The role of nitric oxide in morphine dependence and withdrawal excitation of rat oxytocin neurons

Induction of morphine tolerance and dependence

Rats were anaesthetized with 5% halothane in a mixture of oxygen and nitrous oxide and a 21-gauge stainless steel cannula was stereotaxically implanted in the right lateral cerebral ventricle (3 mm posterior to bregma, 2 mm lateral, 4 mm below the skull surface) and secured with two small screws fixed in the skull and dental cement. This was connected by polythene tubing to a subcutaneously implanted osmotic minipump (Alzet 2001; Alza Corp., Palo Alto, CA, USA) to continuously deliver morphine sulphate solution (Edinburgh Royal Infirmary, Edinburgh, UK) at 10 µg/µL for 40 h then 20 µg/µL for 40 h and finally 50 µg/µL, or vehicle (pyrogen-free distilled water; 1 µL/ h) in naive rats, over 5 days (Rayner et al., 1988). After surgery was completed rats were caged singly.

Blood sampling experiments

To ascertain whether NO mechanisms are altered as a consequence of morphine dependence, morphine-dependent and -naive rats were pretreated with the NOS inhibitor Nω-nitro-l-arginine methyl ester (l-NAME) or the inactive enantiomer Nω-nitro-d-arginine methyl ester (d-NAME) and challenged with a hyperosmotic stimulus. Six days after minipump implantation morphine-dependent and naive rats were anaesthetized with intraperitoneal (i.p.) urethane (1.25 g/kg, 25% w/v) and then given an i.p. injection of either l-NAME (50 mg/kg, n = 5 and 6, respectively; Sigma, UK) or d-NAME (50 mg/kg, n = 6 and 7, respectively; Sigma). A femoral vein was cannulated for intravenous (i.v.) injections and a femoral artery for blood sampling; the other femoral vein was cannulated for infusion of hypertonic saline (via a slow infusion syringe pump). Blood samples (0.3 mL) were withdrawn into heparinized syringes, centrifuged and the plasma separated and stored at − 20 °C for subsequent oxytocin radioimmunoassay. Blood cells were resuspended in 0.9% saline and infused via a femoral venous cannula.

A basal blood sample was taken from all rats 240 min after l-NAME or d-NAME administration. Rats were then infused with i.v. hypertonic saline (2 m, 2.3 mL/h) and five further blood samples taken at 30, 45, 60, 75 and 90 min, at which point the infusion was stopped. All rats then received an i.v. injection of naloxone (5 mg/kg) and a final sample was taken 5 min later.

In a separate experiment to investigate whether endogenous NO is involved in the expression of withdrawal excitation, morphine-dependent rats were treated with the NOS inhibitor, l-NAME or vehicle before naloxone was given. Six days after mini-pump implantation morphine-dependent and -naive rats were prepared for blood sampling as described above. A first basal blood sample was drawn 120 min after the injection of either l-NAME or vehicle (0.9% NaCl, 0.5 mL/kg). Two further basal samples were obtained at 30 and 5 min prior to injection of naloxone (5 mg/kg i.v.; Sigma) at t = 0 min (240 min from l-NAME/vehicle treatment). Three further blood samples were taken at 5, 15 and 60 min after precipitation of morphine-withdrawal with naloxone.

Radioimmunoassay

Oxytocin content in unextracted plasma samples was measured by specific radioimmunoassay (Higuchi et al., 1985). The sensitivity was less than 4.8 pg/mL and intra-assay variation was less than 15% at standard concentrations of 10, 50, 100, 250, 500 and 1000 pg/mL. All samples were measured in the same assay and postnaloxone samples were diluted 1 : 10 in assay buffer prior to assay to ensure measurement on the reliable part of the standard curve.

