Endocannabinoid Identification in the Brain: Studies of Breakdown Lead to Breakthrough, and There May Be NO Hope

See allHide authors and affiliations

Science's STKE  08 Nov 2005:
Vol. 2005, Issue 309, pp. pe51
DOI: 10.1126/stke.3092005pe51


Endocannabinoids are a class of fatty acid derivatives defined by their ability to interact with the specific cannabinoid receptors that were originally identified as the targets of Δ9-tetrahydocannabinol (Δ9-THC), the psychoactive component of cannabis. Endocannabinoids have been implicated in a growing number of important physiological and behavioral events. A full understanding of the functions of endocannabinoids will involve knowing which ones are active, and how they are produced, during any given physical event. However, studying these small lipids in the brain presents many technical challenges. New selective pharmacological tools promise to be very useful in unraveling the complexities of endocannabinoid signaling, but parallel developments from the investigation of the cellular neurophysiology of the endocannabinoid systems highlight the difficulties remaining.

The communications theorist Marshall McLuhan famously observed that "the medium is the message"; in other words, the means by which a signal is delivered is more important than the specific information it carries. Although he had mass media such as television in mind, a less hyperbolic form of his vision could be appropriate for the signaling mediated by the lipid messengers called endocannabinoids. Would a better understanding of this medium tell us more about its messages? Efforts to identify which endocannabinoid is "the" retrograde messenger at any particular synapse will help answer this question. Recent studies indicate that newly developed pharmacological agents (1) will be very valuable in these efforts, but even sharper tools may be needed if these slippery signals are to be pinned down.

The discovery of Δ9-tetrahydocannabinol (Δ9-THC) as the psychoactive ingredient of cannabis led to the characterization (2) and cloning (3) of the heterotrimeric guanine nucleotide–binding protein (G protein)–coupled cannabinoid receptor CB1 in the central nervous system. A second receptor, CB2, is predominant in the periphery (4). Identification of the arachidonic acid derivative anandamide as the first endogenous agonist for CB1 (5) sparked the current widespread excitement about endocannabinoids. CB1-mediated actions, formerly interpreted within the important but restricted framework of drug abuse, are now recognized as contributing to the regulation of learning and memory, as well as feeding, anxiety, and pain. Understanding endocannabinoid-mediated signaling has become correspondingly essential.

Many Candidate Cannabinoids

The discovery of a second endocannabinoid agonist for CB1, 2-arachidonyl glycerol (2-AG) (6, 7), along with a multitude of other CB1-binding ligands [see (8) for a review], ushered in complexities in the interpretation of endocannabinoid signaling. A neurotransmitter system that performs multiple functions generally does so through multiple receptor subtypes and second messengers, not through numerous receptor agonists. Thus, the plethora of endocannabinoids is confusing. The existence of more than one brain CB receptor might help explain things. Indeed, unexpected differences between the actions of the two "gold-standard" CB1 antagonists SR141716A [rimonabant (9)] and AM-251, together with residual cannabinoid agonist–mediated effects in a CB1 knock-out mouse, suggest this possibility (10). The implied "CB3" receptor has not been further characterized, and its existence is controversial (11), although there are other examples of putative cannabinoid receptors that are pharmacologically but not molecularly identified (4).

A partial solution to the many-ligands problem is that some of these molecules are not, or not exclusively, CB1 agonists. Anandamide, for example, is also an agonist at a different receptor: the vanilloid, VR1, receptor (12). Perhaps it functions as an "endovanilloid" rather than an endocannabinoid, although this conclusion is premature.

Despite the unresolved issues, neurochemical assays point to anandamide and 2-AG as the major endocannabinoids in the brain. Stimulation in the striatum causes a selective increase in anandamide production, although not in the production of 2-AG or other endocannabinoids (13). The converse occurs in cultured cortical neurons and hippocampal slices, where stimulation causes Ca2+-dependent 2-AG, but not anandamide, production (14). The major synthesis pathway seems to involve the sequential action of phospholipase C (PLC), which produces diacylglycerol (DAG) from phosphatidylinositol 4,5-bisphosphate, followed by the conversion of DAG to 2-AG by DAG lipase (8). Increases in neurochemically measured 2-AG abundance can be prevented by inhibiting either PLC or DAG lipase (14).

