Minireview: The Role of Nuclear Receptors in Photoreceptor Differentiation and Disease

Abstract

Rod and cone photoreceptors are specialized sensory cells that mediate vision. Transcriptional controls are critical for the development and long-term survival of photoreceptors; when these controls become ineffective, retinal dysfunction or degenerative disease may result. This review discusses the role of nuclear receptors, a class of ligand-regulated transcription factors, at key stages of photoreceptor life in the mammalian retina. Nuclear receptors with known ligands, such as retinoids or thyroid hormone, together with several orphan receptors without identified physiological ligands, complement other classes of transcription factors in directing the differentiation and functional maintenance of photoreceptors. The potential of nuclear receptors to respond to ligands introduces versatility into the control of photoreceptor development and function and may suggest new opportunities for treatments of photoreceptor disease.

Vision depends upon photon capture by an array of rod and cone photoreceptor cell types that are generated in precise patterns over the retina during development. Furthermore, once produced in mammals, these cells must endure for a lifetime in the face of stress-inducing hazards such as daily light exposure and high rates of metabolic activity. Several families of transcription factors, including for example, homeodomain and basic helix-loop-helix proteins, are critical for the development and homeostasis of photoreceptors, as has been discussed elsewhere (1–3). In this review, we consider how nuclear receptors, a family of ligand-regulated transcription factors, contribute to photoreceptor differentiation, survival, and disease.

Nuclear receptors share a general structure consisting of a centrally located DNA-binding domain and a C-terminal ligand-binding domain with associated transcription activation domains (4). A unique feature of nuclear receptors is their potential to respond to lipophilic ligands, which opens up new perspectives on the influence of hormonal or other ligands on photoreceptor differentiation and function. The retina is enriched in nuclear receptors but it is noteworthy that most of the receptors identified are so-called orphans without known ligands. These findings have prompted additional questions about unidentified ligands that remain to be discovered and an emerging interest in synthetic ligands that might eventually find applications as agents for countering retinal disease.

We first introduce the rod and cone photoreceptor types found in mammals and a current model for the transcriptional control of photoreceptor differentiation. We highlight the roles of nuclear receptors at early and late stages in the life of a photoreceptor, based largely on genetic studies in mice. Finally, we consider the role of ligands for nuclear receptors and the involvement of nuclear receptors in retinal disease.

Photoreceptor diversity and vision

Rods, which contain the photopigment rhodopsin, are highly sensitive and mediate vision at night or in dim light. Cones are less sensitive but are suited to mediating vision in bright light or daylight. Cones also facilitate color vision, a property conferred by distinct opsin photopigments with sensitivity to different regions of the visible light spectrum. Thus, the vertebrate retina has one type of rod but several types of cones defined by different opsins. Most mammalian species possess dichromatic color vision, mediated by two cone opsins that respond to medium-long (M, green) or short (S, blue) wavelengths of light. Human color vision is generally trichromatic due to the presence of a third opsin with peak sensitivity to longer (L, red) wavelengths of light (5).

The immature retina not only generates photoreceptor diversity but also distributes rod and cone types in particular patterns over the retinal surface. These patterns differ among species, reflecting adaptation of the visual system to varied lifestyles and habitats (6, 7). For example, in the human retina, S, M, and L cones occur in a mosaic pattern and are concentrated in and around the fovea, a region that lacks rods (8). In mice, S and M opsins are distributed in opposing gradients in cones over the superior-inferior plane of the retina, and cones are interspersed throughout with more numerous rods (Fig. 1A).

Fig. 1.

Photoreceptor phenotypes caused by nuclear receptor mutations. The mouse retina is represented in a simplified flat surface view to show the patterning of rods (pink), M cones (green), and S cones (blue). A, In wild-type mice, cones express M and S opsins in opposing gradients across the superior-inferior (dorsal-ventral) surface of the retina, as indicated to the left of the retina. Superior regions contain M cones and inferior regions, S cones. In mice, cones in midretinal regions express varying amounts of both opsins, although these are not shown here for simplicity. In wild-type mice, the ratio of rods to cones is 97:3. B, Thrb2^−/− mice deleted for TRβ2 fail to produce M cones, and all cones instead shift to an S opsin identity. Thrb2^−/− mice represent a model of blue monochromatic color blindness. C, Rorb^−/− mice, deleted for RORβ fail to produce rods and instead produce an excess of primitive S cone-like cells. D, Esrrb^−/− mice, deleted for ERRβ, produce rods, M cones, and S cones but progressively lose rods at adult ages and represent a model for rod-selective degeneration.

