When the opioid agonist morphine is given chronically and systemically to mice by pellet implantation for 3 days, the animals develop substantial tolerance to the antinociceptive effect of a test dose of morphine given systemically. When the test dose is administered to the spinal cord, however, very little tolerance is observed. We tested six strains of mice differing in the degree to which they develop systemic tolerance to morphine and found that none of them developed significant tolerance to spinal morphine. However, most of these strains did develop substantial spinal tolerance to antinociception induced by the selective μ-agonist [d-Ala2,N-Me-Phe4,Gly5-ol]-enkephalin (DAMGO) and by the selective δ-agonist [d-Pen2,d-Pen5]-enkephalin (DPDPE). Moreover, in naı̈ve animals, the antinociceptive effect of both DAMGO and DPDPE was blocked byd-Phe-Cys-Tyr-d-Trp-Arg-Thr-Pen-Thr-NH2, a selective μ-antagonist, indicating that both agonists mediate antinociception in the spinal cord through μ-receptors. In addition to directly mediating antinociception, however, DPDPE potentiated the antinociceptive activity of DAMGO in the spinal cord of naı̈ve animals, and this antinociception was blocked by the δ-antagonist H-TyrTicPsi[CH2NH]Phe-Thr-OH (TIPPψ), indicating mediation through δ-receptors. In contrast, in tolerant animals, TIPPψ enhanced the antinociception of DAMGO. These results thus demonstrate not only that μ- and δ-opioid receptors interact in naı̈ve animals, but that the nature of this interaction changes during tolerance, from a potentiation to an inhibition. The lack of tolerance to spinal morphine may result from the ability of morphine to act as a partial antagonist at δ-receptors.
There are three major types of opioid receptors in the mammalian central nervous system, μ, δ, and κ, encoded by distinct, although homologous genes (Satoh and Minami, 1995), and they differ in sensitivity to agonists. A preponderance of evidence indicates that opioid antinociception is mediated via the μ-receptor (Fang et al., 1986; Matthes et al., 1996; Sora et al., 1997), although δ- and κ-receptors can also mediate this effect in at least some central nervous system locations and some antinociceptive tests (Porreca et al., 1987; Qi et al., 1990; Sofuoglu et al., 1991). However, several lines of evidence suggest that interactions among different types of opioid receptors may also contribute to the pharmacology of these drugs. This evidence is both pharmacological and biochemical. Pharmacologically, it has been shown that activation of one opioid receptor type can alter the sensitivity to agonist of a different receptor type in the same location, or of the same receptor type in a different location. Examples of the first type of interaction include potentiation of μ-receptor agonist-induced antinociception by δ-receptor agonist in the brain (Heyman et al., 1987, 1989) or spinal cord (Larson et al., 1980; He and Lee, 1998). Examples of the second type of effect include synergistic (mutually potentiating) effects of μ-agonists in the brain and spinal cord or between different regions in the brain (Yeung and Rudy, 1980; Roerig et al., 1984; Bodnar et al., 1991; Rossi et al., 1993).
In addition to this whole animal evidence for opioid receptor interactions, additional support comes from pharmacological and biochemical analysis of cloned opioid receptors expressed in defined cell types. Recent studies by Jordan and Devi (1999) and George et al. (2000) demonstrate that when two different opioid receptor types are expressed in the same cell, their affinities for certain agonists differ from those of the two receptors expressed alone. In both these studies, direct physical interaction of the receptor types into heterodimers or higher order oligomers was indicated by the ability of antibodies to one receptor type to immunoprecipitate the other receptor type. Furthermore, the interacting receptor preparations also displayed altered responses to guanine nucleotides and pertussis toxin, suggesting changes in their signal-transducing properties.
The physiological relevance of such opioid receptor interactions has yet to be demonstrated. But considerable evidence suggests that they could play a role in the development of opioid tolerance. A characteristic property of opioid agonists, constituting one of their greatest clinical liabilities, is that their antinociceptive potency decreases upon chronic treatment. The most widely accepted explanation of opioid tolerance is that it results from changes in the affinity (desensitization) or number (down-regulation) of opioid receptors. Both desensitization and down-regulation have been well documented in certain cultured cell systems (Crain, 1984; Law et al., 1984; Yu et al., 1990), but the evidence that they occur in the brains of whole animals and correlate with tolerance development is more controversial (Tao et al., 1987, 1990; Abdelhamid and Takemori, 1991; Rothman et al., 1991; Polastron et al., 1994; Brodsky et al., 1995; Bernstein and Welch, 1998).
