The second extracellular loop of the prostaglandin EP3 receptor is an essential determinant of ligand selectivity.

The prostaglandin EP3 receptor binds Prostaglandin E2 in a ligand binding pocket formed in part by seven transmembrane alpha-helices. The present studies demonstrate that the second extracellular loop of the receptor is involved in prostanoid ligand recognition as well. Site-directed mutagenesis of seven conserved residues clustered in the amino portion of the second extracellular loop was performed. Receptors with single amino acid substitutions at each of these positions were transiently transfected into HEK293tsA201 cells, their ligand binding profiles assessed, and each receptor was tested for its ability to decrease intracellular cAMP levels. Substitution of Trp199 or Thr202 with alanine resulted in receptors with increases in affinity up to 128-fold for prostanoid compounds with a C1 methyl ester but wild type affinities for natural prostanoid ligands that have a carboxylate moiety at the C1 position. In contrast, substitution of Pro200 with serine caused a loss of selectivity up to 20-fold for naturally occurring prostanoid agonists as compared with the wild type EP3 receptor: the PS200 receptor displayed a decrease in affinity for E-ring compounds and an increase in affinity for F- and D-ring compounds. The EC50 for inhibition of cAMP remained unchanged for each receptor tested.

for ligand recognition (for review, see Ref. 3). Evidence supports the role of residues embedded in the transmembrane as important in receptor-ligand interactions (4). In contrast, the extracellular regions appear to be relatively unimportant for binding of these small ligands, with the exception of a role of an extracellular cysteine disulfide bridge demonstrated in several receptors including the ␤-adrenergic receptor (5) and the thromboxane A 2 receptor (6) as well as a portion of the second extracellular loop of the adenosine A 1 and A 2a receptors (7,8).
Less is known regarding the structural determinants of EP receptor-ligand interactions. Several groups have previously identified the importance of an arginine residue found in transmembrane region VII of the EP 3 receptor and conserved throughout prostanoid receptors (9 -11). Substitution of Arg 329 in transmembrane VII to either Ala or Glu led to a loss of detectable [ 3 H]PGE 2 binding and receptor-mediated inhibition of [cAMP] i (9).
Comparisons of the amino acid sequence between the rabbit EP 3 receptor and the other cloned prostanoid receptors have identified several regions of conservation (9). Fourteen conserved amino acid residues were identified outside the putative transmembrane regions, including six amino acid residues clustered in the amino-terminal portion of the second extracellular loop. We hypothesized that conserved extracellular regions of the EP 3 receptor affect receptor/ligand interactions either directly or indirectly, analogous to the proposed interactions between the extracellular regions and ligands of peptidebinding GPCRs such as neurokinin-1 (13), thyrotropin (14), or [Arg 8 ]vasopressin receptors (15). To test whether this conserved primary structure plays a role in receptor-ligand interaction, a series of point mutants were generated and assayed for their ability to bind a panel of natural and synthetic prostanoid analogs. Findings presented herein provide evidence that the second extracellular loop of the prostaglandin E 2 EP 3 receptor plays a role in ligand selectivity.

