Geoffrey deRosenroll

Topic

Neuronal computations supporting direction selectivity in the mouse retina

Division of Medical Sciences

Date & location

• Friday, May 3, 2024
• 10:00 A.M.
• Medical Sciences Building, Room 150

Examining Committee

Supervisory Committee

• Dr. Gautam Awatramani, Division of Medical Sciences, University of Victoria (Supervisor)
• Dr. Craig Brown, Division of Medical Sciences, UVic (Member)
• Dr. Kerry Delaney, Department of Biology, UVic (Outside Member)

External Examiner

• Dr. Richard Naud, Faculty of Medicine, University of Ottawa

Chair of Oral Examination

• Dr. Tommy Happynook, Department of Anthropology, UVic

Abstract

By the time visual information leaves the eye, it has already passed through multiple layers of neurons organized into feature-selective circuits that work to distill the analogue light signal received by photoreceptors down into diverse spike rate codes in the ganglion cells of the retina, whose axons make up the optic nerve. Taking advantage of the accessibility of the retina relative to the rest of the brain, dissecting these circuits provides great opportunities for the study of neuronal computations. One such circuit is centred around the direction-selective ganglion cell, which spikes robustly when objects move through their receptive field in particular directions and weakly or not at all in the opposite directions owing to inhibition from presynaptic starburst amacrine cells. This is supported by multiple complementary and redundant computations, both in the presynaptic starburst amacrine cells and postsynaptically in the DSGCs. In this thesis, I use computational modelling methods to complement and formalize theories based on empirical studies of these directional mechanisms as well as to form new predictions at the edge of our current understanding of the circuit.

I start by modelling the "space-time wiring" directional mechanism involving the systematic distribution of kinetically distinct bipolar cell inputs along starburst amacrine cell dendrites using physiologically derived bipolar release transients for the first time. Then, moving downstream, I demonstrate how the asymmetric wiring of starburst dendrites to DSGCs is sufficient to drive DS spiking, even in the absence of directional release of neurotransmitters from starbursts. After exploring two mechanisms generating direction-selective responses in DSGCs, I focus on improving our understanding of how the non-directional glutamatergic and cholinergic sources of excitation to DSGC dendrites support the reliable computation of direction. I show that the mediation of glutamatergic inputs by voltage-dependent NMDA receptors at low-contrasts enables a context-dependent switch between modes of neuronal arithmetic: nearly flat addition over contrast and tuning-preserving multiplication over direction. Finally, I examine how the multi-directed corelease of acetylcholine alongside GABA from starbursts means that the dominant excitation to DSGCs is highly spatiotemporally correlated with inhibition in the null directions, ensuring reliable suppression of spiking. Overall, this research highlights multiple examples of how circuit structure and function work together to support consistent neuronal computations over space and time.