An alternative explanation of opioid tolerance is that it results, at least in part, from altered interactions between opioid receptors. In support of this, it has been shown that the synergistic interaction between morphine administered to the brain and spinal cord (Roerig et al., 1984; Roerig and Fujimoto, 1988; He and Lee, 1997), as well as the potentiation of spinal μ-receptors by δ-receptors (He and Lee, 1998), is lost during tolerance. These observations, together with reports that in animals made tolerant to systemic morphine, tolerance develops very slowly to discrete spinal sites (Yaksh et al., 1977;Pfeifer et al., 1989), suggest that during tolerance, an alteration occurs in the interaction between opioid receptors. This altered interaction results in a reduced sensitivity of systemic morphine, although the affinity of morphine for any particular opioid receptor type may be unchanged.
However, in animals made tolerant to systemic morphine, spinal tolerance does rapidly develop to selective μ- and δ-agonists (Porreca et al., 1987; Roerig et al., 1991). Therefore, it seems that tolerance to individual μ- and δ-receptors does occur. Why, then, does tolerance to morphine not rapidly develop under these conditions? In this study, we first confirm that tolerance occurs to selective μ- and δ-agonists, but not to morphine, at the spinal level in several different strains of mice. We then investigate the nature of the interaction between μ- and δ-receptors, and provide a possible explanation for the lack of morphine tolerance.
Materials and Methods
The strains of mice used in this study are listed in Table 1. The sources of these animals were: NIH Swiss-Webster (H-NIH), ND4 (H-ND4), and ICR (H-ICR), Harlan Bioproducts for Science, Indianapolis, IN; Simonsen Swiss-Webster (S-SW), Simonsen Laboratories, Gilroy, CA; Hilltop Swiss-Webster (H-SW), Hilltop, Scottsdale, PA; and Charles River Swiss-Webster (CR-SW), Charles River Laboratories, Wilmington, MA. All six of these strains are derived from an original Swiss strain (two males, seven females) brought to the Rockefeller Institute, New York, NY, in 1926. Thus, these strains are fairly closely related, resulting from genetic changes in various descendant colonies from this original stock.
The abbreviations listed in Table 1 are used to refer to these groups throughout this report. All animals were male and weighed 20 to 25 g, and they were housed for at least 24 h before experiments in a temperature- and humidity-controlled environment, and fed ad libitum. Each mouse was used only once.
The antinociceptive assay was a modification of the radiant tail-flick test described by Tulunay and Takemori (1974). The data were made quantal by designating a positive antinociceptive response as one exhibiting an increased latency to tail-flick at least 3 S.D. above the mean latency of animals not given drug. At least three groups of 10 mice were used to establish dose-response curves and to estimate AD50 values.
Mice were rendered tolerant to morphine by s.c. implantation of one morphine pellet (containing 75 mg of morphine-free base) for 72 h. The degree of tolerance was determined as the ratio of the AD50 value of agonist in morphine-pelleted mice to that of placebo-pelleted mice (Way et al., 1969). The implanted morphine pellet was left intact during the antinociceptive assay.
Drugs were injected i.t. (Hylden and Wilcox, 1980) in a volume of 5 μl/mouse and the latencies were measured 30 min after the injections.
AD50 values and their 95% confidence limits were calculated by the method of Litchfield and Wilcoxon (1949). The apparent pA2 values of CTAP for DAMGO and DPDPE were estimated by the method described by McGilliard and Takemori (1978).
Morphine pellets were provided by the National Institute on Drug Abuse (Rockville, MD). DAMGO, DPDPE, and CTAP were provided by National Institute on Drug Abuse and supplied by Multiple Peptide Systems (San Diego, CA). TIPPψ was a gift from P. Schiller, Clinical Research Institute of Montreal, Montreal, QC, Canada and dissolved in saline with 15% dimethyl sulfoxide. DPDPE was dissolved in 2% 2-hydroxypropyl-β-cyclodextrine (RBI/Sigma, Natick, MA), and other drugs were dissolved in saline.
Antinociceptive Effect of Morphine(s.c. or i.t.) in Naı̈ve and Morphine-Tolerant Animals.
