Ablation of Fatty Acid Transport Protein-4 Enhances Cone Survival, M-cone Vision, and Synthesis of Cone-Tropic 9-cis-Retinal in rd12 Mouse Model of Leber Congenital Amaurosis

Abstract

The canonical visual cycle employing RPE65 as the retinoid isomerase regenerates 11-cis-retinal to support both rod- and cone-mediated vision. Mutations of RPE65 are associated with Leber congenital amaurosis that results in rod and cone photoreceptor degeneration and vision loss of affected patients at an early age. Dark-reared Rpe65^−/− mouse has been known to form isorhodopsin that employs 9-cis-retinal as the photosensitive chromophore. The mechanism regulating 9-cis-retinal synthesis and the role of the endogenous 9-cis-retinal in cone survival and function remain largely unknown. In this study, we found that ablation of fatty acid transport protein-4 (FATP4), a negative regulator of 11-cis-retinol synthesis catalyzed by RPE65, increased the formation of 9-cis-retinal, but not 11-cis-retinal, in a light-independent mechanism in both sexes of RPE65-null rd12 mice. Both rd12 and rd12;Fatp4^−/− mice contained a massive amount of all-trans-retinyl esters in the eyes, exhibiting comparable scotopic vision and rod degeneration. However, expression levels of M- and S-opsins as well as numbers of M- and S-cones surviving in the superior retinas of rd12;Fatp4^−/− mice were at least twofold greater than those in age-matched rd12 mice. Moreover, FATP4 deficiency significantly shortened photopic b-wave implicit time, improved M-cone visual function, and substantially deaccelerated the progression of cone degeneration in rd12 mice, whereas FATP4 deficiency in mice with wild-type Rpe65 alleles neither induced 9-cis-retinal formation nor influenced cone survival and function. These results identify FATP4 as a new regulator of synthesis of 9-cis-retinal, which is a “cone-tropic” chromophore supporting cone survival and function in the retinas with defective RPE65.

Significance Statement

Isorhodopsin, which employs 9-cis-retinal as the light-sensitive chromophore, is known to support rod survival and function in dark-reared Rpe65^−/− mouse model of Leber congenital amaurosis (LCA) that exhibits early cone degeneration. The mechanism regulating 9-cis-retinal formation and the role of 9-cis-retinal in cone-mediated color vision remain largely unknown. Here, we identified FATP4 as a new negative regulator of 9-cis-retinal synthesis in RPE65-null mice. We found that increased 9-cis-retinal did not influence rod function and degeneration, but it significantly enhanced cone survival and function in mice lacking both RPE65 and FATP4. Our findings indicate that 9-cis-retinal functions as a “cone-tropic” chromophore, providing 9-cis-retinal and FATP4 as important therapeutic targets to alleviate cone degeneration and color vision loss in RPE65 mutation-associated LCA.

Introduction

The 11-cis-retinal chromophore linked to rod and cone opsins functions as a molecular switch for initiating phototransduction in response to light stimuli (Wald, 1968; Burns and Arshavsky, 2005; Travis et al., 2007). RPE65, an RPE-specific endoplasmic reticulum (ER)-associated protein (Hamel et al., 1993), is a key retinoid isomerase (Jin et al., 2005; Moiseyev et al., 2005; Redmond et al., 2005) in the canonical visual cycle that generates 11-cis-retinal for rod and cone opsins. RPE65 catalyzes the synthesis of 11-cis-retinol from all-trans-retinyl esters (atREs) synthesized by lecithin:retinol acyltransferase (LRAT) and other enzymes (Ruiz et al., 1999; Kaschula et al., 2006; Kaylor et al., 2015). Mutations of RPE65 are associated with Leber congenital amaurosis (LCA), an early onset severe retinal dystrophy (Gu et al., 1997; Marlhens et al., 1997). While most patients experience night blindness and undetectable rod electroretinograms (ERGs), many patients also exhibit severe cone degeneration at an early age (Jacobson et al., 2005, 2009). Chromatic pupillometry and chromatic sensitivity studies further reveal an early loss of S-cone function, while L- and M- cones remain initially functional (Lorenz et al., 2012). RPE65 also plays an important role in protecting human cones through its second isomerase activity catalyzing synthesis of meso-zeaxanthin (Shyam et al., 2017), one of the three macular pigments (Bernstein et al., 2016). RPE65 is highly concentrated in the macaque central RPE layer localized to the cone-rich area (Jacobson et al., 2007).

Rpe65^−/− mice lack measurable 11-cis-retinoids, displaying massive atRE accumulation, cone opsin mislocalization, and early cone degeneration (Redmond et al., 1998; Rohrer et al., 2005; Znoiko et al., 2005). However, Rpe65^−/− mice still exhibit a minimal visual response from rods, although they lack detectable 11-cis-retinal (Redmond et al., 1998; Seeliger et al., 2001). Fan and colleagues (Fan et al., 2003) have shown that long time dark-rearing of Rpe65^−/− mice increases isorhodopsin and 9-cis-retinal production in the RPE. A 9-cis-retinal is known to form isorhodopsin (Hubbard and Wald, 1952; Crouch et al., 1975), and its administration can restore rod function in Rpe65^−/− mice (Van Hooser et al., 2000). A recent study has also identified dihydroceramide desaturase (DES1) as a second retinol isomerase capable of synthesizing 9-cis-retinol (Kaylor et al., 2013). Additionally, the RPE-retinal G-protein-coupled receptor (RGR) has also been found to generate 11-cis-retinal through a light-dependent mechanism (Chen et al., 2001; J. Zhang et al., 2019; Morshedian et al., 2019).

Despite the existence of the above-described alternative pathways, RPE65 deficiency leads to retinal degeneration including cone death in patients and mouse models. While the exact molecular and cellular mechanisms underlying this phenotype are not fully understood, one potential explanation is that the quantities of 9-cis- and 11-cis-retinals generated by the alternative pathways are insufficient to maintain photoreceptor survival and function. Therefore, identifying the endogenous factors and mechanisms that regulate the synthesis of 9-cis- and 11-cis-retinals is important, not only for understanding the regulatory mechanisms of the syntheses but also for developing a new therapeutic approach.

