Research Article: The serine hydrolases MAGL, ABHD6 and ABHD12 as guardians of 2-arachidonoylglycerol signalling through cannabinoid receptors

Date Published: February , 2012

Publisher: Blackwell Publishing Ltd

Author(s): J R Savinainen, S M Saario, J T Laitinen.


The endocannabinoid 2-arachidonoylglycerol (2-AG) is a lipid mediator involved in various physiological processes. In response to neural activity, 2-AG is synthesized post-synaptically, then activates pre-synaptic cannabinoid CB1 receptors (CB1Rs) in a retrograde manner, resulting in transient and long-lasting reduction of neurotransmitter release. The signalling competence of 2-AG is tightly regulated by the balanced action between ‘on demand’ biosynthesis and degradation. We review recent research on monoacylglycerol lipase (MAGL), ABHD6 and ABHD12, three serine hydrolases that together account for approx. 99% of brain 2-AG hydrolase activity. MAGL is responsible for approx. 85% of 2-AG hydrolysis and colocalizes with CB1R in axon terminals. It is therefore ideally positioned to terminate 2-AG-CB1R signalling regardless of the source of this endocannabinoid. Its acute pharmacological inhibition leads to 2-AG accumulation and CB1R-mediated behavioural responses. Chronic MAGL inactivation results in 2-AG overload, desensitization of CB1R signalling and behavioural tolerance. ABHD6 accounts for approx. 4% of brain 2-AG hydrolase activity but in neurones it rivals MAGL in efficacy. Neuronal ABHD6 resides post-synaptically, often juxtaposed with CB1Rs, and its acute inhibition leads to activity-dependent accumulation of 2-AG. In cortical slices, selective ABHD6 blockade facilitates CB1R-dependent long-term synaptic depression. ABHD6 is therefore positioned to guard intracellular pools of 2-AG at the site of generation. ABHD12 is highly expressed in microglia and accounts for approx. 9% of total brain 2-AG hydrolysis. Mutations in ABHD12 gene are causally linked to a neurodegenerative disease called PHARC. Whether ABHD12 qualifies as a bona fide member to the endocannabinoid system remains to be established.

Partial Text

The discovery of CBRs and their endogenous ligands has greatly accelerated research on cannabinoid actions in the brain. Indeed, CB1R is among the most abundantly expressed and widely distributed GPCR in the brain (Herkenham et al. 1991) (Fig. 1). CB1R unlikely evolved merely to mediate the ‘bliss’ attributed to delta9-tetrahydrocannabinol (THC), the major psychoactive component of Cannabis sativa, nor does nature maintain protein synthesis just for reserve. Instead, the abundance of CB1R in specific types of neurones and its enrichment into pre-synaptic terminals throughout the brain strongly suggests that CB1R evolved specifically to mediate eCB signalling. It is now well established that in response to neural activity, eCBs are produced ‘on demand’ and released from post-synaptic neurones, then activate pre-synaptic CB1Rs in a retrograde manner, resulting in transient and long-lasting reduction of neurotransmitter release at various central synapses (Alger 2002, Freund et al. 2003, Piomelli 2003, Kano et al. 2009). Such a retrograde signalling mode has established a new concept how diffusible lipid messengers, by encaging their cognate GPCRs, can provide both short- and long-term fine-tuning of synaptic efficacy and neural activity. Electrophysiologists have found robust modulation of synaptic plasticity and thus introduced new terminology, such as depolarization-induced suppression of excitation (DSE), and depolarization-induced suppression of inhibition (DSI), both of which are best explained by short-term retrograde eCB signalling inhibiting synaptic release of glutamate and GABA respectively (Kano et al. 2009) (Fig. 2). The presence of molecules of the eCB system, such as the eCBs, CB1R, as well as enzymes involved in eCB metabolism of during neuronal development have been linked to neuronal proliferation, differentiation, migration, axon guidance and synaptogenesis (Bisogno et al. 2003, Keimpema et al. 2010, Argaw et al. 2011). Thus, the eCBs are intimately involved in the physiology of the nervous system.

