Deficiency of thyroid hormone receptor protects retinal pigment epithelium and photoreceptors from cell death in a mouse model of age-related macular degeneration

Age-related macular degeneration (AMD) is the leading cause of vision loss in the elderly. Progressive dystrophy of the retinal pigment epithelium (RPE) and photoreceptors is the characteristic of dry AMD, and oxidative stress/damage plays a central role in the pathogenic lesion of the disease. Thyroid hormone (TH) regulates cell growth, differentiation, and metabolism, and regulates development/function of photoreceptors and RPE in the retina. Population-/patient-based studies suggest an association of high free-serum TH levels with increased risk of AMD. We recently showed that suppressing TH signaling by antithyroid treatment reduces cell damage/death of the RPE and photoreceptors in an oxidative-stress/sodium iodate (NaIO3)-induced mouse model of AMD. This work investigated the effects of TH receptor (THR) deficiency on cell damage/death of the RPE and photoreceptors and the contribution of the receptor subtypes. Treatment with NaIO3 induced RPE and photoreceptor cell death/necroptosis, destruction, and oxidative damage. The phenotypes were significantly diminished in Thrα1−/−, Thrb−/−, and Thrb2−/− mice, compared with that in the wild-type (C57BL/6 J) mice. The involvement of the receptor subtypes varies in the RPE and retina. Deletion of Thrα1 or Thrb protected RPE, rods, and cones, whereas deletion of Thrb2 protected RPE and cones but not rods. Gene-expression analysis showed that deletion of Thrα1 or Thrb abolished/suppressed the NaIO3-induced upregulation of the genes involved in cellular oxidative-stress responses, necroptosis/apoptosis signaling, and inflammatory responses. In addition, THR antagonist effectively protected ARPE-19 cells and hRPE cells from NaIO3-induced cell death. This work demonstrates the involvement of THR signaling in cell damage/death of the RPE and photoreceptors after oxidative-stress challenge and the receptor-subtype contribution. Findings from this work support a role of THR signaling in the pathogenesis of AMD and the strategy of suppressing THR signaling locally in the retina for protection of the RPE/retina in dry AMD.


INTRODUCTION
Age-related macular degeneration (AMD) is the leading cause of blindness in the elderly [1][2][3]. The dry AMD, also known as geographic atrophy, is a form of slowly progressing geographic atrophy of the macula and comprises a majority of AMD cases (∼90%) [1,4]. The disease is characterized by a progressive macular degeneration of the retinal pigment epithelium (RPE) and photoreceptors. There are multiple pathological factors, including aging, oxidative stress, chronic inflammation, and genetic defects. However, oxidative stress/damage to the RPE and the subsequent deterioration of photoreceptors is recognized as the core pathogenic lesion of AMD [1,5,6].
Using a sodium iodate (NaIO 3 )-induced mouse model of AMD [24][25][26], we recently showed that treatment with antithyroid drug protected RPE and photoreceptors from oxidative damage and cell death/necroptosis and preserved retinal function [27]. The present work expanded this research by investigating the involvement of TH receptors (THRs) and the contribution of the receptor subtypes. T3 acts through THRs that belong to the nuclear hormone-receptor superfamily and function as liganddependent transcription factors [9]. Two genes, THRA and THRB, encode related receptors across vertebrate species [9,28]. THRA1 is encoded by the THRA gene, and two THRB-isoform splice variants, THRB1 and THRB2, are encoded by the THRB gene. These receptor subtypes are broadly expressed in a variety of tissues, including the RPE and retina [29][30][31]. However, THRB2 is expressed only in cones in the retina [29,32,33]. THRB2 has been shown to mediate the regulation of TH in cone opsin expression [13,14,34] and cone viability [10,16]. We examined the NaIO 3 -induced cell damage/death of the RPE and photoreceptors in Thrα1 −/− , Thrb −/− (resulting in deletion of both Thrb1 and Thrb2), Thrb2 −/− , and wild-type (C57BL/6 J) mice. Our results show that deficiency of THR significantly diminished the NaIO 3induced cell damage/death and upregulation of the genes involved in cellular oxidative-stress responses, activation of the cell-death signaling, and inflammatory responses.

