Experimental Procedures

Cell Line.

293-SF-MOR cells have been characterized previously (Keith et al., 1996) and were a gift from Dr. Mark von Zastrow (University of California-San Francisco). Briefly, HEK 293 cells were stably transfected with the murine μ-opioid receptor (MOR) cDNA containing the signal FLAG epitope at the amino terminus. Cells were cultured in Dulbeccos's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, 100 μg/ml streptomycin, and 0.025 μg/ml Fungizone.

Flow Cytometric Analysis.

FLAG M2 antibody was labeled directly with fluorescein isothiocyanate (FITC) to an F/P ratio of ∼2.95 as described previously (Keith et al., 1998). For internalization experiments, 293-SF-MOR cells were harvested in 2 mM EDTA/PBS then resuspended in culture media and treated with various drugs for either 1 or 18 h at 37°C. Cells were then chilled to 0°C to stop further receptor internalization and stained with 10 μg/ml FITC-labeled FLAG M2 in 25% FBS for 1 h. Cells were washed two to three times (with 2% FBS/0.1% NaN₃/PBS) and 5,000 to 10,000 cells/sample were analyzed on a FACScan flow cytometer with CellQuest 3.0.1 for acquisition and analysis (Becton Dickinson, Mountain View, CA). The mean fluorescence of unstained cells was subtracted from the mean fluorescence of stained cells before calculating the change in surface receptor number after drug treatment.

cAMP Accumulation Assay.

293-SF-MOR cells were harvested and resuspended in PBS and 1 mM 3-isobutyl-1-methylxanthine for 10 min. Cells were then treated with 5 μM forskolin and various opioid drugs for 15 min in 96-well polypropylene plates. Samples were then sealed and boiled for 5 min, centrifuged at 4000g and supernatants were assayed with an [³H]cAMP assay kit (Diagnostic Products, Los Angeles, CA).

Membrane Preparation.

293-SF-MOR cells were pelleted, frozen at −70°C for at least 30 min, and then resuspended in ice-cold 50 mM Tris-HCl, pH 7, 2.5 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride (homogenization buffer). Cells were disrupted in a Dounce homogenizer and centrifuged at 1000g for 10 min at 4°C. The pellet was resuspended in homogenization buffer, rehomogenized, and centrifuged again at 1000g for 10 min at 4°C. Both supernatants were pooled and centrifuged at 13,000g for 45 min at 4°C. The pellet was resuspended in homogenization buffer, rehomogenized, and centrifuged at 13,000g for 45 min at 4°C. The pellet was resuspended in 50 mM Tris-HCl, pH 7, 0.32 M sucrose and stored at −70°C.

[³⁵S]GTPγS Binding Assay.

[³⁵S]GTPγS binding was performed as described by Befort et al. (1996) with modifications of GDP and [³⁵S]GTPγS concentrations. Briefly, 4 μg of membrane protein was incubated in 50 mM HEPES, pH 7.6, 5 mM MgCl₂, 100 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 0.1% BSA, 1 μM GDP, 0.1 nM [³⁵S]GTPγS, and various opioid ligands for 2 h at 0°C. Membranes were incubated with 10 μM unlabeled GTPγS to determine nonspecific binding. The mixtures were harvested with a Brandel M24RS harvester with presoaked Whatman GT100 GF/B glass filters and washed with ice-cold 50 mM Tris-HCl, pH 7.0. Filters were dried and counted in a Beckman LS1600 scintillation counter with Cytoscint ES (ICN Pharmaceuticals, Irvine, CA).

Materials.

FLAG M2 antibody was purchased from Eastman Kodak (New Haven, CT). [³⁵S]GTPγS (1250 Ci/mmol) was purchased from NEN (Boston, MA). FITC, 3-isobutyl-1-methylxanthine, forskolin, and phenylmethylsulfonyl fluoride were purchased from Sigma Chemical Co. (St. Louis, MO). PTX was purchased from Sigma Chemical Co. (St. Louis, MO) and Calbiochem (La Jolla, CA). Tissue culture supplies were purchased from Omega Scientific (Tarzana, CA). Etorphine, morphine, fentanyl, buprenorphine, and naloxone were gifts from the National Institute on Drug Abuse (Bethesda, MD) and the four stereoisomers of cis-β-hydroxy-3-methylfentanyl were obtained from the Research Triangle Institute (Research Triangle Park, NC).

Results

Agonist-Selective Internalization of μ-Opioid Receptors.

