Acute Activation of Cannabinoid Receptors by Anandamide Reduces Gastrointestinal Motility and Improves Postprandial Glycemia in Mice

The endocannabinoid system (ECS) consists of cannabinoid receptors (CBRs), endocannabinoids, and enzymes involved in their synthesis and degradation. The two main receptors, CB1R and CB2R, are both G protein–coupled receptors differing structurally and by their tissue localization. Whereas CB2Rs are mainly expressed in the immune system, CB1Rs are abundantly present in the central nervous system and to a lower extent in peripheral tissues (1). Arachidonoyl ethanolamide (anandamide) and 2-arachidonoyl glycerol (2-AG) are the two best-characterized endogenous cannabinoids produced in the brain and peripheral tissues that bind CB1R and CB2R with no marked selectivity (2) to activate or inactivate metabolic pathways similarly to Δ9-THC (tetrahydro-6,6,9-trimethyl-3-pentyl-6H-dibenzopyran-1-ol), the major active constituent of marijuana (1,3).

It is now well established that the ECS is involved in many biological functions, including the regulation of energy balance and the pathogenesis of obesity. Obesity is associated with central ECS overactivation, and since activation of CB1R in the brain stimulates food intake, ECS appears to be located at the center of a vicious circle inducing weight gain and related disorders (4,5). However, even if the CB1R-dependent stimulation of food intake may be the main initial cause of various metabolic deregulations associated with obesity, several data collected from animal and human studies using the specific CB1R antagonist SR141716 (rimonabant) suggested that peripheral ECS might also directly regulate energy metabolism (6–8). It was recently reported that activation of peripheral CB1R signaling in key tissues relevant to insulin action is associated with alteration of glucose homeostasis and contributes to diet-induced insulin resistance in mice (9–11). Nevertheless, very little information is available concerning the impact of peripheral ECS activation on intestinal glucose absorption that is however the major determinant of how quickly glucose appears in the circulation during the fed state.

The presence of a functional ECS in the gastrointestinal tract has been pointed out in many species (12–14), and available evidence suggests that cannabinoids reduce gastric and intestinal motility (15,16), but the consequences on plasma glucose appearance were poorly studied. However, consistent data indicate the existence of a positive relationship between gastrointestinal motility and plasma glucose appearance in healthy and diabetic patients (17–19). We therefore investigated the impact of acute activation of CBR by anandamide on postprandial glycemia, recognized as an important component of the overall glycemic control (20,21). Our data showed that activation of gastrointestinal ECS improved postprandial glycemia and further indicated this was associated with neither an increase in glucose clearance nor a decrease in intestinal transport but rather with a reduction in gastrointestinal motility involving both CB1R and CB2R.

Official French regulations (no. 87848) for the use and care of laboratory animals were followed throughout the experimental period. The experimental protocol was approved by the local ethic committee for animal experimentation (BX0622). Eleven-week-old C57BL/6J male mice (JANVIER LABS, Le Genest Saint Isle, France) and global CB1R^−/− mice (gift from Dr. James Pickel, National Institute of Mental Health, Bethesda, MD) were housed individually on a 12/12-h light/dark schedule at 22–23°C with ad libitum access to water and food (Standard Diet AO4; UAR, Epinay-sur-Orge, France). A group of mice was subject to a high-fat diet (30% lard) manufactured by SSNIF (Soest, Germany). After 16 weeks of diet, animals with a weight gain less than +10 g compared with controls were excluded from the study. Diet-induced obesity (DIO) mice (39.1 ± 1.1 vs. 27.3 ± 0.9 g, DIO vs. control) were glucose intolerant and insulin resistant (Supplementary Fig. 2A and B). On the day of each experiment, food was removed from the cages for 6 h (from 8:00 a.m. to 2:00 p.m.).

Anandamide, AM630, AM6545, and exendin 9-39 were purchased from Sigma-Aldrich (Saint-Quentin-Fallavier, France), and SR141716 was provided by Sanofi (Paris, France). Anandamide and CBR antagonists were prepared in 4% DMSO/1% Tween 80 and administered intraperitoneally at 10 mg/kg. In control experiments, animals were injected with vehicle (4% DMSO/1% Tween 80).

