Molecular basis for selectivity of high affinity peptide antagonists for the gastrin-releasing peptide receptor.

Few gastrointestinal hormones/neurotransmitters have high affinity peptide receptor antagonists, and little is known about the molecular basis of their selectivity or affinity. The receptor mediating the action of the mammalian bombesin (Bn) peptide, gastrin-releasing peptide receptor (GRPR), is an exception, because numerous classes of peptide antagonists are described. To investigate the molecular basis for their high affinity for the GRPR, two classes of peptide antagonists, a statine analogue, JMV594 ([d-Phe(6),Stat(13)]Bn(6-14)), and a pseudopeptide analogue, JMV641 (d-Phe-Gln-Trp-Ala-Val-Gly-His-Leupsi(CHOH-CH(2))-(CH(2))(2)-CH(3)), were studied. Each had high affinity for the GRPR and >3,000-fold selectivity for GRPR over the closely related neuromedin B receptor (NMBR). To investigate the basis for this, we used a chimeric receptor approach to make both GRPR loss of affinity and NMBR gain of affinity chimeras and a site-directed mutagenesis approach. Chimeric or mutated receptors were transiently expressed in Balb/c 3T3. Only substitution of the fourth extracellular (EC) domain of the GRPR by the comparable NMBR domain markedly decreased the affinity for both antagonists. Substituting the fourth EC domain of NMBR into the GRPR resulted in a 300-fold gain in affinity for JMV594 and an 11-fold gain for JMV641. Each of the 11 amino acid differences between the GRPR and NMBR in this domain were exchanged. The substitutions of Thr(297) in GRPR by Pro from the comparable position in NMBR, Phe(302) by Met, and Ser(305) by Thr decreased the affinity of each antagonist. Simultaneous replacement of Thr(297), Phe(302), and Ser(305) in GRPR by the three comparable NMBR amino acids caused a 500-fold decrease in affinity for both antagonists. Replacing the comparable three amino acids in NMBR by those from GRPR caused a gain in affinity for each antagonist. Receptor modeling showed that each of these three amino acids faced inward and was within 5 A of the putative binding pocket. These results demonstrate that differences in the fourth EC domain of the mammalian Bn receptors are responsible for the selectivity of these two peptide antagonists. They demonstrate that Thr(297), Phe(302), and Ser(305) of the fourth EC domain of GRPR are the critical residues for determining GRPR selectivity and suggest that both receptor-ligand cation-pi interactions and hydrogen bonding are important for their high affinity interaction.

The gastrin-releasing peptide (GRP) 1 receptor, which mediates the diverse actions of the mammalian bombesin (Bn)related peptide (1,2), GRP, has numerous high affinity peptide antagonists (3)(4)(5). This is in contrast to most other gastrointestinal (GI) hormone/neurotransmitter receptors for which no high affinity peptide antagonists exist (6). These GRP receptor antagonists are now widely used in both in vitro studies (5) and in vivo studies in animals (3,(7)(8)(9)(10)(11) and humans (12). Recent studies show that for many nonpeptide antagonists, differences in amino acids in the transmembrane domains between receptor subtypes are frequently particularly important for determining receptor subtype selectivity (13,14). A recent study (15) shows a similar result with the peptoid antagonist PD168368 for the neuromedin B receptor. However, with peptide antagonists of non-GI hormone/neurotransmitter receptors, interactions with transmembrane regions (16,17) or extracellular domains (18) are important for high affinity interaction or receptor subtype selectivity. Which if any of these results apply to the different classes of GRPR peptide antagonists, at present, is unclear.
GRP and neuromedin B (NMB), mammalian homologues of the amphibian tetradecapeptide bombesin, have structurally related carboxyl termini (19). These peptides mediate a spectrum of biological activities such as stimulating growth of both normal and neoplastic tissues (20 -23), secretion (1), muscle contraction (24), central nervous system effects (including satiety (25), thermoregulation (26), and circadian rhythm (27)), changed developmental (28,29) and immunologic effects (30). These effects are mediated by binding to two structurally and pharmacologically distinct receptors, the GRPR and NMB receptor (NMBR) (31)(32)(33). These two receptors are members of the bombesin receptor family within the G protein-coupled receptor (GPCR) superfamily and share 56% overall amino acid sequence identity (34). Both the GRPR and NMBR are widely distributed in the central nervous system and peripheral tissues including in the GI tract (1,35). Which of the widespread effects of these peptides (1) are important physiologically or in pathologic processes is largely unknown at present. The availability of potent Bn receptor subtype-specific antagonists and a molecular understanding of their basis of action are important steps in addressing these questions.
