Peripheral opiate action on afferent fibres supplying the rat intestine


David Grundy, PhD, Department of Biomedical Science, University of Sheffield, Western Bank, Sheffield S10 2TN, UK.
Tel.: +44 (0)1142 224657; e-mail:


The aim of the present study was to examine the sensitivity of mesenteric afferents supplying the rat small intestine to µ-opioid receptor ligands. Mesenteric afferent discharge was recorded electrophysiologically in response to [D-ALA2, N-Me-Phe4, Gly5-ol]-enkephalin (DAMGO; 100 µg kg−1 i.v.), before and after treatment with the µ-receptor antagonist alvimopan (1 mg kg−1 i.v.). DAMGO markedly stimulated whole nerve mesenteric afferent discharge (P < 0.05), an effect completely blocked by alvimopan. The response of mesenteric afferents to 2-methyl-5-hydroxytryptamine (30 µg kg−1 i.v.), bradykinin (0.1–1 µg kg−1 i.a.) and both low- and high-threshold distension (0–60 mmHg) was unaffected by alvimopan. In chronically vagotomized animals, the low-threshold response to distension was attenuated while the remaining high-threshold response was unaffected by alvimopan. In conclusion, mesenteric afferent fibres are markedly stimulated by µ-opioid receptor agonists, an effect blocked by alvimopan, which may contribute to the gastrointestinal reflex and behavioural responses to opiate treatment or abuse. However, alvimopan did not influence the normal sensitivity of intestinal afferents to chemical and mechanical stimuli that activate different subpopulations of vagal and spinal afferents. Thus, alvimopan may be useful for the treatment of gastrointestinal sequelae following opiate treatment for postoperative or chronic pain.


Opiate treatment for postoperative or chronic pain is frequently associated with adverse effects, including nausea and vomiting and debilitating bowel dysfunction.1 Opiates are generally considered to exert their antinociceptive effect at the level of the central nervous system (CNS). In this respect, intrathecal or epidural injection of morphine produces profound and prolonged analgesia by mimicking the action of endogenous opioids released from neurones in the dorsal horn.2 Opioid receptors are present at both pre- and postsynaptic membranes of spinal nociceptive neurones, where they exert their inhibitory action on pain transmission. However, opioid receptors are not confined to the CNS, but are found in abundance within the enteric nervous system, where they influence both secretory and motor function.3,4 Thus, the ability to dissociate the central from peripheral actions of opiates may have considerable therapeutic benefit.5,6

Recently, opioid receptors have been identified on the peripheral processes of sensory neurones.3 Colocalization studies have confirmed the presence of opioid receptors on VR1-positive visceral sensory fibres.7 Functional studies have demonstrated a powerful inhibitory influence of κ-receptor agonists on the activation of spinal colonic mechanoreceptive afferents.8–10 This inhibition includes the closure of voltage-sensitive calcium channels11 and the blocking of both tetrodotoxin (TTX)-resistant and TTX-sensitive sodium channels, although the latter may include a nonopiate action of some of the ‘κ-selective’ ligands.12 Vagal afferents may show a similar sensitivity to κ-receptor agonists.13 However, vagal afferents may also be a target for µ-receptor agonists as there is considerable literature demonstrating that the antinociceptive actions of low doses of morphine and DAMGO [d-Ala2, N-Me-Phe4, Gly5-ol]-enkephalin are attenuated by vagotomy,14,15 while electrical stimulation induces analgesia in several behavioural pain models.16 This activation of vagal afferent pathways leads in turn to activation of descending inhibitory pathways that suppress pain transmission through the dorsal horn. For this vagally mediated antinociceptive action to occur, opioid-receptor agonists must activate – rather than inhibit – vagal afferents. However, while no such activation was observed in a study of gastric vagal afferents13 a study of intestinal afferents has shown a profound stimulatory effect of µ-, but not κ-, receptor agonists on intestinal afferent firing, an action consistent in dose sensitivity and response duration to that observed in behavioural studies of vagally mediated antinociception.17

Peripherally restricted µ-receptor antagonists that may selectively prevent opioid-induced gastrointestinal effects without influencing analgesia have recently been developed.18 An action of such compounds on vagal pathways that are involved in reflex and behavioural control may contribute to their therapeutic mechanisms. The aim of the current study was therefore to examine the sensitivity of mesenteric afferents supplying the rat small intestine to µ-opioid receptor ligands, and to examine the role of endogenous opioids in regulating mesenteric afferent sensitivity to a range of stimuli designed to activate different subpopulations of vagal and spinal afferents.



