Main

Prostaglandin D2 (PGD2) is an acidic lipid mediator that is derived from arachidonic acid by the sequential action of cyclooxygenase(s) (COX) and PGD2 synthase(s). The COX(s) convert arachidonic acid in a two-step process to first PGG2 and then PGH2. These unstable endoperoxide intermediates are converted to PGD2 by either the haematopoietic or lipocalin PGD2 synthase (Fig. 1). PGD2 is produced in the brain where it might be involved in the regulation of sleep1 and other central nervous system (CNS) activities, which includes pain perception2. In peripheral tissues, the richest cellular source of PGD2 is the mast cell — PGD2 is produced by immunoglobulin E (IgE)-activated mast cells at levels of around 50 ng per 1 × 106 cells3,4. Other leukocyte populations, such as dendritic cells5 and T helper 2 (TH2) cells6, also produce PGD2, but at levels far lower than those produced by mast cells. Although non-mast-cell sources of PGD2 might have important functions in some settings, such as in dendritic cell and TH2 lymphocyte interactions, measurement of PGD2 and its metabolites has been proposed as selective markers of mast-cell activation in clinical asthma7,8,9. The G-protein-coupled receptors (GPCRs) — CRTH2 (also known as DP2), DP1 and TP — that have an important role in mediating the biological effects of PGD2 are detailed in Box 1, and the known natural and synthetic ligands for these receptors are detailed in Table 1.

Figure 1: Pathways of PGD2 metabolism.
figure 1

On cellular activation arachidonic acid is released from membrane phospholipids and converted to the cyclic endoperoxide prostaglandin G2 (PGG2) by the action of cyclooxygenase 1 (COX1) or COX2. The peroxidase activity of the same enzymes leads to the formation of PGH2, which can be converted to a number of biologically active prostaglandins by discrete synthase enzymes that show tissue-specific distribution and might demonstrate preferential coupling to either COX1 or COX2. PGD2 production is dependent on the action of either the lipocalin PGD2 synthase or the haematopoietic PGD2 synthase. Haematopoietic PGD2 synthase is present in mast cells, T helper 2 (TH2) cells and other leukocytes, and it is thought to be responsible for the bulk of PGD2 production during allergic responses. PGD2 is rapidly metabolized (with a half-life of 1.5 min in blood), and the main products that have been detected in vivo are Δ12PGJ2 and 9α11βPGF2. The precise pathway of PGD2 metabolism has not been fully elucidated but it is possible that 13,14-dihydro-15-keto-PGD2 (DK-PDG2), Δ12PGD2 and 15-deoxy-Δ12 14PGJ2 might be formed locally, which is of interest as these compounds, similar to Δ12PGJ2, retain significant chemoattractant receptor-homologous molecule expressed on TH2 cells (CRTH2)-agonist activity.

Table 1 Comparison of the key biological and pharmacological properties of CRTH2, DP1 and TP

It is well established that the presence of an allergen triggers the production of PGD2 in sensitized individuals. In asthmatics, a bronchial allergen challenge leads to the rapid production of PGD2, which can be detected in the bronchoalveolar lavage fluid within minutes, reaching biologically active levels at least 150-fold higher than pre-allergen levels10. Local antigen challenge also stimulates PGD2 production in the nasal mucosa of patients with allergic rhinitis11 and in the skin of patients with atopic dermatitis12. Several lines of evidence support the view that mast cells are the principal sources of PGD2 at sites of allergic inflammation. Cell fractionation studies have shown that PGD2 is produced predominantly by mast cells13, and, in chopped human lung parenchyma, mast-cell activation is a requirement for PGD2 generation14. Also, although both mast cells and basophils produce histamine, basophils do not produce appreciable quantities of PGD2 (Ref. 7). This is relevant when discerning the principal cellular source of PGD2, as PGD2 is produced during the early response to allergen only, unlike histamine, which is produced during the early and late phases15. These data are therefore consistent with the view that mast cells are responsible for the bulk of PGD2 production in an allergic setting. However, PGD2 that is produced by mast cells might provide an essential link between the early phase and the late-phase allergic response by initiating cellular processes that lead to the recruitment and activation of TH2 lymphocytes and eosinophils with associated pathophysiological effects. The mechanisms by which mast-cell-derived PGD2 orchestrates these effects are the focus of this Review.

PGD 2 metabolism

The proposed pathways of PGD2 metabolism are shown in Fig. 1. It is of particular interest that a number of known and putative PGD2 metabolites such as 13,14-dihydro-15-keto-PGD2 (DK-PGD2)16, Δ12PGD2 (Ref. 17), Δ12PGJ2 (Ref. 18), 15-deoxy-Δ12 14PGD2 (Ref. 19), 15-deoxy-Δ12 14PGJ2 (Ref. 19) and 9α11βPGF2 (Ref. 20) retain activity on CRTH2, but are less active on DP1. This implies that the metabolism of PGD2 might be important in defining the pattern of leukocyte activation in allergic diseases and that some metabolites of PGD2 (particularly Δ12PGJ2 (Ref. 18)) might have systemic as well as local effects. These findings also illustrate that CRTH2 is a more promiscuous receptor than DP1 with respect to the diversity of ligands that it interacts with.

11-dehydro-thromboxane B2 is also a weak CRTH2 agonist21, an activity that might be relevant to situations in which this thromboxane metabolite is formed in high concentrations, such as at sites of platelet aggregation.

