The Ultraviolet Opsin Is the First Opsin Expressed during Retinal Development of Salmonid Fishes

Abstract

purpose. To determine the spatial and temporal progression of opsin appearance during retinal development of salmonid fishes (genus Oncorhynchus and Salmo).

methods. Reverse transcription–polymerase chain reaction (RT-PCR) and in situ hybridization with riboprobes against the five classes of opsins present in salmonids (UV, blue, green, red, and rhodopsin) were used to establish the sequence of opsin appearance and the localization of opsins to specific morphologic photoreceptor types.

results. Both detection methods revealed that UV opsin mRNA was expressed first and was followed closely by red opsin mRNA. In situ hybridization results indicated the following opsin sequence: UV, red, rhodopsin, green, and blue. The UV opsin riboprobe labeled single cones, whereas the red and green riboprobes labeled opposite members of double cones. The blue riboprobe started labeling single center cones ∼1 month after initial UV riboprobe labeling, confirming a switch in opsin expression of these cones from UV to blue. All probes first labeled a small patch of cells in the centrotemporal retina, and expression then expanded primarily toward the temporal and dorsal retina, with the exception of the blue opsin which expanded ventrally at first.

conclusions. The sequence of cone opsin appearance in salmonid fishes is similar to that in mammals, in which a violet-blue (SWS1) opsin is expressed first followed by a red (M/LWS) opsin. This sequence is different from that in zebrafish, goldfish, and chick, in which red and green opsins are expressed first. As in mammals, rhodopsin expression in salmonid fishes arises after the first cone opsin. The findings show similarity in the sequence of opsin expression between a group of lower vertebrates, the salmonid fishes, and mammals.

During development of the vertebrate retina, progenitor cells in the outer epithelium differentiate to form two types of photoreceptors: rods and cones. Rods are tubular in shape, are orders of magnitude more sensitive to light than cones, and have a predominant visual pigment (rhodopsin) with maximum sensitivity in the green part of the spectrum (λ_max ∼500 nm). ^{1 2} In contrast, cones show variable morphology (two major types exist: single and double cones, the latter type absent in mammals), ^{3 4 5} and their predominant visual pigment can be most sensitive to ultraviolet (UV, λ_max ∼365 nm), violet (λ_max ∼400 nm), blue (λ_max ∼430 nm), green (λ_max ∼520 nm), or red (λ_max ∼565 nm) light. ^{1 6 7} These categories reflect the class of visual pigment protein (opsin) ⁸ expressed by the cone. In lower vertebrates and, to a lesser degree, in primates, cone photoreceptors form repeating geometric formations termed mosaics. ^{2 3 4} The spatial and chromatic organization of these mosaics is crucial for all aspects of visual function, yet the developmental mechanisms that give rise to them remain unknown. 

In lower vertebrates, cone mosaics may arise from a cascade of cell interactions whereby a given spectral cone type would influence the fate of neighboring uncommitted cells to differentiate into the next spectral type. ⁹ Such sequential development of photoreceptors occurs in the fruit fly, Drosophila melanogaster, where a set of “founder cells” initiates a position-dependent differentiation cascade that results in the mosaic of rhabdoms characteristic of the compound eye of this animal. ¹⁰ For a similar mechanism to operate in lower vertebrates, opsin expression would have to follow a precise developmental sequence, and each opsin class would have to be expressed in a particular cone type occupying a precise position in the unit mosaic. This is the case in cyprinid fishes (goldfish and zebrafish) and in chick, where the order of opsin appearance is rhodopsin, followed by red, green, UV-violet, and blue opsin. ^{11 12 13 14 15 16} In contrast, rodents and other mammals do not form discernible mosaics (a notable exception being the blue cone mosaic in primates), ⁴ and the order of opsin appearance is different from that just stated. In mammals, violet-blue opsin appears first and is then followed by rhodopsin and green and/or red opsin. ^{4 17 18 19 20} A recent study using goldfish has demonstrated that rhodopsin does not play a role in differentiation of cones, as judged by opsin expression. ²¹ A role for the first cone opsin in establishing the chromatic organization of the mosaic through a hypothetical opsin cascade ⁹ has not been ruled out. 

