Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The cellular basis of tolerance to, and dependence upon, many classes of drugs, including opioids, has long defied identification. Tolerance to opioids cannot be explained on the basis of altered metabolism or disposition of the drug (Johnson and Fleming, 1989). Rather, tolerance must be the consequence of changes in some aspect of the function of the cells upon which morphine acts. The concept that tolerance and dependence upon opioids are expressions of individual neuronal adaptation was advanced by Collier (1965, 1966) and has been particularly stressed in recent reviews (Koob and Bloom, 1988; Johnson and Fleming, 1989; Nestler, 1992; Nestler et al., 1993; Fleming and Taylor, 1995).

Adaptive supersensitivity/subsensitivity is a manifestation of a cellular homeostatic mechanism by which a variety of types of excitable cells, including neurons, compensate for chronic changes in the net stimulus they receive (Fleming and Westfall, 1988). When the net change is in the direction of inhibition or decreased activity, the cells become more sensitive to stimuli and/or less sensitive to inhibition, and vice versa. The prolonged inhibition of neurons by a chronically administered opioid agonist produces just such a decrease in neuronal activity (for reviews, see Johnson and Fleming, 1989; Fleming and Taylor, 1995). These sensitivity changes appear and disappear gradually over periods of several days and should not be confused with the rapidly developing and disappearing phenomenon of receptor desensitization. Alterations in three different cellular functions have been associated with chronic adaptive super- and subsensitivity, i.e., 1) changes in receptor density, 2) changes in membrane potential and the Na⁺,K⁺-pump and 3) changes in the adenylyl cyclase transduction system. All three have been firmly established in non-neural tissue and have been given extensive attention in studies of opioid tolerance and dependence in neurons (for reviews, see Fleming and Westfall, 1988; Johnson and Fleming, 1989; Fleming and Taylor, 1995). Reviews of the literature have led to the conclusion that an association between changes in density or affinity of opioid receptors and tolerance or dependence is unlikely (Loh et al., 1988; Johnson and Fleming, 1989).

In every instance in which adequate data have been obtained, adaptive changes in sensitivity to agonists acting through separate receptors and transduction systems (nonspecific supersensitivity and subsensitivity) have been associated with partial depolarization of the cell membrane. The tissues in which this association has been established include skeletal muscle (Fleming, 1971; Sellin and Thesleff, 1980), smooth muscle of the guinea pig vas deferens (Fleming and Westfall, 1975; Hershman et al., 1995), rabbit saphenous artery (Abel et al., 1981), canine colon (Rogers et al., 1993) and neurons of the myenteric plexus (Taylor et al., 1988; Leedham et al., 1992).

Rarely have investigators considered the possibility that tolerance of neurons to opioids might be accompanied by altered sensitivity to nonopioid agonists. The importance of the issue is highlighted by two series of studies that have thoroughly examined the issue, with quite different outcomes.

Implantation of morphine pellets in rats readily induces both tolerance and dependence in LC neurons, which can be demonstrated either in vivo or in slices (Aghajanian, 1978; Andrade et al., 1983; Kogan et al., 1992). The tolerance induced is specific for opioids; that is, it is accompanied neither by subsensitivity to the alpha-2 adrenoceptor agonist clonidine nor by supersensitivity to the stimulatory effect of glutamate. As a consequence, the mechanistic studies have concentrated upon µ-receptor-coupled transduction processes, particularly the involvement of G proteins and cyclic AMP-dependent protein kinase (Duman et al., 1988; Nestler and Tallman, 1988; Nestler et al., 1989; Guitart and Nestler, 1989). Based on these data, Nestler (1992) proposed that an up-regulated cyclic AMP system represents a compensatory response of rat LC neurons to chronic opioid-induced inhibition.