Experiments were performed using male Sprague–Dawley rats (280–350 g), obtained from Charles River, Margate, UK. Rats were allowed, unless stated otherwise, free access to food and water.

General anaesthesia was produced with an intraperitoneal (i.p.) injection of pentobarbitone sodium (60 mg kg−1). In some animals, further ‘top-up’ injections (15 mg kg−1 i.p.) were required to obtain full surgical anaesthesia, indicated by the absence of hind leg withdrawal and corneal reflexes. This level of anaesthesia was maintained over the course of the experiment by infusing pentobarbitone sodium (0.5–1 mg kg −1 min−1) through the jugular vein. Animals were killed at the end of an experiment by an anaesthetic overdose.

Vagotomized animals

In some animals, abdominal vagotomy was performed to eliminate vagal afferent fibres from the mesenteric nerve bundles. A midline laparotomy was performed in isoflurane-anaesthetized rats. The dorsal and ventral vagal trunks were freed along the subdiaphragmatic oesophagus, ligated and sectioned. The abdominal wall and skin were stitched with a polyglycolic acid suture (Dexon; Davis & Geck, Gosport, UK). Following recovery, animals were returned to the housing unit and fed on a liquid diet in water (Complan Original; Heinz Foods, Uxbridge, UK), to which they had been accustomed for 3–5 days prior to surgery, and which was continued for a further 5–10 days before use. To confirm that the subdiaphragmatic vagotomies had been effective in removing the vagal innervation of the jejunum, the peripheral end of the right and left cervical vagus nerves were stimulated at the end of each experiment. In non-vagotomized animals, vagal stimulation caused a profound bradycardia and an increase in intestinal pressure. However, in vagotomized animals, although bradycardia was still observed, vagal stimulation had no effect on jejunal pressure, thus demonstrating the success of the procedure.

Surgical procedures

An incision was made in the neck, and the trachea was cannulated to maintain a patent airway. The right jugular vein was cannulated to enable anaesthetic infusion and the systemic administration of drugs. The left carotid artery was exposed, separated from the vagus nerve, and cannulated with a heparinized catheter (200 units mL−1 heparin in saline) to facilitate arterial pressure recording (Neurolog, NL108; Digitimer, Welwyn Garden City, UK).

Animals' body temperature was maintained at around 37°C by a rectal thermistor-controlled heated blanket. A midline laparotomy was performed and the caecum was excised to create space in the abdominal cavity. A 10-cm length of proximal jejunum was identified (typically 1–5 cm distal of the ligament of Trietz). This was ligated, and incisions were made on the antimesenteric border at each end. The loop was flushed through with saline. A 3.5-cm balloon (diameter 0.8 cm) was passed into the jejunum and secured in place at the aboral end of the 10-cm loop. The balloon was then connected to a barostat (Distender Series IIR, G & J Electronics, Ontario, Canada) to perform distensions of the jejunum. The small abdominal wall incisions were sutured and the muscle and skin of the large abdominal incision were sewn to a steel ring and the resulting well was filled with prewarmed (37°C) liquid paraffin (BDH, Poole, UK).

Nerve preparation and afferent recording

There are typically two mesenteric nerve bundles, which run alongside the two blood vessels located in the mesenteric arcades supplying the jejunum. The connective tissue spanning each arcade was dissected to leave a piece of connective tissue attached to the right side of the arcade. The arcade was then placed on a black perspex platform. With the use of a dissection microscope and illumination from a fibre-optic light source, a single paravascular nerve bundle (approximately 1 cm in length) was exposed by dissection of the overlaying fat and connective tissue. The nerve bundle was severed distal from the jejunal serosa (in order to eliminate the recording of efferent nerve activity), and cleared of fat and connective tissue. It was then attached to one of a pair of platinum recording electrodes, with a sliver of connective tissue wrapped around the other to act as a reference. The electrodes were connected to a single channel 1902 amplifier [Cambridge Electronic Design (CED), Cambridge, UK], and the signal was differentially amplified and filtered. The afferent nerve signal, together with the blood-pressure recording and pressure and volume read-out from the barostat, were passed into a micro 1401 amplifier (CED) and viewed online and captured by a PC running ‘Spike 2’ software (version 4.13, CED).

