Thyroid Hormone Induces a Time-Dependent Opsin Switch in the Retina of Salmonid Fishes

Abstract

purpose. To determine the role of thyroid hormone in inducing the UV (SWS1)-to-blue (SWS2) opsin switch in the retina of two salmonid fishes, the coho salmon (Oncorhynchus kisutch) and the rainbow trout (O. mykiss).

methods. Fish were treated with thyroid hormone (T₄) or the vehicle solution (0.1 M NaOH, control), exogenously or by intraocular injection, at different life history stages. Microspectrophotometry and in situ hybridization with riboprobes against the SWS1 and SWS2 opsins were used to reveal the dynamics of opsin expression in treated and control animals. To assess whether thyroid hormone induced differentiation of retinal progenitor cells into cones, treated and control fish were injected intraocularly with bromodeoxyuridine (BrdU) and the number of proliferating cells in the outer nuclear layer (ONL) determined. These observations were accompanied by histologic counts of cone densities.

results. Thyroid hormone induced a reversible UV-to-blue opsin switch in differentiated single cones of juvenile salmonids (alevin and parr stages), but failed to exert any effect in the retina of older fish (smolt stage). The switch progressed from the ventral to the dorsal retina in clockwise fashion. Thyroid hormone did not induce cone density changes or alterations in the number of BrdU-labeled cells, which were the same in control and treated animals.

conclusions. Thyroid hormone induces a UV (SWS1)-to-blue (SWS2) opsin switch in the retina of young salmonid fishes that is identical with that occurring during natural development. The switch occurs in differentiated photoreceptors, is reversible (maintained by thyroid hormone exposure), and can be induced only before its natural onset. Thyroid hormone did not cause changes in the number of proliferating cells in the ONL. These results conform to the dynamics of thyroid hormone–induced opsin expression in the mouse and are consistent with the opsin plasticity found in differentiated photoreceptors of the fruit fly, Drosophila melanogaster. This work establishes a role for thyroid hormone in triggering opsin switches in the vertebrate retina.

Thyroid hormones (3,5,3′-triiodothyronine, T₃, and its precursor, thyroxine, T₄) play crucial roles during development of most vertebrate organs. ^{1 2} In conjunction with their receptors, thyroid hormones regulate neuronal differentiation and migration, ¹ and their involvement is essential for proper maturation of the brain and related organs such as the cochlea ³ and the retina. ^{4 5 6 7} In the latter organ, thyroid hormones regulate expression of visual pigment proteins (opsins) in differentiating cone photoreceptors of the mouse ^{5 7} and winter flounder. ⁶ It is the combination of cone photoreceptors expressing different opsins that allows an animal to detect color. 

Most vertebrates express a combination of cone opsins maximally sensitive to ultraviolet/violet, violet/blue, green, or green/red light (SWS1, SWS2, RH2, and MWS/LWS opsins, respectively). ⁸ These opsins appear sequentially during retinal development. ^{9 10 11 12 13 14} In the mouse, the ontogeny of opsin expression is such that a violet/blue (SWS1) opsin appears prenatally and is followed by expression of a green (MWS/LWS) opsin around postnatal day 10. ^{5 13} Knockout animals lacking the thyroid hormone receptor TRβ2, or having one that is ligand-binding defective, fail to express green opsin, demonstrating that both the hormone and its receptor are necessary for green opsin expression. ^{4 13} In the winter flounder ⁶ and in the rainbow trout ¹⁵ exposure to exogenous thyroid hormone induces changes in opsin transcription, as has been detected by in situ hybridization with opsin riboprobes and quantitative real-time PCR, respectively. 

Many vertebrates, including humans, ^{12 16} salamanders, ¹⁷ several species of rodent, ^{13 18 19 20} and fish ^{21 22 23 24 25} can coexpress multiple opsins within a cone photoreceptor. In some of these animals, opsin coexpression occurs in cones at the marginal growth zone during retinal development. Ultimately, opsin coexpression in these cones is resolved so that only the latter opsin to appear in the developmental sequence is expressed. In the human retina, for instance, cones coexpressing blue (SWS1) and green/red (MWS/LWS) opsins appear in the retinal margin before cones expressing only green/red opsin. ^{12 16} In mouse, coexpression of violet/blue and green opsins leads to a graded retina with highest expression of green opsin dorsally and lowest ventrally. ^{7 13} This spatial pattern of opsin expression is maintained in the adult by a thyroid hormone gradient. ⁵ A mechanism by which cone spectral phenotypes arise sequentially through opsin coexpression has yet to be elucidated. 

Cone opsin expression may occur sequentially in differentiating cone precursors or may arise through switches in opsin expression within terminally differentiated cones. The growing number of species in which coexpression has been found during retinal development, ^{13 16 19 20 22 23} and the fact that thyroid hormone and its receptors appear to control opsin expression, but not cone proliferation or apoptosis, ^{4 5 7 16 22} suggests that the latter possibility may be common. Indeed, opsin switching has been demonstrated in the rat and the gerbil, where a subpopulation of violet/blue (SWS1) cones gives rise to green (MWS/LWS) cones, ²⁶ and in salmonid fishes. In these animals, single cones switch from an ultraviolet (UV, SWS1)- to a blue (SWS2)-light–sensitive opsin during the early juvenile period. ^{21 22 23} This switch progresses from the ventral to the dorsal retina and is accompanied by the loss of a subpopulation of cones (the corner cones) from the ventral retina. ^{14 22} Of interest, the larval eye of the fruit fly, Drosophila melanogaster, undergoes a similar transformation: terminally differentiated Rh5 photoreceptors switch opsin, while preexisting Rh6 photoreceptors become apoptotic. ²⁷  

Salmonid fishes, like most lower vertebrates, have single and double cones arranged in row-to-square formations termed mosaics. Single cones express the shorter wavelength opsins (UV and blue), whereas double cones express the longer wavelength opsins (green and red, one per double cone member). ^{22 28 29} The presence of the switch and patterned mosaic distribution of cones in the retina of these animals (where every opsin class is associated with a given morphologic cone type) make them particularly attractive models for unravelling regulatory mechanisms of opsin expression. 