In contrast, the chronic implantation of morphine pellets in guinea pigs has been shown to produce, simultaneously, subsensitivity of the isolated LM/MP preparation to the inhibitory effects not only of morphine but also of the adenosine receptor agonist 2-chloroadenosine and the alpha-2 adrenoceptor agonists clonidine and xylazine (Taylor et al., 1988) and supersensitivity to the excitatory effects of nicotine, 5-hydroxytryptamine and potassium chloride (Goldstein and Schulz, 1973; Johnson et al., 1978). This general change in excitability is accounted for by a partial depolarization of the resting membrane potential of S-neurons in morphine-tolerant myenteric ganglia, with no change in the action potential threshold (Leedham et al., 1992). Electrophysiological experiments further indicated that opioid receptor interactions or associated transduction processes were unchanged from control (Leedham et al., 1992).

The striking differences in the results for the rat LC and neurons of the guinea pig LM/MP suggest basic differences in the mechanisms of the development of tolerance/dependence either among different populations of neurons or between species. Presented here is the first report of nonspecific adaptation of central neurons induced by chronic in vivo exposure to morphine. The nTS was chosen for two reasons, as follows: 1) it is the first medullary brainstem relay in the processing of visceroceptive impulses transmitted by primary afferent fibers, including those from the myenteric plexus, and 2) its neurons display spontaneous firing in slice preparations. In particular, neurons of the mnTS were selected for study because these neurons receive the primary sensory input from the gastrointestinal tract. In the absence of information on the pharmacology of guinea pig nTS (in contrast to rat nTS), the study began with an investigation of the response of neurons in control slices to several different agonists. Some preliminary results of these experiments have been presented in abstracts (Malanga et al., 1993a,b, 1994, 1995).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Male albino guinea pigs weighing 200 to 300 g (Hilltop Laboratory Animals, Inc., Scottsdale, PA) were implanted with morphine or placebo pellets (75 mg/pellet s.c.; courtesy of K. H. Davis, Research Triangle Institute, Research Triangle Park, NC), at two pellets/animal, under Innovar anesthesia (0.05 mg/ml fentanyl citrate, 2.5 mg/ml droperidol; 0.15 ml/100 g b.wt. s.c.; Janssen Pharmaceutica, Piscataway, NJ) as previously described (Johnson et al., 1978; Taylor et al., 1988; Leedham et al., 1992). For animals of this age and weight, two pellets were found to be preferable to the four described in previous studies for larger (350-600 g) animals; two pellets in the smaller animals resulted in similar shifts of the dose-response relationship for morphine, 2-chloroadenosine and clonidine in the LM/MP organ-bath preparation. After 7 days, the animals were stunned and decapitated. After craniotomy, the brainstems, from approximately C₁ caudally to the midcollicular level rostrally, were removed and sliced with a vibrating tissue slicer (WPI, New Haven, CT), under ice-cold (4°C) aCSF of the following composition (in mM): NaCl, 126; KCl, 5.0; CaCl₂, 2.4; NaHCO₃, 26; NaH₂PO₄, 1.2; MgSO₄, 1.3; dextrose, 10, pH 7.4; bubbled with 95% O₂/5% CO₂ (carbogen).

Three to five 400-µm slices were taken from each brainstem, at and around (approximately 1 mm caudal and rostral to) the level of the obex, an area including the mnTS. This region was chosen for several reasons. 1) The mnTS in other species contains extensive sensory fibers from the gastrointestinal tract (Satomi et al., 1978; Kalia and Mesulam, 1980) and, therefore, is in the reflex arc that includes myenteric S-neurons. 2) As in other species, the spontaneously active neurons in the guinea pig mnTS were found to have a relatively characteristic firing pattern of slow regular action potentials generated at a frequency of 0.5 to 5 Hz. In contrast, neurons of the ventrolateral nTS, an area associated with respiration, tend to fire intermittently and in a bursting pattern. 3) Experiments, again in other species (Kalia et al., 1984), determined that the highest level of binding of antibodies to Met-enkephalin in the nTS occurred in the mnTS, indicating considerable opioid input to those neurons. Slices were maintained at room temperature (20°C) in carbogenated aCSF for at least 30 min, transferred to a submersion-type perfusion chamber and equilibrated in carbogenated aCSF at 32°C (2 ml/min, gravity-fed) for 30 min before recording.

