A Review of Prostanoid Receptors: Expression, Characterization, Regulation, and Mechanism of Action

Abstract

Prostaglandin signaling controls a wide range of biological processes from blood pressure homeostasis to inflammation and resolution thereof to the perception of pain to cell survival. Disruption of normal prostanoid signaling is implicated in numerous disease states. Prostaglandin signaling is facilitated by G-protein-coupled, prostanoid-specific receptors and the array of associated G-proteins. This review focuses on the expression, characterization, regulation, and mechanism of action of prostanoid receptors with particular emphasis on human isoforms.

Introduction

Prostaglandins are amphipathic, bioactive signaling molecules derived from the oxidation of arachidonic acid. They are involved in a wide range of roles from homeostasis of blood pressure and blood flow, the initiation and resolution of inflammation, the perception of pain, cell survival, and the progression of numerous disease states. These biomolecules act most often as autocrine or paracrine signaling agents and most have relatively short half-lives. Prostanoid signaling is accomplished through specific G-protein coupled receptors (Figure 1). With the exception of a few isoforms with unknown biological function, all known biologically competent receptors are heptahelical, multi-pass membrane proteins, members of the G-protein coupled receptor 1 family (GPCR) and are among the most abundant membrane proteins (Binda et al. 2004). The diversity of prostanoid action is not only defined by specific receptors and their regulated expression, but also to the diversity of G-proteins that most receptors are able to couple to, leading to actuation of different signaling pathways by the same receptor. Although the actions of these receptors are diverse, many show commonalities in their regulation. Each display agonist-induced desensitization that is usually found to be associated with receptor phosphorylation by various protein kinases. Phosphorylation by G-protein receptor kinases (GRK) can lead to arrestin binding that promotes receptor uptake into clatherin-coated pits leading to sequestration into punctate vesicles. This review focusses primarily on human receptors but discusses receptors from other species when information about the human receptors is lacking.

Fig. 1.

Prostanoid metabolic pathway. Gene designations are given for the participating enzymes (rounded boxes), accepted acronyms given for metabolites (ovals), and accepted acronyms for receptors (grey boxes). Receptor binding only indicated for metabolites with an EC₅₀ in the nM range. Enzymes: AKR1B1, aldo-keto reductase 1B1; AKR1C3, aldo-keto reductase 1C3; CBR1, carbonyl reductase 1; FAM213B, prostamide/prostaglandin F synthase; HPGDS, Hematopoietic prostaglandin D synthase; HSA, human serum albumin; PTGDS, prostaglandin D₂ synthase; PTGES, prostaglandin E synthase; PTGES2, prostaglandin E synthase-2; PTGES3, prostaglandin E synthase-3; PTGIS, prostacyclin (PGI₂) synthase; PTGS1, Prostaglandin G/H Synthase 1; PTGS2, Prostaglandin G/H Synthase 2; TBXAS1, Thromboxane A Synthase 1. Note: Figure 1 is in color

Prostaglandin D₂ Receptors

Introduction

There are two distinct types of prostaglandin D₂ receptors found in humans, the prostanoid DP1 receptor (PGD receptor, PGD2 receptor, Gene: PTGDR) and the DP2 receptor (CRTH2, G-protein coupled receptor 44, CD294, Gene: PTGDR2). Both are coupled to G-proteins and upon binding to PGD₂ or its PGJ₂ series metabolic products, they affect intracellular concentrations of cAMP and Ca²⁺ concentrations either directly or indirectly depending on the receptor. Activation is responsible in part for immune regulation, allergic/inflammation responses, mobilization of dendric cells, and impaired PGD₂-induced sleep to name a few. Although the DP1 amino acid sequence is closely related to other prostaglandin receptors it shares only a 22% percent identity with DP2 which is more closely related to chemoattractant receptors such as the N-formylmethionly-leucyl-phenylalanyl (fMLP) and anaphylatoxin C3a receptors (Wang et al. 2018; Hirai et al. 2003).

DP1 receptor

Introduction

The human prostaglandin D₂ receptor DP1 (PTGDR, UniprotKB-Q13258) is translated as a 359 amino acid polypeptide with a calculated molecular weight of 40.3 kDa. There is one additional reported isoform based only on expressed sequence tag (EST) data (Q13258-2) and four coding single nucleotide polymorphic (SNP) variants (R7C, G198E, E301A, and R332Q) (https://genecards.org, Stelzer et al. 2016), none of which involve any known clinical significance or condition. There are no reported X-ray structures, but there are several G-protein coupled receptor database (GPCRdb, https://gpcrdb.org, Pándy-Szekeres et al. 2018) models proposed based on similar proteins (PDB reference 4UHR and 3VG9). Morii and Watanabe (1992) examined the effect of a variety of exo- and endo-glycosidases on the binding of PGD₂ to the DP1 receptor on P₂ membrane prepared from porcine temporal cortex. They found that the exoglycosidases neuramidase, α-manosidase, β-galactosidase, and β-N-acetylhexoamidase reduced the binding of PGD₂ by 6%, 33%, 62%, and 66% respectively, indicating that the integrity of the glycosyl chains was important for ligand binding. Further, Scatchard analysis of the data revealed that modification of the glycosyl chain reduced the affinity of the DP1 receptor for the ligand rather than the maximal binding capacity. Treatment of the P₂ membrane with endoglycosidase N-glycohydrolase F which cleaves the N-glycosyl linkage between the carbohydrate and protein reduced PGD₂ binding by 66%, confirming the presence of N-linked carbohydrate. Extracellular N-linked glycosylation is predicted for N10 with high probability by NetNGlyc analysis (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004). Extracellular N-linked glycosylations are also predicted for N90, and N297 by sequence analysis, but none of these predictions have yet to be confirmed experimentally (Apweiler et al. 2017). Interestingly, treatment with endo-α-N-acetylgalactosamidase, which cleaves only O-linked carbohydrate from proteins, reduced PGD₂ binding by 52%, confirming the presence of O-linked glycosyl chains as well. NetOGlyc analysis (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013) predicts potential O-glycosylation at S238, S254, T240, and S350, all of which are located on the cytoplasmic side of the membrane. Here and throughout this document N-linked and O-linked glycosylation is predicted with these web services. Although the data is strongly suggestive of the presence of both O- and N-glycosylations on the DP1 receptor, one cannot rule out the possibility that at least a portion of the glycosylations may be present on a yet-to-be identified protein associated with the DP1 receptor that served modify the binding affinity for PGD₂.

There are numerous potential phosphorylation sites on the human DP1 based on motifs but none are specifically confirmed experimentally. Utilizing the NetPhos 3.1 server (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999), a conservative minimum score of 0.9 and the availability of sites based on topology predictions (UniProtKB), nine potential sites of phosphorylation are predicted: S145 and S300 on extracellular loops, S254 on a cytoplasmic loop, and S347 on the cytoplasmic, C-terminal domain. There are also additional potential G protein-coupled receptor kinases (GRK) phosphorylation sites on the C-terminal domain predicted by the GPS server (http://gps.biocuckoo.cn/, Xue et al. 2011): S347, S350, S352, T353, S357, and S358. There is experimental evidence for GRK-mediated phosphorylation of the N-terminal of the receptor (see below) (Gallant et al. 2007). The aforementioned methodologies for phosphorylation prediction are used throughout the remaining document.

Expression and Characterization

The DP1 receptor is expressed primarily in the gastrointestinal tract, bone marrow and the immune system, and the gall bladder, but is found in many other tissues such as the retinal choroid, ciliary epithelium, longitudinal and circular ciliary muscles, iris, and brain cerebral cortex (Town et al. 1983; Sharif et al. 2000a; http://www.proteinatlas.org, Uhlén et al. (2015)). Although found in many tissues, this receptor is the least abundant of the prostanoid receptors (Boie et al. 1995).

Ligand binding properties for DP1 have been characterized with human recombinant DP1 expressed in human embryonic cell line HEK293 (Wright et al. 1998; Sawyer et al. 2002), mammalian COS-M6 cells (Boie et al. 1995), and a human immortalized myelogenous leukemia cell line K562 (Hirai et al. 2001), as well as human platelets (Table 1). Binding properties for recombinant mouse DP1 expressed in HEK293 cells (Hata et al. 2005; Hata et al. 2003) as well as combined DP in bovine embryonic tracheal cells have also been examined (Sharif et al. 2003). Although prostanoid receptors exhibit preferences with respect to ligand binding, promiscuous cross reactivity has been reported for several receptors (Abramovitz et al. 2000; Narumiya et al. 1999). Both ligand affinities and efficacies have been examined. [H³]-PGD₂ equilibrium competition assays provide a good measure of relative ligand binding affinities (Table 1). Binding affinities for recombinant receptor show the highest affinity while binding to human platelets is the lowest of all. Binding of PGE₂ and PGF_2α is several orders of magnitude poorer than PGD₂ for all systems examined, supporting the selectivity of the receptor. Binding of the non-enzymatic dehydration product of PGD₂, PGJ₂ is 2-10 times weaker than PGD₂. Signal efficacy upon binding has been measured as the extent of induced cAMP synthesis as compared to the maximal amount produced by PGD₂ at saturation and expressed as the concentration of ligand necessary to produce 50% of the maximal PGD₂ levels (EC₅₀). The relative values for the different ligands binding to the same receptor shown in Table 1 compare well with those for the binding assays indicating that affinity is comparable to efficacy.

Table 1.

Binding affinity (K_i) and signaling efficiency (EC₅₀) of prostanoids for the recombinant DP1 and CRTH2 (DP2) receptors expressed in cell cultures and in human platelets

	Ki (nM)
Receptor/cell line	PGD2	PGE2	PGF2α	9α,11β-PGF2α	PGJ2	Δ12-PGJ2	15-deoxy-Δ12,14-PGJ2	PGI2	Reference
hDP2/K562	61 ± 23	31,000 ± 11,000	2,000 ± 780	-	460 ± 160	7,100 ± 1,600	2,300 ± 400	-	Hirai et al. 2001
hDP2/HEK293	2.4 ± 0.7	4,730 ± 64	395 ± 77	315 ± 92	6.6 ± 0.3	6.8 ± 0.35	3.2 ± 0.3	-	Sawyer et al. 2002
mDP2/HEK293	32 ± 6	-	-	-	-	-	-	-	Hata et al. 2005
mDP2/HEK293	38 ± 6	32,000 ± 400	600 ± 50	-	45.7 ± 0.2	410 ± 20	24 ± 4	-	Hata et al. 2003
hDP2/K562	61 ± 23	3,100± 1100	2,000± 780	-	460 ± 160	7,100 ± 1,300	2,300 ± 400		Hirai et al. 2001
hDP1/K562	45 ± 17	2,300± 1100	>30,000	-	64 ± 13	5,200 ± 1,700	30,000	-	Hirai et al. 2001
hDP1/COS M-6a	1.1	101	1,000	-	-	-	-	-	Boie et al. 1995
hDP1/HEK293	0.6 ± 0.2	107 ± 42	367 ± 85	-	6.6± 0.3	100 ± 13	280 ± 30	-	Wright et al. 1998
hDP1/HEK293	1.7 ± 0.3	307± 106	861 ± 139	-	-	-	-	-	Sawyer et al. 2002
DP/human platelets	81 ± 5	>10,000	18,000 ± 6,460	-	-	-	-	3537	Sharif et al. 2003
DP/human platelets	80.3 ± 5.5	11,000 ± 2,190	32,000 ± 3,500	-	-	-	-	-	Sharif et al. 2000a
DP/human platelets	80 ± 5	11,000 ± 2,100	18,000 ± 6,400	-	-	-	-	-	Sharif et al. 2000b
	EC50 (nM)
Receptor/cell line	PGD2	PGE2	PGF2α	9α,11β-PGF2α	PGJ2	Δ12-PGJ2	15-deoxy-Δ12,14-PGJ2	PGI2	Reference
hDP2/HEK293b	1.6± 0.3	-	456 ± 169	856 ± 322	13.1 ± 5.36	10.8 ± 3.4	11.6 ± 6.3	-	Sawyer et al. 2002
mDP2/HEK293	0.7 ± 0.3	-	-	-	-	-	-	-	Hata et al. 2005
mDP2/HEK293	0.9 ± 0.2	-	-	-	-	-	0.8 ± 0.3	-	Hata et al. 2003
hDP1/HEK293	1.38 ± 0.785	206 ± 156	1,090 ± 975	-	-	-	-	-	Suganami et al. 2016
hDP1/HEK293a	10	500	1000	-	-	-	-	-	Boie et al. 1995
hDP1/HEK293	0.5 ± 0.1	84 ± 1	264 ± 22	-	1.2 ± 0.1	91 ± 18	321 ± 15	-	Wright et al. 1998
bDP/EbTr	74	>1,000	>10,000	-	-	-	-	>10,000	Sharif et al. 2003
hDP/human platelets	98 ± 10	>10,000	>10,000	-	-	-	-	-	Sharif et al. 2000b

Abbreviations: h, human; m, mouse; b, bovine; HEK293, human embryonic kidney cells; K562, human erythroleukemia cells; COS M-6, Green monkey kidney cells; EbTr, bovine embryonic tracheal cells. The K_i values were determined from [³H]PGD₂ displacement assays. The EC₅₀ determined from intracellular changes in cAMP concentrations. a) Values determined from estimated IC₅₀ taken from binding curves, utilizing the Cheng-Prusoff equation (Suganami et al. 2016); b) Ca²⁺ concentrations also measured.

