Given its key function in mouse rod development, it is reasonable to question whether Nrl is of similar importance in the human retina. Interestingly, most of the identified human mutations in NRL are associated with adRP [124–127]. However, Nishiguchi et al. (2004) [128] recently reported two mutations that are likely to encode loss-of-func- tion alleles, and patients with these mutations have clumped pigmentary retinal degeneration and some preservation of blue cone function, both of which are indicative of an analogous function for NRL in human photoreceptor differentiation as in mice.

Nuclear Receptors

Nuclear receptors constitute a large family of transcription factors that function as activators on direct binding of a ligand [129] (see Pharmacological Reviews, Volume 58, Issue 4, for an extensive overview). Those in which a physiological ligand is not known or the receptor is suspected to lack a ligand are listed as “orphan receptors.” Most nuclear receptors homodimerize or heterodimerize in various combinations and are associated with large multisubunit complexes that can regulate transcription in positive and negative manners. Because of their diversity in structure and function and their direct regulation by environmental cues, nuclear receptors influence a large number of cellular responses and gene expression programs. Several nuclear receptors are implicated in photoreceptor development; they include the thyroid hormone receptor beta 2 (TRβ2), several retinoid X receptors (RXRs) and retinoic acid receptors (RARs), and the orphan receptors Nr2E3 (PNR, RNR), and retinoid orphan receptor beta (RORβ). TRβ2, RXRγ, and RORβ are implicated in regulating the timing and selection of opsin expression in developing cones [91, 130–134]. Because of space limitations, however, we focus here on the proposed roles of nuclear receptors on rod development, and much of the current effort has been aimed at understanding the role of Nr2E3 in this process.

Nr2E3

Patients with enhanced S-cone syndrome (ESCS) suffer from progressive vision loss that is initially characterized by night blindness and variability in L and M cone- (L/M cone-) mediated vision and ultimately followed by rod photoreceptor degeneration. What sets this disease apart from most other forms of inherited retinal degenerations is an increased sensitivity to blue light, which is mediated by the S cones. Jacobson and colleagues first described this condition in the early 1990s and noted the similarities in severely affected patients to those with Goldman-Favre syndrome [135–137]. Subsequent psychophysical and electrophysiological studies led to the prediction that patients with ECSC had an unusually high number of S cones, possibly at the expense of the other cones and rod photoreceptors [138–140].

It is now known that mutations in the NR2E3 gene can cause ESCS, Goldman-Favre syndrome, and clumped pigmentary retinal degeneration, and histopathological studies confirmed an overabundance of S-opsin-expressing cells and a lack of rods [141–145]. Expression analysis of NR2E3 during human fetal development showed the transcript and protein to be expressed in a pattern most consistent with postmitotic differentiating rod precursors [146].

rd7 is a naturally derived mutation in Nr2E3, and it serves as an excellent model for the human NR2E3 mutation [141, 147]. As in humans, the rd7 allele is a recessive

allele, although the exact nature of the mutation was not clear until recently [148]. Nevertheless, the long-standing interpretation that the rd7 mutation is a null still holds, and homozygotes have the ECSC phenotype and gradual photoreceptor degeneration. It is also well established that the enhanced S-cone activity is due primarily to changes that occur during retinal development. However, two models exist to explain how Nr2E3 functions in photoreceptor development; they are described next.

One model states that Nr2E3 functions to prevent the excess production of cone photo receptors from RPCs by restricting their proliferation after the normal interval of cone photoreceptor generation [149, 150]. This model makes several predictions. One prediction is that a subset of late embryonic RPCs is committed, or at the very least, specified, to give rise to cones and is actively suppressed from doing so by Nr2E3. Although there is no evidence for a cone-restricted RPC from lineage studies in mice [151], ectopic proliferation is observed in the rd7 retina and it is suggested that these proliferative cells are either S-cone precursor cells that are normally postmitotic or are RPCs fated to give rise to cones [149, 150]. Another prediction is that Nr2E3 is expressed in RPCs after the normal period of cone production. While one study reports Nr2E3 expression in proliferating RPCs at the expected time [149], several others argued that Nr2E3 expression is initiated in postmitotic rod precursors [146, 152, 153], which are also generated during this interval. Expression analyses in zebrafish and mouse suggested that Nr2E3 is transiently expressed in postmitotic immature cones [152, 153], although one study argued that Nr2E3 expression in cones is not transient [149]. Another prediction is that any role for Nr2E3 in rod development is independent of its role in restricting cone production. The evidence given for this is that the size of the S-cone cell population in the rd7 retina does not increase at the expense of L/M cones or rods (the reduction in rods is proposed to occur by apoptosis [149]). While there is general agreement that the L/M cone population is not affected in the rd7 retina, there is disagreement on whether the extra S-cone cells originate from a source of ectopically proliferating cells or from cells normally fated to become rods, which brings us to the next model.