For glucose tolerance test (GTT) and insulin tolerance test, mice respectively received a glucose load (2 g/kg) or insulin intraperitoneal injection (0.5 UI/kg, Actrapid; Novo Nordisk, La Défense, France). For GTT, a d-glucose solution (20% weight for volume) was either injected intraperitoneally (intraperitoneal GTT) or given orally (oral GTT [OGTT]). Oral d-xylose loading test (OXT) consisted in the administration of a d-xylose solution (10% weight for volume; 1 g/kg body weight). Drugs were administered intraperitoneally to 6-h fasting mice 10 min before glucose, xylose, or insulin. Each treatment was tested with a 1-week interval on the same series of mice in the same conditions. Glycemia was measured directly in blood sampled from the tail with a My Life Pura glucose meter (Ypsomed, Paris, France). In some OGTT experiments, blood (25 µL) from tail vein was collected in chilled tubes containing EGTA and DPPIV inhibitor (Millipore S.A.S., Molsheim, France) for determination of insulin, glucose-dependent insulinotropic peptide (GIP), and GLP-1 concentration with a Bio-Plex Mouse Diabetes Assay (Bio-Rad, Marnes-la-Coquette, France). In OXT, some control experiments included oral administration of phlorizin (Sigma-Aldrich) as a sodium-dependent glucose transporter 1 (SGLT-1) inhibitor, 15 min before the first injection. Plasma xylose concentrations were measured with a spectrophotometric kit (Megazyme, Wicklow, Ireland).

Wistar rats (JANVIER LABS) weighing 250–300 g were anesthetized by intraperitoneal injection of 50 mg/kg sodium pentobarbital. Two 5-cm proximal jejunum segments (5 cm from the ligament of Treitz) were taken from each animal and prepared as previously described (22). In brief, sleeves were flushed, everted, and mounted in a 15-mL centrifuge tube. The installation allowed the introduction of buffer into the sac (serosal compartment) and the collection of samples periodically. Sacs were filled with 200 µL buffer consisting of 150 mmol/L NaCl, 1 mmol/L KH₂PO₄, 4 mmol/L K₂HPO₄, and 5 mmol/L fructose, pH 7.4, and immersed in 14 mL of the same buffer (mucosal compartment) added with 0.1 μmol/L or 30 mmol/L d-glucose. [¹⁴C]d-glucose (3.2 µCi/µmole; PerkinElmer, Courtaboeuf, France) was used as a marker to quantify transporter-mediated uptake from the mucosal to the serosal compartment. Fragments were incubated for 40 min at 37°C in the presence of anandamide (5 μmol/L) or vehicle added both in the serosal and mucosal compartment. A 30-µL sample was collected from the serosal compartment every 10 min and replaced with an equivalent volume of glucose-free buffer. Samples were mixed with 4.5 mL of Ultima Gold XR liquid scintillation cocktail (PerkinElmer) and radiolabeled glucose counted by a Packard Tricarb 2000 counter (PerkinElmer).

Transit through the stomach and small intestine was measured by administering a nonabsorbed marker (10% charcoal suspension in 5% gum arabic) as previously described (23). In brief, fasting mice were injected intraperitoneally with anandamide and CBR antagonists respectively 10 and 30 min before gavage with 0.2 mL of the charcoal suspension. Animals were killed by cervical dislocation 25 min later. The entire small intestine was removed, and the distance from the pylorus to the front of the charcoal bolus was measured. Transit rate was calculated as [(distance to charcoal front)/(total length from pylorus to cecum)] × 100 (%).

Total mRNAs were extracted with Tri-Reagent (Euromedex, Souffelweyersheim, France), and 1 µg RNA was reverse transcribed using the Iscript cDNA kit (Bio-Rad). Real-time PCR was performed as described previously (24) using a StepOnePlus real-time PCR system (Life Technologies, Saint-Aubin, France). Primer sequences used for amplification were as follows: forward ccgcaaagatagtcccaatg and reverse aaccccacccagtttgaac for CB1R, forward ccgcaaagatagtcccaatg and reverse aaccccacccagtttgaac for CB2R, and forward ccgcaaagatagtcccaatg and reverse aaccccacccagtttgaac for TATA box binding protein. For each gene, a standard curve was established from four cDNA dilutions (1/10 to 1/10,000) and used to determine the relative gene expression variation after normalization with TATA box binding protein expression.