Our studies in salmonid fishes ²² have shown that the cones in these animals are organized in a square mosaic similar to that in goldfish ¹² and in most other fishes. ³ In contrast to goldfish, zebrafish, and chick, the single cones of Pacific salmonid fishes undergo a switch in opsin from UV to blue. ^{22 23 24} This switch to a longer wavelength–absorbing opsin is analogous to that postulated for mammals (some rodents and human) in which green and red cones are believed to arise from blue cones. ^{5 18 25} The common property of dual opsin cones shared by salmonid fishes and some mammals ^{5 7 18 26 27 28} raises the possibility that the order of opsin expression during development may be the same between these groups. If this is the case, the UV opsin of fish and violet-blue opsin of mammals (both in the SWS1 opsin gene family) ⁸ may play important roles in the chromatic organization of the retina in these animals. 

Methods

Animals

Wild stock chinook salmon (Oncorhynchus tshawytscha) were obtained from the Capilano River hatchery (Vancouver, BC, Canada) at various stages of embryonic development starting at the eyed egg. Two sets of 10 to 20 embryos were collected at each sampling date for opsin transcript detection by reverse transcription–polymerase chain reaction (RT-PCR) and in situ hybridization. 

Salmonid development, like that of other fishes, is primarily a function of water temperature. The developmental stage is therefore routinely characterized by the accumulated temperature units (ATUs), which is the sum of the rearing water temperature, sampled daily, since fertilization. Results of this research are therefore expressed in relation to ATUs. The chinook salmon used were reared at 12°C. At this temperature, the eye spot on the egg appeared at ∼290 ATUs, hatching started at 432 ATUs (all 2 million eggs at the hatchery hatched within a period of 3 days: 432–456 ATUs), and the embryos had completely absorbed their yolk sac (corresponding, in nature, to the swim-up stage from the gravel as an alevin) by ∼810 ATUs. 

Wild stock coho (O. kisutch) and chum (O. keta) salmon were obtained from the Capilano and the Big Qualicum River (Qualicum Beach, BC, Canada) hatcheries, respectively. These salmon were raised in river water that varied in temperature between 3.5 and 8.5°C. Wild stock rainbow trout (O. mykiss) were obtained from the Fraser Valley Trout hatchery (Abbotsford, BC, Canada), and aquaculture stock Atlantic salmon (Salmo salar) were supplied by a hatchery owned by Omega Salmon Group Ltd (Campbell River, BC, Canada). The rearing temperature at these two facilities was 7°C. With the exception of coho salmon, the sampling regimen for the other species was less rigorous than that for chinook salmon. Results from all species were compared, to establish general trends in opsin development in this family of fishes. 

Care and use of experimental animals was approved by the Animal Care Committee of Simon Fraser University, which adheres the guidelines set by the Canadian Council for Animal Care and abides by the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Reverse Transcription–Polymerase Chain Reaction

Embryos were decapitated and the heads immersed in RNA preservative (RNALater; Ambion, Austin, TX) and stored at 4°C. Several hours later, the eyes were separated from the heads and the lenses removed. The resultant eye cups were immersed in fresh solution and stored at 4°C. Opsin cDNAs were generated by RT-PCR amplification of DNase-treated total RNA isolated from homogenized eye cups (RNAqueous-4PCR; Ambion). Reverse transcription was performed at 42°C for 15 minutes. Cycling parameters for the subsequent PCR were: 95°C × 5 minutes; 32 cycles of 95°C × 30 seconds, 56°C × 30 seconds, and 72°C × 1 minutes; and 1 cycle of 72°C × 10 minutes. Primers used for RT-PCR were designed from published opsin sequences to span across introns to distinguish amplicons from DNA contamination. Reaction products were analyzed by electrophoresis on 1% agarose gels in 1 × Tris borate-EDTA (TBE) containing 0.5 μg/mL ethidium bromide, and photographed. The primers used were UV opsin: forward 5′-GGG CTT TGT GTT CTT TGC TG-3′, reverse 5′-GGT ACT CCT CGT TGT TTG TG-3′; blue opsin: forward 5′-ATG AAC ACA ATG AGG TCG AA-3′, reverse 5′-TTA ACC AGC AGA AGA CAC TT; red opsin: forward 5′-AGC AAG ACA AGA CAA CAG AA-3′, reverse 5′-TGA GAG GAT GAC CAC TAT GA-3′; green opsin: forward 5′-ATG CAG AAC GGC ACA GAA GG-3′, reverse 5′-TTA TGC AGG GCC CGC AGA AG-3′; rod opsin: forward 5′-CCC TTT CCA TCT CTC TTT CT-3′, reverse 5′-CCA TGA GTG AGT ACG CCG CC-3′; and β-actin: forward 5′-CCC ATG GAG CAC GGT ATC ATC AC-3′, reverse 5′-GCG TGG GGC AGA GCG TAA CCT TC-3′. 

