I. INTRODUCTION

Top Previous Next References

Prostanoids that consist of the prostaglandins (PG) and the thromboxanes (Tx) are cyclooxygenase products derived from C-20 unsaturated fatty acids (Fig. 1). Prostaglandins contain a cyclopentane ring and two side chains named  and  attached to the ring. According to the modifications of this cyclopentane ring, they are classified into types A to I, in which types A, B, and C are believed not to occur naturally but are produced only artificially during extraction procedures. Prostaglandins G and H share the same ring structure but differ at C-15, having a hydroperoxy and hydroxy group, respectively. Another cyclooxygenase product, TxA, has an oxane ring instead of the cyclopentane ring. Prostanoids are further classified into three series (1, 2, and 3) based on the number of double bonds in their side chains; the series 1 prostanoids contain a 13-trans double bond, the series 2 prostanoids have 5-cis and 13-trans double bonds, and the series 3 prostanoids have 5-cis, 13-trans, and 17-cis double bonds. The prostanoids in series 1, 2, and 3 are synthesized from -homolinolenic acid (8,11,14-eicosatrienoic acid), arachidonic acid (5,8,11,14-eicosatetraenoic acid), and 5,8,11,14,17-eicosapentaenoic acid, respectively. Because arachidonic acid is the most abundant among these precursor fatty acids in most mammals, including humans, the series 2 prostanoids are predominantly formed in their bodies. The above fatty acids are liberated from membrane phospholipids in response to various physiological and pathological stimuli by the action of phospholipase A₂ and are converted to various prostanoids by the sequential actions of cyclooxygenase and the respective synthases. Prostanoids thus formed are released outside of the cells immediately after synthesis. Prostaglandin G, PGH, PGI, and TxA are chemically unstable and are degraded into inactive products under physiological conditions, with a half-life of 30 s to a few minutes. Other PG, although chemically stable, are metabolized quickly. For example, they are inactivated during a single passage through the lung. It is believed therefore that prostanoids work locally, acting only in the vicinity of the site of their production.

View larger version (21K): [in this window] [in a new window]   Fig. 1. Biosynthetic pathways of prostanoids. Formation of series 2 prostaglandins (PG), PGD2, PGE2, PGF2, PGG2, PGH2, and PGI2, and a thromoboxane (Tx), TxA2, from arachidonic acid is shown. The first 2 steps of pathway, i.e., conversion of arachidonic acid to PGG2 and then to PGH2, are catalyzed by cyclooxygenase, and subsequent conversion of PGH2 to each PG is catalyzed by respective synthase as shown. Ring structures of A, B, and C types of PG are shown separately.

Prostanoids exert a variety of actions in various tissues and cells. The most typical actions are the relaxation and contraction of various types of smooth muscles. They also modulate neuronal activity by either inhibiting or stimulating neurotransmitter release, sensitizing sensory fibers to noxious stimuli, or inducing central actions such as fever generation and sleep induction. Prostaglandins also regulate secretion and motility in the gastrointestinal tract as well as transport of ions and water in the kidney. They are involved in apoptosis, cell differentiation, and oncogenesis. Prostanoids also regulate the activity of blood platelets both positively and negatively and are involved in vascular homeostasis and hemostasis. Because prostanoids are produced from fatty acids and hence are generally regarded as hydrophobic compounds, it was thought in earlier times that they were incorporated into the cell membrane and exerted their action by perturbing lipid fluidity. The concept of prostanoid action via prostanoid receptors gradually appeared. This arose from several different lines of studies. First, prostanoids are not as hydrophobic as they were once thought to be and do not incorporate into or permeate the cell membrane (18). Second, each prostanoid has a unique activity profile not exactly overlapping with others, indicating that each prostanoid has a specific site of action. This became apparent by comparing the potencies of various prostanoids and their synthetic analogs on various tissues by bioassay. Studies along this line led to the suggestion of the presence of multiple types of prostanoid receptors in different tissues and cells such as the lung and platelets (7, 66, 179) and culminated in a proposal to classify the prostanoid receptors in 1982 (107). Moreover, various synthetic TxA₂ agonists and antagonists were developed in the late 1970s to early 1980, and using these compounds, a receptor for TxA₂ was identified pharmacologically as the site of competition of TxA₂ agonists and antagonists (for example, see Ref. 95).

The presence of a receptor(s) for prostanoids had also been suggested biochemically. It had been repeatedly reported that the actions of prostanoid were associated with changes in the level of second messengers. Already in the mid 1960s, some prostanoid actions had been noticed to be associated with changes in cAMP levels (see, for example, Ref. 31). Later, the association of the actions of these prostanoids with changes in phosphatidylinositol (PI) turnover and free Ca²⁺ concentrations in the cell were reported. In addition, with the availability of radiolabeled derivatives of prostanoids, it was found in the early 1970s that many tissues and cells contain specific high-affinity binding sites for the prostanoids (117, 183, 184). Although the binding sites identified in earlier studies may not have represented functional receptors, functional correlates of the binding activities to the bioactivity or to the second messenger systems were examined in later studies. Coleman et al. (44) integrated the information obtained by these approaches to propose a comprehensive classification of the prostanoid receptors. They proposed the presence of receptors specific for Tx, PGI, PGE, PGF, and PGD and named them the TP, IP, EP, FP, and DP receptors, respectively. They further classified the EP receptor into three subtypes: EP₁, EP₂, and EP₃, all of which respond to the naturally occurring agonist PGE₂ but differ in their actions and in their responses to various analogs. They later reported a fourth subtype, the EP₄ receptor, which, like the EP₂ receptor, is positively coupled to adenylate cyclase but differs in its response to certain ligands (43). However, none of the receptors had been isolated and cloned until the TxA₂ receptor was purified from human blood platelets in 1989 (239) and its cDNA cloned in 1991 (79). These studies revealed that the TxA₂ receptor was a G protein-coupled rhodopsin-type receptor with seven transmembrane domains. Homology screening in mouse cDNA libraries subsequently identified the structures of all of the eight types and subtypes of the prostanoid receptors. These receptors have been expressed, and their ligand binding properties and signal transductions have been examined. In addition, the tissue and cell distribution of the receptors was studied by Northern blot and by in situ hybridization analyses of their mRNA expression. Correlation of such knowledge with findings accumulated by pharmacological studies using cyclooxygenase inhibitors and using various prostanoid analogs having agonistic and antagonistic activities helps to define the actions of each type of receptor. They also help to reveal novel actions of these receptors. Recently, knockout mice deficient in each receptor have been generated by gene targeting, and implications and significances of prostanoid actions in various physiological and pathophysiological processes are being examined and assessed. This review summarizes current information obtained by these studies. The correlattion of these studies to previous pharmacological works is emphasized to give an overview of the physiological and pathophysiological roles of the prostanoid receptors.

