The endocannabinoid system (ECS) has recently emerged as an important modulator of energy balance (1). Indeed, recent evidence indicates that the ECS modulates several physiologic functions through central and peripheral mechanisms, and that dysregulation of the ECS is linked to abdominal obesity and other risk factors for cardiovascular disease and type 2 diabetes (2). The cannabinoid receptor type 1 (CB1) is widely expressed in the brain, including brain areas associated with the regulation of energy homeostasis (1), and is also present in peripheral tissues, including liver, pancreas, muscle, and adipose tissue (1,2). Genetic or pharmacological reduction of CB1 activity decreases body weight and fat mass to a higher extent than that expected from the reduction of caloric intake alone (1,2). However, whether the metabolic amelioration associated with the administration of a CB1 antagonist is dependent on its appetite suppressant effect is still a matter of debate. Some pair-feeding studies carried out in rodents have revealed that most of the metabolic effects due to chronic (>2 weeks) pharmacological blockade of CB1 or CB1 deficiency cannot solely be explained by a reduction in food intake (3,4,5). However, other studies where metabolic parameters were evaluated when a significant drug-induced anorexia was still present, have reported food-intake dependent metabolic changes (6,7,8,9). In human studies, the CB1 antagonist rimonabant improves dyslipidemia and glycemia parameters to a greater extent than what would be expected by body weight and fat mass reduction (10,11). The mechanisms underlying these effects are still unknown. In order to determine the long-term beneficial effects of CB1 antagonism, we evaluated the overall metabolic action of 4 weeks of peripheral administration of rimonabant in diet-induced obese (DIO) rats.
Overactivity of the endocannabinoid system (ECS) has been linked to abdominal obesity and other risk factors for cardiovascular disease and type 2 diabetes. Conversely, administration of cannabinoid receptor type 1 (CB1) antagonists reduces adiposity in obese animals and humans. This effect is only in part secondary to the anorectic action of CB1 agonists. In order to assess the actions of CB1 antagonism on glucose homeostasis, diet-induced obese (DIO) rats received the CB1 antagonist rimonabant (10 mg/kg, intraperitoneally (IP)) or its vehicle for 4 weeks, or were pair-fed to the rimonabant-treated group for the same length of time. Rimonabant treatment transiently reduced food intake, while inducing body weight loss throughout the study. Rats receiving rimonabant had significantly less body fat and circulating leptin compared to both vehicle and pair-fed groups. Rimonabant, but not pair-feeding, also significantly decreased circulating nonesterified fatty acid (NEFA) and triacylglycerol (TG) levels, and reduced TG content in oxidative skeletal muscle. Although no effects were observed during a glucose tolerance test (GTT), rimonabant restored insulin sensitivity to that of chow-fed, lean controls during an insulin tolerance test (ITT). Conversely, a single dose of rimonabant to DIO rats had no acute effect on insulin sensitivity. These findings suggest that in diet-induced obesity, chronic CB1 antagonism causes weight loss and improves insulin sensitivity by diverting lipids from storage toward utilization. These effects are independent of the anorectic action of the drug.
Methods and Procedures
Male Long–Evans rats were housed individually and maintained on a 12-h light/dark cycle (1:00 am/1:00 pm). They had ad libitum access to water and a high-fat butter diet (HF; Research Diets, New Brunswick, NJ, 4.54 kcal/g; 40% butter; for complete diet composition, see also ref. 12). After 8 weeks on HF diet, the animals were randomized into three groups: rimonabant-treated (RIMO, n = 8), vehicle-treated (VEH, n = 7), and pair-fed (PF, n = 7) by matching body weights, baseline glucose levels, and response to a glucose tolerance test (GTT; see Supplementary Figure S1 online). Pair-fed animals received a daily food amount equal to the mean amount of food consumed by the rimonabant group during the previous 24 h. Body weight and food intake were recorded daily. After 4 weeks of treatment, overnight fasted rats were killed. Tissues and blood were collected for further analysis. All animal protocols were approved by the University of Cincinnati Institutional Animal Care and Use Committee.
