What this is
- Zebrafish retinas contain one rod and four cone subtypes, making them suitable for vision research.
- This study characterizes the photoresponse properties of these photoreceptor types using suction electrode recordings.
- Key findings include the higher sensitivity of rods compared to cones and distinct response kinetics among cone subtypes.
Essence
- Rods in zebrafish exhibit 40–220× higher photosensitivity than cones, but their flash responses are smaller. Green-sensitive cones show 5.5× higher sensitivity than red cones, with response kinetics varying significantly among cone subtypes.
Key takeaways
- Rods demonstrate 40–220× higher photosensitivity than cones, indicating their critical role in low-light vision despite smaller flash responses.
- Green-sensitive cones have 5.5× higher sensitivity than red-sensitive cones, suggesting functional specialization among cone subtypes.
- Response kinetics are fastest in blue- and red-sensitive cones, which are approximately 2× faster than green- and UV-sensitive cones, and 6.6× faster than rods.
Caveats
- The study primarily focuses on adult zebrafish, which may limit the applicability of findings to other developmental stages.
- Flash response measurements were conducted under specific conditions, which may not fully replicate natural environments.
AI simplified
Introduction
Zebrafish is commonly used in life science research including fundamental studies of rod and cone photoreceptor cells. For example, recent studies have explored the following topics: cell fate specification1, mRNA and protein expression profiles2,3, mitochondrial homeostasis4, circadian regulation5–8, regeneration from injury9–12, visual cycle13–16, visual pigments17,18, and live imaging of retinogenesis19 and photoresponses20. Compared to the nocturnal mouse retina which is rod-dominant (97% rods21) and has only green- and UV-sensitive cone subtypes, diurnal zebrafish has cone-dominant retinas at larval stage (92% cones22) and ~ 60%-cone retinas at 3–6 months adult stage23, with four cone subtypes (UV, blue, green, and red). Thus, zebrafish studies can expand our understanding of cone photoreceptor function.
Electrophysiological photoresponses of zebrafish rods and cones have been described in previous studies, mainly by electroretinography (ERG) recordings. In vivo ERG responses have been recorded by placing an electrode just above the cornea of an anesthetized zebrafish24–29, or by inserting an electrode into the eye of an anesthetized adult fish30. Ex vivo ERGs have been recorded also by inserting an glass pipette electrode into an isolated larval eyeball5,31, or by placing the electrode inside the adult eyecup31,32. In some of these ERG studies, the photoreceptor response was obtained by blocking the b-wave pharmacologically5,27,31,32. Single-cell recordings have been reported for rods33, and each cone subtype: UV34, blue35, green36,37, and red cones13,38. However, a complete set of photoresponse characterization from the one rod and four cone classes in the zebrafish retina has not been described yet.
Here, we characterize the flash response properties of rods and cones in adult zebrafish retinas obtained by single-cell suction and ex vivo ERG recordings. Using suction recordings, we collected flash response families and determined spectral sensitivity, intensity-response relationships, and response kinetics from the five subtypes of zebrafish photoreceptor cells. Using ex vivo ERG recordings, we found that rod and cone mixed ERG a-wave is observed in bright light responses. These data provide a solid foundation for future applications of zebrafish in ophthalmology and vision science.
Results
Zebrafish rod and cone cellular morphology and spectral sensitivity
First, we sought to establish the single-cell suction recording method from the five different subtypes of zebrafish photoreceptor cells. Fish were kept in darkness overnight, and retinas were dissected out under dim red and infrared illumination. For cone recordings, the retina was chopped into small pieces by a razor blade, and cells were dissociated by pipetting them in a plastic tube and then transferred into the perfusion chamber. Under a microscope with infrared illumination, we found double cones and single cones on the bottom of the chamber (Fig. 1). For single cones, an outer segment could be drawn easily into the recording pipette. However, for double cones, this was technically challenging because the two outer segments are physically too close to isolate one of them. Therefore, the ellipsoid part of a cone was drawn to obtain inverted current responses, which were re-inverted in the later data analysis. After cell dissociation, many cones had damaged appearance characterized by rough surface texture (Fig. 1f). Those damaged cones produced small sluggish responses, or no response at all. Therefore, recordings were performed only from cones with smooth surfaces.