To correlate changes in the antinociceptive potency of DAMGO or DPDPE with degree of morphine tolerance, we decided to test these selective agonists in several strains of mice differing in the degree to which they develop tolerance after 3-day implantation of morphine pellet. We first characterized these strains with respect to their degree of tolerance to morphine. As shown in Fig. 1, these animals could be grouped into two classes of three strains each, one group designated the high-tolerance group (8- to 10-fold tolerance), and the other the low-tolerance group (2- to 4-fold).
In contrast to the tolerance observed to s.c. morphine, when i.t. morphine was given to placebo and morphine-pelleted animals, a significant degree of tolerance was observed in only one of the six strains (Table 1). There was no significant tolerance in the other five strains. This confirms previous studies (Roerig and Fujimoto, 1988; He and Lee, 1998) and demonstrates that the lack of tolerance to i.t. morphine is not restricted to a single strain of mice.
Antinociceptive Effect of DAMGO and DPDPE in Naı̈ve and Morphine-Tolerant Animals.
As discussed in Introduction, we have previously observed changes in the interaction of μ- and δ-opioid receptors in the spinal cord in H-SW mice during morphine tolerance (He and Lee, 1998). To investigate further the role of δ-receptors in tolerance, we initially assessed the antinociceptive potency of intrathecal DAMGO, a μ-selective agonist, and DPDPE, a δ-selective agonist, administered separately to each of the six strains of mice. As shown in Table 2, both agonists induced antinociception in all the strains of animals tested, and in most cases, the AD50 values were significantly higher in morphine-pelleted animals, indicating cross-tolerance to s.c. morphine. Two of the six strains, the S-SW mice (low-tolerance group), and the H-SW mice (high-tolerance group), showed a low degree of tolerance to DAMGO (i.t.), which was not significant in our analysis. All six strains, however, showed a significant degree of tolerance to DPDPE (i.t.). Thus, even though there is no tolerance to spinal morphine in morphine-pelleted animals, there generally is tolerance to selective agonists.
Antagonism of DAMGO and DPDPE by CTAP.
We previously reported that the selective μ-antagonist CTAP blocked the antinociceptive effect of DPDPE as well as DAMGO in the spinal cord, using Hilltop Swiss-Webster mice (He and Lee, 1998). We repeated this observation here with two additional strains of mice, one from the low-tolerance group (H-ND4) and one from the high-tolerance group (CR-SW). As shown in Fig. 2, CTAP at doses of 0.05 to 0.25 nmol antagonized both DAMGO and DPDPE, and the pA2 values for CTAP against the two agonists were similar, in both strains of mice (Table3). We conclude that in both of these strains of animals, representative of the high- and low-tolerance groups, DAMGO and DPDPE both interact with μ-receptors.
Taken together with the results showing that the spinal cord develops tolerance to both DAMGO and DPDPE, these observations demonstrate that the spinal μ-receptor develops tolerance in morphine-pelleted animals. Because tolerance to spinally administered morphine is not observed, this suggests that morphine's interaction may involve the δ-receptor as well.
Antagonism of DAMGO and DPDPE by TIPPψ.
To assess the role of the δ-receptor in tolerance to spinal DAMGO and DPDPE, we used the selective δ-antagonist TIPPψ. At doses as high as 30 nmol/mouse, TIPPψ had no effect on antinociception of DAMGO (i.t.) in the naı̈ve CR-SW or H-ND4 mice (Fig.3). This observation confirmed that TIPPψ at this dose was not acting on μ-receptors under these conditions. In animals tolerant to systemic morphine, however, TIPPψ at a much lower dose (0.2 or 2.0 nmol/mouse) enhanced the antinociceptive effect of DAMGO about 3-fold in both strains (Fig.4). In the H-ND4 strain, a dose of 2.0 nmol of TIPPψ was sufficient to completely restore the antinociceptive potency of DAMGO (same AD50 as in naı̈ve mice), whereas in the CR-SW strains this dose of TIPPψ partially restored the antinociceptive effect.
These observations suggested to us that in spinal cord of the tolerant CR-SW and H-ND4 mice, the increased AD50 of DAMGO resulted from inhibition by δ-receptors. By removing this inhibition, TIPPψ would increase the antinociceptive potency of DAMGO. In our previous study with H-SW mice, we observed that tolerance in these animals was associated with a loss of potentiation of μ-receptors by δ-receptors that was present in the naı̈ve animals (He and Lee, 1998). To determine whether a similar potentiation occurred in the naı̈ve CR-SW and H-ND4 mice, we compared the AD50 of i.t. DAMGO in the presence and absence of low doses of DPDPE. As shown in Fig. 5, DPDPE significantly enhanced DAMGO antinociception under these conditions, lowering the AD50 4- to 6-fold in each strain of mouse. Furthermore, this potentiation was blocked completely by addition of TIPPψ (30 nmol), a dose that had no effect on DAMGO alone (Fig. 3). Taken together, these results indicate that in the naı̈ve CR-SW and H-ND4 mice, as in the naı̈ve H-SW mice, δ-receptor activation potentiates μ-receptor-mediated antinociception. This potentiation is lost during tolerance in all three strains, becoming in fact an inhibition in CR-SW and H-ND4.