In this study, we assessed whether thyroid hormone (T₄) could induce the switch from UV to blue opsin in the retina of salmonid fishes. After exposing fish to T₄ at different life history stages, we characterized the spatial and temporal dynamics of UV and blue opsin expression by in situ hybridization and microspectrophotometry. We hypothesized that T₄ induces the switch during the early juvenile period (spanning the alevin and parr stages) by repressing UV and enhancing blue opsin expression. We also predicted that treatment of older fish (smolt stage) would either reverse the initial effect ³⁰ or have no consequence, if a closed chromatin state followed the natural opsin switch. ^{2 29} In addition, thyroid hormone could either induce changes in the number of single cones ^{30 31} or not. ²² To investigate an alternative cell death and replacement model, by which UV opsin-expressing cones die and are replaced by newly generated blue cones, and vice versa, we assayed cell proliferation in the retinas of T₄-treated fish by BrdU immunohistochemistry. A better understanding of the role of thyroid hormone in regulating opsin expression and mosaic structure in vertebrates is critical to our understanding of cone differentiation and retinal plasticity. 

Methods

Animals

Wild stock coho salmon (Oncorhynchus kisutch) and rainbow trout (O. mykiss) were obtained from the Capilano River and the Fraser Valley Trout hatcheries (British Columbia, Canada). The fish in this study ranged from the alevin stage to the smolt; their statistics are presented in Table 1 . Salmonid fishes that have recently absorbed their yolk sac and started to feed on their own are called alevins. ²⁸ As they grow, their bodies develop black vertical marks and the fish become known as parr. The parr undergoes a physiological transformation for life in sea water and is then known as a smolt. ²⁸ Nonanadromous species (like the rainbow trout) remain in fresh water throughout life, but their visual system undergoes a transformation similar to that occurring in anadromous (sea-migrating) species, like the coho salmon. ^{22 23} The fish were held at a temperature of 6°C under a 12-hour dark–12-hour light regimen. For each set of experiments, the fish were partitioned into two groups (control and treatment) and held in two sets of identical but separate, self-contained, tanks with bubblers to provide oxygenation. All holding and experimental procedures were approved by the Animal Care Committee of Simon Fraser University, which follows the guidelines set by the Canadian Council for Animal Care and the ARVO Statement for the Use of Animals in Ophthalmic and Visual Science. 

In Vivo Thyroid Hormone Treatments

The fish were exposed to thyroid hormone exogenously or by intraocular injection. Both forms of exposure gave similar results (P > 0.05 for all pair-wise t-tests on analogous retinal sectors), and so the data sets were combined in the statistical analyses. For the exogenous exposures, l-thyroxine (T₄; Sigma, St. Louis, MO) was dissolved in 0.1 M NaOH and added to the water to a final concentration of 300 μg L⁻¹ (treatment). Control fish were treated identically with the exception that only the vehicle solution (0.1 M NaOH) was added to the water. Treatment was performed every day for 6 weeks (alevins) or 2 weeks (parr and smolts) with half of the water in each tank changed daily. Previous research has shown that treatment of rainbow trout with thyroid hormone in this manner for 9 days is sufficient to induce changes in the expression of opsin transcripts. ¹⁵ Five (alevin) and three (parr, smolt) fish from the control and T₄-treated tanks were killed by quick spinal bisection and decerebration at the end of 2 weeks of exposure. To assess whether the effects of thyroid hormone were permanent, we maintained control and T₄-treated fish for 4 weeks (alevins) or 2 weeks (parr, smolt) after the end of exposures and killed, at these times, the same number of animals as described previously. 

In a more direct method of thyroid hormone delivery, we injected three fish at each stage with 5 μL (alevin, parr) or 10 μL (smolt) of 0.3 μg μL⁻¹ T₄ solution (treatment) or 0.1 M NaOH (control). Anesthetized fish were placed in a holder and irrigated with a buffered solution containing 50 mg L⁻¹ MS-222 (anesthetic). Using a micropipette (for the alevin) or a Hamilton syringe (for the parr and smolt), we made a small incision on the top (dorsal) part of the iris and delivered the solution into the eye. The minute incision was then covered with Vet-bond and the fish allowed to recover by switching irrigation to regular water, after which it was returned to the holding tank. Injections were performed for five consecutive days, and the fish (n = 3) were killed from control and treatment tanks on the sixth day. Similar injections have successfully induced repression of blue opsin and premature expression of green opsin in the young mouse. ⁵ Three fish per treatment were further reared for 2 weeks without injections and killed at that time. Retinas from these animals gave similar results to those of exogenously exposed fish left unexposed for the same period, whereas fish killed 1 day after injections gave comparable results to those exposed for 2 weeks exogenously. 