Double cones are known to be composed of red-sensitive (or long-wavelength sensitive) and green-sensitive (middle-wavelength sensitive) cones, and single cones are blue-sensitive (short-wavelength sensitive 2) or UV-sensitive (short-wavelength sensitive 1)39. We confirmed this by spectral sensitivity analysis. The λmax of the long-double cone (also known as principal member), which has a larger ellipsoid, was estimated at 553 ± 2 nm (Fig. 2, Table 1), corresponding to the reported value for red-sensitive cones (565 nm17 or 549 nm40). The λmax of the short-double cone (accessory member) was at 484 ± 2 nm (Fig. 2, Table 1), which is almost identical to the published value for green-sensitive cones (482 nm17,40). Intriguingly, the sensitivity of green-sensitive cones at 675 nm was 10-fold higher than the value predicted from fitting the spectral template. Similar observation was reported for UV-sensitive cones34, which is explained by the ectopically expressed green- or red-sensitive pigment, or both, in a UV cone. However, our data could not be fitted well with a red and green mixed spectrum (Fig. 2c). Blue- and UV-sensitive single cones are known as long and short single cones, respectively3. After cell dissociation, they were hard to distinguish based on morphology but could be easily identified by their characteristic spectral sensitivity (Fig. 2, Table 1): cones showing estimated λmax at 401 ± 2 nm were identified as blue-sensitive, and cones showing estimated λmax at 374 ± 1 nm were identified as UV-sensitive. These values do not agree well with the reported λmax of blue-sensitive cones (411 nm17 or 425 nm40) and UV-sensitive cones (361 nm17 or 342 nm40), but correspond to them. Therefore, in the following experiments, we determined the cell type of single cones based on their spectral sensitivity. In practice, cones were identified by comparing their responses to 405 nm and 450 nm flashes with the same photon density.
For rod recordings, the chopped retinal fragments were directly transferred into the chamber (Fig. 1e). The cell dissociation step was skipped to reduce possible mechanical damage. The λmax of rods was estimated at 505 ± 1 nm (Fig. 2, Table 1), which is almost identical to the reported values for rods (503 nm17 or 502 nm40). Taken together, these results demonstrate successful establishment of the single-cell suction recording method from the five photoreceptor cell types in the zebrafish retina and characterization of their specific action spectra. Hereafter, we characterize the sensitivity and response kinetics of these photoreceptor types.
Microscopic images of dissociated zebrafish photoreceptor cells and a retinal fragment. The right picture in each panel shows a cell after being drawn into the recording pipette. (and) A double cone used for recordings from a red-sensitive cone (), or a green-sensitive cone (). R: a red cone, G: a green cone. (and) A single cone identified as a blue-sensitive cone () or an UV-sensitive cone (). () A retinal fragment used for rod recordings. () Examples of damaged cones with rough surface texture. a b a b c d c d e f
() Spectral sensitivity plots. Sensitivity data were fitted with the Govardovskii nomogram for the α-band of visual pigment absorption using parameters described in Table. Data are mean ± SEM, but error bars are too small to be visible for most of the plots. () Normalized fitting curves with estimated λvalues. () Upper: Residual to the fitting in the lower panel. Lower: plots of green-sensitive cones fitted with red and green mixed spectral template (black). Green (λ484 nm) and red (553 nm) spectral templates were mixed with scaling factors of 2.29 × 10and 2.80 × 10, respectively. However, the curve did not fit well to the plots. a b c 1 max max −2 −4
| λ(nm)max | Estimated flash sensitivity(10pA photonsµm)−3−1−2 | n | |
|---|---|---|---|
| Red-sensitive cones | 553 ± 2 | 5.7 ± 0.5 | 5 |
| Green-sensitive cones | 484 ± 2 | 26 ± 7 | 5 |
| Blue-sensitive cones | 401 ± 1 | 16 ± 1 | 5 |
| UV-sensitive cones | 374 ± 1 | 1.0 ± 0.2 | 9 |
| Rods | 505 ± 1 | 570 ± 42 | 5 |
Intensity-response relationships of zebrafish rods and cones
Flash response families (Fig. 3) were obtained from dark-adapted rods and cones to analyze their maximum response amplitude and sensitivity (Fig. 4, Table 2). The wavelength of stimulation for each cell type was selected based on their respective λmax. However, UV-sensitive cones were stimulated with 405 nm violet light because our equipment did not have a UV light source. Based on the reported absorption spectra of live cells obtained by microspectrophotometry17, the absorption of the UV cone pigment at λmax is 5.5-fold higher than that at 405 nm. Therefore, we added curves (Fig. 4a,b, light purple) and values (Table 2) adjusted for that 5.5-fold difference. Despite their larger outer segment size (Fig. 1e), rods showed the smallest response amplitude among the five photoreceptor subtypes (Fig. 4a, Table 2, Rmax). As expected, rod sensitivity was 40–220-fold higher than that of cones based on their half-saturating flash intensity values (Fig. 4b, Table 2, I1/2). The kinetics of the flash response was clearly slower in rods than in cones, as quantitatively compared in the dim flash response analysis below. Among the four cone subtypes, red- and blue-sensitive cones showed larger responses than green- and UV-sensitive cones (Fig. 4a and Table 2, Rmax). Intriguingly, the I1/2 sensitivity of green-sensitive cones was 5.5-fold higher than that of their fellow red-sensitive cones (Table 2). The I1/2 sensitivity of UV-sensitive cones to 405 nm light was the lowest among the four cone subtypes. However, after the aforementioned adjustment to their λmax, the sensitivity of UV cones was comparable to that of blue-sensitive cones and in between green- and red-sensitive cones (Fig. 4b, light purple curve). The photoresponses of red- and blue-sensitive cones were faster than those of green- and UV-sensitive cones, as compared quantitatively in the next section.
