Introduction

Top Summary Introduction Procedures Results Discussion References

The molecular identification of eight G protein-coupled membrane receptors that mediate the actions of the five primary prostaglandins has helped to explain the myriad of biological effects of these arachidonic acid metabolites. In particular, PGE₂ affects almost every tissue in the body (often in opposing ways), including smooth muscle contraction and relaxation and pro-inflammatory and anti-inflammatory actions. Four subtypes of the PGE₂ receptor, termed EP₁, EP₂, EP₃, and EP₄, have been cloned (Pierce et al., 1995) and shown to couple to different signaling systems. The EP₁ receptor couples preferentially to G_q, the EP₂ and EP₄ receptors couple to G_s, and the six EP₃ receptor carboxyl tail splice variants couple to G_i. Thus, these four receptors respond quite differently to the same physiological ligand.

Surprisingly, when the deduced amino acid sequences for the four EP receptors are aligned and compared, they demonstrate an unexpected degree of divergence. Alignments of all the prostaglandin receptors showed that the EP₂ receptor is more similar to the IP receptor and the DP receptor than to the other three EP receptors. Phylogenetic analysis (Regan et al., 1994; Toh et al., 1995) of receptor sequences led to the conclusion that the prostaglandin receptors evolved from a precursor EP receptor into two subfamilies that differ with respect to their G protein coupling (Fig. 1A). Receptors that preferentially interact with the major endogenous prostaglandins other than PGE₂ must have evolved following the functional division of the EP receptors. Thus, the IP and DP receptors evolved from the EP₂ receptor, and the FP and TP receptors evolved from the EP₃ or EP₁ receptor. If this hypothesis is correct, a relatively small number of amino acids may determine the selective interactions of the prostaglandins with their receptors.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Phylogeny of human prostaglandin receptors. A, Alignment of deduced amino acid sequences of eight cloned human prostaglandin receptors was used to generate the phylogenetic tree (Regan et al. 1994). Receptors linked to Gs are clustered, as are those that are linked to Gq and/or Gi. B, Deduced amino acid sequences of three Gs-linked receptors (EP2, IP, and EP4) were aligned using the University of Wisconsin GCG program. The most highly conserved regions are shown. Underlined residues in the EP2 receptor are those that were mutated, usually to the corresponding residue from the IP receptor. EX, extracellular loop.

The EP₂ and the IP receptors represent an ideal receptor pair for studying the determinants of ligand selectivity. Their seven TMs are more than 60% identical at the amino acid level and they share some common ligands, such as PGE₁. However, PGE₂ is more than 1000-fold selective for the EP₂ receptor, and the stable prostacyclin analog iloprost is more than 1000-fold selective for the IP receptor. We identified approximately 15 residues that were similar in the EP₂ and EP₄ receptor but differed in the IP receptor. To determine whether these amino acid residues were responsible for the preferential activities of PGE₂ and iloprost, the EP₂ receptor sequence was changed at these positions to the corresponding IP receptor amino acid (Fig. 1B) by site-directed mutagenesis and the effect on activation by PGE₂ and iloprost determined in a reporter gene assay. PGE₁, a ligand at both the EP₂ and IP receptors, was used to confirm that the mutant receptors were functionally expressed.

A single amino acid change in TM7 enables iloprost to activate the EP₂ receptor. This mutant receptor is activated by both PGE₂ and iloprost, but not by PGD₂ or PGF₂. This represents the first report of a single amino acid change in a prostaglandin receptor resulting in a gain of function for prostaglandin ligands and is consistent with the proposed phylogeny of the prostaglandin receptors.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. Iloprost and [³H]acetyl Coenzyme A were purchased from Amersham Life Sciences (Arlington Heights, IL). All other prostaglandin compounds were purchased from Cayman Chemical (Ann Arbor, MI). LipofectAMINE, Opti-MEM, and other tissue culture media and serum were purchased from Life Technologies (Gaithersburg, MD). Stripped fetal bovine serum was obtained from Gemini Products (Calabasas, CA). Acetyl coenzyme A, chloramphenicol, and DNase I were purchased from Sigma Chemical (St. Louis, MO).