RESULTS
Deletion of Thrα1 protected RPE and photoreceptors from damage/cell loss induced by NaIO 3 In the previous study, we showed that antithyroid treatment reduces NaIO 3 -induced oxidative damage/cell death of the RPE and photoreceptors [27]. The present work investigated the THR mechanisms involved in NaIO 3 -induced damage. We first examined the contribution of THRA1. Thra1 −/− , Thra1 +/− , and wild-type (C57BL/6 J) mice received a single injection of NaIO 3 (30 mg/kg, i. p.) at postnatal day 30 (P30), and were then analyzed for RPE morphology and photoreceptor integrity at 3 days post NaIO 3 injection. A single administration of NaIO 3 (i.v., i.p., or intraocular injection) induces RPE/photoreceptor damage in experimental animals in a concentration-dependent and time-dependent manner [25,[35][36][37]. The functional and morphological impairments are observed as early as 24 hours, become more severe at 3-7 days, and last up to 4 weeks. We chose 3 days after the treatment as the evaluation time point in the previous work [27] and the present work because the damage is already significant at this time point. RPE morphology and cell loss were evaluated by phalloidin staining for F-actin and DAPI staining for nucleus on RPE whole mounts. Treatment with NaIO 3 induced damage in 50% of the entire RPE area in the wild-type mice. The damaged area was reduced to 25% in Thra1 −/− mice (Fig. 1A, B). In contrast, there was a slight but not significant improvement in Thra1 +/− mice (Fig. 1A, B). The RPE cell number in the central and middle regions was reduced by about 85% and 60%, respectively, in the wild-type mice after NaIO 3 treatment. Thra1 −/− mice showed significantly increased numbers of RPE cells, compared with that in the wildtype and Thra1 +/− mice (Fig. 1C, D). Similar results were obtained in RPE nuclear-number evaluations (Fig. 1D, lower panel). RPE morphology in untreated Thra1 −/− mice was not different from that in the wild-type (data not shown). The protective effect of Thra1 deletion was also observed in mice at a relatively older age. Thra1 −/− and wild-type mice at 7 months received NaIO 3 treatment and were then analyzed for RPE morphology at 2 days post treatment. Phalloidin staining showed that the NaIO 3 Fig. 1 Deletion of Thrα1 protected RPE from cell damage/loss induced by NaIO 3 . RPE morphology and cell loss were evaluated by phalloidin staining for F-actin and DAPI staining for nucleus on RPE whole mounts prepared from Thrα1 −/− , Thra1 +/− , and wild-type mice at 3 days post NaIO 3 injection. A, B Shown are representative low-magnification images of phalloidin staining of the damaged area in the RPE (A) and corresponding quantitative analysis of the damaged area (B). C, D Shown are representative high-magnification images of phalloidin staining and DAPI labeling taken at different regions of the RPE (C) and corresponding quantitative analysis of RPE cell numbers and RPE nuclear numbers (D). Data are represented as means ± SEM for 5-22 mice per group (**p < 0.01, ***p < 0.001, compared with wild-type mice treated with NaIO 3 ).
treatment caused damage in about 88% of the entire RPE area in the wild-type mice, but the damaged area was reduced to 74% in Thra1 −/− mice (p < 0.05, Supplementary Fig. 1). The RPE cellnumber analysis showed similar findings ( Supplementary Fig. 1).
The protective effects of Thra1 deletion on retina/photoreceptors were demonstrated by evaluation of retinal integrity, photoreceptor number, and retinal cell death. H&E staining of the retinal cross sections showed that treatment with NaIO 3 caused severe damage in the photoreceptor layer in the wild-type mice, manifested as a disorganized outer nuclear layer (ONL) and outer segment (OS), reduced nuclear numbers/thickness of the ONL, and shortened OS. The detrimental effects from NaIO 3 treatment were greatly reduced in Thrα1 −/− mice. After NaIO 3 treatment, ONL thickness in the central retina of the wild-type mice was reduced by about 27%, compared with the untreated controls, and deletion of Thra1 nearly completely prevented this reduction ( Fig. 2A). Retinal morphology in untreated Thra1 −/− mice was not different from that in the wild-type ( Fig. 2A). Cone Fig. 2 Deletion of Thrα1 protected rod and cone photoreceptors from cell loss induced by NaIO 3 . Retinal morphology, photoreceptor-layer integrity, and loss of photoreceptors were evaluated by light microscope and morphometric analysis in Thrα1 −/− , Thrα1 +/− , and wild-type mice at 3 days post NaIO 3 injection. Cone density was evaluated by PNA labeling on retinal whole mounts, and photoreceptor apoptosis was evaluated by TUNEL assay. A. Shown are representative light microscopic images of H&E-stained retinal sections, and corresponding quantitative analysis of ONL thickness in the dorsal and ventral regions. RPE retinal-pigment epithelial, ONL outer nuclear layer, INL inner nuclear layer, IPL inner plexiform layer. B Shown are representative confocal images of PNA labeling on retinal whole mounts, and corresponding quantitative analysis. C Shown are representative confocal images of TUNEL labeling on retinal sections and corresponding quantitative analysis. Data are represented as means ± SEM for 3-17 mice per group (*p < 0.05, **p < 0.01, ***p < 0.001, compared with wildtype mice treated with NaIO 3 ).
