Bisretinoids mediate light sensitivity resulting in photoreceptor cell degeneration in mice lacking the receptor tyrosine kinase Mer

The receptor tyrosine kinase Mer is expressed by retinal pigment epithelial (RPE) cells and participates in photoreceptor outer-segment phagocytosis, a process enabling membrane renewal. Mutations in the gene encoding MERTK cause blinding retinitis pigmentosa in humans. Targeted Mertk disruption in mice causes defective RPE-mediated phagocytosis of the outer segments, leading to deposition of autofluorescent debris at the RPE–photoreceptor cell interface, followed by photoreceptor cell degeneration. Here, we show that retinaldehyde adducts (bisretinoid fluorophores) that form in photoreceptor outer segments occupy the unphagocytosed outer-segment debris that accumulates in Mertk−/− mice. Bisretinoids measured by HPLC were elevated in Mertk−/− mice compared with WT animals. Bisretinoids were also more abundant in albino Mertk−/− mice expressing leucine at position 450 of the isomerase RPE65 (Rpe65-Leu450) rather than the variant methionine (Rpe65-450Met) that yields lower bisretinoid levels. In Royal College of Surgeons rats having dysfunctional Mertk, bisretinoids were higher than in WT rats. Intensities of in vivo fundus autofluorescence were higher in Mertk−/− mice than in WT mice and peaked earlier in albino Mertk−/−/Rpe65-Leu450 mice than in albino Mertk−/−/Rpe65-450Met mice. Of note, the rate of photoreceptor cell degeneration was more rapid in albino Mertk−/− mice exposed to higher levels of intraocular light (albino versus pigmented mice) and in mice carrying Rpe65-Leu450 than in Rpe65-450Met mice, revealing a link between bisretinoid accumulation and light-mediated acceleration of photoreceptor cell degeneration. In conclusion, the light sensitivity of photoreceptor cell degeneration arising from Mertk deficiency is consistent with the known phototoxicity of bisretinoids.

Mice homozygous for a targeted deletion of the Mer receptor tyrosine kinase gene (mer kd ; Mertk Ϫ/Ϫ ) (1, 2), mice carrying a chemically induced point mutation in Mertk (3), and rats having a loss of function in Mertk (Royal College of Surgeons (RCS) 3 rats) (4, 5) exhibit recessively inherited progressive degeneration of photoreceptor cells. Mertk is expressed by retinal pigment epithelial cells (RPE), and the loss of photoreceptor cells is caused by an inability of the RPE to phagocytize the outer-segment disc membrane that is shed daily by the photoreceptors. The unphagocytosed outer-segment debris accumulates as autofluorescent extracellular debris at the RPEphotoreceptor outer-segment interface in the subretinal space (2, 4, 6 -10). Early on, studies in the RCS rat assigned the defect to the RPE cell by employing chimeric animals (11) and RPE transplantation paradigms (12), and the disease-causing mutation in Mertk was identified by positional cloning (13). RPE culture systems demonstrated that Mertk functions in the ingestion phase of outer-segment phagocytosis (14). Transfer of WT Mertk to the RPE of RCS rats by viral vector reversed the ingestion defect and served to rescue the photoreceptor cells (15). Mutations in MERTK have also been identified in human subjects with retinitis pigmentosa (16), and although uncommon, these mutations are associated with abnormal patterns of fundusautofluorescencethatincludeareasofincreasedautofluorescence intensity (17,18).
In the presence of Mertk deficiency, electroretinographic recordings can be obtained until degeneration begins at ϳ3 weeks of age in the pink-eyed RCS rat (4,6), with the debris zone becoming apparent at about postnatal day 12. Between 20 and 35 days of age, the volume of this material increases by ϳ30% (6). Loss of the photoreceptor cells in both the mouse and rat models is reflected in thinning of the outer nuclear layer (ONL); in the pink-eyed tan hooded RCS rats, ONL thickness is reduced by 50% at 1 month of age (6). In the RCS rat, the degeneration proceeds most rapidly in the central part of the retina and progresses toward the periphery with time (6). In the pink-eyed and albino rats, the outer-segment debris begins to disappear in posterior central retina at around postnatal day 55, and by day 96, most is gone posteriorly.
