Retinal Light Damage: Structural and Functional Effects of the Antioxidant Glutathione Peroxidase-1

Abstract

purpose. The role of the antioxidant enzyme glutathione peroxidase-1 (GPx1) in protecting the retina against photo-oxidative damage was investigated in GPx1-deficient and wild-type mice.

method. Albino GPx1-deficient and age-matched wild-type mice were examined. Baseline electroretinograms (ERGs) were recorded. Thereafter, mice were exposed to intense light for 12 hours. After a 24-hour recovery in darkness, post–light-insult ERGs were recorded and compared with baseline. Structural effects of light insult were evaluated by retinal histology. Antioxidant expression was investigated by quantitative reverse transcription-PCR (qRT-PCR).

results. Light insult significantly affected ERG responses, with reduced a- and b-wave amplitudes. Structurally, photoreceptor layers were predominantly affected. As expected, GPx1 expression was negligible in GPx1-deficient mice but was upregulated in wild-type mice in response to light insult. Similarly, hemeoxygenase-1 and thioredoxin-1 expression increased significantly in wild-type retinas after light exposure. Catalase, GPx isoforms (GPx2 to -4), peroxiredoxin-6, glutaredoxin-1, and thioredoxin-2 expression was unaffected by GPx1 deficiency and light insult, whereas significant increases in glutaredoxin-2 occurred in non–light-exposed (baseline) GPx1-deficient retinas. Compared with baseline wild-type retinas, lipid peroxidation (TBARS assay), an indicator of oxidative stress, was elevated in baseline GPx1-deficient retinas. Unexpectedly, the light insult induced diminution of retinal function, in terms of ERG amplitude, and structural damage was significantly greater in wild-type than in with GPx1-deficient retinas.

conclusions. The data showing increased oxidative damage in baseline GPx-deficient retina give rise to the hypothesis that increased oxidative stress provides a “preconditioning” environment in which protective mechanisms paradoxically render GPx1-deficient retinas less vulnerable to light-induced oxidative damage. This study identified glutaredoxin-2 as a potential candidate.

Photo-oxidative events are important causative factors underlying light-induced retinal damage. ^{1 2} Light and oxygen are essential for vision, but paradoxically these elements also favor the formation of reactive oxygen species (ROS) that can lead to photochemical damage. ³ Photosensitive molecules in the retina, including the visual pigments, absorb incident light as part of the normal visual process but also generate ROS ¹ as a result of this photochemical event. Consequently, long-chain polyunsaturated acids, particularly docosahexaenoic acid, which is abundant in retinal photoreceptor lipid membranes, are particularly vulnerable to ROS-induced lipid peroxidation. ¹ Because the eye is continuously exposed to ambient light energy, highly efficient retinal defense mechanisms act to protect against photoinduced damage. Antioxidant enzymes, other antioxidants, and neuroprotective agents are known to protect the retina against photo-oxidative damage. ^{2 4} Furthermore, continuous phagocytosis and degradation of retinal photoreceptor outer segment material by the retinal pigment epithelium (RPE) mitigates lipid peroxidative damage. ⁵ However, if the retina is exposed to high-intensity light and/or prolonged continuous light conditions (light insult), oxidative stress develops that triggers retinal photoreceptor apoptosis. ⁴  

Light insult results in increased production of H₂O₂ in the outer retina, ⁶ indicating that ROS accumulation is involved in the development of light-induced retinal damage. The enzymatic antioxidant pathway is of particular importance in abating this oxidative stress. In this pathway, superoxide anion radicals are initially converted to H₂O₂ by superoxide dismutase (SOD), ⁷ then to H₂O by glutathione peroxidase (GPx) and/or catalase. ⁸ Imbalance in this two-step conversion results in the accumulation of H₂O₂, with the consequential formation of highly noxious and reactive hydroxyl radicals through Fenton-type reactions, ⁹ which then initiate rounds of lipid peroxidation. ¹⁰ Immunohistochemical location of SOD, ¹¹ catalase, ¹² and GPx ¹³ to similar regions of the retina, particularly the outer retina and RPE, suggests that the enzymatic antioxidant pathway is involved in the retinal cellular defense against oxidative stress. However, SOD ¹⁴ and GPx ⁸ are now known to consist of multigene families with isoenzymes that have specific molecular, biochemical, and functional characteristics, as well as tissue specific expression patterns. Indeed differential distribution of GPx ² and SOD ¹⁵ isoenzymes throughout the retina, not necessarily including the outer retina, now question the specificity of earlier immunohistochemical findings. ^{16 17} Furthermore, differential expression and regulation of these antioxidant isoenzymes induced by light exposure ^{2 6} suggest that the response of these enzymes in the retina to photo-oxidative stress and their involvement in the enzymatic antioxidant pathway is more complex. The significance of each antioxidant isoenzyme in retinal response to photo-oxidative stress remains to be fully evaluated. 

