Cannabinoid receptor 1 signalling dampens activity and mitochondrial transport in networks of enteric neurones


Pieter Vanden Berghe PhD, Center for Gastroenterological Research, KULeuven, Herestraat 49, Lab 701, B-3000 Leuven, Belgium.
Tel: +32 16 330153; fax: +32 16 345939; e-mail:


Abstract  Cannabinoid (CB) receptors are expressed in the enteric nervous system (ENS) and CB1 receptor activity slows down motility and delays gastric emptying. This receptor system has become an important target for GI-related drug development such as in obesity treatment. The aim of the study was to investigate how CB1 ligands and antagonists affect ongoing activity in enteric neurone networks, modulate synaptic vesicle cycling and influence mitochondrial transport in nerve processes. Primary cultures of guinea-pig myenteric neurones were loaded with different fluorescent markers: Fluo-4 to measure network activity, FM1-43 to image synaptic vesicles and Mitotracker green to label mitochondria. Synaptic vesicle cluster density was assessed by immunohistochemistry and expression of CB1 receptors was confirmed by RT-PCR. Spontaneous network activity, displayed by both excitatory and inhibitory neurones, was significantly increased by CB1 receptor antagonists (AM-251 and SR141716), abolished by CB1 activation (methanandamide, mAEA) and reduced by two different inhibitors (arachidonylamide serotonin, AA-5HT and URB597) of fatty acid amide hydrolase. Antagonists reduced the number of synaptic vesicles that were recycled during an electrical stimulus. CB1 agonists (mAEA and WIN55,212) reduced and antagonists enhanced the fraction of transported mitochondria in enteric nerve fibres. We found immunohistochemical evidence for an enhancement of synaptophysin-positive release sites with SR141716, while WIN55,212 caused a reduction. The opposite effects of agonists and antagonists suggest that enteric nerve signalling is under the permanent control of CB1 receptor activity. Using inhibitors of the endocannabinoid degrading enzyme, we were able to show there is endogenous production of a CB ligand in the ENS.


The intrinsic nerve reflexes that control gut functions such as secretion, absorption and motility arise from coordinated activity in the enteric nervous system (ENS). This nerve network operates largely independent of the brain and is organized in ganglionated-plexi embedded in the muscle layers of the gut wall. Apart from a whole range of classical neurotransmitters, the ENS also contains multiple receptor systems through which it becomes sensitive to circulating hormones (e.g. corticotropin releasing factor)1 or locally released neuromodulators (e.g. motilin, ghrelin, brain-derived neurotrophic factor).2

Another receptor system that is capable of altering ENS function is the cannabinoid (CB) receptor system. Medical application of Cannabis sativa extracts, with D9-tetrahydrocannabinol as the most important active compound,3 has been known for centuries and has been used in traditional medicine to treat gastrointestinal problems or at least to alleviate the symptoms caused by intestinal disorders. Since the discovery of two endocannabinoid receptors and the development of synthetic ligands, CB1 and CB2 receptor activity has been demonstrated in several tissues. Modes of action include inhibition of voltage-gated Ca2+ channels; negative coupling to adenylyl cyclase and opening of A-type and inward rectification of K+ channels.4 In the gastrointestinal tract, more precisely in the ENS (see review),5 CB1 receptors have been identified using different approaches, such as immunohistochemistry, microelectrode recordings,6 acetylcholine release studies7 and muscle contraction studies.8–10 In general, CBs slow intestinal transit11,12 with more pronounced effects in the colon when compared with the small intestine.13

Apart from modulating gut motility (see reviews),11,12 alleviation of visceral pain14 and anti-inflammatory actions,15 the CB receptor system has also been implicated in the control of feeding.16 Based on the observation that CBs significantly increase food intake, the CB receptor system is a potential target for drugs that decrease food intake to treat obesity. CB1 receptor antagonists indeed reduce food intake; AM-251 has been proven effective in animal models17 and rimonabant (SR141716) was shown to reduce weight, waist circumference and high-density lipoprotein cholesterol levels in humans.18 In view of the increasing development of food-intake-related drugs that target the endocannabinoid system, it is crucial to better understand how CB-related drugs act; especially as CB ligands are not devoid of adverse effects.19

The net effect of the CB signalling system depends on many factors including different endogenous ligands, degrading enzymes and constitutively active receptors. It remains elusive whether in the ENS, ligand-independent constitutive receptor activity or endogenous ligands are responsible for CB1 signalling. In this study, we investigated how CB1 receptor activation or inhibition influences enteric nerve network signalling, synaptic vesicle cycling and transport of organelles. We also sought evidence for an endogenous ligand by blocking its degrading enzyme.

