Prostanoid receptor antagonists: development strategies and therapeutic applications

Abstract

Identification of the primary products of cyclo-oxygenase (COX)/prostaglandin synthase(s), which occurred between 1958 and 1976, was followed by a classification system for prostanoid receptors (DP, EP₁, EP₂ …) based mainly on the pharmacological actions of natural and synthetic agonists and a few antagonists. The design of potent selective antagonists was rapid for certain prostanoid receptors (EP₁, TP), slow for others (FP, IP) and has yet to be achieved in certain cases (EP₂). While some antagonists are structurally related to the natural agonist, most recent compounds are ‘non-prostanoid’ (often acyl-sulphonamides) and have emerged from high-throughput screening of compound libraries, made possible by the development of (functional) assays involving single recombinant prostanoid receptors. Selective antagonists have been crucial to defining the roles of PGD₂ (acting on DP₁ and DP₂ receptors) and PGE₂ (on EP₁ and EP₄ receptors) in various inflammatory conditions; there are clear opportunities for therapeutic intervention. The vast endeavour on TP (thromboxane) antagonists is considered in relation to their limited pharmaceutical success in the cardiovascular area. Correspondingly, the clinical utility of IP (prostacyclin) antagonists is assessed in relation to the cloud hanging over the long-term safety of selective COX-2 inhibitors. Aspirin apart, COX inhibitors broadly suppress all prostanoid pathways, while high selectivity has been a major goal in receptor antagonist development; more targeted therapy may require an intermediate position with defined antagonist selectivity profiles. This review is intended to provide overviews of each antagonist class (including prostamide antagonists), covering major development strategies and current and potential clinical usage.

Introduction

The pharmacological classification for prostanoid receptors was developed on the basis that each receptor preferentially recognizes one of the major natural prostaglandins (PGs). Thus, PGD₂ preferentially activates the DP receptor, PGE₂ preferentially activates the EP receptor, with the same applying to PGF_2α/FP receptor and PGI₂ (prostacyclin)/IP receptor (Coleman et al., 1994b). In the case of the TP receptor, both thromboxane A₂ (TXA₂) and its precursor PGH₂ are potent agonists. There are two distinct subtypes of DP receptor, DP₁ and DP₂; the latter has also been called CRTh2 (chemoattractant receptor-homologous molecule expressed on T helper 2 cells). Of the four EP receptor subtypes, EP₁ and EP₃ generally elicit excitatory actions, while EP₂ and EP₄ elicit inhibitory actions on cell function. All prostanoid receptors belong to the G protein-coupled receptor superfamily of cell-surface receptors. Each has seven transmembrane (TM)-spanning segments and may couple to one or more signal-transduction processes. In addition, mRNA splicing variants have been discovered for DP₁, EP₁, EP₃, EP₄, FP and TP receptors (see Pierce and Regan, 1998).

The repertoire of bioactive oxygenated lipids and signal transduction mechanisms is expanded by homodimerization and heterodimerization between different prostanoid receptors. For example, an isoprostane binding site is created by IP/TPα heterodimerization (Wilson et al., 2004). Moreover, the prostamide (prostaglandin-ethanolamide) recognition site appears to result from heterodimerization of wild type and alternatively spliced FP receptor variants (Liang et al., 2008). Prostanoid receptors may also complex with non-prostanoid receptors, for example, EP₁ receptor with β₂-adrenoceptor, but discussion of these interactions is outside the scope of this review. The receptor/second messenger nomenclature used in the review conforms to this journal's Guide to Receptors and Channels (Alexander et al., 2008).

Assay systems and antagonist development

The initial pharmacological differentiation of prostanoid receptors relied heavily on isolated tissue studies, with ileum, trachea and vas deferens of the guinea pig being especially important (Jones et al., 1982; Coleman et al., 1984; 1987; 1994a; Dong et al., 1986; see Chen et al., 2001 for experimental details). Isolated tissue preparations are still used today as they often reflect phenomena that occur in the integrated, living mammal. In addition, they still provide evidence for new receptor entities, as shown by the critical role of the cat iris preparation in the elucidation of prostamide pharmacology (Matias et al., 2004; Woodward et al., 2007; 2008;). Studies on isolated and cultured cells have also been important in building the pharmacological classification of prostanoid receptors (Coleman et al., 1984; Eglen and Whiting, 1988; Woodward et al., 1995a,b;).

Although of low throughput, isolated tissue studies have been critical to the discovery of prostanoid antagonists, particularly for DP₁, EP₁ and TP receptors. However, it is remarkable that antagonists for other receptors (EP₂, EP₃, FP and IP) have been slow to emerge following discovery of the natural ligand/receptor; several reasons may account for this situation. Partial agonism is found in closely related analogues of the natural ligands for DP₁ and TP receptors (Table 1), thereby providing inroads to pure antagonists. However, partial agonism may not always be readily recognized. For example, taprostene, an early analogue of PGI₂ (Müller et al., 1983), was only shown to be an IP partial agonist in 2004 (Chan and Jones, 2004). Furthermore, partial agonism may not always translate into pure antagonism. Thus, some of the many ‘non-prostanoid prostacyclin mimetics’ synthesized (see later) show IP partial agonism (Merritt et al., 1991a; Jones et al., 1997; Seiler et al., 1997; Kam et al., 2001), but IP antagonists have apparently not emerged from this grouping. Moreover, a large range of compounds block TP receptors (see later); this expansiveness may simply not apply to other prostanoid receptors. Finally, there has not been widespread commercial drive to develop EP₂ antagonists, for example, owing to a perceived lack of therapeutic utility (see later).

Table 1.

Prostanoid receptor agonists relevant to defining antagonist profiles

Prostanoid receptor	Full agonist		Partial agonist
	High selectivity	Moderate selectivity
DP1	BW-245C		BW-192C86a
DP2	15(R) PGD2, 15(R)-15-methyl PGD213,14-dihydro-15-oxo PGD2
EP1	ONO-DI-004	17-Phenyl PGE2	Iloprostb,c
EP2	ONO-AE1-259, CAY-10399d	Butaprost-FA, CP-533536e19(R)-hydroxy PGE2f
EP3	ONO-AE-248, SC-46275	Sulprostone, MB-28767	ONO-AP-324g
EP4	ONO-AE1-329, tetrazolo PGE1h	(PGE2)i
FP	Fluprostenol, latanoprost-FA	Cloprostenol	AL-8810j,k,l
IP	Cicaprost	AFP-07, iloprost	Octimibatem, taprostenen
TP	STA2, U-46619		CTA2o, PTA2o,p, U-44069p

Information on the non-referenced agonists may be obtained from Jones (2004) and this journal's Guide to Receptors and Channels edited by Alexander et al. (2008). In older publications, fluprostenol = ICI-81008, cloprostenol = ICI-80996, cicaprost = ZK-96480 and iloprost = ZK-36374.

FA, free acid.

^aSeries of bicyclic-hydantoin prostanoids (Leff and Giles, 1992).

^bDong et al. (1986).

^cBoie et al. (1997).

^dTani et al. (2002).

^eParalkar et al. (2003).

^fWoodward et al. (1993a).

^gJones et al. (1998).

^hAnalogue 19a in Billot et al. (2003).

ⁱUtility on high-sensitivity EP₄ systems.

^jGriffin et al. (1999).

^kWoodward et al. (2007).

^lSharif et al. (2008).

^mMerritt et al. (1991a,b;).

ⁿChan and Jones (2004).

^oArmstrong et al. (1985).

^pJones et al. (1982).

Dramatic progress in antagonist development was seen following cloning of the various prostanoid receptors in the early 1990s: DP₁ (Boie et al., 1995), DP₂ (Hirai et al., 2001), EP₁ (Funk et al., 1993a), EP₂ (Regan et al., 1994), EP₃ (Yang et al., 1994), EP₄ (Bastien et al., 1994), FP (Abramovitz et al., 1994), IP (Namba et al., 1994), TP (Hirata et al., 1991). Stable over-expression of each prostanoid receptor in carrier cell lines allowed high-throughput radioligand binding and functional studies using 96- or even 384-well plate format. Thus, chemical library screening resulted in the discovery of new non-prostanoid scaffolds as leads, from which potent and selective agonists and antagonists were designed. Inspection of Figures 1–8 will reveal the prevalence of aryl-sulphonamido linkages in the various antagonist classes. Non-prostanoid structures provide a practical approach to obtaining new prostanoid-based therapies, as the issues of bioavailability, metabolic stability and cost of synthesis that surround PG analogues are avoided.

Figure 1.

DP₁ receptor antagonists. The natural ligand PGD₂ is shown in the box; the trans-orientation of the α (upper) and ω (lower) side-chains and (S) configuration at C15 are found in primary products of all COX/synthase systems. BW-A868C and ZK-138357 are each composed of four diastereoisomers (chiral centres at C8/C15 and C10/C15 respectively); compound 1 is racemic. Indomethacin is a lead compound for the non-prostanoid antagonists shown in the lower row; N-benzoyl-2-methyl-indol-3-yl-acetic acid templates are shown in red. The 3(S)-enantiomer of MK-0524 has 320-fold lower affinity for the human DP₁ receptor (Sturino et al., 2007).

Figure 8.

TP receptor antagonists. Conversion of PGH₂ to TXA₂ by thromboxane synthase (TXS) is shown in the box; α and ω represent natural 2-series side-chains. The pinane-thromboxane residue (related to 1(S)-α-pinene) is shown in blue; the 6-oxabicyclo(2.2.1)heptane system is in red. AH-23848 has the same α- and ω-chains as GR-32191. Benzenesulphonamide residues present in both prostanoid and non-prostanoid antagonists are shown in cerise. TP antagonists with two types of additional activity are presented. (A) IP agonism is conferred by the diphenylmethyl-heteroatomic unit in the bicyclo[2.2.2]octene analogue EP-157. (B) TXS inhibitory activity is conferred by the pyridin-3-yl residue (green) in isbogrel and ZD-1542 and by a similar replacement for ring A in relatives of GR-32191 (e.g. GR-83783; see text). Additionally, the broken arrows (lower right) typically indicate attachment of part of a TP antagonist to isbogrel (or ridogrel) to generate novel combined TP antagonist/TXS inhibitors; the tether (0–11 carbon units) has also been attached to the left-hand phenyl ring of ICI-192605 (2–8 carbon units) (Ackerley et al., 1995).

Selective agonists for prostanoid receptors

Selective agonists are the pharmacological counterparts of antagonists and, as such, fulfil an important role in receptor characterization. In terms of prostanoid receptors, three of the natural ligands, PGH₂/TXA₂ and PGI₂, are unstable under physiological conditions and are usually replaced by U-46619 and cicaprost respectively (Table 1). In addition, there are potential problems with certain EP agonists synthesized as C1 methyl esters, for example, butaprost (EP₂) and SC-46275 (EP₃); full agonist potency is only realized after (enzymatic) hydrolysis of the ester within the tissue. Ono Pharmaceuticals have synthesized selective agonists for EP₁ (ONO-DI-004), EP₂ (ONO-AE1-259), EP₃ (ONO-AE-248) and EP₄ (ONO-AE1-329) receptors (Suzawa et al., 2000). However, the rather modest potencies of ONO-DI-004 and ONO-AE-248 may restrict their utility in full Schild antagonism protocols (R.L. Jones et al., 2008, submitted). Partial agonists for DP₁, EP₁, EP₃, FP, IP and TP receptors are known (Table 1); their use in antagonist protocols may present difficulties of interpretation.

As drugs, notably for systemic administration, selective prostanoid mimetics are a high-risk proposition because of the myriad of unwanted side effects that may occur. As such, the future of prostanoid-based therapies appears to reside in the main in selective PG synthase inhibitors and prostanoid receptor antagonists.

Antagonist protocols

The Schild protocol for inferring the nature of competition and determining the affinity constant of an antagonist remains the gold standard (see Colquhoun, 2007). It is applicable to both isolated tissue and the ever-increasing number of recombinant (rc) receptor/cell-based assay methods usually involving Ca²⁺ mobilization or cAMP generation. While the latter methods are given to high throughput and precision, care must be taken with high-affinity antagonists in the Ca²⁺ assays as the agonist response is usually measured as the peak of the transient Ca²⁺ signal, which may occur before re-equilibration of antagonist occupancy is complete. The majority of the data given in Table 2 relate to pA₂ values derived using the Schild protocol, with emphasis on human, guinea pig and rat isolated preparations. Binding data (preferably pK_i) are given where functional information is not available. There are considerable binding data on mouse prostanoid receptors, but little affinity data in functional systems.

Table 2.

Affinities of prostanoid receptor antagonists in functional isolated preparations

