NinaB Is Essential for Drosophila Vision but Induces Retinal Degeneration in Opsin-deficient Photoreceptors*

In animals, visual pigments are essential for photoreceptor function and survival. These G-protein-coupled receptors consist of a protein moiety (opsin) and a covalently bound 11-cis-retinylidene chromophore. The chromophore is derived from dietary carotenoids by oxidative cleavage and trans-to-cis isomerization of double bonds. In vertebrates, the necessary chemical transformations are catalyzed by two distinct but structurally related enzymes, the carotenoid oxygenase β-carotenoid-15,15′-monooxygenase and the retinoid isomerase RPE65 (retinal pigment epithelium protein of 65 kDa). Recently, we provided biochemical evidence that these reactions in insects are catalyzed by a single enzyme family member named NinaB. Here we show that in the fly pathway, carotenoids are mandatory precursors of the chromophore. After chromophore formation, the retinoid-binding protein Pinta acts downstream of NinaB and is required to supply photoreceptors with chromophore. Like ninaE encoding the opsin, ninaB expression is eye-dependent and is activated as a downstream target of the eyeless/pax6 and sine oculis master control genes for eye development. The requirement for coordinated synthesis of chromophore and opsin is evidenced by analysis of ninaE mutants. Retinal degeneration in opsin-deficient photoreceptors is caused by the chromophore and can be prevented by restricting its supply as seen in an opsin and chromophore-deficient double mutant. Thus, our study identifies NinaB as a key component for visual pigment production and provides evidence that chromophore in opsin-deficient photoreceptors can elicit retinal degeneration.

Photoreceptors are sensory neurons with specialized lightsensitive compartments, the outer segments of vertebrate cones and rods and the rhabdomeres of insects. These structures are orderly stacks of photosensitive plasma membrane that house the phototransduction machinery. Light sensitivity is mediated by visual pigments, the major protein constituent of these membranes. Visual pigments are G-protein-coupled receptors that consist of an integral membrane protein, referred as the opsin, and an 11-cis-retinylidene chromophore covalently linked to a lysine residue through a Schiff base linkage (reviewed in Ref. 1). Visual pigment biogenesis is a complex process that involves the transport of the opsin through the secretory pathway to photosensitive membranes and the metabolic supply of the chromophore. It is now clear that mutations resulting in ineffective visual pigment biogenesis and pigment regeneration after a bleach are an important cause of retinal degeneration and blindness in humans (for recent reviews see Refs. 2,3).
Drosophila have long served as an animal model to analyze visual pigment biogenesis and study the pathology of associated blinding diseases (for recent review see Ref. 4). The first mutations linked to retinal degeneration were noted in the ninaE gene (neither inactivation nor afterpotential mutant E) (5,6). ninaE encodes the opsin moiety of the flies' major visual pigment (Rh1) 2 and is expressed in photoreceptors R1-R6 (7,8). In ninaE null mutants, rhabdomeres degenerate and their membranes involute into the cytoplasm (9). This phenotype develops light independently and does not require downstream factors of the G-protein signaling cascade (4). Therefore, an essential structural role for opsin in building rhabomeric membranes during photoreceptor morphogenesis has been proposed (9,10). In addition to opsin, the chromophore is also crucial for visual pigment biogenesis. In flies, defects in chromophore production prevent normal expression and stability of the opsin (11,12). In chromophore deficiency, the opsin virtually disappears from photoreceptor cells (12). However, the converse situation to examine the consequences of chromophore deficiency for photoreceptor morphogenesis has not been studied in great detail in the fly.
The chromophore (11-cis-retinal in vertebrates and 11-cis-3-hydroxy-retinal in flies) is a retinoid (C20) derived from carotenoids (C40) by oxidative cleavage and trans-to-cis isomerization of double bonds. Analysis of blind Drosophila mutants has identified several genes required for synthesis of the chromophore (13)(14)(15)(16). Among these mutants, ninaB flies cannot convert carotenoids to retinoids (14,16). The ninaB gene encodes a protein that belongs to a family of structurally related non-heme iron oxygenases (17) that include the mammalian carotenoid-oxygenase BCMO1 and the retinoid isomerase RPE65 (retinal pigment epithelium protein of 65 kDa) (17). These enzymes catalyze two key steps in the pathway for chromophore production, the conversion of carotenoids to retinoids by oxidative cleavage (18) and the trans-to-cis isomerization of the C10,C11 double bond of retinoids (19 -22). Dietary supplementation with preformed retinoids can compensate for BCMO1 deficiency (19), but RPE65 is essential for chromophore production and regeneration (23). In humans, mutations in RPE65 cause Leber congenital amaurosis or the less severe, later onset retinitis pigmentosa (reviewed in Ref. 3).
