Effect of Rpe65 Knockout on Accumulation of Lipofuscin Fluorophores in the Retinal Pigment Epithelium

Abstract

purpose. In all mammalian species examined to date the retinal pigment epithelium (RPE) has been found to accumulate autofluorescent lysosomal storage bodies (lipofuscin) during senescence. Substantial evidence indicates that retinoids in the RPE–retina complex play a major role in RPE lipofuscin formation. Indeed, at least one RPE lipofuscin fluorophore is derived in part from vitamin A aldehyde. However, the precise mechanisms by which retinoids modulate RPE lipofuscin accumulation have not been elucidated. In mice without a functional Rpe65 gene, isomerization of all-trans- to 11-cis-retinol is blocked. Experiments were performed to determine whether this impairment of retinoid metabolism alters RPE lipofuscin accumulation.

methods. RPE lipofuscin fluorophore content was compared in 12- to 13-month-old Rpe65 ^+/+ , Rpe65 ^+/− , and Rpe65 ^−/− mice. Lipofuscin fluorophore content was determined using quantitative fluorometric measurements. RPE lipofuscin content was also estimated with quantitative ultrastructural techniques.

results. In the Rpe65 ^−/− mice, RPE lipofuscin fluorophore accumulation was almost abolished. In addition, a significantly reduced accumulation of lipofuscin fluorophores was also observed in the Rpe65 ^{+/− animals. The inability of the RPE of} Rpe65 ^−/− mice to supply 11-cis-retinal from the RPE to the retinal photoreceptors was accompanied by a massive accumulation of lipid droplets in the RPE that appeared to contain substantial amounts of retinoids.

conclusions. These findings indicate that formation of RPE lipofuscin fluorophores is almost completely dependent on a normal visual cycle. The absence of retinal (both all-trans and 11-cis) in Rpe65 knockout mice drastically reduced formation of lipofuscin fluorophores in these animals. Even an excessive accumulation of retinyl fatty acid esters in the RPE of Rpe65 knockout mice did not contribute to lipofuscin accumulation.

The retinal pigment epithelium (RPE) performs functions essential for photoreceptor cell function and survival, including participation in the retinoid visual cycle, the process by which the visual pigment chromophore, 11-cis-retinal, is photoisomerized to the all-trans configuration during visual transduction and then enzymatically reisomerized to the 11-cis isomer. ^{1 2} Impairment of RPE functions appears to be involved in a number of inherited retinal degenerative disorders ^{3 4 5 6} as well as in age-related retinal degeneration (AMD), one of the most prevalent causes of serious visual impairment in developed countries. ⁷ It is well known that the RPE accumulates massive amounts of autofluorescent lysosomal storage bodies (lipofuscin) during the lifetime of the individual, ^{8 9 10} and the build-up of these intracellular inclusions has been implicated in AMD. ¹¹ Therefore, understanding the mechanisms of RPE lipofuscin formation may eventually provide the basis for preventing vision loss due to AMD. 

Several lines of evidence indicate that vitamin A (retinoids) plays a key role in RPE lipofuscin formation. Animals deprived of retinol necessary for visual pigment synthesis show very little age-related accumulation of lipofuscin in the RPE. ^{12 13} Inhibition of lysosomal protein degradation by the RPE results in a massive accumulation of autofluorescent lysosomal storage bodies in the RPE. ¹⁴ However, retinol deprivation before protease inhibitor treatment prevents development of lipofuscin-like autofluorescence in these inclusions. ¹⁵ The most direct evidence that retinoids are involved in RPE lipofuscin formation was the demonstration that N-retinylidene-N-retinylethanolamine (A2E), one of the RPE lipofuscin fluorophores, can be formed by a reaction between all-trans-retinal and ethanolamine. ¹⁶ Thus, it is likely that all-trans-retinal, generated during visual pigment bleaching and regeneration, is the key retinoid in lipofuscin formation. This mechanism is further illuminated in Stargardt disease, an early-onset form of macular degeneration in which massive accumulation of lipofuscin is evident from childhood, and in abcr knockout mice that also demonstrate early and massive accumulation of lipofuscin. The molecular defect in both these situations is a defective or absent gene encoding Rim protein (RmP). ^{17 18} RmP is an ATP-binding cassette (ABC) transporter protein specific to rod photoreceptor outer segment discs whose apparent substrates are all-trans-retinal ¹⁹ and/or N-retinylidene phosphatidylethanolamine (APE), ¹⁸ a condensation product of retinal with PE that is the apparent precursor of A2PE and, ultimately, of A2E. ^{20 21 22} Impaired removal of either or both of these products from the photoreceptor allows for increased formation of the A2E component of RPE lipofuscin. 