293-SF-MOR cells were treated with various ligands for 1 h at 37°C, chilled on ice, and then stained with FITC-labeled FLAG M2 antibody to quantitate the number of surface receptors by flow cytometric analysis. Figure 2A shows the effects of etorphine, fentanyl, and morphine on surface receptor staining. Etorphine induced a substantial loss of surface receptor staining (∼50%) with low nanomolar potency, whereas morphine only caused a relatively small loss (∼10%) of surface receptor staining at the highest concentration tested. Fentanyl triggered a 35% loss of surface receptors (an amount of internalization that was ∼65% of the maximal internalization caused by etorphine) with an EC₅₀ one order of magnitude greater than that of etorphine. Figure 2B shows the effects of the four stereoisomers of the fentanyl congener cis-β-hydroxy-3-methylfentanyl on surface receptor staining. Only two isomers ofcis-β-hydroxy-3-methylfentanyl, RTI-1a and RTI-1b, induced μ-receptor internalization. The potencies and efficacies of these two compounds for causing internalization were similar to that of etorphine (Table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Flow cytometric analysis of μ-opioid receptor internalization. A and B, 293-SF-MOR cells were treated with various concentrations of drugs for 1 h at 37°C and then chilled to 0°C to arrest further trafficking. Cells were stained with FITC-labeled FLAG M2 monoclonal antibody and analyzed on a FACScan flow cytometer. The mean fluorescence of 10,000 cells minus the mean fluorescence of unstained cells was used to calculate the percentage of surface receptor staining. Values are the means ± S.D. of triplicate determinations in a representative experiment.

Ligand-Selective Up-Regulation of Surface μ-Opioid Receptors.

Table 1 summarizes the effects of the alkaloid ligands on surface receptor staining. Whereas etorphine, RTI-1a, RTI-1b, fentanyl, and morphine induced measurable internalization, the opiate antagonist naloxone induced a 16% increase in surface μ-receptors. Furthermore, RTI-1d and buprenorphine also caused a significant increase in surface receptor staining. RTI-1c showed a tendency to increase surface receptors but this did not reach statistical significance. Longer treatment (18 h) of the cells with buprenorphine or naloxone resulted in a greater increase in surface receptor number than treatment for 1 h (Fig. 3B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Effect of PTX treatment on ligand-induced changes in surface μ-opioid receptor number. 293-SF-MOR cells were pretreated overnight with 100 ng/ml PTX. A and B, percentage of surface receptor staining after 1 h of drug treatment (A) and 18 h of drug treatment (B) was calculated by dividing the mean fluorescence of the cells in each drug treatment by the mean fluorescence of nondrug-treated control and PTX-treated cells. Values are the means ± S.E. of three to nine separate experiments. A single asterisk denotes that the effect of drug treatment on PTX-treated cells is significantly different from that of control cells that were not treated with PTX (Student's t test,P < .05). The double asterisk denotes that PTX treatment caused a greater impairment of the ability of fentanyl to cause internalization compared with that of etorphine (Student'st test, P < .05).

Potencies and Efficacies of Various Opioids for Second Messenger Modulation.

The potencies and efficacies of the alkaloid ligands to inhibit cAMP accumulation in forskolin-stimulated 293-SF-MOR cells are shown in Table 2. Etorphine, RTI-1a, and RTI-1b, which all induced maximal μ-receptor internalization, inhibited cAMP accumulation to the same extent and had subnanomolar potencies. Fentanyl and morphine also were full agonists in this assay and had similar low nanomolar potencies, whereas RTI-1c, RTI-1d, and buprenorphine were partial agonists in this assay. RTI-1c and buprenorphine had similar efficacies (∼80% of maximal inhibition), although buprenorphine was two orders of magnitude more potent than RTI-1c. Naloxone showed no inverse agonist activity in this assay (data not shown).

Potencies and Efficacies of Various Opioids for Activating G Proteins.

It has been shown that only a small fraction of receptors need to be activated to achieve maximal inhibition of adenylyl cyclase activity (Fantozzi et al., 1981). Therefore, it was considered that the [³⁵S]GTPγS binding assay might more realistically reflect the “intrinsic efficacy” of agonists because this assay gives a measure of agonist efficacy at the first level of signal transduction: activation of the G protein. As seen in Table 3, etorphine, RTI-1a, and RTI-1b all stimulated [³⁵S]GTPγS binding to the same extent with low nanomolar potencies. Morphine and fentanyl were partial agonists in this assay, with similar efficacies and potencies. RTI-1c, RTI-1d, and buprenorphine only stimulated between 7 and 21% of maximal [³⁵S]GTPγS, which is consistent with the fact that they were partial agonists for inhibiting cAMP accumulation. Naloxone was a neutral antagonist in this assay (data not shown).

