Presence of functional cannabinoid receptors in human endocrine pancreas
We examined the presence of functional cannabinoid receptors 1 and 2 (CB1, CB2) in isolated human islets, phenotyped the cells producing cannabinoid receptors and analysed the actions of selective cannabinoid receptor agonists on insulin, glucagon and somatostatin secretion in vitro. We also described the localisation on islet cells of: (1) the endocannabinoid-producing enzymes N-acyl-phosphatidyl ethanolamine-hydrolysing phospholipase D and diacylglycerol lipase; and (2) the endocannabinoid-degrading enzymes fatty acid amidohydrolase and monoacyl glycerol lipase.
Real-time PCR, western blotting and immunocytochemistry were used to analyse the presence of endocannabinoid-related proteins and genes. Static secretion experiments were used to examine the effects of activating CB1 or CB2 on insulin, glucagon and somatostatin secretion and to measure changes in 2-arachidonoylglycerol (2-AG) levels within islets. Analyses were performed in isolated human islets and in paraffin-embedded sections of human pancreas.
Human islets of Langerhans expressed CB1 and CB2 (also known as CNR1 and CNR2) mRNA and CB1 and CB2 proteins, and also the machinery involved in synthesis and degradation of 2-AG (the most abundant endocannabinoid, levels of which were modulated by glucose). Immunofluorescence revealed that CB1 was densely located in glucagon-secreting alpha cells and less so in insulin-secreting beta cells. CB2 was densely present in somatostatin-secreting delta cells, but absent in alpha and beta cells. In vitro experiments revealed that CB1 stimulation enhanced insulin and glucagon secretion, while CB2 agonism lowered glucose-dependent insulin secretion, showing these cannabinoid receptors to be functional.
Together, these results suggest a role for endogenous endocannabinoid signalling in regulation of endocrine secretion in the human pancreas.
KeywordsAnandamide 2-Arachidonoylglycerol Beta cell Cannabinoid receptors Fatty acid amidohydrolase Glucagon Human Insulin Islets of Langerhans Somatostatin
cannabinoid receptor 1
cannabinoid receptor 2
fatty acid amidohydrolase
- JWH 133
monoacyl glycerol lipase
N-acyl-phosphatidyl ethanolamine-hydrolysing phospholipase D
The endocannabinoids are relevant homeostatic signals whose dysregulation contributes to obesity and type 2 diabetes [6, 13]. A clinical trial in obese patients treated with Rimonabant, a CB1 antagonist, resulted in effective reduction in body weight, waist circumference and insulin resistance [14, 15]. Additional studies demonstrated that CB1 blockade improves insulin resistance, insulinaemia and glycosylated haemoglobin in obese patients with type 2 diabetes . Since these actions are not totally explained by the weight loss induced by the anorectic actions derived from CB1 blockade, the endocannabinoids may also modulate metabolism in peripheral organs. This hypothesis has been confirmed in animal models where CB1 were found to modulate lipid and glucose metabolism in insulin-sensitive tissues such as the adipose tissue  and the liver . Recent studies extended this notion to the endocrine pancreas, where the endogenous cannabinoid system has recently been identified in mice, rats and the rat insulinoma beta cell line RIN-m5F [17, 18, 19, 20]. Whereas stimulation of CB1 in the rat leads to glucose intolerance, activation of CB2 improves glucose handling after a glucose load [18, 19]. In mice, CB2 modulate calcium oscillations and insulin secretion in vitro . These actions are derived from glucose-induced alterations in endocannabinoid production, as demonstrated in the pancreatic beta cell line RIN-m5F. Thus, elevations of glucose concentration in the culture media are associated with a rise in the levels of both 2-AG and AEA .
To date, no studies have addressed the presence and functional significance of cannabinoid receptors in human endocrine pancreas. However, the clear and conclusive effects of chronic treatment with the CB1 antagonist Rimonabant on insulin resistance in obese humans, with or without type 2 diabetes, clearly suggest the presence of this system in human pancreatic islets [14, 15, 16]. In order to confirm this hypothesis, we examined the presence of functional CB1 and CB2 in isolated human islets, as well as the localisation of the machinery for synthesis and degradation of endocannabinoids in human pancreatic tissue.
