Cannabinoid Type 1 Receptors Are Upregulated During Acute Activation of Brown Adipose Tissue

Brown adipose tissue (BAT) has emerged as a potential target to combat obesity and metabolic disease in humans. Activation of BAT is beneficial for human metabolism at a systemic level because it increases the metabolic rate (1) and is associated with increased lipid and glucose disposal (2–5). BAT can be activated by stimulating the sympathetic nervous system (SNS) with cold exposure or β3-adrenergic receptor (β3-AR) agonists (1,6,7). However, BAT function is also controlled by a number of other factors (8–10), which currently are poorly understood and largely unexplored.

The endocannabinoid system (ECS) and specifically the cannabinoid type 1 (CB1) receptors (CB1Rs) control lipid and glucose metabolism (11). The ECS consists of a network of receptors, their endogenous lipid ligands, and enzymes, which are significant in the brain and many peripheral tissues for modulating complex processes including metabolism. Activation of CB1R promotes the conservation of energy by increasing food intake and inhibiting energy expenditure and thermogenesis, leading to fat mass expansion (12). Conversely, the blockade of CB1R has been found to decrease body weight and fat mass, improve glucose homeostasis and insulin sensitivity, and decrease cardiometabolic risk factors, making CB1R antagonists potential drugs against obesity and diabetes (13). CB1R antagonist rimonabant (SR141716) was previously in clinical use with strong efficacy for weight loss, but was withdrawn because of serious psychiatric adverse effects (14). Recently, novel CB1R antagonists, which act strictly peripherally, have been found to activate BAT in rodents, inducing lipolysis and lipid oxidation, thus improving metabolism (15,16). Activation of BAT via CB1R antagonism without harmful centrally mediated adverse effects could be one way to improve metabolic disease and combat obesity in humans (17).

CB1R physiology and pathology can be studied in vivo using positron emission tomography (PET). An inverse agonist radioligand for CB1Rs, [¹⁸F]FMPEP-d₂, has previously been used to quantify the density of CB1R in the human brain (18–20). So far, this radioligand has not been used to study CB1R of other tissues in humans. Recently, we have shown in rodents that [¹⁸F]FMPEP-d₂ binds to BAT in vivo, indicating high CB1R density in BAT (21). There is also preclinical evidence that CB1R and endocannabinoids are upregulated in BAT after cold or β3-AR activation (22). Here we designed a clinical study aiming to investigate CB1R density in human brown fat using [¹⁸F]FMPEP-d₂-PET imaging in baseline and cold conditions. We also evaluated the effect of obesity on CB1R density in BAT and other tissues including the brain. We found that CB1Rs are upregulated when BAT is metabolically activated by cold, but this response is blunted in overweight subjects. CB1R mRNA expression in human BAT was confirmed from image-guided biopsies. In preclinical pharmacological experiments, CB1R antagonism was shown to increase glucose uptake and lipolysis of brown adipocytes (BAs).

The clinical PET study included 18 healthy males, who were divided into a lean group (n = 9) and an overweight group (n = 9), based on the combination of BMI (lean <25 kg/m²), waist circumference (lean <100 cm), and body fat percentage (lean <20%). The subjects were all Caucasian men, with an average age of 33 years (range 21–54 years). Subjects were determined to be healthy by means of clinical examination, blood tests, and anthropometric measurements (Tables 1 and 2). The study protocol was reviewed and approved by the Ethics Committee of the Hospital District of Southwest Finland and was conducted according to the principles of the Declaration of Helsinki. All study subjects provided written informed consent (Clinical trial reg. no. NCT02941172).

Each subject was studied on 3 separate days after an overnight fast of 8–10 h (Fig. 1A–D). To measure CB1R density, dynamic PET/computed tomography (CT) examinations were performed using the CB1R inverse agonist radioligand [¹⁸F]FMPEP-d₂, once in room temperature (RT) conditions and once during controlled cold exposure. To determine whether the subject had metabolically active BAT, a PET/magnetic resonance (MR) study using the glucose analog [¹⁸F]fluorodeoxyglucose (FDG) was performed during controlled cold exposure. The radioligands [¹⁸F]FMPEP-d₂ and [¹⁸F]FDG were synthesized according to standard operating procedures of the Turku PET Centre (Turku, Finland) as previously described (23,24).