Representative response families from the five zebrafish photoreceptor subtypes obtained by single-cell suction recordings. Flash wavelength and intensity: red-sensitive cones (Red), 575 nm, 32–320,000 photons μm, red trace: 1000 photons μm; green-sensitive cones (Green), 500 nm, 3.2–10,000 photons μm, green trace: 100 photons μm; blue-sensitive cones (Blue), 405 nm, 10–32,000 photons μm, blue trace: 320 photons μm; UV-sensitive cones (UV), 405 nm, 100–320,000 photons μm, violet trace: 3200 photons μm; rods, 500 nm, 1–1000 photons μm, gray trace: 10 photons μm. −2 −2 −2 −2 −2 −2 −2 −2 −2 −2
Intensity-response relationships of zebrafish photoreceptor cells in absolute values () and in normalized values (). The light purple curve is obtained after adjustment for the 5.5-fold higher sensitivity at λcompared to 405 nm for the UV cones. Plots are fitted with Naka-Rushton equation described in Methods and parameters shown in Table.andare illustrated for rod data with dashed lines. Data are mean ± SEM. Numbers of data points are shown in Table. a b max max 1/2 2 2 R I
| (pA)Rmax | (photons μm)I1/2−2 | Hill coefficient | (pA photonsμm)Sf−1−2 | Fractional(10photonsμm)Sf−4−1−2 | n | |
|---|---|---|---|---|---|---|
| Red-sensitive cones | 13 ± 1 | 1600 ± 400 | 1.00 ± 0.02 | 0.011 ± 0.002 | 8.2 ± 1.4 | 19 |
| Green-sensitive cones | 7.5 ± 0.7 | 290 ± 50 | 0.86 ± 0.02 | 0.038 ± 0.006 | 62 ± 12 | 22 |
| Blue-sensitive cones | 11.5 ± 0.7 | 660 ± 60 | 0.89 ± 0.01 | 0.020 ± 0.003 | 17 ± 2 | 18 |
| UV-sensitive cones* | 6.0 ± 0.6 | 5090 ± 460 (930 ± 80) | 0.94 ± 0.02 | 0.0011 ± 0.0002 (0.0061 ± 0.0011) | 2.1 ± 0.3 (12 ± 2) | 18 |
| Rods | 5.7 ± 0.3 | 7.3 ± 0.6 | 1.39 ± 0.04 | 0.43 ± 0.04 | 760 ± 70 | 20 |
Dim flash response kinetics of zebrafish rods and cones
Dim flash responses were collected from the five subtypes of photoreceptor cells to compare their response kinetics (Fig. 5). A dim flash is defined as a flash intensity in the linear region of the intensity-response relationships, which is typically intensity producing a response with amplitude < 30% of Rmax. Based on the appearance of normalized and averaged dim flash responses, the kinetics of the responses was in the following order: red- and blue-sensitive cones > green- and UV-sensitive cones > rods. For quantitative comparison, three kinetic parameters—time to peak (Tp), integration time (Tint), and recovery time constant (τrec)—were obtained for each cell type (Table 3). Integration time is the area of the normalized response, and recovery time constant is the time constant of the exponential fit to the recovery phase of the response (Fig. 5, Kinetic parameters). Based on Tint, the responses of red- and blue-sensitive cones were approximately 2-fold faster than the responses of green- and UV-sensitive cones, and 6.6-fold faster than the responses of rods. Analysis of Tp and τrec showed a similar trend.
