UC San Diego

This dissertation presents findings from two studies investigating the neuroscience of memory. Chapter 1 reports findings from a computational genomics study on Alzheimer's Disease (AD), a common neurodegenerative disease that causes memory loss and dementia. There is hope that genomic information can reveal insights to AD pathophysiology, along with aiding in risk assessment, screening, and diagnosis. This study involved using exome sequencing data from approximately 10,000 individuals (~5000 AD patients) to train neural net (NN) classifiers tasked with estimating the impact of single genomic variants on AD polygenic risk. Together, these chapters and the studies presented therein add to our corpus of knowledge on memory systems, and contribute to our quest of understanding and treating dementia. Chapter 2 covers a study that details a model (supported by experimental results) that can explain how memories can be formed and maintained within neural network synapses, despite continuous and complete molecular turnover of synaptic proteins. Memory formation is thought to involve acute changes to synaptic weights; the ability for synaptic weights to remain stable over long time-periods and undergo evoked change is considered fundamental to our brain’s information storage schema. Yet basic questions regarding synaptic plasticity remain unresolved, including a) how synaptic weights remain stable in the face of perpetual and complete molecular turnover; and b) how such weights can be modified to new stable levels by transient signals. Through a series of computational and biological experiments, we elucidate actin, a protein with unique polymer properties, as a potential central mediator of synaptic weights.