Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Opiates like morphine and endogenous opioid peptides exert their pharmacological and physiological effects through binding to their endogenous receptors, opioid receptors (Minami and Satoh, 1995). Three different types of opioid receptors (µ, , and ) have been demonstrated based on pharmacological, binding and anatomical and molecular data (Law and Loh, 1999). Molecular cloning experiments have confirmed the existence of the proposed opioid receptor types and formed the basis for a torrent of information on their structures and pharmacological properties (Kieffer, 1995).

The receptor proteins consist of about 400 amino acids and have the characteristic seven transmembrane domain structure of G-protein-coupled receptors (Knapp et al., 1995). The seven -helical transmembrane segments are thought to be arranged in a circular manner allowing the macromolecule to form a ligand binding cavity and exposing three intracellular loops and the carboxy terminus to the cytoplasmic milieu and three extracellular loops and the amino terminus to the outside environment (Akil et al., 1996; Fig. 1).

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Schematic drawing of the human µ-opioid receptor. The circle indicates the position of Ser329. TM are denoted by roman numerals.

Opioid receptors are activated by two main kinds of molecules. The first are the opioid peptides, which have structures based on an N terminus of Tyr-Gly-Gly-Phe... A family of twenty endogenous opioid receptor ligands were generated from three precursor proteins. A novel group of peptides has been discovered in the brain and named endomorphins. They are unique in comparison with other opioid peptides by atypical structure and high selectivity toward the µ-opioid receptor (Zadina et al., 1997). A second group of compounds comprises the opiates, the nonpeptide molecules like the morphinans, benzomorphans, phenylpiperidines, diphenylheptanes, and oripavines (Raynor et al., 1994).

Activation of opioid receptors produces a wide array of cellular responses like inhibition of adenylyl cyclase, inhibition of voltage-dependent calcium channels (N, P, Q, and R type), and activation of an inwardly rectifying potassium channel (Law and Loh, 1999). The µ-opioid receptor is of particular clinical and social importance since the more potent analgesic drugs, such as morphine, heroin, fentanyl, and methadone, elicit their beneficial pharmacological effect as well as their addictive liability through activation of the µ-receptor (Matthes et al., 1996).

A powerful approach in mapping the regions involved in drug selectivity is the construction of opioid chimeric receptors. Using receptor chimeras, several groups have reported that the docking sites for the opioid peptides and alkaloids are different (Xue et al., 1995; Meng et al., 1995; reviewed in Law et al., 1999).

Because of the lack of experimentally determined 3D structures, except for rhodopsin, diverse computational strategies are being explored to help bridge this gap for many specific G-protein-coupled receptors (Filizola et al., 1999). Accordingly, models of opioid receptors have been proposed recently. Different procedures like sequence divergence analysis and calculation of putative H-bonding residues combined with distance geometry calculations were used to construct these models (Strahs and Weinstein, 1997; Pogozheva et al., 1998; Filizola et al., 1999). As long as there are no solved 3D structures for opioid receptors, however, additional experimental data of site-directed mutagenesis experiments are necessary to improve the reliability of the complex models, leading to a further optimization of the receptor-ligand complex model (Kanematsu and Sagara, 2001).

Subramanian et al. (2000) revealed a novel binding site model for fentanyl at the µ-opioid receptor using a combination of conformational analysis and ligand docking of a series of fentanyl derivatives. This model suggests that the N-phenethyl of fentanyl group slides through TM-III and TM-VII toward the intracellular end of the cavity. A -hydroxyl substituent is thought to form a hydrogen bond with Ser³³¹.

The goal of this work was to examine the importance of Ser³²⁹ in the human µ-opioid receptor, thereby contributing to define the orientation of the phenethyl ring within the receptor cavity. For clarity, the Ser³²⁹ corresponds to the Ser³³¹ in the clone of Subramanian et al. (2000). Therefore, we mutated this Ser³²⁹ to Ala³²⁹ and determined the EC₅₀ values for GIRK1/GIRK2 channel activation through consecutive activation of wild-type and mutant opioid receptors coupled to G-proteins and expressed in Xenopus laevis oocytes.