Material and methods

Primary myenteric neurone cultures

Cultured myenteric neurones were prepared from adult guinea-pig ileum.20 Guinea-pigs of either sex (250–700 g) were killed by a sharp blow to the head and exsanguination (as approved by the Animal Ethics Committee of the University K.U.Leuven). Longitudinal muscle and myenteric plexus segments were digested in a protease/collagenase solution (30 min, 37 °C). The suspension was spun at 500 g, pellets resuspended and individual ganglia were plated onto glass coverslips. Cultures were kept at 37 °C (5% CO2) and medium (Medium 199 enriched with 10% foetal bovine serum, 50 ng mL−1 7-s nerve growth factor (NGF), 100 U mL−1 penicillin, 100 μg mL−1 streptomycin and 30 mmol L−1 glucose) was changed every 2 days. Experiments were performed between day 7 and 12. CB1 agonists/antagonists were added either to the culture medium (48 h prior to the experiment) or in the experimental HEPES-buffered perfusion solution (composition in mmol L−1: NaCl 150, KCl 5, MgCl2 1, CaCl2 2, Glucose 10, Hepes 10) during imaging.


Cultured myenteric neurones were fixed in 4% paraformaldehyde containing HEPES-buffered solution for 40 min (RT), washed and permeabilized in PBS with 0.5% Triton X-100 containing 4% donkey serum to block a-specific binding sites. Primary antibody solution (24 h at 4 °C) contained: rabbit antisynaptophysin (Dr Jahn, Göttingen, DE, USA) and mouse anti-PGP9.5 (Ultraclone, Cambridge, UK) or rabbit anti-choline acetyl transferase (ChAT; M. Schemann, DE, USA). After washing, secondary donkey antisera were applied: anti-mouse Alexa594 (1 : 1000), anti-rabbit Alexa594 (1 : 1000) and anti-rabbit AMCA (1 : 250; Jackson Immuno Research Labs, West Grove, PA, USA). To identify nitric oxide synthase (NOS)-expressing neurones, we performed NADPH diaphorase stainings (2 h, 37 °C, 0.5 mg mL−1β-NADPH, 0.1 mg mL−1 tetrazolium blue, 0.5% Triton X-100, 0.1 mol L−1 PO4) which have been shown to reliably label NOS-expressing neurones.21 Digital images of NADPH diaphorase stainings were inverted and falsely coloured in green to generate an overlay with the red immunostainings for ChAT.


Total RNA was extracted from 7 days old myenteric neurone cultures and from isolated guinea-pig ileum (positive control) using TRIzol reagent and reverse transcribed to cDNA with Superscript II Reverse Transcriptase. The obtained cDNA served as a template for the PCR reaction, consisting of 40 cycles of amplification (94 °C for 45 s, 56 °C for 1 min, 72 °C for 1 min) with a final elongation of 10 min at 72 °C using 0.5 U of Taq DNA polymerase and 0.4 μmol L−1 of forward (TGATTCAGCGTGGAAATCAG) and reverse (TGACCGTGCTCTTAATGCAG) primers. The PCR product was submitted to agarose gel electrophoresis and visualized with ethidium bromide. The primers were designed, according to the guinea-pig (Gene Bank no. DQ355990) CB1 mRNA sequence, to amplify a 441-bp fragment.

Microscopic imaging

Imaging set-up  Different fluorescent techniques were used to monitor nerve and synaptic activities22 and all fluorescent images were recorded using an inverted Zeiss Axiovert 200 M microscope (Carl Zeiss, Oberkochen, Germany), with TILL Poly V light source (TILL Photonics, Gräfelfing, Germany) and cooled CCD camera (PCO Sensicam-QE, Kelheim, Germany) using Image Pro (Media Cybernetics, Silver spring, MD, USA) for fixed samples and TillVisION (TILL Photonics) for live imaging. All image analysis was performed with custom-written routines in Igor Pro (Wavemetrics, Lake Oswego, OR, USA).

