The European Union recommendations (2010/63/EU) on animal experimentation were followed. All animal experimentations were approved by the Ethic Committee of the University of Malaga (authorization no. 2012–0061A), and followed the 3R's principle. Ten-week-old C57BL/6J male mice were purchased from Charles River (France) and were acclimatized to the animal facility for one week with food and water available ad libitum and lights on between 8:00 and 20:00 h. Mice were then fed a 10% fat diet (control) or a 45% fat diet (45% of Kcal from lard, saturated fats, HFD) for 15 weeks (n = 10 and 30 mice, respectively). Body weight was monitored twice a week and glucose and insulin tolerance assessed by intraperitoneal glucose tolerance (GTT) and insulin tolerance test (ITT), respectively. Briefly, GTT was performed after overnight fasting by injecting 2 g/kg D(+)glucose (Sigma-Aldrich, St. Louis, MO). Blood glucose was monitored from the tail vein at baseline and 15, 30, 60, and 90 min using a glucose meter (Accu-check, Roche Diagnostic). For the ITT, mice were fasted for 6 hours and then injected with 0.5 U/kg of insulin (Humulin, Lilly, France). Blood samples were collected from the tail vein as above and blood glucose was measured at the same time points with the glucometer. Glucose and insulin area under the curve (AUC) was calculated from their corresponding graphs using IMAGEJ software (National Institutes of Health, Bethesda, MA, USA). Once HFD-fed mice showed greater body weight than control mice, glucose intolerance and insulin resistance, animals were randomized to vehicle or treatments and 10 mice were treated with the synthetic cannabinoid Abn-CBD. The herein study with Abn-CBD represents a subset of a larger study that also included the cannabinoid ligand LH-21. The pre-diabetic phenotype of these mice (HFD-vehicle vs. SD-vehicle) has been previously described (22). A detailed scheme of Abn-CBD study design is depicted in Figure 1A.

HFD mice were treated daily for the last 2 weeks of the experimental design with 0.05 mg/kg of Abn-CBD (Cayman Chemical, Ann Harbor, MI) or vehicle. Dosage was selected based on previous literature (15, 23). The stock solution of Abn-CBD was dissolved in ethanol, aliquoted, and stored at −20°C until use. Every day the dose to inject was freshly prepared by diluting an aliquot in saline-Tween80 (1% final ethanol content, 5% Tween80) and then intraperitoneally administered. Body weight and food intake were monitored daily, and the amount of food ingested was converted to kilocalories. Food efficiency was calculated as body weight variation (in grams) over caloric intake (in Kcals). Five days before euthanasia all groups of mice were also daily injected with 75 mg/kg of 5-bromo-2′-deoxyuridine (BrdU) dissolved in saline (Sigma-Aldrich). At the end of the treatment period, mice were euthanized by cervical dislocation and tissues collected for further histological and biochemical analysis. Samples from adipose tissue, pancreas and liver were fixed in 4% PFA by overnight immersion and then paraffin-embedded for histochemistry and immunohistochemistry. In parallel, plasma, adipose, pancreatic and hepatic samples were also snap-frozen and stored at −80°C for biochemistry analysis. The number of animals used for each experiment is detailed in figure legends.

Plasma was obtained by centrifuging (2,000 × g, 4°C, for 10 min) whole blood in EDTA-coated tubes. Plasma levels of insulin and several inflammatory cytokines (IFN-γ, IL-5, IL-6, CXCL1, IL-10) were measured using the Meso Scale Discovery Multi array (Meso Scale Diagnostics, Rockville, MA, USA) in an MSD instrument (SECTOR S 600) equipped with multi-array electrochemiluminescence detection technology (Meso Scale Diagnostics). Plasma levels of leptin (BioVendor, Czech Republic) and HMW Adiponectin (Shibayagi Co., Japan) were measured by ELISA assay in accordance with manufacturer's instructions.

The plasma lipid profile (Triglycerides, total cholesterol, HDL-c, LDL-c, NEFA, and glycerol), blood glucose level and the liver marker alanine transaminase (ALT) were determined by routine laboratory methods using a Cobas Mira autoanalyzer (Roche Diagnostic System, Basel, Switzerland) and reagents from Spinreact (Spinreact S.A.U., Girona, Spain) and Biosystems (Biosystems S.A., Barcelona, Spain). Plasma insulin levels were measured using a commercial ELISA assay according to manufacturer's instructions (Mercodia, Uppsala, Sweden).