Antagonist	Species	Tissue system	Agonist	pA2	Reference
DP1 receptor
AH-6809	Human	Neutrophil/superoxide release	BW-245C	6.55	Wheeldon and Vardey (1993)
			BW-245C	6.59	Lydford et al. (1996a)
		Platelet/aggregation	PGD2	6.27a	Keery and Lumley (1988)
	Cow	Embryonic tracheal fibroblast/cAMP	PGD2	6.36	Ito et al. (1990)
	Rabbit	Saphenous vein	BW-245C	5.93	Lydford et al. (1996c)
BW-A868C	Human	Neutrophil/superoxide release	BW-245C	9.46	Lydford et al. (1996a)
		Platelet/aggregation	BW-245C	9.26	Giles et al. (1989)
		Pulmonary vein	PGD2	7.84	Walch et al. (1999)
		Uterus (non-pregnant)	BW-245C	8.3	Senior et al. (1992)
		Uterus (pregnant)	BW-245C	8.6	Senior et al. (1993)
	Cow	Embryonic tracheal fibroblast/cAMP	BW-245C	8.0	Crider et al. (1999)
	Dog	Nasal vein	BW-245C	7.3	Liu et al. (1996a)
		Tracheal epithelium/Cl- secretion	BW-245C	8.16	Liu et al. (1996b)
	Rabbit	Jugular vein	BW-245C	8.73	Giles et al. (1989)
		Saphenous vein	BW-245C	8.50	Lydford et al. (1996c)
MK-0524 (Laropiprant)	Human	rc-DP1/HEK-293E/binding	[3H]-PGD2	10.5b	Sturino et al. (2007)
		Platelet/cAMP	PGD2	10.05c	Sturino et al. (2007)
ONO-AE3-237	Human	rc-DP1/CHO/binding	[3H]-PGD2	7.74	Torisu et al. (2004c)
S-5751	Human	Platelet/cAMP	PGD2	9.02b,c	Arimura et al. (2001)
	Guinea pig	Platelet/cAMP	PGD2	7.50b,c	Arimura et al. (2001)
ZK-138357	Human	Neutrophil/superoxide release	BW-245C	7.25	Lydford et al. (1996a)
	Rabbit	Saphenous vein	BW-245C	5.05	Lydford et al. (1996a)
	Rat	Peritoneal mast cell	BW-245C	∼6.0	Chan et al. (2000)
Compound 1	Human	Platelet/binding	[3H]-PGD2	6.22	Mitsumori et al. (2003a)
Compound 2	Human	Platelet/binding	[3H]-PGD2	7.62	Mitsumori et al. (2003a)
Compound 3	Human	rc-DP1/HEK-293-Gα15/Ca2+	BW-245C	[∼7.4]	Krauss et al. (2005)
Compound 4	Human	rc-DP1/CHO/binding	[3H]-PGD2	8.27	Torisu et al. (2004c)
Compound 5	Human	rc-DP1/HEK-293/binding	[3H]-PGD2	9.0	Beaulieu et al. (2008)
DP2 receptor
BAY-u3405 (Ramatroban)	Human	rc-DP2/CHO/GTPγS binding	PGD2	7.44	Mathiesen et al. (2006)
		Eosinophil/shape change	PGD2	∼8.0a	Mathiesen et al. (2006)
K-117	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	8.26	Mimura et al. (2005)
K-604	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	7.96	Mimura et al. (2005)
TM-30089d (CAY-10471)	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	8.74	Mathiesen et al. (2006)
				9.22	Ulven and Kostenis (2005)
Compound 6	Human	rc-DP2/pre-B L1.2/Ca2+	PGD2	[6.8]	Bauer et al. (2002)
Compound 7	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	8.64c	Birkinshaw et al. (2006)
Compound 8	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	[9.40]c	Bonnert and Rasul (2004)
Compound 9	Human	rc-DP2/CHO/binding	[3H]-PGD2	7.17	Armer et al. (2005)
		Eosinophil/shape change	PGD2	7.13c	Armer et al. (2005)
		Th2-lymphocyte/chemotaxis	PGD2	7.17c	Armer et al. (2005)
Compound 10	Human	rc-DP2/CHO/Ca2+	PGD2	[8.53]	Fretz et al. (2005)
Compound 11	Human	rc-DP2/HEK-293/binding	[3H]-PGD2	[9.0]	Bonnert et al. (2005c)
EP1 receptor
AH-6809	Human	rc-EP1/HEK-293E/reporter gene	Iloprost	∼6.4	Durocher et al. (2000)
		Pulmonary vein	Sulprostone	5.52	Walch et al. (2001)
	Guinea pig	Ileum	PGE2	6.8	Coleman et al. (1987)
		Ileum	PGE2	7.39	Eglen and Whiting (1988)
		Ileum	PGE1	7.42	Eglen and Whiting (1988)
		Ileum	16,16-DM PGE2	7.59	Eglen and Whiting (1988)
		Trachea	16,16-DM PGE2	7.48	Eglen and Whiting (1988)
		Trachea	17-Phenyl PGE2	7.35	Lawrence et al. (1992)
GW-848687	Human	rc-EP1/not given/reporter gene	PGE2	9.1	Giblin et al. (2007)
MF-266-1	Human	rc-EP1/HEK-293/Ca2+	PGE2	7.8	Clark et al. (2008)
ONO-8711	Mouse	rc-EP1/CHO/binding	[3H]-PGE2	8.77	Watanabe et al. (1999)
	Human	rc-EP1/CHO/binding	[3H]-PGE2	9.22	Watanabe et al. (1999)
ONO-8713	Mouse	rc-EP1/not given/binding	[3H]-PGE2	9.5	Narumiya and Fitzgerald (2001)
SC-19220	Guinea pig	Ileum	PGE2	5.5	Sanner (1969)
		Ileum	PGE2	5.6	Bennett and Posner (1971)
		Trachea	PGF2α	6.6	Farmer et al. (1974)
SC-51089	Human	rc-EP1/HEK-293E/reporter gene	Iloprost	6.94	Durocher et al. (2000)
	Guinea pig	Ileum	PGE2	6.5	Hallinan et al. (1993)
		Ileum	PGE2	6.7	Sametz et al. (2000)
SC-51322	Human	rc-EP1/HEK-293E/reporter gene	Iloprost	8.80	Durocher et al. (2000)
	Guinea pig	Ileum	PGE2	8.1	Hallinan et al. (1994)
		Trachea	17-Phenyl PGE2	8.45	Hung et al. (2006)
Compound 13	Mouse	rc-EP1/CHO/Ca2+	PGE2	8.25c	Naganawa et al. (2006)
Compound 14	Human	rc-EP1/HEK-293E/binding	[3H]-PGE2	8.0	Ruel et al. (1999)
Compound 15	Human	rc- EP1/CHO/Ca2+	PGE2	8.2	Hall et al. (2007b)
EP2 receptor
AH-6809	Human	rc-EP2/COS-7/cAMP	PGE2	∼6.5	Woodward et al. (1995)
		Bronchus	PGE2	5.78	Norel et al. (1999)
	Guinea pig	Trachea	PGE2	5.7e	KJ Ong and RL Jones (unpublished)
EP3 receptor
DG-041	Human	rc-EP3/Chem-1/Ca2+	PGE2	8.09c	Singh et al. (2009)
L-798106	Guinea pig	Aorta	17-Phenyl PGE2f	7.96	Jones et al. (2008)
		Trachea	Sulprostone	7.82	Clarke et al. (2004)
		Vas deferens	Sulprostone	7.48	Clarke et al. (2004)
L-826266	Human	rc-EP3/HEK-293E/Ca2+	PGE2	7.97	R.L. Jones et al. (2008, submitted)
		Erythroleukaemia cell/cAMP	Sulprostone	8.35	Clark et al. (2008)
	Guinea pig	Aorta	17-Phenyl PGE2	7.58	R.L. Jones et al. (2008, submitted)
ONO-AE3-240	Mouse	rc-EP3/not given/Ca2+	PGE2	8.8	Amano et al. (2003)
Compound 17	Human	Erythroleukaemia cell/cAMP	Sulprostone	6.89	Gallant et al. (2002)
Compound 18	Human	rc-EP3/HEK-293E/binding	[3H]-PGE2	7.7	Juteau et al. (2001)
Compound 19	Human	rc-EP3/HEK-293E/cAMP	PGE2	8.22	Belley et al. (2005)
EP4 receptor
AH-23848	Sheep	Ductus arteriosus	PGE2	∼5.2a	Bouayad et al. (2001)
	Human	Middle cerebral artery	PGE2	5.7	Davis et al. (2004)
	Pig	Saphenous vein	PGE2	5.0	Coleman et al. (1994a)
	Rabbit	Saphenous vein	PGE2	4.96	Lydford et al. (1996b)
	Mouse	rc-EP4/CHO/cAMP	PGE2	5.3	Nishigaki et al. (1995)
BGC-20-1531	Human	rc-EP4/HEK-293E/cAMP	PGE2	7.6	Maubach et al. (2009)
		Middle cerebral artery	PGE2	7.8	Maubach et al. (2009)
	Dog	Middle meningeal artery	PGE2	7.7	Maubach et al. (2009)
CJ-023423	Human	rc-EP4/HEK-293/cAMP	PGE2	8.3	Nakao et al. (2007)
	Rat	rc-EP4/HEK-293/cAMP	PGE2	8.2	Nakao et al. (2007)
CJ-042794	Human	rc-EP4/HEK-293/cAMP	PGE2	8.6	Murase et al. (2008b)
	Rat	rc-EP4/HEK-293/cAMP	PGE2	8.7	Murase et al. (2008a)
GW-627368	Human	rc-EP4/HEK-293/cAMP	PGE2b	7.9	Wilson et al. (2006)
		Pulmonary vein	ONO-AE1-329	7.06	Foudi et al. (2008)
	Pig	Saphenous vein	PGE2	9.2	Wilson et al. (2006)
	Rabbit	Saphenous vein	PGE2g	≥8.5	Jones and Chan. (2005)
L-161982	Human	rc-EP4/HEK-293/cAMP	PGE2	∼8.5	Machwate et al. (2001)
		BEAS-2B cell/CRE reporter	ONO-AE1-329	9.14	L.M. Ayer and M.A. Giembycz (unpublished)
		Middle cerebral artery	PGE2	8.4	Davis et al. (2004)
	Rat	rc-EP4/HEK-293/binding	[3H]-PGE2	7.50	Machwate et al. (2001)
		Periosteal cell/cAMP	PGE2	7.0c	Machwate et al. (2001)
MF-498	Human	rc-EP4/HEK-293/cAMP	PGE2	8.77c	Clark et al. (2008)
ONO-AE2-227	Mouse	rc-EP4/CHO/cAMP	PGE2	8.0c	Mutoh et al. (2002)
ONO-AE3-208	Mouse	rc-EP4/not given/binding	[3H]-PGE2	8.89	Kabashima et al. (2002)
Compound 20	Human	rc-EP4/HEK-293E/cAMP	PGE2	8.49c	Burch et al. (2008)
FP receptor
AS-604872	Human	rc-FP/HEK-293E/PI	PGF2α	7.33c	Cirillo et al. (2007)
THG-113	Pig	Retinal blood vessel	PGF2α	[6.34]c	Peri et al. (2006)
THG-113.31	Pig	Retinal blood vessel	PGF2α	[8.00]c	Peri et al. (2006)
THG-113.824	Pig	Retinal blood vessel	PGF2α	[8.96]c	Peri et al. (2006)
THG-113.825	Pig	Retinal blood vessel	PGF2α	7.21c	Peri et al. (2006)
Prostamide F receptor
AGN-204396	Cat	Iris sphincter	Prostamide F2α	5.64	Woodward et al. (2007)
IP receptor
RO-1138452	Human	rc-IP/CHO/cAMP	Carbacyclin	9.0	Bley et al. (2006)
		Pulmonary artery	Cicaprost	8.20	Jones et al. (2006)
	Guinea pig	Aorta	Cicaprost	8.39	Jones et al. (2006)
	Rabbit	Mesenteric artery	Cicaprost	8.12	Jones et al. (2006)
RO-3244794	Human	rc-IP/CHO/cAMP	Carbacyclin	8.5	Bley et al. (2006)
		BEAS-2B cell/CRE reporter	Taprostene	9.24	L.A. Ayer and M.A. Giembycz (unpublished)
Compound 21	Rat	rc-IP/not given/cAMP	(Not given)	[8.12]h	Keitz et al. (2004)
Compound 22	Rat	UMR-106 osteosarcoma cell/cAMP	Iloprost	6.41	Nakae et al. (2005)
Compound 23	Rat	UMR-106 osteosarcoma cell/cAMP	Iloprost	6.32	Nakae et al. (2005)
Compound 24	Human	Platelet membrane/cAMP	Iloprost	7.8c	Brescia et al. (2007)
TP receptor
AA-2414 (Seratrodast)	Human	Bronchus	U-46619	7.7	Itoh et al. (1993)
	Pig	Coronary artery	U-44069	9.0	Imura et al. (1990)
	Guinea pig	Aorta	U-46619	8.5	Zhang et al. (1996)
		Trachea	U-46619	7.69	Ashida et al. (1989)
	Rat	Aorta	U-46619	7.8	Zhang et al. (1996)
AH-23848	Human	Bronchus	U-46619	8.5	Coleman and Sheldrick (1989)
		Bronchus	U-46619	6.9	McKenniff et al. (1988)
		Hand vein	U-46619	8.4	Arner et al. (1991)
		Lung parenchyma	U-46619	8.7	McKenniff et al. (1988)
		Platelet/aggregation	U-46619	8.05	Tymkewycz et al. (1991)
	Guinea pig	Lung parenchyma	U-46619	8.7	McKenniff et al. (1988)
		Trachea	U-46619	8.7	McKenniff et al. (1988)
		Trachea	U-46619	9.76	Tymkewycz et al. (1991)
	Rat	Aorta	U-46619	8.47	Tymkewycz et al. (1991)
		Lung parenchyma	U-46619	6.9	Norman et al. (1992)
		Platelet/aggregation	U-46619	8.19	Tymkewycz et al. (1991)
BAY-u3405 (Ramatroban)	Human	Bronchus	U-46619	8.8	McKenniff et al. (1991)
		Pulmonary vein	U-46619	8.94	Walch et al. (2001)
	Guinea pig	Lung parenchyma	U-46619	7.7	Norman et al. (1992)
		Trachea	U-46619	8.7	McKenniff et al. (1991)
	Rat	Lung parenchyma	U-46619	8.6	McKenniff et al. (1991)
BM-13177 (Sulotroban)	Human	Platelet/aggregation	U-46619	6.31	Karasawa et al. (1991a)
	Guinea pig	Aorta	U-46619	5.58	Karasawa et al. (1991b)
	Rabbit	Jugular vein	U-46619	6.01	Giles et al. (1989)
BM-13505	Human	Hand vein	U-46619	7.9	Arner et al. (1991)
		Platelet/aggregation	U-46619	7.75	Karasawa et al. (1991a)
		Uterus (non-pregnant)	U-46619	7.4	Senchyna and Crankshaw (1996)
	Guinea pig	Aorta	U-46619	6.89	Dubéet al. (1992)
		Aorta	U-46619	7.22	Karasawa et al. (1991b)
		Aorta	U-46619	7.3	Zhang et al. (1996)
		Lung parenchyma	U-46619	7.0	Norman et al. (1992)
		Trachea	U-46619	7.73	Dubéet al. (1992)
		Trachea	U-46619	7.5	Ogletree and Allen (1992)
	Rat	Aorta	U-46619	8.2	Zhang et al. (1996)
		Aorta	U-46619	8.6	Ogletree and Allen (1992)
		Lung parenchyma	U-46619	7.5	Norman et al. (1992)
BMS-180291 (Ifetroban)	Human	Platelet membrane/binding	[3H]-SQ-29548	8.4	Ogletree et al. (1993)
	Guinea pig	Aorta	U-46619	9.8	Zhang et al. (1996)
	Rat	Aorta	U-46619	9.5	Zhang et al. (1996)
CV-4151	Human	Platelet/aggregation	U-46619	5.2	Watts et al. (1991)
	Rabbit	Aorta	U-44069	5.9	Imura et al. (1988)
EP-092	Human	Bronchus	U-46619	7.37	Featherstone et al. (1990)
		Bronchus	U-46619	6.8	McKenniff et al. (1988)
		Lung parenchyma	U-46619	8.9	McKenniff et al. (1988)
		Platelet /aggregation	U-46619	7.73	Tymkewycz et al. (1991)
		Uterine artery	U-46619	8.5	Baxter et al. (1995)
	Guinea pig	Lung parenchyma	U-46619	8.7	McKenniff et al. (1988)
		Trachea	U-46619	7.29	Featherstone et al. (1990)
		Trachea	U-46619	8.7	McKenniff et al. (1988)
		Trachea	U-46619	8.02	Tymkewycz et al. (1991)
	Rat	Aorta	U-46619	8.55	Tymkewycz et al. (1991)
		Lung parenchyma	U-46619	7.1	Norman et al. (1992)
		Platelet aggregation	U-46619	7.80	Tymkewycz et al. (1991)
EP-169	Human	Platelet/aggregation	U-46619	8.30	Tymkewycz et al. (1991)
		Pulmonary artery	U-46619	∼8.2	Qian et al. (1994)
	Guinea pig	Trachea	U-46619	8.77	Tymkewycz et al. (1991)
	Rat	Aorta	U-46619	8.73	Tymkewycz et al. (1991)
		Platelet/aggregation	U-46619	8.48	Tymkewycz et al. (1991)
Glibenclamide	Human	Internal mammary artery	U-46619	6.3i	Stanke et al. (1998)
		Saphenous vein		6.7i	Stanke et al. (1998)
	Dog	Coronary artery	U-46619	6.2	Cocks et al. (1990)
	Rabbit	Aorta	U-46619	6.08a	Pfister et al. (2004)
	Guinea pig	Aorta	U-46619	<5.0	Kemp and McPherson (1998)
	Rat	Aorta	U-46619	6.13	Kemp and McPherson (1998)
GR-32191 (Vapiprost)	Human	Bladder (detrusor)	U-46619	8.27	Palea et al. (1998)
		Bronchus	U-46619	8.77	Featherstone et al. (1990)
		Bronchus	U-46619	8.40	Armour et al. (1989)
		Platelet/binding	[3H]-GR-32191	8.66	Armstrong et al. (1993)
		Pulmonary artery	U-46619	8.18	Lumley et al. (1989)
		Saphenous vein	U-46619	8.93	Furci et al. (1991)
		Umbilical artery	U-46619	8.0	Boersma et al. (1999)
		Uterine artery	U-46619	8.5	Baxter et al. (1995)
		Uterus (non-pregnant)	U-46619	8.6	Senchyna and Crankshaw (1996)
		Uterus (pregnant)	U-46619	8.5a	Senior et al. (1993)
	Guinea pig	Aorta	U-46619	8.77	Lumley et al. (1989)
		Aorta	U-46619	9.4	Ogletree and Allen (1992)
		Trachea	U-46619	8.26	Featherstone et al. (1990)
		Trachea	U-46619	9.43	Tymkewycz et al. (1991)
		Trachea	U-46619	10.0	Ogletree and Allen (1992)
	Rat	Aorta	U-46619	7.87	Lumley et al. (1989)
		Aorta	U-46619	7.49	Furci et al. (1991)
		Aorta	U-46619	8.41	Tymkewycz et al. (1991)
		Aorta	U-46619	8.3	Ogletree and Allen (1992)
		Trachea	U-46619	8.31	Lydford and McKechnie (1994)
GR-83783j	Rat	Aorta	U-46619	7.5	Campbell et al. (1991a)
GR-108774j	Rat	Aorta	U-46619	9.2	Campbell et al. (1991b)
ICI-192605	Human	Platelet/aggregation	U-46619	8.16	Brewster et al. (1988)
		Umbilical artery	U-46619	8.1	Boersma et al. (1999)
		Umbilical vein	U-46619	9.07	Daray et al. (2003)
		Uterus (non-pregnant)	U-46619	9.2	Senchyna and Crankshaw (1996)
	Rat	Aorta	U-46619	8.4	Brewster et al. (1988)
I-PTA-OH	Guinea pig	Lung parenchyma	U-46619	5.6	Norman et al. (1992)
	Rat	Lung parenchyma	U-46619	5.8	Norman et al. (1992)
I-SAP	Human	Platelet/aggregation	U-46619	8.01k	Naka et al. (1992)
KW-3635	Human	Platelet/aggregation	U-46619	8.88	Karasawa et al. (1991a)
	Guinea pig	Aorta	U-46619	7.74	Karasawa et al. (1991b)
L-655240	Guinea pig	Aorta	U-44069	8.0	Hall et al. (1987)
		Aorta	U-44069	8.0	Hall et al. (1987)
ONO-11120	Human	Platelet/binding	[125I]-PTA-0H	7.71c	Narumiya et al. (1986)
		Platelet/aggregation	U-46619	7.49	Tymkewycz et al. (1991)
	Guinea pig	Trachea	U-46619	8.07	Tymkewycz et al. (1991)
	Rat	Aorta	U-46619	7.14	Tymkewycz et al. (1991)
		Platelet/aggregation	U-46619	7.38	Tymkewycz et al. (1991)
ONO-NT-126	Human	Astrocytoma cell/PI	STA2	10.0	Nakahata et al. (1990)
Ridogrelj	Human	Platelet/aggregation	U-46619	5.7	Watts et al. (1991)
	Rat	Tail artery	U-46619	5.5	Janssens et al. (1990)
(±)-S-145 (Domitroban)l	Human	Astrocytoma cell/PI	STA2	8.48	Nakahata et al. (1990)
		Platelet membrane/binding	[3H]-(+)-S-145	9.35	Kishino et al. (1991)
	Rat	Aorta smooth muscle cell/binding	[3H]-SQ-29548	9.5a	Hanasaki et al. (1988)
S-18886 (Terutroban)	Rabbit	Saphenous vein	U-46619	8.9	Cimetière et al. (1998)
SQ-29548	Human	Astrocytoma cell/PI	STA2	8.08	Nakahata et al. (1990)
		Immortalized ciliary epithelial cell/PI	U-46619	7.7m	Sharif et al. (2002)
		Corpus cavernosum	U-46619	9.0	Angulo et al. (2002)
		Umbilical artery	U-46619	7.6	Boersma et al. (1999)
		Umbilical vein	U-46619	7.96	Daray et al. (2003)
		Uterus (non-pregnant)	U-46619	8.2	Senchyna and Crankshaw (1996)
	Pig	Coronary artery	U-46619	8.8a	Kromer and Tippins (1996)
	Rabbit	Aorta	U-46619	7.95	Yoshida et al. (2007)
	Guinea pig	Aorta	U-46619	7.96	Dubéet al. (1992)
		Aorta	U-46619	8.9	Ogletree and Allen (1992)
		Aorta	U-46619	8.5	Zhang et al. (1996)
		Lung parenchyma	U-46619	7.7	Norman et al. (1992)
		Trachea	U-46619	8.70	Dubéet al. (1992)
		Trachea	U-46619	8.9	Ogletree and Allen (1992)
	Rat	Aorta	U-46619	9.2	Zhang et al. (1996)
		Lung parenchyma	U-46619	7.2	Norman et al. (1992)
SQ-30741	Human	Coronary artery	U-46619	7.54	Maassen VanDenBrink et al., 1996)
	Human	Umbilical artery	U-46619	7.0	Boersma et al. (1999)
	Guinea pig	Aorta	U-46619	8.1	Ogletree and Allen (1992)
		Trachea	U-46619	8.6	Ogletree and Allen (1992)
	Rat	Aorta	U-46619	7.9	Ogletree and Allen (1992)
YM-158	Guinea pig	Trachea	U-46619	8.81n	Arakida et al. (1998)
Z-335	Human	Platelet membrane/binding	[3H]-SQ-29548	7.52	Tanaka et al. (1998)
		Platelet/shape changeo	U-46619	8.02	Yoshida et al. (2007)
	Rabbit	Aorta	U-46619	8.64	Yoshida et al. (2007)
ZD-1542j	Guinea pig	Lung parenchyma	U-46619	8.5	Brownlie et al. (1993)
		Trachea	U-46619	8.3	Brownlie et al. (1993)
	Rat	Aorta	U-46619	8.51	Brownlie et al. (1993)

pA₂ values relate to functional assays. Recombinant (rc-) systems: prostanoid receptor followed by the carrier cell line and second messenger measurement. Smooth muscle preparations: contraction or relaxation of induced tone. Platelets: all data from plasma-free platelet suspensions. Where pA₂ values are not available, pK_i/pK_D values derived from radioligand binding are given (italics). Affinity values in square brackets derive from the patent literature.