In flies, an 11-cis configuration of retinoids is also essential for visual pigment biogenesis (12,25,26). Although preformed retinoids such as all-trans-retinal support visual pigment production only in the presence of light, carotenoids can promote visual pigment production light independently (25). Biochemical analysis shows that NinaB from the moth Galleria mellonella combines the activities of a carotenoid oxygenase and a retinoid isomerase in a single polypeptide (25). In this isomerooxygenase reaction, carotenoids such as zeaxanthin are lightindependently converted to one molecule of 11-cis-and one molecule of all-trans-3-hydroxy-retinal. This finding suggests that NinaB and carotenoids are essential for visual pigment biogenesis and photoreceptor morphogenesis. Therefore, we here analyzed the role of NinaB in this process in Drosophila photoreceptors that are amenable for genetic, biochemical, and structural analyses.
For supplementation experiments, flies were raised on carotenoid-depleted media containing the tested supplement, e.g. 500 M all-trans-retinal, 5 M ␤-carotene, or 5 M zeaxanthin. Early pupae were then transferred to fresh vials containing carotenoid-depleted media. Two days after eclosion, adult flies were used for Western blot analysis for Rh1. For investigation of the role of all-trans-retinal in retinal degeneration, freshly eclosed ninaD P245 ; ninaE 17ol flies were transferred to media supplemented with ethanol alone (control) or ethanol plus alltrans-retinal (500 M) for 2 days. Carotenoids were purchased from Wild (Heidelberg, Germany); all-trans-retinal was purchased from Sigma.
Light Avoidance Assay-Foraging third instar larvae of ninaB 360d and Oregon R strains were tested in a light versus dark preference test at 750 lux as described (28). Experiments were conducted in triplicate using 20 larvae for each test.
Estimation of mRNA Levels by Semiquantitative RT-PCR Analyses-RT-PCR analyses for ninaB were carried out as previously described (29).
Extraction of Carotenoids and Retinoids and HPLC Analyses-Lipid compounds from flies were extracted and HPLC procedures were carried out as described previously (17,25,29). All extractions were carried out under a red safety light.
Western Blot Analyses-Western blot analyses of proteins obtained from fly heads were carried out as previously described (29).
Assessment of Retinal Toxicity-Drosophila S2 cells were cultured in Schneider's Drosophila medium (Invitrogen) as previously described (29). Cells were suspended in medium for 24 h at 24.5°C containing either 1% DMSO alone, or either alltrans-retinol or all-trans-retinal, each dissolved in DMSO. Cell viability was estimated by measuring the percentage of cells that failed to stain with Trypan Blue (0.4%, Sigma-Aldrich) after exposure according to the manufacturer's protocol. Cells were then loaded into a hemocytometer chamber, and numbers of non-viable (stained) and viable (non-stained) cells were counted under a microscope.
Test for Enzymatic Activity-For heterologous expression, expression vectors for Drosophila and Galleria NinaB (25), were transformed in the Escherichia coli strain XL1-blue (Stratagene, La Jolla, CA). Bacteria were grown at 28°C to an A 600 of 0.5. Protein expression was induced with L-arabinose (0.02% w/v) for 8 h. Bacteria were harvested by centrifugation and broken with a French press in a buffer containing 50 mM Tricine/ KOH (pH 7.6, 4°C), 100 mM NaCl, and protease inhibitor mixture (Roche Molecular Diagnostics, Pleasanton, CA). The crude protein extract was centrifuged at 20,000 ϫ g for 10 min, 4°C. Reactions were run in 2-ml Eppendorf tubes in a volume of 100 l in 50 mM Tricine/KOH (pH 7.6) with 3% (w/v) 1-S-octyl-␤-D-thiopyranoside in the presence of 20 M zeaxanthin. Retinoids and carotenoids were extracted as previously described (25). Samples were dissolved in methanol/acetonitrile (50:50, v/v) and subjected to HPLC separation on a C18 column (Zorbax Eclipse XDB, 4.6 ϫ 1500 m, 5 m, Agilent Technologies, Santa Clara, CA) followed by mass spectrometry. Detection of 3-hydroxy-retinaloxime was acquired using an LXQ high throughput linear ion trap mass spectrometer (Thermo Scientific, Waltham, MA) interfaced with an atmospheric pressure chemical ionization source and series 1100 HPLC system (Agilent Technologies) consisting of a vacuum degasser, a binary pump, an autosampler with cooled sample tray, thermostatically controlled column compartment, and diode array detector.
