Memory Information Allocation Mechanism within the Brain Partially Revealed
Naoki Matsuo (Associate Professor, Graduate School of Medicine, Osaka University) found that the activity of the initial ensemble is preferentially dedicated to the same learning.
Recent innovative studies have begun to present direct evidence that individual memories reside in the activities of specific spatially distributed neuronal populations within neuronal networks. The next critical question arising from this idea is how specific subsets of neurons are chosen from a large population of neurons to encode a given memory.
We have generated genetically-engineered mice to suppress specific subset of neurons activated during learning. We found that the suppression of neuronal ensembles that are naturally activated during learning results in a failure of the memory retrieval. We further found that the suppression selectively inhibits relearning without disrupting the ability to acquire and retrieve a memory for distinct context. These results indicate that there is a mechanism ensuring that the same neuronal ensemble is engaged for the same learning to strengthen the memory, and it is not substitutable after the ensemble is allocated for the initial learning.
Our results provide substantial insights into the machinery underlying how the brain allocates individual memories to discrete neuronal ensembles and how it ensures that repetitive learning strengthens memory by reactivating the same neuronal ensembles. Our findings could be of some help in understanding the mechanism underlying memory impairments associated with aging and psychiatric disorders.
Our results have been published in Cell Reports on April 16.
External information acquired through daily experiences can be internally represented and stored in the brain across several interacting regions as a memory. Recent innovative studies have begun to present direct evidence that individual memories reside in the activities of specific spatially distributed neuronal populations within neuronal networks.
However, much remains to be elucidated concerning the machinery of memory allocation. For instance, lesion studies suggest that an alternative system can compensate for damage to the primary region employed when animals acquire a memory. However, it is unclear whether functional compensation occurs at the cellular ensemble level. Thus, we investigated whether inhibiting the reactivation of the neuronal ensembles that participated in the initial learning could hinder relearning. If the same subset of neurons is not necessarily assigned to the same learning and an alternative ensemble of neurons can functionally compensate for the inhibited ensemble, animals should acquire and express the fear memory. We investigated which system is working in the brain.
Results of our research
First, we have generated a transgenic mouse in which a synaptic transmission of neuronal ensembles activated by a given behavioral stimulus was selectively and reversibly inhibited via a tetanus toxin light chain (TeNT). The transgenic mice were fear-conditioned to elicit a long-term fear memory and to initiate the synthesis of TeNT selectively in the activated neurons. Mice were returned to their homecages. On the following day, mice were re-exposed to the same chamber without footshocks to evaluate their contextual fear memory by measuring a freezing behavior. The transgenic mice showed significantly less freezing compared with control animals suggesting an impairment of contextual fear memory retrieval in the transgenic mice.
Then, they were retrained in the same chamber (Figure 1). Notably, freezing was not increased compared with pre-retraining duration (Figure 2), indicating that the second training session failed to strengthen the contextual fear memory in these mice. In contrast, control animals exhibited substantially increased freezing after retraining. These results indicate that there is a mechanism ensuring that the same neuronal ensemble is engaged for the same learning to strengthen the memory, and it is not substitutable after the ensemble is allocated for the initial learning.
To examine the specificity of the TeNT-mediated silencing to discrete neuronal representations, we investigated whether the transgenic mice were able to acquire a new fear memory associated with a different context. We found that they showed significantly more freezing during the retrieval test than the period before shock presentation at the retraining session in context.
We have demonstrated the following: 1) Suppression of neuronal ensembles that were naturally activated during fear-conditioned learning impaired the retrieval of the contextual fear memory. 2) Suppression of neuronal ensembles that were activated during fear-conditioned learning hindered relearning of the memory but did not interfere with new learning of a distinct contextual fear memory (Figure 3).
Brain is a tissue that has a flexible system. For example, an alternative system can compensate for damage to the primary region employed when animals acquire a memory. Interestingly, our results revealed that mice did not relearn when the neuronal ensemble engaged in the initial learning was compromised, indicating that functional compensation did not occur.
Established memories can be strengthened by repeated learning. However, the underlying neural mechanism remains to be elucidated. Our result provides remarkable insight because it implies that the same neuronal ensemble is preferentially dedicated to the repetitive learning. This inflexibility of an ensemble could ensure the strengthening of synaptic connections across a specific subset of neurons by repetitive activation, thereby enabling memory enhancement.
Figure 1. A schematic of the experimental design.
In the transgenic mouse, when neuronal activity sufficient to activate the c-fos promoter occurs in the absence of Dox, tetanus toxin is selectively expressed in those neurons activated by the behaviorally relevant events. Animals were trained with fear conditioning in context A during off Dox; they were subjected to second conditioning in either context A or context B in the presence of Dox. Then, mice were subjected to contextual fear memory retrieval testing.
Figure 2. Memory enhancement by re-training.
The percentage of time spent frozen by control mice (red) during the test sessions increased after re-training (i.e. memory was enhanced) while that by transgenic mice (blue) did not.
Figure 3. Specific subsets of neurons are chosen from a large population of neurons to encode a given memory. In the transgenic mouse brain, neurons that were activated during learning expresses tetanus toxin, thereby the neuronal activity is suppressed. Tetanus toxin expression in neurons activated during memory encoding selectively inhibits relearning without disrupting the ability to acquire and retrieve a fear memory for distinct context. These results indicate that there is a mechanism ensuring that the same neuronal ensemble is engaged for the same learning to strengthen the memory, and it is not substitutable after the ensemble is allocated for the initial learning.
To learn more about this research, please view the full research report entitled " Irreplaceability of neuronal ensembles after memory allocation " at this page of the Cell Reports website.