Structures of compounds: 1–5, Figure 1; 6–12, Figure 2; 13–15, Figure 3; 16–19, Figure 4; 20, Figure 5; 21–24, Figure 7.

BEAS, human bronchial epithelium; CHO, Chinese hamster ovary; COS-7, African green monkey kidney; HEK, human embryonic kidney; CRE, cAMP response element; 16,16-DM PGE₂, 16,16-dimethyl PGE₂; PI, phosphoinositide.

^aOur calculation.

^bAppreciable affinity for corresponding TP receptor.

^cpIC_50.

^dInsurmountable antagonism in functional DP₂ systems.

^e1 µM SC-51322 present.

^fReplacement for sulprostone, which had slow onset and offset.

^gPossible interference by EP₂ system.

^hpKi.

ⁱpK_b for non-competitive antagonism.

^jCombined TP antagonist/TXS inhibitor.

^kShape change seen.

^lData for (+)- and (−)-enantiomers in Kishino et al. (1991).

^mNon-competitive antagonism.

ⁿpA₂= 8.87 for LTD₄ antagonism.

^oSuppression of maximum aggregation response.

Inhibition-curve (or Cheng-Prusoff) protocols have been used infrequently in prostanoid receptor studies. Bley et al. (2006) estimated the pA₂ of the IP antagonists RO-1138452 and RO-3244794 using carbacyclin as the fixed-concentration agonist in a human rc-IP receptor – cAMP assay, but failed to use the modified form of the Cheng-Prusoff equation (Craig, 1993; Lazareno and Birdsall, 1993; Leff and Dougall, 1993). These protocols have the advantages of operating over a lower agonist concentration range than Schild protocols and providing direct observation of the rate of onset of antagonism. The latter is important in recognizing the slow approach to steady state that occurs with high-affinity antagonists at low concentration, and also with highly lipophilic antagonists (Jones et al., 2008) that regularly emerge from combinatorial chemistry – high-throughput screening. Ultimately, it is important and even preferable to define the pharmacology in the human target tissue; human rc-receptor assays are a useful accompaniment. Successful drugs require the correct pharmacological attributes, but physical chemical properties are also important.

DP₁ receptor antagonists

Development

Antagonists for, what we now know to be, the DP₁ receptor subtype, were first described in the 1970s and early 1980s. Examples include N-0164 (MacIntyre and Gordon, 1977), diphloretin phosphate (Westwick and Webb, 1978) and desacetyl-1-nantradol (Horne, 1984), but none of these compounds exhibit the potency and selectivity essential for unambiguous receptor classification. The simple xanthone-carboxylic acid AH-6809 (Keery and Lumley, 1988) has sufficient DP₁ affinity (pA₂= 5.9–6.6, Table 2), but it has been mainly employed as an EP₁ antagonist (see later). Indeed, the hydantoin derivative, BW-A868C (Figure 1) has been the only selective, surmountable and competitive DP₁ antagonist (pA₂ > 9 for the human subtype) readily available to pharmacologists (Giles et al., 1989; Lydford et al., 1996c). The N-benzyl substituent is crucial for antagonist activity; in a related series of bicyclic-hydantoin analogues a progression from full agonism to virtually pure antagonism is seen with hydrogen, methyl, ethyl and n-propyl substituents on N¹⁰ (Giles and Leff, 1992). BW-A868C also has low affinity (pA₂= 5.1) for the EP₄ subtype (Lydford et al., 1996c). ZK-138357 (Schering AG) is a moderate-affinity DP₁ antagonist (Table 2) with some structural similarity to BW-A868C.

Recently, the potential pathological role of PGD₂, especially in allergic disorders, has been revived resulting in the discovery and evaluation of highly selective DP₁ antagonists of several structural classes. The most important of these (Figure 1) and their therapeutic applications are described below.

Bicycloheptanes

Chemists in Shionogi have synthesized selective DP₁ antagonists containing a bicyclo[2.2.1]heptane ring system akin to that present in PGH₂ (see inset in Figure 8) (Tsuri et al., 1997; Honma et al., 1998; Mitsumori et al., 2003a). An initial lead was the racemic compound 1 previously shown to be a TP antagonist (Narisada et al., 1988). Subsequent structure–activity relationship (SAR) studies revealed that a 6,6-dimethylbicyclo[3.1.1]heptane (pinane) system could substitute for the bicycloheptane ring (Tsuri et al., 1997; Mitsumori et al., 2003b; Yoshikawa et al., 2005) and carbonylamino or sulphonylamino linkages to the ω-aryl moiety were required for potent DP₁ antagonism (Tsuri et al., 1997; Honma et al., 1998; Mitsumori et al., 2003b). Accordingly, S-5751 has high affinity for the DP₁ receptor (pK_i= 8.8) and is orally active in models of allergy and inflammation in the guinea pig (Tsuri et al., 1997; Arimura et al., 2001; Mitsumori et al., 2003b; Yasui et al., 2008). In addition, lead optimization of the (+)-isomer of compound 1, which has much higher DP₁ selectivity than its mirror-image, led to the 1-methoxy-dibenzo[b,d]furan 2. This antagonist is potent, orally bioavailable and efficacious in guinea pig models of conjunctivitis and allergen-induced bronchoconstriction (Mitsumori et al., 2003a).

Allergan have also filed patents claiming DP₁ receptor antagonism based on a 1(S),4(S)-7-oxabicyclo[2.2.1]heptane scaffold (Krauss et al., 2005). One of these, compound 3, has a pA₂ of ∼7.4 for the human rc-DP₁ receptor and represents a logical structure for lead optimization.

Indole acetic acids

A non-prostanoid exploited for DP₁ antagonism at Ono Pharmaceuticals is the cyclo-oxygenase (COX) inhibitor/non-steroidal anti-inflammatory drug (NSAID), indomethacin (Figure 1). Initial studies showed that the acetic acid moiety could be switched to position 4 on the indole ring (Torisu et al., 2004a). Optimization led to the discovery of two benzoxazines, ONO-AE3-237 and compound 4, with high DP₁ antagonist selectivity and (sub)nanomolar affinity (Torisu et al., 2004b,c,d; Torisu et al., 2005). Administered orally, both compounds effectively suppressed PGD₂- and allergen-induced vascular permeability in the guinea pig conjunctiva (Torisu et al., 2004c). Pharmacokinetic studies on ONO-AE3-237 given by the oral (10 mg·kg⁻¹) and intravenous (1 mg·kg⁻¹) routes to fasted rats afforded plasma half-lives of 7.8 and 9.2 h respectively. The compound has a high volume of distribution indicating good tissue penetration and is 48% bioavailable at a dose of 10 mg·kg⁻¹ p.o. (Torisu et al., 2004c).

Merck Frosst (Wang et al., 2002; Berthelette et al., 2003) and Sanofi-Aventis (Yang et al., 2008) also filed patents claiming DP₁ antagonists with further variation of the acetic acid position on the indole template. Screening of the Merck compound collection identified a difluoro-indole that had low-nanomolar affinity for DP₁ and TP receptors (Sturino et al., 2006) and this, and other analogues, were optimized resulting in two 7-methylsulphone derivatives with high DP₁ affinity (K_i∼2 nM) and at least 100-fold selectivity over other prostanoid receptors. However, both compounds displayed poor pharmacokinetics in the rat, in particular extensive biliary excretion (Sturino et al., 2007). This liability was overcome when the 5-substituent and the 7-methylsulphone on the indole ring were replaced by methylsulphone and fluorine, respectively, to give MK-0542 (laropiprant; Figure 1). Laropiprant has very high affinity for the human rc-DP₁ receptor (pK_i= 10.5), with ∼300-fold lower affinity for the corresponding TP receptor (Sturino et al., 2007). Excellent pharmacokinetic profiles have been found in the rat, dog, monkey and man (Chang et al., 2007; Karanam et al., 2007; Sturino et al., 2007; Lai et al., 2008b). Given orally to healthy male volunteers, laropiprant at single doses up to 900 mg and multiple doses up to 450 mg is rapidly absorbed (T_max= 0.8–2 h), demonstrates dose-proportional systemic exposure, has a half-life of 12–18 h and is generally well tolerated; this pharmacokinetic profile is unaffected by food (Karanam et al., 2007; Lai et al., 2008b). At a dose of 6 mg, laropiprant was effective in antagonizing PGD₂-induced cAMP accumulation in human platelets ex vivo, indicating an interaction with the desired molecular target. Evidence for TP receptor blockade was also detected, but this effect was deemed not to be clinically relevant (Lai et al., 2008b).

Merck Frosst has also disclosed ‘backup’ DP₁ antagonists, in which the indole template present in laropiprant is inverted. The tetrahydropyridoindole 5 (Figure 1) exhibited the best profile (pK_i= 9.0 and 6.8 at DP₁ and TP receptors respectively) and is considered a suitable candidate for development (Beaulieu et al., 2008).

Other structural classes

Certain aminopyrimidines have also been claimed in the patent literature to be DP₁ antagonists but neither in vitro nor in vivo pharmacological data are yet available (Langevin et al., 2007; Stefany et al., 2007).

Therapeutic applications

PGD₂ is an established mediator of allergic disease. It is the major prostanoid released from mast cells (Lewis et al., 1982; Peters et al., 1982) and is also secreted, albeit in lower amounts, by T-lymphocytes of the Th2 subset (Tanaka et al., 2000). In asthma, dermatitis and rhinitis, allergen challenge leads to the rapid production of PGD₂ (Naclerio et al., 1983; Murray et al., 1986; Charlesworth et al., 1991) and PGD₂, itself, can reproduce many symptoms associated with allergic phenomena (see Pettipher, 2008). However, therapeutic/commercial success has not yet been attained with DP₁ antagonists probably because functional and subsequently molecular evidence emerged for a second subtype of PGD₂-sensitive receptor that is strongly implicated in several manifestations of allergic disease including eosinophil infiltration, mucus hyper-secretion and plasma extravasation (see next section).

In terms of allergic inflammation, activation of DP₁ receptors is known to mediate pathological changes in blood flow. In allergic rhinitis, vessels within the nasal mucosa become engorged leading to congestion and the release of plasma proteins, which contribute to enhanced nasal secretions. These effects are mimicked by PGD₂, which explains the limited efficacy of histamine H₁ antagonists in allergic rhinitis (see Pettipher, 2008). Sturino et al. (2007) have shown that laropiprant abolishes the marked increase in nasal airway resistance induced by intranasal instillation of PGD₂ in conscious sheep. Significantly, comparable data have also been obtained in 15 healthy, non-smoking male volunteers in whom laropiprant (25 mg or 100 mg q.d. for 3 days) significantly suppressed PGD₂-induced nasal congestion (Van Hecken et al., 2007). Interestingly, PGD₂ fails to lower diastolic blood pressure (BP) in human volunteers (Heavey et al., 1984) although (BW)-192C86, a DP₁ partial agonist (Gray et al., 1992), is an effective depressor and BW-245C evokes adverse cardiovascular effects consistent with vasodilatation (Al Sinawi et al., 1985). The reason for this discrepancy is unknown, but given that PGD₂ is a potent vasoconstrictor in several species (Jones, 1976; 1978;) and the DP₂ receptor is expressed in the aorta (Nagata and Hirai, 2003) and potentially other blood vessels, its activation may oppose the BP-lowering activity of PGD₂ acting via the DP₁ receptor.

The ability of PGD₂ to promote adverse vasodilatation is also associated with the therapeutic use of niacin (vitamin B3), which, in high doses, is used clinically to lower plasma cholesterol. Niacin may also have anti-oxidant and anti-inflammatory activity and is used, in conjunction with statins, to treat dyslipidaemia (see Kamanna et al., 2008). The adverse effects of niacin, particularly flushing, are due to receptor (GPR109A)-mediated release from Langherhans' cells of PGD₂ and PGE₂, which promote vasodilatation of dermal and cerebrovascular capillaries though activation of DP₁, EP₂ and EP₄ receptors. Merck (Paolini et al., 2008) and Sanofi-Aventis (Harris, 2008) have claimed that a DP₁ antagonist could limit the cardiovascular liability of niacin if given as a combination therapy, thereby providing a better-tolerated drug. Indeed, laropiprant significantly reduced niacin-induced flushing in normal and dyslipidaemic subjects when compared with niacin alone (Cheng et al., 2006; Lai et al., 2007; Paolini et al., 2008). Accordingly, these and other findings led Merck to develop the investigational combination product Cordaptive, which significantly reduced the vascular side effects of niacin in patients with primary hypercholesterolaemia and mixed dyslipidaemia (see Kamanna et al., 2008). However, in April 2008 the US Food and Drug administration did not approve Merck's application to market Cordaptive (see http://www.merck.com/newsroom/press_releases/research_and_development/2008_0428.html), and at the same time, rejected the name of the combination therapy. The US Food and Drug administration's decision not to approve Cordaptive, now renamed Tredaptive, is unclear given that the European Medicines Agency has approved this new extended-release combination therapy (see http://www.emea.europa.eu/humandocs/Humans/EPAR/tredaptive/tredaptive.htm). One possibility is that the beneficial effects of laropiprant are not superior to aspirin (Kamanna et al., 2008) and the risk/benefit ratio is a primary consideration when the long-term effects of a new drug class are unknown.

DP₁ antagonists may also have utility in the treatment of allergic asthma, although, currently, this is controversial (see Pettipher, 2008). Persuasive evidence is available that activated mast cells, through their ability to generate PGD₂, promote T-lymphocyte polarization towards a Th2 phenotype (Faith et al., 2005; Hammad and Lambrecht, 2008). Mechanistically, it is believed that mast cell-derived PGD₂ activates DP₁ receptors on dendritic cells within the respiratory mucosa. This interaction then suppresses the generation of interleukin (IL)-12 (which normally favours a Th1 cell bias) leading to Th2 dominance (Kitawaki et al., 2006; Theiner et al., 2006). A cycle of chronic immunological activation then ensues through further production of IgE and mast cell activation. Indeed, these data are consistent with allergic airway responses being less intense in DP₁ receptor-deficient mice when compared with wild-type animals (Matsuoka et al., 2000). However, the role of DP₁ receptors in regulating allergic reactions is complex. For example, the administration to wild-type mice of the DP₁ agonist, BW-245C, paradoxically reduced pulmonary allergic responses whereas DP₁ receptor null mice were unaffected (Hammad et al., 2007). To explain this apparently contradictory result it has been proposed that DP₁ receptor-mediated inhibition of IL-12 release from dendritic cells during host sensitization promotes Th2 polarization (i.e. is pro-inflammatory). In contrast, during maintained airway inflammation, PGD₂ by suppressing dendritic cell function reduces Th2 cell function (i.e. is anti-inflammatory). Indeed, the DP₁ antagonist, S-5751, attenuated rather than exacerbated allergen-induced inflammation in sensitized guinea pigs (Arimura et al., 2001). Thus, the harmful/protective actions of PGD₂ may depend on when and where it is produced (Pettipher, 2008).

DP₂ receptor antagonists

Development

Studies conducted in the 1970s showed that PGD₂ elicited peripheral vasoconstrictor responses that were unlikely to be due to activation of, what we now understand to be, DP₁, FP or TP receptors (Jones, 1976; 1978;). In particular, 15-oxo PGD analogues (Jones and Wilson, 1978) had unexpectedly high agonist potency. Moreover, evidence for DP receptor heterogeneity was provided in 1985 by the finding that PGD₂ and six related analogues evoked functional responses across a variety of PGD₂-sensitive systems that were not mimicked by the selective DP agonist, BW-245C (Narumiya and Toda, 1985). The results of several other studies also indicated the existence of multiple DP receptors (Woodward et al., 1990a; 1993b; Rangachari and Betti, 1993; Fernandes and Crankshaw, 1995; Rangachari et al., 1995). However, it was not until 1999 that the idea of multiple DP-receptors really began to gain general acceptance. Nagata et al. (1999b) identified a novel molecule expressed on the surface of minor populations of CD4+ T-lymphocytes, which resembled activated Th2 cells in that they released IL-4, IL-5 and/or IL-13 but not interferon-γ on stimulation. This orphan site was named ‘chemoattractant receptor-homologous molecule expressed on Th2 cells’, or CRTh2, as primary sequence analysis showed significant amino acid homology to receptors that mediate chemoattraction (Hirai et al., 2001). Indeed, despite the ability of PGD₂ to interact with the DP₁ subtype and CRTh2 with comparably high affinity (K_i= 45 and 61 nM respectively), the two cognate receptors are quite dissimilar at the amino acid level (Hirai et al., 2001). A follow-up study by the same investigators discovered a mast cell-derived factor that could increase the cytosolic free Ca²⁺ concentration in CRTh2-expressing cells (Nagata et al., 1999a), which was identified as PGD₂ (Hirai et al., 2001). This latter finding coincided with the publication of a pharmacological study in which Monneret et al. (2001) found that PGD₂ was chemotactic for eosinophils and also up-regulated the expression of CD11b and L-selectin by a mechanism that was neither mimicked by BW-245C nor blocked by the DP₁ antagonist, BW-A868C. Thus, a second DP receptor subtype (DP₂, aka CD294) was identified on Th2 cells and eosinophils that mediated responses diametrically opposite to the inhibitory effects classically associated with DP₁ agonism. Given that CRTh2 is now known to be ubiquitously expressed within (Nagata et al., 1999a) and outwith (Hirai et al., 2001; Nagata and Hirai, 2003; Kostenis and Ulven, 2006; Kim and Luster, 2007) the immune system (Sawyer et al., 2002; Nagata and Hirai, 2003), the term ‘DP₂’ is a more appropriate designation and is used throughout this review.

The ability of PGD₂ to act as a chemoattractant for pro-inflammatory cells and to release Th2-like cytokines has resulted in a concerted effort by the pharmaceutical industry to synthesize selective DP₂ antagonists. Indeed, such compounds may be useful in suppressing a myriad of Th2-driven inflammatory pathologies including asthma, otitis, contact dermatitis and rhinitis. At the time of writing, in excess of 90 patents had been filed claiming selective DP₂ antagonists. In the sections below the main structural classes are described and affinity estimates of lead antagonists (Figure 2) are given in Table 2.

Figure 2.

DP₂ receptor antagonists. Inverted 2-methyl-indole-acetic acid residues (compare with 6) are highlighted in red; ramatroban has an extra methylene (C2a). The phenylacetic acid moiety is shown in blue in fenclofenac, a lead molecule for compound 11. Compound 12, K-117 and K-604 contain a tetrahydroquinoline residue (green).

