How specific classes of retinal cells contribute to vision : a computational model

Abstract

The human brain can recreate images by combining parallel streams of information emitted by about one million retinal ganglion cells (RGCs). RGCs exhibit an astonishing functional, anatomical, and molecular diversity and their preference for particular features of the visual scene (contrast, motion, etc.) can be attributed to synaptic connectivity patterns from upstream retinal circuits as well as intrinsic characteristics (such as gene expression, morphological features, membrane properties). However, how these different attributes give rise to distinct functional groups is still largely unknown. In this thesis, we investigated the functional properties of specific RGCs subgroups, sharing gene expression, by applying experimental and theoretical approaches to control their neuronal activity using pharmacogenetics. We hypothesised that modifying their activity may not only affect their individual response but also their concerted activity, thereby elucidating their role in population encoding of visual scenes. To explore this hypothesis, we worked on three main axes: 1. General response characterisation of RGCs in control condition and when their activity is altered through pharmacogenetics. 2. Development of a mathematical model, constrained by empirical data, to unravel the circuit wiring underlying functional diversity. 3. Large-scale simulations of the model on Macular, a novel simulation platform, to explore retinal behaviour to complex stimuli. In this context, we analysed light responses recorded from mouse RGCs and we identified distinct cell types that respond in diverse patterns when their activity is pharmacologically modified. We hypothesised that these various response patterns may arise from lateral interactions between the different RGC types. We tested this hypothesis by means of model definition, mathematical analysis, and numerical simulations and illustrated the role of connectivity patterns in the behaviour of the system. Taken together, our work suggests possible physiological mechanisms underlying the variability of RGCs responses with an emphasis on the role of lateral connectivity on the retinal response.