The diverse activities of PGD2 in the control of local blood flow, bronchial airway calibre and leukocyte function are mediated by high affinity interactions with the GPCRs CRTH2, DP1 and TP. In addition to the effects that are mediated by the GPCRs discussed in this Review, PGD2 can be metabolized to Δ12PGJ2 and 15-deoxy-Δ12 13PGJ2, which have effects relevant to the resolution of inflammation, including the inhibition of cytokine production and the induction of apoptosis. These effects are mediated by intracellular actions that are mediated by both peroxisome proliferator-activated receptor-γ (PPARγ)-dependent22 and PPARγ-independent mechanisms, the latter of which involves the inhibition of nuclear factor-κB (NFκB) activation23. Generally, concentrations of ligand in the micromolar range have been used to observe the intracellular effects of these PGD2 metabolites. As these observations are of pharmacological interest, the physiological significance of these findings has been questioned because of the high concentrations that are needed to induce these effects relative to the concentrations of the PGD2 metabolites that are likely to be formed in vivo24.

Effects of PGD 2 relevant to allergic diseases

Having established that mast-cell-derived PGD2 is produced in high concentrations in response to an allergen challenge, it is interesting to note that PGD2 can mimic a number of key features of allergic diseases. Studies in preclinical species have observed the following features when PGD2 is applied to in vivo preparations, or its overproduction is engineered by genetic manipulation:

  • Vasodilatation leading to erythema (flare) and potentiation of oedema (wheal).

  • Recruitment of eosinophils and TH2 lymphocytes.

  • Modulation of TH2-cytokine production.

  • Bronchoconstriction.

The capacity of PGD2 to cause or potentiate inflammatory responses has been appreciated for a long time. Injection of PGD2 into human skin has been shown to produce a long lasting erythema25,26, to potentiate the effects of other mediators on induration and leukocyte infiltration in human skin26 and to enhance oedema formation in rat skin25. It is most likely that these effects of PGD2, like those of other vasodilator prostaglandins, are due to an increased blood flow to the inflamed lesion27 and are, therefore, most likely to be mediated predominantly by the DP1 receptor. This is based on studies showing that the DP1 agonist BW245C relaxes vascular smooth muscle preparations and the selective DP1 antagonist BWA868C blocks the vasodilator effect of PGD2 (Ref. 28). Although these observations make it clear that DP1 mediates the vascular effects of PGD2, the capacity of PGD2 to promote the cellular changes associated with inflammation is not due to an action on DP1. Studies by Woodward et al.29 in the rabbit eye and guinea pig eye showed that BW245C could mimic the ocular hypotensive effect of PGD2, but it did not share the ability of PGD2 to promote inflammatory changes, particularly eosinophil infiltration. This was the first hint that some of the pro-inflammatory effects of PGD2 occur independently of DP1 activation. Furthermore, this study also ruled out a role for TP in mediating the effect of PGD2 as the TP agonist 9α11βPGF2 was also ineffective in promoting eosinophil accumulation in the eye. The selective effect of PGD2 on eosinophil activation in vivo is consistent with earlier work in which infusion of PGD2 into the dog led to a transient but marked reduction in circulating eosinophils (as a result of activation in the blood) without alterations in the neutrophil count30. The ability of PGD2 to promote the selective accumulation of eosinophils in vivo has also been shown in dog airways31, whereas overexpression of human lipocalin-type PGD2 synthase in transgenic mice leads to enhanced eosinophil recruitment and exaggerated TH2-cytokine production in response to an antigen32.

It has been appreciated for some time that PGD2 can cause potent activation of isolated eosinophils33. However, it was not until recently that the receptor that mediates the chemotactic effect of PGD2 was identified and characterized. In 2001 it was reported that the receptor mediating the activation of eosinophils in response to PGD2 was a novel receptor unrelated to DP1 (Ref. 34). Simultaneously, it was discovered that PGD2 was a ligand for CRTH2 (Ref. 16), which was originally described as GPR44, an orphan chemoattractant-like receptor expressed by TH2 cells. In a series of elegant experiments, Nagata, Hirai and co-workers showed that CRTH2 is selectively expressed by TH2 lymphocytes, eosinophils and basophils35,36, and that PGD2 stimulates chemotaxis of these cell types through a CRTH2-dependent mechanism16. Interestingly, in these studies, the key observation was made that PGD2 is the dominant CRTH2 agonist produced by activated mast cells16.

PGD2 is also a potent bronchoconstrictor and this effect is mediated by the TP receptor37, possibly by the 11-ketoreductase-dependent formation of the TP agonist 9α11βPGF2 (Refs 38, 39). Although exposure to PGD2 leads to bronchoconstriction, DP1 receptors are expressed in bronchial smooth muscle and the selective DP1 agonist BW245C has bronchodilator properties40. Overall, however, the predominant effect of PGD2 in the airways is to cause bronchoconstriction, which indicates that DP1-mediated bronchodilatation is masked by a more powerful bronchoconstrictor effect that is mediated by TP.

Role of DP 1 and CRTH2 in allergic inflammation

It is now emerging that DP1 and CRTH2 have crucial, and complementary, roles in mediating aspects of the allergic response based on pharmacological studies with receptor-selective ligands and studies with genetically deficient mice.

DP1 mediates vasodilatation and polarization of TH2 cells. PGD2 is rapidly produced in response to an inhaled allergen in the nose11 and in the lung10. Haematopoietic PGD2 synthase, DP1 and CRTH2 all show increased expression in the nasal mucosa of patients with allergic rhinitis41. Therefore, it is likely that the activation of DP1 might contribute to the increased nasal blood flow observed in a response to an allergen as a consequence of the DP1-mediated vascular effects that are described above. Intravenous administration of PGD2 to human volunteers leads to intense facial flushing and nasal congestion without having overt effects on the systemic blood pressure or lung function42. This demonstrates that PGD2 has the potential to cause a pattern of vasodilatation that is relevant to the pathology of allergic rhinitis.