Examination of the cellular effects resulting from a particular agonist binding to a specific receptor are frequently discerned through judicious use of highly selective agonists and antagonists for a particular receptor. The most frequently used selective agonists for DP1 are BW245C and TS-022, and there are three commonly used selective antagonists, BWA868C, MK-0524, and S-5751. A listing of agonists and antagonists is given in Table 2.

Table 2.

Prostanoid receptor synthetic agonists and antagonists

Receptor	agonist	reference	antagonist	reference
DP1	BW245C	Rangachari et al. 1995	BWA868C	Rangachari et al. 1995
	TS-022	Arai et al. 2007	MK-0524	Chang et al. 2007
	-	-	S-5751	Arimura et al. 2001
DP2	Indomethacin	Hirai et al. 2002	Ramatroban	Sugimoto et al. 2003
	L-888,607	Gervais et al. 2005	-	-
EP1	ONO-DI-004	Sugimoto and Narumiya 2007	ONO-8713	Sugimoto and Narumiya 2007
	17-phenyl trinor PGE2	Dey et al. 2006	-	-
EP2	ONO-AE1-259	Sugimoto and Narumiya 2007	PF-04418948	af Forselles et al. 2011
	butaprost	Kashmiry et al. 2018	-	-
EP3	ONO-AE-248	Norel et al. 2004	enprostil	Abramovitz et al. 2000
	SC-46275	Savage et al. 1993	M&B 28767	Abramovitz et al. 2000
	MB28767	Kotani et ai. 2000	-	-
	Sulprostone	Jin et al. 1997, Abramovitz et al. 2000	-	-
EP4	ONO-AE1-329	Sugimoto and Narumiya 2007	ONO-AE3-208	Sugimoto and Narumiya 2007
	tetrazolo PGE1	Jones et al. 2009	CJ-042794	Jones et al. 2009
IP	Cicaprost	Jones et al. 2009	RO-1138452	Jones et al. 2009
	Carbacyclin	Jones et al. 2009	RO-3244794	Jones et al. 2009
	iloprost	Jones et al. 2009	-	-
	19(S)-HETE	Tunaru et al. 2016	-	-
TP	U-46619	Abramovitz et al. 2000	AH-23848	Jones et al. 2009
	STA2	Hirata et al. 1996	Ramatroban	Jones et al. 2009
	I-BOP	Hirata et al. 1996	Vapiprost	Jones et al. 2009
	-	-	SQ29548	Wright et al. 1998
	-	-	S-145	Ushikubi et al. 1989b
FP	Fluprostenol	Sharif et al. 2000b	OBE022	Pohl et al. 2018
	Cloprostenol	Anderson et al. 1999	THG113	Peri et al. 2002
	Latanoptost	Abramovitz et al. 2000	AS604872	Jones et al. 2009

The expression of DP1 from post-translational folding and processing in the Golgi to its expression in the plasma membrane has been examined in detail by Binda et al. (2004). Utilizing recombinant proteins expressed in a HEK293 system, they found that lipocalin-type prostaglandin 2 synthase (L-PGDS) (Binda and Parent 2015) helps facilitate the cell surface expression of DP1. Following glycosylation and maturation in the Golgi, DP1 forms a complex with Hsp90 which then interacts directly with L-PGDS as a requirement for cell surface expression, independent of PGD₂ binding. In addition, L-PGDS associates directly with the co-expressed DP1 in the perinuclear region, but not on the plasma membrane, and this association increases the production of PGD₂ by L-PGDS which may represent an intracrine signaling system.

Mechanism of Cell Activation

Signal transduction from the DP1 receptor is reported to occur via a G_αs protein signaling system (see Table 3), resulting in an increase in intracellular concentrations of cAMP (Ito et al. 1990; Boie et al. 1995; Sugama et al. 1989 ; Hirata et al. 1994a; Schratl et al. 2007) which in turn activates protein kinase A (PKA), setting off a number of cellular events. Several studies also report a concomitant increase intracellular Ca²⁺ (Xue et al. 2007; Okuda-Ashitaka et al. 1993), but without an increase in inositol 1,4,5-trisphosphate (IP₃) concentration (Boie et al. 1995). The authors suggest that the increase in intracellular calcium could occur through PKA phosphorylation of the L-type Ca²⁺ channel (LTCC) and the ryanodine receptor (RyR) (Maher et al. 2015; Zaccolo 2009).

Table 3.

Signaling pathways and G-protein association for prostanoid receptors

Receptor/cell line	signaling pathway	G-protein mediation	reference
DP1	cAMP↑ Ca2+↑†	Gαs	Sawyer et al. 2002; Wright et al. 1998
DP2	cAMP↓ Ca2+↑	Gαi	Hirai et al. 2001; Xue et al. 2007
EP1	IP3↑ Ca2+↑	Gαq	Tang et al. 2005; Markovič et al. 2017
	PI3K/AKT/mTOR	Gαi/o	Ji et al. 2010
EP2	cAMP↑	Gαs	Regan et al. 1994a
	IP3↑ Ca2+↑	Gαq/11	Kandola et al. 2014
EP3-I	cAMP↓	Gαi	Kotani et al. 1995
	IP3↑ Ca2+↑	Gαq	Kotani et al. 1995
EP3-II	cAMP↓	Gαi	Kotani et al. 1995; Kotani et al. 1997
	IP3↑ Ca2+↑	Gαq	Kotani et al. 1995; Kotani et al. 1997
	cAMP↑	Gαs	Kotani et al. 1995; Kotani et al. 1997
EP3-III	cAMP↓ Ca2+↑	Gαi	Kotani et al. 1995; Kotani et al. 1997
EP3-IV	cAMP↓	Gαi	Kotani et al. 1995; Jin et al. 1997
	cAMP↑	Gαs	Kotani et al. 1995; Jin et al. 1997
EP3-V	cAMP↓	Gαi	Kotani et ai. 2000
EP3-VI	cAMP↓	Gαi	Kotani et ai. 2000
	cAMP↑	Gαs	Kotani et ai. 2000
EP4	cAMP↑	Gαs	Nishigaki et al. 1996; Regan 2003
	IP3↑ Ca2+↑	Gαq	Fujino et al. 2002
	cAMP↓	Gαi	Leduc et al. 2009
IP	cAMP↑	Gαs	Smyth et al. 1996
	IP3↑ Ca2+↑	Gαq	Smyth et al. 1996
	Ca2+↑†	Gαs	Vassaux et al. 1992
TP
α & β	IP3↑ Ca2+↑	Gαq	Kinsella et al. 1997; Walsh et al. 2000b
α & β	Na+/H+ exchange↑	Gα12,13	Becker et al. 1999
α & β	Ca2+↑†	Gα12,13	Walsh et al. 2000b
α & β	Rho signaling	Gα12,13	Feng et al. 1996
α only	cAMP↑	Gαs	Hirata et al. 1996
α only	IP3↑ Ca2+↑	Gαh	Feng et al. 1996
β only	unknown	Gαh	Feng et al. 1996
β only	cAMP↓	Gαi	Hirata et al. 1996; Ushikubi et al. 1994
FP	IP3↑ Ca2+↑	Gαq	Liang et al. 2008a
	Rho signaling	Gα12,13	Pierce et al. 1999
	cAMP↓	Gαi	Hébert et al. 2005; Ohmichi et al. 1997
	cAMP↑	Gαs	Tachado et al. 1993

† Ca²⁺ is indirectly increased. See text.

Regulation

As is typical for most G-coupled receptors, regulation of the DP1 receptor is accomplished through desensitization and internalization. The regulation of recombinant human DP1 in HEK293 kidney cells with and without recombinant GRKs, arrestins or Ras related proteins (Rab) in the presence or absence of PGD₂ has been examined by Gallant et al. (2007). GRKs are known to mediate G-protein receptor desensitization through phosphorylation of the receptor, leading to high affinity binding to arrestins. Bound arrestins are known to inhibit receptor binding to the associated G-protein and at the same time promote receptor uptake into clatherin-coated pits where the receptors are either degraded or recycled back to the membrane, recycling being controlled by Rabs. When stimulated with 1 μM PGD₂, Gallant et al. (2007) observed that DP1 expressed alone undergoes approximately 25% internalization, leading to a concomitant reduction of activity. When DP1 is co-expressed with GRK2, PGD₂-induced internalization increases by 25% after 2 hours following stimulation but is unaffected by co-expression of either GRK5 or GRK6. In addition, co-expressed GRK2 reduces DP1 signaling as measured by cAMP production by 45% within 15 minutes post-stimulation, whereas co-expressed GRK5 requires 60 minutes to achieve a maximum reduction in cAMP production of 35% and without internalization. Co-expressed GRK 6 has little effect on the signaling. Co-expression with arrestin-3 or arrestin-2 promotes internalization by 53% and 43% respectively. On the other hand, co-expression of Rab4 with DP1 decreases internalization by 64% whereas co-expression with Rab11 has no effect on internalization. These data are consistent with rapid GRK2 phosphorylation of DP1 leading to a slower internalization via arrestins that may be reversed with Rab4, a protein known to be involved in fast endocytic recycling (Wang et al. 1995). Phosphorylation by GRK5 reduces DP1 signaling without subsequent internalization, indicating that the site of phosphorylation may not promote proper arrestin binding but may instead simply reduce G-protein interaction with DP1.

DP2 receptor

Introduction

The human prostaglandin D₂ receptor DP2 (PTGDR2, CRTH2, DL1R, GPR44, UniprotKB-Q9Y5Y4) is translated as a 395 amino acid polypeptide with a calculated molecular weight of 43.3 kDa (Nagata and Hirai 2003). There are no additional reported isoforms and three coding SNP variants reported, V204A (Nagata et al. 1999), F179V, P197T (Hsu et al. 2002) and L281F (https://www.ncbi.nlm.nih.gov/clinvar, Landrum et al. 2016), none of which have any reported effect on function. There are two reported X-ray structures (6D26 and 6D27). Extracellular N-linked glycosylations are predicted for N4 and N25 by sequence analysis (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004) and numerous O-linked sites on both the cytoplasmic and extracellular domains (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013), none of which have been confirmed experimentally. However, Nagata et al. (1999) have shown that treatment of DP2 with endo F increases its mobility on sizing gels, indicating a decrease in mass of 20-30 kDa and confirming the presence of N-linked glycosylation.

There are numerous potential phosphorylation sites on the human DP2 based on motifs, but none are specifically confirmed experimentally (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Eleven potential sites of phosphorylation are predicted: S22 and S27 on the extracellular N-terminal domain, T172 and T174 on extracellular loops, T64 on a cytoplasmic loop, and S318, T344, S345, S346, S352, and S391 on the cytoplasmic, C-terminal domain. There are also additional potential GRK phosphorylation sites in the C-terminal domain predicted for T321, S325, T344, S345, S346, S358, S376, S380, T383, S391, S392, T393, S394, S395 (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The DP2 receptor is expressed primarily in the gastrointestinal tract, brain, endocrine system and muscle, but is also found in many other tissues such as immune tissue (eosinophils, basophils and Th2 lymphocytes in particular), gall bladder, ciliary and both male and female reproductive tissue (http://www.proteinatlas.org, Uhlén et al. (2015)).

Ligand binding properties for DP2 have been characterized with human recombinant protein expressed in both HEK293 (Sawyer et al. 2002; Hata et al. 2005) and K562 (Hirai et al. 2001) cell lines, and mouse recombinant protein in a HEK293 cell line (Hata et al. 2003). Ligand affinities as measured by [³H]PGD₂ equilibrium competition assays and ligand efficacies as measure by the reduction of [cAMP] in either isoproterenol- or forskilin-stimulated cells are given in Table 1. As observed for the DP1 receptor, the DP2 receptor shows a preference for PGD₂, but also cross-reactivity with other ligands such as PGF_2α and 9α,11β-PGF_2α. The relative cross-reactivities are similar to those observed for the DP1 receptor. The most frequently used selective agonists for DP2 are Indomethacin and L-888,607. There is one commonly used selective antagonist, Ramatroban (Sugimoto et al. 2003) (Table 2).

Mechanism of Cell Activation

Signal transduction from the DP2 receptor occurs via a G_αi protein signaling system (see Table 3) resulting in a decrease in intracellular [cAMP] and an increase in intracellular [Ca²⁺] (Sawyer et al. 2002; Hirai et al. 2001; Pettipher et al. 2007). Further, DP2 activation in Th2 cells is mediated through a phosphatidylinositol 3-phosphate kinase (PI3K) and Ca²⁺/calmodulin/calcineurin signaling pathways (Xue et al. 2007).

Regulation

The regulation of DP2 has many similarities to that of DP1 discussed above with some notable exceptions. In addition to their work with DP1, Gallant et al. (2007) also examined the regulation of recombinant DP2 expressed in HEK293 kidney cells. They found that internalization of the DP2 receptor increases 20% in the presence of the PGD2 agonist, 5% less than observed for the DP1 receptor. However, when co-expressed with GRK2, the DP2 receptor’s agonist-induced internalization is increased by 88% compared to an increase of 25% for co-expression of GRK2 with DP1 and the rate desensitization via phosphorylation is considerably slower than observed for DP1. Further, co-expression of GRK5 or GRK6 with DP2 more than doubles the receptor internalization whereas co-expression of either of these GRKs has no effect on DP1 internalization. Co-expression of DP2 with arrestin-2 results in enhancement of agonist-induced internalization by 90%, over twice that observed for DP1, but unlike DP1, co-expression of DP2 with arrestin-3 has no effect on agonist-induced internalization of DP2. As observed for DP1, co-expressed GRKs reduces the DP2 receptor activity (measured by changes in inhibition of forskolin-induced cAMP synthesis) and to a similar degree. Lastly, both DP1 and DP2 recycling is enhanced by Rab proteins, however, for DP2 it is the Rab11 protein that promotes recycling rather than the Rab4 protein that enhances recycling of DP1. These data are consistent with a slow phosphorylation of DP2 by all three GRKs on a time scale comparable to internalization that is in turn reversed by Rab11 activation.

