Differential opioid agonist regulation of the mouse mu opioid receptor.

μ opioid receptors mediate the analgesia induced by morphine. Prolonged use of morphine causes tolerance development and dependence. To investigate the molecular basis of tolerance and dependence, the cloned mouse μ opioid receptor with an amino-terminal epitope tag was stably expressed in human embryonic kidney (HEK) 293 cells, and the effects of prolonged opioid agonist treatment on receptor regulation were examined. In HEK 293 cells the expressed μ receptor showed high affinity, specific, saturable binding of radioligands and a pertussis toxin-sensitive inhibition of adenylyl cyclase. Pretreatment (1 h, 3 h, or overnight) of cells with 1 μM morphine or [D-Ala2MePhe4,Gly(ol)5]enkephalin (DAMGO) resulted in no apparent receptor desensitization, as assessed by opioid inhibition of forskolin-stimulated cAMP levels. In contrast, the morphine and DAMGO pretreatments (3 h) resulted in a 3-4-fold compensatory increase in forskolin-stimulated cAMP accumulation. The opioid agonists methadone and buprenorphine are used in the treatment of addiction because of a markedly lower abuse potential. Pretreatment of μ receptor-expressing HEK 293 cells with methadone or buprenorphine abolished the ability of opioids to inhibit adenylyl cyclase. No compensatory increase in forskolin-stimulated cAMP accumulation was found with methadone or buprenorphine; these opioids blocked the compensatory effects observed with morphine and DAMGO. Taken together, these results indicate that methadone and buprenorphine interact differently with the mouse μ receptor than either morphine or DAMGO. The ability of methadone and buprenorphine to desensitize the μ receptor and block the compensatory rise in forskolin-stimulated cAMP accumulation may be an underlying mechanism by which these agents are effective in the treatment of morphine addiction.

Opioid agonists are the therapeutic choice in the management of severe chronic pain (1). Despite effective analgesic benefit, extended periods of opioid treatment can result in a range of undesirable side effects, not the least of which is the development of tolerance. Opioids such as morphine demonstrate potent and stereoselective effects that are reversed by selective antagonists, suggesting an interaction with a specific membrane receptor (1). Selective, saturable, and high affinity membrane binding sites were subsequently demonstrated by radioligand binding assays on native tissue preparations (1).
Martin et al. (2) proposed the hypothesis of multiple opioid classes and defined receptors as the site of morphine action. The classification of opioid receptors into , , and ␦ was subsequently reinforced by the development of highly selective ligands for each class (3). Recently, the genetic basis for the three opioid classes was obtained, with the molecular cloning of the ␦, , and opioid receptors (for review, see Ref. 4).
Of the three opioid receptor classes, the receptor is thought to be the principal site of analgesic interaction, since most of the clinically relevant opioids used in pain management bind to this receptor with high affinity (5)(6)(7). Studies on the receptor offer potential molecular insights into the cellular basis for tolerance and dependence, serious side effects of prolonged opioid usage which may result from effects on receptor regulation (8).
Whereas morphine induces dependence, methadone is used in the treatment of opioid addiction, despite being a full agonist at the receptor (9). The therapeutic efficacy of methadone has been linked to a lower abuse potential, although it is not established why methadone has lower addictive properties than morphine. Buprenorphine is the other opioid agent used in the treatment of addiction and has not been found to precipitate withdrawal in animal and human studies (9,10). Buprenorphine has been reported to be a partial agonist or mixed agonist/antagonist at the receptor, and these properties have been implicated as the basis for addiction therapy (10).
Despite the clinical importance of tolerance development and dependence, relatively little is known about the molecular and cellular events induced by morphine (8). Also, the cellular events that accompany the therapeutic actions of methadone and buprenorphine in treating addiction are poorly understood.
