Synapses communicate temporal sequences of action potentials between neurons with variant efficacy, allowing the same axon to convey independent messages to multiple post-synaptic targets. Several molecular mechanisms control information flow in neural networks. In the hippocampus, transmission responses are highly variable even at the level of individual synapses across different cell types and changes dynamically on the multiple time scales on which it operates. Modeling synaptic transmission and dynamics requires balancing the model’s interpretability and its ability to espouse experimental data. In this work, the high variability associated with synaptic responses is first considered at a synapse level. Taking a statistical approach to the phenomena, a biophysically tractable gamma-mixture model is developed to characterize postsynaptic responses from single synapse release events recorded by a sensor for the neurotransmitter glutamate as fluorescence transients. Here the development of this modeling framework leads to three different versions of the framework: two bimodal frameworks that take two different approaches to modeling the noisiness of synaptic releases, and a statistically validated unimodal approach. Variational inference techniques are applied to these frameworks through an expectation-maximization algorithm, which operates on the principles of maximum likelihood. This results in the extraction of latent variables for quantal size, number, and release probability, allowing for the characterization of release events at a synaptic level. A system identification approach is taken to capture the diverse types of synaptic response dynamics observed on short time scales. This extends work from previous phenomenological approaches to account for a nonlinearity and the kinetics evolving on multiple time scales present in this phenomenon. Gradient descent methods are used to estimate synaptic kinetics from complex firing patterns such as those observed \textit{in vivo}. The characterized dynamics in synaptic transmission all contribute to the transfer of information between cells and are assumed to strive for maximizing information transfer through reducing redundancies and optimizing cost-efficiency between the required energy input and the information transferred. The postsynapse has a seeming redundancy as it has two glutamate receptors with different detection thresholds, suggesting there should be a benefit to having both receptors; here this idea is explored here through numerical simulation. Taken together with the modeling of observed glutamate release dynamics, this creates an avenue for improved theory for information processing capabilities of synapses.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/41628 |
| Date | 07 January 2021 |
| Creators | Trotter, Daniel |
| Contributors | Naud, Richard |
| Publisher | Université d'Ottawa / University of Ottawa |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
Page generated in 0.0057 seconds