Electron Microscopy and Immunogold Labeling-For ultrastructural analyses, Drosophila heads were fixed in 2.5% glutar-aldehyde in cacodylate buffer, washed with cacodylate buffer, postfixed in 1% osmium tetroxide, and incubated in 1% uranyl acetate. For immunogold labeling, fly heads were fixed in 4% paraformaldehyde, 0.2% glutaraldehyde. Postfixation was done with 8% formaldehyde. All samples were dehydrated by an ethanol series and embedded in Epon. Sections for immunogold labeling were probed with a polyclonal anti-Rh1 rabbit antibody and a secondary anti-rabbit antibody conjugated with gold particles. Microscopy was accomplished with a Philips CM10 electron microscope.
F-actin Staining-Fly heads were fixed in 2% paraformaldehyde for 1.5 h and incubated with increasing sucrose concentrations up to 50% overnight. Then 10-m cryosections were cut, transferred to poly-L-lysine-coated coverslips, and postfixed in 2% paraformaldehyde. Blocking was done in phosphate-buffered saline containing 1% bovine serum albumin and 0.1% saponin for 2 h. Sections were incubated in phalloidinrhodamine overnight at 4°C. Samples were mounted and viewed by confocal microscopy (Leica TCS-SP).

RESULTS
NinaB Is Essential for Larval Vision-In Drosophila, the first light-sensitive structure develops during embryogenesis. The larval Bolwig's organ (BO) entrains the clock of pacemaker neurons and also mediates larval photophobic behavior. The BO is composed of 12 photoreceptors, 8 expressing Rh6 and 4 expressing Rh5 (30). It is not clear how the BO acquires chromophore for the production of these visual pigments. To analyze the role of NinaB in this process, we performed whole mount in situ hybridization with staged embryos. Staining for ninaB mRNA was found at embryonic stage 13 and persisted until stage 16 ( Fig. 1A). At stage 13, staining was localized at the outermost posterior end of the head capsule, directly at the epidermis and in the immediate vicinity of the developing optic lobes. At later developmental stages, the stained cells lost contact to the outer surface of the epidermis and migrated anteriorly toward the midline. At stage 16, stained cells were located on both sides of the pharynx. This spatiotemporal expression pattern of ninaB mirrors the organogenesis of the BO (31). To provide evidence for a role of NinaB in BO function, we analyzed the consequences of NinaB deficiency for larval vision. Larvae are photophobic, and their behavior can be tested by a previously described assay (28). On a test plate that is half illuminated and half dark, wild-type larvae migrated preferentially to the dark side of the test plate. In contrast, ninaB mutant larvae were randomly distributed between the illuminated and dark areas (Fig. 1B). Thus, ninaB mutants showed no photophobic behavior indicating that NinaB is essential for larval light perception.
NinaB Expression Depends on Presence of the Eyes-Later in the life cycle, the Drosophila compound eye develops from the eye imaginal disk. This process is completed with the establishment of visual pigments. ninaB is expressed at late pupal stages coincident with the expression of opsin (29). Based on a genome-wide microarray analysis in sine oculis (so) flies that lack compound eyes, ninaB mRNA is ϳ150-fold enriched in normal eyes (32). RT-PCR analysis confirmed that ninaB mRNA levels were highly reduced in so flies ( Fig. 2A). Some residual ninaB mRNA expression in so flies is likely attributable to the eyelets that develop from the larval BO and prevail in this mutant (33). In accordance with highly reduced ninaB expression, HPLC analysis showed that so flies lacked the chromophore but accumulated carotenoid precursors (Fig. 2C). Thus, ninaB expression and chromophore production depend on the presence of compound eyes. To provide additional evidence that expression of ninaB and chromophore production is dependent on eye formation, we used a fly strain (dpp-Gal4Ͼ ϾUAS-eyless) that develops ectopic eyes. These structures have the characteristics of compound eyes, including groups of fully differentiated ommatidia with a complete set of photoreceptor cells that are light-sensitive (27). RT-PCR analysis revealed that ninaB mRNA expression was induced in thoraces of pupae with ectopic eyes but was not detectable in thoraces of wild-type flies (Fig. 2B). Thus, ninaB expression and chromophore production is governed by eyeless and so, major control genes for eye development.
