A Mouse Model for Studying Cone Photoreceptor Pathologies

Abstract

Purpose.: Due to the low abundance of cone photoreceptors in the mouse retina and the scarcity of alternative animal models, little is known about mechanisms of cone degeneration. Nrl knockout mice develop exclusively the cone-type of photoreceptors. However, the cone photoreceptor layer in Nrl^−/− mice displays an irregular morphology with severe rosette formation. Retinas of Rpe65^−/−;Nrl^−/− mice have no rosettes due to the lack of 11-cis-retinal, but also are not functional. To develop a model with a functional all-cone retina that is morphologically well structured, we generated R91W;Nrl^−/− double-mutant mice, which express a hypomorphic Rpe65 allele (R91W).

Methods.: The following analyses were used to characterize the R91W;Nrl^−/− mice: morphology by light and electron microscopy, protein distribution by immunofluorescence, cone function by electroretinography and optomotor response, RNA levels by RT-PCR, and chromophore levels by HPLC. Cone degeneration was assessed in R91W;Nrl^−/− mice treated with MNU, and in triple R91W;Nrl^−/−;Cpfl1 and quadruple R91W;Nrl^−/−;Cpfl1;rd10 mutant mice.

Results.: The all-cone retina of R91W;Nrl^−/− mice is functional and relatively stable with only very slow age-related degeneration. Using triple and quadruple mutant mice, or a chemical treatment, we demonstrated that cone degeneration could be induced and analyzed in these mice.

Conclusions.: The reduced levels of visual chromophore prevented rosette formation and sustained function in the R91W;Nrl^−/− retina. Thus, the R91W;Nrl^−/− mouse allows study of the etiology of diseases related to cone degeneration in a “morphologically intact” and functional all-cone photoreceptor retina.

Introduction

Loss of high-acuity vision due to degeneration of macular cones is most devastating for everyday life. Despite extensive research, the mechanisms of cone degeneration are only poorly understood and no successful therapy is available for human patients. Only a few models exist and most of these are difficult to investigate due to ethical, methodological, and/or cost-related reasons. ¹ Finding new models for this type of research or for general investigations of cone physiology is even more difficult due to the absence of a macula in nonprimates and to the low numbers of cones in retinas of most common laboratory animals, such as rats and mice. In the mouse retina, rods outnumber cones by a factor of roughly 30. ² On a molecular level, the rod-cone ratio, is determined by a complex network of six transcription factors, including neural retina leucine zipper (NRL) and cone-rod homeobox (CRX). Expression and coordinated activity of these factors control the fate of progenitor cells during retinal development. ³ In the absence of NRL transcription factor, rods are not generated and photoreceptor progenitors are directed toward a cone cell fate resulting in an all-cone retina in Nrl^−/− mice. ⁴ However, Nrl^−/− retinas are dysmorphic, with large and numerous rosette-like structures in the photoreceptor layer. Rosette formation in Nrl^−/− mice depends on the visual chromophore 11-cis-retinal, which is produced by the enzymatic activity of RPE65. ^5,6 The absence of RPE65 and thus the lack of 11-cis-retinal prevents rosette formation in Rpe65^−/−;Nrl^−/− double-mutant mice.^5–7 However, as 11-cis-retinal is required for photon absorption, Rpe65^−/−;Nrl^−/− mice have a nonfunctional retina. 

Recently, we generated an Rpe65^R91W knockin mouse (R91W) with reduced ocular levels of 11-cis-retinal. ^8,9 In humans, this mutation leads to early-onset severe retinal dystrophy ¹⁰ and in mice, to slowly progressing cone degeneration. ^8,9 Levels of the mutant RPE65^R91W protein are reduced by approximately 90% but, because the protein retains enzymatic activity, low amounts of 11-cis-retinal are generated and support photoreceptor function in the retina of R91W mice. ⁹  

We hypothesized that the reduced levels of 11-cis-retinal in R91W mice may be sufficiently low to prevent rosette formation in the absence of NRL, but still be sufficiently high to support retinal function. Thus, we combined R91W and Nrl^−/− mice in the attempt to generate a mouse model with a functional and morphologically intact all-cone retina. We show that R91W;Nrl^−/− mice indeed develop a functional “all-cone” retina with an undisturbed retinal layering. Cone cell death in this double-mutant mouse can be exogenously induced and easily studied. Furthermore, we generated and analyzed triple or quadruple mutant mice to provide models for the analysis of inherited cone degeneration in the absence of rods. This allows the investigation of the consequences of mutations found in patients and provides a valuable tool to test therapeutic approaches for the protection or rescue of cone photoreceptor cells and their function, with the ultimate goal to preserve fine-acuity vision in patients. 