[Ca2+] i– Fluo-4 imaging  Cultured myenteric neurones were loaded with 10 μmol L−1 Fluo-4AM (45 min) and transferred to a coverglass chamber mounted on the microscope stage. Changes in intracellular Ca2+ concentration ([Ca2+]i) are reflected in Fluo-4 fluorescence intensity and recorded at 525/50 nm. Neurones were identified by depolarization (75 mmol L−1 K+, 5 s)20 5 min prior to the experiment. Ca2+ signals were normalized and Ca2+ spikes (if signal rose above: baseline + 5 times the intrinsic noise) counted. Neurones were judged ‘active’ when they displayed at least one Ca2+ spike during 100 s of recording (at 37 °C). Activity over Time (AoT) images were generated by an automated routine that attributes, on a pixel by pixel basis, the maximum value of that pixel in a certain time window. A value was only assigned if a maximum significantly different from the starting value was observed; otherwise the pixel in the AoT image was set to 0, which filtered out all unchanging fluorescence in a given time window (see Fig. 1D). The proportion of active neurones and average spike frequency were compared between control and drug conditions. Because of the duration and variability of the recordings, single dose drug concentrations were chosen based on their maximal effects described in the literature. Although the receptor antagonists (SR 141716, AM251) have been reported to be specific and selective (, it cannot be excluded completely as at high concentrations other low affinity receptor systems may be influenced. The agonists used in this study can also activate CB2 which may underlie some of the differences observed (see Discussion). A local perfusion system (1 mL min−1) allowed us to administer agonist/antagonist solutions for a set period of time.

Figure 1.

 Spontaneous Ca2+ signalling in guinea-pig myenteric neuron cultures. (A) Typical traces of spontaneous [Ca2+]i spike activity in control conditions (blue) and in the presence of tetrodotoxin (TTX, 1 μmol L−1, red) or hexamethonium (Hex, 100 μmol L−1, green). (B) Mean (± SEM) results of the [Ca2+]i spike frequencies in the presence of TTX (red) and hexamethonium (green). (C) Proportion of neurones (± SEM) displaying spontaneous activity during a 100 s recording in TTX (red) and hexamethoinum (green). Neurones were judged active if they displayed at least one [Ca2+]i spike during the recording. The total number of neurones in the field of view was determined by a brief (5 s) K+ depolarization. Significant differences in B and C are marked with an asterisk (*P ≤ 0.05, Student’s t-test) and individual data points are shown in grey. (D) AoT image of a patch of cultured myenteric neurones loaded with Fluo-4 and depolarized with high K+. Bar: 50 μm (E) Following the activity measurements, neurones were identified by high K+ depolarization (as in D), and assigned to be cholinergic (arrows) or nitrergic (arrowheads) based on a their immunohistochemical staining for choline-acetyl transferase (ChAT; in red) or NADPH staining (in green). (F) Proportion of nitrergic (NOS) and cholinergic (ChAT) neurones (± SEM) that took part in the spontaneous network activity. (G) mRNA from primary myenteric neurone cultures and guinea-pig ileum was reverse transcribed and PCR amplified with guinea-pig specific primers. An amplicon of 441-bp was obtained from culture and ileum. No amplification product was detected when the PCR reaction was run without addition of the synthesized cDNA obtained from the RT reaction (−).

FM1-43 imaging  FM1-43 imaging was performed as previously described.22,23 Coverslips containing cultured myenteric neurones were transferred to a recording chamber with two parallel Pt/Ir wires (∼ 10 mm spacing) for electrical field stimulation (40 mA, 1 ms pulses, WPI A385 stimulator; World Precision Instruments, Hertfordshire, UK). Active synapses were labelled in the presence of 10 μmol L−1 FM1-43 by three rounds of 300 stimuli 20 Hz. Destaining, under constant perfusion with HEPES solution, was accomplished by 40 + 400 stimuli and a consecutive exposure to high K+ (Fig. 3A,B). Individual release sites were detected semi-automatically by generating two images: one in which only those pixels appeared that had a net decrease in fluorescence and the other in which the minimum of the first derivative per pixel was coded. These two images were multiplied and after applying a threshold, individual particles were detected and their centre coordinates extracted. ROIs were automatically drawn around these centres and average fluorescence (for a 3.2 by 3.2 μm area) of active release sites (referred to as boutons) was calculated, background subtracted and normalized to the starting value, allowing calculation of fractional release. Release sites that were clustered and could not be separated from each other were excluded from the analysis. All analysis procedures were programmed in Igor Pro (Wavemetrics, OR, USA).

Figure 3.

 Synaptic vesicle recycling in enteric neurones is altered by CB1 receptor activity. (A) Consecutive images of enteric boutons labelled with FM1-43, taken before (1), after 40 (2) and 400 (3) stimuli and at the end of the recording when boutons are completely destained using K+ (4). Bars: 20 μm. (B) Representative destaining trace of an enteric bouton (arrow and circle in 3A). Numbered circles in the tracing correspond to the images in (A). (C) Total fluorescence (in arbitrary units) lost during the destaining was quantified and compared in for the different drug conditions. Error bars are smaller than symbols size. (D) The fractions of dye that were released by electrical stimuli in different conditions. Significant differences in (C) and (D) are marked with an asterisk (*P < 0.05, anova, Bonferroni post hoc test) and individual data points are shown in grey.

