Synaptic plasticity mechanisms that protect and refine local circuits

ABSTRACT Synapses form trillions of connections between billions of neurons in the brain to establish neural circuits that allow us to sense, think, act, learn, and remember. Our goal is to understand how synapse structure supports learning and memory with a focus on dendritic spines, the tiny protrusions that host most of the excitatory

synapses in the brain. While most neuroscientists would agree that synapse growth and retraction are vital for learning and memory, we do not know how these long-term changes in synaptic structure are regulated in the face of ongoing brain plasticity. The synaptic active zone comprises discrete domains where presynaptic

vesicles are docked and released. Postsynaptic responses are restricted to regions within ~100 nm of the vesicle release sites. Our three-dimensional reconstruction from serial section electron microscopy (3DEM) reveals three zones across the synapse: (i) strong active zones (AZs) that have tightly docked presynaptic

vesicles, (ii) weak AZs that have loose or nondocked presynaptic vesicles, and (iii) nascent zones (NZs) that have a thick postsynaptic density but no presynaptic vesicles. At the onset of long-term potentiation (LTP), presynaptic vesicles are rapidly recruited to the NZs, converting them to AZs. Protein filaments shorten and

draw docked presynaptic vesicles closer to the enlarged AZs, and recruit vesicles to dock at weak AZs. This evidence of presynaptic plasticity would increase the area of release and probability of postsynaptic receptor response. The recovery interval following saturation of LTP is 1-4 hours depending on the preparation. During

this interval, new NZs form, primarily on spines containing smooth endoplasmic reticulum, a local resource for regulating calcium and trafficking of lipids, proteins, and organelles. Clusters of spines form in the vicinity of these enlarged spines. We hypothesize that synapse-specific expansion of NZs during LTP provides a basis

for learning and the advantage of spaced over massed learning to establish long-lasting memories. Furthermore, we hypothesize that LTD is driven by the conversion of weak AZs to NZs and ultimately elimination of spines without AZs. To address these hypotheses, we propose multidisciplinary approaches to

investigate NZ and AZ plasticity—including slice physiology, optogenetics, glutamate uncaging, and tomographic 3DEM of synapses along activated axons labeled with APEX. Our Specific Aims are: Aim 1) Determine the specificity of NZ to AZ conversion during synapse enlargement, resource utilization, and spine

clustering underlying the saturation, recovery, and enhancement of LTP. Aim 2) Test whether saturating LTP at an isolated dendritic spine is sufficient to fill NZs and determine the role of PSD-MAGUK proteins and their interaction partners in the recovery of LTP from saturation. Aim 3) Test whether saturation of long-term

depression (LTD) is accompanied by loss of weak AZs and determine the time-course over which LTD recovers from saturation. Outcomes promise new insights about synaptic mechanisms of learning and memory and new targets for understanding and treating learning disabilities.