The nTS in the brainstem slice was visually identified under a dissecting microscope. Neuronal action potentials were recorded extracellularly using glass microelectrodes (10-40 M) filled with 3 M NaCl, amplified with an Axoprobe 1A microelectrode amplifier (Axon Instruments, Burlingame, CA), filtered at 10 kHz and fed either through a Digitimer D.130 spike processor (Medical Systems, Great Neck, NY) for conversion into digital transistor-to-transistor logic pulses or through a DigiData 1200 interface (Axon Instruments) for digitization of voltage signals. Data on spontaneous firing activity were collected, stored and analyzed either with the ISH/PSTH software package (courtesy of Dr. M. R. Palmer, University of Colorado) on an Apple IIE microcomputer (Apple Computer Co., Cupertino, CA) or with pCLAMP software (version 6.0; Axon Instruments) on a Gateway 4DX2-66V personal computer (Gateway 2000, Sioux City, SD). In both instances, data on spontaneous neuronal firing rate were calculated by interval methods over 2- to 4-min collection periods.

Muscimol, bicuculline, 2-chloroadenosine, naloxone, clonidine and idazoxan (all from Sigma) and morphine (National Institute on Drug Abuse) were applied to the slices by superfusion in aCSF, through a gravity-fed valve manifold switching system, at a rate of 2 ml/min. For analysis of the concentration-response data, the IC₅₀ for a given agonist was defined as the concentration of that agonist at which the neuronal firing rate was reduced by 50% from its spontaneous basal value; the value was calculated individually by interpolation for each concentration-response curve. The geometric mean IC₅₀ (Fleming et al., 1972) is the antilogarithm of the mean log[IC₅₀] value for all dose-response curves for a given agonist in neurons from a given treatment group. Because of the apparent excitotoxicity and depolarization block induced by glutamate, excitatory responses were generated in spontaneously active nTS neurons by a modest elevation of the extracellular potassium concentration in the aCSF. Given an extracellular (aCSF) potassium concentration of 5 mM and assuming an intracellular potassium concentration of approximately 100 mM, an elevation of the potassium concentration in the aCSF from 5 mM to 7.3 mM would be expected to change the Nernst potential for potassium by approximately +10 mV, thereby partially depolarizing the neuron. In the experiments comparing the effects of elevated potassium on neuronal activity among treatment groups, no other drug effects on neurons within a given brain slice were assessed after exposure of that slice to high-K⁺ aCSF.

The ² contingency analysis was used to determine the significance of data in the preliminary pharmacological characterizations, e.g., in the assessment of the effects of bicuculline on neuronal responses to morphine (see table 1). For all other statistical comparisons, except where specifically indicated, Student's t test was used after analysis of variance and a value of P < .05 was considered statistically significant. Values in the text and all points in the figures are reported as means ± S.E.M. All statistical analyses were performed and graphical presentations were prepared with Sigma- Stat and SigmaPlot for Windows (Jandel Scientific, San Rafael, CA), respectively.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

In initial studies characterizing pharmacological responses of guinea pig nTS neurons in vitro, recordings were made from 56 neurons in slices from 24 control animals. The drug applications were in random order, and no individual neuron was exposed to all of the agonists or drug combinations. The results are summarized in table 1. Approximately one-half of the nTS neurons tested were reversibly inhibited by GABA (1.0 µM). When the GABA_A receptor-selective agonist muscimol (0.1 µM) was applied, inhibitory responses were observed in all neurons tested. When neurons were exposed to the GABA_A receptor-selective antagonist bicuculline before challenge with any other agonist, a small increase in neuronal activity was observed in approximately one-third of the neurons (table 1). The mean excitatory effect of bicuculline on the neurons was an increase of 23 ± 4% above the basal firing rate. For the remainder of the population, an inhibition of spontaneous activity was also found in nearly one-third of the neurons tested; the remaining neurons showed no change in activity in the presence of bicuculline (table 1). When the effect of bicuculline on the total population of neurons exposed to the drug was examined, there was no significant effect on firing frequency (the mean effect was an increase in firing frequency of <1%).

