Further Delineation of the Two Binding Sites (R*n) Associated with Tachykinin Neurokinin-1 Receptors Using [3-Prolinomethionine11]SP Analogues*

Two binding sites are associated with neurokinin-1 substance P receptors in both transfected cells and mammalian tissues. To further delineate the interactions between the crucial C-terminal methionine of substance P and these two binding sites, we have incorporated newly designed constrained methionines,i.e. (2S, 3S)- and (2S,3R)-prolinomethionines. The potencies of these C terminus-modified SP analogues to bind both sites and to activate phosphatidylinositol hydrolysis and cAMP formation have been measured, together with those of their corresponding sulfoxides and sulfones. The molecular nature of these two binding sites and their selective coupling to effector signaling pathways are discussed in the light of current models of receptor activation. The less abundant binding site is coupled to Gq/11 proteins, whereas the most abundant one interacts with Gs proteins in Chinese hamster ovary cells transfected with human neurokinin-1 receptors. The specific orientation of the C-terminal methionine side chain imposed by these constraints shows that macroscopically χ1 and χ2 angles of this crucial C-terminal residue are similar in both binding sites. However, slight but significant variations in the rotation around the Cγ–S bond yield different either stabilizing or destabilizing interactions in the two binding sites. These results highlight the need of such constrained amino acids to probe subtle interactions in ligand-receptor complexes.

Two binding sites are associated with neurokinin-1 substance P receptors in both transfected cells and mammalian tissues. To further delineate the interactions between the crucial C-terminal methionine of substance P and these two binding sites, we have incorporated newly designed constrained methionines, i.e. (2S, 3S)-and (2S,3R)-prolinomethionines. The potencies of these C terminus-modified SP analogues to bind both sites and to activate phosphatidylinositol hydrolysis and cAMP formation have been measured, together with those of their corresponding sulfoxides and sulfones. The molecular nature of these two binding sites and their selective coupling to effector signaling pathways are discussed in the light of current models of receptor activation. The less abundant binding site is coupled to G q/11 proteins, whereas the most abundant one interacts with G s proteins in Chinese hamster ovary cells transfected with human neurokinin-1 receptors. The specific orientation of the C-terminal methionine side chain imposed by these constraints shows that macroscopically 1 and 2 angles of this crucial C-terminal residue are similar in both binding sites. However, slight but significant variations in the rotation around the C␥-S bond yield different either stabilizing or destabilizing interactions in the two binding sites. These results highlight the need of such constrained amino acids to probe subtle interactions in ligand-receptor complexes.
Receptor proteins are found in multiple conformational states with different levels of energy. There is evidence to suggest that seven transmembrane domain receptors can adopt two distinct macrostates, i.e. the inactive R and active R* forms, and a large spectrum of different microstates, i.e. conformational states (1)(2)(3)(4). There are also numerous examples that seven transmembrane domain receptors can activate various effector systems, especially in cell lines transfected with receptors, a phenomenon related or not to overexpression (5). This observation has led to the proposal that the so-called active R* form of a receptor may in fact hide multiple conformational states, some of them recruiting specific G proteins via a fruitful interaction (1)(2)(3)(4). Most of the current work emphasizes the molecular nature of these R and R* forms of the receptor.
We speculate that if a receptor protein is found in different active three-dimensional structures R* n with regard to its in-teraction with different G proteins, the crucial residues of its ligand (i.e. the so-called pharmacophore) might adopt different orientations when interacting with the different active R* n forms of this receptor. This working hypothesis does not foresee at all the recognition mechanism between the ligand and its receptor, i.e. the processes that lead to the selection or induction of particular conformations of both the ligand and the receptor.