Experimental protocols

After a 10–30-min stabilization period, the viability of the preparation was assessed with a bolus dose of 2-methyl-5-hydroxytryptamine (2-m-5-HT, 30 µg kg−1 i.v.) and a 50-mmHg (held for 5 s) distension of the jejunum. If there was no afferent response to these test stimuli then a different mesenteric nerve bundle was dissected and tested.

Ramp and phasic distensions were produced using a computer-driven barostat with Protocol PlusTM Deluxe software (G & J Electronics). During the ramp distensions, the balloon was inflated so that the pressure in the jejunum rose by 2 mmHg every 4 s (0–60 mmHg over 120 s). Phasic distensions were performed by rapidly inflating the balloon to a preset pressure and holding the pressure for 25 s, after which the balloon was rapidly deflated. Phasic distensions were performed at pressures of 10, 20, 30, 40 and 50 mmHg, separated by a 1-min recovery period, which preliminary studies had shown to generate reproducible afferent responses.

Analyses of data

Data are presented as the arithmetic mean of either the absolute afferent discharge or percentage of control (before vehicle or drug treatment) ± SEM from n-values (which refer to the number of animals per group).

Changes in afferent firing in response to administration of DAMGO, 2-m-5-HT or bradykinin were calculated by measuring the mean peak firing over 5 s and subtracting the mean baseline firing over 30 s immediately prior to drug administration.

Changes in afferent firing during the balloon ramp distensions were calculated by measuring the mean afferent firing rate over 5 s at intrajejunal pressures of 2, 4 and 6 mmHg, up to 60 mmHg, and subtracting baseline discharge (mean over 30 s prior to onset of distensions). Changes in afferent discharge during the phasic distensions were calculated by measuring the whole afferent discharge above baseline over the 25-s distension period. Significant differences between data points were determined by appropriate use of either the Student's paired t-test or a two-way analysis of variance (with Bonferroni corrections where applicable). A probability of P < 0.05 was considered statistically significant.


Drugs used were: 2-m-5-HT (maleate salt) from Research Biochemicals Inc., Poole, UK; pentobarbitone sodium (Sagatal) from Rhône Mérieux Ltd, Harlow, UK; DAMGO from Tocris Cookson Ltd, Bristol, UK; bradykinin from Sigma, Poole, UK; and alvimopan (ADL 8–2698) from Adolor Corporation, Exton, PA, USA. With the exception of alvimopan, each drug was dissolved and diluted in saline (0.9% w/v NaCl in distilled water) to the required concentration. Alvimopan was prepared in 2% dimethyl sulphoxide (DMSO) (Sigma), 30% Captisol® (CyDex, Inc., Overland Park, KS, USA) in sterile water.


Effects of DAMGO on afferent discharge

Intravenous administration of DAMGO (100 µg kg−1) elicited a biphasic activation in whole mesenteric afferent nerve discharge. The response consisted of a rapid, transient burst of firing (mean peak response was 82 ± 7 spikes/s (n = 9) above a baseline discharge of 28 ± 6 spikes/s), followed by a prolonged phase of elevated discharge that lasted several minutes (Fig. 1A).

Figure 1.

(A) Sequential rate histogram depicting the biphasic mesenteric afferent nerve response to the µ-receptor agonist DAMGO (100 µg kg−1 i.v.). A second administration of DAMGO produced a similar, but attenuated, first phase response. (B) Administration of the µ-receptor antagonist alvimopan (1 mg kg−1 i.v.) after the first DAMGO administration inhibited the second phase of the first DAMGO response, and completely blocked the second response. (C) Bar graph depicting the first and second control responses to DAMGO, the response to alvimopan following the first DAMGO response, and the second DAMGO administration (in the presence of alvimopan). Alvimopan significantly attenuated baseline firing (P < 0.05 compared with baseline pre-DAMGO administration). Furthermore, the second response to DAMGO is completely blocked by pretreatment with alvimopan.

In four of the nine studies, DAMGO was administered a second time (Fig. 1A); the second administration elicited a similar, but smaller, activation in afferent nerve discharge of 59 ± 5 spikes/s above baseline discharge (reduction of 35 ± 7%) (Fig. 1A).