The recent cloning and heterologous expression of the receptor have facilitated attempts to determine the biochemical mechanisms of clinically used opioids. In this study we investigated the molecular consequences of prolonged opioid agonist exposure on the cloned murine receptor stably expressed in HEK 1 293 cells. Continuous treatment of the receptor with morphine or DAMGO does not appear to desensitize the receptor, but it does result in an adaptive cellular response that is manifested by an increase in forskolin-stimulated intracellular cAMP levels. In contrast, methadone and buprenorphine treatments result in a pronounced desensitization of the receptor while also blocking the compensatory increases in forskolinstimulated cAMP levels observed with either morphine or DAMGO. These differential actions of opioid agonists at the receptor may explain the mechanism by which morphine induces dependence, whereas methadone and buprenorphine are effective therapeutic agents in the treatment of dependence.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK 293 cells were grown and maintained in minimal essential medium with Earle's salts (Life Technologies, Inc.) containing 10% fetal bovine serum, 1,000 units ml Ϫ1 penicillin and streptomycin sulfate in 10% CO 2 at 37°C. The mouse opioid receptor gene modified with the FLAG epitope (DYKDDDDK) at the amino terminus was a gift from Dr. Mark von Zastrow, University of California at San Francisco. The modified cDNA in the expression vector pcDNA3 (Invitrogen) was stably transfected into HEK 293 cells by a modification of the calcium phosphate protocol (11). Briefly, HEK 293 cell monolayers at approximately 70% confluence in T 75-cm 2 flasks were transfected with 30 g of plasmid. After an overnight incubation, the medium was removed, and the cells were treated with 5 ml of phosphate-buffered saline containing 10% glycerol for 10 min at room temperature. Cells were then washed twice with phosphate-buffered saline and incubated for 3 h at 37°C with growth medium containing 100 g ml Ϫ1 chloroquine. Cells were then washed twice and incubated for 48 h at 37°C in growth medium. Stable transformants were selected in growth medium containing 1 mg ml Ϫ1 Geneticin (Life Technologies, Inc.) and maintained in T 75-cm 2 tissue culture flasks in 10% CO 2 at 37°C.
Radioligand Binding Studies-Receptor binding studies were performed using membranes from stably transfected HEK 293 cells expressing the -FLAG cDNA. Membranes were prepared and receptor binding studies conducted as described (5) and as noted in the table and figure legends. Briefly, cell monolayers were harvested in 6 ml of buffer containing 50 mM Tris-HCl (pH 7.8) containing 1 mM EGTA, 5 mM MgCl 2 , 10 g ml Ϫ1 leupeptin, 10 g ml Ϫ1 pepstatin, 200 g ml Ϫ1 bacitracin, and 0.5 g ml Ϫ1 aprotinin and placed on ice. A cell pellet was prepared by centrifugation at 24,000 ϫ g for 7 min at 4°C and was homogenized in the same buffer using a Polytron (Brinkmann Instruments) at setting 2.5, 30 s. The cell homogenate was centrifuged for at 48,000 ϫ g for 20 min at 4°C, and the resulting cell pellet was homogenized and placed on ice for the binding assays. Binding assays were carried out at 25°C for 40 min in a final volume of 200 l in the presence or absence of competing ligands.
For agonist pretreatment studies, a 10-fold concentrated stock of agonist was diluted into growth medium and added to individual culture flasks. The final concentration of all agonists used in regulation studies was 1 M. Cell monolayers were harvested at the times indicated in the table and figure legends. Pertussis toxin treatments were carried out either in tissue culture flasks or 12-well plates overnight at 37°C with 100 ng ml Ϫ1 pertussis toxin (List Biological Laboratories, Campbell, CA).
cAMP Accumulation Studies-Stably transfected HEK 293 cells were subcultured in 12-well culture plates and allowed to recover for 72 h prior to experiments. For agonist pretreatment and pertussis toxin experiments, the growth medium was replaced for the times indicated in the figure legends with medium containing either ligand or pertussis toxin. After treatment, the medium was removed and replaced with 1 ml of growth medium containing 0.5 mM isobutylmethylxanthine, and the cells were incubated for 30 min at 37°C. The culture medium was then removed and replaced with fresh medium, with or without 10 M forskolin and opioids, and the cells were transferred to 37°C. After 5 min the medium was removed, 1.0 ml of 0.1 N HCl was added, and the monolayers were frozen at Ϫ20°C. For determination of the cAMP content of each well, the monolayers were thawed, placed on ice, sonicated, and the intracellular cAMP levels measured by radioimmunoassay (Amersham plc, Buckinghamshire, UK). Data obtained from the dose-response curves were analyzed by nonlinear regression analysis with GraphPad Prism 2.01 (GraphPad Software, Inc., San Diego, CA).
Radiolabeling of the M2 Monoclonal Antibody-The monoclonal antibody M2 against the FLAG epitope was purchased from Eastman Kodak. Antibody radioiodination was performed by a chloramine-T procedure (12). Briefly, 250 g of M2 antibody was incubated in 200 mM NaPO 4 buffer (pH 7.3) with 0.5 mCi of Na 125 I and the reaction initiated with 100 l of chloramine T (0.5 mg ml Ϫ1 in NaPO 4 buffer). After 30 s at room temperature, the reaction was terminated by the addition of 100 l of sodium metabisulfite (1.25 mg ml Ϫ1 in NaPO 4 buffer). The iodinated protein was separated from free 125 I by column chromatography with Sephadex G-25, and aliquots from the collected fractions were counted in an LKB ␥ scintillation counter and then stored at 4°C.