Cytosolic glutathione peroxidase (cGPx or GPx1) is present in the retina ^{2 16} and has been implicated in the antioxidant response to light insult. ² Although the specific role of GPx1 in protecting the retina from photo-oxidative damage is not known, its involvement in the enzymatic antioxidant pathway is a possible mechanism. Thus, activation of GPx1 in the outer retina may be a crucial factor in the response to photo-oxidative stress, ² mitigating lipid peroxidation and subsequent apoptosis of photoreceptors. The GPx1-deficient mouse ¹⁸ provides a suitable model for examining the role of GPx1 in photo-oxidative damage to the retina, since the specific function of this antioxidant enzyme has been removed via homologous recombination of the gene. We have shown that GPx1 deficiency exacerbates neuropathophysiological events associated with stroke ¹⁹ and head trauma. ²⁰ In the present study the effect of light insult on retinal function and structure was investigated in GPx1-deficient and wild-type mice. On the basis of the role of GPx1 in the enzymatic antioxidant pathway, it was hypothesized that GPx1-deficient mice would be more susceptible to light-induced oxidative damage, reflected in post-insult structural and functional defects. 

Materials and Methods

GPx1-Deficient Mouse

In this study, we used GPx1-deficient mice (Gpx1 ^−/−) of an albinoid background strain (SV129xBALBc/CF1) that have been described by us previously. ¹⁸ These GPx1-deficient mice are histologically and functionally normal, but are particularly susceptible to oxidative stress. ¹⁸ Wild-type (Gpx1 ^+/+) mice of the same mixed genetic background were generated via Mendelian segregation of heterozygous GPx1-deficient matings and maintained as a separate line. All mice were maintained under normal animal house conditions: 12-hour light–dark cycle (maximum light intensity <100 lux), standard murine chow, and water ad libitum. 

Experimental Design and Light Insult

Age-matched (8–14 weeks) male wild-type (n = 33) and GPx1-deficient (n = 35) mice were randomly assigned to light damage or baseline control (non–light damage) groups. All experiments were approved by the Monash Animal Ethics Committee of Monash University and adhere to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

The effects of acute light insult were investigated in retinas of wild-type (n = 16) and GPx1-deficient (n = 15) mice. After baseline retinal function evaluation, the mice were exposed to bright-light (4000 lux) insult for 12 hours. The light insult was delivered at the same time of day (1900–700 hours) in all experiments, to avoid the diurnal effects that are known to influence the degree of light damage. ²¹ During the bright-light exposure, the animals were housed in glass boxes (two per box) within a light chamber, which was coated with highly reflective white paint and illuminated by two 40-W cool white fluorescent tubes mounted 40 cm above the animal boxes. After the bright-light period, the mice were returned to darkness and allowed to recover for 24 hours, then post–light-insult retinal function, morphology, and antioxidant gene expression were investigated. Some mice (n = 4 each for wild-type and GPx1-deficient groups) were allowed to recover for 5 days after light insult, to enable structural effects of the light insult to manifest fully ^{2 22} before morphologic evaluation. 

Retinal function, morphology, and antioxidant gene expression were also investigated in wild-type (n = 12) and GPx1-deficient (n = 13) mice that were not exposed to the light insult and served as the baseline control. Baseline retinal lipids were also measured in retinas from wild-type (n = 5) and GPx1-deficient (n = 7) mice that were not exposed to the light insult. 

Functional Evaluation by Electroretinography

Retinal function was assessed by electroretinogram (ERG). After overnight dark adaptation, the mice were anesthetized by intraperitoneal injection of 100 mg/kg ketamine and 5 mg/kg xylazine, according to Calderone et al. ²³ to minimize irreversible cataract formation. ERGs were recorded from the left eye, after pupil dilation (1% tropicamide and 1.25% phenylephrine HCl), with a chloridized silver wire electrode coated in a 1% methylcellulose solution and located on the central cornea together with mouth reference and tail ground electrodes. Signals were amplified (gain, 8000; band-pass, 1–1000 Hz; PAI-2 physiological amplifier; Pasal Electronics, Warrandyte, Victoria, Australia) and digitized at 4 kHz (DT2801; Data Translation, Marlboro, MA). Light-flash stimuli (0.7 ms duration) produced by a photoflash (60 CT-4; Mecablitz, Metz, Germany) were delivered via a custom-built Ganzfeld dome. A silicon photocell was used to calibrate the luminance of the flash stimulus. ²⁴ A neutral-density filter was used to adjust the stimulus intensity to 1.2 log cd-s m⁻². Two responses were recorded at a stimulus presentation rate of 90 seconds. Signal averaging, digital filtering, and quantification of the recorded ERG responses were conducted post hoc with a custom software system in a personal computer. Digital filtering (cutoff frequency, 45 Hz) removed the high-frequency oscillatory components from the b-wave to ensure reliable detection of the b-wave peak. Quantification of the ERG a- and b-wave amplitudes were measured from prestimulus signal level to a-wave minimum and from a-wave minimum to b-wave maximum, respectively (Fig. 1a) . The implicit times of the ERG a- and b-waves were measured as the time from stimulus onset to wave peak for each (Fig. 1b) . 