To achieve this, we coexpressed GIRK1 and GIRK2 channels together with RGS4, a regulator of G-protein signaling (Kofuji et al., 1995; Doupnik et al., 1997). Coexpression of RGS4, the brain-expressed isoform of RGS proteins, reconstitutes the native gating kinetics by accelerating GIRK1/GIRK2 channel deactivation (Ulens et al., 2000a). This experimentally created model (Ulens et al., 2000b) provides a defined population of functionally active opioid receptor (hMORwt or hMORS329A) and an excellent tool for measuring the efficacy and potency of a ligand for a certain receptor (Pil and Tytgat, 2001).

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Subcloning and in Vitro Transcription of cDNA Clones Encoding GIRK1/2 Channels, Human µ-Opioid Receptors, and RGS4. Plasmids containing the entire coding sequence for the mouse GIRK1 and the mouse GIRK2 channel were subcloned into the vector pSP35T and pBScMXT, respectively, and designated as pSP/GIRK1 (Kobayashi et al., 1995) and pBScMXT/GIRK2 (Kofuji et al., 1995). The polylinker in each of these vectors is flanked by Xenopus globin 5'- and 3'-untranslated regions, resulting in an enhanced protein expression after injection of in vitro transcribed cRNA (Krieg and Melton, 1984). For in vitro transcription, plasmids were first linearized either with EcoRI (for pSP/GIRK1) or with SalI (for pBScMXT/GIRK2). Next the cRNAs were synthesized from the linearized plasmids using the large-scale SP6 mMessage mMachine (for pSP/GIRK1) or T3 mMessage mMachine (for pBScMXT/GIRK2) transcription kit (Ambion, Austin, TX).

Construction of Mutant Human µ-Opioid Receptors. Ser³²⁹ in hMOR was mutated to Ala³²⁹ using the Quickchange site-directed mutagenesis kit (Stratagene, La Jolla, CA). The primers were designed in such way that a silent MluI restriction site was introduced simultaneously: 5'-GCTCTAGGTTACACAAACGCGTGCCTCAACCCAGTCC-3' and 5'-GGACTGGGTTGAGGCACGCGTTTGTGTAACCTAGAGC-3' (codon and complementary codon are underlined; the palyndromic sequence is in bold). Cycling parameters were set according to the manufacture's guidelines. The cDNAs from eight single colonies were digested with MluI to identify possible mutants. All eight clones contained the MluI restriction site, which was introduced by the mutant primers. A 313-base pair fragment containing the desired mutation was isolated by a double restriction digest with NsiI and BglII. The mutant cDNA was then loaded on an agarose gel, the fragment of interest was cut out, gene cleaned (QIAQUICK; QIAGEN, Valencia, CA), and ligated with T4 DNA ligase (Promega, Madison, WI) into the corresponding sites of the hMORwt/pGEMHE. The same mutant fragment was subcloned into pGEM7Zf(+) (Promega) for DNA sequencing (Eurogentec, Seraing, Belgium). For in vitro transcription, the mutant hMORS329A/pGEMHE was linearized with NheI. Next, the capped cRNAs were synthesized from the linearized plasmids using the large-scale T7 mMessage mMachine transcription kit (Ambion).

Experimental Model. X. laevis oocytes were prepared for injection as described (Liman et al., 1992). Oocytes were coinjected with 0.5 ng 50 nl¹ GIRK1, 0.5 ng 50 nl¹ GIRK2, and 10 ng 50 nl¹ RGS4 cRNA, with the addition of 10 ng 50 nl¹ of either hMOR or hMORS329A cRNA. Injected oocytes were maintained in ND-96 solution (composition: 2 mM KCl, 96 mM NaCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES, pH 7.5) supplemented with 50 µg/ml gentamicin sulfate and incubated at 16°C.