Furthermore, insufflation of PGD2 has a greater effect than histamine on nasal congestion in human volunteers43. Treatment with ramatroban (Baynas; Kyorin, Nippon Shinyaku, Bayer Yakuhin), which is now known to be a CRTH2 antagonist (see next section), has no effect on nasal blockage induced by inhalation of PGD2 (Ref. 44), which provides indirect evidence that this effect is DP1-mediated (as ramatroban inhibits both CRTH2 and TP receptors but not DP1). The selective DP1 antagonist S-5751 inhibited early phase increases in nasal pressure in response to antigen in sensitized guinea pigs while the H1 antagonist terfenadine was without effect45. Therefore, at sites of mast-cell-mediated allergic inflammation it is apparent that DP1 activation has an important role in increasing local blood flow leading to hyperaemia and potentiation of oedema formation, which, in the case of allergic rhinitis, contributes to nasal congestion. Although there are currently no clinical data available on the activity of DP1 antagonists in allergic diseases it is of interest that niacin-induced skin flushing is reduced by the selective DP1 antagonist MK-0524 in human volunteers46, which demonstrates that endogenous production of PGD2 leads to DP1-mediated vasodilatation in humans.

Several studies suggest that DP1 might have additional properties relevant to allergic diseases. In mice genetically deficient in DP1 receptors, airway inflammation in response to low-dose antigen is diminished. Compared with wild-type mice, there was a substantial reduction in TH2-cytokine production, eosinophil infiltration, mucus production and airway hyperresponsiveness in DP1-null mice40. Although the effects of DP1 deficiency can be overcome with higher doses of antigen, this study provides compelling evidence that DP1 activation has a crucial role in regulating TH2-dependent airway inflammation. The effect of DP1 deficiency is mirrored by the effect of pharmacological blockade in preclinical models. The DP1 antagonist S-5751 is partially effective in reducing eosinophil and macrophage infiltration into the airways of sensitized guinea pigs exposed to an antigen45. At face value it is difficult to reconcile these effects with the known pharmacology of DP1.

How can these anti-allergic effects of DP1 deficiency or receptor blockade be consistent with the lack of DP1-mediated pro-inflammatory effects on TH2 lymphocytes or eosinophils and the bronchodilator properties mediated by DP1, which would be expected to be bronchoprotective? Paradoxically, the answer could lie in the anti-inflammatory effects afforded by DP1-receptor activation. DP1 is expressed by dendritic cells47,48,49and TH1 cells50, and the DP1 agonist BW245C inhibits cytokine production by these cell types47,48,49. By contrast, DP1 is not expressed at significant levels by TH2 cells and does not significantly affect TH2-cell function51. Therefore, it is proposed that DP1-mediated inhibition of TH1-inducing cytokines such as interleukin-12 (IL12)48 will favour T-cell development towards the TH2 phenotype49 and, therefore, DP1 blockade should halt the polarization of T cells. Of course, the corollary of that is that there is a hypothetical possibility that DP1 blockade might exacerbate TH1-dependent diseases such as rheumatoid arthritis and related autoimmune disorders. In this context it is worth noting that endogenous PGD2 has been proposed to have an important role in downregulating the pro-inflammatory response associated with either irritant or TH1-dependent stimuli. This effect has been observed in preclinical models of non-allergic inflammation, including carrageenan-induced pleurisy52, trinitrobenzene-sulphonic-acid-induced colitis53 and TH1-dependent delayed-type hypersensitivity reactions54, although additional mechanisms involving PPARγ activation (see above) have been proposed in some cases.

CRTH2 mediates direct activation of TH2 lymphocytes and eosinophils. Investigations into the role of CRTH2 in allergic responses have been facilitated over the past few years by the discovery that ramatroban is an effective antagonist of this receptor55. Ramatroban was originally identified as a TP antagonist56 but it is now realized that this drug binds to the CRTH2 receptor with moderate affinity (it has approximately tenfold less potent activity on CRTH2 compared with TP) and blocks responses to the selective CRTH2 agonist DK-PGD2 in vitro and in vivo55,57. Studies with ramatroban suggest that CRTH2 is important in mediating eosinophil accumulation in a number of tissues in response to an allergic challenge including guinea pig nasal mucosa58, mouse airways59 and in mouse skin during contact hypersensitivity reactions60. These effects of ramatroban on eosinophil recruitment are unlikely to be mediated by TP antagonism as eosinophils do not express TP and selective TP antagonists do not influence eosinophil function55,57.

It is also postulated that CRTH2 has an important role in TH2-lymphocyte recruitment and activation. PGD2 has the remarkable property of stimulating the production of the cytokines IL4, IL5 and IL13 by human TH2 cells in the absence of co-stimulation51. This effect is inhibited by ramatroban and is mimicked by the selective CRTH2 agonist DK-PGD2, but not by the DP1 agonist BW245C51. Furthermore, activation of TH2 lymphocytes in response to supernatants from immunologically activated human mast cells is mediated by PGD2 through an action on CRTH2 (Ref. 61). Inhibition of PGD2 production by diclofenac reduces the production of TH2-stimulatory activity by mast cells without affecting the responsiveness of TH2 cells to PGD2 or other stimuli, and CRTH2 blockade leads to a concordant reduction in the response of TH2 cells to mast-cell supernatants61. In vivo, selective CRTH2 agonists enhance allergic responses — injection of PGD2 leads to the exacerbation of allergic responses in the lungs and skin of mice, an effect mimicked by the selective CRTH2 agonist DK-PGD2 (Ref. 62). Taken together, these data suggest that CRTH2 might have an important role in the recruitment of TH2 lymphocytes and other leukocytes to sites of allergic inflammation and, perhaps more importantly, in driving TH2-cytokine production.

