Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The processes underlying the clinically observed tolerance to opioid drugs are complex and may involve learning mechanisms, compensatory changes in neuronal circuits, and desensitization of signal transduction mechanisms (Nestler and Aghajanian, 1997; Ramsay and Woods, 1997). Studies both in vivo and in vitro have documented that the response to opioids desensitize after prolonged exposure to opioids (Cox, 1978). One of the important molecular events that has been shown to be involved in the desensitization of the response to opioids is opioid receptor desensitization. The molecular basis for opioid receptor desensitization was shown to be a reduction in opioid agonist efficacy (Chavkin and Goldstein, 1982, 1984; Porreca and Burks, 1983) that manifests as a reduction in the rate of G protein activation by the agonist-bound receptor complex. For example, continuous infusion of guinea pigs or rats with morphine results in an uncoupling of µ-opioid receptors from associated G proteins as measured biochemically (Werling et al., 1989; Tao et al., 1993), cytochemically (Sim et al., 1996), or electrophysiologically (Christie et al., 1987). The uncoupling of MOR from G proteins is likely to result from the agonist-dependent phosphorylation of MOR mediated by a GRK (Kovoor et al. 1997). Because clinically useful opioids differ in their intrinsic efficacies, the rates of tolerance development for different opioid agonists might be expected to differ (Stevens and Yaksh ,1989a). However, the relationship between agonist efficacy and opioid tolerance is controversial. Stevens and Yaksh (1989a, 1989b) reported that tolerance to the analgesic effects of opioids measured by the hot-plate response of rats was greater for agonists with lower efficacy when continuously infused at equieffective doses. Duttaroy and Yoburn (1995) confirmed that the amount of analgesic tolerance after continuous infusion with opioids was inversely proportional to agonist efficacy but that intrinsic efficacy had no effect on the magnitude of tolerance produced by intermittent administration of opioid agonist to mice.

The complexity of the pharmacological parameters controlling the in vivo situation confounds the analysis of the relationship between efficacy and tolerance because differences in receptor occupancy, drug distribution, metabolism, and receptor selectivity affect the measured response and subsequent changes after prolonged drug administration. Stevens and Yaksh (1989a, 1989b) explained their results by suggesting that agonists with higher efficacy had a larger fraction of spare receptors and, hence, tolerance development to the more efficacious agonists was slower. However, this assumes that agonists with different efficacies produce receptor desensitization at the same rate when they bind the same fraction of receptors. Thus, to better elucidate the molecular mechanisms underlying the relationship between efficacy and receptor desensitization, we used Xenopus laevis oocytes coexpressing the MOR and the G protein-gated inwardly rectifying potassium channel complex (K_IR3.1 and K_IR3.4) to reconstitute a typical neuronal opioid response (North et al., 1987). To reconstitute a homologous receptor desensitization mechanism, we also coexpressed GRKs and -arr 2. Using this gene expression system, we previously showed that coexpression of -ARK2 (or GRK3) and -arr 2 synergistically produced an agonist-dependent homologous desensitization of the ₂-adrenergic receptor, the -opioid receptor, and the MOR activation of the K_IR3 response (Kovoor et al., 1997). Recently, Yu and colleagues (1997) reported that opioids with high receptor efficacies were better able to stimulate MOR phosphorylation than were opioids with low efficacies. Similarly, partial agonists were previously reported to be less effective at promoting -adrenergic receptor phosphorylation by -ARK (i.e., GRK2) than full agonists (Benovic et al., 1988). In the current study, we asked whether µ-opioid agonists with different efficacies produce GRK/-arr 2-mediated desensitization at different rates.

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Chemicals. DAMGO was obtained from Peninsula Laboratories (San Carlos, CA). Fentanyl, buprenorphine, and naloxone were from Research Biochemicals International (Natick, MA). Sufentanyl was a gift from Janssen Pharmaceuticals (Brussels, Belgium). All other chemicals were from Sigma Chemical (St Louis, MO).