Tolerance to the antinociceptive effect of morphine is a very well established phenomenon in most mammals. When the morphine is given systemically, via repeated s.c. injections or implantation of a morphine pellet, substantial tolerance develops within 3 days. In contrast, tolerance develops more slowly to morphine at localized spinal sites (Yaksh et al., 1977; Pfeifer et al., 1989; Stevens and Yaksh, 1989). We have confirmed this finding in six strains of mice differing in their degree of tolerance to s.c. morphine, but also showed that tolerance does develop in most of these strains to the more selective agonists DAMGO and DPDPE. Three of the six strains developed a relatively low degree of tolerance to s.c. morphine (2- to 4-fold), and three a relatively high degree (8- to 10-fold). But only one of the chronically treated strains was tolerant to morphine (i.t.) after 3-day pellet implantation, and the tolerance was barely significant. Yoburn et al. (1990) also reported a rapid development of morphine (i.t.) tolerance to morphine (s.c.), so there may be some strain dependence of this phenomenon. However, the tolerance protocol is also critical.Roerig et al. (1991) report that if the morphine pellet is removed, tolerance to spinally administered morphine develops rapidly.
However, tolerance to the i.t. administration of the δ-selective agonist DPDPE was observed in all six strains, and tolerance to the μ-selective agonist DAMGO was observed in four of the strains (Table2). This tolerance was quite substantial, although it did not correlate with the degree of s.c. tolerance to morphine in these strains. However, it confirms previous reports indicating that more selective opioid ligands develop tolerance to morphine when administered to spinal sites (Russell et al., 1986; Porreca et al., 1987; Roerig et al., 1991).
How could tolerance develop to these selective agonists but not to morphine? To address this question, we first demonstrated that the lack of spinal tolerance to morphine was a general phenomenon, found in several different strains of mice. We then used selective antagonists to confirm the type of opioid receptors these agonists were interacting with. In a previous study using H-SW mice (He and Lee, 1998), we reported that DPDPE-induced antinociception in the spinal cord could be blocked by CTAP, a μ-selective antagonist, even more effectively than by naltrindole, a δ-selective antagonist. These and other results led to our conclusion that DPDPE-induced antinociception was mediated directly through μ-receptors. We repeated and extended these findings in this study. We showed that CTAP was equally effective in antagonizing the action of both agonists in two of the strains of animals used in this study, H-ND4 and CR-SW. Furthermore, the pA2 values for CTAP against DPDPE were similar to those for its antagonism of DAMGO (Table 3). This indicates that DPDPE is acting at the μ-receptor in the spinal cord of these strains, and suggests that its μ-action is a general effect. These results also demonstrate that tolerance to spinal μ-receptors develops in these strains of mice, even though tolerance to morphine is not observed in the spinal cord.
We then showed that although DPDPE mediates antinociception in the spinal cord through μ-receptors, it has a second action, potentiating the action of μ-agonists such as DAMGO (Fig. 5). This potentiation is mediated through δ-receptors, because it was blocked by the selective δ-antagonist TIPPψ (Fig. 5). In the tolerant animals, however, TIPPψ actually enhanced the antinociceptive potency of DAMGO, increasing it to the same level, or close to the same level, as found in naı̈ve animals. Again, this interaction was found in both the high-tolerance and low-tolerance strain of mice tested.
To summarize, our results show that although both the μ-agonist DAMGO and the δ-agonist DPDPE mediate antinociception through μ-receptors in the spinal cord, DPDPE also mediates an interaction between μ- and δ-receptors. In the naı̈ve animal, this interaction is a potentiating one, enhancing DAMGO antinociception, whereas in the tolerant animal, this interaction is an inhibitory one. Although some previous studies, including our own, have demonstrated a potentiation of μ-receptors by δ-receptors at spinal or supraspinal sites (Larson et al., 1980; Porreca et al., 1987; Malmberg and Yaksh, 1992;He and Lee, 1998), to our knowledge this is the first report of an inhibitory interaction in the tolerant animal. However, this observation is consistent with the reports of two other groups that tolerance to morphine is blocked by treatment with δ-antagonist (Abdelhamid et al., 1991; Fundytus et al., 1995), and with a recent report that mice in which the δ-opioid receptor was eliminated by homologous recombination did not develop tolerance to morphine (Zhu et al., 1999).