In healthy and diseased retina, autofluorescent vitamin A-aldehyde adducts that form in photoreceptor outer segments belong to the bisretinoid family of compounds that are internalized by RPE within phagocytosed outer-segment membranes. These fluorophores constitute the lipofuscin of RPE. The long-wavelength autofluorescence exhibited by bisretinoids is readily explained by the extensive systems of conjugated double bonds that populate the vitamin A-aldehyde (A2)-derived side arms of the molecules. The bisretinoid pigments of RPE lipofuscin in humans and rodents include A2E, iso-A2E, all-trans-retinal dimer, A2GPE, and all-trans-retinal dimer-phosphatidylethanolamine (19 -28). The biosynthesis of bisretinoid occurs in photoreceptor outer segments with nonenzymatic condensation reactions between all-trans-retinaldehyde and phosphatidylethanolamine; the 11-cis isomer of retinaldehyde is the light-absorbing chromophore of rhodopsin. For the biosynthesis of A2E, the best known of the bisretinoid fluorophores, the immediate precursor is the phosphatidyl-pyridinium compound A2PE that we have identified in isolated photoreceptor outer segments (29,30). The amount of bisretinoid in healthy photoreceptor cells is not normally sufficient to make an appreciable contribution to fundus autofluorescence because these compounds are continuously transferred to RPE cells through the process of outer-segment shedding and RPE phagocytosis.
Bisretinoids are photoreactive molecules that absorb light in the visible range of the spectrum and generate reactive forms of oxygen (23,26,(31)(32)(33)(34)(35). Evidence derived from fluorescence spectra, age-associated accumulations, and disease phenotypes has shown that fundus autofluorescence in mice and humans originates primarily from bisretinoid lipofuscin in RPE cells. For instance, in both humans and mice, RPE lipofuscin (30, 36 -39) and fundus autofluorescence (40 -43) increase with age. RPE bisretinoid lipofuscin together with fundus autofluorescence also increases with ABCA4 deficiency in humans (44 -46) and mice (24,47). The spectral characteristics of the fluorescence emission generated from bisretinoid lipofuscin in mice are similar to the fundus autofluorescence spectra recorded noninvasively in humans (19,48).
The outer-segment debris that accumulates in the RCS rat retina is considered to figure prominently in events leading to photoreceptor cell degeneration, and although the precise mechanisms are not known, several processes have been proposed. For instance, the debris has been suggested to serve as a diffusion barrier (49), as a cause of vitamin A deficiency (50), as a disruption of the interphotoreceptor matrix (51) and cyclic nucleotide metabolism (52), and as a source of hydrolases (53). However, these mechanisms do not address the composition of the outer-segment debris or its ability to confer light sensitivity.
The Mertk Ϫ/Ϫ mouse provides an excellent platform for assessing the impact of bisretinoids on the health of the photoreceptor cells in which they form. To examine the molecular composition of the outer-segment debris and to explore the cause of photoreceptor cell degeneration in the presence of Mertk mutations, we generated mice carrying null mutations in Mertk in albino and pigmented mice and in mice carrying the Rpe65 amino acid variant (leucine or methionine at residue 450) that confers different levels of bisretinoid formation (22). Here, we demonstrate that bisretinoid fluorophores are the source of autofluorescence in the photoreceptor cell debris that accumulates in Mertk Ϫ/Ϫ mice, and we provide evidence that these photoreactive compounds are linked to mechanisms whereby light aggravates photoreceptor cell degeneration in the Mertk Ϫ/Ϫ mouse.