Studies of allergic responses in mice in which CRTH2 has been genetically deleted have so far produced apparently conflicting results. Satoh et al.63 have reported that allergic skin inflammation and IgE production were significantly diminished in CRTH2-knockout mice, which is consistent with the pharmacological studies with human cells discussed above. By contrast, and rather surprisingly, Chevalier et al.64 found that allergic airway inflammation, in particular eosinophil accumulation and IL5 production, was enhanced in mice deficient in CRTH2. This apparent discrepancy might be related to the unusual receptor-expression pattern of CRTH2 in mice. Unlike humans in which CRTH2 is expressed selectively by TH2 cells and not TH1 cells, in mice CRTH2 is expressed in equal abundance by both TH1 and TH2 cells64,65. Therefore, it is possible that CRTH2-mediated effects on TH1 cells might dominate over those on TH2 cells depending on the setting and variation in experimental protocol, particularly the immunization procedure. This possibility is supported by the observation that that production of IL2 was dramatically reduced in CRTH2-knockout mice64. It is also well established that variations in experimental protocols in mice can lead to profound differences in the nature of the allergic response. For example, airway inflammation can be either mast-cell dependent or mast-cell independent depending on the immunization procedure66. Therefore, it seems that results from the study of allergic responses in mice should be interpreted with caution, particularly those related to CRTH2.

DP1 and CRTH2 act together to drive polarization and activation of TH2 lymphocytes. Based on the proposed roles of DP1 and CRTH2 described above, it is important to note that DP1 and CRTH2, rather than having opposing actions, work together to promote TH2-dependent allergic responses. Activation of DP1 promotes an environment in which polarization of TH2 cells can occur, whereas CRTH2 mediates their recruitment and activation to produce cytokines. Recently, it was discovered that CRTH2 is expressed by central memory TH2 cells in addition to TH2 effector cells, and that thymic stromal lymphopoietin (TSLP)-activated dendritic cells promote the maintenance and polarization of CRTH2+CD4+ TH2 central memory cells67. These central memory TH2 cells express high levels of PGD2 synthase (also cystatin and charcot–leydon crystal protein). In this setting, PGD2 that is produced by dendritic cells and TH2 central memory cells might be important in driving polarized TH2 function through complementary actions of both DP1 and CRTH2. It is worth noting that TSLP levels in the asthmatic airways correlate with disease severity so this finding could be of clinical relevance68. Paradoxically, the DP1 agonist BW245C has been reported to suppress allergic responses in the skin when administered topically at high concentrations69. This effect was associated with the suppression of langerhans cell migration from the epidermis, the suppression of TH2-cytokine production and the enhancement of TH1-cytokine production. Further work is needed to reconcile this effect with other studies that indicate that DP1 activation promotes TH2 polarization, but, as in the case of CRTH2, the apparent differences might be related to the receptor-expression pattern differences in mice, the variations in the immunization procedure or the nonspecific actions of BW245C that occur at high doses. A scheme describing the proposed complementary roles of DP1 and CRTH2 in polarization, recruitment and activation of TH2 lymphocytes is shown in Fig. 2.

Figure 2: Proposed scheme describing the role of mast-cell-derived PGD2 in the polarization and activation of TH2 lymphocytes, effects achieved through combined action on DP1 and CRTH2 receptors.
figure 2

Antigen-induced crosslinking of cell-surface immunoglobulin E (IgE) on mast cells leads to the rapid and abundant production of prostaglandin D2 (PGD2). PGD2 through interaction with the DP1receptor expressed by dendritic cells favours an environment in which T cells are polarized to the T helper 2 (TH2) phenotype by inhibiting the production of TH1-promoting cytokines such as interleukin-12 (IL12). Thymic stromal lymphopoietin also acts on dendritic cells to promote T helper 2 (TH2) polarization, an effect associated with the induction of PGD2 synthase in TH2 cells and further PGD2 production. PGD2 promotes the recruitment and activation of TH2 cells through an action on CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells). TH2 cells, once activated, produce cytokines that mediate a plethora of pathological effects relevant to allergic diseases including IgE production by plasma cells, which, through interaction with mast cells and IgE-facilitated antigen presentation, reinforces the cycle of allergic exacerbations.

Clinical relevance of TP, DP 1 and CRTH2

TP-mediated bronchoconstriction. In patients with asthma, inhalation of PGD2 or its metabolite 9α11βPGF2 leads to bronchoconstriction70, and this effect is inhibited both by ramatroban71,72 and by the more selective TP antagonists GR32191 (Refs 73, 74 and ICI192605 (Ref. 75). At doses that effectively block the bronchoconstrictor effect of PGD2, TP antagonists have been tested for their ability to modify lung function in patients with asthma. GR32191 caused a modest inhibition of early phase bronchoconstriction in response to an allergen in some subjects73, whereas both GR32191 and ramatroban were ineffective against exercise-induced decline in lung function71,76. Chronic treatment of patients with asthma with GR32191 did not lead to an improvement in lung function or reduce symptoms77, which effectively rules out an important role for TP-mediated bronchoconstrictor effects of PGD2 and other prostanoids in asthma.