Although retinoids are clearly involved in RPE lipofuscin formation, the necessity for them to traverse the visual cycle to promote lipofuscin accumulation is not clear. RPE65, a protein preferentially and abundantly expressed in the RPE, ^{23 24} appears to perform an essential role in transformation of all-trans-retinol to 11-cis-retinal in the visual cycle. ²⁵ The absence of functional RPE65 protein severely disrupts the visual cycle, resulting in a deficiency of visual pigment and in an accumulation of retinyl esters in the RPE. ²⁵ Rpe65 knockout mice were used to test the hypothesis that RPE lipofuscin fluorophore formation requires the conversion of all-trans-retinol to 11-cis- and all-trans-retinals. 

Materials and Methods

Rpe65 Knockout Mice

Mice in which the Rpe65 gene was specifically disrupted were generated using gene targeting, as described previously. ²⁵ These knockout (−/−) mice do not express RPE65 protein. Wild-type (+/+) mice and mice heterozygous for the Rpe65 gene disruption (+/−) were used as control subjects. Mice were genotyped using previously described PCR reaction conditions. ²⁶  

The mice were housed under 12 hour cyclic light-dark and were fed a standard commercial mouse diet ad libitum. Illumination was provided by 40-W cool-white fluorescent lamps. Mean illuminance measured on the cage bottoms was 15 to 30 lux during the light phase of each daily cycle. Mice were 12 to 13 months of age at the time the tissues were collected. All procedures involving animals conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Fluorescence Microscopy and Microfluorometry

Quantitative microfluorometry was used to assess the amount of lipofuscin-specific autofluorescence (excitation maximum 380–440 nm; emission maximum 590–650 nm) ^{27 28 29} in the RPEs of Rpe65 ^+/+, Rpe65 ^+/−, and Rpe65 ^−/− mice. The animals were killed by carbon dioxide inhalation and the eyes were immediately enucleated. Enucleations were performed between 6 and 7 hours after the onset of the light phase of the daily light cycle. One eye of each mouse was prepared for either microfluorometry or fluorescence photomicrography, as described previously. ³⁰ Quantitative fluorescence emission intensity measurements were performed with a microscope (Photomicroscope I; Carl Zeiss, Oberkochen, Germany) equipped for epi-illumination and photometry. Fluorescence photomicrography was performed with another microscope (Axiophot; Zeiss). Detailed descriptions of the methods used for the microphotometric measurements and photomicrography are published elsewhere. ³⁰  

In preparations from Rpe65 ^−/− mice, significant green fluorescence emission was observed from fatty acid esters of retinol that had accumulated in the RPE. ³¹ However, unlike the lipofuscin fluorescence, the retinyl ester fluorescence was bleached quite rapidly under the illumination from the microscope. ³¹ Each fluorescence measurement was taken only after the retinyl ester fluorescence had completely faded, which typically occurred after less than 10 seconds in the field of illumination. Control experiments indicated that no change in the lipofuscin-specific fluorescence intensity occurred during a similar period of exposure. 

RPE wholemounts were also prepared as described above for fluorescence photomicrography used to document RPE retinyl ester fluorescence. Fluorescence photomicrographs were made with a microscope equipped for epifluorescence (Axiophot; Zeiss). Fluorescent emissions were stimulated with light from a 50-W high-pressure mercury vapor source. Examination and photography of the specimens for retinyl ester-specific fluorescence were performed, using a ×40 objective (Plan-Neofluor) with a 1.30 numerical aperture, a 395–440-nm band-pass exciter filter, a chromatic beam splitter (FT 460), and a barrier filter (LP 470; all from Zeiss). To decrease the rate of bleaching of the vitamin A by the excitation light beam, a 1.0-optical-density (OD) neutral-density filter was placed in front of the light source for documentation of vitamin A fluorescence. Photomicrography was performed with Elitechrome 100 film (Eastman Kodak; Rochester, NY), with a fixed exposure time. 

