Item

Abstract

Purpose : The mapping of synaptic connectivity patterns using volume electron microscopy (EM) reconstructions has provided insights into the mechanisms of visual information processing in the mouse retina1. However, the neural circuitry of the primate retina differs greatly from that of the mouse. We aim to combine electrophysiological and connectomic analyses to understand how the neural circuitry of macaque retina produces the diverse and unique visual computations performed by retinal ganglion cells (RGCs).

Methods : We presented white-noise stimuli at single-cone resolution to a macaque retina and simultaneously recorded the electrical activity of RGCs using a 512-electrode array. The functional recordings revealed the visual receptive field mosaics of various RGC types: ON and OFF parasol, ON and OFF midget, small bistratified cells (SBC), and previously unidentified types. Single-cone resolution stimuli revealed individual cone locations within the receptive field of each RGC. Using multi-beam serial section EM, we then imaged this functionally characterized retina to collect a large EM volume spanning 800x1000x200 µm3. We aligned the functional with the anatomical data in 2 steps: 1) at the level of RGCs in ganglion cell layer (GCL) and 2) at the level of cones in photoreceptor layer (PRL). For the GCL, we aligned the RGC functional receptive fields with their anatomical dendritic trees. For the PRL, we aligned the cone locations in the functional receptive fields with the cone inner segment locations in the EM volume.

Results : We identified 600 somata in the GCL and 2528 cones in the PRL. Our initial analysis has focused on the anatomical identification of 4 ON parasol, 4 OFF parasol, 2 ON midget, 2 OFF midget RGCs, and 1 SBC in the recording. The bipolar cells and cones that provide synaptic inputs to an SBC are nearly completely identified. Preliminary evidence suggests that we can match many of the cones in the functional receptive field with cones in the EM volume and thus, align the functional to anatomical data within the PRL.

Conclusions : Functional and anatomical data can be aligned 1) at the GCL by locating the functionally recorded RGCs in the EM volume and 2) at the PRL by matching the functionally identified cones with the cones in the EM volume. The immediate next step is to compare the weights of cone inputs to the downstream RGCs estimated using functional and anatomical data.