Genetic Background and Light-Dependent Progression of Photoreceptor Cell Degeneration in Prominin-1 Knockout Mice

Abstract

Purpose.: Mutations in the Prominin-1 (Prom1) gene are known to cause retinitis pigmentosa and Stargardt disease, both of which are associated with progressive photoreceptor cell death. There are no effective therapies for either disorder. The aim of this study was to investigate the mechanism of the retinal degeneration in Prom1-deficient mouse models.

Methods.: We constructed Prom1 knockout mice with two distinct genetic backgrounds of C57BL/6 and C57BL/6xCBA/NSlc, and investigated the photoreceptor degeneration by means of histology and functional tests.. In addition, we examined the effect of light on the Prom1^−/− retina by rearing the mice in the normal light/dark cycle and completely dark conditions. Finally, we investigated if the retinoic-acid derivative Fenretinide slowed the pace of retinal degeneration in these mouse models.

Results.: The Prom1^−/−-knockout mice with both backgrounds developed photoreceptor degeneration after eye opening, but the CB57/BL6-background mice developed photoreceptor cell degeneration much faster than the C57BL/6xCBA/NSlc mice, demonstrating genetic background dependency.. Interestingly, our histologic and functional examination showed that the photoreceptor cell degeneration of Prom1-knockout mice was light-dependent, and was almost completely inhibited when the mutant mice were kept in the dark. The Prom1-knockout retina showed strong downregulation of expression of the visual cycle components, Rdh12 and Abca4. Furthermore, administration of Fenretinide, which lowers the level of the toxic lipofuscin, slowed the degeneration of photoreceptor cells.

Conclusions.: These findings improve our understanding of the mechanism of cell death in Prominin-1–related disease and provide evidence that fenretinide may be worth studying in human disease..

Prominin-1 (Prom1; NCBI Gene ID 8842; also known as RP41, AC133, CD133, MCDR2, STGD4, PROML1) encodes a 120 kDa pentaspan transmembrane glycoprotein,¹ originally identified as a surface antigen of human haematopoietic stem and progenitor cells.² Prom1 is expressed in a number of tissues throughout embryogenesis and adulthood,^1,3–6 and has been implicated in the maintenance of stem cell characteristics and cell proliferation.^6–14 In addition to these functions, Prom1 has been shown to be involved in a number of genetic disorders that cause photoreceptor degeneration, including retinitis pigmentosa (RP; OMIM 268000; autosomal-dominant and autosomal-recessive forms) and Stargardt's type 4 (STGD4; OMIM 603786).^15–17 

Retinitis pigmentosa encompasses a genetically heterogeneous group of inherited disorders causing progressive photoreceptor cell death^15,17 and affects approximately 1 in 4000 people. It usually starts with early rod photoreceptor damage, manifesting as nyctalopia. Later there is loss of peripheral visual field, and in advanced disease loss of central vision due to death of foveal cones. More than 60 causal genes have been identified, including those encoding transcription factors, splicing regulators, membrane proteins, and proteins involved in visual cycle.¹⁷ 

Stargardt's disease is one of the macular dystrophies, which predominantly affect the central retina. The most common form of the dystrophy, Stargardt's type 1 (STGD1; OMIM 248200) is an autosomal recessive disorder that is characterized by macular atrophy with or without flecks at the level of the retinal pigment epithelium^16,18 and has a prevalence of at least 1 in 10,000 births; it is usually diagnosed in the first or second decade of life,^19,20 and is recognized as the most common cause of macular disease in children. Stargardt's type 1 is caused by mutations in the ATP-binding cassette, sub-family A member 4 (ABCA4) gene (NCBI Gene ID 24; also known as RP19, ABCR).²¹ There are four types of Stargardt's disease known so far. Three other forms of macular dystrophy have some phenotypic similarities to STGD1, but some of them show autosomal dominant inheritance; STGD2/3 (OMIM 600110) caused by defects in the ELOVL Fatty Acid Elongase 4 (ELOVL4; NCBI Gene ID 6785) and STGD4 caused by Prom1 genes,^22–24 respectively. Although there have been a number of different therapeutic approaches suggested,²⁵ there are as yet no effective treatments for this group of disorders. 

Mutations in Prom1 have been reported in both RP (autosomal-dominant and autosomal-recessive forms) and STGD4.^15–17 A homozygous frameshift mutation of Prom1 gene, causing a premature termination of the translation, was found in an Indian pedigree and resulted in inherited retinal degeneration.²⁶ A homozygous mutation resulting in a truncated PROM1 protein was identified in a family with severe RP, where affected patients reported night blindness and decreased visual acuity in early childhood with subsequent severe loss of central vision in adult life.²⁷ More recent reports have identified recessive mutations in PROM1 in patients with cone–rod dystrophy.^28,29 In contrast to these autosomal recessive mutations, another mutation, Prom1/R373C (the mutation where the 373rd Arginine residue was substituted with Cysteine), is inherited as an autosomal dominant trait and causes STGD4 as well as bull's eye macular dystrophy (OMIM 608051).²⁴ In this mutation, onset of Prom1-mediated STGD4 varies from 10- to 29-years old with a mean age of 19 years. Thus, Prom1 is associated with a wide range of human retinal dystrophy phenotypes; the disease severity can vary significantly even with individuals who harbor the same disease causing mutation(s). 

