Biphasic Photoreceptor Degeneration Induced by Light in a T17M Rhodopsin Mouse Model of Cone Bystander Damage

Abstract

purpose. To evaluate light-induced retinal damage in transgenic T17M rhodopsin mice as a novel model for bystander cone damage during retinal degeneration.

methods. Mouse eyes were exposed to bright white light (15,000 lux, 2.5 minutes). After exposure, electroretinography was performed on mice dark adapted for 12 or more hours at 0 to 5 days to test photoreceptor response or for 0 to 12 hours to test response recovery. Retinal cryosections were examined by TUNEL staining and outer nuclear layer thickness measurements. Cone morphology was assessed by peanut agglutinin staining in retinal flatmounts and cryosections.

results. T17M retinal function and morphology changed rapidly after exposure to light. Scotopic and photopic electroretinogram responses declined progressively from 0.5 to 3 days. Scotopic response recovery peaked at 50% to 60% of the unilluminated response in 3 hours, indicating an early, rapid decline in scotopic signaling. Photopic responses were near normal or supernormal from 0 to 6 hours. Cell death peaked at 1 day, and outer nuclear layer thickness declined from 1 to 5 days. Disorganized cones were observed at 6 hours, intact and damaged cones were observed at 12 hours and 1 day, but only cone remnants were observed at 3 and 5 days. Light exposure had little to no effect on ERG responses in nontransgenic littermates and other retinal degeneration models.

conclusions. The time course of light-induced T17M retinal damage is biphasic, with an initial decline in rod function within hours followed by bystander cone and rod deterioration within days. The rapid and synchronous induction of damage in this model is attractive for characterizing bystander effects in retinal degeneration.

In tissue injury and disease, the death of cells targeted by physical or chemical insults, infectious agents, or disease-causing mutations is often associated with damage to nontargeted neighboring cells. A striking example of this “bystander” effect occurs in retinitis pigmentosa, a group of retinal degenerative diseases that progress from defects in night and peripheral vision to a loss of central acuity and often complete blindness. ¹ Retinitis pigmentosa has been linked to mutations in more than 45 genes, some of which affect proteins expressed exclusively in rod photoreceptor cells, such as rhodopsin, the α or β subunit of rod cGMP phosphodiesterase, and the α or β subunits of the rod cGMP-gated channel. ^{1 2} Although rods are the primary target in these forms of retinitis pigmentosa, their demise is followed more gradually by cone dysfunction and death. Cone loss occurs initially in the peripheral retina, varies with the mutation, and depends on the extent of rod loss, as demonstrated in patients with rhodopsin retinitis pigmentosa. ^{1 3} Many theories explaining cone deterioration in rod-initiated retinitis pigmentosa have been proposed, ^{4 5 6 7 8 9 10} but the precise mechanisms remain unresolved. Identifying these mechanisms may provide general insights into bystander damage during tissue pathogenesis and lead to therapies that slow or abolish cone loss in retinitis pigmentosa, which greatly diminishes patient quality of life. 

Studies of cone photoreceptor fate in spontaneous and transgenic animal models of rod-initiated retinal degeneration have revealed functional and morphologic changes consistent with bystander effects in cones. ^{11 12 13 14 15 16 17 18} Rod cell death in these models is accompanied by a reduction in cone electroretinogram (ERG) amplitudes, cone outer segment shortening, cytoplasmic densification, axonal elongation, and, ultimately, cone cell death. ^{11 12 13 14 15 16 17 18} As is observed clinically, the timing of cone loss differs among the models, ranging from weeks to years after rods die. Efforts to elucidate the mechanisms of cone damage in these models are limited because degeneration progresses slowly, which precludes examination of early signaling events, and because rod cell death occurs continuously in an unsynchronized fashion, which may obscure biochemical and transcriptional events that link rod death to cone bystander effects. 

Light-induced retinal damage, ¹⁹ also termed photic or phototoxic retinopathy ^{20 21} or photochemical retinal damage, ²² may be considered an alternative model for studying bystander cell loss. Intense illumination induces retinal degeneration accompanied by the death of rods and cones. Because it is easily manipulated, noninvasive, and rapid, light has been used extensively as an experimental trigger of retinal degeneration. Whether light can be used to study bystander photoreceptor loss requires evidence that cone damage depends on rods. In mice, gross degeneration of the outer retina in light-induced damage requires rhodopsin ²³ and is enhanced by knockout mutations in the rod proteins arrestin and rhodopsin kinase. ^{24 25} Although these studies indicated that light-induced damage in mice is rod initiated, they did not include experiments to specifically examine cones, which represent less than 5% of the photoreceptor population ²⁶ and are undetected by gross measures of retinal morphology. 

The present studies were undertaken to identify an experimentally tractable mouse model for examining cone bystander effects. Extreme sensitivity to bright light was recently demonstrated in the T4R rhodopsin dog ²⁷ and the transgenic T17M rhodopsin (T17M) mouse, ²⁸ two animal models of retinitis pigmentosa. Brief illumination times were sufficient to induce fundus changes, shortening of photoreceptor outer segments, loss of rod ERG amplitudes, photoreceptor apoptosis, and thinning of the retinal outer nuclear layer (ONL). The effect of illumination on cones was not specifically examined. Here, we show that light exposure of T17M mouse eyes induces photoreceptor loss in two phases: an early phase restricted to rods that progress within hours after exposure and a late phase involving rods and cones that continues over the next 3 days. We demonstrate that light-induced cone damage in this model is a bystander effect dependent on the T17M rhodopsin mutation in rods because it is absent in nontransgenic littermates and the BALB/c strain often used in phototoxicity studies. We conclude that this mouse model is an attractive model for elucidating bystander cone loss, light toxicity, and therapeutic approaches in retinal degenerative disease. 