In vivo fundus autofluorescence
We imaged in vivo fundus autofluorescence noninvasively using instrumentation comparable with that used clinically. Fundus autofluorescence captured using 488-nm excitation (short-wavelength fundus autofluorescence; SW-AF) originates primarily from the bisretinoids that are amassed in retina as lipofuscin (48). With SW-AF imaging of Mertk Ϫ/Ϫ mice at 4 and 5 weeks of age, the fundus appeared relatively uniform in all mice (54 images, 27 mice) (Fig. 1A). At 6 weeks of age in the albino Mertk Ϫ/Ϫ mice (8 of 9 Rpe65-450Met; 8 of 8 Rpe65-Leu450), dark and bright mottling of the fundus was visible. Dark patchiness was even more pronounced in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice (10 of 12) at 8 weeks, whereas in the agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice (6 of 6), distinct areas of hyperautofluorescence were observed inferiorly (Fig. 1A). In all (10 of 10) albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice at 12 weeks of age, the optic nerve head (ONH) and vasculature were noticeably indistinct. Hyperautofluorescent puncta were also a feature of the SW-AF images in all mouse lines beginning as early as 8 weeks of age; because ocular pigmentation provides a dark background, the puncta were particularly pronounced in agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice and in some eyes persisted at ages of 12 ( Fig. 1A), 15 (not shown), and 24 ( Fig. 1B) weeks. NIR-AF imaging in the agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice revealed that the hyperautofluorescent puncta in these mice were also visible as bright spots in NIR-AF images (Fig.  1B). Whether a given quadrant was affected earlier by patchy nonuniformities was difficult to ascertain, but affected areas typically included posterior central retina. The hyperautofluorescent spots became visible before the appearance of darkened patches, and when central hypoautofluorescence dominated central fundus, autofluorescent spots were still visible peripherally. Note that in some images, fundus autofluorescence photographs in the Mertk Ϫ/Ϫ mice could appear equally bright across ages and genotypes; however, darkening of the internal autofluorescent reference in these cases (rectangle at the top of the image) was a sign that a shorter exposure was sufficient to image the higher fundus SW-AF levels in the mice.
Bisretinoid lipofuscin was measured in vivo by quantifying SW-AF intensities (quantitative fundus autofluorescence, qAF) as reported previously (43). In the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice, qAF reached a peak at age 6 weeks, whereas in Mertk Ϫ/Ϫ /Rpe65-450Met mice, qAF peaked at age 10 weeks; otherwise, the values were essentially at baseline (Fig. 1C). Because SW-AF intensities are modified by the presence or absence of melanin (54), absolute levels of qAF in the albino Bisretinoids and photoreceptor degeneration in Mertk ؊/؊ mice Mertk Ϫ/Ϫ /Rpe65-450Met mice relative to qAF in the agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice could not be compared. Nevertheless, mean qAF normalized to a maximum value of 1 showed that in the latter agouti mice, fundus autofluorescence appeared to undergo a steady climb from age 4 to 24 weeks; the values at 4, 6, and 8 weeks were significantly different from that at age 24 weeks (p Ͻ 0.05, one-way ANOVA with Tukey's multiple-comparison test), whereas qAF in the albino mice peaked at age 10 weeks (Fig. 1D).
As noted above, qAF in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice peaked at 6 weeks of age. The decline in fundus AF that followed was visible in the fundus images ( Fig. 1A) and evident in the plotted qAF values (Fig. 1D). This decrease can be accounted for by bleaching and resolution of the autofluorescent debris. Photoreceptor cells, the source of the AF, are also dying during this period (discussed below).

Fluorescence microscopy
To test for the presence of autofluorescence at the RPEphotoreceptor interface, as exhibited by RCS rats (7, 10), we examined cryostat (horizontal) sections that had been stained with DAPI to label nuclei (Fig. 2). As shown in Fig. 2, albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice exhibited an inherent autofluorescence at the interface between RPE and photoreceptor cells when examined by fluorescence microscopy. The nuclei of the RPE monolayer were situated on the choroidal side of the autofluorescent band. The autofluorescence in the subretinal space was visible at 6 weeks of age in albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice and in the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice, but in keeping with slower disease progression, it was not visible in agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice at this age. Muller cell processes are reported as being present in the photoreceptor debris zone (55,56); thus, uptake of the fluorescent molecules by Muller glia may account for the patterns of autofluorescence that appear to extend across the full thickness of neural retina, particularly in the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice.