The neural retina is formed from multipotent retinal progenitor cells (Fig. 2). Progenitor cells progressively exit the cell cycle, beginning at around midgestation and continuing until the first postnatal week in mice, to produce precursors that are committed to differentiate as photoreceptors as well as other retinal cell types (9, 10). Homeodomain and basic helix-loop-helix transcription factors as well as extracellular signals are often implicated in these early steps in specifying retinal cell fates (1, 10). Cone precursor generation is completed by the time of birth, although M and S opsin patterning is not yet established. Rod precursor generation is more protracted with a peak occurring in the first postnatal week. Postnatally, these immature cones and rods express photopigments, differentiate morphologically, and facilitate visual function by about 2 wk of age in mice.

Fig. 2.

Nuclear receptors and photoreceptor differentiation in mice. Nuclear receptors (genes in blue type) act at several early and later stages of photoreceptor differentiation. In a current model of photoreceptor differentiation, multipotent, proliferative progenitors produce photoreceptor precursors under the control of Otx2 homeodomain factor and other poorly understood factors. These photoreceptor precursors are directed to a rod fate by leucine zipper factor NRL and nuclear receptors RORβ and NR2E3. If the NRL-NR2E3 pathway fails to reach a threshold of activity, the precursor becomes a cone by default. In cones, TRβ2 induces M opsin and suppresses S opsin, depending on thyroid hormone levels and location of the cone on the retina. In the absence of NRL or TRβ2, a photoreceptor precursor differentiates by default as an S cone. Many other factors cooperate to promote photoreceptor differentiation, including transcriptional cofactors, local extracellular signaling factors, and systemic hormonal and vitamin signals. Levels of RA and thyroid hormone can be regulated within the retina by ligand-metabolizing enzymes. Note: 1) Gene locations indicate stages where functions are reported, not necessarily where a gene is first expressed; 2) Expression patterns at early stages are not precisely defined for all genes; some genes may be expressed in dividing progenitor cells and/or early postmitotic cells; 3) Some genes may act at several stages in a cell lineage.

Photoreceptor phenotypes resulting from nuclear receptor mutations

Figure 1 illustrates some of the roles of nuclear receptors in photoreceptor differentiation and homeostasis in mice. Figure 1B depicts a pronounced developmental shift between M and S cone fates caused by deletion of thyroid hormone receptor TRβ2 such that M opsin fails to be expressed, and all cones instead express S opsin. Figure 1C shows a different shift between rod and S cone fates caused by deletion of retinoid-related orphan receptor RORβ. These and other findings discussed below have revealed a surprising degree of plasticity between rod, M cone, and S cone fates during development. Figure 1D indicates a distinct role for estrogen-related orphan receptor β (ERRβ) in the maintenance of the adult retina. ERRβ-deficient mice lose rods in later life and provide a model of retinal degenerative disease. Another orphan receptor NR2E3 (sometimes known as PNR) controls both the differentiation and maintenance of photoreceptors. Table 1 lists additional photoreceptor phenotypes caused by nuclear receptor mutations that are discussed in greater detail below.

Table 1.