Methods

Mice

All mice were given unrestricted access to food and water and raised on cage racks under fluorescent lighting with a 12-hour light/12-hour dark cycle. Light intensity inside cages was less than 10 lux, as measured with a light meter (model 401036; Extech, Waltham, MA). Mice were weaned, and tails were snipped at 21 to 24 days of age. For genotyping, tail-snip DNA was prepared with a PCR kit (Extract-N-Amp Tissue PCR; Sigma-Aldrich, St. Louis, MO). All animal care and use adhered to the guidelines detailed in the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

C57BL/6, BALB/c, and homozygous rd12 mice were obtained from the Jackson Laboratory (Bar Harbor, ME). Transgenic mice expressing human opsin with the T17M or P23H mutations were described previously ²⁸ and were maintained by breeding with C57BL/6. Transgenic mice were crossed with homozygous rhodopsin knockout mice (Rho ^−/−) ²⁹ to produce P23H Rho ^+/− and T17M Rho ^+/− mice, which contain single copies of the transgenic human and endogenous murine opsin genes. Presence of the transgene was confirmed by PCR of tail-snip DNA using the primers 5′-AGTGCACCCTCCTTAGGCA-3′ and 5′-TCCTGACTGGAGGACCCTAC-3′ to generate a 290-bp fragment that was absent in nontransgenic mice. Heterozygous noerg-1 mice containing the C110Y mutation in the mouse rhodopsin gene (a generous gift of Lawrence Pinto, Northwestern University, Evanston, IL) were bred with C57BL/6 (Jackson Laboratory). Mice were genotyped from tail-snip DNA in two PCR reactions pairing the primers 5′-CAAAGAAGCCCTCGAGATAAT-3′ or 5′-CAAAGAAGCCCTCGAGATAAC-3′ with 5′-TCACCACCACCCTCTACACA-3′. Mice heterozygous for the noerg-1 mutation yielded a product with both primer pairs. All experimental mice were pigmented except for BALB/c, which is an albino strain. 

Retinal Light Exposure

Young adult mice (32–34 days old) were used for all experiments. Light exposure was initiated around noon, halfway through the illuminated portion of the lighting cycle. Both eyes were treated with 1% atropine sulfate, 2.5% phenylephrine, and 0.5% proparacaine HCl (Akorn, Buffalo Grove, IL). This treatment was repeated after 1 hour, and, 15 minutes after that, mice were anesthetized for the duration of the procedure with isoflurane (Hospira, Lake Forest, IL) and a gas anesthesia system (Isovet 4; Surgivet, Waukesha, WI). The right eye was illuminated at 15,000 lux for 2.5 minutes with a fiber-optic illuminator containing a 150-W quartz-halogen lamp (model 170-D; Dolan-Jenner, Roxborough, MA). Light intensity at the tip was measured with the Extech (Waltham, MA) light meter. The fiber-optic tip was held approximately 1 to 2 mm from the cornea in a drop of 2.5% hypromellose (Akorn). Left eyes were shielded from light exposure. After light exposure, mice were returned to cyclic lighting or placed immediately in darkness for adaptation. 

Electroretinography

For ERG analysis at 12 hours and later, mice were dark adapted for 12 hours before ERGs were recorded. For analysis of ERG response recovery, mice were dark adapted for the indicated times immediately after light exposure. After dark adaptation, mice were manipulated under dim red light (>630 nm). Mice were anesthetized with intraperitoneal injection of 50 mg xylazine/kg body weight (AnaSed; Lloyd Laboratories, Shenandoah, IA) and 50 mg ketamine/kg body weight (Ketaset; Wyeth, Madison, NJ). Mouse corneas were anesthetized locally with 0.5% proparacaine hydrochloride (Bausch & Lomb, Rochester, NY), and pupils were dilated with 2.5% phenylephrine hydrochloride (Bausch & Lomb). After anesthesia and during and after ERG recordings, mice were placed on a 37°C surface to maintain body temperature. Ground and reference electrodes were inserted subdermally by the hind limb and centered along the nasal ridge, respectively. Gold loop electrodes were placed on each eye with a drop of 2.5% hypromellose. 

Full-field scotopic ERGs were obtained simultaneously from both eyes with a visual electrodiagnostic system (EPIC-2000; LKC Technologies, Gaithersburg, MD) equipped with a Ganzfeld accessory and a Grass flash unit (flash duration ∼10 μs; flash intensity, 2.4 cd · s/m²). Scotopic responses were filtered between 0.05 and 8000 Hz with no notch filter, and four separate recordings were averaged. Mice were then light adapted for 7 minutes, and photopic responses to a 2.4 cd · s/m² stimulus at 2.4 Hz were acquired in the presence of the adapting illumination. Photopic responses were filtered identically to the scotopic responses, and 50 sweeps were averaged after a 10-second preadaptation period. 