UPLC analysis
A representative UPLC chromatogram in Fig. 3 shows the detection of bisretinoids in chloroform/methanol extracts of whole eyecups from albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice (age 6 weeks) (35,57). On the basis of UV-visible absorbance spectra (monitoring at 430 nm) and retention times, absorbance peaks analyzed by UPLC could be assigned to the fluorophores A2GPE ( max ϭ 346 and 438 nm), all-trans-retinal dimer  Bisretinoids and photoreceptor degeneration in Mertk ؊/؊ mice ( max ϭ 298 and 429 nm), and A2PE ( max ϭ 337 and 440 nm) (Fig. 3A). Two peaks attributable to PE-all-trans-retinal Schiff base adducts (N-retinylidene-PE; NRPE) varying in fatty acid composition are also present. Other peaks (labeled A, B, and C) having both UV and visible absorbance maxima ( max ϭ 260 and 418 nm, 329 and 423 nm, and 269 and 436 nm), suggestive of oxidized bisretinoid fluorophores, were also present (Fig. 3). In UPLC profiles under the chromatographic conditions utilized for Fig. 3A, A2E would be expected to elute as an ill-defined peak at a retention time (Rt) of 6.8 min; this peak was not detected at age 6 weeks.

Photoreceptor cell degeneration
Photoreceptor cell degeneration was readily detectable in H&E-stained paraffin sections of the eyes (Fig. 4). As shown previously in mer kd mice (1, 2), photoreceptor cell degeneration was indicated by reduced width of ONL. Specifically, ONL thinning was observed at 6 weeks of age in the albino Mertk Ϫ/Ϫ / Rpe65-Leu450 mice and albino Mertk Ϫ/Ϫ /Rpe65-450Met mice but was not noticeable in agouti Mertk Ϫ/Ϫ /Rpe65-450Met (not shown). At 8 weeks of age in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice, the ONL was reduced to only a single row of cell nuclei (Fig. 4). At the same age, ONL width was also visibly reduced in the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice (Figs. 4  Bisretinoids and photoreceptor degeneration in Mertk ؊/؊ mice and 5) but to a lesser extent. The acidophilic subretinal debris zone was readily visible in H&E-stained sections acquired from the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice (Fig. 4). Photoreceptor cell viability was quantified in Mertk Ϫ/Ϫ mice by measuring ONL thickness. Plotting of ONL thickness as a function of distance superior and inferior to the optic nerve head in the vertical plane revealed that in all three Mertk Ϫ/Ϫ mouse lines, photoreceptor cell degeneration was detected as thinning of the ONL at 8 weeks of age (Fig. 5A). For comparative purposes, we calculated mean ONL area within the segments of retina extending from the ONH to a point 2 mm in the superior and inferior hemi-retina (Fig. 5B). The thinning was most rapid in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice followed by the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice. The reduction in ONL thickness was slower in the agouti Mertk Ϫ/Ϫ /Rpe65-450Met line. Growth of the mouse eyeball (estimated by measuring corneal diameter), with concomitant changes in photoreceptor cell density, reaches a plateau at ϳ8 weeks of age (58). At age 8 weeks, the percent difference (value 1 Ϫ value 2/mean value 1 ϩ value 2 ϫ 100%) between agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice and albino Mertk Ϫ/Ϫ /Rpe65-450Met mice was 98% (p Ͻ 0.05, one-way ANOVA and Sidak's multiple-comparison test), and the percentage difference between albino Mertk Ϫ/Ϫ / Rpe65-Leu450 and albino Mertk Ϫ/Ϫ /Rpe65-450Met mice was 92% (p Ͻ 0.05) (Fig. 5B).