Neurophysiology of Endocannabinoids

Physiological evidence for endocannabinoid signaling came through the study of depolarization-induced suppression of inhibition [DSI; see (15) for a review], a process discovered in the cerebellum (16) and hippocampus (17). DSI was the first rapid retrograde signaling process to be well characterized neurophysiologically. In DSI, a chemical signal is released from principal neurons (pyramidal cells in the hippocampus and Purkinje neurons in the cerebellum) and transiently inhibits evoked inhibitory postsynaptic currents (eIPSCs) mediated by γ-aminobutyric acid (GABA) release from the terminals of intrinsic interneurons. That is, the DSI signal goes against the normal flow of synaptic traffic across the synaptic cleft to receptors on presynaptic nerve terminals, where it reduces synaptic transmitter release. The conclusion that an endocannabinoid mediates DSI follows from the observations that CB1 agonists mimic DSI, CB1 antagonists abolish it (18), DSI is absent in the CB1 knock-out mouse (19, 20), and CB1 receptors are situated precisely on the synaptic terminals of the DSI-sensitive inhibitory interneurons (21, 22). The endocannabinoid realm expanded with the finding that an endocannabinoid also inhibits the release of the excitatory neurotransmitter, glutamate, in a complementary process called depolarization-induced suppression of excitation (DSE) (23).

A steep increase in intracellular Ca2+ concentration ([Ca2+]i) in principal neurons triggers DSI and DSE (16, 17, 24); however, several G protein–coupled receptors (GPCRs), including dopamine D2 receptors (14), metabotropic glutamate receptors (mGluRs) (20, 25), and muscarinic acetylcholine receptors (mAChRs) (26, 27), initiate endocannabinoid release through biochemical processes that are not activated by changes in [Ca2+]i alone. GPCR-dependent endocannabinoid signaling persists for as long as the receptor is activated and, in the case of mGluRs, can produce permanent (and ultimately CB1-independent) changes in synaptic transmission if CB1 stimulation is sufficiently prolonged (28, 29).

Endocannabinoid Identification

Which endocannabinoid is the retrograde messenger? The logic of transmitter identification at conventional chemical synapses between neurons or from nerve to muscle is a staple of introductory neuroscience courses. The candidate transmitter molecule should be packaged in vesicles in presynaptic terminals, released in a Ca2+-dependent way, and be able to be collected from the extracellular space. Appropriate pharmacological agents should mimic or block its effects, and the requisite biochemical machinery (synthetic and degradative enzymes, precursors, and so forth), as well as specific membrane-bound receptors localized in postsynaptic elements, are expected to be nearby.

Lipid messengers, particularly endocannabinoids, pose special problems for the usual program of transmitter identification. They are cleaved from ubiquitous membrane phospholipid precursors and are not stored in vesicles, and no morphological specialization marks their site of release. Anandamide and 2-AG are hydrophobic and thus not readily collected extracellularly. Typically they are measured by membrane lipid extraction followed by liquid or gas chromatography and mass spectroscopy. Although sensitive, these techniques lack the temporal and spatial resolution needed for correlation with cellular physiological actions.

Breakthrough Through Breakdown?

Fatty acid amide hydrolase (FAAH), which in membrane preparations can degrade both anandamide and 2-AG, specifically targets anandamide in intact cells. Inhibition of FAAH enhances the effects of exogenously applied anandamide without influencing those of 2-AG (30). DSI is immune to FAAH inhibition (30), indicating that anandamide does not mediate DSI. Anandamide and 2-AG are both substrates for cyclooxygenase-2 (COX-2) (31), which generates a series of glyceryl compounds from 2-AG. COX-2 inhibitors prolong DSI, offering indirect evidence that 2-AG mediates DSI (30). Interestingly, endocannabinoids are endogenous analgesics (32), and therefore the increases in 2-AG may contribute to the antinociception caused by COX-2 inhibition.

Direct evidence that 2-AG may be the hippocampal DSI endocannabinoid comes from a new study by Makara et al. of 2-AG metabolism (1). Monoglyceride lipase (MGL) is a primary mediator of 2-AG degradation (33). MGL decreases 2-AG abundance in cortical cells and is localized to synaptic nerve terminals near hippocampal pyramidal cells. The synthesis of two specific and irreversible inhibitors for MGL (1) increased the concentration of 2-AG without affecting that of anandamide in rat forebrain slice cultures. Crucially, MGL inhibition prolonged hippocampal DSI, as would be expected if 2-AG were the retrograde signal, and MGL inhibitors extended its lifetime. Hence, the data strongly support the 2-AG hypothesis. The MGL inhibitors are not only valuable new tools for neurophysiologists but will certainly attract the earnest attention of drug companies seeking to exploit the therapeutic potential of endocannabinoids.

Is It Over?

Is 2-AG the hippocampal endocannabinoid? Unfortunately, the case is not yet closed. One problem centers on the biochemical pathway for the synthesis of 2-AG, which, as noted earlier, is thought to proceed through hydrolysis of DAG that has been produced by PLC (8). Neither the PLC inhibitor U73122 nor the DAG lipase inhibitor RHC 80267 affects DSI or other putative 2-AG–mediated physiological processes in hippocampal slices (29, 34). Both inhibitors do prevent the induction of inhibitory long-term depression of inhibitory synaptic responses (29, 34), and RHC 80267 abolishes mAChR-induced endocannabinoid mobilization (34). Thus, a simple lack of efficacy cannot explain the drugs’ inability to reduce DSI.