Rimonabant (SR141716, obtained through the NIMH Chemical Synthesis and Drug Supply Program) was administered daily, intraperitoneally (IP), at a dose of 10 mg/kg, just before the onset of dark. Rimonabant was dissolved in saline with 0.3% of Tween 80 (Sigma-Aldrich, St Louis, MO); the obtained solution was stirred and then sonicated. For the acute insulin tolerance test (ITT), drug or vehicle was administered once, 45 min before the start of the experiment.
Body composition was assessed by nuclear magnetic resonance using an EchoMRI analyzer (EchoMedical Systems, Houston, TX), right before the start of the treatment (day 0) and 17 days after the treatment. Body composition measurements obtained on day 17 were expressed as percentage of change from measurements obtained on day 0.
GTT and ITT
All rats received a GTT (IP injection of 50% dextrose at 2 g/kg) at baseline and on day 18 of the study. On day 22, all rats received an ITT (IP injection of insulin at 0.5 U/kg). The night before the tests, food was withdrawn at 6 pm. The tests were conducted at 9 am the following morning. Blood samples were taken from tail vein at times 0, 15, 30, 45, 60, and 120 min and at 0, 15, 30, and 60 min after the IP injections for the GTT and ITT, respectively. An ITT (0.5 U/kg insulin, IP) was also carried out in eight age-matched, chow-fed, lean rats (300–350 g). Finally, to test the acute effects of a single injection of rimonabant on insulin sensitivity of DIO rats, an additional group of 15 DIO rats received an ITT (0.5 U/kg insulin, IP) 45 min after the administration of a single IP injection of rimonabant or its vehicle.
Plasma and tissue analysis
Blood glucose was measured using a Freestyle (Therasense, Almeda, CA) glucometer. Insulin was measured by radioimmunoassay, as described (13). Plasma leptin and total adiponectin levels from trunk blood were measured with RIA kits (Linco Research, St Charles, MO). Nonesterified fatty acid (NEFA) levels were measured in 5 µl of plasma, using a HR Series NEFA-HR(2) kit (Wako Chemicals USA, Richmond, VA), according to the manufacturer's instructions. Plasma and tissue triacylglycerols (TGs) were measured with the Infinity Triglycerides kit (Thermo Scientific, Waltham, MA). Total plasma cholesterol was measured with Infinity Cholesterol kit (Thermo Scientific). Gene expression was assessed by quantitative reverse transcription–PCR (see Supplementary Methods and Procedures online).
All values are reported as means ± s.e.m. The data were analyzed by one-way or two-way ANOVA, followed by Fisher LSD post hoc tests. P values <0.05 denote statistical significance.
Pharmacological blockade of CB1 reduces body fat independently of its anorectic action
As previously reported (14), obese rats receiving rimonabant had a significant but transient reduction in food intake compared to controls during the first 2 weeks of treatment (Figure 1a). Rimonabant and pair-fed animals had similar body weight loss (Figure 1b). Although lean body mass did not differ among groups (% change of lean mass after 17 days of treatment: VEH, 102.9% ± 0.7; RIMO, 99.4% ± 3.6%; PF, 100.9 ± 1.1%, P > 0.05), rats receiving rimonabant had significantly less body fat compared to both vehicle and pair-fed groups (% change of fat mass after 17 days of treatment: VEH, 102.7% ± 2.5; RIMO, 72.7% ± 4.8%; PF, 85.5 ± 2.9%, RIMO vs. VEH, P < 0.001; RIMO vs. PF, P < 0.05; PF vs. VEH, P < 0.05, respectively). In agreement with their lower fat mass, rimonabant-treated rats had significantly lower plasma leptin compared to both pair-fed and vehicle-treated animals (VEH, 13.0 ± 2.2 ng/ml; RIMO, 6.3 ± 1.0 ng/ml; and PF, 10.6 ± 2.2 ng/ml, RIMO vs. VEH and RIMO vs. PF, P < 0.05). Plasma adiponectin levels did not change among groups (VEH, 6913 ± 795 pg/ml; RIMO, 7851 ± 535 pg/ml; and PF, 7524 ± 649 pg/ml, P > 0.05). Consistent with the lack of anorectic effect at the end of the study, mRNA levels of hypothalamic POMC, NPY, or AgRP were not different (data not shown).
