Zebrafish colour vision: a survey of spectral processing in the zebrafish visual system

The zebrafish possesses a highly complex and utilised visual system. Its input is comprised of four distinct cone types as well as one rod type. However, in larvae, rods are thought to be immature. Accordingly, in their larval form, all visual input to the retina and brain circuits comes from the four cones. These feed into morphologically and functionally distinct and highly diverse circuits to ultimately drive a wide array of visual behaviours. While spatio-temporal processing in larval zebrafish has been studied at considerable depth, comparatively much less is known about their spectral processing. The goal of this thesis was to systematically map the physiological responses of most visual neurons in larval zebrafish – from cones via bipolar and ganglion cells to brain neurons - to stimuli that vary in wavelength. Specifically, we used 2-photon Ca2+imaging of light-driven activity both in the retina and the brain (Chapters 3,4,5). We used an original 2-photon microscope modification to allow for fast multi-plane imaging (Janiak et al, 2019). Stimulation was carried out using a custom-built high-speed monochromator (Belusic et al, 2016) with a high spectral resolution (Chapter 2). Transgenic lines expressing Ca2+-sensors in specific cell populations were generated to selectively observe different neuron types. We imaged cones and bipolar cells in the eye, and ganglion cell terminals as well as the somata of central neurons in the brain. Together, this served to establish a large-scale overview of the spectral computations that are performed at each stage, and how they may aid zebrafish visual functions. In the six chapters of this thesis I: 1. Introduce colour vision in the zebrafish. 2. Describe the construction of the visual stimulator. 3. Describe and discuss the photoreceptor data. Cones transform chromatic signals in a Principal Component Analysis-like manner. This can be part-explained as an adaptation to the spectral characteristics of the visual world (Zimmermann et al, 2018). Previously, such a transformation was thought to occur first in downstream circuits. 4. Describe and discuss the bipolar cell data. Synaptic terminals form several functional clusters that are highly wavelength-dependent, enabling possibilities for complex spectral coding. Spectral opponency is observed in several clusters. Moreover, eye-wide regional specializations are observed, in agreement with prior reports (Zimmermann et al, 2018). I conclude by discussing the functional layering of the inner plexiform layer of the retina in the context of predicted functional wiring from cones. 5. Briefly describe the ganglion cell axonal and brain somatic data. Unlike bipolar cells, their responses are surprisingly uniform. I hypothesize that the response profile is uniquely sensitive to near objects based on data from hyperspectral imaging. 6. Summarise the findings in light of the wider literature. I speculate about the overarching goal of colour vision, relate the findings back to Wilkins & Osorio (2019) and certain logic-related considerations.