Targeting iodothyronine deiodinases locally in the retina is a therapeutic strategy for retinal degeneration

Abstract

Recent studies have implicated thyroid hormone (TH) signaling in cone photoreceptor viability. Using mouse models of retinal degeneration, we found that antithyroid treatment preserves cones. This work investigates the significance of targeting intracellular TH components locally in the retina. The cellular TH level is mainly regulated by deiodinase iodothyronine (DIO)-2 and -3. DIO2 converts thyroxine (T4) to triiodothyronine (T3), which binds to the TH receptor, whereas DIO3 degrades T3 and T4. We examined cone survival after overexpression of DIO3 and inhibition of DIO2 and demonstrated the benefits of these manipulations. Subretinal delivery of AAV5-IRBP/GNAT2-DIO3, which directs expression of human DIO3 specifically in cones, increased cone density by 30–40% in a Rpe65^−/− mouse model of Lebers congenital amaurosis (LCA) and in a Cpfl1 mouse with Pde6c defect model of achromatopsia, compared with their respective untreated controls. Intravitreal and topical delivery of the DIO2 inhibitor iopanoic acid also significantly improved cone survival in the LCA model mice. Moreover, the expression levels of DIO2 and Slc16a2 were significantly higher in the diseased retinas, suggesting locally elevated TH signaling. We show that targeting DIOs protects cones, and intracellular inhibition of TH components locally in the retina may represent a novel strategy for retinal degeneration management.—Yang, F., Ma, H., Belcher, J., Butler, M. R., Redmond, T. M., Boye, S. L., Hauswirth, W. W., Ding, X.-Q. Targeting iodothyronine deiodinases locally in the retina is a therapeutic strategy for retinal degeneration.

Rod and cone photoreceptors degenerate in a wide array of hereditary retinal diseases, such as retinitis pigmentosa and cone–rod dystrophies and age-related macular degeneration. Inherited retinal degenerative diseases affect ∼1 in 3000 individuals worldwide. Age-related macular degeneration is the leading cause of blindness among the aged population. Defects in numerous genes and proteins are linked to retinal degenerative disorders, including those involved in the visual cycle, such as retinal pigment epithelium–specific RPE65 retinol isomerase and lecithin retinol acyltransferase and those involved in phototransduction, such as opsins, subunits of transducin, cGMP phosphodiesterase (PDE)-6, and cyclic nucleotide-gated (CNG) channels. Despite a remarkable heterogeneity of disease-associated genes and pathogenesis, the progressive death of cone photoreceptors ultimately leads to vision impairment and blindness. Light responses mediated by cone photoreceptors are essential for daylight vision, color vision, and visual acuity. There is currently no treatment available for retinal degeneration. Despite wide genetic heterogeneity, the degenerating photoreceptors show common cellular disorder features, including oxidative damage (1, 2), endoplasmic reticulum stress (3, 4), and apoptosis/necrosis (5, 6). These features offer the possibility of targeting cell survival and death regulators and pathways to reduce photoreceptor death, regardless of the genetic origins of the diseases.

Thyroid hormone (TH) signaling regulates numerous physiologic functions, including cell growth, differentiation, and metabolic homeostasis (7, 8). In the retina, TH signaling is well known for its regulation in cone opsin expression; it suppresses expression of short-wave–sensitive opsin 1 (S-opsin) and induces expression of medium-wave–sensitive opsin 1 (M-opsin) (9). Recent studies have implied that TH signaling has a role in cone photoreceptor viability. Using mouse models of retinal degeneration, we found that antithyroid treatment preserves cones. Antithyroid treatment significantly improved cone survival in Rpe65^−/− mice, a model of the pediatric blinding disease Lebers congenital amaurosis (LCA), and Cpfl1 mice with defect in Pde6c, a model of achromatopsia (10). In contrast to the findings from antithyroid treatment, excessive TH signaling induced by treatment with triiodothyronine (T3) or deletion of the T3 degrading enzyme iodothyronine deiodinase (DIO)-3 has been shown to cause cone death (10, 11). This effect was reversed by deletion of TH receptor (11).

In mammals, the thyroid gland predominantly produces the prohormone thyroxine (T4; ∼95%), along with a small amount of the bioactive hormone T3 (∼5%). The metabolism of T4 and T3 is catalyzed by DIO1–3. DIO1/2 convert T4 to T3 by an outer ring deiodination reaction. DIO1 is primarily involved in the conversion of T4 to T3 in the circulation (12). In the peripheral tissues, T4 and T3 are transported to cells, where T4 is converted to T3 by DIO2, and T3 is then transferred to the nucleus and binds to TR, initiating downstream gene expression responses. Intracellular T4 and T3 are degraded by DIO3 to reverse T3 (rT3) and 3,5-diiodo-l-thyronine (T2), respectively (see Fig. 1 for the function of DIO2 and -3). Although thyroid gland function and circulating hormone levels are essential for normal TH function in cells, intracellular TH homeostasis is a highly locally regulated process, controlled by the DIOs (13, 14). In this work, we investigated the effectiveness of targeting DIO locally in the retina to explore the potential of future clinical applications (Fig. 1).

Figure 1.

Targeting DIOs to suppress TH signaling. Circulating T4 and T3 are transported into cells, where T4 is converted to T3 by DIO2. T3 is then transferred to the nucleus and binds to TRs, initiating the downstream responses, and T4 and T3 are metabolized to inactive forms T2 and reverse T3 (rT3) by DIO3. Suppression of intracellular TH signaling can be achieved by overexpressing DIO3 and inhibiting DIO2. TR, TH receptor; RXR, the retinoid X receptor; TRE, thyroid hormone response element.

