Vagal afferent control of opioidergic effects in rat brainstem circuits

Abstract

We demonstrated recently that increasing the levels of cAMP allows opioids to modulate GABAergic synaptic transmission between the nucleus of the tractus solitarius (NTS) and dorsal motor nucleus of the vagus (DMV). Using a combination of electrophysiological, immunohistochemical and biochemical approaches, we provide evidence that vagal afferent fibres dampen cAMP levels within the vagal brainstem circuits via tonic activation of group II metabotropic glutamate receptors (mGluRs). Whole-cell patch-clamp recordings were made from identified neurons of the rat DMV. Following chronic vagal deafferentation, the opioid agonist methionine-enkephalin (ME) inhibited the amplitude of evoked IPSC (eIPSC) in 32 of 33 neurons, without exogenous enhancement of cAMP levels. The ME-induced inhibition was prevented by the group II mGluR-selective agonist APDC. Following perfusion with the group II mGluR-selective antagonist EGLU, ME inhibited eIPSC amplitude in brainstem slices of control rats. Immunohistochemical experiments revealed that, following vagal deafferentation, μ-opioid receptors were colocalized on GABAergic profiles apposing DMV neurons; the number of colocalized profiles was significantly decreased by pretreatment with APDC. Radioimmunoassay and Western blot analysis showed that cAMP and phosphorylated cyclic AMP response element binding protein (pCREB) levels in the dorsal vagal complex were increased following vagal deafferentation. Our data show that by tonically dampening the levels of cAMP within the GABAergic synaptic contacts, activated group II mGluRs prevent the modulation of this synapse by endogenous opioids. These data suggest that the plasticity, hence the response, of central circuits controlling the vagal motor outflow to visceral organs is modulated and finely tuned by vagal afferent fibres.

Sensory information from the viscera is conveyed centrally via vagal afferent fibres to neurons of the nucleus tractus solitarius (NTS). These neurons send the modulated information to both higher CNS centres, which regulate body homeostasis, and to brainstem nuclei, which regulate visceral reflexes (Andresen & Kunze, 1994; Martinez & Tache, 2000; Hay et al. 2001; Jean, 2001; Travagli et al. 2006). The preganglionic vagal motor neurons to the gastrointestinal tract lie in the adjacent dorsal motor nucleus of the vagus (DMV) and receive robust glutamatergic, catecholaminergic and, mainly, GABAergic inputs from the NTS (Travagli et al. 1991, 2006; Sivarao et al. 1998; Rogers et al. 2003; Martinez-Pena y Valenzuela et al. 2004).

Glutamate is the principal neurotransmitter used by vagal afferent neurons; fast glutamatergic synaptic transmission, via activation of ionotropic receptors, is vitally important in the transmission of sensory information to the NTS (Andresen & Yang, 1990; Kawai & Senba, 1996; Smith et al. 1998; Hornby, 2001; Jean, 2001). Several types of metabotropic glutamate receptors (mGluRs) have also been identified within the dorsal vagal complex (DVC, i.e. NTS, DMV and area postrema) at both pre- and postsynaptic sites (Foley et al. 1998; Jones et al. 1998; Hay et al. 1999; Matsumura et al. 1999; Chen et al. 2002; Jin et al. 2004; Chen & Bonham, 2005; Sekizawa & Bonham, 2005) where they have an important role in the modulation of cardiovascular vagal reflexes (Pawloski-Dahm & Gordon, 1992; Foley et al. 1998; Viard & Sapru, 2006). In vitro electrophysiological studies have confirmed that groups II and III mGluRs are present, and modulate synaptic transmission from sensory afferents to the NTS as well as the NTS membrane itself (Glaum & Miller, 1992, 1993a,b; Chen et al. 2002; Jin et al. 2004; Chen & Bonham, 2005). It has been proposed that group II and III mGluRs act principally as autoreceptors, being located primarily at presynaptic sites, where they decrease neurotransmitter release. Both group II and III mGluRs inhibit adenylate cyclase and could have a role in controlling neurotransmission within brainstem circuits (Conn & Pin, 1997; Cartmell & Schoepp, 2000).

Our previous studies have suggested that modulation of GABAergic, but not glutamatergic, synaptic transmission between the NTS and the DMV is dependent upon the ‘state of activation’ of brainstem circuits. More specifically, under normal conditions, neuromodulators such as opioid peptides cannot affect GABAergic synaptic transmission between the NTS and the DMV, probably because of the low resting levels of cAMP within the inhibitory nerve terminals. Indeed, increasing the activity of adenylate cyclase with either forskolin or peptides such as cholecystokinin (CCK) or thyrotropin releasing factor (TRH) induces trafficking of μ-opioid receptors (MORs) to the cell membrane and permits modulation of GABAergic currents by opioid peptides (Browning et al. 2002, 2004; Travagli et al. 2003). We then put forward the hypothesis that as gastrointestinal vago-vagal reflexes can function independently of inputs from higher brain centres (Rogers et al. 2005), the vagal afferent inputs themselves might act to keep cAMP levels low within GABAergic NTS nerve terminals. Accordingly, neuromodulators such as opioid peptides would have no effect on GABAergic transmission at the NTS–DMV synapse unless the afferent input was removed or overcome by coincident activation of the cAMP–protein kinase A (PKA) pathway.

The aim of this study was to investigate whether the effects of opioids on GABAergic brainstem circuits are controlled by tonic sensory vagal afferent inputs via activation of mGluRs.

Methods

Research reported in the present manuscript fully conforms to National Institutes of Health guidelines and was approved by the Pennington Biomedical Research Center, Louisian State University (PBRC-LSU) System Institutional Animal Care and Use Committee.

Retrograde tracing

The neuronal tracer 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) was used to label gastric-projecting DMV neurons, as previously described (Browning et al. 1999). In brief, 14-day-old rat pups of either sex were anaesthetized deeply with isoflurane (2.5% in air, 600 ml min⁻¹; causing abolition of the foot-pinch withdrawal reflex). The abdominal area was cleaned with Novalsan before performing a laparotomy. Crystals of DiI were applied to one gastric region per rat, either to the major curvature of the fundus or corpus, or to the antrum/pylorus. The crystals were embedded in the application site by a fast-hardening epoxy compound before the entire surgical area was washed with warmed, sterile saline. The abdomen was closed with 5–0 suture and the rat allowed to recover for 10–15 days before experimentation.

Electrophysiological recording

Brainstem slices were prepared as previously described (Travagli et al. 1991; Browning et al. 1999). Briefly, rats were anaesthetized deeply with isoflurane before administration of a bilateral pneumothorax. The brainstem was removed and placed in oxygenated, chilled Krebs' solution (see below). A vibratome was used to cut four to six coronal slices (thickness, 300 μm) containing the DVC. Slices were incubated at 35°C for at least 90 min prior to use. A single slice was placed on a custom-made perfusion chamber (volume, 500 μl), kept in place with a nylon mesh and maintained at 35°C by perfusion with warmed Krebs' solution at a rate of 2.5–3.0 ml min⁻¹. Retrogradely labelled (DiI-containing) DMV neurons were identified using a Nikon E600FN microscope equipped with tetramethylrhodamine isothiocyanate (TRITC) epifluorescent filters. Once the identity of a DiI-filled neuron was confirmed, electrophysiological recordings were made under brightfield illumination using Nomarski optics.