photoreceptor density/numbers were assessed by peanutagglutinin (PNA) labeling on retinal whole mounts. After NaIO 3 treatment, cone density in the retinas of wild-type mice was reduced by about 37%, compared with the untreated controls. However, cone density in Thrα1 −/− mice after NaIO 3 treatment was reduced by 13% only, compared with untreated Thrα1 −/− controls (Fig. 2B). Cone density in the untreated Thra1 −/− mice was not different from that in the wild-type (Fig. 2B). Similar to findings in the RPE evaluations, Thra1 +/− mice did not show significant protection against NaIO 3 -induced loss of cones (Fig.  2B). Retinal cell death was evaluated by terminal deoxynucleotidyltransferase dUTP nick-end labeling (TUNEL). After NaIO 3 treatment, a large increase in the number of TUNEL-positive cells was observed in ONL areas of the retinal sections prepared from wild-type mice. However, deletion of Thra1 nearly completely eliminated the TUNEL labeling (Fig. 2C). In this work, we also examined mRNA-expression levels of Thra1 and other THR subtypes in the RPE and retina, and their retinal localization. qRT-PCR analysis of the retinas (prepared from P30 C57BL/6 J mice) showed that Thra1, Thrb1, and Thrb2 were all expressed in the RPE and retina, and that Thra1 was expressed at 1-2-fold higher than Thrb1 and Thrb2 in both tissues (Supplementary Fig.  2A). RNAscope in situ hybridization analysis showed the expression of Thra1 mRNA and Thrb mRNA in all layers of the retina (Supplementary Fig. 2B), similar to the previous findings [29 -31, 38]. No difference in the expression levels of Thrb and Thra1 was observed between the periphery and center of the retina.
Deletion of Thrb protected RPE and photoreceptors from damage/cell loss induced by NaIO 3 We next examined the effects of THRB deletion. Thrb −/− and wildtype mice received a single injection of NaIO 3 (30 mg/kg, i.p.) at P30, and were analyzed for RPE morphology and photoreceptor integrity at 3 days post NaIO 3 treatment. Phalloidin staining of the RPE whole mounts showed that treatment with NaIO 3 caused damage in 25% of the entire RPE area in Thrb −/− mice, which was significantly lower than the 50% damaged area in the wild-type (Fig. 3A, B). RPE morphology in untreated Thrb −/− mice was not different from that in the wild-type (data not shown). The RPE cell number in the central and middle regions was reduced by about 85% and 60%, respectively, in the wild-type mice after NaIO 3 treatment. Thrb −/− mice showed significantly increased numbers of RPE cells, compared with that in the wild-type (Fig. 3C, D). Similar results were obtained in RPE nuclear-number evaluations (Fig. 3C, D).
The protective effects of Thrb deletion on retina/photoreceptors were demonstrated by evaluation of retinal integrity, photoreceptor number, and retinal cell death. The overall retinal morphology was well preserved in Thrb −/− mice after NaIO 3 treatment, compared with that in the wild-type (Fig. 4A). After NaIO 3 treatment, ONL thickness in the central retina of the wildtype mice was reduced by about 27%, compared with untreated controls, and deletion of Thrb completely prevented this reduction (Fig. 4A). Retinal morphology in untreated Thrb −/− mice was not different from that in the wild-type (Fig. 4A). PNA labeling on Fig. 3 Deletion of Thrb protected RPE from cell damage/loss induced by NaIO 3 . RPE morphology and cell loss were evaluated by phalloidin staining for F-actin and DAPI staining for nucleus on RPE whole mounts prepared from Thrb −/− and wild-type mice at 3 days post NaIO 3 injection. A, B Shown are representative low-magnification images of phalloidin staining of the damaged area in the RPE (A) and corresponding quantitative analysis of the damaged area (B). C, D. Shown are representative high-magnification images of phalloidin staining and DAPI labeling taken at different regions of the RPE (C) and corresponding quantitative analysis of RPE cell numbers and RPE nuclear numbers (D). Data are represented as means ± SEM for 5-22 mice per group (**p < 0.01, ***p < 0.001, compared with wild-type mice treated with NaIO 3 ).