The complex formed by substance P ((SP) 1 Arg-Pro-Lys-Pro-Gln-Gln-Phe-Phe-Gly-Leu-Met-NH 2 ) 2 and the human NK-1 tachykinin receptor has been selected for this investigation, as we and others have established that two binding sites associated with this receptor are found both in transfected cell lines (6,7) and in a mammalian tissue (7). These two specific binding sites are not present in equivalent amounts, the ratio being from 1.5/8.5 in transfected cells (whatever radioligand used) (6, 7) to 4/6 in mammalian tissues (7). A controversy still remains whether these binding sites identified both in binding and second messenger coupling assays are canonical active R* n forms, as those determined with constitutive receptors (5,8). Indeed, although in CHO cells expressing the hNK-1 tachykinin receptor the most abundant site labeled by [ 3 H][Pro 9 ]SP is related to cAMP production (9), and the minor one is specifically labeled by [ 3 H]propionyl[Met(O 2 ) 11 ]SP(7-11) correlates with IP production (7), this hNK-1 tachykinin receptor is, however, also coupled to the activation of at least two other path-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Materials-[11-3 H][Pro
Peptide Syntheses-Peptide syntheses were carried out on a 0.1 mmol scale (ABI Model 431A synthesizer), starting from an ␣-p-methylbenzhydrylamine (MBHA) resin (typical substitution, 0.68 mmols/g of resin). All t-butoxycarbonyl amino acids were assembled using dicyclohexylcarbodiimide and 1-hydroxybenzotriazole as coupling reagents. A 10-fold excess of each amino acid was used except for t-butoxycarbonyl-P 3 E Met-OH, which was coupled manually to the MBHA-resin in a 1/1 ratio, and the residual free amino acid groups on the MBHA-resin were then acetylated (acetic anhydride/dichloromethane, 1:5). All attempts to couple the cis isomer t-butoxycarbonyl-P 3 Z Met-OH were unsuccessful. Fmoc-P 3 Z Met-OH was introduced manually in a 1:1 ratio on the MBHA resin after activation as an acid fluoride (20,21). The residual free amino groups on the MBHA-resin were then acetylated. After removal of the Fmoc by piperidine/dimethylformamide, 1:9, all the other tbutoxycarbonyl-amino acids were coupled by the dicyclohexylcarbodiimide-1-hydroxybenzotriazole strategy on the ABI synthesizer. Peptides were cleaved from the resin by anhydrous fluorohydric acid and purified by preparative reverse phase HPLC. The purity of collected fractions was established by analytical HPLC. The sulfoxides were obtained by oxidation of the corresponding crude peptide (2 equivalent NaIO 4 in water, 3 days at 4°C) and purification by preparative HPLC. NMR analysis establishes the presence (1:1 ratio) of both diastereoisomers. The sulfones were prepared by oxidation of the corresponding crude peptide (3 equivalent H 2 O 2 , 30% in water, in glacial acetic acid, room temperature for 15 h).
The purity of all peptides was over 98.5%. Matrix-assisted laser desorption ionization-time of flight analysis was performed by Dr. G 11 ]SP, t R ϭ 6.68 min, MH ϩ (found) ϭ 1405.84. Cell Culture-CHO cells expressing hNK-1 receptors were cultured in Ham's F12 medium supplemented with 100 IU/ml penicillin, 100 IU/ml streptomycin, and 10% fetal calf serum. Cultures were kept at 37°C in a humidified atmosphere of 5% CO 2 . Stable transfections were maintained by geneticin periodic selection.
Measurements of Inositol Phosphate and cAMP Formation-Except when mentioned in the text, PI hydrolysis and cAMP accumulation were determined as described previously (23). Curves were fitted using SIGMA PLOT software (Jandel Scientific, Erkrath, Germany).

RESULTS
Eleven SP analogues modified on the C-terminal position have been tested in binding and second messenger formation studies: SP, [ 11 ]-SP(7-11), K i values are 0.1 and 0.08 nM, respectively) ( Table I).
Although [Nle 11 ]SP retains the activity of SP for NK-1 receptors, as already demonstrated (12), suggesting that the sulphur atom has no role in SP/receptor recognition, we had previously shown that the sulfoxides and sulfone analogues of SP or propionyl-SP(7-11) peptides were not recognized in the same way by NK-1 receptors (7). As shown in Table I 11 ]SP(7-11) binding sites (from 0.08 to 8 nM), [Pro 11 ]SP being the weaker competitor with an affinity of 63 nM.