Mitochondrial imaging  Cultured neurones were loaded with 50 nmol L−1 Mitotracker green (30 min), washed and left to equilibrate for 30 min. Neurones with clear fibre tracts were selected under differential interference optics (DIC) and fluorescence image sequences (200 images, 1 Hz) were recorded from which spatiotemporal maps were generated.24 Proportions of moving and stationary (vertical lines) mitochondria and transport velocity were extracted from the maps.

Data presentation and statistics

Unless stated otherwise, all data are presented as mean ± SEM. In the graphs, individual data points were displayed as small grey symbols and means as larger colour-coded symbols. SEM markers were often smaller than the symbols and therefore hidden. Fluo-4 data were compared with a paired Student’s t-test. Transport and FM1-43 data were compared with anova and Bonferroni post hoc tests. The n-numbers refer to the number of experiments (1 per coverslip) unless mentioned otherwise. Differences were considered to be significant if P < 0.05. Statistical analysis was performed with Microsoft Excel (Microsoft, Redmond, WA, USA), GraphPad (GraphPad Software, San Diego, CA, USA) or SAS Enterprise (SAS Inc., Cary, NC, USA).

Drugs, chemicals and culture media

SR141716a was provided by Sanofi-Aventis (Chilly Mazarin, France). WIN55,212, metanandamide (mAEA), AM-251, URB597, hexamethonium, tetrodotoxin (TTX), collagenase, protease, β-NADPH and tetrazolium blue were purchased from Sigma (Bornem, Belgium) and N-[2-(5-hydroxy-1H-indol-3-yl)]ethyl-5Z,8Z,11Z,14Z-eicosatetraenamide (AA-5HT) from Alexis Biochemicals (San Diego, CA, USA). All fluorescent probes, products for RT-PCR and cell culture-related products were from Invitrogen (Merelbeke, Belgium) except NGF which was purchased from Alomone Labs (Jerusalem, Israel).


Spontaneous Ca2+ signalling in enteric nerve networks

On average, 50% of the neurones were spontaneously active with an average frequency of about 4 Ca2+ spikes per 100 s (based on all control neurones: 686 neurones from 51 coverslips). The vast part of the spontaneous spiking depended on network activity and required nerve conduction as TTX (1 μmol L−1) reduced the Ca2+ spike frequency (1.36 ± 0.3 vs 4.15 ± 0.4 spikes per 100 s, P = 0.0004, n = 6) and the proportion of active neurones (18.1 ± 5.7%vs 60.6 ± 5.7%, P = 0.01) (Fig. 1A–C). The use of the nicotinic receptor blocker hexamethonium (100 μmol L−1) also lowered the spiking frequency (1.33 ± 0.3 vs 2.80 ± 0.3 spikes per 100 s, P = 0.009, n = 6) and the proportion of spontaneously active neurones (19.6 ± 3.6%vs 42.4 ± 6.8%, P = 0.01) (Fig. 1A–C). Combined with the TTX data, this indicates that the majority of Ca2+ spikes originate from reverberating activity in the network. Given the mutually exclusive nitrergic and cholinergic neuronal subgroups in the guinea-pig ENS, we determined the share of each subgroup in the spontaneous network activity by performing anti-ChAT immunohistochemistry and NADPH staining following Fluo-4 recordings (Fig. 1D–E). We found that 23.1 ± 5.5% of nitrergic and 35.6 ± 4.7% of cholinergic neurones present in the culture participated in the spontaneous network spiking (Fig. 1F). The general fraction of nitrergic and cholinergic neurones in these cultures is 13.9 ± 2.7% and 87.1 ± 3.0% respectively. We also examined the presence of CB1 mRNA in guinea-pig myenteric neurone cultures using RT-PCR. A PCR product of the predicted size of 441 bp was amplified from RNA extracted from neuronal cultures (three guinea-pigs). The positive control, with cDNA derived from guinea-pig ileum, was also found at an expected band size of 441 bp. No PCR product was amplified when the PCR reaction was run without the addition of synthesized cDNA obtained from the RT reaction (Fig. 1G).