Whole-cell recordings were made with patch pipettes with a resistance of 3–5 MΩ when filled with potassium gluconate intracellular solution (see below). A single electrode voltage-clamp amplifier was used (Axopatch 1D, Axon Instruments, Union City, CA, USA); data were acquired at 10 kHz, filtered at 2 kHz and digitized via a Digidata 1320 interface (Axon Instruments) before being stored and analysed on a personal computer utilizing pClamp8 software (Axon Instruments) or Mini Analysis software (Jaejin Software, Leonia, NJ, USA). Only those recordings with a series resistance <15 MΩ were accepted.

Drugs were perfused in the bath via a series of manually operated valves at concentrations demonstrated previously to be effective (Conn & Pin, 1997; Mitchell & Silver, 2000; Cartmell & Schoepp, 2000; Chen et al. 2002; Browning et al. 2002, 2004; Jin et al. 2004). Neurons were left to recover fully between drug additions. Antagonists were perfused for at least 5 min prior to reapplication of the agonist; for example, methionine-enkephalin (ME; 10 μm) was applied to observe the inhibition of evoked IPSCs (eIPSCs). After washout for 5 min, CTAP (1 μm) was applied for at least 5 min prior to perfusion with a solution containing both ME and CTAP. Each neuron served as its own control; a neuron was assessed before and after drug addition using either the grouped or paired Students t test or χ²-test with statistical significance set at P < 0.05. A minimum ± 10% variation of the amplitude of evoked currents was arbitrarily taken as indicative of an effect. Results are expressed as means ± s.e.m. Responding neurons only were included in the statistical analysis.

Electrical stimulation

Synaptic currents were evoked using tungsten bipolar stimulating electrodes (WPI, Sarasota, FL, USA) placed in the centralis or medialis subnuclei of the NTS. Pairs of stimuli (0.05–1.0 ms, 10–500 μA, 100–400 ms apart) were applied every 20 s to evoke submaximal IPSCs. No differences in the required stimulus parameters were noted between intact and deafferented groups. The perfusing Krebs' solution contained 1 mm kynurenic acid, to prevent glutamatergic currents, and the pipette solution contained the intracellular Na⁺ channel blocker QX314 (0.5 μm) to prevent antidromically activated action potentials. When recording spontaneous or miniature currents, the intracellular solution contained KCl rather than potassium gluconate. Recordings were conducted at −50 mV.

When experiments with drug pretreatments were conducted, the effects of ME on the eIPSC amplitude were measured either following washout of the drug of interest (i.e. when the eIPSC amplitude returned to baseline) or in the presence of the drug (i.e. against the new baseline, if the drug alone had an effect).

Vagal deafferentation

Both chemical (n = 18) and surgical (n = 15) methods were used to perform unilateral vagal nerve deafferentation. For chemical vagal deafferentation, rats were anaesthetized with a mixture of ketamine–acepromazine–xylazine in saline (80, 1.6 and 5 mg ml⁻¹, respectively, administering 0.1 ml (100 g body weight)⁻¹). Once a deep level of anaesthesia was obtained, the throat and chest were shaved and cleaned with Novalsan. A midline incision from the sternum to the lower jaw was used to expose one of the cervical vagus trunks, which was isolated from surrounding tissue by a small piece of Parafilm. A small (∼2 mm²) piece of absorbable gelatin sponge (Gelfoam, Upjohn, Ontario, Canada) soaked in capsaicin (1% w/v diluted in DMSO: ethanol: saline 1: 1: 8, respectively) was placed on the vagus nerve. After 30 min, the gelatin sponge was removed and the area blotted dry before suturing the incision with 5-0 silk. Sham chemical deafferentations were carried out in a similar manner except a gelatin sponge soaked in the diluent only was placed on the vagus nerve. The rat was allowed to recover for 7 days before experimentation.

Surgical vagal deafferentation was achieved by sectioning the vagal afferent nerve rootlets (afferent rhizotomy). Rats were anaesthetized with a mixture of ketamine–acepromazine–xylazine and, once a deep level of anaesthesia was obtained, they were placed in a stereotaxic frame. Following a dorso-lateral incision at the level of the occipital bone, muscle tissue was blunt dissected to expose an area at the level of the cervical vertebrae; the three supranodose vagal dorsal afferent rootlets on one of the vagal trunks were sectioned under microscopic guidance using a 27-gauge surgical needle. Sham surgical rhizotomies were carried out in a similar manner except the nerve rootlets were exposed but not sectioned. The incision was closed with 5-0 suture and the rats were allowed to recover for at least 4 days prior to experimentation.

Electrophysiological recordings were made from 149 gastric-projecting neurons from 53 rats. In more detail, recordings were made from 35 neurons from 16 control (vagal intact rats), 48 neurons from 15 rats that had undergone surgical vagal deafferentation, 53 neurons from 18 rats that had undergone chemical vagal deafferentation and, finally, recordings were made from 13 neurons from four rats that had undergone sham deafferentation (two surgical and two chemical shams). Neither qualitative nor quantitative differences in response to application of agonist or antagonist were observed with respect to the distinct gastric regions studied (i.e. fundus, corpus or antrum/pylorus), therefore all results were combined.

Although we realize that cutting the slice to perform the electrophysiological recordings implies that the vagus nerve is truncated, we still define the experiments as performed in ‘vagally intact’ rather than ‘acutely vagotomized’ rats because vagal afferent fibres are still viable in the slice preparation. Conversely, for the purposes of clarity, we define the experiments conducted in slices from animals that underwent deafferentation 4–7 days prior to the experiment as ‘deafferented’ rather than ‘chronic deafferented’.

Immunohistochemistry

The brainstems were extracted and sliced (thickness, 200 μm) from 20 rats as described above. Each group contained brainstem slices from more than one rat, and each rat provided samples to more than one group. All treatments were conducted in 20 ml oxygenated Krebs' solution maintained at 35°C. Experiments were carried out in five groups. Group 1: vagal intact, incubation in Krebs' solution for 60 min, n = 6; group 2: vagal intact stimulation of cAMP in forskolin (10 μm) for 60 min, n = 5; group 3: deafferentation, incubation in Krebs' solution for 60 min, n = 6; group 4: deafferentation + activation of group II mGluRs by incubation in APDC (100 μm) for 60 min, n = 7; and group 5: vagal intact + activation of group II mGluRs by incubation in APDC (100 μm) for 60 min, n = 7.

Following incubation, tissue slices were fixed in ice-cold Zamboni's fixative and stored at 4°C overnight. The slices were rinsed subsequently in phosphate-buffered solution (PBS; see below) and transferred into PBS containing 20% sucrose and stored for 3 days (with daily changes of the PBS–sucrose solution). Tissue sections were processed for double-labelling immunofluorescence using rabbit antisera directed against MOR-1 (1: 1000) and mouse monoclonal antibodies against glutamic acid decarboxylase (GAD-67; 1: 200) as previously described (Browning et al. 2002, 2004).

The diluent used for all antibodies was PBS containing 1% bovine serum albumin, 1% normal donkey serum and 0.3% Triton X-100. Sections were incubated with primary antibodies overnight at 4°C, washed in PBS (15min × 3) at room temperature of 22 ± 1°C (RT) and incubated for 1 h at RT with secondary antibodies (1: 1000 goat anti-rabbit Alexa 568 for MOR-1 staining; and 1: 1000 goat anti-mouse Alexa 488 for GAD staining).

Confocal microscopic images were collected by using a Zeiss 510 confocal scanning laser microscope equipped with a Kr–Ar laser and filters for the selective visualization of Texas Red and fluorescein isothiocyanate. Sections were mounted with Fluoromount (Southern Biotechnology Associates, Birmingham, AL, USA) to reduce fading.