Inflammatory effects of DP1 and CRTH2. Although an important role for TP-mediated bronchoconstrictor effects has been ruled out, there is emerging interest in the inflammatory effects of PGD2 that are mediated by DP1 and CRTH2. There is considerable preclinical evidence that both DP1 and CRTH2, acting together, have a key role in the initiation and maintenance of allergic diseases, but, at present, there are no data available on the clinical effect of selective antagonists in patients with asthma or related disorders. The only clinical data available on DP1 antagonists are for MK-0524, which has been shown to inhibit PGD2-mediated vasodilatation in human volunteers46, but this compound has not yet been tested in patients with allergic diseases. The data currently available on the clinical effects of CRTH2 antagonists are restricted to studies with ramatroban, which is non-selective and has only moderate potency. Indeed, although ramatroban is clearly an effective TP antagonist at clinical doses it is not clear whether these doses are high enough to inhibit CRTH2-mediated effects. Despite these shortcomings, ramatroban is approved in Japan under the trade name Baynas for the treatment of perennial rhinitis, and published clinical studies with this compound provide tentative support for a role for CRTH2 in certain aspects of clinical allergy. In patients with perennial allergic rhinitis, treatment with ramatroban over 4 weeks inhibited chronic nasal swelling but did not increase blood flow to the nasal mucosa (which is likely to be DP1-mediated, see above)78. This effect was associated with a partial reduction in rhinitis symptoms. Treatment of asthmatics with ramatroban for 2 weeks inhibited airway hyperresponsiveness to methacholine79. This effect was attributed to TP antagonism but it is possible that blockade of CRTH2 might have contributed to this response.

Effect of COX inhibition on PGD2 production and asthma symptoms. If, as proposed, mast-cell-derived PGD2 has an important role in allergy why are COX inhibitors generally ineffective in the treatment of asthma and related disorders? Bronchial allergen challenge studies have shown that COX inhibitors — non-steroidal anti-inflammatory drugs (NSAIDs) — do not inhibit the early phase or late-phase airway response in asthmatics80. As PGD2 is produced by mast cells in a COX1- and COX2-dependent process it is expected that COX inhibitors would be effective in inhibiting PGD2 production and this would inhibit both the early and late-phase response to some extent. However, NSAIDs also inhibit the production of PGE2, which inhibits mast-cell activation81 and has bronchodilator properties. The bronchoprotective effect of inhaled PGE2 has been demonstrated in patients with asthma after a bronchial allergen challenge82. Consequently, it is likely that the bronchoprotective effects of PGE2 mask the pro-inflammatory effects of PGD2, and that selective inhibition of PGD2 production or action is required to observe a therapeutic benefit in patients with asthma.

The importance of the protective effect of PGE2 is evident in aspirin-intolerant asthma. In asthmatics intolerant to aspirin, inhibition of COX1-dependent production of PGE2 leads to mast-cell activation and leukotriene-mediated bronchospasm83,84. Paradoxically, PGD2 might contribute to aspects of aspirin-intolerant asthma, as aspirin challenge increases PGD2 production, as measured by levels of its metabolite 9α11βPGF2, in these patients7,8. It is unclear why aspirin enhances rather than inhibits PGD2 production in aspirin-intolerant asthmatics, but it might be related to the high induction of COX2 that occurs in the airways of aspirin-intolerant asthmatics85, which is less sensitive than COX1 to inhibition by aspirin. On the basis of these observations, it is likely that the selective inhibition of PGD2 production or action is likely to have a profoundly different pharmacological profile to the non-selective inhibition of the formation of all prostanoids (including PGE2). This is analogous to the situation in the cardiovascular system in which the balance between thromboxane and prostacyclin production is crucial in controlling platelet activation, blood pressure and atherosclerosis — 'selective' inhibition of thromboxane production with low-dose aspirin is cardioprotective, whereas inhibition of COX2-mediated formation of prostacyclin leaves the effects of thromboxane unopposed and increases the risk of myocardial infarction and stroke86. Inhibition of both thromboxane and prostacyclin production with mixed COX1 and COX2 inhibitors (traditional NSAIDs) is generally considered to be neutral with respect to cardiovascular risk.

Genetic variants of DP1 and CRTH2 as risk factors for the development of asthma. Although the biological validation for the supporting roles of both DP1 and CRTH2 in pathogenesis of allergic diseases is encouraging, experience in this therapeutic area has provided a salutary lesson that apparently compelling preclinical biology does not always translate into clinical efficacy. Several approaches have looked promising on the basis of animal data but have not proved to be of clinical benefit to patients.

One therapeutic approach that that has received a lot of attention is anti-IL5. IL5 has a well-recognized role as the main terminal differentiation factor during eosinopoiesis in the bone marrow87. Many preclinical studies have been performed in mice, guinea pigs and monkeys with monoclonal antibodies directed against IL5 (reviewed by Leckie88). Allergen challenge studies in Cynomolgus monkeys have shown that anti-IL5 abolishes eosinophilia and airway hyperresponsiveness89. However, although the preclinical effects of anti-IL5 are compelling, the clinical experience to date has been disappointing90,91.

It is clear that the efficacy of any particular agent in animal models, even non-human primates, does not translate directly to humans. However, both DP1 and CRTH2 differ from most targets studied in asthma so far in that their role in human disease is supported by clinical genetic studies. Genetic variants of PTGDR (the gene encoding DP1) and CRTH2 predispose individuals towards the development of asthma. In the case of CRTH2 an association with severe asthma has been found in African–American and Chinese populations92. The associated polymorphisms in the 3′-untranslated region of CRTH2 lead to increased mRNA stability, which suggests that gain-of-function variants in CRTH2 are causally linked with asthma92. Such individuals are likely to demonstrate an exaggerated chemotactic response to PGD2, leading to enhanced recruitment and activation of TH2 lymphocytes and other leukocytes involved in allergic responses. By contrast, variants of PTGDR have been detected in patients with asthma that have reduced transcriptional activity93. It has been demonstrated in a small cohort of asthmatics that these putative loss-of-function variants have a protective role against the development of asthma92.