Effect of CB1 activity on spontaneous Ca2+ signalling in cultured myenteric neurones

As organization and density of neuronal patches varied from culture to culture, we recorded consecutively from one group of neurones in control and in CB1 drug solutions (Fig. 2A,B). The CB1 blocker AM-251 (10 μmol L−1) significantly increased not only the Ca2+ spike activity (10.2 ± 2.9 vs 3.2 ± 1.0 spikes per 100 s, P = 0.03, n = 5) but also the proportion of active neurones (77 ± 6%vs 46 ± 9%, P = 0.01). Application of the CB1 receptor antagonist SR141716a (1 μmol L−1) also amplified the spike frequency (5.4 ± 1.1%vs 3.0 ± 0.6 spikes per 100 s, P = 0.003, n = 9) in the network, but did not recruit more neurones to be active (55 ± 4%vs 43.9%, P = 0.12) (Fig. 2C). Conversely, the CB1 agonist mAEA (1 μmol L−1) significantly turned down Ca2+ spike frequency (1.7 ± 0.4 vs 4.6 ± 1.3 spikes per 100 s, P = 0.04, n = 7) and reduced the proportion of active neurones (19 ± 5%vs 41 ± 10%, P = 0.02). However, the cannabimetic drug WIN55,212 (10 μmol L−1) did not alter frequency (3.1 ± 1.1 vs 3.8 ± 0.8, P = 0.57, n = 6) or the proportion of active neurones (39 ± 8%vs 56 ± 11%, P = 0.06) (Fig. 2D). CB1 agonists or antagonists did not induce any direct changes in basal cytosolic Ca2+ concentration of the cultured neurones.

Figure 2.

 The frequency of Ca2+ signalling events in cultured networks is controlled by CB1 receptor activity. (A) Fluo-4 images of enteric neurone cultures at rest (Fluo-4) and as identified by activation with 75 mol L−1 K+. The two color images are coded for [Ca2+]i spike frequency (colorscale from 1 to 15 spikes: blue = low, green to red = high frequency) in the presence of control solution and AM-251 (10 μmol L−1). Arrows indicate neurones that were silent in control but active under AM-251. Bars: 20 μm. (B) Typical traces of spontaneous [Ca2+]i spike activity in control conditions (blue) and in the presence of CB1 antagonists (AM-251, 10 μmol L−1, red) or CB1 agonists (mAEA, 1 μmol L−1, green). (C) Mean (± SEM) results of the [Ca2+]i spike frequencies in the presence of antagonists (SR141716, 1 μmol L−1, red) and agonists (WIN55,212, 10 μmol L−1, green). (D) Proportion of neurones (± SEM) displaying spontaneous activity during a 100 s recording in antagonist (red) and agonist conditions (green). Significant differences in C and D are marked with an asterisk (*P < 0.05, Student’s t-test) and individual data points are shown in grey.

Synaptic vesicle recycling

We used FM1-43 imaging to investigate synaptic vesicle recycling under CB1 activity or blockade (Fig. 3A,B). Neurone cultures were incubated for 48 h in control or CB1-related drugs prior to the experiment in which total dye uptake and destaining fractions were determined. The total amount of vesicles used at a release site during loading was enhanced after incubation with mAEA (in arbitrary units: 13.9 ± 0.5*, 776 boutons vs 11.2 ± 0.2, 302 boutons), unaltered by WIN55,212 (1 μmol L−1, 55 boutons) and reduced by the antagonists AM-251 and SR141716a (8.2 ± 0.4*, 86 boutons and 7.6 ± 0.4*, 58 boutons; anovaP < 0.0001, *Bonferroni post hoc test, P < 0.001) (Fig. 3C). No differences could be detected in the fractions released by 40 stimuli, a stimulus that turns over the readily releasable pool of vesicles. However, a significantly smaller fraction was released by electrical stimuli (40 + 400) in the antagonist conditions vs the agonist condition. In other words, the fraction of vesicles that lags behind and can only be set free using high K+ stimulation is larger in the antagonist conditions (Fig. 3D).

Mitochondrial transport

Besides Ca2+ imaging, we used mitochondrial transport imaging as another optical tool to assess nerve cell activity. Transport and relocation of mitochondria is especially important in neurones as these highly polarized cells need to provide energy to their extremities. Two parameters were used to characterize the transport: on the one hand, the number of mitochondria that were transported in the fibres and, on the other hand, the velocity at which they did so (Fig. 4A–E). In control conditions, about 21.5 ± 0.8% of mitochondria were transported during the course of our recordings (n = 18). Incubation (48 h) of the neurones in the presence of either mAEA (10 μmol L−1) or WIN55,212 (10 μmol L−1) significantly decreased the number of transported mitochondria (12.5 ± 1.3%, n = 10 and 7.5 ± 0.7%, n = 4 for mAEA and WIN55,212 respectively). Conversely, incubation with the CB1 blockers AM251 and SR141716a (10 μmol L−1) had a stimulating effect, with 30.6 ± 0.5% (n = 15) and 31 ± 1.3% (n = 6) of mitochondria being transported respectively (Fig. 4F). The velocity of the transport did not differ between control and agonist/antagonist conditions (Fig. 4G).