	II. STRUCTURES OF PROSTANOID RECEPTORS AND THEIR GENES

Top Previous Next References

The structure of the human TxA₂ receptor is shown in Figure 2 as a representative of the prostanoid receptors. It is a protein composed of 343 amino acids and is a G protein-coupled rhodopsin-type receptor with 7 putative transmembrane domains. Since the cloning of this receptor in 1991 by Hirata et al. (79), homology screening based on its sequence was performed in various species, and all of the eight types and subtypes of the prostanoid receptors previously defined pharmacolgically were identified. They include the human and mouse PGD receptor (DP) (21, 80); the mouse, rat, and human PGE receptor EP₁ subtype (22, 63, 244); the mouse and human PGE receptor EP₂ subtype (101, 187); the mouse, human, rat, rabbit, and bovine PGE receptor EP₃ subtype (3, 27, 154, 186, 217, 226, 250); the mouse, human, and rat PGE receptor EP₄ subtype (originally reported as the EP₂ subtype; see below) (6, 13, 84, 193); the mouse, human, bovine, rat, and sheep PGF receptor (FP) (2, 70, 110, 191, 216); the mouse, human, and rat PGI receptor (IP) (20, 103, 149, 152, 197); and the mouse, rat, and bovine TxA receptor (TP) (1, 146, 153). Initially, two species of cloned receptors were reported as EP₂; one was originally cloned by Honda et al. (84) and subsequently by other people (6, 13, 193) and one was later cloned by Regan et al. (187). Although Honda et al. (84) reported their cloned PGE receptor as EP₂ on the basis of its positive coupling to adenylate cyclase, this receptor is insensitive to butaprost, a synthetic PGE derivative, which is inconsistent with the pharmacolgically defined EP₂ receptor (44). On the other hand, the receptor cloned by Regan et al. (187) is sensitive to butaprost. Later, the presence of another PGE receptor subtype with positive coupling to adenylate cyclase was suggested by pharmacological methods (43). The subsequent characterization of the "EP₂" receptor of Honda et al. (84) revealed that it is sensitive to an EP₄-specific ligand, AH23848B (163). Moreover, a mouse homolog of the receptor cloned by Regan et al. (187) was cloned and shown to have properties consistent with the pharmacologically defined EP₂ (101). These results suggest that the receptors cloned originally by Honda et al. (84) in the mouse and subsequently by other groups in other species represent the EP₄, and that cloned by Regan et al. (187) represents the EP₂ subtype.

View larger version (48K): [in this window] [in a new window]   Fig. 2. Structure of human TxA2 receptor. Membrane topology model of 2 isoforms of human TxA2 receptor is shown. Amino acid residues are indicated by single letter code. Two isoforms are different only in carboxy-terminal tail (see sect. IIB and Fig. 4). Amino acid residues conserved by most of prostanoid receptors are shown by solid circles with white letters. N-glycosylation at Asn4 and Asn16 is indicated by -CHO, and putative disulfide bond between first and second extracellular loops is shown.

The amino acid sequence alignment of the eight types and subtypes of cloned mouse prostanoid receptors is shown in Figure 3A. They are aligned based on the homology of the putative seven transmembrane domains of these receptors. There are a total of 28 amino acid residues conserved in and close to these regions. Among them, eight residues are shared also by other families of rhodopsin-type receptors, and they are believed to be involved in the maintenance of structure and/or function of these types of receptors in general. For example, Asp in the second transmembrane domain has been shown in other receptors to be involved in activation of the receptors, by coupling ligand binding to the activation of G proteins (198). Two Cys residues, one in the first and the other in the second extracellular loop, are also conserved. These residues are believed to form a disulfide bond critical for stabilization of receptor conformation and for ligand binding. This function of these residues has been examined in the prostanoid receptors (see section IIIC). In addition to these conserved residues, several features common in the rhodopsin-type receptors are also found in the prostanoid receptors. First, they have one or more consensus sequences for N-glycosylation of asparagine residue(s), Asn-X-Ser/Thr, in the amino-terminal extracellular portion. This motif is found, for example, at Asn-4 and Asn-16 in the amino-terminal portion of the human TxA₂ receptor. Although the TxA₂ receptor protein purified from the platelet membrane was ~57 kDa, the molecular mass calculated from the primary structure is 37 kDa. This difference appears to be accounted for by glycosylation of the receptor. The analysis of the amino-terminal sequence of the purified protein showed that both of these sites were modified (79), and the treatment of the purified protein with N-glycanase reduced the molecular mass of the protein to ~37 kDa (126). Chiang and Tai (37) reported that deletion of the carbohydrate moieties of the human TP receptor by adding tunicamycin during infection of Sf21 cells or mutation of both Asn-4 and Asn-16 abolished the ligand binding to the receptor and suggested that N-glycosylation is crucial for its binding function. Similar potential N-glycosylation sites are also seen in the first extracellular loop of the DP, IP, and EP₂ receptors and in the second extracellular loop of the EP₃ and EP₄ receptors. In addition, as in many other rhodopsin-type receptors, serine and threonine residues, which comprise putative phosphorylation sites, are widely distributed in the cytoplasmic portion of the prostanoid receptors. Phosphorylation of these residues is thought to participate in the desensitization of these receptors, as noted in other rhodopsin-type receptors (77). In fact, agonist-induced phosphorylation (71, 175) and phosphorylation by protein kinase A (PKA) and protein kinase C (PKC) (108) of the human TP receptor are reported to be involved in receptor desensitization. On the other hand, although some rhodopsin-type receptors are palmitoylated at the Cys residue in the carboxy tail and form an additional intracellular loop (24), the consensus sequence for this modification, Leu-X-Cys-(X)n-Arg/Lys- in which the Cys residue is located 11-16 residues distal to the end of the seventh transmembrane domain, is not found in the prostanoid receptors, although IP, EP₂, and EP₄ receptors have a Cys residue at the appropriate position.

View larger version (70K): [in this window] [in a new window]   Fig. 3. A: amino acid sequence alignment of 8 types and subtypes of prostanoid receptors. Amino acid sequences of 8 types and subtypes of mouse prostanoid receptors are aligned. Amino acid residues shared by >5 receptors are shadowed. Putative transmembrane domains are indicated by lines above sequence and numbered. Most carboxy-terminal 64-amino acid residues of EP4 receptor are not shown. B: amino acid sequence alignment of EP3 receptors from different species. Amino acid sequences of corresponding isoform of EP3 receptor from human, bovine, rabbit, rat, and mouse are aligned. Only amino acid residues different from corresponding residues of human receptor are shown for other species. Note that amino-terminal portions of rat and mouse receptors are ~20 amino acids shorter than receptors from other species.

In addition to the features common to other rhodopsin-type receptors, particular motifs specifically conserved among the prostanoid receptors are found in several regions (Fig. 3A). For example, L-X-A-X-R-X-A-S/T-X-N-Q-I-L-D-P-W-V-Y-I-L or homologous motifs are found in the seventh transmembrane domain of most of the prostanoid receptors. Two other sequences, G-R-Y-X-X-Q-X-P-G-T/S-W-C-F and M-X-F-F-G-L-X-X-L-L-X-X-X-A-M-A-X-E-R, are found in the second extracellular loop and in the third transmembrane domains, respectively. Because these highly conserved regions are shared by prostanoid receptors from various species (Fig. 3B), they may participate in the construction of binding domains for structures common to prostanoid molecules. The arginine in the seventh transmembrane domain, which is conserved in all of the prostanoid receptors, was proposed to be the binding site of the carboxyl group of prostanoid molecules by analogy to the retinal binding site, Lys-296, of rhodopsin (79, 155), and has been subjected to extensive mutational analyses (see sect. IIIC). In comparison with the sequences of the transmembrane domains, the alignment of the sequences of intracellular structures is difficult because of their diversity both in composition and in length. However, the conservation of several residues has been found. For example, an Arg or other basic residue is found in the first intracellular loop of all of the prostanoid receptors at analogous positions. A spontaneous mutation of this arginine residue has been found in the TP receptor of patients with a hereditary bleeding disorder and identified as the cause of this disease (see sect. IIIC).

