RPE65 plays an essential role in the processing of retinoids in the visual cycle. This study demonstrates the presence of the RPE65 protein in mammalian and
Xenopus cone photoreceptors as well as in salamander cones, as reported earlier.
3 The protein was not found in the rods of any of these species, further supporting the suggestion that rods and cones have distinct pathways for retinoid processing.
In previous studies, we have detected the RPE65 mRNA in salamander cones, by single cell RT-PCR.
3 However, there are two questions that were not answered by these studies. First, it is not clear whether the RPE65 protein is expressed in cones, even though its mRNA is present. Second, it is unknown whether the expression of RPE65 in cones is a specific feature of amphibians—more specifically, of the salamander—or can be found in other species. The present study addressed these questions by means of immunohistochemistry with a specific antibody to RPE65. Because it is difficult to dissect the retina from RPE contamination, we could not use Western blot analysis to demonstrate reliably the presence of RPE65 protein in the retina. Immunohistochemistry detected the RPE65 protein in cones of all mammalian species analyzed in addition to amphibians. To confirm that the labeling was not due to cross-reactivity of the antibody with another protein, the RPE65
−/− mouse was used as a negative control. The identity of stained cones was determined, not only by the morphology of the stained cells and their positions in the outer retina, but also by double labeling with PNA lectin, a specific marker for cone photoreceptors.
18 19 21 22 In contrast, no staining was observed in rods of any of these species. This confirms that the difference between rods and cones in RPE65 expression is not limited to amphibians.
During the preparation of this manuscript, a paper by Seeliger et al.
23 was published online (http://genetics.nature.com), in which they showed that cones in the retina sections from B6/129S mice are not stained by an anti-RPE65 antibody. However, our results in retina sections showed a specific staining of the cone outer segments, but not of the rods, by an affinity-purified anti-RPE65 antibody. The labeled cones were confirmed by double staining with PNA lectin. This RPE65 labeling provides a better cone morphology and definitive location of the labeled cells in the photoreceptor layer in the retina, when compared with the labeling in retina flatmounts. There are several factors that may explain the discrepancy between our results and that of Seeliger et al.
23 First, our anti-RPE65 antibody was affinity-purified by using an RPE65 peptide epitope. This purified IgG generates more specific and stronger staining than does whole serum. Seeliger et al.
23 did not specify which antibody was used for staining in their study. Because RPE65 levels in cones were substantially lower than that in the RPE, it may require a purified antibody to detect the protein in cones. Second, the labeling procedure could make a difference. They embedded the retina in agarose, whereas we used frozen sections. Third, we used BALB/c mice for RPE65 staining, because BALB/c mice have significantly higher RPE65 expression levels than do either 129S or B6 mice.
20 Seeliger et al.
23 used mice with the B6/129S background, which is expected to produce a lower RPE65 expression than BALB/c mice. The lower expression of RPE65 could make the detection of RPE65 in cones more difficult.
Most vertebrate retinas contain more than one type of cone, with most rodents, such as mouse and rabbit having two types of cones: medium (M) and short (S) wavelength sensitive.
24 25 It has been demonstrated that in the mouse retina, the superior area is dominated by the M cones, whereas the inferior area contains almost all the S cones.
26 In the present study, there was no variation in the density of the RPE65-positive cones between the superior and inferior areas of the mouse retina
(Fig. 3A) . To confirm that both types of cones express RPE65, we double-labeled mouse retina with PNA lectin and the anti-RPE65 antibody, because PNA lectin has been shown to label both types of cones in mouse retinas. The results demonstrate that all lectin-positive cones in the mouse retina were also labeled by the anti-RPE65 antibody. Taken together, these observations show that RPE65 was expressed in both the M and S cones in the mouse retina.
Although the RPE65 gene does not have high homology with any known sequences, there is still a possibility that the labeled protein in cones is encoded by an unknown gene homologous to RPE65, rather than by the RPE65 gene itself. The present study examined this possibility in the retina from the RPE65-knockout mouse. Double labeling of the knockout retina demonstrated no RPE65 labeling in the retina, whereas the lectin-positive cones were present in the knockout mice at ages of 1 and 4 months, confirming that the labeled protein in cones of wild-type animals is indeed encoded by the RPE65 gene.
The RPE is the major source for 11-
cis retinal in the visual cycle, at least in species with a rod-dominant retina (for review see Crouch et al.
27 ). Although rods depend on the RPE for the supply of 11-
cis retinal, it has been suggested that cones, at least those in amphibians, may have an alternative retinoid transporting and processing pathway independent of the RPE.
28 29 Early studies have demonstrated a spontaneous recovery of cone sensitivity (but not of the rod sensitivity) after bleaching in the isolated retinas of turtle,
30 frog,
31 32 and rat.
33 In isolated salamander photoreceptors, Jones et al.
16 have demonstrated that the addition of 11-
cis retinol can restore cone sensitivity but not rod sensitivity, suggesting the presence of 11-
cis retinol dehydrogenase or its homologue in cones.
34 Different from rods, salamander cones have the capacity to transport 11-
cis retinal from the inner segment to the outer segment.
17 These observations support the possibility that there are differences in retinoid processing and transporting systems between rods and cones of some species.
RPE65 is essential for the formation of 11-
cis retinal. Previous studies have shown that the RPE of the RPE65
−/− mouse fails to isomerize all-
trans retinoid to 11-
cis isomers.
8 Our recent experiments showed that the addition of recombinant RPE65 to the tissue homogenate of RPE65
−/− mouse eyecup restored the retinol isomerase activity in the knockout mouse eyecup, although RPE65 alone did not show any in vitro isomerase activity (Moiseyev G, unpublished data, 2000). This result suggests that RPE65 is essential, but not sufficient for isomerization of all-
trans retinoids into 11-
cis isomers and that RPE65 may require binding with its partners to show the isomerase activity. Recent results in the RPE65-deficient dog in which visual function was restored with a gene therapy approach
14 further confirm the essential role of this protein. The exact function of the protein in cones as well as the RPE remains to be determined. The identification of the RPE65 protein in mammalian cones provides additional evidence that mammalian cones may contain a metabolic system that can isomerize retinoids.
The authors thank T. Michael Redmond (Laboratory of Retinal Cell and Molecular Biology, National Eye Institute, National Institutes of Health, Bethesda, MD) for providing the RPE65−/− mice for this study.