Mouse models have been generated to recapitulate the retinal degeneration caused by PROM1. To date, a transgenic mouse expressing the human PROM1/R373C²⁴ and a Prom1-knockout mouse³⁰ have been established and both show retinal degeneration. In both models, the photoreceptor layer (outer and inner segment [OS/IS] layers) and the outer nuclear layer (ONL) start to degenerate at postnatal day 15 (P15), and are absent by 12 months.^24,30 

It is notable that the onset and progress of the diseases vary in patients, even within a single family.^{24,27,28,31,32} Progression of the disease is rapid in some patients, but relatively slow in others. The molecular mechanisms underlying the variation in progression of Prom1-mediated retinal degeneration are, however, still largely unknown, but may reflect variation in environmental factors or modifier genes. 

In this study, we have constructed new Prom1-knockout mice to recapitulate the haploinsufficient-type of photoreceptor degenerative diseases.^22,26 We raised the mutant mice under distinct genetic backgrounds, and found that they showed evidence of photoreceptor cell degeneration with a different rate of progression. Using these models, we investigated the molecular mechanisms of the disease phenotype. 

Methods

Animals and Housing Conditions

All animals in this study were subject to local and national ethical approval and guidance (University College London ethical committee and UK Home Office Regulations for Animal Use; RIKEN CDB Animal Care and Use Committee). All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Two distinct genetic backgrounds (C57BL/6 and a hybrid of C57BL/6xCBA/NSlc) of Prom1-knockout mice were analyzed. The generation of Prom1-deficient mice with C57BL/6 background (RIKEN strain) has been previously reported,¹⁰ and the information has been deposited in the database of RIKEN Center for Developmental Biology ([in the public domain]http://www.cdb.riken.jp/arg/mutant mice list.html [accession number: CDB0623K]). Due to breeding difficulties on this background, in vitro fertilization (IVF) using Prom1 heterozygote sperm and eggs obtained from wild-type CBA/NSlc mice was employed. The resulting heterozygotic sperm and eggs were then used for the second IVF to obtain Prom1^−/− mice. The mice with both backgrounds were used for histologic analyses (Figs. 1A–E, 3A–E), whereas all other experiments were performed only with the C57BL/6xCBA/NSlc background. Prom1^−/− and littermate controls were housed and maintained under standard laboratory conditions with a 12-hour day (160-lux)–night cycle and ad libitum access to food and water. 

Figure 1

 

Outer cell retinal layers progressively degenerate in the Prom1-deficient mice in a genetic background dependent manner. (A–E) Photoreceptor and ONL areas are completely obliterated within 20 days after birth in Prom1^−/− mice with a C57BL/6 background (RIKEN strain). Hematoxylin and eosin (HE) staining was performed on retinal sections of wild-type (A, B) or Prom1^−/− (C, D) mice at P14 (A, C) and P20 (B, D) after birth. Scale bars: 25 μm (A–D). Relative size of each area in the Prom1^−/− to the wild-type is shown in (E), n = 3 for each genotype. (F–N) Hematoxylin and eosin staining of Prom1^−/−-mice retina with a mixed background of C57BL/6xCBA/NSlc. Wild-type (F–I) and Prom1^−/− (J–M) mice retina with a mixed C57BL/6xCBA/NSlc background retinal sections at P14 (F, J), P20 (G, K), P30 (H, L), and P120 (I, M). Scale bars: 25 μm (F–M). (N) Relative size of each retinal cell layer in the Prom1^−/− mice compared with the wild-type mice, n = 3 for each genotype. In (E) and (N), the data are represented with the mean value ± SEM. Asterisks, statistically significant (P value < 0.05). **Degenerated area.

Effect of Light on Prom1 Retinal Degeneration

For dark rearing (darkness luminescence < 0.5 lux), female mice and their litters (P1 or P8) were transferred into a ventilated housing cabinet (Scantainer I-110; Scanbur, Lund, Denmark), which was further modified in house to reduce light exposure from the outside. Dark-housed animals were kept in the cabinet for approximately 4 weeks before analysis. The extent of photoreceptor degeneration was examined at P30. Electroretinograms (ERGs) were performed to assess retinal function. Mice were subsequently culled and the eyes enucleated under dim red light and processed for paraffin sectioning. Animal husbandry was performed under red light (emission spectrum above 600 nm, red lamp, #02580; British Electrical Lamps Limited, London, UK). 

Histology

Eyes were enucleated and fixed in 4% buffered paraformaldehyde (PFA; Sigma-Aldrich, Gillingham, Dorset, UK) in 1X PBS for 30 minutes at room temperature. To allow penetration of the fixative in adult eyes a circumferential cut was made around the posterior margin of the ciliary body and only the posterior eyecup was retained and fixed for another 30 minutes. Eyecups were processed with a series of ascending alcohols; cleared in trichloromethane; then embedded in paraffin. Retina paraffin blocks were cut sagittally at 5-μm thickness using a microtome sections for hematoxylin and eosin (HE) staining were first deaxed in xylene, rehydrated, rinsed in water, and stained. The sections of the central part of the eye were analyzed, and the images were taken at the top part. 