Pharmacological blockade of CB1 decreases plasma and tissue lipid content
Rimonabant treatment significantly reduced plasma TG levels compared to both pair-fed and vehicle-treated rats (Table 1, RIMO vs. VEH and RIMO vs. PF, P < 0.05). Rimonabant also reduced plasma NEFA compared to the other two groups (Table 1, RIMO vs. VEH, P = 0.08; RIMO vs. PF, P = 0.06), whereas no changes were found in plasma total cholesterol levels (P > 0.05). TG content was significantly reduced in the soleus of rimonabant-treated rats as compared to pair-fed animals (Table 1). Moreover, TG content in the EDL of both rimonabant-treated and pair-fed animals was significantly lower as compared to vehicle-treated animals (Table 1). By contrast, TG levels within the liver were not different. We did not detect significant changes in the mRNA levels of several genes of lipid and glucose metabolism (see Supplementary Methods and Procedures online), except for a significant decrease in hepatic stearoyl CoA desaturase-1 expression in rimonabant-treated rats, as compared to vehicle-treated animals (Supplementary Figure S2 online).
Chronic blockade of CB1 increases insulin sensitivity
By design, all groups had similar fasting glucose levels (VEH, 5.9 ± 0.2 mmol/l; RIMO, 6.0 ± 0.2 mmol/l; and PF, 5.9 ± 0.3 mmol/l, P > 0.05), and glucose tolerance (Supplementary Figure S1 online) right before the start of the study. After 18 days of treatment, the rimonabant group had significantly lower fasting glucose than the vehicle group (RIMO, 5.6 ± 0.2 mmol/l vs. VEH, 6.3 ± 0.2 mmol/l; P < 0.05) but comparable fasting glucose as the pair-fed animals (PF, 5.8 ± 0.2 mmol/l; P > 0.05). A GTT revealed no significant differences among the three groups, neither in the glucose excursion nor in the glucose incremental area under the curve (iAUC) (Figure 1c). Similarly, insulin levels were not changed among groups (Figure 1d). Thus, we performed an ITT to determine the effect of chronic CB1 blockade on insulin sensitivity, and compared it to the ITT of age-matched, chow-fed, lean rats (Figure 1e). The effect of insulin on glycemia was similar in DIO vehicle and pair-fed groups. By contrast, the glucose-lowering effect of insulin was greater in rimonabant-treated rats and comparable to the insulin's action in lean, chow-fed rats (Figure 1e). During the ITT, the rate of glucose disappearance (Kd, calculated as the slope from time 0 to 60) was significantly greater in the rimonabant group compared to both vehicle and pair-fed rats (Figure 1f, P < 0.05), and was similar to the one obtained in chow-fed, lean rats (Figure 1f). To evaluate whether a single dose of rimonabant can acutely alter insulin sensitivity in DIO rats, we performed an ITT in DIO rats that received the CB1 antagonist (or vehicle) once, 45 min before the start of the ITT. However, we did not detect any difference in glucose levels (Supplementary Figure S3 online), suggesting that the CB1 antagonist does not acutely improve insulin action.
