Introduction

Top Summary Introduction Procedures Results Discussion References

In G protein-coupled receptors with 7TM, a main ligand-binding crevice is believed to be located between the loops and the outer segments of TM-III, TM-IV, TM-V, TM-VI, and TM-VII (Schwartz and Rosenkilde, 1996). The adrenoceptors were among the first 7TM receptors to be cloned and subsequently characterized in respect of ligand-binding sites. Through a multidisciplinary effort using receptor mutagenesis, affinity cross-linking, and biophysical approaches with fluorescent probes, strong evidence was gathered in favor of the notion that monoamine agonists bind to a pocket located deep in the transmembrane domain of the receptor structure among TM-III, TM-IV, TM-V, and TM-VI (Strader et al., 1994). For example, in the ₂-adrenoceptor, norepinephrine is presumed to bind through its amine function to a conserved aspartic acid residue in TM-III (AspIII:08) and to make an aromatic-aromatic interaction with a phenylalanine residue in TM-VI as the catechol ring of the ligand is oriented by hydrogen bond formation between its hydroxyl groups and two serine residues located one helical turn apart in TM-V (Strader et al., 1994) (Fig. 1). Only recently was the stereospecific recognition of the -OH group of the agonist isoproterenol shown to occur through interaction with an asparagine residue located one helical turn more exterior in TM-VI, AsnVI:20 (Wieland et al., 1996). Mutational mapping of binding sites for other monoamines indicates that they presumably also bind to this deep pocket in the main ligand-binding crevice of their respective receptors through interaction with the corresponding or neighboring residues as identified in the ₂-adrenoceptor (Schwartz, 1994; Strader et al., 1994; Strange, 1996; Strosberg, 1993).

View larger version (67K): [in this window] [in a new window]   Fig. 1.   Helical wheel diagram of the 2-adrenoceptor and helical wheel and serpentine diagram of the tachykinin NK1 receptor. The helical wheel diagrams are built over the rhodopsin structure (Schertler et al., 1993) as interpreted by Baldwin (1993) with TM-III as the central helix and in the counter-clockwise orientation as viewed from outside the cell, which recently was probed experimentally (Elling and Schwartz, 1996). Helix/helix interaction has been optimized on the basis of metal ion site engineering (Elling et al., 1995; Elling and Schwartz, 1996). White letters on black, residues presumed to be involved in agonist binding in the 2-adrenoceptor; black letters on gray, residues previously shown to be involved in substance P binding in the NK1 receptor (Fong et al., 1992; Huang et al., 1994; Kage et al., 1996). In the NK1 receptor, residues probed by alanine substitution and steric hindrance mutagenesis in the current study are shown (white letters on black). Generic numbering system for residues in 7TM receptors indicates helix number (roman type) followed by the residue number (italic type separated by a colon) here and in the text (Baldwin, 1993).

Among 7TM receptors, one of the most thoroughly studied peptide systems is the tachykinin NK₁ receptor. In contrast to monoamine receptors but in analogy with other peptide receptors, several presumed interaction points for the endogenous ligand substance P have been mapped to the amino-terminal extension and to extracellular loops, as well as to the most exterior helical turn of, for example, TM-III and TM-VII (Fig. 1) (Fong et al., 1992; Huang et al., 1994). Extensive mutational analysis performed by several groups has, however, failed to identify an interaction point for substance P in the actual transmembrane domain of the NK₁ receptor corresponding to the site at which monoamines are presumed to bind and activate their receptor (Fong et al., 1992, 1993, 1994a, 1994b; Gether et al., 1993b, 1994a, 1994b; Huang et al., 1994; Strader et al., 1994; Zoffmann et al., 1993) [for a discussion of the effect of substitutions on the hydrophilic, presumed inner face of TM-II, see Huang et al. (1994) and Rosenkilde et al. (1994)]. However, this deep portion of the main ligand-binding crevice is where mutations have indicated that nonpeptide antagonists are binding. Importantly, the identification of many of the presumed interaction points for nonpeptide antagonists has in fact been based on rather small effects on binding affinity. For example, it is generally presumed by others (Fong et al., 1994a; Zoffmann et al., 1993) as well as by the current investigators that the prototype nonpeptide antagonist CP96,345 is interacting with HisVI:17 (His265); however, alanine substitution of HisVI:17 in fact hardly affects the binding of CP96,345. The presumed interaction of CP96,345 with this residue is based on the effect of mutations on analogs of CP96,345, not the compound itself, which obviously is a much weaker augmentation (Fong et al., 1994a; Zoffmann et al., 1993).