Chromatographic analysis of RCS rat
We also analyzed bisretinoids in chromatograms obtained from eyecups of pigmented and albino RCS rats (Fig. 6). HPLC and UPLC were both employed together with columns having different retention properties. Peak authentication was confirmed by UV-visible absorbance and retention times. By HPLC with a C4 column that favors A2PE and all-trans-retinal dimer elution (33), A2PE-attributable peaks were detected in samples obtained from the albino RCS rat while being absent in the albino RCS-rdyϩ control rat (60) (Fig. 6A). The presence of A2E and iso-A2E in the analyte, albeit at low levels, is probably explained by the presence of a phospholipase D-like activity in outer segments (61).
As also demonstrated in the UPLC chromatograms presented in Fig. 6, the Schiff base adduct NRPE (62), which is a precursor in the bisretinoid biosynthetic pathway, presented as multiple pronounced peaks in the RCS rat samples while being less abundant in the WT Long Evans rat (Fig. 6B). NRPE presents as a series of peaks because of species of NRPE differing in fatty acid composition. A2PE and all-trans-retinal dimer were detected in the RCS rat while being absent in chromatograms associated with the Long Evans rat. The A2GPE peak (35) was also greater in the mutant rat.

Discussion
Mutations in MERTK have been identified in human subjects with retinitis pigmentosa and account for about 1% of nonsyndromic autosomal recessive cases (16,63). These mutations in humans cause severe rod-cone degeneration (16,64). Fundus autofluorescence images acquired from these individuals exhibit abnormal patterns that include hyperautofluorescent Bisretinoids and photoreceptor degeneration in Mertk ؊/؊ mice puncta, mottling, increased central autofluorescence, and eventually macular atrophy marked by a pronounced reduction in autofluorescence (17,18). Optical coherence tomography (OCT) reveals a loss of the ellipsoid hyperreflectivity layer; this alteration is attributable to photoreceptor cell degeneration (18). These disease-related features correspond well to the fundus and histometric changes that we observed in the Mertk Ϫ/Ϫ mice.
In color fluorescence micrographs (albino Mertk Ϫ/Ϫ /Rpe65-Leu450 and albino Mertk Ϫ/Ϫ /Rpe65-450Met mice), we observed golden-yellow autofluorescence (excitation, 490 nm) anterior to the RPE; this autofluorescence was consistent with the presence of the previously described debris zone in light microscopic images acquired from RCS rats and Mertk kd mice (2,4,6,65). In a clinically relevant approach, we also detected this autofluorescence by in vivo SW-AF imaging. Increased autofluorescence and mottling were observed, as with SW-AF imaging in human patients carrying mutations in MERTK (17,18). Fundus AF intensity in albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice peaked at 6 weeks of age, a time point that was earlier than in the other mouse lines and that was consistent with the faster rate of bisretinoid formation associated with expression of leucine at position 450 of Rpe65 (22). Conversely, the Rpe65-450Met variant expressed by C57BL/6J and C57BL/6J c2j mice has been shown to slow the regeneration of 11-cis-retinal production (66,67). The retarded kinetics in the presence of the methionine 450 substitution is also associated with reduced bisretinoid in albino C57BL/6J c2j (Rpe65-450Met) versus BALB/cJ (WT Rpe65-Leu450) mice (22), reduced capacity for photon catch (67), and some resistance to light damage (66,68,69). Similarly, in the present studies, albino Mertk Ϫ/Ϫ /Rpe65-450Met (age 8 weeks) mice expressed A2E and A2GPE levels that were lower than in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice. The more rapid photoreceptor cell degeneration in the albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice as opposed to the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice is indicative of a difference associated with bisretinoid levels. Chromatographic measurements of bisretinoid further showed that bisretinoids are elevated in RCS rat eyecups relative to RCS-rdyϩ control rats.