Effect of PTX on Ligand-Induced Changes in Surface μ-Receptor Number.

PTX has been shown to ADP-ribosylate inhibitory G proteins and thereby uncouple GPCRs from their cognate G_i/o proteins (Kurose et al., 1983). Cells were treated overnight with 100 ng/ml PTX, which blocked stimulation of [³⁵S]GTPγS binding by etorphine (176 ± 14% increase in stimulation over basal versus a 9 ± 10% increase in PTX-treated cells; standard deviation, n = 2). PTX treatment also decreased maximal inhibition of cAMP accumulation by etorphine from 82 ± 3% (standard error,n = 9) to 12 ± 3% (standard error,n = 5). Figure 3A shows the change of surface receptor staining induced by a 1-h treatment with etorphine, fentanyl, morphine, buprenorphine, and naloxone in both PTX-treated and untreated cells. PTX treatment alone did not result in a significant change in surface receptor staining compared with untreated cells (data not shown). Etorphine (1 μM) induced a 42 ± 3% loss of surface receptor in PTX-treated cells, whereas the same concentration of drug caused a 63 ± 4% loss of surface receptor in untreated cells (standard error, n = 6). Hence, treatment with PTX inhibited the ability of 1 μM etorphine to cause μ-receptor internalization by 33%. Fentanyl (10 μM) induced a 22 ± 3% loss of surface receptor in PTX-treated cells but a 46 ± 3% loss of surface receptor in untreated cells (standard error, n = 6). Therefore, PTX treatment impaired the ability of fentanyl to cause internalization by 49%, which was significantly greater than the inhibition of etorphine-induced internalization (Student'st test, P < .05). The EC₅₀ values of both etorphine and fentanyl were not significantly different after PTX treatment (data not shown). PTX also attenuated the ability of etorphine to cause a loss of surface receptor after 18 h (72 ± 7% loss of surface receptor in control cells versus a 47 ± 6% loss of surface receptor in PTX-treated cells; standard error, n = 5, Student'st test, P < .05) (Fig. 3B).

Treatment with PTX completely abolished the slight internalization caused by 10 μM morphine; surface staining after 1 h of 10 μM morphine treatment was 90 ± 3% of control surface staining in untreated cells and 101 ± 1% of control surface staining in PTX-treated cells (standard error, n = 6) (Fig. 3A). Furthermore, the increases in surface receptor staining caused by 1-h treatment with either 10 μM naloxone and 10 μM buprenorphine were not significantly affected by PTX (Fig. 3A). Treatment with naloxone induced a 15 ± 2% increase in surface staining in control cells and a 15 ± 2% increase in surface staining in PTX-treated cells (standard error, n = 9). Buprenorphine treatment caused a 7 ± 3% increase in control cells and an 11 ± 0.4% increase in PTX-treated cells (standard error, n = 3). Interestingly, PTX treatment augmented the robust increase in surface receptor staining caused by 18 h of naloxone from a 39 ± 8% increase in surface staining in control cells to a 70 ± 16% increase in PTX-treated cells (standard error, n = 5, Student's t test, P < .05) (Fig. 3B). PTX-treatment also increased the effect of 18 h of buprenorphine treatment (17 ± 10% increase in control cells versus 51 ± 10% increase in PTX-treated cells; standard error, n = 5, Student's t test, P < .05) (Fig. 3B).

Blocking of Agonist-Induced Internalization.

We next determined whether morphine, which caused a slight amount of internalization, and the ligands that did not induce internalization (naloxone, RTI-1c, RTI-1d, buprenorphine) were able to block agonist-induced internalization. Table 4shows the “EC₅₀ values” for these ligands to block internalization induced by 10 nM etorphine (a concentration that is close to etorphine's EC₅₀ for inducing internalization). These ligands blocked etorphine-induced internalization with the following rank order potencies: buprenorphine > naloxone ≫ RTI-1d ≫ RTI-1c = morphine. These potencies generally parallel their rank order of binding affinities to the μ-receptor (Brine et al., 1995; Raynor et al., 1995; P.A.Z., unpublished data).

