Authors :
Presenting Author: Kyle Lillis, PhD – Massachusetts General Hospital
Lauren Lau, PhD – Massachusetts General Hospital
Paige O'Gorman, BS,MS – Massachusetts General Hospital
Kevin Staley, MD – Massachusetts General Hospital
Rationale:
It is tempting to imagine the brain as a network of neurons with stable synaptic connections that perform a fixed computation (akin to a trained convolutional neural network). However, networks of living neurons constantly vary on timescales ranging from milliseconds to years due to processes such as GABAergic inhibition, short- and long-term synaptic plasticity, changes in intra- and extracellular ion concentrations, and metabolic processes. In some cases, these changes can drastically alter the propagation of signals in the network. For example, the epileptic brain can transiently switch from “normal” function to a state of seizure. The nature of these ictogenic transitions at the cellular level remains largely unknown due to challenges in dynamically mapping connectivity with sufficient spatiotemporal sampling.
Methods:
In this work, we use the mouse organotypic hippocampal slice culture, which develops spontaneous recurrent seizures during the first week in culture and can be imaged in its entirety in a single field of view, to continuously monitor changes in functional network connectivity that emerge during transition to seizure. Specifically, we are using the red-shifted voltage sensitive fluorescent protein Varnam-2 to image voltage at >1kHz simultaneously in hundreds of cells. Functional connectivity is inferred by computing a spike-triggered average network response to action potentials in each cell (i.e. when cell X fires, what is the average response in every other cell?). Since voltage sensors are sensitive to subthreshold depolarization and hyperpolarization, we can image, for the first time, excitatory and inhibitory networks. Here we quantify connectivity in 3 epochs of epileptiform activity: postictal, interictal, and preictal. Results:
Preliminary data indicate that, as seizure approaches, excitatory connectivity increases, while inhibitory connectivity decreases. Graph theory-based analyses reveal that mean path length decreases, while clustering and maximum between centrality increase, which suggest the emergence of a small-world network just before transition to frank seizure. Conclusions:
Together, these findings support a model where a small population of key neurons transiently serve to pathologically synchronize the network. These “key cells” represent putative therapeutic targets for anti-epileptic interventions. Funding: NIH R01NS112538, P01NS127769