Results are expressed as means ± SEM. Data were analyzed statistically with GraphPad InStat software (GraphPad Software, Inc., La Jolla, CA) using two-way ANOVA followed by the Tukey post hoc test, or using Student t test. Differences were considered significant at P < 0.05.

To evaluate the impact of CBR activation on whole-body glucose homeostasis, we first assessed glucose tolerance after a single intraperitoneal anandamide injection (10 mg/kg). In order to respect the natural route of nutrient administration, we performed OGTT (2 g/kg). Interestingly, the increase in glycemia in response to glucose ingestion was considerably smaller in mice treated with anandamide compared with control, and this was associated with an improvement of glucose tolerance as illustrated by the AUC_0–2h calculations (Fig. 1).

To determine whether anandamide affected tissue glucose utilization, we measured glucose clearance after intraperitoneal administration (intraperitoneal GTT, 2 g/kg). In these conditions, excluding any involvement of the gastrointestinal tract, results indicated that anandamide induced glucose intolerance (Fig. 2A) and prevents insulin (intraperitoneal injection, 0.5 UI/kg) to rapidly decrease plasma glucose compared with controls (Fig. 2B). Figure 2C also showed that anandamide did not decrease fasting glycemia; rather, drug injection even induced a rapid elevation of t_{0–15 min} plasma glucose levels (+29.69 ± 2.68% for anandamide vs. 11.3 ± 2.87% for vehicle).

To explore the possibility of an alteration of intestinal glucose absorption in anandamide-injected mice, we first performed an OXT (1 g/kg), an investigative tool to study small intestinal function (25). To verify that OXT accurately reflected the efficiency of glucose absorption, OGTT and OXT were also performed in the presence of phlorizin, a potent inhibitor of the SGLT-1 (26). Results confirmed that phlorizin suppressed both plasma appearance of glucose (data not shown) and xylose in response to oral glucose and xylose load, respectively (Fig. 3A). Using this method, we interestingly observed that plasma xylose appearance was strongly delayed in response to anandamide injection compared with vehicle whereas the maximum plasma levels were not different (Fig. 3A), suggesting that anandamide did not alter trans epithelial transport. To further investigate this notion, the direct effect of anandamide on intestinal glucose transport was evaluated using jejunal everted sacs exposed either to low glucose (10 μmol/L) or high glucose (30 mmol/L) concentration to study respectively SGLT-1– and GLUT-2–mediated transport. As expected, glucose transport was strongly inhibited by phlorizin in 10 μmol/L glucose experiments (Fig. 3B) and by the combination of both phlorizin and phloretin in 30 mmol/L glucose experiments (Fig. 3C). On the other hand, we did not observe any effect of anandamide on transporter-mediated uptake of glucose in comparison with vehicle whatever the concentration tested (Fig. 3B and C).

The effect of acute anandamide treatment on gastrointestinal motility was evaluated using the charcoal meal test. The distance traveled by the nonmetabolizable charcoal solution in the small intestine 25 min after gavage was considerably reduced after administration of anandamide compared with vehicle (Fig. 4A), suggesting that the drug decreased either gastric emptying or small intestine motility or both. To determine whether the delay in plasma glucose appearance induced by anandamide was exclusively due to a slower gastric emptying, we next performed a GTT in response to a direct duodenal load. In these conditions, plasma glucose levels were significantly lower at t = 15 min after infusion in anandamide-treated mice compared with control mice (Fig. 4B). Furthermore, we noticed both in duodenal GTT and OXT that anandamide delayed absorption without affecting maximal blood glucose and xylose levels, suggesting that anandamide could reduce small intestine motility. We also tested the possibility that the inhibitory action of the drug could be limited due to its rapid degradation in cells and tissues (27). To verify this hypothesis, two successive doses of anandamide (at t = 10 and t = 0 min) were administered during duodenal GTT. Results showed a prolonged effect of the drug as it further delayed maximal blood glucose associated with a glucose intolerance as indicated by AUC_0–120 (Fig. 4B), suggesting that a steady-state activation of CBR would profoundly alter glucose absorption and plasma clearance. Remarkably, glucose absorption was no more altered by anandamide when duodenal GTT was performed in CB1R^−/− mice (Fig. 4C).

