Gain Control and Sparse Representations in the Olfactory System of the Locust and Fly

Citation

Papadopoulou, Maria (2010) Gain Control and Sparse Representations in the Olfactory System of the Locust and Fly. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/WSYW-KA90. https://resolver.caltech.edu/CaltechTHESIS:05292010-035212206

Abstract

The giant GABAergic neuron (GGN) is a single, paired, non-spiking neuron that arborizes extensively in the mushroom body (MB) of the locust (Leitch and Laurent, 1996), where it overlaps with the dendrites and the axons of Kenyon cells (KCs). KCs are the intrinsic neurons of the MB and are thought to be important for olfactory learning and memory (Davis, 2004). We are interested in understanding the function of GGN in olfactory processing: in particular, its pattern of arborization makes it an attractive candidate for controlling or modulating KC responses to odors, with potential implications for learning and recall. Physiological recordings of KCs in the locust show that these neurons respond sparsely to odors, in marked contrast to their excitatory inputs from the antennal lobe (projection neurons or PNs) (Laurent and Naraghi, 1994; Perez-Orive et al., 2002; Stopfer et al., 2003; Mazor and Laurent, 2005). Inhibition appears to be critical to control KC response threshold, probability and duration during odor stimulation (Perez-Orive et al., 2002). We show that there exists a feedback loop whereby KCs provide excitatory input to GGN , which in return provides inhibitory control of KC excitability. We further demonstrate that manipulating GGN during olfactory stimulation affects odor-evoked subthreshold oscillations, as measured in individual intracellularly recorded KCs, or by a more global measurement of the local field potential. We also assess the influence of GGN by recording from a population of extrinsic MB neurons that receive input from KCs in the β-lobe of the MB (β-LNs).We show that GGN can suppress KC activity to such an extent as to eliminate all spiking in the downstream neurons. With these experiments in the locust, we show that GGN controls the gain of PN-to-KC information transfer and normalizes the output of the KC-population, much reducing its dependence on input strength.

With experiments in Drosophila Melanogaster we try to extend the generality of GGN as a solution for gain control in the MB. Specifically, we carry out intracellular recordings from a Drosophila neuron discovered by Greg Jefferis, which has extensive arborizations throughout the MB calyx and lobes, resembling the locust GGN. We show that this cell is GABAergic, that it is a non-spiking neuron, and that its response to olfactory stimulation is very similar to that of the locust GGN, including a graded response to increasing odor concentration.

Thesis Files