INTRODUCTION

Opiates, opioid compounds, and endogenous opioid peptides act on membrane-bound receptors to produce their effects. The existence of at least three types of opioid receptors, µ, , and , has been demonstrated (1). Mu opioid receptors mediate analgesic effects of many opiates and opioids. Opiates (such as morphine and heroin) are among the most widely abused drugs, mainly because of their actions on the µ opioid receptor to produce euphoria. Development of tolerance and dependence to opiates is mediated largely by the µ receptor. Activation of µ opioid receptors couples via pertussis toxin-sensitive G proteins to various effectors including adenylate cyclase and K⁺ and Ca²⁺ channels (1). Following the cloning of the mouse  opioid receptor (2, 3), several laboratories reported cloning of the µ opioid receptor (4, 5, 6, 7, 8, 9, 10, 11). Hydropathy analysis of deduced amino acid sequences of these clones indicates the presence of seven putative transmembrane domains (TMDs)¹ separated by intra- and extracellular loops, characteristics of G protein-coupled receptors. Availability of µ opioid receptor cDNA clones makes it possible to examine ligand-receptor interactions at the molecular level.

Affinity ligands that form covalent bonds with receptors have been very useful in the elucidation of receptor structure. Specific incorporation of radiolabeled affinity ligand into the receptor followed by SDS-PAGE and fluorography or autoradiography has been used to determine molecular mass of the receptor without prior purification and to examine the nature of carbohydrate moieties (for example, see Liu-Chen et al. (12)). Because of the covalent nature of the bond between the ligand and the receptor, the site of incorporation, thus part of the binding domain, can be precisely determined by peptide mapping and/or determination of amino acid sequences of labeled fragments. For example, [³H]propylbenzilylcholine mustard binds covalently to the m₁ muscarinic receptor. By mapping of [³H]propylbenzilylcholine mustard-labeled receptor fragments and subsequent peptide sequence determination, Kurtenbach et al. (13) identified a conserved Asp in the third TMD to be the residue that formed a covalent bond with the ligand and, thus, to be part of the binding pocket.

-Funaltrexamine (-FNA) (Fig. 1), a fumaramate methyl ester of naltrexamine, was synthesized by Portoghese et al. (14). It was found to have reversible  agonist and irreversible µ antagonist activities in vivo and in vitro (for review, see Ref. 15). Binding of -FNA to opioid receptors in tissue membrane preparations in vitro has also been characterized (16, 17, 18, 19). -FNA binds to brain µ,  and  receptors with IC₅₀ values of 2.2, 14, and 78 n, respectively (16). There is a general agreement in the literature that -FNA binds reversibly, and not irreversibly, to  opioid receptors. -FNA binds irreversibly to µ opioid receptors (16, 17) and at low concentrations (1-10 n) [³H]-FNA covalently labels only µ opioid binding sites with high specificity (20, 21, 22). [³H]-FNA-labeled µ opioid receptors in the brain have broad molecular weight ranges, indicative of glycoprotein nature. Molecular mass values of labeled µ receptors vary among species due to different degrees of glycosylation and the deglycosylated receptor has a molecular mass of ~40 kDa (12).

Chemical structure of -funaltrexamine.

Recently we reported that [³H]-FNA bound irreversibly to the cloned µ, but not , opioid receptor (23). [³H]-FNA-labeled receptors migrated as one broad and diffuse band with a molecular mass of 80 kDa and, upon removal of N-linked carbohydrates, became a sharper band with a molecular mass of ~40 kDa (23). It was thought that -FNA reacts with a Cys residue in the vicinity of the binding site of µ receptors to form a covalent bond (14).

In this study, we localized the region of the µ receptor involved in the covalent binding of [³H]-FNA by cleavage of partially purified labeled receptor with CNBr and determination of sizes of the labeled fragments. We then determined the amino acid residue that formed covalent bond with -FNA by site-directed mutagenesis studies.

EXPERIMENTAL PROCEDURES

Stable Expression of the µ Receptor in CHO Cells

CHO cell lines stably expressing the cloned rat µ receptor were established previously (23), and membranes were prepared for receptor binding and labeling (24).