Light and Electron Microscopy

Eyes from mice of each genotype were prepared for light and electron microscopic analysis. Immediately after enucleation, the eyes were fixed and dissected, as described previously. ^{9 30} After primary fixation, each sample was dissected to obtain a strip of the eyecup along the superior–inferior meridian with the optic nerve head at its center. The tissue was then subjected to secondary fixation in 1% osmium tetroxide and embedded in an epoxy resin. ⁹ Semithin (0.5 μm thick) and ultrathin sections were cut from a central region of each retina centered on the superior–inferior meridian approximately 600 μm superior to the optic nerve head. The 0.5-μm-thick sections were mounted on slides, stained with toluidine blue, and photographed with a microscope (Axiophot, Zeiss). The ultrathin sections were stained with uranyl acetate and lead citrate and examined with an electron microscope (1200EX; JEOL Tokyo, Japan). A series of electron micrographs of contiguous nonoverlapping regions of the RPE from each sample were made at a magnification of ×5000. These micrographs were obtained to represent a minimum of 175 μm of RPE length (measured along the RPE basal lamina) from each eye. 

The mice used in this study have substantial amounts of melanin pigment in the RPE. Unfortunately, the lipofuscin and melanin in these cells were not distinct enough from one another ultrastructurally to enable them to be quantified separately. Therefore, the combined lipofuscin plus melanin content of the RPE was determined for each animal. RPE lipofuscin plus melanin contents were measured from the micrographs with an image analysis system (Metamorph; Universal Imaging, West Chester, PA), to determine the total cross-sectional area of these organelles in each micrograph. Lipofuscin plus melanin in the RPE was defined ultrastructurally as all electron-dense inclusion bodies greater than 0.20 μm in diameter that were distinct from RPE phagosomes, mitochondria, and lipid droplets. ³⁰ Phagosomes in late stages of degradation have ultrastructural features similar to those of lipofuscin. However, the tissues were collected at a time in the light cycle when RPE phagosome content is quite low. ^{32 33} Thus, any contribution of late-stage phagosomes to the lipofuscin content determinations was minimal. Lipid droplets were identified by their uniform light electron density and round profiles. RPE lipofuscin plus melanin contents were determined as the total cross-sectional area of the combined inclusion types per unit of RPE length. 

Statistical Analyses

Analysis of variance was used to assess whether Rpe65 genotype was associated with differences in RPE lipofuscin content. Comparisons between each pair of genotypes were performed using the Student-Newman-Keuls test. ³⁴  

Results

RPE Lipofuscin Autofluorescence

In mice, as in other animals, there is a progressive increase in RPE lipofuscin content during senescence. RPE lipofuscin-specific fluorescence was compared among Rpe65 ^+/+ , Rpe65 ^+/−, and Rpe65 ^−/− mice that were 12 to 13 months old. Figure 1 shows representative fluorescence micrographs of the RPE of wild-type and homozygous knockout mice. Lipofuscin-specific autofluorescence was dramatically reduced in the knockout compared with the wild-type RPE. Quantitative microfluorometry showed that animals that were homozygous for the Rpe65 knockout allele had mean RPE lipofuscin-specific autofluorescence that was only 8.7% of that of age-matched mice that were homozygous for the wild-type Rpe65 allele (P < 0.001; Figs. 1 2 ). RPE lipofuscin-specific fluorescence intensity was also reduced in mice heterozygous for the Rpe65 targeted disruption, but to a lesser degree than in homozygous knockout mice (Fig. 2) . Mean lipofuscin fluorescence intensity in the heterozygotes was 65% of that in the homozygous wild-type mice (P < 0.005). 

