Chronic morphine reduces the readily releasable pool of GABA, a presynaptic mechanism of opioid tolerance

Abstract

Key points

• Chronic treatment with opioids, such as morphine, leads to analgesic tolerance.

• While postsynaptic opioid tolerance is well documented, the involvement of presynaptic mechanisms remains unclear.

• We show that chronic morphine reduces the ability of periaqueductal grey (PAG) neurons to maintain GABAergic transmission.

• This depression of GABAergic transmission was due to a reduction in the effective size of the readily releasable pool.

• This also led to a reduction in opioid presynaptic inhibition; these presynaptic adaptations need to be considered in the development of strategies to reduce opioid tolerance.

Abstract

The midbrain periaqueductal grey (PAG) plays a critical role in tolerance to the analgesic actions of opioids such as morphine. While numerous studies have identified the postsynaptic adaptations induced by chronic morphine treatment in this and other brain regions, the presence of presynaptic adaptations remains uncertain. We examined GABAergic synaptic transmission within rat PAG brain slices from animals which underwent a low dose morphine treatment protocol which produces tolerance, but not withdrawal. Evoked GABAergic IPSCs (inhibitory postsynaptic currents) were less in morphine compared to control saline treated animals. Postsynaptic GABA_A receptor mediated currents and desensitization, presynaptic release probability (P _r), and inhibition by endogenous neurotransmitters were similar in morphine and saline treated animals. By contrast, the effective size of the readily releasable pool (RRP) was smaller in morphine treated animals. While the μ‐opioid agonist DAMGO produced a reduction in P _r and RRP size in saline treated animals, it only reduced P _r in morphine treated animals. Consequently, DAMGO‐induced inhibition of evoked IPSCs during short burst stimulation was less in morphine, compared to saline treated animals. These results indicate that low dose chronic morphine treatment reduces presynaptic μ‐opioid inhibition by reducing the size of the pool of vesicles available for action potential dependent release. This novel presynaptic adaptation may provide important insights into the development of efficacious pain therapies that can circumvent the development of opioid tolerance.

Key points

• Chronic treatment with opioids, such as morphine, leads to analgesic tolerance.

• While postsynaptic opioid tolerance is well documented, the involvement of presynaptic mechanisms remains unclear.

• We show that chronic morphine reduces the ability of periaqueductal grey (PAG) neurons to maintain GABAergic transmission.

• This depression of GABAergic transmission was due to a reduction in the effective size of the readily releasable pool.

• This also led to a reduction in opioid presynaptic inhibition; these presynaptic adaptations need to be considered in the development of strategies to reduce opioid tolerance.

Abbreviations

ACSF

artificial cerebrospinal fluid

AM251

1‐(2,4‐dichlorophenyl)‐5‐(4‐iodophenyl)‐4‐methyl‐N‐1‐piperidinyl‐1H‐pyrazole‐3‐carboxamide

CGP55845

(2S)‐3‐[[(1S)‐1‐(3,4‐dichlorophenyl)ethyl]amino‐2‐hydroxypropyl](phenylmethyl)phosphinic acid hydrochloride

CNQX

6‐cyano‐2,3‐dihdroxy‐7‐nitro‐quinoxaline

EGTA

ethylene glycol tetra‐acetic acid

EPSC

excitatory postsynaptic current

IPSC

inhibitory postsynaptic current

PAG

periaqueductal grey

VGCC

voltage‐gated calcium current

Introduction

The midbrain periaqueductal grey (PAG) is a major site of the analgesic actions of opioids within the central nervous system. Opioids activate analgesic systems within the PAG by a process of disinhibition, that is, a reduction in GABAergic inhibition (Lau & Vaughan, 2014). At the cellular level, the prototypical opioid morphine and μ‐opioid selective agonists such as DAMGO produce disinhibition by two independent mechanisms within the PAG. These include direct postsynaptic inhibition of GABAergic neurons, and presynaptic inhibition of the probability of release (P _r) of GABA from nerve terminals (Chieng & Christie, 1994a, b; Vaughan & Christie, 1997). Together, this pre‐ and postsynaptic opioid disinhibition leads to activation of the PAG descending analgesic pathway which projects via the medulla to the dorsal horn in the spinal cord.

Chronic administration of morphine and other opioids leads to the rapid development of analgesic tolerance, so that larger doses are required to maintain a therapeutic effect (Williams et al. 2013). To date, most cellular studies on tolerance have focused on postsynaptic mechanisms, either in cell lines expressing opioid receptors, or in native brain neurons. These studies have consistently demonstrated that chronic morphine treatment leads to a reduction in postsynaptic coupling of μ‐opioid receptors to inwardly rectifying K⁺ conductances (GIRKs) and voltage gated calcium channels (VGCCs) within the PAG (Bagley et al. 2005a, b; Ingram et al. 2007, 2008; Connor et al. 2015; Wilson‐Poe et al. 2015). In contrast to the robust postsynaptic adaptations, the reported effects of chronic morphine treatment on presynaptic opioid receptor coupling to GABA release within the PAG are equivocal. In these studies, chronic morphine treatment has been shown to have either no effect, produce a decrease, or even an increase in presynaptic μ‐opioid receptor coupling to transmitter release (Ingram et al. 1998; Fyfe et al. 2010; Wilson‐Poe et al. 2015).

Neurotransmitter release is a complex process which is dependent upon a number of factors including, not only P _r, but also release synchrony and the pool of vesicles available for release (Neher, 2015). Interestingly, acute presynaptic activation of GABA_B G_i/o‐G protein‐coupled receptor (GPCRs) modulates both P _r and the size of the readily releasable pool (RRP) in the medial nucleus of the trapezoid body (Thanawala & Regehr, 2013). The impact of chronic morphine treatment on these presynaptic processes has not been examined in previous studies. Therefore, we examined the effect of chronic morphine treatment on GABAergic synaptic transmission within the PAG. Our results indicate that chronic morphine treatment has a profound impact upon the size of the readily releasable pool and this reduces the ability of μ‐opioids to presynaptically inhibit GABAergic synaptic transmission, particularly during burst stimulation. Thus, adaptations in the transmitter release apparatus can lead to a tolerance‐like adaptation at the presynaptic level.

Methods

Male Sprague‐Dawley rats (P22‐29) from Animal Resources Centre (Canning Vale, Australia) were used in all experiments. All animal protocols followed the guidelines of the NHMRC Code of Practice for the Care and Use of Animals in Research in Australia and were approved by the Royal North Shore Hospital Animal Ethics Committee. Animals were housed in groups of three in individually ventilated cages under a 12:12 h light/dark cycle, with environmental enrichment and free access to water and standard rat chow.