In the present study, we have examined the molecular basis for the GRPR selectivity of two different classes of closely related high affinity Bn receptor peptide antagonists: a Bn statine analogue, JMV594 (36), and a Bn pseudopeptide analogue, JMV641 (37). A receptor chimeric and site-directed mu-tagenesis approach was used to identify critical domain and amino acid(s) responsible for these antagonists' selectivity and high affinity for GRPR. In this study we show that the selectivity of peptide antagonists JMV594 and JMV641 for the GRPR over the NMBR depends primarily on interactions with amino acids in the fourth extracellular region of the GRPR. Site-directed mutagenesis studies demonstrate that Thr 297 , Phe 302 , and Ser 305 in this region of GRPR are the critical amino acids for high affinity binding and selectivity of these two antagonists. Computer modeling of this region of the GRPR demonstrates that these amino acids all face inward forward the proposed binding pocket and are all within 5 Å of it, suggesting that cation-and hydrogen bonding interactions between these antagonists and the above three amino acids are essential for receptor subtype selectivity and high affinity interaction.
Construction of Chimeric and Mutant Receptors-The cDNAs of the mouse GRPR and rat NMBR were identical to those described previously (31,32). The cDNA of the wild-type mouse GRPR was cloned between the HindIII site and XbaI site of pcDNA3, and the wild-type rat NMBR was cloned into the EcoRI site of pcDNA3. The GRPR/NMBR chimeras were constructed using the Seamless™ Cloning Kit (38) using hydropathy plots for the GRPR and for the NMBR as described previously (15). Mutant receptors were made by using the QuikChange™ Site-Directed Mutagenesis Kit, following the manufacturer's instructions, except that the annealing temperature was 60°C and the DpnI digestion was for 2 h. Nucleotide sequence analysis of the entire coding region was performed using an automated DNA sequencer (ABI PRISM™ 377 DNA sequencer; Applied Biosystems Inc., Foster City, CA).
Cell Transfection-Balb/c 3T3 cells were seeded in a 10-cm diameter tissue culture dish at a density of 10 6 cells/dish and grown overnight at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin. The following morning, cells were transfected with 5 g of plasmid DNA by the cationic lipid-mediated method (39) using 30 l of LipofectAMINE™ reagent and 20 l of LipofectAMINE™ Plus reagent in serum-free Dulbecco's modified Eagle's medium for 3 h at 37°C. At the end of the incubation period, the medium was replaced with Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin, and 100 mg/ml streptomycin. Cells were maintained at 37°C with a 5% CO 2 atmosphere and were used 48 h later for binding assays.
Preparation of 125 I- [Tyr 4 ]Bn-125 I-[Tyr 4 ]Bn at a specific activity of 2,200 Ci/mmol was prepared by a modification of the methods described previously (40,41). Briefly, 0.8 g of IODO-GEN in chloroform was transferred to a vial, dried under a stream of nitrogen, and washed with 100 l of KH 2 PO 4 (pH 7.4). To this vial, 20 l of KH 2 PO 4 (pH 7.4), 8 g of peptide in 4 l of water, and 2 mCi (20 l) of Na 125 I were added, mixed gently, and incubated at room temperature for 6 min. The incubation was stopped by the addition of 100 l of distilled water and 300 l of 1.5 M dithiothreitol. The iodination mixture was incubated at 80°C for 60 min. The reaction mixture was applied to a Sep-Pak column (Waters Associates, Milford, MA), and free 125 I was eluted with 5 ml of water followed by 5 ml of 0.1% (v/v) trifluoroacetic acid. The radiolabeled peptides were eluted with 200 l of sequential elutions (ϫ 10) with 60% acetonitrile in 0.1% trifluoroacetic acid. The two or three fractions with the highest radioactivity were combined and purified on a reverse-phase, high performance liquid chromatography with a BondaPak column (0.46 ϫ 25 cm). The column was eluted with a linear gradient of acetonitrile in 0.1% trifluoroacetic acid (v/v) from 16 to 60% acetonitrile in 60 min. One-ml fractions were collected and checked for radioactivity and receptor binding. The pH values of the fractions were adjusted to 7 using 0.2 M Tris (pH 9.5), and radioligands were stored in aliquots with 0.5% bovine serum albumin at Ϫ20°C.
