applied to the nerve for 2 minutes and washed. A second response was elicited to ensure the response was reproducible. The antagonist was then perfused onto the nerve for 10 minutes and immediately after, PGE2 (or agonist) in the presence of antagonist was applied and the inhibition recorded. The nerve was then washed and agonist response repeated to ensure the nerve was still viable. Any antagonists that inhibited the PGE2-induced depolarization were then investigated in the human vagus nerve.
The procedure for measuring coughs in conscious guinea pigs has previously been described (22, 23, 31). Male Dunkin-Hartley guinea pigs (250–350 g, Charles River, L'Arbresle, France) were exposed to a 10-minute aerosol of PGE2 (30, 100, or 300 μg/ml) and the number of coughs recorded. Once the optimal concentration of PGE2 had been established, a separate group of animals were dosed with vehicle (0.5% methylcellulose + 0.2% Tween 80 in saline, intraperitoneally) or the EP3 antagonist L826266 (300 mg/kg, interperitoneally) 40 minutes before challenge with PGE2. This high dose was chosen because the pharmacokinetic profile demonstrated low plasma levels and a high degree of plasma protein binding after intraperitoneal dosing (report from Merck Frosst, Montreal, PQ, Canada). It was therefore predicted that this dose would be required to have some impact on the PGE2 response. Indeed the data from a guinea pig pharmacokinetic study and assessment of the impact of plasma binding on antagonist potency suggested that even at this high dose there would only be approximately 0.1 μM of unbound (i.e., active) antagonist. Unfortunately, there was insufficient compound available to dose at a higher level and no other appropriate tool compounds available. Coughs were counted as described above.
The EP1 antagonist GW848687X (24) and the EP4 antagonist GW627368X (26) were kind gifts from GlaxoSmithKline, and the EP3 antagonist L826266 (25) and IP antagonist RO3244794 (30) were kind gifts from Merck Frosst and Roche Palo Alto (Palo Alto, CA), respectively. The FP antagonist AL8810 (27), EP1/2DP antagonist AH6809 (28), and TP antagonist SQ29548 (29) were purchased from Cayman Europe. All antagonists were made in DMSO and stocks stored (1 mM or 10 mM) until required. PGE2, PGF2α, and PGD2 were purchased from Sigma Aldrich (Poole, UK) and dissolved in ethanol to 10 mM. U46619 and Iloprost were purchased from Cayman Europe (Tallinn, Estonia) and dissolved in DMSO to 10 mM. Capsaicin was purchased from Sigma and stored in DMSO at 1 mM until required. A low pH solution was made to pH 5 as previously described (32). Krebs compounds were obtained from BDH (Dorset, UK) and all other chemicals and reagents were from Sigma Aldrich.
Antagonism of prostanoid agonists was analyzed using a two-tailed paired t test, comparing responses to agonist (in the same piece of vagus nerve) in the absence and presence of antagonist. Responses to PGE2 in prostanoid receptor–deficient mice were analyzed using Kruskal-Wallis test for multiple comparisons with Dunns post hoc test, comparing the responses in each prostanoid receptor–deficient group to the wild-type control. Inhibition of the PGE2-induced cough was analyzed using Mann Whitney U test for nonparametric data. Data are presented as mean ± SEM and statistical significance was denoted as P less than 0.05.
Our model of sensory nerve activation has previously been characterized and is predictive of agents that cause cough in vivo (21, 22). Responses to PGE2 in the guinea pig vagus nerve emulate responses in the human vagus nerve (23). In the present study we established a concentration-dependent increase in depolarization to PGE2 (Figure 1) in mouse, guinea pig, and human isolated vagus nerves. With no disparity between the species in the response to PGE2, we deduce that responses in guinea pig and mice are representative of those in human nerves.
After confirming PGE2-induced depolarization in all three species, we investigated the receptor responsible in guinea pig isolated nerves using an array of prostanoid receptor agonists and antagonists. Antagonists at FP (AL8810), EP1/2DP (AH6809), TP (SQ29548), and IP (RO3244794) inhibited their corresponding receptor agonists (PGF2α, PGD2, U46619, and Iloprost, respectively) but did not influence PGE2-induced depolarization (Figure 2A). Vehicle or antagonists at EP1 (GW848687X) and EP4 (GW627368X) had no effect on PGE2; however, the EP3 antagonist (L826266 [0.2 μM]) attenuated depolarization to PGE2 in the guinea pig vagus nerve (Figure 2B). Example traces of the effect of the FP antagonist AL8810 on PGF2α (Figure 2C) and PGE2 (Figure 2D) are presented; AL8810 inhibited the responses to PGF2α but did not attenuate PGE2. After the antagonist was washed out, the agonist responses were recovered after all antagonists. The same range of antagonists was investigated in wild-type mice and the effects were mirrored in both species. The EP3 antagonist significantly inhibited PGE2-induced (10 μM) depolarization of the mouse vagus nerve by 64.8 ± 2.8% (n = 4; P < 0.05), whereas no inhibition by the other antagonists was observed (data not shown).
