RPE65 takes on another role in the vertebrate retina

The high-acuity central vision of humans, and other primates, depends on a region of the central retina called the macula lutea (Latin: yellow spot), containing not only a high concentration of both cone and rod photoreceptors but also a high concentration of xanthophyll carotenoids; hence, its name (1). At the center of the macula is the fovea (Latin: pit) centralis, a region of especially high-acuity vision, containing only cone photoreceptors and Muller glia cells, where, in the human fovea, the density of cones is estimated to reach 1–3 × 10⁵ per square millimeter. In the fovea, there are no overlying interneurons or retinal ganglion cells, as in the rest of the retina. Instead, the axons of the foveal cones are elongated and connect to bipolar cells located on the rim of the foveal pit. Anatomically, these elongated axonal processes form a layer called the fibers of Henle that overlies the cone photoreceptors like a thin cap. It is here that the macular xanthophylls accumulate (2).

It is thought that the protective function of the xanthophylls is twofold: serving as a blue-light filter and as scavengers of reactive oxygen species (ROS), both potentially damaging to cone photoreceptors (3). The electron-rich polyene chains of carotenoids efficiently quench ROS, thereby preventing membrane damage. Three xanthophylls are found in the macula and fovea: two, lutein and zeaxanthin, are abundant in nature, while the third, meso-zeaxanthin, is uncommon elsewhere outside primate and bird retinas (although synthetic meso-zeaxanthin is now commercially available). Both zeaxanthins have a better singlet oxygen-quenching ability than lutein due to an additional conjugated double bond, while meso-zeaxanthin may have the greatest antioxidant ability of the three (4).

Threats to the health of the macula, and vision itself, include various inherited simple Mendelian gene disorders, all relatively rare, but also the far more common condition of age-related macular degeneration (AMD) (5). AMD is a major cause of blindness in older adults, with the risk increasing with age. It is a complex disease, with environmental factors (e.g., smoking) and common and rare variants in >30 genetic loci (e.g., complement/immune-modulatory, lipid metabolism, extracellular matrix genes) contributing to risk and disease pathogenesis (6). A multifactorial disease, AMD not only involves the macular and foveal neural retina but also the underlying retinal pigment epithelium (RPE) and the vascular choriocapillaris. While some surgical and monoclonal antibody therapies are available to treat severe complications (neovascularization), treatment options are limited and cure is currently unavailable.

Epidemiological studies, such as the Age-Related Eye Disease Studies (AREDS and AREDS 2) have found that higher levels of macular xanthophyll intake are correlated with lower risk of developing severe AMD (5) and higher macular pigment optical density (MPOD) is an important phenotypic parameter associated with macular health (1, 3). Accordingly, AREDS 2 proposed preventative measures to mitigate risk of developing AMD, including supplementation with lutein and zeaxanthin, as well as other antioxidants, such as copper and zinc, and ascorbic acid. Xanthophylls are taken up by the gut and transported in the blood (via HDL) to the RPE and taken up into the primate retina by scavenger receptor class B member 1(SCARB1) and CD36 (3). While significant levels of lutein and zeaxanthin, neither of which can be synthesized by animals, are both widely available from dietary sources (e.g., green leafy vegetables, fruits, corn, egg yolks), meso-zeaxanthin, rare in commonly consumed dietary sources, is not (3). Hence, the origin of macular meso-zeaxanthin has been controversial. Some investigators favor a dietary source for meso-zeaxanthin, where trace levels are found in fish skin and egg yolks (7), suggesting that metabolic processes are not necessary to explain its occurrence in the retina, while others hold that it is metabolically transformed in the retina from some other xanthophyll, probably lutein (3). While accumulation of meso-zeaxanthin has been shown to occur in a developmentally regulated manner, first in the RPE/choroid of chicken embryos followed by its appearance in the retina, and nowhere else in the embryo (8), the underlying metabolic mechanism was unknown.