Electrophysiological Recordings. Whole-cell currents from oocytes were recorded 1 day after injection using the two-microelectrode voltage-clamp technique (Geneclamp 500; Axon Instruments, Inc., Union City, CA). Resistances of voltage and current electrodes were kept as low as possible (approximately 200 k) and were filled with 3 M KCl. To eliminate the effect of voltage drop across the bath-grounding electrode, the bath potential was actively controlled. All experiments were performed at room temperature (19-23°C). At the start and end of each experiment, oocytes were superfused with low-potassium (ND-96) solution (composition: 2 mM KCl, 96 mM NaCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES, pH 7.5). During application of increasing concentrations of ligands, oocytes were superfused with high-potassium (HK) solution (composition: 96 mM KCl, 2 mM NaCl, 1 mM MgCl₂, 1.8 mM CaCl₂, 5 mM HEPES, pH 7.5). In HK solution, the K⁺ equilibrium potential is close to 0 mV and enables K⁺ inward currents to flow through inwardly rectifying K⁺ channels at negative holding potentials. A gravity-controlled fast perfusion system (Warner Instrument, Hamden, CT) was used to ensure rapid solution exchanges. Application of opioid ligands did not evoke an increase of the conductance in uninjected oocytes (n = 30). In each experiment, oocytes were clamped at a holding potential of 70 mV for approximately 10 min and superfused with ND-96 solution. Next, the superfusion was switched from ND-96 to HK solution, after which increasing concentrations of an opioid receptor agonist were applied. Each concentration was applied for as long as needed to achieve a steady state GIRK1/GIRK2 current activation. Each ligand concentration was washed out by superfusing with HK solution. During this washout period, the channels return to the control current level as a result of deactivation process that is dramatically accelerated in the presence of RGS4, as described (Ulens et al., 2000a). At the end of each experiment, the oocyte was superfused with HK solution containing 300 µM BaCl₂, causing block of the net GIRK1/GIRK2-gated inward current. Finally, the superfusion was switched back to ND-96 solution to confirm complete reversibility. To avoid that the receptor expression level affects the EC₅₀ values of the investigated agonists in the study, the expression system was standardized as previously described (Ulens et al., 2000b).

Data Analysis. The pCLAMP program was used for data acquisition and data files (Axon Instruments) were imported in Microsoft Excel (Redmond, WA). The percentage of activated current was calculated using the equation below.
and 0% was taken as the control current level. Current percentages were used for the calculation of the EC₅₀ value using the Hill equation.
where I represents the current percentage, I_max the maximal current percentage, EC₅₀ the concentration of the agonist that evokes the half-maximal response, A the concentration of agonist, and n_H the Hill coefficient. Averaged data are indicated as the mean ± S.E.M. and were calculated using n experiments, where n indicates the number of oocytes tested. Statistical analysis of differences between groups was carried out with a Student's t test, and a probability of 0.05 was taken as the level of statistical significance.

Compounds. DAMGO (Sigma-Aldrich, St. Louis, MO), fentanyl HCl [kindly provided by National Institute on Drug Abuse (NIDA), Bethesda, MD), morphine HCl (Federa, Belgium), R004333 (racemic mixture) (Janssens Pharmaceutica, Antwerp, Belgium) were dissolved in HK solution, stored at 4°C, and diluted appropriately for the experiments.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Each receptor (wild-type or mutant) was individually coexpressed with GIRK1/GIRK2 channels and RGS4, mimicking the native neuronal G-protein-mediated pathway of K⁺ channel activation. We used the two-microelectrode voltage-clamp technique to measure the opioid receptor-activated GIRK1/GIRK2 channel response as the increase of the inward K⁺ current at 70mV, evoked by the application of increasing concentrations of opioid ligands. In our study, we examined the potency of DAMGO, morphine, fentanyl, and -hydroxyfentanyl (Fig. 2) on hMORwt and on the mutant receptor hMORS329A.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   Chemical structures of DAMGO, morphine, fentanyl, and -hydroxyfentanyl.

Figure 3 shows representative current traces of agonist-gated currents evoked from oocytes expressing either hMORwt (Fig. 3A) or hMORS329A (Fig. 3B) by DAMGO. Analogously, Figs. 4 and 5 show representative current traces of currents evoked by fentanyl and -hydroxyfentanyl, respectively. Current traces evoked by morphine are not shown. Concentration-response relationships (Fig. 6, A-D) are shown for DAMGO, morphine, fentanyl, and -hydroxyfentanyl, respectively.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Representative current traces evoked from X. laevis oocytes coexpressing GIRK1/GIRK2 channels and RGS4 with hMORwt (A) or hMORS329A (B). Agonist gated currents were evoked at 70mV by application of increasing concentrations of DAMGO.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Representative current traces evoked from X. laevis oocytes coexpressing GIRK1/GIRK2 channels and RGS4 with hMORwt (A) or hMORS329A (B). Agonist gated currents were evoked at 70mV by application of increasing concentrations of fentanyl.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Representative current traces evoked from X. laevis oocytes coexpressing GIRK1/GIRK2 channels and RGS4 with hMORwt (A) or hMORS329A (B). Agonist gated currents were evoked at 70mV by application of increasing concentrations of -hydroxyfentanyl.

