Animals have a diverse set of behaviors that are triggered by specific sensory stimuli, such as motion or looming. In the visual system, this process begins in the retina where the visual scene is divided into nearly 40 parallel information channels before reaching the brain. In the mouse, the superior colliculus is one of the main recipients of retinal output and it mediates tractable visually-guided behaviors such as eye movement, orienting or escaping behaviors. However, it remains unknown how visual signals from individual retinal ganglion cell types are processed by neurons in the superior colliculus to achieve specific computations relevant to behaviors.
To answer this question, we used transgenic mouse lines in which specific types of retino-recipient neurons of the superior colliculus are labeled with Cre recombinase, enabling their monitorization and manipulation. We have established a surgical procedure to record calcium transients from Cre-labeled cells using in vivo two-photon imaging during visual stimulation, and we are categorizing the visual response properties of individual collicular cell types. Using retrograde trans-synaptic viral tracing initiated from Cre-labeled collicular cell types, we have performed two-photon calcium imaging of the labeled presynaptic retinal ganglion cell network. With this approach we are relating the activity of neurons to the activity of connected neuronal networks, and evaluating the degree of convergence and divergence in retino-collicular connectivity. Our work has revealed that one of the collicular cell types examined displays at least three different types of visual responses, possibly suggesting that this cell type can be subdivided in distinct subclasses. In addition, this study disclosed a high degree of convergence in the retino-collicular projections investigated, with 10 retinal ganglion cell types projecting to a single collicular type in an 'unbiased' manner. Lastly, we have investigated the role of individual collicular cell types in a set of visual motor behaviors by ablating specific collicular cells and have identified one cell type whose ablation leads to deficits in looming behaviour, limiting the timely detection of approaching predators from above. In contrast, these cells do not seem to control orienting movements, such as head and eye movements. By linking cell types, circuits and computations, this work will provide mechanistic insight into the circuit basis for parallel processing of visual information and various visual functions in the healthy system, and could disclose novel therapeutic targets for visual motor diseases.
