A threesome to remember: role of the tripartite synapse in persistent cortical memory engrams

Major efforts have been made to uncover how memories are stored in the brain. It is now well established that memories are encoded through learning-induced physical changes in sparse neuronal ensembles, known as engrams. These can be identified by their activation during learning and the expression of immediate early genes (IEGs), such as Fos. Astrocytes actively modulate synaptic function and neuronal communication through their morphology, particularly perisynaptic astrocytic processes (PAPs), which interact with synapses to form the tripartite synapse. Previous research has shown that astrocytes influence memory through G-protein coupled receptor (GPCR) pathways and structural remodelling, thereby affecting synaptic plasticity and memory expression. This thesis examines cortical astrocytes in remote memory, particularly engram cells in the medial prefrontal cortex (mPFC). In Chapter 2, I evaluated whether astrocytic Gs pathway activation in the mPFC affects engram ensembles and remote memory expression. Using chemogenetic manipulation combined with TRAP-based engram labelling, I examined the role of astrocytic Gs signalling during memory formation, consolidation, and retrieval. Activation of this pathway did not alter remote memory expression, nor did it affect engram size or neuronal reactivation during recall. These findings suggest that Gs signalling in cortical astrocytes alone is insufficient to modify memory expression or engram function. In Chapter 3, I assessed the role of astrocytic Gq pathway activation in remote memory. Using iβARK to inhibit and hM3Dq DREADD to stimulate Gq signalling, I found no effect on freezing behaviour during memory tests conducted 28 days after fear conditioning. Despite robust pathway activation, neither suppression nor stimulation influenced long-term memory expression. This indicates that, unlike in the hippocampus, Gq-mediated calcium signalling in mPFC astrocytes is not essential for remote memory formation or recall. In Chapter 4, I investigated the temporal and population-specific dynamics of astrocyte–synapse interactions following learning and their impact on prelimbic cortex (PL) synapses. I introduced Astro-GRASP, a tool to study astrocyte–synapse interactions in specific neuronal populations. Fear conditioning induced time-dependent changes in astrocyte contact with dendritic spines on both engram and non-engram neurons. Early after learning (4 days), astrocyte coverage increased broadly across synapses. At a later stage (4 weeks), astrocytes selectively retracted from synapses on non-engram neurons. Knocking out astrocytic ezrin, which reduces PAP–synapse contact, enhanced remote memory expression. This demonstrates that astrocyte–synapse proximity modulates memory strength. Furthermore, fear conditioning altered input patterns from the caudal anterior cingulate cortex (cACC), increasing inputs onto non-engram neurons while reducing those onto engram neurons. These findings highlight a key role for astrocytic remodelling, particularly in non-engram neurons, during memory consolidation. In Chapter 5, I provided an overview and discussed the findings in the context of recent advances. I addressed three main points: limitations of astrocytic GPCR manipulation, emerging evidence of Fos-expressing astrocytes, and the role of astrocytic functional microdomains. While hippocampal studies suggest GPCR signalling is important, my results show that manipulating Gs and Gq pathways in mPFC astrocytes does not affect remote memory. However, consistent with hippocampal findings, reducing PAP–synapse contact through ezrin knockout enhanced memory expression, reinforcing the importance of astrocyte–synapse proximity. Although learning-associated astrocytes have been identified near engram cells in other regions, my data did not support their presence in the mPFC. Instead, astrocyte PAPs showed time- and synapse-specific responses to learning, suggesting a role for localized astrocytic microdomains in memory processing. Overall, this work advances our understanding of astrocyte–neuron interactions in memory engrams. It shows that while GPCR signalling in cortical astrocytes may have limited influence on remote memory, structural astrocytic changes play a significant role. Understanding these interactions may help explain how brain disorders disrupt memory and could inform the development of new therapeutic strategies targeting astrocyte–neuron communication.