Transient Expression of the µ Receptor in COS-1 Cells

cDNAs of wild type and mutant rat µ receptors were transfected into COS-1 cells with DEAE-dextran-chloroquin method as described previously (25). Cells were harvested, and membranes were prepared as described elsewhere (24).

Labeling of µ Opioid Receptors with [³H]-FNA

Cell membranes suspended in 50 m Tris-HCl buffer, 1 m EGTA and 4 µ leupeptin (TEL buffer, pH 7.5) were incubated with 5 n [³H]-FNA (unless indicated otherwise) in the presence of 200 m NaCl at 37 °C for 75 min as described previously (23). Naloxone (10 µ) was used to define nonspecific binding.

Irreversible Binding of [³H]-FNA

Assay for irreversible binding of [³H]-FNA was performed according to a modification of our published method (12, 21). Briefly, [³H]-FNA-labeled membranes or solubilized labeled preparations (in 1 ml) were precipitated with 50% trichloroacetic acid (final concentration, 10%) on ice for 15 min. One ml of ice-cold 10% trichloroacetic acid was added, sonicated, allowed to stand on ice for at least 15 min, and filtered with GF/B filters under reduced pressure. Filters were washed three times with 5 ml of ice-cold 1% trichloroacetic acid. Radioactivity on the filter was determined by scintillation counting.

Reversible Binding of [³H]-FNA

Reversible binding of [³H]-FNA was determined as the difference between membrane binding and irreversible binding as we described previously (21). For membrane binding, following incubation with [³H]-FNA, the reaction mixture was filtered immediately over GF/B filters under reduced pressure, followed by three 5-ml washes with ice-cold 50 m Tris-HCl buffer (pH 7.5). Radioactivity on the filter was determined by scintillation counting.

Partial Purification of the Labeled Receptor with Antibody Affinity Chromatography

A peptide (µC peptide) was custom synthesized by the University of Pennsylvania Protein Chemistry Laboratory with the sequence CTNHQLENLEAETAPLP, which corresponds to the last 16 amino acids of the C-terminal domain of the µ opioid receptor with an added cysteine residue for conjugation to KLH or BSA (26). KLH or BSA was activated with m-maleimidobenzoyl-N-hydroxysuccinimide ester and passed through a G-25 Sephadex column. Activated KLH or BSA collected was incubated with µC peptide. Reaction mixtures were then used as the antigen.

Two female New Zealand White rabbits (3-3.5 kg) were immunized with the µC peptide according to standard protocols (26). The peptide-KLH conjugate was used in the primary injection, and the peptide-BSA conjugate was used in booster injections. Antiserum was collected 10-14 days after each booster injection.

The µC peptide was conjugated to Affi-Gel-15 for generation of µC-Affi-Gel-15 affinity gel according to the manufacture's instructions. The antiserum generated against the µC peptide was diluted with 1 volume of 10 m Tris-HCl-buffered saline (pH 7.5) and then passed through a 1-ml µC-Affi-Gel-15 column. After extensive washing, adsorbed antibodies were eluted with 0.1  glycine-HCl, 10% ethylene glycol, pH 2.5, and immediately neutralized with 1  Tris. Eluted antibodies were precipitated with ammonium sulfate, desalted over a Sephadex G-25 column equilibrated in 0.1  MOPS buffer (pH 7.5), and concentrated. Recovery of immunoactivity was monitored by immunoprecipitation of [³H]-FNA-labeled µ receptors before and after the purification process with the method of Luthin et al. (27).

Purified antibodies in 2 ml of 0.1  MOPS buffer (pH 7.5) were mixed with 1 ml Affi-Gel-10 at 4 °C for 30 min and then at room temperature for 30 min. Ethanolamine (200 µl, 1 ) was then added to block remaining active sites on the gel matrix. One-half ml of this anti-µC gel was estimated to have a capacity to adsorb 100 pmol of [³H]-FNA-labeled µ receptor based on several purification experiments.