Schröder et al. (2009) examined the effect of DP2’s C-terminal tail on receptor internalization through comparison of the activity of recombinant DP2 with the native 86 amino acid C-terminal tail to that that of a recombinant truncated version with an 8 amino acid tail. They found that the truncated version exhibits enhanced signaling compared to wild type DP2, indicating that the wild type C-terminus inhibits interaction with G_αi to reduce signaling activity. Further, they found that the C-terminal is the site for β-arrestin-2 binding, a prerequisite for internalization, leading to a reduction in signaling. Interestingly, they report that DP2 is not phosphorylated upon agonist binding. This directly contrasts with the Gallant et al. (2007) report where GRKs were found to significantly increase internalization, presumably through phosphorylation of the receptor. Further, this report also indicates that inhibition of PKA with H-89 decreases internalization, again suggesting a phosphorylation event. Further study is clearly required to rationalize these contrasting results.

Regulation on the transcription level has also been examined. MacLean Scott et al. (2018) revealed that the transcription factor GATA3 enhances expression of DP2 during Th2 cell differentiation and within innate lymphoid cells. They also report that over-expression of the NFAT1 transcription factor reduces GATA3 promotor activity. They suggest that a dynamic relationship between GATA3 and NFAT1 competition for the promotor site may play a role in DP-mediated allergic inflammation.

Synergy and Opposing Actions of the DP1 and DP2 Receptors

PGD₂ is generated by mast, Th2, and dendritic cells and is found in high concentrations at sites of inflammation. The DP1 and DP2 receptors act jointly in the development and maintenance of the allergic response.

PGD₂ stimulation of DP2 receptors is directly involved in activation of Th2 lymphocytes, eosinophils and basophils, with little effect on neutrophils (Emery et al. 1989), as well as chemotaxis of these cells to the site of inflammation (Xue et al. 2007; Schratl et al. 2007). Further, efficient synthesis of the microvasculature permeability enhancer cystenyl LTs (LTC₄ LTD₄ and LTE₄) by eosinophils requires simultaneous stimulation of the DP1 and DP2 receptors (Mesquita-Santos et al. 2011). However, these two receptors have opposing effects on eosinophil activation. Binding of PGD₂ to the DP2 receptor activates eosinophils by stimulating the production of the leukocyte integrin CD11b, an event required for adhesion, migration and accumulation at the site of inflammation (Maiguel et al. 2011). Binding of PGD₂ to the DP1 receptor, however, negatively regulates the production of CD11b, serving to modulate the activation of eosinophils (Monneret et al. 2001).

Activation of both DP1 and DP2 receptors by PGD₂ results in the inhibition of apoptosis of particular immune cells, thus prolonging the inflammatory burden in regions of inflammation. Xue et al. (2009) have shown that binding of PGD₂ to DP2 receptors on Th2 cells suppresses annexin V binding, mitochondrial cytochrome C release, and caspase activities associated with cell apoptosis. Peinhaupt et al. (2018) have shown that stimulation of DP1 inhibits the onset of intrinsic apoptosis of eosinophils apparently though the activation of anti-apoptotic proteins such as the Bcl2 family of proteins. Further, they found that DP1 signaling induced the serum response element (SRE) that regulates genes responsible for cytoskeleton production and survival.

Prostaglandin E₂ Receptors

Introduction

There are four known distinct subtypes of PGE₂ receptors: 1) the EP1 receptor (PGE₂ receptor EP1 subtype, Gene: PTGER1), 2) the EP2 receptor (PGE₂ receptor EP2 subtype, Gene: PTGER2, 3) the EP3 receptor (PGE₂ receptor EP3 subtype, PGE2-R, Gene: PTGER3), and 4) the EP4 receptor (PGE₂ receptor EP4 subtype, Gene: PTGER3). Sequence homology between the subtypes is in the 28-33% range (Sugimoto and Narumiya 2007). Expression levels are both tissue and receptor subtype dependent.

EP1 receptor

Introduction

The human prostaglandin E₂ receptor EP1 (PTGER1, UniprotKB-P34995) is translated as a 402 amino acid polypeptide with a calculated molecular weight of 41.8 kDa. There are no additional reported isoforms and three coding SNP variants (A71T, T223M and H256R) (https://genecards.org, Stelzer et al. 2016), none of which involve any known clinical significance or condition. There are no reported X-ray structures. Extracellular N-linked glycosylations are predicted for N8 and N25 by sequence analysis (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004) and numerous O-linked sites on both the cytoplasmic and extracellular domains (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013), none of which have been confirmed experimentally.

There are numerous potential phosphorylation sites on the human PTGER1 based on motifs but none are specifically confirmed experimentally (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Potential sites of phosphorylation are predicted for S69, S150, S238, S249, S260, S262, S265, S282, S285 on cytoplasmic loops, and S390, S393, S394, and S397 on the cytoplasmic, C-terminal domain. There are also potential GRK phosphorylation sites predicted for T382, S384, S389, and S400 (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The EP1 receptor is primarily expressed in the kidney and spleen, but is also found in the gastrointestinal tract, lung, smooth muscle, and the central nervous system (http://www.proteinatlas.org, Uhlén et al. (2015)).

Ligand binding properties for recombinant human EP1 (Abramovitz et al. 2000; Sharif and Davis 2002; Ungrin et al. 2001) and recombinant mouse EP1 (Kiriyama et al. 1997) have been characterized in HEK 293 and Chinese hamster ovary (CHO) cell lines respectively (Table 4). Ligand efficacies for human EP1 were determined by the increase in intracellular [Ca²⁺] accompanying agonist binding (Table 5). The data show that the EP1 receptor exhibits the weakest PGE₂ binding of all EP receptors, but by most reports has efficacy similar to other EP receptors. Interestingly, although the binding of alternate ligands PGE₁ and PGF_2α are 10 and 40-fold weaker than PGE₂ respectively, the differences in efficacies are much smaller. The most frequently used selective agonists for EP1 are ONO-DI-004 and 17-phenyl PGE₂ (Sugimoto and Narumiya 2007; Dey et al. 2006). There is one commonly used selective antagonist, ONO-8713 (Sugimoto and Narumiya 2007) (Table 2).

Table 4.

Binding affinity (K_i) of eicosanoids on the recombinant EP receptors expressed in cell cultures and in human duodenum cells

	Ki (nM)
Receptor/cell line	PGE1	PGE2	PGF2α	PGD2	PGI2 analog	Reference
mEP1/CHO	36	20	1300	-	-	Kiriyama et al. 1997
mEP2/CHO	10	12	-	-	-	Kiriyama et al. 1997
mEP2/CHO	-	1.0	-	-	-	Nishigaki et al. 1995
mEP3/CHO	1.1	0.85	75	-	-	Kiriyama et al. 1997
mEP4/CHO	2.1	1.9	-	-	-	Kiriyama et al. 1997
mEP4/CHO	-	1.0	-	-	-	Nishigaki et al. 1995
hEP1/HEK293	-	9.1	547	5,820	-	Abramovitz et al. 2000
hEP1/HEK293	110	9.9	380	6,040	-	Ungrin et al. 2001
hEP1/HEK293	165	14.9	594	19,500	-	Sharif and Davis 2002
hEP1/HEK293	-	26	594	>19,000	>15,000	Sharif et al. 2003
hEP2/HEK293	-	4.9	964	2,973	-	Abramovitz et al. 2000
hEP3-I/COS7a	19	7.2	1150	-	-	Regan et al. 1994b
hEP3-I/COS7	3.8	1.2	-	-	-	Kotani et al. 1995
hEP3-I/COS-M6a	0.4	0.4	135	1,310	-	Adam et al. 1994
hEP3-I/COS-1a	0.8	0.8	250	250	-	Kunapuli et al. 1994
hEP3-II/COS7	28	14	4648	-	-	Regan et al. 1994b
hEP3-II/COS7	3.9	1.6	-	-	-	Kotani et al. 1995
hEP3-II/COS-M6a	0.33	0.4	85	951	-	Adam et al. 1994
hEP3-III/COS7	11	8.2	599	-	-	Regan et al. 1994b
hEP3-III/COS7	2.6	1.3	-	-	-	Kotani et al. 1995
hEP3-III/COS-M6a	0.28	0.4	36	277	-	Adam et al. 1994
bEP3/HEK293	-	3	24	1,115	5,375	Sharif et al. 2003
hEP3-III/HEK293	-	0.33	38	421	-	Abramovitz et al. 2000
hEP3-IV/COS7	5.5	1.7	-	-	-	Kotani et al. 1995
hEP3-V/CHOb	5.0	5.0	10,000	3,200	-	Kotani et ai. 2000
hEP3-VI/CHO	5.0	8.0	3,200	1,260	-	Kotani et ai. 2000
hEP4/HEK293b	-	0.79	288	1,483	-	Abramovitz et al. 2000
hEP4/HEK293	-	0.9	433	2,139	8,074	Sharif et al. 2003

Table 5.

Signaling efficiency (EC₅₀) of eicosanoids on the recombinant EP receptors expressed in cell cultures and in human duodenum cells

	EC50 (nM)
Receptor/cell line	PGE1	PGE2	PGF2α	PGD2	PGI2 analog	Reference
hEP1/CHO	-	1.98	-	-		Sekido et al. 2016
hEP1/HEK293	9.3	2.90	29	3,200	319	Ungrin et al. 2001
hEP2/CHO	-	3.74	-	-	-	Sekido et al. 2016
mEP2/CHO	-	1.0	-	-	-	Nishigaki et al. 1995
hEP2/COS7c	-	50	-	-	-	Regan et al. 1994a
hEP2/hNPE	-	4.9	964	2973	-	Sharif et al. 2003
hEP3/CHO	-	1.60	-	-	-	Sekido et al. 2016
EP3/gp	-	5.4	-	-	-	Savage et al. 1993
hEP4/CHO	-	4.09	-	-	-	Sekido et al. 2016
hEP4/hNPE	-	46	>10,000	>10,000	>10,000	Crider et al. 2000
hEP4/human duodenum	-	50	-	-	-	Larsen et al. 2005
mEP4/CHO	-	1.0	-	-	-	Nishigaki et al. 1995

Abbreviations: h, human; m, mouse; b, bovine; HEK293 human embryonic kidney cells; COS M-6, Green monkey kidney cells; COS7, monkey fibroblast-like cells; CHO. Chinese hamster ovary cells; COS-1, African green monkey kidney fibroblast-like cells; hNPE human non-pigmented ciliary epithelial cells; gp, guinea pig distal ileum tissue. The K_i values were determined from [³H]PGE₂ displacement assays. The EC₅₀ for recombinant EP1 and EP3 receptors were determined from intracellular changes in [Ca²⁺] and the EC₅₀ for recombinant EP2 and EP4 receptors were determined from intracellular changes in cAMP. The EC₅₀ values for hEP2 and hEP4 expressed in in hNPE cells were determined from intracellular changes in cAMP. The EC₅₀ values determined for gEP₃ gp (guinea pig vas deferens) tissue were determined by changes in the inhibition of twitch contraction values. The EC₅₀ for hEP4 in human duodenum tissue was determined by changes in short circuit current. a) K_i values determined from reported IC₅₀ and K_d utilizing the Cheng-Prusoff equation (Suganami et al. 2016); b) Values determined from IC₅₀ estimated from binding curves and reported K_d utilizing the Cheng-Prusoff equations; c) Value estimated from a graph of [PGE₂] vs. [cAMP].

EP1 has been found to form heterodimers or higher order oligomers with the β2-androgenic receptor (β2AR) airway smooth muscle (ASM) cells and when both are transfected into COS-7 cells. Although EP1 stimulation alone does not appear to have any significant effect on ASM cells itself, in the presence of β2AR the heterodimeric complex forms and stimulation of EP1 serves to alter the β2AR structure leading to a reduction in binding of β2AR to its G_αs protein, resulting in a reduction β3AR-mediated ASM relaxation (McGraw et al. 2006).

Mechanism of Cell Activation

Signal transduction from the EP1 receptor has been controversial with respect to the particular G-proteins involved and the mechanisms that lead to the observed increase in intracellular calcium (Sugimoto and Narumiya 2007; Tsuboi et al. 2002). It is now generally accepted that EP1 interacts with G_αq (Table 3) which in turn activates phosphoinositol-phospholipase C (PI-PLC) causing the release of IP₃ intracellularly, and a subsequent increase in intracellular [Ca²⁺] (Tang et al. 2005; Markovič et al. 2017). Ji et al. (2010) have not only confirmed the interaction of G_αq/11 with EP1 for recombinant human EP1 in a HEK cell line, but have also shown that EP1 can couple to G_αi/o and activate the PI3K/protein kinase B (PKB or AKT)/mTOR kinase signaling pathway in the same experimental system.

Regulation

The regulation of the EP1 receptor is facilitated by several different mechanisms. Nasrallah et al. (2015) examined the expression of EP1 in MCT cells and found this receptor is constitutively expressed and the expression is enhanced following a 24 hr incubation with either PGE₂ or transforming growth factor beta (TGFβ). They found that PGE₂ alone and TGFβ alone increase expression 2.5- and 3.8-fold respectively and combined they increased EP1 expression 7-fold. Further, utilizing MCT cells transfected with EP1 siRNA the effect of PGE₂ on EP1 expression is reduced by 50%, suggesting that the effect of PGE₂ is on the transcriptional level.

