Chronic Antidepressant Treatment Causes a Selective Reduction of μ-Opioid Receptor Binding and Functional Coupling to G Proteins in the Amygdala of Fawn-Hooded Rats

The endogenous opioid system in the brain plays important roles not only in pain relief but also in regulation or modulation of the reward pathway, suggestive of their opioid involvement in acquisition and/or maintenance of addiction to abused drugs (Koob et al., 1998). Apart from a direct action by opioid-related substances such as heroin, the opioid system is also affected by other substances, such as alcohol (Froehlich, 1996; Winkler et al., 1998). Animal experiments comparing alcohol-preferring rodents against their alcoholnonpreferring counterparts suggest that innate changes of endogenous opioid peptides and opioid receptors in certain brain regions might predispose these rodents to consuming alcohol voluntarily (de Waele et al., 1995; Cowen et al., 1998). In addition, clinical trials and application of opioid receptor antagonists in the treatment of human alcoholics (Volpicelli et al., 1995) further substantiated involvement of this system in the self-administration of alcohol.

Surprisingly, a recent survey showed that rather than opioid receptor antagonists, antidepressants (ATDs) are still the leading prescribed medicine in the treatment of alcoholics in the United States (Mark et al., 2003). Apart from reflecting the fact that alcoholism is a polymodal disorder, the survey may also reflect some other facts: first, there may be a high comorbidity between depression and drug addiction; and second, there may be neurochemical similarities in depression and drug dependence (Markou et al., 1998). Targeting multiple neurochemical mechanisms such as serotonin and dopamine systems in the brain mesolimbic region is hypothesized to be part of the mechanism underlying the efficacy of many conventional ATDs, including tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) in the treatment of depression and alcoholism (Weiss and Mirin, 1989; Pettinati et al., 2000; Chen and Lawrence, 2003). However, whether targeting the opioid system is implicated in mechanisms underlying the above-mentioned efficacies of ATDs is less known and has been less extensively explored despite that an association between opioids and depression has been well established.

A series of studies carried out by our laboratory indicate that the opioid system may also be involved in self-administration of alcohol by Fawn-Hooded (FH) rats, a strain of rat exhibiting high alcohol preference and a phenotype of depression (Rezvani et al., 2002). For example, there is an alteration of endogenous opioid peptide precursor mRNA in brain areas in alcohol-naive FH rat compared with alcohol-nonpreferring, depressive Wistar-Kyoto rats (Cowen et al., 1998, 1999a). Opioid peptide precursor genes and μ-opioid receptors in the nucleus accumbens and amygdala were selectively changed by either chronic alcohol consumption or by alcohol withdrawal (Chen and Lawrence, 2000; Cowen and Lawrence, 2001; Djouma and Lawrence, 2002), suggesting that μ-opioid receptors in these regions are dynamically involved in alcohol consumption in FH rats. In addition, volitional alcohol consumption by FH rats was significantly attenuated by the opioid receptor antagonist naltrexone, although tolerance did develop to this in concert with up-regulation of μ-opioid receptors (Cowen et al., 1999b).

The rationale to investigate the role of the opioid system in the efficacy of ATDs to treat alcoholics is first based on an interaction between the ATD and the rewarding process in which potentiation of opioid-induced conditioned place preference by fluoxetine, an SSRI, was obtained (Subhan et al., 2000); and second, there is a distinct opioid receptor adaptation to chronic treatment of ATDs (Vilpoux et al., 2002). As such, the aim of the present study is to investigate whether there is any change of μ-opioid receptor binding and functional coupling to G proteins in brain mesolimbic areas in FH rats after chronic treatment with selected ATDs, using a paradigm in which these ATDs were previously shown to be effective to reduce alcohol consumption (Rezvani et al., 1991; our unpublished data). To do this, we treated FH rats with either desipramine, a TCA, or sertraline, an SSRI, for 10 days before the neurochemical study. By using quantitative autoradiography, μ-opioid receptors labeled with [¹²⁵I]d-Ala²,N-Me-Phe⁴,Met(O)⁵-ol enkephalin (FK33,824) and μ-opioid receptor-mediated G protein coupling labeled by agonist-stimulated [³⁵S]GTPγS binding (Chen and Lawrence, 2000) were mapped in brain sections collected throughout the mesolimbic system.

Materials and Methods

All experiments were performed in accordance with the Prevention of Cruelty to Animals Acts 1986 under the guidelines of the Code of Practice for the Care and Use of Animals for Experimental Purposes in Australia.