The second model states that Nr2E3 functions as a differentiation factor for postmitotic rod precursors and does this in two ways: by repressing the cone gene expression program by acting as a transcriptional repressor and by facilitating the rod gene expression program by acting as a transcriptional activator. The evidence for this model is compelling on several levels. First, regardless of whether Nr2E3 is expressed in RPCs or cones, it is clear that it is expressed in postmitotic rod photoreceptors. Second, in the rd7 retina, many cone messenger RNAs (mRNAs) are highly upregulated and expressed throughout the ONL, which is largely composed of rods as determined by the expression of rod genes [149, 152–154] and by the presence of initially normal scotopic electroretinogram (ERG) responses [147]. Third, several studies have documented that Nr2E3 can function as a transcriptional repressor or activator, and this depends on the context of the cis-acting DNA regulatory sequences and on the protein complexes with which Nr2E3 is associated [150, 152, 153, 155]. It is important to note, however, that many rod genes are not significantly downregulated in the rd7 retina, and this could be due to the persistent expression of Crx, Nrl, and other photoreceptor transcription factors [149, 152–155].

Along these lines, comparison of the Nrl−/− and rd7 photoreceptor phenotypes shows considerable overlap, especially with respect to increases in the number of photoreceptors with

S-cone gene expression characteristics. One difference, however, is that in the Nrl−/− retina, the expression of rhodopsin and other rod-specific phototransduction genes is significantly downregulated. Correlated with this is a lack of Nr2E3 expression [116].

Cheng et al. (2006) [156] developed transgenic mice expressing Nr2E3 from the Crx promoter and S-opsin promoter and analyzed the photoreceptor phenotypes in normal and Nrl−/− mice. Nr2E3 is activated in early-stage rod and cone precursors (photoreceptor precursor stage) when driven by the Crx promoter and is activated in S-cone precursors by the S-opsin promoter. The primary findings of this work are that Nr2E3 is sufficient to block cone gene expression in the presence or absence of Nrl, and it can activate rod gene expression in the absence of Nrl, although not completely. The sum of these observations supports the idea that Nr2E3 suppresses the expression of the cone phenotype, at the same time promoting the rod phenotype in combination with Nrl.

Are these models mutually exclusive? While certain aspects of these models are discordant and need to be resolved, it is possible that both could be correct in their essential aspects. It is important to note that not all photoreceptors in the rd7 mouse are functional cones. In fact, it is estimated that the increase in S-cone number is from approximately 1% to 3% of the total retinal cells [154, 157]. Second, although it is clear that cone mRNA expression is inappropriately activated in the rd7 rod population, cone protein expression appears to be limited [149] and S-opsin protein has not been found in rhodopsin-expressing cells [154]. While the evidence is strong that Nr2E3 has a role in suppressing the cone and promoting the rod gene expression programs in developing rods, the source of the supernumary S cones, whether from cells that underwent ectopic proliferation or from cells normally fated to be rods, still needs to be resolved.

Retinoic Acid/Retinoic Acid Receptors

Vitamin A derivatives (retinoids) regulate many aspects of vertebrate development and homeostasis. They bind to and activate nuclear receptor proteins of the steroid and thyroid hormone receptor superfamily, resulting in derepression and transcriptional activation of target genes. Homoor heterodimers of RARs and RXRs are activated by retinoids with high affinity, whereas unsaturated fatty acids such as docosahexaenoic or linoleic acids function as low-affinity ligands. Retinoic acid-synthesizing enzymes and receptors are present in the vertebrate eye during photoreceptor development [131, 158–162].