retinal whole mounts showed that after NaIO 3 treatment, cone density in the wild-type mice was reduced by about 37%, compared with untreated controls. However, cone density in Thrb −/− mice after NaIO 3 treatment was reduced by 18% only, compared with untreated Thrb −/− controls (Fig. 4B). Cone density in the untreated Thrb −/− mice was not different from that in the wild-type (Fig. 4B). Treatment with NaIO 3 induced a large increase in the number of TUNEL-positive cells in the wild-type mice and deletion of Thrb significantly reduced the number of TUNELpositive cells (Fig. 4C).
Deletion of Thrb2 protected RPE and cones but not rods from damage/cell loss induced by NaIO 3 We then further examined the contribution of THRB2. Thrb2 −/− and wild-type mice received a single injection of NaIO 3 (30 mg/kg, i.p.) at P30, and were analyzed for RPE morphology and Fig. 4 Deletion of Thrb protected rod and cone photoreceptors from cell loss induced by NaIO 3 . Retinal morphology, photoreceptor-layer integrity, and loss of photoreceptors were evaluated by light microscope and morphometric analysis in Thrb −/− and wild-type mice at 3 days post NaIO 3 injection. Cone density was evaluated by PNA labeling on retinal whole mounts, and photoreceptor apoptosis was evaluated by TUNEL assay. A Shown are representative light microscopic images of H&E-stained retinal sections, and corresponding quantitative analysis of ONL thickness in the dorsal and ventral regions. RPE retinal-pigment epithelial, ONL outer nuclear layer, INL inner-nuclear layer, IPL inner plexiform layer. B Shown are representative confocal images of PNA labeling on retinal whole mounts, and corresponding quantitative analysis. C Shown are representative confocal images of TUNEL labeling on retinal sections and corresponding quantitative analysis. Data are represented as means ± SEM for 4-17 mice per group (*p < 0.05, **p < 0.01, ***p < 0.001, compared with wild-type mice treated with NaIO 3 ). photoreceptor integrity at 3 days post NaIO 3 treatment. Phalloidin staining of the RPE whole mounts showed that treatment with NaIO 3 caused damage in 28% of the entire RPE area in Thrb2 −/− mice, which was significantly lower than a 50% damaged area in the wild-type (Fig. 5A, B). RPE morphology in untreated Thrb2 −/− mice was not different from that in the wild-type (data not shown). The RPE cell number in the central and middle regions was reduced by about 85% and 60%, respectively, in the wild-type mice after NaIO 3 treatment. Thrb2 −/− mice showed significantly increased numbers of RPE cells, compared with that in the wildtype (Fig. 5C, D). Similar results were obtained in RPE nuclearnumber evaluations (Fig. 5C, D). The protective effect of Thrb2 deletion was also observed in mice at an old age. Thrb2 −/− and wild-type mice at 17 months received NaIO 3 treatment and were then analyzed for RPE morphology at 2 days post treatment. Phalloidin staining showed that the NaIO 3 treatment caused damage in about 88% of the RPE area in the wild-type mice, but the damaged area was reduced to 75% in the Thrb2 −/− mice (p < 0.05, Supplementary Fig. 3A). The RPE cell/nuclear-number analysis showed similar findings ( Supplementary Fig. 3B).
The protective effects of Thrb2 deletion on retina/photoreceptors were demonstrated by evaluation of retinal integrity, photoreceptor number, and retinal cell death. After NaIO 3 treatment, ONL thickness in the central retina of the wild-type mice was reduced by about 27%, compared with the untreated controls. Unlike that in Thrα1 −/− or Thrb −/− mice, deletion of Thrb2 did not prevent this reduction (Fig. 6A). Retinal morphology in untreated Thrb2 −/− mice was not different from that in the wildtype (Fig. 6A). PNA labeling on retinal whole mounts showed that cone density in the wild-type mice after NaIO 3 treatment was reduced by about 37%, compared with the untreated controls. However, cone density in Thrb2 −/− mice after NaIO 3 treatment was reduced by 16% only, compared with untreated Thrb2 −/− controls (Fig. 6B). Cone density in the untreated Thrb2 −/− mice was not different from that in the wild-type (Fig. 6B). Treatment with NaIO 3 induced a large increase in the number of TUNELpositive cells in the wild-type mice. However, deletion of Thrb2 did not significantly reduce the number of TUNEL-positive cells (Fig. 6C).