Photoreceptor phenotypes associated with nuclear receptor mutations

Gene	Product	Mouse photoreceptor phenotype	Human genetic disease
Nr2e3 (Pnr)	Orphan nuclear receptor	Retinal degeneration, distorted photoreceptor layer; rods express some cone genes in rd7 and knockout strains. (15, 60)	Enhanced S cone syndrome, Goldman-Favre syndrome, reduced rod function with hypersensitive blue cone function; retinitis pigmentosa. (58, 59, 80)
	NR2E3
	Photoreceptor nuclear receptor, PNR
Nr2e1 (Tlx)	Orphan nuclear receptor	Retinal degeneration, distorted photoreceptor layer; defective retinal astrocytes and vasculature in knockout and fierce strains. (36–39)	Not known
	TLX
	Tailless-related receptor
Nr1a2 (Thrb)	Thyroid hormone receptor β	Cone fate shift. Lack M cones; all cones instead express S opsin. (12)	Syndrome of resistance to thyroid hormone; symptoms include rare color visual defects (not systematically studied) (82)
	TRβ
Nr3b2 (Esrrb)	Estrogen-related orphan receptor β	Retinal degeneration; loss of rods in adult retina (61)	Autosomal recessive deafness (visual studies not yet reported) (83)
	ERRβ
Nr1f1 (Rora)	Retinoid-related orphan receptor α	Partial reduction of M and S opsin levels in staggerer mice. (23)	Not known
	RORα
Nr1f2 (Rorb)	Retinoid-related orphan receptor β	Rod fate shift. Lack rods, gain primitive S cones; photoreceptors lack outer segments. (28, 31)	Not known
	RORβ
Nr2b3 (Rxrg)	Retinoid X receptor γ	Extended S opsin distribution in cones in superior retina. (26)	Not known
	RXRγ
Nr2b1 (Rxra)	Retinoid X receptor α	Defective early morphogenesis of neural retina, optic nerve and lens. Defects exacerbated by deletion of RAR isoforms. (43, 44)	Not known
	RXRα
Nr1b2 and Nr1b3, (Rarb and Rarg)	Retinoic acid receptors	Combined deletion causes retinal dysplasia and degeneration; defective pigmented eipithelium and choroid. (42)	Not known
	RARβ and RARγ
Nr2f2 and Nr2f1 (Coup-TF2 and Coup-TF1)	Orphan nuclear receptors COUP-TF2 and COUP-TF1	• Combined deletion causes defective early morphogenesis of retinal pigmented and neural epitheiia; coloboma, microphthalmia. (40)	Not known
		• Individual deletions cause extended S opsin distribution in cones in superior retina. (27)

Nuclear receptors and photoreceptor diversity

Photoreceptors originate from common pools of progenitor cells; however, accumulating evidence indicates that the diversity of rod and cone types is generated in postmitotic photoreceptor precursors with developmentally plastic properties (Fig. 2). Based on loss- and gain-of-function studies (11–14), we proposed that TRβ2 and a basic motif-leucine zipper factor, NRL, control a two-step switch that directs a generic photoreceptor precursor to a rod, M cone, or S cone fate (3). The initial step to become a rod is taken if NRL attains a threshold of expression or activity. If this threshold is not reached, the precursor retains a default S cone fate. In a subpopulation of cones, TRβ2 in the presence of thyroid hormone induces M opsin, thereby setting M and S opsin patterning over the retina.

After the initial fate decision, nuclear receptors further consolidate the rod, M cone, or S cone outcome. In rod precursors, NRL induces expression of the orphan receptor NR2E3. NR2E3 and NRL then cooperate to suppress cone genes (15–18) and, together with the cone-rod homeobox protein CRX and another orphan nuclear receptor NR1D1 (or REVERBα), activate rod gene expression (19–21). In cone precursors, S opsin is induced by RORβ cooperatively with CRX (22), RORα, and the spalt factor SALL3 (23, 24). M opsin induction is triggered by TRβ2 (12) and augmented by RORα (23).

How the selective expression of M or S opsin in cones in different retinal regions is achieved is still a mystery (Fig. 1A). This selection requires not only the induction of one opsin but the concomitant suppression of the other opsin. The principle of spatial segregation of opsins is widely conserved although the patterning and extent of opsin exclusion varies between species. Nuclear receptors are involved in opsin selection in mice, including retinoid X receptor (RXR)γ, which may form heterodimers with TRβ2 to suppress S opsin in the superior retina (25, 26). The orphan receptors chicken ovalbumin upstream promoter transcription factor (COUP-TF)1 and COUP-TF2 also suppress S opsin expression (27). An intriguing possibility is that thyroid hormone, retinoids, or other ligands modulate the activity of multiprotein complexes to impose a quantitatively precise, reciprocal regulation on the M and S opsin genes in different retinal regions.

The induction of NRL and TRβ2 is a crucial step preceding the generation of photoreceptor diversity. RORβ is one factor that induces NRL, such that Rorb^−/− mice, like Nrl^−/− mice, lack rods and instead gain primitive S cone-like cells (Fig. 1C) (11, 28–30). The versatile functions of the Rorb gene in different retinal cell lineages may be mediated by two receptor isoforms (RORβ1 and RORβ2) and by specific interactions with other transcription factors (22, 28, 31). Other studies have begun to unravel signals such as bHLH, notch and bone morphogenetic proteins that control the expression of TRβ2 (32, 33) and COUP-TF (27) in photoreceptor precursors. How such cellular signals and nuclear receptors cooperate in determining photoreceptor patterning is an area of active investigation.