In Situ Hybridization

Fish were killed in the light-adapted state. The left eyeball was removed, the iris and lens discarded, and the remaining eye cup immersed in cryofixative (4% paraformaldehyde in 0.08 M phosphate buffer, pH = 7.4). After a 24-hour fixation at 4°C, the retina was extracted from the eye cup, flattened with small peripheral incisions, and laid over a grid with the optic nerve bud at the center of the grid and the tip of the embryonic fissure pointing downward and perpendicular to the bottom of the grid. This placement procedure ensured the analysis of similar pieces of retina within and between treatments. Each retina was then cut into 4 (alevin), 8 (parr) or 20 (smolt) pieces, and each piece embedded in optimal cutting temperature (OCT) frozen blocks for in situ hybridization as per previous studies. ^{23 32} Sections (7–10 μm thick) obtained from the blocks were collected serially and deposited on alternating slides for parallel processing with the UV and blue riboprobes. These riboprobes were species-specific ²² and their opsin specificity (i.e., exclusive hybridization of a given riboprobe to its corresponding opsin mRNA) has been demonstrated in previous studies. ^{14 22 23} The method of serial deposition of sections permitted comparison of labeling between riboprobes for the same retinal area. We also performed double labeling experiments with the two riboprobes simultaneously to further verify the results obtained by parallel processing of single riboprobes. The primers and method used to generate riboprobes as well as the in situ hybridization protocol have been published previously. ^{22 23} Digital images of sections were acquired with a microscope (E-600; Nikon, Tokyo, Japan) equipped with a digital camera (DXM-100; Nikon) and DIC optics. These images were used to obtain average percentages of single cones that expressed UV opsin mRNA. 

Histology

The eye cup from the right eyeball of the killed fish was immersed in primary fixative (2.5% glutaraldehyde, 1% paraformaldehyde in 0.08 M phosphate buffer, pH = 7.4). After a 24-hour fixation at 4°C, the retina was extracted from the eye cup and rinsed in 0.08 M phosphate buffer, and the pieces were cut according to the same procedure as was used for in situ hybridization. The pieces were then processed for embedding in Epon resin blocks, as described in previous publications. ^{23 33 34} Retinal blocks were cut tangentially in 1-μm steps and the sections stained to reveal the cone mosaic. ²³ Digital images were acquired with the microscope setup just described. 

The density of double cones and single cones (center cones + corner cones) over a 11,520-μm² area was counted by using a grid system on the computer monitor, and the ratio of double to single cones (d/s) was computed for each sector of the retina (d refers to the two-member, double cone pair). Based on this data and the percentage of UV opsin mRNA expressing cones obtained from the in situ hybridization sections, we generated retinal maps illustrating the changes in cone density, d/s ratio, and percentage of UV cones at the various stages. 

Microspectrophotometry

Individual fish from the various treatments were dark adapted overnight. After this adaptation period, the fish was killed, one eye enucleated, and the retina removed under infrared illumination. Small pieces of retina were teased apart and prepared for viewing with the dichroic microspectrophotometer (DMSP) as per previous studies. ^{22 28} The DMSP is a computer-controlled, wavelength-scanning, single-beam photometer that simultaneously records average and polarized transmitted light fluxes through microscopic samples. ^{35 36 37} The DMSP was equipped with ultrafluar objectives (Ultrafluar; Carl Zeiss Meditec, Inc., Dublin, CA): 32/0.4 for the condenser and 100/1.20 for the objective. With the aid of reference measurements recorded through cell-free areas, individual photoreceptor outer segments were illuminated sideways with a measuring beam of a rectangular cross-section of ∼2 × 0.6 μm. Absolute absorbance spectra were computed in 2-nm increments from the obtained transmittances (each spectrum consisted of an average of eight scans). The solid spectra (fits) were derived from experimental data by Fourier filtering. ³⁵  

Bromodeoxyuridine (BrdU) Immunohistochemistry

To assess whether T₄ treatment could induce proliferation of cells in the outer nuclear layer (ONL), treated and control rainbow trout alevins were injected intraocularly with 10 μL of BrdU (Molecular Probes, Eugene, OR) dissolved in 0.08 M PBS to a concentration of 20 mM. Injections took place at the beginning and at the end of a 2-week treatment period. Fish (n = 4) were killed immediately after and 1 week after cessation of treatment. These experiments were intended to evaluate whether thyroid hormone could induce proliferation of progenitor cells which, hypothetically, could give rise to blue opsin–expressing cones (during treatment) or to UV opsin–expressing cones (after cessation of treatment). 

Cryosections were hybridized with either the UV, blue, or rhodopsin riboprobe, ¹⁴ followed by immunohistochemical localization of incorporated BrdU. After riboprobe detection with 5-bromo-4-chloro-3-indolyl phosphate with 4-nitroblue tetrazolium chloride (NBT/BCIP; Roche Diagnostics, Indianapolis, IN), hybridized sections were washed in a 1:1 solution of 4N HCl:PBST (0.08 M phosphate-buffered saline with 0.5% Triton X-100) for 30 minutes. After washes in PBS and PBST, the sections were blocked (10% normal goat serum in 0.08 M PBS with 0.1% sodium azide) for 30 minutes and incubated with the primary antibody, monoclonal PRB-1 mouse anti-BrdU (1:20, Invitrogen/Molecular Probes), for 2 hours at room temperature. They were then washed briefly with PBST and incubated in secondary antibody, goat anti-mouse IgG conjugated to AlexaFluor 488 (1:200, Invitrogen/Molecular Probes), for 2 hours at room temperature. Both antibodies were diluted in PBST with 1% normal goat serum and 0.1% sodium azide. After several washes in PBST, the slides were mounted in glycerol and photographed with a microscope (E-600; Nikon), equipped for fluorescence imaging. In addition to imaging cryosections, BrdU immunohistochemistry was performed on retinal wholemounts from T₄-treated and control fish. 