Electroretinography (ERG)

Electroretinography (ERG) was performed using a ColorDome Ganzfeld ERG (Diagnosys LLC, Cambridge, UK). Briefly, mice were dark-adapted overnight for scotopic measurements and anesthetized under red light illumination by intraperitoneal injection of 50 to 75 mg/kg ketamine (Narketan; National Veterinary Services Ltd., Stoke-on-Trent, UK) and 10% medetomidine (Domitor; National Veterinary Services Ltd., Bracknell, Brackness Forest, UK). Pupils were dilated using 10 mg/kg tropicamide and 0.5% phenylephrine hydrochloride. Animals were kept on a heating mat throughout the procedure to maintain the body temperature at 37°C. Subdermal needle electrodes (CareFusion, Middleton, WI, USA) were inserted as ground and reference electrodes near the hindquarter and between the eyes, respectively. Gold ring electrodes were placed on the corneal surfaces, which were lubricated using carbomer gel (Viscotears; Novartis Pharmaceuticals, Camberley, Surrey, UK). Following a further 10 minutes of dark adaptation the scotopic ERG was simultaneously recorded from both eyes with increasing stimulus strengths using mixed white light from 10⁻⁵ cd.s/m² to 10 cd.s/m² with every 10-fold interval. Four animals per genotype were analyzed in each set of experiments. 

Immunohistochemistry

Following antigen retrieval (boiled in Tris-EDTA pH 9.0) sections were stained with either mouse monoclonal anti-Rhodopsin (Sigma-Aldrich) diluted 1:500 or rabbit polyclonal anti M-opsin (Millipore) diluted 1:500. Secondary antibodies included goat anti-mouse Alexa Fluor 488 or goat anti-rabbit Alexa Fluor 488. Images were collected with LSM 710 confocal microscope (Zeiss, Oberkochen, Baden-Württemberg, Germany). 

Treatment With Fenretinide and A1120

Administration of fenretinide and A1120 was performed as described previously,^33,34 with slight modifications. Briefly, Prom1^−/− mice (P0) were injected intraperitoneally three times a week with 10 mg/kg of fenretinide (N-[(4-hydroxyphenyl)amino]retinal; also known as 4-HPR; CAS number 65646-68-6) or the equivolume of dimethylsulfoxide (DMSO) for 4 weeks. A1120 (2-(4-(2-(trifluoromethyl)phenyl)piperidine-1-carboxamido)benzoic acid; CAS number 1152782-19-8),^35,36 an RBP4 (NCBI gene ID 5950; Retinol Binding Protein 4) inhibitor, was tried at 30 mg/kg (see Discussion). Treated mice were subject to the analyses 2 days after the last injection. 

Quantitative RT-PCR (qRT-PCR)

Total RNA was isolated from whole eye samples using RNeasy mini kit (Qiagen, Crawley, UK) following manufacturer's instructions. Reverse-transcription was performed using Superscript II Reverse Transcription Kit (Life Technologies, Carlsbad, CA, USA). Real-time quantitative RT-PCR was performed with a thermal cycler (7900HT; Applied Biosystems, Waltham, MA, USA). Reagents were obtained from Applied Biosystems and all primers were designed to span an intron to avoid amplifying genomic DNA. 18S was used as the internal control. The primer sequences are available in Supplementary Table S1. Polymerase chain reactions were conducted using at least three separate RNA preparations. 

Electron Microscopy

Eyes were enucleated and fixed in 3% glutaraldehyde and 1% paraformaldehyde in 0.08 M sodium cacodylate-HCl (pH 7.4) for at least 30 hours at 4°C. The cornea and lens were removed and the eye cups oriented and post fixed in 1% aqueous osmium tetroxide for 2 hours, dehydrated by an ascending ethanol series (50%–100%) and propylene oxide, and infiltrated overnight with a 1:1 mixture of propylene oxide and araldite resin (Agar Scientific Limited, Essex, UK). After 8 hours in full resin, the eyes were embedded in fresh resin and incubated overnight at 60°C. Semithin (0.7 μm) sections were cut in the inferior–superior axis passing through the optic nerve head with a microtome (Ultracut S; Leica, Wetzlar, Germany) Ultrathin sections were collected on copper grids (100 mesh; Agar Scientific, Stansted, Essex, UK), contrast-stained with 1% uranyl acetate and lead citrate and analyzed using a JEOL 1010 transmission electron microscope at 80 kV. Retinal sections from three animals per genotype were analyzed. 

Statistics

Electroretinogram (ERG) results were compared using two-way ANOVA on GraphPad Prism V4 (GraphPad software, La Jolla, CA, USA) with P values less than 0.05 considered statistically significant. 

Results

Prom1-Knockout Mice Develop Photoreceptor Cell Degeneration in a Genetic Background Dependent Manner

In order to examine the mechanism of Prom1-mediated photoreceptor degeneration, we originally made a Prom1-deficient mouse with pure C57BL/6 background,¹⁰ in which the wild-type does not develop any degeneration or have retinal abnormalities (Figs. 1A, 1B). At P14, the layered structure, morphology of retinal neurons, and glial cells, in particular morphology of the photoreceptor outer segment, and the numbers of retinal cells, showed that the retina was normal (Figs. 1C, 1E). In contrast, at P20 the photoreceptor cell layer including the outer segment and cell body was completely missing (Figs. 1D, 1E). Thus Prom1-knockout mouse developed photoreceptor cell degeneration (Figs. 1C–E) as observed in the previously reported Prom1-knockout mice.³⁰ 