GIP and GLP-1 are rapidly released from enteroendocrine cells of the small intestine in response to food ingestion. Both stimulate insulin secretion in a glucose-dependent manner in order to tightly control postprandial glycemia (28). As we showed that anandamide prevents hyperglycemia when glucose was given orally, we investigated whether anandamide’s effect on blood glucose was mediated by incretin-dependent stimulation of insulin production. We showed that plasma GIP, GLP-1, and insulin levels were strongly lower in anandamide-injected animals during OGTT (Fig. 5A). In addition, when glucose clearance was measured in the presence of the GLP-1R antagonist exendin 9-39 (to block the action of GLP-1 on pancreas β-cells), the effect of anandamide on glycemia was not prevented (Fig. 5B).

The role of CB1R and CB2R in controlling gastrointestinal motility and postprandial glycemia was approached using different CBR antagonists. The inhibitory effect of anandamide on transit was partially abrogated when mice were pretreated with specific CB1R antagonist SR141716 or AM6545, the action of the latter being restricted to the periphery (Fig. 6A). Whereas the CB2R antagonist AM630 also slightly limited the effect of anandamide, the coinjection of AM630 and SR141716 nearly normalized gastrointestinal transit (Fig. 6A). It was concomitantly observed that the delay of plasma glucose appearance induced by anandamide during OGTT was minored in the presence of AM630 (Fig. 6B). Besides, in the stomach of these animals, molecular analysis revealed the presence of much higher levels of CB2R mRNA compared with CB1R mRNA (Fig. 6C).

To further investigate the contribution of CB2R on postprandial glycemia, OGTT and OXT were performed in CB1R^−/− mice. Plasma glucose and xylose levels were also altered by anandamide during carbohydrate challenge in CB1R^−/− mice (Fig. 7A and B). However plasma xylose appearance was far more reduced by anandamide in wild-type mice than in CB1R^−/− (Fig. 7B), suggesting that the action of the drug was largely mediated by CB1R but also by CB1R-independent mechanisms. In line with this hypothesis, we showed that AM630 partially counteracted the action of anandamide on plasma glucose levels during OGTT performed in CB1R^−/− mice (Fig. 7A). Interestingly, we also observed that CB1R^−/− mice exhibited a much slower gastrointestinal motility compared with wild type and that AM630 strongly improved transit (Fig. 7C). As glucose absorption was no longer altered by anandamide during duodenal GTT in CB1R^−/− mice (Fig. 4C), it could be advanced that the delay in glucose absorption in CB1R^−/− mice observed during OGTT was consecutive to a slowing down of gastric emptying induced by stomach CB2R activation.

The impact of anandamide on whole-body glucose homeostasis was also assessed in DIO mice. These mice were both glucose intolerant and insulin resistant. Injection of anandamide into DIO mice reduced blood glucose appearance during OGTT (Fig. 8A). Xylose appearance was also strongly inhibited by anandamide in the 0–45-min range of OXT (Fig. 8B). It should be pointed out that the administration of the CB1R antagonist SR141716 alone before OXT did not induce any effect on xylose absorption (Fig. 8B), suggesting that intestinal CB1Rs were likely not activated by endogenous cannabinoids in DIO mice. Thus, evidence comes from these findings that the effect of anandamide on glucose appearance was maintained in obese mice. Considering that anandamide was more potent in reducing xylose appearance in DIO versus lean mice (Fig. 8C), we explored the possibility that intestinal ECS activation by anandamide was accentuated in obese mice due to the presence of high levels of CB1R. This assumption was supported by the expression levels of CB1R mRNA in the small intestine mucosa that were threefold higher in DIO than in lean mice (Fig. 8D).