Normalized and averaged dim flash responses of zebrafish photoreceptor cells. Data are mean ± SEM. The numbers of cells for each averaged trace are shown in Table. Definition of kinetic parameters are illustrated in the upper right panel. 3
| (ms)Tp | (ms)Tint | (ms)τrec | n | |
|---|---|---|---|---|
| Red-sensitive cones | 102 ± 5 | 120 ± 8 | 59 ± 5 | 19 |
| Green-sensitive cones | 180 ± 9 | 236 ± 16 | 125 ± 17 | 22 |
| Blue-sensitive cones | 119 ± 6 | 141 ± 7 | 63 ± 6 | 18 |
| UV-sensitive cones | 145 ± 8 | 241 ± 12 | 109 ± 11 | 14 |
| Rods | 413 ± 11 | 833 ± 26 | 486 ± 32 | 20 |
Ex vivo ERG a-wave recordings from the zebrafish retina
Our group previously developed an ex vivo ERG adaptor for mouse retinas41. When used with inhibitors to block signals from downstream neurons and Müller cells (see Methods for details), the adaptor gives the photoreceptor component of the transretinal voltage response, as reported in mouse and human retina studies41–43. Here, we collected ex vivo ERG flash responses from the zebrafish retina (Fig. 6a). Two component flash responses were observed with bright flashes. The slow component detected with low intensities was identified as the rod component, and the fast component observed with 100 photons µm−2 and brighter flashes was identified as the cone component. This threshold intensity of cones is broadly consistent with the intensity-response relationships obtained by single-cell suction method (Fig. 4). Rmax and I1/2 were estimated for rod and cone components by fitting estimated amplitude plots with the Naka-Rushton function. Cone Rmax was 72 ± 10 µV (n = 7, mean ± SEM), which was 2.9-fold larger than the corresponding rod Rmax (25 ± 4 µV). Rod I1/2 was 13 ± 2 photons µm−2, which was about twice the value obtained from single-cell suction recordings (7.3, Table 2). This difference could be caused by absorption or scattering of the test flashes in the acrylic ex vivo ERG adaptor. Cone I1/2 was 425 ± 76 photons µm−2, some 33-fold larger than rod I1/2. This is consistent with the 40-fold difference between the sensitivities of green cones and rods observed in the single-cell suction analysis (I1/2 of 290 and 7.3 photons µm−2, respectively, Table 2). Among responses to 405 nm, 500 nm, and 575 nm flashes at the same 32,000 photons µm−2 intensity, 405 nm flashes gave the largest cone response (Fig. 6c). This is because all cone subtypes respond to 405 nm violet light as shown in their spectral sensitivity plots (Fig. 2; perhaps counterintuitively, sensitivities to 405 nm light are comparable in UV- and red-sensitive cones), and 32,000 photons µm−2 flash is expected to produce saturating or semi-saturating responses from all four cone subtypes (Fig. 6d). Response amplitude to a 575 nm flash, to which blue- and UV-sensitive cones have virtually no sensitivity (Fig. 2), was still 80% of the largest 405 nm flash response, showing that red- and green-sensitive cones are the two dominant sources of the cone component. This observation is consistent with the population ratio of zebrafish cones: red- and green-sensitive cones are roughly twice as numerous as blue- and UV-sensitive cones44. Together, these results demonstrate that the rod and cone responses can be isolated from whole retina ex vivo ERG recordings.
We also established a protocol for isolating the cone component of the ex vivo ERG a-wave using a pair of 500 nm flashes (Fig. 7). A bright initial first flash was given to saturate rods for several seconds, and a second flash was given 2 s later to obtain the cone response (Fig. 7b). A response family was isolated from the paired flash responses (Fig. 7c), and the cone I1/2 was estimated to be 661 ± 75 photons µm−2 (n = 6, mean ± SEM) by fitting the intensity-response relationship (Fig. 7d). This value was 1.6-fold higher than the cone I1/2 estimated from single flash responses (Fig. 6b, 425 ± 76 photons µm−2). We tried to isolate the rod component by subtracting the cone component from the overall rod-cone mixed responses (Fig. 7a). However, a small cone response-like bumps were left on the rod saturating plateau (Supplementary Fig. S1). This suggests that the cone responses elicited by the second flash were a little smaller than the cone responses elicited by an identical single flash. Together with the slightly higher corresponding I1/2 value, this result indicates that cones were desensitized to a detectable degree by the conditioning first flash.
