Putting the Brakes on Fear

Abstract

The extinction of fear is believed to involve inhibitory processes in the amygdala. In this issue of Neuron, Trouche et al. (2013) show that basal amygdala neurons activated by fear conditioning are silenced by local inhibitory interneurons after extinction.

Brains are built to detect, learn, and remember environmental signals for the presence (or absence) of biologically significant events. For example, during Pavlovian fear conditioning, animals learn to fear otherwise innocuous stimuli or places that signal aversive outcomes. Fortunately, these fear memories can be “extinguished” when the learned signals (conditioned stimuli or CSs) occur without any noxious consequences, a process that is fundamental to clinical interventions, such as exposure therapy, for anxiety disorders. Yet the loss of fear that occurs after extinction is fragile; fear relapses with the mere passage of time (spontaneous recovery), changes of context (renewal), and presentation of the aversive US with which the CS had been paired (reinstatement) (Bouton, 1993). Apparently, extinction procedures do not erase fear memories, rather they yield new inhibitory memories that suppress (but do not eliminate) fear to the CS. Understanding the nature of this inhibition is central to improving therapeutic interventions for fear and anxiety, including exposure therapy.

Not surprisingly, the neural mechanism for extinction is believed to involve inhibitory processes in the amygdala, a brain structure that is essential to both the conditioning and extinction of fear (Herry et al., 2010; Maren and Quirk, 2004). One mechanism for fear inhibition that has received considerable support involves prefrontal-amygdala projections that recruit clusters of inhibitory interneurons (ITC cells) interposed between the basal (BA) and central (CE) nuclei of the amygdala. After extinction, ITC cells excited by the infralimbic (IL) division of the medial prefrontal cortex are believed to limit excitatory transmission between BA and CE by directly inhibiting CE neurons that drive fear responses to fear CSs (Likhtik et al., 2008; Quirk et al., 2003).

Although considerable evidence indicates that an IL-ITC circuit maintains extinguished fear, there are both behavioral and neural data that are not readily explained by this model. First, an IL-mediated inhibition of CE (which presumably operates to suppress fear output nonspecifically) by an extinguished CS should block fear to another unextinguished CS when the two stimuli are presented together, but evidence for this is scant (Leung and Westbrook, 2008). Moreover, an IL-mediated suppression of fear by inhibitory ITC cells overlooks the observation that neurons upstream in the basal and lateral amygdala themselves show decrements in activity after extinction (Herry et al., 2008; Repa et al., 2001) that readily renew outside of the extinction context (Hobin et al., 2003). These observations suggest that local inhibition within the BA may selectively (and reversibly) silence neurons in the BA after extinction to suppress fear.

In an article in the current issue of Neuron, Trouche et al. (2013) examined this possibility by labeling neurons in the BA involved in contextual fear conditioning, and then examining whether those neurons are reactivated during memory retrieval after extinction. To this end, they used a TetTag reporter mouse that expresses green fluorescent protein (GFP) under the control of a c-fos promoter when doxycycline is removed from the diet (Reijmers et al., 2007). After labeling neurons during fear conditioning, Trouche and colleagues then assessed ex vivo GFP and Zif expression after a retrieval test to determine whether neurons active during fear conditioning remained active after extinction. Interestingly, they found that roughly 15% of the BA neurons tagged during fear conditioning were reactivated in non-extinguished mice. However, only half that number of neurons were reactivated in animals that underwent an extinction procedure. In other words, extinction training silenced a large proportion of BA neurons that had been active during fear conditioning. They did not observe extinction-induced silencing in either hippocampal area CA1 or IL, suggesting that the silencing was rather specific to the BA. Hence, these results imply that the extinction of fear drives local inhibitory interneurons to establish synaptic contacts with a subset of excitatory BA neurons recruited during fear conditioning.

To further explore this possibility, Trouche and colleagues (2013) examined the co-localization of proteins unique to inhibitory interneurons in active and silenced BA neurons. Interestingly, they found that silenced neurons exhibited significantly greater perisomatic GAD67 labeling, suggesting a proliferation of inhibitory GABAergic synapses on these neurons. In line with this idea, they found that the density of perisomatic parvalbumin (PV) staining was greater in silenced neurons and these changes were only observed in animals undergoing extinction. Together, these data suggest that extinction learning is associated with an increase in the number of new inhibitory synapses onto BA neurons representing the fear memory. Interestingly, Trouche and colleagues (2013) also observed a subset of neurons that were not silenced after extinction, and these cells exhibited higher densities of perisomatic cannabinoid receptor 1 (CB1R) labeling. Because CB1Rs limit GABA release, these receptors suggest a mechanism for sustained activity in neurons that were not silenced by the extinction procedure. Altogether, the data reveal that extinction learning remodels inhibitory synaptic input onto BA neurons to limit the expression of fear.

The reorganization of inhibitory synaptic input onto the specific network of neurons encoding the fear memory is a novel, selective, and direct mechanism for limiting conditioned fear responses after extinction. Although the cellular mechanisms underlying these synaptic changes are not yet understood, a large number of studies suggest that NMDA receptors may be involved (Falls et al., 1992; Zimmerman and Maren, 2010). How NMDA receptors mediate both long-term potentiation of excitatory synapses onto BA neurons encoding fear conditioning and the remodeling of perisomatic inhibition onto these neurons after extinction is a fascinating question. Whatever the mechanism, these data are consistent with the idea that extinction involves new learning that suppresses learned fear responses, rather than erasing the fear memory itself.

Of course, a critically important question concerns how these and other inhibitory mechanisms are themselves silenced during fear relapse. That is, how does fear in response to an extinguished CS renew, for example, when the CS is presented outside the extinction context? One possibility is that the activity of inhibitory interneurons in the BA is context-dependent; the activity of these neurons may be elevated in the extinction context but dampened in a dangerous context. Another possibility is that fear relapse is mediated by BA neurons that remain active after extinction. Clearly, further work is required to understand how target-specific silencing of BA neurons is modulated to allow for the context-dependent expression of fear. It is becoming clear that hippocampal and medial prefrontal cortical projections to basal and lateral amygdala neurons are involved in fear relapse after extinction (Herry et al., 2008; Knapska et al., 2012; Orsini et al., 2011). Whether these circuits ultimately suppress inhibitory activity in the amygdala or drive activity in BA neurons during fear relapse (or both) remains to be examined. Clearly, the use of activity-dependent neuronal tags to track neuronal populations engaged during encoding and retrieval processes is a promising strategy to answer these questions.