EXPERIMENTAL PROCEDURES
Materials-Misoprostol and misoprostol-free acid were gifts of Dr. Paul Collins (Searle). All other prostanoid analogs were purchased from Cayman Chemical (Ann Arbor, MI). Isoproterenol, ascorbic acid, and indomethacin were purchased from Sigma. [ 3 H]PGE 2 was purchased from DuPont NEN. LipofectAMINE, and Opti-MEM were purchased from Life Technologies, Inc.
Site-directed Mutagenesis of the Receptor-Missense mutations were introduced using the polymerase chain reaction (PCR) as described previously (9). The sequence of the flanking oligonucleotides was as follows: upstream oligonucleotide (EP 3 nucleotide 501 coding), 5Ј TG GTG TAC CTA TCC AGG 3Ј; downstream oligonucleotide (EP 3 nucleotide 963 coding), 5Ј CCA GGG ATC CAA TAT CTG G 3Ј.
The internal oligonucleotides used to introduce missense mutations are listed as follows (with underlining indicating nucleotide substitutions): QA198: 5Ј A CAG TAC ACG ATC GCG TGG CCC GG 3Ј; WA199,  5Ј TAC ACC ATC CAA GCT CCC GGG ACG 3Ј; PS200, 5Ј ATC CAG  TGG TCA GGT ACC TGG TGC TTC 3Ј; TA202, 5Ј TC CAG TGG CCT  GGT GCT TGG TGC TTC  PCR fragments encoding the target amino acid substitutions were then subcloned into the hemagglutinin-tagged 77A isoform of the EP 3 receptor in plasmid 77A hemagglutinin wt pRC/CMV, generating the full-length EP 3 77A receptor (9). Two independent clones bearing each amino acid substitution were then isolated and characterized in subsequent experiments. The identity of the mutations was confirmed by sequencing both strands of the PCR amplified region using a Thermo-Sequenase kit (Amersham Life Science, Inc.).
Expression of EP 3 cDNAs in Cell Culture-HEK293tsA201 cells were transiently transfected with plasmids bearing wt or mutant EP 3 cDNA as described (9). Cells were cultured for 72 h, and the medium was replaced every 24 h. At 72 h cells were lysed and membranes prepared as described (16). Protein concentrations were determined by the BCA assay (Pierce).
Ligand Binding Assays-For saturation binding isotherm experiments, 10 -40 g of membrane protein, representing 20 fmol of receptor, was incubated with various concentrations of [ 3 H]PGE 2 , and reactions were stopped by filtration onto glass fiber filters as described (9). For competition binding assays, 10 -40 g of membrane proteins were incubated with 1 nM [ 3 H]PGE 2 and varying concentrations of unlabeled competitor and assayed as described above.
cAMP Measurements-HEK293tsA201 cells were transiently cotransfected with plasmids containing the human ␤ 2 AR and either the wt or mutant EP 3 receptor. cAMP measurements were performed by radioimmunoassay as described (9).
Data Analysis-Saturation binding isotherms, competition binding isotherms, and dose-dependent responses of [cAMP] i were analyzed using Prism (GraphPad, San Diego, CA). K i values were calculated using the method of Cheng and Prusoff (17). Statistical analyses were performed using Instat (GraphPad).

Sequence Alignments of the Second Extracellular Loop
Region of Prostanoids-Sequence alignments of the cloned rabbit EP 3 receptor was performed against all cloned prostanoid receptors (9). A cluster of six conserved amino acid residues was identified in the putative second extracellular loop (Fig. 1). The sequence was identified as Q 198 WPGTWCF, where bold characters represent conserved amino acids. Pro 200 is conserved throughout all cloned prostanoid receptors with the exception of the FP receptor. We tested whether this conserved portion of the receptor plays a role in the prostaglandin EP 3 receptor function and/or structure.
Mutations of Trp 199 and Thr 202 Cause an Increase in the Affinity of Methyl Ester Compounds-TA202 displayed markedly increased affinities for methyl ester compounds of the E series as compared with the wild type receptor despite displaying similar dissociation constants for Fig. 2A, TA202 resulted in a 128-fold increase in affinity for misoprostol as compared with the wild type receptor (K i wt ϭ 1600 Ϯ 350 nM, K i TA202 ϭ 13 Ϯ 3 nM). In contrast, the affinity of TA202 for the carboxylate derivative misoprostol-free acid increased a modest 2-fold (K i wt ) ϭ 6.5 Ϯ 1.9 nM, K i TA202 ϭ 3.3 Ϯ 0.6 nM). The K i values for other natural and synthetic prostanoid agonists with a carboxylate at the C1 position were not statistically different from wild type (Table I). Sulprostone, which has a sulfonamide moiety at C1, had a modest 3-fold increase in affinity. To test if the increased affinity for misoprostol could be extended to other E series methyl ester compounds, TA202 was assayed with a panel of paired methyl ester/carboxylate prostanoid analogs that differed in their substituents at positions which differ between PGE 2 and misoprostol: substitution at the C15 and C16 position and presence or absence of a double bond between C5 and C6. For each compound tested TA202 displayed in-FIG. 1. Identification of conserved residues in the EP 3 receptor. Sequence alignments were carried as described previously (9). Briefly, the predicted amino acid sequence of the rabbit EP 3 receptor was used as the template and aligned against all cloned prostanoid receptors. Conserved residues are indicated by hatched circles and invariant residues by filled circles. Residues with an inset symbol Ⅺ are conserved across the entire superfamily of GPCRs, those without the inset are unique to prostanoid receptors. In the bottom panel, the region of interest is indicated by the one-letter amino acid code of target amino acids. The resulting substitution followed by the position of the residue is also indicated.  Table II in parentheses. f, wt; Ⅺ, TA202.