Morphologic Evaluation by Quantitative Histology

Retinal structure was evaluated by histologic analysis. Enucleated eyes were fixed and embedded in paraffin, and 5-μm sections were taken along the vertical meridian through the optic nerve. After hematoxylin and eosin staining, retinal morphology was qualitatively and quantitatively assessed in identical central (superior and inferior, 200–400 μm from the optic nerve head) and peripheral (superior and inferior, 200–400 μm from the ora serrata) retinal regions. ⁴ Photoreceptor cell loss was quantified by counting rows of photoreceptor nuclei (outer nuclear layer). According to Rohrer et al., ⁴ three measurements were made for each retinal region and averaged to give a single value for the region, then the four values were averaged to give an overall retinal value. 

Analysis of Antioxidant Gene Expression by Quantitative Reverse Transcription–Polymerase Chain Reaction

Retinas were rapidly microdissected from enucleated eyes and snap frozen in liquid nitrogen. Total cellular RNA was extracted from individual retinas (Nucleospin RNA II Extraction Kit; Macherey-Nagel, Düren, Germany) and used to prepare cDNA by reverse transcription using random hexamers and reverse transcriptase (Superscript II; Invitrogen, Carlsbad, CA). Real-time PCR of prepared cDNA were examined over 50 cycles on a sequence detector (model 7500; Applied Biosystems [ABI], Foster City, CA) using either Platinum quantitative PCR supermix UDG (GPx1 to -4; Invitrogen) or TaqMan Fast Universal PCR Master Mix (for all other genes investigated; ABI), VIC or FAM-labeled primer/probe combinations, and ROX dye (ABI). A rodent 18S primer/probe (ABI) was used as a loading reference. Antioxidant gene primers and probes used are given in Table 1 . A Primer Express program was used to create the primers and particular care was taken to ensure that the primers spanned an intron. Data were analyzed with the sequence-detector system software (ver. 1.7; ABI) and gene expression was normalized to 18S rRNA. 

Quantitative Analysis of Retinal Lipid Peroxides: TBARS Assay

Retinas were rapidly microdissected from enucleated eyes and snap frozen in liquid nitrogen. Individual retinas were homogenized in 70 μL 1.15% KCl, after which 100 μL of each of the following was added: 8.1% SDS, 20% acetic acid (buffered to pH 5.3), and 0.82% thiobarbituric acid (TBA). TBA was freshly prepared as a 0.82% solution in 50 mM NaOH. The reaction mixture was incubated at 95°C for 45 minutes according to Hegde et al. ²⁵ Samples were cooled to room temperature and centrifuged at 10,000 rpm for 5 minutes. TBA reactive substances (TBARS) were monitored in the supernatant spectrophotometrically at 532 nm (UVettes; Eppendorf, Fremont, CA). 1,1,3,3-Tetraethoxypropane (Sigma-Aldrich) was used as an external standard ²⁶ and the lipid peroxide level expressed in terms of nanomoles malondialdehyde (MDA) per mg wet weight, according to Ohkawa et al. ²⁷  

Statistics

Results are expressed as mean ± SEM. ERG data were normally distributed, and thus differences in group means were assessed by unpaired t-test with Welch correction when required, or ANOVA and post hoc planned comparisons with the Bonferroni test. Differences in gene expression and quantitative histology were assessed by Kruskal-Wallis (K-W) statistic and post hoc planned comparisons when appropriate. P < 0.05 is considered significant. TBARS data was normally distributed so differences in group-means were assessed by unpaired t-test. 

Results

Retinal Function

ERG responses from GPx1-deficient mice were similar to wild-type mice (Figs. 1a 1b)in baseline recordings. However, analysis of baseline ERG parameters demonstrated a small but significant reduction in mean a-wave amplitude (P < 0.02) and an increase in mean a-wave implicit time (P < 0.002) for the GPx1-deficient compared with the wild-type mice (Figs. 1c 1d) . Amplitude and implicit time of b-wave showed no significant difference between the two groups (Figs. 1e 1f) . 

At 24 hours after light insult, retinal function was investigated under conditions of full retinal dark adaptation. Representative ERG responses are shown in Figures 2a and 2bfor wild-type and GPx1-deficient mice, respectively. In analysis of a- and b-wave amplitudes of the ERG responses (Figs. 2c 2d) , a marked reduction of both was observed in all mice compared with baseline levels. Unexpectedly, the magnitude of the diminution in retinal function, in terms of both a- and b-wave ERG amplitude parameters, was significantly greater in the wild-type mice than in the GPx1-deficient mice. The difference between baseline and post–light-damage ERG a- and b-wave implicit times for wild-type and GPx-1 deficient mice were not statistically significant (Figs. 2e 2f) . 