Discussion

The relationship between G protein activation and receptor internalization in GPCRs is unclear and appears to vary among receptors. Receptors are generally internalized in response to agonist binding (Bohm et al., 1997), although antagonists or antibodies are able to trigger internalization of some receptors (Roettger et al., 1997; Bhowmick et al., 1998; Tolbert and Lameh, 1998; Willins et al., 1999). Studies on β₂-adrenergic and muscarinic receptors have demonstrated that partial agonists cause less internalization than full agonists and the amount of receptor internalization caused by an agonist generally correlates with coupling efficiency (Toews and Perkins, 1984; Thompson and Fisher, 1990; January et al., 1997; Szekeres et al., 1998).

The availability of a variety of ligands for the μ-opioid receptor facilitates the study of the relationship between G protein coupling and receptor internalization. As shown in this study, alkaloid ligands that are full agonists in 293-SF-MOR cells as assessed by [³⁵S]GTPγS binding and cAMP assays (etorphine, RTI-1a, RTI-1b) induce maximal μ-receptor internalization. We also show that morphine and fentanyl, which are partial agonists for stimulating [³⁵S]GTPγS binding, cause significantly less internalization than the full agonist etorphine. Although fentanyl and morphine differ considerably in their ability to induce internalization (66 versus 17% of maximal internalization, respectively), they are similarly efficacious in stimulating [³⁵S]GTPγS binding (∼80% that of etorphine). Selley et al. (1997) also have shown that morphine and fentanyl have comparable efficacies in stimulating [³⁵S]GTPγS binding through μ-receptors, although two other groups have found that fentanyl is more efficacious than morphine in stimulating [³⁵S]GTPγS binding (Traynor and Nahorski, 1995; Emmerson et al., 1996). These discrepancies may result from differences in cell lines, levels of receptor expression, and incubation temperatures. It is of interest to note that the introduction of a methyl and a hydroxyl group (RTI-1a and RTI-1b) can confer full agonist properties to fentanyl both in [³⁵S]GTPγS binding and internalization assays.

A novel finding in this study is that weak partial agonists (buprenorphine and RTI-1d) are able to cause a significant increase in the number of surface receptors, similar to the classical opioid antagonist naloxone. Increases in μ-receptor binding following chronic exposure to naloxone have previously been described in cell lines and in vivo (Zadina et al., 1995; Unterwald et al., 1995; Koch et al., 1998), whereas decreases in μ-receptor binding have been shown to result after chronic buprenorphine treatment in vivo (Belcheva et al., 1993). However, this decrease might represent the failure to dissociate buprenorphine from μ-receptors (Boas and Villiger, 1985). That buprenorphine and RTI-1d, which have low efficacies for activating G proteins, are unable to trigger μ-receptor internalization suggests that the ability of a ligand to activate a certain level of G proteins is a prerequisite for initiating detectable receptor internalization. Furthermore, we have shown that ligands such as buprenorphine can act as agonists with regard to one function (i.e., signaling) and as antagonists for another (i.e., receptor internalization).

In addition to studying the effects of ligands with varying intrinsic activities on surface receptor number, we assessed the effects of blocking G protein function with PTX that ADP-ribosylates G_i/G_o proteins and thus interferes with the ability of these G proteins to be activated by the receptor (Kurose et al., 1983). Other studies that have looked at the effect of PTX on GPCR internalization find either no effect on agonist-induced internalization (Hsieh et al., 1999; Li et al., 1999) or a reduction in the rate or extent of internalization (Van Koppen et al., 1994; Koenig et al., 1997). A previous immunocytochemical study showed that the μ-receptors expressed in HEK cells still exhibit agonist-induced internalization after PTX treatment (Segredo et al., 1997), whereas another group showed internalization of μ-receptors expressed in Neuro_2A cells was blocked after PTX treatment (Chakrabarti et al., 1997). One limitation of immunocytochemical studies is that they are not readily quantifiable. Flow cytometry allows for quantification of changes in cell surface receptors and can reveal even subtle differences in surface receptor number. In this study, we found that PTX treatment impairs the ability of agonists to induce μ-receptor internalization. Furthermore, the ability of fentanyl to cause internalization is significantly more affected by PTX treatment than that of etorphine (49% reduction versus 33% reduction, respectively) and the slight internalization caused by morphine is completely abolished by PTX treatment. Thus, ligands that are extremely efficacious for inducing receptor internalization and causing maximal G protein activation appear to be more resistant to the effects of PTX treatment than less efficacious ligands. The PTX results parallel previous findings that have shown that PTX treatment did not greatly affect the ability of full agonists to cause down-regulation of the μ-receptor (total loss of receptor protein as assessed by radioligand binding), whereas the ability of partial agonists to cause down-regulation was greatly impaired (Yabaluri and Medzihradsky, 1997). Similar to results from other studies (Selley et al., 1998), PTX did not completely block opioid-mediated inhibition of adenylyl cyclase and residual signaling activity may explain the inability of PTX to completely block the internalization caused by etorphine.

