CB1-mediated endocannabinoid tone is enhanced in experimental diet-induced or genetical models of NAFLD, and is characterized by upregulation of adipose tissue and hepatocyte CB1 receptors and increased liver synthesis of anandamide. The pathogenic role of CB1 receptors in NAFLD is supported by the resistance to steatosis of obese mice bearing a global or hepatocyte-specific CB1 deletion, or of rodents administered rimonabant or AM6545, a CB1 antagonist with limited brain penetrance [[4], [5], [6]]. Studies in cultured hepatocytes and liver slices further indicate that the steatogenic properties of CB1 arise from combined hepatocyte activation of SREBP1c-mediated lipogenesis, reduction of fatty acid oxidation via inhibition of AMP kinase, and decreased release of TG-rich VLDL [[4], [5], [7]]. In addition, the adipose tissue may largely contribute to the steatogenic process via CB1-induced release of free fatty acids by adipocytes [[3], [8]] (Fig. 1).

Pharmacological or genetic inactivation of CB1 receptors is also associated with a decrease in serum transaminase levels in obese rodents [[5], [6]]. These observations suggest that besides their steatogenic properties, hepatocyte CB1 receptors might also adversely affect hepatocyte survival. Whether reduced hepatocyte survival may be a consequence of CB1-mediated activation of endoplasmic reticulum stress [5] remains to be investigated.

A potential impact of CB1 receptors on the inflammatory response associated with NASH has been suggested by experiments in obese rats, showing that rimonabant reduces liver inflammation [[5], [6]]. The underlying mechanism remains to be delineated but in hepatocytes, CB1 receptors could contribute to the acute phase response, via activation of CREBH, a liver-specific transcription factor that upregulates acute phase response genes [9]. In addition, extrahepatic sources of CB1 receptors could participate in Kupffer cell activation by LPS. Indeed, activation of CB1 receptors in the intestine increases gut permeability and LPS release in obese mice [10]. In addition, fat CB1 receptors reduce the production of adiponectin, an adipokine which reduces hepatic inflammation [[5], [6]]. In keeping with these data, administration of rimonabant to obese mice reduces gut permeability and endotoxemia, restores adiponectin levels, and improves the hepatic inflammatory response [[5], [6], [10]].

The pathogenic role of CB1 shares striking similarities in ALD and NAFLD, in keeping with common pathological features of liver injury. Mice fed an alcohol-enriched diet display enhanced production of 2-AG by hepatic stellate cells, resulting in paracrine induction of CB1 receptors in hepatocytes [11]. As demonstrated in experimental models of NAFLD, administration of rimonabant or genetic ablation of CB1 receptors promotes resistance to alcohol-induced steatosis, following combined reduction of lipogenesis and enhanced fatty acid oxidation [11].

CB1 receptor expression is upregulated in hepatic myofibroblasts from cirrhotic patients, and profibrogenic effects of CB1 receptors expressed in hepatic myofibroblasts were described in mice undergoing bile duct ligation, chronic exposure to carbon tetrachloride or thioacetamide, and in a model of NASH elicited by prolonged high fat feeding [[12], [13]]. In these models, treatment with rimonabant or genetic ablation of CB1 receptors reduces fibrogenesis compared to control animals. Antifibrogenic properties of CB1 antagonism have been ascribed to growth inhibitory and proapoptotic effects towards hepatic myofibroblasts [12]. Importantly, CB1 antagonism remains operant at advanced stages of chronic liver injury, as shown by the therapeutic effects of rimonabant in rats with full-blown cirrhosis [14].

Liver regeneration is associated with a marked induction of CB1 receptors and increased hepatic production of anandamide [15]. Mice administered rimonabant or lacking CB1 either globally or selectively in hepatocytes, show delayed liver regeneration, as reflected by decreased mitogenic activity of hepatocytes. Reduced hepatocyte proliferation results from inhibition of cell cycle proteins involved in mitotic progression [15].

Endogenous activation of CB1 receptors promote hypotension and splanchnic vasodilation associated with advanced cirrhosis, as a consequence of the enhanced production of anandamide by circulating macrophages and upregulation of CB1 receptors in endothelial cells and mesenteric arteries of cirrhotic rats [16]. Blockade of this vasodilatory tone by rimonabant improves blood pressure, peripheral vascular resistance and portal pressure, and prevents the development of ascites [16].

Activation of CB2 receptors in Kupffer cells limits liver inflammation and reduce alcohol-induced liver injury [17]. Indeed, in alcohol-fed mice, endogenous or exogenous stimulation of CB2 receptors prevents the switch of Kupffer cells to a proinflammatory M1 phenotype, and enhances transition towards the anti-inflammatory M2 phenotype, via a mechanism involving activation of heme oxygenase-1 [17]. Stimulation of Kupffer cell CB2 receptors also blunts alcohol-induced fat accumulation in hepatocytes. The antisteatogenic signal ensues from the reduced paracrine release of steatogenic cytokines (i.e., TNF-α and IL1-β) by CB2-stimulated Kupffer cells [17].

Strinkingly, CB2 receptors display steatogenic properties in HFD-fed mice that derive from indirect CB2-mediated increase in adipose tissue inflammation and insulin resistance [18]. Whether CB2 receptors display protective effects against NASH, as anticipated from their anti-inflammatory and hepatoprotective effects (see below), deserve further studies.

Endogenous or exogenous activation of CB2 receptors has been described as a protective pathway in several models of acute liver injury, in which CB2 receptors undergo early induction in non-parenchymal cells [2]. Interestingly, and similar to CB2 agonists, MAGL inhibition also protects against hepatocyte apoptosis elicited by acute liver injury, by enhancing 2-AG-mediated CB2 signaling [19]. Hepatoprotective properties of 2-AG-mediated CB2 effects result from indirect reduction of oxidative/nitrosative stress in hepatocytes [[19], [20]]. In addition, CB2 receptors expressed in hepatic myofibroblasts also contribute to liver regeneration by accelerating proliferation of hepatocytes via a paracrine interleukin-6 dependent pathway [20].

Although CB2 receptors are not expressed by hepatocytes, their expression has been detected in human hepatocarcinoma cell lines and their activation promote growth inhibition and apoptosis of tumor cells, via an autophagy-dependent pathway [21]. In line with in vitro observations, the CB2 agonist JWH-015 reduces tumor growth in an orthotopic model of hepatocarcinoma xenograft [21].

In contrast to CB1, CB2 signaling has been identified as an antifibrogenic pathway. Thus, CB2-deficient mice show enhanced fibrosis [22], whereas administration of the CB2 agonist JWH-133 to rats with established cirrhosis improves liver fibrosis [23]. Antifibrogenic properties of CB2 receptors rely on combined direct and indirect effects on hepatic myofibroblasts. Indeed, activation of CB2 receptors in hepatic myofibroblasts reduce their density, owing to antiproliferative and apoptotic properties [22]. In addition, CB2 receptors indirectly reduce hepatic myofibroblast accumulation via (i) indirect hepatoprotective effects on hepatocyte survival [20], (ii) anti-inflammatory properties of Kupffer cell CB2 receptors [17] and (iii) downregulation of the production of the profibrogenic cytokine IL17 by Th17 lymphocytes [24] (Fig. 2).