Complementary DNA clones and cRNA synthesis. Except for GRK5, the cDNA clones used and cRNA synthesis methods have been described previously (Kovoor et al., 1997). The rat MOR clone was obtained from Dr. Lei Yu (GenBank accession No, L13069). cDNA for the K_IR3.1 (GIRK1) channel was obtained from Dr. Henry Lester (GenBank accession No. U01071). The K_IR3.4 (GIRK4) clone provided by Dr. John Adelman (GenBank accession No. X83584) and the -arr 2 cDNA (clone provided by Dr. Robert Lefkowitz, GenBank accession No. M91590) were first amplified by the use of Amplitaq DNA Polymerase (Perkin-Elmer Cetus, Norwalk, CT) in a standard polymerase chain reaction using oligonucleotides designed to add a T7 promoter region and a 45-base poly(A)⁺ tail. The rat GRK3 cDNA (GenBank accession No. M87855) was provided by Dr. Shaun Coughlin in the pFROGY vector. The human GRK5 clone (GenBank accession No. L15388) was obtained from Dr. Jeffrey Benovic (Kunapuli et al., 1993). Plasmid templates for all constructs were linearized before cRNA synthesis, and mMESSAGE MACHINE kits (Ambion, Austin, TX) were used to generate capped cRNA from the cDNA templates, which contained either T7, T3, or SP6 promoters to direct synthesis of sense transcripts.

Oocyte culture and injection. Stage IV oocytes were prepared as described previously (Kovoor et al., 1995). cRNA was injected (50 nl/oocyte) using a Drummond automatic microinjector, and then oocytes were incubated at 18° for 3-7 days in normal oocyte saline buffer (96 mM NaCl, 2 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂, and 5 mM HEPES, pH 7.5) solution supplemented with sodium pyruvate (2.5 mM) and gentamycin (50 µg/ml).

Electrophysiology. Oocytes were clamped at 80 mV with two electrodes filled with 3 M KCl having resistances of 0.5-1.5 M using a Geneclamp 500 amplifier and pCLAMP 6 software (Axon Instruments, Foster City, CA). Data were digitally recorded (Digidata 1200, Axon Instruments, and Intel 386PC) and filtered. Membrane current traces also were recorded using a chart recorder. To facilitate the inward potassium current flow through the K_IR3 channels, normal oocyte saline buffer was modified to increase KCl concentration to 16 mM as stated in Results. All drug responses were evaluated in this high K⁺-containing solution. In the experiments in which sufentanyl desensitization was evaluated, the high K⁺ solution in which drug responses were measured also contained 0.1% bovine serum albumin (Sigma). The concentration of NaCl was correspondingly decreased to maintain isoosmolarity. Opioid activation of K_IR3 conductance was measured as described previously (Kovoor et al., 1997).

Statistical analysis. The Student's t test (with two-tailed p values) was used for comparison of the independent mean values. Dose-response curves were fitted to a simple E_max model using the nonlinear regression analysis PCNONLIN v4.2 (SCI Software) to determine the EC₅₀ values and 95% confidence intervals.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

Dose-response relationship of µ-opioid agonists in X. laevis oocytes expressing MOR and K_IR3. As reported previously (Kovoor et al., 1995, 1997), activation of MOR by DAMGO, a selective MOR agonist, elicited a G protein-activated inwardly rectifying potassium conductance in X. laevis oocytes injected with cRNA for MOR and the K_IR3 channel (Fig. 1). MOR-activated K_IR3 channel response was measured as the increase in inward current produced in oocytes clamped at 80 mV. The opioid-induced increase in potassium conductance was blocked by naloxone and required oocyte injection with both MOR and K_IR3.1 cRNA. As reported, coexpression of K_IR3.1 and K_IR3.4 resulted in the formation of heteromultimeric channels, enhanced the agonist-activated response, reduced the amount of channel cRNA required, and greatly attenuated the amount of heterologous desensitization produced on prolonged agonist treatment (Krapivinsky et al., 1995; Kovoor et al., 1997). In this study, we generated cumulative concentration-response curves for activation of K_IR3 by the MOR ligands DAMGO, fentanyl, sufentanyl, buprenorphine, and morphine. To normalize for daily differences in receptor expression, dose-response data are presented as a percentage of the average maximal stimulation produced by DAMGO as measured in the same oocyte batch and under the same conditions.