Electrophysiology

Morphine-treated rats were anaesthetized with urethane (1.25 g/kg, i.p.), a femoral vein and the trachea cannulated and the pituitary stalk and right SON exposed transpharyngeally (Leng & Dyball, 1991). A locally made U-shaped dialysis probe (2 mm membrane length, Spectra/Por RC Hollow Fibres; Spectrum Medical Inc., Great Falls, MT, USA) was bent to position the loop of the membrane flat onto the exposed ventral glial lamina of the SON. A glass micropipette filled with saline (20–40 MΩ) was introduced into the centre of the dialysis probe loop to record extracellular single neuron activity within the SON (Ludwig & Leng, 1997) via a CED 1401 interface attached to a personal computer running Spike 2 software (Cambridge Electronic Design, Cambridge, UK). A bipolar stimulating electrode (Snex-200X; Clarke Electromedical Instruments, Reading, UK) was placed on the pituitary stalk and set to deliver single matched biphasic pulses (1 ms, < 1mA peak to peak) for antidromic identification of SON neurons. Oxytocin neurons were distinguished from vasopressin neurons by their continuous firing pattern and transient excitation in response to i.v. cholecystokinin (20 µg/kg: Renaud et al., 1987). Artificial cerebrospinal fluid was dialysed at 3 µL/min throughout the experiment (pH 7.2; in mm: NaCl, 138; KCl, 3.36; NaHCO₃, 9.52; Na₂HPO₄, 0.49; CaCl₂, 1.26; MgCl₂, 1.18 and urea, 2.16). After withdrawal excitation had been induced by naloxone the dialysis fluid was changed to artificial cerebrospinal fluid containing the NO donor SNP (50 mm; Sigma), then the GABA_A antagonist bicuculline methiodide (2 mm; Sigma) and finally SNP combined with bicuculline (see Fig. 3A). Based on previous experiments we estimate the concentrations of drugs achieved within the SON using this method to be 1 × 10⁻³ of that in the dialysate (Ludwig & Leng, 1997). Sodium nitroprusside has previously been shown to inhibit SON neurons solely through the generation of NO (Liu et al., 1996). One neuron was recorded from each rat.

Statistical analysis

Blood sampling data were analysed using two-way repeated measures analysis of variance (anova) to isolate differences within and between groups. Where the F ratio was significant this was followed by posthoc analysis with Student-Newman-Keuls tests, unless stated otherwise, using SigmaStat^® software (SPSS Science, Chicago, IL, USA). All values are expressed as the mean ± SEM and differences were considered statistically significant if P ≤ 0.05.

The firing rates of identified oxytocin cells were downloaded onto a personal computer using the Spike 2 software package (Cambridge Electronic Design) and analysed using SigmaStat^® software. The mean firing rate of each cell was calculated for 5-min periods immediately before and during peak effects of each treatment unless stated otherwise. Differences were considered significant at P ≤ 0.05 values following paired and Student's t-tests.

Blood sampling experiments

Effects of the nitric oxide synthase inhibitor Nω-nitro-l-arginine methyl ester on hypertonic saline-stimulated oxytocin secretion in morphine-dependent and -naive rats

To allow comparison of the activity of NO mechanisms on stimulated oxytocin neurons in dependent and naive rats, an i.v. infusion of hypertonic saline (Brimble & Dyball, 1977) was given to morphine-dependent and -naive rats. There were no differences in basal plasma oxytocin concentrations between morphine-dependent and -naive rats pretreated with either l-NAME or d-NAME, indicating lack of endogenous NO generation and action on basal oxytocin secretion in naive as well as dependent rats [Fig. 1; before hypertonic saline the mean values were, respectively, 22.5 ± 5.4 and 32.9 ± 6.8 pg/mL, n.s., in d-NAME- and l-NAME-treated naive rats and 24.8 ± 5.5 and 17.9 ± 3.8 pg/mL, n.s., in d-NAME- and l-NAME-treated dependent rats; Fig. 2; before naloxone, the mean values were, respectively, 40.1 ± 7.7 and 46.8 ± 7.4 pg/mL, n.s., in vehicle- and l-NAME-treated dependent rats].