Methods

Mice

All animal experimentation adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the regulations of the Veterinary Authorities of Kanton Zurich, Switzerland. 

The 129S6 (wt), C57BL6 (wt), Nrl^−/− , ⁴ Rpe65^R91W (R91W), ⁹ B6.CXB1-Pde6c^cpfl1/J (Cpfl1), ¹¹ rd10, ¹¹ Gnat1^−/− , ¹² and Rpe65^{−/− 13} mice were housed in the animal facility of the University of Zurich in a 12-hour:12-hour light–dark cycle with access to food and water ad libitum. The R91W;Nrl^−/− , R91W;Nrl^−/−;Cpfl1, and R91W;Nrl^−/−;Cpfl1;rd10 mutant mice were generated using classical breeding schemes. The Nrl^−/− mouse line housed at the University of Zurich was originally homozygous for the rd8 mutation in Crb1. Rd8 was crossed out by breeding to C57BL6 wt mice. Resulting pure Nrl^−/− mice developed rosettes in their retinas similar to Nrl^−/−;Crb1^rd8/rd8 mice (see Supplementary Fig. S1). Genotypes were tested by PCR using genomic DNA from ear biopsies and primer pairs listed in Supplementary Table S1. All the other lines used in this study were free of the rd8 mutation. Each of the following methods has been carried out using three different mice per strain and experimental condition, with the exception of HPLC, optomotor response (OMR), and ERG, where the number of mice is indicated. 

HPLC Determination of Retinoids

Six-week-old mice were dark-adapted for 24 hours (n = 4). All further steps were carried out under dim red light. Mice were killed, and lens and vitreous were removed from the eye through a slit in the cornea. The rest of the ocular tissue, including the retina and eyecup, was snap frozen in liquid nitrogen until further analysis. Retinoid extraction and HPLC analysis was performed as previously described. ¹⁴  

Morphology and Quantification of Retinal Degeneration

Briefly, eyes were marked nasally, enucleated, and fixed in 2.5% glutaraldehyde, nasal and temporal eye halves were separated by a cut through the optic nerve head, and separately embedded in epon plastic as described recently. ¹⁵ For light microscopy, semithin cross-sections (0.5 μm) were counterstained with toluidine blue and analyzed by light microscopy (Zeiss, Axioplan, Jena, Germany). The thickness of the outer nuclear layer was measured on retinal micrographs taken at ×10 magnification using the Adobe Photoshop CS3 ruler tool (Adobe Systems, Inc., San Jose, CA, USA) at indicated distances (μm) from the optic nerve head. 

For electron microscopy, ultrathin sections (50 nm) were stained with uranyl acetate and lead citrate and analyzed using a Philips CM100 transmission electron microscope (Philips Electron Optics, Eindhoven, The Netherlands). 

Immunofluorescence

Mice were perfused with 4% paraformaldehyde (PFA) prepared in PBS (pH 7.4). Eyes were marked nasally by cauterization, enucleated, and processed for cryosectioning, as described earlier in detail. ¹⁵ Dorsoventral cryosections (12 μm) were blocked for 1 hour with 3% normal goat serum or 3% horse serum (containing 0.3% Triton X-100 in PBS), and incubated overnight at 4°C with the following primary antibodies: OPN1SW (1:500, sc-14363; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), OPN1MW (1:500, AB5405; Chemicon, Temecula, CA, USA), CALB (1:500, AB1778; Chemicon), PKC-A (1:1000, P4334; Sigma-Aldrich Corp., St. Louis, MO, USA), SYP (1:100, NCL-L-SYNAP-299; Novacastra, New Castle, UK), VSX2 (CHX10, 1:500; kindly provided by C. Cepko, Harvard University, Harvard Medical School, Boston, MA, USA), POU4F1 (BRN3A, 1:100, MAB1585; Chemicon), CR (1:1000, AB5054; Chemicon), and glial fibrillary acidic protein (GFAP) (1:500, G3893; Sigma-Aldrich Corp.). After washing, slides were incubated with appropriate secondary antibodies labeled with Cy2 or Cy3 (Jackson ImmunoResearch Laboratories, Soham, UK), counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and analyzed by fluorescence microscopy (Zeiss, Axioplan). 

Analysis of Retinal Vasculature in Whole-Mounted Retinas

Briefly, eyes from 6-week-old mice were isolated and incubated for 3 to 5 minutes in 2% PFA prepared in PBS as described recently. ¹⁶ Cornea and lens were removed, and the retina was dissected from the sclera and flat-mounted in PBS. The retina was then stored in methanol at −20°C. Before immunofluorescence analysis, retinal flat mounts were postfixed in 4% PFA for 10 minutes, and blocked with 3% normal goat serum and 0.3% Triton X-100 prepared in PBS for 1 hour. Flat-mounts were incubated overnight with Isolectin IB4 (1:300, I21413; Invitrogen, Zug, Switzerland). After washing in PBS, retinas were mounted and immunofluorescence staining was analyzed with a digitalized microscope (Zeiss, Axioplan) or with a laser scanning confocal microscope (Leica SP5, Wetzlar, Germany). Imaris software (Bitplane AG, Zurich, Switzerland) was used to analyze Z-stacks and to generate XZ projections. 