Retinal Morphology

Evaluation of baseline retinal morphology showed that GPx1-deficient mice had structurally normal retinas (Fig. 3d)compared with wild-type mice (Fig. 3a) . However, the outer segments (p) of GPx1-deficient retinas were shorter than that of wild-type retinas (Figs. 3d 3a , respectively). Quantitatively, there was no significant difference in the number of photoreceptor cells in the outer retinal layers of GPx1-deficient compared with wild-type retinas (Fig. 4) . 

Structural disturbances were observed in the retina after the light insult (Fig. 3) . At 24 hours after light insult, all mice displayed some degree of disorganization of the outer segment (p) layer. The most pronounced structural alteration was a shortening of the outer segments, which was evident only in the wild-type retinas (Fig. 3b) . At 5 days after light insult, marked degeneration of the outer retina was found in wild-type mice (Fig. 3c) , with virtually complete loss of photoreceptor outer segment structures and reduction in the number of photoreceptor cell bodies in the outer nuclear layer (on). Cell body loss ranged from moderate (five to seven rows of nuclei remaining) to extreme (two to three rows of nuclei remaining; Fig. 3c ). In contrast, apart from some disorganization of the outer segments, no obvious change was seen in the retina of GPx1-deficient mice at 5 days after light insult (Fig. 3f) . Quantitative histology (Fig. 4)confirms these qualitative observations. In the outer nuclear layer of wild-type retinas, there was a small but significant reduction in the number of cell rows at 24 hours after light insult, while at 5 days after insult, there was a significant loss of cell bodies, compared with baseline. In contrast, no significant changes in the number of cell bodies were observed in the outer nuclear layer of GPx1-deficient retinas after light insult. 

Retinal Antioxidant Gene Expression

Because numerous antioxidants are affected by photopic injury ²⁸ and alterations of the antioxidant pathway, ²⁹ we investigated the expression of a range of antioxidants that might be altered by light damage and the lack of GPx1. First, the expression of four selenium-dependent GPx isoenzymes was evaluated to determine their relative contributions to GPx function in retinal tissue. Baseline studies demonstrated that all isoforms (GPx1 to -4) were expressed in murine retinas (Fig. 5) . As expected, GPx1 expression in the retina of GPx1-deficient mice was significantly lower (P = 0.0003) than GPx1 levels in wild-type retinas (Fig. 5a) . The approximate 15% residual GPx1 expression in GPx1-deficient retinas most likely reflects background and/or nonspecific primer amplification, since in a prior study we had been unable to detect GPx1 mRNA by Northern blot analysis in all organs investigated, ¹⁸ consistent with homozygous knockout of the GPx1 gene. Mean levels of GPx1, GPx2, and GPx4 expression were similar, whereas GPx3 was expressed at significantly lower levels (P < 0.001) in wild-type retinas. Furthermore, no significant difference was seen in the expression of GPx2, GPx3, and GPx4 between wild-type and GPx1-deficient retinas. 

Analysis of the GPx isoform expression after light insult (Fig. 5)demonstrated a significant upregulation of GPx1 only in wild-type mouse retinas (Fig. 5a ; P = 0.04). There was no change in the expression of the other GPx2 to -4 isoforms in wild-type or GPx1-deficient retinas in response to the light insult (Figs. 5b 5c 5d) . 

Two additional peroxidases—namely, catalase known to remove hydrogen peroxide exclusively ¹² —and peroxiredoxin-6, a recently identified retinal peroxidase, ³⁰ were investigated. Both catalase and peroxiredoxin-6 gene expression (Figs. 6a 6b , respectively) were unaffected by the lack of GPx1 at baseline (P > 0.05), nor by light damage in wild-type or GPx1-deficient retinas (P > 0.05 respectively in both cases). 

The expression of hemeoxygenase-1, a known stress-response gene, ³¹ was unaffected by the lack of GPx1 at baseline but significantly upregulated (∼12-fold; P < 0.001) in response to light damage in wild-type mice (Fig. 6c) . Similarly, GPx1-deficient retinas showed an approximately four-fold upregulation of HO-1 expression compared with GPx1-deficient retinas at baseline (P < 0.01). Because the upregulation in GPx1-deficient retinas was not as extensive as that in wild-type retinas, there was a significant difference in HO-1 expression between wild-type and GPx1-deficient retinas after light damage (P < 0.01 versus light-damaged, wild-type retinas). 

The gene expression of four members of the thiol oxidoreductase family of enzymes was investigated, because differential expression of family members has been observed previously in retinal tissue after light exposure. ³² The expression of glutaredoxin-1 (Grx1; Fig. 7a ), thioredoxin-1 (Trx1; Fig. 7c ), and thioredoxin-2 (Trx2; Fig. 7d ) in GPx1-deficient retinas was not significantly different from wild-type retinas at baseline, unlike glutaredoxin-2 (Grx2; Fig. 7b ), the only family member to show an approximate threefold increase (P < 0.01) in expression at baseline. Trx1 expression increased significantly (P < 0.05) after light damage in wild-type retinas, whereas expression levels for Grx1, Grx2, and Trx2 were not significantly different in wild-type retinas after light exposure. Light damage did not significantly alter the expression of three of the four family members—namely, Grx1, Grx2, and Trx2, in GPx1-deficient retinas—whereas Grx2 expression was significantly reduced in light-damaged GPx1-deficient retinas compared with GPx1-deficient retinas at baseline (P < 0.05). 