[³H]-FNA-labeled µ receptors in CHO cell membranes (100-200 pmol) were solubilized in 3 ml of 2% Triton X-100, 50 m Tris-HCl, 0.9% NaCl, 1 m EDTA, 10 µg/ml each of leupeptin, soybean trypsin inhibitor, aprotinin, and pepstatin, pH 7.5. After being stirred at 4 °C for 2 h, the solubilized mixture was centrifuged at 100,000 × g for 1 h at 4 °C. The supernatant was diluted five times with Tris-HCl-buffered saline (pH 7.5) and filtered with a 0.22-µm Millex-GS filter. The filtrate was loaded at 4 ml/h at room temperature onto a 0.5-ml anti-µC Affi-Gel-10 column, which was equilibrated with Tris-HCl-buffered saline (pH 7.5) containing 0.2% Triton X-100. The column was then washed with 30 ml 0.2% Triton X-100, 1  NaCl in 50 m Tris-HCl, pH 7.5, followed by 10 ml of distilled water. Two ml of 0.5% trifluoroacetic acid, 0.5% Triton X-100 were used to elute the adsorbed receptor at flow rate of 4 ml/h. The eluate was immediately neutralized with 1  Tris base and stored at 70 °C.

Treatment of [³H]-FNA-labeled µ Opioid Receptor with CNBr

The deduced amino acid sequence of the rat µ receptor was analyzed with the PeptideSort program of Sequence Analysis Software Package of Genetics Computer Group, Inc. (Madison, WI) for molecular masses of fragments that could be generated theoretically by CNBr cleavage.

Some aliquots of immunoaffinity-purified preparations were reduced with mercaptoethanol and S-alkylated with 4-vinylpyridine to convert methionine sulfoxide to methionine and break disulfide bonds according to the method described by Fullmer (28). Immunoaffinity-purified receptor (~17 pmol in 400 µl) was adjusted to 1% Nonidet P-40, 0.1  ammonium carbonate, pH 7.8, and 10 µl of 2-mercaptoethanol were added and incubated for 2 h at 37 °C. Thirty µl of 4-vinylpyridine was then added and incubated at room temperature in dark for 2 h.

Partially purified labeled receptor preparations (~0.5 µg protein) with or without prior reduction and S-alkylation were adjusted to 70% formic acid in a volume of 250-500 µl, and CNBr was added (0.02-4 mg/ml) and reacted at room temperature in the dark for 24 h (29). Formic acid was removed under nitrogen. Reaction mixtures were precipitated by 10 volumes of ice-cold acetone and washed with ice-cold acetone. Precipitates were incubated with 2 × Laemmli sample buffer (30) with or without 0.2  dithiothrietol (DTT) at room temperature for 1 h for SDS-PAGE.

SDS-PAGE and Detection of Radioactivity in the Gel by Fluorography

SDS-PAGE was performed with 15.5% gel using the method of Schagger and von Jagow (31) as described previously (12) with ¹⁴C-labeled protein and peptide molecular weight standards. Fluorography was conducted according to the method of Bonner and Laskey (32) as described elsewhere (12).

Oligodeoxynucleotide-directed Mutagenesis

Mutations were introduced into the rat µ receptor (4) with the uracil replacement method of Kunkel et al. (33) using the Muta-Gene kit. Mutants were selected by DNA sequence determination with the dideoxynucleotide termination method of Sanger et al. (34). Mutant µ receptors were subcloned into the HindIII site of the mammalian expression vector pcDNA3.

Opioid Receptor Binding

Opioid receptor binding was performed with [³H]diprenorphine according to our published procedures (24). Saturation experiments were performed with a series of concentrations of [³H]diprenorphine (ranging from 0.05 n to 5 n). Inhibition of [³H]diprenorphine binding by -FNA was performed in the presence of 100 m NaCl with [³H]diprenorphine at a concentration close to its K_d for each receptor and receptor concentrations ±0.1 of K_d. Naloxone (10 µ) was used to define nonspecific binding. Binding data were analyzed with the EBDA program (35).

Protein Determination

Protein contents of membranes and solubilized preparations were determined by the bicinchoninic acid method of Smith et al. (36). Proteins in immunoaffinity-purified materials were quantified with the colloidal gold method of Stoscheck (37).

