DNA Damage and Repair in Light-Induced Photoreceptor Degeneration

Abstract

purpose. Intense light causes photoreceptor death that is greatest in the superior central retina. Short-duration treatment in a light-damage model results in TUNEL-positive photoreceptor nuclei within this region. However, cells lost 10 days after light treatment are fewer than the TUNEL-labeled cells observed earlier. Therefore, this study was undertaken to monitor DNA fragmentation and cell death to explain the discrepancy.

methods. Eyes of dark-adapted rats were light damaged for 4 or 5 hours. DNA fragmentation was measured by TUNEL, laddering, and highly repetitive short interspersed nuclear element (SINE) analysis in dark-adapted, nondamaged control (dark-control) retinas and in retinas collected at 6-hour intervals after light treatment. TUNEL-positive photoreceptor nuclei were counted in these samples along a superior-to-inferior meridian and compared with control and damaged 10-day retinas. Monocytes and DNA polymerase β were monitored by immunohistochemistry.

results. TUNEL-positive staining of photoreceptors was centered around the superior central retina. At 10 days, photoreceptor loss had occurred in this region. In graphs of 6-hour-interval data, two DNA-fragmentation peaks, 24 hours apart, were evident. Monocytes appeared after nuclear damage. Total TUNEL-positive cells under both peaks exceeded the number of photoreceptors lost. The DNA-repair enzyme, polymerase β, was induced in the superior central retina, within photoreceptor inner segments, 24 hours after light treatment, but declined thereafter.

conclusions. One population of damaged cells may mend DNA until the repair mechanism is exceeded and then revert to apoptosis, or, alternatively, two populations may undergo DNA fragmentation 24 hours apart. Either DNA fragmentation is masked at midpoint by temporary repair, or two waves of damage occur, but repair rescues the first set, not the second. Photoreceptors lost are fewer than TUNEL-positive cells. Thus, both possibilities suggest photoreceptor DNA repair. The transient appearance of DNA polymerase β in photoreceptors under these experimental conditions further suggests nuclear repair. Thus, maintenance of in-house DNA-repair mechanisms may provide an alternate approach for the rescue of photoreceptors, as well as other neurons with stress-induced damage. These events may provide useful drug targets to promote photoreceptor survival in various forms of retinal degeneration.

Photoreceptors transduce light to electrical signals that subsequently are processed through retinal synapses. However, excessive light triggers apoptosis. ¹ Light-induced photoreceptor cell death is mediated through rhodopsin, ^{2 3 4} but varies according to the intensity, ^{5 6} duration, ⁷ wavelength of light treatment, ³ body temperature, ^{8 9} light history, and/or prior dark-adaptation interval. ^{10 11 12} Species differ in their response to damaging light, ¹³ and age, ^{6 14} genetics, ^{13 15 16 17} and diet ^{18 19 20 21} can affect the degree of cell loss. Finally, the retinal response varies throughout the day, ²² and can be minimized by limiting the rate of rhodopsin regeneration. ^{4 23 24}  

Noell ^{25 26} recognized that photoreceptor and retinal pigment epithelial (RPE) cell damage is one type of light injury, whereas the other type of injury results only in photoreceptor cell loss. Noell ² first suggested that rhodopsin is the mediator of damage, and Williams and Howell ³ matched the light-damage action spectrum with the green-sensitive rod rhodopsin absorbance curve. Because of this, green-filtered fluorescent light is used in one light-damage model. ^{1 22} Most light-damage models used today rely on the emission spectrum of cool white fluorescent lights, placed either above the animals ^{10 13 18} or surrounding them. ^{27 28} However, xenon illumination and fiber optics have been used when specific wavelengths were required. ^{3 29}  

A light-damage action spectrum in the near ultraviolet has also been described. ³⁰ Although the rat retina is much more susceptible to this short-wavelength light, it shows remarkable morphologic similarities between these two classes of injury. ^{29 31} Recently, Wu et al. ³² used blue light to damage photoreceptors. They demonstrated by electron microscopy that rat rod nuclei undergo apoptosis and that production of 180- to 200-bp fragments and their multimers by internucleosomal cleavage corresponds to peak TUNEL labeling at 8 to 16 hours after light treatment. Blue light enhances light damage by green light. Moreover, blue light has no effect on retinas when rhodopsin is absent, ¹⁷ whereas, with inhibition of photoreversal, damage is less. ^{4 33} Absorption of blue wavelengths by an intermediate of rhodopsin bleaching photoconverts rhodopsin back to its unbleached state and increases the photon-catch capacity, making photoreceptors even more susceptible to further damage. ¹⁷  

Intense light initiates damage and photoreceptor death. The extent of rhodopsin bleaching, its subsequent regeneration, and visual transduction proteins ³⁴ determine the degree of injury. Morphologic changes appear throughout the photoreceptors. Disc vesiculation occurs, mitochondria swell, and the cytoplasm becomes dense. ^{7 35} Finally, nuclear pyknosis occurs, signifying inevitable DNA loss and death. ³⁶ DNA is fragmented, and photoreceptors break down. 

Some photoreceptors may be able to recover from moderate injury. Moriya et al. ⁷ showed that the morphology of cell injury was reversible if the duration of damaging light did not exceed 12 hours, and Specht et al., ³⁷ based on quantification of retinal DNA, reported a biphasic mechanism of rod cell damage with 10% loss after 24-hour exposure and 40% loss during the following 14 days. Other studies by Remé et al. ³⁵ also indicate photoreceptor damage and loss in superior temporal rat retina and damage with recovery in inferior nasal retina. 

These studies suggest repair in photoreceptors, but little is known about this mechanism. It appears that bright light and rhodopsin bleaching trigger a repair response, but physical damage of the DNA molecule may be responsible for its initiation. If DNA is being mended, repair enzymes must be induced at some point. In the current study, a biphasic time course of light-triggered photoreceptor DNA fragmentation was noted, and the DNA-repair enzyme, DNA polymerase β (DNA pol β), was upregulated just before the first wave of nuclear damage. 

