Not Too Excited? Thank Your EndocannabinoidsBradley E. Alger1
Endocannabinoids can mediate neuroprotection, but it is not known how. In this issue of Neuron, Monory et al. use mutant mice and localized viral targeting to produce conditional knockouts of the cannabinoid CB1 receptor. They show that protection against kainic acid-induced seizures and cell death is conferred by CB1Rs on hippocampal glutamatergic nerve terminals.
Epileptic seizures reflect states of pathological hyperexcitability and hypersynchronous activity in large neuronal networks. Broadly speaking, seizures arise from an imbalance of two fundamental antagonistic neuronal motive forces–excitation and inhibition–toward excitation. But the underlying mechanisms may be very complex and, in addition to alterations in the strength of excitatory and inhibitory chemical synapses, may involve electrical gap junctions, neuronal network oscillations, and rewiring of the neuronal circuits. The hippocampus is one of the most seizure-prone brain regions, perhaps because it typically rests near the tipping point of the balance and is susceptible to numerous forms of plasticity. Whatever their etiology, seizures are highly disruptive to normal brain functions, and if severe and prolonged, can lead to very bad outcomes, including neuronal cell death. Intrinsic biological mechanisms that protect against seizures are therefore of great theoretical and practical interest.
Endogenous cannabinoids ("endocannabinoids") are the natural agonists of membrane-bound, G protein-coupled receptors that mediate the actions of drugs, such as marijuana, derived from the cannabis plants. The principal cannabinoid receptor subtype in the CNS, CB1R, is predominantly localized on or near synaptic terminals, and its activation inhibits synaptic transmitter release. The two major endocannabinoids are arachidonyl-ethanolamide (anandamide) and 2-arachidonyl glycerol (2-AG). They are produced by neuronal enzymatic activity and generally serve as intercellular messengers, often traveling in the "retrograde" direction to the incoming synaptic input. CB1Rs are widely dispersed throughout the brain in specific association with well-defined cell types in the different regions, and this accounts for the variety of behavioral effects caused by the exogenous cannabinoids. Several years ago, Panikashvili et al. (2001) reported that experimental closed-head injury produced an elevation in 2-AG and that exogenous administration of 2-AG reduced the brain edema and hippocampal cell death associated with such injuries. This was direct in vivo evidence for a neuroprotective effect of an endocannabinoid. In this issue of Neuron, Monory et al. (2006) now ask and answer novel questions about the cellular mechanisms of endocannabinoid-mediated neuroprotection.
In an earlier investigation, Lutz, Marsicano, and colleagues (Marsicano et al., 2003) reported that mice lacking CB1 (CB1−/−) experienced kainic acid (KA)-induced seizures that were much more severe than those experienced by wt or heterozygotic CB1+/− animals. By inference, the presence of CB1 was protective. Using a Cre/LoxP system in which Cre recombinase was under the control of the CamKIIα gene (not expressed in interneurons), they then created conditional knockout mice in which CB1Rs were deleted in all principal glutamatergic neurons of the forebrain (cortical and subcortical), but were spared in GABAergic interneurons (and cerebellum). The KA-induced seizures experienced by the conditional knockouts were as severe as those in the full CB1−/− mice, and neuronal cell death was also greater in these animals. Thus, the CB1Rs associated with forebrain principal neurons were implicated as major factors for endogenous neuroprotection. As a corollary, the remaining CB1Rs appeared to offer no neuroprotection. Moreover, because the hippocampus is the region that is most vulnerable to KA excitotoxicity (see Ben-Ari and Cossart, 2000), the results suggested that the CB1Rs expressed by hippocampal glutamatergic neurons might be key to neuroprotection. There was no direct evidence for either of these latter two conclusions, however.
In expanding upon their previous work, Monory et al. (2006) turned first to conditional CB1 deletions. They have created two new lines of mice: one lacking CB1 only in cortical (including hippocampal) glutamatergic neurons (Glu-CB1−/−) and one lacking CB1 in forebrain GABAergic interneurons (GABA-CB1−/−). They compared these mice with those lacking CB1 in all forebrain glutamatergic neurons (now referred to as CamKII-CB1−/−). A major finding was that the Glu-CB1−/− mice have the same enhanced seizure vulnerability as the CB1−/− and CamKII-CB1−/− mice, which is consistent with the idea that CB1R on cortical glutamatergic cells are key elements in the defense against KA-induced excitotoxicity. In contrast, the GABA-CB1−/− mice had no seizure-related deficits; their seizure susceptibility was the same as the wt mice. Moreover, the anticonvulsant benzodiazepine, diazepam (which enhances GABA-mediated inhibition), was equally effective in protecting against KA-induced seizures in wt and CB1−/− mice, suggesting that the CB1 system does not influence GABAergic transmission during these seizures.
Is the hippocampus the main recipient of the endocannabinoid-mediated neuroprotection, and if so which neurons are involved?