In PNAS, Shyam et al. (9) now reveal that RPE65, a protein highly preferentially expressed at high levels in the RPE and the well-known retinol isomerase of the vertebrate visual cycle (10–12) (Fig. 1A), is also the lutein to meso-zeaxanthin isomerase (Fig. 1B). Although a remarkable and exciting finding that explains an important phenomenon, the concept that RPE65 is the responsible enzyme should not be entirely surprising, given its identity as a divergent member of a superfamily of carotenoid-cleaving oxygenases (CCOs). CCOs, widespread in all kingdoms of life, include apocarotenoid oxygenases (ACOs) in bacteria, epoxycarotenoid-cleaving enzymes required for abscisic acid biosynthesis in plants [NCEDs (viviparous14s [VP14s])], and carotenoid-cleaving enzymes in plants (CCDs) and animals. Vertebrates express three members of the superfamily: 15,15′-β-carotene oxygenase (BCO1), 9′,10′- β-carotene oxygenase (BCO2), and RPE65. All CCOs are nonheme iron (II) oxygenases, with ferrous iron coordinated by four histidines in a seven-bladed beta-propeller chain fold. Crystal structures of the cyanobacterial Synechocystis ACO, vertebrate RPE65 from Bos taurus (13), plant VP14 from Zea mays, and bacterial resveratrol-cleaving dioxygenase NOV1 from Novosphingobium confirm the conservation of overall fold and active site arrangement in all CCO superfamily members. All CCOs, except for RPE65, oxidatively cleave conjugated polyene substrates. In contrast, the major substrate of RPE65 is all-trans retinyl ester. RPE65 performs a concerted O-alkyl cleavage of the all-trans retinyl ester into retinol and fatty acid, concomitant with carbocation-mediated (or possibly radical cation-mediated) trans- to cis-isomerization of the retinyl moiety (13, 14), a novel function acquired by an ancestral carotenoid oxygenase. Mutations in RPE65 are associated with a rare autosomal recessive blinding disorder called Leber congenital amaurosis, or early onset severe retinal dystrophy (15, 16). Targeted disruption of the Rpe65 gene in the mouse (17) gives rise to a phenotype in which 11-cis retinoids are completely absent in the retinae of homozygous knockout mice, while all-trans retinyl esters accumulate; concomitantly, these mice are extremely insensitive to all but the very highest light intensities.

Fig. 1.

Dual roles of RPE65 retinol isomerase in the primate and avian RPE. (A) In its canonical role, RPE65 catalyzes the concerted ester cleavage of all-trans retinyl esters and the all-trans– to 11-cis–isomerization of the retinyl moiety, as the critical retinol isomerase in the retinoid visual cycle. (B) In its newly described role, RPE65 catalyzes the isomerization of lutein to meso-zeaxanthin by migrating the 4′-5′ double bond (highlighted in red) of lutein to the 5′-6′ position (highlighted in red) of meso-zeaxanthin, thereby converting an ε-ionone ring to a β-ionone ring. The meso-zeaxanthin generated in the RPE is transported by several binding proteins (SR-B1, CD36, and IRBP) to the cone photoreceptor axons, where it is sequestered (indicated by yellow) by GSTP1 to form the macula lutea. CD36, platelet glycoprotein 4; GI, gastrointestinal; GSTP1, glutathione S-transferase (GST P); IRBP, interphotoreceptor retinoid-binding protein (retinol-binding protein 3); LRAT, lecithin/retinol acyltransferase; SR-B1, scavenger receptor class B member 1; Vit A, vitamin A.

