Pankaj Sah is in the Division of Neuroscience, John Curtin School of Medical Research, Australian National University, Canberra, ACT 2601, Australia.
Although we understand how fearful memories are stored in the brain, how they are extinguished remains a mystery. The answers may lie with the cannabinoid compounds our bodies produce.
Cannabinoids such as marijuana and hashish have been used for over a thousand years for medicinal and recreational purposes. The active 'ingredient' of these drugs is 9-tetrahydrocannabinol, which produces effects on nerve cells in the brain by binding to a protein on the neuronal surface, the CB1 receptor1. But of course the receptor is not there simply to detect this externally derived compound: it also binds to 'endogenous' cannabinoids, which are produced naturally by the body. On page 530 of this issue, Marsicano and colleagues2 propose a new role for this 'endocannabinoid' system extinguishing fear-related memories in mice. The finding might have implications for treating anxiety disorders in humans.
We can form memories in several different ways, one of which is Pavlovian conditioning the classic example being that of Pavlov's dogs, which learned to expect food whenever they heard a ringing tone. We all form these types of associations; for instance, we may associate a particular piece of music with our first love affair. But the connection need not always be pleasant. Imagine you are having a quiet walk in a park when you are threatened by an armed person. During the attack you are terrified; your heart races and your palms are sweaty. You run and escape. Later, you may find that entering the same park brings back in detail the memory of the attack, right down to the sweaty palms.
In the lab, the neuronal and molecular mechanisms underlying fearful memories are often studied in animals by using 'fear conditioning'. Here, a neutral or conditioned stimulus, which is typically a tone or a light, is paired with an aversive (unconditioned) stimulus, typically a small electric shock to the foot. After the two stimuli are paired a few times, the conditioned stimulus alone evokes the stereotypical features of the fearful response to the unconditioned stimulus, including changes in heart rate and blood pressure and freezing of ongoing movements. Repeated presentation of the conditioned stimulus alone leads to extinction of the fearful response the animal learns that it need no longer fear a shock from the tone or light.
A large body of work has established that a small, almond-shaped region in the brain, the amygdala, is crucial in acquiring and, possibly, storing the memory of conditioned fear3, 4. It is thought that, at the cellular and molecular level, this learned behaviour requires neurons in the basolateral part of the amygdala, and changes in the strength of their connection with other neurons ('synaptic plasticity') that depend on the NMDA receptor5, which responds to the neurotransmitter glutamate.
The extinction of aversive memories also involves the basolateral amygdala, but the cellular and molecular details are less clear. Infusing antagonists of the NMDA receptor into this region blocks extinction, implying that these receptors are important here, too6. Yet their exact role is not known. It has been proposed that synaptic plasticity is again involved6, but the possible sites of plasticity and the underlying physiology are not known, and NMDA-receptor-dependent plasticity has not yet been correlated with extinction. Moreover, it has been suggested that there are also NMDA-receptor-independent mechanisms of extinction.
Marsicano et al.2 now propose just such a mechanism, which involves the endocannabinoids anandamide and 2-arachidonylglycerol, and their CB1 receptors. These receptors are some of the most abundant neuromodulatory receptors in the central nervous system and are expressed at high levels in the limbic system, cerebellum and basal ganglia8. The classical behavioural effects of exogenous cannabinoids such as sedation and memory changes have been correlated with the presence of CB1 receptors in the limbic system and striatum.
It has been difficult, however, to pin down the physiological role of endocannabinoids and how they are released in these regions. In studies that were the first to reveal such a role, the depolarization of neurons by repetitive activity led to the release of endocannabinoids9, which diffused to the terminals of other neurons and inhibited neurotransmitter release. This effect was transient in the hippocampus and cerebellum9 and long lasting in the striatum10. Yet these changes in neurotransmission have not been connected to any specific behavioural effects. So the study by Marsicano et al.2 represents a leap forward in two areas of neurobiology, in that it clearly implicates the release of endocannabinoids in a well-known, simple learning task. It also links endocannabinoid release to synaptic plasticity.