View larger version (35K): [in this window] [in a new window]   Fig. 1.   Dose-dependent MOR activation by DAMGO, fentanyl, sufentanyl, and morphine of KIR3 expressed in X. laevis oocytes. Dose-response curves for DAMGO, fentanyl, sufentanyl, and morphine were constructed by testing increasing concentrations of agonist in X. laevis oocytes expressing KIR3 and MOR. Oocyte membrane potential was clamped at 80 mV while bathed in normal oocyte saline buffer containing 2 mM KCl, as described in the text. Oocytes then were superfused with a saline buffer in which the KCl concentration was increased to 16 mM, enabling a basal inward current to flow through the inwardly rectifying K+ channels. Cumulatively increasing doses of MOR agonists were applied to the bath, followed by perfusion with the opioid receptor antagonist naloxone (1 µM), which reversed the activation of MOR. The response to each dose of agonist was defined as the increase in inward current over the basal current. The agonist response at each dose of opioid was normalized as a percentage of the average maximal DAMGO response obtained in separate oocytes from the same group. Each point represents the mean response measured in three or four oocytes. Once a set of dose- response measurements for an agonist was performed, the oocyte was discarded. Oocytes were injected with 0.5 or 0.08 ng of cRNA for MOR, and 0.02 ng of each G protein-activated inwardly rectifying potassium channel subunit KIR3.1 and KIR3.4. Higher amounts of receptor cRNA produced a greater opioid response for each agonist. This is particularly evident for morphine, which was a pure antagonist at 0.08 ng of MOR cRNA and a partial agonist at 0.5 ng of MOR cRNA. Inset, maximal response to DAMGO (1 µM) at MOR cRNA concentrations of 0.08 and 0.05 ng of MOR cRNA and the maximal response to sufentanyl (300 nM) at MOR cRNA concentrations of 0.08 ng. EC50 values at 0.5 and 0.08 ng of MOR cRNA injected for each drug along with confidence intervals are summarized in Table 1. Error bars, mean ± standard error and are smaller than the plotted symbols when not shown.

Effect of DAMGO, fentanyl, and morphine treatment on GRK3- and -arr 2-mediated homologous desensitization of MOR. In oocytes expressing only MOR and the K_IR3 channel subunits K_IR3.1 and K_IR3.4, no significant homologous desensitization of the receptor-elicited K_IR3 current was observed (Fig. 2A). However, as demonstrated previously (Kovoor et al., 1997), coexpression of both GRK3 (i.e., -ARK 2) and -arr 2 led to a significant homologous desensitization of the DAMGO-activated K_IR3 current (Fig. 2A). Using oocytes coexpressing GRK3 and -arr 2 with the receptor and channel, we compared the DAMGO-induced desensitization with the desensitization induced by either 1 µM fentanyl or 1 µM morphine, each at the receptor-saturating concentration of agonist. Responses elicited by DAMGO or fentanyl desensitized by ~50% during 8 min of continuous agonist exposure. In the same group of oocytes, the morphine-activated responses showed no desensitization over the same period. To test whether the lack of acute desensitization by morphine was a result of its lower activity, we tested a concentration of DAMGO that evoked a response similar to the maximal morphine response and found that it also produced little receptor desensitization (data not shown). In the absence of either GRK3 or -arr 2, neither fentanyl nor DAMGO treatment produced acute homologous receptor desensitization (Fig. 2C and data not shown).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Effect of GRK3 on desensitization of DAMGO-, fentanyl-, or morphine-elicited MOR responses. A, Left, a representative current trace from an oocyte injected with 0.5 ng of MOR cRNA and 0.02 ng of each KIR3.1 and KIR3.4 shown to illustrate the data acquisition and measurement methods used in this study. Initially, the oocyte was bathed in normal oocyte saline buffer containing 2 mM KCl and clamped at 80 mV. The oocyte was then superfused with a saline buffer containing 16 mM KCl. The elevation in potassium concentration facilitates the measurement by increasing the basal current through the inwardly rectifying potassium channels while the cell is clamped at 80 mV. Subsequent application of 1 µM DAMGO further increased the inward current. Application of the MOR antagonist naloxone (1 µM) reversed the effects of DAMGO and enabled detection of a change in basal current (dashed line). Finally, decreasing the potassium concentration from 16 to 2 mM brought the current back to the initial value. Inset (circle), Respective cRNA injections. Right, identical current trace after base-line subtraction showing only the agonist-induced current. The subtraction was performed by taking the difference between the total inward current and the base-line inward current (dashed line on the left trace). B, Representative traces showing base-line-subtracted responses to sustained application of DAMGO, fentanyl, or morphine (1 µM each), respectively. Oocytes were injected with 0.5 ng of MOR, 0.02 ng of KIR3.1, 0.02 ng of KIR3.4., 0.5 ng of GRK3, and 2 ng of -arr 2. To determine the amount of response desensitization, agonist treatment was followed by 1 µM naloxone perfusion. The amount of desensitization was calculated as the change in response to agonist after 8 min expressed as a percentage of the peak response. Calibration for each trace: 400 nA, 4 min. C, Summary of the percent change in the peak response after acute agonist treatment of oocytes. Treatment of oocytes injected with MOR and KIR3 with 1 µM DAMGO did not produce a significant amount of response desensitization, whereas DAMGO treatment of oocytes from the same batch that also were injected with 0.5 ng of GRK3 and 2 ng of -arr 2 showed a substantial amount of desensitization. Treatment of oocytes coexpressing GRK3/-arr 2 with 1 µM fentanyl, but not with 1 µM morphine, also produced significant desensitization of the agonist response. Error bars, mean ± standard error for four to eight independent determinations. **, p < 0.01 compared with oocytes not injected with GRK3 and -arr 2; ##, p < 0.01 compared with DAMGO or fentanyl.