() Representative flash response family from zebrafish rods and cones obtained by ex vivo ERG recordings. The ERG a-wave was inverted to positive direction for consistency with the single-cell recordings. Flash wavelength and intensity: 500 nm, 1–32,000 photons μm. Red trace was obtained at 100 photons μm. () Intensity-response relationships of the overall response (●) and its estimated rod (■) and cone (▲) components. Rod and cone plots were fitted with Naka-Rushton function with= 25 µV,= 13 photons µm, and= 1.9 for rod data, and with= 72 µV,= 425 photons µm, and= 1.0 for cones. Plots are mean ± SEM (n = 7). () Responses obtained with 405 nm (violet), 500 nm (green), and 575 nm (orange) flash at 32,000 photons μm. () Estimated intensity-response relationships of the four cone subtypes when stimulated with 405 nm flashes, based on their spectral sensitivity (Fig.) and intensity-response relationships obtained with near λflashes (Fig.). A grey line indicates 32,000 photons µm. a b c d −2 −2 −2 −2 −2 −2 R I n R I n max 1/2 max 1/2 max 2 4
Isolation of cone responses by ex vivo ERG recordings. () Control response family. Flash was given one time at 2.2 s (arrow). Flash wavelength: 500 nm. Intensity: 32–32,000 photons μm. Red trace: 320 photons μm. () Responses to paired flashes. Flashes were given two times at 0.2 and 2.2 s (arrows). First flash: 500 nm, 32,000 photons μm. Second flash: same as in (). () Isolated cone responses obtained by clipping 2–4 s in () with subsequent baseline adjustment. () Intensity-response relationship of cone responses. Plots were fitted with Naka-Rushton function with= 72 µV,= 661 photons µm, and= 0.89. Plots are mean ± SEM (n = 6). The estimated cone response fitting curve in Fig.b is redrawn in blue for comparison to demonstrate the slight suppression of the cone responses by the conditioning first flash. a b a c b d −2 −2 −2 −2 R I n max 1/2 6
Discussion
In this study, we have established single-cell suction recordings from the five photoreceptor subtypes in the zebrafish retina (Figs. 1 and 2, Table 1) and collected basic response waveforms and parameters from them (Figs. 3, 4, 5, Tables 2 and 3). We also demonstrated a method to collect and isolate the rod and cone derived responses using the ex vivo ERG adaptor (Figs. 6 and 7). We hope that these results will facilitate and boost the use of zebrafish in vision research45.
Sensitivities varied largely among the four cone subtypes (Figs. 2 and 4, Table 2), with green-sensitive cones showing 5.5-fold higher sensitivity than the fellow red-sensitive cones. A very similar 5.9-fold difference has also been reported for goldfish red- and green-sensitive cones46. In contrast, the three cone subtypes in primate Macaca fascicularis show comparable sensitivity47. What is the molecular mechanism of the large sensitivity difference in fish cones? Based on the previous molecular analyses in zebrafish, phototransduction components of green and red cones are almost identical except for their opsin genes: opn1lw1 and opn1lw2 in red cones and opn1mw1, opn1mw2, opn1mw3, or opn1mw4 in green cones3. Therefore, the lower red cone sensitivity could be ascribed to its visual pigments. Microspectrophotometry analysis of goldfish cones has shown the presence of ~ 30% apo-opsin in red-sensitive L-cones, in contrast to only ~ 3% in green-sensitive M-cones and negligible fraction in blue-sensitive S-cones46. Apo-opsin is formed in cones even in darkness by spontaneous chromophore dissociation from opsin protein48. In salamander red cones, exogenously added 11-cis retinal, which binds to apo-opsin to form stable visual pigment, effectively increases sensitivity by two-fold48. However, this was not observed prominently in goldfish red-sensitive cones46. We will examine the effect of exogenous 11-cis retinal on zebrafish cones in the future. Another possible source of desensitization from pigments is their spontaneous light-independent activity (dark noise). In salamander cones, the frequency of spontaneous pigment activation rate was estimated to be 600 s−1 in L-cones and < 2 s−1 in S-cones. Corresponding to this large dark noise in L-cones, the estimated single photon response of an L-cone is 0.04 pA, which is 5.8-fold smaller than that from an S-cone (0.23 pA)49. In contrast, such frequent thermal activation is not observed in the human red cone pigment (~ 8.8 s−1) when expressed in mouse rods ectopically50. A systematic study of noise properties in zebrafish cone photoreceptors will be needed to clarify the molecular mechanism of sensitivity differences among cone subtypes.