When morphine (0.1 or 1.0 µM) was applied to nTS neurons alone, it was found to decrease firing rate in 40%, increase firing rate in 30% and produce little or no effect in 30% of the neurons (table 1). In contrast, when applied in the presence of bicuculline, morphine inhibited 77% of the neurons tested and excited only 1 of 30 (3%). Furthermore, of the 30 neurons analyzed in table 1, the basal firing rate of nine (30%) was increased in the presence of bicuculline alone, before the application of morphine. This effect of bicuculline is comparable to that observed in neurons exposed to bicuculline before challenge with any agonist (table 1). All effects of morphine were abolished by either washout with aCSF or application of aCSF containing naloxone (1.0 µM). Based on these data and those from studies in rats (Brooks et al., 1992; Rhim et al., 1993), it appears that tonic inhibitory GABAergic tone persists in some nTS neurons in the brainstem slice preparation and that opioids can disinhibit those nTS neurons, resulting in an apparent excitatory response. Therefore, morphine was applied in the presence of 1.0 µM bicuculline in all subsequent experiments. Table 1 also establishes that 2-chloroadenosine (1.0 µM, a nonselective A₁/A₂ adenosine receptor agonist) and clonidine (0.1 µM, an alpha-2 adrenoceptor agonist) produced only inhibition at those concentrations. The inhibitory effect of clonidine was antagonized by 0.1 µM idazoxan, an alpha-2 adrenoceptor antagonist. Illustrated in figure 1 are responses of two different nTS neurons to the application of morphine, clonidine and 2-chloroadenosine. The inclusion of these inhibitory substances in the aCSF bathing the brainstem slice led to a reduction of neuronal firing rate, as indicated by the ratemeter recordings (fig. 1). Both the ratemeter records and action potential recordings illustrated in figure 1 provide information related to the characteristics of the firing rate of these nTS neurons. As indicated under "Materials and Methods," the majority of nTS neurons studied maintained a slow but regular level of spontaneous activity that is illustrated in the action potential recordings (fig. 1). This type of regular spontaneous activity provided a valuable tool for comparison of the action of inhibitory agonists after chronic treatment with morphine.

View larger version (59K): [in this window] [in a new window]   Fig. 1.   Ratemeter and action potential records of the firing of guinea pig mnTS neurons and the effects of inhibitory agonists. A, ratemeter records obtained from the same nTS neuron, illustrating the inhibitory effects of both clonidine (Clon) and morphine (Mor). The action potential recordings provided in A1, A2 and A3 were obtained from this neuron before (A1), during (A2) and after (A3) exposure to 1 µM clonidine. Note that the inhibitory effect of clonidine is antagonized by idazoxan (IDA), whereas the effect of morphine is reversed by naloxone (NAL). B, ratemeter and action potential records obtained from another mnTS neuron from a different slice. Shown is the response of this neuron to 2-chloroadenosine (2-CADO) and morphine. Action potential recordings were obtained from this neuron before (B1), during (B2) and after (B3) exposure to morphine (1 µM). The ratemeter and action potential records illustrate the regularity with which these neurons produce action potentials and the fact that all of the inhibitory agonists used can produce substantial, if not total, inhibition of neuronal activity. The time between ratemeter recording panels illustrated was approximately 20 min.

Excitatory agonists that were applied to the slice preparations included glutamate, substance P and acetylcholine. Although all of these agonists initially increased the firing rate of nTS neurons, the excitatory responses proceeded rapidly to complete loss of action potentials, presumably due to depolarization block. Thus, reliable excitatory responses to these agonists were difficult to obtain and to quantify on a regular basis in this preparation.