Carotenoids and NinaB Are Essential for Rh1 Production during Compound Eye Development-In Drosophila, retinoid production was eye-dependent (see above), and biochemical studies showed that NinaB combines the activities of vertebrate BCMO1 and RPE65 (25). Therefore, we asked whether carotenoids are mandatory for visual pigment biogenesis in Drosophila. To answer this question, we supplemented larvae with either carotenoids or preformed retinoids. The resulting pupae were then transferred to carotenoid-and retinoid-free medium and kept under a 16-h/8-h light/dark cycle. We then measured Rh1 levels in freshly eclosed adult flies by immunoblot analysis. Mature Rh1 has a molecular mass of 32 kDa, indicative of posttranslational processing and translocation to rhabdomere membranes (12). In this experimental setup, carotenoid-supplemented larvae produced Rh1 (Fig. 3A). In contrast, all-transretinal supplementation did not support Rh1 production (Fig.  3A), indicating that Drosophila lacks a transport system for dietary retinoids to support visual pigment biogenesis.
To provide further evidence for the hypothesis that flies lack a transport system for dietary retinoids, we analyzed the consequences of ectopic ninaB expression by using a UAS/GAL4 expression system (34). We established a fly strain that expressed a UAS-ninaB wild-type transgene under the control of a tubulin-GAL4 driver in the whole fly. HPLC analysis revealed that the carotenoid and retinoid contents of these flies were dramatically reduced (Fig. 3, C and D). Immunoblot analysis showed that these flies also lacked Rh1 (Fig. 3B). This finding suggests that chromophore production outside the eyes reduced carotenoids available for chromophore production in the eyes, thereby impairing visual pigment production.
Zeaxanthin Is the Precursor for Chromophore Production-Flies use 3-hydroxy-retinal as chromophore. We found that both non-hydroxylated and hydroxylated carotenoids supported visual pigment production during compound eye development. Previously, it has been shown that ␤-carotene is con-verted via cryptoxanthin into zeaxanthin when fed to larvae, indicating that hydroxylated carotenoids are the direct precursor for chromophore (35). However, tests for enzymatic activity with zeaxanthin and recombinant NinaB failed in a previous study (17). We therefore applied the modified test conditions that were recently established for Galleria NinaB (25). We confirmed that recombinant Galleria NinaB efficiently converts zeaxanthin to 3-hydroxy-retinal (Fig. 4, A and B). Recombinant Drosophila NinaB proved to be less active, but mass spectrometry revealed that 3-hydroxy-retinal was produced (Fig. 4C). Thus, ␤-carotene is first converted to zeaxanthin and then cleaved by NinaB to directly yield 3-hydroxy-retinal.
Both Retinoid Cleavage Products of a Carotenoid Are Converted to the Chromophore-Biochemical analysis showed that NinaB converts one half-site of a carotenoid to 11-cis-retinal, whereas the other half-site remains in the all-trans configuration (25). Accordingly, zeaxanthin supplementation resulted in light-independent Rh1 production (Fig. 5A), and HPLC analysis revealed that 11-cis-3-hydroxy-retinal and all-trans-3-hydroxy-retinal existed in equal molar amounts in flies (Fig. 5B). In illuminated flies, this ratio was altered in favor of the 11-cis-3-hydroxy-retinal stereoisomer (Fig. 5B). This increase of 11-cis-3-hydroxy-retinal indicated that the all-trans-retinal cleavage product is converted in a light-dependent manner to the chromophore. A light-dependent pathway for chromophore production from all-trans-retinal has been described for adult flies (12). However, chromophore production from all-trans-retinal relies on an unspecific diffusion rather than on a protein-facilitated transport process. When we applied a solu-  tion of all-trans-retinal directly on the eyes of adult ninaB mutants, Rh1 production was restored in a light-dependent manner (Fig. 5C). Thus, we conclude that light-dependent trans-to-cis isomerization of retinoids is required for the conversion of the all-trans-retinal cleavage product of carotenoids. This light-dependent pathway ensures that both retinoid cleavage products can be utilized for chromophore synthesis.