In a recent study showing that hepatic CB1Rs contribute to diet-induced hepatic insulin resistance by inhibiting both insulin signaling and clearance, Liu et al. (10) provided evidence that acute peripheral administration of anandamide causes CB1R-mediated glucose intolerance in mice. Although we drew the same conclusions from the present findings obtained after intraperitoneal glucose load, we also intriguingly observed that the same dose of anandamide improved postprandial glycemia during an OGTT. The discrepancies between the results of intraperitoneal GTT experiments (10 and present study) and OGTT experiments reported here are unquestionably due to the impact of anandamide on events related to intestinal glucose absorption that are bypassed when glucose load is administered intraperitoneally. Since concordant findings from this study and literature (10) showed that peripheral CB1R activation by anandamide induced insulin resistance and endogenous glucose production, it could be excluded that the drug prevented hyperglycemia after oral load increasing blood glucose clearance, insulin sensitivity, and/or insulin production. Rather, we presented evidence that anandamide reduced postprandial glycemia, delaying intestinal glucose absorption. This was first demonstrated using oral xylose test (OXT) as a clinical marker for intestinal absorption. Since xylose metabolism is not affected by either insulin secretion or insulin resistance, OXT directly informed about the impact of anandamide on gastrointestinal events without being hindered by the peripheral effects of the drug on glucose clearance. Therefore, it should be considered that two main factors might prevent hyperglycemia after oral glucose loading following anandamide injection: 1) reduction of gastrointestinal motility and/or 2) inhibition of small intestine glucose transport.

The presence of a functional ECS has been identified in the gut. CB1Rs are expressed in the gastrointestinal tract of many species (12,14). Immunohistochemical studies indicate that the enteric nervous system is the main site of CB1R expression and could be the main site of action for cannabinoids in the gastrointestinal tract (13). Myenteric CB1Rs are physiologically involved in the regulation of gastric and intestinal motility and may constitute a physiological “brake” along the gastrointestinal tract in vivo (15,29,30). Interestingly, CB1R and CB2R are expressed in the epithelium of human colonic tissue (31), and our molecular analysis also revealed CB1R and CB2R gene expression in gastric and small intestine mucosa.

It is generally accepted that there is a positive relationship between gastrointestinal motility and glucose absorption. Notably, even minor perturbations in gastric emptying may have a substantial impact on postprandial glycemia in healthy subjects and patients with diabetes. Indeed, more rapid gastric emptying results in a greater initial glycemic response (17–19). The reduction of gastric and intestinal motility by cannabinoids has been well documented in both human and animal studies (15,16), whereas the consequences on glycemic control are poorly described. In our study, reduction of maximal plasma glucose and delay in xylose appearance is consistent with the strong reduction of gastrointestinal motility induced by anandamide. Available evidence indicates that CBR agonists reduce smooth muscle contractility in different regions of the gastrointestinal tract in animals and humans (32). In line with this, distinct effects of anandamide on gastric emptying and small intestine motility could explain why blood glucose appearance was also reduced in duodenal GTT experiments in which the influence of gastric emptying was insubstantial.

Monosaccharides are absorbed through the intestine by a transepithelial transport system. Two types of hexose transporters have been identified in human and murine small intestine: SGLTs and sodium-independent GLUTs (33). When the lumen of the bowel is exposed to high concentrations of glucose, GLUT-2 appears to account for the absorption from the enterocytes after a rapid dynamic process of translocation to the apical membrane. Recent studies suggest that the presence of GLUT-2 at the apical membrane is dependent on the activity of SGLT-1 (34–36). The effect of cannabinoids on intestinal absorption of ³H-2-deoxy-d-glucose (2-DG) has previously been investigated in Caco-2 cells (37). From this study, authors hypothesized that cannabinoids do not interfere with the intestinal GLUT-2–mediated apical uptake of glucose. In accordance with these previous data, our findings obtained from the everted sac method further indicated that activation of intestinal CBR by anandamide does not alter SGLT-1– or GLUT-2–mediated glucose transport. Finally, this strongly suggests that improvement of postprandial glycemia induced by anandamide was not due to a direct inhibition of the intestine mucosa capacity to absorb glucose.