Figure 4.

 Mitochondrial transport in processes of enteric neurones is enhanced by CB1 receptor inhibition. (A) DIC image of myenteric nerve fibres in culture. (B–D) Mitotracker green labelling of mitochondria in myenteric neurone cultures. Consecutive images (time indicated in seconds) of the mitochondria during transport (arrowheads) within the fibre of interest (green, C). Bars: 20 μm. Figure 4B is also available as a supplementary movie (Movie S1). (E) Spatio-temporal map of transport activity in the fibre of interest. Stationary mitochondria appear as vertical lines; transported mitochondria as tilted lines from which their velocity can be derived. (F) The proportion of mitochondria that were transported in either of the CB drug conditions. (G) The velocity of transported mitochondria in the different drug conditions. Significant differences in (F) and (G) are marked with an asterisk (*P < 0.05, anova, Bonferroni post hoc test) and individual data points are shown in grey.

Number of vesicle release sites

To test whether CB1 receptor activity had a long-term influence on the synaptic organization of the enteric network, immunostainings for synaptophysin were performed in cultures treated with CB1 drugs (48 h). We counted all synaptophysin-positive spots per length of fibre, as a measure for the number of vesicle release sites in the cultures. Although AM-251 and mAEA were not able to induce significant changes, WIN55,212 reduced the number of release sites while the antagonist SR141716a caused a significant increase (WIN55,212: 13 ± 1.4, n = 3; mAEA: 18.4 ± 0.88, n = 4; Control: 17.5 ± 0.88, n = 8; AM251: 17.6 ± 0.89, n = 4; SR141716a: 19.6 ± 2.5, n = 4; anova, P = 0.001) (Fig. 5).

Figure 5.

 Number of neurotransmitter release sites can be altered by CB1 receptor activity. (A) PGP9.5 immunostaining of enteric fibres in culture. (B) Synaptophysin-immunostaining, used as a marker for release sites in the enteric neurone cultures. Bar: 20 μm. (C) The number of release sites in the different CB1 drug conditions, expressed as a number per 100 μm of fibre. Significant differences in C are marked with an asterisk (*P < 0.05, anova, Bonferroni post hoc test) and individual data points are shown in grey.

Endogenous ligand or constitutive CB1 activity?

It is still elusive whether the effect of CB1 activity is due to either a constitutive receptor activity or an endogenous ligand activating the receptor upon demand. We addressed this conundrum by using two inhibitors (AA-5HT and URB597, both 100 nmol L−1) of the fatty acid amide hydrolase (FAAH) enzyme that catalyses the degradation of endogenous CBs. If endocannabinoids are produced in the cultures, a FAAH block would result in reinforcement or prolongation of CB1 receptor stimulation akin to the addition of a receptor agonist. We found that in the presence of AA-5HT the Ca2+ spike activity present in the cultured network is significantly reduced (3.0 ± 0.6 vs 5.6 ± 1.2, P = 0.01, n = 9) although this did not result in a smaller proportion of active neurones (63 ± 3%vs 58 ± 7%, P = 0.4). Addition of URB597 similarly resulted in lower spontaneous Ca2+ spike frequency (2.9 ± 0.4 vs 5.2 ± 0.8, P = 0.03, n = 12) without changing the proportion of active neurones (30 ± 4%vs 42 ± 6%, P = 0.16) (Fig. 6).

Figure 6.

 Endogenous ligands are present in the (cultured) enteric nervous system. (A) Representative traces of spontaneous Ca2+ spike activity in control conditions and in the presence of FAAH blockers such as AA-5HT and URB597 (both 100 nmol L−1). (B) Mean (± SEM) results of the [Ca2+]i spike frequencies in control and in cultures exposed to AA-5HT or URB597 (both 100 nmol L−1). (C) Proportion of neurones (± SEM) displaying spontaneous activity during a 100-s recording in control, AA-5HT and URB597 (both 100 nmol L−1) conditions. Significant differences in (B) and (C) are marked with an asterisk (*P < 0.05, Student’s t-test) and individual data points are shown in grey.