Preparation of Opsin Riboprobes

Opsin partial cDNAs were generated by RT-PCR from total RNA isolated from eye cups as described previously. The primers used were: UV opsin partial cDNA forward 5′-GGG CTT TGT GTT CTT TGC TG-3′, reverse 5′-GGT ACT CCT CGT TGT TTG TG-3′ (GenBank accession no. AY214148, our probe corresponds to bases 111-574 of this sequence; http://www.ncbi.nlm.nih.gov/Genbank; provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD); blue opsin partial cDNA: forward 5′-AAA CCT TGG TAG TGG GGA TT-3′, reverse 5′-CAT AGA AGA TAG CAC TGC CC-3′ (accession no. AF425075, our probe corresponds to bases 119–312 of this sequence); red opsin partial cDNA: forward 5- AGC AAG ACA AGA CAA CAG AA-3′, reverse 5′-TGA GAG GAT GAC CAC TAT GA-3′ (accession no.AF425073, our probe corresponds to bases 33–273 of this sequence); green opsin partial cDNA: forward 5′-AAA ATA GGC AAA AGG TTC AC-3′, reverse 5′-TAG ACG GCA AGA CAA TAG TA-3′ (accession no.AF425076, our probe corresponds to bases 1–192 of this sequence); rod opsin partial cDNA: forward 5′-CCC TTT CCA TCT CTC TTT CT-3′, reverse 5′-CCA TGA GTG AGT ACG CCG CC-3′ (accession no.AF425072, our probe corresponds to bases 8–184 of this sequence). The cDNAs were gel purified and cloned into TOPO TA cloning vector pCRII (Invitrogen, Carlsbad, CA) and sequenced by AmpliTaq (Applied Biosystems [ABI], Foster City, CA) dye terminator cycle sequencing (University of British Columbia [UBC] Sequencing Laboratory). The identity of the sequence was confirmed by comparing it to nucleotide sequence databases using the BLASTN program (BLAST www.ncbi.nlm.nih.gov/blast/ provided in the public domain by the National Center for Biotechnology Information, Bethesda, MD). To fabricate the cRNA probe, a PCR fragment containing the partial cDNA insert and an RNA promoter amplified from the pCRII vector was used to generate sense and antisense riboprobes by in vitro transcription. Riboprobes were labeled with digoxigenin (DIG), fluorescein, or biotin (Roche Diagnostics, Indianapolis, IN). 

In Situ Hybridization and Quantification of Opsin Expression

Embryos were decapitated and the heads fixed overnight at 4°C in a solution containing 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS). The tissue was then rinsed in PBS (three times for 30 minutes each), cryoprotected in sucrose solution (30% sucrose, 0.1 M PBS in optimal cutting temperature [OCT] medium) overnight at 4°C, and cryoembedded in 100% OCT medium (Cedar Lane Laboratories, Burlington, ON, Canada). The resultant blocks were cut in 5- to 7-μm steps to reveal (1) transverse sections of the head (coronal cuts of the brain), such that the nasal retina appeared first and the temporal retina last; or (2) longitudinal sections of the head (sagittal sections of the brain), such that the central retina appeared first and the cornea last. The cryosections were used for in situ hybridization with the various opsin riboprobes as per published studies. ²² Briefly, the procedure involved rehydrating the sections, permeabilizing them in 10 μg/mL proteinase K (Sigma-Aldrich, St. Louis, MO) for 5 minutes, followed by exposure to 0.1 M triethanolamine containing 0.25% acetic anhydride, dehydration, and hybridization overnight at 56°C with 1 to 5 μg riboprobe in hybridization solution containing 50% formamide and dextran sulfate. Sections were then treated with RNaseA (Sigma-Aldrich) and incubated with appropriate Fab fragments conjugated to alkaline phosphatase (1:3000; Roche Diagnostics) for 3 hours at room temperature. The riboprobes were visualized using 5-bromo-4-chloro-3indolyl phosphate with 4-nitroblue tetrazolium chloride (NBT/BCIP; Roche Diagnostics). Sense probes were used as negative controls and did not hybridize in any of the retinas. Digital images of sections were acquired with a microscope (model E-600 Nikon, Tokyo, Japan) equipped with a digital camera (DXM-100; Nikon) and DIC optics. 