MATERIALS AND METHODS

Mice, antibodies, and other reagents

The Rpe65^−/− mouse line was generated as has been described (15). The Cpfl1 mouse line was obtained from The Jackson Laboratory (Bar Harbor, ME, USA). The Nrl^−/− mouse line was provided by Dr. Anand Swaroop (National Eye Institute, National Institutes of Health, Bethesda, MD, USA) (16). The Rpe65^−/−/Nrl^−/− (17) and Cngb3^−/−/Nrl^−/− (18) lines were generated by cross-mating. All mice were maintained under cyclic 12-h light–dark conditions. Cage illumination was ∼7 foot-candles (fc) during the light cycle. All animal maintenance and experiments were approved by the local Institutional Animal Care and Use Committee (University of Oklahoma Health Sciences Center) and conformed to the guidelines on the care and use of animals adopted by the Society for Neuroscience and the Association for Research in Vision and Ophthalmology (Rockville, MD, USA).

Rabbit antibodies against mouse M-opsin and cone arrestin (CAR) were provided by Dr. Cheryl Craft (University of Southern California, Los Angeles, CA, USA). Rabbit anti-S-opsin antibody was provided by Dr. Muna Naash (University of Houston, Houston, TX, USA). Mouse anti-rhodopsin antibody 1D4 was provided by Dr. Robert Molday (The University of British Columbia, Vancouver, BC, Canada). Biotinylated peanut agglutinin (PNA) was purchased from Vector Laboratories, Inc. (Burlingame, CA, USA). Rabbit anti-DIO2 antibody was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Rabbit anti-DIO3 antibody was procured from Abcam (Cambridge, MD, USA). Mouse anti-β-actin was obtained from Abcam. Horseradish peroxidase (HRP)-conjugated anti-rabbit or anti-mouse secondary antibodies were purchased from Kirkegaard & Perry Laboratories, Inc. (Gaithersburg, MD, USA). Fluorescent goat anti-rabbit and goat anti-mouse antibodies and streptavidin-Cy3 were procured from Thermo Fisher Scientific (Waltham, MA, USA). Iopanoic acid (IOP) was purchased from TCI Chemicals America (Portland, OR, USA). T4 was purchased from Sigma-Aldrich (St. Louis, MO, USA). Phospholipon 90G was provided by Dr. Bruce Baretz (Lipoid LLC, Newark, NJ, USA). All other reagents were purchased from Sigma-Aldrich, Bio-Rad Laboratories (Hercules, CA, USA), and Thermo Fisher Scientific.

Construction of AAV5-IRBP/GNAT2-DIO3 vectors

The pCDM8-DIO3-WT with DIO3 engineered in a pCDM8 (19) was kindly provided by Dr. P. Reed Larsen at Brigham and Women’s Hospital (Boston, MA, USA). The pCDM8-DIO3-Cys was generated by replacing codons TGA for selenocysteine (Sec)¹⁴⁴ with codons TGC for cysteine (Cys) by using a QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA). Sequences were verified by direct sequencing. The cDNAs were cloned into an AAV vector plasmid containing the chimeric cone-specific IRBP/GNAT2 promoter, which has been shown to direct expression of transgene specifically in cones (20). The vector constructs were subsequently packaged into AAV serotype-5 by plasmid transfection into HEK293T cells by described methods. The resulting AAV5-IRBP/GNAT2-DIO3-WT and AAV5-IRBP/GNAT2-DIO3-Cys were purified and titered (21).

Ocular injection of IOP and AAV-DIO3 vector

The ocular injections were performed under an OPMI VISU 140 surgical operating microscope (Zeiss, Thornwood, NY, USA). Briefly, after anesthesia on ice for 2–3 min and complete dilation with 1% cyclopentolate hydrochloride (Akorn, Lake Forest, IL, USA), a drop of 2.5% methylcellulose was added to the corneal surface to visualize the fundus. A 28-gauge beveled hypodermic needle (BD Biosciences, Franklin Lakes, NJ, USA) was used to puncture the cornea carefully. One microliter of IOP (in DMSO) or AAV-DIO3 vector [in balanced salt solution (BSS); Alcon, Fort Worth, TX, USA] was injected intravitreally or subretinally into 1 eye of each mouse with a NanoFil microsyringe injector system with a 33-gauge blunt needle (Hamilton Co., Reno, NV, USA). The contralateral eye was injected with 1 µl of vehicle. Only eyes with no apparent surgical complications were retained for further evaluation. Table 1 shows information including concentration/titer of the reagents delivered for ocular injection experiments.

TABLE 1.

 Information for ocular delivery experiments

Reagents	Concentration or titer/volume	Vehicle	Mouse line	Delivery route	Evaluation
IOP	50 mM/1.0 μl	DMSO	Rpe65−/−	Intravitreal injection	Rpe65−/− and Rpe65−/−/Nrl−/− mice received injections at P5 and were analyzed for cone density at P25 and cone death at P15, respectively.
			Rpe65−/−/Nrl−/−
IOP	1.0%/10 μl (3×/d)	PBS containing 25% phospholipon 90G	Rpe65−/−	Eye drops	Rpe65−/− mice received eye drops for 20 d, beginning on P5, and were analyzed for retinal function and cone density at the end of the experiment.
AAV5-IRBP/Gnat2-DIO3-WT	1 × 1010 vector genomes/1.0 μl	PBS	Rpe65−/−	Subretinal injection	Mice received injections at P5 and were analyzed for cone density and cell death at P30.
			Cpfl1
AAV5-IRBP/Gnat2-DIO3-Cys	1 × 1010 vector genomes/1.0 μl	PBS	Rpe65−/−	Subretinal injection	Mice received injections at P5 and were analyzed for cone density and cell death at P30.
			Cpfl1

Topical administration of IOP

The IOP eye drops (1.0% IOP in PBS containing 25% phospholipon 90G) was formulated as described by Scheppke et al. (22). Postnatal day (P)5 Rpe65^−/− mice received IOP eye drops or vehicle (25% phospholipon 90G in PBS) (∼10 μl) 3×/d for 20 d. Eyes were collected at the end of the experiments for evaluation of cone density and TUNEL⁺ cells.