Antibody Binding to Cell Monolayers-After agonist treatment of cell monolayers, approximately 200,000 cpm of 125 I-M2 antibody was added to individual wells in 24-well plates in growth medium containing 10% fetal calf serum. Cells were incubated for 30 min at 37°C, the medium aspirated, and the monolayers washed twice with growth medium.
Bound radioactivity was determined after solubilization of the cell monolayer with 0.5 ml of 0.5 N NaOH and counted in a ␥ scintillation counter. Nonspecific radiolabeled antibody binding was determined in the presence of 10 M FLAG peptide (Eastman Kodak) and accounted for 30% or less of the total binding.
Opioids Used in This Study-Cell monolayers were pretreated with the selective receptor agonists DAMGO, morphine, and methadone and the nonselective opioids levorphanol, etorphine, and buprenorphine. The chemical structures of these opioid ligands are shown in Fig. 1.

RESULTS
To investigate the prolonged effects of opioid agonist treatment, we stably expressed the mouse receptor in HEK 293 cells. Pharmacological characterization of the stably transformed cells showed saturable, high affinity binding for selective opioid ligands, which was best defined as a single, noninteracting site (Table I) 0.14 nM (n ϭ 3). No specific radioligand binding was detected in untransfected HEK 293 cells (data not shown). Competition binding experiments with the selective peptide agonist [ 3 H]DAMGO showed that the cloned and expressed mouse receptor shared some pharmacological characteristics with the previously characterized rat and human receptors (6,7). As shown in Table I, selective agonists such as morphine, DAMGO, and methadone showed specific, high affinity membrane binding, as did the nonselective opioid agonists levorphanol, etorphine, and buprenorphine. The affinity for buprenorphine was approximately 3-fold higher than reported previously by Raynor et al. (6). This observed difference in affinity may reflect differences in the surrogate cell lines used (COS-7 in Raynor et al. (6); HEK 293 cells in this study); the presence of the amino-terminal epitope tag; or differences inherent in the cloned and expressed human, rat, and mouse receptors. The rank order of potency for opioid agonists to displace [ 3 H]DAMGO membrane binding at the epitope-tagged mouse receptor is etorphine Ͼ levorphanol Ͼ morphine ϭ buprenorphine ϭ DAMGO Ͼ methadone (Table I).
Previous studies from this and other laboratories have shown that the opioid receptors couple to the inhibition of adenylyl cyclase activity in HEK 293 cells (13)(14)(15). Consistent with the competition binding results, the opioid agonists were found to retain a similar rank order of potency when tested for the ability to inhibit forskolin-stimulated cAMP accumulation ( Fig. 2A). The nonselective opioids etorphine and levorphanol were found to be the most potent, followed by the selective opioids DAMGO and morphine. Buprenorphine and methadone were the least potent for inhibiting cAMP accumulation ( Fig.  2A). However, although buprenorphine appeared to be slightly more potent than methadone in the inhibition of forskolinstimulated cAMP accumulation, the maximal inhibition was significantly different from that of the other opioids examined in this study (p Ͻ 0.05, n ϭ 10). The inability of buprenorphine to inhibit cAMP accumulation maximally in the HEK 293 cells is consistent with the partial agonist activity that has been described by others (9, 10). As shown in Fig. 2B, the opioid agonist inhibition of forskolin-stimulated cAMP was sensitive to pertussis toxin pretreatment. Overnight treatment of cell monolayers with 100 ng ml Ϫ1 pertussis toxin abolished the ability of morphine, levorphanol, or etorphine to inhibit cAMP accumulation. This suggests that the receptor inhibition of cAMP accumulation in HEK 293 cells couples to either the G i or G o class of G proteins. The effects of morphine on the inhibition of forskolin-stimulated cAMP accumulation were blocked by the nonselective opioid antagonist naloxone (Fig. 2C).
[ 3 H]DAMGO binding to the receptor was reduced by 100 M GTP␥S or by overnight treatment with pertussis toxin, in agreement with the effects observed on functional inhibition of cAMP accumulation (Fig. 3).
