Role of aromatic transmembrane residues of the delta-opioid receptor in ligand recognition.

In the present study we examine the role of transmembrane aromatic residues of the delta-opioid receptor in ligand recognition. Three-dimensional computer modeling of the receptor allowed to identify an aromatic pocket within the helices bundle which spans transmembrane domains (Tms) III to VII and consists of tyrosine, phenylalanine, and tryptophan residues. Their contribution to opioid binding was assessed by single amino acid replacement: Y129F and Y129A (Tm III), W173A (Tm IV), F218A and F222A (Tm V), W274A (Tm VI), and Y308F (Tm VII). Scatchard analysis shows that mutant receptors, transfected into COS cells, are expressed at levels comparable with that of the wild-type receptor. Binding properties of a set of representative opioids were examined. Mutations at position 129 most dramatically affected the binding of all tested ligands (up to 430-fold decrease of deltorphin II binding at Y129A), with distinct implication of the hydroxyl group and the aromatic ring, depending on the ligand under study. Affinity of most ligands was also reduced at Y308F mutant (up to 10-fold). Tryptophan residues seemed implicated in the recognition of specific ligand classes, with reduced binding for endogenous peptides at W173A mutant (up to 40-fold) and for nonselective alkaloids at W274A mutant (up to 65-fold). Phenylalanine residues in Tm V appeared poorly involved in opioid binding as compared with other aromatic amino acids examined. Generally, the binding of highly selective delta ligands (TIPPpsi, naltrindole, and BW373U86) was weakly modified by these mutations. Noticeably, TIPPpsi binding was enhanced at W274A receptor by 5-fold. Conclusions from our study are: (i) aromatic amino acid residues identified by the model contribute to ligand recognition, with a preponderant role of Y129; (ii) these residues, which are conserved across opioid receptor subtypes, may be part of a general opioid binding domain; (iii) each ligand-receptor interaction is unique, as demonstrated by the specific binding pattern observed for each tested opioid compound.

In the present study we examine the role of transmembrane aromatic residues of the ␦-opioid receptor in ligand recognition. Three-dimensional computer modeling of the receptor allowed to identify an aromatic pocket within the helices bundle which spans transmembrane domains (Tms) III to VII and consists of tyrosine, phenylalanine, and tryptophan residues. Their contribution to opioid binding was assessed by single amino acid replacement: Y129F and Y129A (Tm III), W173A (Tm IV), F218A and F222A (Tm V), W274A (Tm VI), and Y308F (Tm VII). Scatchard analysis shows that mutant receptors, transfected into COS cells, are expressed at levels comparable with that of the wild-type receptor. Binding properties of a set of representative opioids were examined. Mutations at position 129 most dramatically affected the binding of all tested ligands (up to 430-fold decrease of deltorphin II binding at Y129A), with distinct implication of the hydroxyl group and the aromatic ring, depending on the ligand under study. Affinity of most ligands was also reduced at Y308F mutant (up to 10-fold). Tryptophan residues seemed implicated in the recognition of specific ligand classes, with reduced binding for endogenous peptides at W173A mutant (up to 40-fold) and for nonselective alkaloids at W274A mutant (up to 65-fold). Phenylalanine residues in Tm V appeared poorly involved in opioid binding as compared with other aromatic amino acids examined. Generally, the binding of highly selective ␦ ligands (TIPP, naltrindole, and BW373U86) was weakly modified by these mutations. Noticeably, TIPP binding was enhanced at W274A receptor by 5-fold. Conclusions from our study are: (i) aromatic amino acid residues identified by the model contribute to ligand recognition, with a preponderant role of Y129; (ii) these residues, which are conserved across opioid receptor subtypes, may be part of a general opioid binding domain; (iii) each ligand-receptor interaction is unique, as demonstrated by the specific binding pattern observed for each tested opioid compound.