anova showed a significant increase in oxytocin secretion during hyperosmotic stimulation (P < 0.0001) which was also significantly different between treatment groups (P = 0.003). Thirty minutes after the start of the hypertonic saline infusion oxytocin secretion was increased significantly more in the naive l-NAME-treated rats than in the naive d-NAME-treated rats (by 639.4 ± 148.7 vs. 449.6 ± 56.6 pg/mL, P < 0.05). The oxytocin secretory response to the hypertonic saline infusion in the morphine-dependent rats was delayed in comparison with the naive rats, taking 45–60 min to reach the level at 30 min in naive rats (Fig. 1A). At 45 min, the plasma oxytocin concentration in the l-NAME-treated dependent rats was greater than in the d-NAME-treated dependent rats (Fig. 1A; increases from basal were 620.5 ± 99.4 and 373.2 ± 102.1 pg/mL, respectively; P < 0.05).

To take into account the different time courses of responses in the dependent and naive rats, the plasma oxytocin concentrations at each time point during l-NAME treatment were plotted against the oxytocin values at the equivalent time point during d-NAME treatment for the respective dependent and naive groups. The naive rat data were best fitted by a second order regression (r² = 0.998, P = 0.0001; Fig. 1B). It is clear that l-NAME was no less effective in increasing oxytocin secretion during hyperosmotic stimulation in dependent than in naive rats.

Naloxone given at the end of the hypertonic saline infusion further greatly increased oxytocin secretion within 5 min in all groups, with no differences in plasma concentrations among groups (Fig. 1C).

Electrophysiology

Extracellular single unit recordings were made from eight antidromically identified SON cells, each from a different morphine-dependent rat, for between 1 and 6 h; all cells recorded displayed spontaneous activity and were transiently excited by i.v. administration of 20 µg/kg cholecystokinin (1.0 ± 0.2 spikes/s increase in firing rate, averaged over 5 min before and 5 min after cholecystokinin), typical of oxytocin cells in morphine-naive and -dependent rats (Brown et al., 1996). As previously demonstrated (Bicknell et al., 1988), i.v. administration of 5 mg/kg naloxone caused a robust and sustained increase in the firing rate of five oxytocin cells (by 2.9 ± 0.4 spikes/s). Recordings from the three remaining cells were started after naloxone administration when these cells displayed fast continuous firing at 8.3 ± 1.2 spikes/s, typical of oxytocin cells during morphine withdrawal excitation.

Nω-nitro-l-arginine methyl ester and oxytocin secretory responses to hyperosmotic stimulation

As found previously in anaesthetized rats, there was no evidence for significant endogenous NO restraint of oxytocin neurons under basal conditions (Srisawat et al., 2000) and l-NAME treatment clearly did not precipitate withdrawal excitation of oxytocin neurons in dependent rats (1, 2). Similarly, a NOS inhibitor applied directly onto the SON increased oxytocin neuronal firing rate by only 13%, although intracerebroventricularly administered l-NAME in conscious rats increases basal oxytocin secretion (Kadekaro & Summy-Long, 2000; Srisawat et al., 2000).

Other studies indicate that if stimulation is brief, as with a systemic injection of cholecystokinin (Brown et al., 1996), NO inhibition of oxytocin neurons is still not significantly activated (Srisawat et al., 2000). The hypertonic saline stimulus used in the present study produces a continual increase in oxytocin neuron firing rate in naive rats, by c. 7 spikes/s at 30 min (Leng et al., 2001). At this time in the naive rats l-NAME pretreatment enhanced the oxytocin secretory response by 41% compared with the d-NAME-treated rats, indicating activation of NO restraint on oxytocin neurons (Srisawat et al., 2000) which became ineffective with further osmotic stimulation (Fig. 1). In the morphine-dependent rats the delayed increase in oxytocin secretion with hypertonic saline infusion is likely to reflect incomplete tolerance to inhibitory actions of morphine. Although hyponatraemia reduces the initial response to hypertonic saline infusion (Leng et al., 2001), plasma [Na⁺] is actually increased in dependent rats (140.1 ± 0.5 vs. 135.8 ± 1 mmol/L in naive rats, P < 0.05, t-test). Nor was the delayed response due to greater NO inhibition as l-NAME did not advance the response (Fig. 1). The action of l-NAME in further increasing oxytocin secretion in only the dependent rats with further hypertonic saline infusion indicates that inhibition by NO was at least as effective as in the naive rats.