Quantitative analysis was conducted on tissue sections in each group to identify labelled terminals. In each slice, the area analysed comprised most of the medial third of the DMV (where DMV gastric-projecting neurons are located) and the adjacent ventral border of the NTS; optical sections were taken at two different depths in the slice separated by at least 30 μm, and the average of the number of double-labelled profiles in the two optical section was considered as one measure. Profiles were defined as double labelled for MOR-1 and GAD if both labels appeared in profiles similar in size and geometry that overlapped upon merging of the images of MOR-1 (red) and GAD (green) resulting in yellow colour. Profiles double labelled for MOR-1 and GAD-6 were counted using Metamorph software (Universal Imaging Corp., Downingtown, PA, USA) on single optical layer sections, by an investigator who was unaware of the treatment, and expressed as MOR/GAD area of threshold colocalization. Cross-over compensation with Metamorph software was conducted to correct for fluorescence bleed-through.

In a further group of three rats, we determined whether group II mGluR-immunoreactivity (IR) colocalized with GAD-IR profiles. The brainstems were extracted and fixed in Zamboni's fixative. Brainstem slices (thickness, 40 μm) were treated as above using rabbit antisera directed against mGluRs 2/3 (1: 500 dilution) and mouse monoclonal antibodies against GAD. Qualitative analysis was conducted as above.

Radioimmunoassay

Brainstems were removed from seven rats that had undergone prior surgical vagal deafferentation (see above), and 400-μm thick coronal slices containing the DVC were cut in a Krebs' solution containing the phosphodiesterase inhibitor, isobutylmethylxanthine (500 μm) to limit cAMP degradation. The slices were frozen immediately on dry ice and the right (deafferented) and left (control) side from each brainstem was dissected via micropuncture using cylindrical steel punch tools (diameter, 0.5 mm; Brain Punch Set, Stoelting Co., Wooddale, IL, USA). Total protein was measured with the BCA protein assay kit (Pierce, Rockford, IL, USA). The punched area, spanning the rostro-caudal extent of the DVC, yielded an average of 200 μg total protein per side per rat. Briefly, the right and left DVC were homogenized and cyclic AMP levels were measured in the supernatants by radioimmunoassay (RIA) as previously described (Gettys et al. 1986, 1990, 1991). The assays were calibrated as before (Gettys et al. 1986), and after adjusting for protein concentration of the extracts, cAMP concentration (fmol (ng protein)⁻¹) in the contra- and ipsilateral sections were compared using a paired t test with significance set at P < 0.05.

Western blotting

The brainstem was removed from 18 rats (divided into four groups) 7 days after chemical vagal deafferentation (see above) and a vibratome was used to cut 300-μm thick coronal slices containing the DVC. The slices were frozen immediately and dissected via micropuncture as above. The right (deafferented) and left (control) sides of the DVC were collected and solubilized separately in 60 μl 2% SDS. Total protein assay was carried out as above.

Tissues for Western blots were prepared according to the methods described recently for the DVC (Sutton et al. 2004). An aliquot of the frozen sample was diluted with an equal volume of 5 × loading buffer (final concentration: 50 mm Tris-HCl, pH 6.7, 4% w/v glycerol, 4% SDS, 1% 2-mercaptoethanol and 0.02 mg ml⁻¹ bromophenol blue) and heated at 90°C for 5 min. The samples were then centrifuged at 10 000g, and 40 μg protein was separated by electrophoresis in a 10% Criterion Tris-HCl gel (Bio-Rad, Hercules, CA, USA) and transferred at 4°C to Immobilon-P polyvinylidene difluoride (PVDF) membranes (Bio-Rad) in Towbin-transfer buffer (25 mm Tris, 192 mm glycine, 20% methanol and 0.01% SDS) for 60–90 min. The sandwiched membranes–gel were rinsed in PBS with 0.05% Tween 20 (PBST) and then blocked by agitation in PBST containing 5% blotting non-fat milk for 60 min at RT. The membrane was then incubated with polyclonal rabbit anti-pCREB (1: 750 in PBS containing 0.05% blotting non-fat milk) overnight at 4°C with agitation. The PVDF membrane was washed in PBST for 60 min followed by overnight incubation at 4°C with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG secondary antibody (1: 1000 in PBS containing 0.05% blotting non-fat milk). The membrane was rinsed for 60 min in PBST and the antibody detected with chemiluminescence according to the manufactures' instructions, and visualized by exposure to Kodak BioMax XAR film developed in a Kodak processor (X-OMAT 1000 A). β-actin detection was used for internal control using mouse anti-β-actin (1: 5000) and goat anti-mouse HRP (1: 5000). Films were analysed by computer-assisted densitometry using Quantity One software (version 4.5.0, Bio-Rad), and data were expressed as the optical density of the ratio of pCREB to β-actin.

Drugs and solutions

Krebs' solution contained (mm): NaCl 126, NaHCO₃ 25, KCl 2.5, MgCl₂ 1.2, CaCl₂ 2.4, NaH₂PO₄ 1.2 and dextrose 11 maintained at pH 7.4 by bubbling with 95% O₂–5% CO₂. Intracellular potassium gluconate solution contained (mm): potassium gluconate 128; KCl 10, CaCl₂ 0.3, MgCl₂ 1, Hepes 10, EGTA 1, ATP-Na 2 and GTP-Na 0.25; pH adjusted to 7.35 with KOH. Intracellular potassium chloride solution contained (mm): KCl 140, CaCl₂ 1, MgCl₂ 1, Hepes 10, EGTA 10, ATP-Na 2 and GTP-Na 0.25; pH adjusted to 7.35 with HCl. Zamboni's fixative was composed of: 1.6% (w/v) paraformaldehyde, 19 mm KH₂PO₄ and 100 mm Na₂HPO₄ in 240 ml saturated picric acid, 1600 ml H₂O adjusted to pH 7.4 with HCl. PBS contained (mm): NaCl 115, Na₂HPO₄ 75 and KH₂PO₄ 7.5.

DiI was purchased from Molecular Probes (Eugene, OR, USA), antibodies against GAD-6 and group II/III mGluRs were from Chemicon (Temecula, CA, USA), antibodies against pCREB were from Upstate (Lake Placid, NY, USA) and antibodies against MOR were from Immunostar (Hudson, WI, USA). Neurobiotin and avidin-Texas Red were purchased from Vector Laboratories (Burlingame, CA, USA). l-(+)-2-amino-4-phosphonobutyric acid (L-AP4), 2R,4R-4-aminopyrrolidine-2,4-dicarboxylate (APDC) (RS)-α-cyclopropyl-4-phophonophenylglycine (CPPG), 2s-α-ethylglutamic acid (EGLU) (RS)-α-methylserine-O-phosphate (MSOP) were purchased from Tocris (Ellisville, MO, USA). Unless specified previously, all other chemicals were purchased from Sigma (St Louis, MO, USA).

Results

Effects of opioid peptides on inhibitory synaptic transmission

As previously reported (Browning et al. 2002, 2004), in control animals, perfusion with opioid receptor agonists had no effect on the amplitude of eIPSC unless the slice was pretreated with 10 μm forskolin. The amplitude of the eIPSC in control conditions was 183 ± 16.9 pA compared with 178 ± 12.3 pA in the presence of ME (10 μm, n = 19, P > 0.05) and it was reduced by 29 ± 2.4% following perfusion with forskolin and subsequent application of ME (P < 0.05 versus the first application of ME; Fig. 1A). In these experiments, the paired-pulse ratio, eIPSC2/eIPSC1, was increased from 0.96 ± 0.04 in control to 1.11 ± 0.06 in the presence of ME (P < 0.05; Fig. 1C). The alteration of the paired-pulse ratio suggests a presynaptic site of action of ME.

Figure 1. Opioid peptides modulate differently inhibitory synaptic transmission from deafferented and vagally intact rats.

