Obesity is becoming a global epidemic and the need for development of novel therapeutic interventions is high. The endocannabinoid system (ECS) is regarded as a potential therapeutic target since it regulates energy balance by influencing appetite (Foltin et al., 1988; Jamshidi and Taylor, 2001), intracellular lipolysis and energy expenditure [reviewed in (Cota, 2007; Mazier et al., 2015)]. The ECS consists of cannabinoid receptors, their endogenous ligands, the endocannabinoids, and the enzymes that synthesize and degrade the endocannabinoids. The cannabinoid receptors are G-protein-coupled receptors, comprising the CB1R and CB2R. The CB1R is expressed centrally and in peripheral metabolic tissues including white adipose tissue (WAT), brown adipose tissue (BAT), liver, skeletal muscle and the pancreas. In contrast, the CB2R is mainly expressed in immune cells (Howlett et al., 2002).

The two main circulating endocannabinoids are anandamide (N-arachidonoylethanolamine, AEA) and 2-arachidonoylglycerol (2-AG). Although they are both derived from cell membrane arachidonic acid (AA) derivatives, the levels of these endocannabinoids are differentially regulated. AEA can be generated via hydrolysis of N-acyl-phosphatidylethanolamines (NAPE) by a NAPE-specific phospholipase D (NAPE-PLD) and its degradation is primarily regulated by fatty acid amide hydrolase (FAAH). 2-AG levels are regulated by the biosynthesis enzymes diacylglycerol lipase α and β (DAGL-α and β) and the degradation enzyme mono-acylglycerol lipase (MAGL) (Muccioli, 2010; Bisogno and Maccarrone, 2014; Baggelaar et al., 2018).

Activation of the CB1R in peripheral tissues inhibits fatty acid oxidation resulting in a positive energy balance and thus development of obesity in mice (Ravinet Trillou et al., 2003; Osei-Hyiaman et al., 2005; Arrabal et al., 2015). In addition, an increased tone of the ECS is associated with obesity in humans (Engeli et al., 2005; Bluher et al., 2006). Efforts have been made to reverse obesity by blocking the CB1R by small molecules such as the inverse agonist rimonabant (Sam et al., 2011). Albeit that rimonabant was effective in humans as evident from sustained weight loss and reduction of dyslipidaemia, centrally mediated side effects resulted in removal from the market (Despres et al., 2005; Pi-Sunyer et al., 2006). Nevertheless, blocking the actions of the ECS is still regarded a potent therapeutic strategy [reviewed in (O’Keefe et al., 2014; Simon and Cota, 2017)].

To develop novel therapeutics that target specific aspects of the ECS, it is crucial to obtain more insight in how fast and in which organs the dysregulation of the ECS sets off. In this study, we aimed at elucidating these questions by exposing mice to a high fat diet (HFD) ranging from 1 day up to 18 weeks, which finally causes diet-induced obesity (DIO). We analyzed endocannabinoid levels in plasma as well as the expression of enzymes involved in endocannabinoid synthesis and breakdown in several metabolic organs.

In this study, we demonstrated that HFD feeding increases circulating levels of endocannabinoids, with a rapid initial increase in AEA and a more gradual increase in 2-AG, in the course of DIO development. These changes were accompanied by increased gene expression of the synthesis and degradation enzymes of 2-AG in both WAT and BAT, and with increased expression of the AEA synthesis enzyme Nape-pld in BAT. Taken together, these data indicate that the dysregulation of the ECS in the development of obesity occurs rapidly and that WAT and BAT might contribute to these effects.

The observed increases in endocannabinoids in the development of DIO are in agreement with previous studies in mice that showed increased plasma 2-AG and AEA levels after 9 weeks (D’Eon et al., 2008) and 36 weeks (Pati et al., 2018) of HFD feeding and increased plasma AEA levels in a model for glucocorticoid induced obesity (Bowles et al., 2015). These data also are concordant with data in humans, since obese individuals have higher 2-AG levels compared to lean individuals (Engeli et al., 2005; Bluher et al., 2006; Cote et al., 2007). It was somewhat surprising that a subset of the control group that were not fed a HFD showed very high 2-AG levels in addition to AA levels, the reason of which is currently unclear to us. By performing a second study in mice we confirmed that HFD feeding induced an initial rise rather than decrease in 2-AG levels. Thus, although the reason for the very high 2-AG levels in a subgroup is intriguing, we regarded those mice as biological outliers. Besides 2-AG and AEA, HFD feeding also increased the plasma levels of AA. Since AA is constituent and degradation product of 2-AG and AEA, elevated AA levels may either be a cause or consequence of the increased levels. HFD feeding also increased plasma levels of other N-acylethanolamines, including OEA, PEA, SEA and DEA. These NAEs have other biological targets involved in controlling the energy balance, such as peroxisome proliferator-activated receptors-α (PPARα), PPARγ and G protein-coupled receptor 119 (Fezza et al., 2014).