Whole Cell Radioligand Binding Assays-Competitive binding assays were performed 48 h post-transfection. Disaggregated transiently transfected cells were incubated for 1 h at room temperature in 250 l of binding buffer (pH 7.4) with the ligand 50 pM 125 I-[Tyr 4 ]Bn (2,200 Ci/mmol) in the presence of the indicated concentration of unlabeled peptides. The binding buffer contained 98 mM NaCl, 6 mM KCl, 11.5 mM glucose, 5 mM fumarate, 5 mM glutamate, 5 mM pyruvate, 24.5 mM HEPES, 0.2% (v/v) essential amino acid solution, 2.5 mM KH 2 PO 4 , 1 mM MgCl 2 , 0.5 mM CaCl 2 , 0.2% (w/v) bovine serum albumin, 0.05% (w/v) bacitracin, and 0.01% (w/v) soybean trypsin inhibitor. The cell concentration was adjusted to 0.2-2.6 ϫ 10 6 cells/ml to assure that no more than 20% of the total added radioactive ligand bound. Bound tracer was then separated from unbound tracer by layering 100 l of the binding reaction on top of an oil phase (100 l of Nyosil M20; Nye Lubricants Inc., New Bedford, MA) in a 0.4-ml microcentrifuge tube (PGC Scientific, Frederick, MD) and pelleting the cells through the oil by centrifugation at 10,000 ϫ g in a Microfuge E TM (Beckman Instruments) for 3 min. The supernatant was aspirated, and the pelleted cells were rinsed twice with distilled water. The amount of radioactivity bound to the cells was measured in a Cobra II ␥ counter (Packard Instrument Co.). A 100-l aliquot of the incubation mixture were taken in duplicate to determine the total radioactivity. Binding was expressed as the percentage of total radioactivity that was associated with the cell pellet. All binding values represented saturable binding (i.e. total binding minus nonsaturable binding). Nonsaturable binding was Ͻ15% of the total binding in all experiments. Each point was measured in duplicate, and each experiment was replicated at least three times. Calculation of IC 50 values was performed with a curve-fitting program, KaleidaGraph graphing software (Synergy Software, Reading, PA). Affinity and receptor density were calculated using a least-squares curvefitting program (LIGAND) (42).
Homology Modeling of Extracellular Loop Region-The amino acid sequences of mouse GRPR and rat NMBR were retrieved from the G protein-coupled receptor data base (43). The three-dimensional crystal structure of bovine rhodopsin (1f88) was obtained from the Protein Data Bank (Research Collaboratory for Structural Bioinformatics, Rutgers University) (44). Examination of the hydropathy profile of the fourth extracellular domain of bovine rhodopsin using SYBYL6.6 demonstrated that the hydrophobicity of the sequence is markedly less in the region of the extracellular loop as compared with the adjoining helical regions. This was confirmed by protein sequence analysis (45)(46)(47). Hydropathy evaluation and protein sequence analysis were used to locate the fourth extracellular domain regions in both mouse GRPR and rat NMBR. An additional eight amino acids were included at each end of the putative regions, and the sequences were imported into Deep View, the Swiss-PdbViewer (48) and aligned to the structure of bovine rhodopsin. The threading energy (49) of the alignment was at a minimum and rose if the sequences were shifted in either direction. The orientation of the side chains was optimized to reduce clashes, and the three-dimensional structure of the extracellular loop region was exported to SYBYL as a pdb file.
Binding Site Model-The putative, solvent-accessible binding sites of the extracellular loop of the receptor were explored using the SiteID module in SYBYL6.6. Briefly, the loop region was solvated with water molecules, and clusters of solvent molecules adjacent to solvent-accessible residues were defined as a potential binding site. This demonstrated a possible binding site near the extracellular surface bounded by the extracellular loop and potentially extending into the interior of the helical bundle with a volume of 430 Å 3 . The surface of the solvent cluster adjacent to the extracellular loop was visualized using the SYBYL program Molcad.

RESULTS
Wild-type GRPR and NMBR-The Bn-related natural occurring agonists Bn and GRP ( Fig. 1) had high affinity (IC 50 2.7 nM) for the GRPR (Table I), and neuromedin B (Fig. 1) had high affinity for the NMBR (IC 50 1.2 nM) ( Table I). Bn had a 2-fold and GRP a 12-fold selectivity for the GRPR over the NMBR, whereas neuromedin B had 200-fold selectivity for the NMBR over the GRPR (Table I) Table I). However, we found that each antagonist had a low affinity for the NMBR (IC 50 1,500 -10,000 nM) (Fig. 2, Table I). Therefore, JMV594 and JMV641 had Ͼ5,000 and 3,260 times higher selectivity, respectively, for the GRPR over the NMBR (Fig. 2, Table I). To explore the molecular basis for this GRPR selectivity of JMV594 and JMV641, we first made both loss of affinity GRP chimeric receptors (Fig. 3) and gain of affinity NMBR chimeric receptors (Fig. 4). Then to determine the exact amino acids involved we made GRPR loss of affinity point mutations (Figs. [5][6][7][8] and gain of affinity point mutants of NMBR (Figs. 9 and 10).
