UC San Diego

A comprehensive explanation of neuronal function requires that we understand how cellular structure seen in the brain shapes neuronal activity. For instance hippocampal CA1 synapses do not confine glutamate to the cleft following vesicular release of neurotransmitter but instead allow glutamate to diffuse out into the extracellular space, a phenomenon called 'spillover'. Combining accurate representations of the cellular structure with Monte Carlo simulations of glutamate diffusion in the extracellular space following vesicular release allows us to investigate where glutamate goes after it is released into a synapse. Here we describe a process for the creation of three-dimensional reconstructions of neuropil from two-dimensional EM images. We employed the method to generate three- dimensional reconstructions of the extracellular space from electron microscopy images and subsequent corrections informed by in vivo morphological parameters reported in the literature. Quantitative measurements of the reconstruction are consistent with reports that fixed tissue is shrunken compared to in vivo state with an especially large reduction in extracellular volume fraction. The reconstruction most likely to reflect in vivo conditions has an extracellular volume fraction of 22%, median extracellular width of 40 nm, and glutamate diffusion constant in the extracellular space of 4.5e-6 cm²/sec. As a prelude to simulations of spillover in the reconstruction we constructed a simplified three- dimensional model of hippocampal neuropil and used the model to perform Monte Carlo simulations of spillover following high-frequency burst release of neurotransmitter. The mean radial diffusion distance of a quantum of transmitter was independent of quantal size over the range tested. More than 90% of the diffusing neurotransmitter stays within 2 \[mu\]m of the release site after synaptic vesicular release. Glutamate transporters suppress NMDAR spillover activation almost entirely, while AMPAR spillover activation is nonexistent with or without transporters. Glutamate transporters are not saturated with vesicle size of 3000 glutamate even after 100Hz burst in either model. Our results suggest that glutamate spillover is insignificant in neuropil models with canonical geometry and can be ignored. However, it remains to be seen whether spillover is relevant in the heterogeneous milieu of real neural tissue.