The Retinoid-binding Protein Pinta Is Required to Supply Photoreceptors with Chromophore-Our analysis showed that Drosophila apparently lack a transport system for retinoids.
However, a retinoid-binding protein, designated Pinta, was identified that is expressed in pigment cells of the eyes (36). Pinta is a homolog of the vertebrate cellular retinal binding protein CRALBP that acts downstream of retinoid isomerase RPE65 (37). Pinta mutants lack Rh1, but the biochemical consequences of the pinta null mutation for chromophore production have not been characterized. Therefore, we determined the retinoid composition in the heads of pinta flies by HPLC analysis. Both 11-cis and all-trans-3-hydroxy-retinal were detectable in pinta flies (Fig. 6B). Additionally, 3-hydroxy-retinols existed in amounts comparable to wild-type flies (Fig. 6B). Immunoblot analysis revealed that, despite production of chromophore, Rh1 levels were highly reduced in this mutant (Fig.  6A). This finding indicates that Pinta is not required for chromophore production. Lack of Rh1 in this mutant rather suggests that Pinta acts downstream of NinaB and facilitates the transport of chromophore from its locus of NinaBdependent production to photoreceptor cells.
Rhabdomeres of the ninaB Mutant Do Not Degenerate-Our studies indicated that visual pigment production essentially depends on NinaB. Indeed, Rh1 was absent in ninaB mutants (Fig. 5C). Because photoreceptors degenerate in mutants that that lack opsin, it has been suggested that Rh1 is of critical importance for photoreceptor morphogenesis (9). Therefore, we asked whether photoreceptors of ninaB mutant flies also underwent retinal degeneration. Surprisingly, the overall structure of photoreceptors in chromophore-deficient ninaB mutants was intact (Fig. 7, A and B). Although some photoreceptors developed two rhabdomeres instead of one, this abnormality is clearly distinguishable from retinal degeneration in opsin-deficient mutants, characterized by disintegration of rhabdomeric membranes (9). To exclude the possibility that opsin was translocated to rhabdomeres in the ninaB null mutant and prevented retinal degeneration, we performed immunostaining for Rh1. No evidence for the presence of opsin at rhabdomeres of ninaB mutants was found (Fig. 7D), whereas opsin was readily detectable in wild-type flies (Fig. 7C). Thus, although ninaB null mutants lack Rh1 at rhabdomeres, these membranes did not degenerate as in opsin-deficient mutants.
Chromophore Deficiency Can Prevent Retinal Degeneration in Opsin Null Mutants-Visual pigment deficiency caused by mutations in either ninaB or ninaE genes have dramatically different consequences for photoreceptors. Because the chromophore is present in ninaE mutants (supplemental Fig. S1), whereas in the ninaB mutants it is not (25), we speculated that retinal degeneration in the opsin deficiency is caused by the chromophore. To provide evidence for this hypothesis, we attempted to suppress retinal degeneration by chromophore deficiency. Because the ninaE and ninaB genes are located close together on the same chromosome, we took advantage of the ninaD mutant. ninaD null mutant flies cannot absorb dietary carotenoids and thus lack chromophore (14). We generated an opsin-and chromophore-deficient ninaE;ninaD double mutant by appropriate crossings. We then compared photoreceptor morphology in ninaE single, ninaE;ninaD double mutants, and wild-type flies by F-actin staining of rhabdomeres. Confocal imaging revealed that rhabdomeres in the double mutant were arranged in regular patterns as in wild-type photoreceptors (Fig. 8, A, B, E, and F). In contrast, photoreceptors of the ninaE single mutant exhibited largely disorganized rhabdomeres (Fig. 8, C and D).
