Systemic thyroid hormone is necessary and sufficient to induce ultraviolet-sensitive cone loss in the juvenile rainbow trout retina

The retina of salmonid fishes (trout and salmon) contains several analogous classes of cone photoreceptors [short wavelength-sensitive (SWS), medium wavelength-sensitive (MWS) and long wavelength-sensitive (LWS) cones] and comparable anatomical and functional organisation similar to that of humans (Kunz, 2006). Furthermore, wavelength discrimination studies in teleost fishes and humans suggest that colour vision in salmonid fishes is similar to that of humans (Spekreijse et al., 1991a; Spekreijse et al., 1991b; Wiersma and Spekreijse, 1991; Kitschmann and Neumeyer, 2005; Kunz, 2006; Joselevitch and Kamermans, 2009; Mora-Ferrer and Neumeyer, 2009). One additional and distinct cone type in these fish, the ultraviolet-sensitive (UVS) cone, contains the SWS1 opsin pigment that facilitates detection of polarised light and light in the ultraviolet spectrum (Hawryshyn et al., 1989; Hawryshyn and Hárosi, 1994; Hawryshyn, 2000; Parkyn and Hawryshyn, 2000; Allison et al., 2003). Interestingly, UVS cones are lost over much of the retina during smoltification, a developmental process involving profound changes in physiology, behaviour and morphology that prepare the freshwater juveniles (referred to as parr) for life in a dramatically different environment, the ocean (Hoar, 1988; Kunz et al., 1994; Parkyn and Hawryshyn, 2000). Smoltification occurs just prior to migration to the marine ecosystem and primarily occurs in anadromous salmonids, but land-locked species, such as rainbow trout (Oncorhynchus mykiss) undergo many similar changes, especially that of alterations to their visual system (Hawryshyn et al., 1989; Deutschlander et al., 2001; Veldhoen et al., 2006; Raine and Hawryshyn, 2009). During smoltification, SWS1 opsin expression is downregulated, UVS cones in much of the retina undergo programmed cell death and UV sensitivity is decreased (Browman and Hawryshyn, 1992; Browman and Hawryshyn, 1994a; Kunz et al., 1994; Deutschlander et al., 2001; Allison et al., 2003; Allison et al., 2006; Hawryshyn et al., 2003; Kunz, 2006; Veldhoen et al., 2006; Raine and Hawryshyn, 2009). Smoltification is preceded by an increase in plasma thyroxine (T4) (Hoar, 1988), and exogenous T4 treatment has been demonstrated to elicit precocial loss of UVS cones and body silvering (a morphological change associated with smoltification) in rainbow trout (Browman and Hawryshyn, 1992; Browman and Hawryshyn 1994a; Allison et al., 2006). These studies suggest that systemic thyroid hormone (TH) may induce retinal remodelling in juvenile rainbow trout.

T4 and triiodothyronine (T3) are the two main forms of TH circulating in the blood of teleost fish. T4 is generally found in the greatest concentration in the blood stream and is considered to be a prohormone for the more biologically active form of TH, T3 (Leatherland, 1994; Hulbert, 2000). Deiodination enzymes are important regulators of TH concentrations in the blood and provide either an increased supply of T3 or clearance of both T4 and T3 from the circulation (Eales and Brown, 1993; Leatherland, 1994; Köhrle, 1999; Bianco et al., 2002; Brown, 2005; St Germain et al., 2005; Nunez et al., 2008). Local tissue deiodination is considered an important mechanism for supplying T3 to the tissues of the nervous system, and T3 levels for a target cell can be increased or decreased locally by deiodination via these deiodinases (Darras et al., 1998; Nunez et al., 2008).

TH appears to play a major role in vertebrate growth and development, although the precise nature of its role in teleost fishes is not clear (Leatherland, 1994; Power et al., 2001; Furlow and Neff, 2006; Nunez et al., 2008). In mammals, TH is critical for proper development of the central nervous system, including the retina (Bernal et al., 2003; Ng et al., 2001; Harpavat and Cepko, 2003). The TH receptors (TR), TRα and TRβ, have both been detected in the retina of mice, chicken, the African clawed frog (Xenopus laevis) and winter flounder (Pleuornectes americanus), and have been shown to be involved in opsin expression and cone development (Sjöberg et al., 1992; Cossette and Drysdale, 2004; Mader and Cameron, 2006; Nunez. et al., 2008). TRβ in particular, has been found to play a dominant role in cone development in mice, and TRβ2 knockout mice lack M (green-sensitive) cones and have an altered distribution of S (UV-sensitive) cones across the retina relative to wild-type control mice (Ng et al., 2001; Nunez et al., 2008).