Indole acetic acids

In addition to providing a scaffold for the development of DP₁ antagonists, indomethacin (Figure 1) is also a selective, albeit weak, DP₂ agonist (Hirai et al., 2002; Stubbs et al., 2002). Exploiting this property, Pfizer first reported a benzothiazole derivative 6 that had a pA₂ of 6.8 and was ∼40-fold selective for the DP₂ receptor (Bauer et al., 2002). Subsequently, several patents describing highly potent and selective DP₂ antagonists were filed by AstraZeneca (Baxter et al., 2003a,b; Birkinshaw et al., 2003; Bonnert et al., 2003; 2004; 2005a,b,c; Bonnert and Rasul, 2004). An initial hit was a 7-chloroquinoline derivative of indomethacin (pA₂∼7 for human rc-DP₂ receptor), which also inhibited COX-1 with high potency (Birkinshaw et al., 2006). Inversion of the indole template (Figure 2) and substitution of the 5-methoxy moiety by methyl increased antagonist potency by 23-fold and reduced COX-1 inhibition by a factor of 10 (Birkinshaw et al., 2006). Addition of chlorine at position 8 of the quinoline to give 7 also increased DP₂ affinity by 13-fold (IC₅₀ for inhibition of [³H]-PGD₂ binding = 2.3 nM). This derivative has a preferred biological profile with relatively weak binding to plasma proteins and good bioavailability in rats (76%) and dogs (100%), with half-lives of 1.7 and 5.3 h respectively (Birkinshaw et al., 2006). Compounds having an arylthio substituent on position 3 of the indole nucleus also display very potent antagonism at the DP₂ receptor; compound 8 has a binding IC₅₀ of 0.4 nM (Bonnert and Rasul, 2004).

Related 1-acetic acid derivatives from Oxagen containing methylene or sulphonyl spacers between the aromatic moieties have been reported as potent DP₂ antagonists (Middlemiss et al., 2005a,b,c,d; Armer et al., 2006; Lovell, 2007), including compound 9, which has a K_i of 68 nM and a DP₂/DP₁-selectivity ratio of approximately 150. Functionally, 9 potently inhibits DP₂ receptor-mediated human eosinophil shape change and Th2 cell chemotaxis with IC₅₀ values of 74 and 67 nM respectively (Armer et al., 2005). Moreover, this compound is metabolically stable, has no inhibitory effect on five of the major cytochrome P450 enzymes (1A2, 2C19, 2C9, 2D6, 3A4) and fails to induce CYP3A4, CYP1A and CYP2C9. In rats, 9 is 56% bioavailable and has a half-life of 5.5 h following oral administration. Oxagen has reported the development of a lead compound, ODC9101 (aka OC459), which is in Phase IIa clinical trials for asthma. According to the company's website, ODC9101 has completed safety evaluations, is orally active and suitable for once-a-day dosing. Oxagen has also reported the development of a pre-clinical back-up molecule, OC499, and DP₂ antagonists for non-oral delivery (OC1768) and topical administration (OC2125, OC2184; see http://www.oxagen.co.uk/pdfs/CRTH2summary.pdf). The structures of these compounds have not been disclosed.

Athersys has also described a series of indole acetic acid derivatives with potent DP₂ antagonist activity (Bennani et al., 2006) including substituted 3-benzylphthalazin-1(2H)-ones which have radioligand binding IC₅₀ values in the low nanomolar range; their development status is unknown.

Ramatroban and analogues

Ramatroban (Bay u3405) was originally described as a TP antagonist with a pA₂ of ∼8.8 on human tissues (McKenniff et al., 1991). Later studies revealed DP₂ antagonism (Sugimoto et al., 2003), albeit of lower affinity (pA₂= 7.44; Mathiesen et al., 2006). The structural similarity of ramatroban to compounds 7–9 is clear. These observations were the impetus for the synthesis of compounds with increased selectivity for the DP₂ receptor (Arimura et al., 2003). An example is 10, in which the amide nitrogen atom is included in the tricyclic system. This compound has high affinity for the DP₂ receptor (pA₂= 8.53) with significantly reduced TP receptor-blocking activity (Fretz et al., 2005).

Other minor changes to ramatroban also resulted in compounds with a high degree of DP₂ selectivity. 7TM Pharma reported that N-methylating the sulphonamide or truncating the propionate moiety to acetate produces very selective (>1000-fold over DP₁ and TP), high-affinity DP₂ antagonists with K_i values of 1.9 (TM-30642) and 0.51 nM (TM-30643) respectively (Ulven and Kostenis, 2005). Furthermore, making both modifications to produce TM-30089 (aka CAY-10471) preserved the DP₂ affinity of TM-30643 and further increased the DP₂/TP-selectivity ratio to >10 000 (Ulven and Kostenis, 2005). Interestingly, in functional studies (e.g. [³⁵S]-GTPγS binding/inositol phosphate accumulation; PGD₂-induced eosinophil shape change), ramatroban and TM-30642 are surmountable competitive antagonists whereas TM 30643 and TM 30089 suppress the maximal response in a concentration-dependent manner (Mathiesen et al., 2006). The insurmountable behaviour of TM-30089 and TM-30643 may be due to its slow dissociation from the DP₂ receptor, which also results in long-lasting antagonism (Mathiesen et al., 2006). Whether such pharmacological behaviour occurs in vivo and would be therapeutically advantageous is currently unclear. However, this could be a desirable property as slowly dissociating drugs should act much longer than would be predicted from their plasma half-lives (Mathiesen et al., 2006). 7TM Pharma, in partnership with Ortho-McNeil-Janssen Pharmaceuticals, has a compound in late-stage lead optimization although neither the structure nor the profile of the antagonist has been disclosed (see http://www.7tm.com/News.aspx?M=News&PID=42&NewsID=39).

Phenyl acetic acids

Another NSAID, fenclofenac (Figure 2), provided the starting point for the synthesis of DP₂ antagonists based on a phenylacetic acid template. An initial hit claimed in the original patent filed by Pfizer (Bauer et al., 2002) had a 4-chlorophenylthio substituent resulting in a functional DP₂/DP₁-selectivity ratio of 40. Several other companies, including AstraZeneca, have since filed patents for bis-ether derivatives such as 11, which has a binding pIC₅₀ of 9.0 (Bonnert et al., 2005c).

Tetrahydroquinolines

Millennium and Warner-Lambert (now Pfizer) were the first to disclose DP₂ antagonists within the tetrahydroquinoline class (Awad et al., 2004; Ghosh et al., 2004; 2005; Kuhn et al., 2004). These compounds are unique in that they are non-acidic indicating that a carboxylic acid moiety is not essential for DP₂ antagonism, as previously assumed (see Pettipher et al., 2007). The 4-amino-tetrahydroquinoline 12 (Figure 2) is reported to gain access to the cerebrospinal fluid after oral dosing and also is efficacious in animal models of inflammation at an oral dose of 25 mg·kg⁻¹ (Corradini et al., 2005). However, at the time of writing detailed pharmacological data on non-acidic DP₂ antagonists is sparse. Researchers at Kyowa Hakko Kogyo have reported K_i values of 5.5 and 11 nM for K-117 and K-604 respectively, with minimal interaction with TP or DP₁ receptors at concentrations up to 1 µM (Mimura et al., 2005).

Therapeutic applications

Arguably, allergic inflammation is the primary indication for antagonists that selectively block the DP₂ receptor. Indeed, the gene encoding this receptor shows a particularly strong association with asthma in Chinese and African-American populations (Huang et al., 2004). Moreover, there is good evidence from in vitro and in vivo studies in laboratory animals that PGD₂, acting via the DP₂ receptor, can mediate many of the cardinal features of allergic airways inflammation (see Ulven and Kostenis, 2006; Pettipher, 2008 for detailed reviews). The most important observations that have led to this view can be summarized as follows:

1. PGD₂ and selective DP₂ agonists promote chemotaxis of eosinophils, basophils and CD4+ T-lymphocytes of the Th2 subset and this effect is abolished by a neutralizing anti-DP₂ receptor antibody (Hirai et al., 2001; Monneret et al., 2001).

2. PGD₂ promotes pulmonary eosinophilia in rats; this effect is mimicked by selective DP₂, but not DP₁, agonists and is abolished by ramatroban (Almishri et al., 2005; Shiraishi et al., 2005).

3. In guinea pigs, the DP₂ agonist, Δ¹²-PGJ₂, mobilizes eosinophils from the bone marrow (Heinemann et al., 2003).

4. The DP₂ agonist, 13,14-dihydro-15-oxo PGD₂, promotes pulmonary eosinophilia and exacerbates histopathology in a murine model of allergic asthma (Spik et al., 2005).

5. Ramatroban and DP₂ antagonists devoid of TP receptor-blocking activity reduce pulmonary eosinophilia in several animal species in response to allergen challenge (Nagai et al., 1995; Uller et al., 2007; Pettipher, 2008).

6. PGD₂ promotes the production of Th2 cytokines in vivo including IL-4, IL-5 and IL-13 (Fujitani et al., 2002) and in vitro, this can occur in the absence of allergen or co-stimulatory molecules (Xue et al., 2005).

7. The expression of the DP₂ receptor on eosinophils is up-regulated in atopic individuals (see Kostenis and Ulven, 2006).

8. High concentrations of PGD₂ are present in the airways of asthmatic subjects after antigen challenge (Murray et al., 1986).

9. The expression of the DP₂ receptor on Th2 T-lymphocytes is up-regulated in individuals sensitized to pollen or house dust mite or who have atopic dermatitis (Iwasaki et al., 2002).

The proof of this line of argument is the marketing of ramatroban in Japan under the trade name Baynas for the treatment of perennial allergic rhinitis; its clinical efficacy (e.g. reduction of symptoms and of chronic nasal swelling) has been attributed to DP₂ receptor blockade (Terada et al., 1998).

In addition to anti-allergic indications, DP₂ antagonists may have utility in combating neuropathic pain (Corradini et al., 2005) where the up-regulation of COX-2 and the subsequent formation of PGs are central to disease pathophysiology (Camu et al., 2003).

Hybrid DP₁/DP₂ receptor antagonists

In considering the SAR data described in the preceding sections, indole acetic acids may be a fruitful starting point for antagonists that block both DP₁ and DP₂ receptors. Indeed, in the context of allergic diseases, a hybrid antagonist may exert clinically relevant, beneficial effects that are not achieved when just one DP receptor is targeted. Thus, blockade of the DP₁ subtype would prevent PGD₂ from inhibiting the generation of IL-12 from dendritic cells, thereby inhibiting the polarization of T-lymphocytes to a Th2 phenotype that occurs during host sensitization. Antagonism of the DP₂ receptor would, at the same time, suppress the chemotactic activity of PGD₂ towards eosinophils, basophils and T-lymphocytes and so reduce pulmonary leukocyte burden and inflammatory status. Theoretically, this latter action would overcome any DP₁ receptor-mediated anti-inflammatory effect of PGD₂ on dendritic cells during maintained inflammation (Pettipher, 2008). It is also tempting to speculate that a non-selective DP antagonist that also has TP receptor-blocking activity could be efficacious in allergic asthma. In this scenario, TP receptor blockade would reduce the ability of PGD₂ to induce bronchoconstriction, which in humans is mediated through the TP receptor (Beasley et al., 1989). Again, indole acetic acids and ramatroban analogues may provide good templates for optimization.

EP₁ receptor antagonists

Development

Figure 3 shows the commonly used EP₁ antagonists; while some may be considered as prostanoids, none is structurally close to PGE₂. The first EP₁ antagonist was SC-19220, a dibenzoxazepine hydrazide (Sanner, 1969). Although of low affinity (pA₂= 5.5), SC-19220 proved useful in the early characterization and elucidation of the roles of EP receptors. For example, on guinea pig trachea, 3–50 µM SC-19220 (and indomethacin) suppressed inherent tone, indicating that PGE₂ acting via EP₁ receptors is the likely mediator (Farmer et al., 1974). Also, SC-19220 at 10 µM equally inhibited matching contractions of guinea pig trachea induced by 16,16-dimethyl PGE₂, iloprost and isocarbacyclin, but had no effect on contractions induced by U-46619 or histamine; these results demonstrated that certain PGI₂ analogues could potently activate EP₁ receptors (Dong et al., 1986).

Figure 3.

EP₁ receptor antagonists. The natural ligand PGE₂ is shown in the box. The dibenzoxazepine residue in SC-51322 is shown in blue. Aryl-sulphonamido residues in antagonists with prostanoid and non-prostanoid structures are shown in cerise; ONO-NT-012 contains a styryl-sulphonamido moiety. The 1,2-biaryl-cyclopentene pharmacophores in GW-848687 and MF-266-1 are shown in red. Ring A in GW-848687 is part of a picolinic acid (pyridine-2-carboxylic acid) residue.

SC-19220 (7.5–30 mg·kg⁻¹ i.p.) inhibited carrageenan-induced inflammation in the rat without suppressing PGE₂ levels at the injury site (Barbieri et al., 1977). These and other observations were the stimulus for the synthesis of higher-affinity EP₁ antagonists with potential as anti-inflammatory/analgesic drugs in man; alteration of the acetyl group was the most rewarding strategy. SC-51089 was unusual in the series in releasing hydrazine and was dropped from development (Hallinan et al., 1993). The thioether SC-51322, which is much more potent than the corresponding sulphone (Hallinan et al., 1994), has become the agent of choice for receptor characterization. It behaved competitively over the range 25–625 nM in a human rc-EP₁ receptor – reporter gene assay (pA₂= 8.8; Schild plot slope = 0.91; Durocher et al. (2000). In rat hepatocytes, SC-51322 at 100 nM abolished DNA synthesis induced by PGE₂ or 17-phenyl PGE₂ (Table 1), while 1 µM SC-51322 did not affect responses to either PGF_2α or PGI₂; the selective involvement of EP₁ receptors in the PGE response is clear (Kimura et al., 2000; 2001;). However, not all inferences about EP₁ receptor involvement are as secure. In a study on rat progenitor Leydig cells, 3 and 30 µM SC-51322 suppressed IL-1β expression induced by 10 µM 17-phenyl PGE₂ by 28% and 59% respectively (Walch et al., 2003). Given that this system responded well to 100 nM cloprostenol, a potent FP agonist (Table 1), the action of 17-phenyl PGE₂ and its inhibition by SC-51322 may also have involved FP receptors. SC-51322 at 10 µM did not inhibit PGE₂-mediated inhibition of superoxide generation in human blood neutrophils (EP₂ system) (Kanamori et al., 1997) or inhibit sulprostone-induced contraction of rat femoral artery (EP₃ system) at 1 µM (Hung et al., 2006).

Another widely used EP₁ antagonist is AH-6809. At 0.1–10 µM, it blocked EP₁-mediated actions of PGE₂, 16,16-dimethyl PGE₂ or 17-phenyl PGE₂ in an apparently competitive manner (pA₂= 7.4) (Coleman et al., 1987; Eglen and Whiting, 1988; Lawrence et al., 1992). It did not block EP₃ receptor-mediated contraction of guinea pig ileum (Lawrence et al., 1992) and human pulmonary artery (Qian et al., 1994) at 2 and 5 µM respectively and had no effect on the pre-synaptic EP₃ action of sulprostone on rat trachea at 3 µM (Rackéet al., 1992). However, AH-6809 blocks EP₂ receptors (see later) as well as DP₁ and TP receptors in human-washed platelets (Keery and Lumley, 1988) with pA₂ values of ∼6.3 and ∼5.9 respectively (our calculation). Of further concern is the inhibitory effect of AH-6809 (3–10 µM) on platelet-activating factor (PAF)- and ADP-induced aggregation, which was attributed to inhibition of phosphodiesterase(s) (Keery and Lumley, 1988). In our hands, AH-6809 at 3–10 µM caused similar partial block of the contractile actions of phenylephrine (α₁), histamine (H₁), U-46619 (TP) and ONO-AE-248 (EP₃) on guinea pig aorta (R.L. Jones et al., 2009, submitted). AH-6809 has often been used at even higher concentrations, possibly because of its high water solubility. For example, 30–300 µM AH-6809 inhibited the contractile action of PGE₂ in pig large cerebral artery; EP₁ receptor involvement was inferred (Jadhav et al., 2004). However, the pA₂ corresponding to 30 µM AH-6809 is only 5.2 (our calculation) and the block was insurmountable at the higher concentrations. With more potent and selective EP₁ antagonists now available, it is time to relegate AH-6809 to its place in the historical development of prostanoid antagonists.

The EP₁ antagonists developed by Ono Pharmaceuticals (Figure 3, middle row) demonstrate an interesting progression from the TP antagonist ONO-11120 (see Figure 8; Katsura et al., 1983) to a related pinane analogue (ONO-NT-012) showing EP₁, FP and TP antagonism (and EP₃ agonism), to a bicyclo[2.2.2]octane analogue (ONO-8711) showing EP₁/EP₃ antagonism, and, finally, to the non-prostanoids ONO-8713 and compound 13 with high selectivity for the EP₁ receptor. ONO-8711, ONO-8713 and 13 have K_D values for mouse rc-EP₁ receptors of 1.7, 0.3 and 0.14 nM respectively (Watanabe et al., 1999; 2000; Naganawa et al., 2006). Small modifications to 13 can restore EP₃ antagonist affinity. In rat-cultured mesangial cells, ONO-8713 at 1 µM abolished induction of the transforming growth factor-β-fibronectin cascade elicited by PGE₂ under high-glucose conditions; a COX-2-PGE₂-EP₁ receptor drive was postulated to contribute to deleterious changes in diabetes (Makino et al., 2002). Ohnishi et al. (2001) showed that 10 µM ONO-8713 partially inhibited PGE₂-induced exocytosis in mucous cells from guinea pig antrum (IC₅₀∼1 µM). However, 17-phenyl PGE₂ was a very weak agonist. The authors postulated that EP₁ and EP₄ receptors co-operate to sustain the high exocytotic response to PGE₂. Norel et al. (2004) also showed that 10 µM ONO-8713 partially suppressed the contractile action of sulprostone on human pulmonary vein; functional EP₁ receptors were postulated, even though the selective EP₁ agonist ONO-DI-004 was a very weak agonist. Prostanoid receptor binding data alone do not guarantee functional selectivity at these high antagonist concentrations; inclusion of control agonists (both prostanoid and non-prostanoid) in the system under test is essential.

