Neural mechanisms of sensory encoding after photoreceptor degeneration - PROJECT SUMMARY/ABSTRACT Studies have found that patients with retinal degeneration can lose up to half of all cones before visual deficits are observed, suggesting a nonlinear relationship between cone loss and visual function. Previous studies from our lab have demonstrated that compensation can occur at the level of individual retinal circuits, however it remains unknown if all circuits react in the same way to cone loss. Loss of cone inputs disrupts stimulus detection, which relies on neurons to use both signal and noise to represent stimulus properties. Empirical and simulated neural activity of motion-sensitive cells suggest that, across neural populations, correlations in noise (variability in stimulus-driven responses) can benefit stimulus encoding. However, the impact of noise correlations in population coding and on visual behaviors is not fully understood. Remodeling of retinal circuitry following cone degeneration may disrupt these neural computations, yet current approaches to vision restoration, like stem cell replacement or electrical implants, rely on using the existing neural function after cone loss. Consequently, it is critical to understand neural computation in individual retinal circuits after retinal degeneration. To study this, we will leverage the dependence of the vertical optokinetic reflex (OKR), a visual behavior that tracks global motion in the visual space, on the ON-direction selective ganglion cell (oDSGC). Two types of oDSGCs prefer upward (Superior) or downward (Inferior) motion in visual space and form unique mosaics across the retina, thus represent overlapping regions of visual space. The central hypothesis is that cone degeneration disrupts shared cones between oDSGCs, which decreases noise correlations and decreases fidelity in the OKR response. In these studies, we will use the cone-DTR mouse model, where selective apoptosis of M-opsin cones can be induced in the adult mouse retina. We will characterize correlated noise in neighboring oDSGCs using simultaneous paired recordings to determine the relationship between common inputs and noise correlations in control and cone-deficient mice. These measures will be compared with histological and functional measures of shared cone inputs. We will measure intracellular responses of oDSGCs to determine if noise correlations are driven by inner retinal circuitry, and use pharmacological blockade to identify cell-specific sources of noise correlations. Next, we will investigate the role of noise correlations in the computations underlying the OKR using a model to test if opposing oDSGC type responses are subtracted prior to or after nonlinear pooling. We will define the extent of oDSGC pooling by determining the representation of cones in visual space that elicits the OKR using behavioral assessments of the OKR across stimulus size. Lastly, we will determine if these models can predict the OKR in response to a novel noise correlation stimulus and following cone loss. These studies will improve our understanding of the role of noise correlations in motion detection at the cellular, computational, and behavioral levels.