Human islet isolation
Islets were isolated and purified from human pancreases using the Ricordi method [21, 22]. The present studies were performed in pancreas from four brain-dead, heart-beating, non-diabetic, non-obese (mean BMI 28.3 ± 0.7 kg/m2) adult organ donors (mean age 50 ± 14 years; two women, two men). Cause of death was stroke for two donors and anoxic encephalopathy for the other two. All procedures were performed according to specific legal guidelines and written informed consent was obtained from each donor’s family; the local ethics committee of Carlos Haya Hospital approved and supervised the experiments.
The pancreas was cut into two parts, cannulated and perfused with cold (4–8°C) liberase (Liberase-HI; Roche Molecular Biochemical, Indianapolis, IN, USA). After perfusion, the pancreas was minced, transferred to the Ricordi chamber and digestion carried out at 37°C. Digestion was stopped and the digest diluted with 2 litres dilution solution (Mediatech Cellgro, Herndon, VA, USA). The digest was collected and centrifuged for 3 min at 1,000 rpm (225 g) and 4°C. Pellets were washed and recombined in cold modified University of Wisconsin solution. The tissue digest was purified on a continuous ficoll (Biochrom, Berlin, Germany) density gradient, using the Cobe 2991 cell separator/processor (Gambro, Lakewood, CO, USA).
The fraction containing the purified islet mass was analysed and the purity of the preparation assessed with dithizone. The preparations used in this study had more than 80% purity. Fresh aliquots of 1,000 islet equivalents (IEQs) were snap-frozen for subsequent mRNA quantification and western blotting analysis. The remaining purified islets were cultured at 30,000 IEQs per 75 cm2 non-treated flask (Nunc, Wiesbaden, Germany) in a final volume of 30 ml CMRL-1066 medium (Mediatech Cellgro) supplemented with 10% FCS (v/v), 100 IU/ml penicillin, 100 μg/ml streptomycin, 2.8 μg/ml amphotericin and 2 mmol/l l-glutamine; this was done for 3 to 5 days at 37°C, 95% relative humidity and 5% CO2. Culture medium was replaced every 2 days. After 3 to 5 days of culture the preparations were checked for viability by Trypan Blue exclusion test and found to contain <5% damaged cells.
Immunohistochemistry in human pancreas samples
Immunohistochemical and immunofluorescence studies were performed in human pancreatic tissue obtained from the Pathology Department tissue bank at Carlos Haya Hospital. Biopsy samples were taken from four different non-diabetic, non-obese (mean BMI 27.5 ± 1.5 kg/m2) adult patients (mean age 68 ± 9 years; three women, one man) with pancreatic adenocarcinoma (non-endocrine tumour). All procedures were performed according to specific legal guidelines and written informed consent were obtained from patients; the local ethics committee of Carlos Haya Hospital approved and supervised the experiments. See additional information in Electronic supplementary material (ESM).
NAPE-PLD-, DAGLα- and DAGLβ-specific antibody generation
Polyclonal rabbit antibodies were generated against cannabinoid machinery proteins as described in ESM. Immunising peptides were: (1) a 13-amino acid peptide comprising part of both the C-terminal and the N-terminal region of NAPE-PLD (MDENSCDKAFEET); (2) a 16-amino acid peptide from the C-terminal region of DAGLα (CGASPTKQDDLVISAR); and (3) a 16-amino acid peptide from an internal sequence of DAGLβ (SSDSPLDSPTKYPTLC). Extensive validation studies of these antibodies as immunocytochemistry markers were performed in brain samples (ESM Fig. 1).