In our previous studies, we identified FATP4 as a negative regulator of 11-cis-retinol synthesis catalyzed by RPE65 (Li et al., 2013). FATP4 facilitates transport of long-chain fatty acids across cell membrane and exhibits acyl-CoA synthetase activity toward saturated and monounsaturated long-chain and very long-chain fatty acids (Stahl et al., 1999; Hall et al., 2005; Jia et al., 2007). This single-pass transmembrane protein with ER-localizing and AMP-binding domains (Milger et al., 2006) directly interacts with RPE65 (Li et al., 2020b). Fatp4^−/− mice with wild-type Rpe65 do not display abnormal development and degeneration of photoreceptors (Li et al., 2013). In contrast, FATP4 deficiency in Rpe65-R91W knock-in mouse model of LCA (Samardzija et al., 2008) improves survival and function of cones and rods through increasing the synthesis of 11- and 9-cis-retinals (Li et al., 2020a). However, it is unknown whether FATP4 deficiency can promote 9-cis-retinal formation without the mutant RPE65 and whether it can improve survival and function of cones without 11-cis-retinal. In this study, we aimed to address these questions.

Materials and Methods

Animals

The Fatp4^−/−;Ivl-Fatp4^tg/+ (shown as Fatp4^−/− in this study) and R91W knock-in (KI) mice were provided by Dr. Jeffery H. Miner and Dr. Christian Grimm, respectively (Moulson et al., 2007; Samardzija et al., 2008), and the RPE65-null rd12 mice were purchased from The Jackson Laboratory. The Fatp4^−/− mice are homologous for the L450 alleles of the Rpe65 gene and express transgenic FATP4 in the keratinocytes, but not in the neural retina and RPE (Moulson et al., 2007; Li et al., 2013). The R91W mutation, which has been introduced into the Rpe65 gene of the KI mouse (Samardzija et al., 2008), is one of the most frequent RPE65 mutations associated with LCA (Thompson et al., 2000; Jacobson et al., 2009). We backcrossed the Fatp4^−/− and rd12 mouse lines with wild-type (WT) 129S2/Sv strain mice (Charles River Laboratories) for at least four generations to yield Fatp4^−/− and rd12 mice with the 129S2/Sv genetic background. We crossed the new Fatp4^−/− mice with the new rd12 mice, then intercrossed the heterozygous offspring to generate rd12;Fatp4^−/− mice. R91W;Fatp4^−/− mice on the 129S2/Sv genetic background have been described in the previous study (Li et al., 2020a). Except where noted, mice were maintained in 12 h cyclic light at ∼30 lux. All experiments using animals were performed in accordance with the Association for Research of Vision and Ophthalmology statement for the use of animals in ophthalmic and vision research and the research protocols approved by the Institutional Animal Care and Use Committee for LSU Health New Orleans.

Immunoblot analysis

Protein concentrations in RPE and retinal homogenates from mouse eyecups were determined with the Micro BCA protein assay kit (Thermo Fisher Scientific). After incubating in the Laemmli buffer containing 50 mM dithiothreitol for 10 min at 70°C or room temperature, the homogenates were separated in a 10 or 12% polyacrylamide gel by SDS-PAGE and transferred to an Immobilon-P membrane (MilliporeSigma). The membrane was incubated with blocking buffer, a primary antibody, and a horseradish peroxidase-conjugated secondary antibody against rabbit, mouse, or goat IgG (Jackson ImmunoResearch Laboratories). Before and after incubating with the antibodies, the membrane was washed in phosphate-buffered saline containing 0.1% Tween-20. Antibodies against RPE65 (Jin et al., 2007), FATP4 (Newberry et al., 2003), LRAT (Batten et al., 2004; Golczak et al., 2005), CAR, β-actin, M-opsin, rhodopsin, PKCα (all MilliporeSigma), S-opsin (MilliporeSigma, Santa Cruz Biotechnology), RALDH1 (Proteintech), or mGluR6 (GeneTex) were used as the primary antibodies. Immunoblots were visualized with the ECL Prime Western blotting detection reagent and ImageQuant LAS 4000 (GE HealthCare). Signal intensity of each band was quantified using ImageQuant TL software.

Analysis of retinoids

All tissue manipulations and retinoid analysis were performed under dim red safety light (630 nm, Kodak GBX-2 safelight filter). The retina and RPE in the mouse eyecups were homogenized in 20 mM HEPES buffer containing 150 mM hydroxylamine. Retinoids in the homogenates were extracted with hexane and analyzed by normal phase high-performance liquid chromatography (HPLC), as described previously (Jin et al., 2009). In brief, retinoids in hexane extractions were evaporated, dissolved in 100 µl of hexane, and separated on a silica column (Zorbax Sil, 5 µm, 250 × 4.6 mm, Agilent Technologies) by gradient (0.2–10% dioxane in hexane at 2.0 ml/min flow rate) elution of the mobile phase on an Agilent 1100 HPLC system. Spectral data were acquired for all eluted peaks. Quantitation was performed by comparison of peak areas to calibration curves established with authentic retinoid standards.

Immunohistochemistry

Retinal cryosections prepared from the dorsal-ventral midline of mouse eyes were immunostained as described previously (Sato et al., 2013). Briefly, enucleated mouse eyeballs were fixed with 4% paraformaldehyde in 0.1 M phosphate buffer (PB). After removing the cornea and lens, eyecups were immersed in 15% sucrose in 0.1 M PB for 2 h, in 30% sucrose in 0.1 M PB for 2 h, and then in a 1:1 mixture of 30% sucrose and Optimal Cutting Temperature medium (Sakura Finetechnical) overnight at 4°C. Sections cut on a cryostat were immunostained with the primary antibodies and secondary antibodies. Nuclei were counterstained with 4′-6′-diamidino-2-phenylindole (DAPI) (MilliporeSigma). Images were captured with a Zeiss LSM710 Meta confocal microscope. We recorded 2 × 1, 3 × 2, or 5 × 5 composite images directly in the Tile Scan mode. Numbers of cone cells were counted using ImageJ software (National Institutes of Health).