Nuclear receptors in early retinal differentiation

The mechanisms that determine whether an undifferentiated progenitor cell continues to proliferate or exits mitosis to produce a photoreceptor or other retinal cell type have attracted much investigation (2, 10)(Fig. 2). As a practical benefit, the elucidation of these mechanisms may suggest ways to stimulate regeneration of photoreceptors in retinal degenerative disease or ways to direct stem cells to form photoreceptors for possible therapeutic purposes (34).

One factor required for the production of photoreceptor precursors is the orthodenticle homeobox 2 protein OTX2 (35)(Fig. 2) but nuclear receptors may also influence this process (Table 1). The tailless orphan nuclear receptor TLX is expressed in the retina and brain, where it maintains self-renewal of neural progenitor cells. Tlx-deficient mice display several ocular abnormalities, partly due to imbalances in rates of progenitor proliferation and differentiation (36–38). TLX can act as a repressor and may control self-renewal by regulating cell cycle genes such as Pten, an inhibitor of proliferation. TLX also suppresses Pax2, a homeodomain gene involved in early eye development (39) and influences the differentiation of photoreceptors, astrocytes, and the retinal vasculature (36, 38).

Other nuclear receptors are expressed in the early embryonic retina. Combined deletion of COUP-TF1 and COUP-TF2 causes coloboma, microphthalmia, and malformation of the optic vesicle with a shift in the fates of pigmented epithelial and neural retinal progenitor cells (40). These defects may result partly from dysregulation of Otx2 and Pax6, which are key genes for early retinal development and for the generation of photoreceptors and other cell types. Deletion in mice and overexpression studies in rat retinal explants have also suggested that RORβ influences progenitor proliferation (31, 41). Retinoic acid receptors RARβ and RARγ control the differentiation of the periocular mesenchyme and retinal pigmented epithelium and also influence, possibly indirectly, the formation of the photoreceptor layer in the neural retina (42, 43). RXRα, together with RARα and RARγ, directs the early morphogenesis of the retina, lens, and optic nerve (44). RAR and RXR isoforms in the neural retina may potentially mediate functions in photoreceptors at later stages of differentiation (25, 45).

A few earlier studies explored the role of retinoid, steroid, and thyroid hormone ligands in promoting the differentiation of photoreceptors in embryonic retinal explants (46, 47). There is scope for further investigation of such ligands based on more recent insights into which nuclear receptors are present in the retina. Evaluation might also be extended to synthetic ligands for orphan receptors.

Local metabolic control of ligand concentration

Differentiation in the retina, as in other tissues, is guided by cell-to-cell signals within the local environment. This is typified for example, by notch signaling, which restrains retinal progenitors from differentiating prematurely as photoreceptors (33)(Fig. 2). Another form of local or paracrine-like control within the retina is exerted by enzymes that metabolize ligands for nuclear receptors. Although the vascular circulation is the primary source of endocrine and vitamin-derived ligands, such as thyroid hormone and retinoids, enzymes that activate or inactivate the ligand in the retina can serve as a valve between the systemic signal and the local signal that acts on a target cell during differentiation.

In the circulation, T₃, the biologically potent form of thyroid hormone, is less abundant than T₄, a relatively inactive form of the hormone. In tissues such as brown adipose tissue, an activating enzyme, type 2 deiodinase, amplifies levels of T₃ by conversion from T₄, to meet increased demands for ligand (48). The mouse retina expresses relatively little type 2 deiodinase but is enriched in type 3 deiodinase, a thyroid hormone-inactivating enzyme. Inactivation of type 3 deiodinase causes loss of cone photoreceptors by cell death in neonatal mice (49, 50), suggesting that type 3 deiodinase has a critical role in limiting exposure to ligand to maintain cone survival. Expression of type 3 deiodinase diminishes at older ages when T₃ is necessary to induce M opsin in more mature cones (50, 51).

