Influence of Cannabinoid Receptor Deficiency on Parameters Involved in Blood Glucose Regulation in Mice

Cannabinoids are known to influence hormone secretion of pancreatic islets via G protein-coupled cannabinoid receptor type 1 and 2 (CB1 and CB2). The present study was designed to further investigate the impact of cannabinoid receptors on the parameters involved in insulin secretion and blood glucose recognition. To this end, CB1 and CB2 receptor knockout mice (10–12 week old, both sexes) were characterised at basal state and compared to wild-type mice. The elimination of cannabinoid receptor signalling resulted in alterations of blood glucose concentrations, body weights and insulin levels. Changes were dependent on the deleted receptor type and on the sex. Analyses at mRNA and protein levels provided evidence for the impact of cannabinoid receptor deficiency on the glucose sensing apparatus in the pancreas. Both receptor knockout mouse lines showed decreased mRNA and protein amounts of glucose transporters Glut1 and Glut2, combined with alterations in immunostaining. In addition, pancreatic glucokinase expression was elevated and immunohistochemical labelling was modified in the pancreatic islets. Taken together, CB1 and CB2 signalling pathways seem to influence glucose sensing in β-cells by affecting glucose transporters and glucokinase. These alterations were more pronounced in CB2 knockout mice, resulting in higher blood glucose and lower plasma insulin levels.


Introduction
Hormone secretion of pancreatic islet cells is influenced via G protein-coupled receptors (GPCRs) which are characterised by their seven transmembrane helical domains and their coupling to diverse intracellular signalling pathways [1]. GPCR function is receiving significant interest, since over 30 GPCRs have currently been implicated in the development and progression of beta-cell dysfunction, insulin resistance, obesity and type 2 diabetes mellitus, some of which have successfully been targeted therapeutically [2]. normally occurring in high fat diet-fed mice [15]. Mice deficient in CB 1 displayed no changes in glucose tolerance and insulin sensitivity in association with diet-induced obesity [16]. Furthermore, in CB 2 −/− mice, high fat diet-induced insulin resistance was reduced [17] and a lack of CB 2 -mediated responses also protected mice from both age-related and diet-induced insulin resistance [18]. Another study on diet-induced obesity in CB 2 -knockout and CB 1/2 double-knockout mice reported that mice lacking both of the cannabinoid receptors were lean and resistant to diet-induced obesity. This phenotype was distinct from CB 2 -deficient mice which displayed signs of impaired glucose clearance [19]. Notably, in global models lacking cannabinoid receptors, other organs than pancreatic islets such as liver, skeletal muscle or adipose tissue might be directly or indirectly involved in the regulation of blood glucose levels or insulin resistance. In this context, many peripheral tissues are being influenced by the ECS controlling whole-body metabolism as shown in tissue-specific, genetically modified mice [20]. Still, only one study directly analysed the impact of CB 1 receptor knockout on pancreatic islet function including insulin, glucokinase and glucose transporter 2 [21]. However, components of the glucose sensing machinery such as glucokinase and glucose transporters are essential for islet cell function [22,23]. Notably, glucose transporter 2 (Glut2) is required for glucose stimulated insulin secretion in rodent islets, while GLUT1 plays a major role in human beta-cells [23].
Thus, the present study was performed to investigate the contribution of each cannabinoid receptor deletion in the regulation of hormonal secretion and the glucose sensing apparatus in pancreatic islets. Furthermore, the impact of receptor deletion was addressed in both sexes.

