Key questions of endocannabinoid signalling in the CNS: which, where and when?

In 2001, endogenous cannabinoids (ECs) were identified as retrograde messengers in a form of short-term synaptic plasticity known as depolarization-induced suppression of inhibition (DSI). In this form of plasticity, activity-dependent calcium influx in a postsynaptic cell promotes synthesis and release of ECs that travel backward across the synaptic cleft to activate metabotropic presynaptic type 1 cannabinoid receptors (CB1Rs). Activation of these receptors leads to short-term inhibition of action potential-induced GABA release onto the postsynaptic neuron. As such, DSI represents a mechanism whereby an individual neuron may transiently, but effectively, regulate the release probability of its own inputs.

Motivated largely by initial discoveries in hippocampus and cerebellum, work over the last decade has produced an enormous increase in our understanding of EC-mediated and CB1R-dependent forms of synaptic plasticity. It is now clear that EC-mediated retrograde transmission occurs in many areas of the CNS, that ECs can be mobilized by a variety of specific mechanisms, and that diversity of EC signalling supports a wide range of short-term, long-term, and even metaplastic phenomena at a wide variety of both inhibitory and excitatory synapses.

However, despite extensive progress, it has been persistently difficult for the field to definitively associate specific ECs with specific forms of CB1R-dependent synaptic plasticity. The two primary candidates in multiple types of EC-dependent phenomena are clearly 2-arachidonoylglycerol (2-AG) and N-arachidonoylethanolamide (anandamide). Each of these compounds is a fatty acid derivative that is widely available in the CNS, and each is an effective CB1R ligand. Further, a variety of molecular mechanisms for synthesis and breakdown of each of these compounds has become increasingly well described.

Initial efforts to determine whether 2-AG or anandamide is the retrograde messenger specifically required for DSI relied on the use of enzyme inhibitors to prolong breakdown or prevent synthesis of ECs. In several studies prolonged breakdown of 2-AG increased the duration of DSI, while prolonged breakdown of anandamide did not. On the other hand, a large number of studies that attempted to disrupt EC-mediated signalling by inhibition of diacylglycerol lipase (DAGL), an enzyme necessary to synthesize 2-AG from diacylglycerol, produced highly inconsistent results. Further, it became clear that the pharmacological agents used in these studies were insufficient to distinguish between two isoforms of DAGL (DAGLα and DAGLβ) that have potential for distinct developmental and functional roles.

In this issue of The Journal of Physiology, Yoshino et al. (2011) use a powerful combination of both pharmacological and genetic tools to examine the dependence of DSI as observed in mouse prefrontal cortex (PFC) on DAGL. They find that the magnitude and duration of DSI is increased when degradation of 2-AG, but not anandamide, is inhibited with selective enzyme inhibitors. Similarly, DSI in PFC is blocked by both extracellular and intracellular application of a commonly used DAGL inhibitor. Further, these data are strongly reinforced by the significant observation that basal 2-AG levels are dramatically reduced and DSI is eliminated in DAGLα^−/− but not in DAGLβ^−/− mice.

The report by Yoshino et al. follows two others appearing in 2010 that independently developed DAGLα^−/− and DAGLβ^−/− mice to study the molecular mechanisms of DSI in hippocampus, cerebellum and striatum (Gao et al. 2010; Tanimura et al. 2010). When considered as a group, it is notable that all three studies report that selective removal of DAGLα but not DAGLβ reduces basal 2-AG levels and eliminates DSI in their respective regions of interest. However, the work by Yoshino et al. is unique in that it is the only study to explicitly examine the effects of DAGL removal on DSI in an area of the brain where genetic deletion of DAGLα does not also substantially reduce basal levels of anandamide. Thus, the current study by Yoshino et al. is the first to indicate that intriguing apparent interdependencies between basal levels of 2-AG and anandamide are region specific, and it is also notable for demonstrating unambiguously that changes in basal levels of anandamide are not an essential prerequisite for observing loss of DSI in DAGLα^−/− mice.

While the work of Yoshino et al. and previous authors clearly represents an important step forward for the field, their data also raise a number of intriguing issues that will be important to address in future work. For example, few of these studies reveal a clearly defined functional role for DAGLβ, and none carefully examine potential effects of DAGL removal on constitutive actions of cannabinoids. Similarly, information available to date cannot fully reconcile the extreme sensitivity of DSI to DAGLα with anatomical information that predominantly suggests a subcellular distribution more amenable to modulation of glutamatergic transmission. Finally, all three studies have now made possible a direct comparison of pharmacological vs. genetic (acute vs. chronic) inhibition of DAGL that deserves more attention. Some recent attempts to reconcile poor or non-existent effects of acute inhibition of DAGL with clear impacts of genetic deletion have postulated there may exist separate pools of ‘on-demand’ and ‘pre-stored’ 2-AG (e.g. see Min et al. 2010; Alger & Kim, 2011). In PFC, Yoshino et al. do not observe weak or delayed effects of pharmacological agents that might be expected according to this model; however, they do provide clear indication of apparent changes in physiological properties of PFC neurons that may be uniquely produced by chronic inhibition of DAGL. Overall, continued efforts such as those of Yoshino et al. and others have real potential to provide new mechanistic insights with wide ranging therapeutic implications.