Glucose uptake of BAT in the neck area was measured during cold exposure after a standardized 2-h cooling protocol with cooling blankets (25) (Supplementary Data). [¹⁸F]FDG 148 ± 13 MBq was injected i.v. and a 40-min dynamic PET scan of the cervical region was performed using a 3 T Philips Ingenuity TF PET/MR scanner (Philips Health Care, Amsterdam, the Netherlands). Eight consecutive modified two-point Dixon sequences (mDIXON scan, Philips Health Care) were used to provide anatomical reference in the whole body area and for calculating signal fat fraction maps (see Supplementary Data). Image analysis was performed using Carimas 2.9 software (Turku PET Centre). BAT regions of interest (ROIs) were identified in supraclavicular depots of adipose tissue, confirmed with MR signal fat fraction maps, and glucose uptake was quantified using the Patlak linearization model (26).

To measure CB1R density in different tissues, areas of the neck, abdomen, and brain were scanned using a PET/CT scanner (GE Discovery STE¹⁶; General Electric Medical Systems, Milwaukee, WI) once at RT and once during standardized cold exposure. [¹⁸F]FMPEP-d₂ 152 ± 12 MBq was injected i.v., and dynamic scans of the cervical region (60 min), abdominal area (9 min), and the brain (9 min) were conducted. CT scans of each region were performed for photon attenuation and anatomical reference. Carimas 2.9 software was used for image analysis of BAT, white adipose tissue (WAT), and muscle. ROIs of adipose tissue were manually drawn on the fused PET/CT images, including only voxels with CT Hounsfield units within the adipose tissue range (−50 to −250 Hounsfield units) (27). BAT ROIs were drawn bilaterally in supraclavicular adipose tissue depots, subcutaneous and intraperitoneal WAT ROIs in abdominal regions, and muscle ROIs in the trapezius muscle. Regional time-activity curves were calculated from the dynamic images. Details about data acquisition, PET image analysis, determination of plasma input, metabolite corrections, modeling of the data, and brain image analysis (using SPM12 software; Wellcome Trust Centre for Neuroimaging, London, U.K.) are available in the Supplementary Data. CB1R density of a tissue can be determined by calculating the volume distribution (V_T) or the fractional uptake rate (FUR) of [¹⁸F]FMPEP-d₂ from the dynamic PET images. Briefly, V_T was calculated from the supraclavicular BAT regions by applying the reversible one-tissue compartmental model, using the metabolite-corrected plasma time-activity curve as a model input (21). FUR was also calculated for BAT, WAT, muscle, and brain in order to quantitatively compare CB1R density between the tissues.

Sprague Dawley rats (n = 39, male, 8–10 weeks old, weight 228 ± 28 g) were bred at the animal facility of the University of Turku. All animal experiments were approved by the Regional State Administrative Agency for Southern Finland (ESAVI/3899/04.10.07/2013), and animal care complied with the principles of laboratory animal care and with guidelines of the International Council of Laboratory Animal Science. The animals were housed at 21 ± 3°C, in an atmosphere with 55 ± 15% humidity and with a light period from 6:00 a.m. to 6:00 p.m. All animals had free access to RM1 (E) chow (801002; Special Diets Service, Witham, U.K.) and tap water.

The rats were divided into three groups (n = 13), with each group being administered radioligand alone, radioligand after 10 min of i.v. preadministration of 2 mg/kg β3-AR agonist CL 316243 (Sigma-Aldrich), or 2 mg/kg CB1R antagonist rimonabant (Sigma-Aldrich). Briefly, CL 316243 was dissolved in 0.9% NaCl prior to injection. Rimonabant was dissolved in EtOH (150 µL) and then diluted in biocompatible polar solvent Kleptose (β-cyclodextrin; Apoteket Pharmacy, Uppsala University Hospital, Uppsala, Sweden) to a final EtOH concentration of 20% before injection.