A, representative (averaged) traces from a rat with intact vagus. In control conditions (left trace) perfusion with 10 μm ME does not affect the amplitude of the evoked IPSC. Following perfusion with forskolin 5 min, which did not alter per se the evoked IPSC significantly (middle trace) and washout for 5 min, reperfusion with ME decreased the amplitude of the evoked IPSC (right trace). B, summarized graphical representation of the IPSC paired-pulse experiments in control (intact vagus) rats. Perfusion with 10 μm ME (ME1) did not alter the amplitude of either the first (•) or the second (○) IPSC. Following perfusion with 10 μm forskolin for 5 min and washout, reapplication of 10 μm ME (ME2) decreased the amplitude of the first (•) more than the second (○) IPSC. Parallel lines indicate 5-min intervals. C, histogram summarizing the paired-pulse ratio (IPSC2/IPSC1) in control and following the second application of ME, respectively (ME2), in rats with intact vagus. D, representative (averaged) traces from a rat that underwent surgical deafferentation. In control conditions (left trace) perfusion with 10 μm ME decreased the amplitude of the evoked IPSC. Following perfusion with forskolin for 5 min, which did not alter the evoked IPSC significantly per se (middle trace) and washout for 5 min, reperfusion with ME decreased the amplitude of the evoked IPSC (right trace) to the same extent as before perfusion with forskolin. E, summarized graphical representation of the IPSC paired-pulse experiments in surgically deafferented rats. Perfusion with 10 μm ME (ME1) decreased the amplitude of the first (•) more than that of the second (○) IPSC. Following perfusion with 10 μm forskolin 5 min and washout, reapplication of 10 μm ME (ME2) again decreased the amplitude of the first (•) more than the second (○) IPSC. Parallel lines indicate 5-min intervals. F, histogram summarizing the paired-pulse ratio (IPSC2/IPSC1) in control and following the second application of ME, respectively (ME2), in surgically deafferented rats. *P ≤ 0.05.

In deafferented rats, ME inhibited the amplitude of eIPSC in naive (i.e. not pretreated) brainstem slices. Specifically, following chemical deafferentation, ME (10 μm) reduced the amplitude of eIPSC by 20.7 ± 1.7% from 201 ± 15.1 to 161 ± 13.8 pA (n = 19, P < 0.05). Similarly, following surgical deafferentation, ME reduced the amplitude of eIPSC by 26.7 ± 3.9% from 209 ± 24.1 to 154 ± 18.6 pA (n = 16, P < 0.05; Fig. 1). In these experiments, the paired-pulse ratio, eIPSC2/eIPSC1, was increased from 0.85 ± 0.07 in control to 1.08 ± 0.15 in the presence of ME (P < 0.05). These data suggest that removal of vagal afferent fibres overcomes the dampening of cAMP levels that prevents opioids from modulating GABAergic synapses.

As the results obtained following chemical or surgical deafferentation were similar, the data have been pooled hereafter; the number of chemical or surgically deafferented neurons is, however, specified for each experiment. This observation suggests that the majority of vagal afferent fibres involved in this phenomenon are capsaicin-sensitive rather than Aδ or capsaicin-insensitive C fibres.

In rats that underwent vagal deafferentation, incubation with forskolin (10 μm) had no additional effect on the inhibitory actions of ME. In six neurons (three chemically and three surgically deafferented), ME decreased the amplitude of eIPSC by 26 ± 6.2% (P < 0.05 versus control), following incubation with forskolin, subsequent reapplication of ME decreased the amplitude of eIPSC by 24 ± 4.9% (P < 0.05 versus control; P > 0.05 versus first ME application; Fig. 1B). In these experiments, eIPSC2/eIPSC1 was increased from 0.86 ± 0.07 in control to 1.09 ± 0.16 in the presence of ME (P < 0.05; Fig. 1F). These data suggest that in slices from deafferented rats, in contrast to slices from vagally intact animals, increasing further the activity of adenylate cyclase with forskolin does not induce an increase in the response to ME.

To verify that the ability of ME to inhibit GABAergic synaptic transmission following vagal deafferentation was, indeed, a direct result of the nerve interruption, the ability of ME to inhibit eIPSC was examined in 13 neurons from four sham operated rats (two sham surgical and two sham chemical deafferentations). As in control slices, in all 10 neurons recorded, ME was unable to inhibit the amplitude of eIPSC. Specifically, eIPSC amplitude was 165 ± 18.1 pA in control conditions and 172 ± 18.8 pA in the presence of ME (P > 0.05). In nine out of 10 of these neurons, the presynaptic inhibitory effects of ME on eIPSC were uncovered by incubation with forskolin (forskolin, 184 ± 21.5 pA; ME, 129 ± 11.2 pA; i.e. 28 ± 3.7% inhibition, P < 0.05).

Effects on inhibitory transmission are mediated via presynaptic receptors

As further support for a presynaptic site of action of ME, in addition to the alteration of the paired pulse ratio (see above), the amplitude of the response evoked by pressure ejection of the GABA_A agonist muscimol (100 μm) was unaffected by superfusion with ME (control, 306 ± 71.3 pA; ME 300 ± 69.3 pA; P > 0.05, n = 7 (four chemically and three surgically deafferented; data not shown).

In addition, perfusion with ME reduced the frequency of spontaneous IPSCs (sIPSCs) from 6.52 ± 2.43 to 1.48 ± 0.32 events s⁻¹ (P < 0.05, n = 9, four chemically and five surgically deafferented) without affecting their amplitude (96.7 ± 3.1% of control, P > 0.05, n = 9; data not shown). As the decrease in sIPSC frequency could be due to an effect of ME on either the axon or the cell body, we tested the effects of ME in the presence of 1 μm TTX. The blockade of TTX-sensitive Na⁺ channels significantly decreased the frequency, but not the amplitude, of sIPSCs. Perfusion with ME reduced the frequency of miniature IPSCs (mIPSCs) from 1.1 ± 0.16 to 0.6 ± 0.18 events s⁻¹ (P < 0.05, n = 7, all surgically deafferented) without affecting their amplitude (93 ± 4.0% of control, P > 0.05, n = 7; Fig. 2). Taken together these data indicate a presynaptic site of action of ME to modulate eIPSC amplitude.

Figure 2. Opioid peptides decrease the frequency but not the amplitude of miniature IPSCs in rats that underwent vagal deafferentation.

A, representative traces showing miniature IPSCs in control conditions (top trace). Perfusion with ME (10 μm) decreased the frequency but not the amplitude of spontaneous IPSCs (middle trace). The frequency of mIPSCs recovered towards control levels following washout (bottom trace). B, computer-generated graphics from the same neuron as in A showing the ME-induced alteration in frequency (top trace) but not amplitude (bottom trace) of mIPSCs. C, summary of ME-induced decrease in frequency (left; *P < 0.05) but not amplitude (right) of mIPSCs; n = 7.

Effects on inhibitory transmission are mediated via μ-opioid receptors

Following vagal deafferentation, the μ-opioid receptor antagonists CTOP or CTAP (both 1 μm) abolished the ability of ME to inhibit eIPSCs in all neurons tested. In six neurons (three chemically and three surgically deafferented) in which ME inhibited eIPSC amplitude (27 ± 8.6% inhibition; P < 0.05), perfusion with CTOP, which did not have any effect per se, prevented the ME-induced inhibition (CTOP, 154 ± 8.8 pA; ME + CTOP, 156 ± 12.3 pA; P > 0.05). Similarly, in another six neurons in which ME inhibited eIPSC amplitude by 18 ± 2.4%, perfusion with CTAP, which did not have any effect per se, prevented the ME-induced inhibition in eIPSC amplitude (CTAP, 179 ± 24.3 pA; ME + CTAP, 179 ± 23.0 pA; P > 0.05; Fig. 3).