Electroretinography

Dark-adapted mice were anesthetized with an intraperitoneal injection of ketamine–xylazine mixture, and the pupils were dilated with 1% tropicamide. ERG was recorded from the corneal surface using a silver–silver chloride wire electrode referenced to a subcutaneous electrode in the mouth. A needle electrode in the tail served as the ground. A drop of 2.5% methylcellulose was placed on the cornea to ensure good electrical contact and to prevent corneal desiccation during the entire procedure. ERG recordings were performed in a Ganzfeld dome (Espion e2, Diagnosys) with various intensities of single flash stimuli. For scotopic ERG, single flash stimuli (−4 log cd·s/m² ∼ 2.4 log cd·s/m²) were presented in interstimulus intervals of 0.5–2 min (depending on the stimulus intensity). Four responses were averaged for each step. For photopic ERGs, animals were light adapted for 10 min at a light intensity of 32 cd/m², and ERG responses were obtained with white flashes on the rod-saturating background (32 cd/m²) light (Oh et al., 2008). For recording ERG responses of S- and M-cones, animals were light adapted for 10 min at a white light intensity of 40 cd/m². S-cone ERGs were obtained with xenon flashes with a Hoya U-360 filter on a 40 cd/m² background light, and M-cone ERGs were elicited with 530 nm green light stimuli on the 40 cd/m² background light (Oh et al., 2008). Eight responses to 10 s interval flashes were averaged for each step. Intensity response amplitude data were displayed on log-linear coordinates using the SigmaPlot 11 software.

Experimental design and statistical analysis

All statistical analyses were performed using the SigmaPlot version 11 (Systat Software). Data were expressed as the mean ± standard deviation (SD) of three or more independent experiments. The sample sizes used in this study were based on our previous similar studies (Li et al., 2016, 2020a) and were similar to those generally accepted in the field. Male and female animals were used, and data were pooled from both sexes, except where noted. Significant phenotypic differences between distinct genotypes of mice were determined by single comparisons with an unpaired two-tailed Student’s t test. The significance threshold was set at 0.05 for all statistical tests.

Results

Ablation of FATP4 increased 9cRAL but not 11cRAL in rd12 mice

We recently showed that FATP4 deficiency increased the synthesis of both 11cRAL and 9cRAL chromophores in the Rpe65-R91W mutation knock-in mouse (R91W) model of LCA (Li et al., 2020a). Rpe65^−/− mice have been shown to produce 9cRAL in dark-reared condition (Fan et al., 2003). To investigate whether FATP4 deficiency promoted 9cRAL formation through a RPE65-idependent pathway or via a mechanism involving the intrinsic activity of R91W-RPE65 mutant, we generated RPE65-null Fatp4^−/− (rd12;Fatp4^−/−) mice using the rd12 and Fatp4^−/− mutant lines. We confirmed genotypes of these mutant lines by immunoblot analysis. As shown in Figure 1A, expression level of FATP4 in the rd12 mouse eyecup was comparable with that in the WT eyecup, while it was undetectable in the eyecups of rd12;Fatp4^−/− mouse. RPE65 was undetectable in the rd12 and rd12;Fatp4^−/− eyecups whereas LRAT was expressed at comparable levels in these mice (Fig. 1A).

Figure 1.

Identification of 9-cis-retinal increased in the eyes of rd12 mice lacking FATP4. A, Immunoblot analysis of FATP4, RPE65, and LRAT in the RPE with the indicated genotypes. Actin was detected as a loading control. B, HPLC chromatogram of 9cRAL and atRAL standards converted into the corresponding syn- and anti-isomers of 9-cis or all-trans retinaloximes (Rox). C–E, Representative HPLC chromatograms of retinoids extracted from the eyecups of dark-adapted WT (C), rd12 (D), and rd12;Fatp4^−/− (E) mice. Peaks of all-trans-retinyl esters (atRE), syn-11-cis retinaloxime (syn-11cRox), syn-all-trans retinaloxime (syn-atRox), 11-cis-retinol (11cROL), anti-11-cis retinaloxime (anti-11cRox), all-trans-retinol (atROL), anti-all-trans retinaloxime (anti-atRox), syn-9-cis retinaloxime (syn-9cRox), and anti-9-cis retinaloxime (anti-9cRox) are marked.

We then analyzed retinoids in overnight dark-adapted eyecups of WT and mutant mice by HPLC. To stabilize the highly reactive retinals, we converted the cis- and trans-retinals into the corresponding syn- and anti-isomers of retinaloximes (Roxs) using hydroxylamine (Lee et al., 2016). Figure 1B shows HPLC chromatogram of the Rox isomers converted from the 9cRAL and atRAL standards. Consistent with previously published studies (Fan et al., 2003, 2006), individual eyecups of WT mice contained undetectable or noise level amounts of 9cRAL in the HPLC analysis (Fig. 1C). In contrast, each eyecup of the rd12 mice contained a significantly greater amount of 9cRAL, displaying clear peaks of both syn- and anti-9-cis Roxs (Fig. 1D). Under the same experimental conditions, the peaks of syn- and anti-9-cis Roxs from the rd12;Fatp4^−/− mouse eyecup were clearly higher than those from the rd12 eyecups (Fig. 1E). In contrast to 9cRAL, 11cRAL was undetectable in both rd12 and rd12;Fatp4^−/− mouse eyecups (Fig. 1D,E). This observation is consistent with previous studies demonstrating the absence of 11cRAL in Rpe65 knock-out mice (Redmond et al., 1998; Fan et al., 2003).

FATP4 inhibited both RPE65-independent and R91W mutant RPE65-dependent formation of 9cRAL

To determine whether R91W-RPE65 is involved in the synthesis of 9cRAL, we compared the formation of 9cRAL and other retinoids in overnight dark-adapted eyecups of rd12 and R91W mice. Quantitative analysis of retinoids showed that the amounts of 9cRAL in the R91W mouse eyecup were clearly greater than those in the rd12 eyecup (8.4 vs 5.5 pmol per eyecup), showing a 53% increase in the R91W mouse eyes (Fig. 2A). This result suggests that a portion of 9cRAL is produced through a mechanism involving R91W-RPE65 and may be associated with the enzymatic nature of RPE65 isomerase. RPE65 is not an inherently 11-cis-specific retinoid isomerase and its WT and mutant forms can produce 11-cis, 13-cis, and 9-cis-retinols in vitro (Jin et al., 2007; Redmond et al., 2010; Takahashi et al., 2012).

