Abstracts

Rationale: Many gross anatomical and physiological features of post-traumatic epileptogenesis have been characterized (e.g. mossy fiber sprouting, a latent period of seizure freedom preceding chronic seizures). However, very little is known about the epileptogenic changes in connectivity at the neuronal level. We have developed a method for chronically imaging neuronal activity throughout epileptogenesis, with single-neuron resolution, across a field of view encompassing the entire hippocampal slice. Here we present algorithms and preliminary results for quantifying functional network architecture in these datasets by looking at cells that fire together during interictal periods of relatively asychronous activity. Methods: We used high-sensitivity, red-shifted, genetically encoded calcium indicators (GECIs) to analyze functional connectivity, in spontaneously epileptic organotypic slice cultures, during interictal periods of relatively asynchronous activity. By imaging sparsely-expressed fluorescent proteins using low-magnification objectives, we were able to record from an entire hippocampal slice culture with single-neuron resolution and single action potential sensitivity. Furthermore, interneurons were labelled by co-expression of green fluorescent protein in DLX-expressing cells. For each of ~500 neurons imaged per recording, an output map was computed by identifying "follower neurons" that activated within 100ms of the index neuron. Since the entire slice was imaged, computing output maps inherently generated input maps for all visualized neurons. Using these data we quantify functional network architecture to look at 1) input maps for interneurons vs principal cells, 2) the relationship between input and output maps, 3) the spatial distribution of functional connections. Results: In previous work, we have demonstrated that functional connectivity changes dynamically with network state. For example, pre-ictal disinhibition unveils scale-free functional network connectivity in seizing brain slices (Lillis et. al. 2015). Preliminary data analyzed here suggest that: 1) with the higher sensitivity GECIs used in this study, small-world network structure is apparent in "resting state" calcium imaging data; 2) there is a positive correlation between the number of inputs and outputs a given principal neuron exhibits; and, 3) interneurons have more functional inputs than outputs (presumably because their outputs decrease rather than increase the probability of action potentials in post-synaptic neurons). Conclusions: Our analyses demonstrate that functional network connectivity can be computed from "resting state" calcium imaging data. Although this approach identifies connections based on correlation, which does not guarantee causation, it represents a stride towards the ultimate goal of functional network mapping in living tissue. In a companion poster by Theju Jacob, we apply the functional connectivity algorithms presented here to a computer model of a large-scale seizing network, with known connectivity, to look for a relationship between functional and anatomical connectivity. Funding: We are grateful for funding from the NIH (R01NS034700), which supported this research.