Mutational analysis of binding sites in 7TM receptors as well as most other proteins has been performed through alanine scan mutagenesis (Schwartz, 1994). This procedure, in which the side chain at a particular position basically is just "removed," is in general relatively safe in terms of not disturbing the overall structure of the protein. However, alanine substitutions may not be very effective at locating contact residues in binding sites. Based on the X-ray structure of human growth hormone in complex with its receptor and from the exhaustive mutational analysis of both of these proteins performed by Clackson and Wells (1995), it seems that a surprisingly small fraction of the actual contact residues in the ligand/receptor complex are really important for the binding energy. Alanine substitution of most of the residues in the interface between the hormone and its receptor did in fact not impair significantly the binding affinity (Clackson and Wells, 1995). In the case of 7TM receptors, it could be argued that a peptide such as substance P may obtain most of its binding energy through interactions with residues in the exterior part of the receptor, but it could still activate the receptor through interactions with residues corresponding to those to which monoamines bind. If so, alanine substitutions in the transmembrane segments of the NK₁ receptor would have given false-negative results.

In the current study, the deep part of the main ligand-binding crevice, as defined by the binding site for isoproterenol in the ₂-adrenoceptor, is probed in the NK₁ receptor by both alanine substitution and steric-hindrance mutagenesis (i.e., substitutions with larger side chains, which are assumed to fill up the presumed pocket). The introduction of larger side chains was performed at positions at which our molecular models of the receptor indicated that such substitutions would be "allowed" (i.e., where they would not be expected to harm helix/helix interactions). These molecular models have been optimized by, for example, the construction of a series of interhelical metal ion sites (Elling et al., 1995; Elling and Schwartz, 1996). Four structurally related peptide agonists and nine nonpeptide antagonists, most of which are structurally distinct, were tested in these NK₁ constructs.

	Experimental Procedures

Top Summary Introduction Procedures Results Discussion References

Materials. All peptides were purchased from Peninsula (St. Helens, Merseyside, UK). Nonpeptide compounds for which the structures are shown in Fig. 2 were kindly donated. CP96,345 and CP99,994 were provided by Dr. John A. Lowe III (Pfizer, Groton, CT) (Snider et al., 1991; Rosen et al., 1993). CGP49,823 was provided by Dr. Walter Shilling (Ciba/Novartis, Basel, Switzerland) (Schilling et al., 1993). RP67,580 and RPR100,893 were provided by Dr. Claude Garret (Rhone Poulenc, Paris, France) (Garret et al., 1991; Tabert and Peyronel, 1994). CAM4092 was provided by David Howell (Parke Davis, Cambridge, UK) (Boyle et al., 1994). LY303,870 was provided by Dr. Philip Hipskinds (Eli Lilly, Indianapolis, IN) (Gitter et al., 1995). SR140,333 was provided by Drs. Xavier Edmons-Alt and Jean-Claude Breliére (Sanofi Recherche, Montpelier, France) (Emonds-Alt et al., 1993b). FK888 was provided by Drs. T. Fujii and D. Hagiwara (Fujisawa, Osaka, Japan) (Fujii et al., 1992). Pfu polymerase was from Stratagene (La Jolla, CA). AG 1-X8 anion-exchange resin was from BioRad Laboratories (Hercules, CA). myo-[³H]Inositol (PT6-271), BH reagent (specific activity, 2000 Ci/mmol), and Thermo Sequenase fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP were from Amersham (Little Chalfont, UK).

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Structures of peptide agonists and nonpeptide antagonists for the NK1 receptor used in the current study. Bold, four amino acid residues common to all four tachykinin peptides. The lead compounds for the nonpeptide antagonists CP96,345 (Snider et al., 1991), CP99,994 (Rosen et al., 1993), RP67,580 (Garret et al., 1991), RPR100,893 (Tabert and Peyronel, 1994), LY303,870 (Gitter et al., 1995), and SR140,333 (Emonds-Alt et al., 1993b) were found by screening of chemical files, whereas the initial lead compounds for CGP49,823 (Schilling et al., 1993), CAM4092 (Boyle et al., 1994), and FK888 (Fujii et al., 1992) were peptide-based compounds.