Eye preparation, immunofluorescence labeling, and confocal microscopy

Mouse retinal whole mounts or cross-sections were prepared, and immunofluorescence labeling using antibodies against cone markers was performed (23). For whole-mount preparations, eyes were enucleated, marked at the superior (dorsal) pole with a green dye, for dorsal–ventral orientation, and fixed in 4% paraformaldehyde (Polysciences, Inc., Warrington, PA, USA) for 30 min at room temperature. For retinal cross-sections, mouse eyes were enucleated (the superior portion of the cornea was marked for orientation before enucleation) and fixed in Prefer (Anatech Ltd., Battle Creek, MI, USA) for 15 min at room temperature, and fixed eyes were then stored in 70% ethanol until processing for sections. Paraffin-embedded sections (5 µm thickness) passing vertically through the retina (along the vertical meridian passing through the optic nerve head, to allow an examination of the retina in the dorsal and ventral hemisphere) were prepared with a microtome (Leica, Bannockburn, IL, USA).

Primary antibody incubation (anti-M-opsin, 1:1000; anti-S-opsin, 1:1000; anti-CAR, 1:1000; and anti-DIO3, 1:100) was performed at room temperature for 1 h, followed by incubation with AlexaFluor-488 or -568 or FITC-conjugated secondary antibody. PNA immunohistochemistry was performed with biotinylated PNA (1:250) and streptavidin-Cy3 (1:500). Fluorescent signals were imaged using an IX81-FV500 confocal laser scanning microscope (Olympus, Melville, NY, USA) with FluoView imaging software (Olympus). Cone density was evaluated as described (24). In brief, for retinal whole mounts, images were taken with a ×100 objective on an Olympus microscope, ∼1.0 mm from the optic nerve in the center of the dorsal and ventral areas (Supplemental Fig. 1). The image scale was calibrated and cones were counted in 3 to 4 regions, each with dimensions of 127 μm², with Photoshop software (Adobe Systems, Inc., San Jose, CA, USA). Cone density on retinal cross-sections was evaluated by counting the number of cones present in the dorsal and ventral areas. The averages of the counts were analyzed and graphed with Prism software (GraphPad Software, La Jolla, CA, USA).

TUNEL assay

The TUNEL assays were performed to evaluate photoreceptor apoptotic death with the In Situ Cell Death Fluorescein Detection Kit (Roche Diagnostics, Indianapolis, IN, USA) (18).

Quantitative RT-PCR

Total RNA preparation and reverse transcription were performed as described elsewhere. The gene encoding the mouse or human hypoxanthine guanine phosphoribosyl transferase 1 (Hprt1 or HPRT1) was included as an internal control. The quantitative RT-PCR (qRT-PCR) assays were performed using a real-time PCR detection system (iCycler; Bio-Rad Laboratories, Hercules, CA, USA) and the relative gene expression value was calculated based on the ΔΔC_t method (25). Supplemental Table 1 shows the primers used.

Retinal protein preparation, SDS-PAGE, and Western blot analysis

Retinal protein preparation, SDS-PAGE, and Western blot analysis were performed as has been described (18). After gel separation, transferring, and blocking, blots were incubated with primary antibody overnight at 4°C (anti-DIO2, 1:1,000; and anti-actin, 1:2000) and with HRP-conjugated secondary antibodies (1:20,000) for 1 h at room temperature. SuperSignal West Dura Extended Duration chemiluminescent substrate (Thermo Fisher Scientific) was used to detect binding of the primary antibodies to their cognate antigens. HyBlot CL autoradiography film (Denville Scientific, Inc., Metuchen, NJ, USA) was used to develop the target proteins, and Photoshop CS5 (Adobe Systems, Inc.)was used to analyze the signal density (26).

Weri-Rb1 cell culture, drug treatment experiments, and infection with AAV5-IRBP/GNAT2-DIO3 transgenes

Retinoblastoma cell line Weri-Rb1 (ATCC, Manassas, VA, USA) was cultured in RPMI1640 medium (ATCC) with 15% fetal bovine serum (FBS), 2 mM l-glutamine (Mediatech, Inc., Herndon, VA, USA), 0.1% fungizone (InvivoGen, San Diego, CA, USA), 5 μM plasmocin (InvivoGen), 55 μΜ 2-ME (Thermo Fisher Scientific), and 10 μg/ml insulin (I5500; Sigma-Aldrich) at 37°C in a humidified incubator with 5% CO₂ as described (27). For TH treatment, cells were cultured in RPMI1640 medium with 10% FBS replaced with 10% charcoal-stripped serum (Thermo Fisher Scientific) for 16 h, followed by T4 treatment at various concentrations for another 24 h. For DIO2 inhibitor IOP treatment or TH receptor antagonist 1-850 treatment, cells cultured in RPMI 1640 medium containing 10% charcoal-stripped serum were treated with IOP or 1-850 at various concentrations for 16 h, followed by T4 treatment for another 24 h. At the end of the experiments, the cells were harvested for analysis of M-opsin expression by qRT-PCR.