Materials. [³⁵S]GTPγS (1250 Ci/mmol) was purchased from PerkinElmer Life and Analytical Sciences (Boston, MA). FK33,824 was obtained from Bachem (Bubendorf, Switzerland). [¹²⁵I]Na (2000 Ci/mmol) was purchased from Amersham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK), and iodogen was obtained from Pierce Chemical (Rockford, IL). GDP, GTP, d-Ala²,N-Me-Phe⁴-Gly⁵-ol-enkephalin (DAMGO), and naloxone HCl were obtained from Sigma-Aldrich (St. Louis, MO). [¹⁴C] Microscales were purchased from ARC (St. Louis, MO), and Kodak Biomax AR films were obtained from Eastman Kodak (Rochester, NY). All other reagents were of either analytical or laboratory grade from various suppliers.

Antidepressant Administration. Due to different requirements of brain tissue treatment for [¹²⁵I]FK33,834 binding and [³⁵S]GTPγS-labeled functional coupling to G proteins, two sets of experiments were performed in FH rats (stock parents were obtained from UNC, Chapel Hill, NC). In the first experiment, 14 male FH rats (300–350 g) were assigned into three groups: vehicle (n = 4), desipramine (DMI) (n = 5), and sertraline (n = 5). In the second experiment, 15 male FH rats (300–350 g) were allocated into the same three categorized groups with five rats in each group. The rats in these three groups were treated daily for 10 days by i.p. injection with vehicle (5% DMSO; 1 ml/kg/day) or DMI or sertraline (both 4 mg/kg/day). All rats were housed in groups with a 12-h light/dark cycle and had free access to standard chow and tap water. Six hours after the last injection of drug or vehicle, rats were killed by cervical dislocation and decapitation. The rat brains were quickly removed and frozen over liquid nitrogen for experiment 1 or over supercooled isopentane at –35°C for experiment 2. The frozen brains were stored at -80°C until cutting.

Autoradiography of [¹²⁵I]FK33,824 Binding to μ-Opioid Receptors. Brain sections (14 μm) were cut on a cryostat (Cryocut 1800; Leica Microsystems, Inc., Deerfield, IL) at -18°C at the levels of the nucleus accumbens (NAcc) (bregma, 2.2 to 1.6 mm), the ventral pallidum (VP) (bregma, 0.7 to 0.1 mm), and the ventral tegmental area (VTA) (bregma, -4.5 to -3.9 mm) (Paxinos and Watson, 1986) and stored at -80°C until use. [¹²⁵I]FK33,824, a selective μ-opioid receptor agonist, was used to bind μ-opioid receptors in the present study and the protocol for autoradiographic assay of μ-opioid receptors was as used previously in our laboratory (Cowen et al., 1999b; Djouma and Lawrence, 2002). Briefly, FK33,824 was radio-iodinated and purified by reverse phase chromatography using a phenyl cartridge (Varian, Inc., Palo Alto, CA). To determine μ-opioid receptor binding, slides were equilibrated to room temperature and incubated in 50 mM Tris-HCl (pH 7.4) containing 0.1% bovine serum albumin and 0.1 nM [¹²⁵I]FK33,824 (2000 Ci/mmol) for 60 min at room temperature. Nonspecific binding was determined in the presence 10 μM naloxone. At the end of the incubation period, slides were subjected to 3 × 4-min washes in ice-cold 50 mM Tris-HCl and a dip in distilled water. Slides were then dried under a stream of cold air, desiccated overnight, and then apposed to Biomax AR film (Eastman Kodak) in the presence of [¹⁴C] standard microscales for 8 days before developing.

Autoradiography of Agonist-Stimulated [³⁵S]GTPγS Binding to G Proteins. Brain sections (20 μm) at the level of the nucleus accumbens (bregma, 1.7 to 1.3 mm) and the hippocampus/amygdala (bregma, -2.8 to -2.5 mm) (Paxinos and Watson, 1986) were cut on a cryostat at -20°C and thaw-mounted onto gelatin-chrome alumcoated slides. Slides were dried under vacuum and stored desiccated at -80°C until use. The protocol used for DAMGO-stimulated [³⁵S]GTPγS binding in a previous study (Chen and Lawrence, 2000) was adopted in the present study. Briefly, the brain sections were preincubated in 50 mM Tris-HCl buffer (pH 7.4) containing 3 mM MgCl₂, 0.2 mM EGTA, 100 mM NaCl at 25°C for 10 min, and then exposed to the same buffer with addition of 2 mM GDP at 25°C for 15 min. Thereafter, slides were transferred to the same buffer as used for preincubation but containing 2 mM GDP and 0.04 nM [³⁵S]GTPγS either in the absence (basal) or presence of 10 μM DAMGO (stimulation) at 25°C for 2 h. The nonspecific [³⁵S]GTPγS binding was defined in the presence of 10 μM unlabeled GTPγS in the same incubation medium as used for stimulation study but without GDP. The incubation was stopped by two consecutive washes in ice-cold 50 mM Tris-HCl buffer (pH 7.4) and rinsed briefly in ice-cold deionized water. Slides were dried under a gentle stream of cool air and kept in the desiccant-filled container overnight. Dried sections were apposed to Biomax AR film (Eastman Kodak) in the presence of [¹⁴C] standard microscales for 72 h.