RNA Isolation and Semiquantitative Real-Time PCR

Retinas were removed trough a slit in the cornea and snap-frozen in liquid nitrogen. Total retinal RNA was isolated using an RNA isolation kit (RNeasy; Qiagen, Hilden, Germany) including a DNase treatment; 1 μg RNA was used for reverse transcription using oligo(dT) and M-MLV reverse transcriptase (Promega, Dübendorf, Switzerland). Gene expression was analyzed by real-time PCR on 10 ng template cDNA using PCR polymerase ready mix (LightCycler480 SYBR Green I Master; Roche Diagnostics, Rotkreuz, Switzerland), and a thermocycler (LightCycler480; Roche Diagnostics). Specific primer pairs (Supplementary Table S1) were designed to span a large intronic region or an exon/exon boundary of the target genes, which were chosen based on gene profiling study analyzing selectively labeled retinal cell types ¹⁷ (http://www.fmi.ch/roska.data/index.php [in the public domain]). Signals were normalized to Actb and relative expression was calculated using LightCycler480 software (Roche Diagnostics) using a calibrator sample. 

N-Methyl-N-Nitrosourea Injections and Cell Death ELISA

N-methyl-N-nitrosourea (MNU; Sigma, Buchs, Switzerland) was dissolved freshly in physiologic saline solution and protected from light. The wt and R91W;Nrl^−/− mice received one intraperitoneal injection of 75 μg/g body weight for the experiments shown here. 

Mice were killed at different time points after MNU injection, eyes were either processed as described for morphology or the retinal tissue was isolated and snap frozen in liquid nitrogen. Cell death was determined by Cell Death Detection ELISA (Roche Diagnostics) according to the manufacturer's recommendation. 

Western Blotting

Retinas were homogenized in 100 mM Tris/HCl (pH 8.0) and analyzed for protein content by using the Bradford method. Standard SDS-PAGE and Western blot analysis were performed. Antibodies used were CASP1 (1:10,000, kindly provided by P. Vandenabeele, Ghent University, Ghent, Belgium); GFAP (1:1000, G3893, Sigma), pSTAT1 (1:500, 9171; Cell Signaling, Danvers, MA, USA); STAT1 (1:1000, 9172; Cell Signaling); pSTAT3 (1:500, 9132; Cell Signaling), STAT3 (1:1000, 9131; Cell Signaling); pERK1/2 (1:1000, 9101; NB, Beverly, MA, USA); ERK1/2 (1:1000, 9102; NB), pAKT (1:2000, 9271; Cell Signaling); AKT (1:2500, 9271; Cell Signaling), PI3K (1:4000, D0669; Upstate Biotechnology, Waltham, MA, US), and ACTB (1:10,000, A5441; Sigma). 

Electroretinography

Electroretinograms were recorded from both eyes simultaneously, as previously described. ¹⁸ Briefly, mice were dark-adapted overnight, pupils were dilated with cyclogyl 1% (Alcon Pharmaceuticals, Fribourg, Switzerland) and phenylephrine 5% (Bausch & Lomb Swiss AG, Zug, Switzerland) in dim red light. Mice were subcutaneously anesthetized with ketamine (85 mg/kg; Parke-Davis, Berlin, Germany) and xylazine (4 mg/kg Bayer AG, Leverkusen, Germany). The cornea was kept moist with a drop of methocel 2% (OmniVision AG, Neuhausen am Rheinfall, Switzerland). Light adaptation was accomplished with low background illumination starting 5 minutes before photopic recording. Single white-flash stimuli of intensities ranging from −20 to 25 dB (0.0002–722.0671 cd*s m⁻²) under photopic conditions, were divided into 10 steps. Ten responses were averaged with an interstimulus interval of 4.95 seconds (−20 dB, −14 dB, −10 dB, −4 dB, 0 dB, 5 dB) or 16.95 seconds (10 dB, 15 dB, 20 dB, 25 dB). Traces obtained from left and right eyes of one mouse were averaged for each light intensity. Averages from four independent mice were then used to calculate the mean values to generate traces presented in figures. 