For infection with AAV5-IRBP/Gnat2-DIO3 transgenes, the cells were seeded in a 12-well plate at a density of 5 × 10⁵ cells/well and were grown overnight, and then AAV5-IRBP/GNAT2-DIO3-WT and AAV5-IRBP/GNAT2-DIO3-Cys viruses were added to the wells containing existing growth medium. The concentration of the virus was 1000 multiplicity of infection, with uninfected wells as the negative control. Cells were harvested 48 h after infection and analyzed for DIO3 expression by qRT-PCR.

Scotopic and photopic ERG recordings

Full-field electroretinography (ERG) testing was performed as described (28). In brief, after overnight dark adaptation, animals were anesthetized by intraperitoneal injection of 85 mg/kg ketamine and 14 mg/kg xylazine. ERGs were recorded using an LKC system (Gaithersburg, MD, USA). Potentials were recorded using a platinum wire contacting the corneal surface through a layer of 2.5% methylcellulose. For assessment of scotopic responses, a stimulus intensity of 1.89 log ∣ cd ∣ s ∣ m⁻² was presented to dark-adapted dilated mouse eyes in a Ganzfeld (GS-2000; Nicolet Instruments, Inc., Madison, WI, USA). To evaluate photopic responses, mice were adapted to a 1.46 log ∣ cd ∣ s ∣ m^–2 light for 5 min, and then a light intensity of 1.89 log ∣ cd ∣ s ∣ m^–2 was given. Responses were differentially amplified, averaged, and stored by a Compact-4 signal averaging system (Nicolet Instruments, Inc.).

Statistical analysis

Results are expressed as means ± sem of number of observations. Unpaired 2-sample Student’s t test was used to test for differences between 2 groups of data. Differences were considered statistically significant when P < 0.05. Data were analyzed and graphed with Prism (GraphPad Software).

RESULTS

Intravitreal delivery of DIO2 inhibitor IOP increased cone density in Rpe65^−/− mice

IOP is a potent iodine-containing inhibitor of DIO2 (29, 30) and has been used in an avian model to study the role of TH/DIO2 in filial imprinting (31) and in a zebrafish model to study the role of TH/DIO2 in optic nerve regeneration (32). Before performing the animal experiments, we examined the effects of IOP on T4-induced response in the human retinoblastoma-derived photoreceptor-like cell line Weri-Rb1. These cells express TH receptors and respond to TH with increased expression of OPN1MW (encoding human M-opsin) (33, 34). To examine the effects of IOP, cells were pretreated with IOP at various concentrations for 16 h, followed by T4 treatment for another 24 h. Cells were then analyzed for expression of OPN1MW using quantitative reverse transcription polymerase chain reaction (qRT-PCR). The assays showed that treatment with T4 concentration-dependently induced OPN1MW expression (Supplemental Fig. 2A), and IOP inhibited T4-induced OPN1MW expression in a concentration-dependent manner (Supplemental Fig. 2B). In addition, T4-induced responses were inhibited by a TH receptor antagonist (1-850) (Supplemental Fig. 2C). These results suggest that DIO2 is a key component of the cellular TH signaling regulation in Weri-Rb1 cells, because T4 induced expression of OPN1MW dose dependently in these cells; IOP is an effective DIO2 inhibitor in the Weri-Rb1 cells, because the effect of T4 was dose-dependently inhibited by IOP; and the effect of T4 is mediated via TH receptor, because the effect of T4 was dose-dependently inhibited by a TH receptor antagonist.

Based on the findings from cell culture experiments, we then performed animal experiments. LCA model Rpe65^−/− mice were used in this study. At P5, Rpe65^−/− mice received IOP (50 mM, 1.0 μl) or vehicle (DMSO) via a single intravitreal injection and were evaluated for cone survival/death at P25. We found that treatment with IOP significantly improved cone survival and reduced cone death. As evaluated by PNA and CAR labeling on retinal whole mounts (Fig. 2A, B) and PNA and cone opsin labeling on the retinal sections (Fig. 2C–E), cone density in IOP-treated mice was increased by 30–40%, compared with that in vehicle-treated controls. There was no difference in cone density between vehicle-treated and untreated mice (data not shown), which is consistent with published reports (35, 36).

Figure 2.

Intravitreal delivery of IOP increased cone density in Rpe65^−/− mice. Mice received injections at P5 and were evaluated for cone survival at P25. Representative confocal images of immunofluorescence labeling of PNA (A) and CAR (B) on retinal whole mounts, labeling of PNA (C) and cone opsin (D, E) on retinal sections, and corresponding quantitative analyses. Data are means ± sem of assays in 3–7 mice in each group. The unpaired 2-sample Student’s t test was used to determine differences between IOP- and vehicle-treated mice. *P < 0.05, **P < 0.01. Scale bars, 50 μm.

Intravitreal delivery of IOP reduced cone death in Rpe65^−/− and Rpe65^−/−/Nrl^−/− mice

Effects of IOP on cone death was evaluated by TUNEL assay. We found that the number of TUNEL⁺ cells was reduced by ∼30% in Rpe65^−/− mice treated with IOP, compared with vehicle-treated controls (Fig. 3A). To determine that the reduced number of TUNEL⁺ cells represents a reduction of cone death, we examined the effects of IOP in Rpe65^−/−/Nrl^−/− mice with RPE65 deficiency on a cone-dominant background. Rpe65^−/−/Nrl^−/− mice show a cone defect phenotype similar to that in Rpe65^−/− mice and have been used to study LCA cone defects (17, 37, 38). Rpe65^−/−/Nrl^−/− mice received IOP at P5 and were evaluated for TUNEL⁺ cells at P15. Similar to Rpe65^−/− mice, the number of TUNEL⁺ cells in Rpe65^−/−/Nrl^−/− mice was significantly reduced by IOP treatment (Fig. 3B).