Cannabinoid Receptor Expression in Mouse Pancreatic Islets and Islet Cell Types
In order to assess the putative effects of cannabinoid receptor knockout, the expression of receptors in a murine alpha-cell line and different organs of wild-type (Wt) mice was verified by RT-PCR. The specificity of primers was checked by restriction analysis, and the sizes of PCR products were confirmed (Figure 1a). Immunolabelling of CB 1 was evident in pancreatic islets of Wt and CB 2 −/− mice, while only low levels of immunofluorescence were detected in the exocrine tissue. Most of the labelling was found in the centre of the islet, pointing to a higher expression of CB 1 in pancreatic beta-cells of mouse islets. Double labelling revealed CB 1 colocalisation in glucagon-producing cells as well. The specificity of the primary antibody was verified earlier [24] and is confirmed in the present study ( Figure 1b).  Figure 1. (a) Expression of cannabinoid receptor type 1 (Cnr1) and 2 (Cnr2) mRNA in mouse (m) tissues. RT-PCR followed by gel electrophoresis (left column) revealed the expression of mCnr1 and mCnr2 transcripts in different mouse organs, particularly, in pancreatic tissue including the islets of Langerhans and the mouse alpha-cell line αTC1.9. Restriction digestion of the 175 bp mCnr1 amplicon resulted in defined fragments with molecular sizes of 107 and 68 bp (right column). The restriction analysis of the 188 bp mCnr2 showed defined fragments with molecular sizes of 119 and 69 bp (right column). NTC: nontemplate control; L: 100 bp ladder; LR: low-molecular-range ladder; P: Cnr1 or Cnr2 amplification product; R: restriction fragments. (b) Immunohistochemical staining of cannabinoid receptor type 1 (CB 1 ) in the pancreatic tissue of wild-type (Wt) mice displayed specific labelling of an islet. Glucagon (Gcg, red) and CB 1 (green) double-immunolabelling demonstrated the presence of CB 1 , not only in beta-cells, but also in alpha-cells. At higher magnifications of alpha-cells (2 fold, right panels), weaker CB 1 staining was evident. No immunohistochemical staining of CB 1 was detected in pancreatic islets of CB 1 −/− knockout mice. In all cases, confocal optical sections were merged and are representative for pancreatic islets of the whole pancreatic tissue from three mice per group. Scale bar 20 µm.

Measurement of Body Weight, Blood Glucose, Plasma Insulin and Glucagon
Wt mice showed a mean body weight of 25.24 g (Figure 2a). In comparison, CB 1 −/− mice displayed a significant decrease of mean body weight (22. mice showed increased body weight (29.70 g). In general, the body weight of female mice in all groups was lower than that of male mice (Figure 2b). Compared to blood glucose levels of Wt mice (Figure 2c), we found no differences in the mean values of peripheral blood glucose in CB 1 −/− mice. However, male CB 1 −/− mice showed statistically The same finding was evident in a sex-specific analysis. Interestingly, in female Wt mice, slightly lower insulin levels were observed compared to male Wt mice (p = 0.0832), which became significant between female and male CB 2 −/− mice ( Figure 2f).
The mean plasma glucagon levels were nonsignificantly increased in CB 1 −/− (23.04 pg/mL) compared to Wt mice (16.60 pg/mL, p = 0.2305; Figure 2g). Female CB 1 −/− seemed to be responsible for this increase (Figure 2h).  Figure 2h). Taken together, the elimination of cannabinoid receptor signalling affected body weight, blood glucose as well as plasma insulin and glucagon levels, even though this was differently associated with the type of receptor deleted and the sex. The alterations in blood glucose and plasma insulin levels measured in CB 2 −/− mice were seen in female and male mice, indicating no sex-specific differences. Mean plasma glucagon levels pointed to a slight increase in CB 1 −/− mice. Female CB 1 −/− seemed to be responsible for this increase. Female CB 1 −/− and CB 2 −/− mice showed higher glucagon levels than their respective male counterparts. Values are presented as standard error of the mean (±S.E.M.) with n = 11-23 animals per group or n = 4-12 animals per group when analysing data for male or female Wt and knockout mice separately. * p < 0.05; ** p < 0.01; *** p < 0.001 for overall group comparisons within male or female Wt and knockout mice; † p < 0.05; † † p < 0.01; † † † p < 0.001 for sex-specific comparisons between male and female Wt or knockout mice; unpaired t-test.

Measurements of Transcripts and Proteins of Islet Hormones
The previous observations regarding the shift of metabolic parameters in cannabinoid receptor knockout mice ( Figure 2) led us to investigate the effects of cannabinoid receptor knockout on the function of pancreatic islets. To this end, we determined the transcript levels of the pancreatic islet hormones insulin (Ins1 and Ins2), glucagon (Gcg) and somatostatin (Sst) in pancreatic tissue ( Figure 3). The specificity of PCR products of pancreatic hormones was confirmed in an earlier study [25]. When comparing islet hormone transcript levels between Wt and knockout mice, no statistically significant differences were observed (Figure 3a-d). In addition, there were no significant variations considering sex-specific analyses ( Figure A1a-d).