Optomotor Response

All testing was performed between 12 PM and 3 PM to control for circadian variations in mouse activity levels. A virtual system was used as described previously ^19,20 to measure the OMR of each mouse line. In brief, a virtual cylinder was created with four computer screens (OptoMotry; CerebralMechanics, Lethbridge, Alberta, Canada) surrounding a raised platform on which the mouse was placed and allowed to move freely. The day before testing, mice were preconditioned to habituate them to the testing environment by placing them on the platform for 5 minutes. A video camera was suspended above the mouse and connected to a computer. A vertical grating moving at 12 degrees per second was projected on the four screens, and the frequency of the pattern was adjusted with accompanying computer software (OptoMotry; Cerebral Mechanics). The pattern was presented to the mouse, and an experimenter determined if the mouse had a positive optomotor tracking response (a “yes” response) at each frequency, defined as a reflexive head movement in the same direction and speed as the movement of the grating. The experimenter maintained the center point of the “virtual cylinder” at the mouse's head by manually adjusting a cursor on the accompanying computer screen based on the real-time video feedback. The visual threshold was calculated by the OptoMotry software using a staircase method with stepwise adjustments of the grating frequency up or down depending on the experimenter's “yes” or “no” response. The maximum spatial frequency for each eye was averaged for a combined OMR score, representative of the visual threshold for each individual mouse. Total number of experimental mice for each strain is indicated in brackets: wt (14), Gnat1^−/− (11), Nrl^−/− (14), R91W;Nrl^−/− (9), R91W;Nrl^−/−;Cpfl1 (10), and R91W;Nrl^−/−;Cpfl1;rd10 (6). 

Statistical Analysis

Statistical analysis was performed using Prism4 software (GraphPad, San Diego, CA, USA). All data are presented as mean values ± SDs. The number of samples (n) used for individual experiments is given in the Methods section. Two-way ANOVA followed by a Bonferroni's post hoc test was used to determine significance. P values below 0.05 were considered significant. 

Results

Ocular tissue of 6-week-old R91W;Nrl^−/− mice contained reduced 11-cis-retinal levels (9.1 ± 2.5 pmol per eye) that were approximately five times lower than in Nrl^−/− (42 ± 5.9 pmol) and 40 times lower than in wt (336.1 ± 21.8 pmol) mice. ⁵ Reduced 11-cis-retinal levels were associated with regular retinal layering, well-organized photoreceptor outer segments, and a continuous outer limiting membrane in R91W;Nrl^−/− mice. In marked contrast, Nrl^−/− retinas displayed characteristic rosettes and a wavy, discontinuous outer limiting membrane (Figs. 1a–c). The occasional detection of phagosomes in the retinal pigment epithelium of R91W;Nrl^−/− mice suggested normal phagocytosis of shed cone outer segments (Fig. 1c, bottom, inset, arrow). Recently it was shown that transgenic lines originating from C57BL/6N mice might carry the rd8 mutation in Crumbs homolog 1 (Crb1) gene resulting in rosette formation. ²¹ CRB1 is important for maintenance of adherent junctions between photoreceptors and Müller glia in the outer limiting membrane (OLM). ²² Rosette formation in Nrl^−/− mice was independent of rd8 mutation but solely relied on the presence of the wild-type Rpe65 (Supplementary Figs. S1, S2). 

Figure 1

 

R91W;Nrl^−/− mice develop a normally layered and functional all-cone retina. (a) Retinal morphology of 6-week-old Nrl^−/− (top), R91W;Nrl^−/− (middle), and wild-type (bottom) mice. R91W;Nrl^−/− mice do not develop rosettes in the outer nuclear layer. (b) Immunostaining of short-wavelength cone opsin (S-opsin) in 6-week-old Nrl^−/− (top), R91W;Nrl^−/− (middle), and wild-type (bottom) mice. (c) Retinal morphology of 6-week-old wt, Nrl^−/− , and R91W;Nrl^−/− mice. High-resolution light microscopy images of radial cross sections (top), outer limiting membrane (asterisks, middle), and transmission electron microscopy images of photoreceptor outer segments (bottom). Phagosome in R91W;Nrl^−/− RPE (bottom, arrow in inset). (d) Single flash photopic ERGs demonstrate the functionality of cones in R91W;Nrl^−/− mice (n = 4 per mouse line). (e) Visual acuity of various mouse strains under photopic conditions measured by optomotor response. Shown are means ± SD. Wt (n = 14), Gnat1^−/− (n = 11), Nrl^−/− (n = 14), R91W;Nrl^−/− (n = 9). ***P < 0.001. Scale bars as indicated. Cpd, cycle per degree; PS, photoreceptor segments.