Nitric oxide and morphine withdrawal-induced oxytocin secretion

We have shown here that the direction of action of NO on oxytocin neurons is not altered in morphine dependence either before or after naloxone-precipitated withdrawal. Hence, for withdrawal-induced hyper-excitation of oxytocin neurons, a robust feature of naloxone-precipitated morphine withdrawal in rats (Brown et al., 2000), inhibition of NOS accentuated the excitation (Fig. 2). Morphine withdrawal increases mean arterial blood pressure only slightly (Rayner et al., 1988) although we expect l-NAME to have had a more pronounced effect. However, oxytocin neurons are hardly affected by increases in blood pressure (Leng et al., 1991; Renaud & Bourque, 1991) so the effects of l-NAME on oxytocin neurons is not likely to be a result of blood pressure changes.

Nω-nitro-l-arginine methyl ester increased oxytocin secretion during withdrawal and hypertonic saline infusion in morphine-dependent rats to a similar extent (by 34% at peak of withdrawal and by 33–38% from 45 to 75 min of osmotic stimulation, respectively). Naloxone-precipitated morphine withdrawal resulted in a sustained increase in oxytocin neuron firing rate of 2.9 ± 0.4 spike/s (Fig. 3), similar to increases expected during hypertonic saline infusion (Leng et al., 2001). Hence, the effect of l-NAME (and thus NO) on oxytocin neurons during withdrawal seems to be similar to that in dependent rats stimulated osmotically and, in turn, was not less than the effects of l-NAME in osmotically stimulated naive rats.

The present experiments measuring oxytocin secretion could not distinguish effects of l-NAME on NO production and action on distant or local inputs to oxytocin neurons or on their terminals in the posterior pituitary. The mechanism by which NO production is induced by osmotic stimuli and withdrawal stimulation may involve enhanced glutamate release within the SON (Brown et al., 2000; Leng et al., 2001). As activation of glutamate receptors induces NO production in other systems (for reviews see Fedele & Raiteri, 1999; Kiss & Vizi, 2001), enhanced glutamatergic drive on oxytocin cells during these stimuli may increase NO production to restrain further excitation. Nonetheless, the aggregate effects of NO on secretion by excited oxytocin neurons were evidently not less in morphine-dependent than in naive rats.

In naive rats naloxone has no effect on the firing of oxytocin neurons during hyperosmotic stimulation but increases oxytocin secretion by antagonizing endogenous κ-opioid actions in the posterior pituitary in a similar way in both naive and dependent rats (Coombes et al., 1991; Leng et al., 1997). The similar terminal plasma oxytocin concentrations in all groups after hyperosmotic saline infusion for 90 min, before and after naloxone (Fig. 1), indicate that the maximal firing and secretory capacity of the neurons had been reached and NO inhibition overwhelmed.

Nitric oxide and morphine withdrawal-induced oxytocin cell excitation

The actions of NO on the electrical activity of oxytocin neurons in morphine naive rats are inhibitory (Ozaki et al., 2000; Srisawat et al., 2000; Stern & Ludwig, 2001), as shown here to be the case in dependent rats. The inhibitory actions of NO on oxytocin neurons in naive rats include direct actions on the neurons (Stern & Ludwig, 2001) as well as increasing the frequency and amplitude of fast GABAergic inhibitory postsynaptic currents (Ozaki et al., 2000; Stern & Ludwig, 2001). Oxytocin neurons express soluble guanylyl cyclase subunits, a target for NO (Furuyama et al., 1993), but activating cGMP mechanisms in SON neurons causes excitation (Yang & Hatton, 1999). In mice, histochemical studies indicate that NO produced by oxytocin neurons can act via cGMP on both inhibitory GABA and excitatory glutamate and catecholamine terminals on the neurons rather than directly on the neurons (Vacher et al., 2003). In rats, electrophysiological functional studies have found only facilitatory actions of NO on GABA terminals (Ozaki et al., 2000; Stern & Ludwig, 2001). Nonetheless, if the cGMP signalling mechanism is also present, and appropriately coupled as the major mechanism of NO action (Prast & Philippu, 2001) in excitatory terminals on oxytocin neurons, then it is possible that NO could act to excite oxytocin neurons rather than inhibit them in some circumstances. However, no such circumstance has yet been found and morphine dependence does not, as shown here, ablate or reverse the inhibitory action of NO on oxytocin neurons.