In five experiments, alvimopan was administered 5 min prior to the second DAMGO administration (Fig. 1B). Alvimopan (1 mg kg−1 i.v.) reduced the level of afferent firing to below the initial baseline firing rate prior to DAMGO administration (initial baseline firing rate was 37 ± 11 spikes/s, and following alvimopan treatment in the presence of DAMGO, baseline discharge was 10 ± 3 spikes/s). However, alvimopan administered in the absence of pretreatment with DAMGO had no effect on baseline discharge.

Furthermore, the second administration of DAMGO failed to elicit a response that was highly significant (P < 0.01) compared with the second control response to DAMGO in the absence of alvimopan (Fig. 1C).

Effects of alvimopan on responses to 2-m-5-HT and bradykinin

The effects of alvimopan (10 mg kg−1 i.v.) were examined on the afferent responses to 2-m-5-HT (30 µg kg−1 i.v.) and bradykinin (0.1, 0.3 and 1 µg kg−1 i.a.). Briefly, the afferent response to 2-m-5-HT appeared as an increase in mean peak afferent discharge of 82 ± 7 spikes/s (Fig. 2A, B). Bradykinin evoked dose-dependent increases in afferent discharge of 35 ± 12, 54 ± 15 and 91 ± 21 spikes/s (0.1, 0.3 and 1 µg kg−1 i.a., respectively) above a baseline discharge of 28 ± 6 spikes/s (Fig. 2C).

Figure 2.

(A) Sequential rate histogram depicting a typical mesenteric afferent nerve response to 2-m-5-HT (30 µg kg−1 i.v.). (B) Graph comparing afferent response to 2-m-5-HT in controls and following alvimopan (10 mg kg−1 i.v.). (C) Sequential rate histograms depicting the mesenteric afferent nerve response to bradykinin (0.1, 0.3 and 1 µg kg−1 i.a.) in control and following treatment with alvimopan. These data show that alvimopan has no effect on the afferent responses to 2-m-5-HT or bradykinin.

Alvimopan had no effect on baseline afferent firing in the absence of treatment with DAMGO. In addition, alvimopan had no effect on the afferent responses to 2-m-5-HT (Fig. 2B) or bradykinin (Fig. 2C).

Effect of alvimopan on mechanosensitive afferent responses

Balloon ramp distension produced a biphasic activation in afferent nerve discharge (Fig. 3A). The first phase of the response appeared as a rapid increase in discharge, which reached a plateau at an intrajejunal pressure of approximately 20 mmHg (71 ± 10 spikes/s). This plateau was maintained until the intrajejunal pressures reached approximately 30 mmHg, at which point there was further increase in discharge up to 121 ± 15 spikes/s at 60 mmHg (Figs 3A and 4A).

Figure 3.

Top panels show sequential rate histograms depicting (A) the biphasic increase in afferent discharge evoked by balloon ramp distension of the jejunum and (B) the pressure-dependent increases in afferent discharge during phasic balloon distensions. The bottom panels depict the rises in balloon pressure during the distensions.

Figure 4.

Graphs depicting the effects of alvimopan on the afferent responses to ramp distension in (A) controls; and (B) vagotomized animals. Alvimopan has no effect on the afferent responses to ramp distension in either controls or vagotomized animals.

Phasic balloon distensions induced pressure-dependent increases in afferent discharge (Figs 3B and 5A) of 32 ± 1–103 ± 4 spikes/s (10–50 mmHg mean over 25 s), consisting of an initial dynamic component (first 1–5 s) and a maintained phase of afferent firing.

Figure 5.

Graphs depicting the effects of alvimopan on the afferent responses to phasic distension in (A) controls; and (B) vagotomized animals. Alvimopan has no effect on the afferent responses to phasic distension in either controls or vagotomized animals.

In vagotomized animals, baseline afferent firing rates were not significantly different compared with controls (28 ± 6 spikes/s in control animals, 23 ± 6 spikes/s in vagotomized animals). An attenuation of the early phase (6–25 mmHg, P < 0.05) of the afferent response to balloon ramp distensions was observed and the response had a linear profile (Fig. 4B), as previously observed.19

The response profile to phasic distensions in vagotomized animals was similar to the response observed in control animals, although the mean afferent firing rate over the whole 25-s distension period was lower in vagotomized animals compared with controls (Fig. 5B), largely because of a reduction in the maintained phase of the response.