Construction of mutant receptors. The cDNA encoding the wild-type human NK₁ receptor was cloned into the eukaryotic expression vector pTEJ-8 (Johansen et al., 1990). Mutations were constructed by PCR using either (1) the overlap extension method (Horton et al., 1989) or (2) a combined extension of mutated internal primers by T4 DNA polymerase followed by selective amplification of the mutated DNA strand by PCR (Stappert et al., 1992). The PCR products were digested with appropriate restriction endonucleases, purified, and cloned into the pTEJ8-NK₁. All PCR experiments were performed using pfu polymerase according to the instruction of the manufacturer. All mutations were verified by restriction endonuclease mapping and subsequent DNA sequence analysis using the Thermo Sequenase fluorescent labeled primer cycle sequencing kit with 7-deaza-dGTP on an Alfexpress DNA sequencer according to the manufacturer's instructions (Pharmacia Biotech, Uppsala, Sweden).

Transfections and tissue culture. COS-7 cells were grown in Dulbecco's Modified Eagle's Medium 1885 supplemented with 10% fetal calf serum, 2 mM glutamine, and 0.01 mg/ml gentamicin. The expression plasmids containing the cDNAs encoding the wild-type or mutant receptors were transiently expressed after transfection according to the calcium phosphate precipitation method (Gether et al., 1993a).

Binding experiments. Monoiodinated ¹²⁵I-BH-labeled substance P was prepared and purified by high performance liquid chromatography (Gether et al., 1993a). Transfected COS-7 cells were transferred to culture plates 1 day after transfection. The number of cells per well was determined by the apparent expression efficiency of the individual clones with an aim of 5-10% binding of the added radioligand. Two days after transfection, cells were assayed by competition binding for 3 hr at 4° using 25 pM ¹²⁵I-BH-labeled substance P plus varying amounts of unlabeled ligand in 0.5 ml of 50 mM Tris·HCl buffer, pH 7.4, supplemented with 150 mM NaCl, 5 mM MnCl₂, 0.1% (w/v) bovine serum albumin, and 40 µg/ml bacitracin. Nonspecific binding was determined as the binding in the presence of 1 µM substance P. Determinations were made in triplicate.

Phosphatidylinositol assay. One day after transfection COS-7 cells (0.3 × 10⁶ cells/well) were incubated for 24 hr with 5 µCi of myo-[³H]inositol in 1 ml of inositol-free Dulbecco's 1885 medium supplemented with 10% fetal calf serum, 2 mM glutamine, and 0.01 mg/ml gentamicin per well. Cells were washed twice in PI buffer consisting of 20 mM HEPES, pH 7.4, supplemented with 140 mM NaCl, 5 mM KCl, 1 mM MgSO₄, 1 mM CaCl₂, 10 mM glucose, and 0.05% (w/v) bovine serum albumin and were incubated in 0.5 ml of PI buffer supplemented with 10 mM LiCl at 37° for 30 min. After stimulation with increasing concentration of substance P for 45 min at 37°, cells were extracted with 10% ice-cold perchloric acid followed by incubation on ice for 30 min. The resulting supernatant was neutralized with KOH in HEPES buffer, and the generated [³H]inositol phosphates were purified on BioRad AG 1-X8 anion-exchange resin (Berridge et al., 1983). Determinations were made in triplicate.

Calculations. IC₅₀ and EC₅₀ values were determined by nonlinear regression, and K_i and B_max values were calculated using the Inplot 4.0 software (GraphPAD Software, San Diego, CA).

	Results

Top Summary Introduction Procedures Results Discussion References

In the NK₁ receptor, residues located in the presumed deep part of the main ligand-binding crevice at positions corresponding to the interaction points for agonists in the ₂-adrenoceptor were substituted with alanine or with amino acid residues presenting larger, polar side chains.