Expression of EP1 can also be altered under conditions of low oxygen. Under hypoxic conditions of 5% and 2% O₂, Lee et al. (2007) found that PGE₂ release and EP1 receptor expression is strongly elevated when compared to 21% O₂ conditions. It is not known if the increase in EP1 expression results from a localized increase in PGE₂ levels as described above or is due to another mechanism.

The expression of EP1 is regulated by co-expressed proteins as well. Sood et al. (2014) have found that overexpression of COX-2 increases the membrane expression of EP1 and does so non-transcriptionally. This mechanism provides a feedback loop to resolve inflammation, as COX-2 expression is down-regulated by EP1 stimulation of COX-2 degradation.

EP2 receptor

Introduction

The human prostaglandin E₂ receptor EP2 (PTGER2, UniprotKB-P43116) is translated as a 358 amino acid polypeptide with a calculated molecular weight of 39.8 kDa. There are no additional reported isoforms and two coding SNP variants (C83G and Y285C) (https://www.ncbi.nlm.nih.gov/clinvar/, Landrum et al. 2016), both of which appear to be benign. There are no reported X-ray structures. There are extracellular N-linked glycosylations predicted for N3 and N6 on the N-terminal and N96 and N287 on extracellular loops (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004). There are also O-linked sites predicted for S237 on a cytoplasmic loop and S5 extracellular N-terminal domain (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013). None of these specific glycosylations are confirmed experimentally, however, western blot analysis of HaCaT and COS-7 cells reveal three EP2 reactive bands at 30 kDa, 43-45 kDa, and 51 kDa (Konger et al. 2002), consistent with multiple glycoforms.

There are numerous potential phosphorylation sites on the human EP2 based on motifs but none are specifically confirmed experimentally. Phosphorylation is predicted for S8 and S10 on the extracellular N-terminal domain, S291 on an extracellular loop, S62, S63, S149, S151, S229, S232, S240, and S255 on cytoplasmic loops, and T344 and S353 on the cytoplasmic, C-terminal domain (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). There are also potential GRK phosphorylation sites predicted for the C-terminal domain at S328, S335, T338, T342, S345, S347, and S350 (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The EP2 receptor is widely expressed, albeit in lower abundance when compared to other receptors (http://www.proteinatlas.org, Uhlén et al. (2015)). Ligand binding properties for recombinant human EP2 (Abramovitz et al. 2000) and recombinant mouse EP2 (Kiriyama et al. 1997; Nishigaki et al. 1995) have been characterized in HEK 293 and Chinese hamster ovary (CHO) cell lines respectively (Table 4). Ligand efficacies for human and murine EP2 were determined by the increase in intracellular cAMP (Table 5). The data show that the EP2 receptor exhibits moderate PGE₂ binding, similar to EP3 (see below). The efficacy is like other EP receptors in CHO cells, but exhibit much lower efficiency in COS-7 cells. The most frequently used selective agonist for EP2 is ONO-AE1-259. The most highly selective antagonist for EP2 is PF-04418948 (af Forselles et al. 2011) with greater than 10,000-fold selectivity for EP2 receptors than other prostanoid receptors. There are also several moderately selective antagonists, TG6-129, TG4-155 and TG6-10 available (Ganesh 2014).

Mechanism of Cell Activation

Signal transduction for the EP2 receptor has been shown to occur through both G-protein-dependent and G-protein-independent pathways. The G protein-dependent pathway is similar to that of DP1 (Table 3) in that it is coupled to G_αs (Hirata and Narumiya 2011), leading to activation of adenylate cyclase and the production of cAMP (Honda et al. 1993; Regan et al. 1994a) which in turn activates PKA (Chun et al. 2009; Regan 2003) and exchange proteins that are activated by cAMP (Epacs) (Sands and Palmer 2008). EP2 can also switch from signaling through G_αs to signaling through G_αq/11 in human myometrium, leading to an increase in [Ca²⁺] and pro-inflammatory pathways, promoting a pro-labor condition (Kandola et al. 2014). The G-protein-independent pathway involves the formation of an EP2-β-arrestin1-Src tyrosine protein-kinase complex that leads to the activation of the epidermal growth factor receptor (EGFR) which can then lead to the activation of H-Ras (protein p21), PKB and the MAP kinases ERK1/2 (Chun et al. 2009).

Regulation

The regulation of the EP2 receptor is facilitated by a number of different mechanisms. The effect of various agonists on the expression of prostanoid receptors have been presented. Perchick and Jabbour (2003) examined the effect of upstream prostanoid production on the expression of EP type receptors. Here they overexpressed transfected human cyclooxygenase 2 (COX-2) into the Ishikawa human endometrial cell line to increase the production of PGH₂, the direct precursor for PGE₂. They found that not only was the production of PGE₂ elevated, but that EP2 and EP3 receptor expression was increased with no effect on EP1 and EP4 expression.

Agonist expression can also direct signaling pathways. Agonist-induced conformational changes in the EP2 receptor allows for phosphorylation of specific sites on the C-terminus that alter the receptor response in a phosphorylation site dependent manner (Tobin 2008). Phosphorylation may be accomplished by G-protein receptor kinases (GRK) (Ferguson 2007), protein kinase C (PKC), PKA, or PKB to name a few and the sequence specificity of the kinase determines the site or sites of modification, leading to a specific outcome (Tobin 2008). One such outcome is the binding of β-arrestin which results in the displacement of G_αs, thus halting the cAMP production pathway. Although β-arrestin binding is also known to lead to desensitization through internalization of GPCRs via a clatherin-mediated pathway, expression of recombinant EP2 in 293-EBNA cells (Invitrogen) and COS-1 cells produces little internalization (Desai et al. 2000; Penn et al. 2001). However, when co-expressed with a modified arrestin (ARR2 (R169E)), an arrestin that does not require phosphorylation to bind to GPCRs, EP2 exhibits a four-fold increase in internalization compared to cells co-expressing wild type recombinant arrestin (Penn et al. 2001). These results indicate that the failure of wild type arrestin to cause internalization of EP2 is due to the absence of specific phosphorylation and suggests that the required kinase was not active under the experimental conditions.

Agonist-induced desensitization is a common feature of GPCRs. EP2 is not readily desensitized through phosphorylation as noted above, but over time the formation of PGE₂ metabolic products which have considerably longer half-lives than PGE₂ and lower efficacies, accumulate to effectively compete with PGE₂ for the receptor and in doing so reduce the response to PGE₂ (Nishigaki et al. 1996).

Regulation of EP2 on the transcriptional level has also been observed. It is well established that treatment of macrophages with bacterial lipopolysaccharide (LPS) stimulates macrophages and leads to their releasing large amounts of PGE₂. Ikegami et al. (2001) reported that LPS treatment of cultured mouse macrophages results in a transient five-fold up-regulation of EP2 mRNA at 3 hours after stimulation, which returns to previous levels by 5 hours. They showed further that no protein synthesis is required for the up-regulation and suggest that up-regulation of EP2 mRNA is a direct effect of LPS stimulation. They also observed that changing media promotes up-regulation of EP2 mRNA, but to a lesser degree than observed for LPS and suggest that induction factors in the media may be responsible. Kashmiry et al. (2018) reported that up-regulation of EP2 receptor mRNA in human monocytes and THP-1 cells following LPS stimulation is controlled by the PGE₂ stimulation of the EP4 receptor.

Steroid hormone regulation of the transcription of a large number of genes is well established. Lim and Dey (1997) examined EP2 mRNA expression in the mouse preimplantation uterus as a function of experimentally elevated progesterone (P4) and 17β-estradiol (E2). They found that P4 up-regulates EP2 mRNA and the presence of E2 enhances this effect, while E2 alone down-regulates EP2 mRNA and suggest that the synergy between P4 and E2 on EP2 expression is essential in preparing the uterus for implantation.

EP₃ receptor

Introduction

The human prostaglandin E₂ receptor EP3 (PTGER3, UniprotKB-P43115) is translated as a 390 amino acid polypeptide with a calculated molecular weight of 43.3 kDa. There are 14 reported isoforms and 3 coding SNP variants (N366S, T319M, M169L) (https://genecards.org, Stelzer et al. 2016) of the canonical EP3A sequence, none of which have any reported effect on function. Each of the isoforms is a splice variant of the C-terminal tail and all are identical for the first 359 residues but differ in G-protein interactions and signaling. There are two reported X-ray structures (PDB entry 6AK3 and 6M9T). Extracellular N-linked glycosylation are predicted for N18 and N36 on the N-terminal and with lower confidence N217 and N308 on extracellular loops (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004). There are O-linked sites predicted with high confidence for S20, T22, and S30 on the extracellular N-terminal domain and S369, S371, S373, and S380 on the intracellular C-terminal tail (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013). None of these specific glycosylations are confirmed experimentally, however, site-specific mutagenesis studies of rat EP3 (UniProtKB-P34980) confirm N-glycosylation of N16 and N194 (Böer et al. 2000) which correspond to N18 and N217 in human EP3. Further, the glycosylation for the rat analog is required for correct sorting to the plasma membrane, but not for correct folding of EP3.

There are numerous potential phosphorylation sites on the human EP3 based on motifs but none are specifically confirmed experimentally (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). For all known isoforms, residues 1-359 are identical. High probability phosphorylation sites are predicted for all isoforms at S43 on the extracellular N-terminal domain, and S82, S258 and S270 on cytoplasmic loops. There are also multiple potential GRK phosphorylation sites on the C-terminal domain for all but isoforms EP3-III and EP3-IV (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The EP3 receptor is widely expressed in low abundance, but is highly expressed in smooth muscle, kidney, endometrium, and in adipose tissue (http://www.proteinatlas.org, Uhlén et al. (2015)). Ligand binding properties for recombinant human EP3 isoforms have been characterized in HEK293 (Abramovitz et al. 2000), COS-1 (Kunapuli et al. 1994), COS-7 (Kotani et al. 1995; Regan et al. 1994b) and COS-M6 (Adam et al. 1994) cell lines (Table 4). Ligand efficacies for human recombinant EP3 isoforms were determined in CHO cells by the increase in intracellular [Ca²⁺] (Sekido et al. 2016) and guinea pig vas deferens smooth muscle by changes in twitch contraction values (Savage et al. 1993) (Table 5).

The data show that the EP3 receptor on the average exhibits PGE₂ binding that is greater than most EP receptors, but slightly less than observed for EP4. The average efficacy is similar to other EP receptors. The most selective agonists for EP3 in general are SC-46275, ONO-AE-248, MB28767 and sulprostone (Norel et al. 2004; Savage et al. 1993; Kotani et al. 2000; Jin et al. 1997; Abramovitz et al. 2000) (Table 2). There are two EP3 specific antagonists available, DG-041 and ONO-AE3-240 (Abramovitz et al. 2000) (Table 2).

Mechanism of Cell Activation

Signal transduction by the EP3 receptor is dependent on the isoform present (Table 3). As noted above, there are at least 14 different isoforms, each of which differs only in the composition and length of the C-terminal tail from residue 360 onward (Kotani et al. 1995; Kotani et al. 1997). These differences determine which G-proteins bind and hence determine the signal pathway actuated.

EP3-I (UniprotKB isoform EP3A, identifier P4535115-1) the canonical isoform is coupled to two secondary messenger systems, involving inhibition of cAMP and stimulation of phosphoinositide turnover that is consistent with coupling to G_αi and G_αq proteins respectively (Kotani et al. 1995).

EP3-II (UniprotKB isoform EP3C, identifier P4535115-2) is functionally equivalent to bovine EP3D which is known to be coupled to G_αi, G_αs, and G_αq (Kotani et al. 1995; Kotani et al. 1997) that initiate cAMP repression, an increase cAMP production, and an increase in both inositol 1,4,5-trisphosphate (IP₃) and intracellular [Ca²⁺] respectively.

EP3-III (UniprotKB isoform EP3B, identifier P4535115-3) is coupled to the inhibition of cAMP production (Regan et al. 1994b) with an increase in intracellular [Ca²⁺] (Schmid et al. 1995), does not stimulate IP₃ production (Kotani et al. 1995), and its activity is inhibited by the presence of pertussis toxin, a known inhibitor of G_αi, all of which clearly indicate that it couples to G_αi (Jin et al. 1997).

EP3-IV (UniprotKB isoform EP3D, identifier P4535115-4) is coupled to both stimulation of cAMP production and reduction in cAMP accumulation with no reports of IP₃ generation (Kotani et al. 1995). The activity of this receptor is also inhibited by the presence of pertussin toxin, indicating that the reduction of cAMP is coupled through G_αi (Jin et al. 1997). The production of cAMP is consistent with coupling to G_αs.

EP3-V (UniprotKB isoform EP3E, identifier P4535115-5) is coupled to the inhibition of cAMP production. This fact and the loss of activity in the presence of pertussis toxin indicates a coupling to G_αi (Kotani et al. 2000). Associated changes in intracellular [Ca²⁺] have not been reported.

EP3-VI (UniprotKB isoform EP3F, identifier P4535115-6) like EP3-IV is also coupled to both stimulation of cAMP production and reduction in cAMP accumulation with no reports of IP₃ generation (Kotani et al. 2000). Observed inhibition of cAMP reduction with pertussis toxin and stimulation of cAMP production with the EP3-specific agonist M&B28767 indicate coupling to G_αi and G_αs respectively.