Further investigation of the EP3 antagonist illustrated that L826266 concentration-dependently inhibited PGE2 (Figure 3A). L826266 also completely abolished depolarization induced by sulprostone, an agonist at the EP3 receptor, confirming that L826266 is indeed acting as an EP3 antagonist in this system (Figure 3B). Furthermore, L826266 did not attenuate depolarization induced by 1 μM capsaicin or low pH solution (which both induce a tussive response in vivo) (Figure 3B) illustrating that the antagonist is inhibiting depolarization specifically induced by PGE2. The EP3 antagonist (at a submaximal concentration of 0.2 μM) had a remarkably similar effect in the human vagus nerve, inhibiting PGE2 by 80% (Figure 3C), suggesting that the receptor responsible for PGE2-induced sensory nerve activation is the same in guinea pigs, mice, and humans.
To confirm the role of the EP3 receptor in PGE2-induced sensory nerve activation, we used isolated vagus nerves from prostanoid receptor–deficient mice (Figure 4A). Isolated nerves from wild-type mice depolarized to the expected magnitude, whereas Ptger3−/− mice had a significantly smaller, or completely abolished, response (Figure 4B). All of the mice were genotyped by a standard PCR technique. An example gel of amplification of DNA in the Ptger3 gene in wild-type and Ptger3−/− mouse tissue can be seen in Figure 4C. Primers for the wild type produced a band at 350 bp and Ptger3−/− at 550 bp.
The role of the EP3 receptor in PGE2-induced sensory nerve activation and cough was then investigated in vivo. Guinea pigs were exposed to an aerosol of 30, 100, or 300 μg/ml PGE2 for 10 minutes to establish a suitable concentration (n = 8). The number of coughs were recorded: 30 μg/ml (0 coughs), 100 μg/ml (3.3 ± 3.3 coughs), and 300 μg/ml (12.6 ± 4 coughs). A concentration of 300 μg/ml was selected for use in subsequent experiments. In the study in which we tested the EP3 receptor antagonist the numbers of coughs induced was higher (Figure 5). The reason for this is unknown but could be due to in vivo variation inherent in working with conscious models and/or the additional stress involved in dosing.
Investigation of L826266 on PGE2-induced cough revealed that the antagonist significantly decreased the number of coughs in response to aerosolized PGE2. Although one may expect that the high dose used would block the cough response completely, analysis of the pharmokinetic profile and level of plasma protein binding (performed by Merck Frosst and detailed in Methods) would suggest actual unbound/active levels to be around 0.1 μM. Data from the isolated vagus experiments (Figure 3) show that at this concentration PGE2 responses are only attenuated by approximately 50%, which is remarkably reminiscent of the impact in the cough system. In addition we have data (not shown) in isolated guinea pig trachea showing that 2 μM (i.e., 20 times higher than predicted levels in the cough study) of this compound does not affect PGE2-induced bronchodilation (reported to be through other EP receptors) (33) suggesting that this compound is unlikely to be acting on other EP receptors. Together these data confirm that the EP3 receptor mediates PGE2-induced cough (Figure 5).
Current treatments for airway inflammatory disease, such as inhaled glucocorticoids and long-acting β-agonists, have been associated with significant side effects (3, 4, 7) and furthermore, these treatments are often less effective in certain subpopulations of patients (5, 6). There is therefore an unmet need to develop new, safe drugs, free of adverse side effects, for the treatment of asthma and COPD. PGE2 has been shown to be a bronchodilator and an antiinflammatory agent in several studies in patients with asthma (11–13). However, the development of prostanoid agonists for the treatment of airway inflammatory diseases has been hindered due to airway irritancy and cough (8, 10, 13). Identifying the receptor triggering the reflex cough represents a major step in the development of a PGE2 analog that is a bronchodilator and an antiinflammatory agent but devoid of tussive side effects.
To commence our investigation we measured vagal sensory nerve activation, because the vagus nerves supply the majority of the sensory nerve fibers to the airway, using the isolated vagus nerve preparation. We have previously shown that responses to tussive stimuli in human vagus nerves are representative of those from guinea pig nerves (23). Here, we generated data comparing the responses to PGE2 in mouse, guinea pig, and human species. Therefore we can conclude from these experiments that PGE2 has a remarkably similar effect on human, mouse, and guinea pig vagus nerves. This would suggest that guinea pig and mouse vagi are representative of the effects of PGE2 in human tissue. Moreover, because PGE2 causes both sensory nerve activation and cough in humans and guinea pigs, we have evidence that our model of sensory nerve activity is representative of the human cough reflex.