Retinal Lipid Peroxides

Basal (non–light insult) levels of MDA were assessed in wild-type and GPx1-deficient retinas. Retinal MDA levels (nanomoles MDA/milligram wet weight) were significantly elevated (∼2.7-fold; P < 0.0001) in GPx-1-deficient retinas compared with wild-type retinas (Fig. 8) , suggesting that the lack of GPx1 increases the extent of oxidative lipid damage in GPx1-deficient retinas at baseline. 

Discussion

The GPx1-deficient mouse provides an ideal model for examining the role of the enzymatic antioxidant pathway in regulating intracellular ROS. ^{18 29} The GPx1-deficient mouse has been shown to be unaffected by the lack of GPx1 under normal physiological conditions, but highly susceptible to oxidative stress ¹⁸ and was used in this present study to investigate the role of GPx1 in the retinal response to photo-oxidative stress. In this study, the GPx1-deficient mouse demonstrated normal retinal structure consistent with earlier studies in which other organs in this knockout mouse were also shown to be histologically normal. ¹⁸ However, alterations in retinal function were observed in baseline ERG studies of the GPx1-deficient mouse. Compared with wild-type mice, a small but significant reduction in amplitude and delayed implicit time were found in the ERG a-wave, suggesting that photoreceptor function may be compromised in these mice. It is also interesting to note that the outer segments (p) of GPx1-deficient mice were shorter than that observed in wild-type retinas (Figs. 3a 3d) . These changes may explain the smaller a-wave observed in these mice in the baseline studies. The presence of shorter outer segments suggests that the lack of GPx1 may trigger adaptive changes in the photostasis mechanism or that the photostasis mechanism may be defective in these animals. ³³ Our data are supported by a previous study by Stone et al., ³⁴ in which rats with reduced GPx, induced by deficient dietary selenium, also had altered retinal function characterized by reduced ERG amplitudes compared with dietary normal animals. Our finding that lipid oxidative damage is increased in the GPx1-deficient retina, most likely because of the dysfunctional enzymatic antioxidant pathway, may have contributed to the observed ERG a-wave changes in this study. This notion is supported by the data of Stone et al., ³⁴ who also demonstrated increased lipid peroxides in the lipid-rich photoreceptor outer segments and RPE in their selenium-deficient rats. 

The enzymatic antioxidant pathway in the retina has been implicated in protecting photoreceptors from light damage in studies of mice with a mutated SOD gene. ³⁵ Previous studies with the GPx1-deficient mouse model have demonstrated that perturbation of the antioxidant enzyme pathway, favoring hydrogen peroxide build-up, has pathophysiological consequences. ^{18 29} Upregulation of the GPx1 gene after light insult in wild-type retinas observed in this study and supported by the results of others ² indicates that this antioxidant enzyme is responsive to light-induced photo-oxidative stress. However, the lack of GPx1 did not appear to affect the physiological response of the retina to light insult adversely, since exposure to high-intensity light had a significantly greater impact on the retinas of wild-type mice than on GPx1-deficient retinas. Indeed, the reduction in retinal function, in terms of ERG amplitude, was significantly greater in wild-type mice at 24 hours after light insult. Consistent with these functional effects, retinal structural changes were more marked in wild-type retinas. Indeed, shortening of the outer segments, consistent with photostasis, ³³ was only observed in wild-type retinas and not in GPx1-deficient retinas, at 24 hours after light insult. At 5 days after light insult, there was significant loss of outer nuclear cell bodies in the wild-type retina, consistent with photo-oxidative stress–induced apoptosis of photoreceptors, ^{36 37 38} whereas in GPx1-deficient retinas the number of photoreceptors was unaffected by identical light insult. Although unexpected, these results are supported by those of an earlier study of retinal light damage in rats with suppressed GPx activity induced by dietary selenium deficiency, in which deterioration of the ERG response after light insult was much less than in identically light-affected dietary normal rats. ³⁴  

Our data suggest that the retina of GPx1-deficient mice may be protected or desensitized from the damaging effects of the light insult. A similar protection of retinal cells from light-induced damage has been demonstrated in animals that have been “preconditioned” before light insult, by exposure to relatively bright cyclic light conditions. ^{22 39} The mechanisms involved in this preconditioning effect are not known, but are thought to involve many aspects of retinal cell physiology. ³⁹ Increased oxidative stress in the GPx1-deficient condition may provide such “preconditioning. ” Indeed, our data show that lipids are oxidatively modified in baseline GPx1-deficient retinas, suggesting that lack of GPx1 facilitates increased oxidative stress in retinal tissue before light insult, that in turn may lead to such “preconditioning” (also known as an “adaptive response”). The adaptive response, which refers to the ability of cells or organisms to better resist the effects of a greater toxic agent when first pre-exposed to a lower dose, ⁴⁰ is known to occur in several different organisms and cell-types. ⁴¹ Increased oxidative stress within retinal tissue of GPx1-deficient mice might activate such adaptive processes due to the chronic nature of the sustained oxidative stress. Light insult, in contrast, is an acute (12-hour), intense, photo-oxidative stress, possibly better tolerated in adapted GPx1-deficient retinal cells that are protected against such extreme stimuli. 