Other pharmacophores for potent EP₁ antagonism, with inactivity against cytochrome P450 enzymes and good penetration into the central nervous system being secondary goals (see later), have emerged within the last 10 years. The series reported by Merck (Ruel et al., 1999) contains a tricyclic system akin to that in the Searle series. Again EP₁/EP₃ selectivity can be readily modulated, as shown by replacement of the terminal phenyl group in compound 14 (K_i values for human EP₁ and EP₃ receptors = 10 and 4000 nM) by methyl (770 and 1000 nM). EP₁ antagonists with aryl groups attached to adjacent carbons of a 5-membered ring (cyclopentene, thiophene, pyrrole) have been reported by research groups at GlaxoSmithKline and Merck Frosst. GW-848687 has nanomolar affinity for the human EP₁ receptor, 30-fold lower affinity for the human TP receptor and >400 times lower affinity for other prostanoid receptors (Giblin et al., 2007). The picolinic acid residue in GW-848687 (ring A) is highly acidic (pKa ∼1.0) and there has been considerable SAR work by both research groups on modulating the acidity of this region. The Merck antagonist MF-266-1, with a m-C(CF₃)-(OH)₂ substituent (Figure 3), is a weak acid (pKa = 7.5 for first ionization) that retains high EP₁ affinity (Ducharme et al., 2005; Clark et al., 2008). A m-C(CF₃)₂-OH substituent in the Merck series and a p-C(CF₃)₂-OH substituent in the GlaxoSmithKline series resulted in lower EP₁ affinity (Ducharme et al., 2005; Hall et al., 2007a). Further modifications to ring A (e.g. m-NH(C=O)CH₂Ph/p-(C=O)NHCH(CH₃)Ph substituents) resulted in non-acidic relatives with high EP₁ potency. Compound 15 is also a non-acidic EP₁ antagonist (Hall et al., 2007b).

Therapeutic applications

The upper portion of Table 3 shows that parenteral administration of EP₁ antagonists of different chemical classes suppressed the allodynic/hyperalgesic signs of inflammation in the rat and mouse. Intraplantar injection of these antagonists also opposed pain-producing stimuli, although at the mouse EP₁ receptor AH-6809 had minimal affinity, which suggests an off-target effect. That said, Khasar et al. (1993) showed that PGE₂ and SC-19220 are only mutually antagonistic when they are both injected intradermally and not subcutaneously. More recently, attention has focused on the role of EP receptors in central nociception (see Svensson and Yaksh, 2002; Hefferan et al., 2003; Mebane et al., 2003). Minami et al. (2001) showed that PGE₂ could induce hyperalgesia in the mouse when injected intrathecally over the dose-range 0.00035–350 pmol. Deletion of the EP₃ receptor gene removed the more sensitive component of the hyperalgesia. In addition, the selective EP₃ agonist ONO-AE-248 induced hyperalgesia at relatively low doses. Surprisingly, the EP₁ knock-out mouse showed a hyperalgesic response in the hot-plate test, thereby confounding the role of EP₁ receptors in mediating hyperalgesia (Minami et al., 2001). EP₁ antagonists given intrathecally suppress flinching, and mechanical allodynia and hyperalgesia (Table 3), although in certain cases the doses required for these effects are high. For example, in the studies by Omote et al. (2002), 100 µg ONO-8711 was injected intrathecally in a volume of 10 µl, representing an injectate concentration of 23 mM! Even after dilution in the cerebrospinal fluid, EP₃ receptors are likely to be blocked as well.

Table 3.

Anti-nociceptive and anti-inflammatory activities of EP receptor antagonists

Route/species	Experimental model	Noxious stimulus/site	Antagonist (nominal receptor)	Antagonist dosing	Efficacy	Reference
Oral
Guinea pig	Joint pain (mono-osteoarthritis)	Iodoacetate/shoulder joint	MF-498 (EP4)	30 mg·kg−1	+++	Clark et al. (2008)
	Thermal hyperalgesia	L-902688a/paw	MF-498 (EP4)	0.1–30 mg·kg−1	+++	Clark et al. (2008)
Rat	Writhing response	Acetic acid/peritoneal cavity	SC-19220 (EP1)	50–300 mg·kg−1	++	Drower et al. (1987)
	Behavioural response	Formalin/paw	SC-19220 (EP1)	50–150 mg·kg−1	++	Drower et al. (1987)
	Chronic mechanical/thermal hyperalgesia/allodynia	Nerve constriction/sciatic	ONO-8711 (EP1)	10–100 mg·kg−1 per day	+++	Kawahara et al. (2001)
	Chronic hyperalgesia (weight-bearing)	Freund's adjuvant/knee joint	GW-848687 (EP1)	30 mg·kg−1 per day	+++	Giblin et al. (2007)
	Chronic arthritis (oedema/radiology)	Freund's adjuvant/paw	Compound 20 (EP4)	0.005 mg·kg−1 per day	+++	Burch et al. (2008)
	Local oedema	Freund's adjuvant/paw	MF-266-1 (EP1)	Each 0.008–20 mg·kg−1 daily for 10 days	0	Clark et al. (2008)
			MF-266-3 (EP3)		0
			MF-498 (EP4)		+++
	Mechanical hyperalgesia	Carrageenan/paw	CJ-042794 (EP4)	1–30 mg·kg−1	+++	Murase et al. (2008a)
	Local oedema	Freund's adjuvant/paw	CJ-042794 (EP4)	1–30 mg·kg−1	+++	Murase et al. (2008a)
	Mechanical hyperalgesia	Carrageenan/paw	CJ-023423 (EP4)	3–100 mg·kg−1	+++	Nakao et al. (2007)
	Thermal hyperalgesia	PGE2/paw	CJ-023423 (EP4)	1–29 mg·kg−1	+++	Nakao et al. (2007)
	Chronic inflammation (weight-bearing)	Freund's adjuvant/paw	CJ-023423 (EP4)	10–57 mg·kg−1	+++	Nakao et al. (2007)
	Chronic inflammation (oedema, histology)	Freund's adjuvant/paw	CJ-023423 (EP4)	29–96 mg·kg−1	+++	Okumura et al. (2008)
Mouse	Writhing response	Phenylbenzylquinone/peritoneal cavity	SC-51089 (EP1)	1–30 mg·kg−1	+++	Hallinan et al. (1993)
	Writhing response	Phenylbenzylquinone/peritoneal cavity	SC-51322 (EP1)	1–30 mg·kg−1	+++	Hallinan et al. (1994)
	Migration of Langerhans cells into lymph node	Fluorescein isothiocyanate/skin	ONO-AE3-208 (EP4)	10 mg·kg−1 twice daily	++	Kabashima et al. (2003)
	Ear swelling/histology	UVB/skin of ear	ONO-AE3-208 (EP4)	10 mg·kg−1	++	Kabashima et al. (2007)
Intravenous
Rat	Sensory discharge	Acetic acid/bladder lumen	ONO-8711 (EP1)	1–3 mg·kg−1	++	Ikeda et al. (2006)
	Visceromotor reflex	Distension/bladder	DG-041 (EP3)	10 mg·kg−1	+++	Su et al. (2008a)
Intraperitoneal
Rat	Mechanical hyperalgesia	Freund's adjuvant/paw	AH-23848 (EP4)	0.1–10 mg·kg−1	+++	Lin et al. (2006)
	Thermal hyperalgesia	Freund's adjuvant/paw	AH-23848 (EP4)	0.1–10 mg·kg−1	+++	Lin et al. (2006)
Mouse	Mechanical allodynia	Sulprostone/intrathecal	ZM-325802 (EP1)	0.03 µg·kg−1	+++	Gil et al. (2008)
Topical to skin
Mouse	Oedema	Chronic UVB	ONO-8713 (EP1)	96 nmol	++	Tober et al. (2006)
Intraplantar
Rat	Paw-withdrawal	PGE2/pawb	SC-19220 (EP1)	2.3 nmol	++	Khasar et al. (1993)
	Mechanical hyperalgesia	Incision wound/paw	ONO-8711 (EP1)	4.5–114 nmol	+++	Omote et al. (2001)
	Chronic mechanical hyperalgesia	Partial nerve transection/sciatic	SC-19220 (EP1)	2.3 nmol	+++	Syriatowicz et al. (1999)
	Chronic thermal hyperalgesia	Partial nerve transection/sciatic	SC-19220 (EP1)	2.3 nmol	++	Syriatowicz et al. (1999)
Mouse	Mechanical allodynia	PGE2/paw	AH-6809 (EP1)	100 nmol	0	Kassuya et al. (2007)
			L-826266 (EP3)	0.1–10 nmol	+++
			L-161982 (EP4)	10 nmol	0
		Carrageenan/paw	L-826266 (EP3)	10 nmol	++	Kassuya et al. (2007)
			L-161982 (EP4)	10 nmol	++
	Paw-licking	PGE2/paw	AH-6809 (EP1)	10–100 nmol	0	Kassuya et al. (2007)
			L-826266 (EP3)	3–30 nmol	++
			L-161982 (EP4)	1–30 nmol	+++
Intrathecal
Rat	Mechanical allodynia	Nerve ligation/L5 spinal root	SC-51322 (EP1)	22–220 nmol	++	Hefferan et al. (2003); O'Reilly and Loomis (2007)
	Mechanical hyperalgesia	Incision wound/foot	ONO-8711 (EP1)	114–228 nmol	++	Omote et al. (2002)
	Thermal hyperalgesia	Incision wound/paw	ONO-8711 (EP1)	114–128 nmol	0	Ikeda et al. (2006)
	Flinching (late phase)	Formalin/paw	SC-58109 (EP1)	65–650 nmol	++	Malmberg et al. (1994)
	Mechanical hyperalgesia (late phase)	Carrageenan/paw	ONO-8711 (EP1)	2.28–228 nmol	+++	Nakayama et al. (2002)
	Mechanical hyperalgesia (late phase)	PGE2/intrathecal	ONO-8711 (EP1)	0.11–0.45 nmol	+++	Nakayama et al. (2004)
	Visceromotor reflex	Distension/bladder	L-798106 (EP3)	10–300 nmol	+++	Su et al. (2008b)
			DG-041 (EP3)	10–100 nmol	+++	Su et al. (2008b)
Mouse	Thermal hyperalgesia	PGE2/intrathecal	AH-6809 (EP1)	0.29–2.9 nmol	+++	Uda et al. (1990)
	Writhing response	PGE2/intrathecal	AH-6809 (EP1)	0.29–2.9 nmol	+++	Uda et al. (1990)
	Mechanical allodynia	PGE2/intrathecal	AH-6809 (EP1)	4.2 nmol	0	Minami et al. (1995)
	Thermal hyperalgesia	PGE2/intrathecalc	AH-6809 (EP1)	4.2 nmol	+++	Minami et al. (1995)
	Mechanical allodynia	PGE2/intrathecal	ONO-NT-012 (?)d	3 × 10−5–0.03 nmol	+++	Minami et al. (1995)
	Thermal hyperalgesia	PGE2/intrathecal	ONO-NT-012 (?)	3 × 10−5–0.03 nmol	0	Minami et al. (1995)
Topical to spinal tissue
Rat	Mechanical allodynia	Bicuculline/L5–L6 spine	SC-51322 (EP1)	0.22–17.5 nmol	+++	Zhang et al. (2001)

Upper and lower panels relate to parenteral and topical administration of antagonist.

^aSelective EP₄ agonist.

^bIntradermal.

^cAbout 30-fold higher dose of PGE₂ required compared with mechanical allodynia.

^dSee EP₁ antagonist section for specificity; reported pA₂ of 9.96 in allodynia model is unlikely to be valid because antagonist concentrations in biophase are unknown.

Intravenous ONO-8711 also suppressed afferent nerve discharge to distension of the rat bladder sensitized with acetic acid (Ikeda et al., 2006). The inference from these studies was that PG(E₂) generated locally activates EP₁ receptors on peripheral sensory neurones. Again EP₁/EP₃ selectivity is critical, given that EP₃ agonists augment bradykinin-induced sensory nerve discharge (Kumazawa et al., 1996). Topical ONO-8713 (like celecoxib) inhibited UV-B-induced skin inflammation and tumour development in the mouse (Tober et al., 2006); the higher selectivity of ONO-8713 supports EP₁ receptor involvement. PGE₂ production, but not COX-1/COX-2 expression, was suppressed by ONO-8713; the mechanism is not clear.

Despite continuing synthesis of potent EP₁ antagonists with testing in analgesic/anti-inflammatory models, there has been little solid evidence of clinical efficacy. Sarkar et al. (2003) reported that ZD-6416, which is related to ZM-325802 (Figure 3; Shaw et al., 1999; Jenkins et al., 2001), inhibited upper oesophageal pain threshold to electrical stimulation in human volunteers. However, ZD-6416 does not appear to be particularly useful in the clinical setting (development profile obtained from Pharmaprojects; http://www.pharmaprojects.com).

While the preferred indication for EP₁ antagonists has been for pain, additional therapeutic uses in cancer, osteoporosis, arthritis, and neurodegenerative and renal disorders have been suggested. COX-2 inhibitors are of potential value in reducing colorectal adenomas which, in turn, has created interest in using prostanoid antagonists as an alternative. The cardiovascular risk associated with celecoxib in clinical trials involving colorectal adenoma prevention (Solomon et al., 2005) would intensify interest in using prostanoid antagonists for this indication. ONO-8711 inhibited formation of colonic crypts (Kawamori et al., 2001) and reduced the frequency of polyp formation in APC1309 mice (Watanabe et al., 1999; Kitamura et al., 2003b), and aberrant crypt foci in oxymetazaline-treated mice (Watanabe et al., 1999). Correspondingly, aberrant crypt foci were reduced by 60% in EP₁^−/− receptor mice (Watanabe et al., 1999). COX-2 up-regulation has also been considered a target for drug treatment of pathologies involving neurological injury and neurodegeneration. The downstream effects of COX-2 neurotoxicity have been reported to be EP₁ receptor-mediated (Kawano et al., 2006).

Blood pressure in the spontaneously hypertensive rat was reduced by SC-51322 (10 mg⁻¹·kg⁻¹·day⁻¹, gavage) (Guan et al., 2007). EP₁ receptors appear to regulate BP in the male but not the female mouse (Audoly et al., 1999; Stock et al., 2001). In a more detailed study, Guan et al. (2007) showed that SC-51322 and EP₁-receptor gene deletion blunted pressor responses to 17-phenyl PGE₂ and sulprostone, whereas the pressor response to the PGE₁ analogue MB-28767 (claimed to be ‘a pure EP₃ agonist’) was the same in EP₁^+/+ and EP₁^−/− mice. While we agree with the authors' contention that both EP₁ and EP₃ receptors contribute to the pressor effects, the utility of MB-28767 may be compromised by its moderate TP agonism (Lawrence and Jones, 1992) EP₁-receptor gene deletion also reduced the elevated BP and cardiac hypertrophy following a 4-week infusion of angiotensin (Ang) II in the mouse, and 1 µM SC-51322 markedly attenuated the contractile action of Ang II on the isolated pre-glomerular arteriole (Guan et al., 2007). How these findings for EP₁ receptors integrate with the modest hypertension and reduction of antihypertensive efficacy associated with NSAIDs and COX-2 inhibitors in the human setting (Johnson et al., 1994; Ishiguro et al., 2008) is not clear. Species and gender appear to be highly influential factors.

EP₂ receptor antagonists

Selective EP₂ receptor antagonists are essentially unavailable. This may be partly due to a reluctance to inhibit the potential anti-inflammatory actions of endogenous PGE₂ (see Teixeira et al., 1997) mediated via EP₂ receptors (Noguchi et al., 1999; Nataraj et al., 2001).

Woodward et al. (1995b) showed that AH-6809 has modest affinity for human rc-EP₂ receptors and antagonized PGE₂-induced activation of adenylyl cyclase with a pA₂ of about 6.5. Lower affinities were found for inhibition of PGE₂-induced relaxation of human bronchus (pA₂= 5.78; Norel et al., 1999) and guinea pig trachea (pA₂= 5.6, KJ Ong and RL Jones, unpubl. obs.) (Table 2). This is consistent with its utility for distinguishing EP₁ receptors from other EP subtypes in non-primate pharmacological preparations (Coleman et al., 1987; Eglen and Whiting, 1988; Lawrence et al. 1992). However, given the poor selectivity of AH-6809 in the low micromolar range, findings arising from its use in high concentration as an EP₂ antagonist (e.g. 100 µM in Aronoff et al., 2004) should be approached with caution. K_i values for rc-EP₂ receptors of about 1 µM have been found for some compounds in combinatorial studies (see Murase et al., 2008b); it should be possible to build on these observations.

EP₃ receptor antagonists

Development

EP₃ antagonism in a series of biaryl-acylsulphonamides was reported in 2002 by Merck (Gallant et al., 2002). The lead compound was an AT₁ receptor antagonist 16 (Figure 4) with a binding K_D of 7 µM for the human EP₃ receptor. In a combinatorial approach, reversal of the acylsulphonamide and insertion of an ethylene spacer yielded 17 with an EP₃K_D of 25 nM and minimal binding to other prostanoid receptors. A related analogue, L-798106, at 0.2 µM blocked the pre-synaptic (EP₃) inhibitory actions of sulprostone on guinea pig vas deferens and trachea with pA₂ values of 7.5 and 7.8 respectively (Clarke et al., 2004). L-798106 (0.2–1 µM) also caused parallel displacement of log concentration–response curves for contraction of rat femoral artery by sulprostone (in synergy with phenylephrine) (Hung et al., 2006). L-826266, a chloro analogue of L-798106, has been used in other studies despite the lack of detailed information on its basic pharmacology. Schlemper et al. (2005) showed that L-826266 at 30 µM inhibited both PGE₂- and bradykinin-induced relaxation of guinea pig trachea, and proposed that bradykinin induces de novo synthesis of PGE₂, which then activates EP₃ receptors. However, the specificity of L-826266 at this high concentration is unknown. An alternative explanation is that L-826266 modestly blocks EP₂ receptors thereby allowing the action of PGE₂ on the contractile EP₁ system in the trachea to dominate. Oliva et al. (2006) reported that L-826266 injected (as 0.1 µL of 1.25–5 mM solutions in 20% DMSO/PSS) into the periaqueductal grey matter of the mouse brain suppressed the late hyperalgesic response to intradermal formalin. However, the concentration of L-826266 at the site(s) of action is indeterminate and interpretation of the finding is difficult because similar high ‘doses’ of EP₁ and EP₄ antagonists and AH-6809 also suppressed the formalin response.

Figure 4.

EP₃ receptor antagonists. The AT₁ receptor antagonist, compound 16, is a lead molecule for the biaryl-ene-acyl-sulphonamide antagonists (pharmacophore in red). The left-hand portion of this pharmacophore corresponds to the cinnamic acid moiety in compounds 18 and 19 (see broken brackets). L-826266 is a chloro analogue of L-798106. The lower-middle brackets show modifications to the indole nucleus in the EP₃ antagonist series of DeCode Genetics.

