Recent developments in brain imaging have enabled the emphasis to shift toward population BAY 73-4506 in vivo encoding. However, these studies

have focused almost exclusively on postsynaptic processing of sensory input. Very little is known on a population basis of the functional input delivered by sensory neurons to a given CNS target. The exception to this rule is the olfactory bulb, where defining the response properties of many glomeruli simultaneously at the sensory input level has provided direct insight into spatiotemporal coding of odorant stimuli (Friedrich and Korsching, 1997; Soucy et al., 2009). Here we provide the first systematic description of the form, organization, and dimensionality of the population of visual inputs to the brain using the optic tectum of larval zebrafish as a model. At 6 days postfertilization, zebrafish larvae are translucent and exhibit a repertoire of complex visually guided behaviors, making them an excellent model system for imaging studies of visuomotor transformations (Portugues and Engert, 2009). At this stage check details of development, the larval brain is also small in terms of physical size and number of neurons, allowing activity patterns across a substantial fraction of neurons in the brain to be imaged in a single field of view using optical approaches (Niell and Smith, 2005; Sumbre et al., 2008). The optic tectum, which is used to guide behaviors such as prey capture

and predator and obstacle avoidance (Gahtan et al., 2005), has four distinct retinorecipient laminae, and as a rule, the axon of a single retinal ganglion cell (RGC) is restricted to a single lamina (Xiao and Baier, 2007). To examine the nature of the visual input to the tectum, we fused the genetically encoded calcium sensor GCaMP3 to the synaptic vesicle protein synaptophysin (Tian et al.,

2009) and generated a stable transgenic line of zebrafish that expresses the resulting probe (SyGCaMP3) specifically in RGCs. This allows us to record visually evoked calcium transients in presynaptic terminals of RGC axons in the intact zebrafish brain. The same strategy has been adopted previously using a synaptophysin-GCaMP2 fusion to study population activity of bipolar cells in the zebrafish retina (Dreosti et al., 2009; Odermatt et al., 2012). Furthermore, we developed whatever an unbiased voxel-wise analysis strategy that permits functional characterization of the retinal input independent of RGC axon or tectal neuropil morphology and at a spatial scale below that of a presynaptic bouton. This not only allows visual selectivity to be determined on a voxel-by-voxel basis, but also describes visual input to the tectum on a population basis. We have used these techniques to characterize responses to drifting bars. We have identified three subtypes of direction-selective input and two subtypes of orientation-selective inputs.