Biosynthesis and release of 2-AG can be induced by either depolarization or activation of Gq-coupled GPCRs, typically group I metabotropic glutamate receptors (mGluR1 or mGluR5) or M1/M3 muscarinic acetylcholine receptors (Hashimotodani et al. 2005, Kano et al. 2009). In the hippocampus, 2-AG release is markedly enhanced by simultaneous depolarization and input via Gq-coupled receptors, a phenomenon where PLC-β1 is suggested to serve as the coincidence detector (Hashimotodani et al. 2005). The physiological significance of the coincidence detection is that such a mechanism allows simultaneous sensing of both pre-synaptic (transmitter release) and post-synaptic (membrane depolarization) activities through the Ca2+ dependency of PLC-β1 (Hashimotodani et al. 2005). In principle, PLC activity generates two second messengers, IP3 that triggers Ca2+ release from intracellular stores, and diacylglycerol (DAG) that classically activates protein kinase C. In brain regions endowed for eCB signalling, sn-2-arachidonoyl-containing DAG species are hydrolysed by sn-1-specific lipase (DAGL) to generate 2-AG, the major eCB in brain. Two sn-1-specific DAGL isoforms have been cloned, namely DAGLα and DAGLβ (Bisogno et al. 2003). The cellular expression of the two isoforms was demonstrated to closely reflect 2-AG biosynthesis and release. During neuronal development, localization of DAGLα and DAGLβ changes from pre- to post-synaptic elements, i.e. from axonal tracts in the embryo to dendritic fields in the adult, closely matching with the developmental changes in need for 2-AG synthesis from the pre- to the post-synaptic compartment (Bisogno et al. 2003). Overexpression of DAGLα in mouse neuroblastoma cells results in 2-AG accumulation, whereas knockdown of DAGLα by RNA interference blunts 2-AG production and prevents group I mGluR-stimulated production of this eCB (Jung et al. 2007). A docking platform for efficient coupling of glutamatergic signalling to 2-AG biosynthesis is provided by scaffolds such as Homer proteins that allow interaction of group I mGluRs with the PPxxF domain of DAGLα (Jung et al. 2007). Interestingly, DAGLβ, which based on recent gene ablation studies is not involved in 2-AG generation for retrograde signalling (Gao et al. 2010, Tanimura et al. 2010), also lacks the Homer-interacting PPxxF domain (Jung et al. 2007). Collectively, these studies suggest that DAGLα has specifically evolved to generate 2-AG for retrograde synaptic signalling. Two recent studies using DAGL-knockout (DAGL-KO) mice have provided strong additional support to this idea. DAGLα-KO mice were found to have marked (up to 80%) reductions in 2-AG levels in brain and spinal cord with concomitant decrease in AA levels, whereas DAGLβ-KO animals exhibited either no (Tanimura et al. 2010) or up to 50% reduction (Gao et al. 2010) in brain 2-AG levels. Importantly, several forms of retrograde eCB-mediated synaptic suppression, such as DSE and DSI, were absent from the tested brain regions (hippocampus, cerebellum and striatum) of DAGLα-KO mice but appeared intact in DAGLβ-KO mice brains (Gao et al. 2010, Tanimura et al. 2010). There was no evidence for compensatory changes of DAGLβ in DAGLα-KO mice or vice versa, nor was there evidence for abnormal expression patterns or protein levels of other molecular components of the eCB signalling machinery. Interestingly, eCB control of neurogenesis in the adult hippocampus was compromised in both DAGLα- and DAGLβ-KO mice, as well as in the subventricular zone of DAGLα-KO animals (Gao et al. 2010). In peripheral tissues such as the liver, DAGLβ seems to be the major isoform generating 2-AG (Gao et al. 2010). Collectively, the genetic studies suggest that DAGLα is the major biosynthetic enzyme generating 2-AG for retrograde signalling. It is interesting to note that in some brain regions highly enriched with the CB1R (such as the substantia nigra), immunohistochemistry reveals only sparse labelling with the DAGLα antibodies (Uchigashima et al. 2007, Kano et al. 2009, Tanimura et al. 2010). This would suggest that alternative biochemical routes for 2-AG biosynthesis (Sugiura et al. 2006, Kano et al. 2009) are utilized to generate the CB1R-activating eCB in brain regions with sparse DAGLα expression. Further studies with the DAGL-KO mice should shed more light on this issue.