Chronic dysfunctional antioxidant enzyme mechanisms, due to insufficient GPx1 and the associated increased oxidative stress, may provide the “preconditioning lesion” that activates antioxidant enzymes, other nonenzymatic antioxidants, and/or endogenous neuroprotective agents as the adaptive response to compensate for the lack of GPx1. Hydrogen peroxide is known to induce activation of transcription factors such as NF-κB ⁴² and/or AP1, ⁴³ and we have recently shown that NF-κB activation occurs in fibroblasts, in vitro, derived from GPx1-deficient mice. ⁴⁴ Thus, in the GPx1-deficient retina, H₂O₂-induced activation of these transcription factors could in turn activate a range of protective stress response genes, including other antioxidant enzymes, ^{45 46} which could paradoxically render the retina less vulnerable to acute light-induced damage. 

Retinal expression of the four known selenium-dependent GPx isoenzymes was investigated, to evaluate their relative contributions to GPx function in retinal tissue and to determine whether the other isoforms of GPx contribute to protection of the retina from light insult. Indeed, all four GPx isoforms were expressed in wild-type murine retinas, in agreement with previous studies showing expression of GPx1, ^{2 16} GPx3, ⁴⁷ and GPx4 ⁴⁸ in mammalian retinal tissue. The expression of gastrointestinal glutathione peroxidase (GI-GPx or GPx2) in the retina, particularly at levels comparable to GPx1 and GPx4, is unexpected since GPx2 expression is thought to be restricted to the colon in rodents. ⁴⁹ Therefore, the findings of the present study suggest that GPx2 may be more widely distributed in rodents than originally determined by Northern blot analysis of tissue-specific GPx2 expression. ⁴⁹ Plasma glutathione peroxidase (pGPx or GPx3) shares many functional properties with GPx1, ⁵⁰ but in the present study, retinal GPx3 expression was significantly lower than the other GPx isoforms, consistent with reports that GPx3 is predominantly expressed in ciliary epithelium in the mammalian eye. ⁴⁷ Phospholipid hydroperoxide glutathione peroxidase (PHGPx or GPx4), which acts on lipid hydroperoxides, ⁸ is thought to be primarily associated with the lipid-rich photoreceptor outer segments of the retina, and high basal levels of GPx4, as suggested by Wang et al., ⁴⁸ may be an important factor in protecting against the subsequent light damage observed in this study. 

Upregulation of one or more of the other GPx isoenzymes in the retina could compensate for GPx1-deficiency in this mouse model. However, the findings of this study show that expression of the GPx2 to -4 isoenzymes at baseline were unaffected by GPx1 gene knockout, consistent with findings in other tissues. ⁵¹ Furthermore, the isoforms were unchanged in response to the light insult, indicating that these isoenzymes do not compensate for the lack of GPx1 in this mouse model. In addition, in wild-type retinas no significant change in expression of the GPx2 to -4 isoforms was observed after light insult, suggesting that these isoenzymes are not involved in the acute response of the retina to photo-oxidative stress. Since the expression of the other selenoglutathione peroxidase isoenzymes (GPx2 to -4) is unchanged in the GPx1-deficient retina, the expression of a different peroxidase may be induced to compensate for the GPx1 deficiency. ⁵¹ In addition, we investigated whether the recently described non–selenium-containing glutathione peroxidase, peroxiredoxin 6 (Prx6), which has been identified in mammalian eye ³⁰ and in rodent retina, ¹⁶ is a possible candidate. Our studies now indicate that Prx6 does not compensate for the lack of GPx1, at the level of transcription at least, in GPx1-deficient retina and is therefore an unlikely candidate. 

Other components of the enzymatic antioxidant pathway may compensate for the lack of GPx1. Studies of mouse lens demonstrate that catalase activity provides antioxidative protection against H₂O₂ induced stress in lens epithelium in GPx1-deficient conditions. ⁵² Similarly, in the rat lens, an increase in catalase tends to compensate for the marked decrease in GPx in conditions of diabetes-induced oxidative stress. ⁵³ Furthermore, catalase activity in the RPE has been shown to be an important factor in mitigating ROS generated in the outer retina ⁵ and been shown to mitigate oxidative damage and improve retinal function in postischemic retinas. ⁵⁴ However, the data from this study have now shown that catalase, at the mRNA level at least, is unaffected by light damage in wild-type retinas, and more important, that catalase is unaffected by the lack of GPx1 in this model. Therefore, it is less likely that catalase plays a significant role in the preconditioning of GPx1-deficient retina. 