ERG waveforms were exported in ASCII format and analyzed (Excel; Microsoft, Redmond, WA) with a macro that plots individual waveforms. The macro also calculates b-wave amplitudes from the difference between the minimum of the corneal-negative trough to the maximum of the corneal positive peak and a-wave amplitudes as the difference between the minimum of the corneal-negative trough and the preflash zero baseline. 

Tissue Preparation

Mice were humanely killed by isoflurane overdose followed by cervical dislocation, eyes were excised, and the lens and cornea were removed. For cryosections, posterior cups were fixed for 2 hours on ice in 0.1 M sodium phosphate, pH 7.4, containing 4% paraformaldehyde prepared freshly from a 16% aqueous stock (Electron Microscopy Sciences, Hatfield, PA). After fixation, eyes were dehydrated for 12 to 18 hours at 4°C in 30% wt/vol sucrose, 0.1 M sodium phosphate, pH 7.4, and embedded in optimal cutting temperature compound (Tissue-Tek; Sakura Finetek USA, Torrance, CA). Frozen sections (12-μm thick) through the optic nerve were dried on microscope slides. For retinal flatmounts, the posterior cup was fixed for 1 hour on ice in 4% paraformaldehyde in phosphate-buffered saline (PBS). After fixation, the scleral surface of the posterior cup was placed on a nitrocellulose membrane, four or five radial cuts were made from the periphery toward the optic nerve head, and the resultant segments of the posterior cup were flattened. A small piece of nitrocellulose membrane was applied to the vitreal surface and then peeled from the posterior cup with the retina still attached. The membrane supported the retina during staining and was removed before mounting. 

Histology

Cell death was detected in retinal cryosections with terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling (TUNEL) assay (In Situ Cell Death Detection Kit TMR red; Roche Applied Science, Mannheim, Germany) according to the manufacturer’s instructions. Nuclei were additionally stained by mounting the sections with 4′,6-diamidino-2-phenylindole (DAPI) in medium (Vectashield; Vector Laboratories, Burlingame, CA). Sections were examined on an inverted fluorescence microscope (DMI 6000 B; Leica Microsystems, Bannockburn, IL) equipped with a video camera (Retiga 4000R; QImaging, Tucson, AZ) and image processing and analysis software (Image-Pro Plus; Media Cybernetics, Bethesda, MD). TUNEL⁺ nuclear counts were normalized to the area of the ONL counted. ONL thickness was measured in images of DAPI-stained retinal cryosections on each side of the optic nerve, 250 μm from its center. Distance calibration and measurements were performed with image processing and analysis software (Image-Pro Plus; Media Cybernetics). Values from each mouse were averaged before to obtain a group average. 

Cone morphology was assessed in retinal cryosections and flatmounts by staining with rhodamine-conjugated peanut agglutinin (PNA; Vector Laboratories). Cryosections were washed for 1 hour in PBS, blocked on the slide for 1 hour in PBS containing 3% wt/vol bovine serum albumin (BSA), and stained for 40 minutes with PNA diluted 1:500 in PBS containing 1% BSA. After washing four to five times with PBS, stained sections were treated for 10 minutes with 4% paraformaldehyde in PBS, washed twice with PBS, and mounted in DAPI-containing medium (Vectashield; Vector Laboratories). All steps were performed at room temperature. Enhanced depth-of-field images were acquired in z-stack mode (20–30 planes of approximately 1 μm) and aligned, deconvoluted, and merged with image processing and analysis software (ImagePro SharpStack and AutoQuant X; Media Cybernetics). Retinal flatmounts were blocked for 30 minutes in PBS containing 10% normal goat serum (Sigma), stained for 1 hour with PNA (1:1000) in PBS containing 1% normal goat serum, and washed for 10 minutes in PBS. All steps were performed at room temperature. Flatmounts were placed in medium (Vectashield; Vector Laboratories) on slides with the photoreceptor side facing up and were examined with the fluorescence microscope (DMI 6000 B; Leica Microsystems). 

Eyecups were not oriented by their superior-inferior axis before preparing flatmounts or sectioning. In mouse studies of light-induced damage, eyecups are often oriented to assess regional differences in damage. Damage is typically higher in the inferior than in the superior retina and less pronounced in the periphery, ^{30 31} though there are examples of more symmetrical damage with less than 15% variation within approximately 0.2 to 1 mm of the optic nerve head. ^{24 32} Regional damage in animal models of retinitis pigmentosa and in patients with it has been proposed to arise from a nonuniform distribution of light at the retina, which is determined in part by the directionality of the light source (for example, the use of overhead lighting in animal studies). ^{16 30 33 34 35} More uniform damage is expected from the delivery of light through a fiber-optic source at the corneal surface of a dilated eye, as in our study. Thus, randomly oriented samples were imaged within 0.2 to 1 mm of the optic nerve head. 

Statistical Analysis

ERG amplitudes and mean ONL thickness values from independent mice were averaged, and the SD was calculated (Excel; Microsoft). Data are presented as mean ± SEM. Significance was tested (Excel; Microsoft) with a paired two-tailed Student’s t-test (unpaired for ERG response recovery data) with unequal variance. P < 0.05 was considered significant. 