Previously described correlations between affinity and activity (7,9) have now been extended to a larger population of substance P analogues (n ϭ 53), including those described in this study. No strong correlation can be found either between their affinities for [ 3 11 ]-SP(7-11) binding sites (Fig. 3A, r ϭ 0.808) or between their potencies to stimulate PI hydrolysis and cAMP accumulation (Fig. 3B, r ϭ 0.657, inactive analogues with EC 50 values Ͼ5000 nM were discarded for the correlation analysis). It is also further established that affinity for [ 3 H][Pro 9 ]SP binding sites correlates better with cAMP formation (Fig. 3C, r ϭ 0.922) than with IP formation (Fig. 3D, r ϭ 0.790 11 ]SP(7-11) binding sites highly correlates with their potency to stimulate PI hydrolysis (Fig. 3E, r ϭ 0.975) but not cAMP accumulation (Fig. 3F, r ϭ  0.774).
Further analysis of the results obtained in this study with these eleven SP analogues modified on the C-terminal residue leads to another striking feature. The ratios EC 50(1) /K i1 , which are EC 50(1) (cAMP accumulation) over K i1 ([ 3 H][Pro 9 ]SP binding sites), are quite homogenous; values varying among themselves only from 1-to 3.5-fold (Table I). In contrast, the ratios  11 ]SP(7-11) (K i2 ) binding sites and their related potencies to stimulate cAMP accumulation (EC 50(1) ) and PI hydrolysis (EC 50 (2) 11 ]-SP(7-11) binding sites, its EC 50(2) /K i2 ratio value is 0.9, suggesting a strict relationship between binding and activity parameters. We suspected that the EC 50(2) of compounds with high affinity for [ 3 H][Pro 9 ]SP binding sites may be underestimated, if this more abundant site is not (or weakly) coupled to IP production. It was therefore of interest to investigate whether the potency to stimulate IP formation could be modulated according to the relative amounts of both binding sites (7) and the agonist tested.
Thus, we tested, in PI hydrolysis and cAMP accumulation measurements, the influence of the cell number variation, that is the ratio between the relative site numbers and the amount of agonist on the concentration-response curves.  9 ]SP binding sites, did not present any significant shift in the concentration-response curves, according to assay conditions (Fig. 4). These results establish that the most abundant binding site is not coupled to IP production, otherwise this coupling is far less efficient than the interaction between the minor site and G q/11 proteins. Therefore, the correlation between the affinity for the minor binding site labeled with 11 ]SP (7)(8)(9)(10)(11) and the potency to stimulate PI hydrolysis is even better when only analogues with low affinity (K i Ͼ 50 nM) for the most abundant binding sites labeled with [ 3 H][Pro 9 ]SP are used for linear regression analysis (Fig. 3E, r ϭ 0.975 compared with Fig. 5B, r ϭ 0 Fig. 5A. DISCUSSION This study further supports the concept of the multiplicity of receptor conformations associated with the hNK-1 tachykinin receptor. It has recently been reported that hNK-1 receptor transfected in CHO cells directly activates G␣ q/11 , G␣ s , and G␣ 0 proteins (24) and that there is no cross-talk between cAMP accumulation and IP formation (23,25,26). Therefore, it is likely that EC 50 values of all SP analogues measured in this study arise from direct NK-1 receptor/G protein coupling.