Due to the fact that C57BL/6-background Prom1^−/− mice have difficulty in breeding, we generated the same Prom1-targeted mice with a hybrid background of C57BL/6xCBA/NSlc (see Methods). This new knockout also started to exhibit extensive photoreceptor OS degeneration after P14 (Figs. 1F, 1J, 1N), which coincided with eye opening. However, at P20, in contrast with the Prom1^−/− mice on the C57BL/6 background (Figs. 1D, 1E), the outer plexiform layer (OPL), the outer nuclear layer (ONL), and OS/IS were still preserved with thickness of 74%, 76%, and 61%, respectively, compared with the wild-type mice (Figs. 1G, 1K, 1N). This unexpected result prompted us to analyze the progression of the photoreceptor degeneration chronologically. The Prom1^−/− mice with C57BL/6xCBA/NSlc at P30 showed a substantial degree of retinal degeneration with approximately 22% of photoreceptor nuclei remaining (Figs. 1H, 1L, 1N). By 4 months (P120) the entire ONL and OS/IS had been completely abolished (Figs. 1I, 1M, 1N). On the other hand, the inner nuclear layer (INL) and inner plexiform layer (IPL) were largely unaffected (Figs. 1I, 1M, 1N). 

These histologic observations suggest that PROM1 is essential for the maintenance, rather than the development of retinal structure, and the phenotype with Prom1 deficiency is dependent on the genetic background. 

Expression of Rhodopsin and M-Opsin in the Photoreceptor Cells is Downregulated and Misplaced in Prom1-Deficient Mice

We further analyzed the ultrastructure of the Prom1^−/− in C57BL/6xCBA/NSlc retinal cells by using transmission electron microscopy (TEM). At P20, the photoreceptor OSs of the retina were severely shortened in the Prom1^−/− mice compared with that of wild-type (Figs. 2A, 2B), which is consistent with the histologic findings (Fig. 1K). 

Figure 2

 

The OS of the photoreceptor cells are disordered and Opsin proteins are mislocalised in Prom1-deficient mice. (A–H) Outer segments are disorganized, while other parts of the photoreceptor cells are largely intact at P20. Wild-type (A, C–E) and Prom1^−/−-mice retinae (B, F–H) were analyzed by transmission electron microscope. (A, B) Panorama images of wild-type (A) and Prom1^−/− (B) photoreceptor cells. Note that these are merged images from several individual shots. Original toluidine blue–stained images are available in Supplementary Figure S1. (C–H) Detailed structure of OSs in the wild-type (C–E) and in the Prom1^−/− (F–H) retina was analyzed at the levels labeled in (A) and (B). Tight-like junctions at the border of ONL and IS (red arrowheads in [C, F]), mitochondrial structure in IS (red arrowheads in [D, G]) and the OS structure (E, H) are shown. Scale bars: 2 μm (A, B), 500 nm (C–H). (I–L') Rhodopsin and M-opsin proteins are mislocalized in the Prom1^−/− retina. Wild-type (I, I', K, K') and Prom1^−/− retina (J, J', L, L') of P20 were analyzed by immunohistochemistry using rhodopsin (green in [I, I', J, J']) and M-opsin (green in [K, K', L, L']) antibodies. Ectopic localization is shown with brackets (I–J') and with arrowheads in (L, L'). DAPI (4′,6-diamidino-2-phenylindole) is shown in blue (I', J', K', L'). Scale bars: 100 μm (I–L'). PR, photoreceptor layer.

In the C57BL/6xCBA/NSlc mice, P20 is the beginning stage of the photoreceptor degeneration, and this enabled us to observe the intermediate stages of the degeneration. The tight junction structure at the outer limiting membrane seemed to be largely intact in the Prom1^−/− retina (Figs. 2C, 2F; red arrowheads),³⁷ and the mitochondria were formed normally in the IS (Figs. 2D, 2G; red arrowheads). However, the structure of the OS was severely disorganized and misaligned (Figs. 2E, 2H). This was consistent with the previous observation in the mice with the different genetic background.³⁰ 

We next examined the expression of photoreceptor specific proteins in the Prom1 knockout by immunohistochemistry. At P20, WT retina showed a strong signal of the G-protein coupled receptor (GPCR) rhodopsin correctly localized to the OS in the rod photoreceptor (Figs. 2I, 2I'). In contrast, in the Prom1^−/− retina, the signal was found in the ISs and in the ONL (Figs. 2J, 2J').³⁰ In addition, another GPCR M-opsin, normally localized to the OS of cone photoreceptors (Figs. 2K, 2K'),³⁸ was largely missing in Prom1^−/−, and was partially found in the ONL, with the occasional cells showing positive signal (Figs. 2L, 2L'). 

Taken together these observations demonstrate Prom1 is required for the maintenance of both rod- and cone-photoreceptor cells, and for the correct localization of opsin proteins that act in the OSs. 

Light-Dependent Impairment of Photoreceptor Cells in the Prom1-Deficient Mice

The photoreceptor degeneration in both mice strains of the pure C57BL/6 background and on the C57BL/6xCBA/NSlc background was initiated after P14, when they open their eyes. Thus, we hypothesized that Prom1-mediated photoreceptor degeneration was triggered by light exposure. 