Measurements of Transcripts and Proteins of Islet Hormones
The previous observations regarding the shift of metabolic parameters in cannabinoid receptor knockout mice ( Figure 2) led us to investigate the effects of cannabinoid receptor knockout on the function of pancreatic islets. To this end, we determined the transcript levels of the pancreatic islet hormones insulin (Ins1 and Ins2), glucagon (Gcg) and somatostatin (Sst) in pancreatic tissue ( Figure  3). The specificity of PCR products of pancreatic hormones was confirmed in an earlier study [25]. When comparing islet hormone transcript levels between Wt and knockout mice, no statistically significant differences were observed (Figure 3a-d). In addition, there were no significant variations considering sex-specific analyses ( Figure A1a-d). Western blot analysis of insulin showed no change in protein content in CB1 −/− , but a significant reduction of 48% in CB2 −/− mice ( Figure 4a). Considering sexes, a decrease of insulin was evident in both groups, with a significant decrease displayed in female CB2 −/− mice ( Figure A2a). In contrast, total glucagon protein levels were not altered in CB1 −/− and CB2 −/− (Figure 4b). But in male CB2 −/− mice, an increased glucagon level was measured when compared to the Wt, whereas in female knockout mice, a decrease was evident ( Figure A2b). Overall, higher somatostatin protein amounts were detected without reaching the level of significance in CB1 −/− mice ( Figure 4c). Interestingly, this trend was seen in female CB1 −/− mice only ( Figure A2c).  Western blot analysis of insulin showed no change in protein content in CB 1 −/− , but a significant reduction of 48% in CB 2 −/− mice ( Figure 4a). Considering sexes, a decrease of insulin was evident in both groups, with a significant decrease displayed in female CB 2 −/− mice ( Figure A2a). In contrast, total glucagon protein levels were not altered in CB 1 −/− and CB 2 −/− (Figure 4b). But in male CB 2 −/− mice, an increased glucagon level was measured when compared to the Wt, whereas in female knockout mice, a decrease was evident ( Figure A2b). Overall, higher somatostatin protein amounts were detected without reaching the level of significance in CB 1 −/− mice ( Figure 4c). Interestingly, this trend was seen in female CB 1 −/− mice only ( Figure A2c).

Changes in Key Components of the Glucose Sensing Machinery
To provide possible explanations for changes in blood glucose levels in cannabinoid receptor knockout mice, we investigated the expression of glucose transporter Glut1 (encoded by Slc2a1) and Glut2 (encoded by Slc2a2), as well as the glucose sensor glucokinase (Gck) in pancreatic tissue.
Compared to Wt, the gene expression of Glut1 was significantly downregulated in CB1 −/− and CB2 −/− mice (a decrease of 47% in CB1 −/− and of 71% in CB2 −/− , Figure 5a). The same pattern was reflected in female and male groups ( Figure A1e). In accordance, protein levels were lower in both knockout mice, but showed significant changes in CB2 −/− only (a reduction of 30% in CB1 −/− , p = 0.0853 and of 41% in CB2 −/− , Figure 5b). Nonsignificant decreases were observed in male and female groups ( Figure A3a). In immunohistochemistry, Glut1 was presented in the pancreatic islet with enhanced immunolabelling in the centre of the islet mainly occupied by beta-cells. Colabelling of Glut1 with glucagon revealed a colocalisation of Glut1 in alpha-cells as well, albeit with weaker staining ( Figure  5c). The reduced immunolabelling of Glut1 was evident in both knockout mice.