Despite the presence of these conserved sequences, overall homology among the prostanoid receptors is quite limited, ranging from ~20 to 30%. It is worth noting that there is only this limited extent of homology even among the four subtypes of PGE receptors, which makes it difficult to get any insight into the areas determining the ligand binding specificity of each receptor, simply by comparing the sequence of one receptor with those of the receptors for other prostanoids. On the other hand, the homology of a given type or subtype of receptor among various species is considerably higher. For example, the sequence homology between human and mouse IP, TP, EP₁, EP₃, EP₄, and FP receptors is 79, 76, 84, 84, 88, and 89%, respectively, and the homology among human, bovine, rabbit, rat, and mouse EP₃ receptors ranges from 84 to 97% (Fig. 3B). There are differences in the translation initiation sites of some receptor types and subtypes in different species, which affect the length of the amino-terminal extracellular domain of the receptor. For example, this portion of the human, bovine, and rabbit EP₃ receptor is 20 amino acids longer than that of the rat and mouse receptors. Conversely, this portion of the human IP receptor is 30 amino acids shorter than that of the rat and mouse receptors. Differences in the actions of prostanoids between species are well known. For example, EP-157, a PGI analog, acts on human and horse platelets as an agonist, whereas it is an antagonist when examined in pig and rat platelets (10). Rabbit platelets are known to differ from human, cat, and canine platelets in their response to CTA₂ and PTA₂, which are both Tx analogs (30). The potency of ONO-11120, an antagonist of the TP receptor, is two orders of magnitude lower in rabbit platelets than in human platelets (156). These species differences may be due to differences in the structure of the respective receptors, despite high sequence homology among receptor homologs from various species. The molecular basis for the difference in sensitivity of the rat and human TP receptor to I-BOP, a TP agonist, has been examined in the rat/human chimeric TP receptor (see sect. IIIC).

Chromosomal localizations of the genes encoding the mouse and human prostanoid receptors have been determined. The genes encoding the mouse DP, EP₁, EP₃, EP₄, FP, IP, and TP receptors were mapped to chromosomes 14, 8, 3, 15, 3, 7, and 10, respectively (91, 225). The genes encoding the human EP₁, EP₃, EP₄, FP, IP, and TP receptors were mapped to chromosome bands 19p13.1, 1p31.2, 5p13.1, 1p31.1, 19q13.3, and 19p13.3, respectively (59, 165). The gene encoding the human DP receptor and those encoding the mouse and human EP₂ receptors have not yet been mapped. These studies showed that the EP₁, EP₄, IP, and TP receptor genes are localized in chromosomal segments of each animal previously found homologous between the mouse and the human (28, 50, 229). Furthermore, it is notable that the loci of the EP₃ and FP receptor genes are in close proximity in both the mouse and human chromosome, suggesting that the distal mouse chromosome 3 is the homologous segment of the short arm of human chromosome 1 and that the EP₃ and FP receptor genes evolved by gene duplication.

The structure of a prostanoid receptor gene was first clarified for the human TP receptor, which contains three exons separated by two introns, one in the 5'-noncoding region and the other at the end of the sixth transmembrane domain (165) (Fig. 4). This exon-intron relationship appears to be conserved in other types of prostanoid receptors and across various species, such as in the mouse and the human DP receptor (21, 80), the mouse EP₁ receptor (15), the mouse EP₂ receptor (104), the human EP₃ receptor (186), the human and mouse EP₄ receptor (8, 62), the mouse FP receptor (75), and the human IP receptor (167). The first intron is located upstream of the ATG start codon in the reported prostanoid receptor genes, except in the mouse EP₄ receptor gene, in which it is located 16 bp downstream of the translational start site (8). There are additional exons encoding carboxy-terminal tails in some of the prostanoid receptors, and alternative splicing of these exons further creates several isoforms. This was observed in the mouse, rat, bovine, rabbit, and human EP₃ receptors (3, 27, 90, 115, 154, 161, 186, 219, 226), the human TP receptor (185), and the ovine FP receptor (180). This alternative splicing occurs in the carboxy-terminal region after the seventh transmembrane domain and creates various receptor isoforms that differ only in their carboxy tails. The isoforms of the EP₃, FP, and TP receptors have almost identical ligand-binding specificities among each receptor. However, isoforms of the bovine EP₃ and human TP receptors are coupled to different G proteins and induce different signaling pathways (82, 154), and those of the mouse EP₃ receptor are different in their efficacy of G protein coupling (219), in their sensitivity to agonist-induced desensitization (161), and in their extent of constitutive activity (74, 157). The differences in constitutive activity have also been reported for isoforms of the human EP₃ and ovine FP receptors (94, 180). Because the carboxy-terminal domains of some EP₃ receptor isoforms are similar between species whereas others do not show such homology, the possible existence of at least seven EP₃ receptor isoforms in any given species has been proposed (186). In fact, eight human EP₃ receptor isoforms have recently been reported (115). These results may suggest the presence of other isoform(s) of the TP receptor, since there is no homology in the carboxy-terminal domains of the two human TP receptor isoforms and the mouse TP receptor. Another variant receptor has been reported in the rat EP₁ receptor (174). This variant receptor was generated by failure of splicing in the sixth transmembrane domain in exon 2 and is suggested to have a seventh transmembrane domain that is not homologous to the seventh transmembrane domain highly conserved among all members of the prostanoid receptors. When expressed in cultured cells, this receptor shows similar ligand binding specificity with the EP₁ receptor but is defective in signal transduction and suppresses signaling through other PGE receptors. The result that splice variants are found only in EP₁, EP₃, FP, and TP and not in DP, EP₂, EP₄, and IP coincides with the results of phylogenetic analyses of the prostanoid receptors, which show that the former and latter groups of receptors form different branches in receptor evolution (see sect. IIC).

View larger version (9K): [in this window] [in a new window]   Fig. 4. Organization of human TxA2 receptor gene and generation of receptor isoforms by alternative splicing of its mRNA. Three exons of TP receptor gene are indicated by white boxes (top). Amino acid coding regions of cDNA are shown by boxes (middle and bottom), and those corresponding to putative transmembrane domains are indicated by solid boxes and numbers. Note that alternative usage of third exon generates 2 isoforms of receptor, TP and - that differ in their carboxy-terminal tails.

Analyses of the 5'-flanking region and the first intron of prostanoid receptor genes have revealed several consensus sequences in the cis-acting regulatory elements (Fig. 5). Basal promotor motifs for transcription such as the TATA box and CCAAT box have been identified in the 5'-flanking region of the transcription initiation site of some of the prostanoid receptor genes. The promotor regions of the mouse EP₄, the human EP₃, and the human EP₄ receptor genes have a TATA box, a TATA-like box, and two CCAAT boxes, respectively. On the other hand, the human TP, the human IP, the mouse FP, and the mouse EP₁ receptor genes lack these conventional motifs. A variety of other regulatory elements have also been found in the promotor region of the prostanoid receptor genes. The human TP receptor gene contains SP-1 binding sites, AP-2 consensus sequences, a phorbol ester response element (TRE), acute-phase reactant regulatory elements (APRRE), a c-myc binding motif, and a glucocorticoid response element (165). The human IP receptor gene shows AP-2 consensus sequences, APRRE, human polyoma virus JC promotor elements (JCV), c-myb binding motifs (c-Myb), an AP-1 binding site, SP-1 binding sites, and a glucocorticoid response element (GRE) (167). There is an AP-1 site and an AP-2 site in the mouse EP₁ receptor gene (15) and a sis-inducible factor (SIF) binding element, E boxes, AP-2 sites, an interferon (IFN)- responsive element (-IRE), a c-Myb, and a GC box in the human EP₃ receptor gene (115). The human EP₄ receptor gene contains several responsive motifs for proinflammatory agents such as NF-IL6, NFB, and H-apf-1 in addition to a Y box, AP-1 sites, and AP-2 sites (62). The mouse EP₄ receptor gene contains AP-1 sites, AP-2 sites, SP-1 sites, an NFB element, an E box (MyoD), an NF-IL6 element, a GRE, and Pit-1 sequences (8). The mouse EP₂ receptor gene contains NF-IL6 and NFB elements as well as a progesterone response element (PRE) (104). These motifs are well correlated with the result that the EP₂ gene is regulated by both proinflammatory and hormonal stimuli. Interestingly, the transcriptional initiation sites of the EP₂ gene are different between the macrophage and the uterus, suggesting alternative promotor usage in these tissues (104).