A recent study of light damage induced changes in gene expression in the murine retina indicates that complex cellular events are initiated by photo-oxidation. ²⁸ Numerous genes are upregulated in response to light insult, with the balance of pro- and antiapoptotic gene product activities determining survival of retinal cells. ²⁸ In particular, several antioxidants expressed in the retina are known to be increased after photopic injury, including thioredoxin, ³² ceroplasmin, ⁵⁵ metallothionein, ³⁷ and hemeoxygenase-1 (HO-1). ⁵⁶ In agreement with previously published data, our study suggests an important role for both hemeoxygenase-1 and thioredoxin-1 in response to oxidant-mediated light injury, because we observe an approximately 12- and 4-fold increase in gene expression respectively after light exposure in wild-type retinas. Of importance, the upregulation of HO-1 was attenuated in GPx1-deficient retinas after light exposure, whereas that of Trx1 was unchanged by light damage, highly suggestive that oxidative stress is lessened in light damaged GPx1-deficient retinas, possibly due to a “preconditioning” event. Perturbation of the expression of any of these antioxidant genes due to GPx1-deficiency could potentially influence the proapoptotic antiapoptotic balance and in this manner “precondition” retinal cells, to improve survivability after light insult. Initially, we suspected that the expression of HO-1 may be of particular significance in this regard, since HO-1 has been found to be upregulated in a compensatory response in selenium deficient rat liver, ⁵⁷ although this may be not be directly caused by the associated GPx-deficiency. ⁵⁸ Furthermore, HO-1 is also thought to play a role in the ability of retinal glial cells to protect photoreceptors from oxidative damage. ³¹ However, as this study now shows, a compensatory upregulation of HO-1 expression in GPx-1 deficient retinas before light damage was not observed, making this antioxidant less likely as a “preconditioning” candidate in GPx1-deficient retinas. 

This study has shown a significant upregulation in the expression of the mitochondrial isoform of glutaredoxin (Grx), namely glutaredoxin-2 (Grx2), in GPx1-deficient retinas before light exposure. Grx2 is a protein belonging to the thioredoxin superfamily and known to be important in the regulation of mitochondrial redox status and the regulation of cell death via apoptosis. ^{59 60} Upregulation of Grx2 in response to the lack of GPx1 could lead to “preconditioning” so that retinal cells are afforded greater protection against light damage, because overexpression of Grx2 has been shown to inhibit cytochrome c release and caspase activation, both important events in the protection against apoptosis. ⁶⁰ Furthermore, previous studies have confirmed that Grx is upregulated in response to increased H₂O₂. ^{61 62} It is therefore conceivable that the lack of GPx1 with resultant increases in H₂O₂ within retinal mitochondria, may upregulate Grx2 which in turn “precondition” retinal cells to improve survivability after light insult. However, it is still possible that other antioxidants not explored in this study, contribute to the adaptive response in GPx1-deficient retinas. 

In summary, contrary to expectation, the retinas of GPx1-deficient mice were not more adversely affected by light insult than were wild-type retinas. Unexpectedly, GPx1-deficient retinas appeared to be protected to some extent from the functional and structural effects of light insult. Given the necessity to control for photo-oxidative stress in the general physiological functioning of the retina ³ and the numerous antioxidative mechanisms available to achieve this, ^{2 4 28} it is conceivable that GPx1-deficiency could be compensated for by “adaptive responses” to the chronic oxidative stress of the GPx1-deficient retina, resulting in desensitization to subsequent light insult. One possible candidate identified by this study is Grx2, a mitochondrial redox-sensitive antioxidant known to attenuate apoptosis. In addition, it is possible that ambient light, together with a defective antioxidant system, may provide the chronic oxidative stress in GPx1-deficient retinas, leading to the adaptive response. Dark-rearing of animals from birth could potentially remove this oxidative stress in the retinas of GPx1-deficient animals and circumvent the adaptive response. This hypothesis is currently under investigation. 

 Submitted for publication July 24, 2005; revised December 15, 2005, and January 18, 2006; accepted March 29, 2006.

 Disclosure: A.D. Gosbell, None; N. Stefanovic, None; L.L. Scurr, None; J. Pete, None; I. Kola, Merck & Co., Inc. (E); I. Favilla, None; J.B. de Haan, 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.

 

Representative baseline ERGs from (a) wild-type and (b) GPx1-deficient mice. (a, double-headed arrows) a- and b-wave amplitude measurements; (b, horizontal arrows) implicit time measurements. These ERG parameters were measured at baseline in all wild-type (n = 28) and GPx1-deficient mice (n = 28). Comparison between wild-type and GPx1-deficient measurements was by unpaired t-test, with the Welch correction if required: (c) a-wave amplitude, *P = 0.02; (d) a-wave implicit time, **P = 0.002; (e) b-wave amplitude, P = ns; (f) b-wave implicit time, P = ns.

Table 1.