Deletion of THR abolished/diminished NaIO 3 -induced geneexpression upregulation in the RPE and retina
To explore the mechanisms underlying THR signaling suppressioninduced protection, we examined expression of the genes involved in oxidative-stress responses, apoptosis/necroptosis pathways, and inflammatory responses. Wild-type, Thrα1 −/− , and Thrb −/− mice received a single injection of NaIO 3 (30 mg/kg, i.p.) at P30, and were then analyzed for gene expression in the RPE and retinas at 1 day post NaIO 3 treatment. NaIO 3 treatment significantly induced expression of these genes in the RPE (Fig. 7) and retina (Fig. 8) of the wild-type mice, and deletion of Thrα1 or Thrb abolished or significantly suppressed the upregulation of the gene expression induced by NaIO 3 (Figs. 7 and 8).
Treatment with THR antagonist MLS reduced ARPE-19 cell death and hRPE cell death after NaIO 3 treatment The involvement of THR in NaIO 3 -induced RPE damage/cell loss was also examined in an in vitro cell culture model. The ARPE-19 cells and hRPE cells cultured in RPMI-1640 medium were treated with NaIO 3 at 5 and 10 mM for ARPE-19 cells and 20, 30, and 40 mM for hRPE cells in the absence and presence of various concentrations of the THR antagonist MLS for 24 hours [39,40], and were then analyzed for cell viability by MTS assay. MLS is a Fig. 6 Deletion of Thrb2 protected cones but not rods from cell loss induced by NaIO 3 . Retinal morphology, photoreceptor-layer integrity, and loss of photoreceptors were evaluated by light microscope and morphometric analysis in Thrb2 −/− and wild-type mice at 3 days post NaIO 3 injection. Cone density was evaluated by PNA labeling on retinal whole mounts, and photoreceptor apoptosis was evaluated by TUNEL assay. A Shown are representative light microscopic images of H&E-stained retinal sections, and corresponding quantitative analysis of ONL thickness in the dorsal and ventral regions. RPE retinal-pigment epithelial, ONL outer nuclear layer, INL inner nuclear layer, IPL inner-plexiform layer. B Shown are representative confocal images of PNA labeling on retinal whole mounts, and corresponding quantitative analysis. C Shown are representative confocal images of TUNEL labeling on retinal sections and corresponding quantitative analysis. Data are represented as means ± SEM for 4-17 mice per group (*p < 0.05, ***p < 0.001, compared with wild-type mice treated with NaIO 3 ).
member of the methylsulfonylnitrobenzoate-containing series and it inhibits THR's interaction with the coactivator steroid-receptor coactivator 2 and antagonizes T3-activated transcription [39]. The experimental data showed that treatment with NaIO 3 concentration-dependently reduced viability of the ARPE-19 cells and hRPE cells, and treatment with MLS preserved viability of these cells in a concentration-dependent manner ( Supplementary  Fig. 4).

DISCUSSION
In the previous study, we have shown that treatment with antithyroid drug nearly completely preserved RPE and photoreceptors from cell damage/death in mice (C57BL/6) treated with NaIO 3 [27], and reversed upregulation of the genes involved in cellular-stress responses and cell death. The present work evaluated the effects of deletion of Thrα1, Thrb, or Thrb2 to understand the involvement of THRs and the contribution of the different receptor subtypes. We show that deletion of Thrα1, Thrb, or Thrb2 protected RPE and photoreceptors from cell damage/ death caused by NaIO 3 treatment, and significantly suppressed NaIO 3 -induced gene-expression alterations. In addition, treatment with THR antagonist effectively reduced NaIO 3 -induced cell death of ARPE-19 cells and hRPE cells in a cell culture system.
It is worth mentioning that a single administration of NaIO 3 (i.v. or i.p.) selectively induces RPE/retinal damage in experimental animals (wild-type C57BL/6) with more severe damage in the central and middle regions of the RPE/retina and less damage in the periphery [27,35,36]. The TH-receptor knockout mice showed a similar phenomenon, e.g., more severe damage in the center, as  that in wild-type mice. It is not known at this time how the central area of the RPE/retina is more severely affected, and the phenomenon merits further investigation.