Materials and Methods

Animals

Albino (Sprague-Dawley) rats (male, 150–175 g), obtained from Charles River (Raleigh, NC), were acclimated for 1 week (12 hours dark:12 hours dim overhead fluorescent light cycle; 0.12 μE/s · m² [7.1 lux] within cages; onset 6:00 AM) and then placed in darkness for 48 hours before light treatment. Rats were light stimulated for 4 or 5 hours and then returned to darkness for up to 10 days. Animals were fed rat chow and given water, ad libitum. 

Treatment of the animals followed a protocol that was approved by the LSU Institutional Animal Care and Use Committee and was in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Guiding Principles in the Care and Use of Animals of the National Institutes of Health (Publication 86-23). 

Light Source

An 8-light array of 10-in. circular fluorescent lights (22-W; Cool White, FC8T9/CW; General Electric, Fairfield, CT; 270 μE/s · m²; 18 klux) formed the light stimulator. Dark-adapted rats were placed within 5-in.-diameter plastic tubes divided into two chambers containing one animal each, which were then placed within the fluorescent light array for 4 or 5 hours. Experiments were always begun between 8:00 and 10:00 AM. A fan blew room air through the chamber, keeping animals at room temperature throughout light stimulation. Before light treatment, pupils were dilated with tropicamide (Mydriacyl; Alcon, Fort Worth, TX). One group of rats was kept in darkness and served as the control. 

Histology and TUNEL Reaction

After light stimulation, animals were killed by CO₂ asphyxiation at 6-hour intervals for 5 days or after 10 days, and retinas were collected for TdT-mediated fluorescein-12-dUTP nick-end labeling (TUNEL). Corneas were slit and eyes kept overnight in 10% neutral buffered formalin (Mallinckrodt, Paris, KY) at 4°C. Eyes were oriented by the posterior ciliary arteries visible on the back of the globe ³⁵ and hemisected through the optic nerve along the superior-inferior meridian. Lens halves were removed and the superior corneas notched to facilitate orientation of sections. ³⁸ After an additional hour in fixative, tissue was dehydrated through ethanol to xylene and then paraffin. Ten-micrometer-thick sections were affixed to glass slides and immunohistochemistry or the TUNEL reaction was performed with a TUNEL kit (Promega, Madison, WI), as described in the kit protocol. Propidium iodide was used as a nuclear counterstain. Thus, FITC-labeled, TUNEL-positive nuclei appeared yellow, whereas all other nuclei were stained red when sections were viewed by fluorescence microscopy. 

To determine the extent of photoreceptor cell loss after light treatment and DNA fragmentation, some animals were held for 10 days to ensure the complete loss of cells with DNA fragmentation. Tissue was fixed in glutaraldehyde and formaldehyde, postfixed in OsO₄, and embedded in plastic according to conventional histologic techniques. ³⁹ These half eyes were oriented as stated earlier. One-micrometer-thick sections were cut and stained with toluidine blue and viewed by light microscopy. The thickness (number of cells) of the outer nuclear layer (ONL) across each retina was recorded and compared with dark-adapted, untreated retinas. 

Immunohistochemistry

Sections from these time courses were also studied by immunohistochemistry to determine the presence of macrophages and DNA-associated enzymes. Mouse anti-rat ED1 monoclonal antibody (Serotec, Oxford, UK) was used for macrophages, mouse anti-rat monoclonal proliferating cell nuclear antigen (PCNA or cyclin) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) for cell proliferation, and mouse anti-rat monoclonal DNA polymerase β (pol β) antibody (Alpha Diagnostics International, San Antonio, TX) for DNA repair. Deparaffinization and rehydration of sections for ED1 labeling and brief formalin fixation of 20-μm-thick frozen sections for PCNA and DNA pol β labeling were followed by antigen retrieval, ⁴⁰ quenching of endogenous peroxidase activity (for PCNA and DNA pol β), blocking, and incubation with primary antibody (ED1, 1:50; PCNA and DNA pol β, 1:10) for 36 hours at 21°C. ED1-treated sections were reacted with sheep anti-mouse IgG-FITC (1:200; Amersham, Piscataway, NJ) for 24 hours, coverslipped in anti-fade solution (Vector Laboratories, Burlingame, CA), and examined by fluorescence microscopy. PCNA- and DNA pol β-treated sections were incubated with biotinylated goat anti-mouse IgG (1:100, 1 hour), followed by avidin-biotin complexing (Vectastain ABC Elite kit PK-6102, as per directions; Vector Laboratories, Burlingame, CA), and visualized with the diaminobenzidine (DAB) reaction (Peroxidase Substrate kit, SK-4100, as per directions; Vector Laboratories) by conventional bright-field microscopy. 

Western Blot Detection of DNA Pol β and PCNA

Retinal samples were homogenized at 4°C in sterile phosphate-buffered saline (pH 7.2; Life Technologies, Gaithersburg, MD) that contained broad-spectrum serine, cystine, and calpain protease inhibitors (78410 Halt Cocktail; Pierce, Rockford, IL). After proteins in each sample were quantitated by a Bradford microassay, 20 μg was separated on Trisbase-glycine (TG)-SDS mini-gels and blotted onto nitrocellulose membranes, as previously described. ⁴¹ Blots were probed with mouse monoclonal anti-DNA pol β (DNPB11-M; Alpha Diagnostics International) ⁴² or anti-PCNA (cyclin; PC10, sc-56; Santa Cruz Biotechnology) ⁴³ primary antibodies. Bound primary antibodies were detected by an anti-mouse IgG peroxidase-linked secondary antibody and were developed with a Western blot analysis system according to the manufacturer’s instructions (RPN2132, ECL Plus; Amersham, Arlington Heights, IL). TG-SDS gel bands for DNA pol β and PCNA (cyclin) were detected at 39 and 36 kDa, respectively. 

DNA Laddering

The apoptotic DNA ladder kit from Roche Molecular Biochemicals (Indianapolis, IN) was used as directed. Briefly, retinas were isolated and cells lysed in binding buffer. DNA was collected on glass fibers with chaotropic salts, rinsed, and eluted into buffer. Samples were aliquoted and adjusted to equal DNA (4–5 μg total DNA per sample), as determined by nucleic acid quantification (Nucleic Acid Quicksticks; Clontech, Palo Alto, CA), and run on 1% agarose gels. Low-DNA-mass ladders (10068-013; Life Technologies, Rockville, MD) were run in adjacent lanes. DNA ladders were visualized under UV light and the 180- to 200-bp bands quantitated. 