View larger version (14K): [in this window] [in a new window]   Fig. 5. Transcriptional cis-elements in promotor regions of prostanoid receptor genes. Promotor regions of human EP4 receptor gene (A) (62) and human IP receptor gene (B) (165) are shown.

The above findings suggest that the expression of prostanoid receptors is regulated by several factors through action on cis-acting regulatory elements of the respective receptor genes. However, only limited information is presently available regarding this regulation. As noted above, the promotor region of the TP receptor gene has a TRE (165). In accordance with this result, treatment with phorbol esters induced TP receptor expression in human erythroleukemia (HEL) cells, a megakaryocyte-like cell line (108, 150), and in CHRF-288 megakaryoblastic cells (57). On the other hand, Kinsella et al. (108) reported that glucocorticoids and inducers of the acute-phase response, such as interleukin (IL)-1, IL-6, lipopolysaccharide (LPS), C-reactive protein, and tumor necrosis factor (TNF), could not induce this gene expression in HEL cells, despite the presence of a GRE and APRRE in its promotor region. However, in contrast to HEL cells, glucocorticoids and IL-6 induced TP receptor expression in rat cultured vascular smooth muscle cells (224). These differences in the regulation of gene expression of the TP receptor may result from differences in species or cell types. Remarkably, Halushka and colleagues reported that testosterone induced gene expression of the TP receptor in HEL cells (129) and increased platelet thromboxane A₂ receptor density and the aggregation response in humans (4). Gene expression of the prostanoid receptors induced by phorbol esters was examined in cultured cell lines of monocytoid and lymphoid lineage (141). When these cells were stimulated by phorbol esters, EP₄ receptor gene expression was upregulated in THP-1 and U 937 cells (monocytoid cell lines), and Raji cells (B-cell line), and was downregulated in MOLT-4 and Jurkat cells (T-cell lines). This effect of phorbol esters on THP-1 cells was specific to the expression of the EP₄ receptor gene, and expression of the other prostanoid receptors, such as the EP₁, EP₂, EP₃, IP, and DP receptors, was not upregulated. EP₄ receptor expression was also upregulated in NIH 3T3 fibloblast and RAW 264.7 macrophage cell lines after stimulation with serum or bacterial LPS, respectively (8). Although EP₂ receptor expression was also upregulated in RAW 264.7 cells, that of the IP receptor was downregulated in both NIH 3T3 and RAW264.7 cell lines. Katsuyama et al. (100) showed that in the J774.1 macrophage cell line, LPS treatment augumented the expression of both EP₂ and EP₄, but the increase in EP₂ expression was much more drastic. Furthermore, the simultaneous addition of IFN- only inhibited LPS-induced expression of EP₂, but not that of EP₄. These results suggest a possibility that gene expression of prostanoid receptors is regulated in the body under various physiological and pathopysiological conditions. However, such in vivo regulation of receptor expression has not yet been demonstrated.

The prostanoid receptors thus consist of eight types, each encoded by different genes. These receptors can be grouped into three categories on the basis of their signal transduction and action: the relaxant receptors, the contractile receptors, and the inhibitory receptors. The relaxant receptors, which mediate increases in cAMP and induce smooth muscle relaxation, consist of the IP, DP, EP₂, and EP₄ receptors. The contractile receptors consist of the TP, FP, and EP₁ receptors, which mediate Ca²⁺ mobilization and induce smooth muscle contraction. The EP₃ receptor is an inhibitory receptor that mediates decreases in cAMP and inhibits smooth muscle relaxation (for detailed discussion, see sect. IIB). Sequence homology among these functionally related receptors is higher than those between the receptors from the three separate groups. The overall homology among the relaxant receptors is between 32 and 44%, and the homology in the transmembrane domains among contractile receptors is ~50%. In contrast, the homology of the EP₃ receptor with any receptor from the other two groups remains below 30%. Toh et al. (233) used a computer-assisted method and performed a more detailed sequence comparison of the prostanoid receptors. They also included receptors for other types of lipid mediators, namely, human and guinea pig platelet-activating factor (PAF) receptor (85, 151) and human lipoxin A receptor (61) in their analysis and constructed a phylogenetic tree to infer the evolutional relationship among the lipid mediators. They found that the prostanoid receptors constitute a distinct cluster within the rhodopsin-type receptors, while PAF and lipoxin receptors belong to another cluster shared by peptide receptors such as those for tachykinin, bradykinin, and endothelin. The prostanoid receptor cluster was further divided into three subclusters: cluster 1 consists of the relaxant receptors, EP₂, EP₄, IP and DP; cluster 2 consists of the contractile receptors EP₁, FP, and TP; and cluster 3 consists of the inhibitory receptor EP₃ (Fig. 6). A similar phylogenetic tree was reported by Reagan et al. (187) and Boie et al. (21). These results suggest that the cyclooxygenase pathway may have been initiated as a system composed of PGE and its receptor, where the subtypes of the PGE receptor then evolved from this primitive PGE receptor to mediate different signal transduction pathways and that receptors for other PG and Tx subsequently evolved from functionally related PGE receptor subtypes by gene duplication. The results also suggest that the prostanoid receptors evolved differently from receptors for other lipid mediators. Recently, Yokomizo et al. (252) reported the identification of the leukotriene B₄ receptor. This receptor is another G protein-coupled rhodopsin-type receptor that was previously isolated as an orphan chemoattractant receptor. It shows significant homology not only to the lipoxin A receptor but also, like the lipoxin receptor, to peptide receptors such as somatostatin receptor type 3 and the IL-8 receptor, but not to the prostanoid receptors. These findings suggest that the cyclooxygenase pathway and the lipoxygenase pathway may have evolved independently and then integrated into the arachidonate cascade.

View larger version (9K): [in this window] [in a new window]   Fig. 6. Molecular evolution of prostanoid receptors. Phylogenetic tree of prostanoid receptors was constructed by sequence comparison of prostanoid receptors from various species.