Detailed imaging procedures, data acquisition, and analysis are described in the Supplementary Data. Briefly, each rat was sedated, and anesthesia was maintained throughout the imaging studies. Animals were positioned in the PET/CT scanner with the BAT in the center of the field of view. [¹⁸F]FMPEP-d₂ 9.7 ± 2.0 MBq (corresponding in all cases to <0.1 µg/kg) was administered i.v. in the tail vein (n = 24, or n = 8 from each treatment group). Each rat was then examined by PET/CT scanning over 120 min. V_T of [¹⁸F]FMPEP-d₂ in BAT in each rat was calculated in the PMOD kinetic modeling module (PKIN; PMOD Technologies, Zurich, Switzerland) using the one-tissue compartment model as previously described (21).

[¹⁸F]FDG 20.1 ± 1.6 MBq was administered i.v. in the tail vein in another group of rats (n = 15, or n = 5 from each treatment group). Each animal was then examined by PET/CT scanning over 60 min. The glucose utilization (in micromoles per 100 g per minute) in BAT was estimated by fitting the PET data to a [¹⁸F]FDG two-tissue compartment model, using a lumped constant of 1.3.

After the PET scanning, the animals were sacrificed (120 min after [¹⁸F]FMPEP-d₂ injection [n = 24] or 60 min after [¹⁸F]FDG injection [n = 15]). Tissues were excised and measured in a Wizard² Automatic Gamma Counter (PerkinElmer, Turku, Finland). Measured radioactivity was corrected for decay, weight of the organ, and background, and it was expressed as percentage of the injected dose/gram of tissue.

Nine of the 18 subjects who underwent PET imaging gave additional written consent for acquiring BAT biopsy samples from the supraclavicular neck region (five lean subjects, four overweight subjects). In sterile operating room conditions with an anesthesiologist monitoring the procedure, biopsy samples of BAT were taken by a plastic surgeon through one small skin incision, using local anesthesia. The anatomical location of BAT was predetermined with [¹⁸F]FDG PET/MR images. After removal, samples were immediately snap frozen into liquid nitrogen and stored at −70°C until analysis.

Human multipotent adipose-derived stem cells (hMADSs) were obtained from C. Dani (Université Côte d’Azur, CNRS, Inserm, iBV, Faculté de Médecine, Nice, France) and differentiated into BAs as described previously (28). Human white adipocytes (WAs) were obtained from PromoCell and differentiated according to manufacturer instructions.

Total RNA was isolated using TRIzol (human cells) or NucleoSpin RNA XS (human BAT samples; MACHERY-NAGEL). cDNA synthesis was performed with ProtoScript First Strand cDNA Synthesis Kit (New England BioLabs) according to manufacturer instructions. Human cannabinoid receptor mRNA expression was analyzed using an ABI 7900HT Fast Real-Time PCR System with SybrGreen (Roche), and expression was calculated as 2^-dCt relative to the TATA-box binding protein. PCR details are described in the Supplementary Data. mRNA of the CB1R was measured at baseline and after incubation with 1 μmol/L noradrenaline (NA) for 16 h.

To measure lipolysis, differentiated hMADS were washed twice with lipolysis medium (DMEM21603; Life Technologies) supplemented with 2% w/v fatty acid–free BSA (Sigma-Aldrich) followed by incubation with lipolysis medium containing 100 nmol/L each antagonist/inverse agonist (CB1 SR141716A; CB2 SR144528), each agonist (CB1 ACEA; CB2 JWH133), and/or 1 μmol/L NA at 37°C and 5% CO₂ for 4 h. Antagonists were added 20 min prior to NA treatment. All substances were purchased from Tocris Bioscience. Fifty microliters of cell culture media was collected per well and incubated with 50 µL of Free Glycerol Reagent (Sigma-Aldrich) for 5 min at 37°C. Absorption was measured at 540 nm. Glycerol release was calculated with glycerol standard (Sigma-Aldrich) and normalized to protein content. To measure glucose uptake, hMADSs were differentiated in 12-well plates, starved for 16 h, treated with indicated substances (see above) for 20 min, and glucose uptake assay (catalog #136955; Abcam) was performed according to manufacturer instructions. Antagonists were added 20 min prior to NA treatment. Experiments were performed with four independent cell cultures.