Double immunofluorescence and immunohistochemistry
Paraffin-embedded sections of human pancreases were analysed for the presence of CB1, CB2 and FAAH in alpha (glucagon), beta (insulin) and delta (somatostatin) pancreatic islet cells by double immunofluorescence. Sections were incubated overnight at room temperature with mouse anti-insulin (dilution 1:200; Sigma-Aldrich Quimica S.A., Madrid, Spain), anti-glucagon (1:200; Sigma) or anti-somatostatin (1:200; Genetex, San Antonio, TX, USA) antibody and a rabbit anti-CB1 (dilution 1:100; ABR—Affinity Bioreagents, Golden, CO, USA), anti-CB2 (1:100; ABR) or anti-FAAH (1:50; Cayman Chemical, Ann Arbor, MI, USA) antibody. After extensive washes in PBS, the sections were incubated for 2 h at room temperature in a secondary anti-mouse IgG–FITC antibody (dilution 1:200; Sigma) and a secondary anti-rabbit IgG–Cy3 antibody (1:300; Jackson Immunoresearch Laboratories, West Grove, PA, USA). Finally, the sections were washed in PBS and analysed under epifluorescence microscopy (Olympus Europa, Hamburg, Germany). Additional immunohistochemistry studies were done to determine the levels of NAPE-PLD, DAGLα, DAGLβ, FAAH, MAGL, insulin and chromogranin A. In all cases, the specificity of the immunostaining was confirmed by omission of the first antibody or the use of preadsorption of primary antibodies with the immunising peptide. Digital photographs were taken with an Olympus BX41 microscope (Olympus).
Real-time quantitative PCR
Real-time quantitative PCR was used to measure CB1 (also known as CNR), CB2 (also known as CNR2) and NAPE-PLD mRNA expression according to Hansson et al. . Briefly, total RNA was isolated from snap-frozen pellets containing 1,000 IEQs from three different donors by using Trizol reagent (Gibco BRL Life Technologies, Baltimore, MD, USA). All RNA samples had A260:280 ratios of 1.8 to 2.0. Total RNA from each sample and random hexamers were used to generate first strand cDNA using transcriptor reverse transcriptase (Roche Applied Science, Indianapolis, IN, USA). Negative controls included reverse transcription reactions omitting reverse transcriptase. The cDNA obtained was used as the template for real-time quantitative PCR with an iCycler system (Bio-Rad, Hercules, CA, USA) using the QuantiTect SYBR Green PCR kit (Qiagen, Hilden, Germany). A comparative analysis of the expression of both cannabinoid receptors in human islets was made using reference total RNA samples from commercial standards of both human cerebellum and human leucocytes (BD Biosciences, Palo Alto, CA, USA). Primers for PCR reactions and annealing temperatures are shown in the ESM. Quantification was carried out with a standard curve run at the same time as the samples with each reaction run in duplicate. Absolute values from each sample were normalised with regard to beta actin.
Western blot analysis
Frozen cell pellets from human pancreatic islets (1,000 IEQs) were suspended in 50 μl SDS sample buffer containing dithiothreitol and heat-denatured for 5 min at 95°C. For immunoblotting, equal amounts of protein from lysates (10 μl per lane) were subjected to 7.5% (w/v) SDS-PAGE and homogeneous transfer to nitrocellulose membranes controlled by Ponceau red staining. Blots were preincubated for 1 h at room temperature with PBS containing 0.1% (v/v) Tween 20 and 2% (w/v) albumin fraction V from bovine serum (blocking buffer). For protein detection, each blotted membrane lane was incubated separately with the specific CB1 (1:250; ABR), CB2 (1:250; ABR), FAAH (1:100; Cayman), MAGL (1:500; kindly donated by D. Piomelli, Department of Pharmacology, University of California, Irvine, CA, USA), DAGLα (1:100), DAGLβ (1:100) and NAPE-PLD (1:75) antibodies from rabbit, diluted in PBS containing 0.1% (v/v) Tween 20 and 2% (w/v) albumin fraction V from bovine serum. This was done overnight at room temperature. The specific protein bands were visualised using the enhanced chemiluminescence technique (Amersham International, Amersham, UK) and an imaging system (Auto-Biochem; LTF Labortechnik, Wasserburg/Bodensee, Germany). Western blots showed that each primary antibody detected a protein of the expected molecular size. As additional controls, blotted membrane lanes were incubated with the primary antibody preadsorbed with the corresponding immunising peptide.