Furthermore, it has been demonstrated that deiodinase (D2) activity is present in the retina of killifish and rainbow trout and that this activity can be altered with T4 treatment (Orozco et al., 2000; Plate et al., 2002). Recently, exogenous T4 in rainbow trout parr was shown to upregulate retinal TRβ and downregulate D2 across the whole retina within two days of treatment, prior to significant downregulation of SWS1 expression (Raine and Hawryshyn, 2009). This study provides evidence that exogenous T4 may be directly involved in initiating SWS1 downregulation and subsequent UVS cone loss in juvenile rainbow trout.

However, while several lines of evidence implicate T4 in UVS cone loss, the precise role and mechanism of T4 in eliciting this developmental phenomenon is unresolved. It is still unclear if T4 alone is sufficient to induce UVS cone loss, as previous T4 exposure methods may have introduced confounding variables such as stress effects from daily handling. Cortisol and other stress-related factors can modulate circulating T3 levels in vertebrates (Kühn et al., 1998), which in this case could act in concert with T4 to trigger retinal remodelling. Furthermore, the role of endogenous T4 in natural UVS cone loss has not been examined. The present study utilises a slow-release implant technique to examine the role of exogenous and endogenous TH in retinal remodelling in juvenile rainbow trout. We demonstrate that exogenous T4 treatment is sufficient to induce SWS1 downregulation and UVS cone loss, and that blocking endogenous T4 with the anti-thyroid agent PTU delayed retinal remodelling during development, suggesting that endogenous T4 is necessary for this process.

Rainbow trout (Oncorhynchus mykiss Walbaum) parr were obtained from Rainbow Springs Trout Hatchery (Thamesford, ON, Canada). For all of the experiments, fish were held at a mean temperature of 15°C in constantly flowing aerated water. The 12 h:12 h light:dark cycle fluorescent lighting used in these experiments was spectrally comparable with Veldhoen et al. (Veldhoen et al., 2006). Fish were fed a low ration (1% g body mass day^–1) of a commercial trout food (Classic Sinking Feed, Martin Mills, Elmira, ON, Canada) to slow growth and prevent potential premature smoltification. Fish care and treatment were performed in accordance with the Canadian Council for Animal Care regulations and the Queen's University Animal Care Committee.

For the purposes of this study, the terms ‘parr’ and ‘smolt’, which are commonly used to represent pre- and post-migratory stages, respectively, of anadromous salmonid species, were used here to represent the comparable developmental stages in rainbow trout. A ‘true’ smolt undergoes all of the behavioural, physiological and morphological changes necessary for survival in the ocean whereas a land-locked species like rainbow trout undergoes a ‘pseudo’-smoltification process (see Hoar, 1988). However, because anadromous rainbow trout (steelhead trout) were not available for study, and because rainbow trout are a popular and easily available model organism that exhibit similar changes in their visual system to ‘true’ smolts (Allison et al., 2006; Veldhoen et al., 2006; Raine and Hawryshyn, 2009), the latter were used in this study.

A coconut oil implant was employed to expose rainbow trout parr to exogenous T4 and PTU. Use of the implant allows the fish to be held under optimal flow-through conditions in an effort to remove the potential effects of stress resulting from the previously used immersion method, where fish were held in static water and handled daily during water changes. The coconut oil implant method used in the current study was based on a method used previously for altering other blood hormone levels in salmonids (Flett and Leatherland, 1989; Vijayan and Leatherland, 1989; Holloway and Leatherland, 1997). This is the first time exogenous T4 and PTU have been delivered via a coconut oil implant. For this reason, a pilot study was employed to determine the optimal amount of coconut oil and T4 required to elevate plasma T4 levels to appropriate levels for the duration of the experiment (data not shown).