Although the genetic studies highlighting that an 'antagonist surrogate phenotype' protects against asthma are encouraging, the importance of DP1 and CRTH2 in the pathogenesis of allergic diseases will only be known with certainty when carefully controlled clinical trials with potent and selective antagonists have been completed. The following section describes the progress in identifying such antagonists.

Identifying selective CRTH2 antagonists

The chemical literature on DP1 antagonists has been reviewed elsewhere94,95,96. This review of the chemical literature will focus on more recently described CRTH2 antagonists, as these compounds target the receptor that mediates the direct pro-inflammatory effects of PGD2 on the recruitment and activation of TH2 lymphocytes, eosinophils and basophils. Representative compounds from the literature and patents have been selected to illustrate the classes of CRTH2 antagonists currently known.

Identifying potential chemical start points — ligand approaches. Simple modifications to PGD2 and screening of known and related prostanoids have allowed some basic ligand-binding structure–activity relationships (SARs) to be generated97 (Fig. 3).

Figure 3: PGD2 and related prostanoids.
figure 3

The prostaglandin D2 (PGD2) methyl ester has a hundredth of the affinity for CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) compared to PGD2 (319 nM and 2.4 nM, respectively), which indicates that the carboxylic-acid group present in PGD2 is likely to make a key contribution to the binding affinity of PGD2 at the CRTH2 receptor. PGF retains moderate binding affinity for CRTH2 (395 nM), whereas PGE2 has much lower affinity (4,730 nM), which potentially indicates a preference for the hydroxyl group on the cyclopentyl ring to be at the 9 position for optimal binding. Interestingly, modification to the 15-hydroxy group as in 14-dihydro-15-keto-PGD2 (DK-PGD2) does not affect CRTH2 binding (2.9 nM) but does provide high selectivity over DP1.

Of more interest to the drug discovery chemist was the discovery that non-prostanoid structures present in known NSAIDs have affinity for CRTH2 (Ref. 97) (Fig. 4). Several NSAIDs showed moderate binding affinity for CRTH2, such as fenclofenac and sulindac sulphide, but were also active on DP1. Indomethacin, however, demonstrated selective binding for CRTH2 (binding values vary between 25 nM and 8,000 nM depending on assay conditions97,98,99,100) and, interestingly, was shown to be a potent CRTH2 agonist in functional assays97,98. The results with these NSAIDs were replicated when the mouse CRTH2 receptor was studied99,100. Interestingly, in these studies, zomepirac (Fig. 4), which retains the same p-chlorobenzoyl group present in indomethacin, was shown to bind mouse CRTH2 with moderate affinity (3,300 nM) but weakly antagonized the effects of indomethacin100.

Figure 4: NSAIDs used as starting points for the design of selective CRTH2 antagonists.
figure 4

Several non-steroidal anti-inflammatory drugs (NSAIDs) showed moderate binding affinity for CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells), such as fenclofenac and sulindac sulphide (4,953 nM and 3,450 nM respectively), but were also active on DP1. Indomethacin, however, demonstrated selective binding for CRTH2 (binding values vary between 25 nM and 8,000 nM depending on assay conditions) and was shown to be a potent agonist of the receptor (15–50 nM) in functional assays. Zomepirac was subsequently shown to have antagonistic activity in vitro.

The first identification of a potent CRTH2 antagonist was reported by Bayer when it was found that ramatroban (Fig. 5) possessed CRTH2 antagonistic activity55. Ramatroban bound CRTH2 with an IC50 of 100 nM and inhibited PGD2-mediated human eosinophil migration (IC50 of 170 nM). As ramatroban is marketed as an anti-allergic drug in Japan, these findings provided encouragement that a small-molecule oral CRTH2 antagonist was a realistic and tractable therapeutic goal.

Figure 5: Ramatroban analogues with CRTH2 antagonist activity.
figure 5

Moving the pendant amide nitrogen present in ramatroban into the core ring structure led to compound 1, which demonstrated potent (3 nM) antagonistic activity in vitro. By transposing the fused benzene and cyclohexyl rings a series of compounds were identified with moderate binding activity and selectivity as exemplified by compound 2 (CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) binding Ki = 250 nM, TP binding Ki = 1,500 nM). Further modification of compound 2 by shortening the propionic acid group to an acetic acid group gave compound 3, a more potent and selective analogue (CRTH2 binding Ki = 30 nM, TP binding Ki = >20,000 nM) demonstrating once again the strong preference for the acetic-acid group compared to propionic-acid groups in potent CRTH2 antagonists. By simply N-methylating the sulphonamide NH (compound TM30642, CRTH2 binding Ki = 1.9 nM, TP binding Ki = 3,000 nM) or by truncating the propionic-acid group to an acetic-acid group (compound TM30643, CRTH2 binding Ki = 0.51 nM, TP binding Ki = 540 nM) or a combination of both modifications (compound TM30089, CRTH2 binding Ki = 0.6 nM, TP binding Ki = >10,000 nM), three potent CRTH2 binding compounds with high selectivity over TP binding were identified with improved selectivity over ramatroban (CRTH2 binding Ki = 4.3 nM, TP binding Ki = 4.5 nM).