Several decades ago, Dowling observed that vitamin A is essential for development and maintenance of photoreceptors in cats [163]. Culture experiments in chick and rodents showed subsequently that the primary metabolite retinoic acid stimulates photoreceptor differentiation and survival [164–166]. In zebrafish, retinoic acid treatment resulted in precocious rod differentiation, whereas inhibition of synthesis using citral led to a delay in photoreceptor differentiation [167]. Similarly, retinoic acid injections into pregnant rats late during pregnancy accelerated rod differentiation after birth [168]. Another ligand for RXRs, docosahexaenoic acid (DHA), is a major structural lipid of retinal photoreceptor outer segment membranes. In rat retinal cultures, DHA promotes maturation of CRX-positive photoreceptor precursors [169]. Thus, these studies indicate that RAR signaling is sufficient and required for photoreceptor differentiation in vertebrates.

At which stage does retinoic acid act to regulate rod development? In vitro, double-labeling experiments with BrDU and rod markers indicated that retinoic acid acts on proliferating RPCs to increase the proportion of rods. This effect was accompanied by a decrease of amacrine cells, suggesting that retinoic acid instructs the rod fate in RPCs by compromising the generation of other cell types [164]. Conversely, in zebrafish, treatment with retinoic acid increased rhodopsin expression, but did not alter cell numbers, arguing against an effect on cell fate [170]. However, Khanna et al. showed that retinoic acid can directly transactivate the Nrl promoter and promotes expression of NRL in mammalian retinal cells in vitro [171]. Thus, retinoic acid indeed could act as an early extrinsic signal to instruct RPCs to adopt the rod fate. In agreement with this, Li et al. observed an increase of CRX expression in retinoic acid-treated human retinoblastoma cells [172]. In sum, these observations suggest that the outcome of retinoic acid action on rod development is species dependent.

Interestingly, in mouse retinal explants physiological concentrations of retinoic acid have also been shown to induce rod-specific cell death when RPE is present [173]. Furthermore, since retinoic acid can induce expression of cone and rod photoreceptorspecific genes in retinoblastoma cell lines [172, 174–176], it might exert differential effects that activate distinct mechanisms in both cell types: alone to regulate rod differentiation or in combination with TRβ2 to regulate cone differentiation [131].

EXTRACELLULAR FACTORS AND SIGNAL

TRANSDUCTION PATHWAYS

Wnt/Frizzled Pathway

Wnts are secreted glycoproteins that bind to Frizzled transmembrane receptors and regulate multiple developmental processes such as embryonic patterning, cell polarity, cell fate determination, and proliferation. Many Wnts, Frizzleds, and pathway modulators are expressed during retinal development, and more recently, Wnt/Frizzled signaling has been shown to regulate several aspects of eye development in vertebrates [177, 178].

Wnt-inhibitory factor (WIF-1) is a secreted antagonist of the Wnt/b-catenin pathway that is present in the extracellular matrix (ECM) and binds to Wnts in the extracellular space [179, 180]. It contains a signal sequence, a WIF domain, and five epidermal growth factor (EGF) repeats. Hunter et al. (2004) [180] identified WIF as an ECM component in the developing mouse retina. During the period of photoreceptor differentiation, WIF-1 is expressed in the subretinal space and interphotoreceptor matrix. Other potential partners of the Wnt pathway that are expressed appropriately are Wnt4, LRP6, and Frizzled-4. Furthermore, WIF-1 and Wnt4 appear to interact with each other [180]. In rat retina dissociated cultures, WIF-1 decreases, and Wnt4 promotes rod differentiation, suggesting that these components could modulate rod differentiation and maintenance. However, Frizzled-4 mutants do not show a photoreceptor phenotype, and Wnt/β-catenin activity in the photoreceptor layer of postnatal mice is not detected in Wnt/reporter mice [181, 182] (our unpublished observations). Furthermore, conditional inactivation of b-catenin in the mouse does not result in an obvious change of rhodopsin immunoreactivity [183].