The ability of buprenorphine and naloxone to cause up-regulation of surface receptors after 1 h does not appear to be affected by PTX treatment. However, PTX treatment causes a considerable augmentation of the increase in surface receptor number induced by 18-h treatment with buprenorphine or naloxone. Thus, PTX clearly modulates the ability of ligands to regulate surface receptors, impairing the internalization mechanism while potentiating the ability of antagonists and low-efficacy partial agonists to increase surface μ-receptors. This is consistent with studies showing that heterotrimeric G proteins are involved in protein trafficking (Helms, 1995). Additionally, ADP-ribosylation of G proteins may alter the association of G proteins with their cognate receptors and thus modify the accessibility of proteins involved in trafficking.

Finally, the ability of various ligands to block etorphine-induced receptor internalization was assessed. Ligands block internalization with potencies that parallel their binding affinities for the μ-receptor (buprenorphine > naloxone ≫ RTI-1d ≫ RTI-1c = morphine). Morphine has been reported to induce enkephalin release in vivo and its ability to block internalization may contribute to its physiological actions (Olive and Maidment, 1998). Perhaps more importantly, the observation that buprenorphine also can block internalization with a high potency and potentially up-regulate μ-receptors in vivo may aid in explaining its utility as a treatment for drug addiction.

Overall, this study provides a number of insights into the relationship between G protein activation and regulation of μ-opioid receptor number on the cell surface. Although there is no strict correlation within groups, full agonists for G protein activation induce maximal internalization, whereas high-efficacy partial agonists for G protein activation (morphine and fentanyl) induce partial internalization. The data suggest that strong agonists are able to put the receptor in a conformation that is favorable for both activating G proteins and entering the endocytic route. Agonists such as etorphine and DAMGO, which efficiently trigger internalization, have been shown to cause more μ-receptor phosphorylation than morphine and might allow for a state of the receptor wherein adaptor proteins (such as β-arrestins) bind and subsequently direct the receptor into the clathrin-mediated endocytic pathway (Yu et al., 1997; Ferguson et al., 1998). Two articles have demonstrated that morphine is able to efficiently induce internalization of the μ-receptor if G protein-coupled receptor kinase or β-arrestin is overexpressed (Whistler and von Zastrow, 1998; Zhang et al., 1998). Ligands that are not able to elicit a threshold activation state (i.e., antagonists and weak partial agonists) may put the receptor in a conformational state that is unfavorable for entering the endocytic pathway. Indeed, we have shown that the partial agonists buprenorphine and RTI-1d, as well as the antagonist naloxone, not only fail to induce internalization but also cause an up-regulation of surface μ-receptors.

In a recent article by Whistler et al. (1999), it was suggested that the relative activity versus endocytosis (RAVE) value of a drug is predictive of the ability of that drug to induce tolerance and/or dependence (Roth and Willins, 1999; Whistler et al., 1999). Drugs with a high RAVE value (e.g., morphine) cause more tolerance in certain dosing paradigms than drugs that have RAVE values much lower than morphine (e.g., etorphine and methadone). Although this is an intriguing and provocative theory, drugs with high specificity for the μ-opioid receptor should be tested to further support this hypothesis. For instance, etorphine has a very high affinity for the δ- and κ-opioid receptors in addition to the μ-receptor (Raynor et al., 1994) and methadone can act as a noncompetitive antagonist atN-methyl-d-aspartate receptors, a property shown to block opioid tolerance (Davis and Inturrisi, 1999). Although there are no data available concerning the tolerance and dependence liabilities of the isomers ofcis-β-hydroxy-3-methylfentanyl, these compounds will undoubtedly prove to be excellent tools with which to explore the relationship between μ-receptor internalization and tolerance and dependence to different opioid drugs given their high μ-receptor selectivity (>15,000 times more selective for μ-receptors than δ- and κ-receptors for RTI-1a and RTI-1b) (Brine et al., 1995; Wang et al., 1995). Although the precise role of ligand-regulated trafficking of μ-receptors in adaptational processes after acute and chronic opioid treatment remains to be determined, differences in trafficking will surely be found to contribute to the unique pharmacological profiles of the different opiate drugs.