Static secretion of insulin, glucagon and somatostatin in isolated islets
After 5 days in vitro, total IEQs and acinar contamination were determined by staining with dithizone. Viability was checked by Trypan Blue exclusion. Less than 5% of cells within each islet were non-viable and no acinar tissue was adhered. Human cultured islets were then washed with Hanks’ solution and suspended in a serum-free medium. Groups of 50 IEQs were plated on sterile 24-well plates (Nunc) with 1 ml medium containing (mmol/l): 115 NaCl, 10 NaHCO3, 5 KCl, 1.1 MgCl2, 1.2 NaH2PO4, 2.5 CaCl2 and 25 HEPES; plus 1% albumin and either 3 mmol/l or 11 mmol/l glucose. Additionally, one of the following drugs was added to the medium (Tocris Bioscience, Bristol, UK): 2-AG (10−5 to 10−6 mol/l); AEA (10−8 to 10−7 mol/l); arachidonyl-2′-chloroethylamide (ACEA), a specific CB1 activator  (10−7 to 10−9 mol/l); 3-(1′,1′-dimethylbutyl)-1-deoxy-Δ8-tetrahydrocannabinol (JWH 133), a specific CB2 activator  (10−7–10−9 mol/l); or N-(piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM251), a selective CB1 antagonist  (10−7 mol/l). Concentrations of cannabinoid receptor agonists and antagonists were selected according to the reported Ki value of each compound. Each experimental condition was assayed in quadruplicate and islets isolated from four different donors were employed. The islets were incubated at 37°C and 5% CO2 for 1 h, after which the medium containing the islets was collected in 1.5 ml tubes with 520 kIU aprotinin and centrifuged for 5 min at 1,000×g and 4°C. The supernatant fractions were divided into aliquots and stored at −80°C for further quantification of insulin, glucagon and somatostatin by means of specific commercial radioimmunoassays or enzyme immunoassay kits. The cellular pellets were frozen at −80°C to measure total protein amount (see ESM). Secretion of insulin, glucagon and somatostatin from each well was normalised with the protein content of their 50 IEQs.
Analysis of islet 2-AG levels
Because the immunohistochemical analysis revealed that the islets contained the machinery for synthesis and degradation of 2-AG but not AEA, we measured intracellular levels of 2-AG in islets incubated either in low (3 mmol/l) or high (11 mmol/l) glucose concentrations. Seven-hundred IEQs per well were incubated and 1 h after incubation islets were centrifuged and assayed in triplicate for 2-AG content using quantitative GC-MS in a TRACE GC/MS 2000 system with electron impact ionisation detector (Finnigan, San Jose, CA, USA; see ESM).
Results are expressed as the mean ± SEM. The significance of differences between the groups was evaluated by either Student’s t test (two-tailed, paired groups) or one-way ANOVA followed by Newman–Keuls test as post hoc (multiple group comparison). A p value of <0.05 was considered significant.
Human islets express CB1 and CB2 mRNA
mRNA expression of CB1 and CB2
CB1: β-actin (×10−4)a,b
CB2: β-actin (×10−4)a,b
8.63 ± 1.07
0.0191 ± 0.0010
2.95 ± 0.81
0.0224 ± 0.0110
0.0148 ± 0.0001
1.86 ± 0.56
Double immunofluorescence revealed the cell subtype containing both CB1 and CB2 in human islets
CB1 and CB2 as well as the enzymes for synthesis and degradation of endocannabinoids were detected by Western blot analysis in human islets
High glucose increases 2-AG content within islets
In vitro stimulation of CB1 and CB2 modifies insulin, glucagon and somatostatin secretion
Figure 4c shows how stimulation of CB1 with both selective (ACEA) and natural (AEA and 2-AG) CB1 agonists increased insulin release from islets cultured under high glucose concentrations. This effect was also observed at low glucose concentrations (3 mmol/l) and was antagonised by the selective CB1 antagonist AM251  (Fig. 4d). CB2 stimulation with the selective agonist JWH 133 lowered glucose-dependent insulin release (Fig. 4c), an effect mediated by CB2, since it was found to be antagonised by the selective CB2 antagonist AM630 (data not shown).
Regarding somatostatin secretion, we were unable under our experimental conditions to find any specific effect of the CB2 agonist JWH 133 in islets cultured at 11 mmol/l glucose (Fig. 5d). However, delta cells from islets incubated at 3 mmol/l glucose responded to the CB1 agonist ACEA with a potent release of somatostatin (Fig. 5c). Because CB1 are not located in somatostatin cells, the effects of this CB1 agonist on somatostatin are probably derived from the potent stimulation of glucagon and insulin secretion.
