Synaptic Integration, Calcium Dynamics, and Plasticity in Striatal Spiny Projection Neurons

Abstract

Habits enable effortless navigation of complex environments by applying previously reinforced goal-directed behaviors learned through repetition. Yet, goal-directed and habit-learning may also be coopted in neural disorders of substance addiction. The basal ganglia—a group of interconnected brain regions lying below the cortex—underlie habit and goal-directed learning, but maladaptive function of the basal ganglia is implicated in addiction and substance abuse disorders. The striatum is the input nucleus of the basal ganglia and, as such, integrates excitatory synaptic inputs from the cortex and thalamus, reward-related dopamine inputs from mid-brain, and local inhibitory inputs to regulate basal ganglia circuits and related behaviors. The striatum is organized into patch and matrix anatomical compartments, which exhibit key, yet often overlooked differences relevant to learning. Synaptic plasticity—the experience dependent change in connection strength between neurons—underlies learning and memory throughout the brain. Spiny projection neurons (SPNs), the principal neurons of the striatum, are key sites of striatal synaptic plasticity. Striatal synaptic plasticity supports goal-directed and habit learning by reinforcing experiences (represented by cortical inputs) that produce rewarding outcomes (represented by dopaminergic inputs). However, the cellular mechanisms for SPNs to recognize spatiotemporal patterns in cortical inputs and change connection strengths for relevant patterns are unclear. Synaptic plasticity requires intracellular calcium elevation, and the magnitude and duration of calcium influx has been hypothesized to determine the direction—either potentiation or depression—of synaptic plasticity. However, most synaptic plasticity findings are from in vitro slice experiments conducted with regular, repeated inputs unlike naturally occurring, variable in vivo activity. One of the most important questions in neuroscience is how spatiotemporally distributed and variable in vivo-like inputs generate synaptic plasticity. In this dissertation I use computational methods to integrate ex vivo mechanisms with in vivo-like conditions to answer this question and determine the distribution of synaptic plasticity in response to cortical activity. Simulation experiments evaluate the central hypothesis that in vivo-like patterns of synaptic input will support striatal synaptic plasticity. First, I test the hypothesis that spatiotemporal patterns of synaptic input will produce nonlinear, spatially specific spine calcium signals. I show that calcium transients in dendritic spines encode spatiotemporal patterns of synaptic inputs, and I discover a key role for inhibitory synaptic inputs—they enhance the stimulus specificity of spine calcium for active versus inactive synapses, which suggests inhibition may support learning by restricting synaptic potentiation to active synapses while preventing it for inactive synapses. Second, I evaluate the effect of spatiotemporal patterns of synaptic input with in vivo-like variability on synaptic plasticity. I show that synaptic plasticity is highly robust to trial-to-trial variability in spike timing. Further, I uncover distinct spatiotemporal synaptic activity patterns associated with potentiation and depression. Lastly, I investigate whether SPNs from patch and matrix compartments of the striatum exhibit intrinsic cellular differences and evaluate their contributions to striatal activity. I show that patch and matrix SPNs exhibit intrinsic differences in excitability and synaptic integration, suggesting that distinct functional associations of patch and matrix compartments are not simply due to network differences but also cell-type differences. Together, these studies reveal new insights into striatal and basal ganglia circuit function. My work suggests key mechanisms for how striatal SPNs can encode environmental and body contexts—represented by sensory, association, and motor cortex—by nonlinear integration of coactive synaptic inputs converging from all areas of cortex to the striatum. This nonlinear integration produces distinct calcium traces that can produce plasticity specific to those coactive synapses, thus underlying learning. Inhibitory function can strengthen the specificity of plasticity to generate persistent changes and support stable and long-lasting learning. Cell-type specific differences in distinct anatomical compartments of the striatum can bias striatal activity towards patch compartments, which are specifically important for reinforcement learning. My work predicts that plasticity is robust to in vivo trial-to-trial variability and suggests a generalizable spatiotemporal plasticity rule showing how the spatial organization of dendritic synaptic connectivity influences learning throughout the brain. Together these results contribute to advances in understanding the cellular level of learning and can support future discoveries into preventing aberrant learning associated with addiction.