Analysis of Rat Retinal Low-Molecular-Weight DNA Fragments by using Rat SINE-Specific Primers

Total retinal DNA was isolated and purified on silica gel membranes (69504; Qiagen, Valencia, CA), and a DNA-primed polymerase chain reaction (PCR) was performed using 1 μg total DNA per sample. The following rat-specific highly repetitive short interspersed nuclear element (SINE) repeat primers were used: rat SINE ID sense 5′-GGC TGG GGA TTT AG-3′ and rat SINE antisense 5′-TTC GGA GCT GGG GA-3′. PCR conditions were 94°C for 3 minutes (initial melt), followed by 15 cycles of 94°C for 10 seconds (PCR cycle melt), 52°C for 30 seconds (anneal), and 72°C for 3 minutes (amplification), followed by a final extension of 94°C for 7 minutes. Forty-five microliters amplified DNA reaction mixture was run on a standard 1.3% agarose TBE gel (Tris-borate, 89 mM; EDTA, 2 mM; pH 8.0) and analyzed after staining 3 hours with 1 μg/mL ethidium bromide. Signal strengths were quantitated by a molecular imager (GS-250; Bio-Rad, Hercules, CA) and the DNA data signal-analysis package provided with the imager. 

Cell Counting

Whole sections of eye, including cornea and retina from the superior margin, through the optic nerve, to the inferior edge, were used to determine the number of TUNEL-positive nuclei or surviving cells within the ONL. Occasionally, the optic nerve was removed as the block face was trimmed to produce a flat, complete eye section. In these cases, sections from the region just adjacent to the optic nerve head show an uninterrupted band of nuclei across the ONL. Orientation was verified by a notched superior cornea in all sections. Sequential fields along the retina were photographed at 400× and saved as digital images, and the number of fluorescing nuclei counted every 150 μm. These measurements were then plotted to provide retinal DNA-fragmentation profiles. Similarly, the numbers of photoreceptor nuclei were counted in plastic sections of 10-day posttreatment retinas. These profiles represent the thickness (number of nuclei) of the ONL along the retina. The superior margin was always oriented to the left, the inferior margin to the right. 

Three-dimensional plots were generated from retinal profiles obtained at 6-hour intervals. Each profile was generated as has been described, then stacked in sequence along a time axis with commercial software (S-plus software; MathSoft/Statistical Sciences, Seattle, WA) to produce three-dimensional surface plots with color draping. This software allowed tilting and rotation to best observe retinal regions and times of DNA fragmentation. 

Bleb Quantification

The same paraffin sections used to generate the TUNEL-positive photoreceptor nuclei counts were used to quantify nuclear blebbing throughout the light-damage time course. The numbers of fluorescing blebs were counted every 150 μm along the retinas from the superior to the inferior margins. Blebs were counted along a vertical line oriented to the ONL thickness and centered on each field. All field counts for each retina were added to derive the total for that time point. 

Results

Photoreceptor Nuclei in Light-Damaged Retinas

The thickness of the ONL was observed from the superior retinal margin and at the optic nerve head, through the optic disc, to the inferior edge in unstimulated eyes. Except for regions adjacent to retinal margins, the ONL was consistently 11 to 12 nuclei thick in control animals (Fig. 1A) . The distribution of photoreceptor nuclei (ONL thickness) across the dark-control retinas of is shown in Figure 2 . 

Ten days after light treatment, photoreceptor nuclei had disappeared, leaving few ONL nuclei within the superior central, sensitive region of the retina (Fig. 1B) . Most damaged photoreceptor material (nuclei and inner and outer segments) was gone. Nuclear loss from the ONL after 4 and 5 hours of light treatment was illustrated in nuclear-thickness plots, showing a greater sensitivity to light in the superior central retina (Fig. 2) . 

Retinal TUNEL-Positive Response

Throughout these time courses, only photoreceptor nuclei demonstrated DNA fragmentation, with the exception of occasional RPE cells, which became TUNEL-positive 12 to 24 hours after the first peak of photoreceptor cell labeling. TUNEL-positive nuclei appeared in different forms throughout our time course. Initially, nuclei appeared to be slightly fluorescent with a brighter peripheral ring and a dimmed center. Later, centers filled and nuclei became uniformly brighter, finally becoming so bright that images of adjacent cells fused to form a yellow mass unless the illumination intensity was decreased (Fig. 1C) . Near the end of the process of nuclear degradation, small blebs appeared and greatly increased in amount as the number of labeled nuclei declined. Finally, only massive numbers of these fluorescent blebs remained. These first appeared as beadlike strands extended toward the RPE (Fig. 1D) , but later appeared as small assemblages among the outer segments (Fig. 1E) or within the debris zone (Figs. 1G 1H) . Occasionally, blebs were observed within large cells among the outer segments (Fig. 1F) . Blebbing occurred from approximately 24 to 96 hours after light treatment, peaking from 54 to 78 hours (Fig. 3) . The maximum number of blebs recorded from a single retina was 563 at 60 hours after light treatment; the maximum number of blebs recorded in a single frame was 22. 

TUNEL Time Course

The numbers of TUNEL-positive photoreceptor cell nuclei were counted from the superior margin of the retina, through the optic nerve, to the inferior margin in each retina in each 19-point time course and normalized to the greatest value. Figures 4A and 4B illustrate the TUNEL labeling in two different time courses, each after 5 hours of light. Two peaks of labeling occurred approximately 36 hours apart. The level of labeling between the two high points was low, but never completely disappeared. Additional plots were constructed to show the spatial distribution of labeled nuclei in 50-nm steps along each retina. These 6-hour-interval plots were then stacked to produce a three-dimensional plot for each 19-point time course (Fig. 1K) . This plot illustrates a highly labeled area that appeared in the superior-central, sensitive region of the retina. In this region, DNA fragmentation was found throughout the 12-cell-thick ONL, whereas the inferior retina showed little fragmentation. Rotation of this graph to allow viewing across time revealed two distinct peaks of TUNEL-positive photoreceptor nuclei at 18 to 24 hours and at 36 to 42 hours (Fig. 1L) . This demonstrates the temporal and spatial patterns of DNA fragmentation approximately 20 hours and 40 hours after light treatment. The region of highest TUNEL labeling corresponded to the region of photoreceptor cell loss. 