The molecular machinery needed to release and process 2-AG signalling is present in islets
Four major contributions result from this study. First, CB1 and CB2 are present in human islet cells with a specific distribution. While CB1 are in both alpha and beta cells, CB2 are located in somatostatin-secreting delta cells. In humans, CB2 levels are much lower than CB1 levels throughout the endocrine pancreas. This finding supports the existence of species-specific differences in the distribution of cannabinoid receptors: while mice and rats produced functional CB2 in beta cells, humans do not [17, 19]. This may be relevant for future analysis of cannabinoid receptor-dependent actions on islet physiology. Second, while the net effects of CB1 agonists on islet secretion are stimulatory, actions of CB2 agonists are inhibitory, in agreement with previous studies in rat and mice cells [17, 20]. CB1 stimulation enhances insulin secretion (both at low and high glucose concentrations). On the other hand, CB2 stimulation reduces insulin secretion, an effect apparently not mediated by somatostatin secretion. Third, at low glucose concentrations, CB1 agonists stimulate glucagon secretion from alpha cells and somatostatin from delta cells. Alpha cells are extremely sensitive to CB1 agonists, since they respond in the nanomolar range. However, the effects of CB1 stimulation on somatostatin secretion are not direct, since delta cells do not produced CB1. This effect is probably mediated through glucagon-dependent somatostatin release, as a negative feedback mechanism, by which tight control of glucagon secretion would be achieved. Finally, analysis of the machinery for biosynthesis and degradation established that in the human pancreatic islet the endocannabinoid synthesised is 2-AG. We deduced this from the fact that both isoforms of DAGL, the enzymes that produce 2-AG, were found. The second endocannabinoid, AEA, seems to be produced outside the islets, as we were unable to detect the presence of NAPE-PLD, the main enzyme producing AEA. However, AEA seems to be active in islets because it was found to modulate insulin secretion. Moreover, the islets abundantly produce FAAH, the main AEA-degrading enzyme (ESM Fig. 3). One possibility is that AEA may be transported to the endocrine pancreas by the blood . According to studies in humans and rodents, elevated circulating levels of AEA result in hyperglycaemia, insulin resistance and elevated insulin levels through its multiple actions in the pancreatic islets, the adipose tissue and the liver [13, 18, 19, 36 and present results].
Overall, the present study suggests that intra-islet 2-AG production may serve as a local modulator of insulin and glucagon secretion in cells producing CB1 and CB2. 2-AG may be degraded in the islet, since we found relevant production of MAGL enzyme. This hypothesis is supported by the finding of 2-AG in the islets and by the relevant modulation that glucose concentration exerted on its intracellular levels. Our observation is in agreement with a previous report in rat insulinoma cells  and correlates with the 2-AG-dependent insulin secretion found in islets cultured under high glucose concentrations. Disappearance of 2-AG from islets under low glucose concentrations can be interpreted as a removal of a stimulatory signal needed for insulin secretion. Stimulation of CB1 under this low glucose condition is able to promote insulin release, but this response is associated with a potent counter-regulatory glucagon secretion.
The effects on insulin and glucagon secretion that were observed after CB1 stimulation are both stimulatory, indicating a desynchronisation of both types of cells. One potential explanation is the pleiotropic actions of AEA in the endocrine pancreas, as described in rodent islet cells . In these cells almost one third of the beta cells stimulated with AEA showed opposite patterns of calcium oscillation coupled to insulin secretion, indicating multiple mechanisms of action (; A. Nadal and P. Juan-Picó, unpublished results). Additional mechanisms may be the interference induced by AEA on GAP junctions  involved in electrical beta cell coupling . However, the fact that the selective blockade of CB1 with AM251 attenuated the stimulation of both insulin and glucagon secretion supports a more relevant contribution of the CB1.
Finally, we found that CB2 are located in somatostatin-secreting cells. Since this peptide is a potent intra-islet modulator of insulin and glucagon secretion , it is feasible to expect somatostatin-dependent responses to cannabinoid receptor agonists. Under our experimental conditions, CB2 activation did not change somatostatin secretion in 11 mmol/l glucose, suggesting that CB2-dependent inhibition of insulin secretion is not related to CB2-induced somatostatin release. Further research is needed to clarify the role of this receptor in delta cells.