Serial sections were collected cyclically on poly-l-lysine–coated slides for in situ hybridization treatments with each of the four cone opsin probes and the rod probe. These slides, along with positive and negative controls (made up of slides containing retinal sections from older fish), were processed simultaneously, permitting comparison of results between probes. To elucidate the order of opsin appearance, absence or presence of label was noted for each probe, and counts of labeled cells were conducted at early stages of opsin expression over a similar stretch of radially oriented retina (20-μm length of photoreceptor layer). This stretch was chosen by careful examination of all sections and selecting consecutive sections that showed the highest number of labeled cells for each probe. The selection was possible because the site of initiation and expansion dynamics were similar for all early-appearing opsins (see the Results section). Based on the dimensions of the eye at each stage analyzed, the number of sections required to cut it entirely, the extent of labeling of the sections, and the presence of ocular landmarks (the embryonic fissure and the lens), we were able to deduce the approximate pattern of opsin expression during development. 

Immunocytochemistry

Embryos were collected for immunocytochemical analysis using the mouse zpr-1 monoclonal antibody (Zebrafish Monoclonal Facility, University of Oregon, Eugene, OR). These samples were embedded in OCT medium and sectioned as described previously. After three washes for 5 minutes with 50 mM PBS, the cryosections were blocked with PBST (2% normal goat serum in PBS/0.3% Triton X-100) for 1 hour. These were then incubated with the zpr-1 antibody (1:100) in PBST for 2 hours at 4°C, washed in PBST, and incubated in goat anti-mouse IgG conjugated to FITC antibody (1:500) in PBST for 1 hour. After several washes with PBST, the slides were mounted and photographed with the microscope (E-600; Nikon), equipped for fluorescence imaging. 

Photoreceptor Ultrastructure

The stages at which opsin mRNA was first detected by RT-PCR and in situ hybridization were analyzed for photoreceptor ultrastructure with a transmission electron microscope (TEM; H7600; Tokyo, Japan). Whole heads were fixed overnight at 4°C in primary fixative (2.5% glutaraldehyde and 1% paraformaldehyde in 0.06 M PBS), followed by postfixation in secondary fixative (1% osmium tetroxide in 0.06 M PBS) for 1 hour at 4°C. After a brief wash, the tissue was dehydrated through a series of solutions of increasing ethanol concentration, infiltrated with mixtures of propylene oxide and Epon resin, and embedded in 100% Epon resin. ² Blocks were cut with an ultramicrotome (UC6; Leica, Deerfield, IL), to obtain 70-nm-thick sections that were collected on 200-μm grids. The grids were stained with 2% uranyl acetate followed by lead citrate, and the sections were photographed by TEM. 

Results

Opsin Expression by RT-PCR

In chinook salmon, expression of UV opsin mRNA by RT-PCR was first detected at 288 ATUs (2/8 amplifications). This partial transcript was 600 bases long and was probably the unspliced version of the 464 base-long transcript first detected at 317 ATUs (Fig. 1) . From 317 ATUs onward, all amplifications resulted in UV opsin product, with increased intensity of the band as a function of total ATUs (Fig. 1) . At 326 ATUs, expression of green opsin transcript was clearly visible in half of the samples (4/8 amplifications). At 368 ATUs the green opsin band was more prominent than at 374 or 432 ATUs, even when the loaded sample volume was three times more for the latter two dates. Red opsin transcript first appeared at 336 ATUs (1/8 amplifications) but only consistently at 368 ATUs (Fig. 1) . Blue opsin transcript was the last to be expressed, ∼432 ATUs (Fig. 1) . Rhodopsin transcript was present at 317 ATUs, concomitant with the appearance of UV opsin transcript (Fig. 1) . These results suggest the following sequence of opsin appearance: rod, UV, green, red, and blue. 