Perhaps surprisingly, we found that the amplitude of the rod response was smaller than that of the cones (Fig. 4a, Table 2Rmax), even though rods have the largest outer segments among the five photoreceptor subtypes (Fig. 1). The rod amplitude, 5.7 pA, is consistent with previously published zebrafish rod data (5.5 pA)33. Reported rod amplitudes measured by the suction method are varied among fish species: 12.4 pA in skate51, 9.2 pA in goldfish46, 36.7 pA in striped bass52, and 3.5 pA53 or 4.9 pA54 in carp. Zebrafish and carp belong to the same Cyprinidae family, hence small rod response might be a common feature in this family. Consistently, the rod component in the ex vivo ERG a-wave was 25 μV which is only about 1/3 of the corresponding cone component (72 μV, Fig. 6b). The adult zebrafish retina has 40%-rod 60%-cone ratio23, therefore the estimated ratio of individual rod and cone amplitudes is 25/40:72/60 ≈ 0.521:1. The major source of ex vivo ERG a-wave was found to be red- and green-sensitive cones (Fig. 6c), and average membrane current Rmax of those cells (13 and 7.5 pA, Table 2) is 10.3 pA. Thus, the estimated rod response amplitude is 10.3 × 0.521 = 5.36 pA, which is close to the Rmax determined by single-cell suction recordings, 5.7 pA (Table 2). This suggests that the small rod response in single-cell suction recordings does not originate from the loose fitting of the recording pipette specifically to a long and straight rod outer segment. What is the molecular mechanism of the smaller responses in zebrafish rods? One possibility is the circadian regulation of the response amplitude. Rods mediate nighttime vision due to their 40–220-fold higher sensitivity compared to cones (Fig. 4b, Table 2, I1/2). However, all results presented in this study were obtained only in daytime when rod function is supposed to be suppressed. Fish rods and cones show dynamic day-night regulation in multiple biological processes including rod-cone gap junction55, retinomotor movement56, and disappearance of ERG b- and d-waves after the disorganization of cone ribbon synapses in midnight5. These regulations work to increase photosensitivity of the retina in nighttime, probability of photon capture in rod outer segments, and suppress cone function, respectively. Accordingly, the rod response might be boosted specifically in nighttime to enhance night vision. To address this question directly, we are currently collecting day-night comparison data from rods and cones. Another possible reason for the relatively small rod responses is that the perfusion solution used in this study might not be particularly suitable for rods because of the different metabolism in rods and cones57–59. This point could be clarified by in vivo ERG recordings of pharmacologically isolated photoreceptor a-waves from adult zebrafish.
Dim flash response kinetics were faster in red- and blue-sensitive cones than in UV- and green-sensitive cones. The response shutoff process in red cones is thought to be accelerated by the pigment-derived noise discussed above. This noise would work as equivalent background light in darkness to drive the calcium-feedback mechanisms, which would upregulate the response shutoff through mechanisms mediated by recoverin/S-modulin and guanylate cyclase activating proteins (GCAPs)60–62. Another possible factor modulating the response kinetics could be the different expression patterns of phototransduction proteins among the four cone subtypes. At mRNA level, grk7b is enriched specifically in UV-sensitive cones among the four photoreceptor Grk subtypes, and arr3b is enriched in UV- and blue-sensitive cones among the four photoreceptor arrestin subtypes2. However, these expression patterns do not provide a simple explanation for the quicker response in blue-sensitive cones over UV- and green-sensitive cones. To resolve this issue, more quantitative insights will be required, particularly at the protein level. For example, Zang and colleagues have shown corresponding but different circadian regulations of mRNA and protein for Grk7a and Arr3a in adult zebrafish eyeballs6.
