Among the more remarkable computational feats performed by neurons is the extraction of information from light. In the eye, retinal neural circuits process images. One of the fundamental challenges for eyes is to maintain constancy of neural product over ambient illuminations ranging from starlight to the brightest noontime sun. To cover this range separate classes of photoreceptors have evolved, rods for nocturnal vision, and cones for diurnal vision. Each in turn is served by dedicated sets of interneurons forming intra-retinal rod and cone networks. Mammals such as cats, rabbits, rats and mice provide satisfactory models for rod system circuitry. In primates cone types and cone circuits have evolved for color vision. Common laboratory mammals have a poorly developed color vision and a low density of cones. The zebrafish model provides a cone dominated retina, rich in color vision, and with the opportunity to study cone networks. Zebrafish is advantageous for ease of genetic manipulations and for extensive libraries of mutants and transgenics. It is for these reasons that, over the past decade, this lab and others have worked to develop zebrafish as a model for electrophysiological and anatomical studies of visual system circuitry. Membrane receptors for neurotransmitters are a vital component of neural circuits. In retina the receptor expression pattern is often a unique signature for different neural types, and these receptor signatures dictate the functional properties of circuits. One initial research focus was neurotransmitter responses of individual zebrafish retinal neurons, either dissociated or in retinal slice. For the former, the method was fluorescent voltage probe recording. For the latter patch electrode recording was used. Glutamatergic actions on cone bipolar cells revealed a variety of mechanisms, including metabotropic receptors, AMPA-kainate receptors, and transporter-associated chloride channels. The distribution of GABA transporters, and a variety of ionotropic GABA receptors on neural types was later uncovered. Some bipolar cells express a GABA transporter, a marker for GABAergic neurotransmission. Zebrafish horizontal cells lack GABA receptors, but a subset transport GABA and are GABAergic. As seen in voltage-probe techniques, GABA transporter signals are very large, and may even serve as a GABA sensor mechanism. In parallel with studies of neurotransmitter receptors, a morphological library of horizontal, bipolar, and amacrine cell types was developed, both through patch microelectrode staining, and through 'diolistic' staining, each technique in retinal slice. Another research phase examines neural light responses. A flattened, perfused eyecup preparation provides microelectrode access to horizontal cell and amacrine cell light responses. Cell bodies, dendrites and axons of horizontal cells were revealed in wet epifluorescence microscopy following microelectrode injection of alexafluor 594. Physiological studies reveal 6 spectral types, including trichromatic UV color opponent cells, with dominant UV cone signals being opposed by blue and green cone signals, but reinforced by red cone signals. The axons of UV trichromatic cells are longer and the dendritic fields wider than other spectral types. The morphology resembles anatomical H3 types. A UV color-opponent physiology has not been reported for horizontal cells of other species. Tetrachromatic responses were also seen. These were depolarized by UV, hyperpolarized by far blue, depolarized by blue-green, and hyperpolarized by yellow or red. As the spectral responses of horizontal cells contain so many different cone signals, the 570nm-peaking red cones, the 480nm-peaking green cones, the 410nm-peaking blue cones, and the 362nm-peaking UV cones, a model was devised to infer the signal composition. This consists of the sum of four saturable Hill functions, one for each cone spectral type. It is a three-dimensional response-wavelength-irradiance function. The model quantifies the stimulus color calculations that horizontal cells, or other zebrafish retinal neurons perform. Physiological studies of amacrine types revealed four temporal patterns: 1) Depolarizing transients at ON and at OFF. 2) Sustained depolarization. 3) A hyperpolarizing or biphasic ON response followed by a transient OFF depolarization. 4) Color opponent responses with response sign determined by wavelength. The dendritic stratification pattern of each physiological type was reconstructed from image stacks of microelectrode injected cells in Neurolucida. Each physiological type revealed unique dendritic branching: ON-OFF cells are almost exclusively bistratified within the retinal inner plexiform layer (IPL), though with several patterns of bistratification; ON cells are monostratified near the middle, or just below the middle of the IPL; OFF cells are monstratified in the distal IPL, near amacrine cells (sublamina a); color opponent cells are monostratified in the proximal IPL, near ganglion cells (sublamina b). Amacrine cell dendritic fields range from 150 to 300 microns. Some are radially symmetric. Others branch profusely only to one side of the cell body. Still others have one or more axon-like processes extending away from a central dendritic arbor, typically in the OFF IPL sublamina. The responses of ON-OFF are dominated by red cones, ON and OFF types mix red and green cone signals. Color opponent cell sample all cone types in various patterns. Recently it has become possible to study the color properties of larval ganglion cells. The most common type is UV triphasic, similar to H3 horizontal cells. It is excited by red and UV stimuli, but inhibited in the mid-spectral range. Altogether zebrafish lives up to the expectation of rich processing networks for wavelength discrimination. A further research initiative examines the labeling of retinal neurons through forward transgenic insertions of reporter genes. Typically transgenic larval eye fluorescence was noted by other labs. We determine the neural types leading to eye fluorescence. GE4a was developed in the Fumihito Ono Lab, NIAAA. In addition to populations of amacrine and ganglion cells, there is label in a select horizontal cell type, perhaps H2. The gene insertion is in a non-coding region of chromosome 14. y245 is a Gal4:UAS line developed in the Harry Burgess lab, NICHD. Red and green cones label brightly, as does a Muller cell population. More faintly marked are amacrine and ganglion cells. The transgene insertion appears to involve the Musashi1 promoter on chromosome 8. The mutation is also associated with a slow retinal degeneration of UV cones, reduced sensitivity, and aberrant spectral pattern, seen in the isolated cone responses of both larvae and adults. The thyroxin beta 2 receptor (trb2) is required for the differentiation of long wavelength cones. The Rachel Wong lab developed the transgenic crx:mYFP-2A-trb2 which overexpresses this receptor early in development, leading to the overproduction of cones with long wavelength opsin expression, a directed, dominant, transgenic mutant. In physiology, spectral sensitivity shifts towards long wavelengths for both cone PIII and ON-bipolar ( b2) ERG components. The effect persists from larvae to adults. Both photoreceptor and bipolar physiology are altered, suggesting far reaching changes in retinal processing are controlled by the trb2 nuclear receptor.