Currently, it is unknown which organs contribute to the increased plasma endocannabinoid levels in HFD-induced obesity (Hillard, 2018). We showed that body weight positively correlates with plasma endocannabinoid levels, albeit that the correlation with AEA (R²= 0.654) is stronger than with 2-AG (R²= 0.073). Since body weight differences in the range of approximately 30–50 g, as observed in this study, are mainly caused by differences in body fat (van Beek et al., 2015), it was considered likely that engulfment of lipids by adipocytes and/or expansion of the adipocyte pool would contribute to the increase in endocannabinoids. Insulin resistance, which is closely linked to increased intracellular lipid deposition (Tchernof and Despres, 2013) is also associated with a dysregulated ECS (Gruden et al., 2016). By performing gene expression analysis in metabolically active organs, we could demonstrate that expression of enzymes involved in endocannabinoid synthesis increased in WAT as well as BAT. This is in full agreement with a previous study in which 3 and 8 weeks of HFD feeding, with a diet closely resembling the HFD used in our experiments, resulted in increased local levels of endocannabinoids (AEA and 2-AG) in BAT (Matias et al., 2008). The increase in plasma 2-AG coincided with increased gene expression of DAGLα and DAGLβ in both WAT and BAT. Given the different time-course of expression, where the increase in Daglα seems to precede the increase in Daglβ, we postulate that DAGLα may be responsible for the initial rise in 2-AG, while DAGLβ may mediate the late increase in 2-AG. Similarly, the increase in plasma AEA coincided with increased gene expression of its synthesizing enzyme NAPE-PLD in BAT. Moreover, plasma AEA correlated positively with Nape-pld expression in BAT but not in WAT. It is therefore likely that BAT rather than WAT contributes to the rise in AEA levels. In this respect, it is interesting that expression of Nape-pld is higher in BAT than in WAT, as evident from lower Ct values. The expression levels of most of the enzymes showed a sharp increase in the first week of HFD feeding, which coincided with the timing of the largest increase in lipid deposition in BAT. In BAT, ABHD4, and GDE1 were shown to also be involved in AEA synthesis and their expression respond to BAT activating agents (Krott et al., 2016), although we did not find the expression levels of these enzymes to coincide with the increase in circulating AEA and NAEs levels with prolonged HFD feeding. Of note, expression of the endocannabinoid degradation enzymes Faah and Mgll in the adipose tissues were either undetectable or increased. Although we have not been able to measure actual enzyme activities due to technical reasons, it is tempting to speculate that net whole body endocannabinoid synthesis exceeds degradation since circulating levels increase in the course of DIO. Synthesis enzymes of 2-AG in liver and skeletal muscle were only transiently increased, and the synthesis enzyme of AEA was decreased in skeletal muscle. Thus, although we cannot exclude the contribution of other organs as source for plasma endocannabinoid levels (e.g., brain and intestine), our data suggest that WAT and BAT are likely important organs that release 2-AG and AEA levels in HFD-induced obesity.

It is interesting to speculate on the cellular source within the adipose tissue depots that is involved in endocannabinoid synthesis. HFD-induced development of DIO causes accumulation of macrophages in WAT (van Beek et al., 2015) as well as BAT (Van den Berg S. M., unpublished). Macrophages are able to produce AEA (Di Marzo et al., 1996), and we found a positive correlation between macrophage marker Cd68 and Nape-pld expression in BAT as well as with AEA levels in plasma. However, the concentration of macrophages in adipose tissue is relatively low, even in obesity, and stronger positive correlations were found between the lipid content in BAT and both Nape-pld expression and plasma AEA levels. Therefore, adipocytes likely contribute substantially more to the circulating endocannabinoid pool than macrophages. This hypothesis is corroborated by previous findings that AEA co-localizes with adiposomes or lipid droplets in vitro (Oddi et al., 2008) and that specific deletion of NAPE-PLD in adipocytes of mice decreased levels of PEA, OEA and SEA in WAT, despite increased inflammation and influx of macrophages (Geurts et al., 2015). In our study, HFD feeding increases the cellular mRNA levels of the synthesizing enzymes in adipose tissue. In addition, expansion of the total number of adipocytes in the time course of DIO further increases whole body expression of these enzymes. The rapid increases in gene expression in BAT may be explained by the rapid whitening of BAT as induced by HFD feeding (Shimizu et al., 2014). Indeed, BAT lipid droplet content positively correlated with Nape-pld expression and AEA plasma levels. Furthermore, these data are in line with the recent observation that acute activation of BAT decreases Nape-pld expression (Krott et al., 2016). Interestingly, the concentration of 2-AG and AEA is higher in BAT than WAT (Krott et al., 2016), suggesting at least a role of BAT in determining circulating endocannabinoid levels. Collectively, it is likely that lipid-filled adipocytes rather than macrophages within the adipose tissues contribute to circulating plasma endocannabinoid levels.