To determine the effects of agonist pretreatment on the mouse receptor, the reduction of [ 3 H]DAMGO binding was used as a measure of receptor-G protein coupling. Using a 3-h agonist pretreatment protocol, the effects of the opioid agonists morphine, levorphanol, the synthetic peptide DAMGO, and the highly potent opioid etorphine were examined. In addition, two opioids currently used in the treatment of opioid addiction, methadone and buprenorphine, were also studied. Pretreatment with five of the six opioid agonists reduced high affinity [ 3 H]DAMGO binding (Fig. 3). This suggests that pretreatment with these three opioids results in an apparent altered coupling of the receptor with G proteins in the HEK 293 cells. Since high affinity [ 3 H]DAMGO binding is also reduced by overnight pertussis toxin treatment, the effects of the agonist pretreatment impair the coupling of the receptor to the G i or G o class of G proteins in the HEK 293 cells. DAMGO pretreatment was found to have no effect on high affinity [ 3 H]DAMGO binding (102 Ϯ 18%, n ϭ 3). Buprenorphine and etorphine pretreatments appeared to abolish [ 3 H]DAMGO binding relative to control binding (99 Ϯ 1% reduction, n ϭ 4 for buprenorphine; 99 Ϯ 0.3% reduction, n ϭ 3 for etorphine); and methadone, levorphanol, and morphine pretreatments reduced radiolabeled agonist binding (reductions of 79 Ϯ 13%, 73 Ϯ 6% and 35 Ϯ 8%, respectively, n ϭ 3). The effects of buprenorphine were stereoselective, as (ϩ) buprenorphine, the inactive stereoisomer, did not bind to the mouse receptor (data not shown).
To determine the effects of agonist pretreatment on -FLAG receptor density, two independent studies were performed. The binding of [ 3 H]diprenorphine, a nonselective opioid antagonist, was found to be decreased by four of the six agonist pretreatments examined, in agreement with the [ 3 H]DAMGO studies (Fig. 4A). Morphine and DAMGO pretreatments had negligible effects on [ 3 H]diprenorphine binding (reductions of 9 Ϯ 11% and 19 Ϯ 8%, respectively). And, whereas levorphanol reduced antagonist binding by 28 Ϯ 4%, methadone, buprenorphine, and etorphine pretreatments resulted in marked reductions in [ 3 H]diprenorphine binding (61 Ϯ 12%, 99 Ϯ 1%, and 68 Ϯ 8%, respectively). Since etorphine and buprenorphine have been proposed to be highly lipophilic opioids (10), the dramatic reductions in [ 3 H]diprenorphine binding may result from an irreversible component of the ligand binding that occurred during the 3-h pretreatment, making direct radioligand measurements inaccurate. To assess the effects of agonist pretreatment on receptor density independently of radioligand binding, an iodinated monoclonal antibody against the aminoterminal FLAG epitope was used. If the continued presence of the pretreatment agonist (e.g. buprenorphine) affected the affinity of [ 3 H]diprenorphine binding, then the use of the radiolabeled M2 antibody should provide an alternative means to measure the cell surface receptor density. At present, the amino terminus of the mouse receptor is predicted to be an extracellular site not known to be directly involved in ligand binding (3,16,17). Fig. 4B demonstrates that the binding of the radiolabeled M2 antibody to intact cell monolayers is unaffected by pretreatment with the agonists levorphanol (101 Ϯ 7% of control, n ϭ 4), buprenorphine (109 Ϯ 7% of control, n ϭ 8), or morphine binding (95 Ϯ 3% of control, n ϭ 3), whereas DAMGO resulted in small decreases in antibody binding (83 Ϯ 5% of control, n ϭ 4). In contrast, the 3-h pretreatment of cells with 1 M etorphine resulted in a reduction of M2 binding to 64 Ϯ 6% (n ϭ 8) of the untreated control (Fig. 4B). The antibody binding studies indicate that despite the ability of levorphanol,  Fig. 5, the functional consequences of agonist pretreatment were examined on the inhibition of forskolinstimulated cAMP accumulation. Fig. 5A shows that the morphine inhibition of forskolin-stimulated adenylyl cyclase is dosedependent over the concentration range of 10 Ϫ11 to 10 Ϫ6 M. In untreated monolayers, a maximal inhibition of cAMP accumulation (75 Ϯ 8%, n ϭ 8) was observed at 1 M morphine. Three-hour or overnight morphine pretreatment appeared to have little additional effect on the EC 50 of the morphine doseresponse curve (Fig. 5A). However, the pretreatments did appear to enhance the maximal inhibitions observed at the 1 M morphine concentration (93 Ϯ 5%, n ϭ 7 and 92 Ϯ 5%, n ϭ 4, inhibition of forskolin-stimulated cAMP accumulation for 3-h and overnight