Experiments were then performed to compare pharmacological sensitivity between neurons in preparations from animals chronically treated with morphine and control slice preparations. Recordings were obtained from 88 neurons from 34 morphine-treated animals and 79 neurons from 32 control animals. The morphine-treated animals were implanted with morphine sulfate pellets 7 days before the day of the experiment. Control animals, except the seven used for the potassium experiments, received placebo pellets. It has been repeatedly shown that other tissue preparations from unimplanted and lactose-implanted guinea pigs do not differ in their responsiveness to any of the agonists used here (Goldstein and Schulz, 1973; Johnson et al., 1978). Experimental and control animals did not differ significantly in mean weight at the time of pellet implantation. During the following 7 days, the placebo group gained approximately 10% in body weight, whereas the weight of morphine-implanted animals did not change significantly.

The spontaneous basal firing rate of control neurons was 2.20 ± 0.20 Hz and that of the neurons from morphine-treated guinea pigs was 2.47 ± 0.24 Hz; these rates were not significantly different. Pharmacologically induced responses were normalized to the firing rate existing just before application of the agonist. For muscimol, this firing rate was in the absence of any drug; for all other agonists, this was the firing rate in the presence of 1.0 µM bicuculline. Bicuculline caused a small increase in basal firing rate in one-third of the neurons (table 1) but had minimal effects on the mean basal firing rate of the total population of neurons. There was a slight trend toward an increase in basal firing rate of neurons with time. However, linear regression analysis indicated that this had no significant effect on the percent change in rate induced by any agonist (P > .2 for all comparisons).

Preliminary experiments indicated that cumulative concentration-response curves could be readily obtained for the inhibitory effects of muscimol but, due to marked acute desensitization, not for those of morphine or 2-chloroadenosine. For example, when morphine was administered in a cumulative fashion (0.1 to 1.0 µM), 1.0 µM produced only 20 to 30% and 10⁵ M only 50 to 60% inhibition in slices from placebo-treated animals. In contrast, when administered as a single concentration without prior exposure to the agonist, a morphine concentration of 1.0 µM inhibited firing by a mean of 60 to 80%, with a substantial number of neurons being totally inhibited, as illustrated in figure 1. The use of noncumulative concentration-response curves was impractical in this preparation because of the long periods of washout and recovery required between concentrations. Consequently, comparisons of pharmacological sensitivity were made with single concentrations.

In experiments in which muscimol was used as an agonist, muscimol (0.3 µM) was always tested first. After washout and recovery of basal firing rate, bicuculline (1.0 µM) was added to the superfusate. After the base-line firing rate had stabilized (2-3 min), morphine (1.0 µM) was added. Figure 2 presents the results. In control neurons, these concentrations of muscimol and morphine produced a mean inhibition of neuronal firing of approximately 50 and 70%, respectively. However, in neurons from morphine-treated animals, the inhibitory effects of both agonists were significantly less, indicating subsensitivity (tolerance).

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Comparison of single-concentration effects of muscimol and morphine on control mnTS neurons and mnTS neurons from chronically morphine-treated animals. Muscimol (0.3 µM) inhibited control neurons by 52 ± 8% (n = 15 cells from eight animals) and chronically morphine-treated neurons by 26 ± 6% (n = 13 cells from five animals). Morphine (1.0 µM) plus bicuculline (1.0 µM) inhibited controls by 74 ± 8% (n = 16 cells from eight animals) and cells from morphine-treated animals by 44 ± 10% (n = 13 cells from five animals). * P < .05.

Cumulative concentration-response curves were also obtained with muscimol in slices from placebo-treated animals and animals implanted with morphine pellets (fig. 3). The concentration-response curve was shifted to the right, characteristic of subsensitivity, in the morphine-treated group. The shift at the IC₅₀ was 5.6-fold (P < .001). The mean response to single concentrations of muscimol at 0.3 µM (fig. 2) falls very close to the responses predicted from the concentration-response curves for each group. Thus, for an agonist that does not induce desensitization under these experimental conditions, tolerance (subsensitivity) can be comparably demonstrated with concentration-response curves and with single concentrations.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Cumulative dose-response curves for muscimol in mnTS slice preparations from naive and chronically morphine-treated animals (controls, n = 14 neurons from six animals; morphine-treated, n = 18 neurons from eight animals). There is a 5.6-fold rightward shift of the muscimol curve at the geometric mean IC50 in the chronically morphine-treated group (P < .001).