DNA Laddering

Retinas from these time courses were also used to look for DNA fragmentation laddering. The density of the 180- to 200-bp band at the bottom of each lane (proportional to the quantity of fragmented DNA) was compared with the same regions in the dark-control lanes and plotted (Fig. 5) . This band was selected for analysis, because it represents the nucleosome-sized fragments characteristic of apoptosis. ^{6 37 44} Two peaks emerged, one at approximately 30 hours, the other at 48 to 60 hours. 

Retinal Low-Molecular-Weight DNA Fragments Detected with Rat SINE-Specific Primers

In this technique, used for the analysis of DNA fragmentation, the highly repetitive SINEs are used to determine the quantity of short DNA fragments. SINE DNA sequences are species specific and are integral genetic markers for nuclear DNA. Therefore, because of the unique repetitive character of rat SINE sequences, mitochondrial DNA fragments are not amplified and measured with this technique, allowing assessment of only nuclear-specific DNA damage. A time course for low-molecular-weight, mononucleosome DNA fragments, suggestive of apoptosis, is shown in Figure 6 . Increases in the relative signal intensity to 5.6-fold and 10.6-fold over the dark control were apparent at 24 and 60 hours, respectively (n = 3, P < 0.05, ANOVA). 

Distribution of ED1-Positive Macrophages within the ONL after Light Treatment

The occurrence and distribution of macrophages were observed in retinal sections from the same eyes that were used for the TUNEL-labeling time courses. There were no ED1-positive macrophages found in the retina until approximately 42 hours after light treatment, when a few cells appeared deep in the photoreceptor layer. These were large, approximately 20 μm in diameter, with centrally located nuclei approximately 10 μm in diameter (Fig. 1I) . Very gradually, the number increased, peaked at approximately 72 hours, and then declined sharply. No immunopositive cells were observed at 84 hours. These ED1-positive cells appeared and accumulated within the photoreceptor debris zone (ONL and inner and outer segments) in the central superior region of the retina, where the greatest TUNEL labeling at 24 and 48 hours and maximal photoreceptor cell loss at 10 days occurred. At 72 hours, these cells reached their highest density at approximately 14 cells per 400× microscope field (Fig. 1I) . Autofluorescent red-orange clumps of photoreceptor outer segments appear within these cells. No ED1-positive cells were observed in the photoreceptor layer near the retinal margins. 

A subset of smaller, flatter, ED1-positive cells with spottily labeled cytoplasm appeared within the choriocapillaris near the RPE cells. These were concentrated near the light-damaged region, but also extended in smaller numbers to the retinal margin (not shown). Similar ED1-positive cells were present near the base of the ciliary bodies, as well as outward toward the iridocorneal angle (not shown). Finally, another subset of large, spherical ED1-positive cells, approximately 20 μm in diameter, appeared singly at regular intervals along the retinal-vitreous border, proximal to the light-damaged region (Fig. 1J) . These had also disappeared by 84 hours after treatment. 

TUNEL-Positive Cells and Final Cell Loss

For two 4-hour light-treatment time courses, retinal sections were selected at the times of the two TUNEL-positive peaks, and the numbers of TUNEL-positive nuclei were counted. These numbers, when summed, were proportional to the total TUNEL-positive photoreceptors for that specific time course. This was compared with total photoreceptors from control retinas to arrive at the percentage of TUNEL-positive nuclei. Counts of photoreceptor nuclei remaining in retinas 10 days after light treatment, when compared with total photoreceptor nuclei in dark-adapted, nondamaged control (dark-control) retinas (n = 38), provided the percentage of photoreceptor nuclei lost. In these two time courses, an average of 54% of photoreceptors became TUNEL positive, but an average of only 31% were lost (Table 1) . 

Occurrence, Distribution, and Localization of Proliferating Cell Nuclear Antigen and DNA Pol β

Western blot analysis of left retinas of dark-control rats and retinas taken at 24, 30, and 45 hours after light treatment (n = 3) revealed a uniform, minimal presence of PCNA. However, DNA pol β increased fourfold over dark-control levels at 24 hours, diminished twofold by 30 hours, and returned to dark levels at 45 hours (Fig. 7A) . When the right retinas of these same animals were examined by immunohistochemistry, no PCNA was detected. DNA pol β was not detected in dark-control retinas, but was localized to photoreceptor inner segments only in the superior central, sensitive region at 24 hours after treatment (Fig. 7B) . Thirty- and 45-hour retinas were immunonegative; negative controls were colorless (images not shown). 

Discussion

Light-Damage Model

The ideal light-damage model would use light with an emission spectrum that spans rhodopsin absorption with sufficient intensity to trigger injury in photoreceptors, but minimize duration of the stimulus to reduce animal stress. However, the damage should only affect a portion of the retina, allowing accurate, consistent assessment of the degree of cell loss. Finally, photoreceptor loss must occur by apoptosis. We considered these characteristics in the development of our light-damage model. 

Cool white fluorescent light has an emission spectrum produced by two phosphors. The output spans the midportion of the visible spectrum, with peaks at approximately 480 and 585 nm (Fig. 8A) , but diminishes rapidly toward 400 and 700 nm. ⁴⁵ Thus, the only wavelengths available to the rat retina during our light treatment were those within the emission spectrum. 

Only 2.7% of rat photoreceptors are cones. ⁴⁶ The remaining cells are green-sensitive 500-nm maximum wavelength (λ_max) rods. ³ Thus, any cone response would be difficult to detect. The action spectrum of retinal light damage matches a 500-nm λ_max rhodopsin-absorbance curve, ³ suggesting that light damage is mediated through the green-sensitive rod photoreceptor. It was rod cell degeneration that we observed. Therefore, only wavelengths from the fluorescent source that are absorbed by the 500-nm λ_max rods can trigger damage. 

