Effects of cannabinoids in in vitro and in vivo models of epilepsy

Although clinical data testing cannabinoids as antiepileptic agents in humans remains limited, cannabinoid receptor agonists and antagonists have been directly tested for effects on various aspects of epileptiform activity in a great many in vitro and in vivo models of epilepsy.

For example, one of the earlier studies to specifically implicate the CB1Rs in anticonvulsant action of cannabinoids used an electroshock model of a secondarily generalized seizure. In this model, a corneal shock is used to produce hind limb extension in rodents. Intraperitoneal injection of either WIN55,212-2 or Δ9-THC produces dose dependent inhibition of this phenomenon in a manner that can be prevented by prior injection of a CB1R antagonist (Wallace et al., 2001). Additional work revealed that a CB1R antagonist alone can reduce seizure threshold, suggesting a possible role for cannabinoid tone (Wallace et al., 2002). This later result seems consistent with a recent case study indicating that rimonabant produced partial seizures in an epileptic patient that had been seizure free for many years (Braakman et al., 2009).Further, in 2003 (Wallace et al.) reported that treatment of rats with either Δ9-THC or the synthetic CB1R agonist WIN55212-2 essentially eliminated seizures in a pilocarpine model of epilepsy. Notably, these compounds were more effective at preventing seizures in pilocarpine treated rats than two conventional antiepileptics, phenytoin and phenobarbital. Conversely, application of a CB1R antagonist increased both seizure duration and frequency in epileptic but not in control animals. Finally, these authors reported increased expression of CB1R in hippocampal membranes (see alsoFalenski et al., 2007; Falenski et al., 2009), and increased levels of 2-AG during seizures.

Later work with the pilocarpine model revealed that WIN55,212-2, AEA, and 2-AG reduced the frequency of both spontaneous and miniature excitatory postsynaptic currents, but only in epileptic tissue. WIN55,212-2 was also shown to inhibit bursts of EPSCs observed in granule cells secondary to photolysis of glutamate in a CB1R dependent manner, while population discharges produced by antidromic activation of mossy fibers were similarly effected (Bhaskaran & Smith, 2010a). These data suggested the possibility that sprouted mossy fibers may have heightened sensitivity to cannabinoid agonists in this model. Additional work from this group, also noted that AEA can enhance glutamate release in the pilocarpine model in the presence of a CB1R antagonist via apparent activation of TRPV1 receptors. Consistent with this observation, the TRPV1 agonist capsaicin enhanced miniature and spontaneous frequency of excitatory postsynaptic currents only in mice that developed epilepsy (Bhaskaran & Smith, 2010b).

Additional intriguing effects of cannabinoids have been noted in an in vitro models of acquired epilepsy and status epilepticus as observed in hippocampal neuronal culture. For example, in 2006 (Blair et al.) used this preparation to demonstrate that low μM concentrations of WIN55,212-2 can produce CB1R dependent anticonvulsant effects against both spontaneous and recurrent epileptiform discharges, and against status epilepticus induced by application external solution containing low concentrations of magnesium. Again, in this preparation, cannabinoid agonists were more effective than conventional antiepileptics phenytoin and pentobarbital. Additional work with this model showed that 2-AG and methanandamide also produce dose and CB1R dependent inhibition of status epilepticus as produced by low magnesium (Deshpande et al., 2007a), that application of a CB1R antagonist alone will produce status epilepticus in cultures that display spontaneous recurrent epileptiform discharges one day after a three hour exposure to low magnesium, but not in control cultures (Deshpande et al., 2007b), and that prolonged exposure of magnesium treated cultures to WIN55,212-2 will not only down regulate CB1R expression and negate its anticonvulsant effects, but indeed will produce a dose dependent increase in the frequency of spontaneous recurrent epileptiform discharges (Blair et al., 2009). These results simultaneously reinforce some aspects of the findings with the pilocarpine model, and yet also begin to highlight some of the difficulties that will likely complicate any straightforward use of cannabinoid receptor agonists as therapeutic agents in human epilepsy.