Chronic morphine treatment and behavioural testing

We utilized a drug administration protocol that has previously been shown to produce behavioural antinociceptive tolerance without the adaptations that occur during opioid withdrawal in both adolescent and adult rats (Ingram et al. 2007, 2008; Wilson‐Poe et al. 2015). Rats were administered either morphine (5 mg per kg of body weight), or saline twice daily (09.00 and 16.00 h) for 2 days (days 1 and 2), and were returned to their home cage after each injection. Animals were used for experiments on the following 2 days (days 3 and 4). Morphine (5 mg ml⁻¹ in saline), or saline were injected subcutaneously at a volume of 0.1 ml per 100 g of body weight. Animals were treated in groups of three (saline:morphine animal numbers = 2:1 and 1:2 on alternate weeks).

Electrophysiology

For in vitro experiments, rats were deeply anaesthetized with isoflurane and decapitated. The brain was quickly removed and submerged in ice‐cold artificial cerebrospinal fluid (ACSF, in mm; 126 NaCl, 2.5 KCl, 1.4 NaH₂PO₄, 1.2MgCl₂, 2.4CaCl₂, 11 glucose, and 25 NaHCO₃). In each animal, two to three coronal midbrain sections (300 μm) that contained the PAG were cut in ice‐cold ACSF using a vibratome (VT1000S, Leica Microsystems, North Ryde, NSW, Australia). PAG slices were maintained at 34°C, submerged in ACSF and equilibrated with 95% O₂ and 5% CO₂.

Slice recordings

In most experiments, slices were then individually transferred to the recording chamber (volume 0.8 ml) and superfused continuously (2.5 ml.min⁻¹) with 34°C ACSF. Ventrolateral PAG neurons (see inset in Fig. 1 A) were visualized using Dodt‐tube optics on an upright microscope. Whole‐cell patch clamp recordings of PAG neurons were obtained using an Axopatch 700B amplifier (Molecular Devices, Sunnyvale, CA, USA). The internal solution contained (in mm): 140 CsCl, 10 EGTA, 5 Hepes, 2 CaCl₂, 1 MgCl₂, 2 MgATP, and 3 QX‐314; with pH 7.3 and osmolality 270–290 mOsm l⁻¹. Series resistance (<20 MΩ) was compensated by 80% and continuously monitored during experiments. Recordings of evoked inhibitory postsynaptic currents (IPSCs) were obtained from PAG neurons, voltage‐clamped at −65 mV, in the presence of the non‐NMDA glutamate receptor antagonist 6‐cyano‐7‐nitroquinoxaline‐2,3‐dione (CNQX, 5 μm) and the glycine receptor antagonist strychnine (5 μm). Evoked IPSCs were elicited using a monopolar glass stimulating electrode filled with ACSF. In order to standardize the level of stimulation of synaptic inputs the tip of the stimulating electrode (1 MΩ) was placed 150 μm lateral to the recording electrode, at a depth 50 μm into the slice. In addition, a stimulation strength of 50 V (0.2 ms duration) was used in all experiments, except for those measuring the stimulus–response relationship (2.5–90 V). Single and paired (inter‐stimulus interval 70 ms) evoked IPSCs were elicited every 6 s. Train evoked IPSCs (20, or 50 stimuli at rates of 20 s⁻¹ and 100 s⁻¹) were elicited every 30 or 60 s. In train recovery experiments, a series of IPSCs were evoked at 0.05, 0.1, 0.2, 0.5, 1, 2, 4.5 and 10 s following each train.

Figure 1. Chronic morphine treatment reduces evoked IPSCs.

A, representative averaged traces of paired evoked IPSCs (inter‐stimulus interval 70 ms) at increasing stimulus intensities (5, 20, 50, 90 V) in PAG neurons from animals which had undergone saline, or morphine treatment. The inset shows the ventrolateral subdivision of the caudal‐intermediate PAG where neurons were located. B, summary plot of the stimulus–response relationship for evoked IPSC amplitude and paired pulse ratio in PAG neurons from saline and morphine treated animals (n = 12, 11). C, summary plot of the paired pulse ratio of evoked IPSCs for neurons from saline and morphine treated animals (n = 26, 20) at a mid‐range stimulus intensity (50 V). D, representative traces of evoked IPSCs in neurons from saline and morphine treated animals at the mid‐range stimulus intensity (25 consecutive traces with overlaid average). E, summary plot of the coefficient of variation (CV) of evoked IPSCs for neurons from saline and morphine treated animals at the mid‐range stimulus intensity (averaged response = thick line, n = 26, 22). In B, ^* P < 0.05, ^** P < 0.01 for saline versus morphine. In C and E, box and whiskers plots depict the 25th–75th percentile, and the min–max values, respectively. [Color figure can be viewed at wileyonlinelibrary.com]

Isolated neuron recordings

In some experiments, slices were treated with protease XIV (0.2 mg ml⁻¹) for 30–60 min at 32°C and subsequently with protease X (0.2 mg ml⁻¹) for 10–20 min at 32°C. After enzyme treatment, the slices were kept in the normal ACSF for 1 h. The ventrolateral region of the PAG was identified with a binocular microscope, and was micro‐punched out from the slice. These PAG sub‐sections were mechanically dissociated onto culture dishes (3801, Falcon, Dickinson, USA) with fire‐polished fine glass Pasteur pipettes filled with a solution containing (in mm): 150 NaCl, 5 KCl, 1 MgCl₂, 2 CaCl₂, 10 glucose, and 10 Hepes (pH was adjusted to 7.4 with Tris‐base), and kept at room temperature (22–24°C). Single neurons were visualized on an inverted microscope for whole‐cell patch clamp recordings at room temperature (22–24°C). GABA and muscimol were applied by a flow pipe positioned adjacent to the neurons.

Analysis

All recordings were filtered (5 kHz low‐pass filter) and sampled (20 kHz) for analysis (Axograph X, Axograph Scientific Software, Byron Bay, Australia). For analysis, evoked IPSCs were averaged over 6–8 consecutive episodes. For the assessment of single and paired (inter‐stimulus interval 70 ms) evoked IPSCs, the amplitude of evoked IPSCs was measured as the peak of the IPSC relative to a 0.2 ms baseline interval immediately preceding the stimulus artefact. The coefficient of variation was measured as the standard deviation of the amplitude of 25 consecutive evoked IPSCs divided by their mean amplitude.

For the assessment train evoked IPSCs, 20 or 50 stimuli at rates of 20 and 100 s⁻¹ were used. In these experiments charge transfer, rather than the amplitude of evoked IPSCs was measured because altered release synchrony can potentially lead to underestimation of IPSC charge transfer. The phasic charge transfer of evoked IPSCs during train stimulation was measured as the area under each IPSC relative to a 0.2 ms baseline interval immediately preceding that IPSC and the following IPSCs (see i in Fig. 4 B). The tonic charge transfer was calculated as the area under the phasic IPSC relative to the baseline preceding the first evoked IPSC in the train (see ii in Fig. 4 B). The total charge transfer of evoked IPSCs during train stimulation was calculated as the sum of the phasic and tonic components.