pretreatments, respectively). Levorphanol also inhibited forskolin-stimulated cAMP accumulation in a dosedependent manner, with a maximal inhibition in control cells of 76 Ϯ 6% (n ϭ 6) at 1 M. In contrast to the effects of morphine pretreatment on morphine inhibition of cAMP accumulation, pretreatment of the HEK 293 cells with levorphanol did not appear to enhance the maximal cAMP inhibition (82 Ϯ 8%, n ϭ 6, for 3-h pretreatment; 81 Ϯ 7%, n ϭ 4, for overnight pretreatment), while having little effect on the observed agonist EC 50 (Fig. 5B). In agreement with the morphine and levorphanol results, DAMGO (Fig. 5C) inhibition was also dose-dependent, with a maximal inhibition of 73 Ϯ 5% at 1 M (n ϭ 4). Unlike either morphine or levorphanol pretreatment, DAMGO pretreatment appeared to shift the EC 50 for cAMP inhibition to higher concentrations, with the control EC 50 ϭ 3.3 Ϯ 1.9 nM (n ϭ 4) and DAMGO-treated EC 50 ϭ 8.4 Ϯ 1.4 nM (n ϭ 7). DAMGO pretreatment did not appear to enhance the maximal inhibition of cAMP accumulation by DAMGO (Fig. 5C). Levorphanol and DAMGO 3-h pretreatments resulted in a rightward shift (Fig. 5D) in the EC 50 values for morphine inhibition of cAMP accumulation (10 Ϯ 1.3 nM, n ϭ 8, for DAMGO pretreatment and 37 Ϯ 2 nM, n ϭ 8, for levorphanol pretreatment). Neither pretreatment appeared to enhance the maximal inhibition of cAMP accumulation by morphine. Thus, although morphine and levorphanol pretreatments can reduce high affinity agonist binding, as assessed with [ 3 H]DAMGO, these agonists do not cause a functional desensitization of the receptor.
The inability of morphine, DAMGO, and levorphanol treatments to desensitize the response to forskolin-stimulated cAMP accumulation could be the result of the HEK 293 cells lacking critical cellular factors needed for opioid receptor desensitization. To test this possibility, HEK 293 cells were pretreated with etorphine, an extremely potent opioid (18 -20). Etorphine pretreatment desensitized the mouse receptor in HEK 293 cells (Fig. 6A). The 3-h etorphine exposure resulted in a rightward shift of the etorphine dose-response curve (control EC 50 ϭ 0.044 Ϯ 0.001 nM, n ϭ 6; treated EC 50 ϭ 2.1 Ϯ 1 nM, n ϭ 7) as well as a substantial reduction in the levels of maximal inhibition (control, 86 Ϯ 3%; treated, 14 Ϯ 12%). These results suggest that the mouse receptor expressed in HEK 293 cells can undergo desensitization when pretreated with the potent opioid etorphine. Although pretreatment of -FLAG-expressing HEK 293 cells with the clinically used analgesic morphine did not desensitize the receptor, a pronounced receptor de- sensitization was observed after pretreatment with methadone and buprenorphine, opioids used in the treatment of morphine and heroin dependence (Fig. 6, B and C). A 3-h pretreatment of cell monolayers with methadone resulted in a marked rightward shift in the EC 50 for the methadone inhibition of forskolin-stimulated cAMP accumulation ( Fig. 6B; methadone control EC 50 ϭ 9.6 Ϯ 1.5 nM, n ϭ 5; 3-h treated EC 50 ϭ 230 Ϯ 3.7 nM, n ϭ 5). The 3-h methadone pretreatment reduced the maximal inhibition of cAMP accumulation (control, 84 Ϯ 6%, n ϭ 7; 3-h pretreated, 55 Ϯ 6%, n ϭ 5). Buprenorphine pretreatment (3 h) also resulted in receptor desensitization (Fig. 6C) at buprenorphine concentrations above 10 Ϫ8 M, with the maximal inhibition of cAMP accumulation reduced from 67 Ϯ 3% (n ϭ 14) for control to 2 Ϯ 8% (n ϭ 5) for treated cells. No effect on forskolinstimulated cAMP accumulation was found with (ϩ) buprenorphine (data not shown). The three opioid agonists etorphine, methadone, and buprenorphine also appeared to desensitize the effects of morphine on cAMP accumulation (Fig. 6D) by markedly reducing the ability of maximal concentrations of morphine to inhibit forskolin-stimulated cAMP accumulation. Methadone pretreatment appeared to have a pronounced effect on the observed EC 50 for morphine inhibition, with an approximately 200-fold shift to higher agonist concentrations when compared with control ( Fig. 6D; 3-h treatment EC 50 ϭ 467 Ϯ 19 nM, n ϭ 3, compared with Fig. 4A; control EC 50 ϭ 2.3 Ϯ 1.5 nM). Thus, the opioids methadone and buprenorphine are able to desensitize the mouse receptor to morphine, with rightward shifts in the EC 50 for inhibition and a reduction in the maximal levels of inhibition after pretreatment. The pretreatment effects of these three opioids differ markedly from the effects observed on the mouse receptor after pretreatment with morphine, levorphanol, or DAMGO.
