We generated transgenic zebrafish Tg (Gnat2:gal4-VP16/UAS:nfsB-mCherry) that express E. coli nitroreductase and mCherry in cone photoreceptor cells. In developing embryos, the first transgene expression (reported by mCherry) was detected in the pineal gland at approximately 22–24 hours post-fertilization (hpf) (Fig. 1A). By 48 hpf, mCherry was detected in the retina, at which time retinal cone photoreceptor cells were developed (Fig. 1B). The intensity of transgene expression in the pineal gland and retina increased during embryonic stages (between 56 and 120 hpf; Fig. 1C–F). We also observed non-specific auto-fluorescence in the yolk and cardiovascular areas (Fig. S1).

We examined the cell types that express the transgene in the pineal gland and retina in young adult animals (2 months old) by co-localization of mCherry with various opsin antibodies. In the pineal gland, we observed strong immunoreactivity of zpr-1 antibody (which labels double-cone photoreceptor cells) (Fig. 2A). The expression of mCherry co-localized with zpr-1, suggesting that the transgene is expressed in double-cone-like photoreceptor cells. We did not detect blue-cone or UV-cone opsin expression in the pineal gland (Fig. 2B, C). Weak rhodopsin expression was detected, but it did not co-localize with mCherry (Fig. 2D). In the retina, mCherry was detected in all the cone cell types, and it co-localized with different cone opsin antibodies (zpr-1, blue- and UV-cone opsin antibodies, respectively) (Fig. 2A–C). We did not observe co-localization of mCherry and rhodopsin antibodies (Fig. 2D).

We examined transgene expression in zebrafish at different ages. In developing larvae and young adults (1 month old), the transgene was expressed in cone photoreceptor cells in the pineal gland and retina. In the pineal gland and retina, co-localization of mCherry and zpr-1 antibodies was observed (Fig. 3A).

In zebrafish older than 3 months, the expression of the transgene in the retina was reduced in comparison to the expression in young adult animals. Often, the expression of the transgene was restricted to the central retina, near the optic nerve. Co-localization of mCherry and zpr-1 immunoreactivity was detected in the pineal gland and central retina (Fig. 3B). By 8 months, the expression of the transgene was completely diminished in the retina, i.e., no mCherry expression was detected, whereas its expression in the pineal gland remained unchanged (Fig. 3C). During this time period (between 3 and 8 months of age), we did not observe cell death in the retina, indicating that the loss of mCherry was due to diminished transgene expression, rather than cone photoreceptor cell apoptosis.

In adult zebrafish older than 8 months, the transgene is only expressed in the pineal gland. This provides a tool for selective ablation of pineal photoreceptor cells without affecting retinal photoreceptor cells. We ablated pineal photoreceptor cells by treating the transgenic fish (>8 months) with metronidazole. After 4 metronidazole treatments, the expression of the mCherry in the pineal gland was diminished (Fig. 4A–C). We observed cell death (by TUNEL labeling) in the pineal gland. We did not obverse cell death in the pineal gland or retina in untreated transgenic fish or wild-type animals (Fig. 5). Cell loss was seen in the pineal gland in metronidazole-treated transgenic animals (by DAPI labeling, Fig. S2).

Depletion of pineal photoreceptor cells did not affect zebrafish visual sensitivities. When measured at 6 pm (in the normal LD, when the fish are most sensitive to light stimuli) ^[11], the absolute behavioral visual thresholds of wild-type fish (AB strain), metronidazole-treated AB fish, untreated- and metronidazole-treated transgenic fish were –6.2±0.3, –6.0±0.3, –6.1±0.3, and −6.2±0.2 log units, respectively (Fig. 6A). This suggests that the absolute rod sensitivity was not affected after pineal photoreceptor cell ablation. The behavioral cone sensitivities measured using narrow-band color filters were also similar in wild-type AB fish, metronidazole-treated AB fish, untreated and drug-treated transgenic zebrafish. When examined using a 625-nm wavelength cut-off filter, for example, the visual thresholds of wild-type AB fish, metronidazole-treated AB fish, untreated- and metronidazole-treated transgenic fish were −3.2±0.2, −3.1±0.2, −3.0±0.3, and −3.1±0.2 log units, respectively. When examined using a 500-nm wavelength cut-off filter, the behavioral visual thresholds of wild-type AB fish, metronidazole-treated AB fish, untreated- and metronidazole-treated transgenic fish were −5.7±0.3, −5.7±0.3, −5.4±0.2, and −5.6±0.3 log units, respectively (Fig. 6A).

