Tools for Higher Dimensional Study of the Drosophila Larval Olfactory System

Abstract

Olfaction is one of the most behaviorally important sensory modalities, yet it is one of the least studied. Many organisms rely on olfactory cues as the primary sensory input for decision making. And yet, study of olfactory perception has been hindered by a lack of tools for efficiently delivering diverse odorants as well as an inability to understand the properties of the olfactory space. This thesis will describe several tools developed to provide better control and flexibility over olfactory cues and to to provide improved methods for analyzing and interpreting neural activity data. Odors evoke responses from olfactory receptor neurons (ORNs), which are processed in olfactory glomeruli in a brain region called the antennal lobe in insects and the olfactory bulb in vertebrates. These signals are then relayed by projection neurons (PNs) to higher brain centers, including mushroom body in insects and the piriform cortex in vertebrates. We choose as a model system the Drosophila larva, an organism with a centralized brain and layered neuronal architecture but with significantly lower numerical complexity in comparison with vertebrates and even the adult Drosophila. We describe tools used for analysis of neural calcium imaging data including motion correction of recordings of various regions of the larval brain and semi-automated segmentation of multi-neuronal recordings of dendritic calcium activity. We apply these strategies to the study of larval local neurons that form synapses with ORNs and PNs in the antennal lobe. We measure diverse odorant responses to a population of five local neurons but see no effect of these neurons on the olfactory population code. We describe a model developed to estimate parameters for odorant-ORN dose-response curves that incorporates animal-to-animal variability. Finally, we describe a powerful tool for imprinting arbitrary activity patterns on the ORN population code of the larva. This is achieved by identifying a panel of odorants tuned to activate specific ORNs and developing a microfluidic mixing device that allows for arbitrary mixing of up to ten odorants simultaneously. We demonstrate this method by imaging odor-evoked responses of Kenyon cells, neurons that receive olfactory input from PNs and project it to the mushroom body. We are able to identify functional connections of Kenyon cells and can estimate relative connection strengths. We further use the pattern generator to study the Keystone interneuron, another inhibitory antennal lobe neuron. We find that the Keystone's neuropil integrates mixture of odorants nonlinearly, with specific regions of the antennal lobe showing either enhancement or suppression of mixtures, dependent on the molecular structure of the odorants delivered. The olfactory pattern generator will support higher dimensional study of the larva olfactory system by allowing the number of available olfactory stimuli to be comparable to or even greater than the number of olfactory neurons. With these tools, we can develop a better understanding of olfaction and improve our ability to study high dimensional sensory spaces.