Responses to single concentrations of 2-chloroadenosine (1.0 µM) (fig. 4) and clonidine (0.1 µM) (fig. 5) were obtained in slices from control animals and morphine-treated animals. In each instance, 2-chloroadenosine or clonidine was administered in the presence of bicuculline and after applications of, and recovery from, morphine (1.0 µM), also in the presence of bicuculline. Note that 2-chloroadenosine and morphine were applied to the same neurons to obtain figure 4; clonidine and morphine were applied to the same neurons to obtain figure 5. Each of the agonists produced mean inhibition of firing rate of approximately 60% in control slices. In contrast, each agonist produced mean inhibitory effects of only about 20% in neurons from morphine-treated guinea pigs, indicating subsensitivity.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Comparison of single-concentration effects of 2-chloroadenosine (2-CADO) and morphine on control mnTS neurons and mnTS neurons from chronically morphine-treated animals. Both agonists were applied in the presence of 1.0 µM bicuculline. 2-Chloroadenosine (1.0 µM) inhibited control neurons by 63 ± 11% and neurons from chronically morphine-treated animals by 22 ± 9%. Morphine (1.0 µM) inhibited control cells by 60 ± 13% and cells from morphine-treated animals by 18 ± 10%. For all four bars, n = 10 neurons from four control or four morphine-pretreated animals. *P < .05, relative to control.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Comparison of single-concentration effects of clonidine and morphine on control mnTS neurons and mnTS neurons from chronically morphine-treated animals. Both agonists were applied in the presence of 1.0 µM bicuculline. Clonidine (0.1 µM) inhibited control neurons by 60 ± 6% and those from chronically morphine-treated animals by 20 ± 5%. Morphine (1.0 µM) inhibited control neurons by 66 ± 7% and neurons from chronically morphine-treated animals by 25 ± 6%. The n values for both drugs were 22 neurons from 7 control animals and 24 neurons from 11 morphine-pretreated animals. *P < .05, relative to control.

The excitatory effect of elevated extracellular potassium on spontaneous neuronal activity was also compared between control nTS neurons and nTS neurons from morphine-treated animals (fig. 6). Switching from aCSF with normal potassium (5 mM) to aCSF with elevated potassium (7.3 mM) increased the spontaneous firing rate an average of 58 ± 13% over the spontaneous basal firing frequency in control nTS neurons (n = 21) and 152 ± 48% over basal in morphine-treated nTS neurons (n = 25). This excitatory effect of potassium-induced depolarization on neuronal activity was significantly greater in neurons from the morphine-treated animal group (P < .05). Other excitatory agonists were not included in the experiments comparing sensitivity because of the "depolarization block" already discussed.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Comparison of the effects of elevation of the potassium concentration in aCSF from 5.0 to 7.3 mM on mnTS neurons in slices from control and chronically morphine-treated animals. The elevated potassium concentration increased the spontaneous firing rate by 58 ± 13% over the basal rate in control preparations (n = 21 neurons from seven animals) and by 152 ± 48% in neurons from animals implanted with morphine pellets (n = 25 neurons from six animals). * P < .05