Materials

[³H]-FNA (2 batches with specific activities of 15.4 and 19.2 Ci/mmol, respectively) and -FNA were supplied by the National Institute on Drug Abuse. Naloxone HCl was provided by DuPont-Merck Co. [³H]Diprenorphine (35 Ci/mmol) was purchased from Amersham Corp.; ¹⁴C-labeled protein and peptide standards were from Life Technologies, Inc.; m-maleimidobenzoyl-N-hydroxysuccinimide ester, 4-vinylpyridine, and CNBr from Sigma; Affi-Gel-10, Affi-Gel-15, and Muta-Gene kit were from Bio-Rad; Triton X-100 was from Boehringer Mannheim; Nonidet P-40 (Surfact-Amps^TM P-40) and bicinchoninic acid protein assay reagents were from Pierce; the vector pcDNA3 was from Invitrogen (San Diego, CA); and Protein-Gold protein assay reagent was from Integrated Separation System (Natick, MA).

RESULTS

Immunoaffinity Purification of [³H]-FNA-labeled µ Receptors

In order to obtain more definitive information from peptide mapping studies, we performed the partial purification of [³H]-FNA-labeled µ receptors from solubilized labeled CHO cell membrane preparations by immunoaffinity chromatography.

Protein contents and specific [³H]-FNA labeling of the µ receptor were determined in labeled CHO cell membrane preparations and in immunoaffinity-purified preparations. CHO cell membranes and immunoaffinity-purified preparations contained ~0.6 and ~2500 fmol of [³H]-FNA-labeled µ receptors/µg of protein, respectively (Table I). Thus immunoaffinity chromatography purified approximately 4000-fold. The yield was approximately 60%. Calculated based on the molecular mass of the µ opioid receptor protein (44 kDa) (4), [³H]-FNA-labeled µ receptors represent ~11.0% of proteins following purification (Table I).

Table I. Purification of [3H]-FNA-labeled µ opioid receptors by immunoaffinity chromatography One hundred to 200 pmol of the rat µ receptor expressed in CHO cell membranes were labeled with [3H]-FNA, solubilized, and purified as described under ``Experimental Procedures.'' [3H]-FNA-labeled µ receptors Purity fmol/µg protein % CHO cell membranes 0.63  ± 0.07a 0.0028  ± 0.0003 Immunoaffinity-purified µ receptors 2503  ± 391 11.1  ± 1.7 a  Data are expressed as mean ± standard error of the mean (n = 3).

Cleavage of [³H]-FNA-labeled µ Receptors with CNBr

CNBr cleaves at carboxyl side of methionine (29). Based on the deduced amino acid sequence of the rat µ receptor, CNBr cleavage will theoretically generate fragments of the following mass (in Da, from the N to the C terminus): 149, 6320, 747, 1945, 1172, 3405, 2434, 1163, 4668, 296, 4546, 1335, 1115, 2060, and 13379 (Fig. 2).

When Immunoaffinity-purified [³H]-FNA-labeled µ receptors were cleaved with CNBr, and the reaction mixture was incubated with a sample buffer containing DTT before SDS-PAGE, a major labeled fragment of 4.5 kDa and a minor labeled band of 7 kDa were detected (Fig. 3A, lane 2). Prior reduction with mercaptoethanol and S-alkylation with 4-vinylpyridine, which reduced methionine sulfoxide to methionine and disrupted disulfide bonds, yielded a single 4.5-kDa labeled fragment and improved CNBr cleavage efficiency (Fig. 3A, lane 3 versus lane 2). CNBr experiments were performed with different concentrations of CNBr (0.02, 0.1, 0.2, 1, 2, 3, and 4 mg/ml in 0.25-0.5 ml) for ~0.5 µg of protein. With 7-day film exposure, at 0.02 and 0.1 mg/ml, there was no visible labeled band beside the receptor, and at 0.2 mg/ml, there was a faint labeled band of 4.5 kDa (not shown). At 1-4 mg/ml, the intensity of 4.5-kDa labeled band was increased (2 and 4 mg/ml shown in Fig. 3A). At 4 mg/ml, there was no evidence of any smaller labeled bands (Fig. 3A, lane 4) compared with treatment with 1 or 2 mg/ml (Fig. 3, A, lane 3, and B, lane 2). These results strongly suggest that CNBr cleavage was carried out to completion.

H]-FNA-labeled µ receptor treated with CNBr.