Once we established that PGE2 activates isolated vagal sensory nerves, we proceeded to investigate pharmacologically the receptor responsible for PGE2-induced depolarization using a range of prostanoid antagonists. The EP3 antagonist L826266 was the only antagonist investigated that attenuated responses to PGE2 in both guinea pig and mouse. Furthermore, we have provided evidence that this antagonism is mirrored in the isolated human nerve, illustrating that the EP3-mediated sensory nerve activation, elicited by PGE2, is an effect that is translated across species. Capsaicin and a low pH solution were not inhibited by the EP3 antagonist, suggesting that the antagonist is specifically inhibiting PGE2-induced sensory nerve activation rather than acting as a general inhibitor of nerve depolarization.
Although significant pharmacological evidence that the EP3 receptor is responsible for PGE2-induced sensory nerve activity was generated, we then corroborated these data by using isolated nerves from mice devoid of one of the prostanoid receptors. The Ptger3−/− mice, lacking EP3 receptors, had a diminished response to PGE2, whereas responses in nerves from other prostanoid receptor–deficient mice (Ptger1−/−, Ptger2−/−, Ptger4−/−, Ptgfr−/−, Ptgdr−/−, Ptgir−/−, Tbxa2r−/−) depolarized to the same magnitude as wild types. This research culminated in the demonstration that L826266 inhibited PGE2-induced cough in our in vivo guinea pig model. Taken together these studies confirm that cough induced by PGE2 is caused mainly, if not solely, by activation of the EP3 receptor.
PGE2 is commonly presumed to be proinflammatory and has been implicated in several inflammatory disease conditions, including rheumatoid arthritis (34) and UVB-induced cutaneous inflammation (35). However, in humans, the lung appears to be unique in that PGE2 has beneficial effects (8, 9, 11–13) as well as unwanted effects (8, 13). The receptor(s) responsible for the beneficial effects seem to differ from the EP3 receptor causing the detrimental cough. Using human airway smooth muscle and rodent models, there is evidence that the EP2 receptor is responsible for the PGE2-induced bronchodilation (36–38). Antiinflammatory effects of PGE2 have been identified in both human (12, 13) and rodent models (39). One isolated study has suggested that, in a mouse model, activation of the EP3 receptor could suppress ovalbumin-induced inflammation (40). However, the wealth of information available, especially from human cell–based assays, would suggest that the receptor(s) responsible for the antiinflammatory actions of PGE2 are the EP2 and/or the EP4 receptor subtype (41–45).
Rationale for the development of a PGE2 analog that does not activate the EP3 receptor is not just limited to cough, as the EP3 receptor mediates other detrimental effects of PGE2 in the airways. Although PGE2-mediated bronchodilation is believed to be via the EP2 receptor, bronchoconstrictor effects of PGE2 have been shown to act through EP1 and EP3 (33). What is more, the EP3 receptor has been implicated in the chemotaxis of mast cells to sites of inflammation (46).
In summary, we have established in this current study that PGE2-induced sensory nerve activation and cough are mediated via the EP3 receptor. Our research, and other published data, suggests that the unwanted effects of PGE2 (proinflammatory effects, bronchoconstriction, and cough) are mediated by the EP3 receptor and beneficial effects are likely to act via a different receptor. This research provides considerable support for the development of a PGE2 analog that does not activate the EP3 receptor, thereby providing a novel therapy for airway inflammatory diseases that is antiinflammatory and a bronchodilator but devoid of tussive side effects.
The authors thank the following for their intellectual contribution and kind gifts: D. Slipetz, M. Gallant, R. Forget (Merck Frosst) for L826266; R. McArthur-Wilson (GlaxoSmithKline) for GW848687X and GW627368X; Roche Palo Alto for RO3244794; S. Narumiya (Kyoto University) and T. Maruyama (ONO Pharmaceuticals) for the prostanoid receptor-deficient mice. The authors also thank M. Catley, D. Clarke, M. Collins, E. Hardaker, D. Hele, J. Nasra, and S. Wong for their help and advice with the experiments. They also thank M. Johnson and T. Nials (GlaxoSmithKline) for their support and encouragement.
Supported by GlaxoSmithKline (S.A.M., M.A.B.), and a project grant from the Medical Research Council (MRC), UK (G0800196) (M.A.B.). The research was also funded by a project grant from the MRC, UK (G0800195). The human tissue work was funded by an Experimental Medicine Grant (G0502019) from the MRC, UK. The human tissue experiments in this study were undertaken with the support of the NIHR Biomedical Research Unit in Advanced Lung Disease at the Royal Brompton and Harefield NHS Foundation Trust and Imperial College London and partly funded by the NIHR Biomedical Research Unit funding scheme.
Originally Published in Press as DOI: 10.1164/rccm.200903-0388OC on September 3, 2009
Conflict of Interest Statement: S.A.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.A.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.G.B. received $5,001–$10,000 from GlaxoSmithKline, $5,001–$10,000 from Sound Pharma, and $5,001–$10,000 from AstraZeneca in consultancy fees; $1,001 to $5,000 from Vectura, $1,001–$5,000 from GlaxoSmithKline, and $1,001–$5,000 from Schering Plough in advisory board fees; and $10,001– $50,000 from AstraZeneca and $10,001–$50,000 from Novartis in industry-sponsored grants.