Using a combination of immunohistochemical staining and double in situ hybridization for CB1 and the vesicular glutamate transporter, VGluT1, Monory et al. identified the mossy cell, a glutamatergic cell type in the dentate gyrus, as a focal point, by virtue of its relatively high and consistent expression of CB1 and suspected role in epileptogenesis (Ratzliff et al., 2002). Mossy cells (not be confused with mossy fibers, which are the granule cell axons) provide a strong excitatory drive to the granule cells, and their synaptic output is suppressed by cannabinoids. To see whether these cells are involved in behavioral seizure protection, the authors injected a Cre-expressing adeno-associated virus (AAV-Cre) into the mossy cell region of the dentate gryus of CB1-floxed mice. This produced a CB1 deletion confined largely to this region and nearby CA1 and CA3. Amazingly, these mice suffered significantly worse seizures when treated with KA than wt or AAV-GFP virus-treated animals. This is the first solid evidence that CB1Rs confined to a limited cell group can play a significant role in epileptogenesis. Hence, the powerful combination of molecular lesioning methods employed by Monory et al. strongly supports the emerging picture that a strong epileptogenic influence causes excessive output of glutamate, which, if unchecked, leads to development of seizures and widespread neuronal death. Normally, however, glutamate output and its deleterious sequelae are limited by the inhibitory actions of endocannabinoids on glutamatergic terminals. These findings fuel hope of capitalizing on the endocannabinoid-mediated neuroprotection to develop novel therapies for treatment of seizures or stroke.
The finding of unaltered seizure susceptibility in GABA-CB1−/− mice, while not necessarily a paradox, is a bit unexpected. Epileptic activity can readily be induced in experimental models by decreasing the strength of GABA inhibition, and this mechanism may contribute to the human disease. As noted earlier, anticonvulsants often enhance GABAergic responses. The naive expectation would be that activation of CB1Rs on GABAergic terminals, and the resultant suppression of GABA transmission, would exacerbate seizures. On these grounds, CB1 absence in the GABA-CB1−/− mice would be protective.
Why is removal of the CB1Rs on GABAergic terminals not associated with an obvious seizure phenotype?
For one thing, the CB1Rs are not associated with all GABAergic interneurons, but only a well-defined subset: the cholescystokinin (CCK)-containing basket cells (Freund et al., 2003). The other major basket cell type, the parvalbumin (PV)-containing cells (PV and CCK groups are mutually exclusive), do not express CB1, and evidently the activity of these and other non-CB1-expressing interneurons was sufficient to account for the anticonvulsant actions of diazepam. The GABA-CB1−/− animals will serve as most interesting subjects for unraveling the behavioral functions of the prominent GABAergic eCB system.
Obvious questions stemming from the Monory et al. (2006) study include whether endocannabinoids are also protective in other seizure models, and what exactly explains the neuroprotection. Inhibition of glutamate release is a likely mechanism, and the CamKII-CB1−/− mice do release glutamate more readily than wt mice (Marsicano et al., 2003), but the mechanism of CB1R-mediated neuroprotection has not been unambiguously demonstrated. Some conclusions are complicated by species and preparation differences (e.g., Mechoulam and Lichtman, 2003, for a brief overview). There are other subtle issues as well.
Where does the neuroprotective endocannabinoid come from, and what is its identity?
Monory et al. (2006), carefully checked to see that their GABA-CB1−/− animals lacked CB1Rs on the interneuron terminals by testing for the reduction of inhibitory responses by neuronally released endocannabinoids ("DSI") and showing that it was absent. Surprisingly, they do not report having done the analogous test on the Glu-CB1−/− or CamKII-CB1−/− animals. Reduction of excitatory responses by neuronally released endocannabinoids (DSE), in other words, should have been absent. One must assume that they thought of doing this test, so its omission is noteworthy. Deletion of CB1R on the glutamate terminals was confirmed by showing that a synthetic CB1R agonist did not suppress excitatory transmission in CamKII-CB1−/− or Glu-CB1−/− animals (although it did in wt). The question, therefore, is not whether the mutants are the genuine articles, but how and how readily the endocannabinoids that normally activate CB1Rs on glutamate terminals are mobilized. Unlike DSE in the cerebellum, which is prominent and easily induced by physiologically relevant stimuli (Kreitzer and Regehr, 2001), DSE in the hippocampus is not generated by stimuli that produce DSI (Ohno-Shosaku et al., 2002) and at best only modestly reduces glutamate release. Endocannabinoids are generated by both Ca2+- and G protein-coupled receptor pathways, which may differ biochemically (Edwards et al., 2006). It will be important to work out the biochemical and indeed the cellular source of neuroprotective endocannabinoids.
What is the neuroprotective endocannabinoid?
Glutamate stimulates production of 2-AG but not anandamide in the hippocampus (Stella et al., 1997). Moreover, a recent report finds CB1Rs expressed on the hippocampal glutamatergic axons and a key biochemical component of the 2-AG synthetic pathway positioned on dendritic spines directly across from the terminals (Katona et al., 2006). The 2-AG system is therefore ideally positioned and poised to mediate neuroprotection. The prediction is that 2-AG levels should be increased during seizures. Yet direct endocannabinoid measurements by Marsicano et al. (2003) did not confirm this expectation: KA treatment increased anandamide but not 2-AG levels. It seems there are still subtleties regarding endocannabinoids, their regulation and function, that provide ample opportunities for future discoveries.
Finally, it was once cynically remarked that the lesion approach to understanding the brain is like trying to learn how a television set works by disabling one tube at a time with a hammer (those unfamiliar with "tube" in this context should consult an older person). Seeing the picture dissolve into a mass of wavy lines following the breaking of a particular tube might lead to the conclusion that it had functioned primarily as the "wavy line suppressor"; the intricacy of the truth obscured by the dramatic effects caused by the malfunction. It is worth bearing in mind that, although the hammers and tubes grow smaller, such concerns do not entirely go away.
Source: ScienceDirect - Neuron : Not Too Excited? Thank Your Endocannabinoids