The all-cone retina of R91W;Nrl^−/− mice was functional and showed a photopic b-wave amplitude that was twice as high as in wt at the highest stimulus intensity (Fig. 1d). However, the reduced levels of 11-cis-retinal in R91W;Nrl^−/− mice, most probably, caused a slight reduction in light sensitivity and b-wave amplitudes when compared with Nrl^−/− mice (Fig. 1d). The R91W;Nrl^−/− retina was not only functional, but also generated useful vision, even though photopic optomotor response tests showed that visual acuity was reduced by approximately 50% as compared with wt mice. Despite the presence of five times higher 11-cis-retinal levels, the visual acuity in Nrl^−/− mice was only slightly above the value in R91W;Nrl^−/− mice. As expected, Gnat1^−/− mice (no functional rods) showed a visual acuity similar to wt mice (Fig. 1e). Thus, combining the Nrl knockout with the R91W mutation resulted in the generation of a mouse with a morphologically normal and functional all-cone retina. 

To test whether the all-cone outer retina influences organization of the inner retina, we analyzed expression of specific cell markers using immunofluorescence (Fig. 2a). The M- and S-cone opsins (OPN1MW and OPN1SW) were detected almost exclusively in outer segments of wt and R91W;Nrl^−/− mice. In R91W, cone opsins were mislocalized to synaptic terminals, as reported previously. ⁸ Some rosettes in Nrl^−/− retinas were positive for both cone opsins, as reported before. ²³ Calbindin (CALB), protein kinase C alpha (PKC-A) and visual system homeobox 2 (VSX2, also known as CHX10), synaptophysin (SYP), and calretinin (CR) were used as markers for horizontal, bipolar, and synaptic terminals and amacrine cells, respectively. They showed no overall difference between rod-dominated (wt and R91W) and the all-cone R91W;Nrl^−/− retinas. In contrast, protein localization in Nrl^−/− was irregular and followed the contours of rosettes (Fig. 2a). The POU domain, class 4, transcription factor 1 (POU4F1, also known as BRN3A) localized to the ganglion cell layer in all strains. No clear signs of gliosis were visible in 6-week-old mice, as judged by GFAP immunofluorescence. However, increased GFAP immunoreactivity and higher Gfap mRNA and protein levels were observed in older R91W;Nrl^−/− and Nrl^−/− mice (data not shown). Thus no overall differences were observed between wt and R91W;Nrl^−/− mice in cell organization of the inner retina. This was corroborated by gene expression analysis of retina-specific cell markers in wt, Nrl^−/− , and R91W;Nrl^−/− mice (Fig. 2b; for validation of selected genes see Supplementary Fig. S3). Overexpression of Opn1sw was detected in both Nrl^−/− and R91W;Nrl^−/− mice. Expression of Opn1mw was slightly increased but generally comparable between all-cone and wt retinas (Fig. 2b). These findings are consistent with reports showing that the lack of Nrl leads to increased Opn1Sw expression and enhanced S-cone but not the M-cone activity, as measured by ERG. ^4,24,25 No difference among the three strains was found for Vsx2 (Chx10), a marker for bipolar cells. Markers for amacrine cells (Myf2) and ganglion cells (Pou4f1) were similarly expressed in wt and R91W;Nrl^−/− , but were significantly reduced in Nrl^−/− mice at later time points (statistics are presented in Supplementary Table S2). These data are consistent with a recent study reporting loss of cells in the ganglion and inner nuclear layer of Nrl^−/− mice. ²³ Atrophic axon terminal arbors, reduced higher-order branchlets, and smaller dendritic fields were reported for horizontal cells in dysmorphic Nrl^−/− mice, ²⁶ suggesting that horizontal cells may suffer directly from the irregular arrangement of cones. Notably, mitochondrial amidoxime reducing component 1 (Marc1), a marker gene for horizontal cells, was significantly reduced in Nrl^−/− , whereas expression in R91W;Nrl^−/− was similar to wt. These data suggest that the normal tissue lamination found in R91W;Nrl^−/− mice supports survival and normal physiology of second- and third-order neurons downstream of the all-cone nuclear layer. 

Figure 2

 

R91W;Nrl^−/− mice show normal expression and localization of the different retinal cell types. (a) Immunostaining for cell marker proteins in retinal cryosections of 6-week-old wt, R91W, Nrl^−/− , and R91W;Nrl^−/− mice. Nuclei were counterstained with DAPI. Scale bar as indicated. (b) mRNA levels of marker genes for individual retinal cell types analyzed by semiquantitative real-time PCR. Shown are means ± SD. n = 3. Myf6, myogenic factor 6.

Vascular disorganization and leakage were reported in retinas of Nrl^−/− mice. ²³ It is not known whether this vascular phenotype is a direct consequence of altered oxygen requirements of the cone-only retina, which were predicted to be approximately one-fourth of that in the wt in darkness, ²⁷ or whether it is secondary to rosette formation. At 6 weeks of age, all three vascular plexi were comparable between R91W;Nrl^−/− and wt mice (Fig. 3). In Nrl^−/− mice, however, especially the outer plexus was disturbed and distances between plexi were reduced. Thus, not the absence of rods, but the disturbed retinal lamination altered gross vascular architecture in Nrl^−/− mice, as this phenotype was absent in the R91W;Nrl^−/− retina. 