The size of CNBr-generated labeled fragments suggests that the site is within Ser¹⁶²-Met²⁰³ (calculated mass, 4,668 Da) or Ala²⁰⁶-Met²⁴³ (calculated mass 4,546 Da) (Fig. 2; also see Fig. 6). Ser¹⁶²-Met²⁰³ corresponds to the second intracellular loop and TMD 4. Ala²⁰⁶-Met²⁴³ corresponds to the second extracellular loop and the N-terminal half of the TMD 5. 

H]-FNA as determined in this study.

Since a disulfide bond was postulated between the first and second extracellular loops, we used a CNBr treatment of nonreduced and nonalkylated samples with or without post-treatment DTT reduction to differentiate the two possibilities. With DTT in the SDS-PAGE sample buffer, CNBr cleavage of [³H]-FNA-labeled µ receptors generated one major labeled fragment of 4.5 kDa and a minor labeled band of 7 kDa (Fig. 3, A, lane 2, and B, lane 2). Without DTT in the sample buffer, only a 7-kDa labeled fragment was detected (Fig. 3B, lane 3). These results are consistent with the notion that the labeled fragment of Ala²⁰⁶Met²⁴³ (calculated mass, 4,546 Da) forms a disulfide bond with the unlabeled fragment of Gly¹³¹-Met¹⁵¹ (calculated mass, 2,434 Da) (Fig. 2; also see Fig. 6) for a combined mass of ~7 kDa.

In Fig. 3, A and B, control labeled receptors did not enter the 15.5% gel. However, when 10% gel was used, these receptors had an identical molecular mass as labeled receptors that were not exposed to 70% formic acid (not shown).

Determination of the Amino Acid Involved in [³H]-FNA Covalent Binding by Site-directed Mutagenesis

Site-directed mutagenesis studies were carried out within Ala²⁰⁶-Met²⁴³ to determine the residue that is involved in covalent bond formation with -FNA. The fumaramate group of -FNA reacts with -SH most readily, but it may also react with other functional groups in the order of -NH₂ > -OH > -COOH.² We generated mutants of the µ receptor, in which several possible amino acid residues from amino acid 206-243 of the µ receptor were mutated individually, and examined the irreversible binding of [³H]-FNA to these mutant receptors. The following mutants were generated, T208A, K209R, S214A, C217A, S222A, K233R, C235S, and C235A, and mutation was confirmed by nucleotide sequence determination.

All mutants except C217A retained high affinity [³H]diprenorphine binding. Cys²¹⁷ has been postulated to form a disulfide bond with Cys¹⁴⁰ (4, 5, 6, 7, 8, 9, 10, 11). Irreversible binding of [³H]-FNA to T208A, K209R, S214A, S222A, C235S, and C235A mutants was similar to that of the wild type. As an example, irreversible binding of [³H]-FNA to the K209R mutant is shown in Fig. 4.

H]-FNA to wild type and mutant rat µ opioid receptors.

In contrast, the K233R mutant displayed no specific irreversible binding of [³H]-FNA (Fig. 4), although this mutant bound [³H]diprenorphine, -FNA, and [³H]-FNA with affinities similar to those of the wild type receptor (Fig. 5 and Table II). In addition to arginine substitution, we generated three other K233 mutants: K233A, K233H, and K233L. [³H]-FNA did not bind irreversibly to any of these K233 mutants (Fig. 4). In contrast, all three mutants bound [³H]diprenorphine, -FNA, and [³H]-FNA with high affinities similar to the wild type µ receptor (Table II and Fig. 5). These results indicate that Lys²³³ of TMD5 is the residue that forms a covalent bond with [³H]-FNA (Fig. 6).

H]-FNA to wild type and mutant rat µ opioid receptors.