There are eight additional EP3 isoforms reported in the literature of which only EP3.e and EP3.f have been partially characterized as reducing [cAMP] upon PGE₂ activation (Schmid et al. 1995). Interestingly sequences for these two are not listed in UniProtKB, however, P435115-11 and P435115-10 have identical sequences to EP3.e and EP3.f respectively with the exception of an additional exon sequence VANAVSSCSNDGQKGQPISLSNEIIQTEA (360-388) to the N-terminal side of the defining C-terminal sequence.

Regulation

Regulation by receptor internalization is isoform dependent. In HEK293 cells expressing human recombinant EP3-I, EP3-II , EP3-III EP3-IV, EP3-V, EP3.e, and EP3.f, all receptors are found to be located primarily on the cell surface under non-stimulating conditions (Bilson et al. 2004). Isoforms EP3-III, EP3-IV, EP3-V, EP3.f are also observed intracellularly. When stimulated with PGE₂, isoform EP3-I translocate robustly to intracellular punctate vesicles along with β-arrestin. EP3.f behaves similarily to EP3-I but is internalized to a lesser extent. Isoforms EP3-II, EP3-V, EP3-VI internalize upon stimulation to a lesser extent than EP3-I and this is not accompanied by β-arrestin migration into the vesicles. Isoforms EP3-III and EP3-IV do not internalize in response to PGE₂ binding. This is expected, as the C-terminal tails of these isoforms contain no Ser of Thr residues, and thus the phosphorylation required for β-arrestin facilitated internalization cannot occur.

Receptor response to agonist activation of EP3 isoforms I-IV has been examined in detail in terms of inhibition of cAMP production (Jin et al. 1997). Isoforms EP3-I and EP3-II exhibit typical agonist concentration-dependent behavior whereas EP3-III exhibits full constitutive activity and EP3-IV exhibits partial constitutive activity. In contrast, agonist response for both the EP3-III and EP3-IV receptors in terms of increased intracellular [Ca²⁺] is agonist dependent (Regan et al. 1994b; Schmid et al. 1995; Sekido et al. 2016).

Regulation of EP3 on the transcriptional level has been reported without reference to specific isoforms. Human leukemic T cells of the HSB.2 T cell line express EP2, EP3 and EP4 receptor subtypes. Exposure of these cells to the T cell mitogen concanavalin A (Con A) increases interleukin 6 (IL-6) secretion in response to PGE₂ while at the same time downregulates the expression of EP3 mRNA (Zeng et al. 1998). In contrast, forced overexpression of the ERG transcription factor in immortalized PCA cell line DU145 results in the upregulation of EP3 while simultaneously increasing IL-6 secretion (Merz et al. 2016).

EP4 receptor

Introduction

The human prostaglandin E₄ receptor EP4 (PTGER4, UniprotKB-P35408) is translated as a 488 amino acid polypeptide with a calculated molecular weight of 53.1 kDa. There are no reported isoforms and one reported coding SNP variant (V294I) which is not reported to have any effect on function (https://www.ncbi.nlm.nih.gov/clinvar/, Landrum et al. 2016). There are two reported X-ray structures (PDB 5YHL and 5YWY). Extracellular N-linked glycosylation is predicted with high confidence for N7 and two with lower confidence, N177 on an extracellular loop and N482 on the intracellular C-terminal domain (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004). There are also numerous O-linked sites predicted for both the N- and C-terminal domains as well as on one extracellular loop (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013). None of these specific glycosylations are confirmed experimentally. There are five known phosphorylation sites on the intracellular C-terminal domain, S370, S374, S377, S379, and S382 found in various combinations that are involved in to β-arrestin binding and internalization (Neuschäfer-Rube et al. 2004). There are additional potential serine phosphorylation sites predicted for S11, S13, S19 on the extracellular N-terminal, S222, S252, S259 on a cytoplasmic loop, and S371, S437, S440, S442, S443, S448, S460, and S480 on the C-terminal domain (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Additional potential GRK phosphorylation sites are predicted for S366, S442, S443, S450, and S484 with high probability (http://gps.biocuckoo.cn/, Xue et al. 2011). None of these phosphorylations have been confirmed experimentally.

Expression and Characterization

The EP4 receptor is widely expressed with the highest amounts in smooth muscle, and the immune system (http://www.proteinatlas.org, Uhlén et al. (2015)). Ligand binding properties for recombinant human EP4 have been characterized in HEK293 cell lines (Abramovitz et al. 2000; Sharif et al. 2003) (Table 4). Ligand efficacies for human recombinant EP4 were determined in CHO cells by the increase in intracellular cAMP obtained by enzyme immunoassay (Sharif et al. 2003) and response to stimulated short circuit current (SSC) and slope conductance (SG) in tissue from human duodenal biopsies (Larsen et al. 2005) (Table 5). The data show that the EP4 receptor exhibits the tightest PGE₂ binding of all EP receptors, however, the average efficacy is significantly less than observed for other EP receptors. The most selective agonists for EP4 are ONO-AE1-329 and tetrazolo PGE₁, (Sugimoto and Narumiya 2007; Jones et al. 2009) and the most selective antagonists are ONO-AE3-208 and CJ-042794 (Sugimoto and Narumiya 2007; Jones et al. 2009) (Table 2).

Mechanism of Cell Activation

Similar to EP2, signal transduction for the EP4 receptor has been shown to occur through both G-protein-dependent (Table 3) and G-protein-independent pathways. G protein-dependent coupling to G_αs (Nishigaki et al. 1996), leads to activation of adenylate cyclase and the production of cAMP which in turn activates PKA (Regan 2003) that phosphorylates downstream proteins, including the cAMP response element binding protein (CREBP) (Takayama et al. 2002). EP4 has also been shown to switch from signaling through G_αs to signaling through a cAMP independent and PI3K- dependent mechanism, that leads to phosphorylation of extracellular signal-related kinases (ERKs), which in turn leads to the expression of early growth response factor-1 (EGR-1) that is not observed in EP2 signaling (Fujino et al. 2003; Fujino and Regan 2006). PI3K-dependent signaling may also result in phosphorylation of PKB which in turn regulates cell survival through protection from apoptotic stimuli (George et al. 2007). Similarly, EP4 stimulation also activates the T-cell-factor (Tcf)/lymphoid enhancer factor (Lef), known to be activated by the G_αq-coupled FP_B prostanoid receptor through a PI3K-dependent signaling pathway (Fujino et al. 2002). EP4 signaling can also occur through coupling with G_αi/o, resulting reduction of intracellular cAMP via inhibition of adenylate cyclase. Leduc et al. (2009) examined the differential coupling of EP4 to various G-proteins and found that selectivity was dependent on the identity of the agonist binding to the receptor. They also found that PGE₂ was the most selective for G_αs signaling whereas PGF₂ and PGE₁-OH were more selective for G_αi/o signaling and β-arrestin binding.

Regulation

As observed for EP2, Nishigaki et al. (1996) found that EP4 is desensitized through agonist-induced GRK-mediated phosphorylation. However, the two receptors differ in sensitivity to agonist where EP4 is more rapidly desensitized than EP2 and is much less responsive to PGE₂ metabolic products than EP2, suggesting that EP4 is involved in a short-term and rapid response while EP2 is involved in a more long-lasting response. Similar to EP2, desensitization of EP4 leads to replacement of G_αs with a β-arrestin isoform that leads to internalization into punctate vesicles. However, the phosphorylation state of EP4 is not as critical to the internalization as it is for EP2 (Penn et al. 2001). In fact, Desai et al. (2000) have shown that mutant EP4 receptors with Ala substituted for each and every Ser and Thr in the C-terminal domain internalized to the same extent as wild type EP4 (Desai et al. 2000) indicating that phosphorylation is unnecessary for internalization. However the same study showed that EP4 mutants with C-terminal tails truncated before residue 383 had a reduced internalization and those truncated at the proximal end of the C-terminus (residue 350) did not internalize at all, indicating that the presence of the C-terminal tail is necessary for internalization.

Regulation of EP4 on the transcriptional level has also been observed in mouse macrophages. As noted above for EP2, LPS treatment of cultured macrophages results in changes in EP receptor mRNA expression. However, the mechanisms and results for EP4 are quite different. In contrast to the LPS-induced upregulation of EP2, Ikegami et al. (2001) report that EP4 is down-regulated within 3 hours of treatment and the EP4 mRNA expression levels drop to less than 10% of the control levels. The mechanism apparently involves the LPS-induced over-production of PGE₂ which in turn mediates EP4 stimulated cAMP production that results in a reduction of gene expression. The likely regulatory mechanism under inflammatory conditions would involve an interaction of cAMP-induced CREBP and NFκB transcription factors with the EP4 gene.

In human monocytes the transcriptional regulation is quite different (Kashmiry et al. 2018). Here, LPS treatment alone results in a nearly three-fold increase in EP4 mRNA rather than the decrease observed in the mouse study and in the presence of added PGE₂ the increase is significantly lower at 25% above control.

In human glioblastomas EP4 transcription is regulated in part by the specificity protein 1 (Sp-1) transcription factor (Kambe et al. 2008). Phosphorylation of Sp-1 reduces its ability to bind to DNA which in turn leads to the suppression of EP4 transcription. Phosphorylation of Sp-1 is known to be facilitated by several kinases including ERK. The fact that ERKs are known to be activated by EP4 signaling (Fujino et al. 2003) suggests that this mechanism may represent a negative feedback loop.

IP receptor

Introduction

The human prostacyclin (PGI₂) receptor hIP (PTGIR, UniprotKB-P43119) is translated as a 386 amino acid polypeptide with a calculated molecular weight of 41.0 kDa. No X-ray structures have been reported. There are four other reported isoforms of which three represent structures utilizing fewer than the total exons found in the canonical sequence and one with an additional exon not found in the canonical form. All are predicted to have a least one transmembrane helix and the biological functions are currently unknown. There are 18 reported SNP variants, eight of which exhibit various degrees of biological dysfunction and represent less than 2% (n = 1,761) of the hIP receptors examined (Stitham et al. 2002; Stitham et al. 2011). One N-linked extracellular glycosylation is predicted with high confidence for N7 and with lower confidence for N78 and N203 (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004) and one high confidence cytosolic O-linked predicted for S337 (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013). Only glycosylation at N7 and N78 have been experimentally confirmed (Zhang et al. 2001). Although specific glycosylation of other residues has not been confirmed experimentally, glycosylation at multiple unknown sites have been reported (Miggin et al. 2002; Donnellan and Kinsella 2009) and glycosylation of the C-terminal region, possibly S337, may be involved in membrane localization (Smyth et al. 1998). One prenylation site is predicted by PrePS ((http://mendel.imp.ac.at/sat/PrePS/index.html, Maurer-Stroh et al. 2007) for C383, and confirmed experimentally (Miggin et al. 2002). Palmitoylation was originally reported for C308 and C311, however, the publication has since been retracted (Miggin et al. 2003). Predictions for palmitoylation with CSS-Palm (http://csspalm.biocuckoo.org/, Zhou et al. 2006) reveal high confidence predictions at C5, C308, C309, and C-311 with the highest for C5. However, C5-C165 and C92-C170 disulfides have been confirmed experimentally, potentially ruling out a C5 palmitoylation (Giguère et al. 2004). Further, it has been shown that formation of these disulfides is required for expression and may also be involved in dimerization and oligomerization of the receptor. There are potential serine phosphorylation sites predicted for S14 on the extracellular N-terminal, S268 and S269 on an extracellular loop, T230 on a cytoplasmic loop, and S328, S337, and S374 on the C-terminal domain (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Only phosphorylation at S328 has been confirmed experimentally (Smyth et al. 1998) and shown to be accomplished by PKC (Smyth et al. 1998; Smyth et al. 1996). There are 11 potential high probability GRK phosphorylation sites predicted but as yet not experimentally confirmed (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The human IP receptor (hIP) is widely expressed with the highest amounts in lung, muscle, female tissues, and the immune system (http://www.proteinatlas.org, Uhlén et al. (2015)). Ligand binding properties for hIP have been characterized in platelet membranes by displacement of [³H]iloprost, an IP-specific agonist (Jones et al. 2009; Siegl et al. 1979; Tsai et al. 1989; Crider et al. 2001) (Tables 2 and 6) instead of PGI₂ due to the short half-life of the latter. Ligand efficacies for hIP have been determined in platelet membranes (Fitscha et al. 1987; Stürzebecher and Losert 1987; Kobzar et al. 2001), vascular smooth muscle (Angulo et al. 2002; Hadházy et al. 1986; Baxter et al. 1995, lymphocytes (Kilfeather et al. 1984), and recombinant human IP expressed in COS-1 (Crider et al. 2001) and MEG-01 (Tunaru et al. 2016) cells (Table 6). Although other PG receptors are present in the human tissue samples, the EC₅₀ values clearly indicate that any cross-reactivity of PGI₂ with the other receptors would result in negligible response. The most selective agonists for IP are cicaprost, carbacyclin, and iloprost, the former two being slightly more potent than the latter (Jones et al. 2009) (Table 2). Interestingly, the eicosanoid 19(S)-hydroxy-eicosatetraenoic acid (19(S)-HETE) also serves as a moderately potent hIP agonist and stimulates the production of cAMP (Tunaru et al. 2016). The most selective antagonists are RO-1138452 and RO-3244794 (Jones et al. 2009) (Table 2).

Table 6.