Radial sections (n = 10 per retina) were used to count the number of BrdU-labeled cells in the ONL over a 443-μm stretch of centrodorsal retina. This area of the retina was chosen because, at the alevin stage, it is not yet affected by the naturally occurring UV-to-blue opsin switch. ^{14 22} The action of exogenous T₄ could thus be assessed without confounding effects from the natural switch. The mean from all retinas was computed for each treatment and expressed as a density per millimeter. 

Results

Cone Mosaic and Density after Thyroid Hormone Treatment

Both the coho salmon and rainbow trout had two types of cones (singles and doubles) arranged in row to square mosaics (Fig. 1) . Each double cone was composed of two cells apposed together by a double membrane partition and, at the level of the inner segment, had elliptical cross-section. The orientation of neighboring double cone elliptical cross-sections was such that they could form primarily a row mosaic (Fig. 1A)or a square mosaic (Figs. 1C 1E 1G) . In either mosaic type (Figs. 1A 1C) , the partitions of neighboring double cones formed a cruciform pattern with a single (center) cone located in the middle (at the hypothetical intersection of partitions). Other single cones, when present, faced the partitions (Figs. 1A 1C 1E) ; these are known as the corner cones. ^{23 32} We noted that, due to axial changes in double cone morphology (especially in the centrotemporal retina), the row mosaic at the ellipsoid level changed into a square mosaic at the nuclear level (Fig. 1B) . ³² Double cones in other retinal regions were less likely to change axial disposition (Figs. 1C 1D 1E 1F 1G) . 

The alevin retina had primarily square mosaics complete with corner cones (double to single cone [d/s] ratio ∼1). However, as the fish grew, there was a progressive loss of corner cones from the ventral retina (Fig. 1G)culminating in total loss (Fig. 1H , d/s ∼ 2) in the lower half of the ventral retina by the parr stage. The mosaic patterns were the same in the control fish and in those treated with thyroid hormone. 

Induction of an Opsin Switch by Thyroid Hormone in the Single Cones of Young Salmonid Fishes

Control alevin coho exposed to the vehicle solution (0.1 M NaOH) for 2 weeks had retinas in which the majority of the single cones (all in the dorsal retina and >97% in the ventral retina) expressed UV opsin mRNA exclusively (Figs. 2A 2B 2C 2D) . This result contrasted with the retina of thyroid hormone (T₄)–treated fish where the majority (>77%) expressed blue opsin mRNA, regardless of whether they were center or corner cones (Figs. 2E 2F 2G 2H) . Exposures continued for an additional 4 weeks during which the lower ventrotemporal retina of both control and T₄-treated fish experienced an incremental loss of corner cones (up to ∼60%). After this, the alevins left untreated for 4 weeks revealed a retina where all the single cones in the dorsal sector (Figs. 2I 2J)and most of the single cones in the ventral sector (Figs. 2K 2L)expressed UV opsin mRNA. This reversal in opsin expression occurred for both center and corner cones, with the lower ventral retina having lost most of the corner cones at this stage (Figs. 2K 2L) . These results were similar for rainbow trout alevins, although thyroid hormone action was even more pronounced (leading to all single cones expressing blue opsin mRNA in the ventral retina after T₄ treatment for 2 weeks) but corner cone loss was reduced compared with the coho retina (only up to ∼20% of the single cone population disappeared). 

Larger (parr) coho salmon treated with the vehicle solution for 2 weeks showed a mixed retina where center and corner cones could express UV or blue opsin mRNA or both (Figs. 3A 3B) . In the centrodorsal retina, most center cones tended to express blue opsin mRNA, whereas most of the corner cones expressed UV opsin mRNA. The lower half of the ventral retina had no corner cones and the remaining center cones expressed blue opsin mRNA exclusively (Figs. 3C 3D) . By contrast, after 2 weeks of T₄ treatment, all single cones (with the exception of the odd corner cone) expressed blue opsin mRNA (Figs. 3E 3F 3G 3H) . After the end of exposure, control parr left untreated for 2 weeks showed retinas in which UV opsin mRNA expression was primarily restricted to the corner cones (Fig. 3I)whereas most single cones expressed blue opsin mRNA (Fig. 3J) . Fish that had been treated with T₄, then left untreated for 2 weeks, had UV opsin expression primarily restricted to corner cones in the dorsotemporal retina (Fig. 3K) . Compared with the alevin, recovery of UV opsin mRNA expression was significantly lower (Figs. 3M 3O)and restricted to areas where expression was highest in the control. Nearly all the single cones in this retina expressed blue opsin mRNA exclusively (Figs. 3L 3N 3P) . 

Histologic and in situ hybridization results for the alevin and parr retinas are summarized in Figure 4 . Blue opsin expression began in the ventral retinas of the control fish and expanded nasally toward the dorsal retina. The loss of corner cones expanded peripherally, both nasally and temporally, toward the dorsal retina, and the extent of this nonreversible loss did not correlate with changes in single cone opsin expression. This result is further illustrated by the results of T₄ treatment. Cone densities and d/s ratios were statistically the same between analogous sectors of control and treated retinas at any stage (P > 0.05 for all pair-wise t-tests). 