Deletion of Thrα1 protected RPE, cones, and rods, and suppressed upregulation of the stress-/death-response genes in the RPE and retina. This is consistent with the broad expression of Thrα1 in the RPE and retina ( [30,31], also see Supplementary Fig.  2), and the functional role of this receptor subtype [38,41]. Of note, the heterozygous deletion of Thrα1 did not achieve a protection, suggesting that the remaining 50% of the receptors are able to fulfill the regulation. The regulation of THRA1 in the viability of RPE and photoreceptors has never been documented. This is the first study to evaluate the regulation/contribution of THRA1 signaling in the RPE and retina using mouse models. We show that THRA1 was involved in NaIO 3 -induced RPE and retinal degeneration and deletion of Thra1 effectively protected RPE and photoreceptors.
Deletion of Thrb (resulting in deletion of both Thrb1 and Thrb2) protected RPE, rods, and cones, and suppressed upregulation of the stress-/death-response genes in the RPE and retina. This is consistent with the broad expression of Thrb in the RPE and retina ( [30,31], also see Supplementary Fig. 2), and the functional role of these receptor subtypes [38,41]. Because THRB2 is expressed only in the cones in the retina [29,32,33], and Thrb2 deletion has been documented in cone protection against T3-induced cell death [10] and in mouse models of LCA and achromatopsia [16], we also included Thrb2 −/− mice in this study to learn more about Thrb2 deletion-associated protection in an NaIO 3 -induced mouse model of AMD. Deletion of Thrb2 protected RPE and cones, similar to that in mice with Thrb deletion. As mentioned above, the deletion of Thrb implies that both the THRB1 and THRB2 isoforms are missing. Based on the nature of the deletion, if both THRB1 and THRB2 isoforms are involved, we expect to see more/additive protection in Thrb −/− mice than in Thrb2 −/− mice. In this study, however, we did not see additive protection in the RPE and cones in Thrb −/− mice, compared with that in Thrb2 −/− mice. The protection levels against NaIO 3 -induced damage/cell death in the RPE and cones were not different between Thrb −/− mice and Thrb2 −/− mice (see Figs. 3 and 5 for RPE, and Figs. 4B and 6B for cones). These observations may suggest that the protection in the RPE and cones in Thrb −/− mice is mainly mediated by deficiency of THRB2. As mentioned above, the role of THRB2 in cone opsin expression and T3-induced cone death has been well documented. The findings from the present work provided evidence showing the role of THRB2 in a different model of cone degeneration (NaIO 3induced cone death). The regulation of THRB/THRB2 in the RPE is little understood, and to our knowledge, this work for the first time demonstrates a role of THRB/THRB2 signaling in RPE stress/ damage.
Rods were protected in Thrb −/− mice but not in Thrb2 −/− mice (see Fig. 4A and 6A). This observation is consistent with the previous findings showing that THRB2 is present only in cones in the retina [29,32,33], and may suggest a critical contribution of THRB1-mediated damage of rods in the model. Interestingly, deletion of Thrb2 protected cones and RPE, despite the presence of THRB1 in these cell types. These observations may suggest a predominant contribution of THRB2-mediated damage in these cell types, although the contributions of THRB1 cannot entirely be excluded. Structurally, THRB1 and THRB2 isoforms differ in their Nterminus; codons 1-94 of THRB1 are encoded by exons that are not present in THRB2; and THRB1 is identical to THRB2 in the C-terminus [7,42]. How such structural differences contribute to the functional variation is not known at this time, and merits further exploration. Together, our data support the view that THRB/THRB2 signaling regulates cell damage/death of the RPE after oxidative-stress challenge, and that in the retina, it is likely that THRB1 signaling regulates cell damage/death of the rods, whereas THRB2 signaling regulates cell damage/death of the cones. Nevertheless, more definitive information about the role of the THRB1 isoform would come only from the use of Thrb1 −/− mice, and this could be our next-step effort.
It should be noted that the protective effects of Thra1 and Thrb2 deletion were also observed in mice at relatively older ages (see Supplementary Fig. 1 for 7-month-old Thra1 −/− mice, and Supplementary Fig. 3 for 17-month-old Thrb2 −/− mice). However, the protection from older Thra1 −/− and Thrb2 −/− mice was somewhat reduced, compared with that in the young mice (see Fig. 1A, B for 1-month-old Thra1 −/− mice and Fig. 5A, B for 1month-old Thrb2 −/− mice). More severe damage in older mice after NaIO 3 treatment, as reported previously [27] and as shown in the present study (see Supplementary Figs. 1 and 3), may contribute to reduced protection. It might also be related to a reduced protection capacity in the older mice, which merits further investigation.
