Where does the engram come from? Study of prefrontal cortex inputs during memory consolidation

The formation and storage of memories has been under deep investigation for several decades. Nevertheless, the precise contribution of each brain region involved in this process and the interplay between them across memory consolidation is still largely debated. Although memory encoding requires the hippocampus (HPC), it is becoming apparent that other brain regions are also important in early phases of memory formation, such as the medial prefrontal cortex (mPFC). In addition, engram cells have been found in the mPFC at memory encoding: these cells represent a close approximation of the "memory trace" insofar as they are activated at the time of learning, as they undergo enduring molecular modifications after learning, and as their reactivation drives memory recall. In contrast to HPC engram cells, mPFC engram cells are thought to be kept silent during a recent recall and until the memory is fully consolidated, becoming active only at remote recall. To explain the dynamics of these mPFC engram cells, we hypothesized that inputs to the mPFC could be differentially activated over the course of memory consolidation.

We first determined that the prelimbic cortex (PL), a subregion of the mPFC, was specifically activated during the encoding phase of a fear memory in mice by measuring the expression of the Immediate Early Gene cFos, transcribed upon neuronal activation. We then traced the inputs of PL cells that were active during encoding, the putative PL engram cells, in order screen for relevant connections for further investigations. To achieve this, we used an activity-dependent monosynaptic retrograde tracing method based on rabies tracing. Then, to determine the activation patterns of these specific PL inputs, we measured their activity throughout consolidation of a contextual fear memory using an unbiased retrograde tracer coupled with cFos immunostaining. We further tested their functional relevance by chemogenetically inhibiting these projections during the different learning phases. This approach confirmed the roles of the Entorhinal Cortex and the Basolateral Amygdala inputs during the encoding phase of a fear memory to later recall remote memories. In addition, we found that the Claustrum (CLA) to PL projection was also required during the encoding phase, but to recall recent memories specifically. Moreover, the Insular Cortex (INS) to PL projection was necessary during recent recall. Eventually, we observed that the CLA and INS manipulations led to a modification of PL engram cells reactivation during recent recall.

Overall, our results suggest that there is a functional shift in PL inputs during the course of memory consolidation, which is also impacting how PL engram cells are reactivated. In addition, we observed that the activity of PL inputs is part of a broader brain network, both through the presence of axon collaterals targeting other brain regions and by the existence of parallel and perhaps redundant pathways that could ensure