Direct Evidence for a T₄-Induced Opsin Switch within Single Cones

The average number of cones labeled by the UV and blue opsin riboprobes in the T₄-treated retina of alevin and parr exceeded the total number of single cones found in the control retinas, suggesting that a population of single cones in the T₄-treated retina coexpressed UV and blue opsin mRNAs. Furthermore, the number of cones labeled with the blue opsin riboprobe in the T₄-treated retinas was greater than the difference in UV riboprobe-labeled cones between the control and T₄-treated retinas implying an upregulation of blue opsin mRNA in some of the residual UV opsin–expressing cones. Double in situ labeling with the UV and blue riboprobes confirmed the presence of single cones expressing both mRNAs in T₄-treated and control retinas (Fig. 5A) . Further evidence of a switch in opsins was obtained by microspectrophotometry. 

Measurements of visual pigment absorbance from the outer segments of isolated single cones revealed the presence of the two opsins (Fig. 5B) . Cones were found with peak absorbance (λ_max) in the UV at ∼380 nm but with increased absorbance at ∼455 nm, denoting a mixture of UV and blue visual pigments. Other cones had a graded absorbance along the outer segment characterized by blue absorbance at the base (λ_max ∼ 455 nm) shifting to violet absorbance (λ_max ∼ 405 nm) toward the tip (Fig. 5B) . These absorbance profiles had uncharacteristically large bandwidths at half maximum (HBW), especially for UV opsin–expressing cones, ^{17 38} in accordance with the presence of two visual pigments, as demonstrated previously. ²³ The peak absorbances of the visual pigments were also shifted toward longer wavelengths compared with the control, ^{22 23} in agreement with a switch in chromophore from vitamin A₁ to vitamin A₂ as a result of T₄ treatment. ²⁸ These absorbance profiles, together with the double label in situ hybridization results, provide direct evidence of an opsin switch induced by thyroid hormone that parallels events occurring during natural development. ^{14 23}  

Effect of Thyroid Hormone on Opsin Expression in the Retinas of Older Juveniles

The retina of the smolt was characterized by a lack of UV opsin mRNA expression regardless if the fish was a control or treated with T₄ for 2 weeks (Fig. 6) . All the single cones expressed blue opsin mRNA. Likewise, fish left untreated for an additional 2 weeks had no change in single cone opsin expression: all single cones expressed blue opsin mRNA (Fig. 6) . These results were the same in fish treated with T₄ by intraocular injection for five consecutive days indicating that the mode of delivery was not responsible for the negative results. Cone densities remained invariant regardless of treatment for analogous retinal sectors. 

Effect of Thyroid Hormone on Retinal Cell Proliferation and Topography

The number of BrdU-labeled cells in the centrodorsal retina of the rainbow trout alevin was statistically the same regardless of treatment (average [per millimeter] ± SD in the ONL was: 48 ± 16 [2 week control], 42 ± 19 [2 week T₄ treatment], 45 ± 15 [control 1 week posttreatment], and 51 ± 22 [1 week post-T₄ treatment]; ANOVA P > 0.05). These proliferating cells were most abundant near the retinal periphery (Figs. 7A 7B 7C 7D) , forming a ring of fluorescence around the retina. Analysis of sections further revealed that the BrdU label was predominantly located in the ONL, vitread to the cone inner segments and overlapping those of rod photoreceptors and their nuclei (Figs. 7E 7F 7G 7H 7I 7J 7K 7L 7M 7N 7O 7P 7Q 7R 7S 7T 7U 7V 7W 7X 7Y) . The labeling pattern was the same regardless of the treatment. 

Discussion

Regulation of Opsin Expression in Differentiated Single Cones by Thyroid Hormone

Our study shows that thyroid hormone induced a switch in UV-to-blue opsin expression in differentiated single cones of young salmonid fishes. Opsin switches in other animals likely involve differentiating protocones in growth zone areas, as these are the sites where sequential (new) opsins are first observed. ^{4 7 12 16 26} Although opsin switches probably occur in the peripheral growth zone of the salmonid retina as well, ²² our results clearly show that differentiated cones (as judged by morphology and functional synaptic connections leading to UV sensitivity) ³³ switch opsins throughout the entire retina. This finding, along with previous observations on the absence of switches between opsin families within double cones, ^{22 23} demonstrate that opsin expression is differentially regulated between morphologically distinct cone types. 

Our results corroborate previous findings showing that UV and blue opsin expression is not linked to single cone position in the retinal mosaic of salmonid fishes. ^{14 22 23 29} In contrast to earlier reports on rainbow trout claiming that thyroid hormone decreased the overall UV sensitivity of the animal due to corner cone loss, ^{30 31} cone densities in our study were not different between control and treated animals (Fig. 4) . In addition, the number of proliferating ONL cells was statistically the same regardless of treatment (∼45 mm⁻¹) further suggesting that thyroid hormone does not alter cone densities. Our results explain the trends in opsin downregulation measured by quantitative PCR in the retina of rainbow trout ¹⁵ and agree with the findings in mouse where thyroid hormone regulates opsin expression but does not induce cone apoptosis. ^{4 5 7} Nonetheless, the temporal dynamics of thyroid hormone action were such that center cones were targeted first and corner cones switched opsin expression thereafter (Fig. 3) . This was especially the case in the ventral retina where corner cones started to disappear as the opsin switch was taking place (Fig. 4) . This trend, previously observed during natural growth of salmonid fishes, ^{22 23} suggests differences in the expression of transcription factors that mediate opsin regulation and cell fate in center versus corner cones. 