Owing to the broad regulation of TH signaling, deletion of Thra1 or Thrb has effects on many tissues in experimental animals. Examples of Thra1-deletion effects include impaired cardiovascular/heart function [43], abnormal body temperature [44], impaired development of brain tissues [45] and bone tissues [46], and impaired hearing system/auditory function [47]. Examples of Thrb deletion include neuronal-behavior defects/cerebellar defects [48], altered metabolism [49], and deafness [50]. Deletion of Thrb2 specifically causes cone defects [13]. Serum T3 and T4 levels in the knockout mice are elevated by about 0.5-1.5-fold, compared with that in the wild-type mice [13,47,51]. Although we cannot absolutely rule out the possibility, it is less likely that an unrelated preconditioning in these knockout-mouse lines affects the retinaldegeneration phenotype/the effects of the receptor deletion. The RPE and retinal morphology/integrity are not different between the untreated wild-type mice and the receptor-knockout mice at the ages studied (see Figs. 2, 4 and 6).
In summary, this work demonstrated the involvement of THR signaling in cell damage/death of the RPE and photoreceptor after oxidative-stress challenge. The work also provided insights into the regulation/contribution of the different THR subtypes in cell viability. Findings from this work support a role of THR signaling in the pathogenesis of AMD and the view of targeting THR signaling locally in the retina for protection of RPE and photoreceptors in dry AMD.

Treatment of NaIO 3
NaIO 3 treatment was performed as described previously [37]. Briefly, mice received a single injection of NaIO 3 (30 mg/kg, i.p.) at P30 or other ages as indicated, and were then analyzed for RPE and photoreceptor damage/cell death at 3 days post NaIO 3 injection.
Eye preparation, immunofluorescence labeling, confocal microscopy, and retinal morphometric analysis The RPE whole mounts were prepared for immunofluorescence labeling. Briefly, eyes were enucleated and fixed in 4% paraformaldehyde (PFA, Polysciences, Inc.) for 1 h at room temperature, followed by removal of the cornea, lens, muscles, and retina. The RPE sheets (the sclera-choroid-RPE sheets) were then fixed in 4% PFA for another 1 hour at room temperature, followed by washing (PBS, 5 min, 3x) and blocking with 10% FBS in 0.5% Triton X-100 in PBS for 1 hour at room temperature. The RPE sheets were then stained with Alexa Fluor® 594 phalloidin (1:40) for 30-45 min at room temperature and DAPI (1 ng/mL) for another 30 min at room temperature, followed by washing (PBS, 5 min, 2x). The RPE whole mounts were made by transferring the sheets onto the slides, followed by mounting with Hard medium (H-1500, Vector Laboratories).
The retinal whole mounts and cross sections were prepared for immunofluorescence labeling, as described previously [11]. For retinal whole-mount preparations, eyes were enucleated, marked at the superior pole with a green dye, and fixed in 4% PFA for 30-60 min at room temperature, followed by removal of the cornea and lens. The eyes were then fixed in 4% PFA in PBS for 4-6 h at room temperature, and retinas were isolated and the superior portion was marked for orientation with a small cut. For retinal paraffin sections, eyes were enucleated (the superior portion of the cornea was marked with green dye prior to enucleation) and fixed in Prefer (Anatech Ltd.) for 25-30 min at room temperature. Sections (5-µm thickness) passing vertically through the retina (along the vertical meridian passing through the optic-nerve head) were prepared using a Leica microtome (Leica Biosystems), and were used for hematoxylin and eosin (H&E) staining. For retinal cryosections, eyes were fixed in 4% PFA for 1 hour, and the cornea and lens were then removed, followed by fixing the eye cups in 4% PFA for 3 hours. The eye cups were then soaked in graded concentrations of sucrose overnight at 4°C. After being embedded with an optimal cutting-temperature (OCT) compound, 5-μm retinal sections were prepared using a Thermo Scientific CryoStar NX70 Cryostat. Prior to blocking on retinal sections, antigen retrieval was performed in 10 mM sodium citrate buffer (pH 6.0) for 30 min in a 70°C water bath. Retinal whole mounts or sections were blocked with Hanks' balanced salt solution containing 5% BSA and 0.5% Triton X-100 for 1 h at room temperature or overnight at 4°C. Peanut-agglutinin (PNA) immunohistochemistry was performed using biotinylated PNA (1:250) and streptavidin-Cy3 (1:500).