Lipid peroxidation was determined by measuring thiobarbituric acid-reactive substances (TBARS). Malondialdehyde (MDA), a natural bi-product of lipid peroxidation reacts with thiobarbituric acid (TBA) to generate a MDA-TBA adduct that can be easily quantified at 532 nm in a spectrophotometer. Tissue samples were homogenized with ice-cold Tris-HCl buffer (150 mM KCl, 50 mM Tris, pH 7.4) supplemented with butylated hydroxytoluene to avoid artificial peroxidation during the test. The supernatant was incubated with MDA to obtain the TBARS. The absorbance was measured (VERSAmax, Molecular Devices LLC, San Jose, CA, USA) and interpolated in a standard curve using malondialdehyde-bisdiethyl-acetal (MDA, Sigma-Aldrich). The final values were expressed as nanomoles of TBARS per milligram of tissue.

Islets of Langerhans were isolated from two mice in each group by using the collagenase digestion method, as previously described (24). Islets were then cultured for 20–24 h in RPMI-1640 medium supplemented with 11 mM glucose (Invitrogen, CA, USA), 2 mM glutamine, 200 IU/ml penicillin, 200 μg/ml streptomycin and 8% fetal bovine serum stripped with charcoal-dextran (Invitrogen). For GSIS experiments islets were pre-incubated at 37 °C for 2 h in Krebs-bicarbonate buffer solution containing 14 mM NaCl, 0.45 mM KCl, 0.25 mM CaCl₂, 0.1 mM MgCl₂, 2 mM HEPES and 3 mM glucose, and equilibrated with 95% O₂: 5% CO₂ at pH 7.4. Size-matched islets, five in each well from a 24-well plate, were seeded in 0.5 mL fresh buffer containing 3 mM glucose or 11 mM glucose. Then islets were incubated for 1 h at 37°C, 5% CO₂. After incubation, 1% bovine albumin was added to each well, and the plate was cooled at 4°C for 15 min to stop insulin secretion. Media were then collected and stored at −20°C until insulin measurement by ELISA (Mercodia, Uppsala, Sweden), according to the manufacturer's instructions.

Number of islets and islet area were assessed by morphometric analysis in pancreatic sections. Paraffin-embedded pancreases from each mouse were cut at four different levels and stained with haematoxylin and eosin. Low-magnification photomicrographs were taken in an Olympus BX41 microscope (Olympus Corporation, Tokyo, Japan) and analyzed by the ImageJ software.

Determination of hepatic glycogen was performed according to previous reports with some modifications (25, 26). Briefly, the liver samples (300–500 mg) were transferred to test tubes containing 30% KOH (w/v) and boiled for 1 h until complete homogenization. Na₂SO₄ was then added, and the glycogen was precipitated with ethanol. The samples were centrifuged at 800 g for 10 min, the supernatants were discarded, and the glycogen was dissolved in hot distilled water. Ethanol was added and the pellets obtained after a second centrifugation were dissolved in distilled water in a final volume of 25 ml. Glycogen content was measured by treating a fixed volume of sample with phenol reagent and H₂SO₄. Absorbance was then read at 490 nm with a spectrophotometer (VERSAmax, Molecular Devices LLC).

Tissues (100 mg of adipose tissue) were dissected and mRNA isolated using Trizol Reagent (Sigma-Aldrich) and RNeasy Mini Kit (Qiagen) following manufacturer's instructions. Retrotranscription was performed using SuperScript IV RT (Thermo Fisher Scientific Inc., Waltman, MA, USA) and mRNA expression were analyzed in an Applied Biosystems^® 7500 fast using Fast Advanced Master Mix (all from Thermo Fisher Scientific) and appropriate FAM-labeled Taqman primers and probes for Cxcl1, Il10, and Ucp1. VIC-labeled primers and probe were used for housekeeping genes.