At the beginning of the current experiment, the rainbow trout parr were 8.7±0.7 g wet mass. The coconut oil implant consisted of 20 μl of coconut oil per gram of mean body mass of the fish. Forty fish were allocated to each of four aquaria. Each of these four tanks corresponded to one of the four experimental treatments: a control treatment with only the vehicle [dimethyl sulfoxide (DMSO)] in the implant; a T4 treatment (L-thyroxine sodium salt pentahydrate; Sigma-Aldrich, St Louis, MO, USA) utilising 3 μg T4 μl^–1 coconut oil; a PTU treatment (6-propyl-2-thiouracil; Sigma-Aldrich) of 3 μg PTU μl^–1 coconut oil; and a T4/PTU treatment implanted with 3 μg of T4 and 3 μg of PTU μl^–1 coconut oil. PTU prevents endogenous production of T4 and would therefore decrease systemic levels of T4 in the trout and serve as a contrary treatment to T4. The T4 implant was designed to produce blood plasma T4 levels similar to those obtained using the typical immersion treatment method utilising 300 μmol l^–1 T4 in solution. It was predicted that T4 treatment would induce premature smoltification and associated retinal changes whereas PTU treatment would prevent natural smoltification. The combined T4/PTU treatment was predicted to induce changes similar to those seen with T4 alone, as exogenous T4 delivery should not be affected by PTU.

Complete sampling from each group occurred after four weeks of treatment, at which point experiments with T4 and T4/PTU implants were terminated. Only the control and PTU-implanted fish were retained past the four-week sampling point in order to examine the effects of PTU exposure on natural smoltification. The ration level was increased to 4% g body mass day^–1 after the four-week sampling point to allow rapid growth and development of the fish in the two remaining treatments. Fish were visually monitored for changes in size, body colouring and rounding of the rostrum indicative of natural pseudo-smoltification, which occurred by the ninth week of treatment. The experiment was terminated at this point.

Fish were dark adapted prior to euthanasia with 150 mg l^–1 tricaine methanesulfonate (MS-222; Crescent Research Chemical, Phoenix, AZ, USA). Following euthanasia, wet mass and fork length (tip of snout to fork in tail) measurements were immediately recorded. Blood was then quickly collected after transverse severance of the tail, slightly caudal to the anal fin. Heparinised microcapillary tubes (Fisher Scientific, Ottawa, ON, Canada) were used to collect blood from the caudal blood capillaries. Following cervical transection, the eyes were quickly removed and the neural retina was subsequently dissected free of the retinal pigment epithelium from both eyes under deep red illumination. The isolated retina from the right eye (N=5 fish per treatment) was placed in 4% paraformaldehyde to preserve the tissue for assessment of UVS cone presence using immunohistochemistry on the whole retina. The left isolated retina (N=5 fish per treatment) was immediately placed in RNAlater (Ambion Inc., Austin, TX, USA) for gene expression analysis. The remaining fish in each group were used for blood collection and length and mass measurements.

The blood-containing hematocrit tubes were removed from ice and centrifuged at 1500 × gravity at 4°C to separate the blood plasma from the cellular components. 100 μl of plasma was pooled from 2–3 fish to generate five replicates for each treatment. This was the minimum volume of plasma required for T4 analysis performed by a diagnostic lab using a standard radioimmunoassay procedure validated with spiked trout plasma (Texas Vet Med Diagnostic Laboratory, College Station, TX, USA). Plasma samples were stored at –20°C until T4 analysis.

Fixed whole retinas were immunostained with a mouse polyclonal antibody to the N-terminal portion of the rainbow trout SWS1 opsin (Allison et al., 2006). All solutions were prepared in phosphate buffer (PB, pH 7.4), and all steps were performed at room temperature unless otherwise noted. Retinas were incubated with PB containing 5% normal goat serum and 0.3% Triton-X (both from Sigma-Aldrich) to reduce non-specific binding. Retinas were incubated overnight at 4°C with the primary SWS1 opsin antibody diluted 1:5000 in PB containing 1% goat serum and 0.3% Triton-X. The tissue was rinsed in fresh PB and incubated for one hour in the secondary antibody (Alexa Fluor 594 goat anti-mouse, Invitrogen, Carlsbad, CA, USA) diluted 1:750 in 1% goat serum with 0.3% Triton-X. The tissue was then rinsed with fresh PB and counterstained with Alexa Fluor 488 peanut lectin-agglutinin (PNA; Invitrogen). PNA has been found to non-specifically label cones in many vertebrate species, including rainbow trout, and was used as a positive control (Blanks and Johnson, 1984; Allison et al., 2006). The immunostained whole retinas were mounted photoreceptor-side up with Prolong Gold mounting medium (Invitrogen) on FisherFrost Plus histology slides with a cover slip.