Retinal Morphology and Ultrastructural Analysis of RPE Lipofuscin Content

Despite the absence of a functional Rpe65 gene in the knockout mice, there was only a moderate loss of photoreceptor cells in these animals relative to the normal control animals (Fig. 3) . The mean number of photoreceptor nuclei in 100-μm-long cross-sections of the retinas were 169 ± 13 in the Rpe65 ^+/+ mice compared with 110 ± 10 in the Rpe65 ^−/− animals. In addition, although there was no visual pigment chromophore in the Rpe65 ^−/− mice, the rod outer segment morphology of the remaining photoreceptors in these animals appeared normal (Figs. 3 4) . 

Although its characteristic autofluorescence is an identifying feature of lipofuscin that accumulates in the RPE during senescence, it is possible that not all components of lipofuscin are autofluorescent. ^{15 35} Therefore, ultrastructural analyses were performed in an attempt to assess whether Rpe65 genotype influences the accumulation of not only the fluorescent constituents of lipofuscin, but of the total volume of lipofuscin in the RPE. Unlike in humans, ⁸ the lipofuscin in the mice was not distinct enough from RPE melanin to allow it to be quantified independently from the latter organelles. Thus, the combined RPE lipofuscin and melanin contents were determined in the Rpe65 ^+/+, Rpe65 ^+/−, and Rpe65 ^−/− mice. The Rpe65 ^−/− animals had a mean lipofuscin-plus-melanin content that was 26% less than that in the wild-type mice (P < 0.01; Figs. 4 5 ). The mean RPE content of these organelles in the Rpe ^+/− mice was intermediate between those of the Rpe65 ^+/+ and Rpe65 ^−/− animals, although the differences between the Rpe65 ^+/− mice and the other two groups did not meet the P < 0.05 criterion level for statistical significance. 

Accumulation of Lipid Droplets in the RPE of Rpe65 ^−/− Mice

Under normal conditions, the RPEs of most mammalian retinas, including that of the mouse, contain few if any lipid droplets. However, in the Rpe65 ^−/− mice, numerous large lipid droplets were present throughout the RPE (Fig. 4) . The size and number of the droplets were greater in the 1-year-old animals analyzed in this study than in the 15-week-old animals analyzed in the original description of the Rpe65-deficient phenotype. No such lipid droplets were observed in the Rpe65 ^+/+ or Rpe65 ^+/− mice (Fig. 4) . Redmond et al. ²⁵ concluded that the accumulation of lipid droplets in the knockout animals correlates with an overaccumulation of retinyl esters but did not present direct evidence that the lipid droplets contain retinyl esters. 

RPE Vitamin A Autofluorescence

To determine whether the lipid droplets in the Rpe65 ^−/− animals were a reservoir for storage of excessive retinyl esters, RPE wholemounts were evaluated with fluorescence microscopy to assess the presence and distribution of retinol–retinyl esters. Under conditions optimized for visualization of retinol–retinyl ester–specific autofluorescence, flatmounted RPEs of the Rpe65 ^−/− mice had numerous spherical inclusions that produced a bright green emission when the samples were illuminated with UV-blue light (Fig. 6A) . The sizes and distribution of these fluorescent spherical inclusions were consistent with sizes and distributions of the lipid droplets seen with electron microscopy (Fig. 4B) . No such inclusions were observed in the Rpe65 ^+/+ or Rpe65 ^+/−mice (Fig. 6) . In mice with the latter genotypes, only a faint yellow lipofuscin-specific autofluorescence could be seen under these conditions (Fig. 6) . 