GIP release is triggered by the arrival of nutrients in the upper small intestine, where it is produced by enteroendocrine K cells located predominantly in the duodenal epithelium (38), and it is well recognized that exogenous GLP-1 and GIP inhibit glucose absorption by slowing gastric emptying (39,40) or reducing intestinal motility (41). However, it is very unlikely that incretin variations were responsible for alteration of glucose absorption in our study since GIP and GLP-1 plasma levels were reduced after anandamide injection and blockade of GLP-1 receptors by exendin 9-39 did not modify glycemic profile. Rather, it is very likely that hormone secretion was reduced due to the slow delivery of glucose into the duodenum.

To date, available evidence suggests a role for CB2R in the regulation of gastrointestinal motility. CB2R has been showed to inhibit defecation in mice (42) and to reduce the gastrointestinal transit in a model of intestinal inflammation (43,44). It has also been demonstrated that CB2Rs were expressed in the stomach and that activation of the receptor with exogenous agonists reduces gastric motility (45,46). In the current study, we first observed that CB2R gene expression was significant in the stomach. Then, we showed that part of the effects of anandamide on glycemia during glucose challenge observed in wild-type mice persisted in CB1R^−/− and were partially counteracted by the CB2R antagonist AM630. In addition, the action of anandamide on gastrointestinal transit was abrogated only by the combined injection of SR141716 and AM630, suggesting that anandamide could reduce transit and thereby plasma glucose appearance, activating both receptors. Nevertheless, the problem of the respective contribution of CB1R and CB2R in regulating gastric emptying and small intestine motility was not resolved. The case of CB1R^−/− mice is also of particular interest since these animals exhibited a naturally low transit rate that could be strongly improved by AM630, suggesting the existence of a CB2R-mediated inhibitory tone. However, in these conditions, AM630 could also have exerted a CB2R inverse agonist effect, as already described in CHO cells (47). Finally, our data suggested that the regulation of gastrointestinal motility by CB2R deserves to be further studied especially in pathological conditions since an enhanced expression of these receptors has been reported during obesity and endotoxic inflammation (31).

In our study, the inhibitory effect of anandamide on glucose appearance was amplified in DIO mice and associated with high mRNA CB1R expression in the intestine of these animals. Nevertheless, recent data revealed that obesity was associated with a decrease in anandamide content in the stomach of obese mice and the intestine of Zucker rats (15,48), suggesting that ECS activity is downregulated in the gastrointestinal tract of obese animals (49), contrary to other peripheral organs (5,50). In line with this, we also showed that SR141716 administered alone to DIO mice before OXT did not modify xylose appearance, whereas it was strongly inhibited with exogenous anandamide. This suggests that even though CB1R expression was increased, it is very unlikely that CB1Rs were tonically activated by endocannabinoids in DIO mice.

In summary, our findings demonstrated that acute activation of intestinal ECS reduced postprandial glycemia. Reduction of plasma glucose appearance was not due to an alteration of intestinal glucose transport but rather to the inhibition of gastrointestinal transit consecutive both to CB1R and CB2R activation. Identifying precisely the respective role of CB1R and CB2R in each part of the gastrointestinal tract could help to develop new pharmacological tools to improve postprandial glycemia, an important component of the overall glycemic burden (20,21). Further studies are also necessary to determine whether the endocannabinoid tone is modified in the gastrointestinal tract of DIO mice and the consequences on postprandial intestinal glucose delivery.

Acknowledgments. The authors thank Serge Monier (Plateforme de Cytométrie) for excellent technical assistance.

Funding. This work was supported by funds from the Regional Council of Burgundy and European Union (FEDER).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.T.-F., L.D., J.G., and P.D. designed, performed, and analyzed the experiments and wrote the manuscript. T.M. performed some experiments. B.V. analyzed the data and reviewed the written draft of the manuscript. P.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.