	III. PROPERTIES OF PROSTANOID RECEPTORS

Top Previous Next References

Each of the eight types and subtypes of prostanoid receptors shows selective ligand-binding specificity that distinguishes it from the others. In previous studies, this specificity was mostly characterized by the selective responses of tissues to prostanoids and their analogs in a group of bioassay systems. Indeed, each of the prostanoid receptors was initially identified by its preferential responsiveness to a particular type of naturally occuring prostanoid and was subsequently characterized by various synthetic prostanoid analogs that have been synthesized in an attempt to selectively mimic or inhibit particular prostanoid actions. The chemical structures of PG analogs frequently used in such analyses are shown in Figure 7. For example, the EP receptor was initially characterized in the guinea pig ileum and fundus, dog fundus, chick ileum, and cat trachea as a site at which PGE₂ showed the most potent agonistic activity among the prostanoids (107). This receptor was then subdivided into two on the basis of its sensitivity to antagonism by SC-19220, one designated as the EP₁ receptor, exerting its action in the guinea pig ileum and the dog and guinea pig fundus, and the other in the cat trachea and chick ileum. The receptor in the cat trachea and chick ileum was then divided into EP₂ and EP₃ on the basis of their different sensitivities to AY23626 (11-deoxy-PGE₀) and sulprostone (45). The fourth receptor, EP₄, was later discovered as a receptor with smooth muscle relaxant effects similar to EP₂ but different from EP₂ on the basis of the weak potency of AH13205 and its antagonism by AH23848 (43). This approach of receptor characterization has thus been quite valid and successful. However, these studies only provided qualitative rather than quantitative information as to the ligand-binding properties of the receptors, firstly because most bioassay tissues contain more than one type of prostanoid receptor, and the compounds tested show summed action on various receptors, and also because the efficacy of action is different in different bioassay systems. Different degrees of responsiveness of the same receptor type in different species were also noted. Cloning of the prostanoid receptors has enabled the homogeneous expression of each type of receptor from the same species and made the evaluation of ligand-binding characteristics of each receptor as well as the cross-reactivity of prostanoid compounds over several types of receptors possible. Kiriyama et al. (109) used cultured cells expressing each of the eight types of mouse prostanoid receptors to examine the binding affinities of 33 prostanoids and their analogs to each receptor by determining the inhibition constants (K_i) values for the specific radioligand binding to the receptor. Such systematic analyses are not available for the receptors from other species. However, similar K_i values for radioligand receptor binding have been shown or can be calculated for some of the human receptors. These results are summarized below and are shown in Table 1.

View larger version (27K): [in this window] [in a new window]   Fig. 7. Structures of synthetic prostanoid analogs. A: PGD2 and DP ligands. B: PGE2 and EP ligands. C: PGF2 and a FP ligand. D: PGI2 and IP ligands. E: TxA2 and TP ligands. Binding affinities of these compounds to 8 types of mouse prostanoid receptors are shown in Table 1.

The mouse DP receptor showed an affinity for ligands in the order of PGD₂ > BWA868C, BW245C > STA₂. Their K_i values were 21, 220, 250, and 1,600 nM, respectively. This order of affinity differs from the properties reported for the cloned human DP receptor, which showed almost equal ligand binding affinities for PGD₂, BW245C, and BWA868C at 1.1, 0.9, and 1.7 nM, respectively (21). On the contrary to this difference in binding affinities between the two species, agonist potencies of BW245C and PGD₂ are almost the same. They both act as full agonists for the DP receptor, and their EC₅₀ values for cAMP elevation were 0.54 and 6.8 nM, respectively, for the mouse receptor and 0.7 and 6 nM, respectively, for the human receptor. A presumed DP receptor antagonist, BWA868C, can evoke a limited response, indicating that this compound is a partial agonist. Consistent with its K_i value for the human receptor, BW868C showed a pK_B of 8.7 for BW245C-induced relaxation in the rabbit jugular vein (67). Thus the DP receptor only binds its own putative ligands with high affinity. The binding affinities of other prostanoids and their analogs are at least more than two orders of magnitude lower than these compounds. On the contrary, PGD₂ bound to the mouse FP receptor with an affinity comparable to that for the mouse DP receptor, a K_i value of 47 nM (Table 1), indicating that PGD₂ may act on the FP receptor. In fact, PGD₂-induced bronchoconstriction in the anesthetized dog has been suggested to be mediated by the FP receptor (42).

The rank order of affinity for the mouse EP₁ receptor was 17-phenyl-PGE₂, PGE₂, sulprostone, iloprost > PGE₁ > misoprostol, M&B-28767 > 11-deoxy-PGE₁ > PGF₂. Their K_i values were 14, 20, 21, 21, 36, 120, 120, 600, and 1,300 nM, respectively (Table 1). Some EP agonists such as 16,16-dimethyl-PGE₂, GR-63799X, butaprost, 1-OH-PGE₁, and AH13205 did not show any significant binding to this receptor. SC-19220 and AH-6809, known as antagonists for the EP₁ receptor (44), showed no affinity for the mouse EP₁ receptor. The human EP₁ receptor bound PGE₂ with a dissociation constant (K_d) value of 1 nM, and bound other PG in the rank order of PGE₂ > PGE₁ > PGF₂ >> PGD₂ (63). The human receptor also bound AH6809 with a calculated K_i of 333 nM. Weak affinities were also noted for SC19920 and butaprost; the respective K_i values were calculated at 4.5 and 33 µM. This species difference is important as these compounds are frequently used to determine if a action of PGE is mediated by the EP₁ receptor. It should also be mentioned that SC-19920 has a procainelike local anesthetic action (181). It is also noteworthy that 17-phenyl-PGE₂, which is considered to be a relatively specific agonist for the EP₁ receptor, bound to the mouse EP₃ receptor with a higher affinity than to the EP₁ receptor (Table 1).

The rank order of affinity of the EP ligands for the mouse EP₂ receptor was PGE₁, PGE₂ > 16,16-dimethyl-PGE₂ > 11-deoxy-PGE₁ > butaprost > AH13205, misoprostol > AH-6809. Their K_i values were 10, 12, 17, 45, 110, 240, 250, and 350, respectively (Table 1). In addition, this receptor bound with low affinity to two TP ligands, I-BOP and STA₂, and one IP ligand, isocarbacyclin, with low affinity; their K_i values were 220, 220, and 1,000 nM, respectively. 19R(OH)-PGE₂ was reported to be a specific agonist for the EP₂ receptor (248). However, this ligand showed no affinity for the mouse EP₂ receptor and had only a weak affinity for the FP receptor. Butaprost showed affinity only to the EP₂ receptor, indicating its high selectivity for this receptor. No significant binding was observed with other EP agonists such as 17-phenyl-PGE₂, sulprostone, M&B-28767, GR63799X, or 1-OH-PGE₁. The human EP₂ receptor binds PGE₂ and PGE analogs similarly to the mouse receptor, with a rank order of PGE₂ > PGE₁ > 16,16-dimethyl-PGE₂ > 11-deoxy-PGE₁ > butaprost > AH13205, 19R(OH)-PGE₂>1-OH-PGE₁, M&B-28767 > sulprostone = 0. Prostaglandin E₂, 1-OH-PGE₁, AH13205, and butaprost work as full agonists of the human receptor with EC₅₀ values of 43, 2,000, 3,100, and 5,800 nM, respectively.