Expression of Cone Opsins by In Situ Hybridization

The sequence of cone opsin expression revealed by RT-PCR in chinook salmon was different from that revealed by in situ hybridization of opsin riboprobes on cryofixed retinas. In this case, expression of UV and red opsin mRNAs was first detected at 350 ATUs. Of 20 embryos processed in the range of 350 to 362 ATUs, 2 had retinas that labeled exclusively with the UV riboprobe, whereas 7 labeled with both the UV and red riboprobes. A one-way ANOVA performed on average cell counts over a 20-μm stretch of photoreceptor layer revealed that the number of UV cones (average ± SD: 1.2 ± 1.8) was not significantly different from that of red cones (0.8 ± 1.3; F_1,39 = 0.689, P = 0.412). In subsequent days, however, red cones became more numerous than UV cones; for instance, at 432 ATUs (Fig. 2) , the average number of UV cones (9.8 ± 3.3) over a 20 μm stretch of photoreceptor layer was significantly lower than the number of red cones (13 ± 3.2; F_1,19 = 4.76; P = 0.043). At higher ATUs (>1 week after hatching), UV and red cones showed similar labeling intensity (Fig. 2) . Labeling with the green riboprobe was first detected at 456 ATUs, with multiple cells clearly labeled at 468 ATUs (Fig. 2) . Both the red and green riboprobes labeled one member of the double cone, forming a square mosaic pattern similar to that established by the UV cones (Figs. 2M 2N 2O) . Blue opsin mRNA was first detected at 690 ATUs (Fig. 3A)and the label was restricted to single cones that occupied the center position (Figs. 3B 3C)in the square mosaic formation characteristic of cone photoreceptors in the salmonid retina. ^{2 22} These labeling patterns are consistent with microspectrophotometric observations of visual pigment absorbance from the various morphologic types of photoreceptors in the retina of salmonid fishes, ^{22 29} attesting to the specificity of the probes. The order of cone opsin appearance obtained by in situ hybridization was therefore: UV, red, green, and blue. 

Spatiotemporal Dynamics of Cone Opsin Expression

The first opsin mRNAs to appear (UV and red) were found within a small patch of cells located in the centrotemporal retina (Figs. 4A 4B 4C) . That these cells were committed cone photoreceptors was verified by immunocytochemical labeling with the zpr-1 antibody (Figs. 4D 4E) . In salmonid fishes, we have found that this antibody labels cone photoreceptors (Figs. 4F 4G) , as is the case in zebrafish and goldfish (though, in these species, it primarily labels double cones). ²¹ At the time of first opsin mRNA detection by in situ hybridization, the cones had not yet developed outer segments, although the inner segments were clearly visible (Fig. 4H) , which was not the case in retinas at the time of first opsin mRNA detection by RT-PCR (Fig. 4I) . 

With higher ATUs, expression proceeded toward the temporal and dorsal retina with little expansion toward the ventral retina. This pattern of expansion was visible in both transverse and longitudinal sections, as illustrated by UV and red riboprobe labeling (Figs. 5 6 7) . At 468 ATUs, UV label increased from the nasal (Fig. 5A)to the temporal retina (Fig. 5I) . In accordance with this pattern, longitudinal sections revealed strong labeling in the central and temporal retina with diminished to no labeling in the ventral and nasal retina (Fig. 6) . The same pattern was observed for red opsin mRNA expression (Fig. 7) . By 690 ATUs, UV and red opsin mRNAs were detectable in all parts of the retina except in the peripheral area ventral to the embryonic fissure. At this stage, green opsin mRNA was distributed over a large area of the centrotemporal retina after expansion dynamics similar to that of the UV and red opsins (data not shown). 