View larger version (19K): [in this window] [in a new window]   Fig. 6.   Concentration-response curves for GIRK1/GIRK2 channel activation by increasing concentrations of DAMGO (A), morphine (B), fentanyl (C), and -hydroxyfentanyl (D). Agonist-gated currents were evoked from X. laevis oocytes coexpressing GIRK1/GIRK2 and RGS4 with hMORwt () or hMORS329A (). The agonist-gated increase of GIRK current at each concentration was normalized to a maximal response of 100%. Each point represents the average current activation evoked from five to eight oocytes (mean ± S.E.M.).

Table 1 summarizes EC₅₀ values calculated for these agonists via the receptors and also shows the structure of the amino acid at position 329. Averaged data are indicated as the mean ± S.E.M. and were calculated using five to eight experiments. DAMGO, a µ-opioid-selective peptidic agonist was more than 4 times less potent via the mutant hMORS329A receptor (EC₅₀ value: 111.1 ± 22.3 nM) compared with the wild-type receptor (EC₅₀ value: 25.6 ± 5.0 nM). Morphine showed a 5-fold increase in potency for the mutant receptor (EC₅₀ value: 28.0 ± 6.1 nM) than for the wild-type receptor (EC₅₀ value: 148.6 ± 16.3 nM). The EC₅₀ value of fentanyl on the mutant receptor (9.3 ± 0.3 nM) was more than 7 times lower compared with the EC₅₀ value of fentanyl on the wild-type receptor (64.9 ± 7.8 nM). On the wild-type receptor, hydroxyfentanyl (EC₅₀ value: 38.1 ± 2.1 nM) was more potent than fentanyl (EC₅₀ value: 64.9 ± 7.8 nM), whereas fentanyl activates GIRK1/GIRK2 channels with a significantly higher potency than hydroxyfentanyl on the hMORS329A mutant receptor (EC₅₀ values: 9.3 ± 0.3 nM for fentanyl and 24.5 ± 0.5 nM for hydroxyfentanyl).

The EC₅₀ value of hydroxyfentanyl on the mutant receptor was only slightly, albeit significantly, lower compared with the EC₅₀ value on the wild-type receptor. The EC₅₀ values for DAMGO and hydroxyfentanyl are significantly different between the wild-type and the mutant receptor.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The serine 329 residue is invariant across the µ-, -, and -opioid receptors and lies in a region of high amino acid homology. The corresponding positions of this serine in the human - and -opioid receptor are position 323 and 311, respectively (Pogozheva et al., 1998). This amino acid is located in the middle of the seventh transmembrane helix and is surrounded by other hydrophilic residues. The alanine-mutant was preferred because alanine possesses no hydrogen-bonding properties and for reasons of similarity in length of the side chain (Table 1).

All opioid receptor models have a ligand-binding cavity that is partially covered by the extracellular loops. Several groups have reported the participation of these loops in the binding of opioid peptides, like DAMGO (Minami et al., 1995; Onogi et al., 1995). Mutational studies showed the importance of Lys³⁰³, Val³¹⁶, Trp³¹⁸, and His³¹⁹ around the third extracellular loop of the µ-opioid receptor to discriminate between µ- and -opioid receptors (Seki et al., 1998; Ulens et al., 2000b). When these four residues are engineered into the -opioid receptor, the resultant -opioid receptor mutant can bind DAMGO with high affinity (Seki et al., 1998). Mutation of the hydrophilic Asn²³⁰ in the second extracellular loop of the µ-opioid receptor did not alter the potency of DAMGO (Pil and Tytgat, 2001).