Opiates are strong analgesic compounds currently used in the treatment of pain. They also have euphoriant action and strong addictive properties. Opiates, as well as endogenous opioid peptides, exert their biological action through three classes of membrane receptors, known as , ␦, and . The recent cloning of three genes encoding these receptors (for a review, see Ref. 1) has shown that opioid receptors belong to the G protein-coupled receptors (GPRs) 1 superfamily with a seven transmembrane domain topology and share high protein sequence identity, particularly in transmembrane and intracellular regions. The availability of opioid receptor clones allows to examine the mechanisms of action of opioid ligands at the molecular level, and here we investigate the structural basis of receptor-opiate interactions in the mouse ␦-opioid receptor (mDOR, see Refs. 2

and 3).
It is believed that the seven putative transmembrane domains of GPRs are folded as helices and tightly associated within the membrane to form the receptor protein core. The binding site of GPRs which recognize small biogenic amines has been best characterized (4). Three-dimensional computer modeling relying on bacteriorhodopsin structure as a scaffold and the analysis of amino acid conservation in sequence alignments were used in conjunction with site-directed mutagenesis data to identify residues which interact with the small nonpeptidic ligands. The binding site was shown to lie within the helical bundle, about 15 Å from the extracellular surface and spanning Tms II to VI, with a central role for Tm III. Other GPRs bind peptide ligands with sizes ranging from tripeptides (e.g. N-formyl peptide or thyrotropin-releasing hormone) to proteins (follicle-stimulating, luteinizing, and thyroid-stimulating hormones). Mutagenesis experiments demonstrated that, in addition to transmembrane residues, extracellular domains participate in ligand recognition (for reviews, see Refs. 5 and 6), indicating that the peptide binding site extends beyond the transmembrane pocket for these receptors.
Opioid receptors are GPR family members endowed with an extremely large ligand repertoire (7). They bind a wide variety of structurally diverse molecules, including small rigid alkaloid compounds, synthetic peptides, as well as a family of endogenous peptides. Opioid receptors are therefore expected to interact with their ligands at multiple sites, both extracellular and intramembranous, an assumption which is supported by initial point mutagenesis and chimeric studies (8 -18). We have focused on the role of transmembrane domains in the ␦ receptor. In a previous study, we have demonstrated that the conserved Asp residue in Tm III, although not interacting directly with the positive charge of opioid ligands, is critically involved in maintaining intact binding properties of the receptor (19). Other reports have shown implication of charged residues found in Tm II and Tm VI in opioid recognition (8,9). In this study, we have used our knowledge from studies of other GPRs and receptor modeling to establish a putative topology of the receptor binding site. We have identified a number of aromatic amino acid residues as candidates for opiate recognition and we have examined the contribution of their side chains to the binding of a large set of structurally distinct opiate molecules, using a site-directed mutagenesis approach. We provide evidence that a cluster of aromatic residues may represent a major area of interaction for most ligands. In addition our results demonstrate a distinct mode of interaction for each ligand under investigation, ruling out the possibility of a unique opioid binding site.

EXPERIMENTAL PROCEDURES
Materials-All ligands tested were obtained from Sigma unless otherwise stated. Bremazocine was from Research Biochemicals International and BW373U86 was kindly provided by K. J. C. Chang (Burroughs Wellcome Co., Research Triangle park, NC); TIPP and TIPP were provided by Astra Pain Control (Montreal, Canada); [ 3 H]naltrindole (specific activity; 44.5 Ci/mmol) was provided by A. Borsodi (Szegued, Hungary), and [ 3 H]diprenorphine (specific activity, 37 Ci/mmol) was obtained from Amersham Corp.