Table II. Binding affinities of [3H]diprenorphine and -FNA to wild type and mutant µ opioid receptors Saturation binding of [3H]diprenorphine, inhibition of 0.2-0.4 n [3H]diprenorphine binding by -FNA and saturation reversible binding of [3H]-FNA were performed as described under ``Experimental Procedures.'' Data were derived from three independent experiments and expressed as mean ± standard error of the mean (n = 3). [3H]Diprenorphine [3H]-FNA Kd Bmax  -FNA IC50 Kd Bmax n fmol/mg protein n n fmol/mg protein Wild type 0.23  ± 0.04 2430  ± 183 3.3  ± 1.1 0.38  ± 0.08 3623  ± 232a K233R 0.21  ± 0.04 805  ± 233 5.8  ± 1.6 0.83  ± 0.03 2099  ± 82 K233A 0.24  ± 0.03 367  ± 22 3.9  ± 0.9 0.50  ± 0.04 764  ± 46 K233H 0.24  ± 0.04 399  ± 13 7.9  ± 0.5 1.26  ± 0.22 711  ± 27 K233L 0.45  ± 0.05 337  ± 52 5.6  ± 1.3 0.78  ± 0.06 740  ± 21 K209R 0.22  ± 0.05 850  ± 155 6.0  ± 1.8 NDb ND a  Bmax of reversible binding plus the maximal level of irreversible binding. b  ND, not determined.

DISCUSSION

In this study, we demonstrated that the region of [³H]-FNA incorporation in the rat µ receptor was Ala²⁰⁶-Met²⁴³ by peptide mapping of labeled receptor fragments (Fig. 6). The residue involved in covalent bond formation was then determined to be Lys²³³ by site-directed mutagenesis studies (Fig. 6). To the best of our knowledge, this is the first report that demonstrates directly the site of interaction between an affinity ligand and an opioid receptor. In addition, this study provides molecular basis for use of -FNA in opioid pharmacology.

Several laboratories, including ours, have been carrying out chimeric receptor studies and site-directed mutagenesis studies on opioid receptors to characterize ligand-receptor interaction (23, 25, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50). Results from chimeric receptor studies reveal regions in the receptor that are important for binding of certain ligands. Site-directed mutagenesis studies demonstrate that a certain amino acid residue is important for binding of agonists and/or antagonist. However, there is no direct information as to how the ligand is situated in the binding pocket. Through the use of an affinity ligand in approaches analogous to the one described in this study that one can determine directly the side chain of an amino acid residue in a receptor that interacts with a functional group of a ligand. This approach complements chimeric receptor and mutagenesis studies. Since -FNA is a fairly rigid compound, determination of the amino acid involved in covalent bond formation provides an anchoring site for molecular modeling of binding of -FNA and related morphinans and the µ receptor.

Peptide mapping of G protein-coupling receptors has been problematic, due to several factors including hydrophobic nature of the proteins, inaccessibility of certain residues by enzymes or chemicals and possible interference of detergents used for solubilization. Nevertheless, peptide mapping of affinity ligand-labeled receptors has been successfully carried out to depict the receptor binding domains of ₂- and ₂-adrenergic, neurokinin and m₁ muscarinic receptors (13, 51, 52, 53) among many others. In these studies, careful analysis with enzymes and chemicals, different doses and incubation periods was performed. Whether this technique is applicable to a given receptor can only be determined empirically.

CNBr cleavage was used successfully in the identification of binding site peptides of ₂-adrenergic (52) and m₁ muscarinic (13, 54, 55) receptors. We believe that mapping of CNBr-generated labeled fragments of [³H]-FNA-labeled µ receptors yielded reliable results for the following reasons. Immunoaffinity chromatography purified [³H]-FNA-labeled µ receptors to approximately 10% purity, which represents ~4000-fold purification over labeled membrane preparations. With higher purity, CNBr cleavage yielded more definitive results (29). Cleavage with CNBr was conducted with up to 4 mg/ml CNBr for 0.5 µg of protein in 500 µl, which is in great excess compared with standard protocols of 50 to 1 molar ratio of CNBr to Met residues (29). However, these concentrations of CNBr were in the same range as those in studies on ₂-adrenergic (52) and m₁ muscarinic (13, 54, 55) receptors. The high concentrations of CNBr needed is probably due to the hydrophobic nature of the receptors and the presence of detergent. Thus, the 4.5-kDa labeled band most likely represents the final product of CNBr cleavage. In addition, with the SDS-PAGE method of Schagger and von Jagow (31), the 3- and 6.2-kDa molecular mass markers were resolved very well with 15.5% gel (Fig. 3, A and B). Since the labeled band migrated between these two markers, it is in the range that molecular mass can be determined with confidence. Moreover, among the fragments that theoretically can be generated by CNBr cleavage, only two with masses of ~4.5 kDa. Fragments of ~4.5 kDa could be clearly resolved in the 15.5% SDS-PAGE used in this study from those of 3405 and 6320 Da, which are the two theoretical CNBr fragments closest to 4.5 kDa in molecular mass (Fig. 1). These peptide mapping results, thus, provide a good starting point for mutagenesis studies.