In this study, we investigated the role of CB1 receptors in the control of ENS activity. We confirmed, by RT-PCR, that CB1 receptors were expressed in these cultures and found that baseline activity can be modulated in either direction using CB1 receptor agonists and antagonists. Other nerve activity, such as mitochondrial transport, was altered similarly and also synaptic vesicle turnover was influenced by CB1 modulation. The fact that activity and transport increase with CB1 antagonists suggests tonic inhibition, which is mediated by CB1 receptor activity. However, increased activity under antagonists does not provide a conclusive evidence for endogenous endocannabinoid tone as some antagonists, at least in vitro and at high concentrations (e.g. SR141716a), also have inverse agonist properties and may silence constitutive CB1 receptor activity. We found evidence for an endogenously produced ligand in the ENS by using two different inhibitors for FAAH, the endocannabinoid degrading enzyme, which is not interfering with the constitutive CB1 activity.

The CB system has been shown to play an important role in energy homeostasis25 and in several gastrointestinal disorders.12 Obesity, a multifactorial disorder, has become a prominent life-threatening disease with high co-morbidity. Besides invasive surgical techniques, pharmacological approaches are receiving increasing attention, as interference with orexigenic pathways proves useful to treat obesity. The endocannabinoid system has been implicated in the control of food intake and CB1 receptor antagonists (rimonabant/SR141716a) have been shown promising to reduce appetite in obese patients.18 Although the importance of endocannabinoid signalling is clear from several studies, it remains incompletely understood how exactly and to what extent it controls ENS function. Different CB signalling constituents were shown to be present in the gastrointestinal tract. Expression of CB1 receptors was shown in cholinergic but not NOS-expressing neurones of the mouse colon.10,13 Using specific primers for the guinea-pig CB1 sequence, we found CB1 PCR product in tissue and in the primary nerve cultures derived from guinea-pig ileum. CB1 receptors are also expressed in submucous neurones and the degrading enzyme FAAH was shown to be expressed in rat myenteric ganglia.5 Recently, it was reported that CB2 receptors are expressed not only by immune cells but also in rat enteric nerves.26 Based on functional in vitro experiments, CB1 receptor activity was found to inhibit acetylcholine release and as a consequence reduce smooth muscle contractility and fast excitatory postsynaptic potentials.7,27,28 More recent reports show that CB1 receptors regulate both excitatory and inhibitory neurotransmission and that their effects on motility result from reducing excitation and activating inhibitory pathways.10,29In vivo, the net effect of CB1 receptor activity is to slow down intestinal motility (see review).11 In this study, we show that the network activity does not discriminate between excitatory or inhibitory nerves. As the majority of nerves and vesicular release in the ENS is excitatory in nature, we predict that the net effect of higher spontaneous activity and higher number of release sites will be stimulatory as well.

The endocannabinoid system is tonically active

Using enteric neurone cultures, we show that CB1 activity negatively regulates ENS activity, reducing spontaneous Ca2+ spike frequency and mitochondrial transport. We were able to both increase activity by using receptor antagonists (SR141716a and AM-251) and induce inactivity by adding a CB1 agonist (mAEA). This clearly shows that enteric nerve activity is under subtle but tonic control of an endogenous CB system. WIN55,212 acts somewhat differently, but this may be due to the differences in specificity or desensitization. Although toxicity of some agonists has been reported, this could not explain why WIN55,212 even used at a 10-fold higher concentration had less effect than mAEA. WIN55,212 is not a selective CB1 agonist but a general cannabimetic, which may also activate CB2 receptors, whether this effect counterbalances the CB1 effect of WIN55,212 remains to be determined. The increase in spontaneous activity under CB1 blockade reflects the release of a tonic brake on Ca2+ spike activity. As referred to earlier, this may result from constitutive receptor activity or from endogenous ligands.

Increased activity in enteric neurone cultures, induced by brain-derived neurotrophic factor, has been associated with an increase in number of release sites.2 Similarly the antagonist SR141716 caused an increase while WIN55,212 reduced the synaptophysin-positive spots, suggesting that enteric network activity causes new release sites to be established. This is contrary to Kim et al.30 who describe an inhibition of synaptic pruning by WIN55,212 in the hippocampus.