Figure 3. Opioid peptides inhibit amplitude of evoked IPSCs via μ-opioid receptors only.

Histogram summarizing the involvement of μ-opioid receptors in the inhibition of IPSCs from rats that underwent vagal deafferentation (surgical deafferentation or capsaicin treatment). Perfusion with ME (10 μm) inhibited IPSC amplitude; pretreatment with forskolin (10 μm) did not elicit any further inhibition upon reapplication of ME. The ME-induced inhibition of IPSC amplitude was mimicked by the μ-opioid receptor agonist, DAMGO (1 μm) and both ME and DAMGO effects were prevented by the μ-opioid receptor antagonist, CTAP (1 μm). By contrast, the δ-opioid receptor agonist DPDPE (1 μm) and the κ-opioid receptor agonist U50,488 (1 μm) were without effect. *P < 0.05

In 18 neurons from deafferented rats (eight chemically and 10 surgically deafferented), the selective μ-opioid receptor agonist DAMGO (1 μm) inhibited eIPSC amplitude from 237 ± 21.4 pA in control to 185 ± 17.9 pA in the presence of DAMGO (i.e. 22 ± 1.8% inhibition; P < 0.05). In these experiments, eIPSC2/eIPSC1 was increased from 0.88 ± 0.15 in control to 0.98 ± 0.16 in the presence of DAMGO, (P < 0.05).

In four of these neurons (one chemically and three surgically deafferented), the DAMGO-induced inhibition (control, 232 ± 47.5 pA; DAMGO, 160 ± 25.2 pA; P < 0.05) was prevented by incubation with CTOP (CTOP, 214 ± 44.9 pA; DAMGO + CTOP, 235 ± 9.1 pA; P > 0.05), whereas in a further seven neurons (four chemically and three surgically deafferented) the DAMGO-induced inhibition (control, 249 ± 38.1 pA; DAMGO, 206 ± 35.7 pA, i.e. 19 ± 1.9% inhibition; P < 0.05) was prevented by incubation with CTAP (CTAP, 220 ± 38.1 pA; DAMGO + CTAP, 214 ± 39.7 pA; P > 0.05). In contrast, neither the δ-opioid receptor agonist, DPDPE (1 μm) nor the κ-opioid receptor agonist, U50,488 (1 μm) were able to inhibit eIPSC amplitude (control, 238 ± 28.0 pA; DPDPE, 240 ± 28.2 pA; and control, 245 ± 28.1 pA in versus U50,488, 252 ± 32.5 pA; for both P > 0.05, n = 4 chemically and n = 4 surgically deafferented; Fig. 3). These data indicate that μ-opioid receptors only are involved in the ME-induced inhibition.

The presynaptic μ-opioid receptors do not require trafficking to the nerve terminal membrane

Previously, we showed that activation of the cAMP–PKA pathway results in the trafficking of μ-opioid receptors to the GABAergic nerve terminal membrane via a brefeldin A-sensitive pathway (Browning et al. 2004). Following vagal deafferentation, however, the Golgi-disrupting agent brefeldin-A (5 μm) (Law et al. 2000; Chaipatikul et al. 2003) did not affect the ability of ME to inhibit eIPSC amplitude. Specifically, in seven neurons recorded from vagally deafferented rats (four chemically and three surgically deafferented), ME initially inhibited eIPSC amplitude (control, 220 ± 46.6 pA; ME, 165 ± 30.6 pA, i.e. 24 ± 3.6% inhibition; P < 0.05). Following incubation with brefeldin-A, however, the magnitude of the ME-induced inhibition was unaltered (i.e. 26 ± 4.3% inhibition; P < 0.05; P > 0.05 versus first ME application; data not shown). These data suggest that the effects of ME on slices from deafferented rats do not require additional receptor trafficking.

Measurement of intracellular cAMP and CREB phosphorylation

To test the hypothesis that vagal afferent fibres modulate cAMP levels and the activation state of PKA, intracellular cAMP levels and the phosphorylation state of CREB were measured in slices from vagally intact and surgically deafferented sides of 25 rats.

When expressed relative to the total protein content in each slice, steady-state cAMP levels were increased by deafferentation from 0.74 ± 0.18 to 1.16 ± 0.20 fmoles cAMP (ng protein)⁻¹ (n = 7, P < 0.05; Fig. 4).

Figure 4. Deafferentation increases the levels of pCREB and cAMP in the DVC.

A, immunoblots of pCREB and β-actin from micropunctures of the DVC obtained from control (vagally intact) and deafferented rats. Quantitative analysis of the immunoblots from four groups (total of 18 rats) expressed as a ratio of pCREB/β-actin. Note that the pCREB/β-actin ratio is higher in deafferented rats than in control rats. B, histogram illustrating the summarized results for cAMP and pCREB. Note that deafferentation increased the levels of both cAMP (expressed as fmoles (ng total protein)⁻¹and pCREB (expressed as optical density ratio of pCREB/β-actin,). *P < 0.05.

In four groups (total 18 rats) that underwent unilateral vagal deafferentation, the levels of pCREB were measured by Western blot and expressed as optical density relative to that of β-actin in the control (vagally intact) and deafferented side. As the assay is less sensitive than the RIA, we combined the protein extracts from all the rats in each group. Following deafferented, the optical density of pCREB increased relative to control from 0.7 ± 0.1 to 0.8 ± 0.1 (P < 0.05; Fig. 4). These data suggest that the levels of cAMP within the dorsal vagal complex are dampened by tonically active vagal afferent inputs.

Role of metabotropic glutamate receptors in presynaptic inhibitory effects of opioids

In four out of five neurons from control (vagally intact) rats, in which ME did not inhibit eIPSC amplitude (control, 157 ± 18.1 pA; ME, 162 ± 18.2 pA; P > 0.05), following incubation with the non-selective group II/III mGluR antagonist CPPG (100 μm), which did not have any effect per se, subsequent reapplication of ME inhibited eIPSC amplitude (control, 162 ± 19.7 pA; ME, 135 ± 18.1 pA, i.e. 17 ± 1.1% inhibition; P < 0.05). In these experiments, eIPSC2/eIPSC1 was increased from 0.72 ± 0.02 to 0.77 ± 0.02 in the presence of CPPG and CPPG + ME, respectively (P < 0.05). In a further six neurons in which ME had no effect on eIPSC amplitude (control, 179 ± 18.7 pA; ME, 182 ± 21.0 pA; P > 0.05), following incubation with the selective group II mGluR antagonist EGLU (200 μm), reapplication of ME inhibited eIPSC amplitude (control, 188 ± 16.1 pA; ME, 135 ± 11.3 pA, i.e 28 ± 4.6% inhibition; P < 0.05). In contrast, in a further five neurons, incubation with the selective group II mGluR agonist APDC (100 μm) did not unmask the ability of ME to alter eIPSC amplitude (0.1 ± 2.8% inhibition; P > 0.05). Similarly, in four of five neurons (control, 171 ± 34.2 pA; ME, 181 ± 35.3 pA; P > 0.05), incubation with the selective group III mGluR antagonist MSOP (500 μm), which did not have any effect per se, did not unmask the ability of ME to alter eIPSC amplitude (3 ± 3.1% inhibition; P > 0.05; Fig. 5).

Figure 5. Role of mGluR agents on ME-induced inhibition in IPSC amplitude.