The mouse EP₃ receptor bound most of EP ligands with a rank order of affinity of sulprostone, M&B-28767, PGE₂, PGE₁, 11-deoxy-PGE₁, GR63799X, 16,16-dimethyl-PGE₂, 17-phenyl-PGE₂ > misoprostol, AH13205 > 1-OH-PGE₁. Their K_i values were 0.60, 0.68, 0.85, 1.1, 1.5, 1.9, 1.9, 3.7, 67, 82, and 330 nM, respectively (Table 1). In addition, the mouse EP₃ receptor bound three IP ligands, iloprost, carbacyclin, and isocarbacyclin, and one TP ligand, STA₂, with K_i values comparable to those for their respective receptors. This receptor also bound two other IP ligands, beraprost and cicaprost, with K_i values of 110 and 170 nM, respectively. Furthermore, this receptor bound PGF₂, I-BOP, and PGD₂ with K_i values of 75, 100, and 280 nM, respectively. These findings are in good agreement with the reported agonist order of potency of some of these compounds in rabbit cortical collecting tubule cells: PGE₂, PGE₁, 16,16-dimethyl-PGE₂ > carbacyclin, PGF₂ > PGD₂ (211). Although sulprostone, M&B-28767, 16,16-dimethyl-PGE₂, and 11-deoxy-PGE₁ also bound to other receptors, they showed the highest affinities for the EP₃ receptor (Table 1). Sulprostone showed affinities for both the EP₁ and FP receptors. M&B-28767, which is known as an EP₁ and EP₃ receptor agonist, also bound to the FP receptor with a K_i value of 124 nM. 16,16-Dimethyl-PGE₂ bound to the EP₂ and EP₄ receptors with the highest affinity out of all the PGE analogs. 11-Deoxy-PGE₁ showed affinities to the EP₂, EP₄, and FP receptors. Misoprostol, known as an EP₂ and EP₃ receptor agonist, showed K_i values of 118, 254, 66.8, and 66.8 nM for the EP₁, EP₂, EP₃, and EP₄ receptors, respectively (Table 1). AH13205, a known EP₂ agonist, also bound to the EP₃ receptor with a higher affinity. On the other hand, GR63799X showed high affinity only to the EP₃ receptor, indicating its high selectivity for this receptor. The human EP₃ receptor binds PGE₂ and M&B-28767 with K_d and K_i values of 0.7 and 0.2 nM, respectively (3). Curiously, this receptor also binds AH6809 with a calculated K_i of 1.3-4.7 nM. This is in contrast to the mouse EP₃ receptor that does not show binding affinity to AH6809.

The rank order of affinity of the ligands for the mouse EP₄ receptor was PGE₂, PGE₁ > 11-deoxy-PGE₁, 16,16-dimethyl-PGE₂, misoprostol > 1-OH-PGE₁, GR63799X, M&B-28767 > 17-phenyl-PGE₂. Their K_i values were 1.9, 2.1, 23, 43, 67, 190, 480, 500, and 1,000 nM, respectively (Table 1). This rank order is in good agreement with previously reported findings. For example, the rank order of potency for the EP₄ receptor in the fetal rabbit ductus arteriosus was PGE₂ >> misoprostol > GR63799X >> AH13205, and the equieffective molar ratios of these ligands were 1, 145, 685, and >100,000, respectively (209). The values for the same ligands for the mouse EP₄ receptor were 1, 59, 294, and >2,000, respectively (Table 1). In addition to these EP ligands, STA₂ bound to this receptor with a K_i value of 350 nM. As for the human EP₄ receptor, only qualitative information based on a radioligand displacement experiment is available (13). This experiment showed the equal binding affinities of the human EP₄ receptor to PGE₂ and PGE₁ and a relatively high affinity to M&B-28767; their respective IC₅₀ values were ~1 and 9 nM, respectively. This receptor showed only very weak affinity to butaprost and AH6809, with IC₅₀ values of >10 µM.

The mouse FP receptor only bound to PGF₂ and fluprostenol with high affinity; their K_i values were 3.4 and 3.7 nM, respectively. Some prostanoids can cross-react with this receptor, but with at least a 10-fold lower affinity than the above two compounds, with a rank order of PGD₂, 17-phenyl-PGE₂ > STA₂, I-BOP, PGE₂, M&B-28767 > 16,16-dimethyl-PGE₂, sulprostone > U-46619, 19R(OH)-PGE₂. Their K_i values were 47, 60, 97, 100, 100, 124, 350, 580, 1,000, and 1,000 nM, respectively (Table 1). Fluprostenol could only bind to the FP receptor, indicating the high selectivity of this ligand. The result that a variety of non-FP ligands show relatively high binding affinities for this receptor indicates that the ligand-binding specificity of the FP receptor is broader than previously suspected. This is more marked in the human FP receptor than in the mouse and showed a similar rank of binding affinity of PGF₂, fluprostenol > PGD₂ > PGE₂ > U-46619 > iloprost, with respective calculated K_i values of 2.1, 2.7, 5.4, 65, 112, and 920 nM, respectively (2).

The rank order of affinity of the ligands for the mouse IP receptor was cicaprost, iloprost, isocarbacyclin > beraprost, PGE₁ > ONO1301 > carbacyclin > 11-deoxy-PGE₁. Their K_i values were 10, 11, 15, 16, 33, 47, 110, and 1000 nM, respectively (Table 1). A similar rank order of binding affinity was found in the human IP receptor, where iloprost >> carbacyclin > PGE₂ >> PGD₂, PGF₂, and U-46619 (20, 103). This is in good agreement with the reported rank order of potency of ligands, cicaprost, iloprost > carbacyclin, in platelets from several species (10). Isocarbacyclin, beraprost, and ONO-1301 also showed high affinities for the IP receptor, as previously reported (96, 228, 249). Interestingly, all of the IP ligands used in this study bound to the EP₃ receptor with K_i values ranging from 22 to 740 nM (Table 1). Among these ligands, iloprost, carbacyclin, and isocarbacyclin showed affinities comparable to those found for the IP receptor. This result suggests the possibility that IP ligands act on the EP₃ receptor. Cross-reaction of IP ligands on the EP₃ receptor was recently suggested in the presynaptic EP₃ receptor in guinea pig vas deferens (227). In fact, carbacyclin has been reported to act on the EP₃ receptor (211). Only iloprost could also bind to the EP₁ receptor (Table 1); the actions of this compound on the EP₁ receptor have already been reported (53).

The rank order of affinity of ligands for the mouse TP receptor was I-BOP, S-145 > GR32191, SQ29548, STA₂ > U-46619. Their K_i values were 0.56, 0.68, 13, 13, 14, and 67 nM, respectively (Table 1). This rank order and K_i values correspond well to previously reported results. For example, Morinelli et al. (145) reported a rank order of I-BOP > SQ29548 > STA₂ > U-46619, with respective IC₅₀ values of 2.2, 4.7, 17, and 62 nM in ligand-binding competition experiments on human platelets. Other ligands known to act on other types of prostanoid receptors had no affinity for the TP receptor, except for M&B-28767; M&B-28767 bound to the receptor with a K_i value of 1,300 nM. It has been reported that PGD₂ and PGF₂-induced bronchoconstriction in humans is mediated by the TP receptor (46). It has also been reported that PGF₂ and PGE₂ contract the rat aortic ring via the TP receptor (56). However, PGD₂, PGF₂, and PGE₂ showed no affinity for the TP receptor in the mouse. Thus the TP receptor is quite specific for putative TP ligands. On the other hand, STA₂ bound to the EP₃, EP₂, and EP₄ receptors with K_i values of 23, 220, and 350 nM, respectively (Table 1), and I-BOP bound to the FP, EP₃, and EP₂ receptors with K_i values of 100, 100, and 220 nM, respectively (Table 1). Although there have been no reports stating that TP ligands act on these receptors, these results should be taken into consideration when performing experiments using these compounds.

Signal transduction pathways of prostanoid receptors have been studied by examining agonist-induced changes in the levels of second messengers (cAMP, free Ca²⁺, and inositol phosphates), and by identifying G protein coupling by various methods. These results are summarized in Table 2. These studies, which combined the results from cultured cells expressing individual cloned prostanoid receptors and those obtained from native receptors in tissues, not only confirmed the previous biochemical findings in crude systems, but also revealed several novel characteristics of the prostanoid receptors.