Two theoretical models, based upon the two-state allosteric ternary complex model (27) accounting for promiscuity or multiple receptor-effector coupling, could depict results reported herein. In the three-state model of agonist action, the receptor exists under one inactive (R) and at least two active (R* and R**) states, each active state coupling to a single G protein type (1,2). This model predicts that different agonists would have different affinity for R* and R** and therefore that the associated pharmacology with both effector pathways could be different. A particular agonist (A) acts to displace the receptor equilibria from unoccupied (R, R*, and R**) to occupied (AR, AR*, and AR**) forms (1). Thus, all the occupied forms are enriched according to the affinity constant of the agonist for the three different receptor states R, R*, and R** (1). As there is a finite number of receptors, this model assumes that in terms of efficacy, the two pathways are dependent on one another, which is definitively not the case with the hNK-1 tachykinin receptor. Of the fifty-three tachykinin analogues already tested, some of them (represented by [Nle 11 ]SP in this study) have an efficacy in cAMP accumulation response lower than others, but this efficacy is not correlated to a higher efficacy in A, E, and F) or  53 (B, C, and D)  IP formation response. Thus, there is no mutual depletion of active receptor states, as expected from the model. In the isolated pathways of the three-state model, which may be applied when a receptor is coupled to one G protein in an assay system and a different one in another assay system, the R* and R** states are now independent of one another (1). Under these conditions, one agonist may exhibit different efficacy and potency orders between the two pathways. Data presented herein better fit these isolated pathways of the three-state model. Indeed, the maximal binding capacities differ for the two binding sites, favoring the existence of these independent R* and R** states for the hNK-1 tachykinin receptor. However, working with CHO cells transfected with a single cDNA of human NK-1 receptors, this model would thus imply the hypothetic existence of two distinct receptors or the compartmentalization of receptor signaling pathways that may artificially create conditions in which two pathways could operate in an isolated manner (28 -30). But, it has to be mentioned that identical affinity could be measured for [Pro 9 ]SP and propionyl[Met-(O 2 ) 11 ]SP (7)(8)(9)(10)(11) 11 ]SP (7)(8)(9)(10)(11) in binding assays at 37°C (40-min incubation) or 4°C (180-min incubation), indicating that desensitization and internalization processes should not play a major role in the differential pharmacology associated with these two binding sites (data not shown).

FIG. 3. Correlations between affinities and activities by linear regression analysis with 22 (
In the model assuming agonist trafficking of receptor sig-nals, an agonist selectively leads to the coupling of the receptor to one G protein over another through particular conformations of the receptor (3,4). In this model, the receptor protein adopts a gaussian distribution of multiple conformations, which is specifically shifted by agonists to enrich particular states characterized by their relative affinity for G proteins (3,4). It is also assumed that different agonists may differentially activate signaling pathways not only in terms of realized efficacy but also of potential efficacy (31). For the hNK-1 tachykinin receptor, it has to be noticed that the ratio between EC 50 values of different SP analogues to activate cAMP accumulation (EC 50(1) ) over IP formation (EC 50(2) ), which describes their intrinsic potential efficacy (31) 11 ]SP(7-11) (K i1 /K i2 ) (Fig. 6B). Linear regression analysis reveals two populations of peptides (Fig. 6B) that do not fit the previous discrimination of SP-like and septide-like families (32,33), based only on affinity criteria. Septide, [pGlu 6 , Pro 9 ]SP (6 -11), and [P 3 E Met(O 2 ) 11 ]SP (a SP-like peptide) belong to the first family (A) (Fig. 6B, filled  circles), whereas NKA (a septide-like peptide) and [P 3 E Met 11 ]SP (a SP-like peptide) are members of the second family (B) (Fig.  6B, hollow circles). This correlation analysis suggests that for the same binding affinity ratio (K i1 /K i2 ), analogues of family B have either lower potential efficacy than analogues of family A to activate cAMP accumulation or higher efficacy to stimulate IP production. Therefore, it might be more judicious to associ- ate these two binding sites with their fruitful interaction with G protein, i.e. the less abundant binding sites with G q/11 and the most abundant with G s .
Whether these two binding sites represent, in view of the current models of receptor activation, different conformational states of the same receptor in very slow equilibrium (Ͼ120 min, the time course of binding studies) or distinct receptor proteins differently coupled to G proteins as suggested from other studies (34 -36) is a question that remains to be addressed.