It has been estimated that retinal irradiance in the albino is 2 orders of magnitude more than in pigmented eyes (70). It is thus significant that photoreceptor cell degeneration was also accelerated in the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice relative to the agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice, indicating that light modulates the degeneration. The difference in the timeline of photoreceptor cell degeneration due to ocular pigmentation probably also explains why normalized qAF peaked at 6 weeks in the albino Mertk Ϫ/Ϫ /Rpe65-450Met mice while continuing to rise in the agouti Mertk Ϫ/Ϫ /Rpe65-450Met mice. Reports of studies in the RCS rat also implicate light as a factor aggravating photoreceptor cell degeneration. For instance, RCS rats with pigmented eyes or RCS rats raised in darkness exhibit a slower rate of degeneration than cyclic-light-reared melanin-deficient RCS rats (4,6,71).
ONL thinning due to loss of photoreceptor cells also occurs more rapidly in inferior retina than in the superior hemisphere in both pigmented RCS rats (6) and pigmented Mertk null mutant mice (2). Similarly, we observed that ONL thinning occurred faster in the inferior hemiretina of albino Mertk Ϫ/Ϫ / Rpe65-Leu450 mice. Here the difference was 14% at 6 weeks of age (p Ͻ 0.05). The propensity for light damage inferiorly in Mertk mutant mice is consistent with inferior damage observed in animals raised at elevated light levels (72), in transgenic mice expressing the P23H mutation (73), and in patients with sectoral retinitis pigmentosa (74). In many of these cases, the more rapid progression of disease in inferior retina is attributed to the modifying effects of light.
The light sensitivity expressed by Mertk mutant mice is readily accounted for by bisretinoid phototoxicity. Specifically, bisretinoids have the propensity to photogenerate reactive oxygen species such as singlet oxygen and to photodegrade into aldehyde-bearing fragments and, specifically, the dicarbonyls methylglyoxal and glyoxal that are responsible for cell damage (26,75). This bisretinoid photodegradative process is responsible for the lower bisretinoid levels measured in lightversus dark-reared mice (35) and for the finding that mice burdened with age-and genotype (albino Abca4 Ϫ/Ϫ mice)-associated increased bisretinoid (A2E and all-trans-retinal dimer) exhibit increased susceptibility to retina light damage (76). Similarly, Figure 6. Representative chromatograms illustrating the detection of bisretinoid lipofuscin pigments in RCS rat eyecups. A, posterior eyecups from albino dystrophic RCS rats and control rats (RCS-rdyϩ) were analyzed by reverse-phase HPLC with a C4 column and monitoring at 430 nm (age 2 months). Compounds were identified on the basis of UV-visible absorbance and retention times recorded using authentic standards. Detection of A2E (Rt ϭ 4.2 min), iso-A2E (Rt ϭ 5.3 min), atRAL dimer (Rt ϭ 11.8 min), and A2PE (Rt ϭ 23.2 min). B, UPLC analysis of eyecups obtained from pigmented RCS rats and WT Long Evans rats (430-nm monitoring; age 2 months; detection of A2GPE (Rt ϭ 37.0 min), atRAL dimer (Rt ϭ 60.2 min), and A2PE (Rt ϭ 94.3 min)). Insets (top), UV-visible absorbance spectra of chromatographic peaks corresponding to bisretinoids. mAU, milli-absorbance units.
Bisretinoids and photoreceptor degeneration in Mertk ؊/؊ mice the accumulation of bisretinoids in photoreceptor cell outer segments with Mertk deficiency probably explains the lightassociated retinal degeneration expressed by Mertk Ϫ/Ϫ mice. Support for links among bisretinoid accumulation, adverse effects of light, and photoreceptor cell degeneration is provided by the evidence indicating that the rate of photoreceptor cell degeneration was more rapid in mice exposed to higher levels of intraocular light (albino versus agouti) and by the finding that ONL thinning in albino Mertk Ϫ/Ϫ /Rpe65-Leu450 mice advanced more rapidly in inferior retina. Consistent with the premise that the light-associated effects of Mertk deficiency operate through bisretinoid was the observation that ONL thinning was more rapid in mice having higher levels of bisretinoid (Rpe65-Leu450 versus 450Met).