Figure 2.

Comparison of ocular retinoid and RALDH1 expression profiles in WT and rd12 mice with or without FATP4. Quantification of 9-cis-retinal (A), 11-cis-retinal (B), all-trans-retinol (C), and all-trans-retinyl esters (D) in the eyecups of overnight dark-adapted WT, Fatp4^−/−, rd12, rd12;Fatp4^−/−, R91W, and R91W;Fatp4^−/− mice. E, Immunoblot analysis showing expression of RALDH1 in the retina and RPE of WT, rd12, and rd12;Fatp4^−/− mice. F, Expression levels of RALDH1 in the retina or RPE of rd12 and rd12;Fatp4^−/− mice are expressed as fold of RALDH1 expression level in the WT retina or RPE.

We then assessed the effect of FATP4 deficiency on the formation of 9cRAL in the rd12 and R91W mouse eyes by measuring the contents of 9cRAL in overnight dark-adapted eyecups of rd12;Fatp4^−/− and R91W;Fatp4^−/− mice. Lack of FATP4 increased the quantity of 9cRAL in both rd12 and R91W mice. The rd12;Fatp4^−/− and R91W;Fatp4^−/− eyecups contained 8.8 or 14 pmol 9cRAL per eyecup, respectively, indicating that FATP4 deficiency resulted in a 59.8 or 66.7% increase in 9cRAL synthesis in the rd12 or R91W mice. In contrast, lack of FATP4 did not significantly alter the synthesis of 11cRAL, all-trans-retinol and atRE in the eyes with the same genotypes for the Rpe65 gene (Fig. 2B–D).

As the amounts of total retinoids and atRE in the rd12 and rd12;Fatp4^−/− mouse eyecups were significantly greater than those in WT mice, we tested whether expression of retinaldehyde dehydrogenase (RALDH) was changed in the retina and RPE of the mutant mice (Harper et al., 2015). As shown in Figure 2E,F, the expression levels of RALDH1 in the retina and RPE of rd12 and rd12;Fatp4^−/− mice were ∼2–2.5-fold higher than those in the WT mouse retina and RPE.

FATP4 deficiency did not alter rod function and degeneration in rd12 mice

Previous studies have shown that maintaining Rpe65^−/− mice in constant darkness can either increase 9cRAL synthesis or reduce rod degeneration, depending on the duration in darkness (Fan et al., 2003, 2005). As 9cRAL was increased by FATP4 deficiency in rd12 mice (Figs. 1, 2), we tested whether rod degeneration was reduced in rd12;Fatp4^−/− mice, as compared with rd12 mice. Immunoblot analysis showed that expression levels of rhodopsin in the 2- and 4-month-old rd12;Fatp4^−/− retinas were similar to those in age-matched rd12 retinas (Fig. 3A,B). Immunohistochemical analysis of rhodopsin also showed similar degrees of rod outer segment (OS) degeneration in 2-month-old rd12 and rd12;Fatp4^−/− mice (Fig. 3C). Furthermore, rd12 and rd12;Fatp4^−/− mice exhibited comparable visual function of rods in scotopic ERG tests (Fig. 3D,E). These data indicate that although FATP4 deficiency induced a ∼60% increase in the 9cRAL synthesis, it is not enough to improve rod survival and function in young adult rd12 mice. These results are in agreement with the previous studies showing undetectable rod a-wave amplitude in 1 d dark-reared Rpe65^−/− mice and the similar degrees of rod degeneration in dark-reared Rpe65^−/− and Lrat^−/− mice (Fan et al., 2003, 2008), with the latter mice contain a significantly smaller amount of isorhodopsin compared with Rpe65^−/− mice (Fan et al., 2008).

Figure 3.

Comparable rod degeneration and scotopic ERGs in rd12 and rd12;Fatp4^−/− mice. A, B, Immunoblot analysis of rhodopsin (Rho) in the retinas of 2-month-old (A) or 4-month-old (B) WT, rd12, and rd12;Fatp4^−/− mice. Rho monomer, dimer, and trimer are indicated. Histograms show relative expression levels of rhodopsin (including monomer, dimer, and trimer) in WT, rd12, and rd12;Fatp4^−/− mice at the indicated ages. C, Immunohistochemical analysis of Rho in 2-month-old WT, rd12, and rd12;Fatp4^−/− retinal sections. The outer and inner nuclear layers (ONL and INL) were counterstained with DAPI (blue). OS, outer segments; IS, inner segments. D, Representative scotopic ERG responses of rods in overnight dark-adapted WT, rd12, and rd12;Fatp4^−/− mice to the indicated light flashes. E, Amplitudes of scotopic ERG a- and b-waves evoked with the indicated flashes in 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. Error bars show SD (n = 10). F, G, Immunoblot analyses of PKCα (F) and mGluR6 (G) in the neural retina and RPE of 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. Actin was used as loading control. Histograms show relative expression levels of PKCα and mGluR6 in the neural retinas of the WT and mutant mice.

As b-wave amplitudes of scotopic ERGs in rd12 and rd12;Fatp4^−/− mice were markedly lower than those in WT mice (Fig. 3D,E), we investigated whether ON bipolar cells (ON-BC) were degenerated in rd12 and rd12;Fatp4^−/− mice. ON-BC are known to produce b-wave responses (Stockton and Slaughter, 1989; Hood and Birch, 1996). Immunoblot analyses showed that expression levels of the ON-BC synaptic proteins, mGluR6 and PKCα, in the retinas of 2-month-old rd12 and rd12;Fatp4^−/− mice were similar to those in the age-matched WT retina (Fig. 3F,G). These results may be consistent with the slight (15–20%) reduction of the ONL thickness in rd12 and rd12;Fatp4^−/− mice (Fig. 3C) and suggest that the diminished b-wave responses in rd12 and rd12;Fatp4^−/− mice are due to the significant reduction of rod phototransduction activity (a-wave amplitudes) rather than degeneration or reduced synaptic function of ON-BC in these mutant mice.