DeCode Genetics has described a series of related molecules containing an indole nucleus, from which DG-041 (Figure 4) was selected for clinical investigation (Singh et al., 2009). DG-041 had a IC₅₀ of 8.1 nM in a EP₃/Ca²⁺ flux FLIPR assay; corresponding values in DP₁ and DP₂ assays were 131 and >10 000 nM respectively (Singh et al., 2009). SAR studies involved modifications to the terminal aryl moieties, together with the indole unit. Compounds with an inverted indole nucleus retain high EP₃ affinity (Zhou et al., 2009a,b;), as do indolones and hexahydro-indolones (O'Connell et al., 2009). Saturation of the α.β-double bond also produced highly potent EP₃ antagonists, while further saturation of the remaining double bond in the hexahydro-indolone residue resulted in marked reduction in affinity. Much of DeCode's work was directed towards improving water solubility: predicted n-octanol/water partition coefficients (ClogP) for DG-041, L-798106 and L-826266 are 6.6, 6.9 and 7.4 respectively (ChemAxon freeware). In this context, our recent studies on L-798106 and L-826266 have shown a slowly developing block of EP₃ agonist-induced contraction on guinea pig aorta, affording pA₂ values of 7.96 and 7.58 respectively after 3-h contact (Jones et al., 2008). As expected, highly potent antagonists such as BMY-180291 (TP, pA₂= 9.8) and doxepin (histamine H₁, pA₂= 9.6) also had slow onsets at low-nanomolar concentrations. However, the slow onsets of L-798106 and L-826266 may be related to their high lipophilicity rather than their (moderate) receptor affinity.

Merck-Frosst researchers have also identified EP₃ antagonism in ortho-substituted cinnamic acid derivatives, which correspond to the left-hand portions of the ene-acyl-sulphonamide antagonists shown in Figure 4. Compound 18 has a binding K_D for the human EP₃ receptor of 20 nM (Juteau et al., 2001), while compound 19 has the highest binding affinity (3 nM) and behaved as a pure antagonist in a human EP₃/adenylyl cyclase assay (pA₂= 8.22; Belley et al., 2005). Compound 19 is also highly lipophilic (ClogP = 8.29).

ONO-AE3-240 is reported to be a highly selective EP₃ antagonist (mouse EP₃/EP₁ selectivity ratio = 2500; Amano et al., 2003), but its structure has not been disclosed.

Therapeutic applications

Like most prostanoid receptors, the EP₃ receptor has been implicated in pain of various aetiologies. These include allodynia produced by HIV-1 glycoprotein gp 120 (Minami et al., 2003) and PGE₂ (Kassuya et al., 2007), acute herpetic pain (Takasaki et al., 2005), thermal hyperalgesia (Oka et al., 1994) and formalin-induced hyperalgesia (Oliva et al., 2006). A major role for EP₃ (and IP) receptors has been claimed in endotoxin-induced enhancement of pain perception (Ueno et al., 2001). Intravenous DG-041 also suppressed the visceromotor reflex to bladder distension in the rat (Su et al., 2008a); the authors defined this antagonist as brain non-penetrant, but no pharmacokinetic evidence was presented. Intrathecal application of L-798106 and DG-041 also produced a long-lasting suppression of the visceromotor reflex, while intracerebroventricular administration produced only a transient reduction (Su et al., 2008b). However, EP₃ agonists are known to exert gastrointestinal cytoprotection and attenuate gastric acid secretion in animal models (Bunce et al., 1990; Kunikata et al., 2002). Although it is widely assumed that there is a similar involvement of EP₃ receptors in man, an extensive search of the literature provides only circumstantial evidence for this. For example, while misoprostol is clinically useful in suppressing gastroduodenal erosion (see Hawkey, 2000), its selectivity is not high enough to infer the involvement of EP₃ receptors (EP₃∼ EP₂∼ EP₄ >> EP₁ for misoprostol-free acid; Abramovitz et al., 2000). A reduction in myocardial ischaemic damage was also achieved with EP₃ agonist treatment in the rat (Zacharowski et al., 1999). Thus, it is feasible that the therapeutic use of EP₃ antagonists will result in a side effect profile at least comparable to that of COX inhibitors. This possibility does not seem to have hindered development work on other uses of EP₃ antagonists as discussed below.

It has been suggested that endogenous PGE₂ activates EP₃ receptors on stromal cells surrounding a tumour causing the release of vascular endothelial growth factor, which then promotes angiogenesis and tumour growth (Amano et al., 2003). Injection of the EP₃ antagonist ONO-AE3-240 around sarcoma-180 tumours in the mouse markedly suppressed these effects in a manner similar to EP₃ receptor gene-deletion; ONO-8711 (EP₁ antagonist) and ONO-AE3-208 (EP₄ antagonist) were ineffective. In contrast, EP₃ agonists acting on all three mouse EP₃ receptor isoforms expressed in HEK-293 cells caused cell clustering and inhibited their proliferation via a G₁₂-RhoA pathway (Macias-Perez et al., 2008).

A further possible therapeutic application of an EP₃ antagonist is the treatment of pre-term labour. Based on the known activity of misoprostol (Sanchez-Ramoz et al., 1997) and sulprostone (Fruzzetti et al., 1988), it appears that EP₃ receptor stimulation produces cervical ripening, a critical event that precedes parturition. An EP₃ antagonist could be effectively combined with a tocolytic, such as an EP₂ agonist (Senior et al., 1993), to provide therapy for pre-term labour. The EP₃ receptor has been uniquely associated with febrile responses (Ushikubi et al., 1998). However, development of an EP₃ antagonist for treating fever seems unlikely, given that low-cost COX inhibitors are highly effective in reducing body temperature.

Finally, DG-041 has shown promise in the treatment of peripheral cardiovascular disease. Activation of EP₃ receptors on human platelets enhances aggregation induced by a variety of agents (in the presence of a TP antagonist); inhibition of adenylyl cyclase/priming of protein kinase C is thought to be the mechanism (Matthews and Jones, 1993; Vezza et al., 1993). In addition, there is increased bleeding tendency and increased susceptibility to thromboembolism in the EP₃ receptor knock-out mouse (Ma et al., 2001). DG-041 at 0.03–3 µM inhibited the enhancement by sulprostone of ADP- or collagen-induced aggregation in human platelet-rich plasma (PRP) (Heptinstall et al., 2008; Singh et al., 2009). The pA₂ value of 8.3 (our calculation from Heptinstall et al., 2008 data) is probably an underestimate of the affinity owing to plasma protein binding of DG-041. In the rat, DG-041 at 5 or 60 mg·kg⁻¹ (by gavage; co-administration with clopidogrel) inhibited enhancement of platelet aggregation induced by PGE₂ex vivo; there was no increase in bleeding time compared with clopidogrel alone (Singh et al., 2009). A similar profile was obtained with DG-041 in Phase I trials in healthy subjects (reported in Heptinstall et al., 2008).

EP₄-receptor antagonists

Development

The first EP₄ antagonist to be reported was AH-23848 (Coleman et al., 1994a), a close relative of the selective TP antagonist GR-32191 (Figure 8). Ligand binding studies on human rc-receptors indicated low EP selectivity for AH-23848, with Ki values for EP₁, EP₂, EP₃, EP₄ and IP receptors being 45, 50, 4.4, 14 and >100 µM respectively (Abramovitz et al., 2000). However, its selectivity in functional assays appears to be higher and it has been of considerable utility in differentiating the EP₂, EP₄ and IP agonist activities of prostanoid ligands. For example, Jones and Chan (2001) used AH-23848 at 30 µM to demonstrate that the PGI₂ analogues cicaprost and AFP-07 relax certain vascular preparations by activating both EP₄ and IP receptors. Moreover, Lai et al. (2008a) showed that pulmonary artery smooth muscle cells from monocrotaline-treated rats have a reduced IP receptor density and that iloprost-induced cAMP elevation is blocked by AH-23848, and therefore likely to be due to activation of EP₄ receptors.

AH-23848 has since been overtaken by more potent agents. Antagonists in the major group contain an acyl-sulphonamide unit (Figure 5) and show some similarity to the ene-acyl-sulphonamide EP₃ antagonists. Indeed, L-161982 is a methyl analogue of the lead molecule 16 for the Merck EP₃ antagonists. L-161982 was reported to have a pA₂ of about 8.5 in rc-EP₄ receptor – adenylyl cyclase assays (Machwate et al., 2001). In a rat native EP₄ assay, L-161982 inhibited PGE₂-induced cAMP accumulation with an IC₅₀ of about 30 nM, while forskolin-induced cAMP accumulation was unaffected by 10 µM L-161982. In similar functional assays, CJ-023423 afforded pA₂ values of 8.3 and 8.2 for human and rat rc-EP₄ receptors using the Schild protocol (Nakao et al., 2007). In binding assays, CJ-023423 showed weak affinity for human EP₂ receptors and PAF receptors, while interactions with other prostanoid receptors and a range of non-prostanoid receptors were minimal.

Figure 5.

EP₄ receptor antagonists. L-161982 is a methyl analogue of compound 16 in Figure 4. Acyl-sulphonamido residues are shown in red. The bonds indicated by asterisks in MF-498 are subject to oxidative/hydrolytic attack in vivo; the corresponding substituents in compound 20 prevent these transformations.

GW-627368 has pKi values for binding of 7.0 and 6.8 for human rc-EP₄ and TP receptors respectively; binding to other prostanoid receptors is minimal (Wilson et al., 2006). Correspondingly, a pA₂ of 7.9 was obtained for GW-627368 in a human rc-EP₄ receptor – adenylyl cyclase assay, with good evidence for competition. On human pulmonary vein, GW-627368 had a pA₂ of 7.06 against ONO-AE1-329 (Table 2; Foudi et al., 2008); its higher affinity against PGE₂ may have been due to opposing contractile activity. GW-627368 had a higher affinity on piglet saphenous vein (pA₂= 9.2) with a linear Schild plot up to a concentration ratio of about 60; further rightward shift of the agonist curve was insignificant owing to PGE₂ activating a less sensitive EP₂ relaxant system (Wilson et al., 2006). GW-627368 at 10 µM did not affect TP receptor-induced contraction under the same conditions. Results for rabbit saphenous vein were discrepant: Wilson et al. (2006) showed that 10 µM GW-627368 did not affect PGE₂-induced relaxation implying the presence of an EP₂ system only, while Jones and Chan (2005) found a right-shift of about 1 log unit with 1 µM GW-627368, consistent with the presence of EP₂ and EP₄ systems; GW-627368 did not antagonize relaxation induced by either the selective EP₂ agonist ONO-AEI-259 or the PGI₂ analogue, taprostene in the latter experiments. MF-498 (Clark et al., 2008), which is quite similar in structure to GW-627368, undergoes oxidative/hydrolytic metabolism at the three regions indicated in Figure 5; compound 20 was considerably more resistant to attack while retaining high EP₄ antagonist affinity (Burch et al., 2008). The most recent addition to this group is BGC-20-1531 (Maubach et al., 2009). It exhibits surmountable antagonism of PGE₂-induced relaxation of human cerebral and middle meningeal and dog carotid and middle meningeal arteries in vitro, while having no effect on PGE₂ (probably EP₃)-induced contraction of human coronary, pulmonary and renal arteries.

A second looser group of amide-containing EP₄ antagonists is represented by ONO-AE2-227, ONO-AE3-208 and CJ-042794 (Figure 5). Binding studies indicate that ONO-AE2-227 and ONO-AE3-208 retain considerable affinity for EP₃ receptors (Ki = 21 and 30 nM; Mutoh et al. 2002; Kabashima et al. 2003) and this must be borne in mind when interpreting in vivo data (see later). CJ-042794 behaved competitively in functional assays (Schild protocol) involving human and rat rc-EP₄ receptors (Murase et al., 2008a,b;). Its EP₄/EP₃ selectivity ratio is very high, but it still retains measurable affinity for human EP₂ receptors (pK_i= 6.2).

Therapeutic applications

Recent studies involving parenteral administration of several EP₄ antagonists (AH-23848, CJ-023423, CJ-042794, MF-498, ONO-AE3-208) have clearly demonstrated a major involvement of EP₄ receptors in small-animal models of inflammation (Table 3). Joint pain, mechanical and thermal hyperalgesia and oedema were markedly suppressed, often equivalent to the efficacy of selective COX-2 inhibitors such as rofecoxib. A peripheral site of action seems likely given that intraplantar administration of L-161982 suppressed carrageenan-induced mechanical allodynia in the mouse (Kassuya et al., 2007). Whether block of the EP₄ receptor alone produces an effective anti-inflammatory drug in man remains to be seen. Moreover, PGE₂ may have a protective role in inflammation. Takayama et al. (2002) showed that PGE₂ suppressed chemokine production stimulated by lipopolysaccharide in human macrophages; L-161982 at 100 nM blocked this action.

Prostanoid EP₄ receptors, indeed nearly all EP receptor subtypes, have been implicated as contributors to colon tumorigenesis caused by excessive production of PGE₂ (Fujino and Regan, 2003; Majima et al., 2003; Masataka et al., 2003; Mutoh et al., 2006). The selective EP₄ agonist ONO-AE1-329 (0.1–1 µM) increased colony formation in the human colon cancer cell line HCA-7 (Mutoh et al., 2002) and L-161982 blocked PGE₂-induced proliferation of HCA-7 cells (Cherukuri et al., 2007). In both the oxymetazoline model of aberrant crypt foci (putative preneoplastic lesions) and the Min mouse model of intestinal polyp development, the EP₄ antagonist ONO-AE2-227 produced about a 67% reduction in the appropriate scores (Mutoh et al., 2002). Aberrant crypt foci were similarly reduced in EP₄^−/− mice. In the APC1309 mouse, ONO-AE2-227 had a preferential effect on polyp size, while the EP₁ antagonist ONO-8711 had a more pronounced effect on polyp number, and a combination of antagonists behaved additively (Kitamura et al., 2003b). In the context of (tumour) vascularity, ONO-AE3-208 reduced IL-1β-induced angiogenesis in the mouse cornea at an oral dose of 1 mg·kg⁻¹ o.d. (Kuwano et al., 2004). Finally, pretreatment of mouse mammary tumour cells with AH-23848 and ONO-AE3-208 followed by washing and immediate injection into immunologically compatible mice reduced pulmonary tumour score (Fulton et al., 2006). These comprehensive studies provide encouragement that EP₄ receptor antagonists may provide a safer replacement for COX-2 inhibitors in treating colon cancer. As, however, all EP receptor subtypes have been implicated in colon cancer, a pan-PGE synthase inhibitor may be more effective in preventing the key cell proliferative and angiogenic events. Arguably, and somewhat paradoxically, the EP₄ receptor has also been shown to participate in the maintenance of intestinal homeostasis by preserving mucosal integrity. In both EP₄^−/− mice and wild-type mice treated with an EP₄ antagonist (ONO-AE3-208), susceptibility to the development of colitis was reported in a model of inflammatory bowel disease (Kabashima et al., 2003; Narumiya, 2003).

There has been interest in using EP₄ agonists for their anabolic effects on bone (Raisz, 2006). The EP₄ antagonist L-161982 at a dose of 10 mg·kg⁻¹·day⁻¹ reversed bone formation induced by PGE₂ in the rat, without affecting its diarrhoeal action (Machwate et al., 2001). Also, human mesenchymal stem cells in culture secreted PGE₂ via COX-2, which was associated with production of bone morphogenetic protein-2 (BMP-2), a factor that stimulates differentiation of precursor mesenchymal cells into mature bone. The selective COX-2 inhibitor, NS-398, and the EP₄ antagonist, ONO-AE3-208, (concentration not specified) suppressed BMP-2 expression (Arikawa et al., 2004). These findings identify bone loss as a potential side effect of EP₄ antagonists.

In vitro, EP₄ vasodilator systems typically exhibit high sensitivity to PGE₂ and this property is reflected in the role of PGE₂ in maintaining the open state of the ductus arteriosus during gestation in human and animal species. Strong expression of EP₄ (and IP) receptors is found in ductus tissue from the newborn infant and child (Leonhardt et al., 2003). Towards the end of gestation, smooth muscle cells within the ductus migrate to the endothelial lining where they form intimal cushions. Declining PG levels at birth result in ductus constriction, thereby bringing the intimal cushions into close contact and effecting permanent closure. Intimal cushion formation is also driven by EP₄ receptor activation (Yokoyama et al., 2006). The COX inhibitor, indomethacin, is commonly used to treat failure of ductus closure, but its efficacy is poor in a substantial proportion of cases, perhaps related to suppression of intimal cushion development (see Ivey and Srivastava, 2006). AH-23848 blocked PGE₂-induced relaxation of the rabbit and sheep isolated ductus arteriosus preparations (Smith et al., 1994; Bouayad et al., 2001), while ONO-AE3-208 constricted the ductus of fetal and neonatal rats in vivo (Momma et al., 2005). Whether an EP₄ antagonist would be better than a COX inhibitor in treating patent ductus in premature infants is not yet clear.

A second potential role for an EP₄ vasodilator system is the genesis of vascular headache in migraine. EP₄ antagonists block PGE₂-induced relaxation of human-isolated middle cerebral artery (Davis et al., 2004; Maubach et al., 2009) and the picture has been enlarged to include the interaction of endogenous PGE₂ with calcitonin gene-related peptide release from trigeminal nerves (Maubach et al., 2009). The use of AH-23848 and L-161982 also has provided evidence for a role of EP₄ receptors in substance P release by stretching of the renal pelvic wall (volume expansion) leading to activation of renal sensory afferents and subsequent diuresis/natriuresis (reno-renal reflex) (Kopp et al., 2004).

FP receptor and prostamide receptor antagonists

Development

A variety of FP receptor antagonists have been reported in the past, but none has stood up to rigorous analysis. Thus, PGF_2α-1-dimethylamine and PGF_2α-1-dimethylamide (Maddox et al., 1978; Stinger et al., 1982) exhibited no meaningful FP antagonist or agonist activities (Sharif et al., 2000; Woodward et al., 2008). Phloretin (Kitanaka et al., 1993) was non-selective and very weak in blocking fluprostenol-mediated phosphoinoside turnover in rat A7r5 vascular smooth muscle cells (Sharif et al., 2000). Similarly glibenclamide (Delaey and Van de Voorde, 1995) was a weak, non-selective prostanoid antagonist (Sharif et al., 2000).

AL-3138 and AL-8810, both PGF_2α analogues (Figure 6), have received a degree of acceptance as FP antagonists (Griffin et al., 1999; Sharif et al., 2000). AL-8810 proved useful in studying PGF_2α-mediated up-regulation of the orphan nuclear receptor Nur 77 (Liang et al., 2004). However, further studies reveal that they are neither potent nor selective. For example, AL-8810 appears to block TP receptors (Hutchinson et al., 2003), an effect that has been confirmed in human rc-TP receptor stable transfectants (A.H. Krauss and D.F. Woodward, unpublished). In addition, both agents are FP partial agonists. This can be seen in the original report of AL-8810 on phosphoinositide turnover (Griffin et al., 1999) and is corroborated by the finding of myogenic activity in the mouse uterus (Hutchinson et al., 2003). AL-8810 induced a more pronounced Ca²⁺ signal in human rc-FP stable transfectants (Y. Liang and D.F. Woodward, unpublished) and was a full agonist in the cat isolated iris preparation (Woodward et al., 2007); the latter action was not blocked by a prostamide antagonist (Woodward et al., 2007; see later), so it presumably reflects FP receptor agonism.