After consideration of all the findings in this study and placing them within the context of endocannabinoid physiology, we propose a major role for CB1 and 2-AG in the regulation of islet physiology and energy homeostasis. CB1 stimulation was able to increase insulin and glucagon secretion, as well as lipogenesis in liver and adipose tissue, while it blocked the incorporation of glucose into the muscle cells, leading to a ‘saving cycle’: reduced energy expenditure and increased energy storage [6, 7, 12, 13]. Enhanced insulin release may help to facilitate the incorporation of glucose into adipocytes, while enhanced glucagon release may sustain high plasma glucose levels, favouring this ‘saving cycle’. Under this hypothesis, CB1 can be considered a new ‘thrifty gene’. Its physiological and pharmacological profile is strikingly similar to that of the μ-opioid receptor, a recently proposed contributor to the ‘saving cycle’  and a major partner of CB1 signalling in the brain . In light of this potential role, and supported by clinical trials using the CB1 antagonist, the endocannabinoid system can be considered a promising target for pharmacological development in diabetes and complicated obesity.
This study was supported by Consejerias de Salud e Innovación, Junta de Andalucia (SAS 144/04 and SAS 0064/05), Instituto de Salud Carlos III (grants 03/0178, 07/0880 and 07/1226), Redes Temáticas ISCIII-RETIC RD06/001 and RD06/0015/008 and MEC (grants SAF 2004-07762, SAF 2005-08014, SAF 2006-12863, BFU2005-01052). P. Juan-Pico has a PhD scholarship from Instituto de Salud Carlos III. G. Milman was supported by US National Institute on Drug Abuse. The authors are grateful to: J. M. Mellado (Human Islet Isolation Facility, Carlos Haya Hospital, Malaga, Spain) for kindly helping in human islet isolation; D. Piomelli (Department of Pharmacology, University of California, Irvine, CA, USA) for providing rabbit anti-MAGL and anti-NAPE-PLD; J. Romero (Research laboratory, Fundacion Hospital Alcorcon, Madrid, Spain) and K. Mackie (Department of Psychiatry and Brain Sciences, Indiana University, Bloomington, IN, USA) for kindly donating CB1 and CB2 blocking peptides; I. Sanchez (Regenerative Medicine Laboratory, IMABIS Foundation, Malaga, Spain) for technical assistance; J.C. Aledo (Department of Biochemistry and Molecular Biology, University of Malaga, Spain) for helping in antibodies generation; and A. Garcia-Ocaña (Division of Endocrinology, University of Pittsburgh, PA, USA) for initial support in immunohistochemistry.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.
- 1.Devane WA, Dysarz FA 3rd, Johnson M, Melvin LS, Howlett AC (1998) Determination and characterization of a cannabinoid receptor in rat brain. Mol Pharmacol 34:605–613Google Scholar
- 3.Stella N, Schweitzer P, Piomelli D (1992) A second endogenous cannabinoid that modulates long-term potentiation. Nature 388:773–778Google Scholar
- 14.Van Gaal LF, Rissanen AM, Scheen AJ, Ziegler O, Rossner S, RIO-Europe Study Group (2005) Effects of the cannabinoid-1 receptor blocker rimonabant on weight reduction and cardiovascular risk factors in overweight patients: 1-year experience from the RIO-Europe study. Lancet 365:1389–1397PubMedCrossRefGoogle Scholar
- 15.Pi-Sunyer FX, Aronne LJ, Heshmati HM, Devin J, Rosenstock J, RIO-North America Study Group (2006) Effect of rimonabant, a cannabinoid-1 receptor blocker, on weight and cardiometabolic risk factors in overweight or obese patients: RIO-North America: a randomized controlled trial. JAMA 295:761–775PubMedCrossRefGoogle Scholar
- 24.Hillard CJ, Manna S, Greenberg MJ et al (1999) Synthesis and characterization of potent and selective agonists of the neuronal cannabinoid receptor (CB1). J Pharmacol Exp The 289:1427–1433Google Scholar
- 27.Pettit DAD, Harrison MP, Olson JM et al (1998) Immunohistochemical localization of the neural cannabinoid receptor in rat brain. J Neurosci Methods 51:391–402Google Scholar