Immunostained retinas were visualised with a Zeiss Axioskop 2 compound microscope (Carl Zeiss Canada Ltd, Toronto, ON, Canada) equipped for epifluorescence and photographed with a Retiga EXi 12-bit CCD cooled monochromatic digital camera (QImaging, Surrey, BC, Canada). Images were captured using Northern Eclipse 7.0 (Empix Imaging, Mississauga, ON, Canada) software. Monochromatic images were pseudocoloured using ImageJ 1.40g (W.S. Rasband, National Institutes of Health, Bethesda, MD, USA) according to the fluorophore used (for example, tissue labelled with Alexa 594 secondary antibody was pseudocoloured red). Images were cropped and adjusted for brightness and contrast using Adobe PhotoShop 7.0 (Adobe Systems, Mountain View, CA, USA) but image content was not altered.

Each retina was homogenised individually with a Retsch MM301 mixer mill (Retsch GmBH & Co. Ltd KG, Hann, Germany). Total RNA was isolated from RNAlater-protected retinas using an Absolutely RNA miniprep kit (Stratagene, La Jolla, CA, USA). Quantification of total RNA isolations was performed using RiboGreen (Invitrogen Canada Inc., Burlington, ON, Canada) following the manufacturer's protocol and using an MX3000P real-time quantitative PCR thermocycler (Stratagene). Briefly, master mixes were generated containing TE buffer, Ribogreen and known concentrations of RNA to generate a standard curve using duplicate 25 μl volumes. A mastermix for each unknown was generated from 1 μl of each unknown, Ribogreen and TE buffer to run 25 μl per well in triplicate. The fluorescence signal generated from the RNA standard curve was used to assess the concentrations of the unknown samples. 1 μg total RNA of each of the retinal RNA isolates was then reverse-transcribed to cDNA using a Superscript III First Strand Synthesis SuperMix kit for qRT-PCR (Invitrogen).

Quantitative RT-PCR (qRT-PCR) analysis of individual total RNA samples from each retina was performed for each of the O. mykiss gene specific primers designed for SWS1, SWS2, TRα, TRβ, D2 and a housekeeping gene, ribosomal protein L8 (L8) (Table 1). All primer pairs were validated in previous studies (Raine et al., 2004; Veldhoen et al., 2006; Raine and Hawryshyn, 2009). L8 has been demonstrated to be effective as a house-keeping gene, as it does not vary significantly during TH treatment and development in rainbow trout or anurans (Shi and Liang, 1994; Callery and Elinson, 2000; Crump et al., 2002; Veldhoen et al., 2006; Raine and Hawryshyn, 2009). Primer optimisation reactions were run to determine the optimal concentration of primers to use in the qRT-PCR reactions; the resulting optimal equimolar amounts of primer varied between 100 and 300 μmol l^–1 (Raine and Hawryshyn, 2009). Standard curves were run for each primer pair using serial dilutions of rainbow trout parr retinal cDNA. The magnesium chloride concentration in the standard curve reactions was altered between 3 and 5 mmol l^–1 as needed, to generate repeatable R² values greater than 0.98 and reaction efficiencies between 90 and 100% (Raine and Hawryshyn, 2009).