The absorbance curve of this rhodopsin (Fig. 8A) was plotted from the Dartnall nomogram. ⁴⁷ A point-by-point multiplication of the rhodopsin curve and the fluorescence-emission curve yielded the rhodopsin absorption curve for that light source. Because rod pigment cannot absorb more light than is actually present, the new absorption curve was plotted so that the peak of absorption (100% at 500 nm) just contacted the light-emission spectrum (Fig. 8A) . Comparison of the areas under both curves demonstrates that the wavelengths responsible for rod photoreceptor light damage constitute only 34% of all wavelengths (measured as lux) emitted from the fluorescent light source. 

Our model of light damage, which subjects rats to a relatively brief 5-hour stimulus, induced only photoreceptor cell death, centered around the superior central sensitive region ³ of the retina, 1.2 mm from the optic nerve. Although DNA damage occurred within 24 hours, nuclear loss from the ONL appeared complete by approximately 36 to 48 hours. Later time points demonstrated little additional cell loss. Thus, our light treatment did not uniformly destroy all photoreceptors, nor did it require extended lengths of time to produce photoreceptor death (Fig. 8B) . However, the time course from initial stimulus to cell disappearance, in our model, allowed adequate time to monitor cellular events as degeneration proceeded in the Sprague-Dawley rat. 

Retinal Region of Highest Sensitivity

Photoreceptors throughout the retina respond to damaging light, ^{1 2 27 31 48 49 50} and those with extensively fragmented DNA undergo apoptosis and removal within 10 days. ^{51 52} In rats exposed to light in circular chambers that present uniform stimulation to the retina, the central portion of the superior retina (the sensitive region ³ ) showed extensive cell loss. We observed a 29% loss of photoreceptors after 4 hours of light treatment and a 56% loss after 5 hours of light. Many studies in rodents have demonstrated that this is the major site of rodent photoreceptor cell loss. ^{2 13 29 53 54 55} We have analyzed whole retinas, but it is likely that the retinal light responses described in this study were derived chiefly from this sensitive region. 

Photoreceptor Death by Apoptosis

After damaging light, apoptosis is initiated, resulting in photoreceptor cell death. ^{1 5 56} Because necrosis can also cause DNA fragmentation, it is not simple to demonstrate death by apoptosis. Several events must be demonstrated, including (1) ultrastructural evidence that nuclear DNA is condensing while the other organelles remain undamaged with intact membranes; cell labeling or disappearance that is not random throughout the tissue, but is confined to a specific subset of cells; and DNA that is broken into mononucleosome-length fragments of 180- to 200-bp multimers. ^{6 37 44}  

Photoreceptor nuclei begin to display DNA condensation within 24 hours after light treatment, ⁵⁷ indicating that the initiation of apoptosis coincides with the first wave of DNA fragmentation. In this study, the TUNEL-labeled retinal cells were exclusively photoreceptors; TUNEL-positive nuclei appeared only within the ONL, primarily within the region where cells were later lost. This suggests a cell-type-specific event and not a regional necrotic response involving many cell types. We have also demonstrated by SINE analysis that DNA fragmentation into short strands was abundant, that DNA-nucleosome fragments were also abundant in the shortest lengths, and that fragmented DNA distributed itself along gels in 180- to 200-bp multimers to form a ladder, suggesting specific, nonrandom generation of DNA fragments. Thus, the results shown in the present study were the consequences of the initiation of light-induced photoreceptor apoptosis. 

Two Waves of DNA Damage

DNA fragmentation is assessed by a variety of methods, including internucleosomal fragmentation that produces 180- to 200-bp DNA fragments or their multimers, ⁵⁸ which are separated by agarose gel electrophoresis. Histologic stains, such as cresyl violet, hematoxylin and eosin, and toluidine blue, also demonstrate changes within nuclei, such as an outer dense ring or unusual density changes that reflect chromatin condensation, ^{59 60} although some nuclei normally exhibit denser regions with these stains. DNA-specific fluorescent dyes (propidium iodide, 4′,6′-diamino-2-phenylindole [DAPI], and the Hoechst stain ⁶¹ ) also identify condensed nuclear areas. Labeling of the terminal nicked 3′-OH ends of DNA breaks also demonstrates DNA damage, ^{62 63} but this method cannot reveal strand lengths to distinguish necrosis from apoptosis. Although this TUNEL technique is a good indicator of damage, DNA would, theoretically, also label at single-strand breaks. Interpretation is limited if additional methods are not used. Electron microscopy demonstrates morphologic changes within the nucleus. ^{60 64} Moreover, induction and activation of certain enzymes can also suggest apoptosis. However, enzymes associated with apoptosis, such as caspases, do not indicate DNA damage. The direct method to determine whether DNA has been fragmented is to demonstrate nonrandom fragmentation and accumulation of 180- to 200-bp fragments. Therefore, we combined detection of TUNEL-labeled photoreceptor nuclei and quantification of retinal DNA by length to assess the nature of light-induced photoreceptor DNA damage. 

We monitored the appearance of damaged photoreceptor DNA through 19 time points and demonstrated that fragmentation occurred in a biphasic manner. Histologic analysis of TUNEL-labeled sections, Southern blot analysis, quantification of DNA ladders, and analysis of SINE repeats showed two waves of DNA fragmentation. 

Cell counts of TUNEL-positive photoreceptor nuclei revealed a second wave of nuclear damage occurring from 36 to 54 hours after light treatment. The technique for labeling the 3′-OH DNA ends of damaged DNA does not distinguish between single- and double-strand breaks, but labeling of double-strand breaks is much more efficient. ⁶⁵ Thus, analysis of TUNEL-positive cells cannot determine whether the two waves of nuclear damage have different strand-break ratios. However, in single- and double-strand break analyses in light-damaged retinas with and without antioxidants present, Specht et al. ³⁷ suggest a two-phase process in which initial damage could be mediated by active oxygen species, implying random single-strand damage, followed by nonrandom enzymatic fragmentation. 

Gel electrophoresis of DNA fragments from the retinal time courses produced ladders. Presence of a 180- to 200-bp band has been considered an indicator of apoptosis. ^{6 37 44} Density analysis of this band indicated a relatively brief initial wave, a decline, and a more prolonged wave of fragmentation approximately 24 hours later in a pattern similar to that shown by TUNEL analysis. However, unlike the TUNEL technique, laddering can only separate DNA fragments (i.e., DNA with double-strand breaks). The TUNEL labeling of retinas suggests that the second wave may only appear to be broader than the first because of the occurrence of numerous nuclear blebs near the end of the cell-death phase. The laddering technique is incapable of separating fragmented nuclear DNA from the same DNA later localized within blebs, and so an artificially elevated trailing edge may follow actual nuclear fragmentation. 