Figure 4. Chronic morphine treatment reduces train evoked IPSCs.

Raw traces of IPSCs evoked during repetitive stimulation in PAG neurons from saline (A and B) and morphine treated (C and D) animals (n = 8 per treatment group). Data are shown for trains of 20 stimuli evoked at 20 s⁻¹ (A and C) and 100 s⁻¹ (B and D). Summary plots of phasic charge transfer of evoked IPSCs (E) and tonic charge transfer (F) in saline and morphine treated animals at both stimulus frequencies. The inset in B depicts the calculation of the phasic (i) and tonic (ii) components of charge transfer. In E and F data are also shown for slices from morphine treated animals which were incubated in a cocktail of GPRC antagonists (Antags; see text, n = 7). [Color figure can be viewed at wileyonlinelibrary.com]

The effective pool size for action potential evoked transmitter release was measured using prolonged, high frequency train stimulation (50 stimuli at 100 s⁻¹) (Neher, 2015). Using this approach, high frequency stimulation was used to rapidly deplete the pool, with 100 s⁻¹ being the maximum frequency at which individual IPSCs could be resolved. The train length used for these experiments was 50 stimuli as this was found to be the number required to achieve a steady state level of release (balance between release and replenishment, see Fig. 5 A and D and Results). Pool size was first calculated by linear back‐extrapolation to the y‐axis from the final 10 points of the plot of cumulative evoked IPSC charge transfer; this gave the estimate RRP_SMN (Schneggenburger et al. 1999). We also made a correction to this linear back‐extrapolation to account for the dependence of vesicle replenishment upon the number of empty release sites, as this increases throughout the train (Thanawala & Regehr, 2013). To do this, the summed capacity for replenishment was calculated as the integral of the normalized differential of IPSC charge transfer during the train. This was then transformed by linear back‐extrapolation to obtain a corrected y‐intercept; this gave the estimate RRP_TR. Pool size was also calculated using a different approach which includes a simple kinetic correction for vesicle replenishment (Ruiz et al. 2011). In this approach, the decay in charge transfer of evoked IPSCs during train stimulation was fitted by the sum of a decaying exponential, to account for release of the pre‐existing pool of vesicles (A ₀e^{− n / K}), and a recovery exponential, to account for the release of newly recruited vesicles (A ₁[1 − e^{− n / K}]). In this estimate pool size, RRP_RBT, was estimated as the integral of the release from the pre‐existing pool of vesicles. The probability of release, P _r, was calculated as the first IPSC in the train divided by each of the estimates of pool size.

Figure 5. Chronic morphine treatment reduces the readily releasable pool size.

A and B, plots of the phasic charge transfer of evoked IPSCs during train stimulation (100 s⁻¹) in neurons from saline (A) and morphine treated (B) animals (n = 9 each treatment group). The data were fitted by the sum (continuous line) of an exponential decay (dashed line) and recovery (dotted line). C and D, plots of cumulative phasic charge transfer of evoked IPSCs in neurons from saline (C) and morphine treated (D) animals (error bars not included for clarity). RRP size estimates are shown by linear back‐extrapolation from the last 5 IPSCs in the train (RRP_SMN continuous lines, RRP_TR dashed line). The dotted lines are RRP_SMN back‐extrapolations from the 5 IPSCs before the 20th, 30th and 40th evoked IPSCs. E–G, summary plots of the phasic charge transfer of the first evoked IPSC in the train (E), RRP size (F) and probability of release, P _r (G). In G–I, ^* P < 0.05, ^** P < 0.001 for saline versus morphine. [Color figure can be viewed at wileyonlinelibrary.com]

To determine the time course of evoked IPSCs during train stimulation and in recovery experiments, the charge transfer of evoked IPSCs was fitted by a single exponential and the decay time constant was compared between neurons from saline and morphine treated animals using a sums‐of‐squares F‐test (Prism, Graphpad Software, La Jolla, CA, USA). Statistical comparisons were made using Student's t test (unpaired, factor = vehicle/morphine treatment), or two‐way ANOVA (between‐subjects factor = vehicle/morphine treatment; within‐subjects factor = train stimulus number, or pre/post‐DAMGO) (Prism). When ANOVA main effects, or interactions were significant post hoc comparisons were made using the Bonferroni adjustment for multiple comparisons. Differences were considered significant when P < 0.05. All data shown as means ± SEM.

Simulations

To explore the effect of reducing P _r and RRP size we used a simple simulation of train evoked IPSCs, as described previously (Thanawala & Regehr, 2013). The size of IPSCs in a train is given by:

$$\mathit{IPSC}{}_1=\mathit{IPSC}{}_0(1-P_r+P_{r}R),$$ (1)

$$\mathit{IPSC}{}_n=\mathit{IPSC}{}_{n-1}(1-P_r)(1-R)+P_r{N}_{0}R,$$ (2)

when there is replenishment at a rate R which is proportional to the extent of depletion, and N ₀ is the total number of release sites. Representative values of P _r and R were estimated by fitting eqns (1) and (2) to the summary data in Fig. 8 C and D using a least‐sums‐of squares error approach (Axograph). Simulated plots of cumulative and evoked IPSCs were then generated for normalized data with representative decreases in P _r and RRP using these equations (Axograph).

Figure 8. Simulation of the effect of reducing P _r and RRP size on evoked IPSCs.

Simulated plots of cumulative evoked IPSC amplitude (A) and evoked IPSC amplitude raw traces (B). Data are shown before and after a reduction in both P _r and RRP size (35% and 25% reduction), and after a reduction in P _r alone (35% reduction). Baseline data were normalized, with P _r of 0.1 and replenishment rate, R, of 0.03 (estimated by fitting to the baseline data in Fig. 7 C and D). [Color figure can be viewed at wileyonlinelibrary.com]

Drug solutions

Morphine sulfate was obtained from the National Measurement Institute (Lindfield, NSW, Australia). DAMGO, GABA, muscimol, picrotoxin, strychnine hydrochloride, protease XIV, protease X and N‐[2‐[4‐(2‐methoxyphenyl)‐1‐piperazinyl]ethyl]‐N‐2‐pyridinylcyclohexanecarboxamide maleate (WAY100635) were obtained from Sigma (Castle Hill, NSW, Australia); 1‐(2,4‐dichlorophenyl)‐5‐(4‐iodophenyl)‐4‐methyl‐N‐1‐piperidinyl‐1H‐pyrazole‐3‐carboxamide (AM251), CNQX, gabazine (SR95531), LY341495, naloxone hydrochloride, QX‐314 bromide, from Abcam Biochemicals (Cambridge, UK); CGP55845 and DPCPX were from Tocris Cookson (Bristol, UK). For in vitro experiments, stock solutions of drugs were made in distilled water, or DMSO then diluted (1:3000–10,000) to working concentrations in ACSF immediately before use and applied by superfusion to the slice chamber.