In earlier work, Bernstein and coworkers (8) had shown that meso-zeaxanthin first appears in the RPE/choroid of dark-incubated chicken embryos at embryonic day (E) 17, increasing as the embryo developed further and appearing in the retina at E19, but nowhere else in the embryo. In PNAS, Shyam et al. show (9) that this period of development correlates with the significant up-regulation of RPE65 mRNA and protein, also required for production of 11-cis retinal chromophore for the developing photoreceptors (Fig. 1A). Supply of HPLC-purified lutein to HEK293T cells heterologously expressing RPE65 revealed accumulation of meso-zeaxanthin, in a dose- and time-dependent manner, while control cells or cells exposed to isomerically pure zeaxanthin do not. Furthermore, primary cultures of chicken RPE, which (unlike mammalian RPE cells) retain robust expression of RPE65, also accumulated meso-zeaxanthin when incubated with lutein, also in a dose- and time-dependent manner. To test the hypothesis in an in vivo system, pharmacological inhibition of RPE65 by ACU-5200-HCl, a competitive inhibitor of RPE65 and an analog of emixustat (18), injected into the yolk sac at E17 and at E19 blocked meso-zeaxanthin accumulation in injected embryos in a dose-dependent manner.

Docking of lutein into a homology model for chicken RPE65 indicates that the ε-ionone ring lies in proximity to the histidines coordinating the iron center, with the C4′/C5′/C6′ segment, where the relevant double-bond migration occurs, close to the histidines. Binding of the ε-ionone ring in the pocket may be facilitated by hydrogen bonding with Glu417 and Trp331 present in the cavity and/or aromatic ring stacking with phenylalanine residues also present. Double-bond migration from the 4′-5′ position to the 5′-6′ position is proposed to be mediated by coordinated acid-base catalysis. This double-bond migration confers the additional conjugated double bond (11 instead of 10 in lutein) to the resultant meso-zeaxanthin. Given what we know about the catalytic potential of the nonheme iron center, catalyzing double-bond cleavage of CCOs or O-alkyl cleavage of the ester bond of all-trans retinyl esters (in the case of RPE65) (13, 14, 18), this is a reasonable conjecture. Alternatively, a carbocation mechanism, as in the trans- to cis-isomerization of retinoids (13, 14, 18), may be responsible. The catalytic efficiency, as indicated by the rate of accumulation of meso-zeaxanthin in the experiments presented, is low and likely much slower than that for the trans- to cis-isomerization of retinoids, already considered to exhibit low turnover. A direct comparison of substrates, between all-trans retinyl ester and lutein, would be informative.

Despite the low turnover, resultant meso-zeaxanthin is transported across the interphotoreceptor space to the cone photoreceptors (Fig. 1B), where it is thought to be specifically sequestered by glutathione S-transferase [GST P1 (GSTP1)], the zeaxanthin-binding protein of the human macula (19). Although RPE65 is expressed throughout the RPE, and cone photoreceptors are dispersed throughout the peripheral retina, the concentration of/exclusive presence of cones in the fovea may serve to increase the relative prominence of meso-zeaxanthin in this region. In conclusion, while the primary role of RPE65 is as retinol isomerase in the visual cycle, it is not inconceivable that it should have a secondary role as the lutein to meso-zeaxanthin isomerase. Although double-bond migration of lutein to meso-zeaxanthin is different from trans- to cis-isomerization of retinol, the actual bond broken by RPE65’s iron center is the O-alkyl linkage of the retinyl ester. So, consistent with the divergent functional path that RPE65 has taken from its fellow BCOs and CCOs, lutein to meso-zeaxanthin isomerization may be yet another string to the bow of this versatile nonheme iron center.

Finally, these exciting findings will provoke much further study in this important area of vision research with implications in AMD, a disease of major impact on quality of life of older adults. Other important questions remaining include the following: (i) the precise cellular localization(s) of the meso-zeaxanthin (photoreceptor cell bodies, photoreceptor axons, muller glia, other cells?), (ii) trafficking mechanisms of the meso-zeaxanthin from the RPE to its final destination, and (iii) whether there is reduced accumulation of meso-zeaxanthin in the maculae of individuals with RPE65-associated retinal dystrophies and/or unaffected carriers of a mutant RPE65 allele. Interestingly, with respect to the last point, an SNP in RPE65, along with SNPs in many other genes in carotenoid and lipid metabolism and transport, is associated with MPOD variability in the Carotenoids in Age-Related Eye Disease Study (20).