Binding affinity (K_i) and signaling efficiency (EC₅₀) of eicosanoids on the recombinant IP receptors expressed in cell cultures and in human platelets, lymphocytes and smooth muscle

	Ki (nM)
Receptor/cell line	19S-HETE	PGI2 analog	PGE2	PGF2a	PGD2	Reference
hIP/plateletsa	-	1600	> 1 x 107	> 1 x 107	> 1 x 107	Siegel et al. 1979
hIP/plateletsa	-	470	> 1 x 105	-	6,300	Tsai et al. 1989
hIP/plateletsb	-	22	-	-	-	Tsai et al. 1989
hIP/platelet membrane	-	1,398 ± 724	53,708 ± 2,136	50,000	>140,000	Crider et al. 2001
hIP/MEG-01a	660	-	-	-	-	Tunaru et al. 2016
	EC50 (nM)
Receptor/cell line	19S-HETE	PGI2 analog	PGE2	PGF2a	PGD2	Reference
hIP/human plateletsc	-	5.4-11.6	-	-	-	Fitscha et al. 1987
hIP/human plateletsc	-	0.81	-	-	-	Stürzebecher et al. 1987
hIP/human plateletsc	-	1.9 ± 0.2	-	-	-	Kobzar et al. 2001
hIP/smooth muscled	-	32	-	-	-	Angulo et al. 2002
hIP/smooth musclee	-	15	-	-	-	Hadházy et al. 1986
hIP/smooth musclef	-	12.7	-	-	-	Baxter et al. 1995
hIP/lymphocytesf	-	6.3	-	-	-	Kilfeather et al. 1984
hIP/COS-1g	-	40	-	-	-	Tunaru et al. 2016
hIP/MEG-01g,h	567	-	-	-	-	Tunaru et al. 2016

Abbreviations: h, human; m. mouse; MEG-01, Human megakaryoblastic leukaemia cells; COS-1, African green monkey kidney fibroblast-like cells. The K_i values were determined from IP specific agonist [³H]iloprost displacement assays, as PGI₂ itself has a very short half-life. The EC₅₀ were determined from intracellular changes in cAMP concentrations or a physical change. a) K_i calculated from reported IC₅₀ of agonist vs. [³H]iloprost using Cheng-Prusoff equation (Suganami et al. 2016); c) K_i corrected for PGI₂ hydrolysis, value was 470 nM without correction; c) EC₅₀ determined from ADP-induced platelet aggregation; d) EC₅₀ determined from arterial relaxation; e) EC₅₀ determined from femoral artery relaxation; f) EC₅₀ determined from uterine artery relaxation g) EC₅₀ determined from increase in cAMP; h) EC₅₀ estimated from fluorescence of cAMP probe vs. [PGI₂] for recombinant hIP expressed in MEG-01 cells.

hIP receptors have the ability to form homodimers, homooligomers and heterodimers with other GPCRs. Homodimerization and oligomerization occurs through disulfide linkages, specifically C5-C165 and C92-C170, and monomers, dimers and oligomers are expressed on the cell surface in well-defined ratios (Giguère et al. 2004). Furthermore, the polymerization process occurs intracellularly and is independent of agonist stimulation. Whether polymerization is a requirement for initial cell surface expression is currently unknown. Heterodimers of hIP and the human thromboxane receptor hTPα have been reported by Wilson et al. (2004). Expression of both receptors in HEK293 and human aortic smooth muscle cells (hAMSC) were demonstrated and when activated with a TPα agonist, rather than the expected increase in IP₃ and intracellular [Ca²⁺] normally associated with TP_α stimulation, an intracellular cAMP increase is observed as would be expected for the IP receptor. Clearly the G-protein specificity of the heterodimer is defined by the IP receptor subunit.

Mechanism of Cell Activation

Similar to EP2 and EP4 receptors, signal transduction for the hIP receptor occurs through both G_αs and G_αq protein pathways (Table 3). Iloprost initiates cAMP production (EC₅₀ = 0.1 ± 0.03 nM) and is favored over IP₃ production (EC₅₀ = 43.1 ± 10 nM) in HEK293 cells expressing recombinant hIP, resulting in the activation of PKA and PKC and their associated pathways respectively (Smyth et al. 1996). Further, basal receptor phosphorylation rapidly occurs and does so only when agonist concentrations exceeds 1 nM, well above that required for cAMP production, indicating that the PKC and not the PKA pathway is responsible. Phosphorylation assays in the presence of PKA- and PKC-specific inhibitors confirm these observations. Although not specifically proven, these results strongly suggest that phosphorylation alters the G-protein specificities and hence the preferred pathway (Smyth et al. 1998).

hIP signal transduction in preadipocytes is somewhat different (Vassaux et al. 1992). Both systems produce cAMP at low concentrations of IP agonist, carbaprostacyclin (cPGI₂) in the case preadipocytes. However, in preadipocytes, the cPGI₂ also produces a transient increase in intracellular [Ca²⁺] that does not depend on cAMP production, extracellular [Ca²⁺] or calcium channel blockers, and is independent of IP₃ production. The pathway for the increase in [Ca²⁺] is not currently known, but the latter observations suggest that the mechanism for this increase differs from the IP₃-independent pathway suggested for DP1.

Regulation

Zhang et al. (2001) show that the regulation of expressed hIP is modulated by the glycosylation state of the hIP receptor. They found that the extent of glycosylation at both N7 and N78 and in particular the glycosylation of N78 is paramount to the efficient binding of receptor agonist. Lack of glycosylation at N7 itself has little effect (≤ 10%) on iloprost binding. However, lack of glycosylation at N78 increases the iloprost K_d seven-fold and lack of glycosylation at both N7 and N78 increases the K_d to undetectable levels, effectively inhibiting binding. Glycosylation also has an effect on signaling efficiency. Similar to the pattern observed for the binding efficiency, lack of glycosylation at N7 reduces cAMP production efficiency by increasing the EC₅₀ value nearly 10-fold, lack of glycosylation at N78 increases EC₅₀ over a 1000 fold, and lack of glycosylation at both sites eliminates cAMP signaling altogether. Interestingly, although the wild type receptor exhibits activation of phospholipase C (PLC) and IP₃ production, deglycosylation at either N7 or N78 eliminates this signal pathway.

The isoprenylation state of C383 also has a direct effect on hIP signaling (Miggin et al. 2002). Although the isoprenylation state does not alter iloprost binding, isoprenylation is required for efficient binding of both G_αs and G_αq proteins and for this reason is required for competent signaling through both the cAMP and IP₃ pathways.

Regulation of hIP also occurs through independent pathways that are agonist stimulated and typically involve phosphorylation of the C-terminal as well. As noted above, elevated concentrations of the hIP agonist iloprost stimulate the activation of PKC, resulting in the phosphorylation of the hIP receptor and the production of IP₃ (Smyth et al. 1996). Later work reveals that the PKC release results in the phosphorylation of S328 and that this event reduces the production of cAMP to effectively basal levels while increasing the production of IP₃ 2-3 fold (Smyth et al. 1998). These results strongly suggest that this phosphorylation alters the G-protein affinity from G_αs to another G-protein, presumably G_αq. It should be noted that phosphorylation at S328 does not account for all agonist-induced receptor phosphorylation and the function of these phosphorylation sites is currently unknown.

G-protein preferences also depend on interactions with other regions of the receptor. Chimeric constructs of IP and TP receptors reveal that the 116-134 intracellular loop of the IP (intracellular loop 2) receptor is necessary for G_αs-mediated cAMP production and that deletion of the YLYAQ sequence from this region not only drastically reduces cAMP production, but also enhances the intracellular [Ca²⁺] through a G_αq signaling pathway (Chakraborty et al. 2013). Further, the intracellular loop 209-235 (intracellular loop 3) is involved in both G_αs and G_αq binding.

Sequestration of hIP through internalization is another known pathway for regulation (Giovanazzi et al. 1997). Although internalization of other PG receptors is mediated by GRK-mediated phosphorylation and β-arrestin binding, hIP receptor internalization has been shown to be independent of GRK-2, -3, -5, -6 and arrestin-2 (Smyth et al. 2000). Further, iloprost-induced internalization requires agonist concentrations (EC₅₀ = 27.6 ± 5.7 nM) similar to that required for PKC-induced phosphorylation of S328 but is independent of this phosphorylation event. Sequestration of hIP proceeds through a dynamin-dependent clathrin-coated vesicular endocytotic pathway and a dynamin-independent pathway (Smyth et al. 2000). One mechanism explaining the dynamin-dependent hIP internalization pathway involves the Rab5a protein, a Ras-like GTPase (O'Keeffe et al. 2008). Upon binding GTP, the now active Rab5a binds to agonist (circaprost)-bound hIP and in concert with dynamin, internalizes the Rab5a-receptor complex into punctate endocytotic vesicles until 30-40% of hIP is internalized. The internalization is clearly directed by Rab5a, as over co-expression of Rab5a with hIP significantly increases internalization. Rab5a is released upon hydrolysis of GTP and the hIP receptor remains internalized for several hours post-stimulation. By 4 hours after stimulation, 50% of the sequestered receptor returns to the surface. Upon prolonged stimulation (4-8 h), 50% of the internalized hIP fails to return to the surface and thus likely undergoes degradation through trafficking to lysosomes. Interestingly, the C-terminal tail, directly involved in other PG receptor internalizations, is not involved in facilitating Rab5a-mediated endocytosis itself, but instead appeared to play a role in hIP sorting following the Rab5a-mediated endocytosis. Additional mechanisms have been presented showing the involvement of the related Rab11a and Rab4a proteins (Wikström et al. 2008). hIP has been shown to localize to both Rab11a- and Rab4a-positive vesicles in response to cicaprost binding. As observed for Rab5a, Rab11a also interacts directly with hIP, however, the C-terminal tail of hIP is required for interaction with Rab11a, unlike the Rab5a interaction. On the other hand, Rab4a does not directly interact with hIP, nor does it direct hIP to Rab4a positive vesicles.

Although the precise pathway for sequestration is currently unknown, it has been shown that glycosylation of the extracellular N7 and not the extracellular N78 is required for internalization (Zhang et al. 2001). Internalization is also affected by the prenylation state of C383. Farsenyl or geranylgeranyl prenylation of C383 is also required for efficient internalization of the IP receptor, possibly by allowing the insertion of the prenyl group into the membrane thus forming a fourth cytoplasmic loop that aids internalization (Miggin et al. 2002).

As noted above, hIP undergoes internalization in response to agonist stimulation and a portion of the internalized receptors are later recycled back to the membrane while some are transported to lysosomes for degradation. As reported for other proteins, including other GPCR receptors (Hanyaloglu and von Zastrow 2008), targeting to lysozomes requires prior polyubiquitination of the protein. Polyubiquitinated hIP has been observed in HEK cells and is required for transport to lysozomes, but it is not known if polyubiquitination occurs prior to internalization or post-internalization (Donnellan and Kinsella 2009) [99].

Regulation at the transcriptional level has also been reported. Turner and Kinsella (2010) report that estrogen significantly increases the transcription of hIP in primary aortic smooth muscle cells and that the increase is modulated through the binding of the ERα nuclear estrogen receptor to the hIP promotor region on the DNA itself. The ERβ nuclear receptor is apparently not involved. The authors postulate that this interaction in part explains the vasodilatory effect of estrogen as well as its anti-atherogenic effects (see also Tostes et al. 2003). Eivers and Kinsella (2016) report that androgens also confer cardioprotective effects through stimulation of hIP transcription in human erythroleukemia (HEL) cells and human endothelial cells (EA.hy926). Utilizing dihydrotestosterone (DHT) as stimulus, hIP mRNA levels were significantly increased, 3.9-fold in the case of HEL cells. Further, in combination with low serum cholesterol, DHT enhances the binding of the sterol regulatory element binding protein (SREBP1) to the hIP promoter thus increasing transcription. The mechanism by which low cholesterol also helps facilitate this process is thoroughly discussed by Heemers et al. (2006).

TP receptor

Introduction

The human thromboxane A₂ (TXA₂) receptor is expressed as two isoforms referred to as hTRXα and hTRXβ. hTRXα (TBXA2R, UniprotKB-P21731) is translated as a 343 amino acid polypeptide with a calculated molecular weight of 37.5 kDa whereas hTRXβ (UniprotKB-P21731-2) is translated as a 407 amino acid polypeptide with a calculated molecular weight of 44.2 kDa (Hirata et al. 1996). The two isoforms are identical for the first 328 amino acids and differ in their C-terminal tails due to differential splicing of exon 3, leading to vastly different intracellular signaling. There are numerous coding SNP variants reported (R60L, P28S, C68S, V80E, E94V, A160T, V176E, V217I, D238G, V241G, S283C, and D304N) of which several result in receptor dysfunction. R60L results in a receptor with a defective response to its natural agonist TXA2 (Hirata et al. 1994a). R60L, V241G, and D304 N result in susceptibility to platelet-type bleeding disorder 13 (https://www.ncbi.nlm.nih.gov/clinvar, Landrum et al. 2016). S283C and D238G lead to impaired platelet aggregation (https://www.ncbi.nlm.nih.gov/clinvar, Landrum et al. 2016). Both receptors are seven helix transmembrane proteins. There are no reported X-ray structures and one reported structural model for the alpha isomer (PDB reference 1LBN). Two N-linked extracellular glycosylations are predicted with high confidence for N4 and N16 (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004) and confirmed definitively for hTRXα and potentially for hTRXβ by Walsh et al. (1998). Confirmation for hTRXα was obtained through examination of N to Q site directed mutagenesis of recombinant receptor. Treatment of HEK293 cells expressing individual isomers with tunicamycin, a potent inhibitor of N-glycosylation, reduces the expression of both isomers indicating that glycosylation is not only present on both isomers but is also required for proper folding in the ER. Further, treatment of the same cell lines individually with endo-H, an enzyme that cleaves off the glycol portion of N-linked sugars, reduces the ligand binding efficiencies by approximately 50%, confirming the presence of N-linked glycosylation on both isoforms and indicating that the glycosylation is involved in ligand binding. There are also numerous high confidence cytosolic O-linked sites predicted (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013). Specific sites for these glycosylations have not been experimentally confirmed. There is potential phosphorylation predicted for both isomers, T286 on the second extracellular loop, S139 and S239 on cytoplasmic loops, and S324 on the C-terminal domain. Both isomers have additional and different phosphorylation sites predicted for the C-terminal tail (hTRXα: S329 and S331; hTRXβ: S330, S355, S360 and S404) (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Although in vivo and in vitro phosphorylation of both receptors has been confirmed (Kinsella et al. 1994; Habib et al. 1999), not all potential sites have been confirmed experimentally. Deletion and substitution mutants have confirmed hTRXα phosphorylation at T337 (Kelley-Hickie and Kinsella 2004) and S329 (Walsh et al. 2000a) and hTRXβ phosphorylation at S145, S239, S357 (Kelley-Hickie and Kinsella 2006), and T399 (Kelley-Hickie and Kinsella 2004). Numerous additional GRK phosphorylations are also predicted but not experimentally confirmed (http://gps.biocuckoo.cn/, Xue et al. 2011).

