Project Summary
Understanding how the brain stores and accesses information is one of the fundamental goals of
neuroscience. This question has been addressed across many scales, spanning detailed genetic analyses of
expression patterns of key molecules supporting neuronal communication to functional recordings of intact brains
in behaving animals. Over the past 50 years, synaptic plasticity has emerged as a leading cellular pathway
underlying learning and memory. During synaptic plasticity, dynamic regulation of AMPA-type glutamate
receptors (AMPARs) bidirectionally tunes synaptic strength, leading to long-lasting changes in the efficacy of
chemical communication between neurons. Synaptic plasticity is active in nearly every region of the brain and
plays a role in diverse processes, from innate fear behaviors to high-level cognition and memory. Despite the
prevalence and importance of synaptic plasticity, we still lack basic knowledge regarding how information is
distributed in synapses across the brain during learning and behavior, and even less regarding which specific
synaptic changes are necessary for long-term memory, mainly due to technical difficulties arising from the
immensely complex nature of synaptic networks. Here, we present a suite of novel methodologies that breaks
through these barriers. Our approach leverages transgenic labeling of endogenous synaptic proteins and in vivo
two-photon microscopy to enable visualization of synaptic plasticity in real time in behaving mice. Using deep
network learning, we will develop algorithms to automatically detect and track how the strength of millions of
individual synapses changes during learning. This will enable exploration of circuit-specific learning mechanisms
within discrete cell types and specific presynaptic inputs. Ultimately, this pioneering approach has the potential
to provide an unprecedented view of synapses in behaving animals, enabling new discoveries regarding how
dynamic regulation of synaptic strength encodes learning and memory.