Endocannabinoid receptors reduce neurotransmitter release by inhibiting influx of Ca2+ ions in the presynaptic terminal.31 Apart from exocytosis, the retrieval of vesicles or endocytosis is also under tight control of Ca2+ ions. Here, using FM1-43, we investigate what effect CB1 activity has on synaptic vesicle cycling. The total number of used vesicles, reflected in the amount of FM1-43 dye accumulated during the staining, is reduced by antagonists. This may be counterintuitive at first, but can be explained if either (i) vesicle recycling became more efficient or (ii) there is compensation. In the case of increased efficiency, the same vesicles would be used over and over again, which would limit dye uptake during staining. In the second alternative case, CB1 receptor expression or signal transduction could be altered to compensate for reduced CB1 activity during antagonist incubation, which would result in enhanced CB1 activity upon antagonist washout. To test the likelihood of either hypothesis, we normalized the fluorescent signals and determined the fractions released by electrical stimuli in different conditions. In the case of higher efficiency, larger fractions would be set free as all dye is present in a small amount of vesicles. However, the opposite proved true, which argues in favour of compensatory CB1 activity. This is an important issue as it shows that prolonged exposure to antagonists may have important rebound effects. Whether this is due to enhanced expression or improved transduction will require rigorous testing. Tolerance or rebound phenomena have also been described for prolonged exposure to agonists. Basilico et al.32 show that after 5 h incubation with WIN55,212, the electrically evoked responses are no longer reduced and in human preparations, SR141716 enhancement of twitch responses was even stronger after WIN55,212 treatment.33

We also monitored mitochondrial transport in nerve processes as a measure of intracellular nerve activity. To the best of our knowledge, this is the first report that shows an influence of CB1 receptor activity on mitochondrial transport. Whether this is directly due to receptor signalling or rather a consequence of network activity will need to be determined. In an earlier report, we have shown that high K+ depolarization24 shuts down all transport in enteric neurones, which argues against a direct link between activity or at least depolarization and mitochondrial transport. However, it may well be that increased activity due to CB1 receptor blockade falls within the range for it to enhance mitochondrial transport, whereas the unphysiological K+ stimulus first activates and then stalls neurones in a paralysed state of inactivity.

Endogenous ligands

CB1 receptor effects are complex and apparent conflicting results have been reported; most probably owing to different aspects of CB1 signalling including constitutive activity, development of tolerance and interaction with other biochemical pathways. Although the presence of anandamide and 2-arachidonoylglycerol has been measured by HPLC in extracts from the intestinal wall,13 it is still under debate whether these endogenous ligands are necessary to activate CB1, and if so, where exactly they originate from. The fact that low levels of CB1 blockers are sufficient to cause intestinal propulsion11,12 supports an important role for constitutive CB1 receptor activity. In contrast, the anandamide reuptake inhibitor, AM-404, potentiates anandamide’s inhibitory effect on ACh release, which argues for the existence of a ligand.34 We used another pharmacological approach to investigate whether an endogenous ligand is involved. The breakdown of the endogenous ligand anandamide is catalysed by FAAH, which was revealed in the intestine by HPLC13 and immunohistochemistry.5 Here, we show that blockade of FAAH by AA-5HT or URB597 shuts down spontaneous Ca2+ spike activity, similar to exogenous CB1 ligands. Although the source and nature of the ligand is not known, these experiments prove that it is definitely endogenous to the ENS as its effect is even present in an isolated, cultured setting. As FAAH catalyses the degradation of anandamide but not 2-AG,35 it is most likely that indeed anandamide is the ligand for the CB1 receptors in the ENS.

In conclusion, this paper demonstrates an important role for CB1 receptors in regulating the activity state of the ENS. Activation of CB1 receptors dampens spontaneous network activity in both inhibitory and excitatory neurones, while receptor inhibition releases a brake for nerve signalling. Transport of mitochondria, necessary for energy supply at distant loci (eg. synaptic contacts), is suppressed and enhanced by agonists and antagonists, respectively. The opposite effects of agonists and antagonists suggest continuous control and the FAAH inhibitors indicate that this is, at least in part, due to an endogenous tone of endocannabinoids produced by enteric neurones or glia. Finally, changes in synaptic vesicle cycling suggest that tolerance develops presynaptically as fewer vesicles are turned over after a prolonged exposure to CB1 receptor blockers. We conclude that interference with this modulating CB1 system may be a good option to interact with the ENS and tune endogenous activity. Still, the long term effects such as development of tolerance and changes in synaptic density will need further testing in a dedicated set of experiments.


SR141716 was a gift from Sanofi-Aventis, France. This work is supported by grants from the FWO (Scientific Research Foundation, Flanders, Belgium) and BOF (KULeuven, Belgium).

Competing Interests

The authors have no competing interests.