After engineering mice to lack the CB1 receptor, Marsicano et al. first showed that although these animals could learn and later recall the association of a tone with a foot shock, they could not extinguish the memory. A drug that antagonizes the CB1 receptor similarly prevented extinction in wild-type mice. The authors then found that during the extinction protocol (exposure to the tone alone), the levels of both anandamide and 2-arachidonylglycerol were raised in the basolateral amygdala in mutant and normal mice. This implies that a process involving activation of the CB1 receptors by endocannabinoids is essential in the extinction of conditioned fear.
Next, in experiments with slices of normal mouse brains, the authors looked at neurons in the basolateral amygdala that can release GABA (an inhibitory neurotransmitter). They found that low-frequency stimulation of these neurons leads to a long-term reduction in the release of GABA, which in turn leads to less inhibition of the connecting 'pyramidal' neurons. This long-term 'depression' a type of synaptic plasticity was completely blocked by the CB1-receptor antagonist, and absent in CB1-deficient mice. These findings suggest that the endocannabinoids reduce GABA release in the basolateral amygdala, thereby helping to extinguish the fear-conditioned response. In mammals, the neurons that release GABA are largely interneurons, which can be divided into several populations on the basis of their expression of certain proteins and peptides (such as cholecystokinin). The role of endocannabinoids in reducing GABA release fits with the finding that CB1 receptors in the basolateral amygdala are present on the terminals of cholecystokinin-containing interneurons11, 12.
This is an entirely new cellular and molecular mechanism for extinction. But how does it tie in with the NMDA receptors? There seems little doubt that activation of these glutamate receptors in the basolateral amygdala is somehow required for extinction6. But Marsicano et al.'s brain-slice experiments were performed with blocked glutamate receptors, showing that the endocannabinoid-mediated synaptic plasticity they report does not need the NMDA receptors. So we have yet to find out how these receptors are involved in extinction.
It has been argued that the neuronal circuitry underlying fear conditioning has similarities to that responsible for fear-related clinical conditions, such as post-traumatic stress disorder4. Behavioural therapies for these conditions including systematic desensitization and imagery therapies share features with extinction. The finding that the endocannabinoids contribute to extinction raises the possibility that drugs that target these molecules and their receptors could be useful new treatments for anxiety disorders. Finally, there is much anecdotal evidence of patients using cannabis heavily in the early stages of psychiatric illness. This has often been thought to contribute to acute illness. But it seems possible that it may instead be a form of self-medication for the sometimes extreme anxiety that these people experience.
References: 1. Matsuda, L. A. et al. Nature 346, 561-564 (1990)
2. Marsicano, G. et al. Nature 418, 530-534 (2002)
3. LeDoux, J. E. Annu. Rev. Neurosci. 23, 155-184 (2000)
4. Davis, M. & Whalen, P. J. Mol. Psychiatry 6, 13-34 (2001)
5. Blair, H. T., Schafe, G. E., Bauer, E. P., Rodrigues, S. M. & LeDoux, J. E. Learn. Mem. 8, 229-242 (2001)
6. Walker, D. L. & Davis, M. Pharmacol. Biochem. Behav. 71, 379-392 (2002)
7. Santini, E., Muller, R. U. & Quirk, G. J. J. Neurosci. 21, 9009-9017 (2001)
8. Herkenham, M. et al. J. Neurosci. 11, 563-583 (1991)
9. Wilson, R. I. & Nicoll, R. A. Science 296, 678-682 (2002)
10. Gerdeman, G. L., Ronesi, J. & Lovinger, D. M. Nature Neurosci. 5, 446-451 (2002)
11. Katona, I. et al. J. Neurosci. 21, 9506-9518 (2001)
12. McDonald, A. J. & Mascagni, F. Neuroscience 107, 641-652 (2001)