Modeling-The modeling procedure has been previously described in detail for the modeling of the angiotensin AT II receptor (20). Briefly, the mDOR three-dimensional model was constructed following a threestep strategy: (i) a model of the human rhodopsin receptor was built from the resolved structure of bacteriorhodopsin (21); (ii) from that, a model of the best well studied hamster ␤2-adrenergic receptor was built from the obtained human rhodopsin model using a sequence alignment that results from homologies existing between all GPR Tm regions. Many amino acid guide points could be retained in each Tm helix to allow next pertinent use of homology building procedures. As for the building of the opsin receptor, only the seven Tm were built. (iii) This ␤2-adrenergic receptor model was finally used as a template for the modeling of the mouse ␦-opioid receptor itself. The whole modeling was performed with the BIOSYM software using the "Insight Discover," and "Homology" modules (Biosym Technologies, San Diego, CA).
Mutagenesis of mDOR-mDOR cDNA (3) was subcloned in HindIII-BamHI sites of the mammalian expression vector pCDNAI/Amp (Invitrogen). Single stranded DNA for mutagenesis was prepared from pCDNA-mDOR using the helper phage M13KO7 (Promega) and the Escherichia coli strain NM 522 (Stratagene) and was used as template for site-directed mutagenesis. Sequences of oligodeoxynucleotides were chosen to replace the following residues of mDOR receptor either by phenylalanine: Tyr 129 (Tm III) and Tyr 308 (Tm VII) or by alanine: Tyr 129 (Tm III), Trp 173 (Tm IV), Phe 218 and Phe 222 (Tm V), Trp 274 (Tm VI). Oligodeoxynucleotides containing the mutant codon were annealed to the single-stranded template, followed by an elongation with the Klenow fragment and ligase. Heteroduplex plasmid DNA was then used to transform the repair-deficient E. coli strain BMH71-18 mutS (Clontech). Transformants were selected by growth on LB plates supplemented with ampicillin (100 g/ml), and mutations were confirmed by manual DNA sequencing (Sequenase kit; U. S. Biochemical Corp.) in both directions. Sequences were also confirmed by automated DNA sequencing (373A DNA, Perkin-Elmer) using fluorescently labeled nucleotides (Taq DyeDeoxy terminator cycle sequencing kit, Perkin-Elmer).

Expression of Wild-type mDOR and Mutant Receptors in COS Cells and Ligand
Binding-COS-1 cells (1.5 ϫ 10 6 cells/140-mm dish) were transfected with purified plasmids (35 g/dish) using the DEAE-dextran method. After 72 h growth in Dulbecco's modified Eagle's medium with 10% fetal calf serum, the cells were harvested and membranes were then prepared as described previously (3). For binding experiments, various amounts of membrane proteins of mDOR and mutant receptors, ranging from 20 to 100 g, were diluted in 50 mM Tris-HCl,

RESULTS
Opioids are cationic compounds with strong hydrophobic character. Structure-function studies have shown that a positively charged nitrogen atom located at a fixed distance from a phenolic ring, a common structure between alkaloids and opioid peptides, represents the active or "message" part of opioid molecules (see Ref. 22). The positive charge being an absolute requirement for ligand binding, the existence of a salt bridge has long been postulated to be the major interaction between the ligand and its receptor site. In preliminary work (19), we tested the hypothesis that this interaction occurs at Asp 128 in mDOR. This negatively charged residue is present in Tm III of all GPRs that bind small cationic neurotransmitters and is known to interact directly with catecholamines in the ␤2-adrenergic receptor. Our results showed that, although this Asp residue is a structural component of the binding site, it does not interact directly with opioid ligands. Another study has shown that Asp 95 (Tm II), the only other anionic residue present in transmembrane domains, does not represent a common attachment point for opioids (8). The lack of evidence for a main electrostatic interaction between opioids and the ␦ receptor strengthens the hypothesis that non-ionic interactions are predominantly responsible for ligand chelation in the binding site.