Figure 3.

Intravitreal delivery of IOP reduced cone death in Rpe65-deficient mice. Rpe65^−/− and Rpe65^−/−/Nrl^−/− mice received injections at P5 and were evaluated for cone death at P25 and P15, respectively. Representative confocal images of TUNEL⁺ cells in retinal sections and the corresponding quantitative analysis in Rpe65^−/− (A) and Rpe65^−/−/Nrl^−/− (B) mice. Data are means ± sem of assays from three to four mice in each group. The unpaired 2-sample Student’s t test was used to determine differences between IOP-treated and vehicle-treated mice. *P < 0.05. Scale bars, 30 μm.

Topical delivery of IOP improved cone survival in Rpe65^−/− mice

The effectiveness of IOP was further demonstrated by topical delivery. P5 Rpe65^−/− mice received IOP in eye drops (1.0% IOP in PBS containing 25% phospholipon 90G) 3×/d for 20 d. The mice were then evaluated for cone survival and death. We found that cone density was significantly higher in IOP-treated mice than in vehicle-treated controls (Fig. 4A). The number of TUNEL⁺ cells was reduced by ∼33% in Rpe65^−/− mice treated with IOP (Fig. 4B). To learn whether the treatment has any effects on photoreceptor function, we used ERG to examine retinal light responses. The analysis showed that the scotopic and photopic ERG responses were not different between the IOP- and vehicle-treated mice, and the photoreceptor function was not improved in the IOP-treated mice (Supplemental Fig. 3).

Figure 4.

Topical delivery of IOP increased cone density and reduced cone death in Rpe65^−/− mice. Mice received topical administration of IOP at P5 and were analyzed for cone density and TUNEL⁺ cells at P25. A) Representative confocal images of immunofluorescence labeling of PNA on retinal whole mounts and corresponding quantitative analysis. B) Representative confocal images showing TUNEL⁺ cells on retinal sections and corresponding quantitative analysis. Data are represented as means ± sem of assays from 6 to 7 mice in each group. The unpaired 2-sample Student’s t test was used to determine differences between IOP- and vehicle-treated mice. *P < 0.05, **P < 0.01.

Subretinal delivery of AAV5-IRBP/GNAT2-DIO3 improved cone survival in Rpe65^−/− mice

DIO3 degrades T3 and T4. Thus, we examined whether overexpression of DIO3 is protective. The enzyme contains the Sec residue at position 144, which is critical for the enzyme’s activity, but limits its expression (19, 39). Therefore, we generated 2 transgene constructs, AAV5-IRBP/GNAT2-DIO3-WT and AAV5-IRBP/GNAT2-DIO3-Cys. The Cys mutant was constructed by substitution of the codons TGA (for Sec) with codons TGC (for Cys). We first examined expression of the transgenes in Weri-Rb1 cells. Cells were infected with viral vectors and analyzed for DIO3 expression by qRT-PCR at 48 h after infection. The assays showed that the levels of DIO3-WT and DIO3-Cys in cells infected with the viral vectors were increased by ∼70- and ∼1000-fold, respectively, compared with uninfected controls (Supplemental Fig. 4).

To test these transgenes in vivo, P5 Rpe65^−/− mice received subretinal injections of AAV5-IRBP/GNAT2-DIO3-WT, AAV5-IRBP/GNAT2-DIO3-Cys, or vehicle (PBS) and were evaluated for cone survival/death at P30. The expression of DIO3 was examined by immunofluorescence labeling with anti-DIO3 antibody and colabeling with cone and rod markers. The assays showed a significant amount of immunoreactivity of DIO3 in viral vector–injected mice, compared with vehicle-treated controls (Fig. 5A). Colabeling with PNA and with rhodopsin confirmed its cone localization, with signals enriched in the inner segment area (Fig. 5B, C).

Figure 5.

Expression of hDIO3 in cones of Rpe65^−/− mice after subretinal delivery of AAV5-IRBP/GNAT2-DIO3. Rpe65^−/− mice at P5 received AAV5-IRBP/GNAT2-DIO3-WT (1 × 10¹⁰ vector genomes, 1.0 μl), AAV5-IRBP/GNAT2-DIO3-Cys (1 × 10¹⁰ vector genomes, 1.0 μl), or vehicle (PBS) via subretinal injection. Retinal expression/cone localization of hDIO3 and cone density was analyzed at P30. Representative confocal images of DIO3 labeling in the mouse retina (A), colabeling with PNA (B), but not with rhodopsin (C). OS, outer segment; IS, inner segment. Scale bars, 30 μm.

Cone density evaluation revealed that cone survival in mice treated with AAV5-IRBP/GNAT2-DIO3 was significantly improved. As analyzed by PNA labeling on retinal whole mounts, treatment with AAV5-IRBP/GNAT2-DIO3-WT increased cone density by ∼23 and 47% in the dorsal and ventral areas, respectively, and treatment with AAV5-IRBP/GNAT2-DIO3-Cys increased cone density by ∼32 and 103% in these regions, respectively (Fig. 6A). Similar results were obtained from evaluations performed on retinal sections (Fig. 6B). The Cys mutant appeared to induce a more significant rescue than that induced by the wild-type.

Figure 6.