Extracellular Chimeric Receptors-Four loss of affinity GRPR chimeric receptors were made with the extracellular domains of NMBR substituted for the comparable domains in GRPR (Fig. 3), and four potential gain of affinity NMBR chimeras were made with the extracellular domains of GRPR substituted into NMBR (Fig. 4). Substitution of the first, second, and third extracellular domain in the GRPR by the comparable domain from the NMBR did not alter the affinity for JMV594 or JMV641, and each chimeric GRPR had similar affinities to wild-type GRPR ( Fig. 3 and Table II). Substitution of the fourth extracellular domain in the GRPR by the comparable domain of the NMBR decreased the affinity 1,400-fold (from 2.2 Ϯ 0.05 to 3,100 Ϯ 120 nM) for the statine antagonist, JMV594, and 1,200-fold (from 0.46 Ϯ 0.03 to 530 Ϯ 22 nM) for the pseudopeptide antagonist, JMV641 (Fig. 3, Table II). When the reverse study was performed by substituting in the NMBR the extracellular domain of the GRPR to attempt to gain affinity, replacement of the fourth extracellular domain increased the affinity for JMV594 by 300-fold (from Ͼ10,000 to 29 Ϯ 0.58 nM) and with JMV641 caused an 11-fold gain in affinity (from 1,500 Ϯ 40 to 140 Ϯ 4.2 nM) (Fig. 4, Table II). Replacement of the first, second, or third extracellular domain of the NMBR by the comparable domain of the GRPR had no effect on the affinity of JMV594. However, substitution of the third extracellular domain had a small effect on the affinity for JMV641, increasing affinity less than 1-fold ( Fig. 4, Table II). These results from the study of chimeric receptors demonstrated that the fourth extracellular domain was principally involved in determining selectivity for GRPR over NMBR for these two structurally different peptide antagonists.
GRPR Fourth Extracellular Domain Mutants (Loss of Affinity Point Mutants)-To identify which amino acid(s) in the fourth extracellular domain of GRPR are responsible for the high affinity for JMV594 and JMV641, the amino acid differences and identities were compared between the mouse GRPR and rat NMBR in the fourth extracellular domain (Fig. 5). The rat, mouse, and human NMBR are identical in the fourth extracellular domain (33,50). The human and mouse GRPR are identical also in this region and differ from the rat by one conservative substitution of isoleucine for valine in position 303 (33). In the fourth extracellular domain, 11 amino acid differences were present, occurring at positions 290 -291, 293, 295, 297, 299 -300, and 302-305 of GRPR, which are comparable with positions 291-292, 294, 296, 298, 300 -301, and 303-306 of NMBR (Fig. 5). To study the 11 amino acid differences in the fourth extracellular domains of these receptors, we first made six GRPR loss of affinity group point mutants (Fig. 5). Two of the six GRPR group point mutants caused a decrease in affinity for both antagonists (Fig. 6). The GRPR mutant with Thr 297 of the GRPR replaced by Pro 297 from the comparable position of NMBR (Thr 297 3 Pro 297 ) and a second mutant with Phe-Val-Thr-Ser 302-305 of the GRPR replaced by Met-Ile-Val-Thr 302-305 , decreased the affinities for JMV594 by 17-and 90-fold, respectively, and for JMV641 by 2-and 48-fold, respectively ( Fig. 6 and Table III). The other four GRPR group point mutants, including replacement of Tyr 290 -His 291 of the GRPR by Phe-Asn 290,291 , Ser 293 of the GRPR by Lys 293 , Val 295 of the GRPR by Ile 295 , and Met-Leu 299,300 of the GRPR by Leu-Gly from NMBR had no effect on the affinity for either antagonist (Fig. 6, Table  III). However, neither of the two group changes alone (i.e. 2-100-fold decrease) caused a decrease in affinity equal to the Ͼ1,100-fold decrease in affinity seen for each antagonist when the entire fourth extracellular GRPR domain was replaced by that from the NMBR (Table III). Therefore, a series of point mutations and combinations were made to identify which other amino acids were important for determining the antagonists' selectivity.