Standard curves were run in triplicate and unknowns were run in duplicate. Master mixes were used whenever possible to reduce the possibility of pipetting error. Each 10 μl reaction contained Supermix-UDG (Invitrogen), ROX and 2 μl of 20-fold diluted retinal cDNA. The thermocycling program was 50°C for 2 min, 95°C for 2 min followed by 40 cycles of 15 s at 95°C, 30 s at 60°C and 30 s at 72°C. A dissociation curve was generated at the conclusion of each qRT-PCR run to ensure that all products generated during the thermocycling program produced only one distinct DNA peak that dissociated at a consistent temperature. DNA amplification was performed using a MX3000P real-time quantitative PCR thermocycler (Stratagene), and the resultant amplification curves and data analysis were performed by the accompanying MXPro v. 4.0 software using the efficiency-corrected comparative quantitation method (Pfaffl, 2001). A relative method of qRT-PCR quantification was determined to be most relevant for analysis of the experimental data, because changes relative to the control fish were the target of the investigation and absolute copy numbers of RNA were not required for this purpose.

Statistical analysis was performed on all data using SigmaPlot 11 software (Systat Software Inc., San Jose, CA, USA). One-way analysis of variance (ANOVA) was used to compare data from all treatment groups for each method of analysis. The Holm–Sidak method was subsequently employed to detect differences between the treatment groups. A probability level of 0.05 was used in all cases to indicate significance. All data were expressed as means ± standard error. All qRT-PCR data were expressed as means ± standard error relative to control retina.

After four weeks of treatment, control fish retained parr bars and a sloped rostrum, while T4-treated fish exhibited body silvering and a rounded rostrum indicative of smoltification (Fig. 1). As predicted, PTU-treated fish were similar in external morphology to the control fish. T4/PTU-treated fish displayed morphological changes comparable with those of the T4-treated group, complete with body silvering and rounded snouts (Fig. 1). No significant differences in wet mass or fork length were observed between groups after four weeks of treatment (Table 2).

While the control and PTU-treated fish appeared similar after four weeks of treatment, the two groups were morphologically distinct by week nine of the experiment (Fig. 1). After nine weeks the control fish appeared to have gone through the metamorphic process of smoltification, displaying the external morphology of a juvenile rainbow trout, including rounded rostrum, faded parr bars and multi-coloured, iridescent scales similar to that of adult fish. The PTU-treated fish, which retained their parr bars, exhibited more morphological similarity to that of the control fish at the beginning of the study (Fig. 1). The fork length and wet mass of PTU-treated fish after nine weeks was also significantly less than that of the controls (Table 2).

Plasma T4 levels were significantly elevated in fish treated for four weeks with T4 or T4/PTU implants as compared with control and PTU-treated fish (Fig. 2). Control and PTU-treated fish exhibited low plasma T4 levels at 0.46±0.28 ng ml^–1 and 0.23±0.23 ng ml^–1, respectively, while T4- and T4/PTU-treated fish had significantly higher plasma T4 values at 269.33±46.61 ng ml^–1 and 60.80±29.35 ng ml^–1, respectively. Interestingly, the plasma T4 levels from fish in the T4/PTU-treatment group were dramatically lower than those that received only T4 treatment (P<0.05).

After nine weeks of treatment, control fish had low plasma T4 levels at 0.1 ng ml^–1, and in the PTU-treatment group plasma T4 levels were below the detection threshold (Fig. 2).

After four weeks of treatment, SWS1 expression significantly decreased in both the T4- and T4/PTU-implanted fish relative to implant controls (Fig. 3, P<0.05). PTU-treated fish showed modest upregulation of SWS1 at both the 4-week and 9-week time points when compared with the control fish but this trend was not significant at either time point (Fig. 3, P=0.103 at four weeks and P=0.208 at nine weeks). When the individual qRT-PCR data for each fish were compared with the corresponding fish size and UVS cone immunostaining results, a pattern emerged, suggesting that SWS1 upregulation in PTU-treated fish may be dependent on fish size; smaller fish had higher SWS1 levels (Table 3).

As some previous studies have suggested that SWS1 downregulation coincides with SWS2 upregulation in the rainbow trout retina (Cheng et al., 2006; Cheng and Flamarique, 2007), relative transcript levels for SWS2 (blue) opsin were also examined. SWS2 did not exhibit any significant changes in transcript level over the course of the experiment in any treatment group or time point examined (Fig. 3, P=0.224).