Figure 6.

FP receptor and prostamide receptor antagonists. The natural ligand PGF_2α and its 1-ethanolamide derivative, prostamide F_2α, are shown in the box. AL-3138 and AL-8810 are FP partial agonists in many systems; α= corresponding side-chain in PGF_2α. The THG analogues are peptides: amide (CO-NH) residues are shown as red bars. The AGN analogues (upper right) are prostamide receptor antagonists; C1-amide residues are shown in blue.

A series of octapeptides (THG-131 derivatives, Figure 6) have been claimed to possess selective FP antagonist activity (Chemtob and Peri, 2006; Peri et al., 2006). In particular, THG-113.31 at 1 µM markedly inhibited PGF_2α-induced contraction of pig retinal blood vessels, while having minimal effect on contraction to 17-phenyl PGE₂, U-46619, phenylephrine, Ang II and endothelin-1; the inhibition of PGF_2α contraction was insurmountable. In addition, THG-131.31 inhibited the associated phosphoinositide hydrolysis with an IC₅₀ of about 30 nM. However, THG-131.31 competed poorly with [³H]-PGF_2α for binding to human rc-FP receptors (∼13% at 10 µM). Several other studies have cast doubt on the utility of THG-131.31. At 10 µM, THG-113.31 showed fairly weak antagonism of the contractility of longitudinal and circular strips of sheep myometrium induced by PGF_2α; there was no effect on PGE₂ contractions (Hirst et al., 2005). Also, 10 µM THG-113.31 had no effect on PGF_2α-induced contraction of human pregnant myometrium, while inhibiting spontaneous and oxytocin-induced contractions at much lower concentrations (Friel et al., 2005). Finally, Doheny et al. (2007) showed that THG-113.31 at 10–50 µM enhanced BK_Ca channel opening in isolated myocytes from human uterus, an effect that was reversed by iberiotoxin. Several chemically simpler peptidomimetics in the THG series showed more potent block of PGF_2α-induced contraction than THG-113.31 (Figure 6, Table 2; Peri et al., 2006); it would be of interest to have full pharmacological profiles.

Currently, the most convincing FP antagonist is the non-prostanoid AS-604872 (Cirillo et al., 2007). It has K_i values of 35, 158 and 323 nM for human, rat and mouse rc-FP receptors and its selectivity was 20-fold for EP₂ receptors and greater than 300-fold for other prostanoid receptors. AS-604872 showed no agonist activity in a human FP receptor – inositol phosphate assay and had an IC₅₀ of 47 nM against PGF_2α. In vivo, AS-604872 (1–30 mg·kg⁻¹, i.v.) inhibited PGF_2α-induced uterine contraction in the non-pregnant rat; inhibition of oxytocin-induced contraction was slight.

Prostamide research originated from studies on neutral PGF_2α analogues and notably bimatoprost (17-phenyl PGF_2α-1-ethylamide), an effective anti-glaucoma drug (Woodward et al., 2003; 2004;). As previously mentioned, neutral PGF_2α analogues show only weak affinity for FP receptors (Maddox et al., 1978; Schaaf and Hess, 1979), a profile subsequently confirmed for −OH, −OCH₃ and −CON(R)₂ C1-substitutes (Woodward et al., 2000; 2008; Matias et al., 2004). The pharmacology of bimatoprost was similar, with no meaningful activity at FP receptors but pronounced activity in certain preparations such as cat lung strip, cat iris, rabbit uterus, and human ciliary smooth muscle cells (Liang et al., 2003; Woodward et al., 2003b; Matias et al. 2004; Chen et al. 2005). The pharmacology of bimatoprost appeared indistinguishable from that of PGF_2α-1-amides, but this could not be rationalized until the discovery that anandamide (arachidonic acid-1-ethanolamide) was a substrate for COX-2 (Yu et al., 1997).

PGE₂-1-ethanolamide (prostamide E₂) was the first prostamide to be discovered and was identified as the major product following addition of anandamide to rc-COX-2 or cells expressing COX-2, but not COX-1 (Yu et al., 1997). Subsequently, more extensive studies demonstrated that COX-2 oxidizes anandamide to endoperoxide intermediates, which are converted by specific PG synthases to the various prostamides (Kozak et al., 2002; Koda et al., 2004; Yang et al., 2005; Moriuchi et al., 2008). Despite being different terminal biosynthetic products, the evidence to date suggests that prostamides D₂, E₂ and F_2α interact with a single receptor to exert their effects. Prostamide F_2α and its analogues are, however, about 10 times more potent than prostamides D₂ and E₂ (Woodward et al., 2007).

Initial pharmacological characterization of the prostamides, for example prostamide E₂ (Ross et al., 2002), relied on agonist studies. In the context of PGF analogues, FP agonists (17-phenyl PGF_2α and PGF_2α) and bimatoprost produced Ca²⁺ signals in entirely different cells in a cat iris smooth muscle cell preparation (Spada et al., 2005), suggesting the existence of a receptor with a distinct preference for prostamide F agonists. The receptor structure appears to involve heterodimerization of the wild-type and alternative mRNA splicing variants of the FP receptor, both encoded by PTGFR, the FP receptor gene (Liang et al., 2008). This FP/alt-FP co-expression is analogous to the isoprostane binding site which is formed as a result of IP/TP receptor heterodimerization (Wilson et al., 2004).

The strategy for discovering a prostamide antagonist was to identify antagonists in cluster 2 of the prostanoid receptor evolutionary tree (Narumiya et al., 1999) and then form corresponding C1-amides. Cluster 2 contains the TP receptor, whose stable agonists have been a starting point for antagonists at other receptors (DP₁, see Figure 1; EP₁, see Figure 3). Based on the oxabicycloheptane analogue BMS-180291 (Figure 8) (Webb et al., 1993), two prototype prostamide antagonists AGN-204396 (Figure 6) and AGN-204397 were identified (Woodward et al., 2007; 2008;). These agents showed good prostamide F/FP selectivity, but were of low affinity (pA₂∼5.5) and also blocked TP receptors. Substituting oxygen at C3 dramatically enhanced prostamide affinity, by as much as 100-fold for AGN-211334 and AGN-211335 (Wan et al., 2007; Liang et al., 2008; Woodward et al., 2008). AGN-211334 and AGN-211335 potently inhibited prostamide F_2α and bimatoprost responses in cat iris preparations, but did not alter responses to FP agonists (Wan et al., 2007; Liang et al., 2008). AGN-211334 blocked the increase in conventional aqueous humour outflow produced by bimatoprost in the human perfused anterior segment preparation (Wan et al., 2007), thereby demonstrating that the effects of bimatoprost in the human eye are prostamide receptor-mediated; previously, it has been suggested that bimatoprost's activity is dependent on deamidation to the FP-active free acid in ocular tissue (Camras et al., 2004). AGN-211335 blocked the secondary Ca²⁺ wave, myosin light chain phosphorylation, and Cyr 61 up-regulation induced by bimatoprost in the FP/alt-FP prostamide system (Liang et al., 2008). These second-generation prostamide antagonists are likely to be sufficiently potent for in vivo studies.

Therapeutic applications

Prevention of pre-term labour (tocolysis) is probably the only therapeutic modality where a FP antagonist may be of value. This is a serious and unmet medical need given that premature birth accounts for 60–80% of perinatal deaths (Goldenberg, 2002). Parturition is prevented in FP^−/− mice (Sugimoto et al., 1997) confirming a significant role for PGF_2α (see Challis et al., 2002 for a review of PG involvement). THG-113.31 is tocolytic in the pregnant sheep (Hirst et al., 2005) and may be useful for delaying pre-term birth (Olson, 2005). AS-604872 suppressed spontaneous uterine contractions in late-term pregnant rats and delayed preterm birth caused by mefipristone in pregnant mice; AS-604872 appeared to be more effective than the β₂-adrenoceptor agonist ritodrine (Chollet et al., 2007). It is not yet clear whether these findings will translate into an effective drug in human reproduction.

Prostanoid FP receptors have been implicated in cardiomyocyte hypertrophy (Pönicke et al., 2000; Xu et al., 2008) and cancer (Fujino and Regan, 2001) and may play a role in regulating water and solute transport based on in situ hybridization studies in the mouse kidney (Saito et al., 2003). The utility of a selective FP antagonist is, however, unclear.

The therapeutic utility of prostamide antagonists is similarly uncertain. Increased anandamide levels may result in the formation of prostamides as major products in inflammation and infection (Glass et al., 2005). Testing of AGN 211334, or a close congener, in relevant animal models should provide evidence for or against a functional role of prostamides.

IP receptor antagonists

Development

IP receptor antagonists are a recent development based on the potential role of PGI₂ in mediating pain (Bley et al., 1998). In an extensive series of studies, two structurally distinct classes of selective IP antagonist emerged from focused chemical library screening and synthetic chemistry (Bley et al., 2006) (Figure 7). The 2-(phenylamino)-imidazoline series is represented by RO-1138452 and compound 21 (Clark et al., 2004; Keitz et al., 2004), while the N-substituted phenylalanine series has a ‘traditional’ carboxylate in the phenylalanine residue and is typified by RO-3244019 (Fitch et al., 2004) and its difluoro analogue RO-3244794 (Bley et al., 2006).

Figure 7.

IP receptor antagonists. The natural ligand PGI₂ (prostacyclin) is shown in the box. 2-(Phenylamino)-imidazoline moieties are shown in blue and phenylalanine residues in red (S-configuration in compound 24). RO-3244794 is a difluoro analogue of RO-3244019.

RO-1138452 has high affinity for human native (platelet) and rc-IP receptors, with pK_i values of 9.3 and 8.7 respectively in studies utilizing [³H]-iloprost, although it also displays considerable affinity for PAF (7.9) and imidazoline (8.3) receptors (Bley et al., 2006). In functional studies in platelets (Jones et al., 2006), pA₂ values were lower than K_i values obtained in radioligand competition studies with platelet membrane preparations and for inhibition of carbacyclin-induced cAMP formation in cells over-expressing rc-IP receptors (Bley et al., 2006); the difference was attributed to protein binding in studies involving PRP. The pA₂ values obtained in isolated blood vessel preparations (human pulmonary artery 8.20, guinea pig aorta 8.39 and rabbit mesenteric artery 8.12) were intermediate between values obtained in PRP and plasma membranes/cells in buffer (Jones et al., 2006). The slight suppression of the cicaprost maximum response seen with higher concentrations of RO-1138452 was attributed to functional antagonism emanating from the (albeit weak) EP₃ agonist action of cicaprost. However, RO-1138452 displayed an insurmountable antagonist profile in studies of chemokine release from human airway epithelial cells using taprostene (see Table 1) as IP agonist (Ayer et al., 2008). Moreover, RO-1138452 inhibition of taprostene-induced cAMP response element-dependent transcription was not reversed over a 20-h ‘washout’ period. These data could not be ascribed to covalent receptor inactivation, allosterism or a state of antagonist hemi-equilibrium and may be due to a pseudo-irreversible interaction with the IP receptor (Ayer et al., 2008).

The N-substituted phenylalanines 22 and 23 (Nakae et al., 2005) are weak IP antagonists, while the greater potency of 24 is associated with as S-configuration in the phenylalanine residue (Brescia et al., 2007). RO-3244794 has no meaningful activity at EP₁, EP₃, EP₄ and TP receptors; no data were provided for DP₁, DP₂, EP₂ or FP receptors (Bley et al., 2006). While RO-1138452 did not block EP₂ receptor-mediated relaxation in guinea pig aorta (pA₂ < 6.0; Jones et al., 2006), RO-3244794 had a pA₂ of 6.92 for the human rc-EP₂ receptor expressed in HEK-293 cells (S.M. Hill and M.A. Giembycz, unpubl. data). Compound 24 did not bind to human EP₂ and EP₄ receptors (Brescia et al., 2007). Strictly speaking, the missing data related to these compounds (e.g. RO-1138452 at DP₂ and FP receptors; Jones et al., 2006) need to be addressed. In the context of systemic drug design, RO-3244794 has much greater oral bioavailability than RO-1138452 (51% vs. 0.7% in the rat) (Bley et al., 2006).

Therapeutic applications

Pain has been the initial focus for IP antagonists based on IP receptor agonist effects and IP receptor distribution (reviewed by Bley et al., 1998). Altered pain perception and inflammation were observed in IP^−/− mice (Murata et al., 1997). Subsequently, IP antagonists were shown to reduce pain responses in models where prostanoids have been implicated. These include acetic acid-induced abdominal constriction, mechanical hyperalgesia produced by carrageenan and pain associated with models of osteoarthritis and inflammatory arthritis (Bley et al., 2006; Pulichino et al., 2006). Such analgesic effects are accompanied by anti-inflammatory properties, which is invariably the case. In a collagen-induced arthritis model in mice, 21 produced effective inhibition when administered as a pretreatment, but was inactive when given after the initiation of the arthritis (Pulichino et al., 2006). RO-3244794 and indomethacin were equi-effective in reducing carrageenan-induced rat paw oedema and more effective than rofecoxib in inhibiting the foot weight distribution change associated with intra-articular injection of monoiodoacetate (Bley et al., 2006). Ostensibly, these results imply that IP receptors provide a singular target that would result in drugs that are at least as effective as NSAIDs and COX-2 inhibitors. The role of EP receptors in pain/inflammation has already been discussed. Despite a satisfactory preclinical profile, strong circumstantial evidence suggests that EP₁ antagonists were a failure in clinical trials. This, in turn, tends to cast doubt over the clinical prognosis for IP antagonists. Given the number of prostanoids that may be released locally and the even greater number of target receptors available, it may be regarded as counter-intuitive to propose that inflammation can be attributed to only one type of prostanoid receptor.

A role for PGI₂ in mediating the sensitized release of substance P from rat dorsal root ganglion neurons has been suggested based on studies with 23 (Nakae et al., 2005). Related to this is the potential use of IP antagonists for bladder disorders. In rat models of bladder function, RO-3244019 dose-dependently decreased bladder contraction frequency and increased micturition threshold and voiding interval (Cefalu et al., 2007). RO-3244019 was also effective in treating neurogenic detrusor overactivity arising from spinal cord injury in the rat (Khera et al., 2007). Again, these results are of uncertain predictive value for clinical success, because indomethacin exhibited good activity in these models. The IP antagonist, BAY-73-1449 (Figure 7), was effective in acutely reducing shunt vessel blood flow in a rat model of portal hypertension (Bexis et al., 2008); full details of its pharmacology are unavailable.

Finally, side effects originating from perturbation of the TXA_2/PGI₂ balance could be greater with IP antagonists than with COX-2 inhibitors, including hypertension, stroke, myocardial infarction and atherosclerosis. Nephrotoxicity and K⁺ and Na⁺ retention also may occur (Nasrallah and Hébert, 2005), because prostacyclin synthase (PGIS) and COX-2 deficiencies produce similar renal toxicity. This suggests a protective role for PGI₂, but it is important to note that no marked renal phenotype occurs in TP^−/− or IP^−/− mice (Breyer and Breyer, 2000; Yokoyama et al., 2002; Nasrallah and Hébert, 2005).

TP receptor antagonists

Development

Elucidation of the structures of PGH₂ and TXA₂ (Figure 8) was soon followed by a range of carba/thia bicyclic analogues (see Wilson and Jones, 1985). One aim was to produce a chemically stable TP agonist (e.g. U-46619, STA₂); another was to investigate the potential for TP receptor antagonism. Thus, pinane-TXA₂ (PTA₂) was reported to block both constriction of cat coronary artery and aggregation of human platelets elicited by U-46619 (Nicolaou et al., 1979). However, this analogue often behaves as a partial agonist, inducing a 60% maximal activation in some TP systems (Jones et al., 1982; Tymkewycz et al., 1991). More useful TP antagonists emerged from additional modification of the ω-chain. The pinane analogue ONO-11120 (Figure 8) is a true antagonist (Katsura et al., 1983), while EP-045 (Jones et al., 1982) and EP-092 (Armstrong et al., 1985) were developed from the PG endoperoxide analogue, 9,11-etheno PGH₂, also a partial agonist (Jones et al., 1982). SQ-29548 (Ogletree et al., 1985) and BMS-180291 (Ogletree et al., 1993) have a hybrid bicyclic system (oxabicyclo[2.2.1]heptane) and differ from the natural agonists in the cis-orientation of the α and ω-chains. While GR-32191 (Lumley et al., 1989) and ICI-192605 (Brewster et al., 1988) still retain a prostanoid skeleton, other TP antagonists do not, including L-655240 (related to indomethacin; Hall et al., 1987), BM-13505 (daltroban; Yanagisawa et al., 1987), KW-3635 (Karasawa et al., 1991a,b;), AA-2414 (seratrodast; Ashida et al., 1989) and even the K_ATP-channel blocker glibenclamide (Cocks et al., 1990). The attractively simple chemistry involved in the synthesis of daltroban has spurred the development of higher-potency, non-prostanoid antagonists containing a (p-halo)-benzenesulphonylaminomethyl residue at a critical distance from the carboxylate, for example, Z-335 (Tanaka et al., 1998) and S-18886 (terutroban; Cimetière et al., 1998). A similar residue in prostanoid molecules such as S-145 (domitroban; Mihara et al., 1989), I-SAP (Naka et al., 1992) (Figure 8), ONO-NT-126 (p-bromo; Nakahata et al., 1990) and oxa- and thia-bicyclo[3.1.0]hexane derivatives (Kamata et al., 1990) also confers high TP affinity.