Two candidate factors that may account for these differences are the retinoic acid receptor, RXRγ, and the retinoid-related orphan receptor, RORβ. RXRγ heterodimerizes with its receptor partner TRβ₂ to regulate opsin expression ³⁹ and, in the mouse, suppresses SWS1 opsin. ⁴⁰ In contrast, monomers of RORβ activate the mouse SWS1 opsin promoter. ⁴¹ It is possible that single cones in the salmonid retina express RORβ initially, to induce UV opsin expression, but that this factor is downregulated, starting with the center cones, as levels of thyroid hormone and its receptors increase. ^{42 43} Under this scenario, RORβ may play a dual role in maintaining UV opsin expression and cell survival in corner cones, as mice lacking RORβ undergo accelerated programmed cell death. ⁴¹  

Time Dependence of Thyroid Hormone Activity and Potential Mechanisms of Action

The effects of thyroid hormone on the salmonid retina were restricted to the younger juvenile stages (alevin and parr) and were absent in the smolt. During development in salmonid fishes, levels of thyroid hormone receptors and ligand (T₄, T₃) increase before full yolk sac absorption. ^{42 43} The progressive increase in receptor levels expands the stage (yolk sac alevin) ⁴³ when nearly all single cones express UV opsin. ¹⁴ Toward the end of yolk sac absorption, there is a sudden increase in thyroid hormone, ⁴² and center cones in the ventral retina commence switching from UV to blue opsin. ¹⁴ Although there are at present no measurements of thyroid hormone and related receptors in the retina of young salmonid fishes, the temporal expression of these variables bears resemblance to similar changes occurring during retinal development of the mouse. ^{5 7} In this animal, unliganded TRβ2 represses green opsin expression in the embryo whereas, postnatally, the liganded form is necessary for green opsin expression. ⁷ This expression is predominant in dorsal cones, coincident with the highest levels of thyroid hormone, ⁵ and resulting in a dorsoventral gradient of diminishing green opsin content of individual cones. ^{7 13} In analogy with the mouse, provided relative levels of blood plasma thyroid hormone and overall number of receptors in salmonid fishes translate into similar levels in the retina, it may be that the unliganded TRβ2 receptor (or other isoform) represses blue (SWS2) opsin expression at the embryonic stage (egg and yolk sac alevin), whereas the liganded form promotes SWS2 expression at later stages via the opsin switch. 

The reversibility of the opsin switch induced by exogenous thyroid hormone declined from the alevin to the smolt stage (Figs. 4 6) . This decline correlated well with the natural progression of the switch occurring in the retina of control fish. Thyroid hormone was thus incapable of regulating opsin expression after the switch had occurred naturally. These results are consistent with those from mouse where the hormone is ineffective at changing opsin expression once repression of the violet/blue (SWS1) opsin has been established. ^{5 7} Induction of either hyperthyroidism or hypothyroidism had no effects on opsin expression in adult mice ⁷ or in postmetamorphic winter flounder ⁶ once photoreceptor cells had fully differentiated. 

Effects of Thyroid Hormone on Cell Proliferation and Opsin Expression

The similarity in patterns of BrdU incorporation between treatments indicates that thyroid hormone does not enhance proliferation of retinal progenitors in the ONL, as may be expected if the changes in opsin expression observed in this study were due to differentiation of new cones expressing UV and/or blue opsin. Although we cannot rule out the possibility that some of the BrdU-labeled cells could differentiate into single cones, our results demonstrate that such a mechanism cannot account for the magnitude of opsin change observed between control and treated fish. This conclusion is supported, indirectly, by studies in goldfish where full differentiation of cones requires on the order of 3 weeks. ⁴⁴ The alterations in opsin expression observed in this study occurred far too quickly (in as little as 3 to 5 days after treatment initiation, data not shown) for differentiation of opsin-expressing cones to take place. Together, these observations are consistent with induction of opsin switches in single cones as a result of T₄ treatments, as demonstrated by our in situ hybridization and microspectrophotometry results. 

The number of BrdU-labeled cells found in this study are in the range reported by other investigators for the retina of rainbow trout alevin using an antibody against proliferating cell nuclear antigen (PCNA). ⁴⁵ Likewise, the BrdU localization pattern (in the ONL, coincident with rod opsin mRNA expressing cells) is similar between studies and suggests that the labeled cells belong to the rod lineage, as first established for the goldfish retina. ⁴⁶  

²  Present affiliation: Department of Biochemistry and Molecular Biology, University of British Columbia, Vancouver, British Columbia, Canada.

 Supported by Natural Science and Engineering Research Council of Canada (NSERC) Operating Grant 238886 (INF). CLC is the recipient of a Natural Sciences and Engineering Research Council (NSERC) PhD Canada Research Fellowship and KJG is the recipient of a NSERC MSc Canada Research Fellowship.

 Submitted for publication August 12, 2008; revised December 10, 2008; accepted April 13, 2009.

 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.

 

Table 1.

 

Size of the Fish Used in the Study

Fish and Treatment	Weight (g)	Length (cm)
Alevin
Coho
TH	0.37 (0.088)	3.89 (0.24)
Control	0.36 (0.091)	4.01 (0.30)
Post-TH treatment	1.13 (0.35)	4.95 (0.58)
Control posttreatment	1.17 (0.23)	5.10 (0.47)
RT
TH	0.11 (0.039)	2.73 (0.29)
Control	0.11 (0.016)	2.69 (0.31)
Post-TH treatment	1.01 (0.28)	4.70 (0.34)
Control posttreatment	1.04 (0.24)	4.65 (0.27)
Coho parr
TH	6.72 (0.76)	9.46 (0.72)
Control	7.12 (1.64)	9.30 (1.08)
Post-TH treatment	8.25 (2.32)	10.1 (0.74)
Control posttreatment	9.31 (1.36)	10.2 (0.67)
Coho smolt
TH	125 (28.1)	24.0 (2.12)
Control	118 (25.0)	24.2 (1.60)
Post-TH treatment	128 (26.5)	24.3 (1.19)
Control posttreatment	126 (20.8)	24.1 (2.06)
RT alevin-to-parr
TH (BrdU)	2.5 (0.52)	6.6 (0.28)
Control (BrdU)	2.6 (0.43)	6.5 (0.34)
Post-TH treatment (BrdU)	2.7 (0.42)	6.8 (0.36)
Control posttreatment (BrdU)	2.6 (0.48)	6.7 (0.31)