In situ hybridization studies combined with immunohistochemical studies with properly validated antibodies have revealed the ultrastructural localization, molecular composition and synaptic organization of the apparatus needed for 2-AG-mediated retrograde signalling at various central synapses. An excellent review covering details of this topic is available (Kano et al. 2009) and therefore only the general aspects are briefly discussed here. The emerging picture is strikingly similar regardless of the brain region and mammalian species studied. At the glutamatergic hippocampal synapses of both rodents and humans, DAGLα is concentrated in dendritic spine heads around the post-synaptic density throughout the hippocampal formation, whereas CB1R is strategically situated at the pre-synaptic axon terminals on the opposite side of synaptic cleft (Katona et al. 2006, Kano et al. 2009, Ludányi et al. 2011). In the striatal medium spiny neurones, DAGLα and the Gq-coupled GPCRs (mGluR5 and M1) are all enriched on the somatodendritic surface (Uchigashima et al. 2007), whereas CB1R localization is pre-synaptic and the receptors are enriched on the GABAergic axon terminals and relatively low abundance on the corticostriatal glutamatergic terminals (Uchigashima et al. 2007). Relative enrichment of DAGLα on the post-synaptic membrane at glutamatergic synapses and CB1Rs at the inhibitory terminals instead of excitatory terminals seems to be one hallmark of retrograde eCB signalling at various central synapses (Fig. 2). Such an arrangement likely provides the neuroanatomical and physiological basis for observations where the stimulation thresholds required to evoke DSI are generally much lower than those needed to observe DSE (Kano et al. 2009, Yoshida et al. 2011).

By analogy to classical neurotransmission, the magnitude and duration of eCB signalling are tightly regulated by the balanced action between the enzymes that synthesize and hydrolyse these lipid messengers. The hydrolysis of eCBs is principally carried out by four enzymes belonging to the metabolic serine hydrolase family. Solid experimental evidence supports the primary role of fatty acid amide hydrolase (FAAH) in the inactivation of AEA both in vitro and in vivo (Ahn et al. 2009). Indeed, comprehensive inactivation of FAAH in rodents either by genetic or pharmacological means results in marked (>10-fold) elevations in brain AEA levels, and as a consequence, CB1R-dependent analgesia in various models of acute and chronic pain (Ahn et al. 2009). Importantly, no tolerance of CB1R function and behavioural responses takes place after chronic FAAH inactivation (Ahn et al. 2009, Schlosburg et al. 2010).

The major enzymatic route for 2-AG inactivation in brain is via hydrolysis generating AA and glycerol as the end products. MAGL was the first hydrolase implicated in 2-AG degradation both in vitro and in vivo (Dinh et al. 2002, 2004, Saario et al. 2004, Labar et al. 2010b). In many tissues and cell types, MAGL is detected both in soluble and membrane preparations. Traditionally, MAGL functions as a key lipolytic enzyme in the mobilization of lipid stores for fuel or lipid synthesis. Accordingly, MAGL was originally purified, and subsequently cloned from the adipose tissue (Karlsson et al. 1997, Labar et al. 2010b). A recent study has illuminated a pathophysiological, eCB-independent role for MAGL as well. Activity-based protein profiling (ABPP) of various cancer cells has identified MAGL overexpression as the key metabolic switch orchestrating cancer cell malignancy by redirecting lipids from storage sites towards biosynthesis of cancer promoting signalling lipids such as eicosanoids and lysophospholipids (Nomura et al. 2010). It was only after discovery of the CBRs and their endogenous ligands that MAGL was linked to the eCB system.