Subretinal delivery of AAV5-IRBP/GNAT2-hDIO3 increased cone density in Rpe65^−/− mice. Rpe65^−/− mice at P5 received AAV5-IRBP/GNAT2-DIO3-WT (1 × 10¹⁰ vector genomes, 1.0 μl), AAV5-IRBP/GNAT2-DIO3-Cys (1 × 10¹⁰ vector genomes, 1.0 μl), or vehicle (PBS) via subretinal injection and were evaluated for cone survival at P30. Representative confocal images of immunofluorescence labeling of PNA on retinal whole mounts (A) and retinal sections (B) and the corresponding quantitative analyses. Data are means ± sem of assays from 5–12 mice in each group. The unpaired 2-sample Student’s t test was used to determine differences between transgene- and vehicle-treated mice. *P < 0.05, **P < 0.01. Scale bars, 30 μm.

Subretinal delivery of AAV5-IRBP/GNAT2-DIO3 improved cone survival in Cpfl1 mice

Cpfl1 with defect of Pde6c is a naturally occurring achromatopsia/cone dystrophy model with rapid and severe cone degeneration (40, 41). We previously showed that treatment with antithyroid drug preserved cones in these mice (10). We examined the effects of subretinal delivery of AAV5-IRBP/GNAT2-DIO3. We treated Cpfl1 mice with virus at P5 and evaluated cone density at P30. We found that mice treated with DIO3 transgenes showed significantly improved cone survival. The treatment increased cone density by ∼30% (Fig. 7A). Similar results were obtained from analysis performed on retinal sections (Fig. 7B).

Figure 7.

Subretinal delivery of AAV5-IRBP/GNAT2-hDIO3 increased cone density in Cpfl1 mice. Mice received subretinal injections of AAV5-IRBP/GNAT2-DIO3 constructs at P5 and were evaluated for cone survival at P30. Representative confocal images of immunofluorescence labeling of PNA on retinal whole mounts (A) and retinal sections (B) and corresponding quantitative analysis. Data are means ± sem of assays from 4–12 mice in each group. The unpaired 2-sample Student’s t test was used to determine significance between transgene vector- and vehicle-treated mice. *P < 0.05, **P < 0.01, ***P < 0.001. Scale bars, 30 μm.

Enhanced expression levels of DIO2 and Slc16a2 in cone degeneration retinas

Suppressing TH signaling by use of antithyroid drugs or through targeting DOIs has been shown to protect cones. The main question to be asked is how suppression of TH signaling preserves cones. It has been shown that excessive TH causes cone death in both healthy and degenerating cones (10, 11). The finding that suppressing TH signaling protects cones undergoing degeneration suggests that either there is elevated TH signaling in degenerating cones/retinas, or that the degenerating cones become more sensitive to TH, or both. As an initial effort to address the question, we examined the expression levels of DIOs in cone degeneration retinas. We used Rpe65^−/−/Nrl^−/− retinas to facilitate the biochemical detection of cellular alterations in cones. We found that the mRNA and protein levels of DIO2 in Rpe65^−/−/Nrl^−/− mice were significantly higher than levels in age-matched Nrl^−/− mice. The levels of DIO2 mRNA in Rpe65^−/−/Nrl^−/− mice increased by about 2- to-4-fold (Fig. 8A), whereas the levels of DIO2 protein increased by ∼50% (Fig. 8B). There was no difference in Dio1 expression between Rpe65^−/−/Nrl^−/− and Nrl^−/− mice (Fig. 8C). Dio3 expression was increased in P15 mice, but was unchanged in P30 mice (Fig. 8D). In contrast, the expression levels of the genes encoding the cone phototransduction proteins M-opsin (Opn1mw) and S-opsin (Opn1sw) were significantly lower in P30 Rpe65^−/−/Nrl^−/− mice (Fig. 8E, F). In this work, we also examined the expression of Dio2 in another cone degeneration model, Cngb3^−/−/Nrl^−/− mice with cone cyclic nucleotide-gated channel β (CNGB)-3 deficiency on the Nrl^−/− background, a model of achromatopsia/cone dystrophies (10, 18), and found increased expression of Dio2 (Fig. 8G). In addition, we examined expression of the gene encoding the TH transporter monocarboxylate transporter 8 (Slc16a2) (42, 43) and found that the expression of Slc16a2 was significantly higher in Rpe65^−/−/Nrl^−/− mice, compared with age-matched Nrl^−/− controls (Fig. 8H).

Figure 8.

Enhanced expression of DIO2 and Slc16a2 in cone degeneration retinas. Retinal expression levels of DIOs were analyzed in Rpe65^−/−/Nrl^−/− and Nrl^−/− mice. A) Increased levels of Dio2 in retinas of P15 and P30 Rpe65^−/−/Nrl^−/− mice. Shown are results of the qRT-PCR assays. B) Increased levels of DIO2 in retinas of P15 Rpe65^−/−/Nrl^−/− mice. Shown are representative images of the immunoblot detections and the corresponding densitometric analyses. Actin was used as a loading control. C, D) Unchanged levels of Dio1 (C) and increased levels of Dio3 (D) in Rpe65^−/−/Nrl^−/− retinas. E, F) Decreased levels of Opn1mw (E) and Opn1sw (F) in Rpe65^−/−/Nrl^−/− retinas. G) Enhanced levels of Dio2 in Cngb3^−/−/Nrl^−/− retinas. H) Enhanced level of Slc16a2 in Rpe65^−/−/Nrl^−/− retina. C–H) Results of the qRT-PCR assays. Data are means ± sem of assays from four to nine mice in each group for mRNA analysis [Nrl^−/− at P15 (n = 9); Rpe65^−/−/Nrl^−/− at P15 (n = 4); Nrl^−/− at P30 (n = 6); Rpe65^−/−/Nrl^−/− at P30 (n = 4); Cngb3^−/−/Nrl^−/− at P30 (n = 6)] and from 6 mice in each group for protein analysis. The unpaired 2-sample Student’s t test was used to determine differences between Rpe65^−/−/Nrl^−/− and Nrl^−/− mice. *P < 0.05, **P < 0.01, ***P < 0.001.