Data Analysis. Autoradiographic images on developed films were subsequently quantified, using a Microcomputer Imaging Device M4 image analysis (Imaging Research, St. Catherines, ON, Canada), by comparison of optical density, under constant illumination, of the autoradiograms compared with the [¹⁴C] standard microscales (Chen and Lawrence, 2000). The optical density of [¹²⁵I]FK33,824 binding as well as the basal and agonist-stimulated [³⁵S]GTPγS binding was expressed in dpm per square millimeter of targeted nucleus and was measured from four consecutive sections in each region in each rat. The results are also expressed as the net percentage of increase of agonist-stimulated [³⁵S]GTPγS binding over the basal. Unless otherwise indicated, data are reported as mean value ± S.E.M.

Statistics. The statistics software program SigmaStat (SPSS Inc., Chicago, IL) was used throughout. One-way analysis of variance (ANOVA) followed by Dunnett's test or Kruskal-Wallis ANOVA on ranks followed by Dunn's test was performed for the comparison of radioligand binding parameters between DMI-treated or sertraline-treated groups and their vehicle control. The comparison of difference between basal versus stimulated [³⁵S]GTPγS binding, as well as differences between treatment groups was determined by a two-way ANOVA followed by Bonferroni t test. A one-way ANOVA was also used to compare the percentage of net increase of [³⁵S]GTPγS binding between the different treatment group and the control. A significance level of P < 0.05 was used throughout.

Results

Effects of Treatment with DMI or Sertraline on [¹²⁵I]FK33,824 Binding to μ-Opioid Receptors. A wide-spread distribution of specific [¹²⁵I]FK33,824 binding (>95% of total binding) to μ-opioid receptors was found at all levels in brain sections from all three treatment groups (Fig. 1). High binding was observed in the caudate-putamen (CPu) (in patches), fundus striatum (FStr), dentate gyrus, dorsomedial geniculate nucleus, and olivary pretectal nucleus. In addition, there was intermediate binding in the amygdaloid complex, the frontal and parietal cortex [Cx(FP)], lateral (LS) and medial septum, VP, and the VTA. After 10 days of consecutive drug treatment, binding of [¹²⁵I]FK33,824 in the amygdala, hippocampal CA1 and CA3 areas, the occipital and temporal cortex [Cx(OT)], dorsomedial geniculate nucleus, olivary pretectal nucleus, olfactory tubercle, and superior colliculus were significantly reduced by DMI (-11 to -27%) and sertraline (-29 to -50%) compared with the vehicle control (Fig. 2). In addition, a smaller but statistically significant reduction of [¹²⁵I]FK33,824 binding was found in the dentate gyrus, NAcc, and VTA.

Effects of Treatment with DMI or Sertraline on DAMGO-Stimulated [³⁵S]GTPγS Binding to G Proteins. The topographic distribution of μ-opioid receptor mediated-[³⁵S]GTPγS binding stimulated by DAMGO at the level of the NAcc and hippocampus has been previously documented (Chen and Lawrence, 2000) and was similar in the present study. Figure 3 shows representative autoradiograms of [³⁵S]GTPγS binding at the level of the NAcc and the hippocampus, which exhibited a similar pattern of binding between the three treatment groups. Briefly, a statistically significant increase of DAMGO-stimulated [³⁵S]GTPγS binding over the basal was observed in all brain regions across three treatment groups (Fig. 4). The net increase of DAMGO-stimulated [³⁵S]GTPγS binding in the amygdala, which reached +60% over the basal activity of vehicle control, was significantly decreased by -25% in both DMI- and sertralinetreated rats (Table 1). In contrast, the [³⁵S]GTPγS binding in other brain regions was not significantly affected by either antidepressant. In addition, in the sertraline-treated group there was no alteration of basal [³⁵S]GTPγS binding compared with the vehicle-treated group, whereas the basal [³⁵S]GTPγS binding in several regions, including the CPu, the cingulate cortex, Cx(FP), and NAcc was significantly elevated in DMI-treated rats (Fig. 4).