Low-magnification images were taken under the Olympus MVX10 dissection microscope equipped with Image-Pro 6.3 software (Media Cybernetics, Inc.). High-magnification images were taken with 60X or 100x objectives on the FV1000 confocal laser-scanning microscope equipped with FluoView imaging software (Olympus, Melville). ImageJ software (https://imagej.net/) was used to analyze the damaged area on the RPE whole mounts. For quantification of RPE cell numbers and RPE nuclear numbers, images from four quadrants in the central, middle, and peripheral regions were counted and normalized to the number in onesquare millimeter. Evaluation of cone density on retinal whole mounts was performed as described previously [11,52]. For retinal morphometric analysis, retinal cross sections stained with H&E were used for morphometric analysis to evaluate ONL integrity/rod survival, as described previously [11,53].

TUNEL assays
Terminal deoxynucleotidyltransferase dUTP nick-end labeling (TUNEL) was performed on paraffin-embedded retinal sections, using an in situ celldeath fluorescein-detection kit (Sigma-Aldrich, Catalog#: 11684795910), as described previously [54]. Immunofluorescence signals were imaged using an Olympus FV1000 confocal laser-scanning microscope. TUNEL-positive cells in the outer nuclear layer passing through the optic nerve were counted and averaged from at least three sections per eye, from 3 to 5 mice per condition.

RNA isolation and quantitative real-time PCR
The mouse RPE cells were isolated as described [55]. Total RNA preparation and reverse transcription were performed as described previously [56]. The gene encoding the mouse hypoxanthine guanine phosphoribosyl transferase 1 (Hprt1) was included as an internal control. Supplementary Table 1 shows the primers used. The quantitative real-time 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 ΔΔCt method, as described previously [56].

RNAscope in situ hybridization
RNAscope in situ hybridization was applied to examine the expression/ localization of mRNA levels of Thra1 and Thrb in the retina, as described [57,58]. The assays were conducted using RNAscope® 2.5 HD Detection Reagent-Red Kit (Advanced Cell Diagnostics, Catalog #: 322360), as per the manufacturer's instructions. Briefly, cryosections of the mouse retinas (5 μm) were hybridized with the target probes for Thra1 (Catalog #: 531731, Advanced Cell Diagnostics) and for Thrb (Catalog #: 544331, Advanced Cell Diagnostics) at 40°C for 2 hours, with negative and positive controls (Catalog #: 31004 and Catalog #: 313911). Thra1 probe targets the region between 1836 and 2336 of the Thra1 mRNA (NM_178060.4). Thrb probe targets the region between 32 and 461 of the Thrb mRNA (NM_001113417.1), with potential detection of both Thrb1 and Thrb2. The slides were then counterstained with 50% hematoxylin blued with 0.02% ammonia water, dried, and mounted, and images were acquired using an Olympus microscope.

ARPE-19 and hRPE cell culture and drug treatment
ARPE-19 (ATCC, Manassas, VA) cells and hRPE cells (kindly provided by Dr. Goldis Malek at Duke University) were cultured in RPMI-1640 medium (ATCC) with 10% FBS, as described [59,60]. To examine the effects of THR antagonist MLS on NaIO 3 -induced cell death, cells were cultured in RPMI-1640 medium with 10% FBS for 24 hours and were then treated with NaIO 3 at various concentrations in the absence and presence of MLS for another 24 hours, followed by MTS assay for evaluation of cell viability.

MTS assay
The cell-viability/-proliferation assay (MTS) was performed using One Solution Cell Proliferation Assay kit (CellTiter 96 ® AQueous One Solution, Promega, Madison, WI, USA), as per the manufacturer's instruction. The results of MTS assay were obtained by measuring absorbance at 490 nm with a fluorescence-plate reader (Molecular Devices, Sunnyvale, CA, USA). All assays were performed in triplicate and experiments were repeated three times.

Statistical analysis
The results are expressed as means ± SEM of the number of mice. Power analysis was performed to choose the sample size. The analysis indicates that a sample size of 3-6 mice/group for evaluations of retinal degeneration in the mouse retinas will provide at least 80% power (1-β) for a two-sided, two-sample t-test at a 0.05 alpha level. One-way ANOVA was used to analyze for significance within sets of data, and two-tailed Student's t-test was used for differences between two groups of data. Differences were considered statistically significant when p < 0.05. Statistical tests for every figure are justified as appropriate. Data were analyzed and graphed using GraphPad Prism ® software (GraphPad Software, San Diego, CA).