M-cone vision is improved in rd12 mice lacking FATP4

Light adaptation in photopic vision involves significant depletion of the visual pigments and cones maintain their visual function even after the vast majority of the visual pigments bleached (Jones et al., 1993; Burkhardt, 1994). In addition, the cone OS disks containing cone opsins are directly associated with the interphotoreceptor matrix containing interphotoreceptor retinoid-binding protein that has been shown to support cone vision (Jin et al., 2009; Parker et al., 2009). These facts suggest that a small increase in the synthesis of 9cRAL, a functional isochromophore (Crouch et al., 1975; Fan et al., 2003), may improve cone-mediated visual function in the rd12;Fatp4^−/− mice. We therefore tested this possibility by recording photopic ERG responses of WT, rd12, and rd12;Fatp4^−/− mice under a rod-saturating background light. The mice were adapted to the rod-saturating light (32 cd/m²) for 10 min, before inducing ERG responses with a series of increasing achromatic flashes. As shown in Figure 4A–C, amplitudes of a- and b-waves elicited with flash intensities bigger than 1 log or 0 log cd.s/m² were significantly higher in rd12;Fatp4^−/− mice than rd12 mice. These results suggest that cone visual function is partially rescued in rd12 mice by FATP4 deficiency. To confirm this result, we measured b-wave implicit times in photopic and scotopic ERGs of rd12 and rd12;Fatp4^−/− mice. Previous studies have shown that patients with incomplete achromatopsia caused by mutations in cone-specific genes, such as CNGA3, CNGB3, and PDE6C, display markedly decreased b-wave amplitudes and prolonged b-wave implicit times in photopic ERGs (Trankner et al., 2004; Jimenez-Siles et al., 2023). We found that b-wave implicit times of photopic, but not scotopic, ERGs of rd12;Fatp4^−/− mice were significantly shorter than those of rd12 mice, while photopic b-wave implicit times of WT mice were far shorter than those of rd12 and rd12;Fatp4^−/− mice (Fig. 4D). These results and the comparable expression levels of the synaptic proteins (Fig. 3F,G) suggest that the improvement of both a- and b-wave responses in rd12;Fatp4^−/− mice is due to partial rescue of cone visual function, rather than enhancement in synaptic function or less degeneration of the ON bipolar cells.

Figure 4.

FATP4 deficiency improved M-cone but not S-cone vision in rd12 mice. A, Representative photopic ERG responses of 8-week-old WT, rd12, and rd12;Fatp4^−/− mice to a series of increasing flashes (1–100 cd·s/m²) of white light under a rod-saturating background light. B, C, Amplitudes of photopic a- and b-waves evoked with the indicated white light flashes. D, Times from stimulus onset to peaks of photopic b-waves evoked with the indicated white light flashes in WT, rd12, and rd12;Fatp4^−/− mice. Asterisks indicate statistically significant differences between rd12 and rd12;Fatp4^−/− mice (*p < 0.02); error bars denote SD (n = 6–7). E, Representative ERG responses of M-cones in 8–10-week-old WT, rd12, and rd12;Fatp4^−/− mice to the indicated flash intensities of 530 nm green light. The ERG responses were evoked with a series of increasing stimuli (log −1∼1 cd·s/m²) of 530 nm light. F, Amplitudes of M-cone b-waves evoked with the indicated flash intensities of the green light. G, Amplitudes of S-cone ERG b-waves evoked with the indicated flash intensities of 360 nm UV light.

To identify the types of cones that contribute to the improvement of the above photopic ERG responses in rd12;Fatp4^−/− mice, we performed ERG tests with a series (−1∼1 log cd·s/m²) of 530  nm green or 360  nm UV light stimuli under a rod-saturating background light (40 cd/m²). As shown in Figure 4E,F, the b-wave responses elicited with 530 nm light flashes bigger than 0 log cd·s/m² were significantly improved in 8-week-old rd12;Fatp4^−/− mice, as compared with age-matched rd12 mice. These results and the scotopic ERGs (Fig. 3D,E) suggest that FATP4 deficiency has improved visual function of M-cones in rd12 possibly via the increase in formation of 9cRAL. We then evaluated S-cone visual function by measuring S-cone ERG responses induced with 360 nm UV light flashes. In contrast to the M-cone ERGs, S-cone ERGs were not substantially improved in rd12;Fatp4^−/− mice, as compared with rd12 mice (Fig. 4G).

FATP4 deficiency mitigated M-cone degeneration in the superior retina of rd12 mice

Previous studies have suggested that the rod system is the source of vision in RPE65 deficiency (Seeliger et al., 2001). We therefore investigated whether the enhanced M-cone ERGs are correlated with improved preservation of M-cones in rd12;Fatp4^−/− mice. We first analyzed expression levels of M-opsin in whole retinal homogenates by immunoblot analysis. We observed that the abundance of M-opsin in 1-month-old rd12;Fatp4^−/− retina was significantly greater than that in age-matched rd12 mice (Fig. 5A,B). We then analyzed expression patterns of M-opsin in the inferior and superior halves of the rd12 and rd12;Fatp4^−/− retinas because M- and S-cones in the mouse retina are concentrated in two opposite areas and form gradients along the inferior-superior axis (Szel et al., 1993; Calderone and Jacobs, 1995). In 2-month-old WT 129S2/Sv strain mice, ∼90% of M-opsin are present in the superior half of the retina (Li et al., 2020a). Immunoblot analysis showed that expression level of M-opsin in the superior, but not inferior, retina of 2-month-old rd12;Fatp4^−/− mice were clearly higher than that in age-matched rd12 mice (Fig. 5C). Quantitative analysis of the immunoblots revealed that the content of M-opsin in the superior halves of the rd12;Fatp4^−/− retina was nearly 2.5-fold greater than that in the rd12 superior retina (Fig. 5D). In agreement with these results, immunohistochemistry showed that the number of M-cones preserved in the superior retina of 2-month-old rd12;Fatp4^−/− mice was greater than that in age-matched rd12 superior retina (Fig. 5E).