To understand which amino acids were important in   (Fig. 9). These three amino acids (i.e. Pro 298 , Met 303 , and Thr 306 ) in NMBR are in the position comparable with that of the amino acids in GRPR that were the key amino acids for high selectivity of GRPR for the antagonists identified in the loss of affinity studies (Fig. 9). For the statine antagonist JMV594, two single amino acid mutants (Pro 298 3 Thr 298 and Thr 306 3 Ser 306 ) caused ϳ3-fold gain in affinity, and the combination mutant (Pro 298 , Met 303 , Thr 306 3 Thr 298 , Phe 303 , Ser 306 ) demonstrated a Ͼ130-fold gain in affinity (Fig. 10, Table III). This increase in affinity was only 2.5-fold less than the gain of affinity seen when the entire fourth extracellular domain of the GRPR was substituted into the NMBR (Fig. 10 and Table III). With the pseudopeptide JMV641, the substitutions of these three amino acids singularly or together into NMBR showed a different result. The substitution of Thr 298 for Pro 298 alone was the most critical amino acid for the gain in affinity for JMV641, causing a 5-fold increase in affinity ( Fig.  10 and Table III). In contrast to the statine analogue JMV594, with the pseudopeptide JMV641 the replacement of Thr 306 in the NMBR by Ser 306 from the comparable position in the GRPR had no effect. Furthermore, the combination replacement of all three of these key fourth extracellular domain amino acids of  the NMBR (Pro 298 , Met 303 , Phe 306 ) resulted in no greater gain in affinity than the effect of replacing the Pro 298 in NMBR alone ( Fig. 10 and Table III), whereas in the case of the statine antagonist JMV594 it caused a 40-fold gain in affinity over any single replacement ( Fig. 10 and Table III).
Binding Site Model-To attempt to gain additional insight into why certain amino acids within the fourth extracellular domain of GRPR were important for determining selectivity for these two antagonists, the proposed binding sites of the antagonists to the GRPR in the fourth extracellular domain of GRPR were analyzed using modeling programs (Fig. 11). The modeling results based on the crystal structure of bovine rhodopsin (44) revealed that Arg 288 , Ser 289 , Tyr 290 , Tyr 292 , Ser 293 , Glu 294 , Val 295 , Asp 296 , Thr 297 , His 301 , Phe 302 , and Ser 305 in the fourth extracellular domain of GRPR all contained atoms that were within 5 Å of the putative binding site (Fig. 11). Six of these 12 amino acids in the GRPR (i.e. Tyr 290 , Ser 293 , Val 295 , Thr 297 , Phe 302 , and Ser 305 ) are different from the comparable amino acids in the NMBR (Fig. 5). The three key amino acids (i.e. Thr 297 , Phe 302 , and Ser 305 ) that were found to be the primary determinants for the specificity of these two antagonists for the GRPR all projected into the interior of the putative binding pocket, with most atoms within a 5-Å distance of the projected binding pocket (Fig. 11). DISCUSSION In general, the roles of GPCRs mediating the action of most GI hormones/neurotransmitters in physiological or in pathological processes are still unclear. This is in large part because, for many, specific receptor antagonists do not exist, and for others, only recently have high affinity antagonists been developed (51). The antagonists for these receptors generally fall into one of three types: nonpeptide antagonists (the largest group); peptide antagonists; or, in a few cases, peptoid antagonists, which have features of both peptides and nonpeptides (52,53). There are a number of studies of the molecular basis of action of nonpeptide antagonists for various GI hormone receptors (14, 54 -57); however, there are only a few studies for peptide an-  tagonists (14,16,17,55). This has occurred in large part, because potent peptide antagonists have been described for only a few GI hormone/neurotransmitter receptors (51). One of these exceptions is GRPR, which mediates the actions of the mammalian bombesin-related peptide, GRP, in the central nervous system and peripheral tissues (1)(2)(3)58). Six different classes of peptide receptor antagonists are described for the GRPR (3, 4, 36, 37, 59 -62), some with sufficient potency and stability to be used recently for in vivo studies to determine GRP's action in humans (12). At present, for each of these peptide GRPR antagonists, their molecular basis of action (the molecular determinants of their receptor selectivity or for their high affinity receptor interaction) is unknown. In this study, we examined the molecular basis of action of two of the most potent GRPR peptide antagonists, the statine analogue, JMV594, and the pseudopeptide analogue, JMV641 (36,37).
In the present study, we not only found that each of these peptide antagonists had an affinity as high as GRP or bombesin for the GRPR as reported previously (36,37) but also found that they were highly selective for the GRPR over the other mammalian Bn receptor, the NMBR. Each of the antagonists had Ͼ3,000-fold higher affinity for GRPR than NMBR, despite the fact that these receptors share an ϳ50% overall amino acid identity (2,33). The analyses of both the loss of affinity GRPR chimera and gain of affinity NMBR chimera support the con-clusion that differences in the fourth extracellular domains of these two receptors play the major role in determining selectivity of these two different classes of peptide antagonists.