Whatever the explanation may be, the highly homogenous family of peptides modified at the C-terminal methionine led us to conclude that slight, but significant, differences exist in the orientation of the sulphur substituents (C␥-S rotation), in the two binding sites. This conclusion arises from comparison of the potencies (EC 50 (1) and EC 50(2) ) of the two constrained 3-prolinomethionine analogues [P 3 E Met 11 ]SP and [P 3 Z Met 11 ]SP ( 1 and 2 angles) and the potencies of the sulfoxides and sulfone analogues (C␥-S rotation), compared with those of SP. The pseudorotation cycle of the five-membered pyrrolidine led to a fluctuation of the 1 angle between Ϫ90°and Ϫ150°for the 3S-isomer and ϩ90°and ϩ150°for the 3R isomer of 3-prolinomethionine. The two extreme values (Ϫ150°and ϩ150°) are close to that of the trans rotamer (t, 180°) for the corresponding nonconstrained amino acid (Fig. 7). The two other values of 1 for the prolinomethionine (Ϫ90°and ϩ90°) are close to the gauche(-) rotamer (g Ϫ , Ϫ60°) and gauche(ϩ) rotamer (g ϩ , ϩ60°), respectively, for methionine (Fig. 7). With a prolinomethionine substituted in position 4, the 1 value can only oscillate between Ϫ30°and ϩ30°, values close to the g Ϫ and g ϩ rotamer for methionine (Fig. 7). These rotamers are excluded in the hNK-1 tachykinin receptor, because SP(6 -11) analogues substituted with both diastereoisomers of 4-prolinomethionine were completely inactive in guinea pig ileum bioassay (18), a tissue already depicted for the presence of the two binding sites (37 11 ]SP both 1 ϭ Ϫ60°a nd 2 ϭ ϩ60°are excluded because of a gauche interaction between the C-terminal carboxamide and the pyrrolidine moiety. Thus, the ligand can only adopt an orientation correspond-ing to 2 ϭ 180°. This rotamer is accepted by both binding sites. Rotation around the C␥-S bond controls the spatial distribution of the methyl and the lone pairs (sulfides) or the oxygen atoms (sulfoxides, sulfones). These orientations can be deduced from comparison of the potencies of the sulfoxide and sulfone analogues with those of SP for the two signaling pathways, because each peptide [Met(O) 11 11 ]SP, both orientations of the oxygen atom are present in a 1/1 ratio allowing conclusive comparison with the sulfone analogues. Comparison of the potencies of the sulfone analogues to stimulate PI hydrolysis or cAMP formation led us to conclude that rotation around the C␥-S bond results into two different spatial distributions of the sulphur substituents in the two binding sites. Furthermore, one oxygen of the sulfone must establish a stabilizing interaction in the minor binding sites; [P 3 E Met(O 2 ) 11 ]SP and [P 3 E Met(O) 11 ]SP being more potent than SP and equipotent to SP in stimulating PI hydrolysis, respectively. In contrast, one oxygen of the sulfone must develop a destabilizing interaction in the major binding sites, because [P 3 Z Met(O 2 ) 11 ]SP is about 15 times less potent than SP, whereas the diastereoisomeric mixture of the sulfoxides [P 3 Z Met(O) 11 ]SP is far less potent than SP in stimulating cAMP production. It has been established that the oxygens of a sulfone molecule are less polar than the oxygen of the sulfoxide analogues (38, 39) (this difference is still noticeable with an eleven-residue peptide, cf the HPLC elution profiles of the corresponding sulfide, sulfoxide, and sulfone analogues). It is envisageable that the stabilizing interaction involves / stacking, whereas the nature of the destabilizing interaction must be electrostatic with an anionic cavity in the vicinity.
In conclusion, these constrained prolinomethionines turned out to be valuable tools to highlight the importance of the methionine in the SP-receptor complex and to analyze the conformational states of the hNK-1 tachykinin receptor.