In our study, we found no evidence for a contribution of decreased degradation pathways of 2-AG and AEA in adipose tissues determining plasma levels of endocannabinoids. Specifically, we were unable to detect any expression of AEA degradation enzyme Faah in WAT, BAT and muscle, and found decreased liver Faah expression after 18 weeks of HFD feeding. This is in line with the fact that FAAH was reported to play an important role in obesity. Notably, a missense polymorphism in the FAAH gene is associated with obesity in humans (Sipe et al., 2005) and FAAH deficient mice have increased AEA levels in, e.g., the liver and show increased fat mass and body weight (Tourino et al., 2010). On the other hand, Bartelt et al. (2011) found that HFD feeding for 16 weeks in mice caused a decrease in Faah expression and FAAH enzymatic activity in WAT which was accompanied by increased AEA in this tissue. To what extend catabolism of AEA and 2-AG by adipose tissue determines circulating levels of these endocannabinoids warrants further study.

It is tempting to speculate on the biological role of the increases of 2-AG and AEA in the time course of HFD-induced obesity. Endocannabinoids are known to decrease insulin sensitivity (Gruden et al., 2016) and to reduce sympathetic responses by inhibiting noradrenergic signaling (Quarta et al., 2011; Krott et al., 2016), and thereby decrease lipolysis in WAT and thermogenesis in BAT. Possibly, in case of acute lipid overload, as mimicked by a switch from regular chow to a HFD, initial accumulation of lipids in WAT and BAT drives the synthesis pathways of endocannabinoids that can have autocrine and even paracrine effects on these organs to inhibit sympathetic signaling. This sequence of events reduces intracellular lipolysis in BAT and WAT, thereby resulting in reduced thermogenesis in BAT and increased triglyceride storage in WAT. Such as a feed-forward mechanism may thus allow the body to store excess lipids effectively in adipose tissues. Interestingly, we have shown that inhibition of endocannabinoid signaling by strictly peripheral CB1R antagonism activates BAT and reduces adiposity in HFD-fed mice (Boon et al., 2014). In theory, HFD induced lipid accumulation can lead to increased endocannabinoid synthesis which attenuates BAT and WAT activity and results in a positive energy balance.

The time course of HFD-induced obesity plasma endocannabinoid levels rapidly rise as most probably explained by increased synthesis pathways in adipose tissue depots. We speculate that this sequence of events may attenuate sympathetic signaling in these tissues by CB1R agonism, which would result in reduced thermogenesis and increased storage of excess lipids in WAT. Given the fact that strictly peripheral CB1R antagonism activates BAT and reduces adiposity in mice, we anticipate that strategies inhibiting CB1R selectively on (brown) adipocytes or reducing endocannabinoid synthesis by adipocytes may be a worthwhile strategy to pursue in combating obesity and associated disorders.

EK and VK performed the experiments, analyzed the data, wrote the manuscript, and contributed to the discussion. VK developed and validated the UPLC-MS/MS method to quantify endocannabinoids, NAEs and AA in murine plasma. BM and RvE analyzed the data, contributed to the discussion, and reviewed/edited the manuscript. SvdB designed the study, performed the in vivo experiment and kindly provided the samples for analysis, and reviewed and edited the manuscript. MdW and EL designed the study and reviewed and edited the manuscript. KN, SK, AH, and TH contributed to the discussion and reviewed and edited the manuscript. AR-M and TC reviewed and edited the manuscript. MvdS, PR, and MB designed and supervised the project, contributed to the discussion and reviewed and edited the manuscript.

AR-M and TC are employees of Lilly. The remaining 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.