Motion Processing in the Larval Zebrafish Tectum

Abstract

Larval zebrafish are highly visual animals that display a diverse repertoire of visually guided behaviors. Five days after their birth, they start tracking and hunting moving prey, a behavior that likely requires underlying neural circuits to analyze motion. The optic tectum, the largest structure in the zebrafish brain, is known to be involved in prey capture behavior. The specific role of this structure in motion processing is still an open question. The larval tectum receives processed inputs from direction selective retinal ganglion cells (DSRGCs). How do these inputs influence the responses of tectal neurons? Do local tectal circuits further affect tectal responses to motion? To study this, we performed in vivo two-photon calcium imaging on populations of tectal neurons and in vivo whole cell recordings while presenting larvae with moving stimuli. We show that a substantial fraction of tectal neurons are sensitive to the direction and speed of moving stimuli. Direction selectivity (DS) in these neurons is weakly correlated with RGC inputs and strongly correlated with local inhibition. The inhibition comes from the null direction of the recorded neurons and appears to be mediated by direction selective inhibitory neurons. Our data demonstrates the presence of a tectal circuit for computing the direction of motion, whose motif resembles the DS circuit in the vertebrate retina. What roles do excitatory and inhibitory tectal neurons have in motion processing? To explore this, we recorded motion responses from labeled glutamatergic and GABAergic neurons in conjunction with a newly generated pan-neuronal transgenic line expressing the genetically encoded calcium indicator GCaMP6s. We show that excitatory and inhibitory tectal neurons display a matching degree of selectivity to motion. DS inhibitory neurons seem to cluster into two populations preferring head-directed or tail-directed motion. In contrast, DS excitatory neurons form three overlapping clusters. The preferred directions of these clusters appear to be phase shifted with respect to those of DSRGC inputs. Our results show that rather than being a simple relay center for processed retinal inputs, the tectum builds direction selective responses by employing a network of highly selective interneurons. This processing appears to transform the representation of motion direction by RGCs into distinct representations by subpopulations of tectal neurons.