Substitutions of ProIII:08 (Pro112). The important, conserved aspartic acid in TM-III of monoamine receptors, which most convincingly has been implicated in binding of the amine function through complementary chemical modifications performed on the ligand and receptor (Strader et al., 1991), is a proline residue in the NK₁ receptor, ProIII:08 (Pro112). This residue was substituted by alanine, aspartic acid, and histidine, of which the latter two would be expected to occupy considerably more space than the pyrrolidine ring of the proline. Furthermore, these two residues would change the local environment considerably by introducing polarity and possibly charge in front of TM-III in this part of the main ligand-binding pocket (Fig. 1). Because ProIII:08 (Pro112) does not correspond to one of the highly conserved proline residues of the transmembrane helices in 7TM receptors, it was not expected that substitution with a "normal" amino acid residue in general would perturb the receptor structure by altering the backbone configuration at this site. This assumption was confirmed by the normal agonist binding and signal transduction observed in these constructs (see below).

View larger version (26K): [in this window] [in a new window]   Fig. 3.   Competition binding curves for four nonpeptide compounds in the wild-type (WT) NK1 receptor and in two mutant forms with substitution of ProIII:08 (Pro112) that corresponds to the main monoamine interaction point in TM-III, AspIII:08 (Asp113). CP99,9994 and RPR100,893 represent the group of compounds that are seriously affected by the steric hindrance mutations, and FK888 and LY303,870 represent compounds that are less or not at all affected (see tables). Binding experiments were performed as whole-cell binding on transiently transfected COS-7 cells using 125I-BH-labeled substance P as radioligand.

Substitution of ThrV:09 (Thr201) and IleV:12 (Ile204). In TM-V of the ₂-adrenoceptor serine residues located at two positions one helical turn apart, SerV:09 (Ser204) and SerV:12 (Ser207) are believed to be involved in the formation of hydrogen bonds to the hydroxyl groups of the catechol ring (Strader et al., 1994). In the NK₁ receptor, these positions are occupied by a threonine residue, ThrV:09 (Thr201), and an isoleucine residue, IleV:12 (Ile204), respectively (Fig. 1). ThrV:09 was mutated to alanine, isoleucine, or histidine, whereas IleV:12 was probed by only histidine substitution and only in combination with a similar exchange at position V:09. Similar results were obtained by either "removing" the polar side chain of ThrV:09, as in the alanine substitution, or by introducing the slightly larger and apolar side chain of isoleucine at this position (Table 2). Among all the ligands tested, the binding of only two nonpeptide antagonists, CAM4092 and LY303,870, was impaired by these two mutations (Table 2). In fact, the binding of three of the peptide agonists was improved 2-4-fold (Table 2). In contrast, introduction of a histidine residue at position V:09 relatively selectively reduced the binding affinity of the two structurally related nonpeptide antagonists, CP96,345 and CP99,994 (Table 2). Interestingly, residue V:09 is located in an i + 4 position [i.e., one helical turn] below HisV:05 (His197), which previously has been demonstrated to be a crucial contact point for CP96,345 (Fong et al., 1993; Gether et al., 1993b, 1994b).

Substitutions of PheVI:20 (Phe268). In TM-VI, two residues have been implicated in agonist binding in the ₂-adrenoceptor: PheVI:17 and AsnVI:20. The major interaction site for the catechol ring itself, PheVI:17 (Phe290) is in the NK₁ receptor already occupied by a residue with a relatively large side chain [i.e., the histidine residue HisVI:17 (His265)] (Fig. 1). This position has been probed extensively with multiple different substitutions, including phenylalanine, and in no case was any effect on substance P binding observed, whereas the binding of nonpeptide antagonists of several classes was impaired by these mutations (Fong et al., 1994a; Zoffmann et al., 1993). Therefore, we did not further address HisVI:17.

Steric hindrance mutagenesis of alanine residues in the transmembrane segments. The occurrence of an alanine residue on the inner face of a transmembrane helix will conceivably create a hole in the protein structure due to the small size of the side chain. This hole can then be occupied by a side chain from another, interdigitating helix, or it can function as part of the binding site for a receptor ligand. Two such alanine residues were probed by steric hindrance mutagenesis through substitution with valine (i.e., introduction of two extra methyl groups).