Results

Chronic morphine treatment reduces basal GABAergic synaptic transmission

We examined whether basal GABAergic synaptic transmission within the PAG is altered by a chronic low dose morphine treatment protocol that produces analgesic tolerance, but not withdrawal‐type behaviours (Ingram et al. 2008; Wilson‐Poe et al. 2015). To assess the effect of chronic morphine, animals were treated with morphine (5 mg kg⁻¹) or its vehicle saline twice daily for 2 days, and recordings of evoked IPSCs were made from ventrolateral PAG neurons in midbrain slices on the following 2 days. Increasing the stimulus intensity led to an increase in the amplitude of evoked IPSCs which differed between morphine and saline treated animals (Fig. 1 A and B, F(7,147) = 4.1, P < 0.001, n = 12, 11). The increase in evoked IPSC amplitude with stimulus intensity was less in PAG neurons from morphine compared to saline treated animals (Fig. 1 B, P < 0.01–0.05 saline versus morphine at 50–90 V).

At a mid‐range stimulus intensity (50 V), both paired pulse facilitation and depression were observed in neurons from saline and morphine treated animals when IPSCs were evoked at a short inter‐stimulus intervals (Fig. 1 A and C, n = 26, 20). At a mid‐range stimulus intensity (50 V), the paired pulse ratio of evoked IPSC amplitudes did not differ between saline and morphine treated animals (Fig. 1 C, P > 0.05). Furthermore, the paired pulse ratio of evoked IPSCs did not vary with stimulus intensity (Fig. 1 A and B, F(4,84) = 0.67, P > 0.05, n = 12, 11). At this stimulus intensity, the amplitude of single evoked IPSCs in neurons from both saline and morphine treated animals displayed variability (Fig. 1 D, n = 26, 20). The coefficient of variation of the amplitude of evoked IPSCs was not significantly different between neurons from morphine compared to saline treated animals (Fig. 1 E, P > 0.05). Together, these observations suggest that chronic morphine treatment produced a depression of evoked IPSCs.

Chronic morphine treatment does not affect quantal GABAergic synaptic transmission, or GABA_A receptors

The difference between evoked IPSCs in saline and morphine treatment animals may have been due to adaptations in postsynaptic GABA_A receptors. We examined this by assessing the response of acutely isolated ventrolateral PAG neurons. Fast application of the GABA_A agonist muscimol (0.1–1 μm) produced concentration dependent inward current (Fig. 2 A and B). The concentration dependence of the steady‐state muscimol current did not differ between neurons from saline and morphine treated animals (Fig. 2 C, F(2,36) = 0.8, P > 0.05, n = 6–8 per concentration in each treatment group). At the highest concentration tested, the muscimol induced current displayed desensitization (Fig. 2 A and B). The extent of muscimol desensitization did not differ between neurons from saline and morphine treated animals (Fig. 2 D, P > 0.05, n = 7 each for saline and morphine). In addition, the decay time constant of muscimol desensitization did not differ between neurons from saline and morphine treated animals Fig. 2 E, P > 0.05).

Figure 2. Chronic morphine treatment does not alter the response to exogenous muscimol and GABA.

Raw current traces during application of muscimol (0.1, 0.3 and 1 μm) to isolated PAG neurons from saline (A) and morphine treated (B) animals (n = 6–9). C, concentration–response curves of the steady state current produced by muscimol in isolated PAG neurons and by GABA in neurons in intact PAG slices (concentration in μm). D, the extent of desensitization during application of muscimol (1 μm) to isolated PAG neurons and of GABA (1000 μm) to neurons within intact PAG slices; shown as a percentage of the initial peak current. [Color figure can be viewed at wileyonlinelibrary.com]

The difference between saline and morphine treated animals may also have been due to other factors influencing postsynaptic GABA_A receptor activation such as uptake via GABA transporters in the slice preparation. We examined this by assessing the response of neurons to exogenous application of GABA in slices. GABA (10–1000 μm) produced a concentration dependent inward current (Fig. 2 C). The concentration dependence of the steady‐state GABA current did not differ between neurons from saline and morphine treated animals (Fig. 2 C, F(2,26) = 1.7, P > 0.05, n = 7, 8 at each concentration for saline and morphine). At the highest concentration tested, the GABA‐induced current displayed desensitization, and the extent of this did not differ between neurons from saline and morphine treated animals (Fig. 2 A, B and D, P > 0.05). It might be noted that the extent of muscimol desensitization in isolated neurons differed from that of GABA desensitization in slices (Fig. 2 D, F(1,26) = 6.7, P < 0.05).

We also examined whether the quantal nature of GABAergic synaptic transmission was altered by morphine treatment by comparing TTX‐resistant miniature IPSCs between treatment groups. In the presence of TTX (300 nm), the spontaneous rate of miniature IPSCs was not significantly different between neurons from morphine and saline treated animals (Fig. 3 A and C, P > 0.05, n = 17, 15), although there was a large within‐group variability as reported previously (Bobeck et al. 2014). The amplitude of miniature IPSCs was not significantly different between neurons from morphine and saline treated animals (Fig. 3 B and D, P > 0.05). In addition, miniature IPSCs in neurons of both morphine and saline treated animals displayed similar kinetics, with rise‐times and half‐widths which were not significantly different (Fig. 3 B, E and F, P > 0.05). Together, these results suggest that the impairment of baseline GABAergic transmission by chronic morphine treatment is not mediated by alterations in postsynaptic GABA_A receptors.

Figure 3. Chronic morphine treatment does not alter quantal IPSCs.