Results

ERG Response in T17M and Nontransgenic Mice

We examined the ERG responses of untreated mice at 1 month of age to ensure scotopic (primarily a rod-initiated response) and photopic (primarily a cone-initiated response) signals could be detected (Fig. 1) . The scotopic response in T17M mice was significantly lower than that of nontransgenic littermates, confirming previous reports. ^{28 36} Scotopic a- and b-wave amplitudes were 32% and 67% of nontransgenic values, respectively (T17M a-waves 160 ± 11 μV, b-waves 814 ± 36 μV, n = 10; nontransgenic a-waves 505 ± 26 μV, b-waves 1207 ± 68 μV, n = 8). These results imply a loss of rod-dependent signaling in T17M mice consistent with retinal degeneration. Photopic a- and b-wave amplitudes were 73% and 76% of nontransgenic values, respectively (T17M a-waves 15 ± 1 μV, b-waves 120 ± 12 μV; nontransgenic a-waves 20 ± 1 μV, b-waves 164 ± 16 μV). Photopic responses, which were obtained with a strobe stimulus in the presence of background illumination to isolate cones, closely matched responses in mice lacking rod function obtained under similar conditions. ^{37 38} Thus, photopic responses reflected the activity of cones, which showed degeneration at 1 month in T17M mice. 

ERG Responses in Dark-Adapted T17M Mice after Light Exposure

To test whether cone damage accompanies light-induced rod damage in T17M mice, we exposed their right eyes to bright white light. At various times after light exposure, mice were dark adapted, and ERG analysis was performed under scotopic and photopic conditions. The averaged scotopic ERG responses of illuminated eyes were diminished at all time points compared with simultaneously measured unilluminated contralateral eyes (Fig. 2) . Photopic responses also were diminished at all time points, though to a lesser extent than scotopic responses. No decrease was observed in illuminated nontransgenic littermates (data not shown). 

To evaluate the decline in retinal function, the ratio of mean ERG responses (illuminated/unilluminated) was plotted as a function of time after light exposure (Fig. 3) . At 0.5 days, scotopic a- and b-wave amplitude ratios (Figs. 3A 3B , left panels) were significantly depressed in T17M mice (0.33 ± 0.08 [P = 0.0005] and 0.26 ± 0.06 [P = 0.0004], respectively; n = 6), indicating a substantial loss of rod function. At this earliest time point, the photopic a-wave amplitude ratio was decreased significantly as well (0.52 ± 0.11, P = 0.006), consistent with loss of cone function (Fig. 3A , right panel). In contrast, the photopic b-wave amplitude ratio (Fig. 3B , right panel) at 0.5 day was unaffected (0.93 ± 0.20, P = 0.73). From 1 to 3 days after light exposure, scotopic and photopic responses decreased to their minimum observed values (Figs. 3A 3B) . From 3 to 5 days, scotopic and photopic response ratios remained low, with values ranging from 0.08 to 0.11 and 0.18 to 0.23, respectively. These results imply that rod and cone function decline coordinately from 0.5 to 3 days after light exposure. The low level of activity at the latest time points indicated that the bright light exposure was insufficient to completely abolish photoreceptor function, at least during the period examined. ERG responses of nontransgenic littermates were unaffected by light exposure because the amplitude ratios in these mice were close to 1.0 throughout the time course, and no significant differences were detected between illuminated and unilluminated eyes (Figs. 3A 3B , open circles). 

ERG Response Recovery after Light Exposure

The dramatic decrease in rod function at 12 hours after light exposure raised the question of whether functional changes could be observed at earlier times. The brief stimulus allowed us to examine ERG response recovery after a variable period of dark adaptation starting within minutes of light exposure. This approach is conceptually similar to analysis of dark adaptation, ³⁹ though light of greater intensity and duration is used to induce damage. Our approach also differed in that individual mice were examined at a single time point to minimize potential complications from light-induced damage during ERG analysis. Averaged ERG responses during the recovery of T17M mice and nontransgenic littermates are shown in Figure 4 , and the corresponding time course of amplitude ratios is shown in Figure 5 . 

Light exposure effectively reduced the rod response in nontransgenic and T17M mice (Fig. 4) . A photobleach of greater than 93% was estimated from the y-intercept of a linear fit to the first three points (Fig. 5) . The actual degree of bleaching was probably higher given the intensity and duration of the light stimulus. Recovery from this photobleach in nontransgenic animals was characterized by an early linear phase followed by a more prolonged nonlinear phase, as described in typical dark adaptation studies. ³⁹ However, the two phases were interrupted by a dip in response ratio at approximately 2 hours. This dip might have resulted from rod outer segment shedding induced by the bright stimulus, as observed within 2 hours of light exposure in the T4R dog. ²⁷ Recovery of scotopic responses in T17M mice was initially similar to that in nontransgenic mice but subsequently showed progressive slowing. T17M a- and b-wave amplitude ratios reached peak values of 0.62 ± 0.10 and 0.52 ± 0.06, respectively, at 3 hours. At 6 hours, the a- and b-wave ratios declined further to 0.46 ± 0.09 and 0.32 ± 0.05, respectively, and remained unchanged at 12 hours within experimental error. By contrast, nontransgenic scotopic amplitude ratios were more than 1.0 at 12 hours, indicating full ERG recovery. These results are consistent with an irrecoverable loss of 50% to 70% of rod function in T17M mice within 6 hours of exposure to bright light. 