In wild-type zebrafish, the behavioral visual sensitivity fluctuates between the day and night, i.e., the fish are most sensitive to light in the afternoon and least sensitive at night and in the early morning. From 6pm to 3am, the behavioral visual threshold increases approximately 2.0 log units ^[11]. This pattern of visual threshold fluctuation persists in animals kept in DD and LL conditions, suggesting the involvement of endogenous circadian clocks (see also Fig. S3). We measured visual sensitivity of Tg (Gnat2:gal4-VP16/UAS:nfsB-mCherry) fish at different times in the day and night while the fish were kept in DD. Similar to wild-type zebrafish, in the transgenic fish the visual sensitivity fluctuated between the subjective day and night (Fig. 7A). In DD, the fish were least sensitive to light at subjective night and early morning. During the day, visual sensitivity levels gradually increased. By 6pm, visual sensitivity increased to levels approximately 1.5–2.0 log units above the sensitivity levels measured in the early morning hours (3am), similar as seen in wild-type fish. In the evening, the visual sensitivity levels began to decrease. It continually decreased at night, and reached the least sensitive level at 3am (Fig. 7A). Similar patterns of visual threshold fluctuations were observed when measured using the narrow-band 625 nm and 500 nm wavelength filters. In both cases, the transgenic fish were most sensitive in the subjective late afternoon (7pm) and least sensitive in the early morning (3am and 7am) (Fig. 7B, C). Between these two time points, the difference in behavioral visual threshold was 1.7±0.4, 2.4±0.3, 2.4±0.3 log units, respectively, when tested under white, red and green light, similar as seen in wild-type animals (Fig. 6B).

In the absence of pineal photoreceptor cells (after metronidazole treatments), the circadian rhythms of behavioral visual sensitivity were dampened out. When tested using white light (n = 12), for example, the behavioral visual thresholds increased at most time points, except at 3am and 7am when the visual thresholds were already at the highest levels (Fig. 8A). Between the day and night, the difference in visual threshold in metronidazole-treated fish was reduced to approximately 0.9±0.3 log unit (Fig. 6B). Metronidazole-treatments produced no effects in the circadian rhythms of behavioral visual sensitivity in untreated transgenic or wild-type animals (Fig. 6B; S3).

The circadian rhythms of retinal cone sensitivity (measured using 625 nm and 500 nm wavelength narrow-band filters, respectively) were also dampened out after metronidazole treatments (Fig. 8B, C). This was largely due to visual threshold decreases (except in the late afternoon, when the fish were already most sensitive to light). When examined using red light, for example, at 3am all the tested fish (n = 12) showed decreased visual threshold levels when compared to the visual thresholds measured in untreated transgenic fish (Fig. 8B). When examined using green light, at 11pm and 3am all the tested fish (n = 12) showed decreased visual threshold levels (Fig. 8C). Between the subjective day and night, in metronidazole-treated fish the difference in behavioral visual thresholds were decreased to 1.2±0.3 and 1.3±0.2 log units, respectively, when measured using 625 nm and 500 nm wavelength filters (Fig. 6B). Metronidazole-treatments produced no effects in the circadian rhythms in cone sensitivity in untreated transgenic and wild-type animals (Fig. 6B; S3).

We investigated whether light exposures could restore the circadian rhythms of visual sensitivity in metronidazole-treated transgenic zebrafish. In these experiments, the fish received 25-min light exposures at 7am (room fluorescent light). Thereafter, the fish were kept in DD for 24 hours. At 7am on the following day, the fish were tested for behavioral rod and cone sensitivity using white light, red (625 nm) and green (500 nm) light, respectively. The measurements were repeated in 4-hour intervals, and completed at 7am on the following day.

In 8 of 12 fish tested (with white light), the circadian rhythms of behavioral visual sensitivity were restored. During a 24-hour period of subjective day and night, the fish were most sensitive to light in the subjective afternoon, and least sensitive at subjective night and early morning, similar to the patterns of visual threshold fluctuation in wild-type and untreated transgenic control animals (Fig. 7A). In the remaining fish (n = 4), two fish showed no threshold fluctuations between the day and night, and two fish showed abnormal fluctuation patterns (i.e., they were most sensitive to light at night or in the early morning). After the short light exposures, the metronidazole-treated transgenic zebrafish also showed restored circadian rhythms of cone sensitivity when tested with red and green light (9 out of 12 fish in red light, and 8 out of 12 fish in green light). They were sensitive to light stimuli in the afternoon and not sensitive at night and in the early morning (Fig. 7B, C). The remaining fish showed no visual threshold fluctuation or shifted patterns of fluctuation.

We evaluated the effects of pineal photoreceptor cell ablation on circadian rhythms of other activities in the transgenic fish, such as locomotion. Locomotor behaviors were assessed by monitoring the swimming activity of the fish. The fish were constrained in a 2-liter container (L×W×H: 30×10×5.5 cm) filled with 0.5 liter system water (see also ^[26]). Five vertical lines were drawn on the wall of the container at an equal distance. In a 3-min period, we counted the number of lines the zebrafish crossed. The experiments were conducted at different times in the day and night while the fish were kept in DD. In control groups (untreated transgenic fish), the fish was most active in the day and least active during the night (Fig. 9). At 9am, for example, swimming activities of the fish were approximately 3-fold higher than the swimming activities measured at midnight. In metronidazole-treated transgenic fish, however, swimming activities were decreased during the day and night. We did not observe fluctuations in locomotor behaviors between the day and night (Fig. 9).