Based on work in the Rdh8 Ϫ/Ϫ /Abca4 Ϫ/Ϫ mouse (77), it has been suggested that, apart from adverse effects of bisretinoids, all-trans-retinal is directly damaging to photoreceptor cells. In the Mertk mutant mouse, Abca4 and retinol dehydrogenases are functional; thus, all-trans-retinal does not accumulate. Rather, all-trans-retinal is reduced to nontoxic retinol, and the all-trans-retinal that escapes reduction forms bisretinoid. The difference between Mertk mutant mice and WT mice rests in the handling of the shed OS. Specifically, phagocytosis of the outer-segment debris is essential for photoreceptor cell survival. Of note, at 8 weeks of age, 11-cis-retinal levels are highest in agouti Mertk Ϫ/Ϫ /Rpe65-Leu450 as compared with the two other Mertk Ϫ/Ϫ mouse lines (Fig. 5D). Thus, there is potential for release of more abundant all-trans-retinal in agouti Mertk Ϫ/Ϫ /Rpe65-Leu450 mice, yet outer nuclear layer thickness is better preserved in these mice.
There are other apparently disparate retinal disorders that are accompanied by accumulations of autofluorescent material in the outer retina and/or in the subretinal space as in the Mertk Ϫ/Ϫ mice. For instance, in a mouse model of retinal detachment, outer-segment debris that is not phagocytosed by RPE becomes visible as foci of elevated autofluorescence (78). Within autofluorescence rings that often characterize fundus autofluorescence images in retinitis pigmentosa (79 -83), photoreceptor cell viability is reduced, and qAF can be higher than at equivalent fundus positions in healthy eyes (84). Autofluorescent flecks visible in the fundus of STGD1 patients serve as yet another example. We have shown that fundus flecks in STGD1 are a manifestation of groups of degenerating photoreceptor cells (85). This conclusion is buttressed by the finding that flecks visualized in SW-AF images correspond to hyperreflective lesions extending radially within photoreceptor-attributable bands in spectral-domain OCT scans. The hyperreflectivity of flecks in spectral-domain OCT scans co-localizes with reduced or absent RPE melanin autofluorescence in NIR-AF images; loss of this signal is indicative of RPE atrophy. Abnormally increased SW-AF is also associated with outer segments that form the core of photoreceptor cell folds or rosettes in degenerating mouse retina (78,86). As shown here in Mertk Ϫ/Ϫ mice, these findings indicate that the process of photoreceptor cell degeneration may involve increased bisretinoid formation. Under all of these conditions, the increased bisretinoid may have the potential to aggravate disease processes.
As was reported in human MERTK-associated disease (18), we observed hyperautofluorescent puncta in fundus SW-AF images of Mertk Ϫ/Ϫ mice. Similar granular-like hyperautofluorescent foci have been described in other mouse models of retinal degeneration (78,(87)(88)(89)(90)(91)(92)(93). In some cases, these foci have been assigned to hyperreflective lesions in the outer nuclear layer-attributable band of OCT scans and to circularly arranged ONL nuclei in histological sections (78), whereas in other work, it was suggested that macrophages within the subretinal space were the source of the hyperautofluorescence (18,94,95). In a mouse model of retinal detachment, we found that autofluorescent puncta visible in the fundus exhibited emission spectra (emission peaks at 581 and 629 nm with excitations at 488 and 561 nm, respectively) that were consistent with an origin from the bisretinoid fluorophores that form in photoreceptor cell outer segments (48,78). In the present work, macrophages are unlikely to account for the hyperautofluorescent spots because F4/80 staining (Fig. 1C) is consistent with single cells rather than cell aggregates of a size sufficient to account for fundus spots.