A, representative traces of IPSCs evoked in a gastric-projecting DMV neuron under control (vagally intact) conditions (left). Note that ME (10 μm) has no effect on IPSC amplitude. Following superfusion with the group II mGluR-selective antagonist, EGLU (200 μm), subsequent reapplication of ME inhibited IPSC amplitude. Histogram summarizing the effects of mGluR agents on ME-induced inhibitions in IPSC amplitude (right). Under control (vagally intact) conditions ME has no effect on IPSC amplitude (the data represent the pooled means ± s.e.m. of control experiments). In four out of five neurons, ME inhibited IPSC amplitude following incubation with the non-selective group II/III mGluR-selective antagonist, CPPG (100 μm; filled bars). Similarly, in six out of six neurons following incubation with the selective group II mGluR antagonist, EGLU (200 μm), ME decreased IPSC amplitude. By contrast, in four out of five neurons ME had no effect on IPSC amplitude following incubation with the selective group III mGluR antagonist, MSOP (500 μm). In five out of five neurons, ME had no effect on IPSC amplitude following incubation with the group II mGluR agonist APDC (100 μm). By comparison, ME inhibited IPSC amplitude in four out of five neurons following incubation with the GABA_B receptor-selective antagonist, saclofen (300 μm). *P < 0.05 versus ME alone. B, representative traces of IPSCs evoked in a gastric-projecting DMV neuron from a rat that underwent vagal deafferentation (left). Note that following incubation with the group II mGluR-selective antagonist, APDC (100 μm), ME (10 μm) has no effect on IPSC amplitude (cf. Fig. 1, A). Histogram summarizing the effects of mGluR agents on ME-induced inhibition of IPSC amplitude in vagally deafferented DMV neurons (right). Under control deafferentation conditions (open bar), ME (10 μm) inhibited IPSC amplitude (the data represent the pooled means ± s.e.m. of control experiments). Following incubation with the group II mGluR-selective agonist, APDC (100 μm), ME was unable to inhibit IPSC amplitude in any of the nine neurons tested. By contrast, incubation with either the group II mGluR-selective antagonist EGLU (200 μm; n = 5/5 neurons) or the group III mGluR-selective agonist, L-AP4 (100 μm; n = 8/8 neurons) had no effect on the ability of ME to inhibit IPSC amplitude in any of the neurons tested. By comparison, neither the GABA_B receptor-selective agonist baclofen (30 μm) nor the CCK receptor selective antagonist lorglumide (1 μm) affected the ability of ME to inhibit IPSC amplitude (n = 10 and n = 8 neurons, respectively). *P < 0.05 versus ME control.

Following vagal deafferentation, application of group II but not III mGluR agonists blocked the ability of ME to inhibit eIPSC amplitude. In nine neurons from vagal deafferented rats (five chemically and four surgically deafferented), prolonged incubation (120 min) with the selective group II mGluR agonist APDC (100 μm) abolished the ability of ME to inhibit eIPSC amplitude (APDC, 138 ± 13.2 pA; APDC + ME, 138 ± 11.1 pA; P > 0.05). In contrast, in five neurons from surgically deafferented rats, the group II mGluR antagonist EGLU (200 μm) did not alter the ability of ME to inhibit eIPSCs (EGLU, 324 ± 26.6 pA; EGLU + ME, 251 ± 22.3 pA, i.e. 22.6 ± 2.1% inhibition; P < 0.05). Similarly, in a further nine out of 10 neurons from deafferented rats (five chemically and four surgically deafferented), incubation with the selective group III mGluR agonist L-AP4 (100 μm; 120 min) did not alter the ability of ME to inhibit eIPSCs (L-AP4, 184 ± 24.4 pA; L-AP4 + ME, 141 ± 22.1 pA, i.e. 26 ± 3.3% inhibition; P < 0.05; Fig. 5). These data suggest that blockade of the tonic activation of group II mGluRs, either with pharmacological antagonists or following the ablation of vagal afferent fibres, uncovers the inhibitory effects of ME on the GABAergic NTS–DMV synapse.

Immunohistochemical localization of group II mGluR, MOR and GAD

Glutamate is the main neurotransmitter utilized by vagal afferent fibres to convey visceral sensory information to the NTS. It is well known that glutamate interacts with both ionotropic and metabotropic glutamate receptors (mGluRs). Of the mGluRs present in the DVC, only group II and III mGluRs are capable of decreasing the levels of cAMP.

In three rats, we observed qualitative immunohistochemical colocalization of group II mGluRs and GAD on profiles throughout the rostro-caudal extent of the DVC (Fig. 6).

Figure 6. 2/3 mGluR and GAD-IR colocalize on DVC profiles.

2/3 mGluR-IR, red; GAD-IR, green; profiles double-labelled for 2/3 mGluR and GAD, yellow. A, low-magnification micrograph illustrating a portion of the DVC. B, high-magnification micrograph illustrating the detail of several 2/3 mGluR-IR profiles colocalized with GAD-IR (arrowhead) within the NTS. The magnification is insert ‘b’ shown in A. C, high-magnification micrograph illustrating the detail of several MOR-IR profiles colocalized with GAD-IR (arrowhead) within the DMV. The magnification is insert ‘c’ shown in A. Scale bar, 20 μm in A; 2 μm in B and C.

As reported previously (Browning et al. 2002, 2006), few profiles double-labelled for MOR-1 and GAD-6 were observed in the DVC from control rats (50.1 ± 28.3; n = 6). In slices taken from rats that underwent surgical vagal deafferentation, however, the colocalization of MOR-1 and GAD-6 was enhanced compared to control (128.8 ± 64.14 (co-localized profits or double labelled profiles), n = 6; P < 0.05), and was similar to the number of double-labelled profiles measured in control rats following incubation with 10 μm forskolin (100 ± 42.5 (co-localized profiles or double labelled profiles); n = 5).

Whereas incubation of slices from control (vagally intact) rats with the group II mGluR agonist APDC (100 μm) did not affect the absolute number of double-labelled profiles (41.3 ± 21.6; n = 7), incubation with 100 μm APDC in slices from deafferented rats significantly decreased the number of double-labelled profiles (22.4 ± 12.31; n = 7; P < 0.05).

To eliminate the potential bias of large interday variability in the absolute number of labelled profiles, the same slice served both as control (intact side) and as treatment measurement (deafferented side), and we expressed the results as deafferentation/control ratio. Thus, following deafferentation, the ratio of colocalized profiles increased to 4.07 ± 1.19 whereas APDC decreased the ratio of colocalized profiles to 0.65 ± 0.12 (P < 0.05; Fig. 7). These data indicate that vagal group II mGluRs colocalize with GAD-6-positive profiles in the DVC and that afferent inputs modulate the colocalization of MOR-1 and GAD-6 via group II mGluRs.

Figure 7. Vagal deafferentation increases expression of MOR on GABAergic profiles.

MOR-IR, red; GAD-IR, green; profiles double-labelled for MOR and GAD, yellow. A, low-magnification, micrograph illustrating a portion of the DVC from the control (vagally intact) side of the brainstem. B, high-magnification micrograph illustrating the detail of a MOR-IR profile colocalized with GAD-IR (arrowhead). The magnification is insert ‘b’ shown in A. C, low-magnification micrograph illustrating a portion of the DVC from the contralateral (deafferented) side of the same slice as in A. D, high-magnification micrograph illustrating the detail of several MOR-IR profiles colocalized with GAD-IR (arrowhead). The magnification is insert ‘d’ shown in C. Scale bar, 20 μm in A and C; 2 μm in B and D. E, histogram illustrating the summarized results of the immunohistochemical experiments expresses as number of colocalized profiles. The number of colocalized profiles has been calculated on the whole field of view as in A or C. *P < 0.05 versus deafferentation. F, histogram illustrating the ratio of colocalized profiles in deafferented versus control slices. Note the increased ratio following vagal deafferentation; pretreatment with the group II mGluR agonist APDC significantly reduced the ratio (*P < 0.05).