Three-dimensional Modeling of mDOR and the Role of Aromatic Residues-We have used three-dimensional computer modeling to identify intramembranous amino acid residues that may face the inner side of the helices bundle and account for ligand binding. Interestingly, we found that the best candidates are predominantly aromatic residues, and their side chains are presented on Fig. 1. On the model we observe that Trp 173 , Phe 218 , Phe 222 , and Trp 274 are clustered and form an aromatic pocket spanning Tms IV, V, and VI. The putative binding site extends over Tms III and VII, where residues facing the pore acquire a more hydrophilic character. Among these, the amphiphilic Tyr 308 residue seems to bridge Tm VII and Tm III. Hydrophilic residues have also been identified as putative anchor points for ligand on this model, such as Asn 131 and His 278 (not shown) or Asp 128 (Tm III), which we have investigated earlier. In the empty receptor, some of the identified aromatic residues seem to be in close contact with other amino acid side chains. Trp 274 may interact with Phe 270 within Tm IV through aromatic-aromatic interaction. Polar interhelices interactions may also be seen with a possible interaction of the OH group of Tyr 308 or the indole NH group of Trp 274 with the carbonyl of Asp 128 via hydrogen bonds.
Residue Tyr 129 , which is adjacent to the critical Asp 128 residue, does not face the binding pocket in this model, but rather participates in reinforcing helical III intrinsic stability through aromatic-aromatic interactions occurring between Tyr 129 , Phe 133 , and Phe 137 . Interestingly, in another model of mDOR derived from bovine opsin crystallographic data, 2 the side chain of Tyr 129 seems more likely to be located inside the pore due to a different position of Tm III relative to adjacent helices. In addition, this tyrosine residue is present in muscarinic m3 and TRH receptors and was shown to play a critical role in ligand recognition (23,24). We therefore have included examination of the role of Tyr 129 in our study despite its outside positioning in our model.
Mutagenesis of the Receptor and Choice of Opioid Ligands-Point mutant receptors constructed in this study are shown on Fig. 2. We have replaced the identified tryptophan (Tms IV and VI) and phenylalanine (Tm V) residues by alanine (W173A, F218A, F222A, and W274A), thus removing the side chains which seem to constitute the aromatic pocket proposed by the model. We have modified the tyrosine residues in Tms III and VII into phenylalanines (Y129F and Y308F) to evaluate the 2 M. Hibert, personal communication.

FIG. 2.
Site-directed mutagenesis of aromatic amino acid residues of mDOR. A schematic representation of the putative seven transmembrane domain topology of the receptor is shown and residues mutated in this study highlighted. The single-letter amino acid code is used. The numbers next to mutated amino acids refer to their positions within the mDOR sequence. possible contribution of the hydroxyl groups. Finally we have replaced tyrosine 129 by alanine (Y129A) to remove the entire phenol moiety and ascertain the role of this aromatic ring in Tm III. We then have examined the binding properties of each mutant receptor. Since opiates and opioid peptides constitute a group of compounds with high structural diversity, we have tested a repertoire of ligands with representative molecules in each class, including endogenous peptides, synthetic peptide analogues, or alkaloid compounds, agonists or antagonists, nonselective or ␦-selective ligands (Fig. 3).
Expression  Fig. 4. Each mutant receptor bound at least one ligand with an affinity comparable with that of the WT receptor, suggesting that the general receptor conformation is maintained in the mutated receptors.
Tyrosine 129 Critically Contributes to the Opioid Binding-Mutant Y129F exhibits reduced affinities for the ligands under study, with effects being very modest (2.8-fold for DPDPE) or more significant (12-fold for bremazocine). This result suggests a general contribution of a hydrogen bond to ligand binding. Remarkably, a stronger reduction (91-fold) is observed for deltorphin II, a ␦-selective agonist isolated from frog skin which does not contain the canonical YGGF N-terminal sequence typical of opioid peptides and therefore may bind in a distinct manner. For this particular peptide, the OH group of Tyr 129 may represent a major anchor point. Further replacement of tyrosine by alanine shows contributions of both the hydroxyl group and the aromatic ring of the tyrosine side chain. Binding of long endogenous peptides (␤-endorphin and dynorphin A), a synthetic peptide (TIPP) or ␦-selective alkaloids (BW373U86 and naltrindole) at Y129A mutant is affected similarly to binding at Y129F mutant, suggesting that the aromatic ring does not contribute to recognition of these compounds. In contrast short endogenous peptides (Leu-and Met-enkephalin), synthetic peptides (DADLE and DPDPE) as well as nonselective alkaloids (bremazocine and naloxone) exhibit affinities for Y129A up to 90-fold lower than those for Y129F. Deltorphin II binding is also reduced 5-fold more and appears as the most affected ligand, with a total decrease in affinity of 430-fold relative to the WT receptor. For these latter compounds, therefore, the aromatic ring of Tyr 129 participates in ligand binding. Altogether, these results underscore a general involvement of Tyr 129 side chain in ligand recognition, with a distinct contribution of both the OH group and the aromatic ring depending on the ligand under study.