Changes in Key Components of the Glucose Sensing Machinery
To provide possible explanations for changes in blood glucose levels in cannabinoid receptor knockout mice, we investigated the expression of glucose transporter Glut1 (encoded by Slc2a1) and Glut2 (encoded by Slc2a2), as well as the glucose sensor glucokinase (Gck) in pancreatic tissue.
Compared to Wt, the gene expression of Glut1 was significantly downregulated in CB 1 −/− and Figure 5a). The same pattern was reflected in female and male groups ( Figure A1e). In accordance, protein levels were lower in both knockout mice, but showed significant changes in CB 2 −/− only (a reduction of 30% in CB 1 −/− , p = 0.0853 and of 41% in CB 2 −/− , Figure 5b). Nonsignificant decreases were observed in male and female groups ( Figure A3a). In immunohistochemistry, Glut1 was presented in the pancreatic islet with enhanced immunolabelling in the centre of the islet mainly occupied by beta-cells. Colabelling of Glut1 with glucagon revealed a colocalisation of Glut1 in alpha-cells as well, albeit with weaker staining (Figure 5c). The reduced immunolabelling of Glut1 was evident in both knockout mice. and Wt values were defined as 100%. * p < 0.05; *** p < 0.001; unpaired t-test. Scale bar 20 µ m.
Glut2 transcript levels were, although not statistically significant, decreased in knockout mice (Figure 6a). A nonsignificant reduction of Glut2 transcripts was seen in both male and female mice ( Figure A1f). Considering Glut2 protein, a significant decrease was measured for CB1 −/− and CB2 −/− mice (Figure 6b). An overall significant decrease of protein was seen in both sexes ( Figure A3b). Immunohistochemistry revealed the labelling of Glut2 to be predominant in pancreatic islets and, more specifically, in pancreatic beta-cells restricted to the cell membrane (Figure 6c). Knockout mice yielded weaker immunostaining of the membranes, but increased cytoplasmic labelling, as seen by small dots in the cytoplasm. In CB2 −/− , this altered pattern was more accentuated, showing more accumulations in the cytoplasm and a modified appearance of the membrane structure. Glut2 transcript levels were, although not statistically significant, decreased in knockout mice (Figure 6a). A nonsignificant reduction of Glut2 transcripts was seen in both male and female mice ( Figure A1f). Considering Glut2 protein, a significant decrease was measured for CB 1 −/− and CB 2 −/− mice (Figure 6b). An overall significant decrease of protein was seen in both sexes ( Figure A3b). Immunohistochemistry revealed the labelling of Glut2 to be predominant in pancreatic islets and, more specifically, in pancreatic beta-cells restricted to the cell membrane (Figure 6c). Knockout mice yielded weaker immunostaining of the membranes, but increased cytoplasmic labelling, as seen by small dots in the cytoplasm. In CB 2 −/− , this altered pattern was more accentuated, showing more accumulations in the cytoplasm and a modified appearance of the membrane structure. Gck transcript levels were significantly increased in both CB1 −/− and CB2 −/− compared to Wt mice (Figure 7a). This increase seemed to be sex-independent ( Figure A1g). However, protein analysis revealed a significant increase of Gck protein in CB1 −/− only (Figure 7b), which was restricted to female CB1 −/− mice ( Figure A3c). Gck immunolabelling in pancreatic islets of Wt mice displayed a heterogeneous cytoplasmic staining, where it appears to surround the nuclei in a ring-like manner (Figure 7c). CB1 −/− displayed a heterogeneous distribution pattern of Gck labelling with accumulations Gck transcript levels were significantly increased in both CB 1 −/− and CB 2 −/− compared to Wt mice ( Figure 7a). This increase seemed to be sex-independent ( Figure A1g). However, protein analysis revealed a significant increase of Gck protein in CB 1 −/− only (Figure 7b), which was restricted to female CB 1 −/− mice ( Figure A3c). Gck immunolabelling in pancreatic islets of Wt mice displayed a heterogeneous cytoplasmic staining, where it appears to surround the nuclei in a ring-like manner ( Figure 7c). CB 1 −/− displayed a heterogeneous distribution pattern of Gck labelling with accumulations around the nuclei with areas of high and low density in the cytoplasmic region. In CB 2 −/− , the heterogeneous staining of Gck was even more prominent and seemed to be reinforced by distinct accumulations in the perinuclear region, displaying a semilunar appearance (Figure 7c). around the nuclei with areas of high and low density in the cytoplasmic region. In CB2 −/− , the heterogeneous staining of Gck was even more prominent and seemed to be reinforced by distinct accumulations in the perinuclear region, displaying a semilunar appearance (Figure 7c).