The original impetus for this study was to understand why mice treated chronically with morphine rapidly develop tolerance to s.c. morphine but not to i.t. morphine (Table 1). Our results provide one possible explanation for this. As a nonselective agonist, morphine interacts with both μ- and δ-receptors. However, previous studies have shown that its action in systems that contain a pure population of δ-receptors, such as NG108-15 cells, is as a partial agonist (Law et al., 1983). Thus, it can have an antagonistic effect on δ-receptors at certain concentrations. It may in this way block the inhibitory action of spinal δ-receptors on μ-receptors in the to systemically tolerant animal, preventing the development of spinal tolerance.
The studies reported here do not imply any particular mechanism of μ-δ interaction. An early model proposed that the two receptors were physically associated (Rothman and Westfall, 1982), and very recently it has been reported that μ- and δ-receptors (George et al., 2000) as well as δ- and κ-receptors (Jordan and Devi, 1999), may form heterodimers that have ligand-binding properties different from pure populations of either receptor. Therefore, one possible mechanism of tolerance is that spinal δ-receptors physically associate with μ-receptors, resulting in a reduction in affinity of agonist for μ-receptor. However, the evidence for receptor affinity changes as the basis of tolerance is controversial. Many studies have reported no change in μ-receptor affinity after chronic treatment with morphine or other μ-agonist (Hollt et al., 1975; Rogers and el-Fakahany, 1986; Tao et al., 1987, 1990; Besse et al., 1992;Polastron et al., 1994), whereas a few studies have reported either increased (Abdelhamid and Takemori, 1991) or decreased (Rothman et al., 1991) affinity. However, most of this work was carried out on brain tissue, rather than spinal cord.
An alternative possibility, equally consistent with our data, is an interaction downstream from the receptor. For example, δ-receptor activation, in the highly tolerant animal, could stimulate a pathway that antagonizes at some point a pathway stimulated by μ-receptor activation. This kind of inhibition would not be accompanied by any detectable changes in opioid receptor number or affinity.
A second question concerns the change that promotes the altered μ-δ interaction. One possibility is that during tolerance, a change occurs in the conformation of either the μ-receptor and/or the δ-receptor, altering their physical interaction. Such a change could occur, for example, as a result of modification of the receptor by phosphorylation (Bernstein and Welch, 1998). An alternative possibility is that the change results from an up-regulation of an endogenous δ-system. The activated δ-system then inhibits the μ-receptor indirectly, through second messenger interaction. A role for an endogenous agonist in μ-opioid receptor action can also be inferred from the results of studies with μ-opioid receptor “knockout” mice, which have a greater sensitivity to thermal pain than normal animals (Sora et al., 1997). It has also been reported that δ-receptors in brain are up-regulated by chronic morphine treatment (Abdelhamid and Takemori, 1991).
It must be emphasized, however, that our data draw a clear-cut distinction between tolerance at the spinal level and systemic tolerance. Only one of the six strains developed tolerance to i.t. morphine, and although all strains did develop significant tolerance to i.t. DPDPE, and four of the strains to i.t. DAMGO, we did not see a correlation between degree of i.t. tolerance to either agonist, on the one hand, and s.c. tolerance to morphine, on the other. There is considerable evidence that spinal and supraspinal opioid receptors can interact synergistically to induce antinociception, and that opioid tolerance may result from an alteration of this synergism (Yeung and Rudy, 1980; Roerig and Fujimoto, 1988; Roerig et al., 1984). The endogenous opioid dynorphin seems to play an important role in this system, because dynorphin can partially restore synergism in tolerant animals (He and Lee, 1997). Our work here, on the other hand, suggests a role, at the spinal level, for an endogenous δ-system. By modulating δ-receptor inhibition of μ-receptors, this system could contribute to the spinal component of spinal-supraspinal synergism.
This work was supported in part by National Institute on Drug Abuse Grants DA02643 and DA10.
- Received July 30, 2001.
- Accepted September 25, 2001.
- The American Society for Pharmacology and Experimental Therapeutics