This result has both similarities with and differences from studies on the interaction of peptide antagonists and agonists with other GPCRs. Studies of several GPCRs demonstrate that the receptor extracellular domains can be an important receptor region for determining high affinity ligand binding (63). However, only a few studies have explored whether the determinants of high affinity interaction or selectivity for a peptide antagonist are due to interactions with the receptor extracellular domains. Such an interaction is not important for determining high affinity interaction of the peptide antagonist D-Arg-[Hyp 3 ,D-Phe 7 ]bradykinin (NPC567) with the B 2 bradykinin receptor (16) or BQ-123 with the endothelin A receptor (14). However, it is important for the high affinity interaction of the peptide antagonists JMV179 with the human CCK-A receptor (55) and [Sar 1 ,Ile 8 ]angiotensin II with the AT 1 receptor (18). For a number of peptide GI receptor neurotransmitter/hormone agonists, high affinity receptor binding or receptor subtype selectivity depends on interaction with receptor extracellular domains, including CCK-8 with the CCK-A or CCK-B receptor (64), bradykinin with B 2 bradykinin receptor (65), and substance P with the neurokinin-1 receptor (66). However, for other peptide agonists such as the high affinity interaction of  NMB with the NMBR (67), endothelin (ET) with the ET A receptor (57), and neuropeptide Y with neuropeptide Y Y 1 receptor (68), the high affinity interaction or selectivity is primarily determined by amino acids in the transmembrane regions. Similarly, with many nonpeptide antagonists such as the interaction of L365,260 with the CCK-B receptor (54), losartan with the AT 1B angiotensin II receptor (69), bosentan with the ET A receptor (57), and L161,664 with the neurokinin-1 receptor (70), the extracellular domains do not contain determinants for high affinity interactions or receptor subtype selectivity.
To determine which amino acids in the fourth extracellular domain of the GRPR account for the high affinity of these two antagonists for the GRPR and its selectivity for this receptor over the NMBR, we performed a comparative alignment of the amino acids in this region between the GRPR and the NMBR. Our results support the conclusion that the threonine residue in position 297 of GRPR instead of a proline in the NMBR, a phenylalanine in position 302 of GRPR instead of a methionine in NMBR, and a serine in position 305 of GRPR rather than a threonine in NMBR are the key amino acid differences responsible for the high affinity of the two peptides for the GRPR and their high selectivity for the GRPR over the NMBR. Whereas the individual replacement of these three amino acids caused only a 10-fold change in affinity for each antagonist, a combined substitution of these three residues by the comparable NMBR amino acids led to potentiated effect with a marked decrease in affinity (ϳ500-fold) for each of the antagonists. This result demonstrates that amino acid differences at the both the amino and carboxyl terminus of the fourth extracellular domain are involved in determining the high affinity and selectivity for the GRPR of both antagonists. This result has both similarities and differences from studies of binding of other peptide ligands to GI hormone/transmitter GPCRs. It is similar to results with interaction of the peptide antagonist [Sar 1 ,Ile 8 ]angiotensin II (18) with the AT 1 receptor, in which a cooperative interaction between His 24 , Tyr 26 , and Ile 27 in the first extracellular domain is required for high affinity ligand interaction or for the peptide antagonist D-Arg-[Hyp 3 ,D-Phe 7 ]bradykinin's interaction with the B 2 bradykinin receptor, in which the ligand's interaction with multiple amino acids in the sixth transmembrane domain is required for high affinity interaction and selectivity. Similarly, with almost all studies on peptide agonists, such as bradykinin's interaction with the bradykinin 1 or 2 receptor (65); CCK-8 interaction with the CCK-B receptor (64); and substance P, neurokinin A, or neurokinin B interaction with the neurokinin-1 receptor (66), cooperative interactions with a number of residues in the same extracellular domain are required for high affinity interaction and selectivity for the one receptor subtype. In contrast to these cooperative interactions, with some peptide antagonists such as the interaction of the pentapeptide BQ123 with the ET A receptor (14) or JMV179 with the CCK-A receptor (55) and with many nonpeptide antagonists such as L365,260 interaction with the CCK-B receptor (54) The solvent-accessible amino acids with any atoms within 5 Å of a possible binding site are colored yellow, with the atoms of these amino acids within 5 Å colored red. The blue residues reside farther from the solvent-accessible site, which is illustrated as a transparent shape, defined as an envelope enclosing a cluster of water molecules surrounding the solvent-accessible residues. Val 303 and Thr 304 are at the rear (lipid-facing side) of the helical region and are therefore not accessible in this model. The critical amino acids Thr 297 , Phe 302 , and Ser 305 , which differ between GRPR and NMBR are all within 5 Å of the possible binding site. receptor (56), a single amino acid difference between receptor subtypes accounts primarily for the high affinity or selectivity of the antagonist.