DAMGO and other peptides not only interact at the extracellular surface but also extend within the receptor to interact deep within the receptor cavity. Similar effects have been observed with other peptide receptors (Krystek et al., 1994). Site-directed mutagenesis of Tyr³²⁶ to Phe³²⁶ gave rise to a 20-fold decrease in K_i value of DAMGO (Mansour et al., 1997). Mutation of Asn³³² to Asp³³² even eliminated detectable binding of radiolabeled DAMGO, but an additional mutation of Asp¹¹⁴ to Asn¹¹⁴ could partially restore the affinity, indicating a structural relation between these two residues (Xu et al., 1999b). The Ser³²⁹ mutation we made is located in the middle of these (Tyr³²⁶ and Asp³³²) mutations. The mutation of Ser³²⁹ to Ala³²⁹ caused a more than 4-fold decrease in potency for DAMGO, suggesting the existence of hydrophilic interactions between DAMGO and the serine residue. This result provides additional evidence that DAMGO forms hydrophilic contacts deep within the binding cavity of the seventh transmembrane helix. The carboxyl terminal hydroxyl group is a possible candidate to form a hydrogen bond with Ser³²⁹. Little is known, however, about the docking sites of DAMGO in the µ-opioid receptor as nonpeptidic ligands form the main focus of receptor models (Kanematsu and Sagara, 2001; Filizola et al., 1999). Chimeric and mutational data (reviewed in Law et al., 1999) only point out the differential binding domains for peptide and nonpeptide ligands.

The potency of morphine was increased 5-fold by mutating the wild-type hMOR to hMORS329A. This result is indicative for strong hydrophobic interactions between the mutated receptor and the small alkaloid agonist. Assembling mutagenesis data together with different opioid receptor binding pocket models can locate the site of interaction. Morphine and fentanyl, both nonpeptidic agonists, possess both a protonatable nitrogen and an aromatic group that are commonly thought to mimic the N-terminal tyrosyl moiety of opioid peptides. All receptor models agree that Asp¹⁴⁷ in the third transmembrane helix is the key anchor point of binding morphine and fentanyl to the µ-opioid receptor by ion pairing between the carboxylate group of Asp¹⁴⁷ and the ammonium ion moiety of the agonist (Habibi-Nezhad et al., 1996). Site directed mutagenesis confirmed the importance of this conserved residue (Li et al., 1999). The protonated nitrogen of the opioid ligand may also gain stabilization through cation  interactions with the neighboring Tyr residue, Tyr¹⁴⁸ (Xu et al., 1999a; Kanematsu and Sagara, 2001; Mo et al., 2002). Several groups point out that the 3-hydroxyl group of the morphine tyramine moiety forms an H-bond with His²⁹⁷ in the sixth transmembrane helix (Spivak et al., 1997; Pogozheva et al., 1998). Although this para-hydroxyl substituent contributes to high affinity binding, McFadyen et al. (2000) showed that is not critical.

The aliphatic 6-hydroxyl group was first thought to interact with Tyr³²⁶ (Mansour et al., 1997) or with Asn²³⁰ (Pogozheva et al., 1998). Our group showed that the potency of morphine increased when this Asn²³⁰ residue was replaced by a hydrophobic amino acid (Pil and Tytgat, 2001). We concur with a recent alignment that places the 6-hydroxyl group of morphine outside the binding pocket and suggests that the hexene ring of morphine is surrounded with Lys²³³, Tyr¹⁴⁸, Asn²³⁰, and Trp³¹⁸ (Chen et al., 1996; McFadyen et al., 2000). This last mentioned residue plays an important role in the selectivity of several drugs for the µ-opioid receptor (Xu et al., 1999a; Ulens et al., 2000b).

According to Pogozheva et al. (1998), Ser³²⁹, together with Asn⁸⁶, Asp¹¹⁴, Ser¹⁵⁴, Asn³²⁸, and Asn³³², is a part of a highly conserved polar cluster. Surprisingly, mutation of this Ser³²⁹ to Alanine, thereby disturbing this putative cluster, can enhance the potency of morphine toward the mutant receptor. In a recent model, Kanematsu and Sagara (2001) found that when the quaternary ammonium head group was placed to form electrostatic interactions with Asp¹⁴⁷, the phenyl ring of morphine was located close to the phenyl ring of Tyr³²⁵ (this residue corresponds to Tyr³²⁶ in our clone) in TM-VII. We conclude that additional hydrophobic contacts between Ala³²⁹ in the mutant receptor and the phenyl ring of morphine can explain the increase in potency of morphine via the hMORS329A receptor (Fig. 7A).