Molecular Determinants of EP 3 Receptor Ligand Selectivity 13476
creased affinity for the methyl ester compound from 38-to 128-fold as compared with wild type; however, the affinities for the carboxylate analog of each pair was not statistically different from wt (Table II). Fig. 2B summarizes the selectivity ratios for the various drugs tested.
The identity of Trp 199 is not highly conserved among prostanoid receptors as tyrosine, phenylalanine, or alanine may be found at this position. Nonetheless, substitution of Trp 199 with alanine resulted in a similar phenotype to the TA202 receptor. WA199 displayed a 9-fold increase in affinity for misoprostol and a 3-fold increase in affinity for sulprostone while the K i values for other compounds tested remained unaffected ( Table I).
Mutation of Pro 200 Affects Ligand Selectivity-Proline 200 is conserved among the EP 1 , EP 2 , EP 3 , EP 4 , DP, IP, and TP receptors across species. In the FP receptor, this position is occupied by a serine residue. Pro 200 of the EP 3 receptor was mutated to Ser, and its binding characteristics were determined. The K d PS200 ϭ 4.1 Ϯ 0.5 nM versus K d wt ϭ 1.4 Ϯ 0.3 nM suggested a modest loss of [ 3 H]PGE 2 affinity for PS200 (twotailed p value Ͻ 0.01). Competition binding isotherms resulted in a loss of specificity for PS200 as determined using PGE 2 , PGD 2 , and PGF 2␣ . The PS200 receptor displayed a 4-fold loss in affinity for PGE 2 ; however, it gained 3-and 5-fold in affinity for PGF 2␣ and PGD 2 , respectively (Table I, Fig. 3). Similarly, results obtained with the synthetic PGE analogs sulprostone and misoprostol-free acid demonstrated a 5-7-fold decrease in affinity, while misoprostol displayed a 5-fold increase in affinity. Thus, the overall pattern for the PS200 substitution was a loss in affinity for compounds that bind with high affinity to wt and an increase for compounds that bind with lower affinity to wt receptor. These results suggest Pro 200 plays a crucial role maintaining ligand binding selectivity.
Substitution of Gln 198 , Trp 203 , Cys 204 , and Phe 205 Does Not Affect Ligand Selectivity-Substitution with alanine at the positions Gln 198 , Trp 203 , Cys 204 , and Phe 205 did not affect the binding profile of these receptors despite absolute conservation of the primary structure at these positions among all cloned prostanoid receptors. Membranes prepared from cells transfected with each receptor were tested in saturation binding (data not shown) and competition binding isotherms with the panel of natural and synthetic prostanoid analogs, and no statistically significant changes in affinity or order of agonist potency were observed (Table I). These results argue against the role of each individual residues in receptor-ligand interactions.

Summary of competition binding isotherms of wild type and TA202
receptors to methyl ester compounds K i values were averaged from three to five independent experiments Ϯ S.E. The number of independent experiments is indicated in parentheses next to the K i value. The two-tailed p values were determined by comparing the average values of K i wt versus K i mutant .  Table I.

Molecular Determinants of EP 3 Receptor Ligand Selectivity 13477
transduction properties of each receptor variant were determined. As compared with the wild type receptor (EC 50 ϭ 480 pM), the various receptor EC 50 values were: QA198 ϭ 550 pM, WA199 ϭ 360 pM, PS200 ϭ 500 pM, TA202 ϭ 370 pM, WA203 ϭ 400 pM, CA204 ϭ 370 pM, FA205 ϭ 200 pM) and did not display any statistically significant differences in receptor evoked signaling in three independent experiments.