Retinal Can Provoke Cell Damage-Our results with the ninaE;ninaD double mutant showed that lack of chromophore can suppress the degeneration phenotype of the ninaE null mutant. Conversely, this suggests that the presence of chromophore is required for retinal degeneration in the opsin-deficient photoreceptors. To provide evidence that the chromophore is toxic to cells, we treated Drosophila Schneider S2 cells with all-trans-retinal. Trypan blue staining of cells revealed that all-trans-retinal caused cell death in a dose-de- . C, application of all-trans-retinal onto the eyes rescues defects in Rh1 production of the ninaB mutant. Western blot analysis for Rh1 with total protein extracts of fly heads after application of all-trans-retinal directly onto the eyes. All-trans-retinal was dissolved in ethanol (100 M) and applied directly to the compound eyes using a fine brush. Rh1 levels were determined 2 days after application of all-trans-retinal. ninaE 17ol (rh1 null) and wild-type flies reared on standard corn medium were used as controls. Tubulin was used as the loading control. pendent manner (Fig. 9, A-C). Cells treated with the same amount of all-trans-retinol remained viable, suggesting that chromophore toxicity depends on the aldehyde group. To provide in vivo data that the chromophore can cause retinal degeneration, we treated adult ninaE;ninaD double mutants with all-trans-retinal. All-trans-retinal can bypass defects in chromophore production in adult ninaD mutant flies when applied in very high amounts (12,14). After supplementation with alltrans-retinal, flies were kept for 3 days in a light/dark cycle to allow light-dependent chromophore synthesis from this precursor. Photoreceptor ultrastructure was then analyzed by electron microscopy. The all-trans-retinal-treated double mutants developed retinal degeneration. Their rhabdomeric membranes involuted into the photoreceptor cell lumen (Fig. 9E). In contrast, double mutants treated only with solvent displayed intact rhabdomeres (Fig. 9D). These results support the idea that the chromophore is the causative agent for retinal degeneration in opsin-deficient photoreceptors.

DISCUSSION
Successful production of visual pigments is essential for photoreceptor morphogenesis and function. These G-protein-coupled receptors consist of a protein moiety (opsin) and a covalently linked 11-cis-retinylidene chromophore. In mammals, RPE65 catalyzes trans-to-cis isomerization of retinoids, the key step for chromophore production and recycling (3). RPE65 belongs to the same enzyme family as mammalian carotenoid-oxygenases and insect NinaB. Recently, we biochemically demonstrated that NinaB combines the activities of a ret-inoid isomerase and a carotenoid-oxygenase in a single protein (25). Here we show that, instead of retinoids and RPE65 as in mammals, it is carotenoids and NinaB that are essential for visual pigment production in Drosophila. ninaB mutant flies exhibit impaired chromophore production and consequently lack visual pigments. However, they do not show signs of retinal degeneration as described for opsin-deficient photoreceptors. This observation can be explained by our demonstration that retinal degeneration is caused by chromophore itself in opsindeficient ninaE photoreceptors. Retinal degeneration can be prevented by restricting chromophore supply as shown in an opsin and chromophore-deficient double mutant. Thus, our study establishes NinaB as a key component for Drosophila vision and provides evidence that NinaB-dependent chromophore production induces retinal degeneration in opsin-deficient photoreceptors.
Chromophore Production Is Eye-dependent-In vertebrates, BCMO1 converts carotenoids to those retinoids required for vision and gene regulation. The resulting retinoids are then transported, taken up by target tissues, and metabolized to reti-  nyl esters for storage or to retinoic acid for gene regulation. Retinoids in the eyes also are isomerized by RPE65 for chromophore production and regeneration. Drosophila require retinoids only for vision (11), and NinaB combines the activity of its mammalian counterparts in a single protein (25). During embryogenesis, NinaB was expressed in the developing BO, and behavioral testing showed that light reception was impaired in ninaB mutant larvae. HPLC analysis in so flies that lack compound eyes revealed that carotenoids accumulated, but retinoids were absent. RT-PCR analysis additionally indicated that NinaB expression was highly reduced in this mutant. This result is consistent with the outcome of a genome-wide microarray analysis that revealed a strong reduction of ninaB expression in this mutant (32). Additionally, we found that ninaB expression was induced in the thoraces of flies that developed ectopic eyes. These structures, being comprised of fully differentiated light-sensitive photoreceptors, must have produced visual pigments (27). Thus, ninaB expression and chromophore production is restricted to the locus of retinoid action in the eyes.