Effect of GRK5 substitution for GRK3 on MOR desensitization. We also tested whether another member of the GRK family, GRK5, was able to modulate MOR activity (Fig. 3). Brief treatment (8 min) with either DAMGO, fentanyl, or morphine (1 µM) did not produce significant desensitization in MOR-activated K_IR3 responses in oocytes that had been coinjected with cRNA for GRK5 and -arr 2. However, prolonged DAMGO or fentanyl treatment did produce significant desensitization in oocytes coexpressing both GRK5 and -arr 2. In this experiment, oocytes were exposed for 12-14 hr to either DAMGO, fentanyl, or morphine in normal oocyte saline buffer and then the drug was removed by perfusion with saline buffer for 5 min. The amount of receptor desensitization produced by each agonist was assessed by measuring the response to a challenge with 1 µM DAMGO; this is important because the degree of desensitization may otherwise vary depending on the efficacy of the test agonist.

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Effects of pretreatment with DAMGO, fentanyl, or morphine on GRK5-mediated desensitization of MOR responses. Control oocytes were injected with a mixture of the cRNAs: 0.5 ng of MOR, 0.02 ng of KIR3.1, and 0.02 ng of KIR3.4. Some of the oocytes also were injected with 1 ng of GRK5 and 3 ng of -arr 2. Insets (circles), cRNA injected; the KIR3.1/KIR3.4 complex is represented as KIR3, and all recordings were performed 3-4 days after injection. A, Representative current traces from oocytes untreated with agonist (control) or pretreated with DAMGO, fentanyl, or morphine (1 µM) for 12-14 hr. Top set of four traces, from oocytes expressing MOR and KIR3. Bottom set of traces, from oocytes also coexpressing GRK5 and -arr 2. For these traces, oocytes were clamped at 80 in normal oocyte buffer and then perfused with a challenge dose of 1 µM DAMGO in oocyte buffer containing 16 mM KCl. Dashed lines, application of 1 µM naloxone in the same buffer. The amplitude of the DAMGO response was defined as the portion reversed by naloxone (arrow to dashed line). The KIR3 currents were reversed by perfusion with high Na+-containing normal oocyte buffer. Calibration for each trace is 500 nA, 2 min. B, Summary of the desensitization of DAMGO-elicited MOR responses after 12-14-hr pretreatment with DAMGO, fentanyl, or morphine compared with matched, untreated oocytes. Error bars, mean ± standard error for three or four independent determinations. **, p < 0.01 compared with oocytes not coexpressing GRK5 and -arr 2.

View larger version (41K): [in this window] [in a new window]   Fig. 4.   Effect of pretreatment with DAMGO, sufentanyl, or morphine on GRK3-mediated desensitization of DAMGO-elicited MOR. A, Comparison of DAMGO and morphine treatment on GRK3-mediated desensitization of MOR. Oocytes were treated with either 1 µM DAMGO or 1 µM morphine for 12-14 hr and tested for MOR desensitization as described in Fig. 3. Conditions were the same as described above except 1 ng of GRK3 cRNA was substituted for GRK5 cRNA. Insets (circles), summarize the cRNA injected. After 12-14-hr pretreatment with either DAMGO or morphine, oocytes were clamped at 80 in normal oocyte buffer for 5 min to wash out residual agonist and then perfused with a challenge dose of 1 µM DAMGO in oocyte buffer containing 16 mM KCl. The amplitude of the DAMGO response was defined as the portion reversed by naloxone. Bars, mean amplitude of the naloxone-reversible DAMGO response compared with the DAMGO response in matched oocytes not pretreated with either DAMGO or morphine (percent control response). B, Comparison of DAMGO and sufentanyl treatment on GRK3-mediated desensitization of MOR. The experiment was performed exactly as described above except that oocytes were pretreated for only 20 min. with either 400 nM DAMGO or 300 nM sufentanyl. Error bars, mean ± standard error for four to seven independent replicate determinations. **, p < 0.01 compared with oocytes not injected with GRK3 and -arr 2.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Here, we report that in addition to regulation by GRK3, MOR also may be uncoupled in an agonist-dependent manner by a GRK with different properties, GRK5. We also show that the highly efficacious µ-opioid agonists DAMGO, sufentanyl, and fentanyl produce a GRK- and -arr-mediated uncoupling of the MOR. In our investigation, efficacy directly correlated with GRK3- and -arr 2-mediated receptor uncoupling. The highly efficacious ligands DAMGO, fentanyl, and sufentanyl caused a greater amount of receptor desensitization than morphine at saturating concentrations of agonist.