Site-directed mutagenesis. Missense mutations were introduced by the Kunkel method (Kunkel, 1985) using a Muta-Gene kit purchased from BioRad (Richmond, CA). The human EP₂ receptor cDNA (Regan et al., 1994) was placed in pcDNA3 (InVitrogen, Carlsbad, CA) for all mutagenesis and expression studies. Oligonucleotides were purchased from Genosys (The Woodlands, TX) or synthesized in-house on an Oligo1000 oligonucleotide synthesizer (Beckman Instruments, Fullerton, CA). Mutations were verified by DNA sequence analysis using a Sequenase kit (Amersham Life Sciences, Arlington Heights, IL). The double mutant EP₂ S120G L304Y was constructed from two single mutants by using the internal ApaI restriction site and standard protocols.

Expression of EP₂ cDNAs in cell culture. CV-1 cells were transiently transfected with plasmids carrying the wild-type or mutant EP₂ cDNA using lipofectamine. The CRE-CAT reporter plasmid TESblgIICRE(+)NHSE was obtained from Dr. Pamela Mellon of The Salk Institute (La Jolla, CA). This construct contains an 18-base-pair CRE from the promoter of the  subunit gene for the human glycoprotein hormone linked to the herpes simplex virus thymidine kinase promoter, which drives bacterial CAT gene expression (Delegeane et al., 1987). Thus, CAT enzyme activity is dependent on and proportional to cAMP levels in the cells. Fifty thousand cells were plated in wells of a 24-well plate and transfected with 125 ng of receptor plasmid, 250 ng of reporter plasmid, and 6 µg of lipofectamine per well. Cells were fed with Dulbecco's modified Eagle's medium containing 20% stripped fetal bovine serum at approximately 5 hr. Cells were dosed with drug approximately 24 hr after feeding, and assayed for CAT activity approximately 18 hr after dosing.

CAT assay. Medium was removed by aspiration and the cells washed twice with ice-cold phosphate-buffered saline (1× = 0.02% KCl, 0.02% KH₂PO₄, 0.8% NaCl, 0.216% Na₂HPO₄·7H₂O) without calcium or magnesium. Lysis buffer (50 µl) containing 1% Triton X-100, 1 mM Tris·HCl, pH 7.8, and 2 mM EDTA, pH 8.0, and 0.4 mg/ml DNase I was added and cells lysed on ice with periodic shaking for 45 min. Reaction mix (50 µl) containing 40 µM [³H]acetyl coenzyme A, 60 µM acetyl coenzyme A, 30 µM HCl, 2 mM chloramphenicol, 200 mM Tris·HCl, pH 7.8, and 4 mM EDTA, pH 8.0, was added to the lysate and the mixture incubated for 90 min at 37°. The reaction was stopped with 100 µl of 7 M urea and the entire volume (200 µl) transferred to scintillation vials. One milliliter of scintillant (0.8% 2,5-diphenyloxazole in toluene) was added and the vials shaken to mix the phases. The phases were allowed to separate for 15 min before reading in a scintillation counter, to allow the [³H]acetylated chloramphenicol product to partition into the nonpolar phase (Nielsen et al., 1989). Samples were assayed in triplicate and average dpm values were obtained.

Data analysis. Because PGE₁ is the common ligand for EP₂ and IP receptors, all values were expressed as a percentage of the maximum PGE₁ value. The basal value for each dose-response curve was subtracted from all dpm values. The highest PGE₁ value for each receptor on each assay day was considered 100%. Dose-response curves were generated using KaleidaGraph (Abelbeck/Synergy Software, Reading, PA) by least-squares fits to this equation: response = maximum response + (minimum response  maximum response)/[1 + (concentration of ligand/EC₅₀)]. The data are reported as mean ± standard error of three to 12 independent experiments.

	Results

Top Summary Introduction Procedures Results Discussion References

To identify key residues for ligand discrimination, mutants were screened for CAT activity in a CRE-CAT reporter gene assay, using 0.1, 10, and 100 nM concentrations of PGE₁, PGE₂, and iloprost (data not shown). Because PGE₁ activity should not be affected by mutations that alter ligand selectivity between the EP₂ and IP receptors, it was used to assess the functionality of the mutant receptors. Mutants that were not activated by PGE₁ were assumed to be inappropriately expressed or improperly assembled and were not pursued further. The majority of mutants were unremarkable, in that they retained the ability to signal in response to PGE₁ and PGE₂ and did not gain the ability to signal in response to iloprost (Table 1). Active mutants were assayed for CAT activity over a complete range of doses from 1 nM to 10 µM.