Figure 3

 

R91W;Nrl^−/− mice develop a normal retinal vasculature. Retinal whole mounts (top) and xz-stacks (bottom) from 6-week-old mice show that the retinal vasculature of R91W;Nrl^−/− was indistinguishable from wild-type mice, whereas the vascular plexi in Nrl^−/− mice were disturbed (see xz-stacks). D, deep; I, intermediate; P, primary vascular plexus.

Our analyses revealed that the R91W;Nrl^−/− mouse has a functional and morphologically normal all-cone retina, thus providing a valuable tool to study pathologies associated with cone photoreceptors (Figs. 1 1552–3). To induce cone degeneration, we used the chemotoxic compound MNU. Dose-response experiments (data not shown) established 75 μg/g body weight of MNU as the optimal dose for the R91W;Nrl^−/− mouse. Appearance of pyknotic nuclei was evident as early as 12 hours post injection and peaked between 24 and 48 hours in both wt and R91W;Nrl^−/− retinas (Fig. 4a). Quantification of cell death (Fig. 4b) and analysis of proteins (Fig. 4c) known to be involved in degeneration of rod-dominated retinas ²⁸ showed that timing of cell death and molecular response to MNU treatment was similar for rods and cones. However, because more cells survived in R91W;Nrl^−/− than in wt mice (Fig. 4a [12D]), cones may be more resistant to MNU toxicity than rods. 

Figure 4

 

The MNU induces cone cell death in R91W;Nrl^−/− mice. Six to 7-week-old wt and R91W;Nrl^−/− mice were analyzed at different time points after MNU application as indicated. (a) Retinal morphology showing progressive cone degeneration in R91W;Nrl^−/− similar to rod degeneration in wt mice. Note more surviving cells in all-cone retinas of R91W;Nrl^−/− 12 days after MNU treatment. (b) Relative quantification of photoreceptor cell death. Levels of free nucleosomes generated during apoptotic cell death were determined using an ELISA-based assay. Shown are means ± SD. n = 3. (c) Protein signaling determined by Western blotting in wt and R91W;Nrl^−/− mice at various time points after MNU treatment as indicated. Ctrl, control.

To monitor cone degeneration in a disease setting, we generated R91W;Nrl^−/−;Cpfl1 triple-mutant mice. Cpfl1 mice harbor a mutation in the cone PDE6 (α') catalytic subunit ²⁹ and cones most likely degenerate due to cGMP overload. ³⁰ Most cones in triple-mutant mice were lost between 6 and 12 weeks of age, especially in the inferior retina; however, degeneration was slow and some cones still persisted at 1 year of age (Figs. 5a, 5b). In addition, a strong photopic ERG and modest visual acuity was recorded in 8- to 9-week-old R91W;Nrl^−/−;Cpfl1 mice (Figs. 5c, 5d). In 8-month-old R91W;Nrl^−/−;Cpfl1 mice, the b-wave amplitude further diminished (Fig. 5c, overlay). The ectopic expression of rod PDE6 subunits in Nrl^−/−;Cpfl1 mice ³¹ may reduce cGMP levels, thereby supporting prolonged cone survival and function in R91W;Nrl^−/−;Cpfl1 mice. To completely ablate PDE activity, we generated quadruple R91W;Nrl^−/−;Cpfl1;rd10 mice, which carry an additional mutation in the beta subunit of rod PDE6. ¹¹ In these mice, retinal function was abolished (Figs. 5c, 5d) and the inferior retina degenerated with only one to two rows of photoreceptors left at 6 weeks of age (Figs. 5a, 5b). Again, the inferior retina was more affected. We also noted a slow degeneration in R91W;Nrl^−/− mice, with both retinal hemispheres similarly affected. 

Figure 5

 

Models of inherited cone degeneration on the basis of the all-cone R91W;Nrl^−/− mouse. (a) Retinal morphology and (b) measurement of the thickness of the outer retina in R91W;Nrl^−/− , R91W;Nrl^−/−;Cpfl1, and R91W;Nrl^−/−;Cpfl1;rd10 mice at 6 weeks, 12 weeks, and 1 year of age. Quadruple-mutant mice show fastest degeneration, especially in the inferior retina. Shown are means ± SD. n = 3. (c) Photopic ERG recordings of a single flash response series of 8- to 9-week-old R91W;Nrl^−/−;Cpfl1 and R91W;Nrl^−/−;Cpfl1;rd10, and comparison with indicated mouse lines at 25 db light intensity (overlay). The overlay shows recordings from 8- to 9-week-old mice, plus an 8-month time point for the R91W;Nrl^−/−;Cpfl1 mouse line. n = 4. (d) Optomotor response in 6- to 7-week-old mice of the following lines: wt (n = 14), R91W;Nrl^−/− (n = 9), R91W;Nrl^−/−;Cpfl1 (n = 10), and R91W;Nrl^−/−;Cpfl;rd10 (n = 6). Shown are means ± SD.