Results are presented as the mean ± SD. A paired two-tailed Student t test (α = 0.95) was used for assessing differences between baseline and cold conditions within a group. Unpaired two-tailed student t test (α = 0.95) was used for assessing differences between human lean and overweight groups, and differences between treatment groups of rats and for cell experiments. Pearson’s correlations were used to study associations between [¹⁸F]FMPEP-d₂ FUR in human brain and BAT, BAT glucose uptake, and uncoupling protein-1 (UCP1) expression. Statistical analyses were performed using IBM SPSS 23.0 software.

Compared with baseline conditions, acute cold exposure increased the FUR of the CB1R radioligand in supraclavicular BAT by threefold in lean subjects (P = 0.006) (Fig. 1E), indicating higher CB1R density. Interestingly, overweight subjects exhibited low CB1R density at baseline, which increased during cold exposure (P = 0.026) (Fig. 1E), but only reached the baseline levels of lean subjects. Importantly, the uptake of [¹⁸F]FMPEP-d₂ in BAT correlated strongly with the degree of functional BAT activity, as measured by its glucose uptake in cold (R = 0.89, P < 0.001) (Fig. 1F). Additionally, cold markedly increased energy expenditure in both lean and overweight subjects (Table 2). The V_T and FUR of [¹⁸F]FMPEP-d₂ are both indices of CB1R density in tissue. The BAT V_T correlated with the BAT FUR under RT conditions (R = 0.80, P < 0.001) and cold conditions (R = 0.73, P = 0.001); hence, the FUR is a suitable index for CB1R density when V_T is unavailable (for details, see Supplementary Data).

To verify our imaging findings, we analyzed image-guided BAT biopsy samples from nine study subjects, specifically studying BAT markers UCP1, β3-ARs, CB1R, and CB2R. We found that mRNA expression of CB1R was higher compared with that of CB2R (P = 0.039), but no significant difference was found between lean and overweight subjects (Fig. 1G). mRNA expression of UCP1 was significantly higher in lean than in overweight subjects (P = 0.02) (Fig. 1I). When biopsy data from lean (n = 5) and overweight (n = 4) subjects were pooled, UCP1 mRNA expression correlated with BAT [¹⁸F]FMPEP-d₂ uptake (R = 0.73, P = 0.027) (Fig. 1J) as well as glucose uptake under cold conditions (R = 0.76, P = 0.028) (Fig. 1K).

The central nervous system is essential in mediating cannabinoid signaling and BAT activation. When we quantified the CB1R density of the brain at RT, CB1R density was 23% lower in overweight than in lean subjects (Fig. 2A and C). Cooling increased CB1R density in the pooled group, (P = 0.033), specifically in the areas of the midbrain, pons, and parietal lobe. Furthermore, a positive correlation was found between [¹⁸F]FMPEP-d₂ uptake in BAT and brain gray matter in cold conditions (P < 0.0005) but not at baseline (Fig. 2D and E, midbrain region).

In overweight subjects, the uptake of [¹⁸F]FMPEP-d₂ in WAT both subcutaneously and intraperitoneally was significantly lower compared with that in lean subjects, whereas uptake in muscle was similar in both groups (Fig. 2A). Interestingly, in lean subjects, cold exposure induced a tendency of higher CB1R density in abdominal intraperitoneal fat (P = 0.07) (Fig. 2A).