Discussion
The present work focuses on the impact of deletion of cannabinoid receptor types CB 1 or CB 2 on pancreatic islet hormone secretion and glucose metabolism. Until now, the expression and distribution of both cannabinoid receptors in pancreas was controversial in mice, showing the presence of CB 1 mainly on nonbeta-cells including alpha-or delta-cells [26][27][28]. However, CB 1 and CB 2 receptors were also found in islet beta-cells [10,11]. Another group confirmed the presence of CB 1 but neglected that of CB 2 in mouse islets [29]. In accordance with studies of Li and colleagues [10,11], we detected mRNA of both CB 1 and CB 2 in mouse pancreatic islets. CB 1 immunolabelling was restricted to pancreatic islets, displaying a stronger signal in beta-cells and a weaker signal in alpha-cells.
Knockout of CB 1 led to decreased body weight in mice without any notable impact on islet hormones in plasma and pancreas. Similar to our findings, lower body weights were reported in CB 1 -/mice compared to CB 1 +/+ mice, whereas plasma insulin and glucose levels were not different between strains [30]. In contrast, another study described an upregulation of insulin gene expression in CB 1 -/mice and an increased level of pro-insulin compared to those of age-matched CB 1 +/+ mice [21]. As shown in rodent models in general, pharmacological inhibition of the CB 1 downregulated food intake and body weight [31]. In line with these findings, anti-obesity effects were reported as a result of chronic CB 1 receptor antagonist treatment [32] or for whole-body CB 1 knockout mice [15,30]. Knockout of CB 1 led to decreased body weight in mice without any notable impact on islet hormones in plasma and pancreas. Similar to our findings, lower body weights were reported in CB 1 −/− mice compared to CB 1 +/+ mice, whereas plasma insulin and glucose levels were not different between strains [30]. In contrast, another study described an upregulation of insulin gene expression in CB 1 −/− mice and an increased level of pro-insulin compared to those of age-matched CB 1 +/+ mice [21].
As shown in rodent models in general, pharmacological inhibition of the CB 1 downregulated food intake and body weight [31]. In line with these findings, anti-obesity effects were reported as a result of chronic CB 1 receptor antagonist treatment [32] or for whole-body CB 1 knockout mice [15,30]. Interestingly, the generation of a cell-specific male CB 1 R knockout mouse model (beta-CB 1 R −/− ) allowed us to determine the role of CB 1 specifically in beta-cells of CB 1 −/− mice [6]. Contrary to our findings, beta-CB 1 R −/− mice showed no changes of body weight and a significant increase of fasting plasma insulin levels. Similar to the previous study [6], a lowered fasting blood glucose level was also observed in the male CB 1 −/− mice in our study. In view of these findings, our knockout data must be considered in the context of a whole-body metabolism, taking into consideration the well-established interplay between pancreas and liver [33]. In this context, the liver is known to express CB 1 and CB 2 receptors [34]. The CB 2 −/− mice showed decreased plasma insulin levels combined with increased blood glucose levels in both sexes, and increased body weights in male mice. Another study reported in two-month-old male CB 2 −/− mice similar body weight values to those of normoglycaemic, age-matched, wild-type mice [18]. Increased body weight of CB 2 −/− mice was measured in other studies [19,35], which is in accordance with our data considering male CB 2 −/− mice only.
Sex-specific differences were measured for serum glucagon levels in our study, displaying higher levels in female CB 1 −/− mice and, in general, higher levels in female knockout mice compared to male knockout mice. This strongly supports a sex-specific impact of receptor loss, which has not been considered so far, since most studies used male rodents. However, there is an increasing number of studies suggesting a sexually dimorphic function of the CB 1 receptor [36], indicating sex-specific differences within the cannabinoid-regulated biology [37]. Beyond sex-specific characteristics in cannabinoid receptor pharmacology, there is growing preclinical evidence for an impact of gonadal hormones, particularly estradiol, on cannabinoid receptor density or function [38], providing a possible explanation for our sex-specific findings.
In feeding mouse models, CB 2 receptor deletion was associated with impaired glucose clearance [19], supporting our findings of elevated blood glucose and hormonal differences in CB 2 −/− . Using diet-induced obesity mice, CB 2 receptor agonists were shown to be efficacious in reducing body weight and obesity-associated metabolic parameter, e.g., insulin. Furthermore, acute administration of CB 2 -agonist JWH-015 produced a significant improvement in glucose clearance [39]. We propose that the deletion of CB 1 or CB 2 in our study resulted in an impairment of glucose metabolism, since key components of the glucose sensing machinery, such as glucose transporters and glucokinase, were affected in both knockout mouse lines. Sand rats represent a well-established model of nutritionally-induced, noninsulin-dependent, type 2 diabetes. The loss of immunostaining for the Glut2 glucose transporter in the plasma membrane of the pancreatic beta-cells became evident when these animals subsequently developed hyperglycaemia [40]. Furthermore, in type 2 diabetic Goto-Kakizaki rats, which were characterised by changes in blood glucose and insulin, Glut2 was reduced, and changes in its distribution patterns were evident [41]. Accordingly, glucose transporter 2 was reduced in both knockout mouse lines in the present study, although in CB 2 −/− the reduction seemed to be reinforced by cytoplasmic accumulations of Glut2 immunolabelling, indicating a reduced number of membrane-bound functional