 Data are the mean (±SD). For any given group (n = 10), the treatment and control statistics were not significantly different (t-test with α = 0.05). For BrdU detection the fish (n = 4) were larger alevins (alevin-to-parr). In these experiments, TH or vehicle treatment lasted 2 weeks and posttreatment 1 week. RT, rainbow trout; TH, thyroid hormone (T4) treatment for 2 weeks (or intraocular injections for 5 days); post-TH treatment, fish left untreated for 4 weeks (alevin) or 2 weeks (parr, smolt) after cessation of TH treatment; control, fish that underwent the same treatment regimen as the TH-treated fish but with the vehicle solution (0.1 M NaOH) instead.

Figure 1.

 

Micrographs of cone mosaic formations in the light-adapted retina of coho salmon parr and related schematics of the unit mosaic in each retinal location. (A) Row mosaic from the temporal retina, at the level of the double cone ellipsoids, showing a full complement of corner cones (white asterisks); center cones (c) are located at the imaginary intersection of neighboring double cone partitions (white arrowheads), whereas corner cones face the double cone partitions. (B) Same area of retina as in (A) but at the nuclear level, where double cones form a square mosaic. A schematic of the mosaic unit, obtained by tracing several cones in (A), is presented to the right of (B). The same layout was used for the other photomicrograph pairs (C–H). (C, D) Full square mosaic from the centrodorsal retina at the double cone ellipsoid and nuclear level, respectively. (E, F) Full square mosaic from the centronasal retina at the double cone ellipsoid and nuclear level, respectively. (G) Square mosaic from the centroventral retina showing a partial loss of corner cones (black asterisks indicate locations with missing corner cones); a few corner cones remained (white asterisk). (H) Square mosaic from the distal ventral retina showing a complete absence of corner cones (black asterisks). In every sector of the retina, cones are smaller and more closely packed toward the distal (peripheral) retina. Scale bar, 10 μm.

Figure 2.

 

Micrographs of sections from the retina of alevin coho salmon after in situ hybridization with species-specific UV and blue (BL) riboprobes. (A, B) After 2 weeks of hormone vehicle exposure, all the single cones in the dorsal retina of the control, regardless of whether they were center or corner cones, were labeled with the UV riboprobe (A, white arrows), but none were labeled with the blue riboprobe (B, black arrowheads). (C, D) The majority (>97%) of single cones in the ventral retina of the control labeled with the UV riboprobe (C), but failed to label with the blue riboprobe (D). (E–H) After 2 weeks of thyroid hormone exposure; most of the single cones in the dorsal (E) and ventral (G) retina failed to label with the UV riboprobe (black arrowheads), but all labeled with the blue riboprobe (F, H, black arrows); note that there is a partial loss of corner cones in sections (G) and (H) (black asterisks). (I–L) Single cones in the retina of the thyroid hormone–exposed alevin that were left untreated for 4 weeks, labeled exclusively with the UV riboprobe (I, dorsal retina with a full complement of corner cones; K, ventral retina without corner cones). No cones in the dorsal (J) or ventral (L) retina labeled with the blue riboprobe. Double black arrowheads: rods (r) that occupied the space where corner cones used to be. NT on the side panel refers to nontreated (for a given number of weeks after exposure). Other symbols and nomenclature are as in Figure 1 . Scale bar, 10 μm.

Figure 3.

 

Micrographs of sections from the retina of coho salmon parr after in situ hybridization with the UV and blue riboprobes. (A, B) The dorsal retina of the control, after 2 weeks of vehicle exposure, shows single cones labeled with the UV (A) or the blue (B) riboprobe, although the UV riboprobe labeled primarily corner cones and the blue riboprobe labeled primarily center cones. (C, D) The lower half of the ventral retina in the control had no corner cones, and the center cones labeled exclusively with the blue riboprobe (D). (E–H) After thyroid hormone exposure for 2 weeks, all the single cones labeled with the blue riboprobe (F, dorsal retina; H, ventral retina) and all but the odd corner cone in the dorsal retina (E) failed to label with the UV riboprobe (E, G). (I, J) The dorsal retina of the control, 2 weeks after the end of vehicle exposure, showed labeling similar to that at the end of exposure (A, B); nonetheless, labeling by the UV riboprobe was further restricted to corner cones (I), whereas the blue riboprobe labeled most center and corner cones (J). (K–P) Two weeks after the end of thyroid hormone exposure, the highest numbers of UV-labeled single cones were found in the temporal retina (K), whereas other areas of the retina (e.g., the nasal, M) showed only the odd corner cone labeling with the UV riboprobe. Most of the single cones in both areas labeled with the blue riboprobe (L, N). (O–P) In the ventral retina, most corner cones disappeared, and the remaining center cones labeled exclusively with the blue riboprobe (P). Symbols and nomenclature are as in Figure 2 . Scale bar, 10 μm.

Figure 4.