Discussion

General investigations of cone physiology or in vivo analyses of pathological mechanisms involved in the degeneration of cones in the central macula are hampered by a lack of suitable mammalian models with a cone-rich retina. Here we established the R91W;Nrl^−/− mouse that displays a well-ordered all-cone retina with a normal retinal vasculature and a strong photopic function that generates useful vision. R91W;Nrl^−/− retinas also retain wild-type–like expression levels of cellular markers of the inner retina, in marked contrast to Nrl knockouts, which may lose cells of the inner nuclear layer (INL) and ganglion cell layer (GCL) with time. ²³ Thus, the preserved retinal structure in the outer nuclear layer (ONL) of R91W;Nrl^−/− mice also benefits cells in the inner retina, which might allow a facilitated analysis of cone-mediated signal transduction pathways. 

We show that cone degeneration can be induced in R91W;Nrl^−/− mice by different means, allowing monitoring not only of the progression, but also cone-specific molecular mechanisms of retinal degeneration. The preserved retinal organization makes this model ideal for transplantation studies in which cone delivery, integration, and survival can be assessed to establish therapeutic approaches potentially applicable to the human macula. The strong retinal function makes the R91W;Nrl^−/− mouse especially useful also to test various other strategies to rescue cones in a cone-rich environment. Importantly, the success of the treatment can be analyzed in vivo via assessment of retinal function. So far, this was not achievable in mice, as most models for cone degeneration are predominantly models for achromatopsia with no remaining cone function. Other models are characterized by secondary cone cell death where cone degeneration is a late and slowly progressing event that is further influenced by surrounding rods. 

Age-related macular degeneration is characterized by gradual loss of central vision due to the photoreceptor degeneration within the macular region. Several aspects of AMD, such as RPE atrophy, lipofuscin accumulation, increase of A2E in senescent mice, and general photoreceptor degeneration, were previously simulated and studied in several mouse models. ³² However, the lack of the macula in mice hindered the molecular analysis of the loss of central vision caused by cone degeneration. By using the all-cone retina of R91W;Nrl^−/− mice, this aspect can now be analyzed. However, a careful interpretation of the experimental data has to be undertaken not to oversimplify the complexity of AMD and other macular degenerations. Anatomically, the all-cone retina of R91W;Nrl^−/− mice is composed of two cone types, whereas the human retina contains three types. The predominant cone type in R91W;Nrl^−/− mice is the blue-light–sensitive cones, ⁴ which are absent from the human macula. Additionally, the absence of parafoveal rods, which are the first cells to die in AMD, also has to be considered. ³³  

In humans suffering from enhanced S-cone syndrome, the only retinal disease with a gain in number of cone photoreceptors, distorted retinal laminar architecture, and rosette presence have been documented. ^34,35 Similarly, mice lacking NRL have increased numbers of cones and develop rosettes in the ONL. This pathology leads to aberrant photoreceptor packing, abnormal association between photoreceptors and retinal pigment epithelium, retinal detachment, changed dendritic fields and axon terminals in horizontal cells, vascular leakage, and ganglion cell death. ^23,25 The exact mechanism of how rosettes form in these retinas has not been resolved, but inspection of the retinal morphology of Nrl^−/− mice suggests deficits in the formation of the OLM, a notion in line with a recent study that proposed that absence of rods disturbs OLM formation. ³⁶ Crucial for the formation of the OLM is the CRB1 protein that regulates localization of adherens junctions between photoreceptors and Müller glia cells. Recently it was shown that transgenic lines originating from C57BL/6N mice (such as Nrl^−/− ) might carry an rd8 mutation in Crb1 (Crb1^rd8 ), which can cause rosette formation when homozygously present. ²¹ However, rosette appearance in the Nrl^−/− mouse was independent of the rd8 mutation but strictly depended on the Rpe65 status; that is, different combinations of Rpe65 and Crb1 alleles were tested on the Nrl^−/− background (see Supplementary Fig. S1) and rosette formation was prevented only in mice with reduced RPE65 levels. Several groups have already shown that rosette formation is prevented in Nrl^−/− mice when no functional RPE65 protein is present and that rosette formation can be reestablished by the exogenous application of 11-cis-retinal. ^5–7 This suggests the involvement of some retinoid intermediates, most probably cis-retinoids, in rosette morphogenesis. Collectively, our data show that low levels of cis-retinoids are tolerated in the Nrl^−/− retina and that normal retinal lamination can be established in the absence of rods. 