Several species of G proteins have been reported to participate in signaling via the TP receptor. These proteins include G_q (204), G_q and an 85-kDa unidentified G protein (113), G_q and G_i2 (240), and G₁₂ and G₁₃ (166). There have been several pharmacological studies suggesting the heterogeneity of TP receptors, not only among various tissues (73, 168) but also within a single cell type, such as in the blood platelet (54, 223). The above observations, together with the existence of isoforms of the TP receptor described in the previous section, may explain the multiplicity of signal transduction pathways that are activated via this receptor. It is interesting in this respect that the recently identified TP receptor mutant in human platelets with a point mutation at Arg-60 (81) cannot induce aggregation but can induce shape change in platelets and the activation of phospholipase (PL) A₂ (65). Hirata et al. (82) examined the signal transduction pathways of the two splicing isoforms of TP. They found that both isoforms couple to PLC activation equally well, but couple oppositely to adenylate cyclase. One isoform, TP, activates adenylate cyclase, whereas the other, TP, inhibits adenylate cyclase. The Arg60Leu mutation impairs PLC activation by both isoforms; it impairs adenylate cyclase activation by TP but retains the ability of TP to inhibit the cyclase. On the basis of these findings, Hirata et al. (82) suggested that the pathway linked to adenylate cyclase inhibition, or other pathway(s) not affected by the above mutation of TP, may be involved in some of the TP-mediated platelet responses, such as shape change and PLA₂ activation. The cloned FP receptor is also coupled to the activation of PLC via G_q. Functional coupling of the FP receptor with G_q was shown by an experiment using anti-G_q antibodies (92). In NIH 3T3 cells, PGF₂ induces DNA synthesis through this pathway (246). Although coupling was not observed with G_i or G_s in FP receptor-transfected Chinese hamster ovary (CHO) cells, PGF₂ has been shown to inhibit gonadotropin-stimulated cAMP formation in luteal cells (232).

The species of G protein to which EP₁ receptors are coupled remains unidentified. The EP₁ receptor mediates PGE₂-induced elevation of free Ca²⁺ concentration in CHO cells. This increase in free Ca²⁺ concentration was dependent on the availability of extracellular Ca²⁺ and accompanied by a barely detectable PI response (244). This observation suggests that the EP receptor may regulate Ca²⁺ channel gating via an unidentified G protein. The EP₂ and EP₄ receptors are coupled to G_s and mediate increases in cAMP concentration. The major signaling pathway of the EP₃ receptor is inhibition of adenylate cyclase via G_i. However, it is noteworthy that the splice variants of the EP₃ receptor described above are coupled to different signaling pathways and that one of these isoforms negatively regulates G protein activity. For example, the bovine EP₃A receptor is coupled to G_i and induces the inhibition of adenylate cyclase, whereas the EP₃B and EP₃C receptors are coupled to G_s and act to increase levels of cAMP. The EP₃D receptor is coupled to G_q, in addition to G_i and G_s, and evokes a pertussis toxin-insensitive PI response (154). Moreover, the bovine EP₃C receptor has been seen to demonstrate a novel type of receptor-G protein interaction, in addition to the conventional stimulation of G_s. When an agonist is bound to this receptor, the activity of G_o is directly inhibited due to an apparent increase in its affinity for GDP but not for GTP (160). The finding that the carboxy-terminal tail of a receptor has an important role in determining G protein coupling specificity explains the previously reported multiplicity of signal transduction pathways that reportedly operate via the EP₃ receptor (44, 69, 159). Such mechanisms where G protein specificity is determined by the carboxy-terminal tail may also work in the signal transduction of other rhodopsin-type receptors.

The IP receptor has been known to stimulate adenylate cyclase. However, expression studies revealed that it mediates not only a rise in cAMP levels but also PI responses (152). Prostaglandin I₂ has been reported to induce the elevation of free Ca²⁺ concentration in several lines of cultured cells (243, 247). IP receptor-induced PI responses in CHO cells were not inhibited by either pertussis toxin or cholera toxin, suggesting that the G_q family of G proteins is likely to be participating in this response (152). The DP receptor is coupled to G_s and mediates increases in cAMP concentrations. No PI response was observed in DP receptor signaling (21, 80), although the stimulation of the human DP receptor expressed in HEK 293 cells induced a transient increase in intracellular free Ca²⁺ concentration possibly via a cAMP system (21).

Curiously, in the expression systems of cloned receptors, some prostanoid-evoked responses occur at much lower ligand concentrations compared with their K_d values. For example, iloprost-stimulated elevations in cAMP levels, in IP receptor-transfected CHO cells, had an EC₅₀ value of 100 pM (152). Similarly, PGE₂ decreases cAMP levels in EP₃ receptor-transfected CHO cells with an IC₅₀ value of 100 pM (219). These values are 45- and 30-fold lower, respectively, than their K_d values for binding. The reason for this discrepancy is not clear, but it may reflect differences in the efficacy of postreceptor signal transduction mechanisms among various cells. In fact, variations in coupling efficacy of the IP receptor were observed in platelets of various species (10). In canine cortical collecting tubule cells, picomolar concentrations of PGE₂ antagonize vasopressin actions and induce a decrease in cAMP levels (177). Such differences in the efficacy of signaling are also seen between different signal transduction pathways via the same receptor. For example, the EC₅₀ value of an iloprost-stimulated PI response in an expression system was 100 nM, which is three orders of magnitude higher than that of iloprost-stimulated elevations of cAMP levels.

Domains and amino acid residues involved in ligand binding and signal transduction have been examined by creating mutant receptors by site-directed mutagenesis. Particular attention has been paid to the residues conserved in the prostanoid receptors and in the rhodopsin-type receptors in general. As descrived above, the seventh transmembrane region of the prostanoid receptors has a highly conserved motif, i.e., L-X-A-X-R-X-A-S/T-X-N-Q-I-L D-P-W-V-Y-I-L-X-R. Funk et al. (64) examined the role of the three amino acid residues, L291, R295 and W299, in this region of the human TP. The point mutants TP-W299L, R295Q, W299R and L291F all lost their binding to the antagonist [³H]SQ-29548. In addition, three of the mutants, R295Q, W299R, and L291F, also failed to show binding to the agonist I-BOP, whereas W299L showed binding to I-BOP and another agonist, U-46619, with affinities indistiguishable from the wild-type receptor. These results demonstrated the importance of the seventh transmembrane domain in ligand binding to the thromboxane receptor and indicated that agonists and antagonists may be recognized differently. The importance of the arginine and other charged residues in the seventh transmembrane domain was examined also in the EP₃ receptor (11, 87). One study (11) showed that mutations of this arginine of rabbit EP₃, R329A and R329E, both abolished [³H]PGE₂ binding to the receptor, whereas the mutation of an aspartic acid in this region, D338A, showed no alteration in ligand-binding properties. On the other hand, the D338A mutant was defective in signal transduction, showing no decreases in cAMP when activated with up to a 1 µM concentration of sulprostone, an EP₁/EP₃ agonist. The other study (87) showed that mutation of the corresponding arginine residue (R309) of mouse EP₃ to glutamate or valine also led to loss of the ligand binding, whereas mutation of this residue to lysine did produce the binding with higher affinity. These results indicate that the seventh transmembrane domain is involved in both ligand binding and transduction processes. The R295 of human TP, the R329 of rabbit EP₃, and the R309 of mouse EP₃ corrrespond to the arginine conserved by all the prostanoid receptors. Because modification of the carboxylic group of prostanoid molecules usually reduces the agonistic potencies of these compounds (see, for example, Ref. 201), discussion of these results led to the suggestion that this conserved arginine makes a charge-charge interaction with the carboxy moiety of the prostanoid ligand. However, the affinities of carboxy methyl esters of prostanoid molecules to their receptors differ from one receptor to another, and even in the same receptor from one species to another. For example, the K_i value of PGE₂ methyl ester for [³H]PGE₂ binding to rabbit EP₃ is 1,600 nM, which is 1,000 times higher than that of PGE₂, 1.6 nM. On the other hand, its K_i value of the mouse EP₃ receptor is only 10 times higher than that of PGE₂ in the same receptor. These findings are difficult to interpret if the difference in the binding affinities between the free carboxylic acid and methyl ester is determined solely by interaction of the respective group with the arginine residue. This issue has been clarified by the work of Audoly and Breyer (12), in which they introduced point mutations into another conserved motif in the second extracellular loop of the rabbit EP₃ receptor and examined binding properties of the mutant receptors. When the first tryptophan (W199) or threonine residue (T202) in the sequence of Q¹⁹⁸-W-P-G-T-W-C-F was replaced by alanine, the affinity for carboxy methyl esters of PGE derivatives was increased by up to 128-fold, and the selectivity ratios comparing the K_i values of the methyl esters with the free carboxylic acids were reduced greatly from a few hundred times to ~10 times or less. The authors discussed these findings saying that the preference for free carboxylate derivatives is primarily determined by this region in the second extracellular loop, which may work itself as a part of a ligand-binding domain or as a filter or bait to the transmembrane binding pocket.