The ex vivo ERG method provided pharmacologically isolated a-waves from rods and cones. This approach is advantageous to the single-cell suction method in several ways. One is the higher signal-to-noise ratio of transretinal recordings. Typical signal-to-noise ratio was 30–40 in the raw trace of our ex vivo ERG recordings. In contrast, it was only 5–10 in the single-cell suction recordings, requiring averaging 5–10 responses to improve the signal-to-noise ratio. This point is particularly important if analyzing small responses from degenerating retinas from disease model animals. For example, we previously obtained transretinal voltage signals from severely degenerated Rpe65-/-Grk1-/- mouse retinas from which suction recordings were not possible63. Secondly, for transretinal recordings, the retina is kept mostly intact. This allows us to study not only a-wave responses from photoreceptor cells but also b-wave responses, produced mainly by ON bipolar cells, as well as the slow PIII component from Müller glial cells, by comparing photoresponses in the presence of different set of inhibitors41,64. ERG recordings from the intact retina also enable the study of the interaction between cones and Müller cells in the retinoid metabolism65. Thirdly, our ex vivo ERG adaptor could be set up in a commercial ERG system as well64, and recently, a detailed method for recording from the zebrafish retina with the Diagnosys Espion ERG system was published by another group66.
Compared to the rod-dominant mouse retina (97% rods21), zebrafish has cone-dominant retina at larval stage (92% cones22) and ~ 60%-cone retina at 3–6 months adult stage23. Many human eye diseases including age-related macular degeneration and Leber congenital amaurosis type 2 lead to cone degeneration. In addition, in retinitis pigmentosa, preservation of the remaining cone function after rod death is of great interest in ophthalmology research. Understanding the physiology of cones in health and disease will advance our understanding of these diseases. Studies of cone-enriched zebrafish retinas in combination with genetic manipulations, single-cell genomics, proteomics, functional imaging, and electrophysiology hold exciting potential for addressing many fundamental questions in visual sciences67,68.
Methods
Chemicals
DL-2-amino-4-phosphonobutyric acid (DL-AP4) was obtained from Tocris Bioscience (Bristol, UK). All other chemical reagents were obtained from Sigma-Aldrich (Saint Louis, MO).
Animals
All experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by the Institutional Animal Care and Use Committee of UC Irvine. Authors complied with the ARRIVE guidelines. Fish were purchased from the Zebrafish International Resource Center at University of Oregon and kept at room temperature (22 °C) under scheduled 14 h light-10 h dark illumination cycle. Previous ERG study reported a red-shift in the rod spectral sensitivity when zebrafish were kept in cold water (22–25 °C), which was explained by the formation of porphyropsin69. However, we did not see obvious shifts in our spectral sensitivity data (Fig. 2, Table 1). Fish were fed commercial fish foods (Micro Pellets, Hikari, Hyogo, Japan) two or three times a day. Prior to the experiments, each zebrafish was transferred into a 250 mL plastic beaker with approximately 150 mL water and kept in darkness overnight for complete dark adaptation of photoreceptor cells. Fish were euthanized by rapid cooling in ice-cold water followed by decapitation.
Single cell suction recordings
Eyes were enucleated, and the cornea and lens were removed using a pair of micro scissors and fine forceps under a stereomicroscope with an infrared illuminator and infrared scopes. The retinas were peeled off in Ringer’s solution (104 mM NaCl, 2 mM KCl, 1.6 mM MgCl2, 1.5 mM CaCl2, 30 mM NaHCO3, and 10 mM glucose, pH 7.3 by bubbling with 95%O2/5%CO2) and chopped into small pieces randomly with a razor blade. For rod recordings, retinal fragments were directly transferred into the perfusion chamber. For cone recordings, cones were dissociated from the retina by pipetting the retinal fragments in a plastic tube with a wide-bore glass pipette, typically for 15 times, and then immediately transferred into the recording chamber. Perfusion was temporarily halted for 10 min until cells were settled down on the bottom of the chamber. During a recording session, cells were kept under constant perfusion of Ringer’s solution (1.2 mL/min, 22 °C) and imaged under infrared illumination by bright field microscopy without using contrast enhancement techniques (e.g. differential interference contrast, phase contrast). A single rod or cone was drawn into the suction pipette for flash response recordings. Whenever possible, only the outer segment part of the cell was drawn. However, the outer segments of red- and green-sensitive cones were found to be too close to draw only one of them. In that case, the ellipsoid part of the inner segment was drawn to obtain inverted current responses, which were re-inverted digitally in the subsequent data analysis. Both recording and reference electrodes were filled with electrode solution in which 30 mM NaHCO3 in Ringer’s solution was substituted with NaCl and HEPES (129 mM NaCl, 2 mM KCl, 1.6 mM MgCl2, 1.5 mM CaCl2, 10 mM HEPES, and 10 mM glucose, pH 7.3 by NaOH). The tip of the recording electrode was heat polished to achieve the desired inner diameter to obtain tight fitting to the target cell: 3 µm for rods and UV-sensitive cones and 4 µm for red-, green-, and blue-sensitive cones. The cell was exposed to flash stimuli delivered from a custom-made light-emitting diode (LED) system70. To obtain sufficient intensity for violet light, a 405 nm LED (M405L4, Thorlabs, Newton, NJ) was added to the system, and beams from two LEDs were merged with a dichroic mirror (DMLP425R, Thorlabs). Spectral spreads of bandpass filters and the 405 nm LED are shown in Table. S1. Light intensity was calibrated with an optometer (350 Linear/Log optometer, UDT Instruments, San Diego, CA). Membrane current signals were amplified with an amplifier (Axopatch 200B, Molecular devices, San Jose, CA), low pass filtered (8-pole Bessel at 30 Hz; Model 3382, KrohnHite, Brockton, MA) and imported to a computer through a digitizer (Digidata 1440A, Molecular Devices). About 20% delay in Tp values is expected by the 30 Hz low-pass filter. Data analysis was done with Clampfit 10 (Molecular Devices). Other details are written and recorded in our previous video71.