Role of other adenylate cyclase-coupled receptors in presynaptic inhibitory effects of opioids

We also examined the influence of other adenylate cyclase-coupled receptors in rats that underwent vagal deafferentation. As we have previously demonstrated that CCK can uncover presynaptic inhibitory effects of ME (Browning et al. 2004) via its actions to activate adenylate cyclase, we examined whether a tonic CCK input to GABAergic nerve terminals was disrupted following vagal deafferentation. We thus assessed the actions of the selective CCK_A receptor antagonist lorglumide (1 μm) to antagonize the inhibitory effects of ME. In eight out of nine neurons from vagal deafferented rats (five chemically and three surgically deafferented), prolonged incubation with lorglumide did not alter the ability of ME to inhibit eIPSC amplitude (lorglumide, 176 ± 25.2 pA; lorglumide + ME, 124 ± 19.4 pA, i.e. 30 ± 3.5% inhibition; P < 0.05; Fig. 5). These data indicate that the uncovering of the ME effects in deafferented rats is not due to the disruption of a tonic CCK-mediated input.

Finally, to investigate whether tonic activation of presynaptic GABA_B receptors inhibits cAMP levels within GABAergic nerve terminals, we examined the ability of the selective GABA_B receptor antagonist saclofen (300 μm) to uncover presynaptic inhibitory effects of ME in control (vagally intact) brainstem circuits. In five neurons in which ME had no initial effects (control, 104 ± 14 pA; ME, 101 ± 14 pA; P > 0.05), following incubation with saclofen, subsequent reapplication of ME inhibited eIPSC amplitude in four out of the five neurons (control, 124 ± 14.6 pA ME, 92 ± 16.4 pA, i.e. 28 ± 6.2% inhibition; P < 0.05). However, following vagal deafferentation (five chemically and five surgically deafferented) prolonged incubation with the selective GABA_B receptor agonist baclofen (30 μm) did not abolish the ability of ME to inhibit eIPSC amplitude (control, 233 ± 35.8 pA; ME, 178 ± 30.7 pA, i.e. 26 ± 3.6% inhibition; P < 0.05; Fig. 5). These data suggest that the role of GABA_B receptors is to modulate, but not to control, the afferent vagal input.

Role of endogenous opioids in regulating inhibitory synaptic transmission

The analgesic dipeptide kyotorphin has been shown to promote the release of ME from nerve terminals independently from binding to opioid receptors (Shiomi et al. 1981; Rackham et al. 1982). In slices from control (vagal intact) rats, perfusion with kyotorphin (10 μm) had no effect on the amplitude of eIPSCs (control, 141 ± 20.8 pA; kyotorphin, 138 ± 22.2 pA; P > 0.05; n = 5). Following incubation with forskolin (10 μm), however, subsequent reapplication of kyotorphin resulted in a decrease in eIPSC amplitude (control, 163 ± 3.0 pA; kyotorphin, 119 ± 6.4 pA, i.e. 27 ± 4.2% inhibition; P < 0.05, n = 3/3 neurons tested). In rats that underwent vagal deafferentation, kyotorphin was able to decrease eIPSC amplitude in eight out of nine neurons (four chemically and four surgically deafferented) without the need for application of forskolin (control, 302 ± 55.7 pA; kyotorphin, 242 ± 56.1 pA, i.e. 26 ± 5.7% inhibition; P < 0.05; data not shown). In these experiments, eIPSC2/eIPSC1 was increased from 0.69 ± 0.1 to 0.79 ± 0.1 in control and in the presence of kyotorphin, respectively (P < 0.05). These data suggest that even in deafferented rats the kyotorphin-induced release of endogenous opioids is capable of modulating the GABAergic NTS–DMV synapse.

Discussion

This study demonstrates that the inhibitory action of opioid peptides on the NTS–DMV GABAergic synapse requires the ongoing activation of mGluRs by vagal afferent fibres to be overcome. Afferent glutamatergic input impedes the effects of opioid peptides via tonic activation of group II mGluRs; removal of this tonic input via experimental vagal deafferentation or, in physiological conditions, by overcoming vagal afferent signalling relieves the persistent, mGluR-mediated inhibition of cAMP, allowing the interaction of enkephalins with MORs on GABAergic nerve terminals. Our conclusions are supported by electrophysiological, anatomical and biochemical evidence.

We demonstrated previously that neuromodulators such as opioid peptides are normally unable to inhibit GABAergic NTS–DMV synaptic transmission and few, if any, MORs are located on GABAergic NTS–DMV profiles (Browning et al. 2002). Following activation of the cAMP–PKA pathway, MORs are rapidly (within minutes), although transiently (1 h), translocated to the GABAergic nerve terminal membrane where they can interact with opioid peptides (Browning et al. 2004). This type of cAMP-mediated modulation is characteristic of NTS–DMV GABAergic synapses as adjacent glutamatergic contacts are unaffected (Browning et al. 2004). Such observations led us to question why GABAergic, but not glutamatergic, transmission within the DVC is regulated in this manner. As gastrointestinal vago-vagal reflexes can function independently of inputs from higher brain centres (Rogers et al. 2005), we hypothesized that vagal afferent inputs themselves might dampen cAMP levels within these brainstem circuits. Accordingly, neuromodulators such as opioid peptides would have no effect on GABAergic synaptic transmission unless the afferent input was removed/overcome by concomitant activation of the cAMP–PKA pathway. Indeed, activating the cAMP–PKA pathway with hormones, such as CCK or TRH, or with cAMP–PKA activators such as forskolin or 8Br-cAMP, exposes the NTS–DMV GABAergic synapse to modulation by opioid peptides (Browning et al. 2004).

Here, chemical C-fibre ablation (Holzer, 1991; Raybould & Holzer, 2000) or surgical deafferentation resulted in inhibition of eIPSCs and mIPSCs by ME without prerequisite activation of cAMP–PKA at the GABAergic NTS–DMV synapse. Indeed, further activation of adenylate cyclase by forskolin did not elicit any additional inhibitory effects of ME on eIPSCs, and the effects of ME were not antagonized by the Golgi-disrupting agent brefeldin-A. These observations suggest that, unlike in vagally intact rats, MOR insertion on GABAergic NTS terminals has already occurred in deafferented rats, possibly to its maximal extent. Pharmacological experiments with CTAP/CTOP and DAMGO confirmed that the inhibitory effects of ME were the result of actions at MORs; alterations of the paired-pulse ratio and decrease of the mIPSC frequency, but not of a muscimol-induced current, confirmed the presynaptic site of action.

The effects of ME on deafferented animals were due to removal of the brake on cAMP levels by tonic mGluR vagal inputs. Our RIA and Western blot analysis of cAMP and pCREB levels (used as a marker for cAMP concentration; Sutton et al. 2004) within the DVC indicated that vagal deafferentation increased the levels of both, suggesting that in normal conditions vagal afferent input tonically inhibits cAMP levels within the NTS. Although many signalling factors affect the phosphorylation state of CREB, when viewed in conjunction with the increased cAMP levels, we feel confident that the observed changes in pCREB were also due to increased cAMP–PKA activity.