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Pharmacological examination of the guinea pig mnTS in vitro suggests that the responsiveness of these neurons is similar to that reported for the rat mnTS with the agonists tested. A large percentage of neurons tested were inhibited by GABA and excited by glutamate, whereas substitution of a GABA_A-selective agonist, muscimol, for GABA yielded inhibition of spontaneous firing in all neurons tested. In addition, these inhibitory responses were antagonized by bicuculline, a GABA_A-selective antagonist, suggesting that this inhibition is mediated by activation of GABA_A receptors. The lack of uniform response among all mnTS neurons to the endogenous neurotransmitter GABA may be due to activation of more than one receptor subtype with different cellular actions and subcellular localizations. For example, presynaptic regulation of GABA release in the mnTS may be mediated by GABA_B receptors (Brooks et al., 1992), the activation of which may reduce endogenous GABA release and consequently disinhibit, or excite, the neuron observed. Such an idea is further supported by the fact that bicuculline exposure leads to different changes in neuronal activity as well. The fact that bicuculline produces excitation in a proportion of guinea pig mnTS neurons is consistent with the idea that the slice preparation contains some active inhibitory tone. Given the neural circuitry that remains active in the slice preparation, it is not surprising that nonselective agonists exert mixed effects when applied by superfusion, because the balance of excitatory to inhibitory effects reflects both the localization of different receptors and the level of ongoing synaptic activity in the neuron of interest. The data obtained with micromolar and lesser concentrations of agonists strongly suggest that the guinea pig mnTS neurons possess GABA_A, µ-opioid and A₁ or A₂ adenosine receptors and alpha-2 adrenoceptors functionally coupled to inhibitory cellular mechanisms. These conclusions are supported by the relative selectivity of the agonists and by the use of selective antagonists (bicuculline, naloxone and idazoxan).

Morphine alone inhibited slightly less than half of the guinea pig nTS neurons to which it was applied. In the presence of bicuculline, a GABA_A-selective antagonist, however, morphine inhibited the spontaneous firing of 77% of the neurons tested. This shift in the ratio of excitatory to inhibitory effects of morphine in the presence of a GABA_A receptor antagonist is most convincingly explained by an opioid-mediated disinhibition of the neurons, similar to that observed in the ventral tegmental area (Johnson and North, 1992) and in the hippocampus (Zieglgänsberger et al., 1979). In both structures, opioids preferentially inhibit smaller GABAergic inhibitory interneurons, resulting in excitation of the larger projection neurons through a disinhibition that is blocked by GABA_A-selective antagonists. The effects obtained with bicuculline alone would also support such a conclusion. This interpretation of the data from the guinea pig nTS is consistent with both anatomical and pharmacological data from other systems, including the rat nTS (Brooks et al., 1992; Rhim et al., 1993).

The spontaneous basal firing rate of guinea pig nTS neurons tended to be higher but was not changed significantly by chronic treatment of the animals with morphine. In other spontaneously active in vitro central nervous system preparations, such as the rat LC, basal activity has been reported to be either increased (Kogan et al., 1992) or unchanged (Andrade et al., 1983) by chronic morphine treatment, which is known to result in the development of pharmacological tolerance to the opioids. The fact that tonically active inhibitory processes exist within the brainstem and appear to persist in the brainstem slice preparation suggests that the regulation of spontaneous frequency is a complex physiological process. In addition, the fact that chronic treatment with morphine might be expected to produce different degrees of excitation and inhibition via direct and indirect effects on mnTS neurons further complicates the regulation of spontaneous activity after chronic exposure to morphine. The regulation of spontaneous activity of mnTS neurons under control conditions as well as after morphine treatment can be adequately evaluated only by using intracellular recording.

Comparison of single-concentration applications of inhibitory agonists and cumulatively increasing concentrations indicated that guinea pig nTS neurons rapidly desensitized to morphine and 2-chloroadenosine but not to muscimol. To avoid the complication of acute desensitization in determinations of pharmacological sensitivity, single concentrations of agonists were used to determine the effect of chronic morphine pellet implantation on sensitivity. The results were unequivocal. Chronic exposure of animals to morphine led to subsensitivity (tolerance) of nTS neurons in vitro to inhibitory effects mediated through µ-opioid receptors, GABA_A receptors, A₁/A₂ adenosine receptors and alpha-2 adrenoceptors. Concentration-response curves with muscimol, to which significant acute desensitization could not be detected, confirmed that the subsensitivity could also be demonstrated as a significant shift of the curve to the right. The results indicate that the opioid tolerance that develops in the guinea pig nTS is nonspecific among a variety of pharmacologically unrelated inhibitory agonists. These findings differentiate the opioid tolerance that develops in the guinea pig mnTS from opioid tolerance in the rat LC, which is highly opioid-specific (Christie et al., 1987; Kogan et al., 1992), and suggest that the adaptive change that occurs in guinea pig mnTS neurons chronically exposed to morphine is of a physiologically more general nature, occurring beyond the level of the cell-surface receptors.