Long-term disruption of MAGL was achieved either by genetic ablation (MAGL-KO) or chronic treatment with the MAGL-selective inhibitor JZL184 (Chanda et al. 2010, Schlosburg et al. 2010), and in both studies the general outcome was similar. In contrast to the antinociceptive effects typically observed after acute MAGL inhibition (Kinsey et al. 2009, Long et al. 2009), animals with chronic MAGL blockade had normal pain responses in several pain models. Moreover, these mice lacked several behavioural responses, such as hypothermia, hypomotility, or catalepsy, typically observed after administration of cannabinoid agonists such as THC. In addition, MAGL-KO mice had decreased body weight, thus resembling the lean phenotype of CB1R-KO mice or animals treated with CB1R antagonists such as rimonabant (Di Marzo et al. 2001). Moreover, eCB-mediated short-term synaptic suppression was compromised (Schlosburg et al. 2010).

In 2005, the first homology model of MAGL was presented based on 3D structure of chloroperoxidase from Streptomyces lividans as the template (Saario et al. 2005). As a result of poor sequence homology between MAGL and chloroperoxidase, the homology model offered limited insights into the overall structure and organization of MAGL protein. However, as the central core of the α/β hydrolase superfamily members is highly conserved, the model offered first insights into the MAGL active site with the catalytic triad (S122-D239-H269), previously identified based on mutagenesis studies (Karlsson et al. 1997). The model also suggested that two cysteine residues (C208 and C242) were located within a close distance from the active site and it was suggested that one or both of these cysteines were potential targets of the maleimide-based inhibitor NAM (Saario et al. 2005). Site-directed mutagenesis studies have provided experimental support for these predictions (Labar et al. 2010a,b).

ABHD6 is a newly discovered post-genomic protein and relative little is known on its physiological functions. High expression of ABHD6 has been reported in certain forms of tumours suggesting that ABHD6 might serve a new diagnostic marker of these tumours (Li et al. 2009, Max et al. 2009). However, its knockdown in cancer cells did not inhibit tumour cell growth (Max et al. 2009). Based on hydropathy analysis and biochemical studies, ABHD6 appears to be an integral membrane protein (Blankman et al. 2007) that possesses typical α/β-hydrolase family fingerprints such as the lipase motif (GHSLG) and a fully conserved catalytic triad (postulated amino acid residues S246-D333-H372). The active site is predicted to face cell interior. Such an orientation suggests that ABHD6 is well suited to guard the intracellular pool of 2-AG (Fig. 4). A recent study (Marrs et al. 2010) provided the first evidence to link ABHD6 as a bona fide member to the eCB system by demonstrating that ABHD6 controls the accumulation and efficacy of 2-AG at the CBRs. In neurones, ABHD6 was detected both at mRNA and protein level and its pharmacological inhibition led to activity-dependent accumulation of 2-AG in neuronal cultures. In the mouse cortex, ABHD6 localization in neurones was mainly post-synaptic, often juxtaposed with the pre-synaptic CB1Rs. ABHD6 was expressed also in many principal glutamatergic neurones, some GABAergic inteneurones, as well as astrocytes but not in resident microglia (Marrs et al. 2010). However, in a microglial cell line (BV-2), ABHD6 was enriched in mitochondrial fraction and its knockdown reduced the hydrolysis of 2-AG in intact cells with concomitant sensitization of 2-AG-stimulated and CB2R-dependent cell migration. In murine cortical slices, pharmacological ABHD6 inhibition facilitated the induction of CB1R-dependent long-term synaptic depression by otherwise subthreshold stimulation. Interestingly, CB1R-dependent short-term synaptic depression (DSI or DSE) remained unaltered following ABHD6 inhibition. The postulated physiological role of ABHD as a regulator of 2-AG levels at the site of production of this eCB is schematically illustrated in Figure 4.