In order to examine the effect of light on the photoreceptor degeneration in Prom1-knockout mice, the C57BL/6 mice were reared in complete darkness and compared with mice kept under normal light-dark cycle (see Methods for details). At P30, the Prom1^−/− kept under normal light cycle indeed had no photoreceptor cells (OS, IS, ONL in Figs. 3A, 3B). In contrast, the animals reared under the dark conditions still had some photoreceptor cells remaining, and their OS of comparable length with the wild-type control (Figs. 3C, 3D). Similar observations were made in the Prom1^−/− mice with the C57BL/6xCBA/NSlc background (Figs. 3E–H); the Prom1^−/− mice reared under the normal light-dark cycle developed a severe photoreceptor degeneration (Figs. 3E, 3F), while the same Prom1-depleted animals under the dark condition retained a significant proportion of the photoreceptor layer (Figs. 3G, 3H). These observations suggest that the photoreceptor degeneration in Prom1-knockout mice developed in a light-dependent manner both in C57BL/6 and C57BL/6xCBA/NSlc backgrounds. 

Figure 3

 

Retinal degeneration in Prom1-knockout mice is light-dependent. Both C57BL/6 (A–D) and C57BL/6xCBA/NSlc (E–H) retinae were significantly protected from the photoreceptor degeneration when the animals were reared in dark conditions. Wild-type littermate controls (A, C, E, G) and Prom1^−/− (B, D, F, H) mice (C57BL/6xCBA/NSlc) were reared under normal light (A, B, E, F) or complete darkness (C, D, G, H) conditions from P8 until P30. Scale bars: 50 μm (A–H). (I–L) Ultrastructural analysis by transmission electron microscopy. The C57BL/6xCBA/NSlc animals were reared as in (H) and fixed at P30. Scale bar: 2 μm. Detailed structure at the levels denoted in (I) is shown in (J–L). Mitochondria (red arrowhead in [K]) and tight-like junctions (red arrowheads in [L]) are shown. Scale bar: 500 nm. (M–N') The misocalization of rhodopsin (green in [M, M']) and M-opsin (green in [N, N']) was reduced in the dark-reared mice. The Prom1^−/− animals were reared in the dark condition and analyzed by immunohistochemistry. DAPI is shown in blue in (M', N'). Scale bars: 100 μm (M–N'). (O) Quantification of the cells in which rhodopsin and M-opsin are mislocalized to ONL. Several squares of 100 × 100 μm were randomly chosen within the IS and ONL and the numbers of the positive cells for each protein were counted. The data of the wild-type and Prom1^−/− mice reared in the normal light-dark conditions were collected from Figures 2I through 2L'. The data are represented with the mean value ± SEM. Three retinal tissues for each were analyzed. Asterisks, statistically significant (P < 0.05).

We further analyzed the ultrastructure of the photoreceptor cells by transmission electron microscopy (TEM). The OS of the Prom1^−/− retina reared under dark conditions partially retained the structure of the cells (Figs. 3I–L, compared with Figs. 2B, 2F–H). In addition, as examined by immunohistochemistry in the Prom1^−/− mice, less mislocalized signal of rhodopsin and M-opsin staining was found in the IS compared with those under the light/dark condition (Figs. 3M–O, compared with Figs. 2J, 2J', 2L, 2L'). Together these observations indicate that the Prom1^−/− photoreceptor cell structure can be retained when the retina is protected from light. 

We next investigated the function of photoreceptor cells in the C57BL/6xCBA/NSlc Prom1^−/− mice. To this end, we examined the a- and b-waves of the ERG. An ERG is conducted to measure the electrical responses of the retinal cells upon external stimuli, and may be used to evaluate progression of retinal diseases.³⁹ An ERG consists of two peaks; the initial peak (a-wave; Fig. 4A) reflects the physiological health of the rod or cone photoreceptors and the second (b-wave; Fig. 4A), the inner layers of retina, including bipolar, amacrine and Müller cells.⁴⁰ We performed both scotopic (Figs. 4A–F) and photopic (Figs. 4G, 4H) ERGs to investigate the function of rod and cone photoreceptors, respectively. 

Figure 4

 

Prom1^−/− mice show maintained their photoreceptor function when reared in dark. (A–D) Representative recordings of scotopic ERG responses (a-waves, light blue in [A]; and b-waves, orange in [A]). C57BL/6xCBA/NSlc wild-type littermate (A, B) or Prom1^−/− mice (C, D) reared under the normal light/dark condition (A, C) or in the complete darkness from P8 until P30 (B, D) were subject to ERG using seven light amplitudes within the range of 10⁻⁵ and 10¹ (cd.s.m⁻²) under scotopic conditions. (E, F) Recordings of a-waves (E) and b-waves (F) from the wild-type under the normal light/dark (black lines and circle dots; n = 5) or the dark (green lines and circle dots; n = 6), Prom1^−/− under the normal light/dark (red lines and rectangular dots; n = 8) or the dark (blue lines and triangular dots; n = 3) conditions were summarized. Note that the horizontal axes are displayed in logarithmic scale (G, H).

First, we performed scotopic ERGs on the P30 mice (C57BL/6xCBA/NSlc) reared under the normal light/dark condition. The Prom1^−/− mice showed marked impairment of a- and b-waves (Fig. 4C, and red lines in Fig. 4E, 4F). 