Biphasic nuclear damage was also demonstrated by quantitating the amount of fragmentation within one specific portion of the rat genome. Mammalian genetic material contains unique repetitive DNA elements in nuclear DNA. One type is the highly repetitive SINE repeats, present in at least 10,000 random copies per eukaryotic genome. ^{66 67 68} SINE analysis is very size specific, detecting DNA of only mononucleosome length. This highly sensitive technique, capable of detecting a very small number of breaks, can be used to compare mitochondrial and nuclear DNA damage and repair kinetics. ^{69 70} To assess the integrity of the nuclear genome (the intactness of the DNA double strand), nuclear-specific SINEs were used as primer sites for DNA-based PCR reactions. ^{41 70 71} Rat-specific SINE primers were used for extraction and amplification. Subsequent DNA quantitation demonstrated two peaks of low-molecular-weight, mononucleosome-sized DNA fragments approximately 36 hours apart in the 6-hour-interval retinal time courses. As in the laddering experiments, the second wave of fragmentation was longer than the first. This may be from continued endonuclease activity or from detection of persistent DNA fragments within the blebs. 

Upregulation of DNA-Repair Enzyme Expression

Bright light triggers a sequence of events that can lead to photoreceptor DNA strand breaks, and alterations in the nucleus lead to cell death through apoptosis. However, the presence of antioxidants such as ascorbic acid ^{72 73} and dimethylthiourea ²⁸ diminishes the amount of damage to photoreceptor DNA and preserves retinal morphology. ^{74 75} Finally, the reduction of all-trans retinol dehydrogenase activity by bright-light treatment is oxidative in nature. ²⁸ Conversely, the presence of the nucleoside 8-hydroxydeoxyguanosine, an indicator of the oxidation of 8-hydroxyguanine, is unaffected by intense light. ⁴⁴ Thus, these studies suggest an oxidative component in cell death, but it is not known whether oxidation directly triggers apoptosis. However, once DNA is damaged, repair mechanisms are present to correct the damage. ⁷⁶  

Four major DNA-repair mechanisms have been identified. The process of homologous recombination followed by nonhomologous end-joining repairs double-strand breaks. There are three other mechanisms used in DNA excision repair: base-excision repair, nucleotide-excision repair, and mismatch repair. ⁷⁶ The mismatch-repair mechanism repairs incorrect base pairing from replication or recombination. Nucleotide-excision repair removes large helix-distorting molecules, and enzymes of the base-excision repair mechanism replace single damaged bases that result from hydrolytic damage and endogenous oxidative activity. ⁷⁷  

Base-excision repair in higher eukaryotes can occur by two different mechanisms: one is dependent on DNA pol β, a small DNA-repair enzyme; the other is dependent on PCNA, a cyclin involved in DNA proliferation. ⁷⁸ PCNA is involved in cell proliferation and cell-cycle control. ⁷⁹ It is required for tethering DNA polymerases to the DNA template for DNA synthesis, ⁸⁰ and for some repair, where it interacts specifically with DNA pol δ. ⁷⁸  

It has now been determined that there are at least 130 DNA-repair genes in the human genome: 16 are DNA polymerases (catalytic subunits), and three others possess PCNA-like DNA-damage sensor activity. ⁷⁶ Five of these polymerases (α, β, γ, δ, and ε) have been well characterized. ⁸¹ However, most do not function in the repair of nicked DNA: pol α repairs 20- to 70-nucleotide gaps and is inactive on nicked DNA; pol δ functions during semiconservative DNA replication; pol ε is involved with long-patch repair and replication; and pol γ repairs mitochondrial DNA. Pol β, however, repairs only DNA with very short gaps ⁸² and is upregulated during oxidative stress. ⁸³ Moreover, two nuclear DNA-repair enzymes, apurinic-apyrimidinic endonuclease and pol β, have been shown to increase (340% and 110%, respectively, above the control) in cultured human RPE cells that become apoptotic after exposure to rod outer segments. ⁸⁴ Increases in DNA pol β indicate activation of a repair mechanism. 

Finally, it has been reported that there is no incorporation of [methyl-³H]thymidine or [8-³H]deoxyguanosine into retinal DNA after light damage, ⁴⁴ suggesting to us that any new incorporation of radionuclides into repaired DNA would be minimal and difficult to detect. Thus, we have examined retinas after light damage for the presence of PCNA and DNA pol β, and our immunochemical analyses provided additional evidence for a light-induced DNA-repair mechanism within photoreceptors. 

Dark-control retinas and retinas at 24, 30, and 45 hours after light treatment were selected to span the period of the first DNA-fragmentation wave. PCNA was not detected by whole-retina Western blot analysis or immunohistochemistry, suggesting that cell proliferation had not been activated by light treatment. DNA pol β, however, was detected in retinal sections at 24 hours after light treatment. Because dark-control and 45-hour retinas showed no pol β immunolabeling, this polymerase was induced, followed by a decline back to control levels. Localization was restricted to photoreceptor inner segments, the site of protein synthesis in these cells. Moreover, pol β labeling occurred only in the superior retina, in the sensitive region where the most severe light damage occurs in rodents. ³ This suggests that at 24 hours after light treatment, the nicked DNA-repair enzyme, pol β, has been and/or is being synthesized and readied for transport to the nucleus. This event precedes destruction of nuclei and photoreceptor cell loss. Thus, the appearance of pol β in the sensitive region, within inner segments, at 24 hours suggests activation of intrinsic light-induced photoreceptor DNA repair. 