A, representative raw traces of spontaneous miniature IPSCs in neurons from saline and morphine treated animals (n = 17, 15). B, averaged miniature IPSCs in neurons from neurons represented in A (average of 224 and 281 IPSCs over 100 s in saline and morphine neurons). Summary plots of the rate (C), amplitude (D), rise‐time (E) and half‐width (F) of spontaneous miniature IPSCs in neurons from saline and morphine treated animals. In C–E, box and whiskers plots depict the 25th–75th percentile, and the min–max values, respectively. [Color figure can be viewed at wileyonlinelibrary.com]

Chronic morphine treatment reduces GABAergic synaptic transmission during repetitive activation

In intact animals, PAG neurons fire action potentials spontaneously at rates of up to 30 s⁻¹, and when activated, at rates of over 100 s⁻¹ (Behbehani et al. 1990). Thus, examination of single evoked IPSCs might not reflect the dynamic state of GABAergic synapses. We therefore compared the effect of repetitive stimulation at lower (20 s⁻¹) and near‐maximal (100 s⁻¹) physiological frequencies in neurons from saline and morphine treated animals (at the mid‐range stimulus intensity, 50 V). At the low stimulus frequency (20 s⁻¹), the phasic charge transfer of evoked IPSCs was maintained during train stimulation in neurons from both saline and morphine treated animals (Fig. 4 A, C and E, n = 10, 8 for saline, morphine). The time course of phasic charge transfer throughout a 20 s⁻¹ train did not differ between neurons from saline and morphine treated animals (F(19,304) = 1.4, P > 0.05). At the high stimulus frequency (100 s⁻¹), there was a gradual depression of evoked IPSCs in neurons from both saline and morphine treated animals (Fig. 4 B, D and E, n = 8, 8 for saline, morphine). The time course of phasic charge transfer throughout a 100 s⁻¹ train differed between neurons from saline and morphine treated animals (F(19,266) = 4.6, P < 0.0001). The decrease in evoked IPSC phasic charge transfer followed an exponential time course at this frequency; however, the decay time constant (τ) was similar in neurons from both treatment groups (Fig. 4 E, τ = 48 ± 11 ms and 61 ± 12 ms in saline and morphine, P > 0.05). In addition, the phasic charge transfer of evoked IPSCs decreased to a similar extent in neurons from saline and morphine treated animals (saline = 22 ± 3% and morphine = 21 ± 3% of the first evoked IPSC, P > 0.05). Thus, chronic morphine treatment does not alter the time course, or extent of depression during low and high frequency stimulation.

Train stimulation led to the development of a tonic current which was greater at the higher stimulus frequency (Fig. 4 A and C versus B and D). This tonic current could have been due to summation of IPSCs, or spill‐over onto adjacent extrasynaptic GABA_A receptors. We examined these alternatives by estimating the tonic current by superposition of single IPSCs at a high frequency (100 s⁻¹). The amplitude of the tonic current obtained by superposition of twenty single IPSCs was only 24 ± 3% and 20 ± 5% of that observed during a train in saline and morphine treated animals, respectively (P < 0.0001). Thus, the tonic current build up during train stimulation was most likely due to spill‐over of GABA onto extrasynaptic GABA_A receptors which were not activated during single evoked IPSCs. We have recently shown that there is a picrotoxin sensitive, but gabazine (10 μm) insensitive GABA_A mediated tonic current within the PAG (Lau et al. 2014). However, addition of gabazine (10 μm) abolished both the phasic IPSCs and tonic current elicited by train stimulation in PAG neurons from both saline and morphine treated animals (n = 10 each). Thus, the phasic IPSCs and tonic currents evoked by train stimulation were likely to be mediated by pharmacologically similar, but spatially distinct GABA_A receptor populations.

At some central synapses, a common pool of vesicles delivers synchronous phasic IPSCs and desynchronized GABA_A mediated currents (Lu & Trussell, 2000; Otsu et al. 2004). If the depression of evoked IPSCs in morphine treated animals was due to desynchronization, this would be observed as an increase in the tonic current during train stimulation. While the increase in the tonic charge transfer during low frequency (20 s⁻¹) train stimulation differed between neurons from saline and morphine treated animals (F(19,266) = 3.2, P < 0.0001), there was no difference in the tonic charge transfer at any individual IPSC within the train (Fig. 4 A, C and F, P > 0.05 for IPSC1–20). The increase in the tonic charge transfer during high frequency (100 s⁻¹) train stimulation also differed between neurons from saline and morphine treated animals (Fig. 4 B, D and F, F(19,266) = 6.1, P < 0.0001). At this frequency, the tonic charge transfer was greater in neurons from saline compared to morphine treated animals from IPSC5–20 (Fig. 4 F, P < 0.05). The tonic charge transfer at the end of the high frequency train was positively correlated to the phasic charge transfer of the 1st evoked IPSC amplitude in the train (linear regression, r ² = 0.51 and 0.56 for saline and morphine treated animals), and the slope of this relationship was not significantly different between neurons from saline and morphine treated animals (P > 0.05). This indicates that desynchronization was unlikely to account for the depression of evoked IPSCs in morphine treated animals.

Reduced GABAergic synaptic transmission following chronic morphine is not due to enhanced GPCR mediated presynaptic inhibition

Another potential mechanism underlying the morphine treatment induced depression of evoked IPSCs may have been presynaptic inhibition via endogenously released transmitters. We therefore examined the effect of incubating PAG slices from morphine treated animals in a cocktail of antagonists for GPCRs which are known to presynaptically inhibit synaptic transmission within the PAG, including μ‐opioid, GABA_B, cannabinoid CB1, 5‐HT1A, adenosine A1 and group II/III mGluR receptor antagonists (naloxone, CGP55845, AM251, WAY100635, DPCPX, LY341495, all 1 μm, n = 7). During high frequency (100 s⁻¹) stimulation, the phasic charge transfer of evoked IPSCs throughout the train did not differ between control and antagonist cocktail incubated slices from morphine treated animals (Fig. 4 E, F(19,247) = 0.8, P > 0.05). Likewise, the tonic charge transfer of evoked IPSCs throughout the train did not differ between control and antagonist incubated slices from morphine treated animals (Fig. 4 F, F(19,247) = 0.2, P > 0.05). Thus, presynaptic GPCR mediated inhibition was unlikely to make a contribution to the depression of evoked IPSCs in morphine treated animals during high frequency stimulation.

Chronic morphine treatment depletes the readily releasable pool

We next examined whether the changes in train‐evoked IPSCs induced by chronic morphine treatment were due to presynaptic adaptations in the readily releasable pool (RRP). As these experiments require maximal depletion of the pool of vesicles available for release, we used the highest train frequency at which individual evoked IPSCs could be resolved (100 s⁻¹). As observed with single and short train evoked IPSCs, the phasic charge transfer of the 1st evoked IPSC in trains of 50 stimuli was less in neurons from morphine compared to saline treated animals (Fig. 5 A, B and E, P < 0.05, n = 9 per treatment group). In addition, the evoked IPSC phasic charge transfer declined during train stimulation, with a decay time constant which was not significantly different between neurons from saline and morphine treated animals (Fig. 5 A and B, τ = 128 ± 14 ms and 113 ± 11 ms in saline and morphine, P > 0.05).