What then is the role of the conserved arginine in the seventh transmembrane region? Chang et al. (34, 35) pointed out that an arginine residue can form both ionic bonding and hydrogen bonding, acting as a hydrogen donor, and they evaluated the contribution of these two types of bonding by the conserved arginine, on ligand binding and transduction of the EP receptors. In one experiment, they examined the binding and potency of three compounds, PGE₂, PGE₂ methyl ester, and 1-OH-PGE₂ to the EP receptors. These compounds act as a negative charge, a hydrogen acceptor, and a hydrogen donor, respectively. They found that PGE₂ and the PGE₂ methyl ester showed almost the same potency to the four EP receptors, whereas the potency of 1-OH-PGE₂ was much lower, indicating that hydrogen bonding is enough for an agonist to exert its action. They also tested this hypothesis by mutating this charged arginine either to the noncharged but polar Gln or Asn, or to the nonpolar Leu, and examining the binding and potency of PGE₂ and its analogs to these mutant receptors. The affinity of PGE₂ was hardly affected by the mutations to Gln or Asn but decreased by ~40 times by the Leu substitution. These results suggest that, although the arginine can provide both ionic interactions and hydrogen bonding, the ionic interaction is not essential and the hydrogen bonding alone can support ligand binding. These results also indicate that some PG analogs with carboxy modifications interact with the receptors only via hydrogen bonding. Based on this hypothesis, Negishi et al. (158) compared the potencies of PGE₂ and sulprostone in the signal transduction of EP₃ to G_i and G_s. Although sulprostone showed equal potency to PGE₂ for G_i activation, its potency to activate G_s was more than 10 times weaker than PGE₂. Moreover, sulprostone showed the lower binding affinity to the G_s-coupled EP₃ receptor than to the G_i-coupled receptor, and PGE₂ failed to bind to the G_s-coupled EP₃D-R332Q mutant receptor and to activate its pathway. Although these authors indicate from these findings that the hydrogen bonding interaction may not be enough for signal transduction to G_s and suggest that the ligand-binding properties of the prostanoid receptors can be different depending on the species of G protein coupled to the receptor, the difference of the reactivity between PGE₂ and sulprostone to the G_s-coupled EP₃ receptor may be caused by interactions other than the C₁ moieties of these molecules and the arginine residue in the seventh transmembrane domain, because the sulfonamide of the C₁ moiety of sulprostone can be ionized at neutral pH.

Another question raised by several studies is whether the cysteine residue in the second extracellular loop forms a disulfide bond important for receptor structure. Audoly and Breyer (12) substituted alanine for Cys-204 in this region of rabbit EP₃ and found no change in the receptor's binding affinities for PGE₂ and its analogs. This is in contrast to the findings reported for TP (36, 47). In these studies, serine was substituted for different cysteine residues in the human TP receptor. Among the substitutions, substitution of Cys-105 in the first extracellular loop and Cys-183 in the second extracellular loop (which is at an analogous position to Cys-204 in rabbit EP₃) completely abolished ligand-binding activity. Because Cys residues at analogous positions are proposed to make a disulfide bond in other rhodopsin-type receptors (52) and because the TP receptor loses ligand-binding activity after reduction with dithiothreitol or sulfhydryl alkylation (55), these authors suggested that Cys-105 and Cys-183 make an essential disulfide bond. The authors also found that mutation of Cys-102 also affected the ligand-binding activity. In addition, D'Angelo et al. (47) further found that the Cys223Ser substitution retained ligand binding but abolished agonist-induced Ca²⁺ mobilization activity. Because this cysteine residue is not conserved by most of the other prostanoid receptors, the implication of this binding in the G protein coupling of other receptors remains obscure. On the other hand, the involvement of a single conserved amino acid residue in the G protein coupling of a prostanoid receptor has become apparent from a study on an inherited disorder. Hirata et al. (81) analyzed a hereditary bleeding disorder and found that it was associated with the Arg-60 to Leu mutation in the first intracellular loop of the human TP receptor. The receptor with this mutation showed normal binding properties but was defective in PI turnover. A subsequent study by the same authors (82) revealed that this mutation affected the PI turnover mediated by G_q in both TP receptor isoforms but did not inhibit the G_i-mediated decrease of adenylate cyclase in the -isoform of this receptor, suggesting that this region is involved in coupling with G_q but not with G_i. This arginine is conserved at analogous positions in all of the prostanoid receptors. Which type of G protein coupling this arginine residue is involved with in each receptor is not known at present.

As described, each type and subtype of the prostanoid receptors shows specific ligand-binding properties, and the same receptor from different species sometimes shows different binding properties. Domains conferring ligand-binding specificity have been examined in chimeric receptors composed of two receptors with different selectivities. As described above, IP shows high-affinity binding to prostacyclin analogs such as iloprost and carbacyclin as well as PGE₁, but not to PGE₂ or other types of prostanoids, whereas DP shows selective binding to the type D PG. Kobayashi et al. (114) constructed chimeric DP/IP receptors and examined the domains conferring specificity to each receptor. The binding specificity of IP was found to be determined by recognition of both the ring structure and the side chain configuration and that the latter is primarily recognized by the sixth and seventh transmembrane regions, whereas the domain that recognizes the former seems to be located elsewhere and can accomodate the rings not only of PGI and PGE, but also of PGD. On the other hand, selective binding of DP appears to be determined by the first three transmembrane regions. A similar line of study was published recently by Kedzie et al. (105). These authors also made use of a high degree of homology between the relaxant group of prostanoid receptors. They introduced point mutations to amino acid residues conserved in the EP₂ and EP₄ receptor but not in the IP receptor and examined the residues conferring responsiveness to IP ligands. They found that a Leu304Tyr mutation in the seventh transmembrane region of the EP₂ receptor enhanced the potency of iloprost ~100-fold, almost equal to that of PGE₁. This may be consistent with the above proposal by Kobayashi et al. (114) that the sixth and seventh transmembrane regions are responsible for accomodation of the -chain of IP ligands. A question still remains, however, as to the mechanism of the selectivity of the IP receptor, because IP can bind both iloprost and PGE₁ but not PGE₂, while their mutant receptor binds PGE₂ more preferentially than iloprost and PGE₁. Chimeric receptors have also been used to examine the amino acid residues responsible for ligand-binding difference among the receptors from different species. The rat TP binds the agonist I-BOP with about a 10-fold higher affinty than human TP. Dorn et al. (58) constructed chimeric rat/human TP receptors and examined the domain determining this property. They found that the portion from the amino terminus to an area in the first transmembrane regio