In this work, we also examined the expression of these genes in Rpe65^−/− mice, compared with that in wild-type mice to investigate TH signaling in a degenerating retina containing a mixture of rods and cones. The assays showed that the expression levels of Dio2 and Dio3 were significantly higher in Rpe65^−/− mice than that in the wild-type mice; there was no significant difference in Dio1 and Slc16a2 expression levels between Rpe65^−/− and wild-type mice; and, as reported (44), the expression levels of Opn1mw and Opn1sw were significantly lower in Rpe65^−/− mice (Supplemental Fig. 5).

DISCUSSION

We investigated the effects of targeting DIOs locally in the retina and showed that treatment with a transgene vector for overexpression of DIO3 or a DIO2 inhibitor improved cone survival in mouse models of cone degeneration. Our work suggests that targeting DIOs locally in the retina represents a first step toward the development of a therapeutic strategy.

Inhibition of DIO2 protects cones

The protective effects of DIO2 inhibition were demonstrated by the use of the DIO2 inhibitor IOP. Intravitreal or topical delivery of IOP significantly increased cone density and reduced cone death in Rpe65^−/− and Rpe65^−/−/Nrl^−/− mice, indicating that DIO2 can be effectively targeted to suppress TH signaling. However, we did not observe a significant improvement in the cone light responses. The scotopic and photopic ERG responses were not different between the IOP- and vehicle-treated mice. Possible explanations could be 1) the level of the cone number increase is not sufficient for an increase in the sum of the cone light responses recorded by photopic ERG, and 2) the deficiency of RPE65 and the associated opsin mistrafficking are not corrected by the treatment (Fig. 2D, E); therefore, the resultant cone functional impairment is not corrected.

DIO2 has been well characterized as a critical component in the control of cellular T3 level/TH signaling. These include its regulation in adaptive thermogenesis in brown adipose tissue (45), in normal myogenesis and muscle regeneration (46), and in hearing function and cochlear development (47). A recent study showed that inhibiting DIO2 using IOP accelerates reinnervation of the optic tectum after optic nerve crush in zebrafish (32). Our work holds promise for cone protection by DIO2 inhibition locally in the retina.

Overexpression of DIO3 protects cones

The protective effects of DIO3 overexpression were demonstrated with subretinal delivery of AAV5-IRBP/GNAT2-DIO3-WT and AAV5-IRBP/GNAT2-DIO3-Cys. This is the first effort to suppress TH signaling by overexpression of DIO3. Like other DIOs, DIO3 is a Sec-containing enzyme. The Sec residue is critical for the enzyme’s activity, but limits its expression (19, 39). Replacing the codons TGA (for Sec) with codons TGC (for Cys) was shown to enhance the expression by about 50-fold while reducing the activity by ∼6-fold (39). Therefore, the Cys mutant has been used in heterologous expression studies (48). We showed that both the wild-type and the Cys mutant transgenes induced effective expression of DIO3 in the mouse retina; the Cys mutant transgene induced DIO3 expression ∼10-fold higher than that induced by the wild-type transgene. After subretinal injection, Rpe65^−/−, Rpe65^−/−/Nrl^−/−, and Cpfl1 mice showed significantly increased cone density and reduced cone death. The Cys mutant appeared to induce better rescue than did the wild-type, which might be attributable to higher expression of the mutant.

The controlling role of DIO3 in local tissues, including retinal and other neuronal tissues, has been well documented in studies of Dio3^−/− mice. These mice display auditory defects and premature cochlear differentiation (49), loss of cones (11), and cerebellar abnormalities (50), and all of these defects are rescued or partially rescued by TH receptor deletion. In nonneuronal tissues, DIO3 depletion has been shown to cause massive apoptosis of muscle cells and impairment of the muscle regeneration and repair process (51). The controlling role of DIO3 in local TH signaling regulation and cell survival has also been supported by its induction/upregulation under a variety of pathologic conditions, critical illnesses, and regeneration processes. Induction of DIO3 was observed in patients with liver and skeletal muscle injury (52) and in animal models of injury/diseases, including rat models of brain damage/ischemia and hypoxia (53, 54), rat models of heart failure (55, 56), and a zebrafish model of fin regeneration (57). The significance of such induction regulation has been attributed to decreased expenditure, as part of a homeostatic mechanism, and reduced cell metabolism in the face of limited oxygen supply and stressful conditions (14, 58, 59). Induction of DIO3 has been suggested to be a survival factor in muscle stem cell proliferation and lineage progression (59). The present work demonstrates for the first time the promise of DIO3 overexpression via transgene delivery for cell protection.

It is important to note that the rescue achieved by overexpression of DIO3 or inhibition of DIO2 is partial. This partial rescue may be associated with 1) the drug treatment efficiency, which is associated with the drug bioavailability and distribution; 2) the requirement to target multiple TH components; and 3) the contributions from non-TH factors. It is important to emphasize that although TH signaling plays a role in cone degeneration, it is not the sole factor. Other non-TH factors could contribute to the disease progression. Also, the initial pathologic conditions will not be corrected by TH inhibition. Critical among these are the requirement of cone photoreceptors for the 11-cis-retinal chromophore missing in the Rpe65^−/− mice (15). Similarly, the pathologic elevation of the cellular cGMP and the subsequent over activation of the CNG channel in the Pde6c-deficient Cpfl1 mice (40, 41) would not be corrected by TH suppression treatment.