Mutations of Tryptophan Residues in Tms IV and VI Strongly Affect the Binding of Specific Ligand Classes-Bind-
ing of the endogenous peptides ␤-endorphin, dynorphin A, Leuand Met-enkephalin at W173A mutant receptor is strongly reduced (21-43-fold), whereas TIPP, BW373U86, naltrindole, and bremazocine affinities remain unaffected. K i values for DADLE and DPDPE significantly decrease, but to a lesser in Tm VII as a hydrogen bond donor by phenylalanine replacement. A decrease in ligand binding is observed for most ligands (5-10-fold), with the exception of bremazocine, TIPP, and Met-enkephalin binding, which remains essentially unaffected. Interestingly, Leu-enkephalin binding is affected by 5-fold, underscoring a distinct mode of interaction for the two highly structurally related enkephalin peptides. Mutant F218A appears weakly involved in ligand binding, since affinities do not decrease more than 5-fold. Binding to the mutant F222A is significantly reduced for two peptidic ligands, DADLE and deltorphin II (14.5-and 13.5-fold decrease affinity). Otherwise all other opioid compounds are less affected (Ͻ7-fold). Amino acid substitutions at positions 308, 218, and 222 therefore modify the binding properties of the receptor, although to a lesser extent than mutations at residues 129, 173, and 274.

DISCUSSION
Three-dimensional modeling has allowed us to identify transmembrane aromatic amino acid residues as possible candidates for opioid binding at the ␦ receptor, and we have investigated their role by site-directed mutagenesis. It is important to note that the mutations under study generally affect the binding of some, but not all investigated opioid ligands. We can therefore assume that structural modifications, that would dramatically change the general conformation of the receptor binding site, do not take place in the mutated receptors. Consequently, our results indicate that amino acid residues Tyr 129 , Trp 173 , Phe 218 , Phe 222 , Trp 274 , and Tyr 308 indeed participate in ligand recognition. Whether alterations of ligand binding observed in this study arise from the disruption of direct ligandreceptor interactions or rather from subtle effects on helical packing is not known.
Role of Tyrosine Residues in Tms III and VII-Compared with other mutations examined in this study, modifications of Tyr 129 (Tm III) induce major impairment of ligand binding. This amino acid residue therefore appears as a major component of the binding site. A tyrosine residue is present at homologous position in m3 muscarinic and TRH receptors and was shown critical in ligand recognition for both receptors (23,24). Thus, in this report, we bring further support for the implication of this particular residue within the seven transmembrane topology. Interestingly, in a previous study (19), we have demonstrated that the adjacent Asp residue (Asp 128 ) is a critical determinant of the binding site conformation. Therefore, the altered binding properties of Y129A and D128N mutant receptors (Ref. 19 and Fig. 5) indicate that both adjacent Tm III residues play an important role in ligand recognition, consistent with structure-function studies, which indicate that helix III plays a central role in GPRs (6).