The effects of alvimopan (10 mg kg−1 i.v.) were investigated on the responses to ramp and phasic distensions in both controls (Figs 4A and 5A) and vagotomized animals (Figs 4B and 5B). The afferent responses to both ramp and phasic distensions were unaffected by treatment with alvimopan 10 mg kg−1 (n = 4). Furthermore, no effect on baseline firing rate was observed.


Eastwood & Grundy17 provided the first electrophysiological data demonstrating an excitatory effect of opioid receptors on intestinal afferent sensitivity. They described a profound and long-lasting, dose-dependent stimulation of mesenteric afferent discharge following administration of µ- and δ-selective agonists. The dose range of DAMGO that produced a marked stimulation of mesenteric afferent discharge was similar to that which caused a vagally dependent increase in tail-flick latency in a behavioural test of analgesia, the hot-plate test.20 Moreover, the persistence of opioid action in these behavioural studies over a 20-min observation period was also similar to the maintained increase in afferent firing described by Eastwood & Grundy. Similarly, the κ-receptor agonist U-50488 neither stimulated mesenteric afferents nor influenced tail-flick latency, implying that these receptors do not give rise to vagally mediated antinociception.20 Previous studies have shown that κ-receptor agonists inhibit baseline discharge and responses to distension of gastric vagal mechanoreceptors.13,17

The current study has confirmed the mesenteric afferent sensitivity to DAMGO and furthermore shown that this effect is antagonized by a peripherally restricted µ-receptor antagonist, alvimopan. Indeed, alvimopan, administered during DAMGO-induced augmented afferent firing, immediately reduced the level of firing to below the pre-DAMGO baseline firing rate. Alvimopan is a recently developed opioid antagonist whose activity is restricted to receptors in the periphery, including the gastrointestinal tract.18 It has recently been shown to reverse opioid-induced bowel dysfunction without reducing analgesia or precipitating withdrawal symptoms in nonsurgical patients receiving opioid treatment for chronic pain.6 In addition, alvimopan is suggested to normalize bowel dysfunction in patients recovering from abdominal surgery with postoperative pain.5

This effect of alvimopan probably arises primarily from antagonism of opioid receptors in the ENS, which when activated modulate the release of both excitatory and inhibitory enteric neurotransmitters, leading to decreased propulsive motility and inhibition of neurogenic secretion of water and electrolytes.4 However, part of the effect of alvimopan may also arise from an additional action at the level of the extrinsic afferent nerve terminals. Data from the present study, together with that from an earlier investigation using the same experimental model,17 suggest that opioids may stimulate gut afferents and this may contribute to the reflex motor and secretory action of opioids.

The observation that µ receptors are expressed on sensory neurones indicated that opioids may have a direct effect on afferent excitability. An indirect action following release of mediators from enteroendocrine cells or mast cells cannot be ruled out at this stage. However, the action of opioids is traditionally considered to be one of inhibition following G-protein-mediated inhibition of adenylate cyclase, inhibition of voltage-dependent calcium channels or activation of an inwardly rectifying potassium channel.21 If these transduction pathways were to underlie the afferent activation seen in the present study, one would have to propose that it resulted from disinhibition following the blocking of release of some inhibitory mediator responsible for ongoing suppression of baseline afferent firing. Indeed, in this model, treatment with N-type calcium channel blockers gives rise to a small elevation in baseline discharge, indicating that there is some inhibitory influence on afferent discharge.22 However, the time-course of this increase is very different from the short latency for activation seen following opioid administration in the present study. Indeed, the latency for the DAMGO response is similar to the circulation delay, that is the time required for the agonist to reach the gut from the site of administration into the jugular vein. This latency is also similar to that seen following 5-hydroxytryptamine (5-HT) and cholecystokinin (CCK), which have been shown previously to exert a direct excitatory effect on afferent firing.23,24 It is therefore unlikely that such a rapid and powerful activation would result from disinhibition or an indirect effect of DAMGO, mediated through the release of other transmitters. The alternative to disinhibition therefore is that opioids can exert a direct excitatory effect on afferent firing. In this respect, there is growing literature showing that opioids can evoke a direct receptor-mediated excitation, both of central and peripheral neurones including primary cultures of dorsal root ganglion neurones.25,26 In these cells, stimulation occurs at lower concentrations of opioids, but inhibition becomes dominant as the concentration reaches micromolar levels. This observation leads to the proposal that there might be separate stimulatory and inhibitory versions of each opioid receptor subtype. However, recent data showing that a switch from stimulation to inhibition may depend on the concentration of glycolipid GM1 gangliosides indicates that the same receptor may mediate both effects depending on how the receptor is coupled to Gi or Gs.27 The different dose sensitivity for excitatory and inhibitiory actions of opioids may explain some of the anomalous observations relating to pain management and in particular the ability of low doses of antagonist to augment morphine's antinociceptive potency.28