Discussion

Based on association of the opioid system in volitional alcohol consumption in FH rats (Cowen et al., 1998, 1999b; Chen and Lawrence, 2000; Djouma and Lawrence, 2002) and the efficacy of ATDs in reducing ethanol consumption in this strain (Rezvani et al., 1991; our unpublished data), the present study investigated the effect of desipramine and sertraline on μ-opioid receptors in FH rat brain. After chronic drug treatment, binding of [¹²⁵I]FK33,824 to μ-opioid receptors was reduced across many mesolimbic nuclei. In contrast, the coupling efficiency of μ-opioid receptors to G proteins was reduced only in the amygdala. Functional alteration of μ-opioid receptors induced by chronic antidepressant treatment in the amygdala, a brain nucleus associated with learning, emotion, and reward (Baxter and Murray, 2002), may relate to the behavioral effects of these drugs in FH rats.

[¹²⁵I]FK33,824 predominantly labels the high-affinity state of μ-opioid receptors (Cowen et al., 1999b; Djouma and Lawrence, 2002). In this study, significant attenuation of [¹²⁵]IFK33,824 binding was induced by treatment with desipramine or sertraline across many brain regions; however, sertraline exerted more robust effects than desipramine. The attenuation seems region-specific because [¹²⁵I]FK33,824 binding was unchanged in the FStr, the nucleus of the horizontal limb of the diagonal band, medial septum, and VP. The reduction of binding to cortical μ-opioid receptors without change in the caudate-putamen by desipramine agrees with a previous study (Reisine and Soubrie, 1982). Bidirectional alteration of μ-opioid receptors was reported from a study examining [³H]naloxone binding in rat brain after treatment with paroxetine or reboxetine. Thus, paroxetine increased μ-opioid receptor binding in cingulate and insular cortices, dorsal endopiriform nucleus (4 days), and olfactory tubercle (21 days), but it decreased binding in thalamus (21 days). Reboxetine increased binding in the amygdala (4 days) and hippocampus and thalamus (21 days), but it decreased binding in the dorsal raphe (4 days) (Vilpoux et al., 2002). Apparently, the adaptation of μ-opioid receptors to ATD treatment showed time dependence and drug dependence; however, other factors, including radioligand, strain of rat, and time chosen to collect brains, could also contribute.

It is noteworthy that the time (6 h after the last drug injection) we chose to collect the brains was when the concentration of drug might reach a pharmacological range in the brain (Hrdina and Dubas, 1981; Fuller et al., 1995) or the pharmacological effect could be manifested in the treated rats (Chen and Lawrence, 2003). A major reason for choosing the present sampling time is based on consideration that μ-opioid receptors may undergo a second adaptation (withdrawal effect) after long-term treatment with ATD, because we knew this receptor system showed a dynamic response to alcohol withdrawal in FH rats (Chen and Lawrence, 2000; Djouma and Lawrence, 2002). However, the interpretation of any alteration of μ-opioid receptor binding must be taken cautiously due to potential interactions between μ-opioid receptors and TCAs or SSRIs (Wahlstrom et al., 1994), even though such an interaction has been questioned (Wong et al., 1985). Lines of evidence do, however, support that the decrease of μ-opioid receptor binding from the present finding may reflect a down-regulation of μ-opioid binding sites. First, a differential and region-specific attenuation of μ-opioid receptor binding was obtained in brain regions where a similar drug distribution in these areas has been demonstrated (Hrdina and Dubas, 1981; Tremaine et al., 1989). Second, there was no match of the reported drug distribution and the decrease of functional coupling of μ-opioid receptors to G proteins. Finally, pharmacological down-regulation of μ-opioid receptors by a drug-induced elevation of endogenous opioids, such as enkephalin, in specific brain regions is possible (De Felipe et al., 1985). Therefore, the region-specific decrease of μ-opioid receptors is likely to be a down-regulation within the neural network or an indirect alteration of μ-opioid receptor binding derived from a conformational change of the receptor by an interaction of the ATDs with the nearby membrane of the receptor (Somoza et al., 1981). Indeed, the apparent down-regulation of binding in the present study may represent drug-induced heterodimerization of μ-opioid receptor subunits with other receptors, the affinity of [¹²⁵I]FK33,824 being reduced in the heterodimer compared with the native receptor. The ability of opioid receptors to exist as heterodimers is well established (Jordan and Devi, 1999; George et al., 2000). One way to clarify whether changes in binding reflect changes in protein would be to replicate the present study quantifying μ-opioid receptors by immunohistochemical approaches. In a similar manner, the use of antagonist radioligands would be less susceptible to drug-induced changes in the conformational states of receptors; however, it is important to note that one of the reasons for choosing an agonist radioligand was to facilitate comparison with agonist-stimulated binding of [³⁵S]GTPγS.