To better understand CB1R physiology and the suitability of this tracer for BAT imaging, we conducted further PET studies in rats during pharmacological activation of BAT and blockade of CB1R. Similarly to humans, [¹⁸F]FMPEP-d₂ uptake markedly increased in interscapular BAT after i.v. administration of the β3-AR agonist CL 316243 (Fig. 3A). This clear increase persisted also when taking possible alterations of [¹⁸F]FMPEP-d₂ metabolism into account by kinetic modeling (Fig. 3B). The in vivo activation of BAT by CL 316243 was confirmed by increased glucose utilization (Fig. 3C), and BAT perfusion (Fig. 3D). Preadministration of the CB1R antagonist rimonabant inhibited the [¹⁸F]FMPEP-d₂ uptake in BAT (Fig. 3B) and brain at basal conditions (P < 0.0001), indicating that the uptake was receptor mediated rather than nonspecific. However, CB1R antagonism with an i.v. dose of 2 mg/kg did not significantly alter interscapular BAT glucose uptake or modulate perfusion (Fig. 3C and D).

Receptor mRNA expression analyzed in a human BA cell line (hMADS) and primary human WAs revealed significantly higher CB1R mRNA expression in BAs compared with WAs. NA stimulation significantly increased the mRNA expression of CB1R in BAs, but significantly decreased it in WAs. No changes were observed in CB2R mRNA expression with NA (Fig. 3E and F).

We further studied the effect of cannabinoid receptor agonism and antagonism on the function of human BA in vitro. Pharmacological blockade of CB1R increased glucose uptake, whereas CB1R stimulation, CB2R stimulation, and CB2R blockade had no effect (Fig. 3G). Activation of human BA with NA increased glucose uptake, and CB1R antagonism further enhanced this increase (Fig. 3G). Again, CB1R or CB2R agonism and CB2R antagonism combined with NA had no effect (data not shown). Lipolysis is another hallmark of BA activation, so we analyzed whether CB receptors can modulate glycerol release from human BAs. Blocking the CB1R significantly increased lipolysis, whereas CB1R stimulation had no effect (Fig. 3H). CB1R blockade after maximal stimulation of BA with NA increased lipolysis, albeit not significantly. Taken together, these data show that CB receptors are expressed in human BAs and that CB1R blockade can positively modulate BA function cell autonomously.

To our knowledge, this is the first study investigating CB1R expression and function in human BAT in vivo. Our imaging results show that acute cold exposure markedly increases [¹⁸F]FMPEP-d₂ binding in supraclavicular BAT depots of lean healthy men, indicating increased CB1R density when BAT is activated. CB1R mRNA expression in human BAT was also confirmed from samples obtained by image-guided biopsies. A physiological increase in receptor density suggests that the ECS plays a role in the activation of human BAT.

Our results are consistent with those of a previous study in mice showing that stimulation of BAT by cold and β3-AR agonism increased endocannabinoid levels in BAT (22). Moreover, the activation of primary BAs induced transcription of Cnr1, the gene encoding the CB1R (22). CB1R agonism promotes a positive energy balance (12), hence with their findings Krott et al. (22) hypothesized a negative feedback mechanism, where endocannabinoids, their enzymes, and receptors are upregulated during BAT activation as a potential autoregulatory loop to inhibit thermogenesis. In line with this, another recent study (29) measured increased plasma endocannabinoid levels of healthy lean men after mild acute cold exposure. Kantae et al. (29) did not find correlations between human BAT activity and plasma cannabinoids. However in this study, with our dynamic and highly sensitive PET imaging methods, we show a positive association between CB1R density and glucose uptake in human BAT during cold exposure. UCP1 mRNA expression measured from BAT biopsy samples was also positively associated with CB1R density and glucose uptake in BAT (n = 9, pooled lean and overweight subjects). An acute cold stress may stimulate the ECS to provide more CB1Rs for endocannabinoid binding in BAT in order to inhibit excess energy expenditure and return homeostasis toward a positive energy balance.

