Rapid escape responses are critical for predator avoidance in fish. Yet, while short-latency C-start (SLCs) circuitry is well-known (e.g., Mauthner and related cells), neurons integral to long-latency C-starts (LLCs) remain uncharacterized. In this dissertation, I identify neurons critical for LLC through the genetic and laser ablations of neurons in transgenic lines.

Although transgenic lines provide powerful tools for implicating neurons in behavior, they suffer a number of limitations. Transgene expression is frequently broad, incompletely mapped, or off-target, making it difficult to accurately compare en masse or

to other modalities. I addressed this by designing a UAS reporter that suppresses off-target expression through microRNA binding and building a digital atlas from hundreds of transgenic zebrafish lines. By co-imaging and registering lines with a broadly expressed structural marker, the Zebrafish Brain Browser aligns expression to within approximately one cell diameter allowing rapid and accurate comparison of expression, identification of transgenes, and prediction of genetic overlap in almost any set of cells in the larval zebrafish brain. Other modalities (e.g., neural activity and anatomic segmentation) were also incorporated from Z-Brain, another popular zebrafish brain atlas, by a novel multichannel secondary registration. Together, this work increases the fidelity, interoperability, and accessibility of brain atlases and provides a powerful platform for the dissection of neural circuits in larval zebrafish.

Using these tools to design and analyze genetic ablations, I performed a 'circuit-breaking' screen to identify neurons underlying LLC behavior. Three of the screened lines reduced LLC probability by >50%. These lines labeled two shared cell clusters: one adjacent to the locus coeruleus (LC) and another in the dorsal hindbrain. Through laser ablation and optogenetic stimulation, LC-adjacent neurons were shown to be both necessary and sufficient for LLC startle. Projections of individual LC-adjacent neurons were characterized by a novel genetic intersection approach. These neurons were strikingly homogeneous, projecting bilaterally to midbrain and hindbrain structures. From this work, I hypothesize that ipsilateral hindbrain projections activate premotor neurons, while contralateral neurites subserve reciprocal inhibition. For the first time, I have identified a core component of the circuit mediating long-latency C-starts, an ethologically important behavior in zebrafish.