Signal transduction experiments. The results presented show that neither alanine substitutions nor steric hindrance mutations in the deep part of the main ligand-binding pocket affected the affinity of peptide agonists on the NK₁ receptor as determined in ligand-binding experiments. Nevertheless, theoretically the peptides could obtain most of their binding energy through interactions with residues located in the more exterior parts of the receptor but still perform the actual activation of the receptor through interaction with residues located deep in the transmembrane domain. However, as shown in Fig. 4 and in Table 4, substance P was able to stimulate phosphatidylinositol turnover in all these mutant receptors with a similar EC₅₀ value as observed in the wild-type NK₁ receptor. In the mutant receptors, the E_max value for the substance P-induced phosphatidylinositol turnover was also similar and the same as that of the wild-type receptor, except for the construct with the double histidine substitution in TM-V and the alanine and histidine substitution of PheVI:20, which demonstrated a 3-5-fold reduction in maximal response to substance P (Table 4).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Dose-response curves for stimulation of phosphatidylinositol (PI) turnover in response to substance P in the wild-type (WT) NK1 receptor and in mutant forms of this with substitution of ProIII:08 (Pro112) to aspartic acid and histidine, respectively.

	Discussion

Top Summary Introduction Procedures Results Discussion References

One of the major problems in mutational mapping of binding sites is the difficulty in ruling out false-negative results (Schwartz et al., 1995). This is demonstrated, for example, in the case of growth hormone and its receptor, for which the actual binding site now is known due to X-ray analysis of the receptor/ligand complex (Clackson and Wells, 1995). False-negative results are mainly a problem when alanine substitutions are used because this procedure basically creates only an extra "hole" in the presumed binding pocket through removal of the side chain. The effect of an alanine substitution is highly dependent on the relative contribution to the binding energy of the replaced residue. A residue may be part of the ligand/receptor interface without in fact contributing much to the total binding energy (Clackson and Wells, 1995). However, the introduction of a larger side chain in a presumed binding pocket conceivably would cause many more problems for the ligand by impairing the interaction not only with the mutated residue but presumably also with neighboring residues due to the incongruence that would be created in a larger part of the interface. In the current study, this is reflected in the observed 100-1000-fold decrease in affinity for a number of nonpeptide antagonists as the result of the introduction of larger and polar residues, such as aspartic acid and histidine, in the deep part of the main ligand-binding crevice between TM-III and TM-V in the NK₁ receptor (Tables 1 and 2). It is on the bases of these significant, positive "hits" that the lack of effect of such substitutions on the binding of substance P and related tachykinin peptide agonists as well as a group of nonpeptide antagonists is viewed as strong evidence in favor of the notion that the peptides and certain nonpeptides do not use this deep pocket as part of their binding site in the NK₁ receptor.

Implications for nonpeptide antagonist binding. The steric hindrance mutants in the deep pocket among TM-III, TM-V, and TM-VI result in a very substantial loss of affinity for a number of the nonpeptide antagonists acting on the NK₁ receptor (Tables 1 and 2). These results combined with the results of a number of previous studies performed with some of these compounds suggest strongly that the binding sites for compounds such as CP96,345, CP99,994, RP67,580, RPR100,893, and CAM4092 are in fact located in this pocket (Schwartz, 1994; Schwartz et al., 1995; Strader et al., 1994). For compounds such as CGP49,823 and LY303,870, the results point in the same direction, but the effects of the steric hindrance mutations were not as consistent. FK888 was resistant to all kinds of mutational mapping performed in the deep pocket, but together with LY303,870, it was the ligand that was most seriously affected by the alanine substitution at position VI:20 located at the exterior border of the pocket, indicating that these compounds may be binding mainly in the more shallow part of the main ligand-binding crevice (Fig. 5). It should be noted that false-positive results are still difficult to rule out. Thus, it is impossible to differentiate between the effect of substituting an actual contact residue between the receptor and a ligand and the effect of substituting a second row residue (Schwartz et al., 1995). However, the fact that the peptide agonists could bind normally and activate signal transduction in these constructs indicates that the overall structure of the receptor was not perturbed.