A commonly used method to estimate the effective pool size is linear back‐extrapolation of the plot of cumulative evoked IPSC phasic charge transfer (Schneggenburger et al. 1999). In order to obtain a reliable estimate of pool size, the train frequency and length must be sufficient to produce sufficient pool depletion (>60–90%) and reach a steady state level of release (Neher, 2015). This estimate of pool size, RRP_SMN, increased with the length of the train and approached an asymptote at train lengths of 40–50 stimuli, indicating that this was sufficient to obtain a reliable estimate of pool size (Fig. 5 C and D). Under these conditions, RRP_SMN was greater in neurons from saline compared to morphine treated animals (Fig. 5 F, P < 0.0001).

The rate of vesicle replenishment is dependent upon the number of available (empty) release sites, and this will increase during rapid train stimulation (Neher, 2015). Thus, RRP_SMN might have been an underestimate of pool size because it assumes that the rate of vesicle replenishment is constant throughout a train. We therefore also estimated pool size with a correction to the linear back‐extrapolation estimate based upon the increase in capacity for replenishment throughout a train (Thanawala & Regehr, 2013). This corrected estimate, RRP_TR, was greater than RRP_SMN in neurons from both saline and morphine treated animals (Fig. 5 C and D). Nonetheless, RRP_TR was still greater in neurons from saline compared to morphine treated animals (Fig. 5 F, P < 0.0001). We also used another approach that accounts for vesicle replenishment throughout a train (Ruiz et al. 2011). In this approach, the plot of phasic charge transfer was fitted by the sum of an exponential decay and an exponential recovery, to account for release of the pre‐existing and newly recruited vesicles, respectively (Fig. 5 A and B). This corrected estimated pool size, RRP_RBT, was similar to that observed for the RRP_TR, and was greater in neurons from saline compared to morphine treated animals (Fig. 5 F, P < 0.0001). The probability of release using the uncorrected and corrected estimates of pool size, P _r‐SMN, P _r‐TR and P _r‐RBT, was not significantly different between neurons from saline and morphine treated animals (Fig. 5 G, P > 0.05).

The difference in pool size between neurons from saline and morphine treated animals may have been due to changes in the rate of vesicle re‐filling during rapid pool depletion. To examine this we estimated the rate of vesicle replenishment by determining the recovery of evoked IPSCs following train stimulation. The recovery of the phasic charge transfer of single evoked IPSCs displayed an exponential recovery with time constants that were not significantly different between neurons from saline and morphine treated animals (Fig. 6, P > 0.05, saline = 1.1 ± 0.3 s, n = 6, morphine = 0.9 ± 0.4 s, n = 7).

Figure 6. The rate of recovery following train stimulation is unaffected by chronic morphine treatment.

The charge transfer of single evoked IPSCs at a range of time points following train stimulation. Data are averaged across all neurons from saline and morphine treated animals (n = 6, 7). Data are also shown for the first evoked IPSC in the train. [Color figure can be viewed at wileyonlinelibrary.com]

Chronic morphine alters μ‐opioid modulation of train evoked IPSCs

RRP analysis has recently been used to demonstrate that presynaptic GABA_B receptor activation modulates synaptic transmission by inhibiting both release probability and the size of the readily releasable pool (Thanawala & Regehr, 2013). We therefore examined whether the μ‐opioid receptor activation has a similar effect on GABAergic synaptic transmission in the PAG. In saline treated animals, a maximal concentration of μ‐opioid receptor agonist DAMGO (3 μm) produced a reduction in the charge transfer of the first evoked IPSC in the train (Fig. 7 A and E, P < 0.001, n = 7). The effect of DAMGO on the charge transfer of evoked IPSCs, however, varied throughout the train (F(50,600) = 1.4, P < 0.05). Thus, DAMGO produced an increase in the time constant of decay of evoked IPSCs (Fig. 7 A, P < 0.01, τ = 112 ± 10 ms and 238 ± 85 ms for pre‐ and + DAMGO IPSCs). The effect of DAMGO on pool size, RRP_SMN, was estimated by linear extrapolation of the cumulative IPSC plot (Fig. 7 C). In saline treated animals, DAMGO produced a reduction in RRP_SMN, plus a reduction in release probability, P _r‐SMN (Fig. 7 F and G, RRP_SMN: P < 0.01, 77 ± 5% of pre‐DAMGO, P _r‐SMN: P < 0.001, 56 ± 7% of pre‐DAMGO).

Figure 7. Chronic morphine reduces the DAMGO induced reduction in RRP size.

A and B, plots of the phasic charge transfer of evoked IPSCs during train stimulation (50 at 100 s⁻¹) in neurons from saline (A) and morphine treated (B) animals, prior to (Pre) and then during superfusion of DAMGO (n = 7 per treatment group); data were fitted by an exponential decay. C and D, plots of cumulative phasic evoked IPSC charge in PAG neurons from saline (C) and morphine treated (D) animals, prior to and then during DAMGO; RRP_SMN size estimates are shown by linear extrapolation from the last 10 IPSCs in the train. E–G, are charts of the effect of DAMGO on phasic charge transfer of the first evoked IPSC in a train (E), the pool size, RRP_SMN (F), and release probability, P_{r ‐SMN} (G), in PAG neurons from saline and morphine treated animals. ^* P < 0.05, ^** P <  0.01 and ^*** P < 0.0001 for pre‐ versus + DAMGO in each treatment group. [Color figure can be viewed at wileyonlinelibrary.com]

Given that chronic morphine treatment reduced the size of the readily releasable pool, we therefore examined whether this affected μ‐opioid modulation of GABA release probability and pool size. In morphine treated animals, DAMGO produced a reduction in the charge transfer of the first evoked IPSC (Fig. 7 B and E, P < 0.001, n = 7). In these animals, the effect of DAMGO on evoked IPSC charge transfer varied throughout the train (F(50,600) = 1.4, P < 0.05). Thus, DAMGO produced an increase the time constant of decay of evoked IPSCs (Fig. 7 B, P < 0.01, τ = 111 ± 11 ms and 206 ± 64 ms for pre‐ and + DAMGO). The effect of DAMGO on RRP_SMN, however, differed between neurons from saline and morphine treated animals (F(1,12) = 4.9, P < 0.05). Thus, DAMGO did not alter RRP_SMN in neurons from morphine treated animals (Fig. 7 D and F, P > 0.05; 97 ± 12% of pre‐DAMGO). The effect of DAMGO on P _r‐SMN did not differ between neurons from saline and morphine treated animals (F(1,12) = 0.1, P > 0.05). Thus, DAMGO produced a reduction in release probability in neurons from morphine treated animals (Fig. 7 G, P < 0.001, 49 ± 5% of pre‐DAMGO).

