Natural Odors and Mixture Encoding in the Fly Olfactory System
Author: O'Connell, Thomas F.
Year: 2027
Degree: Dissertation (Ph.D.)
Advisor: Hong, Elizabeth J.
Committee Members: Dickinson, Michael H.; Meister, Markus; Siapas, Athanassios G.; Hong, Elizabeth J.
Option: Neurobiology
DOI: 10.7907/41ax-d324
Abstract
Natural odors are typically complex mixtures of many volatiles, but this stimulus space is poorly understood, and olfactory neuroscience has largely proceeded unmoored from this critical context. I analyze the commercial Volatile Compounds in Food (VCF) database, to form one of the broadest descriptions of natural odors. We use this VCF data first to argue that natural volatiles are at much lower concentrations than typical in olfactory neuroscience.
I develop a method to image the outputs of all ~50 Drosophila olfactory receptor neuron (ORN) types, where all of one type project to a structure called a glomerulus, with stereotyped position and responses. I develop a method to identify and segment all glomeruli, using a panel of 24 odors. Our lab also developed a technique to image all of the Kenyon cells (KCs), which are two synapses downstream and involved in learned olfactory behaviors.
First, we image complete ORN and KC responses to a diverse panel of 17 monomolecular odors, and find that there is a consistent remapping of odor representations from ORNs to KCs, beyond the simple decorrelation we expected. We ask if this remapping makes odors that co-occur in the same natural sources appear more similar, by comparing to VCF distances, and we find that it does partially align KC odor representations with these VCF distances. I adapt a biophysically-plausible KC model to show that part of this remapping may be explained by known structure in the connections into the KCs.
Next, we explore mixture encoding with two complex five-component mixtures, with naturalistic distributions of component intensities, again focusing on the ORN to KC transformation. We find that surprisingly few KCs respond to the mixtures, despite so many responding selectively to components. This mixture suppression does not seem to be inherited from the ORNs, where glomeruli that respond to components tend to respond to the mix.
Files
- oconnell_thomas_2026_final.pdf (application/pdf)