View larger version (62K): [in this window] [in a new window]   Fig. 5.   Outside-in views of the main ligand-binding crevice of the 2-adrenoceptor and substance P NK1 receptor with presumed contact residues for agonists (black letters on white). In the NK1 receptor, these residues have been identified through receptor mutagenesis in the amino-terminal segment (Asn23, Gln24, and Phe25) (Fong et al., 1992) in the most exterior parts of TM-III (HisIII:04, His108) (Fong et al., 1992) and TM-VII (TyrVI:02, Tyr287) (Huang et al., 1994) and by affinity cross-linking in exterior loop 2 (Met181) using photoreactive, radiolabeled substance P (Kage et al., 1996). Steric hindrance mutagenesis in the current study indicates that many nonpeptide antagonists bind in the deep pocket corresponding to the presumed binding site for monoamines in their receptors but that this is not the case for the peptide agonists and, importantly, that several nonpeptide antagonists, such as FK888, LY303,870, and SR140,333, also do not map to this pocket. PheVI:20 (Phe268), which corresponds to the presumed interaction point for the -OH group of isoproterenol in the 2-adrenoceptor, AsnVI:20 (Asn293), has been implicated as a possible contact point for both substance P and a number of nonpeptide antagonists in the current study (see Table 3).

Implications for peptide agonist binding. As shown in the current study, it is possible to fill out the deep pocket of the main ligand-binding crevice of the NK₁ receptor without affecting the binding or function of substance P or the binding and function of three other high affinity peptide agonists of this receptor. Of the five main presumed interaction points for agonists in the ₂-adrenoceptor, only one, position VI:20 (which is the most exteriorly located site) seems to be shared by substance P in the NK₁ receptor. Previously, site-directed mutagenesis and affinity cross-linking have been used to identify presumed contact points for substance P in the amino-terminal exterior segment of the receptor, in extracellular loop 2 close to the disulfide bridge connecting the top of TM-III, and at the most exterior parts of TM-III and TM-VII (i.e., in the more "shallow" part of the main ligand-binding crevice) (Fong et al., 1992; Huang et al., 1994; Kage et al., 1996; Li et al., 1995) (Figs. 1 and 5). The substitution of a number of residues down along the inner, hydrophilic face of TM-II in the NK₁ receptor does impair the ability of substance P to compete with radiolabeled antagonists seriously (Huang et al., 1994). However, the affinity of the peptide as measured in homologous binding assays and bioassays is hardly affected (Huang et al., 1994; Rosenkilde et al., 1994). Thus, it is unlikely that these deeply located residues are part of the high affinity binding site for substance P (Rosenkilde et al., 1994).

Does PheVI:20 (Phe268) represent a common interaction point for peptide agonists and nonpeptide antagonists?. It has been surprisingly difficult to locate common interaction points in the NK₁ receptor for substance P and competitive nonpeptide antagonists, although a relatively large number of interaction sites have been identified for each ligand (Gether et al., 1993b; Schwartz, 1994; Schwartz et al., 1995; Strader et al., 1994). PheVI:20 seems to be the first good candidate for such a residue, although the effect on peptide affinity is only 11-fold and the binding of the homologous peptides was not affected (Table 3). It cannot be excluded that the generally negative effect on ligand binding of substitution of the naturally occurring, large phenylalanine residue with the smaller alanine residue could be a result of an indirect effect. The facts that substitution of PheVI:20 with the equally large or even larger side chains of tryptophan and histidine had almost negligible effects on the binding of all 13 tested ligands and that most compounds were negatively affected by the alanine substitution at this position could be taken as support for the notion that the effect of the alanine substitution is indirect. However, it could equally well be argued that both tryptophan and histidine could substitute well for the natural phenylalanine residue in a direct receptor/ligand interaction. In the overall picture of presumed interaction points for substance P, PheVI:20 fits rather well between the residues at the outer segments of TM-III and TM-VII and Met181 located next to the disulfide bridge from the top of TM-III, which by affinity cross-linking has been implicated as a major interaction site for substance P (Kage et al., 1996) (Figs. 1 and 5). Thus, it seems likely that PheVI:20 could form part of the "inner wall" or "floor" of the binding site for substance P in the outer portion of the main ligand-binding crevice of the NK₁ receptor. If the effect of alanine substitution of PheVI:20 on the binding of the nonpeptide antagonists is also considered to be a direct effect, then this residue is the first example of the otherwise surprisingly elusive common interaction point for substance P and nonpeptide antagonists. However, as previously discussed, it is a matter of opinion whether such a single residue should be considered to represent a case of overlap in binding site in a common rigid mold or instead forms part of different binding sites occurring in different receptor conformations (Schwartz et al., 1995). We favor the latter possibility, although the answer to this question will not be known until the actual three-dimensional structures of the various receptor/ligand complexes have been determined.