Previous genetic studies ruled out all retinal cells, including photoreceptors and pigment cells, as the site of NinaB action (16,38). Instead, immunolocalization of a GFP reporter expressed under control of the ninaB promoter indicated neuronal and glial cells of the lamina and medulla of the optic lobes as the locus of NinaB action (16). This observation is compatible with the marked reduction in ninaB expression in so mutant flies, because not only their compound eyes but also the lamina of their optic lobes are greatly reduced (39). Such chromophore production in glial and neuronal cells appears evolutionally conserved. Ganglion cells of the vertebrate retina can also be light-sensitive and express melanopsin (40), the closest vertebrate homolog of insect visual pigments. Additionally, Müller glial cells of the vertebrate retina have been implicated in visual chromophore production for cone photoreceptors (41).
A Carotenoid-dependent Pathway for Chromophore de Novo Synthesis-Carotenoids and retinoids are highly lipophilic molecules, suggesting that specific membrane transporters and binding proteins exist that mediate the transport of these compounds within the body. In Drosophila, specific transporters for carotenoids have been identified that mediate absorption of dietary carotenoids (14,16,29). The ninaD gene encodes a scavenger receptor essential for the accumulation of dietary carotenoids at larval stages (29). During photoreceptor morphogenesis, acquired carotenoids are transported to ninaB-expressing cells in a process that depends on the Santa Maria protein (16). However, the fly genome does not encode molecular components for the retinoid transport systems described in vertebrates, e.g. the serum retinol-binding protein RBP4 and its receptor STRA6 (42,43). Only carotenoids such as zeaxanthin and ␤-carotene consistently support visual pigment production during compound eye development. A previous study showed that ␤-carotene is converted to zeaxanthin, indicating that this xanthophyll is directly cleaved to 3-hydroxy-retinoids (35). Using a recently established protocol (25), we now show that Drosophila NinaB can convert zeaxanthin to 3-hydroxyretinoids. Thus, hydroxylated carotenoids are the direct precursors for chromophore production. In contrast, preformed retinoids failed to support this process in Drosophila, even though all-trans-retinal was supplemented in 100-fold higher amounts relative to carotenoids. This inability to transport retinoids was also evidenced by the phenotype resulting from ectopic ninaB expression. When we expressed UAS-ninaB under a tub-GAL4 driver, ectopic production of 3-hydroxyretinoids reduced the levels of carotenoids available for chromophore production in the eye with consequent impairment of visual pigment biogenesis. Only when applied in very high amounts to chromophore-deficient adult flies can all-trans-retinal be utilized for visual pigment production (4,14). But this effect relies on nonspecific diffusion of these compounds rather than on a protein-facilitated transport process. When we applied a solution of all-trans-retinal directly on the eyes of ninaB mutants, visual pigment production was restored in a light-dependent manner. Physiologically, this light-dependent isomerization of retinoids likely plays a role in the conversion of the all-trans-stereoisomer produced by carotenoid conversion via NinaB. As recently shown, NinaB converts one half-site of a carotenoid to 11-cis-stereoisomer, whereas the other half-site exists in the all-trans configuration (25). Accordingly, 11-cis-3hydroxy-retinal and all-trans-3-hydroxy-retinal should have existed in equimolar amounts in carotenoid-supplemented dark-reared flies. In illuminated flies, this ratio was altered in favor of the 11-cis-3-hydroxy-retinal stereoisomer. The lightdependent pathway for all-trans to 11-cis isomerization ensures that both retinoid cleavage products can be used for chromophore production. Although retinoids are not transported well within the Drosophila body, the retinoid-binding protein Pinta is evidently required for visual pigment production. Pinta is expressed in the pigment cells and is a homolog of the vertebrate CRALBP that acts downstream of RPE65 (36). Previously, Pinta was suggested to support reduction and oxidation of retinoids for visual chromophore production by binding all-trans-retinol (36). However, our biochemical analysis showed that pinta mutant flies showed a similar 3-hydroxy-retinoid composition as wildtype flies and that all-trans-retinol is absent in flies. Thus, the endogenous 3-hydroxy-retinoid that binds to Pinta remains to be determined. Because pinta mutants produced chromophore, but Rh1 levels were highly reduced, Pinta is likely required for the transport of chromophore from its place of NinaB-dependent production to photoreceptor cells.