transporters. In diabetic rodents, impaired glucose-stimulated insulin secretion was associated with a markedly reduced expression of Glut2, regulated on both the mRNA and protein levels [23], thereby supporting the well-accepted fact that Glut2 is required in order to maintain normal glucose homeostasis and normal endocrine pancreas function. Notably, in renal tissue, the Glut2 expression and translocation changes are under the regulation of the CB 1 receptor [42]. Thus, CB 1 affected its dynamic membrane translocation and modulated glucose reabsorption. High glucose and ACEA (a CB 1 agonist) treatment caused an increase of perinuclear translocation of Glut2 in renal MDCK II cells, whereas under CB 1 antagonism with JD5037, a decrease of perinuclear Glut2 was demonstrated. Inhibition of CB 1 also downregulated the Glut2 expression in renal cells [42]. The aforementioned processes could play a role in our CB 1 −/− and perhaps CB 2 −/− mice with lower levels of glucose transporters incorporated into the cell membrane. Thus, we hypothesise that the genetic knockout of CB 1 , but also CB 2 , has the potential to influence the trafficking of glucose in beta-cells by affecting the glucose transporters Glut1, and especially Glut2 at transcriptional and protein levels, as well as their translocation. Glut1 is a high affinity and low K m transporter responsible for glucose influx into beta-cells which becomes activated at low glucose concentrations, allowing rapid equilibration to occur of extra-and intra-cellular glucose. Glut1 was also affected after CB 1 or CB 2 loss. In this context, a direct functional link between the transporters responsible for glucose uptake and the capacity of beta-cells to increase insulin secretion was reported [43]. However, the detailed underlying mechanisms in association with cannabinoid receptors have been investigated neither in rodent nor in human beta-cells. CB 1 was shown to regulate the Glut2 expression, its membrane translocation and activity by a signalling mechanism that involves elevating cytosolic Ca 2+ levels and activating the upstream modulator of Glut2, protein kinase C-β1 [42,44]. Glucokinase plays a critical role in glucose homeostasis, and is described as a candidate diabetes mellitus gene [45,46]. In type 2 diabetic rodent models, a decrease of pancreatic glucokinase expression was paralleled by a reduced staining in the pancreatic islet [40,41], which is contrary to our observations. Glucose functions as a modulator of the pancreatic glucokinase, directly affecting levels in beta-cells, and stimulates insulin secretion [45]. Thus, the hyperglycaemia that occurs in CB 2 −/− mice may be responsible for the higher level of glucokinase in the pancreatic islets, which might be a way to compensate for lower plasma insulin levels. Nonetheless, this does not give a plausible explanation for changes seen in CB 1 −/− mice, as their blood glucose levels were comparable to those of Wt mice.
So besides glucose, insulin regulates beta-cell glucokinase expression [47]. In accordance with this, a recent study investigated an interaction between the CB 1 and the insulin receptor, and showed an increased expression of glucokinase in mouse beta-cells after silencing the CB 1 receptor, an effect which was lost when the insulin receptor was missing. Furthermore, CB 1 −/− islets in pancreata from overnight-fasted CB 1 −/− mice displayed increased glucokinase expression compared to islets from CB 1 +/+ mice [21].
In summary, the present study underscores the importance of CB 1 and CB 2 signalling for pancreatic islet cell function with underlying different roles of CB 1 and CB 2 , which is supported by [48]. CB 1 or CB 2 receptor knockout resulted in alterations in the glucose sensing recognition apparatus, including glucose transporters and glucokinase, which could explain metabolic changes. In beta-cells, glucokinase is the first enzyme phosphorylating glucose, and is thus the primary determinant for the glycolytic flux rate [22]. Surprisingly, our CB 2 −/− mice revealed the most alterations, although it is known that CB 2 is expressed to a much lower extent (100-fold difference) than CB 1 [6,8,27,49]. Notably, the altered glucose sensing in pancreatic beta-cells, combined with lower insulin levels and higher blood glucose in CB 2 −/− mice, should be taken into consideration in view of the development of pharmacological cannabinoid receptor antagonists in diabetes therapy. Otherwise, it should be remembered that only one of the two receptors in our mice was deleted, while the other was functionally active. Hence, we cannot exclude a compensatory mechanism of the other cannabinoid receptor type. So, further studies will be needed to clarify different roles of CB 1 and CB 2 signalling and their mechanism in pancreatic islet cells. 12 female, 11 male) mice and their wild-type littermates (CB 1 +/+ , Wt; n = 11: 4 female, 7 male) were bred as previously described [12,14]. The mice, taken from our own breeding colony, had a C57BL/6N background. All animals were housed in a temperature-and humidity-controlled vivarium with a 12-hr light-dark cycle (L:D = 12:12, light on: 06:00 a. m.), and had access to food and water ad libitum, feeding on a standard diet. Mice at 10-12 weeks of age were used. Food was taken away 3 h before mice were killed under deep anaesthesia during the light period (10:00-14:00 a.m.). Body weight was determined and blood glucose levels from tail-tip samples were analysed with a measuring device (MediSense Precision, Wiesbaden, Germany). Blood samples were acquired by heart ventricle puncture. After centrifugation, the supernatant was stored at −80 • C. Pieces of organs were stored at −80 • C. For gene expression analysis, pieces of pancreata were immediately preserved in RNAlater (Ambion Inc., Austin, TX, USA). Stored probes of alphaTC1.9 cells were used for RT-PCR.