The threonine, serine, and phenylalanine residues found to be important in determining GRPR selectivity for both classes of antagonists in the present study have been reported in several studies in other GPCRs to play a critical role in determining high affinity interaction and selectivity of the ligand for a GPCR subtype (16,17,57,64,69). A threonine in the sixth transmembrane domain of the bradykinin 2 receptor is required for high affinity interaction with the peptide antagonist D-Arg-[Hyp 3 ,D-Phe 7 ]BK (NPC567) (16), two threonine residues in the fifth transmembrane domain of the m3 muscarinic receptor are critical amino acids for high affinity interaction with agonists (acetylcholine, carbachol) (71), and a serine in the third transmembrane domain of the AT 1A AII receptor is one of the critical amino acids required for high affinity interaction with the peptide antagonist [Sar 1 , Ile 8 ]AII (72). In the former two studies (16,71), it was proposed that the mechanism of the enhanced affinity due to the threonine or serine was by enhancing hydrogen bonding with the ligand. The presence of a phenylalanine residue in either a transmembrane domain or extracellular domain in other GPCRs also has been shown to play an important role in high affinity ligand-receptor interaction and ligand GPCR selectivity (16,17,57,64,69). The presence of a phenylalanine in the sixth transmembrane domain of the bradykinin 2 receptor (Phe 261 ) is necessary for high affinity and selectivity of the peptide antagonist NPC567, and this effect was proposed to be mediated by an amino-aromatic or aromatic-aromatic interaction (16). A phenylalanine residue in the second transmembrane domain of the CCK-A receptor (Phe 107 ) or two phenylalanine residues in the second extracellular domain of the CCK-B receptor are critical amino acids for high affinity interaction with peptide agonist CCK-8 (64). Similarly, with substance P interaction with the neurokinin-1 receptor (66), endothelin with the ET A receptor (57), and the peptide antagonist [des-Arg 10 -Leu 9 ]kallidin with the bradykinin 1 receptor (17), phenylalanines in either extracellular domains or transmembrane regions are essential for high affinity interaction and selectivity. Phenylalanine, similar to other aromatic amino acids, characteristically interacts with ligands by functioning as a strong locus of cation-binding (73). The cationbinding occurs through the side chains from the aromatic amino acids such as phenylalanine, tyrosine, or tryptophan (74). In our study, the substitution of methionine from the comparable position of NMBR for phenylalanine in the GRPR resulted in a marked decrease in affinity for JMV594 and JMV641, supporting the conclusion that cation-interactions are important for the high affinity and selectivity of both of the peptide antagonists.
At present, it is not possible to compare the receptor structural determinants of high affinity interaction and GRPR selectivity of these two different classes of peptide antagonists with that for either the agonist bombesin or GRP, peptide agonists of which both antagonists are close structural analogues. There are no studies on the molecular determinants of either bombesin or GRP's high affinity interaction with the GRPR or of the selectivity of GRP for GRPR over the NMBR. However, our study does demonstrate that these two different antagonists have both similarities and differences in their molecular determinants of high affinity GRPR interaction and selectivity. They are similar in that with both the statine antagonist JMV594 and the pseudopeptide antagonist JMV641, the amino acid differences in the fourth extracellular domains of NMBR and GRPR are primarily responsible for their high affinity and selectivity for GRPR. They are also similar in that of the 11 amino acid differences in this domain between GRPR and NMBR, the three most important for each antagonist for high affinity interaction are Thr 297 , Phe 302 , and Ser 305 in the GRPR. Furthermore, with each antagonist, these three amino acids had a potentiating effect on enhancing affinity, with the combination having a much greater effect than any of the three substitutions alone. These results suggest that for both antagonists cation-interactions and hydrogen bonding through serine and threonine receptor residues are of primary important in determining high affinity interaction. However, some results suggest that these two different classes of antagonists have some important differences in the molecular determinants of their GRPR selectivity. First, for the pseudopeptide antagonist, JMV641, the GRPR loss of affinity and NMBR gain of affinity chimeric studies showed that differences in second and third extracellular domains were of some importance in determining its GRPR selectivity, whereas for the statine analogue, JMV594, they were not important. Second, within the fourth extracellular domain, which was the most important domain for GRPR selectivity for both antagonists, the presence of Thr 297 , Phe 302 , and Ser 305 in the GRPR was relatively more important for the statine antagonist JMV594's selectivity than for the pseudopeptide antagonist, JMV641. With both NMBR gain of affinity chimeric receptors made by substituting the extracellular domain of NMBR with that from GRPR and with the gain of affinity combination point mutant, [Thr 298 ,Phe 303 ,Ser 306 ]NMBR, the gain in affinity was much greater for JMV594 than JMV641. Specifically, with the replacement of the fourth extracellular domain of NMBR by that of GRPR ([e4-GRPR]NMBR) there was a 30fold greater gain in affinity for JMV594 than JMV641 (i.e. 345-versus 11-fold increase, respectively). Similarly, with insertion of Thr 298 , Phe 303 , and Ser 306 in NMBR, there was a Ͼ20-fold greater increase in affinity for JMV594 than JMV641 (Ͼ133-versus 6.5-fold increase, respectively). These results demonstrate that while the molecular determinants of receptor subtype selectivity for these two closely related classes of GRPR antagonists is generally similar, there are also some important differences.