The puzzle is further complicated by a report that mGluR or mAChR stimulation does not generate endocannabinoid synthesis in hippocampal neurons cultured from mice lacking PLCβ1 (35). In wild-type mice, GPCR activation during a slight increase in [Ca2+]i suggested that the Ca2+-dependent PLC isoform PLCβ1 is a "coincidence detector" for the production of endocannabinoids. A sensitive bioassay confirmed that GPCR agonists increase DAG levels in wild-type neurons, so the critical PLCβ1 is evidently neuronal, as would be expected if 2-AG were being produced. Surprisingly, DSI, the first response thought to be mediated by 2-AG, is absolutely normal in PLCβ1−/− mice. Another Ca2+-dependent pathway (36) might be responsible for 2-AG synthesis in DSI, although this pathway should also be sensitive to DAG lipase inhibition. Endocannabinoid mobilization must occur in less than 200 ms to account for DSI (37), and a future challenge will be to determine whether 2-AG synthesis can take place on this time scale.


Just as DSI was beginning to look like the unusual, rather than the prototypical, endocannabinoid-mediated response, because of its independence from PLC and DAG lipase, a previously discarded candidate DSI messenger has been resurrected, and this could change everything (38) (Fig. 1). Reportedly, either application of an antagonist of nitric oxide (NO) synthesis or of an NO scavenger inhibits hippocampal DSI. Guanylate cyclase is often an effector of NO actions, and a guanylate cyclase inhibitor also prevented DSI. The implication is that the suppression of eIPSCs by NO is blocked by CB1 antagonists, placing NO somewhere upstream of CB1 activation. As to whether the actual retrograde messenger would be NO or the endocannabinoid, two scenarios are possible: Both might be synthesized and released from the pyramidal cell and interact to cause CB1 activation. A more radical notion is that NO from the pyramidal cell stimulates endocannabinoid production within the interneurons. The endocannabinoid would then activate interneuronal CB1 receptors; in other words, CB1 would be an autoreceptor. NO would then be the true retrograde DSI signal, and the endocannabinoid would act as a local effector. This hypothesis is suggested by the finding that in some neocortical interneurons, Ca2+ influx triggers the synthesis of an endocannabinoid that activates somatic CB1 autoreceptors (39). The postulated role of NO in DSI is intriguing, but much work lies ahead to see if it can really account for all of the previous data.

Fig. 1.

Summary of new data on endocannabinoids in hippocampal suppression of inhibition. (A) Current thinking supports the model that 2-AG is synthesized and released from the pyramidal cell (PYR, top) by at least two distinct pathways: One is initiated by an influx of Ca2+ through voltage-gated calcium channels (the DSI mechanism), whereas the other, initiated by the activation of GPCRs, does not require a large increase in [Ca2+]i. The DSI mechanism is independent of PLC and DAG lipase. Neurochemical evidence indicates that both PLC and DAG lipase are implicated in 2-AG synthesis activated by G proteins, and the general scheme (solid arrow) is corroborated by a mouse lacking PLCβ1, but not by pharmacological experiments (dotted arrows). CB1 cannabinoid receptors are located in the presynaptic nerve terminal membrane of GABAergic interneurons (INs) that synapse on the pyramidal cell. The activation of CB1 can reduce GABA release by inhibiting Ca2+ influx through voltage-gated N-type Ca2+ channels, and may also antagonize vesicular release by direct effects on the release process. The degradative enzyme for 2-AG, MGL, resides in the membrane and cytosol of the interneuron, where its inhibition will allow 2-AG to activate CB1 for a longer period. (B) A new hypothesis states that an increase in [Ca2+]i in the pyramidal cell activates nitric oxide synthase (NOS) and that the NO produced diffuses to the interneuron, where it activates guanylate cyclase (GC). Cyclic GMP (cGMP) initiates 2-AG synthesis through an unknown mechanism. The local 2-AG activates the CB1 receptors. As in (A), MGL inhibitors will prolong the action of 2-AG.

The Future

Progress in understanding endocannabinoid signaling will be aided by the new MGL inhibitors (1). As recent work on PLCβ1 and NO shows, however, the uncharted territory still hides major surprises. The synthetic pathways for the physiologically relevant 2-AG are unclear, as is the relation between endocannabinoid synthesis and release. Sorting out the details will demand new molecular approaches. Optical techniques such as the use of caged anandamide (37) will permit precise manipulation of the systems at the cellular and subcellular levels. The ability to visualize the actual translocation of an endocannabinoid from its site of production to the CB1 receptors would be very helpful. Detectors that can sense and identify endocannabinoids in situ in real time are a dream, but maybe not an impossible one. There is much to be done, and the apparent success in developing an anti-obesity drug by manipulating the endocannabinoid system (40), as well as the potential for treating smoking and diabetes, suggest that a deeper understanding of this medium of neuronal communication may prove very rewarding.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
View Abstract

Navigate This Article