Theoretically, changes of μ-opioid receptors should affect the coupling efficiency of the receptor to G proteins, quantifiable by agonist-stimulated [³⁵S]GTPγS binding. With the exception of the amygdala, however, this proved not the case from the present study, despite that alterations of μ-opioid receptor binding were observed in many brain nuclei. It is not clear at this stage what mechanisms underlie the mismatch of decreased μ-opioid receptors and the lack of change in stimulated [³⁵S]GTPγS binding. Notwithstanding that possibilities including differential receptor reserves and the ratios of receptor over the G proteins could play a role, the existence of G protein-dependent and -independent down-regulation of μ-opioid receptors might contribute to disassociation of parallel changes of receptor binding and G protein coupling (Pak et al., 1999). The net increase of stimulated [³⁵S]GTPγS binding can also be affected by the basal binding to the radioligand. Previous studies demonstrated chronic desipramine treatment could facilitate G protein-activated adenylyl cyclase that is mediated by G_s subunits without changing the levels of G_s in some cell lines (Chen and Rasenick, 1995). In our study, such an influence on G_s coupling might reflect by a region-specific elevation of basal binding of [³⁵S]GTPγS in striatum and cortex. This increase of basal [³⁵S]GTPγS binding mediated by G_s subunits could be activated via increased endogenous adenosine by desipramine (Moore et al., 2000).

Whereas a direct interaction of TCAs on G_o subunits was reported (Yamamoto et al., 1992), most studies still support that ATDs are less likely to directly affect the levels of G_i/G_o in the form of protein or mRNA (Chen and Rasenick, 1995). The present study further substantiated no direct effect of ATDs on the G_o/G_i system, especially on G_i/G_o-mediated coupling to μ-opioid receptors because there was no drug distribution-related or μ-opioid receptor alteration-related changes of G protein coupling, except the amygdala. As such, the mechanism underlying the decrease of the receptor-mediated functional coupling to G proteins in the amygdala could be due to a genuine down-regulation of μ-opioid receptors in this region. In contrast, the ability of DAMGO to stimulate [³⁵S]GTPγS binding is unchanged in other nuclei, despite an apparent down-regulation of receptors. It would therefore seem that there is possibly a difference in the mechanism behind the reduced binding of [¹²⁵I]FK33,824 in the amygdala compared with other structures, such that the reduced binding in regions where coupling is not affected may reflect conformational changes caused by drug-induced heterodimerization. For example, DAMGO can act as an agonist at heteromeric μ-opioid/somatostatin receptors (Pfeiffer et al., 2002) and therefore, although the affinity of [¹²⁵I]FK33,824 binding to such a receptor may be reduced, the DAMGO-stimulated binding of [³⁵S]GTPγS may be unchanged. Clearly, further studies are required to clarify whether this is the case; however, it remains a feasible explanation for the present data. Also of importance would be to determine DAMGO-stimulated binding over a range of agonist concentrations in addition to the presumably supramaximal concentration used in the present study. It is clearly possible that more subtle alterations in receptor coupling may be unmasked under these conditions.

It is interesting that the amygdala was specifically and uniquely affected by desipramine or sertraline at the levels of μ-opioid receptor binding and its functional coupling to G proteins, which could be part of the mechanism for ATDs to improve behavior and/or reduce ethanol consumption in FH rats. The amygdala has been implicated in the modulation of memory storage and emotional aspects of behavior, including fear and anxiety, as well as states associated with consummatory responses or in the formation of stimulus reinforcement associations (Baxter and Murray, 2002). Although it is not clear what the net effect of the decrease of the μ-opioid receptor function is in the amygdala or in brain areas innervated such as the NAcc (Yu and Han, 1990), indirect evidence suggests that inhibition of μ-opioid receptor function in this region may reduce the anxiolytic effect of ethanol (Wilson et al., 2003), via modulation of GABAergic neurotransmission in the amygdala (Hyytiä and Koob, 1995).

In summary, the present data demonstrate that chronic antidepressant treatment in FH rats can regulate μ-opioid receptors, and this regulation seems to be functionally relevant within the amygdala. As such, the opioid system within the amygdala may represent a locus of action for some of the therapeutic effects of antidepressant medication.