We next performed scotopic ERGs on the Prom1^−/− mice raised under complete darkness during the period between P8 and P30. Strikingly, the a-wave and b-wave of the scotopic ERG in the dark-reared Prom1^−/− mice were significantly improved compared with those reared in the normal light/dark condition, and the b-wave reached almost the same level as that of the wild-type retina (Figs. 4E, 4F; blue lines). The amplitudes of a- and b-waves in the dark-reared wild-type mice appeared to be the same as those in the wild-type mice reared in the normal light/dark cycle, suggesting that the light stimulation per se did not influence the condition of the wild-type retina (Figs. 4A, 4B, 4E, 4F; black and green lines). Taken together, this set of analysis demonstrated that Prom1^−/− photoreceptor cells are functionally impaired upon light exposure. 

We additionally performed photopic ERGs. In agreement with the scotopic ERGs, the amplitudes of both a-waves and b-waves in the photopic ERGs were reduced in the Prom1^−/− mice reared under the normal light/dark cycle, but maintained if the knockout mice were protected from the light exposure (Figs. 4G, 4H). 

Altogether these structural and functional examinations indicate that photoreceptor cells in the Prom1^−/− mice are very vulnerable to light and damaged in a light-dependent manner. 

Expression of Genes Involved in the Visual Cycle is Affected in the Prom1 Knockout Mice

Light-dependent retinal degeneration is caused by a number of factors (see Refs. 41 and 42 for review). Since some of these (e.g., Abca4, Rdh8, Rpe65, rhodopsin) are highly associated with the visual cycle, we investigated the effect of the Prom1 gene knockout on the visual cycle. 

The visual cycle is a process, which maintains the levels of the chromophore in photoreceptor OSs.¹⁶ Phototransduction starts with the conversion of 11-cis-retinal into all-trans-retinal on exposure of the photoreceptors to light. The all-trans-retinal is then recycled initially in the photoreceptors and later in the retinal pigmented epithelium (RPE) cells to reform 11-cis-retinal. All-trans-retinal is first reduced to all-trans-retinol (Vitamin A), a process which requires ABCA4,^19,43,44 RDH12,^44–47 and RDH8.⁴⁴ All-trans-retinol is then transported with IRBP to the RPE cells, where it is converted, with the aid of lecithin retinol acyltransferase (LRAT) to all-trans-retinyl-ester and further to all-trans-retinol with the help of the isomerase RPE65. This all-trans-retinol is oxidized into 11-cis-retinal by RDH5/11 and is transported back to the rod cells with interphotoreceptor retinoid-binding protein (IRBP; also known as retinol-binding protein 3 [RBP3]), where it forms a complex with rhodopsin (see Refs. 16, 48, 49 for review). 

In order to investigate how Prom1 is involved in this visual cycle, we examined the expression of visual cycle genes in PROM1-deficient mice. We isolated the eyes from the P14 wild-type or Prom1^−/− mice and performed qRT-PCR analysis. Expression of Abca4 and Rdh5/12/14 was significantly reduced in the Prom1^−/− retina. In contrast, the expression of rhodopsin, IRBP, LRAT, S-opsin, Rdh11, and Rdh13 were not significantly reduced compared with that in the wild-type tissue, while M-opsin was rather upregulated (Fig. 5). This data suggests that PROM1 is required for the maintenance of the expression levels of Abca4 and Rdh12, and is consistent with the idea that PROM1 is involved in the regulation of the visual cycle, especially at the reducing step of all-trans-retinal to all-trans-retinol. 

Figure 5

 

Prom1 is essential for the expression of some of the visual cycle–related genes. The relative expression levels of the indicated genes were analyzed by qRT-PCR analysis. Quantitative RT-PCR was performed for C57BL/6xCBA/NSlc wild-type and P14 Prom1-deficient retina. The expression level of each gene in the wild-type retina was defined as 1. Three retinal tissues were analyzed. Primer sequences are indicated in Supplementary Table S1. Asterisks, statistically significant (P < 0.05).

Fenretinide Can Partially Prevent the Photoreceptor Degeneration

Failure of the reduction of all-trans-retinal to all-trans retinol results in the formation of all-trans retinal dimer and A2E (a lipofuscin fluorophore; also known as N-retinylidene-N-retinylethanolamine),^48,50–53 both of which induce inflammation and subsequent photoreceptor cell dysfunction. The accumulation of these compounds has been observed in Abca4^−/−, Rdh8^−/−, and Rdh12^−/− mice.^{33,44,53–55} 

The synthetic retinoid derivative fenretinide reduces the availability of retinoids to the eye by its interaction with retinol binding proteins³³ (RBP; RBP1-4; NCBI gene IDs 5947-5950),^56,57 and thereby reduces accumulation of A2E in RPE. As fenretinide has been investigated in animal models of Abca4-mediated Stargardt diseases,^33,34,58 we investigated whether fenretinide may also reduce photoreceptor cell death in Prom1^−/− mice. 

First, we examined the effect of fenretinide on the Prom1^−/− mice. We injected 10 mg/kg of fenretinide three times a week from P0 for four weeks. At P30, compared with the vehicle-injected control, fenretinide injected Prom1^−/− mice had thicker photoreceptor segments and ONL, and increased number of photoreceptor nuclei (Figs. 6A–E). A similar result was obtained with a higher dosage (20 mg/kg) of Fenretinide (M.D., S.B., S.O., unpublished data, 2014). This finding suggests that Fenretinide protected, at least in part, the photoreceptor cells from Prom1-deficient mediated degeneration. 