Frozen liver samples were sliced in a cryostat, attached to microscope slides, and air-dried at room temperature for 30 min. Liver sections were then stained in fresh Oil red O for 10 min, rinsed in distilled water and immediately counterstained with haematoxylin for 1 min. Photomicrographs were taken on an Olympus BX41 microscope and the Oil red O staining intensity was quantified by using Image J software.

Liver fibrosis was evaluated at the histological level by staining collagen fibers with Masson's trichrome staining. For that purpose, paraffin-embedded sections were hydrated and stained by using the Masson's trichrome kit (Casa Álvarez Material Científico S.A., Madrid, Spain) according to manufacturer's instructions.

Paraffin-embedded tissue sections (3 μm) were dewaxed, hydrated, and treated with antigen unmasking solution, citric acid based (Vector Laboratories Inc., Burlingame, CA, USA) for 20 min in a steamer and then 20 min to cool down. Sections were washed thrice with phosphate buffered saline (PBS). Endogenous peroxidase was quenched with 2% H₂O₂ in PBS for 30 min with agitation and endogenous biotin, biotin receptors, and avidin binding sites were blocked by avidin/biotin blocking kit according to manufacturer's instructions (Vector Laboratories Inc.). Immunohistochemistry was performed by incubating overnight at 4°C with primary antibody 1/100 (anti-insulin, Sigma-Aldrich; anti-F4/80, Abcam, Paris, France; anti-pNFKB, Abcam), rinsed thrice with PBS, followed by HRP polymer-conjugated Goat anti-Rat/Mouse polyclonal antibody (1 h) and finally rinsed thrice again and developed with diaminobenzidine substrate. Slides were rinsed in tap water, lightly counterstained with Mayer's haematoxylin, rinsed in ammonium chloride dehydrated and mounted with DPX medium (Shandon, Pittsburgh, Pennsylvania, USA). Specific primary antibodies were substituted with PBS or non-immune isotype-matched sera as the negative control.

For immunofluorescence primary antibody (anti-insulin 1/100 overnight, Santa Cruz Biotechnology Inc., Dallas TX, USA; anti-5-bromo-2′-deoxyuridine 1/100 overnight, Sigma-Aldrich; anti-αSMA, 1/100 overnight, Santa Cruz Biotechnology) incubation was followed by anti-rabbit IgG-AlexaFluor488 and/or anti-mouse IgG-AlexaFluor568 (1/1000; Thermo Fisher Scientific), for 1 h at room temperature. Slides were coverslipped and protected from photobleaching by Fluoroshield Mounting Media (Sigma-Aldrich). Photomicrographs were taken on an Olympus BX41 and further processed by Image J software to quantify signal intensity.

Apoptosis was evaluated by the TUNEL technique using the in situ apoptosis detection kit (Roche) according to manufacturer's instructions. Images were analyzed using ImageJ software. Number of TUNEL-positive cells were normalized to islet area.

Data are expressed as mean ± standard error of the mean (S.E.M.) for data fitting a normal distribution and median ± min to max for non-normal distributions. The statistical significance of differences in mean or median values was assessed by Student t-test or analysis of variance (ANOVA) followed by Tukey's multiple comparison test for normal distributions and by Mann-Whitney test or Kruskal-Wallis test followed by Dunn's multiple comparison test for non-normal distributions. All analyses were performed with GraphPad Prism 6.07 or 7.04 (GraphPad Software, San Diego, CA, USA). A p < 0.05 was considered significant.

