A cartography of learning-induced synaptic potentiation using the genetically encoded SynActive toolbox

During learning tasks, specific neuronal populations are recruited and are thought to be responsible for the physical storage of memories (Josselyn and Tonegawa, 2020). However, each neuron has thousands of synaptic contacts with its peers, and these synapses can adapt to the strength of incoming stimuli individually and bidirectionally, playing a fundamental role in the plasticity underlying the formation of long-term memories (Rogerson et al., 2014). Despite this, knowledge of how individual synaptic contacts are involved in this process is still limited.To address this gap, our lab developed a method called SynActive (SA) to label and manipulate potentiated dendritic spines, the postsynaptic elements of excitatory synapses where most of plasticity occurs. This dual targeting strategy exploits Untranslated Regions (UTRs) of the Arc mRNA and a short postsynaptic Density (PSD)-binding synthetic peptide to allow activity-dependent expression of any reporter protein at dendritic spines undergoing translation-dependent potentiation, improving expression and  retention of these proteins at the postsynaptic density (Gobbo et al., 2017). In the first proof of principle application of the SynActive experimental strategy, the expression of the SA reporter in the mouse brain was achieved by a transgenic  approach, via in utero electroporation of embryos.In order to facilitate the use of the Synactive strategy, the first objective of my thesis was to develop a set of  adeno-associated viral vectors (AAVs) directing the expression of SA-based genetically encoded reporters. Using in vivo delivery of SA-based reporters via AAV vectors, we aimed (i) to create a map of potentiated synapses in the hippocampus induced by associative learning, and (ii) to analyze the proteomic signature of learning-induced synaptic potentiation.To achieve the first goal, we used an SA-based vector encoding a  shortened version of PSD95, the most abundant scaffolding protein of the excitatory postsynaptic density (Cheng et al., 2006), fused with a fluorescent protein, to selectively detect the subset of potentiated spines among all dendritic spines. Taking advantage of a Tet-ON inducible system (Sun et al., 2007), we temporally confined the expression of the SA-construct to the formation of a contextual fear memory (encoding). Then, by imaging the SA-positive dendritic spines, we produced a database of the distribution and geometry of potentiated spines along different branches of the dendritic tree. We found notable differences in their distribution between discrete regions of the hippocampus and sections of the same neuron.Next, we constructed a SA-based proteomic bait, by tagging the full length PSD-95 protein fusing it to a FLAG epitope (Einhauer and Jungbauer, 2001) to define the proteomic signature of dendritic spine potentiation. Exposure to contextual fear conditioning triggered the expression of FLAGged PSD-95 at the level of potentiated spines, thus demonstrating the validity of this tool for future analysis of the PSD-95 interactome and analyzing changes in the proteic content specific to potentiated synapses.These findings offer insights into the distribution, geometry, and molecular composition of potentiated dendritic spines, which are thought to play a crucial role for learning and memory.