Retinoic acid (RA) is derived from vitamin A in the diet. Several RA-metabolizing enzymes produce RA asymmetrically over the superior-inferior plane of the retina, which has been thought to reflect a role for RA in the early morphogenesis of the eye (52, 53). Recent studies also suggest a role for these enzymes at postnatal stages of cone maturation. The expression of RA-synthesizing (Raldh3) and RA-catabolizing (Cyp26a1 and Cyp26c1) enzymes in the immature mouse retina is under the control of the VAX2 homeodomain factor. Vax2^−/− pups display reduced expression of S opsin and an abnormally extended distribution of M opsin in the inferior retina (45). These abnormalities may arise from impaired generation of RA because administration of RA to pregnant mice partially rescues S opsin expression in neonatal Vax2^−/− progeny. In retinal explants, RA may induce not only photoreceptor markers (46) but also cell death (54). Thus, stringent control over the local concentration of both T₃ and RA ligands may be required to balance differentiation and cell death rates in retinal development.

Paradox and potential: ligands for orphan nuclear receptors

The majority of nuclear receptors identified in the retina, including COUP-TF1, COUP-TF2, RORβ, RORα, NR2E3, TLX, and ERRβ are orphans without known physiological ligands and thus present a paradox of how they serve as receptors. Some of these orphan proteins may yet become adopted if they are found to bind natural high-affinity ligands. Another speculation is that lower affinity but higher abundance ligands may stabilize some orphan receptors in an active conformation (55). For example, RORβ, a constutively active orphan receptor, possesses a large ligand-binding pocket that fortuitously accommodates small hydrophobic molecules in vitro (56). Conceivably, endogenous ligands could include hydrophobic components present in specific retinal cell lineages.

There is considerable interest in synthetic ligands for orphan nuclear receptors as an approach to treatments for disease. High-throughput screens have identified small molecules that bind to NR2E3 (57), which is mutated in human retinal diseases (58, 59) (Fig. 3). One agent (compound 11a) was further shown to stimulate changes in gene expression patterns in mouse retinal explants (60). Another candidate orphan receptor for such approaches is ERRβ, which is required for long-term survival of rods in adult mice (61). ERRβ does not bind natural estrogens but responds to a synthetic agonist DY131 and inverse agonists including 4-hydroxytamoxifen (62). Transactivation of the rhodopsin promoter by NRL, CRX, NR2E3, and ERRβ is enhanced by DY131 and suppressed by 4-hydroxytamoxifen. In tests on retinal explants from Crx^−/− mice, a model of retinal degeneration, DY131 reduced cell death rates and recovered some rhodopsin expression (61).

Fig. 3.

Natural and artificial ligands for nuclear receptors in the retina. The examples shown represent the natural ligand for TRβ2 (triiodothyronine) and synthetic agonists for NR2E3 (11a) and ERRβ (DY131).

COUP-TF2 and RORβ have been reported to bind RA in vitro (63, 64), leading to speculation that RA might act not only through RAR isoforms but also through orphan receptors if RA can attain sufficiently high local concentrations in vivo. RORα and NR1D1 are capable of binding cholesterol and heme, respectively (55, 65). However, the biological relevance of these interactions in the retina remains unknown. Questions about the potential for some orphan receptors to bind diverse small molecules and whether such molecules act as stabilizing factors or true ligands that regulate receptor conformation and activity have been discussed elsewhere (55, 66).

Regarding potential therapeutic applications, high-throughput screening may help to identify more selective, higher affinity compounds for use in targeting orphan receptors in the retina. Human retinal degeneration associated with photoreceptor dysfunction is a frequent cause of untreatable forms of vision loss (3, 67). Thus, the investigation of novel ligands may open new avenues for treatment of retinal as well as metabolic or other diseases (66).

Photoreceptor survival in the mature retina

Mammals, unlike some avian and amphibian species, lack the natural ability to regenerate photoreceptors (7, 34). Human photoreceptors must maintain their function for a lifetime. Consequently, the dysfunction or death of photoreceptors that occurs in many degenerative diseases leads to irreversible loss of vision. Furthermore, photoreception makes high metabolic demands that require stringent homeostatic control throughout life to limit the risks of damage from oxidative stress. Thus, there is great interest in understanding the mechanisms that maintain photoreceptor viability during adulthood and aging.