EP₂ activity. The EP₂ receptor demonstrated function, as determined by CAT activity, in response to PGE₁ and PGE₂ (Fig. 2A). EC₅₀ values of 34.0 ± 11.0 nM for PGE₁ and 25.5 ± 6.6 nM for PGE₂ are in the range of previously published values for the EP₂ receptor (Regan et al., 1994; Woodward et al., 1995; Nishigaki et al., 1996). The EP₂ receptor responds to iloprost (Fig. 2A), carbacyclin (Fig. 3), PGD₂, and PGF₂ (Fig. 2E) only at micromolar concentrations, as previously reported.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   Activation of wild-type and mutant human EP2 receptors. Receptors were expressed in CV-1 cells and assayed by the CRE-CAT assay. Drugs used were PGE1 (), PGE2 (), Iloprost (), PGD2 (), and PGF2 (). Points, mean ± standard error of the noted number of experiments performed with triplicate determinations. A, The human EP2 receptor (12 experiments) has EC50 values of 34.0 ± 11.0 nM for PGE1 and 25.5 ± 6.6 nM for PGE2. The EC50 value for iloprost is >10 µM. Basal activity was approximately 10,000 dpm, with a 5.0 ± 2.4-fold increase in the PGE2 signal. B, Mutant EP2 L304Y (eight experiments) has EC50 values of 99.0 ± 29.5 nM for PGE1, 33.3 ± 11.2 nM for PGE2, and 128.9 ± 38.3 nM for iloprost. Basal activity was approximately 15,000 dpm, with a 2.6 ± 0.5-fold increase in PGE2 signal. C, Mutant EP2 S120G (four experiments) has EC50 values of 47.4 ± 21.1 nM for PGE1, 33.0 ± 11.7 nM for PGE2, and >10 µM for iloprost. Basal activity was approximately 17,000 dpm, with a 3.1 ± 0.7-fold increase in the PGE2 signal. D, Mutant EP2 S120G L304Y (eight experiments) has EC50 values of 36.0 ± 10.0 nM for PGE1, 37.9 ± 12.3 nM for PGE2, and 142.1 ± 43.4 nM for iloprost. Basal activity was approximately 10,000 dpm, with a 3.4 ± 0.9-fold increase in the PGE2 signal. E, Wild type EP2 receptor (three experiments) shows activity only in response to PGD2 or PGF2 at concentrations  1 µM. The response to PGE1 from A is included for reference. F, Mutant EP2 S120G L304Y (three experiments) shows activity in response to PGD2 or PGF2 a only at concentrations  1 µM. The response to PGE1 from D is included for reference.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Activation of EP2 and EP2 S120G L304Y mutant receptors by prostacyclin analogs. Receptors were expressed in CV-1 cells and assayed by the CRE-CAT assay. Drugs used were iloprost () and carbacyclin (). Points, mean ± standard error of four experiments performed with triplicate determinations. A, The EP2 receptor shows only weak stimulation by iloprost and carbacyclin. The EC50 values are > 10 µM. B, The EP2 S120G L304Y receptor has EC50 values of 189.8 ± 89.0 nM for iloprost and 1705.0 ± 734.1 nM for carbacyclin.

Residue 304 is a key determinant of selective ligand interaction. The mutant receptor EP₂ L304Y was activated in response to PGE₁, PGE₂, and iloprost (Fig. 2B). The EC₅₀ values are 99.9 ± 29.5 nM for PGE₁, 33.3 ± 11.2 nM for PGE₂, and 128.9 ± 38.3 nM for iloprost. The potency of iloprost is increased at least 50-fold because of the mutation at position 304.

2

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Comparison of signal magnitude of the EP2 L304Y and EP2 S120G L304Y mutant receptors. Points, mean ± standard error of four experiments with triplicate determinations, performed on cells transfected in the same experiment with the two mutant cDNAs. The basal activity was approximately 25,000 dpm for EP2 L304Y and 10,000 dpm for EP2 S120G L304Y, with the average increase over basal for iloprost of 2.6-fold and 3.6-fold, respectively.