Photopic a-wave amplitude ratios for nontransgenic mice after light exposure decreased immediately to approximately 0.3, rose linearly over the next 2 hours, and approached a plateau at later times (Figs. 4 5) . The b-wave response was more complex and included a peak in recovery at approximately 2 hours that was slightly higher than the unilluminated value (Fig. 4) . More rapid recovery of the photopic response might have been expected because mouse cones recover fully within seconds of a brief, subsaturating flash. ⁴⁰ However, the photopic response to a longer supersaturating exposure, as used here, has not been described. The initial decrease in ERG response and the slow recovery might have arisen in part from cone pigment photobleaching and regeneration, though changes in postreceptoral signaling might also have contributed. Strikingly, from 1 to 6 hours, T17M photopic b-wave amplitude ratios were higher than in nontransgenic mice whereas a-wave ratios were similar (Fig. 5) . The T17M photopic b-wave response peaked at approximately 2 hours, as in nontransgenic mice, though to a greater extent. These results point to an increase in T17M postreceptoral signaling from 1 to 6 hours that was supported by the more pronounced oscillatory potentials of T17M photopic responses during this period (Fig. 4) . The similarity of the a-wave amplitude ratios in T17M and nontransgenic mice suggests that T17M cone function was less affected than rods for at least 6 hours after light exposure. One explanation for the “supernormal” b-waves in T17M mice could be the derepression of cone signaling when rods are desensitized. ⁴¹ Together with the scotopic observations, these findings support an early phase of T17M damage during which rod function declined precipitously while cone function was close to normal, with increased postreceptoral signaling. 

Other Retinal Degeneration Models Less Affected by Light Exposure

We demonstrated previously that rod sensitivity to light exposure was attributed to the T17M rhodopsin mutation because transgenic P23H rhodopsin mice were insensitive to a 5000 lux, 2.5-minute light exposure. ²⁸ To extend this finding with the illumination protocol used in the present study and to test whether cone function was affected, we examined BALB/c mice, often used as a model in phototoxicity studies, and transgenic P23H rhodopsin and noerg-1 (C110Y rhodopsin) mice, which model autosomal dominant retinitis pigmentosa. At 5 days after light exposure, no significant differences in the scotopic or photopic ERG responses of illuminated and unilluminated eyes were observed in BALB/c mice (Fig. 6) . A slight effect was observed in P23H mice (scotopic a-wave amplitude ratio 0.87, P = 0.004, n = 9; photopic b-wave amplitude ratio 1.18, P = 0.04, n = 9) and C110Y mice (photopic a-wave amplitude ratio 0.82, P = 0.004, n = 10). Other ERG responses in these mice did not differ significantly with light exposure. These results suggest the illumination protocol induces minimal rod or cone photoreceptor damage in these mouse lines. Light exposure did not alter the scotopic or photopic response in rd12 mice, which are defective in the visual cycle because of a stop mutation in the RPE65 gene (Fig. 6) . During the course of this study, we became aware that the photopic response in rd12 (Hauswirth W, personal communication, October 2008) and other RPE65 null mice ⁴² is caused solely by rod activity; hence, these results cannot address whether light exposure affects cone function. Nevertheless, the behavior of rd12 mice supports the claim that hypersensitivity to light damage is a unique attribute of the T17M rhodopsin mutation and not a general characteristic of retinal degeneration. 

Photoreceptor Cell Death and Clearance

To establish whether the loss of ERG responses was accompanied by morphologic changes, we examined retinal cryosections for photoreceptor cell death and clearance of photoreceptor nuclei. As determined by TUNEL analysis (Fig. 7) , cell death in illuminated eyes was observed in the ONL. TUNEL⁺ nuclei were observed in the ONL of illuminated T17M eyes but rarely in unilluminated or nontransgenic eyes (Fig. 7) . The number of TUNEL⁺ nuclei, normalized to the ONL area, was significantly increased from 12 hours to 3 days after light exposure, with a peak value of 3.3 ± 0.9 × 10³ nuclei · mm⁻² at 1 day (Fig. 7M) . This corresponded to a small fraction (∼8%) of all photoreceptor nuclei, based on a nuclear density of 4.7 × 10⁵ nuclei · mm⁻² in the ONL measured from DAPI-stained cryosections. Cell death in T17M eyes occurred without a significant change in the ONL thickness at 6 and 12 hours after light exposure (Fig. 7N) . At later times, a steady decrease in ONL thickness was observed and became significant at 1 and 3 days (Fig. 7N) . No changes in ONL thickness were apparent in nontransgenic mice. These results are consistent with a lag between photoreceptor cell death and clearance after light-induced damage in T17M mice, as observed in other forms of retinal degeneration. They also suggest that processes other than photoreceptor cell death and clearance cause the 60% to 80% decrease in rod ERG response observed at 12 hours (Fig. 3) . 