Even though the all-cone retina of R91W;Nrl^−/− mice is functional and morphologically normally layered, long-term survival of photoreceptors is reduced. The slightly increased basal levels of pSTAT3, GFAP, and CASP1 (Fig. 4c, line “ctrl”) suggest moderate retinal stress that may lead to the activation of degenerative processes with time. The absence of rods may affect these processes in several ways. First, the all-cone retina of R91W;Nrl^−/− mice expresses five times less Nxnl1 mRNA then wt (data not shown). The Nxnl1 gene encodes for rod-derived cone viability factor, which was proposed to be secreted mainly by rod cells to support cone photoreceptor viability. ^37,38 Second, lack of rods may impair interactions of cone outer segments with the RPE leading to a reduced nutrition of cones from the choroid/RPE. ³⁹ Cone starvation may be the consequence, which has been proposed as a cause for nonautonomous cone cell death in models of RP. ³⁹ Further experiments are needed to define the mechanisms leading to the observed age-dependent cone loss in R91W;Nrl^−/− retinas. 

Moderate and fast degeneration was observed in the R91W;Nrl^−/−;Cpfl1 and R91W;Nrl^−/−;Cpfl1;rd10 models, respectively. The R91W;Nrl^−/−;Cpfl1 triple-mutant mouse harbors an additional mutation in cone phosphodiesterase (Pde6c), which leads to complete loss of cone function in the normal rod-dominated retina. ⁴⁰ In R91W;Nrl^−/−;Cpfl1 mice, however, retinal function is reduced but not fully ablated. It has been recently demonstrated that cones in Nrl^−/−;Cpfl1 mice also express rod PDE6 catalytic subunits that can partially replace the nonfunctional cone PDE alpha subunit. ³¹ Potentially, this may also reduce toxic cGMP levels and support cone viability and function in R91W;Nrl^−/−;Cpfl1 mice. To completely ablate retinal function and potentially accelerate degeneration, we generated quadruple R91W;Nrl^−/−;Cpfl1;rd10 mice, which carry an additional mutation in the beta subunit of rod PDE6. This resulted in the complete loss of retinal function and in a severe retinal degeneration with most cone photoreceptors lost in the inferior retina within the first 6 weeks of age. This was not due to the developmental deficits, as 2-week-old mice showed a normal ONL thickness with no apparent signs of degeneration (data not shown). 

The prominent cone degeneration in the functional R91W;Nrl^−/−;Cpfl1 retina is especially useful for rescue experiments. Between 6 and 12 weeks of age, 50% of cones are lost in the inferior retina of these mice. Because this loss of photoreceptors is reflected by a corresponding loss of function, treatment success can be directly evaluated by ERG in young mice with a high statistical power. 

In summary, we present a mouse with a functional, normally layered all-cone retina that can be used to study various aspects of cone degeneration. We need to emphasize that the R91W;Nrl^−/− mouse is not a bona fide model for AMD, but facilitates the investigation of various aspects of this and other human diseases that affect cones in the rod-free central macula. The main advantage of the R91W;Nrl^−/− mouse model over most rodent models for inherited cone degeneration is that the R91W;Nrl^−/− retina has a well-organized all-cone retina with sustained cone visual function. Molecular mechanisms regulating cone cell death can thus be studied in the absence of confounding signals from dying rods. Given the importance of the central high-acuity cone vision for humans and the presence of numerous diseases threatening its function and the survival of macular cones, the retina of the R91W;Nrl^−/− mouse may not only facilitate investigations of cone-related mechanisms but may also significantly contribute to the development of therapeutic strategies to preserve central, cone-mediated vision. Such strategies may include neuroprotective treatments, gene therapy approaches using viral and nonviral delivery systems, photoreceptor or stem cell transplantation, and the use of biodegradable scaffolds to support regenerative processes. 

Supplementary Materials

pdf S1 

Acknowledgments

The authors thank Andrea Gubler, Cornelia Imsand, and Christel Beck for their excellent technical support, and Anand Swaroop (University of Michigan) for providing the Nrl^−/− mice. The authors alone are responsible for the content and writing of the paper. 

Supported by the Swiss National Science Foundation (SNF #31003A_133043), University of Zurich Forchungskredit, Vontobel Foundation, and SwissLife. 

Disclosure: M. Samardzija, None; C. Caprara, None; S.R. Heynen, None; S. Willcox DeParis, None; I. Meneau, None; G. Traber, None; C. Agca, None; J. von Lintig, None; C. Grimm, None