Animals and Tissue Sampling
Isolated islets were prepared for qualitative mRNA analysis. Briefly, the mouse was killed by cervical dislocation, and the pancreas was perfused by injection of 3 mL of Collagenase-P (1 mg/mL; Roche, Mannheim, Germany) in Hank's buffered salt solution (HBSS) containing 25 mM HEPES and 0.5% (w/v) BSA into the common bile duct. Subsequently, the perfused pancreas was digested in 2 mL of collagenase solution for 9-10 min at 37 • C. With the help of a cannula (18G × 11 2), islets were mechanically detached from exocrine parts and washed for several times with HBSS. Finally, islets were purified by hand picking in RPMI 1640 supplemented with 10% FCS, 100 U/mL penicillin and 100 µg/mL streptomycin. Total islet RNA was extracted with TRIzol.

RNA Extraction, DNAse Treatment and Real-Time RT-PCR
Total RNA was isolated using a standard protocol for TRIzol extraction (TRI Reagent ® , Sigma-Aldrich GmbH, Taufkirchen, Germany). DNAse treatment (DNA-free™; Ambion Inc. Austin, TX, USA), and reverse transcription (Promega Inc., Madison, WI, USA) was carried out as indicated according to the manufacture's protocols.
Real-time RT-PCR was carried out using 7900HT Fast Real-Time PCR system (Applied Biosystems, via Thermo Fisher Scientific, Karlsruhe, Germany). Primer sequences are listed in Supplementary  Table S1. For quantification of the relative expression levels of target genes, the ∆∆-C t method [50] was used. The expression of beta-actin (Actb) was used to normalise the target genes. The identity of amplicons was verified by restriction analysis and agarose gel electrophoresis.

Confocal laser Scanning Microscopy
Fluorescence-labelled pancreatic tissues from three Wt, CB 1 −/− and CB 2 −/− mice of both sexes were analysed by confocal laser scanning microscopy (Leica TCS SPE, Wetzlar, Germany). Viewing the whole tissue, 6-17 islets per mouse were randomly scanned. Images consisting of 2048 × 2048 pixel were recorded using a 40× oil-immersion objective and a zoom factor of 1.5. CLSM images consisted of stacks of three consecutive virtual sections which were combined in a single image using the maximum projection of the image analysis software Fiji. Borders of glucagon containing alpha-cells were determined by using an Auto Local Threshold and a Gaussian Blur filter. By an automated operator, the staining was detected in the converted binary image and defined as region of interest (ROI).

Measurements of Plasma Insulin and Glucagon
Insulin and glucagon concentrations were measured in mouse plasma using a Mouse Ultrasensitive Insulin ELISA (ALPCO, Salem, NH, USA) and a Mouse Glucagon ELISA (Crystal Chem, Zaandam, Netherlands) according to the manufacturer's instructions.

Statistical Analysis
For statistical analyses, CB 1 −/− or CB 2 −/− mice were compared to Wt mice (including sex differences) using an unpaired t-test (Prism 8, GraphPad Software Inc., San Diego, CA, USA). Unpaired t-test was chosen to compare the mean values obtained in Wt and knockout mice, as they showed differences in  (± S.E.M.) with n = 4-12 mice per group defining Wt values as 100%. * p < 0.05; *** p < 0.001 for comparisons between male or female Wt and knockout mice; unpaired t-test.