The findings that K233R, K233A, K233H, and K233L mutant µ receptors bound diprenorphine and -FNA with high affinity indicate that these mutants all retain opioid receptor conformations. Thus, the inability of these mutants to bind -FNA irreversibly is due to lack of covalent bond formation, but not their inability to bind -FNA in the binding pocket. IC₅₀ values of -FNA in inhibiting [³H]diprenorphine binding to the µ receptor and K233R, K233A, K233H, and K233L mutants are 6-9-fold of K_d values of [³H]-FNA reversible binding to the wild type and mutant receptors. The reasons may be 2-fold. First, when the radiolabeled concentration is about its K_d, IC₅₀ values of competing reversible ligands are approximately two times of their K_i values according to Cheng and Prusoff (56) equation. Second, it is possible that diprenorphine and -FNA are oriented differently in the binding pocket and/or have different reversible binding kinetics, resulting in higher IC₅₀ values of -FNA in inhibiting [³H]diprenorphine binding than K_d values of [³H]-FNA reversible binding.

Although K233R, K233A, K233H, and K233L mutants of the µ receptor bound diprenorphine and -FNA with high affinity similar to the wild type, expression levels varied among these mutants (Table II), which were all lower than that of the wild type. In binding experiments, this variation was taken into account. For each receptor, inhibition of [³H]diprenorphine by -FNA was carried out with the [³H]diprenorphine concentration at about its K_d and the receptor concentration of ±0.1 K_d. We have observed a similar variation in expression level among chimeric µ/ and µ/ receptors (25, 49). Expression levels differed among chimeras, yet it was quite consistent for a certain chimera. The reason for this difference is not clear, probably due to difference in secondary and/or tertiary structures of cDNA and mRNA molecules, rate of translation and rate of protein processing.

It is intriguing to note that for the wild type and all mutants, B_max values calculated from [³H]-FNA binding were higher than those of [³H]diprenorphine binding, although the rank order of expression levels among the wild type and all mutants remained the same: the wild type > K233R > K233A ~ K233H ~ K233L. The ratios of the B_max values of [³H]diprenorphine binding to those of [³H]-FNA binding were 0.67, 0.38, 0.48, 0.56, and 0.46 for the wild type, K233R, K233A, K233H, and K233L, respectively. Two possibilities may account for this discrepancy. First, [³H]diprenorphine obtained from Amersham Corp. (batch 64) required high performance liquid chromatography purification to attain high specific binding. The external standard method was used to determine the amount of [³H]diprenorphine by comparing peak areas of 210-nm absorbance. This method may introduce ~10% variation. Specific activity of [³H]diprenorphine was determined based on the amounts of radioactivity and [³H]diprenorphine in a small aliquot. Since the amount of radioactivity was very high in the aliquot, the experimental error may be large. Calculation of B_max based on the determined specific activity may introduce relatively large errors. Second, COS-1 cells used for [³H]-FNA binding was much lower than those used for [³H]diprenorphine binding in the passage number. Whether this difference in the passage number affects expression levels remained to be determined. We have found previously that for a given clone, the expression level is fairly consistent from one transfection experiment to another when COS-1 cells are identical or similar in the passage number (25, 49).

CNBr cleavage of [³H]-FNA-labeled receptors yielded a single labeled peptide band. Mutation of Lys²³³ to Arg, Ala, His, or Leu eliminated irreversible binding of [³H]-FNA to the µ receptor. These observations indicate that Lys²³³ is the primary, and most likely the only, site of -FNA covalent incorporation. Thus, -FNA binds to the µ receptor in a specific orientation, and the fumaramate side chain is fixed in space rather than rotating freely, since if the side chain freely rotates, -FNA will label multiple sites. -FNA is most likely to orient in the binding pocket of the µ receptor similar to its congeners, such as naltrexone, morphine, and naloxone for three reasons. First, our kinetic studies demonstrated that reversible binding of -FNA to the µ receptor occurs prior to the formation of covalent bond (21). Second, -FNA has similar binding affinity for the µ receptor as naltrexone,³ its parent compound, suggesting interactions with identical or similar binding epitopes. Third, a major consideration in support of the notion that the morphinan portion of -FNA is situated in the binding pocket when irreversible binding occurs is that opioid ligands that have higher affinity for the µ receptor are more potent in inhibiting irreversible binding of -FNA. In this regard, the order of potency in inhibiting [³H]-FNA irreversible binding is sufentanil > CTAP > morphine > DADLE (12, 20), which is similar to the order of their affinities for the µ receptor.