Discussion

Dietary vitamin A deficiency has long been known to dramatically retard lipofuscin accumulation in the rat RPE. ¹² Since this discovery, a growing body of evidence has accumulated indicating that visual cycle retinoids are directly involved in the formation of at least the autofluorescent constituents of RPE lipofuscin. ^{12 13 15 16 21 35 36 37 38} However, questions remain as to the precise mechanisms by which retinoids regulate RPE lipofuscin accumulation. In this study, absence of a complete visual cycle, as evident in the Rpe65-deficient mouse, drastically reduced the accumulation of lipofuscin in mice, despite an excessive accumulation of vitamin A esters in the RPE. Thus, cycling between all-trans- and 11-cis-retinal was necessary for significant accumulation of lipofuscin to occur. These data are consistent with the report that impaired removal of all-trans-retinal from the photoreceptors in abcr-knockout mice is accompanied by a dramatically increased accumulation of A2E. ¹⁸ The dramatically decreased lipofuscin fluorophore accumulation could not be attributed to photoreceptor cell loss; the densities of photoreceptor cells in 12- to 13-month-old Rpe65 knockout mice were approximately 65% of control densities, and the remaining photoreceptors retained normal-appearing outer segments. These findings are consistent with previous demonstrations that photoreceptor cell number and morphology are well conserved after long-term dietary vitamin A deprivation that results in depletion of retinoids from the retina. ^{39 40}  

A model to explain the role of retinoids in RPE lipofuscin formation was proposed a number of years ago. ²¹ Shown in Figure 7 is an updated model that takes into account the results of the present study and the evidence that abcr mutations can influence RPE lipofuscin accumulation. In this model, a fraction of the all-trans-retinal generated in the photoreceptor outer segments during photopigment bleaching reacts with amines in the outer segments to generate precursors of RPE lipofuscin fluorophores. ^{21 22} These compounds are taken up by the RPE during normal outer segment phagocytosis where they are modified and accumulate in secondary lysosomes. 

A prediction of this model is that disruption of the vitamin A visual cycle would result in a reduction in RPE lipofuscin fluorophore formation. That has been shown to be true: Rpe65 knockout mice, incapable of isomerizing all-trans-retinol into 11-cis-retinal, ²⁵ show very little lipofuscin fluorophore accumulation. In other words, absence of retinal flux prevents accumulation of lipofuscin fluorophores. Unable to be isomerized, large amounts of retinyl fatty acid esters accumulate in the RPE of the knockout mice, ²⁵ especially in the aged animals used in this study. These lipid droplets are not normally present in the RPE of mice, and retinoid fluorescence and biochemical measurements show that they contain retinyl esters. Retinoid-containing lipid droplets also form in the RPE of mice administered large doses of all-trans-retinyl ester, ⁴¹ suggesting that the formation of lipid droplets in the knockout mice is secondary to excessive accumulation of retinoids in the RPE and is an indirect effect of the genetic defect. It is clear that all-trans-retinyl esters, even at high levels, do not participate in the generation of RPE lipofuscin fluorophores. RPE lipofuscin is composed of a mixture of fluorophores, including A2E. ^{16 27} The latter compound makes only a small contribution to total RPE lipofuscin fluorescence. ²⁷ Because lipofuscin-specific fluorescence is almost totally absent in the Rpe65 ^−/− mice, the findings support the hypothesis that all the many different lipofuscin fluorophores, ²⁷ and not just the A2E that has been identified, ¹⁶ are derived from reactions of retinal with other photoreceptor outer segment amines. 

Previous studies have suggested that not all the molecular constituents of RPE lipofuscin may be autofluorescent and that at least some of the nonfluorescent components may accumulate in RPE lysosomes, independent of the retinoid-derived fluorophores. ^{15 27} The relative contribution of nonfluorescent components of RPE lipofuscin-like inclusions can be assessed ultrastructurally, as was attempted in this study. However, unlike in the human RPE, ⁴² lipofuscin granules in the mouse RPE were not ultrastructurally distinct from melanin granules. Thus, to test the hypothesis that lipofuscin volume and fluorescence are reduced in the knockout mice, we measured the combined melanin-lipofuscin content of the RPE. Consistent with the fluorescence measurements, the combined volumes of these organelles were reduced in the knockout mice. However, because of the contribution due to melanin, it was not possible to determine the degree to which this reduction correlated with the magnitude of reduction in fluorescence intensity. To determine more precisely the degree to which lipofuscin volume is reduced as a result of the Rpe65 mutation, the study should be repeated in albino mice homozygous for Rpe65 disruption (not currently available). The results of the present study allow us to conclude, however, that even if reduced amounts of secondary lysosomes accumulate during senescence in the Rpe65 ^−/− mice, these secondary lysosomes do not contain autofluorescent retinoid derivatives. 