Figure 5.

Alleviation of M-cone degeneration in the superior retina of rd12;Fatp4^−/− Mice. A, Immunoblot analysis of M-opsin in the retinas of 1-month-old WT, rd12, and rd12;Fatp4^−/− mice. Actin was detected as loading control. B, Expression levels of M-opsin in rd12 and rd12;Fatp4^−/− retinas are normalized with actin levels and shown as percent of M-opsin levels in the WT retina. C, Immunoblot analysis of M-opsin in the inferior or superior halves of 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. D, Normalized expression levels of M-opsin in the superior or inferior retinas of rd12;Fatp4^−/− mice (in C) are shown as fold of M-opsin levels in the rd12 superior or inferior retinas. Asterisk indicates significant differences between rd12 and rd12;Fatp4^−/− mice; n.s. denotes no significance. E, Immunohistochemistry of M-opsin in the superior and inferior retinas of 2-month-old WT, rd12, and rd12;Fatp4^−/− mice.

Alleviation of S-cone degeneration in the superior retina of rd12 mice lacking FATP4

Immunoblot analysis showed that ∼80% of total S-opsin was present in the inferior half of 1-month-old WT 129S2/Sv strain retina. However, S-opsin in the same regions was reduced to undetectable level in both rd12 and rd12;Fatp4^−/− mice of the same age (Fig. 6A,B). While ∼20% of total S-opsin were present in the superior half of WT retina, the overwhelming majority of S-opsin disappeared in the same area of rd12 mice; in contrast, a significant amount of S-opsin remained in the superior retina of rd12;Fatp4^−/− mice (Fig. 6A–C). The quantity of S-opsin in the rd12;Fatp4^−/− superior retina was approximately threefold of that in the rd12 superior retina and was ∼30% of WT level in the same area (Fig. 6A–C). As WT superior retina contained only 20% of total S-opsin, the amount of S-opsin remaining in the rd12;Fatp4^−/− retina was ∼7% of total S-opsin in the WT retina.

Figure 6.

FATP4 deficiency mitigated S-cone degeneration in the superior retina of rd12 mice. A, Immunoblot analysis of S-opsin in the inferior or superior halves of 1-month-old WT, rd12, and rd12;Fatp4^−/− retinas. Note that the majority of S-opsin is distributed in the inferior half of WT retina, while in the rd12 and rd12;Fatp4^−/− inferior retinas S-opsin is almost undetectable. B, Percentages of S-opsin included in the inferior or superior halves of WT retinas in A. Levels of S-opsin in the inferior and superior retinas of rd12 and rd12;Fatp4^−/− mice are shown as percent of S-opsin levels in WT. C, S-opsin level in the WT superior retina is set as 100% and the levels of S-opsin in the superior retinas of rd12 and rd12;Fatp4^−/− mice are shown as percent of S-opsin levels in WT. D, Immunohistochemical analysis of S-opsin in the superior and inferior retinas of 1-month-old WT, rd12, and rd12;Fatp4^−/− mice. E, Immunoblot analysis of S-opsin in the indicated micrograms of total retinal homogenates from 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. F, Immunohistochemistry of S-opsin in the superior and inferior retinas of 2-month-old rd12 and rd12;Fatp4^−/− mice. Scale bars denote 100 μm.

Consistent with the results of immunoblot analysis, immunohistochemistry showed that the number of S-opsin-positive cones (S-cones) in the superior retina of 1-month-old rd12;Fatp4^−/− mice were more than twofolds of that in the superior retina of age-matched rd12 mice (Fig. 6D). Compared with the superior retina, the inferior retinas of both rd12 and rd12;Fatp4^−/− mice exhibited much more severe degeneration of S-cones, while the density of S-cones in the WT inferior retina was significantly higher than that in the WT superior retina (Fig. 6D). The number of S-cones in the rd12;Fatp4^−/− inferior retina was slightly greater than that in the rd12 inferior retina, but only a very small population of S-cones remained in the inferior retina (Fig. 6D). In addition, immunohistochemistry revealed a significant mislocalization of S-opsin in rd12 and rd12;Fatp4^−/− mice, as evidenced by localization of S-opsin to the cell body and synaptic areas of S-cones (Fig. 6D). These phenotypes of S-cones might contribute to the lack of significant difference in S-cone ERGs between rd12 and rd12;Fatp4^−/− mice.

To confirm the observations described above and to determine whether FATP4 deficiency slows down the progression of S-cone degeneration in rd12 mice, we analyzed S-opsin expression in 2-month-old rd12 and rd12;Fatp4^−/− mice. Immunoblot analysis showed that S-opsin was reduced to an undetectable level in the rd12 retina, while in the rd12;Fatp4^−/− retinas, S-opsin was still detectable, although the signal was very weak (Fig. 6E). In agreement with these results, immunohistochemistry showed that the number of S-cones remaining in the inferior and superior retinas of rd12;Fatp4^−/− mice were clearly greater than those in the same retinal regions of rd12 mice (Fig. 6F). These observations suggest that the progression of S-cone degeneration in rd12 is faster than that in rd12;Fatp4^−/− mice.