Arginine 302 is a key residue for prostaglandin activity. The conserved arginine residue in TM7 has been proposed to be the counterion for binding of the carboxyl group of prostaglandin compounds. Previous studies have demonstrated that alterations of this residue in EP₃ and TP prostaglandin receptors alters ligand binding and signaling (Funk et al., 1993; Huang and Tai, 1995; Negishi et al., 1995; Audoly and Breyer, 1997; Chang et al., 1997). To confirm the significance of this residue in receptor-ligand interactions for the G_s-coupled branch of the prostaglandin receptor family, this arginine has been substituted with a neutral residue (R302Q) and a negatively charged residue (R302E) to evaluate the change in function of these mutants.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Activation of EP2 R302Q and EP2 R302E mutant receptors. Receptors were expressed in CV-1 cells and assayed by the CRE-CAT assay. Drugs used were PGE1 (), PGE2 (), and iloprost (). Points, mean ± standard error of four experiments performed with triplicate determinations. A, The EP2 R302Q mutant has EC50 values of 1123.2 ± 346.5 nM for PGE1 and 238.3 ± 53.8 nM for PGE2. Iloprost was not active. The basal activity was approximately 5000 dpm, with a 2.7 ± 0.5-fold increase in the PGE2 signal. B, The EP2 R302E mutant has EC50 values of 694.3 ± 210.8 nM for PGE1 and 949.8 ± 297.2 nM for PGE2. Iloprost was not active. The basal activity was approximately 5000 dpm, with a 2.5 ± 0.4-fold increase in the PGE2 signal.

	Discussion

Top Summary Introduction Procedures Results Discussion References

The proposed prostaglandin receptor phylogeny (Regan et al. 1994; Toh et al. 1995) led us to hypothesize that only a few amino acids determine the selective interactions of the prostaglandins with their receptors. The approximately 100-fold increase in the activity of an IP receptor-selective agonist with a single amino-acid substitution confirms this hypothesis and is consistent with the proposed phylogeny.

The phylogenetic analysis highlighted candidate amino acids for gain of function mutations that could be rapidly screened with our functional assay (Fig. 1B). This approach avoided the difficulties of interpreting loss of functional or binding activity, which can result from defective receptor synthesis and assembly rather than a change in ligand-receptor interactions. Only one recent study that swapped regions of the IP and DP receptors identified a gain of function, although this was not specific for a single type of prostaglandin (Kobayashi et al., 1997). A nonspecific enhancement of activity could result from a general change in the activation kinetics of the receptors.

The role of residue 304 in ligand activation is specific to prostacyclin receptor agonists. The EP₂ L304Y and EP₂ S120G L304Y mutants acquire the ability to respond to the IP ligand iloprost, but not to the DP ligand PGD₂ or the FP ligand PGF₂ (Fig. 2F). Thus, the specific gain of function seen in the EP₂ L304Y and EP₂ S120G L304Y mutants is likely caused by an alteration in the vicinity of TM7, at a site that specifically interacts with iloprost.

It seems likely that residue 304 is sensitive to structural alteration of the  side chain. Prostacyclin and its stable analog, iloprost, have a constrained configuration because of an additional ring that is not present in PGE₂. PGE₁, which activates both EP₂ and IP receptors, has a much more flexible  chain and presumably can adopt a configuration that enables interaction with either receptor. The iloprost  chain, which differs from prostacyclin and PGE₂, does not seem to interact with residue 304 because carbacyclin, another prostacyclin analog that has an  chain identical to prostacyclin and PGE₂ and a constrained  chain, also gains function with the EP₂ S120G L304Y mutant (Fig. 3). It is interesting to note that the potency of carbacyclin relative to iloprost for the mutant is similar to its relative potency in binding to the IP receptor (Boie et al., 1994). The data are thus consistent with the L304Y substitution's enhancing the ability of the prostacyclin  chain to interact with the EP₂ receptor. One explanation for the result is that the hydroxyl moiety of tyrosine provides a new contact point for prostacyclin and its analogs, compensating for one that prostaglandins with a less constrained  side chain possess.