Zebrafish (Danio rerio, wild-type AB and transgenic fish were maintained as previously described ^[34]. Normally, the fish were kept in cyclic light-dark (LD; room fluorescent light, from 07∶00 to 21∶00). For the DD experiments, the fish were removed from the LD at 21∶00 the day before the experiment and subsequently kept in the DD. All the experiments were performed according to the NIH guidelines for animals in research.

The previously identified 3.2 kb zebrafish guanine nucleotide binding protein (G protein), alpha transducing activity polypeptide 2 (gnat2) promoter fragment was digested from a plasmid construct ^[17] with XhoI and BamHI. The restriction fragment was ligated into the XhoI and BamHI sites of a modified pT2KXIG vector ^[35] that contained a custom multiple cloning site and an SV40 late poly(A) sequence ^[36]. A 654 bp Gal4-VP16 fusion ^[37] was then ligated into the BamHI and NotI restriction sites between the promoter and poly(A) sequence. The resulting expression construct was purified with the QIAGEN Plasmid Maxi Kit (Qiagen, Valencia, CA), followed by phenol/chloroform (1∶1) extraction. Tol2 transposase mRNA was in vitro transcribed from the pCSTZ2.8 plasmid ^[35] using the SP6 mMessage mMachine kit (Ambion, Foster City, CA).

The pT2KXIG-gnat2:Gal4-VP16 expression construct and Tol2 transposase mRNA were resuspended in nuclease-free water at a combined concentration of 25 ng/µl each, and co-injected into 1–4 cell stage wild-type AB strain embryos. Injected fish were out-crossed to the Tg(UAS-E1b:nfsB-mCherry)jh17 line ^[37] and the progeny were screened for fluorescence to identify founders and to establish the Tg(gnat2:Gal4-VP16) nt21 line. Out-crosses of double-transgenic to wild-type fish produced fluorescence in 25% of the progeny, which is consistent with single insertions of each transgene. For the described research, the transgenic fish (embryos or adults) were obtained by selecting individuals that expressed mCherry at 3 dpf from crosses between the Tg(Gnat2:gal4-VP16/UAS:nfsB-mCherry) transgenic fish and wild-type AB zebrafish.

Metronidazole (Sigma-Aldrich, St. Louis, MO) was dissolved in regular tank water (Instant Ocean Salts in distilled water, 3 g/gal, pH = 7.2). The fish were treated at 28° in 1.0 liter tank water containing 10 mM metronidazole. After 24 hours of treatment, the fish were removed from the treatment tank and transferred into the recovery tank, which contained regular tank water. The fish were treated 4 times (in 2-day intervals) before the behavioral experiments were performed. This scheme of metronidazole treatment did not cause lethality in adult zebrafish. Complete ablation of the pineal photoreceptor cells after the metronidazole treatments were confirmed by examining the pineal gland (in whole-mount embryos or cryostat sections of adult brain tissues) with fluorescent microscopy (i.e., lack of mCherry) and TUNEL labeling (i.e., evidence of cell death).

Procedures for immunolabeling were similar as previously described ^[38]. Embryos, adult eyes or brain tissues were fixed in 4% paraformaldehyde and embedded in OCT. Cryostat sections (adult retinas or pineal glands) were collected at 12 µm in thickness. The following primary antibodies were used: mouse monoclonal anti-double cone opsin (zpr-1, 1∶200); [University of Oregon], rabbit polyclonal anti-blue opsin (1∶250); rabbit polyclonal anti-UV opsin (1∶1000); rabbit polyclonal anti-rhodopsin (1∶5000) ^[38]. Secondary antibodies (Alexa Fluor 488, goat anti-mouse or rabbit) were used at a dilution of 1∶500. TUNEL labeling was performed using the ApoAlert DNA Fragmentation Assay Kit (Clontech, Mountain View, CA).

Methods for measuring zebrafish visual sensitivity were similar as previously described ^[39]. The test apparatus consisted of a circular transparent plastic container surrounded by a drum rotating at 10 rpm in either clockwise or counterclockwise directions. The inside of the drum was covered by white paper marked with a black segment. The drum was illuminated from above with white light, maximum intensity, log I = 425 µW/cm2. The intensity of the light was adjusted by changing neutral density filters in 0.5 log unit steps.

Normally, the fish swim slowly along the wall of the container. However, when encountered by the black segment rotating outside the container, the fish display robust escape responses, i.e., as soon as the black segment comes into view, the fish immediately turns and rapidly swims away. Usually, a judgment of whether the fish saw the black segment can be made in less than 10 seconds. To measure the behavioral visual sensitivity, we recorded the minimum light intensities that evoked behavioral escape responses. At least 5 escape responses in 8 encounters were required to record a threshold response. The measurement was performed under different light wavelengths: white light (rod sensitivity), red light (white light that passes a 625-nm wavelength narrow band cut-off filter; red cone sensitivity), green light (white light that passes a 500-nm wavelength narrow band cut-off filter; green cone sensitivity).