Identifying potential chemical start points — utilizing receptor information. A physicogenetic approach to seven transmembrane-spanning receptors has been used by researchers at 7TM Pharma to classify these receptors with respect to the physicochemical properties of the key amino-acid residues located in a common core ligand-binding site defined from the crystal structure of rhodopsin101. When this approach was applied to CRTH2, the closest related receptors identified were chemokine-like receptor 1 (CMKLR1), angiotensin II receptor type 1 and 2 (AGTR1, AGTR2), in contrast to standard phylogenetic methods of analysis, which highlighted the cysteinyl leukotriene receptors (CysLTs), BLTs (leukotriene B4 receptors, BLT1 and BLT2), and formyl-peptide-receptor-like receptors (FPRLs) as the closest relatives. By focusing on the physicogenetically related AGTR1 and AGTR2 receptors, the existing knowledge around ligands and site-directed mutagenesis data associated with these receptors, a pharmacophore model specific for CRTH2 was derived and used in an in silico screening exercise. Virtual screening of a diverse 1.2 million compound library identified 600 compounds, which were tested at a concentration of 10 μM for CRTH2 binding activity. TM27632 and TM3170 (Fig. 6) were identified as chemically distinct, low micromolar active hits, and further compounds with binding potencies as low as 20–30 nM were also found. Interestingly, the screening of 40 AGTR1 and AGTR2 ligands gave a 25% hit rate, with candesartan (Fig. 6) being the most potent hit (2,100 nM), but because the first approach did not rely on any specific AGTR-ligand scaffold or structural input the hits identified from the 600 compounds screened were selective CRTH2 ligands, which did not require substantial optimization to remove any unwanted AGT-receptor activity101.

Figure 6: Physicogenetically identified CRTH2 antagonists.
figure 6

A physicogenetically based method that classifies seven transmembrane receptors with respect to the physicochemical properties of the key amino-acid residues located in a common core ligand-binding site to CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) identified angiotensin II receptor type 1 (AGTR1) and AGTR2 receptors as likely to share similar binding properties to CRTH2. Screening of a focused compound library led to the identification of TM27632 and TM3170 as CRTH2 ligands with low micromolar affinity, whereas selected screening of AGTR1 and AGTR2 ligands identified candesartan as the most potent (2,100 nM) ligand for CRTH2.

Site-directed mutagenesis of the highly homologous mouse CRTH2 has been studied and provides key information on how prostanoid and non-prostanoid ligands bind102. PGD2 and indomethacin interact with distinct but overlapping sets of residues within the CRTH2 binding pockets, and CRTH2 resembles chemoattractant receptors more closely than other prostanoid receptors in terms of ligand-binding parameters. Amino-acid residues were selected for mutation on the basis of CRTH2-receptor modelling and those key residues common to the well-studied and related formyl peptide receptor (FPR) and C5a chemotactic receptors. It was found that by sequentially replacing His106 (TM III), Lys209 (TM V), Glu268 (TM VI) and Arg178 (extracellular loop II) with alanine PGD2 binding was reduced, but the His106Ala and Glu268Ala mutants bound indomethacin and ramatroban with similar affinity to the wild-type receptor. The Arg178Ala mutant and Lys209Arg modification reduced binding of PGD2, indomethacin and ramatroban by fourfold to tenfold, indicating that these residues are important for the binding of both prostanoid and non-prostanoid ligands. Interestingly, the Tyr261Phe mutant retained PGD2 binding but reduced binding of both indomethacin and ramatroban indicating, again, the existence of overlapping but distinct binding pockets for these two classes of ligand. Ligand-docking studies with PGD2 and indomethacin suggested that Lys209 might interact with the carboxylate group of each ligand and that for prostanoid ligands the 9-hydroxyl group on the cyclopentyl ring forms a hydrogen bond to Glu268.

Structural classes of CRTH2 antagonists

Indole acetic acids. Once indomethacin was identified as a potent ligand for CRTH2 with agonistic activity, it provided a template from which to try and derive potent and selective CRTH2 antagonists devoid of COX activity. The first indication that this was possible came from a Pfizer screening patent that identified the compound indole 4 (Fig. 7) as an antagonist with 40-fold selectivity over DP1 (Ref. 103). Researchers at AstraZeneca have subsequently patented a range of indole-acetic-acid derivatives104,105,106,107,108,109,110 (Fig. 7).

Figure 7: Indole-acetic-acid based compounds with CRTH2 antagonist activity.
figure 7

Indole 4 was the first-identified CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) antagonist within this structural class with 40-fold functional selectivity for CRTH2 over DP1. N-aryl substituted analogues such as compound 5 demonstrated moderate functional antagonist activity in an in vitro calcium mobilization assay (pA2 = 6.8). Subsequent transposition of the acetic-acid and aryl groups led to more potent compounds such as 6, which bound CRTH2 with a pIC50 of 8.15 and introduction of a hetero-spacer group (S, SO2, O) between the aromatic group and the indole at position 3 generated highly potent compounds such as 7, which bound CRTH2 with a pIC50 of 9.4. The corresponding transposed indole 3-acetic-acid compounds with the same spacer groups from the indole at position 1 have also been reported. Interestingly, some of the extended sulphonamide analogues such as compound 8 retain binding affinity for both CRTH2 (Ki = 222 nM) and DP1 (Ki = 86 nM), whereas truncated compounds such as 9 are CRTH2 selective.

Oxagen have patented 1-acetic-acid derivatives with either methylene or sulphonyl spacers between the indole and aromatic groups from the indole-3 position111,112, and have also disclosed the corresponding transposed indole-3-acetic-acid compounds with the same spacer groups from the indole-1 position113,114. More detailed information on the SAR of truncated sulphonamides has been disclosed with compound 9 (Fig. 7), which demonstrated highly selective CRTH2-binding Ki of 68 nM and functional antagonist activity on human eosinophil shape change (IC50 = 74 nM) and human TH2-cell chemotaxis (IC50 = 67 nM)115. Indole 9 was also shown to be metabolically stable in vitro and devoid of any COX, cytochrome P450, cytotoxicity or human ERG (ether-a-go-go related gene) liabilities. Pharmacokinetic studies in rats with compound 9 was shown to be 56% bioavailable with an oral half-life of 5.5 hours115.