Adaptive sensitization or overshoot of adenylyl cyclase activity has been implicated in the cellular mechanisms of opioid tolerance (21,22), and a recent study on the cloned rat receptor suggested that the physiological consequences of receptor sensitization, rather than receptor desensitization, may be critical to tolerance development (22). We examined whether a physiological adaptation phenomenon occurred after opioid agonist pretreatment of HEK 293 cells by determining the levels of forskolin-stimulated cAMP after agonist pretreatment. As shown in Fig. 7, a 3-h pretreatment with 1 M morphine or DAMGO leads to an approximately 3.4 Ϯ 0.3-fold increase (ANOVA, n ϭ 17, p Ͻ 0.001) and 2.9 Ϯ 0.2-fold (n ϭ 12, p Ͻ 0.001) in forskolin-stimulated intracellular cAMP accumulation, respectively, for morphine and DAMGO, when com-

FIG. 4. Agonist pretreatment effects on [ 3 H]diprenorphine binding or 125 I-M2 monoclonal antibody binding to membranes prepared from HEK 293 cells stably expressing the -FLAG
cDNA. Panel A, cell monolayers were treated for 1 h, 3 h, or overnight with 1 M morphine, levorphanol, DAMGO, methadone, buprenorphine, or etorphine. After treatment, the monolayers were harvested and assayed for [ 3 H]diprenorphine binding. The percent reduction in specific radioligand binding was calculated from the corresponding untreated control monolayers. The results presented are the means Ϯ S.E. of three or more independent experiments, all assayed in triplicate with 1 M diprenorphine used to define nonspecific binding. Panel B, 125 I-M2 monoclonal antibody binding to -FLAG-expressing HEK 293 cell monolayers. HEK 293 cells were plated in 24-well plates and pretreated with the appropriate agonist for 3 h at 37°C. After treatment, the medium was aspirated, the monolayers were washed with growth medium, and approximately 250,000 cpm of 125 I-M2 monoclonal antibody was added. After a 30-min incubation at 37°C, the monolayers were washed, solubilized in 1 N NaOH, and counted in a ␥ scintillation counter. Total M2 binding was typically 2,500 -3,000 cpm for untreated cells; nonspecific binding, determined in the presence of 10 M FLAG peptide, was 10 -20% of the bound counts. The results are presented as percent of untreated control monolayers and are the mean Ϯ S.E. of at least three experiments. Statistical significance was determined by paired Student's t test with significance defined as p Ͻ 0.05. pared with untreated cells. This increase was not observed with levorphanol (n ϭ 8), methadone (n ϭ 12), or buprenorphine (n ϭ 9) pretreatment, which was significantly different from the effects of morphine or DAMGO (ANOVA, p Ͻ 0.001). A combination of morphine and DAMGO resulted in no additive increase in the adaptive cAMP response above that seen with the individual agonists (data not shown), suggesting that morphine and DAMGO may use a common mechanism in causing the increases in intracellular forskolin-stimulated cAMP. The effect of the potent opioid etorphine (n ϭ 9) was unique among the opioid agonists examined in this study, since the pretreatment reduced the intracellular cAMP levels by approximately 50% but was statistically insignificant compared with the individual effects of levorphanol, methadone, and buprenorphine (ANOVA, p Ͼ 0.05). These results indicate that the opioid agonists differ in their ability to cause compensatory increases in forskolin-stimulated cAMP with the mouse receptor, as only morphine and DAMGO pretreatments resulted in an adaptive, or overshoot, response.