The relative expression levels of the TRα, TRβ and D2 transcripts were quantified and compared with time-matched control fish (Fig. 3). There were no significant differences in TRα expression levels across treatment groups or time points. However, fish in all three groups (T4, PTU, T4/PTU) treated for four weeks demonstrated a significant elevation in TRβ transcripts relative to time-matched control fish.

There were no changes in TRβ mRNA levels in the PTU-treated fish after nine weeks of treatment when compared with time-matched controls. The level of D2 mRNA was significantly lower in both the T4 and T4/PTU treatments relative to the control fish at four weeks post-implant. Although there did appear to be an increasing trend in the level of transcript for D2 in the PTU-treated group after four weeks when compared with the control fish, this trend was not statistically significant. Both control and PTU-treated fish displayed similar D2 expression levels after nine weeks of treatment.

Immunostaining of whole retina with anti-SWS1 opsin demonstrated that UVS cones were present in the dorsal retina of both the control and PTU-treated fish, while both the T4- and T4/PTU-treated fish demonstrated UVS cone loss in the dorsal retina at the conclusion of the 4-week treatment period (Fig. 4). UVS cones were also missing from the ventral retina of T4 and T4/PTU-treated fish by four weeks (data not shown). The ventral retina of all five of the control fish was also devoid of UVS cones by week four of the experiment (Fig. 5). This suggests that naturally occurring UVS cone loss prior to smoltification begins in the ventral hemisphere. By contrast, the UVS cone population was retained in the ventral retina of two out of the five fish treated for four weeks with PTU (Fig. 5). A comparison of individual fish sizes, SWS1 opsin gene expression and UVS cone location in the retina demonstrated that smaller PTU-treated fish (under 10 g wet mass) exhibited higher SWS1 expression levels and more ventral UVS cone retention than larger PTU-treated fish (Table 3).

Control and PTU-treated fish exhibited greater differences in UVS cone distribution across the retina after nine weeks of treatment, when control fish resembled smolts. The control fish possessed immunolabelled UVS cones only in the dorsal-temporal quadrant of the retina, while the PTU-treated fish displayed additional ventral and dorso-nasal labelling (Fig. 6).

The current study demonstrates that systemic T4 is necessary and sufficient to induce UVS cone loss in the juvenile rainbow trout retina. Exogenous T4, administered via slow-release implant, triggered precocious smoltification, including SWS1 downregulation and UVS cone loss. The necessity of endogenous T4 in naturally occurring UVS cone loss was revealed using exogenous treatment with the anti-thyroid agent PTU. Decreased T4 production following PTU treatment attenuated natural SWS1 opsin downregulation and UVS cone loss in the retina of rainbow trout smolt. This is the first study to show that T4 is involved in both induced and natural SWS1 downregulation and UVS cone loss in rainbow trout. Finally, qRT-PCR analysis demonstrates that several components of TH regulation are present in the rainbow trout retina and that expression of these components changes with T4 treatment, suggesting that TH may act directly on the retina to induce UVS cone loss during smoltification.

The T4-induced downregulation of SWS1 and corresponding UVS cone loss seen in the present study is in keeping with several prior studies in juvenile rainbow trout (Browman and Hawryshyn, 1992; Browman and Hawryshyn, 1994a; Deutschlander et al., 2001; Allison et al., 2006; Veldhoen et al., 2006; Raine and Hawryshyn, 2009). However, previous studies involved the immersion of fish in static solutions containing exogenous T4. This method requires daily handling of the fish in order to ensure the removal of accumulated biological waste products and to provide a relatively constant supply of hormone. Increases in circulating cortisol levels associated with stress, including handling stress, have been found to interact with TH in fishes (Kühn et al., 1998). Thus, there is the potential for cortisol to be generated using this treatment method. This raises the possibility that cortisol is necessary in combination with exogenous T4 treatment to induce precocious UVS cone loss. The use of a coconut oil implant to provide long-term exposure of rainbow trout parr to exogenous T4 in this study reduced the potentially confounding effects of stress inherent in the previously utilised T4-treatment method, allowing the fish to be held under optimal conditions with minimal researcher-induced stress. This provides strong evidence that exogenous T4 alone induces UVS cone loss in rainbow trout.