 

Retinal maps (n = 5) of cone distributions in the retina of coho salmon alevin (A) and parr (C) and rainbow trout alevin (B). Histologic counts of the number of cones (single and double cones) were performed at each sampling time shown, whereas in situ hybridization counts with the UV and blue riboprobes were not performed at 4 (28 days) and 6 (42 days) weeks of initial exposure but were obtained at all other sampling times shown. Each retinal sector has three (or two) numbers arranged in a column. The first is the average number of double cones (in thousands per square millimeter), the second is the average ratio of double cones to single cones (d/s), and the third (when present) is the average percentage of single cones that labeled with the UV riboprobe. The absence of the latter statistic at 14 days and in postexposure times indicates a retinal sector in which all single cones were blue. In situ processed retinas are also color-coded into general categories corresponding to the percentage of single cones that labeled with the UV riboprobe as detailed in the legend at the bottom right. For each retina, the area within the red perimeter had substantial loss of corner cones (>30% loss, d/s ratio > 1.3). The scale bar (above the legend) in the coho salmon alevin represents 3.3 (14 days), 4 (28 days), 4.5 (42 days), and 5 (post 28 days) mm. In the rainbow trout, it represents 2.6 (14 days), 3.1 (28 days), 3.6 (42 days), and 4.2 (post 28 days) mm. In the coho salmon parr, the scale bar is 5.1 (14 days) and 5.5 (post 14 days) mm. Other symbols: D, dorsal; N, nasal.

Figure 5.

 

Single cone opsin transcripts and related visual pigments in the transitional retina of coho salmon. (A) Micrograph of a section from the dorsal retina of 2-week thyroid hormone–treated coho parr after double-label in situ hybridization with the UV and blue riboprobes. Shown are corner cones expressing UV opsin mRNA (red, white arrows), center cones expressing blue opsin mRNA (blue, black arrows), and corner cones expressing both UV and blue opsin mRNAs (purple, green arrows). (B) Single cones switching from UV to blue opsin in thyroid hormone–treated fish show a variety of absorbance profiles indicative of coexpression of opsins. The top profile in (B) was recorded from the midsection of a cone’s outer segment that showed a predominant UV visual pigment with λ_max ∼ 380 nm, and an absorbance “hump” around 455 nm (black arrow) indicative of blue visual pigment presence. The bottom part of the curve in (B) illustrates recordings obtained from a single cone in which the outer segment base had a blue visual pigment with λ_max ∼ 455 nm, whereas the tip showed a UV-blue visual pigment absorbance mixture peaking at ∼405 nm. For clarity, absorbance curves were displaced along the vertical axis by adding 0.02 units to successive curves. Symbols and nomenclature as in Figure 2 . Scale bar, 10 μm.

Figure 6.

 

Micrographs of sections from the retina of coho salmon smolt after in situ hybridization with the UV and blue riboprobes. (A–D) Neither the dorsal (A, B) nor the ventral (C, D) retina of the control, 2 weeks after hormone vehicle exposure, show any single cones labeled with the UV riboprobe (A, C), but all label with the blue riboprobe (B, D). (E–H) Single cones in the dorsal (E, F) and ventral (G, H) retina, after 2 weeks of thyroid hormone exposure, label exclusively with the blue riboprobe (F, H). Note that micrographs corresponding to ventral retina do not have corner cones. (I, J) Two weeks after cessation of vehicle treatment, all the single cones in the dorsal retina of the control label exclusively with the blue riboprobe (J). (K, L) Likewise, all the single cones in the dorsal retina of the thyroid hormone–treated fish, 2 weeks after the end of treatment, label exclusively with the blue riboprobe (L). (M, N) Micrographs of radial section from the centroventral retina of the smolt show exclusive perinuclear labeling of single cones with the blue riboprobe (N). d, double cone; n, nucleus; all other symbols and nomenclature as in Figure 2 . Scale bar, 10 μm.

Figure 7.

 

Micrographs of retinal wholemounts and sections from the light-adapted rainbow trout alevin after immunolocalization of BrdU alone (wholemounts) or in conjunction with opsin riboprobes (sections). (A–D) Micrographs of nasal (A, B) and temporal (C, D) regions of wholemounts show highest BrdU incorporation (double white arrowheads, fluorescence) at the periphery of the retina regardless of treatment (A, C; control; B, D; thyroid hormone treated). In addition, labeled cells are found vitread (lower focusing depth) to the cone photoreceptor layer (C). (E–H) Oblique sections of control fish retina showed that the BrdU labeled cells resided vitread to the cone layer and, in particular, to the inner segments of single cones expressing UV opsin mRNA (white arrowheads; E, F). BrdU-labeled cells were found in the ONL coincident with cells expressing rod opsin mRNA (double black arrowheads; G, H). In these photomicrographs and the pairs that follow (I–Y), the same structures are shown in bright-field (left) and fluorescence images (right). (I–P) Radial sections of control fish retina after 2 weeks of hormone vehicle exposure (I–L) showed the same pattern of BrdU labeling in the ONL as that in thyroid hormone–exposed fish (M–P). Fluorescent cells were located vitread to the single cones, which express UV (I, J) or blue (M, N) opsin mRNA in control and thyroid hormone–treated fish, respectively. Fluorescent cells were confined to a layer of prominent rod opsin mRNA expression (K, L, O, P). (Q–Y) Oblique sections illustrate that 1 week after the end of treatments, both control and post–thyroid hormone–exposed fish showed the same pattern of BrdU-labeled cells as described previously (E–P). Rod, labeling with rod opsin riboprobe; rpe, retinal pigment epithelium. Other symbols and nomenclature are as in Figure 2 . Scale bars: (A–D) 187 μm; (E–Y) 25 μm.

The authors thank the staff at the Capilano River and Lower Fraser River hatcheries for the fish, and Christina Gulbransen for help with the husbandry.