Ramatroban analogues. Although ramatroban is a non-selective CRTH2 antagonist with only moderate potency, it also provided an attractive starting point from which to derive selective CRTH2 antagonists as, unlike indomethacin, it already had antagonist functionality. Shionogi disclosed a range of ramatroban analogues in the patent literature116 and, subsequently, Actelion have patented analogues in which the pendant amide nitrogen in ramatroban has been moved into the core ring structure, with compound 1 (Fig. 5) having 3 nM antagonist CRTH2 activity in an in vitro calcium mobilization assay117.

7TM Pharma have also demonstrated that only minor changes to the structure of ramatroban can lead to highly selective CRTH2 antagonists without TP activity118 (Fig. 5). Subsequently, it was shown that the functional activities of these three antagonists were different — ramatroban and TM30642 demonstrated competitive antagonism of PGD2, whereas both TM30643 and TM30089 showed non-competitive antagonism, possibly due to CRTH2 blockade mediated by an orthosteric site of the receptor119.

More recently, researchers at Athersys have used a 'reverse scaffold' approach to isosteric ramatroban analogues120 (Fig. 5).

Aryl acetic acids. Aryl acetic acids were first identified as CRTH2 ligands from fenclofenac (Fig. 4) and as antagonist templates in the same Pfizer screening patent that identified the first indole acetic acid CRTH2 antagonists. In that patent the compound biaryl thioether 10 was identified as a CRTH2 antagonist with 40-fold functional selectivity for CRTH2 over DP1 (Ref. 103) (Fig. 8). Subsequently, several companies have filed patents on the aryl-acetic-acid template as CRTH2 antagonists, for example, AstraZeneca have patented biaryl-acetic-acid derivatives121,122,123,124 such as compound 11 (Fig. 8) with a CRTH2-binding pIC50 of 8.2 (Ref. 122) and biaryl-ether-acetic-acid derivatives125,126such as compound 12 (Fig. 8) with a CRTH2-binding pIC50 of 9.0 (Ref. 125).

Figure 8: Aryl acetic acid CRTH2 antagonists.
figure 8

The biaryl thioether 10 was one of the first-identified CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) antagonists within this structural class with 40-fold functional selectivity for CRTH2 over DP1. Biaryl-acetic-acid derivatives such as compound 11 with a CRTH2 binding pIC50 of 8.2 and biaryl-ether-acetic-acid derivatives such as compound 12 with a CRTH2 binding pIC50 of 9.0 have been patented.

Non-acidic CRTH2 antagonists. The non-acidic CRTH2 antagonists identified so far are dominated by the tetrahydroquinoline structural class. It is most likely that this class of compounds was derived from screening campaigns within pharmaceutical companies as there has been a flurry of patent publications all outlining very similar structures within a short time frame. The class itself is important because they highlight that potent and selective CRTH2 antagonists can be identified that lack what was assumed to be an essential carboxylic acid within their structure. Although the lack of a chemical handle from which to derive salts might prove to be a hindrance in terms of solubility and pharmaceutical development of this class of compound they do offer the ability to cross the blood–brain barrier and treat CNS disorders without resorting to a carboxylic-acid prodrug approach.

The first patent applications in the tetrahydroquinoline class were made by Millennium (now believed to be licensed to Sanofi–Aventis) followed closely by Warner–Lambert (now Pfizer). Millennium disclosed a range of analogues including compound 13 (Ref. 127) (Fig. 9) and followed this with a further application claiming more extended amide compounds128. Warner–Lambert (Pfizer) have patented similar structures129,130 (Fig. 9) and, interestingly, have filed use patents demonstrating penetration into cerebral spinal fluid after an oral dosing of compound 14 (Fig. 9) and efficacy in models of inflammatory and neuropathic pain at oral doses of 25 mg per kg131.

Figure 9: Non-acidic CRTH2 antagonists.
figure 9

Tetrahydroquinoline 13 was one of the first non-acidic CRTH2 (chemoattractant receptor-homologous molecule expressed on TH2 cells) antagonists identified. Subsequently compound 14 has been demonstrated to penetrate into the cerebrospinal fluid after oral dosing and possess efficacy in both inflammatory and neuropathic pain models. K117 demonstrated an IC50 of 7.8 nM against prostaglandin-D2-induced human eosinophil chemotaxis and was selective for CRTH2 in ligand-binding experiments when compared with several prostanoid receptors and prostanoid-generating enzymes, including cyclooxygenases. Interestingly, the close analogue K376 was shown to be a CRTH2 agonist indicating the requirement for an aromatic-group-based amide from the tetrahydroquinoline nitrogen atom to maintain CRTH2-antagonist activity.

Researchers at Kyowa Hakko Kogyo have also published some preliminary SAR information within this series132 (Fig. 9).

Conclusions

It is now becoming clear that PGD2 is an important mediator of allergic responses. The high concentrations produced in response to an allergic stimulation combined with its highly potent activity result in PGD2 having a dominant role in mediating mast-cell-dependent activation of TH2 lymphocytes, an effect mediated by CRTH2. Therefore, PGD2 produced by mast cells might provide an essential link between the early phase and late-phase allergic responses and such antagonism of PGD2 provides an attractive target for therapeutic intervention. There are a number of potent and selective CRTH2 antagonist series identified with drug-like properties and results of ongoing clinical trials in asthma and allergic rhinitis are awaited with interest.