Chasing “fear memories” to the cerebellum

The cerebellum has long been known as a center of fine motor control and sensory-motor integration. It also is appreciated as essential for the acquisition and expression of conditioned eye-blink responses, a type of discrete sensory-motor learning (1, 2). Recently, the cerebellum also has been implicated in acquiring (3–5) and expressing (6) “emotional” associative learning, which is commonly believed to depend entirely on forebrain regions. Specifically, the learning and memories for an association between neutral and fear-eliciting stimuli, also known as “fear conditioning”, often are said to reside in the basolateral amygdala (ref. 7, for an alternative view see ref. 8). An elegant study by Sacchetti and colleagues in this issue of PNAS (9) adds a new dimension to the recently described role of the cerebellum in emotional learning and memory by revealing that the cerebellum is essential for consolidating memories for both tone and contextual fear conditioning for several days longer than the basolateral amygdala. Thus, their findings are not only relevant to the current debate regarding the sites of permanent storage of fear memories, but more importantly, they expand our knowledge of the topographical and chronological composition of the brain network(s) responsible for consolidating this type of memory.

The term “fear memory” typically is used to describe the plasticity in the brain underlying the long-lasting modifications in the behavior of animals that had undergone fear conditioning. Fear conditioning is a long-lasting (10) form of classical (Pavlovian) conditioning where animals learn to associate a nonfear eliciting stimulus, a conditional stimulus (CS), with an aversive stimulus, an unconditional stimulus (US), that generates a fear state expressed by a multitude of unconditional responses (URs) (11). As a result of learning the CS–US association, animals start to respond to the CS alone with conditional responses (CRs). CRs were not previously generated by the CS and may be similar to the fear-related URs. The magnitude of the CRs usually indicates the strength of the memory for the fear conditioning training. In studies investigating the neural substrates of fear memories in mammals, the CS is typically a tone, a light, or a place (tone, light, or contextual fear conditioning, respectively) and the US is an aversive stimulus, such as an electrical shock to the feet or the tail of animals. By using tone fear conditioning, during which rats also acquire fear to the environmental context in which the tone-foot shock pairing occurs, Sacchetti and colleagues (9) demonstrate for the first time that consolidation of memories for tone fear conditioning depends on the interpositus nucleus of the cerebellum for at least 4, but not 8, days after training and to the vermis of the cerebellar cortex for at least 8, but not 16, days after training. They further show that consolidation of memories for contextual fear conditioning depends only on the cerebellar cortex for at least 4 days. The reported effects are specific to fear memory consolidation, because the functional integrity of the cerebellar regions was disrupted temporarily at a defined time after the rats had acquired the CS–US association (12), as assessed by freezing behavior, a characteristic immobility that rats display in situations of intense fear (13). Furthermore, the observed impairments in freezing behavior at the retention test (when the CS was presented alone) did not result from an impaired performance of the freezing response, as the test was conducted when the rats were not under the effects of the inactivating drug.

The cerebellum is involved in consolidating fear memories for over a week after the training.

The finding that the cerebellum is involved in consolidating fear memories for over a week after the training are a significant advancement in our understanding of the brain regions processing and storing fear memories, as current investigations center on fear conditioning-induced plasticity in the medial geniculate, hippocampus, amygdala, and auditory cortex. It is hypothesized that during fear conditioning, the basolateral amygdala complex (BLC, including the lateral, basolateral, and basomedial nuclei) receives CS information (integrated representation of the spatial context from the hippocampus, tone-specific information from the medial geniculate and the auditory cortex, or light-specific information from higher-order visual cortical areas), associates this information with the US, and permanently stores the CS–US associations (7, 14). This hypothesis is based mainly on the observation that pretraining or pretest lesions or inactivation (both general and N-methyl-d-aspartate receptor specific) of this brain region lead to decreased freezing behavior (15–20) or to decreased potentiated startle (21), two CRs typically used to assess memory for fear conditioning. In addition, the basolateral amygdala develops long-term potentiation-like plasticity as a result of auditory fear conditioning (22).

However, the hypothesis that the BLC is the permanent storage site for fear memories is inconsistent with evidence from several lesion, pharmacological, electrophysiological, and imaging studies. Rats with BLC lesions can acquire contextual fear memories and express them by avoiding a place previously paired with foot shock (23, 24). The observed impairment in conditional freezing in the absence of the BLC can be caused by impairment in the ability to generate this CR, decrease in the strength of the acquired memory, or a combination of both. Consistent with the latter are the observations that BLC-lesioned rats are impaired in expressing unconditional freezing (25, 26), and that activity in the BLC after fear conditioning can modulate the strength of the fear memory (27, 28). Moreover, rats with complete BLC lesions can show memory savings expressed as a modest, but significant, context-specific increase in conditional freezing after a single foot shock in a context previously associated with foot shock (the increase was not present in BLC-lesioned rats that were initially trained in a different context, ref. 29). Furthermore, plasticity induced by fear conditioning is not unique to the BLC. After auditory fear conditioning, associative increase in the neuronal responses to the CS also is observed in the medial geniculate and the auditory cortex (30–32). Additionally, an imaging study in rats revealed specific increases in activity during fear conditioning in multiple brain regions, including brainstem auditory nuclei and the cerebellum (3).

The known temporal characteristics of fear memory consolidation are also inconsistent with the BLC being the permanent repository of fear memory. Temporarily inactivating the BLC for several hours is memory impairing when administered up to 4 days after tone and contextual fear conditioning; however, similar inactivation of the perirhinal cortex is memory impairing even at 8 days (27, 28). Combined with the fact that lesions of the perirhinal cortex also abolish conditional freezing (33), these findings suggest that fear memories are elaborated in the perirhinal cortex after being processed by the BLC. Because the BLC and perirhinal cortex are concurrently involved in fear memory consolidation for the first 4 days, these findings are also consistent with the hypothesis that the BLC modulates storage of emotional memories occurring in cortical sites (34). Collectively, the reviewed arguments against the hypothesis that the BLC stores the fear memory suggest that fear memories are processed and possibly stored in multiple brain regions. In this context, the findings from the current study by Sacchetti and colleagues indicate that the interpositus nucleus and the cerebellar cortex are part of a wide network of brain regions involved in consolidating memory for fear conditioning. More intriguingly, the cerebellar cortex, along with the perirhinal cortex, is involved in the consolidation of memory for tone fear conditioning for the longest duration currently reported, thus making it a plausible site of long-term storage of this memory.

In conclusion, Sacchetti and colleagues present evidence that builds on previous findings to expand our understanding of how emotional memories are elaborated and stored in the brain. It is revealed that consolidation of fear memories is an integrated process that involves network(s) of brain regions, including the cerebellar cortex and interpositus nucleus, with defined temporal dynamics. A remaining question is whether the plasticity seen in many brain regions during acquisition of fear conditioning is elaborated by parallel and independent networks or is a unitary dynamic process that shifts through different brain regions over time.