A crystal-clear zebrafish for in vivo imaging

Abstract

The larval zebrafish (Danio rerio) is an excellent vertebrate model for in vivo imaging of biological phenomena at subcellular, cellular and systems levels. However, the optical accessibility of highly pigmented tissues, like the eyes, is limited even in this animal model. Typical strategies to improve the transparency of zebrafish larvae require the use of either highly toxic chemical compounds (e.g. 1-phenyl-2-thiourea, PTU) or pigmentation mutant strains (e.g. casper mutant). To date none of these strategies produce normally behaving larvae that are transparent in both the body and the eyes. Here we present crystal, an optically clear zebrafish mutant obtained by combining different viable mutations affecting skin pigmentation. Compared to the previously described combinatorial mutant casper, the crystal mutant lacks pigmentation also in the retinal pigment epithelium, therefore enabling optical access to the eyes. Unlike PTU-treated animals, crystal larvae are able to perform visually guided behaviours, such as the optomotor response, as efficiently as wild type larvae. To validate the in vivo application of crystal larvae, we performed whole-brain light-sheet imaging and two-photon calcium imaging of neural activity in the retina. In conclusion, this novel combinatorial pigmentation mutant represents an ideal vertebrate tool for completely unobstructed structural and functional in vivo investigations of biological processes, particularly when imaging tissues inside or between the eyes.

Figure 1. Generation of crystal, a fully transparent combinatorial pigmentation mutant.

(a) Left: schematic diagram of the main populations of pigment cells in wild type (WT) zebrafish. Centre: cell lineages generating the different populations of pigment cells. Right: mutations affecting genes controlling either pigment cell formation (nacre^w2/w2 and roy^a9/a9) or melanin production (alb^b4/b4) used to generate the crystal mutant. RPE, retinal pigment epithelium. (b–d) Pigmentation phenotypes of wild type, single mutant, casper and crystal zebrafish at adult (>3 month old, (b)), embryonic (3 dpf, (c)) and larval (7 dpf, (d)) stages. Red dashed boxes indicate crystal mutants. Insets on the right display eye pigmentation phenotypes. 3 dpf and 7 dpf zebrafish treated with 200 μM PTU are shown at the bottom of (c,d). Note that the optical transparency of crystal fish is higher than that of wild type, single mutant, casper and PTU-treated fish.

Figure 2. PTU impairs zebrafish behaviour and visual system function.

(a) Schematic diagram illustrating the optomotor response behavioural assay. Individual 5 dpf larvae were positioned in a petri dish containing Danieau solution. An LCD screen controlled by a computer was used to display black and white square-wave gratings moving in 4 directions (red arrows) at the bottom of the petri dish. Each larva was tested 5 times in total (each trial lasted 6 seconds) and scored according to the trials it responded to (i.e., fish turns and swims in the direction of the moving gratings). (b) Quantification of the optomotor response assays for 5 dpf wild type (WT, black), crystal (blue) and PTU-treated (red) larvae (n = 25 larvae in each group). The frequency distribution (left), cumulative frequency distribution (centre) and mean ± SEM (right) of number of responsive trials per larva are reported. Note that PTU-treated larvae show a significant decrease in the number of trials they respond to. ns, non-significant; ***p < 0.001; one-way ANOVA with post-hoc Tukey’s HSD test. (c) Functional calcium imaging of tectal cells and retinal ganglion cell axons expressing GCaMP5G (green) in 4 dpf Tg(elavl3:GCaMP5G) larvae. Distance of right eye from projection screen is 3 cm. Recordings are performed from 3 Z-planes (approximately 20 μm total volume thickness) in the contralateral optic tectum at 4 Hz image acquisition rate. Dashed box shows the angles of moving bars relative to zebrafish larva orientation. Mean ΔF/F₀ images of calcium recordings in untreated (control) and PTU-treated larvae followed by mapping of direction-selective (DS) and orientation-selective (OS) voxels are displayed. Np, neuropil; SPV, stratum periventriculare; DSI, direction selectivity index; OSI, orientation selectivity index. Scale bar is 40 μm. (d) Average number (top) and relative frequency (bottom) of DS, OS, visually responsive and non-DS/non-OS voxels per Z-plane in control and PTU-treated 4 dpf larvae (n = 12 larvae in each group). Criteria used to identify DS and OS voxels are reported at the top. Note the dramatic reduction in visually responsive, DS and OS voxels following 200 μM PTU treatment. Error bars are ± SEM. ***p < 0.001, unpaired two-tailed Student’s t test.

Figure 3. Improved optical accessibility of zebrafish larvae in whole-brain light-sheet imaging using crystal.

(a) Volumetric imaging of the larval zebrafish brain with light-sheet microscopy in 4 dpf nacre (top), casper (middle) and crystal (bottom) Tg(elavl3:GCaMP6f) larvae (n = 4–5 larvae in each group). Six different volume sections per larva out of 450 (total xyz volume of 798 × 623 × 283 μm³) are displayed. L, left; A, anterior. Scale bar is 200 μm. (b) Single volume sections of the brain (100 μm Z plane depth) in nacre (left), casper (middle) and crystal (right) larvae. Note the dark region between the eyes of nacre and casper larvae due to the excitation light being absorbed or reflected by pigments present on the surface of the eyes. L, left; A, anterior. Scale bar is 100 μm. (c) Insets showing the labelling of amacrine and ganglion cells in the left retina of the crystal larva (right) compared to nacre (left) and casper (middle) larvae, where no GCaMP6f fluorescence is detected in the eyes. T, temporal; N, nasal. Scale bar is 50 μm. (d) 3D reconstructions of the brain (lateral view) of nacre (left), casper (middle) and crystal (right) Tg(elavl3:GCaMP6f) larvae shown in (a). Note the improved optical accessibility (~18% of brain volume) allowed by crystal. P, posterior; V, ventral. Scale bar is 100 μm. (e) Average imaged brain volume in nacre, casper and crystal 4 dpf Tg(elavl3:GCaMP6f) larvae. Error bars are ± SEM. ns, non-significant; *p < 0.05; **p < 0.01; one-way ANOVA with post-hoc Tukey’s HSD test.

Figure 4. Calcium imaging of visually evoked neural activity in the retina using crystal.

(a) Two-photon functional calcium imaging of amacrine cells and ganglion cell dendrites expressing GCaMP6f (green) in 4 dpf crystal Tg(elavl3:GCaMP6f) larvae (n = 8 larvae). Distance of the eye from LCD screen is 2 cm. Recordings are performed from 2–4 Z-planes (approximately 20 μm total volume thickness) at 4 Hz image acquisition rate. Dashed box shows the angles of moving gratings relative to zebrafish larva orientation. INL, inner nuclear layer; GCL, ganglion cell layer; IPL, inner plexiform layer. Scale bar is 20 μm. (b,c) Mean ΔF/F₀ image of a representative calcium recording (b) followed by voxel-wise analysis of direction and orientation selectivity of visual responses (c). DSI, direction selectivity index; OSI, orientation selectivity index. (d) ΔF/F₀ calcium traces during a representative tuning experiment from the 6 selected regions of interest (ROIs) shown in (b). Polar plots showing the tuning profiles (i.e., integral ΔF/F₀ responses to different stimuli) of individual ROIs are reported on the right. Stimulus epochs are shown in grey. Dark arrows indicate the different directions of gratings motion. The blank-screen null condition is indicated by a ‘-’ sign.