Taken together, the more rapid photoreceptor cell degeneration in the Mertk Ϫ/Ϫ albino versus agouti mice on the same genetic background (Mertk Ϫ/Ϫ /Rpe65-450Met) together with the faster degeneration in the Mertk Ϫ/Ϫ /Rpe65-Leu450 versus 450Met mice (of the same albino coat color) are consistent with the premise that bisretinoid accumulation in the outer segments of Mertk Ϫ/Ϫ mice is responsible for light-mediated aggravation of photoreceptor cell degeneration. This work illustrates that photoreceptor cells can serve as an aberrant source of heightened fundus autofluorescence due to increased bisretinoid formation in impaired photoreceptor cells. These findings are not only relevant to human MERTK disease but are also relevant to our understanding of the paracentral rings of elevated AF in retinitis pigmentosa (96) and/or the bright AF flecks that characterize recessive Stargardt disease (44,97).
Pink-eyed inbred dystrophic (RCS) rats, black-eyed dystrophic RCS rats (RCS-pϩ), and pink-eyed congenic RCS-rdyϩ rats that are WT at the rdy locus were obtained from Dr. Matthew LaVail and raised in a 12-h light/12-h dark environment at ϳ15 lux. The institutional animal care and use committee of Columbia University approved all procedures.

Fundus imaging
The pupils of anesthetized mice were dilated, the cornea was lubricated, and fundus autofluorescence images (55°wide-field lens; 0.98-mm detection pupil) were acquired with a confocal scanning laser ophthalmoscope (Spectralis HRA, Heidelberg Engineering, Heidelberg, Germany) (43). After visual pigment was bleached (20 s), laser power was set at ϳ280 microwatts and sensitivity at 100, nine successive frames were acquired at 488-nm excitation with the high-speed mode (138 images in 41 mice), and nonnormalized frames were saved. A mean of 100 frames were obtained at 790-nm excitation (sensitivity 105) with high-resolution ART (automatic real-time) mode and resized with Photoshop CS4 (Adobe Systems, San Jose, CA) to 768 ϫ 768 pixels, the same as high-speed mode images. Near-IR reflectance images (820 nm) were also acquired. Mice were not serially imaged; each image was acquired from a different mouse.

qAF
As described (43), mean gray levels (GLs) were calculated from eight predefined segments around the optic disc, and blood vessels were excluded by histogram analysis. qAF at 488-nm excitation was calculated by normalization to the GL of the reference after subtraction of zero light (GL 0 ) and inclusion of a reference calibration factor. Fluorescence intensities at 790 nm were calculated by subtracting the minimal GL of optic nerve head measured by ImageJ (a public domain, Java-based image-processing program developed at the National Institutes of Health).
For 11-cis-retinal quantification, mouse eyecups (1 eye/ sample) were homogenized and derivatized using O-ethylhydroxylamine (100). Retinal O-ethyloxime was extracted with hexane and resuspended in acetonitrile. The Waters Acquity UPLC-MS system was used with a CSH C18 column (1.7 m, 2.1 ϫ 100 mm; Waters) and gradients of water (A) and acetonitrile (B) with 0.1% of formic acid as follows: beginning at 60% B, holding for 5 min, followed by a linear increase to 70% B over 55 min, followed by a linear increase to 100% B over 10 min (flow rate of 0.3 ml/min).

Histological analysis
Mouse eyes were marked with tissue dye and fixed (4% paraformaldehyde, 20% isopropyl alcohol, 2% TCA, and 2% zinc chloride), and 5-m H&E-stained paraffin sections were prepared. The sections most centrally located within the ONH were imaged digitally. ONL width was measured at 200-m intervals in the vertical plane and plotted as distance (mm) superior and inferior to the ONH. ONL area was calculated using the measurement interval of 0.2 mm multiplied by the sum of ONL thicknesses in superior and inferior hemiretina (ONH to 2.0 mm in superior and inferior hemiretina). To compare ONL thicknesses in superior versus inferior hemiretina, we calculated the mean of thicknesses measured at 0.2-mm intervals from the ONH to the position 2.0 mm in superior and inferior retina.

Statistical analysis
Statistical analysis was performed using GraphPad Prism version 6 (GraphPad Software, Inc., La Jolla, CA); p Ͻ 0.05 was considered significant.