Figure 6

 

The visual cycle inhibitor Fenretinide delays PROM1 mediated retinal degeneration, and the retinal functions are partially preserved. C57BL/6xCBA/NSlc wild-type (A, B) or Prom1^−/− (C, D) mice treated with DMSO (vehicle) (A, C) or Fenretinide (B, D) were analyzed at P30. Scale bars: 50 μm (A–D). (E) Quantification of the photoreceptor nuclei in the ONL. (F–K) Fenretinide treatment shows an improvement of the ERG response. Responses of a-waves (F–H) and b-waves (I–K) were recorded under scotopic conditions in the wild-type retina treated with control DMSO (black lines and circle dots in [F, I]; n = 5) or Fenretinide (FEN; gray lines and rectangle dots in [F, I]; n = 3) and Prom1^−/− retina treated with control DMSO (black lines and circle dots in [G, J]; n = 4) or FEN (red lines and rectangle dots in [G, J]; n = 3). (H, K) The relative ERG scores upon the FEN treatment compared with those treated with the control DMSO were replotted from (F, G) and from (I, J) at 1 and 10 cd.s/m² (denoted as 0 and 1 in the horizontal axis; [I]) or at all data points (K). Note that the horizontal axes in (F–K) and the vertical axes in (I, K) are displayed in logarithmic scale. Asterisks, statistically significant (P < 0.05). The data are represented with the mean value ± SEM.

Next, we evaluated the effect of Fenretinide on the function of Prom1^−/− retina by examining the scotopic ERG responses. One caveat in this examination is that Fenretinide induces night blindness as a side effect, because the function of rhodopsin is partially blocked and the visual cycle is slowed down.³⁴ Consistently, the a-wave amplitude was decreased in the Fenretinide-treated retina (Fig. 6F; gray line) compared with the nontreated retina (Fig. 6F; black line), indicating that rod function is affected by Fenretinide. In contrast, in the Prom1^−/− retina, which were almost insensitive to the stimulus unless treated (Fig. 6G, black line), the rod responses were improved when treated with Fenretinide (Fig. 6G, red line). This effect was more evident when the results (Figs. 6F, 6G) were replotted against the control (DMSO-treated) retina as the reference (Fig. 6H). This observation suggests that Fenretinide, despite its inhibitory effect on normal rod cell function, improves rod function in Prom1^−/− mice, possibly by reducing cell death. Likewise, the b-waves were decreased (Figs. 6I, 6K) when Fenretinide was treated in wild-type retina, but were improved in the Prom1^−/− retina upon treatment with Fenretinide (Figs. 6J, 6K). 

Together the analyses indicate that the treatment with Fenretinide maintains the structure of photoreceptor cells, which is accompanied by a partial recovery of function. 

Discussion

Genetic Background Influences the Progression of Photoreceptor Degeneration

This study demonstrates that variations in the genetic background significantly influence development of photoreceptor cell degeneration mediated by knockout of the Prom1 gene. We examined the mutants with two genetic backgrounds; a pure C57BL/6 and a hybrid C57BL/6xCBA/NSlc. The Prom1^−/− mice of the pure C57BL/6 background exhibited complete photoreceptor degeneration within 1 week after eye opening (P20), while the retinal degeneration in the C57BL/6xCBA/NSlc mice developed much more slowly (Fig. 1). The same mouse strain showed the same degeneration profile, suggesting that the variations in photoreceptor degeneration seem to be dependent on the genetic background, rather than the individual mice. Variations in disease severity are reported in patients with retinal disease caused by both autosomal-recessive (AR)^26,28,29 and autosomal-dominant (AD) Prom1 mutations.³² Mouse knockout models are good models for human disease caused by AR Prom1 mutations, but are not as useful as for studying the mechanism of AD disease. However, it is possible that environmental or genetic modifiers may be common to both types of disorder. 

A number of factors, both environmental and genetic, may account for this variation. In contrast to the studies in man, the environmental factors can be carefully controlled in experimental studies in mouse models; the mice used in our experiments were reared under carefully standardized conditions. Therefore, most of the variation in the progression of the photoreceptor degeneration is likely to be due to the genetic factors. The mechanisms underlying these strain differences are unknown and their identification will be challenging. The two types of the mice used in our experiments will be a good model system to identify such genetic factors for making this variation. 

It should also be noted that there is diversity even in the same genetic background.³⁰ The previously reported Prom1^−/− mice with the same C57BL/6 background started the degeneration around P14, but even after 6 months the mice still had some remaining photoreceptor cells,^10,30 which is much slower than our knockout mice with the same C57BL/6 background employed in our study (Figs. 1A–E). This may be due to minor genomic variations raised by the difference of the mouse source.³⁰ More detailed study of these various mouse models will help identify these presumed genetic modifiers. 

The divergence of the severity in the mutant has been defined as “expressivity.”^59–61 It has been suggested that expressivity is caused by modifier genes, which quantitatively influence the effect of the loss of the targeted gene. In the case of Prom1, Abca4, and Rdh12 may play such roles. This view is supported by the fact that the severity of light-induced photoreceptor damage is dependent on the expression level⁶² or genetic polymorphisms^63,64 of the Rpe65 gene. Further work is needed to identify the modifier gene(s) for Prom1, including the investigations of genomic variation, for example, single-nucleotide polymorphism (SNP) and epigenetic status of the genomes in the different individuals or genetic backgrounds. 