FATP4 deficiency slowed down progression of cone degeneration in rd12 mice

Cones lacking cone opsins do not form the outer segments, but they survive for several months in mouse model (Xu et al., 2022). To determine the impacts of FATP4 deficiency on the progression of degeneration of M- and S-cones in rd12 mice, we performed immunoblot and immunochemical analyses of cone arrestin (CAR) in rd12 and rd12;Fatp4^−/− mice at different ages. CAR is expressed in both M- and S-cones. Also, it is localized to the OS, cell body, and synaptic region of cones. Therefore, its immunostaining can assess degeneration of whole cone cells. Immunoblot and immunohistochemical analyses showed that, in contrast to M- and S-cones, CAR-positive cones are relatively evenly distributed in the superior and inferior retinas of 2-month-old WT mice (Fig. 7A,C). In the inferior retina of 2-month-old rd12 mice, CAR protein and CAR-positive cells were reduced to almost undetectable levels (Fig. 7A–C), confirming that most cones (including M- and S-cones) are degenerated in the inferior retina of rd12 mice at 2 months of age. The lack of detectable CAR in the rd12 inferior retina may reflect the rapid degeneration of S-cones, which are dominant in the inferior retina, and is consistent with previous studies showing rapid degeneration of S-cones in Rpe65^−/− and Lrat^−/− mice (Znoiko et al., 2005; T. Zhang et al., 2011). In contrast to the inferior retina, the superior retina of 2-month-old rd12 mice contained some CAR protein and CAR-positive cells (Fig. 7A–D). Importantly, FATP4 deficiency significantly alleviated degeneration of CAR-positive cells in both inferior and superior retinas of 2-month-old rd12 mice (Fig. 7A–D). The numbers of CAR-positive cones preserved in both superior and inferior retinas of 2-month-old rd12;Fatp4^−/− mice were 1.7-fold greater than those in age-matched rd12 mice (Table 1). The protective effect of FATP4 deficiency on cone survival in the superior retina of rd12 mice was persistent until at least 5 months of age (Fig. 7E). As a result, the difference in number of CAR-positive cones preserved in the superior retinas of rd12 and rd12;Fatp4^−/− mice was enlarged to 2.5-fold at 5 months of age (Table 1).

Figure 7.

FATP4 deficiency slowed cone degeneration rates in rd12 mice. A, Immunoblot analysis of cone arrestin (CAR) in the indicated micrograms of the superior and inferior retinal homogenates from 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. Actin was detected as loading control. B, Normalized expression levels of CAR in the superior or inferior retinas of 2-month-old rd12 and rd12;Fatp4^−/− mice are shown as percent of CAR levels in the superior or inferior halves of WT retina. C, Immunohistochemical analysis of CAR in retinal sections from 2-month-old WT, rd12, and rd12;Fatp4^−/− mice. ON, optic nerve head; sup, superior; inf, inferior. D, Higher-magnification images of the areas of rectangles shown in C. E, Immunohistochemistry of CAR in the superior retinas of 5-month-old rd12 and rd12;Fatp4^−/− mice.

Table 1.

Numbers of CAR-positive cones in the superior and inferior retinal sections^a

Age	rd12	rd12;Fatp4−/−	p values	Fold of rd12
2-month	Superior 69 ± 9	Superior 117 ± 7	<0.01 n = 3	1.7
	Inferior 3 ± 1	Inferior 5 ± 2	<0.15 n = 3	1.7
5-month	Superior 39 ± 8	Superior 96 ± 6	<0.01 n = 3	2.5
	Inferior 0	Inferior 1	$\mathrm{\infty }$	$\mathrm{\infty }$

^aThe retinal sections were taken from the dorsal-ventral midline of the eyes.

Discussion

Identifying the endogenous factors that support cone survival and function in physiological and pathological conditions is critically important not only for understanding the regulatory mechanisms of daytime color vision, but also for the development of therapeutic interventions alleviating cone degeneration and central vision loss in patients with cone degeneration. In the present study, we provided compelling evidence that (1) FATP4 is a negative regulator of 9cRAL formation in the eyes that lack RPE65 or have hypomorphic Rpe65 alleles associated with LCA; (2) endogenous 9cRAL is a cone-tropic chromophore critical for survival of M- and S-cones in the superior retina of rd12 mice; and (3) 9cRAL promotes visual function of M-cones in rd12 mice.

Although the light-independent formation of 9cRAL was first observed >2 decades ago in dark-reared Rpe65^−/− mice (Fan et al., 2003), the underlying molecular mechanism of 9cRAL synthesis remains unclear. Consistent with previous studies (Van Hooser et al., 2000; Fan et al., 2003), 9cRAL was undetectable in overnight dark-adapted eyes of WT mice and Fatp4^−/− mice with wild-type Rpe65 alleles (Fig. 1). In contrast, both rd12 and R91W knock-in (KI) mouse eyes contained significantly increased amounts of 9cRAL under the same dark-adapted conditions (Figs. 1, 2). Rpe65^−/− and R91W-KI mice have been shown to produce a massive accumulation of atRE in the RPE (Redmond et al., 1998; Samardzija et al., 2008) due to lack or severe reduction of RPE65 function, which hydrolyzes atRE to synthesize 11-cis-retinol (Moiseyev et al., 2003). We observed that the amounts of atRE in overnight dark-adapted rd12 and R91W-KI mouse eyes with or without FATP4 were 20-fold greater than those in WT and Fatp4^−/− mice, while the amounts of atROL in dark-adapted rd12 and rd12;Fatp4^−/− mouse eyes were twofolds greater than those in the WT and Fatp4^−/− mouse eyes (Fig. 2C,D). Importantly, RALDH1 that catalyzes the synthesis of all-trans retinoic acid (RA; Harper et al., 2015) was markedly upregulated in the retina and RPE of rd12 and rd12;Fatp4^−/− mice (Fig. 2E,F). These data suggest that excessive RA signaling may contribute to the 9cRAL formation through its nuclear receptors, such as RA receptors (RARs) and retinoid X receptors (RXRs; von Lintig et al., 2021), in the RPE of rd12 and R91W mice. All six RAR and RXR subtypes are expressed in the RPE cells (Dwyer et al., 2011), suggesting that excessive formation of RA may cause transcriptional induction of the enzyme(s) involved in the synthesis of 9cRAL to compensate for the shortage of 11cRAL supply in rd12 and R91W mice. RA is not a functional chromophore of the visual pigments in rods and cones. Therefore, excessive RA formation cannot directly support visual function of rd12 and R91W mice, although it may improve photoreceptor survival via its transcriptional activity (Amamoto et al., 2022).