Upregulation of DIO2 and TH transporter in degenerating retinas

The expression of DIO2 and Slc16a2 were increased in Rpe65^−/−/Nrl^−/− retinas, which contrasts with the reduced expression of Opn1mw and Opn1sw in these mice and Rpe65^−/− mice (17, 37, 44). The elevated expression of TH components was also observed in the Cngb3^−/−/Nrl^−/− mouse model of cone dystrophy/achromatopsia. In this study, we also examined the expression of these genes in Rpe65^−/− mice with a mixture of rods (rod-dominant) and cones. The observation of Dio2 expression was similar to that in Rpe65^−/−/Nrl^−/− mice (i.e., the expression of Dio2 was increased). The observed elevation of Dio2 expression in Rpe65^−/− mice is less likely to be contributed solely by the changes in cones. Instead, it may imply the alterations in both rods and cones. Unlike that in Rpe65^−/−/Nrl^−/− mice, the expression of Slc16a2 was not altered in Rpe65^−/− mice, suggesting that the expression of Slc16a2 is upregulated in cones, but not in rods, and that the regulation of Slc16a2 expression is different between cones and rods. Taken together, TH signaling appears to be locally enhanced in degenerating retinas, although the underlying mechanism remains to be determined.

Upregulation of DIO2 has been reported in several other pathologic conditions (60, 61). This upregulation could be a result of the pathologic conditions, which could in turn further contribute to or worsen the disease progression. The observation that suppressing TH signaling reduces cone death suggests that the elevated TH signaling is harmful for cones. The view that high TH is detrimental has been documented in several pathologic conditions. In the sensory/neuronal system, excessive TH signaling induces death of cone photoreceptors (10, 11), auditory defects/cochlear degeneration (49), and cerebellar abnormalities (50). TH signaling has also been implicated in other neurodegenerative conditions, including Alzheimer’s disease (62, 63). Recently, a population-based study showed that high serum T4 values are associated with an increased risk of age-related macular degeneration (64), suggesting a role of TH signaling in the pathogenesis of age-related macular degeneration. Nevertheless, the finding that the TH components were upregulated explains why suppressing TH signaling protected cones. The view that low TH signaling is protective is receiving increasing attention, as is the protective effect of suppression of cellular TH components, detailed above.

Future perspectives/therapeutic significance

The eye has advantages over other organs in disease management because of its accessibility to topically delivered therapeutic eyedrops (22, 65). We demonstrated that targeting DIOs locally in the retina protects cones. These findings provide a promise for development of a novel strategy for retinal degeneration management. The significance and future perspectives include the following: the development of a DIO3 activator/inducer would benefit intracellular TH inhibition and cone protection; the development of more potent DIO2 inhibitors would benefit intracellular TH inhibition and cone protection; and suppressing cellular TH signaling in the retina could be achieved by using inhibitors separately or in combination. We anticipate that TH inhibition at multiple targets will lead to better protection. In addition, the findings of cone protection by TH signaling suppression will encourage us to explore whether these approaches can be used for protection of rods. Compared to our knowledge regarding the role of TH signaling in cones, we know little about the role of TH signaling in rods. Nevertheless, it has been shown that excessive TH signaling is harmful to rods as well. T3 treatment reduced the thickness of outer nuclear layer (ONL)/number of nuclei in the ONL in wild-type and retinal-degeneration model mice (10, 11). Moreover, treatment with IOP was shown to improve reinnervation of the optic tectum after optic nerve crush in zebrafish (32), suggesting that the protection is a general feature. It should be pointed out that we and others have shown that suppressing TH signaling by antithyroid drug treatment has no negative effect on cone/rod structure and function or survival during development and in adult mice (9, 66, 67). Thus, the ocular delivery of a TH inhibitor is perhaps less likely to result in detrimental effects. It is our hope that future patients with retinal degeneration will benefit from the use of TH inhibitor eyedrops.

In summary, targeting DIOs locally in the retina protected cones in mouse models of retinal degeneration. Our findings suggest that targeting DIOs locally in the retina represents a novel strategy for cone protection and serves as the first step toward management of retinal/cone degeneration. If local suppression of TH signaling can effectively preserve even a small population of cones, the outcome would be significant and could help patients with retinal degenerative diseases to maintain their quality of life and independence.

Glossary

AAV

adeno-associated virus serotype

CAR

cone arrestin

CNG

cyclic nucleotide-gated

Cys

cysteine

DIO

deiodinase iodothyronine

ERG

electroretinography

GNAT

guanine nucleotide-binding protein G(t)

HRP

horseradish peroxidase

INL

inner nuclear layer

IOP

iopanoic acid

IRBP

interphotoreceptor retinoid-binding protein

LCA

Lebers congenital amaurosis

ONL

outer nuclear layer

P

postnatal day

PNA

peanut agglutinin

qRT-PCR

quantitative reverse transcription–polymerase chain reaction

Sec

selenocysteine

TH

thyroid hormone

T4

thyroxine

T3

triiodothyronine

AUTHOR CONTRIBUTIONS

F. Yang, H. Ma, M. R. Butler, J. Belcher, and S. L. Boyle performed research and analyzed data; F. Yang, H. Ma, and X.-Q. Ding designed the research and wrote the manuscript. T. M. Redmond, S. L. Boyle, and W. W. Hauswirth contributed to the research design and wrote the manuscript.