Differences in antagonist affinity constants between platelet and vascular smooth systems have stimulated much debate about the existence of TP receptor subtypes (Mais et al., 1985; 1988; Swayne et al., 1988; Morinelli et al., 1989; Masuda et al., 1991; Tymkewycz et al., 1991; Folger et al., 1992). Species heterology, the difficulty of distinguishing partial agonism from functional antagonism, and incomplete equilibration of high-affinity antagonists have been confounding factors. For example, antagonism of U-46619 by BMS-180291 on human platelets was surmountable for the shape-change response, but insurmountable for aggregation; the rate of aggregation was also slowed (Ogletree et al., 1993). Similar profiles had been reported previously for EP-092 (Armstrong et al., 1985) and GR-32191 (Lumley et al., 1989) on human platelets, and for EP-169 and AH-23848 on human and rat platelets, but not on rabbit platelets where their affinities are lower (Tymkewycz et al., 1991). However, against the slow-acting high-affinity TP agonist EP-171 (Jones et al., 1989), GR-32191 did not alter the aggregation rate (Lumley et al., 1989). It is likely that slow dissociation of a high-affinity antagonist from the TP receptor retards U-46619 occupancy in the early stage of the aggregation response thereby favouring the disaggregation process and insurmountability; in contrast, shape change, which does not fade, allows a true measure of the equilibrium state. In pig platelets, longer pre-incubation times were required for S-145 than SQ-29548 owing to the smaller association rate constant of the former antagonist (Mihara et al., 1989). BMS-180291 also had a Schild slope greater than unity on guinea pig aorta (Zhang et al., 1996). However, in the low nanomolar range BMS-180291 requires up to 2 h to reach steady state on the aorta (Jones et al., 2008) and the non-ideality may simply reflect retarded diffusion of a high-affinity ligand (pA₂= 9.8, Table 2) through the extracellular space.

Radioligand binding studies of TP receptors coincided with these functional measurements. Two saturable binding sites were identified on human platelets using [³H]-9,11-epoxymethano PGH₂, the first radioligand developed for the TP receptor (Armstrong et al., 1983; Pollock et al., 1984). Binding to the more abundant site was displaced by EP-045 at concentrations similar to those required to inhibit [³H]-phosphatidate formation, Ca²⁺ elevation and aggregation induced by the TP agonist. A less abundant higher-affinity binding site was also identified, but it did not show the characteristic preference for a 15(S) configuration in the natural prostanoid ω-chain. The presence of high- and low-affinity binding sites for agonist (but not antagonist) ligands was also evident using [³H]-trimetoquinol (Ahn et al., 1988) and [³H]-SQ-29548 (Hedberg et al., 1988). Further studies showed that the high-affinity site was associated with the platelet shape change (and increase in cytosolic Ca²⁺), while the lower-affinity site was associated with aggregation [and activation of phospholipase C (PLC)] (Dorn, 1989; Takahara et al., 1990). [³H]-GR-32191 played an important role in the elucidation of these relationships by binding reversibly to the ‘shape change site’ and irreversibly to the ‘aggregation site’ (Takahara et al., 1990). It is difficult to explain this irreversibility given that GR-32191 does not obviously contain a chemically reactive group (Figure 8). Detailed studies have shown that 30-min exposure of human platelets to GR-32191 resulted in about 50% loss of binding sites for either [³H]-GR-32191 or [³H]-SQ-29548, while neither SQ-29548 nor BM-13177 affected B_max. It was speculated that GR-32191 binds to internalized TP receptors (Armstrong et al., 1993); the zwitterionic nature of GR-32191 at neutral pH may be relevant. A light-activated, covalent-bonding TP antagonist, azido-BSP, also discriminated these platelet sites by blocking aggregation but not shape change induced by U-46619 (Zehender et al., 1988). The subsequent identification of a second TP receptor isoform (TPβ) from a human umbilical vein endothelial cDNA library (Raychowdhury et al., 1994) and the detection of mRNA for the α and β isoforms in human platelets (Hirata et al., 1996) would appear to complete the argument. However, these isoforms, which arise by alternative gene splicing and differ only in their cytoplasmic tails, do not show the ligand discrimination typical of the high- and low-affinity binding sites. Finally, significant expression of the TPα isoform only was found in human platelets (Habib et al., 1999).

The seventh transmembrane domain (TM-7), which is strictly conserved in all the TP receptors characterized to date, is critical to TP agonist and antagonist function. Point mutations in this domain in the human TP receptor severely suppressed the binding of SQ-29548 (Funk et al., 1993b). Chimeric substitutions of the human TP receptor with the corresponding TM 1, 2 and 4 from rat resulted in modest suppression of SQ-29548 binding, lesser suppression of I-BOP (TP agonist) binding, and a poor correlation between the data sets (Dorn et al., 1997).

Several different inhibitory properties have been combined with specific TP antagonism, either by chance or deliberately. For example, TP antagonists based on PGH₂ with diphenylmethyl-oxime (e.g. EP-157, Figure 8) or diphenylmethyl-azine residues in the ω-chain were found to activate IP receptors in platelet and vascular systems (Armstrong et al., 1986; 1989; Jones et al., 1993). A diaryl-hetero(cyclic) moiety is critical to the IP agonism (Jones et al., 1993). Related compounds lacking a prostanoid ring system (e.g. octimibate) showed similar profiles (Merritt et al., 1991a,b;), with BMY-45778 being the most potent of a large series of non-prostanoid prostacyclin mimetics synthesized by Bristol-Myers Squibb (Meanwell et al., 1994; Seiler et al., 1997). Accurate estimation of the TP antagonist affinities of many of these compounds is difficult owing to their high lipophilicity, slow onset/offset (Jones et al., 1997) and, in certain instances, an ability to inhibit (non-prostanoid) Gq-PLC-driven responses (Chow et al., 2001).

Combining TP receptor antagonism with thromboxane synthase (TXS) inhibition has been an extensively investigated strategy, with the aim of balancing antagonistic/inhibitory activities several hours after dosing. TXS inhibition is expected to divert PGH₂ to PGD₂ and PGI₂ (Vermylen et al., 1981; Smith, 1982), which both inhibit human platelet activation, and usually requires the presence of either a (N)-imidazole as found in dazoxiben (Randall et al., 1981) or a pyridin-3-yl group as in ridogrel (Hoet et al., 1990) and the related CV-4151 (isbogrel, Figure 8, Imura et al., 1988). ZD-1542 (Brownlie et al., 1993) is a pyridin-3-yl derivative related to ICI-192605, while GR-83783 (Campbell et al., 1991a), a relative of GR-32191, has a 4-(pyridin-3-yl)-phenyl moiety (Figure 8). The (N)-benzimidazole KW-3635 did not inhibit cow platelet TXS at 100 µM (Miki et al., 1992). Several groups have successfully expanded the biaryl region of ridogrel (Cozzi et al., 1994) or combined sulotroban/daltroban moieties with ridogrel/isbogrel moieties (Figure 8), resulting in GR-108774 (Campbell et al., 1991b), CGS-22652 (Bhagwat et al., 1993) and compound 36 in Soyka et al. (1993). In addition, Zeneca have used various tethers to connect the whole or part of the ICI-192605 nucleus to either a dazoxiben or an isbogrel nucleus (Figure 8) (Ackerley et al., 1995). BM-531 and BM-573 are combined TP antagonist/TXS inhibitors lacking a carboxylic acid group (Dognéet al., 2001; Rolin et al., 2001).

Finally, YM-158 has similar high affinity for TP and cys-LT receptors in guinea pig trachea (Arakida et al., 1998) (see later).

Therapeutic applications

The discovery of thromboxanes was Nobel prize-winning research. Thromboxane A₂ is undoubtedly important in regulating cardiovascular homeostasis. Its biosynthesis by platelets and other tissues and highly potent actions on platelets and blood vessels provided a strong impetus for the development of TP antagonists (Patrono, 1990; Patscheke, 1990; Davis-Bruno and Halushka, 1994). However, these agents have not been a success to date. Clear evidence of benefit was not established in early studies (Misra, 1994), but economics was also a major factor. Low-dose aspirin, which targets platelet TXA₂ synthesis by irreversibly acetylating COX-1, has proven benefits (Patrignani et al., 1982; Fitzgerald et al., 1983; Reilly and Fitzgerald, 1987). More potent TP antagonists, with superior pharmacokinetic profiles, were developed, but these also do not appear successful at the commercial level. The reasons are not entirely clear. At the pharmacological level, a TP antagonist would appear preferable to low-dose aspirin, especially in the light of recent events surrounding COX-2 inhibitors.

1. The clinical side effects associated with COX-2 inhibitor therapy reveal the TXA_2/PGI₂ balance as more delicate to perturbation and more important than was previously believed.

2. Low-dose aspirin does not always have the required TXS/PGIS selectivity (Knapp et al., 1988; Patscheke, 1990).

3. Isoprostanes are alternative endogenous human TP receptor agonists and hence blockable by TP antagonists: their non-enzymatic formation would not be affected by aspirin, other COX inhibitors or TXS inhibitors (Kawikova et al., 1996; Gardan et al., 2000; Janssen et al., 2001).

4. TP antagonists exhibit cardio-protective effects that are not shared by aspirin (Gomoll and Ogletree, 1994; Grover et al., 1994).

Given these considerations, a potent, highly selective TP antagonist may be worth revisiting in large-scale clinical trials on cardiovascular disease. As oxidative stress and resultant formation of isoprostanes is now linked to atherogenesis (Dognéet al., 2005), an initially unappreciated dimension to the TXA₂/PGI₂ balance and cardiovascular risk is made manifest. Specifically, the clinical outcomes would be related to stroke, heart attack and angina. Thromboxane A₂ has also been implicated in hypertension occurring in pregnancy and TP antagonists (/TXS inhibitors) have been proposed as treatments for pre-eclampsia (Keith et al., 1993; Dognéet al., 2006).

In addition to the vasculature, TXA₂ potently stimulates other smooth muscles to contract. Thus, TP antagonists have been proposed as potential therapeutic modalities for asthma. Seratrodast (Figure 8) has received marketing approval for treatment of asthma in Japan (see Rolin et al., 2006). Ramatroban and seratrodast have also been evaluated in Phase III clinical trials in the USA (Dognéet al., 2002). The therapeutic rationale is to ameliorate the marked bronchoconstriction produced by TXA₂ and its involvement in bronchial hyperesponsiveness. However, TXA₂ and other prostanoids that activate TP receptors (e.g. PGD₂) are not the only powerful bronchoconstrictors generated in asthma and it is unlikely that a TP antagonist alone would be adequate therapy in most patients. In this context, SQ-29548 and the cys-LT antagonist montelukast acting alone inconsistently inhibited contraction of human lung slices in culture challenged with antigen; a combination of the antagonists was much more effective, while an H₁ antagonist was ineffective (Wohlsen et al., 2003). A combined TP/cys-LT antagonist (e.g. YM-158) may have greater clinical efficacy.

Activation of EP₃, FP, TP and possibly EP₁ receptors causes contraction of the pregnant human myometrium, while DP₁, EP₂ and IP receptors mediate relaxation (Senior et al., 1993). Given this scenario, the potential of selective TP antagonists for treating labour-associated disorders, such as pre-term labour, would seem low. Recent studies on human myometrial specimens obtained at parturition provide a new perspective. It appears that there is a marked alteration in prostanoid receptor functional dynamics at term, prior to and following the onset of labour (Fischer et al., 2008). Comparing the effects of U-46619, PGE₂ and PGF_2α, a loss of FP but not TP responsiveness was apparent following the onset of labour (Fischer et al., 2008). As PGE₂ produces a net inhibition of myogenic activity, this leaves TXA₂ as the only COX-derived product capable of exerting a contractile effect during labour. The role of TXA₂ in human parturition may be greater than previously envisaged.

The involvement of TXA₂ in inflammatory bowel disease may be significant (Rampton and Collins, 1993). The aetiology of ulcerative colitis and Crohns' disease is not fully understood, but initiating factors in pre-disposed individuals elicit severe and prolonged inflammation of the gut mucosa. Early studies with ridogrel and picotamide, combined TP antagonist/TXS inhibitors, claimed some clinical improvement associated with reduced TXA₂ (TXB₂) release (Rampton and Collins, 1993). Given the multifactorial nature of these diseases and the number of other eicosanoids purported to be involved, pronounced beneficial effects of TP antagonists alone seem unlikely.

Evidence is available that activation of the TP receptor is implicated in chronic persistent cough (CPC). This is a detrimental and debilitating condition that serves no obvious function (French et al., 2002), afflicts 9–33% of the population in Europe and North America and for which no satisfactory treatment is currently available (see Chung and Pavord, 2008). Thus, the mechanisms underlying CPC and the identification of new anti-tussive agents present a grossly neglected and unmet clinical need.

In humans, PGs have high tussive potency suggesting that they may be released locally in a variety of respiratory diseases where cough is a characteristic symptom. In subjects with asthma, both indomethacin and ozagrel (OKY-046; a TXS inhibitor) increase the threshold for cough when compared with placebo, suggesting that TXA₂ may be one of the cyclo-oxygenase products that sensitize the cough reflex (Fujimura et al., 1995). Indomethacin and ozagrel also attenuate cough in subjects with hypertension treated with angiotensin-converting enzyme inhibitors (Fogari et al., 1992; Malini et al., 1997; Umemura et al., 1997). Thus, the potential application for TP antagonists in alleviating CPC is clear. Indeed, in an animal model of asthma-related cough, the TP antagonist, seratrodast (Figure 8) significantly suppressed cough induced by capsaicin (Xiang et al., 2002). Moreover, in 16 patients with stable chronic bronchitis, seratrodast (80 mg b.i.d. for 28 days) significantly increased the threshold for capsaicin-induced cough (Ishiura et al., 2003). Finally, ramatroban is reported to attenuate cough in subjects with cough variant asthma (Kitamura et al., 2003), which could be due to blockade of both TP and DP₂ receptors (Gardiner and Browne, 1984).

A role for TXA₂ in cancer of the female breast and genital tract has also been suggested. Findings are reminiscent of the TXA_2/PGI₂ balance in the cardiovascular system: TXA₂ promotes tumour growth and metastasis, while PGI₂ exerts a protective effect by maintaining vascular and platelet homeostasis (Honn and Meyer, 1981; Nigam et al., 1992). A significant role in ovarian, cervical and other gynaecological cancers may exist (Nigam et al., 1992), but there remains insufficient evidence to assign a pivotal role.

Thromboxanes have also been implicated in a number of other conditions, including glomerulonephritis (Patrono, 1990; Wardle, 1999), allergic conjunctivitis (Woodward et al., 1990b), allergic rhinitis (Misra, 1994), diabetes (Dognéet al., 2006), septic shock and pulmonary embolism (Ghuysen et al., 2005). In summary, while numerous potential uses have been proposed over several years, clinical and economic success has been modest to date despite the availability of extremely potent and long-acting TP antagonists. What are the future therapeutic prospects? An argument could be made for superiority over low-dose aspirin. In order to establish this, extensive clinical trials would be essential. The bigger concern for the pharmaceutical industry would be ‘how much better would a TP antagonist perform compared with low-dose aspirin?’ If the difference is small, economics will dictate the fate of excellent TP antagonists. Perhaps the brightest new star in the galaxy of therapeutic options is the treatment of pre-term labour, given that TXA₂ is now known to be the major prostanoid involved in parturition. The only potential side effect that may occur with TP antagonist therapy is possible enhancement of hypersensitivity reactions (Narumiya, 2003). This would not be a major burden under most circumstances.

Therapeutic future of prostanoid receptor antagonists

The future for selective prostanoid antagonists does not appear to provide an encouraging scenario. Potent and selective antagonists for every prostanoid receptor are now available, with the exception of the EP₂ subtype. Some selective antagonists have received regulatory approval as drugs but these instances tend to be few and not worldwide. In this category are DP₂ and TP antagonists. Drug approval filings/development have been discontinued on numerous TP antagonists, EP₁ antagonists and, most recently, an EP₄ antagonist. Arguments in favour of more potent and irreversible antagonists have been put forward but, in the global scheme of drug research, the current drugs seem adequate at the very least.

DP₂ antagonists apart, of those drugs designed as selective for a single prostanoid receptor, the TP blockers still appear to be the best prospect. This would require revisiting cardiovascular studies and daring to directly compare with low-dose aspirin. The use of the TP antagonists for treating pre-term labour and even perhaps primary and secondary dysmenorrhoea should be contemplated in the light recent evidence (Fischer et al., 2008). This is, however, a rather narrow spectrum of utility and does not fulfil three decades of high expectation.

Straightforward consideration of the current situation does not favour the widespread therapeutic utility of highly selective prostanoid receptor antagonists. Stated simply, where COX inhibitors are clinically effective, selective antagonists are not. This is not really astonishing. There are five major prostanoids biosynthesized by COX and additional active metabolites (e.g. 13,14-dihydro-15-oxo PGD₂, 19(R)-OH PGE₂). The isoprostanes are formed by non-enzymatic oxidation of arachidonic acid (Morrow et al., 1990; 1994;). Finally, there are the PG-ethanolamides (prostamides) and PG-glyceryl esters, which are COX-2 products of the endocannabinoids anandamide and 2-arachidonyl glycerol. Expectations that one prostanoid and one dedicated receptor play a dominant, all-important role in most disease processes are lofty. It is unlikely that this all distils down to one important receptor activity in most diseases: a receptor widely controlling synergy or solely mediating a critical pathophysiological event.

Individual PGs and their receptors may play compensatory, fail-safe roles. Individual prostanoids may act sequentially to initiate and sustain disease states. They may subserve complementary roles. The role of DP₁ and DP₂ receptors in allergy, notably allergic rhinitis, provides an excellent example of two receptors behaving in a complementary manner to initiate and maintain the disease state. The combination of DP_1/DP₂ antagonism in a single molecule appears a very promising therapeutic approach (Pettipher, 2008) and one that appears feasible. Equally so, combining TP and EP₃ antagonistic properties to prevent both platelet activation and vasoconstriction in cardiovascular disease states appears to be an achievable goal. Prostanoid-based drugs for pre-term labour may require even more versatility, with ideal therapy perhaps embodying. the following attributes in a single molecule: (i) TP antagonism, (ii) EP₃ antagonism to block cervical ripening and (iii) EP₂ agonism to provide a tocolytic effect. Such a molecule would be a formidable medicinal chemistry challenge. Further challenges relate to diseases where COX inhibitors are widely and successfully used, because the ideal spectrum of antagonist properties may not be entirely clear. If the promise of prostanoid-based therapeutics is ever to be fulfilled, the role of prostaglandins may need to be carefully thought-out on a disease-by-disease basis.

Glossary

Abbreviations:

Ang II

angiotensin II

BMP

bone morphogenetic protein

BP

blood pressure

ClogP

predicted n-octanol/water partition coefficient

COX

cyclo-oxygenase

CPC

chronic persistent cough

PG

prostaglandin

EP^+/+/EP^−/−

system involving wild-type/homozygous gene-deleted EP receptor

FA

free acid

NSAID

non-steroidal anti-inflammatory agent

PAF

platelet-activating factor

PGIS

prostacyclin synthase

PLC

phospholipase C

PRP

platelet-rich plasma

rc

recombinant

SAR

structure–activity relationships

TM

transmembrane

TXA₂

thromboxane A₂

TXS

thromboxane synthase

Conflicts of interest

The authors affirm no conflicts of interest.