Blue labeling was first observed in the centrotemporal retina (Fig. 3 , the same area where the mRNAs of other opsins first appeared). By 768 ATUs, blue label was present across the lower half of the ventral retina, nasal to the embryonic fissure (Fig. 3D) . This labeling was opposite that of the UV label (Fig. 3E) , which remained present in the upper three quarters of the retina. The spatial progression of blue opsin mRNA expression was therefore different from that of the other cone opsins. 

Rhodopsin Expression

Rhodopsin mRNA expression was first detected in the centrotemporal retina (Figs. 8A 8B)and it progressed temporally and dorsally (Figs. 8C 8D 8E) , as per most of the cone opsins. Localization of rhodopsin mRNA was present in the budding inner segments at early stages of expression (Figs. 8A 8B) , and then became concentrated in small cell bodies located more vitreal and between cone inner segments (Figs. 8F) , in accordance with rod inner segment positioning in the light-adapted retina of salmonid fishes. ^{2 22}  

Other Salmonid Species

Figure 9shows a summary of RT-PCR and in situ hybridization results for all the salmonid species studied. Regardless of the rate of development, several observations appeared to be common. First, there was a delay of at least 1 day between detection of an opsin transcript by RT-PCR and corresponding detection by in situ hybridization. Second, the UV and red opsin mRNAs were always among the first transcripts to be detected by in situ hybridization. Third, detection of green opsin mRNA by in situ hybridization was much delayed compared with detection by RT-PCR; by the latter technique, green opsin mRNA was detected before or around the time of UV and red opsin mRNA expression. Fourth, except for the Atlantic salmon, blue opsin mRNA was detected last and, by in situ hybridization, it appeared after hatching and closer to full yolk sac absorption (swim up, alevin, stage) or later. In contrast, by in situ hybridization, mRNAs for UV and red opsins appeared up to 2 weeks before hatching, and those for rhodopsin and green opsin appeared just before or around the time of hatching. We have shown that, unlike Pacific salmonids, the young Atlantic salmon shows blue cones shortly after hatching, ²² in agreement with the early expression of this opsin reported here for this species. 

Discussion

Both the RT-PCR and in situ hybridization results show UV and red opsins to be expressed almost simultaneously during salmonid retinal development. Our detailed observations of chinook salmon embryos and results from rainbow trout and Atlantic salmon (see Fig. 9 ) further indicated that UV opsin expression preceded that of red opsin. The salmonid UV opsin belongs to the family of opsins that also encompasses the violet-blue opsins of mammals (the SWS1 opsin gene family). ⁸ Red opsins from both groups of vertebrates also belong to the same (M/LWS) opsin gene family. ⁸ Hence, the sequence of cone opsin expression from SWS1 to M/LWS is the same for salmonids and mammals. Also as in mammals, ^{19 20} rhodopsin expression in salmonid fishes appears to be delayed with respect to that of the first cone opsin (Fig. 9 , in situ hybridization results), despite the presence of precocial rods that form a square mosaic around the single center cones. ² In cyprinid fishes, ^{9 13} precocial rods express the first opsin (rhodopsin) and form rows that lead the opsin expression front. 

Another point in common between salmonid fishes and mammals is the presence of dual opsin-expressing cones during retinal development. In salmonid fishes, the center cones, which, like the rest of the single cones, express UV opsin at hatching (Fig. 2M) , ^{22 23 24} progressively start changing over to blue opsin closer to the time of yolk sac absorption (Fig. 3 ; the salmonid blue opsin belongs to the SWS2 opsin gene family). ⁸ This switch originates in the same area where other opsin transcripts are first detected and eventually reaches the dorsal retina. In mammals (rodents, primates), the first cones to appear express an SWS1 (violet/blue) opsin. ^{4 17 18} Later on, dual opsin expressing cones are found, and it is believed that a subpopulation of these gives rise to the longer wavelength (red and green opsin) cones by opsin switch. ^{5 17 18 28} Coexpression of photoreceptor opsins has not been reported in goldfish, ¹³ zebrafish, ³⁰ or chick. ¹⁶  