Like ABHD6, ABHD12 is a recently identified post-genomic protein with poorly defined physiological function. Its potential role as a brain 2-AG hydrolase was revealed using ABPP with mouse brain proteome and it was estimated that at the bulk brain level ABHD12 accounts for approx. 9% of total 2-AG hydrolase activity (Blankman et al. 2007). Currently, 2-AG is the only known substrate for ABHD12 but it is possible that the enzyme utilizes also other substrates. As far as we are aware, 2-AG hydrolase activity is the only feature so far potentially linking ABHD12 to the eCB system. Based on hydropathy analysis and biochemical data, ABHD12 appears to be an integral membrane protein whose active site is predicted to face the lumen/extracellular space (Blankman et al. 2007). Typical α/β-hydrolase domain protein fingerprints, including the lipase motif (GTSMG) and catalytic triad (predicted amino acid residues S148-D278-H306), are fully conserved both in rodent and human ABHD12 primary structure. It was recently reported that mutations in the ABHD12 gene that are predicted to compromise catalytic activity severely, are causally linked to a neurodegenerative disease called PHARC (polyneuropathy, hearing loss, ataxia, retinitis pigmentosa, and cataract) (Fiskerstrand et al. 2010). Based on this surprising finding, the authors suggested that ABHD12 must perform essential physiological functions in the nervous system and that PHARC may serve as a human ABHD12 KO model (Fiskerstrand et al. 2010). However, tissue 2-AG levels and metabolism were not explored in this study and thus the possible connection to the eCB system clearly requires further studies. Interestingly, ABHD12 transcripts are highly expressed in various brain regions and specifically in microglia, but are also abundant in related cell types such as macrophages and osteoclasts (Fiskerstrand et al. 2010). There is also increasing appreciation for the potential role of the eCB system, and microglial CB2Rs in particular, as regulators of immune function in the CNS (Cabral et al. 2008). Thus, as schematically illustrated in Figure 5, ABHD12 appears to be well suited to guard 2-AG-CB2R signalling in brain microglia as well as in related cell types of peripheral tissues.

Delicate physiological regulatory mechanisms have evolved to maintain the balance between the ‘on demand’ biosynthesis and degradation of the signalling competent pool of 2-AG in the CNS. Significant progress has been made during the last few years in dissecting out details of the synaptic architecture and molecular components of the eCB system intimately involved in the physiology and function of the nervous system. Recent milestones include the elucidation of MAGL crystal structure, as well as the availability of selective pharmacological and genetic tools to specifically target key enzymes involved in the generation (DAGLα) and degradation (MAGL) of 2-AG. Insights into MAGL crystal structure open new avenues to exploit MAGL function in further detail. From the perspective of rational drug design, the shape of the hydrophobic tunnel leading to the catalytic site suggests ‘a high druggability of the protein’ (Bertrand et al. 2010), offering possibilities for further development of potent and specific MAGL inhibitors. However, one may ask the question whether there is further need for such inhibitors as potential therapeutics. The answer might be yes if we consider the recently disclosed pathophysiological role of MAGL in promoting cancer cell malignancy (Nomura et al. 2010). In addition, specific pharmacological tools are needed to explore MAGL function further. On the other hand, the answer might be no if we consider MAGL inhibitors as potential therapeutics to alleviate pain, for example. Direct CBR1 agonists like THC produce analgesia in various pain models but their therapeutic use is limited because of undesired psychotropic effects. Prolonging and amplifying the eCB tone by inhibiting their enzymatic metabolism has therefore emerged as an alternative strategy to manipulate the eCB system for possible clinical benefit (Lambert & Fowler 2005, Mackie 2006, Hohmann 2007, Saario & Laitinen 2007, Petrosino & Di Marzo 2010). We have just learnt that chronic MAGL inhibition leads to 2-AG overload and functional antagonism of the eCB system, both at the molecular and behavioural level (Lichtman et al. 2010). Would partial MAGL inhibition result in pain relief without desensitization of the eCB system? Might FAAH be a better molecular target of the eCB system for pain relief? This reasoning is supported by findings that the analgesic effects of a FAAH inhibitor persist after long-term administration and no apparent desensitization of CB1R function takes place after chronic FAAH inactivation (Ahn et al. 2009, Schlosburg et al. 2010). We know very little on the physiological or pathophysiological roles of ABHD6 and ABHD12. The postulated causal link between ABHD12 mutations and the neurodegenerative disease PHARC should stimulate further research that will clarify whether ABHD12 is a molecular component of the eCB system. As always, new important findings tend to raise more questions than to provide final answers. This in mind it appears that researchers on the eCB field will be busy also during the forthcoming years.

There is no conflict of interest.