Histologic Evidence for Cone Damage

To test whether light exposure affected cone morphology, we stained retinal flatmounts and cryosections with rhodamine-conjugated PNA, which labels the sheath of cone inner and outer segments and cone pedicles (Fig. 8) . Cryosections were mounted with nuclear stain (DAPI) to identify retinal layers. Well-developed cone inner and outer segments were apparent in control T17M eyes (Figs. 8A 8B) , and cone pedicles were clearly defined in the outer plexiform layer (Fig. 8B , arrowheads). At 6 hours after light exposure, intact cones with shortened outer segments were observed. Outer and inner segments were abnormally clustered in flatmounts and disorganized in cryosections. At 12 hours and 1 day, outer segments were lost from many cones. By 3 days, PNA-stained debris was abundant, with very recognizable outer or inner segments. PNA staining in the outer plexiform layer was diffuse, and discrete pedicles were no longer apparent. At 5 days, only a few areas were stained with PNA in flatmounts, and no cones were detected near the optic nerve head in either flatmounts or cryosections, though a few cones were found in the retinal periphery. Cone pedicles were absent. Cone morphology in nontransgenic mice was unaffected by light exposure (data not shown). These results indicate that light-induced damage in T17M mice is associated with cone deterioration and has the same time course as the clearance of photoreceptor nuclei. 

Discussion

Analysis of the underlying mechanisms that govern the retinal response to injury and disease faces considerable challenge because of the complexity of the tissue, which in the mouse includes seven major neuronal and glial cell types, resident microglial, vascular endothelial, and retinal pigment epithelial cells, ^{43 44} and the dynamic nature of the response, which can evolve from damage to a single afflicted cell type to wholesale restructuring of the retina on a time scale from days to years. ⁴⁵ The study of bystander cone damage in rod-initiated retinal degeneration has the potential to simplify this analysis in that the fate of a single cell type is examined under conditions in which a different cell type is initially targeted. Identifying the cellular and molecular processes that govern cone damage under these conditions may help to elucidate the retinal response to more complicated pathogenic stimuli. To this end, we have developed a mouse model for studying bystander cone damage. The model is based on the earlier observation that the T17M mouse model of retinitis pigmentosa is hypersensitive to bright light. ²⁸  

Several lines of evidence indicate that cone damage in the model represents a bystander effect. First, light exposure of T17M mouse eyes affects rods and cones, even though the genetic lesion targets rod cells. The 17-kb human opsin transgene in T17M mice includes 4.8 kb upstream of the transcriptional start site, ³⁶ which exceeds the 4.1-kb mouse opsin upstream region sufficient for rod-specific expression. ^{46 47} Thus, T17M opsin expression is likely limited to rods, supporting the bystander model. Second, nontransgenic littermates and BALB/c mice are unaffected by illumination under the conditions used, and two additional mouse models of rhodopsin RP (P23H rhodopsin and noerg-1 mice) are only slightly affected. These results indicate that cone damage depends on the rhodopsin transgene and is allele specific. Third, two phases of degeneration were observed after light exposure. In the first phase, rod function (scotopic data) decreased by 60% to 70% from 0 to 6 hours, with only a slight loss of cone function (photopic data). The functional decrease was not attributed to photoreceptor cell death, which increased only slightly by 6 hours, or to clearing of nuclei from the ONL, which was unchanged. In the second phase, a coordinate loss of cone and remaining rod function was observed from 0.5 to 3 days, accompanied by photoreceptor cell death, ONL thinning, and cone deterioration. These results fit a model in which light exposure results in significant loss of rod function within several hours, with bystander effects accounting for the damage to cones and rods at later times. 

The biphasic progress of damage in the T17M model is emphasized in a logarithmic transform of the functional data, which linearizes exponential components (Fig. 9) . In this analysis, responses from the dark-adapted ERG and ERG recovery assays are plotted together to reveal photoreceptor function up to 3 days, when the minimum observed response was reached. The scotopic plot clearly reveals a biphasic decline in function starting immediately after light exposure. Fitting with a biexponential curve (Fig. 9A)yielded a decay process with a t _1/2 of 2 to 3 hours for the first phase, complete in 12 to 15 hours after exposure, and a t _1/2 of approximately 0.5 days for the second phase, complete in approximately 3 days. This analysis implies that the mechanism of rod loss differs in the two phases. The photopic plot is also biphasic, with an initial upward sloping phase (based solely on a-waves because b-waves were supernormal) and an exponential decay beginning at 6 hours (Fig. 9B) . The similarity of the scotopic and photopic kinetics in the second phase raises the possibility that rod and cone loss after 12 to 15 hours occurs by the same mechanism. It is surprising that light exposure does not immediately abolish rod function. This finding may reflect the time required to accumulate toxic metabolites, alter ion gradients, or amplify signaling events that lead to rod cell dysfunction. Biochemical and histologic examination of illuminated T17M mice during the first few hours after light exposure may identify molecular and cellular processes that precede rod cell death, which has been a longstanding goal in understanding phototoxicity in the retina. 

A striking observation in the analysis of ERG response recovery was the increase in photopic b-wave responses from 1 to 3 hours after light exposure. The a-wave responses during this period were not significantly increased, suggesting that the increased b-wave response was attributed to effects on postreceptoral signaling in the cone-mediated pathway. Supernormal photopic b-wave responses have been observed in the P347L rhodopsin pig and the rhodopsin knockout mouse. ^{37 48} Although such responses have been attributed to defects in retinal development, ⁴⁸ their rapid onset after light exposure in T17M mice suggests that they may also result from a loss of rod function, perhaps because of derepression of cone synaptic signaling. ⁴¹ Further characterization of this model may reveal changes in rod-cone interactions that account for the growth in cone b-wave amplitudes. 