Other contact points are necessary for the discrimination of prostaglandin ligands. The iloprost potency at the IP receptor is approximately 100-fold greater than its potency at EP₂ L304Y. There must also be amino acids responsible for reducing the potency of PGE₂ at the IP and other non-EP receptors. The region of the binding pocket that enables the selective interactions of PGE₂, as well as PGD₂ and PGF₂, may be distinct from TM7 and is likely involved in interactions with the cyclopentane ring. A recent study with chimeric IP/DP receptor constructs suggested a role for TM3 of the DP receptor in cyclopentane ring interactions (Kobayashi et al., 1997). These authors also suggested that TM6 or TM7 is responsible for  chain recognition, which is consistent with our results. In another study, mutagenesis of a conserved hydroxy amino acid in TM6 of the EP₃ receptor seemed to alter the selectivity of that receptor (Negishi et al., 1995). The approach we used here can be applied to other receptor pairs (the EP₂ and DP receptors for instance) to identify significant determinants of cyclopentane ring interactions. A comparison of EP₂ and EP₄ receptors might also determine why PGE₂ is approximately 10-fold less potent at the EP₂ receptor than at other EP receptors.

Residue 120 in TM3 probably influences ligand activity by interacting with residue 304. The EP₂ S120G mutation alone has wild-type pharmacology, which indicates that this residue is not involved in defining a ligand interaction site (Fig. 2C). However, EP₂ L304Y exhibits a loss of signal magnitude that is rescued by the double mutant, EP₂ S120G L304Y, which indicates that the TM3-TM7 interactions are important for optimal receptor function (Fig. 4). The elevated basal activity of the single mutants may reflect reduced stability of the receptor in the inactive state, which is stabilized by G protein coupling or the double mutant. Further evidence that these two amino acids interact is the fact that the changes at these positions between EP₂ and IP receptors are complementary in side chain length and polarity.

Arg302, which is located in TM7 and corresponds with arginine residues that have been implicated in binding of the prostaglandin carboxylic acid moiety, is predicted on the basis of modeling to be situated at the opposite end of the binding pocket from TM3 and TM6. As expected for an amino acid that is conserved in all prostaglandin receptors, mutagenesis of this residue resulted in a nonselective reduction of agonist activity (Fig. 5). The retention of signal transduction, however minor, with the Arg302 mutants EP₂ R302Q and EP₂ R302E is intriguing. The result is consistent with PG ligands having many receptor contact points, as previously discussed. Mutagenesis of the corresponding residue in other prostaglandin receptors has resulted in a range of responses from total loss of binding or signaling to retention of some signaling. In some studies (Funk et al., 1993; Huang and Tai, 1995) binding alone was evaluated, which may underestimate any residual function present in the TP and EP₃ mutants. In one case, functional studies using the synthetic EP₃ analog sulprostone supplemented binding studies and demonstrated that mutation of the conserved arginine in the EP₃ receptor resulted in complete loss of function (Audoly and Breyer, 1997). However, other studies have demonstrated retention of some EP₃ function when PGE₂ but not sulprostone is the agonist (Negishi et al., 1995; Chang et al., 1997).

There are only a few previous examples of a single amino acid change altering the selectivity of a receptor for endogenous ligands. In the somatostatin receptor subtype 5, substitution of Phe265 in TM6 with tyrosine increased the affinity of the 14-amino-acid form of somatostatin, so that it was comparable with its affinity for somatostatin receptor subtypes 1-4 (Ozenberger and Hadcock, 1995). Mutation of Tyr129 in TM2 of endothelin receptor A to histidine, the corresponding amino acid in endothelin receptor B, enhanced endothelin-3 binding so that its affinity was similar to that of endothelin receptor B (Krystek et al., 1994; Lee et al., 1994). These receptors, as well as the EP₂, should support study of the evolution of ligand-receptor pairs. One would predict that the emergence of the particular amino acid changes that are described and novel endogenous ligands occurred at similar stages of evolution.

In summary, a single residue in TM7 of the EP₂ receptor that is changed in the IP receptor determines the activity of stable prostacyclin analogs. This result is consistent with a model of prostaglandins interacting with their receptors at a universal contact point (TM7 arginine) and at other residues, probably in a pocket formed by TM7, TM6, and TM3 that enables ligand discrimination. The significant role of a single amino acid in the selectivity of the EP₂ receptor is consistent with the hypothesis that the IP receptor evolved from the EP₂ receptor. In effect, the mutant receptor represents a molecular "missing link" in the evolution of the IP receptor from the EP₂ receptor.