The bipolar cell terminal as a selective spatio-temporal filter

The visual signal impinging on the photoreceptor array is passed through a series of spatio-temporal filters at the level of bipolar and ganglion cells before being passed to the brain as ~20 parallel representations of the visual world. Key processing steps occur within the retina’s two synaptic layers, the outer and inner plexiform layers (OPL, IPL). In the OPL of the mouse retina, dendrites of ~10-12 different types of bipolar cells integrate visual information forwarded from photoreceptors, under modulatory control of horizontal cells, setting up the initial building blocks from which complex “trigger features” are subsequently extracted by retinal ganglion cells in the IPL. Here, lateral inputs from amacrine cells directly impinge on ganglion cell and bipolar cell synapses further shaping the visual signal. To understand the visual signal harbored in BC terminals is therefore pivotal to our understanding of retinal circuit function. We characterized the ex-vivo light-response properties of mouse BC types by calcium imaging of synaptic terminals in the whole-mount retina and found that BCs can be reliably clustered based on their temporal properties and their propensity to generate spikes alone. The identified functional BC outputs are organized in a kinematopic map across different IPL strata, with fastest projecting centrally and slower cells projecting to either edge. Moreover, the fastest BC types generate all-or-nothing spikes, thereby enhancing time-precise processing of visual signals in the retina. But bipolar cells typically possess in the order of 20-30 individual axonal terminals, with potentially different intrinsic properties, at different electrotonic distance from the axon, each under modulatory control by a potentially different subset of amacrine cells. This suggests that individual bipolar cell terminals belonging to the same cell may represent differential space-time filters, thereby vastly expanding the wealth of signal diversity available to RGCs towards the extraction of highly specific trigger features. Indeed, already very basic properties of individual synapses, such as presynaptic bouton size fundamentally impact temporal processing by individual terminals (work on fish, collaboration with L Lagnado). Using 2P calcium imaging of light-evoked signals within individual terminals belonging to the same mouse BC we show that mouse bipolar cells systematically multiplex visual stimuli into distinct parallel channels represented by individual terminals (collaboration with R Smith and WR Taylor). Preliminary data suggests that key differences between signals forwarded by different terminals extend to the kinetic and polarity domain, rather than the spatial domain. Finally, we also estimate spatiotemporal receptive fields in individual postsynaptic sites along RGC dendrites under the same experimental conditions used for recording from BC terminals and complement our approach with population imaging of light-evoked calcium responses within RGC somata towards a more complete understanding of temporal processing in the mammalian retina