Multiple phases of photoreceptor loss during retinal degeneration have been documented in other animal models of rod-mediated retinitis pigmentosa. In rd mice, rod function is lost within 28 days of birth, but cones are retained for months. In rd10 mice, all rod function is lost by postnatal day 40, whereas cones are detected at 9 months. ¹⁵ In the P347L pig, most rods and some cones die within 6 weeks of birth; the remaining rods and approximately 50% of the cones die between 6 and 12 weeks, and the remaining attenuated cones are lost slowly, with detectable function up to 87 weeks. ^{14 49} The early phase of cone damage in these studies is most comparable with the rapid decline in cone function and morphology within 3 days reported here. We established a clear deterioration of cone morphology after light exposure, including the loss of PNA-stained outer and inner segments, which has been taken as evidence of cone cell death. ^{10 50} However, these criteria do not exclude the possibility that cone cells with greatly attenuated outer and inner segments might still have been present, perhaps contributing to the low but detectable photopic ERG responses at 3 to 5 days in our study. Studies that identify cones by nuclear morphology ²⁶ or other methods are needed to establish whether cone cell death occurs and to reveal additional phases of light-induced cone bystander damage in the T17M mouse. 

The primary advantage of the T17M model over others used to study bystander effects is that damage is induced rapidly and synchronously. Loss of rod cells in rod-initiated animal models of retinal degeneration is unsynchronized and ongoing; therefore, many phases of cell degeneration coexist. More rapid degeneration can be induced by light, but illumination periods of hours to days are typically used to accumulate sufficient damage to detect by histologic or biochemical methods. ^{24 25 51} By contrast, retinal degeneration and cone damage in T17M mice can be induced by 2.5 minutes of illumination, which is critical for studying early events in bystander cone loss. Although cone loss is apparent at 1 month even in unilluminated mice (Fig. 1) , the loss induced by bright light exposure is much more rapid, which should enhance the detection of biochemical or histologic changes that correlate with cone bystander damage. The degenerative loss of cones might be further slowed by dark rearing. We expect the rapid initiation of bystander damage in this model will be valuable for testing therapies that preserve cones in the face of rod loss. The model is likely to be most relevant for understanding cone damage in the human central and peripheral retina, in which cone density is similar to that of the mouse retina and in which cone loss is initially observed in retinitis pigmentosa. It may also be helpful in identifying mechanisms of cone loss in the fovea, a rod-free region of approximately 0.2-mm diameter in the center of the macula that is responsible for central acuity. Rod loss in the surrounding parafoveal region of the macula may initiate cone damage throughout the macula by mechanisms similar to those in the periphery, although the timing and severity of the damage may vary because of differences in the relative density of rods and cones. Similar approaches are possible in other animal models of rod-initiated retinal degeneration that exhibit bright light hypersensitivity, such as the T4R dog or the arrestin-rhodopsin kinase double-knockout mouse. ^{24 25 27}  

Previous studies of mutation-specific light hypersensitivity in animals with rhodopsin retinitis pigmentosa raised the concern that damage might be induced in patients with rhodopsin retinitis pigmentosa during routine funduscopic examination or retinal surgery, each of which exposes the retina to bright light. ^{27 28} These studies acknowledged that the risk is relatively low for some patients with retinal degeneration, which is supported by the modest functional deficits after light exposure in P23H and C110Y rhodopsin mice and the lack of an effect in rd12 mice. For a subset of patients, however, our evidence for light-induced cone damage in T17M mice greatly reinforces the need for precaution in the clinic and in daily life. Indeed, disease progression in persons with certain mutations of rhodopsin retinitis pigmentosa may depend more on brief durations of exposure to intense illumination than on constant exposure to low light levels. Given that cone dysfunction profoundly affects patient quality of life, patients are urged to avoid situations in which even momentary exposure to intense illumination is possible. 

 Supported by an unrestricted grant from Research to Prevent Blindness to the University of Florida Department of Ophthalmology; the Retina Research and Education Fund; the Charlie Mack Overstreet Fund for Retinal Diseases; and National Eye Institute Core Grant EY08571.

 Submitted for publication November 5, 2008; revised December 10 and 19, 2008; accepted March 25, 2009.

 Disclosure: M.P. Krebs, None; D.A. White, None; S. Kaushal, None

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

 

Figure 1.

 

ERG responses in untreated mice showing robust rod and cone function. T17M mice (n = 10) and nontransgenic littermates (n = 8) were dark adapted for 12 hours or more, and ERG analysis was performed under scotopic and photopic conditions. Each panel shows the mean ERG (dark line) superimposed on traces for individual mice (gray lines). Only data from the left eye are shown; simultaneously recorded right eye responses were identical within error. NT, nontransgenic.

Figure 2.

 

ERG responses declined rapidly after light exposure in T17M mice. After illumination of the right eye, mice were dark adapted for 12 hours or more, and ERG analysis under scotopic (A) and photopic (B) conditions was performed. The interval between light exposure and ERG analysis is indicated. Mean ERG responses of 6 to 11 mice are shown.