An important finding of the present study is the identification of FATP4 as a new regulator of 9cRAL synthesis in rd12 mice. We observed that ablation of FATP4 did not induce the formation of 11cRAL, but it significantly increased the formation of 9cRAL in rd12 mice in a light-independent mechanism (Figs. 1, 2). We also confirmed that lack of FATP4 in the RPE increased the formation of 9cRAL in dark-adapted R91W mice (Li et al., 2020a). As a single transmembrane protein, FATP4 facilitates transport of long-chain fatty acids across cell membrane. It also exhibits enzymatic activity promoting synthesis of fatty acyl-CoA esters (Jia et al., 2007), which have been shown to suppress the peroxisome proliferator-activated receptor gamma (PPARγ) transcription factor (JØrgensen et al., 2002). These studies suggest that FATP4 deficiency may upregulate PPARγ. All three subtypes of PPARs are expressed in the human RPE of adult donors (Dwyer et al., 2011). PPARs form heterodimers with RXRs to regulate gene transcription (Ziouzenkova and Plutzky, 2008). RA is not only an activating ligand for RARs, but also a high affinity ligand for PPARβ/δ (Shaw et al., 2003). Ablation of PPARβ/δ results in aberrant pigmentation in the RPE (Choudhary et al., 2016). Interestingly, 9cRAL synthesis is significantly increased in the eyes of Rpe65^−/− mice with decreased melanin synthesis (Fan et al., 2006). These observations and our data suggest that ablation of FATP4 may further promote RA signaling linked to induction of 9cRAL synthesis in the RPE of rd12 mice. On the other hand, it is also possible that FATP4 regulates 9cRAL synthesis as well as cone survival and function through a mechanism/s that has not been identified yet. Nevertheless, it will be interesting to investigate whether there is any relationship between FATP4 and melanin synthesis.

Another important finding of this study is the identification of the endogenous 9cRAL as a cone-tropic chromophore that supports M-cone vision and survival of both M- and S-cones in the superior retina of RPE65 deficient mice. Early in vitro or ex vivo studies have established 9cRAL as a photosensitive chromophore of isorhodopsin (Crouch et al., 1975; Pepperberg et al., 1976). The endogenous 9cRAL has also been identified as a chromophore of isorhodopsin that supports scotopic vision in dark-reared Rpe65^−/− mice (Fan et al., 2003). In mouse retina, only ∼4% of photoreceptors are cones (Carter-Dawson and LaVail, 1979), and cones are less sensitive to light (Fain and Dowling, 1973). In addition, rods entrap limited amount of 11cRAL chromophore in R91W mice (Samardzija et al., 2009). Based on these studies, we predicted somewhat improvement of rod function in rd12;Fatp4^−/− mice because FATP4 deficiency increased the synthesis of 9cRAL. However, both a- and b-wave amplitudes of scotopic ERGs in rd12;Fatp4^−/− mice were similar to those in rd12 mice (Fig. 3). Consistent with this observation, rd12 and rd12;Fatp4^−/− mice exhibited comparable degrees of rod degeneration (Fig. 3). Unexpectedly, we observed that M-cone ERGs were significantly improved in rd12;Fatp4^−/− mice, as compared with rd12 mice (Fig. 4). This improvement was supported by a substantial increase in preservation of M-cones in rd12;Fatp4^−/− mice (Fig. 5). A possible explanation for our results is that 9cRAL is a cone-tropic chromophore possibly due to its stable or higher binding affinity for cone opsins. This possibility is supported by previous studies demonstrating that nonbleached 11-cis chromophores of salamander cone pigment, but not rhodopsin, can be substituted with 9cRAL under dark condition (Kefalov et al., 2005). In addition, structural difference in rod and cone OS membrane disks may also contribute to effective supply of 9cRAL to cone opsins. Unlike the rod disks, which are physically sequestered by the surrounding rod plasma membrane, cone disks are directly associated with the interphotoreceptor retinoid-binding protein that plays a pivotal role in supporting cone vision (Jin et al., 2009; Parker et al., 2009). While we cannot rule out the possibility that rods are involved in the green light-evoked ERGs, the significant increase in cone survival (Figs. 5–7) is consistent with the “cone-tropic” function of 9cRAL.

Administration of 9cRAL or 9-cis-retinyl acetate has been shown to improve cone survival and function in Rpe65^−/−, Gnat1^−/−;Rpe65^−/−, or Rpe65^−/−;Rho^−/− mice (Znoiko et al., 2005; Maeda et al., 2009; Tang et al., 2010). In human clinical trial of QLT091001/9-cis-retinyl acetate treatment, improvement of both rod and cone visual function is also observed in some patients with RPE65 or LRAT mutations (Koenekoop et al., 2014; Scholl et al., 2015). These studies have demonstrated the ability of 9cRAL to improve cone vision. In addition, these studies support the potential mechanism that FATP4 deficiency improves cone survival and function through increasing 9cRAL formation, rather than FATP4 deficiency itself. However, these studies have not identified 9cRAL as a cone-tropic chromophore, possibly due to difficulty in isolating the cone-specific effect of large amounts of exogenous 9cRAL and QLT091001, which have exhibited much wider effects, including side effects.

The expression level of S-opsin and the number of S-cones remaining in the superior retina were increased more than twofolds in rd12;Fatp4^−/− mice, compared with rd12 mice (Fig. 6). However, we failed to observe improvement of S-cone function in rd12;Fatp4^−/− mice (Fig. 4). The exact reason for this is currently unknown. Approximately 80% of S-opsin were present in the inferior half of WT retina; however, S-opsin was reduced to undetectable levels in the rd12;Fatp4^−/− inferior retinas (Fig. 6). Therefore, total amount of S-opsin in 1-month-old rd12;Fatp4^−/− mice was <7% of total S-opsin in WT retina. Furthermore, a significant portion of S-opsin was mislocalized in rd12;Fatp4^−/− mice (Fig. 6). These phenotypes might contribute to the lack of improvement of S-cone function in rd12;Fatp4^−/− mice.

Slowing the progression of cone degeneration and vision loss could greatly extend the years of improved quality of life in patients. FATP4 deficiency significantly slowed down the progression of cone degeneration. As a result, the number of cones surviving in 5-month-old rd12;Fatp4^−/− mice was 2.5-fold greater than that in age-matched rd12 mice (Fig. 7 and Table 1). These findings suggest that targeting FATP4 and the endogenous 9cRAL is a therapeutic strategy to delay cone degeneration and vision loss in retinal dystrophies associated with shortage in 11cRAL synthesis.