The conclusion that -FNA binds to the µ receptor in a specific orientation with the fumaramate side chain fixed in space is in accord with earlier findings of Portoghese and colleagues (57, 58) suggesting that [³H]-FNA labels one site, but not multiple sites, in the µ receptor. There are stringent structural requirements for irreversible binding of -FNA to occur. The 6-isomer of -FNA (-FNA) acts reversibly at the µ opioid receptor and can protect the receptor against irreversible blockade by -FNA (57). While the trans form (in the double bond of fumaramate) can irreversibly block the µ opioid receptor, the cis form acts only as a reversible ligand (57). When x-ray crystal structures of -FNA and -FNA were superimposed, it was found that the fumaramate double bond of -FNA was more than 2 Å away from that in -FNA and in the wrong orientation for nucleophilic attack to take place (58). Such strict spatial requirement in the alignment of the electrophilic group (fumaramate) in -FNA with the nucleophile in the receptor strongly suggests that -FNA reacts with only one residue. In addition, fumaramate group has only moderate reactivity compared to nitrogen mustard and isothiocyanate, which are functional groups commonly found in affinity ligands.

Lys²³³ is a conserved residue among µ, , and  opioid receptors. Thus, the selectivity of -FNA irreversible binding for the µ opioid receptor appears to be derived from tertiary structural differences, but not primary sequence variations. Using µ/ chimeric receptors, we (23) reported that the region of TMD 6, third extracellular loop, and TMD 7 of the µ receptor was important for irreversible binding of -FNA. In addition, results from µ/ and µ/ chimeric receptor studies indicate that the same region is important for binding selectivity of morphine (43, 47), a morphinan compound with a structure similar to -FNA. It is likely that this region of the µ receptor confers µ-specific receptor conformation for morphinans.

Since it was thought that -FNA formed a covalent bond with a Cys residue (14) in the µ opioid receptor, we also examined irreversible binding of [³H]-FNA to several Cys mutants: C79S, C159S, C190S, C190A, C251S, C292S, C292A, C321S, C330S, and C330A. All mutants had high affinity for -FNA and displayed similar levels of specific irreversible binding of [³H]-FNA as the wild type µ receptor. Thus, together with the finding on Lys²³³, Cys residues are not likely to be involved in covalent bond formation with -FNA.

It has been thought that the double bond of the fumaramate of -FNA reacts with the µ receptor by Michael addition. Lys²³³ lies at the extracellular end of TMD 5. A covalent bond between the double bond of the fumaramate of -FNA and the -amino group of lysine provides an anchoring site for molecular modeling of interaction of the µ receptor with -FNA and related morphinans, such as naltrexone, naloxone, and morphine. The fumaramate moiety of -FNA appears not to contribute to binding affinity, since naltrexone, devoid of the fumaramate group, binds to the µ receptor with similar high affinity as -FNA.⁴ Thus, Lys²³³ is not likely to be part of the binding pocket for other morphinans. The distance from the reactive electrophilic group to the morphinan rings of -FNA is approximately 5 Å. This covalent bond formation, coupled with -FNA structure, will allow us to generate model-based hypotheses to probe amino acids that are critical for the binding of morphinan compounds. Of particular interest are amino acid residues that interact with the phenolic hydroxyl moiety and the protonated nitrogen of -FNA and related morphinan compounds. Befort et al. (50) found that D147A mutation in the µ opioid receptor did not affect binding affinities of diprenorphine and naloxone, a finding that is inconsistent with the notion that Asp¹⁴⁷ forms ion-pairing with protonated nitrogen. Site-directed mutagenesis studies to determine amino acids interacting with the phenolic hydroxyl moiety and the protonated nitrogen are currently being conducted.