Coordinated Production of the Chromophore and the Opsin Is Required for Photoreceptor Morphogenesis-In flies, the chromophore is of importance not only for phototransduction but also for targeting of Rh1 to rhabdomere membranes via the secretory pathway (19). Because rhabdomeres of opsin-deficient mutant degenerate (4), we were interested in the consequences of the ninaB null mutation for photoreceptor structure. Surprisingly, we found that NinaB-deficient photoreceptors were structurally intact. In ninaB flies, some photoreceptors developed two instead of one rhabdomere. This abnormality might be attributed to a requirement of Rh1 in the organization of the cortical actin cytoskeleton during rhabdomere formation in photoreceptor cell development (10), but it is clearly distinguishable from retinal degeneration described for mutants that lack the opsin protein (9). Because we used the ninaB 360d null mutant, we can exclude the possibility that residual amounts of chromophore prevented retinal degeneration. We further confirmed by immunocytochemistry that opsin was not transported to rhabdomeres in this mutant. The absence of opsin at rhabdomere membranes is in accord with results from the chromophore-deficient ninaG mutant (15). We conclude that functional Rh1 is not required for structural integrity of rhabdomeres.
Because lack of opsin is not the proximal cause of retinal degeneration, we speculated that chromophore is the causative agent. It has been previously shown that retinal degeneration in the ninaE mutant can be prevented by transient expression of a wild-type ninaE transgene at the late pupal stage (44) when ninaB expression and chromophore production is initiated during eye development (29). A need to coordinate opsin and chromophore production is also evidenced by our finding that photoreceptor degeneration in a ninaE null mutant can be rescued by genetically induced chromophore deficiency. Conversely, supplementation with high amounts of all-trans-retinal of the adult ninaE/ninaD double mutant did result in retinal degeneration. Dose-dependent toxicity of all-trans-retinal was also found in Drosophila S2 cells. No such effect was found when cells were incubated with similar amounts of all-transretinol. These results implicate the aldehyde functional group of the chromophore in this toxicity. This agrees with the known chemistry of retinaldehyde. It can form Schiff base adducts with phosphatidylethanolamine, which in turn can react with a second retinal molecule to produce N-retinylidene-N-retinyl eth-anolamine, which has been implicated in the pathogenesis of certain human retinal diseases (45). In addition to forming toxic adducts, depletion of components such as phosphatidylethanolamine from rhabdomeres may interfere with the structural integrity of these membranes. Finally, reactivity of an aldehyde might also lead to interactions with other cell components, including amino groups of metabolic intermediates, DNA, and proteins. Such toxicity has been recently suggested to contribute to retinal damage in mammals under conditions wherein the all-trans-retinal photoproduct accumulates (46). Accordingly, abolition of chromophore production in mammals prevents light-induced retinal degeneration (47). HPLC analysis revealed that ninaE mutants produced chromophore despite opsin deficiency, although overall retinoid content was reduced as compared with wild-type controls (supplemental Fig. S1). Thus, retinaldehyde toxicity is likely an immediate cause of the retinal degeneration in Drosophila as well.
In summary, our analysis establishes NinaB as a key component for visual pigment production and vision in Drosophila. For chromophore production, carotenoids rather than retinoids must be transported to ninaB-expressing neuronal and glial cells in the lamina of the optic lobes. NinaB then converts carotenoids directly to the chromophore in a combined isomerization-oxygenase reaction. Downstream in the pathway, the retinoid-binding protein Pinta is then required to supply photoreceptors with the chromophore. Toxicity resulting from lack of opsin production is in turn not due to a structural requirement of Rh1 for rhabdomere biogenesis, but rather to the resulting free chromophore. Both ninaE and ninaB are governed by so and ey, major control factors for eye development. This spatiotemporal coordination of chromophore and opsin production optimizes visual pigment biogenesis and protects the structural integrity of photoreceptors from excess chromophore. Thus, our findings open the possibility of using Drosophila mutants to screen for genetic suppressors and drugs that reduce the retinal degeneration and blindness resulting from retinaldehyde toxicity.