Endocannabinoids are produced on demand, acting primarily in the brain, but exerting important regulatory effects on metabolism in adipose tissue (30). During cold exposure, CB1Rs were also upregulated in the brain, specifically in the areas of the midbrain, pons, and parietal lobe, which are closely related to the sympathetic control of BAT function. Furthermore, CB1R density in the midbrain correlated positively with BAT CB1R density in cold, but not warm conditions. These results indicate a relationship among the ECS, the SNS, and BAT. The midbrain region includes the hypothalamus, which is one key site for controlling homeostasis and energy expenditure and is where endocannabinoids play a major regulatory role (11). The parietal lobe receives and processes sensory input, including temperature, whereas the pons is a significant signaling route controlling autonomic functions (31). Temperature is sensed in peripheral tissues, and information is received and processed in these areas of the central nervous system, after which efferent sympathetic outflow in the form of NA is increased to BAT, resulting in increased thermogenesis (6). Endocannabinoid signaling in the brain and in BAT seems to be upregulated acutely in cold to modulate a suitable thermogenic response.

In overweight subjects, CB1R density in BAT was low and the increase in cold was blunted, merely reaching the baseline values of lean subjects, possibly reflecting the generally reduced activity of BAT in overweight subjects. However, CB1R density was also significantly lower in abdominal WAT depots and in the brain compared with lean subjects. This suggests a broader downregulation of the CB1Rs, signifying the impaired regulation of the ECS in overweight subjects, which is in line with the results of previous studies. Excessive activation of the ECS is associated with obesity (12), and a negative association between CB1R density in the brain and BMI has been reported previously (19,20). Obese subjects have increased circulating endogenous cannabinoid levels, whereas mRNA expression of CB1Rs is lower in the WAT of obese subjects compared with lean subjects (32,33). Moreover, higher plasma endocannabinoid levels are related to increased abdominal adiposity and cardiometabolic risk factors (34,35). The CB1R antagonist rimonabant resulted in weight loss in obese patients (13), demonstrating that blocking the overactive ECS could improve metabolism. These findings exhibit the negative feedback loop of the ECS; chronically high amounts of circulating endocannabinoids in obesity are associated with fewer CB1Rs in brain and adipose tissue.

In addition to studying the physiological effects of cold on CB1R signaling in human BAT, we conducted pharmacological studies targeting the CB receptors in rodents and in a human BA system. Previous preclinical evidence suggests that CB1R blockade enhances BAT function. In mice, CB1R antagonists blocked the inhibition of β3-AR, leading to increased activation in BAT, lipolysis, and the uptake of fatty acids from plasma (36). Moreover, adipocyte-specific CB1R deletion results in the browning of WAT, the promotion of a thermogenic program, and an increase in alternatively activated macrophages, which increase local NA levels (37). In human BA, we showed that CB1R antagonism increased glucose uptake and lipolysis, but CB1R agonism, CB2R antagonism, or CB2R agonism did not have any effect. During increased NA availability, we also measured increased CB1R mRNA expression in BA, but not in WA. Our results add to the existing evidence that CB1Rs are significant in regulating BA function in an SNS-dependent manner.

Interestingly, in rats we did not measure any appreciable increase in glucose utilization in BAT up to 60 min after CB1R antagonist administration, which is in disagreement with previous findings (36,38,39). This discrepancy may be explained by differences in methodology, such as the acute i.v. dose given here. Longer lasting plasma exposure to the CB1R antagonist might be required for inducing a recordable increase in glucose utilization in BAT in rat. Unfortunately, the effect of CB1R antagonism in humans could not be investigated in this study for ethical reasons. No CB1R antagonist is currently in clinical use for humans and the subjects in this study already received the maximal acceptable annual radiation dose considered safe for healthy volunteers. Future studies need to be performed to investigate whether peripheral CB1R antagonism with novel compounds could activate BAT in humans.