Significance of Photoreceptor Biphasic DNA Fragmentation

The significance of the biphasic retinal DNA fragmentation is not clear. The second wave may represent a class of macrophages that undergo apoptosis within the ONL. ⁵¹ However, macrophages were not detected in the retina until 42 hours after light treatment. Their appearance peaked at 72 hours, in agreement with other studies, ³⁵ and well past the time course for the second peak of DNA fragmentation. In rat photoreceptor apoptosis induced by administration of N-methyl-N-nitrosourea, TUNEL-positive photoreceptor cells were found 24 hours later, but ED1-positive macrophages peaked at 7 and 21 days. ⁸⁵ This further suggests that macrophages appear to remove cellular debris only after DNA fragmentation and photoreceptor death occur. Thus, the second wave of damage does not result from the presence of macrophages. 

If the TUNEL-positive cells corresponding to the first and second waves of DNA fragmentation are different sets of damaged photoreceptors, then the sum of both waves represents the total number of cells to be lost. This sum should equal the number of missing cells observed 10 days after light treatment. Table 1 compares these numbers, showing that more cells were TUNEL positive than were lost. 

Conversely, both peaks of DNA damage could occur within the same photoreceptors. These two waves were spaced approximately 24 hours apart. During that interval there was almost no DNA fragmentation and no noticeable blebbing, indicating nuclei that were TUNEL positive during the first wave of fragmentation had not yet broken down. Therefore, many TUNEL-positive nuclei observed at the first peak must now be TUNEL negative. 

Both possibilities strongly support the presence of DNA repair within light-damaged photoreceptors. There is evidence for such a mechanism. In studies in which the stimulating light is only at, or slightly above, the threshold amount that induces photoreceptor damage, photoreceptors recover from light damage, generally within a 2-week period. ^{7 29 35 86 87 88} In another form of neuronal injury, after bilateral common carotid artery occlusion in the gerbil, by day 3, after 6 minutes of ischemia, one third of the hippocampal neurons within the CA3 region are brilliantly TUNEL positive, and yet, by day 7, all TUNEL label has gone and there is no decrease in cells in the CA3 region. ⁸⁹ In this instance, neurons containing a very high number of DNA fragments within a highly specific region of the hippocampus appear to be damaged for a single day. Thereafter, there is no evidence of DNA breakage. Because the cell number does not decrease, the fragmented DNA must be repaired. In another study of ischemic mouse and gerbil brains, Tobita et al. ⁹⁰ showed an early increase in single-strand DNA breaks, followed by delayed DNA fragmentation, suggesting early initiation of a DNA-repair mechanism. Our data suggest photoreceptors possess a similar mechanism. 

Finally, there is additional evidence for a biphasic light-induced retinal response. Organisciak et al. ⁹¹ and Specht et al. ³⁷ quantified single- and double-strand DNA damage in rat retina, in the presence or absence of the antioxidant dimethylthiourea after light treatment. They showed by DNA quantification a 10% loss of visual cells after 24 hours of light treatment and a second 40% loss during the following 14 days in darkness, suggesting that some cells could recover from moderate damage. They suggested that damage by activated oxygen species produces the early phase, while the second phase results from apoptosis. 

Conclusions

Although bright light triggered apoptosis in photoreceptor cells, nuclear DNA fragmentation occurred in two waves, approximately 24 hours apart, and was followed by the appearance of monocytes. By 10 days, loss of photoreceptor cells was shown to be less than anticipated, based on DNA-fragmentation studies. Also, expression of the DNA-repair enzyme pol β was upregulated at 24 hours after light treatment, but only within the retinal region of highest sensitivity. We suggest that photoreceptors are endowed with an active DNA-repair mechanism that is partially successful in rescuing damaged cells, and which accounts for the two apparent waves of fragmentation and lower-than-expected cell loss. Thus, maintenance of in-house DNA-repair mechanisms may provide an alternate approach for the rescue of photoreceptors, as well as other neurons with stress-induced damage. These events may provide useful drug targets to promote photoreceptor survival in various forms of retinal degeneration. 

 Supported by National Eye Institute Grant R01 EY05121, the Eye, Ear, Nose, and Throat Foundation, New Orleans, Louisiana; and the Ernest C. and Yvette C. Villere Chair for Retinal Degeneration, New Orleans, Louisiana.

 Submitted for publication September 25, 2001; revised April 12, 2002; accepted June 3, 2002.

 Commercial relationships policy: N.

 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.

 

Retinal response to excessive light. Images (A–J) Superior-central, sensitive region of the retina, 1.2 mm from optic nerve. (A, B) Plastic 1-μm-thick sections showing the effect of damaging light on retina. (A) Photoreceptor nuclei appear as dark blue spheres in stacks of 12 in this dark-control retina. (B) Sensitive region of retina, 10 days after light treatment. Upper arrowhead: Bruch’s membrane; lower arrowhead: base of photoreceptor layer. GC, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; IS, photoreceptor inner segments; ONL, outer nuclear layer; OS, photoreceptor outer segments. (C–H) Fluorescence microscopy of 10-μm-thick paraffin sections of retinas at the indicated times after light treatment. This time course illustrates the fate of TUNEL-positive photoreceptor nuclei as fragmented DNA is first labeled, then blebbed from photoreceptors to become phagocytized by cells within the debris zone (DZ). (C) All retinal nuclei fluoresced red from the nuclear counterstain propidium iodide. TUNEL-positive nuclei, fluorescing green from FITC, appeared yellow because of double labeling. As endonucleases began DNA fragmentation, nuclear margins became TUNEL-positive rings, gradually filling in until affected nuclei were uniform and bright. Finally, blebbing occurred as DNA-containing packets were expelled from the cells (arrowheads 1–4). (D) Blebbed DNA packets often collected in linear strands aligned toward the RPE (arrowheads). (E) Early blebs accumulated among outer segments (green band, top). Some nuclei appeared large because of TUNEL fluorescence intensity. (F) Occasionally, throughout DNA and cellular fragmentation, clumps of TUNEL-positive DNA were found in large spherical cells in the distal photoreceptor layer. (G) Some TUNEL-positive nuclei remained. Undamaged propidium iodide-labeled nuclei were present (arrowheads). Horizontal arrowhead, top right: original position of Bruch’s membrane. (H) Nuclear fragmentation was complete. Arrowhead: Bruch’s membrane. (I, J) Fluorescence microscopy of 10-μm-thick paraffin sections of retina labeled with ED1, an indicator of monocytes and macrophages. (I) ED1-positive retinal cells (green fluorescence, arrowheads) appeared from 42 hours to 78 hours after light treatment. Red-orange clumps were unstained, autofluorescent masses of photoreceptor outer segment fragments. Large, spherical, ED1-positive cells appeared at 78 hours, evenly spaced at 50- to 75-μm intervals, along the vitreal (V) edge (J, arrowheads). Magnification bars, 50 μm. (K, L) Three-dimensional plots of the number of TUNEL-positive nuclei across the retina at 6-hour intervals after light treatment. Each time point is plotted to show the number of labeled nuclei within each ONL nuclear stack at 50-μm intervals from the superior edge of the retina at left, through the optic nerve, to the inferior retinal margin at right. Each microscope field along the retina was 150 μm wide; three equidistant measurements were taken within each field. Position along the retina can be measured from the optic nerve, located at 0 eccentricity, toward the superior or inferior margins. Within the region of highest sensitivity, all 12 nuclei within a stack exhibited damage (vertical axis). The graphs for each time point were stacked through the z-axis to create the three-dimensional array. A 90° rotation of the graph (in K) illustrates two peaks of retinal TUNEL-positive nuclei at approximately 24 hours and 40 hours throughout this 6-hour-interval time course (L). Dark-control (D) and 0 hours at right, 96 hours and 10 days (10d) after light treatment at left.