The significant contribution of Tyr 129 in ligand binding seems to be in contradiction with its outward orientation in the proposed model (see Fig. 1B). One hypothesis may be that the possible implication of Tyr 129 in aromatic stacking suggested by the model (see ''Results'') is important to maintain helix III conformation and that Tyr 129 plays a structural role. This interpretation, however, is not consistent with the observation that the aromatic ring of Tyr 129 does not seem to be involved in the binding of some ligands (␤-endorphin, dynorphin A, TIPP, BW373U86, and naltrindole), which rather interact with the OH group of the tyrosine residue. A more likely possibility to explain the decreased binding potency of opioid ligands at Y129F and Y129A mutants is that the amino acid side chain faces the inner core of the helices bundle and act as an anchor point for these ligands. Refinement of the model is now required to assess this hypothesis. In particular, revision of the alignment procedure and consideration of mutagenesis data on ␤2 receptor may lead to repositioning of helix III.
Although less drastically involved in receptor binding properties, Tyr 308 (Tm VII) contributes significantly to the binding of most ligands tested in this study. This residue is also found in TRH receptor, bradykinin receptor, and in amine binding GPRs and was shown important for agonist and antagonist binding in m3 muscarinic receptor (23). Also a histidine residue is present at homologous position in adenosine A1 receptor and mutation into leucine demonstrated its critical role in ligand recognition (25). Finally, position 308 in mDOR corresponds to that of a lysine residue in the opsin family, where it represents the attachment point of retinal, the covalently bound ligand of this receptor family. Thus, the OH group of Tyr 308 , which seems to be generally involved in opioid binding at the ␦ receptor, appears to play a role similar to that of homologous residues in other GPRs.
Role of the Aromatic Cluster Spanning Tms IV, V, and VI-Tryptophan residues 173 (Tm IV) and 274 (Tm VI) play a critical role in the binding of specific groups of opioid compounds. These amino acids are highly conserved among GPRs. Site-directed mutagenesis studies have shown their involvement in ligand recognition at m3 muscarinic receptor for both residues (26), as well as at bovine rhodopsin and angiotensin AT1 receptors for tryptophan residue in Tm VI (27,28). Our study shows that these two conserved tryptophan residues also play a role in binding at mDOR, consistent with the idea that these residues may represent common binding site determinants in GPRs despite the wide diversity of the ligands.
Mutations of phenylalanine residues of Tm V poorly affect ligand binding compared with other mutations examined in this study. Particularly, F218A represents the only mutation in this study that barely affects ligand binding. A phenylalanine residue is also present at this position in other GPRs. At present a role for this amino acid has not been documented, but the involvement of a serine residue, which is found at homologous position in receptors that bind small biogenic amines (Ser 204 in ␤2-adrenergic, Ser 199 in D1 dopamine, Ser 201 in ␣2-adrenergic), has been clearly demonstrated. The serine amino acid residue was suggested to interact with the catechol moiety of the ligand, both from three-dimensional modeling studies (4)  (Fig. 1B), and the side chains of mutated amino acid residues are colored. Each color represents the extend of affinity decrease due to the mutation. Results from the previously studied D128N mDOR mutant receptor (19) are also represented in the figure. and mutagenesis experiments where a 25-fold affinity decrease is described following serine to alanine substitution (29,30). Our results indicate that Phe 218 in the ␦ receptor does not participate to ligand binding to the same extent as the homologous serine residue in catecholamine receptors. Modification of Phe 222 into alanine alters binding potency of a larger number of ligands compared with F218A mutation. Taken together, results from mutations of the two phenylalanine residues in Tm V suggest that Phe 222 is located closer to the binding site than Phe 218 , which is consistent with our model (Fig. 1A). Furthermore the overall weak effect of both F218A and F222A mutations, relative to the important alterations of ligand binding arising from modifications at Tms III, IV, and VI, indicates that opioid ligands interact weakly with Tm V and bind more centrally relative to the putative binding pocket.