Recent years have also brought interesting results regarding the effects of cannabinoid receptor agonists on threshold for seizures as induced in mice by injection of the GABA antagonist pentylenetetrazole (PTZ). Specifically, acute administration of ACPA increased threshold for PTZ induced seizures while administration of the CB1R antagonist AM251 effectively decreased seizure threshold and reversed the anticonvulsant effect of ACPA (Shafaroodi et al., 2004). Intriguingly, this study also reported that the opioid receptor antagonist naltrexone produced an additive effect with low doses of AM251 in blocking the anticonvulsant action of ACPA. Interestingly later work with this model indicated that extremely low doses of AM251 actually potentiated rather than blocked anticonvulsant action of another CB1R agonist ACEA (Gholizadeh et al., 2007), and that similar effects could be produced by L-arginine, a nitric oxide precursor (Bahremand et al., 2009).

While the data above make a compelling case for the potential to manipulate epileptiform activity through modulation of cannabinoid mediated signaling, it is important to highlight two additional lines of evidence that may be of particular interest in considering the potential of endocannabinoids as antiepileptic agents. The first is based on work with febrile (fever induced) seizures. In this model of epilepsy long lasting increases in inhibitory synaptic transmission, CB1R expression, and DSI are all induced, along with long term limbic hyperexcitability, by brief exposure to high core body temperatures at postnatal day ten (Chen et al., 1999; Chen et al., 2003). Unexpectedly, later work found that a novel tetanus induced potentiation of DSI could be occluded by prior febrile seizures, and that the primary effects of febrile seizures themselves could also be prevented by application of a CB1R antagonist (generally considered a proconvulsant agent) during the initial insult (Chen et al., 2007; commentary by Lutz & Monory, 2008). Consistent with this idea, the same group has more recently shown that a single application of a CB1R antagonist can, if delivered rapidly after an insult, also prevent long term development of hyperexcitability in a model of epilepsy based on fluid percussion injury of the head (Echegoyen et al., 2009). Although sufficiently rapid delivery post insult may not always be an easy therapeutic goal to reach, these findings convincingly open up very intriguing and novel ideas about largely untested potential for prophylactic therapeutic interventions human epilepsies that may effectively short circuit epileptogenesis with brief, but precisely timed, inhibition of CB1R dependent signaling (Armstrong et al., 2009). Nevertheless, it is important to note that a recent study, using a different model of epilepsy, has failed to show similar efficacy of a cannabinoid antagonist delivered early during an insult in disrupting epileptogenesis (Dudeket al., 2010).

Finally, it is important to cover fascinating results about the role of cannabinoids and their receptors in the development of kainic acid (KA) induced seizures. In this model of epilepsy, systemic application of kainic acid drives excitatory synaptic activity in hippocampal circuits resulting in short term excitotoxic cell death and longer term hyperexcitability (Ben-Ari & Cossart, 2000). In 2003, (Marsicano et al.) demonstrated that CB1R-/- mice were more susceptible to KA induced seizures than wild type littermates, and further that this heightened sensitivity was maintained in mice where the CB1R was selectively removed from all principal forebrain neurons, but not from associated interneurons. This result suggested that activation of CB1Rs on excitatory neurons was essential to protect from KA induced seizures, however it could not provide a definitive answer as to whether activation of CB1Rs on GABAergic neurons was also required. A subsequent study did reveal a likely role for brain derived neurotropic factor CB1R dependent protection from KA induced seizures (Khaspekov et al., 2004). Importantly, these authors went on several years later to develop additional conditional knockouts of the CB1R that were selective to cortical glutamatergic or GABAergic neurons (Monory et al., 2006). In this study it became clear that selective removal of CB1 from glutamatergic neurons produced a dramatic reduction in CB1R expression in the inner molecular layer of the dentate gyrus that was ultimately associated with hilar mossy cells. Further, this loss of CB1Rs only in glutamatergic neurons, like the conditional knockouts in principal neurons before, continued to dramatically increase the susceptibility of mice to KA induced seizures, while strikingly, selective removal of CB1Rs from GABAergic neurons did not. Overall, these selective knockout animals continue to be one of the more elegant tools developed to study the role of cannabinoid receptors in epileptogenesis, and these papers continue to provide an excellent demonstration of a specific case in which selective activation of CB1Rs on glutamatergic rather than GABAergic terminals is required for antiepileptic action of endogenous cannabinoids.