Figure 2.

 

Number of nuclei present within the ONL (outer nuclear layer) in dark-control retinas (•, n = 38, SE), and in 4 hours (⋄, n = 5, SD) and 5-hour light-treated retinas (○, n = 3, SD) counted 10 days later. The number of photoreceptor cell nuclei per stack (thickness) is plotted every 150 μm on a section along a meridian from the superior edge, through or adjacent to the optic nerve, to the inferior retinal margin for each curve. Dashed line: position of the optic nerve. Arrow: center of the highly sensitive region. When compared with dark-control animals, 28.9% and 55.5% of photoreceptors were lost in the 4- and 5-hour light-treated animals, respectively.

Figure 3.

 

Number of photoreceptor nuclear blebs occurring across an entire retina at each 6-hour time point in a 5-hour light-damage time course. Each time point represents one animal. The maximum number of blebs recorded from a single retina was 563 at 60 hours after light treatment. The maximum number of blebs recorded in a single frame was 22. End of light treatment is indicated as 0 hours. The dark-control (D) is shown at left; the 10-day (10d) posttreatment value is at right.

Figure 4.

 

TUNEL labeling of photoreceptor cell nuclei at 6-hour intervals after light treatment. Total labeled photoreceptor cell nuclei for each time point were counted and plotted. (A) and (B) represent two separate time courses. Although the onset of DNA fragmentation varies, two increases in DNA damage were observed after light treatment: one at 24 to 30 hours and one at 36 to 54 hours. These two peaks occur approximately 36 hours apart. TUNEL-positive nuclei appear up to approximately 60 hours after light treatment, after which no significant nuclear labeling is seen. Dark-control values are indicated at left, 10-day (10d) posttreatment value at right.

Figure 5.

 

Analysis of photoreceptor nuclear DNA fragmentation. DNA laddering was measured in dark-control retinas (D) and at 6-hour intervals after 5 hours of light treatment. The DNA gel ladders are shown at the top with a parallel low-DNA-mass ladder marker in lane M. Marker (bp) fragment sizes are indicated. ( ) Region where the mononucleosome-sized DNA fragments were quantitated. Density measurements of the 180- to 200-bp bands of these ladders at each time point are shown.

Figure 6.

 

Effects of 5-hour light treatment on adult rat retinas shown by rat highly repetitive SINE PCR of total retinal DNA. Rat SINE-specific primers were used with PCR to detect low-molecular-weight and mononucleosome-sized DNA fragments to estimate the amount of DNA fragmentation. One gel is shown at the top. Density measurements for each time point were collected for three complete courses and averaged. Bars indicate SD, n = 3.

Table 1.

 

Analysis of TUNEL Labeling after 4-Hour Light Treatment, Assuming Two Different Sets of Damaged Photoreceptor Cells

	Time Courses			Mean (%)
		1	2
Total TUNEL-positive photoreceptor cells (% of dark controls, n = 38)
TUNEL peak 1	43	22
TUNEL peak 2	20	23
Total TUNEL labeling (% TUNEL peak 1+ 2)
	63	45	54
Total photoreceptor cell loss at 10 days after treatment (% of dark controls)
	29	32	31
Total photoreceptor cells repaired (% total TUNEL labeling minus total cell loss)
	34	13	23

Figure 7.

 

Localization of DNA pol β and PCNA (cyclin) in control and light-treated rat retinas. (A) Relative levels of DNA pol β and PCNA with immunolabeled Western blot analysis of whole retinas: dark-control and at 24, 30, and 45 hours after light treatment. Samples were normalized to make dark-control values equal 100%. (B) Immunohistochemistry of DNA pol β. The superior central sensitive region of the retina, 1.2 mm from the optic nerve in dark-control retinas (left) was compared with the same region in retinas 24 hours after light treatment (middle) and a corresponding inferior retinal region of the same section 24 hours after treatment (right). Arrowheads: intense immunopositive label corresponding to photoreceptor inner segments.

Figure 8.

 

Characteristics of the light-damage model. (A) Graph of emission spectrum of cool white fluorescent light (F, ▵), absorption curve for rat 500-nm λ_max rod rhodopsin (500 rod, ○) and absorption curve for rhodopsin in fluorescent light (•). These bulbs contain two phosphors that peak at 480 and 585 nm, and exhibit five very narrow mercury spikes. Rat rhodopsin absorbs 34% of the available fluorescent light (100%). (B) Light-damage time course. Profiles of the number of photoreceptor nuclei across the retina along a meridian from the superior edge, through the optic nerve, to the inferior edge. This nuclear layer is 12 nuclei deep in healthy rats. Times are hours after light treatment. Only the superior retina was affected, and once cell loss occurred at 36 to 48 hours, no additional damage appeared.

The authors thank Hilary Thompson of the Department of Ophthalmology, LSU Health Science Center, New Orleans, Louisiana, for discussions and statistical analysis of data, and for expert construction of the three-dimensional plots.