The faster rate of GRK-mediated desensitization caused by treatment with either DAMGO, fentanyl, or sufentanyl compared with that produced by morphine suggests that the more efficacious agonists induce a conformation of the receptor that is a better substrate for the kinase. Because G subunits are required for recruitment of GRK3 to the plasma membrane and activation of the kinase (Pitcher et al., 1995, Muller et al., 1997), the more efficacious ligands would be expected to activate the GRK3 more effectively than does morphine. However, we demonstrated here that GRK5, which differs from GRK3 in that it is not bound or activated by G, also produces a faster rate of MOR desensitization after DAMGO or fentanyl treatment. The more efficacious agonists thus allow more effective desensitization of the receptor independent of G release. The slower rate of GRK5-mediated desensitization evident in our study is consistent with the results of Kunapuli et al. (1994), who showed that purified GRK5 had a lower V_max than GRK2 at other G protein-coupled receptor substrates.

Sternini et al. (1996) and Keith et al. (1996) reported that in the HEK 293 cell line and in neurons in vivo, high efficacy agonists such as etorphine produced MOR internalization. However, morphine, which effectively activated the receptor, did not produce MOR internalization in HEK 293 cells or in vivo. Recent experiments suggest that in addition to uncoupling the ₂-adrenergic receptor from G proteins, phosphorylation of the ₂-adrenergic receptor by receptor kinases followed by -arr binding also is important for receptor internalization (Goodman et al., 1998). -Arr serves as an adapter protein that links the agonist-bound receptor to the cellular internalization machinery. Using the X. laevis oocyte expression system, we have shown a synergistic action of the GRKs and -arr on MOR desensitization. If phosphorylation of the MOR by receptor kinases followed by -arr binding to MOR also is important for MOR internalization, then our demonstration that morphine only very weakly stimulates GRK-mediated arrestin binding to MOR may provide a mechanistic explanation for the observations of Keith and colleagues. Under conditions in which the levels of receptor expression are high or when the amount of receptor activation is greatly amplified by the effector system used to monitor receptor activity, poorly efficacious agonists such as morphine may be expected to produce maximal responses. However, for poorly efficacious ligands, the fraction of bound receptors that are in the active state at any time would be small, and this would reduce the probability for GRK phosphorylation and subsequent -arr binding. In support of that hypothesis, we observed that the concentration of DAMGO that produced a response similar to that evoked by saturating doses of morphine also produced little receptor desensitization. Our results predict that the minimal dose of etorphine or DAMGO producing the same response as the dose of morphine tested also would fail to produce receptor internalization.

There are two discrepancies between the conclusions presented in the current study and the recent report by Yu et al. (1997). They found that sufentanyl was more efficacious than DAMGO in activating MORs expressed in oocytes and concluded that agonist-dependent receptor phosphorylation occurred in oocytes without adding GRK-cRNA. Under the conditions in which no cRNA for GRK or -arr is injected (Yu et al., 1997), we previously established that the response desensitization that occurs is heterologous and occurs via adaptive changes downstream of the receptor (e.g., without receptor phosphorylation) (Kovoor et al., 1995). In oocytes coexpressing MOR and the 5-hydroxytryptamine_1A receptor, stimulation of either receptor resulted in heterologous desensitization of the subsequent response to the other and injection of the nonhydrolyzable GTP analogue guanosine-5'-O-(3-thio)triphosphate did not affect the rate of desensitization (Kovoor et al., 1995). Furthermore, the heterologous desensitization of receptor/K_IR3 coupling was greatly attenuated by the coexpression of the K_IR3 channel subunits K_IR3.1 and K_IR3.4 (Kovoor et al., 1997). After testing a series of agonists with differing efficacies at MOR in the X. laevis oocyte expression system, they determined that the amount of desensitization of the agonist-induced responses was related directly to the efficacy of the agonist and concluded that the more efficacious agonists produced a greater receptor desensitization. However, Yu et al. (1997) did not show that receptor phosphorylation resulted in homologous receptor desensitization because their results simply indicate that heterologous desensitization (e.g., changes occurring downstream of the receptor, presumably at the channel) also is dependent on agonist efficacy.