Numerous inherited, environmental, or age-related factors can compromise photoreceptor survival in adulthood. Recent studies suggest that nuclear receptors contribute to the maintenance as well as the earlier differentiation of photoreceptors. In adult mice, ERRβ is critical for long-term maintenance of rods (61). In ERRβ-deficient mice, rods and cones are formed but rods are lost at older ages (Fig. 1D). By 1 yr of age, almost all rods are lost whereas cones largely survive. Evidence indicates that the Esrrb gene is a target of NRL (68), suggesting that NRL-dependent pathways control the maintenance of mature rods as well as the earlier differentiation of rods discussed in previous sections (17, 69). Moreover, synthetic agonists for ERRβ have the potential to restimulate rod genes in mouse models of retinal degeneration (61).

Transcriptional regulation by nuclear receptors in the retina

Understanding the function of nuclear receptors in photoreceptors requires elucidation of the underlying gene-regulatory networks. Although changes in retinal mRNA expression patterns have been identified in mice with mutations in nuclear receptors (17, 28, 38, 69) and in vitro in thyroid hormone-treated retinoblastoma cells that have some cone characteristics (70), much remains to be learned about the primary target genes and mechanisms of transcriptional regulation. Nuclear receptor-binding sites have been identified upstream of, or within, several retinal target genes. These sites generally contain one or two copies of the consensus binding motif for nuclear receptors, 5′-AGGTCA-3′ (4), often correlating with whether the receptor may bind as a monomer, such as RORβ, or dimer, such as NR2E3 (Table 2) (16, 22, 29, 30, 38–40, 71).

Table 2.

Examples of retinal target genes and response elements for nuclear receptors

Nuclear receptor	Target gene	Response element	Type of element and response (Reference)
RORβ	Opn1sw		Monomeric 5′-AT-rich; positive (22)
	−295 site 1	AATATAGGTCA
	−242 site 2	GATTGAGGTCA
RORβ	Nrl		Monomeric 5′-AT-rich; positive (29, 30)
	−785	GAAAATGTAGGTCA
TLX	Pax2		Monomeric; negative (39)
	−82	GACAAGTCATCC
TLX	Pten		Monomeric; negative (38)
	−1681	GATAAGTCACTT
NR2E3	Idealized element	AGa/gTCAAAa/ga/gTCA	Dimeric DR1; negative (16)
NR2E3	Prph2		Dimeric DR1; positive (71)
	−15,766	AAGTCACAATTCA
COUP-TF	Pax6		Dimeric DR1; negative (40)
	3′- UTR	TGTTCACAGTCCA

Notes: The consensus motif AGGTCA is underlined. Location of the element relative to the transcription start site of the mouse target gene is noted (in bp, where defined); 3′-UTR, 3′-untranslated region; DR1, direct repeat with 1-bp spacing. The idealized NR2E3 binding site was determined by repeated selection for in vitro binding of NR2E3 to DNA.

The context of a binding site in the target gene contributes to determining how a nuclear receptor interacts with other factors to produce cell-specific activation or repression of the gene. For example, RORβ binds target DNA elements specifically but is a weak transcriptional activator. However, RORβ cooperates with more potent homeodomain activators CRX and OTX2 to stimulate expression of the Opn1sw (S opsin) and Nrl genes, respectively (22, 29, 30). Combinatorial actions may also be mediated by protein-protein interactions. COUP-TF need not always bind directly to target genes but may interact with SP1 factor to enhance the activation of Otx2 in the retina (40). Likewise, RARβ activates its own RARβ2 promoter in the Rarb gene through an autoregulation loop that is enhanced by TLX (72).

Nuclear receptor activity in photoreceptors also depends upon interactions with cofactors and chromatin-modifying complexes. Thus, NR2E3 and a retinal corepressor RetCoR form multiprotein complexes with histone deacetylases (73). TLX may regulate cell cycle control genes, in part, by interaction with the atrophin 1 corepressor and histone lysine-specific demethylase LSD1 (38, 74). Another type of coregulator is Pias3, an E3 SUMO ligase that contributes to rod differentiation by sumoylation of NR2E3 (75) and/or NRL (76). Pias3 may also modify M and S cone differentiation in mice (77).