The compensatory effects of morphine and DAMGO on adenylyl cyclase were blocked by the addition of other opioid agonists. Fig. 7 shows that when cell monolayers were pretreated with morphine in combination with other opioid agonists such as levorphanol (n ϭ 3, p Ͻ 0.001), methadone (n ϭ 3, p Ͻ 0.001), etorphine (n ϭ 2, p Ͻ 0.001), or buprenorphine (n ϭ 3, p Ͻ 0.001), the forskolin-stimulated increase of adenylyl cyclase activity was reversed. In a similar manner, DAMGO pretreatment in combination with other opioid agonists also resulted in a reduction in the forskolin-stimulated levels of intracellular cAMP (Fig. 7). No statistically significant differences were observed for the effects of methadone when compared with the combinations of methadone and morphine (n ϭ 3, p Ͼ 0.05) or methadone and DAMGO (n ϭ 3, p Ͼ 0.05). These results demonstrate that the opioid agonists levorphanol, etorphine, methadone, and buprenorphine are capable of blocking the compensatory increases in intracellular cAMP observed after chronic treatment with morphine or DAMGO. DISCUSSION In the current study we have stably expressed an epitopetagged mouse receptor (23) in HEK 293 cells and examined the pharmacological and functional consequences of prolonged agonist exposure. The mouse receptor binds the selective agonists morphine, DAMGO, and methadone with high affinity, as well as the nonselective opioids levorphanol, etorphine, and buprenorphine. In agreement with earlier studies on the heterologous expression of cloned opioid receptors in HEK 293 cells (13)(14)(15), the high affinity agonist binding of the expressed mouse receptor was sensitive to the effects of GTP␥S and pertussis toxin pretreatments. The sensitivity to pertussis toxin indicates that in the HEK 293 cells, the expressed mouse receptor couples to a member of the G i or G o family of G proteins. Such a result is consistent with immunoprecipitation studies on other surrogate cell lines that stably express the cloned opioid receptors (24).
The most significant findings of the current study are the demonstrations that the opioid agonists differentially regulate the mouse receptor and that the two opioids currently used in the treatment of addiction, methadone and buprenorphine, desensitize the receptor. The agonist pretreatment studies presented here show that morphine, levorphanol, etorphine, methadone, and buprenorphine effectively reduce high affinity agonist binding, although not to the same extent. In contrast, pretreatment with DAMGO had no effect on high affinity agonist binding. Buprenorphine and etorphine were found to be the most potent opioids in that both decreased high affinity agonist and antagonist binding, but only etorphine clearly down-regulated the receptor in HEK 293 cells. Of considerable importance is the demonstration that the two opioids currently used in the treatment of addiction, methadone and buprenorphine, both potently desensitize the mouse receptor, whereas morphine and levorphanol, which are highly addictive, did not desensitize the receptor. This suggests that the therapeutic role of methadone and buprenorphine in treating addiction may somehow be related to the ability to desensitize the receptor. Further differences between the opioid agonists were noted on the pretreatment sensitization of adenylyl cyclase. Morphine and DAMGO pretreatments substantially increased forskolin-stimulated intracellular cAMP levels, and this process was blocked by methadone and buprenorphine, which themselves did not cause the response. These results suggest both opioid agonists and antagonists are capable of reversing the supersensitization events seen after morphine or DAMGO pretreatment, indicating that the cellular basis for withdrawal may be more complex than previously proposed (21,22).
The interplay among tolerance, functional desensitization, receptor down-regulation, and the compensatory increases observed in intracellular second messengers is intricate. All of these phenomena have been proposed to play a role in opioid tolerance and dependence. The development of opioid tolerance has been reported to result from a loss of opioid responsiveness, or a desensitization of opioid action, rather than from a significant receptor down-regulation (25,26). To develop a cellular paradigm for tolerance development, a number of studies have been conducted on the effects of prolonged opioid treatment with opioid receptor-expressing cell lines. In cells that endogenously express opioid receptors, the results of receptor downregulation and desensitization studies have been complicated by the presence of one or more opioid receptor classes in the cell lines studied (27). Recently, studies on the cloned human and rat receptors in the COS-7 transient expression system have demonstrated that chronic agonist exposure results in neither receptor down-regulation nor desensitization (6). Arden et al. (13) were able to demonstrate down-regulation and phosphorylation of the rat receptor in stably transfected HEK 293 cells; however, it is unclear whether the down-regulation event was accompanied by receptor desensitization. Chakrabarti et al. (28) stably expressed the rat receptor in a murine neuroblastoma cell line and examined the effects of morphine and DAMGO pretreatment on opioid receptor binding and adenylyl cyclase inhibition. The results of this study (28) indicated that both morphine and DAMGO were able to reduce the maximum number of receptor binding sites and, following extended treatment, diminished the maximal levels of opioid-mediated inhibition of cAMP. The differences in receptor regulation observed in the studies of Raynor et al. (6) and Arden et al. (13) suggest that agonist regulation of the receptor may be a function of the surrogate cell lines or of the species isoforms of the receptor. None of these studies (6,13,28) addressed the actions of buprenorphine or methadone on receptor regulation.
The results presented here demonstrate that the desensitization of the receptor by buprenorphine is independent of down-regulation or internalization of the receptor, since cell surface labeling of the receptor with a radiolabeled monoclonal antibody revealed no changes in antibody binding after buprenorphine treatment. The loss of high affinity agonist and antagonist binding after buprenorphine pretreatment may be due to the continued presence of the opioid on the receptor, with the continuous activation of the receptor resulting in functional desensitization. Such a notion would be consistent with previous studies that have demonstrated a slow off-rate for buprenorphine binding to the opioid receptor (10).