These results raise the question as to the mechanism of this vagal control of the GABAergic NTS–DMV synapse. Vagal afferent terminals use glutamate as their principal neurotransmitter (Andresen & Kunze, 1994; Aylwin et al. 1997; Smith et al. 1998). Although ionotropic transmission is vital for the integration of sensory information, its mechanisms and fast timescale mean it is unlikely to be responsible for the constant modulation of cAMP levels. However, several types of mGluR have already been identified (anatomically and electrophysiologically) within the NTS, both pre- and postsynaptically (Glaum & Miller, 1992, 1993a,b; Jones et al. 1998; Hay et al. 1999; Foley et al. 1999; Matsumura et al. 1999; Jin et al. 2004). We extended these previous observations by showing that group II mGluRs are present on GABAergic profiles apposed to both NTS and DMV neurons. Furthermore, electrophysiological experiments have demonstrated that electrical stimulation of the tractus solitarius, by activating mGluRs, decreases the efficacy of GABAergic transmission in interneurons impinging on NTS baroreceptor neurons (Chen & Bonham, 2005), providing a mechanistic explanation for the known role for mGluRs in shaping brainstem cardiovascular reflexes (Pawloski-Dahm & Gordon, 1992; Foley et al. 1999; Viard & Sapru, 2006).

It is well established that (1) group II and III mGluRs act principally as presynaptic autoreceptors, decreasing neurotransmitter release; and (2) both groups are negatively coupled to adenylate cyclase (Conn & Pin, 1997; Cartmell & Schoepp, 2000). These accepted functions make both mGluR groups potential candidates for controlling the cAMP levels in brainstem circuits. Here, following blockade of group II, but not group III mGluRs, ME inhibited eIPSC amplitude in control brainstem slices. Acute blockade of tonically activated group II mGluRs by the selective antagonist EGLU increased cAMP–PKA activity allowing the expression of MORs on GABAergic nerve terminal membranes where they inhibited synaptic transmission. Conversely, incubation with group II mGluR agonists prevented the ME-induced inhibition of GABAergic synaptic transmission in deafferented brainstem slices. These observations indicate the presence, in control animals, of a tonically active glutamatergic input that, via interaction with group II mGluRs, modulates the availability of MOR on the NTS–DMV GABAergic synapse.

Our immunohistochemical analyses agree with these electrophysiological conclusions. MORs are not normally colocalized on GAD-IR terminals within the DVC (Browning et al. 2002, 2004); incubation of brainstem slices with group II or mixed group II/III mGluR antagonists increased the number of profiles colocalizing MOR and GAD. Similarly, following vagal deafferentation, MORs and GAD appear colocalized in nerve terminals throughout the DMV, including on assumed close synaptic contacts with identified gastric-projecting DMV neurons. Conversely, incubation of deafferented brainstem slices with group II mGluR agonists abolished this colocalization.

Although other receptors negatively coupled to adenylate cyclase and similarly located within the NTS circuits undoubtedly contribute to the modulation of subdiaphragmatic sensory afferent fibres and to the overall control of the cAMP–PKA pathway (Brooks et al. 1992; Rhim et al. 1993; Page & Blackshaw, 1999; Browning & Travagli, 2001; Blackshaw, 2001; Glatzer & Smith, 2004), none appears to have such a pivotal role as group II mGluRs. We hypothesize that other vagally released neurotransmitters that activate receptors negatively coupled with adenylate cyclase, such as GABA acting on GABA-B receptors, contribute to the ‘dampening’ of the GABAergic circuitry between NTS and DMV. Relieving this inhibition, for example by pretreating the specimen with saclofen, allows the levels of cAMP to rise and uncovers the modulation of the synapse. If this type of modulation was the principal means of controlling the NTS–DMV GABAergic circuits, then perfusion of slices from deafferented rats with a GABA-B agonist, for example, would occlude any effects of ME. However, our data with baclofen argue against the primary control of this synapse by GABA-B receptors, which appear to have a modulatory role only during concomitant activation of mGluRs.

Physiological significance

The NTS receives sensory information from the viscera and is involved in the regulation of homeostasis and visceral reflexes; while our data are configured for the study of gastrointestinal circuits, they may also provide a template for studies of homeostasis and integrative feeding-related circuits. The GABAergic NTS–DMV synapse is tonically activated in gastrointestinal brainstem circuits (Sivarao et al. 1998) and is probably the most relevant modulatory synapse in gastrointestinal reflexes (McCann & Rogers, 1993; Travagli et al. 2003). Previously, we assumed that several neurotransmitters/modulators were unable to modify this GABAergic synapse (Browning & Travagli, 2001, 2003; Browning et al. 2002) until we demonstrated that the ‘state of activation’, determined by the levels of cAMP in these inhibitory synapses, governed their regulation (Browning & Travagli, 2001; Browning et al. 2004).

Although a 20%–30% decrease in the amplitude of GABAergic currents to DMV may seem relatively small, one must keep in mind that DMV neurons are spontaneously active, both in vivo and in vitro (for review see Travagli et al. 2006). As their membrane potential is close to threshold for action potential firing, even a small shift of the membrane potential, such as that induced by a small reduction in GABA currents, has profound effects on the vagal output (for review see Travagli & Rogers, 2001).

The present study establishes that vagal afferent inputs control this state of activation, via ongoing activation of mGluRs present on, or adjacent to, GABAergic nerve terminals. Such regulation may provide a convenient, on-demand control of the principal NTS–DMV synapse within gastrointestinal vagal brainstem circuits. As these vagal afferent terminals already produce and release glutamate, this mechanism of regulation also provides an efficient modulation of vagal reflexes without the need for additional neurotransmitters.

While keeping in mind that afferent inputs to gastrointestinal vagal circuits appear to be organized differently from those devoted to cardiovascular function, these results raise fundamental questions about the structure of brainstem vagal circuits. Different components of cardiovascular brainstem vagal circuits exhibit disparate complements of mGluRs (Foley et al. 1998, 1999; Hay et al. 1999; Chen et al. 2002; Jin et al. 2004; Chen & Bonham, 2005; Mueller et al. 2005; Sekizawa & Bonham, 2005; Viard & Sapru, 2006). Both excitatory and inhibitory synaptic transmission onto second-order presumed baroreflex NTS neurons are affected by mGluRs; both group II and III presynaptic inhibitory mGluRs have been identified in vagal brainstem circuits, yet no evidence of postsynaptic receptors on NTS neurons themselves has been observed (Glaum & Miller, 1993a; Chen et al. 2002; Jin et al. 2004; Chen & Bonham, 2005). In the present study, we found pharmacological evidence of group II but not group III mGluRs on presumed presynaptic GABAergic nerve terminals in gastrointestinal circuits.

Moreover, this type of afferent-controlled modulation occurs only with GABAergic and not glutamatergic transmission, suggesting that underlying differences exist in the organization of excitatory and inhibitory central brainstem circuits. Differences may exist either in the expression of group II mGluRs on inhibitory but not excitatory nerve terminals, or in the pathways and connections of vagal afferent inputs themselves. The former would suggest a hitherto unknown differentiation in the pharmacology of NTS neurons; the latter would suggest basic differences in the circuitry of vagal afferent neuronal inputs. Unfortunately, immunohistochemical approaches to distinguish gastric from cardiovascular or respiratory NTS GABAergic nerve terminals are beyond current technical capabilities. Although it remains to be determined whether particular vagal afferent neurons are responsible for this type modulation, our chemical and surgical deafferentation data suggest that the majority of vagal afferent fibres involved in this phenomenon may be unmyelinated.

Based on our results, we are tempted to speculate that the sensory information from the gastrointestinal tract to the NTS–DMV circuits is regulated by the fine balance of glutamate-activated ionotropic or metabotropic receptors, and it is this balance that determines the ability of endogenous substances (e.g. opioid peptides) to modulate the GABAergic NTS–DMV synaptic connection.