DISCUSSION
Using site-directed mutagenesis studies of the prostaglandin EP 3 receptor, the present results demonstrate that the second extracellular loop plays an integral role in receptor-ligand interaction and particularly in ligand selectivity. The gain of agonist affinity observed for several mutant receptor phenotypes argues strongly against gross perturbation of receptor structure by these amino acid substitutions as the cause of the altered ligand binding phenotypes. The precise role of the second extracellular loop in receptor-ligand interaction is unknown. One possible explanation for the observed phenotype is that the second extracellular loop forms part of the binding pocket and is in direct contact with the bound ligand. Alternatively, the effects of these mutations may be the result of an indirect role of the second extracellular loop. This is analogous to the Ca 2ϩ , Na ϩ , or K ϩ channels where it has been proposed that extracellular loops fold into the ion channel pore and interact with the transmembrane helices (18). A third possibility is that the loop is important for the overall conformation of the receptor without direct interaction with the transmembrane helices. This interpretation may be consistent with the idea that modification of Trp 199 , Pro 200 , or Thr 202 induces a general relaxation of the receptor conformation, preventing it from discriminating structurally related prostanoid analogs.
Several models supporting a direct role of extracellular domains of peptide/amino acid binding GPCRs have been described in the literature. In the case of the receptor for parathyroid hormone, it was suggested that amino acid residues near the extracellular surface of the transmembrane helices play a "filter" role allowing discrimination between various ligands (19). This concept could be extended to the EP 3 receptor, where it is conceivable to consider Thr 202 or Trp 199 as filters which reduce affinity for compounds with a C1 methyl ester. Elimination of Thr 202 or Trp 199 side chains results in the elimination of a "gate" preventing methyl ester prostanoids from accessing the binding cleft. This model argues in favor of a two-step process in terms of receptor-ligand interaction. It has been suggested for the metabotropic glutamate receptor, mGLUR1, that the extracellular region of the receptor may act as a primary point of receptor-ligand interaction or "bait" and subsequently facilitate presentation to the transmembrane ligand pocket (20). Similarly, this model may be applied to the second extracellular loop of the EP 3 receptor, whereby Trp 199 , Pro 200 , and/or Thr 202 attract ligands and present them to the binding pocket. The filter and bait models are not mutually exclusive and may be complementary to one another.
It is interesting to note that mutation of Cys 204 , which is conserved among all cloned prostanoid receptors, caused no detectable change in ligand binding affinity or receptor activation. These results are in sharp contrast to the results presented for the thromboxane A 2 receptor, where replacement of Cys 183 to Ser (analogous position as Cys 204 ) led a complete loss of binding and signaling (6). These authors suggested that a critical disulfide bond existed between Cys 183 and Cys 105 (transmembrane domain III) of the thromboxane A 2 receptor; however, results presented here argue against the importance of a putative disulfide bridge involving Cys 204 in the EP 3 receptor.
It is also of interest that within this highly conserved aminoterminal portion of the second extracellular loop, substitutions at the other absolutely conserved positions Gln 198 , Trp 203 , Cys 204 , and Phe 205 did not detectably affect ligand binding or signal transduction. It may be that substitutions at several positions are required to disrupt receptor function. Alternatively, an untested function (e.g. internalization/recycling) may have been altered.
We propose a revised version of the model currently described for the EP 3 receptor (21). We had previously shown that the EP 3 and EP 4 receptors displayed increased affinities for carboxylate compounds versus their methyl ester derivatives (9,12). Furthermore, a large body of literature has suggested that the carboxylate moiety of C1 interacts with the positive side chain of arginine in transmembrane VII. The above results with TA202 suggest that a negative charge on C1 of prostanoids is not required for high affinity interactions with the EP 3 receptor as shown for the TA202 receptor. Based on the high degree of homology of the second extracellular loop among cloned prostanoid receptors, it is conceivable these findings may be generalized to some or all of the prostanoid receptors. This may suggest that a revised three-dimensional model incorporating the extracellular loop regions is required to interpret receptor-ligand interactions.