One limitation of this study is that the data exhibit only short-term changes in the ECS and BAT, lacking the long-term effects. When combining our data and others (22,29), it seems that, acutely, BAT can produce endocannabinoids and modulate the density of CB1Rs available for binding them in order to adapt to changes in sympathetic tone. We speculate that during prolonged cold exposure, Cnr1 mRNA expression may increase, but some desensitization of ECS signaling may also occur, similar to long-term changes seen in obesity. Other long-term adaptations such as expansion of BAT volume and activity and browning of WAT may also affect the response. Cold acclimation studies in humans or repeated β3-AR stimulation experiments in rodents are required to understand possible chronic changes and adaptations of the ECS and BAT function.

The CB1R PET radioligand [¹⁸F]FMPEP-d₂ has previously been used for neuropsychiatric brain studies, and its use as a surrogate biomarker for BAT was previously reported in rats (21), but the potential of using it to study human BAT physiology has previously been unexplored. Here, in activated BAT of both humans and rats, we observed a strong increase in [¹⁸F]FMPEP-d₂ uptake. Binding was blocked by rimonabant, confirming a CB1R-mediated mechanism, whereas nonspecific or off-target binding is negligible. In rats, in addition to increased CB1R binding, we estimate that an increase in perfusion occurred based on the increase in glucose extraction rate. Increased perfusion of BAT is a consequence of increased oxidative metabolism (25,40), and whereas higher perfusion could also result in more radioligand delivery to the target tissue, perfusion alone could not explain the sixfold increase in [¹⁸F]FMPEP-d₂ binding. Furthermore, we observe active retention of the radioligand in BAT throughout the PET scan, in the form of increased V_T (an index of specific binding). Therefore, the current PET data can be explained by transiently increased expression of the CB1R, in response to cold or β3-AR–mediated activation in BAT.

In conclusion, acute adrenergic activation of BAT increases CB1R density in BAT in humans and rodents. This upregulation may be a negative feedback response of the ECS to inhibit excessive energy expenditure and restore homeostasis, and is likely mediated via the central nervous system. In overweight subjects, CB1R density in BAT, WAT, and the brain was significantly lower compared with lean subjects, reflecting impairment of the endogenous cannabinoid system in obesity. We conclude that endocannabinoid signaling via the CB1R is significant in the activation and regulation of human BAT, and targeting CB1R could provide a prospective way to treat obesity and metabolism.

Acknowledgments. The authors thank all of the study subjects enrolled in this study for cooperation and all the technical staff of Turku PET Centre for assistance.

Funding. The study was financially supported by the Academy of Finland (307402, 259926, 265204, 292839, and 269977), the Paulo Foundation, the Instrumentarium Foundation, the Turku University Hospital Research Funds, and the European Union (EU FP7 project 278373, DIABAT). The study was conducted within the Finnish Centre of Excellence in Cardiovascular and Metabolic Diseases, which was supported by the Academy of Finland, the University of Turku, Turku University Hospital, and Åbo Akademi University. T.G. was supported by Deutsche Forschungsgemeinschaft (DFG) grant GN 108/1-1. A.P. was supported by DFG grant RTG 1873.

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

Author Contributions. M.L. designed the studies, performed clinical imaging experiments, analyzed data, and prepared the manuscript. O.E. designed the studies, obtained funding, conducted the preclinical imaging experiments and data analysis, and edited the manuscript. T.G. conducted preclinical in vitro studies and data analysis and edited the manuscript. V.O., M.B., K.K., J.T., M.U., and M.H.-S. processed and analyzed data. J.H. supervised and contributed to image analysis. T.N. and M.T. conducted acquisition of human tissue biopsy samples. S.L. provided radioligands for imaging experiments. A.P. obtained funding, supervised the studies, and edited the manuscript. K.A.V. and P.N. designed the study, obtained funding, supervised the performance of the studies and data analysis, and edited the manuscript. All authors contributed to the critical revision of the manuscript and approved the final version. P.N. 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.

Prior Presentation. Parts of this study were presented at the Keystone Symposium on Obesity and Adipose Tissue Biology, Keystone, CO, 22–26 January 2017.