The origin of opsin expression and its spatial progression in the salmonid retina are different from those described for goldfish ¹³ and zebrafish. ^{9 11} The origin of expression in the centrotemporal retina of salmonid fishes differs from the centroventral position described for goldfish. ¹³ With the exception of the blue opsin, expression of the other opsins in salmonid fishes then progresses toward the temporal and dorsal retina, in similar fan-shape dynamics as described for cyprinids. ⁹ Expression of blue opsin in salmonids seems to follow a wave of single cone transformation (from UV-to-blue opsin expression) that starts in the centroventrotemporal retina and progresses in the opposite direction to the other opsins. Differences in patterns of expression between opsins have also been observed in the chick retina. ¹⁵ In this animal, green and red opsins are first detected slightly nasal to the optic disc while violet and blue opsins are expressed in cell patches at this site and in the temporal retina. Expression then proceeds peripherally with a bias toward the temporal retina. Rhodopsin, in the chick, is first detected surrounding the embryonic fissure in the centroventral retina, and expression is then confined to the ventral retina. ¹⁵ The reasons for these differences in opsin dynamics within and between species are unknown. 

Analogous discrepancies in opsin sequence and time of appearance to those reported in this study derived from RT-PCR versus in situ hybridization have also been found in the chick. ¹⁶ It has been suggested that inhibitory pathways, perhaps acting through the retinal pigment epithelium, maintain opsin expression at very low levels in the retina at early stages of development. ¹⁶ Such opsin transcripts could be amplified by RT-PCR, but their importance to early photoreceptor differentiation is unknown. In salmonid fishes, there are no morphologic signs of photoreceptor differentiation at the time when opsin messages are first detected by RT-PCR. The purpose for the upregulation of the green opsin message, observed at 368 ATUs (Fig. 1) , is also puzzling. It is possible that such upregulation is due to the transient expression of different opsin isoforms, as occurs during retinal development of the zebrafish. ¹⁴ The reason for such isoforms, which have very close absorbance profiles, ¹⁴ is also unknown. 

 Supported by Natural Science and Engineering Research Council (NSERC) Operating Grant 238886 (INF).

 Submitted for publication April 19, 2006; revised September 8, 2006; accepted December 4, 2006.

 Disclosure: C.L. Cheng, None; K. J. Gan, None; I. Novales Flamarique, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Opsin transcripts detected by RT-PCR during retinal development of chinook salmon. Bands shown are representative of at least half of the amplifications for any given date. Right: the gene families to which the various salmonid opsins belong.

Figure 2.

 

Micrographs of cryosections from the retina of chinook salmon labeled by in situ hybridization with one of the four cone opsin riboprobes (UV, red, green, or blue). (A–D) At 374 ATUs, only UV and red opsin mRNAs were detected (arrows: labeled cells). (E–H) At 432 ATUs, the number of cells labeled with the red riboprobe was, on average, slightly higher than that labeled by the UV riboprobe. (I–L) At 468 ATUs, a small number of green-labeled cells were visible. (M–P) Oblique sections from embryos at 672 ATUs show that the UV riboprobe labeled the single cones, whereas the green and red riboprobes labeled opposite members of the double cone pair (arrowheads: the membrane partition that separates the two members of a double cone). At this stage of development, there was still no labeling with the blue riboprobe. rpe, retinal pigment epithelium. Scale bar, 11 μm.

Figure 3.

 

Micrographs of cryosections from chinook salmon at various ATUs showing blue and UV opsin mRNA labeling. (A, B) Radial sections showing increased labeling of single cones in the centroventrotemporal retina with incremental ATUs; d, double cone. (C) Tangential section showing that the blue label was restricted to the center cones of the square mosaic. (D, E) Serial longitudinal sections showing labeling by the blue riboprobe in the ventral retina (D) and corresponding label by the UV riboprobe in the dorsal retina (E). White arrows: limits of labeling; black arrows: individual labeled cells; l, lens; ef, embryonic fissure. ATUs are shown on the bottom left of each panel. Other nomenclature as in Figure 2 ; retinal directions (D, dorsal; N, nasal) in (D) apply to (E). Scale bars: (A–C) 16 μm; (D, E) 160 μm.

Figure 4.