The mesenteric nerve bundle is a mixed nerve containing both vagal and spinal primary afferents as well as intestinofugal fibres projecting to the prevertebral ganglia.29 However, using a range of stimuli that activate different subpopulations of vagal and spinal afferents it is apparent that endogenous µ-receptor ligands do not influence the ability of these endings to respond to chemical or mechanical stimuli. Thus, the response to 5-HT3 receptor stimulation with 2-m-5-HT, which is specific for some vagal mucosal afferents, is unaffected by alvimopan. Similarly bradykinin's action on spinal afferents is unchanged following alvimopan. Distension activates various populations of low threshold, high threshold and wide-dynamic range mechanosensitive afferents that convey information via vagal and spinal pathways, yet the response to distension was also unchanged by alvimopan. Moreover, chronic vagotomy, performed to eliminate vagal afferents from the mesenteric bundles, clearly attenuated the low threshold component of the response to distension, but the remaining response that can be attributed to spinal afferents was also unchanged.

Thus, alvimopan does not influence the sensitivity of gut afferents to either chemical mediators or mechanical stimulation. As 2-m-5-HT is specific for vagal afferents in this preparation,23 while bradykinin acts on spinal afferents,18 and because the distension protocol employed will act on both low- and high-threshold mechanoreceptors, this implies that endogenous opioids do not normally modulate the sensitivity of either vagal or spinal afferent pathways. However, as the sensitivity to DAMGO is markedly attenuated in animals that have undergone truncal vagotomy to eliminate the vagal afferent population from within the mesenteric bundle,17 this would suggest that vagal rather than spinal afferents are stimulated by exogenous opioids. As discussed above, there is a wealth of evidence that spinal afferents are inhibited by opioids, in particular κ opioids.

Vagal afferents contribute to the reflex repertoire of gut motor and secretory function. Vagal afferents also contribute to behavioural responses to food, including anorexia, nausea and vomiting.30 These effects of opioids on vagal afferent firing may contribute to opioid bowel dysfunction, although an action at the level of the nucleus of the solitary tract may also contribute, as pre- and postsynaptic opioid receptors operate at this level, similar to their action in the dorsal horn.31

It has been shown previously that DAMGO and CCK activate the same population of vagal afferent fibres.17 As CCK-sensitive afferents are also stimulated by certain nutrients in the lumen, especially lipids, and are involved in the regulation of feeding behaviour this suggests that feeding behaviour and antinociception may be linked. Intestinal nutrients release CCK to act on vagal afferent nerve terminals, and the reflex response to this includes satiety and gastric relaxation.32 The latter is part of a feedback mechanism that matches gastric emptying to the digestive and absorptive capacity of the intestine. Gastric relaxation is also part of the sequelae of postoperative ileus, which can be exacerbated by treatment with opioids to control postoperative pain.5 From the observations made in the present study, it is conceivable that vagovagal reflexes triggered by opioids may contribute to this dysmotility state and that alvimopan may act at this site to reverse ileus.

In conclusion, the effect of alvimopan may be mediated in part by antagonism of µ-opioid receptors on sensory nerve terminals, which are activated by opiate treatment for postoperative or chronic pain. An action at these sites would be expected to contribute to the reversal of opiate-induced dysmotility and nausea and vomiting, but may also interfere with vagally mediated antinociception. Given the centrally acting analgesic action of opioids, one would not expect interference with antinociception to compromise pain management. Moreover, as alvimopan has no effect on afferent chemical and mechanical sensitivity it may normalize gut function without affecting the normal sensory traffic generated by mechanical and chemical stimuli from within the bowel.