It is possible that other molecular constituents (e.g., lipids, proteins) of the photoreceptor outer segments, in addition to the retinoid-derived compounds, are involved in RPE lipofuscin fluorophore formation. This possibility is supported by studies showing that specific elimination of photoreceptors from the retina early in life results in greatly reduced accumulation of lipofuscin fluorophores during senescence. ^{43 44} Because the photoreceptors degenerate slowly in the Rpe65 knockout mice, it could be argued that the reduced lipofuscin fluorophore content was due at least in part to reduced RPE phagocytosis of nonretinoid precursors from the outer segments. This seems unlikely, however. The photoreceptors degenerate only gradually in the Rpe65 knockout mice ²⁵ and significant numbers remain in the 12- to 13-month-old animals, whose RPE still contain outer segment–derived phagosomes. If phagocytosis of these retinoid-deficient outer segments contributed significantly to RPE lipofuscin fluorophore formation, fluorophore accumulation would not be expected to be virtually abolished in the knockout mice as it was. 

In conclusion, the absence of a visual cycle flux of 11-cis- and all-trans-retinal, even in the presence of a large amount of retinyl esters in the RPE, reduces the formation of lipofuscin in the RPE. Thus, it is possible that less severe mutations in RPE65, not otherwise causing retinal dystrophy but instead partially reducing the efficiency of 11-cis-retinal production in the visual cycle, may confer a protective benefit. Such a benefit is evident from light damage studies in the mouse Rpe65 L450M variant ⁴⁵ and the Rpe65 knockout mouse. ⁴⁶ Similarly, allelic variation in RPE65 may contribute to individual differences in rates of lipofuscin accumulation and in risk for developing AMD. 

 Supported in part by an unrestricted grant from Research to Prevent Blindness, Inc.

 Submitted for publication March 23, 2001; revised June 11 and July 17, 2001; accepted July 26, 2001.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked“ advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Fluorescence micrographs of RPE wholemounts from (A) Rpe65 ^+/+ and (B) Rpe65 ^−/− mice. Micrographs were obtained using a filter combination optimized for visualizing lipofuscin fluorescence. Scale bar, 100 μm.

Figure 2.

 

Quantitative measures of RPE lipofuscin-specific fluorescence in mice with the three genotypes. Bars represent mean ± SEM of fluorescence intensities determined relative to a standard.

Figure 3.

 

Representative light micrographs of the central–superior retinas of (A) Rpe65 ^+/+ and (B) Rpe65 ^−/− mice. Scale bar, 50 μm.

Figure 4.

 

Representative electron micrographs of the RPE of (A) Rpe65 ^+/+ and (B) Rpe65 ^−/− mice. Arrows: Representative electron-dense inclusion bodies that were measured to obtain the data shown in Figure 3 . Numerous large lipid droplets (L) not normally present in the RPE were observed in the Rpe65 ^−/− mice. Scale bar, 2 μm.

Figure 5.

 

Quantitative ultrastructural measures of RPE lipofuscin plus melanin content in mice with the three genotypes. Bars indicated mean ± SEM of values for each group.

Figure 6.

 

Fluorescence micrographs of RPE wholemounts from (A) Rpe65 ^−/− and (B) Rpe65 ^+/+ mice. Micrographs were obtained under conditions optimized for visualization of vitamin A fluorescence. Scale bar, 100 μm.

Figure 7.

 

Proposed model for roles of Rpe65 and retinoids in RPE lipofuscin fluorophore formation. At-RDH, all-trans-retinol dehydrogenase; LRAT, lecithin-retinol acyltransferase; 11-cis RDH, 11-cis-retinol dehydrogenase; CRALBP, cellular retinaldehyde-binding protein. RPE65 is required for the generation of 11-cis-retinol from all-trans-retinyl esters present in the RPE, although the precise role of RPE65 in this process has not been elucidated.

The authors thank Laura Marler for assistance with the ultrastructural analyses and Shirley Yu for genotyping the animals used in the study.