The above observations indicate that while DAMGO reduces both release probability and pool size in saline treated animals, it only reduces release probability in morphine treated animals. We used a simple model to simulate the impact of these difference on train evoked IPSCs (Thanawala & Regehr, 2013). Simulation of a reduction in P _r and RRP size of 35% and 25%, similar to that observed for saline treated animals, led to a decrease in the cumulative evoked IPSC plot (Fig. 8 A). Simulation of a reduction in P _r alone (35%) led to a lesser decrease in the cumulative evoked IPSC plot (Fig. 8 A). A reduction in both in P _r and RRP size led to a greater reduction in the first evoked IPSC than that produced by reducing P _r alone (Fig. 8 B). Furthermore, while inhibition of evoked IPSCs was observed later in the train when both P _r and RRP size were reduced, this inhibition was largely abolished when P _r alone was reduced (Fig. 8 B). This suggests that the maintenance of inhibition throughout train stimulation is dependent upon modulation of pool size, and this was reduced by morphine treatment.

Chronic morphine reduces μ‐opioid inhibition of short‐burst evoked IPSCs

We finally examined whether the change in μ‐opioid presynaptic inhibition had an effect under more physiological conditions, as rapid neuronal firing only occurs in short period bursts in the in vivo preparation (Behbehani et al. 1990). DAMGO (3 μm) produced a reduction in both the phasic evoked IPSCs and the tonic current evoked during short burst trains (20 stimuli at 100 s⁻¹) in neurons from both saline and morphine treated animals (Fig. 9 A and B, n = 11, 12 for saline and morphine treatment groups, respectively). We therefore quantified the effect of DAMGO on the total charge transfer, including both the phasic IPSC and the tonic component (Fig. 9 C). The DAMGO induced inhibition of total charge differed between neurons from saline and morphine treated animals (F(1,420) = 207, P < 0.0001). The DAMGO induced inhibition of total charge transfer was less in neurons from morphine compared to saline treated animals only from stimuli 3–20 in the train (Fig. 9 C, P < 0.05). Thus, morphine treatment reduces the DAMGO induced inhibition of total IPSC charge transfer, particularly during repetitive stimulation.

Figure 9. Chronic morphine treatment reduces the DAMGO induced inhibition of total charge transfer during repetitive stimulation.

Representative raw traces of IPSCs evoked during repetitive stimulation (20 at 100 s⁻¹) in PAG neurons from saline (A) and morphine treated (B) animals prior to and then during superfusion of DAMGO (n = 11, 12 neurons for saline and morphine). C, plot of the inhibition of total charge transfer by DAMGO throughout a train, in PAG neurons from saline and morphine treated animals; P < 0.01–0.05 for saline versus morphine at evoked IPSC3–17 and 19–20. [Color figure can be viewed at wileyonlinelibrary.com]

Discussion

The present study has demonstrated that chronic morphine treatment reduces the size of the pool of GABA vesicles readily available for action potential dependent release within the midbrain PAG. This depletion of pool size led to a reduction in μ‐opioid presynaptic inhibition of GABAergic synaptic transmission which was more apparent during repetitive stimulation. This suggests that chronic morphine treatment reduces presynaptic opioid inhibition within GABAergic nerve terminals as a result of adaptations in the neurotransmitter release machinery.

Chronic morphine‐induced depression of GABAergic synaptic transmission is due to a reduction in RRP size

In the present study, chronic morphine treatment produced a depression of GABAergic synaptic transmission onto PAG neurons. The chronic morphine induced depression of evoked IPSCs was not due to changes in postsynaptic GABA_A receptors because the amplitude of quantal GABAergic IPSCs and the response to exogenous GABA and muscimol were unaffected by morphine treatment. In addition, the depression was not due to extrinsic modulation by endogenously released neurotransmitters because it was unaffected by addition of antagonists for a range of GPCRs that have presynaptic actions within the PAG. These observations suggest that chronic morphine treatment leads to intrinsic presynaptic adaptations in GABAergic nerve terminals within the PAG.

A number of intrinsic presynaptic adaptations may have led to the observed depression of GABAergic synaptic transmission. This depression was not due to a reduction in the probability of release because chronic morphine had no effect on the paired pulse ratio and co‐efficient of variation of evoked IPSCs. The lack of a role of release probability was also consistent with the lack of effect of chronic morphine on the rate of spontaneous miniature IPSCs as observed in prior studies that have used a similar low dose morphine treatment regimen (20 mg kg⁻¹ total) that produce analgesic tolerance (Fyfe et al. 2010; Wilson‐Poe et al. 2015). It might be noted that these observations differ from studies that have observed an increase presynaptic excitability using higher dose regimens (300 mg kg⁻¹ total) that produce both tolerance and dependence/withdrawal (Ingram et al. 1998; Hack et al. 2003; Bagley et al. 2011). This suggests that the specific presynaptic adaptations within the PAG are dependent upon the chronic treatment regimen, and this might distinguish between the mechanisms associated with tolerance and withdrawal. The depression of evoked IPSCs could also have been due to desynchronization of release because a common pool of vesicles delivers synchronous phasic currents and desynchronized tonic currents at some central synapses (Lu & Trussell, 2000; Otsu et al. 2004). This was unlikely to be the case because morphine treatment reduced both the phasic IPSCs and the asynchronous GABA_A mediated tonic current which developed during high frequency stimulation.

A number of approaches were used to measure the size of the pool of vesicles available through action potential dependent release, including linear back‐extrapolation of the plot of cumulative evoked release and two forms of correction to account for replenishment of the vesicle pool during prolonged stimulation (Schneggenburger et al. 1999; Wesseling & Lo, 2002; Ruiz et al. 2011; Thanawala & Regehr, 2013; Neher, 2015). Analysis of high frequency, prolonged train evoked GABAergic IPSCs indicated that the chronic morphine induced depression of GABAergic synaptic transmission was largely due to a reduction in the size of the readily releasable pool of vesicles, although a role for release probability cannot be excluded. It should be noted there were a number of factors which limited interpretation of RRP analysis in our preparation, including the relatively low release probability and GABA_A receptor saturation, both of which affect estimation of pool size (Neher, 2015). These could not be overcome because increasing release probability leads to saturation of GABAergic synaptic transmission within the PAG (Aubrey et al. 2017). Another potential limitation was that GABA_A receptor desensitization during rapid train stimulation may have led to an underestimate of pool size (Taschenberger et al. 2005). While there are no good tools to limit GABA_A receptor desensitization (Neher, 2015), we found that the kinetics of miniature IPSCs and the desensitization to exogenous GABA/muscimol application were unaffected by chronic morphine treatment.