When needed, the second eyecup from the fish was transferred to a 35 mm diameter culture dish with 2 mL of Ringer’s solution and stored at room temperature (22 °C) in a dark box supplied with humidified 95%O2/5%CO2 gas until used for recordings.
Intensity-response relationships were fitted with Naka-Rushton hyperbolic function:
where Rmax is maximum response amplitude (pA for single-cell suction recording data, µV for ex vivo ERG data), I is the flash intensity (photons µm−2), n is the Hill coefficient, and I1/2 is the half-saturating flash intensity (photons µm−2).
The λmax values for rods and cones (Table 1) were estimated by fitting the sensitivity plots (Fig. 2) with the Govardovskii nomogram for the α-band of visual pigment absorption72. Data points used for fitting were 405–500 nm for UV-sensitive cones, 405–550 nm for blue-sensitive cones, 450–650 nm for green-sensitive cones, 525 nm to 675 nm for red-sensitive cones, and 450–675 nm for rods.
The estimated λmax (Table 1) of blue- (401 ± 2 nm) and UV-sensitive cones (374 ± 1 nm) did not agree well with the reported values (blue: 411 nm17 or 425 nm40 and UV: 361 nm17 or 342 nm40). A possible cause of these gaps would be the lack of data points in the UV region (Fig. 2a). The data could be improved by adding UV LEDs in our current equipment or by using a mercury or xenon arc lamp69 as a light source.
Ex vivo ERG
Eyes were collected as described above. The retinas were peeled off in Ringer’s solution supplemented with 20 µM DL-AP4 and 2 mM sodium aspartate73 to block the postsynaptic components of the photoresponse, and with 100 µM BaCl2 to suppress the slow glial PIII component74. Because aspartate deactivates glutamate input to both ionotropic and metabotropic receptors, DL-AP4 which blocks only the metabotropic receptor32 is probably not required. However, as DL-AP4 is usually used in our mouse studies, it was added in our early attempts of zebrafish recordings and was kept in the formula throughout this study for consistency. The same supplemented solution was used for perfusion. The retina was placed in an ex vivo ERG adaptor (Ocuscience, Henderson, NV), perfused with Ringer’s solution (1.2 mL/min, 22 °C), and exposed to calibrated flash stimuli (see above). Trans-retinal voltage was amplified with a differential amplifier (DP-311A, Warner Instruments, Hamden, CT), low pass filtered (8pole Bessel at 30 Hz) and imported to a computer through a digitizer. Data analysis was done with Clampfit 10. Other details are written and recorded in our previous paper and video articles41,64.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
This work was supported by UCI Gavin Herbert Eye Institute 20/20 Society Pilot Research Grant (S.S.) and by NIH grant EY033810 (V.J.K.). The authors acknowledge support to the Gavin Herbert Eye Institute at the University of California, Irvine from an unrestricted grant from Research to Prevent Blindness and from NIH core grant P30 EY034070. The authors also thank Dr. Yohey Ogawa (Washington University in St. Louis) for discussion and advice in the interpretation of our data.
Author contributions
S. Sato, and V.J. Kefalov contributed to the conception and design of this research. S. Sato performed experiments and analyzed data. S. Sato and V.J. Kefalov wrote the manuscript. All authors approved the final version of the manuscript.
Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Contributor Information
Shinya Sato, Email: shinyas1@hs.uci.edu.
Vladimir J. Kefalov, Email: vkefalov@hs.uci.edu
References
Associated Data
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.