SWS1 downregulation and UVS cone loss were attenuated in PTU-treated fish after both four and nine weeks of treatment, demonstrating that endogenous TH is necessary for these events. Treatment with PTU consistently prevents TH synthesis and significantly decreases plasma TH levels in all teleost species examined (for reviews, see Leatherland, 1994; Cyr and Eales, 1996). By nine weeks, the effects of PTU treatment were demonstrated by undetectable levels of plasma T4, lack of change in body colouration and decreased growth when compared with the control fish. Differences between retinas of control and PTU-treated fish were most evident at the nine-week time point. By this time, the PTU treatment had most likely prevented the endogenous production of TH associated with natural smoltification; these fish were smaller and retained their parr marks, while control fish were larger, with colouration resembling adult fish. In other studies, treatment of salmonids with 25 mg l^–1 (Sullivan et al., 1987) or 30 mg l^–1 (Ebbesson et al., 1998) PTU has been shown to reduce plasma total T4, body growth, food conversion efficiency, pigment changes and the ability to osmoregulate in seawater. Ebbesson et al. found a decrease in plasma total T3 levels and a corresponding decrease in plasma growth hormone (GH) with PTU treatment (Ebbesson et al., 1998). The results of these studies provide support for the lack of both body growth and changes in body pigment observed in the nine-week PTU-treated trout in the current study. This also provides supporting evidence that a sufficient systemic dose of PTU was released from the implant in the PTU-treated trout to inhibit T4 production.

While there was a trend toward a higher level of SWS1 expression in the PTU-treated fish relative to the control fish at both sampling times, this trend was not statistically significant. In an effort to illuminate results that are slightly obscured by the high variation in SWS1 opsin gene expression in the four- and nine-week PTU-treated groups, results for each individual fish were collated based on body size, SWS1 opsin expression relative to the control parr and UVS cone distribution (Table 3). These results demonstrate that smaller fish were most likely to exhibit higher expression levels of SWS1 and a greater distribution of UVS cones across their retina. At the start of the experiment, some ventral UVS cone loss was evident in some of the same population of parr (data not shown), and the onset of UVS cone loss has been suggested to occur early in the parr stage (Cheng et al., 2007). PTU treatment was not expected to trigger regeneration of UVS cones already lost via natural processes but merely to arrest additional endogenous T4-induced loss after the initiation of PTU treatment. Therefore, early SWS1 downregulation and UVS cone loss in the experimental fish may have partially masked effects of the PTU treatment. Long-term PTU treatment begun with much younger and smaller parr (<5 g) would be necessary to substantiate this hypothesis.

Fish treated with both T4 and PTU resembled T4-treated fish in all measurements examined, including body morphology, SWS1 expression levels and UVS cone loss. These results were expected, as PTU acts directly on the thyroid tissue to inhibit endogenous T4 synthesis but PTU was not expected to affect the high circulating T4 levels resulting from exogenous administration. Interestingly, plasma T4 levels were noticeably lower in T4/PTU-treated fish than in the T4-only fish, although the reason for this is not clear. It is tempting to interpret the lower plasma T4 level in the T4/PTU fish as a result of the PTU treatment but there is no direct evidence to support this. Nevertheless, the exogenous T4 level administered in these T4/PTU fish was sufficient to induce precocious smoltification and retinal remodelling.

The present study demonstrates that manipulation of systemic T4 levels in rainbow trout leads to changes in retinal gene expression of the TH-regulating components examined. Specifically, expression of the nuclear receptor TRβ was significantly upregulated in fish treated with T4, PTU and T4/PTU implants for four weeks, prior to complete natural developmental loss of UVS cones from most of the central retina. D2 expression was downregulated following both T4 and T4/PTU treatments, while expression levels in PTU-treated fish were not different from controls. Alternately, TRα expression levels did not differ between any of the treatment groups, and there were no differences in gene expression of the TRs or D2 between PTU-treated and control fish after nine-weeks of treatment, when complete natural developmental loss of UVS cones from most of the central retina was completed.