Light-Dependent Photoreceptor Degeneration in Prom1-Mutant Mice and Possible Therapies for Prom1-Causing Diseases

It has been shown that STGD1 patients are sensitive to light, and undergo light-protective treatments.⁶⁵ This is consistent with the light-dependent photoreceptor degeneration found in the Prom1^−/− mice, raising the possibility that Prom1 is involved in the visual cycle function. Our qRT-PCR analysis revealed the reduction of Abca4 and Rdh12 expression at the onset of the photoreceptor degeneration, both of which are involved in the visual cycle (Fig. 5). These two genes may therefore be downstream targets of Prom1. An alternative possibility is that Prom1-mutant cells start to degenerate first, and Abca4 and Rdh12 expression is downregulated as a secondary effect of photoreceptor cell death. We would favor the former possibility, as the reduction of Abca4 and Rdh12 was highly selective, Abca4 and Rdh12 are downregulated, whereas expressions of other genes were unchanged (Fig. 5), suggesting that the gene regulation by PROM1 is independent of the photoreceptor degeneration. 

How does PROM1 control the expression of these genes? As PROM1 is a transmembrane protein, PROM1 may transduce the extracellular information or stimulation intracellularly. Future transcriptome studies will provide systematic information about downstream target genes and possible upstream regulatory pathways. 

Accumulation of all-trans-retinal dimer and A2E has been reported in the STGD1 model mice with mutations of Abca4.^51,66,67 A2E is the major fluorophore of lipofuscin and high levels of lipofuscin are seen on fundus autofluorescence imaging in patients with Stargardt disease. As Prom1 deficiency in our mouse models results in downregulation of Abca4 and Rdh12 gene expression, it is likely that ensuing effects on the visual cycle will lead to accumulation of lipofuscin such as all-trans-retinal dimer and A2E in the Prom1^−/− mutants, although the direct measurement is yet to be performed. 

Fenretinide, originally recognized as an anticancer drug and later applied to retinal disease,^68,69 has been investigated in mouse models of STGD1 disease.³⁶ The potential human use of Fenretinide will depend on the balance between the benefit (desired effect) to inhibit the accumulation of the toxic lipofuscin,^56,57 and the systemic and ocular side effects. Fenretinide use in animal models (Figs. 6F, 6H, 6I, 6K, Supplementary Figs. S2A, S2C),⁷⁰ and in man⁷¹ leads to rod dysfunction. On the other hand, the Prom1-mutant mice treated with Fenretinide showed improved retinal structure (Figs. 6A–E) and improved scotopic ERG (Figs. 6G, 6H, 6J, 6K) indicating that the positive effect on photoreceptor degeneration outweighed the inhibitory effect on rod function. 

However, the recovery of the photoreceptor function from our Fenretinide treatment experiments still needs to be improved. In particular, the photopic ERG responses were not significantly improved by Fenretinide (Supplementary Fig. S2). One of the possibilities for this limitation is the method of administration. Recent studies have shown Fenretinide has very poor oral bioavailability due to low solubility and low permeability.^72–74 Therefore, for ocular diseases, a controlled-release formulation, which provides drug release over a longer period of time, may be advantageous.⁷⁵ We have tried another RBP binding compound A1120, which has been used in STGD1 mouse models.^35,36 The drug was however, less effective than Fenretinide (Dellett M, Ohnuma S-I, unpublished data, 2013). As some primary amines and GPCR/adenylate cyclase inhibitors have recently been introduced to improve the photoreceptor survival in Abca4^−/−; Rdh8^−/− mice,^48,76 it would be interesting to explore their effects in Prom1^−/− mice. 

In addition to its effect on the visual cycle, it has also been shown that PROM1 interacts with Protocadherin-21 (NCBI Gene ID; 92211; PCDH21, CDHR1),^24,77 which has also been implicated in the photoreceptor degeneration.⁷⁸ The systematic identification of PROM1-interacting proteins will help to better understand the direct function of PROM1. 

In conclusion, we have provided evidence of light-dependent photoreceptor degeneration in the Prom1^−/− mice, and have identified possible downstream signaling targets. Further investigation of the downstream target genes and interacting proteins will allow a better understanding of the mechanisms of photoreceptor cell death and more targeted therapeutic approaches in the future. 

Acknowledgments

The authors thank Peter Munro for his assistance with the transmission electron microscope, and Lisa Haigh and the members of the Ohnuma's laboratory for comments and discussions. In vitro fertilization for Prom1-mutant mice was performed with the assistance of TAKEDA Pharmaceuticals. 

Supported by grants from RP Fighting Blindness (S-IO; Buckingham, Bucks, UK), a RIKEN internal fund (TK; Kobe, Japan), Foundation Fighting Blindness (ATM; Owings Mills, MD, USA), National Institute for Health Research UK (ATM, GAL; Moorfields Biomedical Research Centre, London, UK), and Medical Research Council (G0900002 [SB, GAL], Swindon, UK). 

Disclosure: M. Dellett, None; N. Sasai, None; K. Nishide, None; S. Becker, None; V. Papadaki, None; G.A. Limb, None; A.T. Moore, None; T. Kondo, None; S.-I. Ohnuma, None