Genome-wide analyses using chromatin immunoprecipitation (ChIP) and next-generation sequencing (ChIP-Seq) will aid in delineating direct target genes of nuclear receptors in photoreceptors. ChIP analysis combined with expression profiling has revealed distinct functions for NR2E3 during photoreceptor differentiation (17, 18, 69). Although there is debate about a role for NR2E3 in cone proliferation (78), evidence indicates that NR2E3 and NRL, a direct inducer of NR2E3 (17), function together primarily to suppress cone genes and induce rod genes in early photoreceptor precursors, thereby tipping the differentiation fate from a cone to a rod, as discussed above.

Nuclear receptors and photoreceptor disease

Although mutations in many nuclear receptor genes produce photoreceptor or ocular phenotypes in mice, only a few, so far, are known to be associated with human retinal disease (Table 1). The most prominent example is the orphan receptor gene NR2E3, which was first shown to be involved in enhanced S-cone syndrome, an autosomal-recessive condition with marked increases in blue cone sensitivity (58, 79). Subsequently, mutations in NR2E3 have been identified in clinical presentations with a spectrum of overlapping phenotypes including photoreceptor dysfunction and/or retinal degeneration (59, 80, 81). The underlying disease mechanism is believed to be the aberrant expression of S opsin and other cone genes in rods in the absence of NR2E3 function (18, 79).

Mutations in the human THRB gene in the syndrome of resistance to thyroid hormone are associated with varied endocrine and neurological symptoms. Although photoreceptor function has not been investigated specifically, monochromatic color blindness has been noted in a rare case of recessive inheritance of this syndrome (82). Mutations in human NR3B2 (encoding ERRβ) are associated with autosomal-recessive deafness (83) but studies of vision have not yet been reported.

Ligand-deficiency disorders

As predicted from the phenotypes of receptor mutations, deficiency of T₃ or RA ligands also produces photoreceptor defects in mice. Congenital hypothyroidism in mice produces cone opsin abnormalities (84–86) related to those caused by deletion of TRβ2 (12) and in rats may produce other retinal defects (87). However, there has been little specific investigation of photoreceptor function in human congenital thyroid diseases. A few studies have reported that infants with potentially inadequate exposure to thyroid hormone during development in utero exhibit lower visual contrast sensitivity (88) and that premature infants with low thyroid hormone levels display impaired visual processing (89).

Vitamin A deficiency in rodent embryos causes eye malformations consistent with a lack of RA ligand at early developmental stages because RAR deletions cause similar defects (43). Furthermore, rat and human fetal retinal explants respond to RA with changes in photoreceptor differentiation markers (46, 47). However, elucidation of the role of RA ligand at later stages of photoreceptor maturation is complicated because vitamin A, the precursor of RA, is also the precursor of the retinal chromophore required by photopigments for phototransduction activity. Thus, vitamin A deficiency at postnatal ages produces night blindness and photoreceptor degeneration because of loss of chromophore. Future studies might aim to distinguish more specifically the role of RA ligand at later stages of photoreceptor maturation and maintenance. Recently, it was suggested that RA ligand contributes to cone opsin patterning in neonatal mice (45).

Concluding remarks

Nuclear receptors are critical for the function of endocrine, metabolic, and other systems (4). In this review, we suggest that nuclear receptors are also essential in the visual system, in the differentiation and homeostatic maintenance of photoreceptor cell types.

Why do nuclear receptors feature so prominently in the visual system? It may be pertinent that the developmental process that generates photoreceptor diversity and patterning in the retina is notably plastic and has been adapted differently across vertebrate species (6, 7). Great variations exist in the proportions and patterning of rods and cone types found in different species. Even among mammals, photoreceptor patterning varies considerably, reflecting the crucial role of vision for the survival of species in widely divergent habitats. We may speculate that nuclear receptors are well suited to the control of this pliable developmental system because of their inherently versatile functions. The ability of nuclear receptors to interact with other classes of transcription factors and chromatin-modifying complexes as well as their sensitivity to external ligands may allow these proteins to act as key integrators of diverse signals in this dynamic cellular environment.

Finally, although we have focused on the mammalian retina, much may be learned from other vertebrate or invertebrate species (90). Many of the nuclear receptors and ligands discussed above are also found in the amphibian, avian, or fish retina (91–94).

NURSA Molecule Pages^†:

• Nuclear Receptors: PNR | TLX | TR-β | ERR-β | ROR-α | ROR-β | RXR-γ | RXR-α | RAR-β | RAR-γ | COUP-TFII | COUPTFI.