 

Micrographs of cryosections and EPON-embedded sections from the retina of chinook salmon. (A) UV opsin expression started at ∼374 ATUs in the centroventrotemporal retina (rectangle). (B) Labeled UV cones (black arrows) within the rectangle in (A). (C) Labeled red cone (black arrow) within an analogous area to that in (A), micrographed from the following (serial) section. (D–G) Labeling by fluorescence with the zpr-1 antibody in 374 ATU embryos (D, E) showing restricted label to the differentiating photoreceptor layer, and similar labeling in 10 g chinook smolts showing label of cone photoreceptors (F, G); d, double cone; s, single cone. Black and white arrows: the same structures under bright-field and fluorescence imaging, respectively. (H, I) Radial sections showing photoreceptor ultrastructure in embryos at 374 ATUs (H) and 317 ATUs (I). (H, arrows) budding inner segments; n, nucleus. Other nomenclature as in Figure 2 . Magnification bar: (A) 160 μm; (B–G) 16 μm; (H) 2.1 μm; (I) 2 μm. Retinal directions: C, central; D, dorsal.

Figure 5.

 

Micrographs of transverse head cryosections from 468 ATU chinook salmon showing UV opsin mRNA labeling in (A, B) nasal, (C ,D) centronasal, (E, F) central, (G, H) centrotemporal, and (I, J) temporal regions of the retina. For each set of panels, the one on the right shows the area encompassed by the rectangle depicted in the adjacent panel on the left. Nomenclature as in Figures 3 and 4 ; retinal directions in (A) apply to (A, C, E, G, I). Scale bar: (A, C, E, G, I) 160 μm; (B, D, F, H, J) 16 μm.

Figure 6.

 

Micrographs of longitudinal head cryosections from 468 ATU chinook salmon showing UV opsin mRNA labeling in (A, D) central, (B, E) centrotemporal, and (C, F) temporal regions of the retina. The retinal area within the rectangle depicted in each of the top panels is shown on the corresponding bottom panels. Retinal directions: D, dorsal, N, nasal (in A) hold for (A–C). Nomenclature as in Figure 3 . Scale bar: (A–C) 160 μm; (D–F) 16 μm.

Figure 7.

 

Micrographs of transverse and longitudinal head cryosections from 468 ATU chinook salmon showing red opsin mRNA labeling in (A, B) nasal, (C, D) temporal, and (E, F) centrotemporal and centrodorsal regions of the retina. For each set of panels, the one on the right shows the area encompassed by the rectangle depicted in the adjacent panel on the left. Nomenclature as in Figures 3 and 5 ; retinal directions in (A) apply to (C). Scale bar: (A, C, E), 160 μm; (B, D, F) 16 μm.

Figure 8.

 

Micrographs of transverse and longitudinal head cryosections from chinook salmon at various ATUs showing rhodopsin mRNA labeling. (A, B) Initial site of label in the centrotemporal retina at 432 ATUs. (C, D) Enhanced labeling in the centrotemporal retina 3 days after first rhodopsin mRNA detection. (A–D) Right: the area encompassed by the rectangle depicted in the adjacent left panel. (E) Longitudinal section showing that early rhodopsin label stretches along the centrotemporal retina. (F) Labeling of discrete cell inner segments located between and vitread to the double cone ellipsoids. Nomenclature as in Figures 2 3 and 5 ; retinal directions in (A) apply to (C). Scale bar: (A, C, E) 160 μm; (B, D, F) 16 μm.

Figure 9.

 

Approximate onset of expression (arrows) of the various opsin mRNAs for each species of salmonid fish investigated. U, R, G, B, and r correspond to UV, red, green, blue and rod opsin mRNAs, respectively. Also shown are the ranges of ATUs at which main developmental stages (eyed egg, hatching into a yolk sac embryo , and full yolk sac absorption and swim-up as an alevin ) occurred for each species. Chum embryos at 411 and 428 ATUs (available previous samples to those reported) did not reveal any opsin mRNA expression by RT-PCR and in situ hybridization, respectively. Similarly, no opsin mRNA expression was observed by in situ hybridization in 425 ATU Atlantic salmon embryos. RT-PCR analysis was not performed for Atlantic salmon.