Ten-week-old male C57Bl/6J mice were fed a high fat diet (HFD) or standard diet (SD) for 15 weeks. Mice fed HFD weighed significantly more than SD-fed mice (37.9 ± 0.9 vs. 32.6 ± 1.2 g respectively; Figure 1B). Diet induced obese (DIO) mice were glucose intolerant and insulin resistant but they did not show impaired fasting glucose [103 ± 7 vs. 108 ± 4 mg/dl, SD- and HFD-mice, respectively (22)]. DIO mice were then randomized to daily intraperitoneal injections of vehicle (HFD-vehicle mice) or 0.05 mg/kg of Abn-CBD (HFD-Abn-CBD mice) for 2 weeks (Figure 1A). Dosage for Abn-CBD was selected based on an ipGTT on lean mice (Supplementary Figure 1) and previous literature (15). After 2 weeks of treatment, body weights of HFD-vehicle and HFD-Abn-CBD mice were comparable (36.5 ± 0.8 and 38.9 ± 1.7 g, respectively), while SD-fed mice weighed 29.8 ± 1.8 g (Figures 1B,C). Both HFD-Abn-CBD and HFD-vehicle mice remained glucose intolerant (Supplementary Figure 2). Calorie intake per mouse was significantly higher in the HFD-Abn-CBD than in the HFD-vehicle group (Figure 1D), although calorie per gram (Figure 1E) and food efficiency was comparable in both groups (Figure 1F). In obesity, adiposity positively correlates with leptin plasma levels. HFD-vehicle mice had a 2.8-fold increase in plasma leptin levels compared to lean mice (Figure 1G). In agreement with their increased body weight and food consumption (Figures 1C,D), HFD-Abn-CBD mice had greater leptin levels (4.2-fold higher compared to lean mice) although differences found between HFD-vehicle and -Abn-CBD mice did not reach statistical significance (p = 0.07) (Figure 1G). Adiponectin, a hormone secreted from the adipose tissue to regulate glucose homeostasis, is known to be reduced in obesity (27). Plasma adiponectin levels were significantly lower in HFD-fed mice compared to SD-fed mice, independently of treatment (Figure 1H).

As DIO is associated with elevated triglycerides, free-fatty acids and cholesterol in circulation, we therefore analyzed the plasma lipid profile of SD-fed, HFD-vehicle, and HFD-Abn-CBD mice. Although plasma triglyceride levels in HFD-fed mice were similar to those found in SD-fed mice, total cholesterol was 1.3-fold higher in HFD-fed compared to SD-fed mice (Table 1). High density lipoprotein (HDL) and non-esterified fatty acid (NEFA) content in plasma were also significantly higher in HFD-fed mice than in plasma from SD-fed mice (1.5- and 1.6-fold, respectively; Table 1). There were no significant differences in low density lipoprotein (LDL) levels in plasma, neither of glycerol levels between HFD-fed and SD-fed mice (Table 1). The lipid profile found in plasma from HFD-Abn-CBD mice was comparable to the one found in HFD-vehicle mice, including high cholesterol, elevated HDL and NEFA (1.25-, 1.5- and 1.6-fold higher, respectively, than in SD-fed mice; Table 1).

After 2 weeks of treatment with vehicle or Abn-CBD, HFD-vehicle mice were hyperinsulinemic compared to SD-fed mice (Figure 2A) while blood glucose levels were comparable (Figure 2B). These data suggest that mice remained insulin resistant as before starting the treatment but had not yet developed diabetes. Remarkably, mice treated with Abn-CBD had a similar insulinemia than SD-vehicle (Figure 2A) while maintaining blood glucose at the same level as SD- and HFD-vehicle mice (Figure 2B). We analyzed the pancreatic islets of Langerhans both at the morphometric and functional level (Figures 2C–F). No changes in number of islets were detected among groups (Figure 2E). However, mice fed a HFD showed a strong tendency toward larger islets than SD-fed mice (p = 0.06) while this hypertrophy was absent in HFD-Abn-CBD islets (Figure 2D). Treatment with Abn-CBD had no effect on total insulin secretory capacity, as shown by a static in vitro glucose-stimulated insulin secretion assay (Figure 2F). In DIO, loss of beta cell mass is associated with increased inflammation and oxidative stress in beta cells that leads to beta cell death. Accordingly, we found increased intra-islet apoptosis (Figure 3A), phosphorylation of p65 (p-NFκB) (Figure 3B) and a slight non-significant increase in pancreatic TBARS (Figure 3D) in HFD-vehicle when compared to SD-Vehicle. Importantly, Abn-CBD significantly reduced pancreatic content of TBARS (Figure 3D), intra-islet pNFκB staining (Figure 3B) and macrophage infiltration (Figure 3C), greatly lessening apoptosis (Figure 3A). Moreover, Abn-CBD induced beta cell proliferation, as measured by BrdU positive staining of beta cells, compared to SD-fed mice (Figure 3E). Thus, Abn-CBD protected beta cell mass without altering beta cell function.