In contrast, the electrical excitation of locus coeruleus neurons, somato-dendritic release of noradrenaline and local release of glutamate during morphine withdrawal are suppressed by systemic or local infusion of NOS inhibitors (Hall et al., 1996; Pineda et al., 1998; Javelle et al., 2002) or by local infusion of a soluble guanylyl cyclase inhibitor (Sullivan et al., 2000). Thus, it is envisaged that withdrawal excitation of locus coeruleus neurons involves glutamate release and action through N-methyl-d-aspartate receptors, increased intracellular [Ca²⁺] and hence nNOS activation (Garthwaite et al., 1989; Hall et al., 1996). The action of NO in further stimulating excitatory synapses and supporting withdrawal excitation of locus coeruleus neurons (Manzoni et al., 1992; Uzbay & Oglesby, 2001) contrasts with the present study showing inhibitory actions of NO on oxytocin neurons during withdrawal.

There may be a modification of the interaction of NO with the GABA input onto oxytocin neurons in morphine-dependence. Firstly, the reduction by SNP of the firing rate of withdrawn oxytocin neurons (Fig. 3) was less marked than previously reported for SNP inhibition of basal oxytocin neuron firing rate in naive rats (Stern & Ludwig, 2001). This may well simply be a consequence of a reduced effect of NO per se as the neurons are strongly excited, as indicated by the oxytocin secretion data (Fig. 1). Alternatively, or additionally, NO generated by SNP may be less effective because of competition with endogenous NO generated by NOS as the neurons are excited. Secondly, despite the increased firing rate during morphine withdrawal, bicuculline increased oxytocin neuron firing rate to an extent (28% increase; Fig. 3) similar to that seen in naive rats (24% increase), under identical recording conditions, previously reported by Stern & Ludwig (2001). Thus, there is continuing GABA restraint of oxytocin neuron firing even during withdrawal excitation, similar to the GABA input to oxytocin neurons during osmotic stimulation (Leng et al., 2001). Thirdly, bicuculline did not diminish the inhibitory action of SNP on firing rate during withdrawal excitation (Fig. 3), in contrast with a significant reduction in the inhibitory effect of SNP on the firing rate of oxytocin neurons in naive rats in the presence of bicuculline (Stern & Ludwig, 2001). However, this may, like the less marked effect of SNP alone, reflect that, under the conditions of strong withdrawal excitation, endogenous NO is maximally effective on the GABA terminals and the additional NO generated by SNP acts only directly on the oxytocin neurons for this reason.

In conclusion, it seems that the effects of inhibiting NOS generation in neurons excited by precipitated morphine withdrawal will depend upon whether the action of NOS and NO in the normal regulation of these neurons is excitatory, as in the locus coeruleus, or inhibitory, as in the regulation of oxytocin neurons. The excitation of oxytocin neurons during morphine withdrawal activates NOS which then restrains the excitation, as in other conditions in which they are strongly excited. Our evidence does not indicate that changes in NOS and NO mechanisms underlie morphine dependence and withdrawal excitation in oxytocin neurons. The hyper-secretion of oxytocin in the rat model of precipitated morphine withdrawal is accentuated by a NOS inhibitor and this questions whether NOS inhibitors should be considered for use in controlling other withdrawal phenomena (Uzbay & Oglesby, 2001). The present study cautions that NOS inhibitors should not be expected to prevent all opiate withdrawal symptoms and, indeed, that some may be enhanced.