Both animal studies and clinical trials have highlighted the ability of CB1 antagonists to increase insulin sensitivity, when these compounds are administered chronically (1,2,10,11). Such metabolic amelioration is, in part the direct result of the ability of this class of compounds to reduce caloric intake and fat mass. In agreement with previous reports (reviewed in ref. 2), we have confirmed the anorectic and weight-reducing effect of rimonabant, and found that rimonabant-treated rats have reduced body fat and circulating leptin levels relative to pair-fed controls. This suggests that the anorectic action of CB1 antagonism only partially accounts for the loss of fat mass. Indeed, the anorectic effect of rimonabant is transient (7–15 days), whereas its effect on fat mass loss is persistent for the duration of the treatment. In this regard, and in agreement with other studies (14), we noticed that an overnight fast (a standard step in the correct performance of GTT and ITT) induced a greater, although nonsignificant, weight loss in the rimonabant group (Figure 1b). This is consistent with recent reports describing an increase in energy expenditure caused by the administration of CB1 antagonists (3,15,16). Our findings also confirm that pharmacological blockade of CB1 improves the plasma lipid profile (4,10,17). Furthermore, we demonstrate that these effects are not secondary to the anorectic action of the drug, as calorie restriction by pair-feeding did not change TG levels. Interestingly, rimonabant treatment lowered TG content in both the soleus and EDL muscles, oxidative and glycolytic type muscles, respectively. In particular, we found that pair-feeding (calorie restriction) favors decreased TG content in the EDL but not in the soleus. This raises the possibility that CB1 antagonism might regulate lipid metabolism in oxidative and glycolytic fibers via different mechanisms. However, 4 weeks of rimonabant administration did not decrease hepatic TG content in DIO rats. Similarly, a recent report has described a rimonabant-induced decrease in intramyocellular lipids associated to a lack of modulation of hepatocellular lipid content. These effects were independent of the anorectic action of the compound (4). Other studies, conducted with higher doses and longer duration of treatment, have reported decreased hepatic lipid content by histological analysis (18). Nonetheless, in our study, rimonabant decreased the hepatic expression of stearoyl CoA desaturase-1, a pivotal enzyme in the regulation of hepatic lipogenesis, whose activity is inversely correlated with insulin sensitivity (19).
Our study also demonstrates that a 4-week long treatment with rimonabant improves insulin sensitivity, independent of the appetite-suppressing effect of the drug (Figure 1e,f). Despite the improved ITT, rimonabant did not improve GTT (Figure 1c), suggesting that CB1 antagonism might ameliorate only specific components of the glucose-stimulated insulin response. Nonetheless, we show that rimonabant treatment restores insulin-mediated glucose-lowering action to that of lean, chow-fed rats (Figure 1e,f).
Recent reports have suggested that the effects of CB1 antagonism on insulin sensitivity might be mediated by the reduction of food intake (6,7). This study is the first to clearly demonstrate that the improved insulin-sensitivity does not depend on the drug-induced anorexia. This discrepancy, as previously mentioned, is in part explained by the different length of treatment characterizing those studies. Moreover, the adipose mass loss and the metabolic changes that we have described are in agreement with other recently published investigations in which such effects resulted to be independent from the anorectic action of rimonabant (3,4).
In our study, improved insulin sensitivity was associated with reduction in fat mass and muscle lipid content. Since, in obesity, insulin resistance is strongly linked to increased lipid deposition in skeletal muscle (20), we speculate that the effect of rimonabant to decrease TG content in muscle is directly implicated in the amelioration of insulin-dependent glucose disposal. Finally, previous investigations have suggested that CB1 antagonism improves glucose homeostasis by increasing the expression and circulating levels of adiponectin (reviewed in ref. 2). However, we and others (21,22) did not observe changes in adiponectin levels after pharmacological blockade of CB1.
In conclusion, we report that DIO rats chronically treated with a CB1 antagonist have a decrease in body fat, which is partly independent of the anorectic action of the drug. This effect is accompanied by a decrease in circulating and intramuscular lipid levels and improved insulin-stimulated glucose disposal. These findings therefore point to novel mechanisms through which the ECS might affect lipid and glucose metabolism of obese individuals.
Supplementary material is linked to the online version of the paper at http:www.nature.comoby
We thank Kay Ellis for help with the hormone assays. We also thank Joyce Sorrell, Kathi A. Blake Smith, Michael Segrist, and Mouhamadoul Toure for their technical assistance. This study was supported by NIH grant DK 17844 (to S.C.W.), NIH/NIDDK grant 5P01 DK 56863 (to R.J.S.), and DoD grant W81XWH-06-2-0019 (to R.J.S.), as well as ADA research award and NIH/NIDDK grant DK66058 (to S.O.).
D.C. and S.C.W. have been a consultant for and received honoraria from Sanofi-Aventis US.