View larger version (40K): [in this window] [in a new window]   Fig. 7.   Visualization of the 3D pharmacophore of morphine (A) and fentanyl (B) in the wild-type receptor. Morphine and fentanyl are shown in yellow. The transmembrane helices are denoted by roman numerals. The side chains of Asp147 and Ser329 are represented in ball-and-stick form with oxygen, carbon, and nitrogen colored in red, black, and blue, respectively. A, the quaternary nitrogen of morphine (shown in blue) interacts with Asp147, whereas the phenyl ring lies close to Ser329. B, the piperidine nitrogen of fentanyl (colored in blue) forms electrostatic interactions with Asp147, whereas the N-phenethyl group of fentanyl extends within the binding cavity to Ser329. Molecular graphics were generated with MOLSCRIPT.

The potency of fentanyl was increased 7-fold by mutating the wild-type receptor to hMORS329A. As compared with morphine, we observe a similar change in EC₅₀ value, indicating a possible similar hydrophobic effect with a similar structural feature. According to the model of Pogozheva et al. (1998), the conformation with the phenyl ring of the phenethyl fragment oriented toward the extracellular surface is the most reasonable choice. In contrast with this model, a novel study examining the binding mode of a series of fentanyl derivatives places the N-phenylpropanamide group toward the extracellular side of the cavity (between TM-III and TM-VI), whereas the N-phenylethyl group slides through TM-III and TM-VII toward the intracellular end of the cavity (Fig. 7B). The residues close to the phenylethyl moiety in TM-VII are Asn³³⁰, Ser³³¹, Asn³³⁴, and Pro³³⁵ (Subramanian et al., 2000). These residues correspond to Asn³²⁸, Ser³²⁹, Asn³³², and Pro³³³, respectively, in our human receptor. The mutation of Ser³²⁹ to a lipophilic Ala³²⁹ can create a more favorable vicinity for the N-phenethyl group of fentanyl to interact deep within the binding cavity, thereby clarifying the increased potency of fentanyl on hMORS329A.

Surprisingly, a rather small, albeit significantly, increase in potency was observed with -hydroxyfentanyl on HMORS329A compared with HMORwt. We expected a decreased potency on the assumption that Ser³²⁹ serves as a hydrogen-bonding partner with the -hydroxyl group of the hydroxyfentanyl derivatives, as suggested by Subramanian et al. (2000). It is unlikely that Ser³²⁹ forms a hydrogen bond with this hydroxyl group, but an interaction with another part of the N-phenethyl groups remains possible as Ala³²⁹ exert a favorable influence. Tang et al. (1996) hypothesized that Tyr¹⁴⁸ contributes to the binding of ohmefentanyl via hydrogen bonding with the -OH group. Mutating this residue to Alanine, however, did not show a strong effect on the binding of ohmefentanyl (Xu et al., 1999a). Other possible hydrogen bonding partners thought to be close to the -hydroxyl group are Thr¹²⁰, Asn¹⁵⁰, and Tyr³²⁶ (Subramanian et al., 2000). Mutation of Tyr³²⁶ to Phe³²⁶ resulted in a decreased affinity for morphine, DAMGO, and fentanyl, but no fentanyl derivatives were tested (Mansour et al., 1997). Further mutagenesis studies are needed to elucidate the precise hydrogen-bonding partner.

-Hydroxyfentanyl activates GIRK1/GIRK2 channels through the wild-type µ-opioid receptor with a significantly higher potency than fentanyl. Thus, introducing the hydroxyl group enhances the potency of fentanyl, providing additive evidence for a hydrophilic interaction between the receptor and this hydroxyl group. Remarkably however, on the hMORS329A receptor fentanyl is more potent compared with -hydroxyfentanyl. We hypothesize that -hydroxyfentanyl forms a hydrogen bond with the receptor in such a way that the phenyl ring does not extend to the same extent as fentanyl within the binding cavity. This model can explain the observation of the smaller increase in potency for -hydroxyfentanyl on the mutated receptor.

In conclusion, the present studies demonstrate the dual role of Ser³²⁹ in the potency of various agonists. The potency of DAMGO, a peptide agonist, decreased, whereas the potency of nonpeptide agonists, like morphine, fentanyl, and -hydroxyfentanyl, increased when this residue was mutated to alanine. These results shed also new light on the orientation of the phenethyl ring of fentanyl within the receptor cavity and on previous models of ligand interaction.