Several previous studies have demonstrated TR expression in the vertebrate retina (Sjöberg et al., 1992; Harpavat and Cepko, 2003; Mader and Cameron, 2006; Applebury et al., 2007; Ng et al., 2009; Raine and Hawryshyn, 2009). Both TRα and TRβ are expressed in the retina of another fish, the winter flounder (Pleuronectes americanus), in addition to rainbow trout (Mader and Cameron, 2006; Raine and Hawryshyn, 2009). Upregulation of retinal TRα gene expression has been shown to accompany TU treatment (an anti-thyroid agent similar to PTU) using the whole fish immersion method in post-metamorphic flounder (Mader and Cameron, 2006). Conversely, changes in TRα were not apparent in the present study. However, this may be a result of several differences between the two studies, including treatment delivery method, PTU/TU concentration, species and developmental stage examined. In winter flounder, the increase in TRα gene expression did not correspond to an increase in levels of the corresponding protein, suggesting that functional TRα may be regulated by post-transcriptional mechanisms and not involved in retinal changes accompanying flounder metamorphosis (Mader and Cameron, 2006). Interestingly, Mader and Cameron did note an increase in TRβ expression in the retina of post-metamorphic flounder compared with the pre-metamorphic condition (Mader and Cameron, 2006). As flounder metamorphosis is a TH-driven process, this latter finding suggests a role for TRβ in the developing flounder retina.

TH receptors are expressed in the developing mouse retina and multiple studies in mouse models of TH disruption provide compelling evidence that TRβ2 (a mammalian splice variant of TRβ) regulates opsin expression during mammalian retinal development (Ng et al., 2001; Roberts et al., 2006; Applebury et al., 2007; Pessôa et al., 2008; Lu et al., 2009). Mouse cone photoreceptors express one of two opsins, a short-wavelength-sensitive S-opsin and a middle-wavelength-sensitive M-opsin (Wikler et al., 1996). Endogenous T3 binding to TRβ2 is necessary to inhibit S-opsin expression and turn on M-opsin expression in developing mouse retina in vivo (Roberts et al., 2006; Pessôa et al., 2008). Mouse S-opsin belongs to the SWS1 opsin subfamily (Yokoyama and Yokoyama, 1996) and suggests that TH regulation of SWS1 via TRβ may be an evolutionarily conserved vertebrate feature.

Previous studies postulated a correlation between SWS1 downregulation and SWS2 opsin upregulation in the retina of juvenile salmonids, and suggested that UVS cones do not degenerate but instead transform into blue cones (Cheng et al., 2006; Cheng et al., 2009; Cheng and Flamarique, 2007). Although this phenomenon was not specifically investigated in this study, the expression of SWS2 opsin at the four- and nine-week sampling periods did not negatively co-vary with SWS1 expression. In the present study, SWS2 was always slightly but not significantly upregulated after four weeks of treatment, regardless of whether relative SWS1 expression levels were reduced (T4, T4/PTU treatment) or increased (PTU treatment). Therefore, the current data do not support the hypothesis of negative co-regulation between SWS1 and SWS2. The present study, however, examined transcript levels across the whole retina and this approach could mask any potential regional variation. Thus, although this study does not provide any evidence to support the claim that SWS1 and SWS2 opsins are co-expressed, neither do the results provide any evidence to unequivocally dispute the claim.

The present study demonstrates that exogenous T4 alone is sufficient to induce UVS cone loss in the juvenile rainbow trout retina, and that endogenous T4 is necessary to trigger natural SWS1 downregulation and corresponding UVS cone loss. The TH-regulating TRs and D2 are expressed in the trout retina and their expression changes with systemic T4 manipulation. This is the first study to provide evidence for a role of endogenous T4 in natural UVS cone loss in salmonid fishes.

The authors would like to thank Dr Duncan McKenzie for facilitating access to the Texas Vet Med Diagnostic Laboratory, College Station, Texas.

 
• D1–D3

  deiodinase 1–3

 
• DN

  dorsal nasal

 
• DT

  dorsal ventral

 
• PB

  phosphate buffer

 
• PNA

  peanut lectin-agglutinin

 
• PTU

  propylthiouracil

 
• T3

  triiodothyronine

 
• T4

  thyroxine

 
• TH

  thyroid hormone

 
• TR

  thyroid hormone receptor

 
• TRα

  thyroid hormone receptor isoform alpha

 
• TRβ

  thyroid hormone receptor isoform beta

 
• UVS

  ultraviolet-sensitive