Elevations of extracellular potassium would be expected to depolarize the neuronal membrane by lowering the Nernst potential for potassium, because in most excitable cells the membrane potential is determined largely, but not entirely, by the resting potassium conductance. This depolarization resulted in a greater increase in spontaneous firing in mnTS neurons in preparations from animals chronically treated with morphine than in controls, suggesting that the morphine-tolerant mnTS neurons were more excitable. These data are consistent with the hypothesis that the development of opioid tolerance in these neurons is associated with a general change in the excitability of the cell. It must be noted, however, that depolarization in this manner affects not only the neurons of the mnTS but, obviously, all neurons in the slice preparation. Given the functional circuitry remaining intact within the slice, this experimental procedure could lead to mixed effects and is, therefore, only a crude method of assessing neuronal excitability in the nTS in this preparation. Nevertheless, it is consistent with the subsensitivity to several inhibitory agonists being due to a general change in cellular excitability.

One autonomic circuit that regulates abdominal visceral activity consists of the nTS, dorsal motor nucleus of the vagus, myenteric plexus and nodose ganglion, which, in turn, projects back to the nTS. Endogenous opioid peptides may modulate the activity of many, if not all, of those neural components. Opioid tolerance has been demonstrated in both myenteric S-neurons and mnTS neurons of the guinea pig. Of particular importance is the finding that, in both of these neuronal populations, opioid tolerance is pharmacologically nonspecific, suggesting a mechanism of development of long-term, heterologous tolerance distinct from the cellular events believed to underlie rapid, homologous, receptor-based desensitization (Fleming and Taylor, 1995). In myenteric neurons, the nonspecific tolerance to inhibitory agonists and supersensitivity to excitatory agonists is clearly associated with a partial depolarization of the myenteric S-neurons (Leedham et al., 1992). Furthermore, in non-neuronal cells, nonspecific adaptive sensitivity changes are due to membrane depolarization secondary to decreased function of the Na⁺,K⁺-pump (Hershman et al., 1993, 1995; Rogers et al., 1993). The possible role of membrane potential and the Na⁺,K⁺-pump in opioid tolerance in central nervous system neurons will be addressed in future experiments.

In the rat LC, chronic treatment with morphine results in a highly specific opioid tolerance, which is associated with alterations in the G protein-coupled adenylyl cyclase system at virtually every point within that transduction pathway (reviewed in Nestler et al., 1993). In contrast, with chronic morphine treatment we observe the development of a nonspecific opioid tolerance both in the myenteric plexus and now in the nTS of the guinea pig. At least two possibilities exist to explain these observed differences. First, the differences in the specificity of opioid tolerance may reflect a true species difference in the cellular mechanisms of tolerance development between rats and guinea pigs. Second, the differences in tolerance specificity may reflect cellular differences in the two neuronal populations, the LC and the nTS. These are important considerations for future investigation.

Several conclusions can be drawn from these data. First, neurons of the guinea pig mnTS in vitro are pharmacologically similar to those of the rat. Second, the opioid narcotic morphine exerts mixed effects on guinea pig mnTS neurons through both direct inhibition of spontaneous firing and synaptic disinhibition of basal activity, presumably through suppression of basal GABAergic tone that exists in the slice preparation. Third, chronic treatment of guinea pigs with morphine results in the development of tolerance in the mnTS, which is not pharmacologically specific to the opioids but extends to a wide variety of pharmacologically unrelated agents. Fourth, the opioid tolerance that develops in the guinea pig mnTS is apparently associated with a more generalized change in neuronal excitability or responsiveness, suggesting that this tolerance is due to a change in the cellular physiology of the neuron and is not confined to a single receptor pool or transduction process.