Low-grade chronic inflammation is associated with DIO-related complications. Accordingly, plasma of HFD-vehicle mice showed significantly higher levels of interleukin 6 (IL-6; 3.2-fold increase; Figure 4A) and CXCL-1 (2.4-fold increase; Figure 4B) compared to SD-fed mice, while interleukin 5 (IL-5) levels trended to be 1.6-fold higher (Figure 4C). Of importance, treatment with Abn-CBD restored the levels of IL-6, CXCL-1, and IL-5 to those found in SD-fed mice (Figures 4A–C). HFD-fed mice did not show any change in either interferon gamma (IFNγ; Figure 4D) nor interleukin 10 (IL-10; Figure 4E) compared to SD-fed mice, independently of the treatment.

Since obesity is strongly associated to inflammation at white adipose tissue (WAT) (28), we analyzed the inflammation occurring in visceral (VAT) and subcutaneous adipose tissue (SAT). HFD-fed mice showed an expected increase in adipocyte size compared to SD-fed mice, independently of treatment (Figure 5A). We found an increase in the presence of crown-like structures in HFD-vehicle WAT upon staining with F4/80, which was significantly decreased in WAT from Abn-CBD-treated mice (Figure 5B). Analysis of cytokine expression showed a significant reduction of Il10 expression in VAT (Figure 5C), and a significant increase in the expression of Cxcl1 in both VAT (p = 0.07) and SAT from HFD-vehicle mice (Figures 5D,E, respectively). Interestingly, treatment with Abn-CBD prevented the alterations observed in cytokine expression in VAT (Figures 5C,D) and SAT (Figure 5E), showing that Abn-CBD protects WAT from diet-induced inflammation. We also wanted to explore whether Abn-CBD could promote a shift toward browning in WAT. For this purpose we analyzed Ucp1 expression in BAT, VAT, and SAT. While no significant differences were found in brown adipose tissue among groups (Figure 5F), HFD-vehicle mice showed a significant decrease in Ucp1 expression in VAT and SAT that was counteracted by Abn-CBD (Figures 5G,H).

In the liver the accumulation of triglycerides leads to NAFLD that eventually leads to liver failure (9). Herein, DIO mice accumulated triglycerides in liver independently of treatment, displaying 2-fold increase in lipid content compared to lean mice, as shown by oil red-O staining (Figure 6A). Furthermore, liver from HFD-vehicle mice exhibited fibrosis, as determined by a significant increase in collagen content compared to SD-fed mice (Figure 6B) as well as early makers of liver fibrosis, such as α-SMA (Figure 6C). Interestingly, the liver from HFD-Abn-CBD mice had significantly lower collagen than HFD-vehicle mice (Figure 6B) and reduced staining of α-SMA (Figure 6C). Moreover, liver glycogen content was comparable in both HFD- and SD-fed vehicle-treated mice (Figure 6D). Although not reaching statistical significance (p = 0.07), HFD-Abn-CBD mice displayed a 2.2-fold higher content of hepatic glycogen than HFD-vehicle mice (Figure 6D). In order to assess liver damage, we measured plasma levels of alanine aminotransferase (ALT). Levels of ALT in DIO mice were comparable to those found in lean mice, independently of treatment (Figure 6E).

As HFD-Abn-CBD mice displayed reduced liver fibrosis concomitant with lower levels of several circulating pro-inflammatory cytokines compared to HFD-vehicle mice, we further assessed the degree of liver inflammation. Liver from HFD-vehicle mice had increased p-NFκB compared to liver from SD-fed mice, while liver from HFD-Abn-CBD mice had significantly reduced p-NFκB levels comparable to those in liver from SD-fed mice (Figure 7A). Additionally, liver from HFD-vehicle mice showed a significant 3.3-fold increase in the number of macrophages (F4/80⁺ cells) compared to SD-fed mice (Figure 7B). Treatment with Abn-CBD greatly reduced the number of macrophages to the levels found in SD-fed mice (Figure 7B). We then determined hepatocytes viability measured as apoptosis and proliferation. Hepatocytes from HFD-vehicle mice were undergoing significantly more apoptosis than those from SD-fed mice, which was reverted upon Abn-CBD treatment (Figure 7C). In addition, Abn-CBD increased hepatocytes proliferation when compared to HFD-vehicle, but not over those levels found in hepatocytes from SD-fed mice (Figure 7D).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.