Figure 3.

 

Time course of the ERG response decline showing coordinate loss of scotopic and photopic function. The ratio of mean a-wave (A) and b-wave (B) amplitudes from illuminated and unilluminated eyes of T17M mice (closed circles) and nontransgenic littermates (open circles), calculated from the ERG data in Figure 2 . Dashed lines indicate the ratio expected if light exposure has no effect, values represent mean ± SEM, and asterisks indicate a significant difference (P < 0.05) between illuminated and unilluminated eyes. n = 6–11.

Figure 4.

 

ERG response recovery during dark adaptation within the first 12 hours after illumination. Compared with nontransgenic mice, which exhibit recovery of rod function showing rod functional loss accompanied by near normal or supernormal cone function in T17M mice. After light exposure in the right eye, T17M or nontransgenic mice were dark adapted for the indicated times before ERG analysis under scotopic (A) and photopic (B) conditions. Mean ERG responses of 3 to 7 mice are shown.

Figure 5.

 

Time course of ERG response recovery. Mean a-wave (A) and b-wave (B) ERG amplitudes ratios were determined for T17M mice (closed circles) and nontransgenic littermates (open circles) by dividing the mean amplitudes in illuminated eyes (Fig. 4)by mean amplitudes from fully dark-adapted, unilluminated eyes (Fig. 1) . Dashed lines indicate the ratio expected if light exposure has no effect, values represent mean ± SEM (n = 3–7).

Figure 6.

 

Other mouse models of retinal degenerative disease show little to no damage 5 days after light exposure. Mean scotopic (A) and photopic (B) ERG responses of 5 to 10 mice are shown.

Figure 7.

 

Photoreceptor cell death and clearance after light exposure. (A–L) Retinal cryosections stained by TUNEL assay (red) and with DAPI were prepared from nontransgenic littermates at 1 day after light exposure (A, B) or T17M mice at 6 hours (C, D), 12 hours (E, F), 1 day (G, H), 3 days (I, J), or 5 days (K, L) after light exposure. Abundant TUNEL⁺ nuclei were evident in the ONL of illuminated eyes at 12 hours and 1 day (F, H) and were rare in nontransgenic littermates (A, B) and unilluminated T17M eyes (C, E, G, I, K). Results are typical of three independent experiments. Scale bar, 50 μm. (M) Quantitation of TUNEL⁺ nuclei in illuminated (solid circles) or control (open circles) eyes showing a peak at 1 day. The nuclear count was normalized to the ONL area. (N) Light exposure caused clearance of photoreceptor nuclei, as determined by ONL thickness in DAPI-stained cryosections. Average ONL thickness values from illuminated eyes were normalized to values from unilluminated eyes and plotted as a function of time after exposure for T17M mice (closed circles) and nontransgenic littermates (open circles). The dashed line represents the expected ONL thickness ratio expected if illumination has no effect. Values shown in (M) and (N) represent mean ± SEM (n = 2–3). (N) Asterisks denote a significant (P < 0.05) difference in exposed and control eyes.

Figure 8.

 

Histologic analysis of cone photoreceptor fate after light exposure in T17M mice. Cone morphology in retinal flatmounts (left) and cryosections (right) was examined by staining with rhodamine-PNA (red). Cryosections were also stained with DAPI (blue) to detect nuclei. Images, obtained by wide-field fluorescence microscopy, correspond to a single focus plane (flatmounts) or a merged, deconvoluted z-stack (cryosections). Unilluminated eyes (A, B) showed well-formed cone outer and inner segments and distinct pedicles (B, arrowhead). Outer and inner segments were disorganized and progressively deteriorated at 6 hours (C, D), 12 hours (E, F), and 1 day (G, H) after light exposure. Pedicles remained intact (D, F, H). At 3 (I, J) and 5 (K, L) days, intact cones could not be identified, and pedicles were disrupted (J, L). PNA staining was substantially decreased in flatmounts at 5 days (K). The gross morphology of inner nuclear and inner plexiform layers was unchanged by light exposure. Retinal layers are indicated for one cryosection. IS + OS, photoreceptor inner and outer segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer. Scale bars: flatmounts, 50 μm; cryosections, 25 μm.

Figure 9.

 

Biphasic photoreceptor loss after light exposure in T17M mice. Functional data from the dark-adapted ERG and ERG response recovery assays (Figs. 2 3)were compiled in a semilogarithmic plot to emphasize kinetic components of light-induced damage. F _+light/F _−light represents the ratio of function measured by either assay with or without light exposure. The degeneration of scotopic (A) and photopic (B) responses is biphasic, with an inflection at approximately 0.25 days (arrows). Circles: dark-adapted ERG. Triangles: ERG response recovery. Open shapes: a-Waves. Closed shapes: b-Waves. F _+light/F _−light for ERG response recovery data was obtained by dividing T17M and nontransgenic amplitude ratios (Fig. 3) . Data were fitted with a biexponential (A) or a monoexponential (B) decay curve.

The authors thank Larry Pinto for the gift of noerg-1 mice, Ken Lake and Andy Neeley for genotyping, and Clay Smith and Al Lewin for comments on the manuscript.