The chronic morphine induced depletion of the pool of ready releasable GABA may have been due to a number of factors. It was unlikely to be due to a reduction in the rate at which vesicles were replenished because the recovery rate of evoked IPSCs following train stimulation was unaffected by morphine treatment. It was also unlikely to be due to enhanced presynaptic excitability because of the lack of effect of chronic morphine treatment on release probability (see above). However, chronic morphine increases the postsynaptic excitability of PAG neurons (Chieng & Christie, 1996), and the resultant increase in spontaneous action potential induced transmitter release could lead to a pool depletion. Recently, it has been shown that pool size is dependent upon calcium influx into presynaptic terminals (Thanawala & Regehr, 2013). While morphine treatment does not affect the magnitude of VGCC current in PAG neurons, these studies have been carried out on postsynaptic ion channels (Connor et al. 1999; Bagley et al. 2005a), which may differ from presynaptic GABAergic terminals. It might be noted that the lack of effect of chronic morphine treatment on spontaneous miniature IPSC rate (see above) is inconsistent with a reduction in pool size. Evoked and miniature IPSCs are mediated by distinct Ca²⁺‐dependent and independent processes within the PAG (Vaughan & Christie, 1997; Aubrey et al. 2017). Given the dependence of pool size on Ca²⁺ influx (see above), alterations in pool size might not be equally reflected in measures of evoked and miniature IPSCs. The precise presynaptic mechanisms underlying the observed reduction in pool size remain to be determined, and we cannot rule out other mechanisms including a reduction in action potential propagation within nerve terminals, a loss of nerve terminals, or even silencing of synapses.

μ‐Opioid modulation of RRP size is reduced by chronic morphine treatment

Analysis of high frequency, prolonged train evoked GABAergic IPSCs also indicated that acute DAMGO application to slices produced a reduction in release probability in neurons from saline treated animals. This is consistent with prior studies which have observed that the μ‐opioid inhibition of single evoked IPSCs is associated with an increase in the paired pulse ratio of evoked IPSCs, and a reduction in the frequency of spontaneous miniature IPSCs (Vaughan & Christie, 1997; Vaughan et al. 1997). Interestingly, DAMGO also produced a decrease in the effective pool size in these animals. This is consistent with the demonstration that presynaptic GABA_B receptor activation and blockade of VGCCs reduces both release probability and effective pool size in other brain regions (Schneggenburger et al. 1999; Lou et al. 2008; Thanawala & Regehr, 2013). Given that activation of μ‐opioid and GABA_B receptors inhibits VGCCs within the PAG (Connor et al. 1999), it is possible that the effect of DAMGO on pool size was mediated by a reduction in presynaptic Ca²⁺ influx via VGCCs.

Chronic morphine treatment largely abolished the DAMGO induced reduction in effective pool size, but not the reduction in release probability. The difference between the effect of DAMGO on evoked IPSCs (and cumulative evoked IPSCs) in saline and morphine treated animals was reflected in models simulating inhibition of both release probability and pool size, versus inhibition of release probability alone (Thanawala & Regehr, 2013). One explanation for this is that depletion of the pool of vesicles available for release restricts the subsequent actions of DAMGO, such that it only acts to reduce release probability. Given the above stated dependence of pool size on presynaptic Ca²⁺ influx, it might also be speculated that this reduction in opioid modulation of pool size was due to the reduction in opioid receptor coupling to VGCCs produced by chronic morphine treatment (Bagley et al. 2005a; Connor et al. 2015).

While PAG neurons can fire action potentials at high rates in both in vivo and in vitro preparations, burst firing is only sustained for periods much shorter than those required for the above RRP experiments (Behbehani et al. 1990; Sanchez & Ribas, 1991). It was observed that chronic morphine treatment reduced the DAMGO induced inhibition of GABAergic total charge transfer during short burst stimulation, and that this became more prominent after the first few stimuli. This implies that the reduction in basal pool size produced by chronic morphine treatment could have a greater impact on GABAergic synaptic transmission under conditions of elevated neuronal firing within the PAG. The functional role of these changes remains to be determined because the source of the opioid sensitive GABAergic inputs was not identified in the present study. In this regard, it has been shown that GABAergic inputs to PAG neurons arise from both local interneurons and other brain regions such as the amygdala (Reichling & Basbaum, 1990; Oka et al. 2008; Tovote et al. 2016), although the opioid sensitivity of these inputs is not known.

Functional implications

The midbrain PAG has a crucial role in the analgesic actions of opioids, and in the tolerance to their analgesic actions (Jacquet & Lajtha, 1976; Chieng & Christie, 1996; Lane et al. 2005). There has been a lack of consensus on the effect of chronic morphine treatment on μ‐opioid presynaptic inhibition of GABAergic synaptic transmission within the PAG (Ingram et al. 2008; Fyfe et al. 2010; Bagley et al. 2011; Wilson‐Poe et al. 2015). We examined this by using a chronic morphine treatment protocol which produces tolerance to the analgesic effects of morphine, but does not produce withdrawal, or dependence (Ingram et al. 2007). While this experimental design does not directly identify the mechanisms underlying tolerance, it indicates that there are specific presynaptic adaptations within the PAG which are associated with the development of opioid analgesic tolerance.

Prior studies have demonstrated that μ‐opioids acutely inhibit GABAergic synaptic transmission by reducing the probability of transmitter release from nerve terminals within the PAG (Chieng & Christie, 1994b; Vaughan & Christie, 1997). The present observations suggest that, in addition to their effect on release probability, μ‐opioids reduce the size of the pool of GABA readily available for release. Interestingly, chronic morphine treatment led to a depression of GABAergic synaptic transmission which was largely due to a reduction in the pool of vesicles available for release. Under these conditions, μ‐opioids acutely reduced release probability, but had little effect on pool size. Functionally, this led to a reduction in presynaptic inhibition of GABAergic synaptic transmission, particularly during burst stimulation. This suggests that chronic morphine treatment reduces μ‐opioid presynaptic inhibition by reducing the pool of vesicle available for action potential‐dependent release. These findings have implications for the development of more efficacious pain therapies that can circumvent the development of opioid tolerance.

Additional information

Competing interests

The authors declare that they have no competing interests associated with this paper.

Author contributions

All of the experiments were carried out in the Pain Management Research Institute Research Laboratories, Kolling Institute of Medical Research, Northern Clinical School, The University of Sydney and Royal North Shore Hospital, St Leonards, New South Wales 2065, Australia. The study was conceived and designed by C.W.V. and A.W.P. Experimental data acquisition, analysis and interpretation of the data was by A.W.P., H.J.J. and C.W.V. Drafting and revision of the manuscript was by A.W.P., H.J.J. and C.W.V. All authors have approved the final version of the manuscript. All authors agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by Australian National Health and Medical Research Council Project Grant 1002680 (C.W.V.) and NIH Fellowship F32 DA034464 (A.W.P.).

Author's present address

Adrianne Wilson‐Poe: Pain Center, Department of Anesthesiology, Washington University School of Medicine, 660 South Euclid Ave, Campus Box 8054, St Louis, MO 63110, USA.