However, activation of the receptor is not the sole requirement for desensitization since morphine, DAMGO, and levorphanol are more potent and effective than buprenorphine in inhibiting forskolin-stimulated cAMP accumulation. Clearly, buprenorphine must stimulate the receptor to induce a cascade of cellular events resulting in a desensitization of the receptor which does not occur with these other opioid agonists. This may be due to buprenorphine binding differently to the receptor. Preliminary mutagenesis studies suggest that buprenorphine has different requirements for binding to the receptor than morphine, DAMGO, or levorphanol, since point mutations at key amino acids which abolish morphine, DAMGO, or levorphanol binding do not affect buprenorphine binding (29).
In the current study etorphine induced cellular responses distinct from the other opioids examined, as a 3-h pretreatment resulted in the down-regulation and complete desensitization of the receptor while also lowering the compensatory increases observed for forskolin-stimulated cAMP accumulation. This unique agonist profile may be due to the extremely potent effects of etorphine and may also reflect the lipophilic nature of this opioid, as membrane binding studies point to a component of etorphine binding which is essentially irreversible (30). The in vivo actions of etorphine, as assessed in animal and human behavioral studies, also indicate that etorphine is profoundly more potent than morphine by a factor of as much as 500 -10,000-fold, depending upon the species tested and behavioral paradigm employed (18,20). It has also been noted that although etorphine has a short duration of action in man, it is capable of suppressing abstinence in morphine-dependent subjects, suggesting that etorphine may be effective in the short term curbing of opioid withdrawal (20). However, etorphine has been reported as being highly addictive in rodents and primates (19) and has been reported as having abuse potential in man (20). Studies on the ability of etorphine to induce tolerance and addiction indicate that tolerance can occur in rats and monkeys, but no evidence of tolerance has been observed in humans (19). Thus, the unique and complex biochemical actions of etorphine observed in the current study are also mirrored by the unusual effects reported for in vivo behavioral studies in a variety of species.
Methadone pretreatment also desensitized the receptor and prevented the adaptive sensitization of forskolin-stimulated cAMP accumulation by morphine. As observed for buprenorphine, the desensitization appeared to be independent of receptor down-regulation. Methadone, like morphine, is a full agonist at the mouse receptor (9). The ability of methadone to desensitize the receptor may also be due to its ability to bind differently to the receptor than morphine, DAMGO, or levorphanol. Consistent with this notion, recent studies have delineated distinct domains of the rat receptor involved in the binding of -selective alkaloids and peptides (17,31).
The inability of morphine and levorphanol to desensitize the receptor suggests that uncoupling of the receptor from adenylyl cyclase is unrelated to the tolerance development associated with these agonists. Both agonists induce dependence in rodents (1), yet only morphine causes an adaptive sensitization of adenylyl cyclase (21,22). These results suggest that the enhanced stimulation of forskolin-stimulated cAMP accumulation is unlikely to be the biochemical basis for morphine addiction. A recent study on adaptive sensitization in HEK 293 cells has shown that the type VI isoform of adenylyl cyclase may be involved, suggesting that in these cells the compensatory response occurs downstream of the receptor (32).
The ability of buprenorphine and methadone to desensitize the receptor functionally may be critical for the therapeutic efficacy of these opioids in the treatment of addiction. Morphine and heroin dependence may be a consequence of the continued activation of the receptor, with subsequent long term cellular changes in the nervous system. These prolonged neurochemical alterations may result from the inability of opioids such as morphine to desensitize the receptor. The therapeutic benefit derived from methadone or buprenorphine treatment may reflect the ability of these agents to interfere with the effects initiated by morphine, even in its continued presence, thereby interrupting a cascade of cellular events. However, methadone and buprenorphine are not opioid antagonists and do not precipitate withdrawal, thereby making them useful in the treatment of narcotic addiction. Furthermore, buprenorphine treatment does not appear to induce dependence (10) despite having agonist properties at the opioid receptor. Buprenorphine and morphine have been proposed to induce their in vivo analgesic action in animals through distinct cellular mechanisms (33). The therapeutic actions and long term consequences in man may also be due to distinct cellular actions.
Our findings show that chemically distinct opioid agonists can bind to the receptor and produce different cellular consequences. The studies presented here reveal that the addictive agents, such as morphine, and the opioids used in the treatment of addiction, buprenorphine and methadone, can be distinguished by different effects on receptor regulation. These studies provide a model system for understanding the molecular and cellular basis of dependence and tolerance, thereby facilitating studies on the design and development of new opioids that are potent analgesics but devoid of addictive properties.