Aromatic Residues Are Part of a General Opioid Binding Domain-Our results seem to indicate a correlation between the potency of ligands to interact at mutated receptors and their selectivity degree toward the ␦ subtype. We have observed that affinity of some highly ␦-selective compounds (BW373U86, naltrindole, and TIPP) is weakly affected by our set of mutations while binding of compounds with low ␦ selectivity (bremazocine, naloxone, ␤-endorphin, and dynorphin A) is altered at most positions and to a larger extent (Fig. 4). This indicates that residues identified in this study constitute major anchor points for nonselective opioids, while additional specific interactions may take place for ␦-selective compounds in other regions of the receptor.
Our study does not provide evidence for the existence of distinct agonist and antagonist binding sites in the region embedded between Tms III and VII, that we have investigated. Each amino acid substitution affects the binding of at least one agonist and one antagonist, suggesting largely overlapping sites for both ligand types in the transmembrane aromatic domain.
Altogether these observations suggest that the set of aromatic residues proposed by our model is not involved in agonist/antagonist-or ␦ selectivity, but rather constitute a general binding domain for opioids. Consistent with this idea is the fact that all six aromatic amino acids, identified by three-dimensional modeling as potential key residues, are conserved across ␦-,and -opioid receptors. We might therefore expect these residues to play similar roles in and receptors, and further mutagenesis experiments may confirm this assumption. Finally, the suggestion that transmembrane aromatic residues examined here are part of a general binding domain raise the possibility that these residues interact with the ''message'' part of opioid molecules (22), which is common to all tested compounds. Further ligand-docking studies on a refined threedimensional model should provide precise indications in support of this hypothesis.
Distinct Binding Fingerprints for Each Opioid Ligand-We have used a schematic representation of the putative aromatic site proposed by three-dimensional modeling to illustrate effects of mutations for the binding of each ligand under investigation (Fig. 5). An interesting observation is that the point mutations, altogether, alter the binding of every ligand with a specific pattern. This allows to define a unique binding fingerprint for each ligand and indicates that every opioid compound interacts at the ␦ receptor using a distinct combination of interactions with aminoacid residues of the binding domain. Indeed, and although two of our mutations allow to group structurally related ligands (W137A and W274A), we clearly observe specific features which make each ligand-receptor interaction unique. These include the striking reduction of deltorphin II binding at Y129F mutant as compared with other ligands, the equal involvement of both tryptophan residues for DPDPE binding only, and the unique property of TIPP in binding at W274A mutant with increased affinity. In line with this idea also is the observation of different interaction modes for the two highly similar peptides Leu-and Met-enkephalin. Binding of Met-enkephalin is more reduced at the W173A mutant relative to Leu-enkephalin, while its affinity remains unchanged at Y308F, a mutation that significantly affects Leuenkephalin binding. Thus, our study demonstrates for the first time that there is no unique binding site for opioids.
Conclusion-Evidence has accumulated that aromatic residues may interact with various chemical groups and undergo intra-or intermolecular bonds. Aromatic-aromatic interactions have been suggested to be important in maintaining receptor structure and driving conformational changes in GPRs (31). Also, the existence of amine-aromatic interactions between a ligand and its receptor site has been postulated based on crystallographic data for both acetylcholinesterase (32) and the phosphotyrosine recognition SH2 domain of v-src oncogene (33), as well as from mutagenesis studies of the tachykinin NK1 receptor (34). Numerous aromatic residues are present in transmembrane domains of opioid receptors, some of them being conserved in all GPRs (35). Our study shows that the side chains of Tyr 129 , Trp 173 , Phe 222 , Trp 274 , and Tyr 308 residues provide favorable interaction sites for opioid compounds and that their contribution may be distinct from that of their homologous counterparts in other GPRs. Our results further suggest that these transmembrane aromatic residues, which are conserved across opioid receptor subtypes, may represent common anchor points for opioid ligands in the ␦ receptor. Finally, the overall picture of opioid binding at the mutated receptors supports the idea of a general transmembrane binding domain with no single binding site. Studies of complementary modifications of both receptor and ligand structures, in ␦ as well as in and receptors, may help in the future to elucidate the precise mechanisms of ligand-receptor interaction at aromatic residues in transmembrane domains of opioid receptors.