Expression and Characterization

The human TP receptor (hTP) is widely expressed with the highest amounts in the endocrine system, female tissues, and the immune system (http://www.proteinatlas.org, Uhlén et al. (2015)). Most tissues express both isoforms with hTRXα predominating over hTRXβ. Levels of hTRXα mRNA expression are relatively constant throughout most tissue types while mRNA levels of hTRXβ vary significantly (Miggin and Kinsella 1998).

Many ligand binding and efficacy studies do not specify which particular isoform was under study, albeit hTRXα is the likely candidate. Further, tissue studies of TP receptors would be sampling both isoforms simultaneously. For such studies hTP is used to designate either or both human isoforms. Fortunately, the ligand binding properties of the two isoforms were found to be identical (Hirata et al. 1996) and thus differentiation between isoforms is unnecessary for ligand binding studies (Table 7). Efficacy studies are a bit more complex, as these receptors can couple to an array of G-proteins and in a isoform-dependent manner for which the cellular response may differ. Ligand binding properties for hTP have been characterized with recombinant receptor in HEK293 (Abramovitz et al. 2000; Capra et al. 2013) and COS-m6 (Hirata et al. 1996) cell lines as well as in human platelet membranes (Modesti et al. 1989; Miki et al. 1992; Armstrong et al. 1993; Dorn 1989) by displacement of synthetic radiolabeled agonists (Table 7). Ligand binding to recombinant murine TP expressed in CHO cells (Sawyer et al. 2002; Kiriyama et al. 1997) has also been examined. Ligand efficacies for hTP have been determined using recombinant receptor in COS-m6 (Hirata et al. 1996), CHO (Hirata et al. 1996, 1994b), and HEK293 cell lines (Capra et al. 2013) as well as in human hand vein (Arner et al. 1991), lung (McKenniff et al. 1988) tissues and platelets (Dorn 1989, 1991; Mayeux et al. 1988; Tymkewycz et al. 1991; Ushikubi et al. 1989b), all with synthetic agonists (Table 7). TXA₂ cannot be used as an agonist itself, as it is an unstable AA metabolite with a half-life of about 30 seconds (Ricciotti and FitzGerald 2011). PGH₂ can also bind and activate TP receptors (Gluais et al. 2005), but its short in vitro half-life of 5 min (Yu et al. 2011) severely limits its usefulness in such studies. Conversion of PGH₂ to multiple prostaglandins products in vivo shortens the half-life even further. The most commonly used selective TP agonists are U-46619, I-BOP, and STA₂ (Abramovitz et al. 2000; Hirata et al. 1996) and the most commonly used selective TP antgonists are SQ29548 and S-145 (Wright et al. 1998; Ushikubi et al. 1989b). Ah-23848, Ramatroban, and Vaproprost have also been used (Jones et al. 2009) (Table 2).

Table 7.

Binding affinity (K_i) and signaling efficiency (EC₅₀) of prostanoids and agonists on the recombinant TP receptors expressed in cell cultures and in human platelets, lung, and vascular smooth muscle

	Ki (nM)
Receptor	U46619	I-BOP	STA2	PGE2	PGF2a	Reference
hTP/HEK293a	35 ± 5	-	-	29,000 ± 6,702	8,700 ± 670	Abramovitz et al. 2000
hTP/HEK293b	100	16	-	2000c	5000c	Capra et al. 2013
hTP/COS-m6c	-	25	-	-	-	Hirata et al. 1996
mTP/CHOd	67	0.56	14	-	-	Sawyer et al. 2002, Kiriyama et al. 1997
hTP/plateletse	16 ± 5		-	-	-	Modesti et al. 1989
hTP/plateletse	39 ± 4.7	-	-	-	-	Miki et al. 1992
hTP/plateletsf	221 ± 21	-	87 ± 19	-	-	Armstrong et al. 1993
hTP/plateletsg	8.5 ± 2.5	0.21 ± 0.03	-	-	-	Dorn 1989
hTP/vascular smooth muscleg	88 ± 10	0.18 ± 0.02	-	-	-	Dorn 1991
hTP/purified in Tris/asolectine	19,900	-	870	-	-	Ushikubi et al. 1989a
	EC50 (nM)
Receptor	STA2	I-BOP	U46619	TXA2	PGH2	Reference
hTPα/COS-m6h	2	-	-	-	-	Hirata et al. 1996
hTPβ/COS-m6h	2	-	-	-	-	Hirata et al. 1996
hTPα/CHOi	-	17	-	-	-	Hirata et al. 1996
hTPβ/CHOi	-	NA	-	-	-	Hirata et al. 1996
hTP/CHOh	4	-	-	-	-	Hirata et al. 1994b
hTP/HEK293h	-	25	99	-	-	Capra et al. 2013
hTP/hand veinj	-	-	40	-	-	Arner et al. 1991
hTP/lung parenchymalk	-	-	32	-	-	McKenniff et al. 1988
hTP/bronchiolar ringl	-	-	16	-	-	McKenniff et al. 1988
hTP/plateletsm	-	0.209 ± 0.024	-	-	-	Dorn 1989
hTP/plateletsn	-	-	-	163 ± 21	45 ± 2	Mayeux et al. 1988
hTP/plateletsn	-	-	4.8 ± 0.2	-	-	Tymkewycz et al. 1991
hTP/plateletso	90	-	-	-	-	Ushikubi et al. 1989b

Abbreviations: h, human; HEK293, human embryonic kidney cells; COS M-6, Green monkey kidney cells; CHO. Chinese hamster ovary cells. The K_i values were determined from [³H]agonist displacement assays. The EC₅₀ determined from intracellular changes in IP₃, cAMP, [Ca²⁺] or the indicated physiological changes. NA= no activity. a) K_i determined from the displacement of [³H]SQ29548; b) IC₅₀ estimated from a graph of agonist vs. [³H]SQ29548 and the K_i value calculated using Cheng-Prusoff equation (Suganami et al. 2016); c) IC₅₀ estimated from graph of agonist vs. [³H]S-145 and the K_i value calculated using Cheng-Prusoff equation; d) K_i determined from the displacement of [³H]S-145; e) K_i determined from the displacement of [³H]U46619; f) K_i calculated from reported IC₅₀ of agonist vs [³H]SQ29548 using Cheng-Prusoff equation; g) K_i for high binding site calculated from IC₅₀ of agonist vs. [¹²⁵I]-BOP using Cheng-Prusoff equation; h) EC₅₀ determined from IP₃ generation; i) EC₅₀ determined from cAMP generation determined from titration curves; j) EC₅₀ determined from % maximal contraction of vein ring segments; k) EC₅₀ determined from % maximal contraction of lung parenchymal strip; l) EC₅₀ determined from % maximal contraction of bronchiolar ring segments; m) EC₅₀ determined from [Ca²⁺] increase; n) EC₅₀ determined from initiation of platelet aggregation; o) EC₅₀ determined from inhibition of PGI₂ analog iloprost (3 nM) induced cAMP increase.

Mechanism of Cell Activation

The TP receptor is rather promiscuous in its G-protein coupling and thus the ultimate effect of activating the TP receptor is dependent on the availability of particular G-proteins which in turn is cell type and tissue dependent (Table 3). Early work characterized hTP receptor in platelets as coupling with G_αq (Shenker et al. 1991), known to stimulate IP₃ and diacylglycerol (DAG) production via PLC-β and a subsequent increase in intracellular [Ca²⁺] and PKC. Later work with recombinant proteins expressed in HEK293 cells reveals that two specific members of the G_αq family, G_α11 and G_α16, couple to hTP (Kinsella et al. 1997; Walsh et al. 2000b). Although both hTP isoforms were shown to couple through the G_αq family, only hTRXβ also coupled to G_αi, which leads to a decrease in cAMP production by inhibiting adenylate cyclase, whilst hTRXα appears to also couple with G_αs, as stimulation with the I-BOP agonist leads to an increase in cAMP production (Hirata et al. 1996). The interaction between hTP and the specific G-protein appears to be regulated by different domains of the receptor where residue R60 of hTP is involved in the binding of G_αq and G_αs, but not G_αi (Hirata et al. 1996; Hirata et al. 1994b). Both hTRXα and hTRXβ couple to G_α12, and quite likely to G_α13 as well (Offermanns et al. 1994; Becker et al. 1999), and upon stimulation, intracellular [Ca²⁺] increases in an IP₃/DAG independent manner (Walsh et al. 2000b). Apparently, G_α12 couples both hTP isoforms to verapamil-sensitive, L-type Ca²⁺ channels. The hTRXα receptor couples to G_α13, leading to an increase in Na⁺/H⁺ exchange. Agonist binding to the alpha isoform is enhanced when coupled to G_α13 (Becker et al. 1999). Activation of G_α12 or G_α13 also leads to activation of guanine nucleotide exchange factor (GEF) for Rho, p115RhoGEF, that initiates Rho-mediated signaling (Kozasa et al. 1998). Both hTP isoforms also bind to the high molecular weight (70 kDa) G_h protein (Vezza et al. 1999). Upon stimulation, the hTRXα-G_h complex enhances IP₃/DAG production via activation of PLC-d1 and ultimately an increase in intracellular [Ca²⁺] and active PKC (Feng et al. 1996). The hTRXβ-G_αh complex does not activate IP₃/DAG production and its biological function has yet to be determined. Interestingly, G_αh also functions as a tissue transglutaminase that is inhibited by GTP binding, suggesting that the β-isoform may be involved in regulation of the transglutaminase function.

Regulation

Regulation of expressed hTP is modulated by the glycosylation state of the receptor by multiple mechanisms. Walsh et al. (1998) reported that reduction of N-glycosylation of hTRXα through the use of N to Q mutants reveals that loss of N4, N16 or both glycosylations results in a reduction of agonist binding by 53%, 42% and 92% respectively. The reductions are caused by a reduction in maximal binding (Bmax) and not by changes in affinity. The reduction in Bmax is not a function of membrane expression, as the expression of mutants lacking either N4 or N16 glycosylation are the same as wild type. In addition, the membrane expression of the double mutant is reduced by 45% compared to wild type. As would be expected, N-glycosylation also affects the second messenger signaling. Intracellular [Ca²⁺] mobilization by hTRXα for which glycosylation is removed from N4, N16 or both showed reductions in [Ca²⁺] of 22%, 11%, and 58% respectively, significantly less than the reduction in ligand binding. Intracellular cAMP production is likewise reduced in the absence of glycosylation at N4, N16 or both where reductions of 12%, 25%, and 58% respectively were observed. Similar results would be expected for hTRXβ, with the exception of cAMP production, as the structures are identical with the exception of the cytosolic C-terminal tail. This expectation is supported by the observation by Walsh et al. (1998) that tunicamycin, a potent inhibitor of N-linked glycosylation, treatment of HEK293 cell lines individually expressing recombinant hTRXα and hTRXβ reduces Ca²⁺ mobilization by 84% and 88% respectively. The larger reduction in Ca²⁺ mobilization for hTRXα in this experiment compared to the double mutant is likely due to retention of residual signal transducing activity by the double mutant.