To attempt to gain additional insight into why these three amino acids of the fourth extracellular domain are important in determining the high GRPR selectivity of these two antagonists, three-dimensional modeling of this region of the GRPR was undertaken. Modeling of the GRPR was performed based on the results with bovine rhodopsin, which has recently been crystallized and characterized by x-ray diffraction (75). In rhodopsin, the fourth extracellular domain is well defined and solvent-accessible (75). The hydropathy profile of the sequence is markedly more hydrophilic in the loop region than in the adjoining transmembrane helical regions. This profiling was used to determine the ends of the loop region in GRPR, and the sequence of the extracellular loop was threaded onto the threedimensional structure of rhodopsin. When the sequences were aligned, the loop regions coincided, and any shift in the alignment resulted in a higher threading energy, indicating a less favorable environment for the sequence. In this model, the critical Thr 297 , Phe 302 , and Ser 305 residues in the fourth extracellular domain of the GRPR were found to face the interior of a binding pocket formed primarily by the loop of the fourth extracellular domain (Fig. 11). Our receptor model indicates that Arg 288 , Ser 289 , Tyr 290 , Tyr 292 , Ser 293 , Glu 294 , Val 295 , Asp 296 , Thr 297 , His 301 , Phe 302 , and Ser 305 in the fourth extracellular domain of GRPR are within the 5 Å of the putative binding site and could interact with the antagonists (Fig. 11). However, Arg 288 , Ser 289 , Tyr 292 , Glu 294 , Asp 296 , and His 301 are in a comparable position in the NMBR and therefore are un-likely to be important in the selectivity of the JMV594 and JMV641 for the GRPR. Of the other six amino acids (Tyr 290 , Ser 293 , Val 295 , Thr 297 , Phe 302 , and Ser 305 ), our mutagenesis studies show that only Thr 297 , Phe 302 , and Ser 305 are important in determining the selectivity of JMV594 and JMV641 for the GRPR over the NMBR. Substitution of the other three amino acids (Tyr 290 , Ser 293 , and Val 295 ) by the comparable amino acids from the NMBR (Phe 291 , Lys 294 , and Ile 296 ) caused no change in the affinities of JMV594 and JMV641, suggesting that either the backbone substitutions of these amino acids were not involved in ligand interaction or the comparable amino acid from NMBR was sufficiently similar to effectively replace the GRPR amino acid. Therefore, we conclude that Thr 297 , Phe 302 , and Ser 305 , which are the critical amino acids in determining the selectivity of JMV594 and JMV641, each are facing and within 5 Å of the proposal binding pocket.
In conclusion, our receptor chimeric gain and loss of affinity studies showed that the fourth extracellular domain of the GRPR was the principal receptor region responsible for the high affinity and selectivity of the statine antagonist JMV594 and the pseudopeptide antagonist JMV641 for the GRPR over the NMBR. Our mutagenesis studies show that Thr 297 , Phe 302 , and Ser 305 in the fourth extracellular domain of the GRPR were the key amino acid differences in determining the GRPR selectivity and high affinity interaction of these two antagonists. To our knowledge, this is the first study that reveals the molecular basis of action between the GRPR and any of its classes of high affinity peptide antagonists. The results from these studies allowed us to obtain a more detailed picture of the receptorligand interaction and propose a model that could account for important receptor-ligand interactions. The availability of this model could be of use for studying the molecular basis of the interaction of this class of receptor not only with other peptide antagonists but also with natural agonists and peptoid antagonists.