Direction-Selective Neurons in the Larval Zebrafish Optic Tectum: Functional Characterization and Cohort Reliability

Abstract

The larval zebrafish optic tectum and anterior pretectum together form one of the most tractable vertebrate circuits for linking single-neuron visual response preferences to single-synapse-resolution anatomy. This thesis applies that capability to one well-characterized circuit: the direction-selective population of the optic tectum and pretectum. The aim is to produce a quantitative, atlas-anchored description of this population in a single fish, F2 (the target of an ongoing same-specimen electron-microscopy reconstruction) and to test how reliably each F2-level finding holds across a 24-fish cohort imaged in the same laboratory. Three questions structure the analysis of the two-photon calcium-imaging data. First, we ask what the functional structure of direction-selective (DS) cells looks like across F2's imaging volume. Second, if the F2 motif is reproducible across a 24-fish comparison cohort spanning incross and outcross genetic backgrounds and 6 vs. 7 days post fertilization. Third, we ask whether the contralateral-retinotopic hemispheric asymmetry predicted for L→R- vs. R→L-preferring cells holds when the analysis is restricted to genuinely motion-sensitive cells rather than cells responding to static visual content. F2 recovers a population of DS cells comprising roughly one fifth of its responsive neurons. The 24-fish cohort recovers a comparable DS fraction, which is broadly consistent with prior tectal estimates extended to the broader imaging volume. In every cohort fish, the L→R-preferring cells concentrate in the fish's right tectal hemisphere and R→L-preferring cells in the left. The asymmetry is preserved when the laterality statistic is recomputed over only motion-sensitive cells, identified using the static-visual pauses in the stimulus, demonstrating that the hemispheric pattern reflects genuine motion processing rather than static visual drive. Functional clustering over the full 32-scene stimulus profile further shows that DS classification partitions cells along an axis largely orthogonal to the full-scene clustering axis which supports a mixed-selectivity interpretation that DS cells have preferences and are not one-feature detectors. We created a 3D moveable atlas of the larval zebrafish brain with the activated DS cells to view the spatial clustering. Together these analyses constitute a same-fish functional scaffold that the F2 EM reconstruction can be anchored to, and establish the computational half of a two-arm function-to-connectome project.