Regulation through desensitization occurs through both homologous (receptor specific) and heterologous (highly specific for other receptors) modalities, each isoform taking a different pathway (Parent et al. 1999; Murray et al. 1990). Habib et al. (1997) examined the agonist-dependent phosphorylation of both hTP_α and hTP_β transfected into HEK293 cells. Using the specific stable TP-specific agonist U46619 they found that both isoforms are phosphorylated with similar time and dose dependency and phosphorylation reduces both the production of IP₃ and the increase in intracellular [Ca²⁺]. Current thought is that the phosphorylation prevents proper docking of G-proteins to the receptor, thus preventing or reducing signal transmission. Further, their data suggest that the phosphorylation leading to desensitization is mediated primarily by kinases other than PKA or PKC. However, they did report that PKC and PKA marginally influence receptor phosphorylation, but not at residues that influence desensitization and suggest that phosphorylation by these kinases may be involved in a heterologous desensitization. In a related set of experiments, Zhou et al. (2001) have shown that treatment of HEK293 cells expressing recombinant hTPα with the TP-specific agonist I-BOP induces both receptor phosphorylation and Ca²⁺ release in a time- and dose-dependent manner. Following pretreatment (10 min) with I-BOP, a second challenge with I-BOP abolishes Ca²⁺ induction, clearly demonstrating homologous desensitization. Further, co-transfection of hTPα with GRK5 and GRK6 augments the I-BOP phosphorylation and inhibits the I-BOP-induced Ca²⁺ release. Without pre-stimulation with I-BOP, the GRKS do not facilitate receptor phosphorylation, showing that GRKs must be recruited to the receptor by agonist-induced phosphorylation before GRK-facilitated phosphorylation can occur. Wang et al. (1998) suggest a potential source for the initial phosphorylation. They report that a cGMP-dependent G kinase phosphorylates the TP receptor that in turn inhibits IP₃ release and increases intracellular [Ca²⁺]; specific α or β isoforms were not investigated. They also show that PKA is not involved. Inactivation of hTPβ and not hTPα is also facilitated by internalization. Parent et al. (1999) report that following stimulation with U46619 of hTPβ transfected in HEK293 cells, the receptor internalizes to a to a plateau of about 40% in 2-3 hours whereas transfected hTPα do not internalize after 3 hours of stimulation. Further internalization requires the presence of GRK (GRK2 and potentially either GRK 3, GRK5, or GRK6), arrestin-2 or -3 and dynamin, suggesting the involvement of clatherin-coated pits that lead to punctate vesicle formation. Overexpression of arrestins significantly promotes internalization whereas GRK co-expression only increases internalization slightly, suggesting that arrestins are the limiting factor for internalization. Utilizing hTPβ transfected into HEK293 cells, Kelly-Hickie and Kinsella (2006) show that PKC does indeed phosphorylate hTPβ and this leads to 20% partial and transient agonist-induced impairment of activity. In addition, they observed that GRK-mediated, agonist-induced phosphorylation leads to a sustained desensitization. Utilizing specific serine to alanine mutations in transfected hTPβ receptors they found that S145 on a cytoplasmic loop is the target of PKC phosphorylation and that S239 and S357 on the C-terminal tail are the targets for GRK phosphorylation. In a complementary study Kelley-Hickie et al. (2007) demonstrated that like hTPβ, phosphorylation of S135 by PKC partially and transiently impairs hTPα signaling. Unlike hTPβ, GRK phosphorylation of the C-terminal tail is not involved in sustained desensitization. Instead they found that both PKC and cGMP-dependent protein kinase (PKG) phosphorylated hTPα at T337 and S331 respectively, leading to profound desensitization of hTPα, each kinase of which is stimulated through the TP signaling pathway.

Heterologous desensitization occurs through both the desensitization of the G-protein rather than the receptor itself and through the action of other receptors. Examples of the former have been reported by several groups. Both G_α12 an G_α13 are known signal transducers for hTP. They have been shown to be phosphorylated by PKC in response to increasing cAMP levels created by other receptor signaling and that this phosphorylation results in the loss of hTP signaling (Manganello et al. 1999; Offermanns et al. 1996). The phosphorylation reduces the affinity of the α subunit for the βγ subunits and thus prevents signaling of both receptor isoforms through the affected G-protein (Kozasa and Gilman 1996). Murray et al. (1990) have reported that stimulation of the hIP receptor with receptor specific agonists results in a PKC-independent desensitization of only one of the two isoforms hTP, the one with strong binding to the antagonist GR3291 shown by others to be the hTPα receptor (Habib et al. 1999). These results were confirmed by Walsh et at (Walsh et al. 2000a) and they further show that the desensitization of hTPα occurs through direct PKA phosphorylation of S329 which is located on the C-terminal domain of the hTPα receptor.

FP receptor

Introduction

The human prostaglandin F_2α receptor (hFP, PTGFR, UniprotKB-P43088) is expressed as a 395 amino acid transcript with a calculated molecular weight of 40.0 kDa. There are six reported splice isoform variants (Liang et al. 2008a) obtained from mRNA data that differ only after residue 266, each of which are calculated to have only 6 transmembrane helices rather than the seven found in the canonical isoform (Cserzo et al. 2002). No X-ray structures are available. Although none of these variants have been found naturally occurring at the protein level to date, all have been transfected into HEK293 cells and are translated into functional protein (Liang et al. 2008a; Vielhauer et al. 2004) (see below). There are two reported coding SNP variants (Q8L and L218S) (https://www.ncbi.nlm.nih.gov/clinvar/, Landrum et al. 2016) neither of which has been reported to affect function. There are two N-linked glycoforms predicted for N4 and N19 but are not experimentally confirmed (http://www.cbs.dtu.dk/services/NetNGlyc/, Blom et al. 2004). However, N-glycosylation of the FP receptor in rats has been shown to be necessary for biological activity (Kitanaka et al. 1994). O-glycosylation is predicted for S234 (http://www.cbs.dtu.dk/services/NetOGlyc/, Steentoft et al. 2013) and has not been observed experimentally. There are potential phosphorylation sites predicted for S29, S94, and S279 on extracellular loops, S62 on a cytoplasmic loop, and S337 and S341 on the C-terminal intracellular domain (http://www.cbs.dtu.dk/services/NetPhos/, Blom et al. 1999). Over a dozen potential GRK phosphorylation sites are predicted for the C-terminal domain as well (http://gps.biocuckoo.cn/, Xue et al. 2011). None of these phosphorylation sites have been confirmed experimentally.

Expression and Characterization

The human FP receptor (hFP) is widely expressed with the highest amounts in smooth muscle, female tissues, urinary bladder, and the gall bladder (http://www.proteinatlas.org, Uhlén et al. (2015)). Only the cannonical isoform (hFP) has thus far been shown to produce an intracellular signaling in response to PGF_2α or the analog and specific agonist fluprostenol (see below). Agonist-induced signaling for recombinant variant 1 (isoform 2) and 4 (hFPs, isoform 5) have been tested and results show that the latter does not produce an intracellular signal in response to PGF_2α (Liang et al. 2008a) and the former does not even bind the agonist (Vielhauer et al. 2004). These findings make sense, as it has been shown that the seventh transmembrane domain, missing in all variants, is required for agonist binding (Neuschäfer-Rube et al. 2003). Ligand binding properties for FP have been characterized for human recombinant receptors, for recombinant mouse receptor (mFP), as well as bovine receptor (bFP) in corpus luteum tissue homogenates by displacement of radiolabeled agonists (Table 8). Ligand efficacies for recombinant hFP, mFP and rat receptor (rFP) as well as in rat hepatocytes and astrocytes have been determined with a variety of agonists. FP receptors are the least selective of the prostanoid receptors; reported EC₅₀ values are given in Table 8. The most selective agonists are Fluprostenol, Cloprostenol, and Latanoprost (Abramovitz et al. 2000; Anderson et al. 1999; Sharif et al. 2000b) whereas the most highly selective antagonists are OBE022, THG113 and AS604872 (Pohl et al. 2018; Peri et al. 2002; Jones et al. 2009) (Table 2).

Table 8.

Binding affinity (K_i) and signaling efficiency (EC₅₀) of prostanoids and agonists on the recombinant FP receptors expressed in cell cultures and in various tissues

	Ki (nM)
Receptor	PGE2	PGF2a	PGD2	fluprostenol	Iloprost	U46619	bimatoprost	Reference
hFP/HEK293a	119 ± 12	3.2 ± 0.3	6.7 ± 0.5	-	11 ± 1	241 ± 8	-	Abramovitz et al. 2000
mFP/CHOa	100	3	47	-	-	1000	-	Narumiya et al. 1999, Wheeldon and Vardey 1993
hFP/COS-M6b	85	2.8	7.0	3.5	1200	146	-	Abramovitz et al. 1994
bFP/corpus leuteumc	110	10	17	-	-	-	-	Anderson et al. 1999
hFP-altFP4 dimer	-	100	-	-	-	-	-	Liang et al. 2008a
bFP/corpus leuteuma	3400 ± 710	130 ± 6	2500 ± 760	98 ± 9	-	-	6310 ± 1650	Sharif et al. 2003
	EC50 (nM)
Receptor	PGE2	PGF2a	PGD2	fluprostenol	Iloprost	U46619	bimatoprost	Reference
hFP/HEK293a	-	10 ± 1.7	-	-	-	-	-	Kunapuli et al. 1997
rFP/rat myocytesa	-	50 ± 12	-	-	-	-	-	Kunapuli et al. 1998
hFP/COS7b	-	60.0	-	-	-	-	-	Neuschäfer-Rube et al. 2003
hFP/HEK293c	-	11.2	-	-	-	-	-	Woodward et al. 2000
rFP/A7r5a	2607 ± 270	30.90 ± 2.82	222.0 ± 71.4	4.45 ± 0.19	NA	5900 ± 1230	-	Griffin et al. 1998
hFP/HEK293c	-	-	-	-	-	-	4762	Woodward et al. 2003
mFP/Swiss 3T3a	-	-	-	4	-	-	-	Sharif et al. 2000b
rFP/hepatocytes	-	1000	-	50	-	-	-	Meisdalen et al. 2007
hFP/HEK293	-	29 ± 2	-	4.6 ± 0.4	-	-	168 ± 165	Sharif et al. 2003

Abbreviations: h, human; HEK293, human embryonic kidney cells; COS7, monkey fibroblast-like cells; CHO. Chinese hamster ovary cells; A7r5, rat aorta cells; Swiss 3T3, Swiss albino mouse embryo tissue cells. The K_i values were determined from [³H]PGF_2α displacement assays. The EC₅₀ determined from intracellular changes in [IP₃] or [Ca²⁺]. a) IC₅₀ estimated from a graph of agonist vs. [³H]PGF_2α and the K_i value calculated using Cheng-Prusoff equation (Suganami et al. 2016); b) K_i value calculated from reported IC₅₀ for displacement of using Cheng-Prusoff equation; c) EC₅₀ determined from change in [IP₃].

Mechanism of Cell Activation

The hFP receptor couples primarily through G_aq leading to an IP₃/Ca²⁺ signaling pathway (Table 3) (Breyer et al. 2001; Abramovitz et al. 1994). Potential coupling to G_αs has also been reported as an explanation for the increase in cAMP that follows agonist-induced desensitization of FP in bovine iris sphincter (Tachado et al. 1993). Coupling with other G-proteins has not been reported. However, the rabbit kidney recombinant FP receptor known to couple G_αi has a 98.3% sequence homology to hFP, suggesting a potential for G_αi coupling. Recombinant ovine FP (FP_A) expressed in HEK293 cells has been reported to couple through G_α12/13, resulting in stimulation of the small G-protein Rho (Pierce et al. 1999). However, the lower overall sequence homology between ovine FP_A and hFP (83.9%) make it less likely that hFP might also couple with G_α12/13, especially since the C-terminal domains that interact with the G-proteins have a lower sequence homology (78.7%).

As noted above, there are six truncated variants of FP, none of which have been shown to exhibit biological signaling activity when expressed alone. However, an hFP dimer with variant 4 (hFP-altFP4) expressed in HEL293 cells has been shown to be activated by the prostamide PGF_2α analog bimatoprost whereas hFP alone shows less than 20% the response and altFP4 shows no response at all to this amide analog (Liang et al. 2008a). Further, both the hFP monomer and the heterodimer exhibit identical signaling responses to PGF_2α. Clearly, the addition of the altFP species extends the agonist repertoire to amide analogs, lending a possible biological function for these receptor isoforms. Although not yet shown, heterodimer formation with the other altFP isoforms may also modify the agonist specificity of the FP receptor.

Regulation

Regulation of hFP has been shown to be modulated in several different ways. Tachado et al. (1993) confirmed the coupling of bovine FP to G_αq leading to the activation of PLC. They also observed that PGF_2α stimulation of FP leads to short-term desensitization that results in an increase in cAMP and attenuation of PLC activity. Both events involve PKC, known to activate adenylate cyclase. The authors suggest a possible explanation involving FP switching from G_αq signaling to G_αs signaling, the latter known to activate adenyl cyclase, thus increasing basal cAMP. The suggestion is qualified by the potential for a yet-to-be discovered FP receptor that normally signals through G_αs. It is interesting to note that the potential phosphorylation sites on the C-terminal domain are high probability sites for PKC phosphorylation and that phosphorylation of other receptors in this region is known to modify G-protein binding specificity. Agonist-induced hFP desensitization has also been observed for recombinant hFP expressed in HEK293 and NIH3T3 cells (Kunapuli et al. 1997), but the connection to either cAMP or PKC has not been explored. FP regulation on the transcriptional level has also been reported. Liang et al. (2008b) examined regulation of hFP expression in human uterine myocytes. Here they found that hFP mRNA expression is negatively regulated by progesterone and cAMP and upregulated by the inflammatory cytokines IL-1β and TNFα that in turn activate the transcription factor NFkB and PKC, the latter also shown to participate in the upregulation of hFP. The results explain the decrease in myometrial expression of hFP during pregnancy and the increase associated with labor.

Conclusion

In this review the expression, characterization, regulation, and mechanism of action of prostanoid receptors are summarized with a focus on human receptors. Prostanoid receptors control numerous biological functions not only through the diversity in prostaglandins themselves, but through the diversity of receptor isoforms and the heterogeneity of G-protein coupling. Commonality is found in regulation where agonist-induced receptor desensitization is accomplished through phosphorylation by PKA, PKB, PKC or a variety of different GRKs. Although much is known about prostanoid receptor signaling, there are still many missing pieces. It is our hope that this review will help underscore these areas and stimulate research to find these missing pieces.

Availability of data and material

Not applicable

Compliance with ethical standards

Conflicts of interest/Competing interests

The author declares that he has no conflicts of interest or competing interests.

Ethics approval

Not applicable

Consent to participate

Not applicable

Consent for publication

Not applicable

Code availability

Not applicable