Furthermore, under the conditions used by Yu et al., the high levels of MOR cRNA and low levels of channel expression (caused by the limiting expression of endogenous K_IR3.5) are likely to have resulted in spare opioid receptors for DAMGO, sufentanyl, and fentanyl. In X. laevis oocytes injected with cRNA for only the K_IR3.1 subunit (as done by Yu et al.), the expression of functional channels is dependent on the formation of heteromultimers between the exogenously expressed K_IR3.1 subunit and the endogenously expressed K_IR3.4 homologue K_IR3.5 (Krapivinsky et al., 1995, Hedin et al., 1996). In oocytes injected only with cRNA for the K_IR3.1, it has been shown that channel, and not receptor, is limiting (Hedin et al., 1996). This conclusion is supported by the lower EC₅₀ values obtained (Yu et al., 1997). Under conditions in which there are spare opioid receptors, it thus is surprising that sufentanyl produced greater responses than DAMGO (Yu et al., 1997). We failed to replicate that finding.

Previously, Blake et al. (1997) stated that treatment of MORs expressing HEK 293 cells with buprenorphine abolished the ability of other µ opioids to inhibit adenylyl cyclase, and they interpreted these results to argue that buprenorphine produced a profound receptor desensitization. This conclusion is in direct contrast to our results, which indicate that less efficacious agonists produce less receptor desensitization. To reconcile this discrepancy, we tested 10 µM buprenorphine and found that it did not activate K_IR3 channels at any of the levels of functional MOR. Buprenorphine is an agent that binds the MOR with very high affinity but very low efficacy, and we found that 1 µM buprenorphine acted as a pseudoirreversible antagonist consistent with previous findings (Thomas and Hoffman, 1993). Thus, the reduction in response previously reported may have resulted from receptor antagonism rather than desensitization.

There is an apparent contradiction in the relationship between ligand efficacy and receptor desensitization observed in this study and the opposite relationship between opiate tolerance and ligand efficacy observed in vivo (Stevens and Yaksh, 1989a, 1989b; Duttaroy and Yoburn, 1995). Highly efficacious agonists produced less tolerance in vivo while producing greater GRK-mediated receptor phosphorylation and receptor desensitization in vitro. The discrepancy can be reconciled if we assume that a very large fraction of spare receptors mediates the analgesic response to the highly efficacious opioids but that morphine has relatively few, if any, spare receptors in vivo. That assumption is supported by previous studies demonstrating a large fraction of spare opioid receptors in vivo (Chavkin and Goldstein, 1981; Perry et al., 1982). Furthermore, a drug with low efficacy such as morphine may act as a full agonist in tissues that express an excess of MOR and act as a partial agonist in tissues with lower receptor expression. After the desensitization or inactivation of an identical number of receptors, a less efficacious drug will show a greater shift in the dose-response curve than a more efficacious drug with an extremely large receptor reserve based on mass action kinetic principles. The in vivo tolerance studies were confounded because the amount of tolerance produced by prolonged exposure to an agonist was evaluated by examining the shift of the dose response curve to the same agonist. Under these conditions, a reduction in receptor reserve would be expected to produce a greater shift in the dose response of a partial agonist than a full agonist. A clearer measure of the degree of change at the MOR after prolonged treatment with various toleragens would have been obtained by evaluating the shifts in the dose-response curves for a single, receptor-selective test agonist.

In studies using established cell lines, heterologous gene expression systems, or real neurons, efforts to define the relationship between agonist efficacy and tolerance or desensitization have been hindered by the confounding presence of spare opioid receptors. In the current study, we deliberately used a system in which there were no spare receptor receptors for any of the opioids tested (increases in MOR cRNA injection gave proportional increases in agonist-activated maximal response). Thus, further work is required to define the molecular changes in opioid signaling after prolonged in vivo exposure to drug. Nevertheless, the results obtained using in vitro model systems predict that highly efficacious opioid agonists produce a higher rate of receptor phosphorylation in animals treated with drug for extended times.