 

Primers and Probes Used in Quantitative RT-PCR Analysis of Retinal Antioxidant Gene Expression

Gene	Sense Primer	Antisense Primer	Probe
GPx1	CTC ACC CGC TCT TTA CCT TCC T	ACA CCG GAG ACC AAA TGA TGT ACT	ACC CCA CTG CGC TCA TGA CCG A
GPx2	GTG GCG TCA CTC TGA GGA ACA	CAG TTC TCC TGA TGT CCG AAC TG	CCT GGT AGT TCT CGG CTT CCC TTG CA
GPx3	CAT ACC GGT TAT GCG CTG GTA	CCT GCC GCC TCA TGT AAG AC	CAC CGG ACC ACA GTC AGC AAC GTC
GPx4	TGA GGC AAA ACT GAC GTA AAC TAC A	GCT CCT GCC TCC CAA ACT G	TGG TTT ACG AAT CCT GGC CTT CCC CT
Catalase	TTC AGA AGA AAG CGG TCA AGA AT	GAT GCG GGC CCC ATA GTC	CAC TGA CGT CCA CCC
HO-1	AGA TGA CAC CTG AGG TCA AGC A	TTG TGT TCC TCT GTC AGC ATC AC	CTA AGA CCG CCT TCC
Prx6	CCT GAA GAG GAA GCC AAA CAA	CGG AGG TAT TTC TTG CCA GAT G	TCC CTA AAG GAG TCT TCA C
Grx1	CCC TTC CCA CTC CTG CAT T	GGA GGT TGA GGC TGA GAA CAC T	ACT GCC CTT ACT TAG C
Grx2	TTT GTC AAT GGA CGA TTT ATT GGA	GCA GCA ATT TCC CTT CTT TGT G	CGC ACG GAC ACT C
Trx1	TGC AGA GGG CCA AAG TTC A	TGG AAC TGG AGG AAC AAG TAG CT	TTC TCA GCA TCC ATA CGG
Trx2	CAG CCT CTG GCA CAT TTC CT	GTT CGG CTT CTG GTT TCC TTT	CCT GCC TCT GCT TGA

Figure 2.

 

Representative ERG responses from light-damaged mice: (a) wild-type and (b) GPx1-deficient mouse at baseline and 1 day after light insult. ERG parameters were measured in wild-type (n = 16) and GPx1-deficient (n = 15) mice at baseline (before light insult) and 1 day after light insult, so that the effect of light damage on retinal function could be quantified: (c) a-wave amplitude; (d) b-wave amplitude; (e) a-wave implicit time; (f) b-wave implicit time. ANOVA reveals significant differences in group means for a-wave (P < 0.0001) and b-wave (P < 0.0001) amplitudes. Post hoc comparisons: *P < 0.001 baseline versus light-damaged amplitudes for wild-type and GPx1-deficient; #P < 0.001 comparing light damaged wild-type versus light damaged GPx1-deficient mice. Differences in a- and b-wave implicit times were not statistically significant.

Figure 3.

 

Representative retinal histology. (a) Wild-type baseline; (b) wild-type 1 day after light insult; (c) wild-type 5 days after light insult (extensive loss of outer nuclear cell bodies shown in this case); (d) GPx1-deficient baseline; (e) GPx1-deficient 1 day after light insult; (f) GPx1-deficient 5 days after light insult. (Retinal layers: rpe, retinal pigment epithelium; p, photoreceptor; on, outer nuclear; op, outer plexiform; in, inner nuclear; ip, inner plexiform; gc, ganglion cells; nf, nerve fiber.) All sections stained with hematoxylin and eosin.

Figure 4.

 

Quantification of retinal morphology by counting number of rows of nuclei in the outer nuclear layer from retinas of wild-type and GPx1-deficient mice at baseline and 24 hours and 5 days after light insult, thereby enabling light-damage–initiated photoreceptor cell death to be assessed. Kruskal-Wallis statistic = 16.07, P = 0.007. Significant post hoc comparisons: *wild-type baseline versus wild-type 1 day after light insult, P < 0.05; **wild-type baseline versus wild-type 5 days after light insult, P < 0.01 (the number of retinas quantified from each group are shown).

Figure 5.

 

Retinal GPx isoform gene expression for baseline wild-type (n = 8) and GPx1-deficient (n = 7) mice, and 1 day after light insult wild-type (n = 8) and GPx1-deficient (n = 8) mice. (a) GPx1 isoform expression. Kruskal-Wallis statistic = 23.10, P < 0.0001; significant post hoc comparisons: #P = 0.0003 comparing baseline levels wild-type versus GPx1-deficient, **P = 0.04 baseline versus light–damaged levels for wild-type; (b) GPx2 isoform expression; Kruskal-Wallis statistic = 1.20, P = ns; (c) GPx3 isoform expression; Kruskal-Wallis statistic = 6.70, P = ns; (d) GPx4 isoform expression; Kruskal-Wallis statistic = 2.50, P = ns.

Figure 6.

 

Retinal gene expression for (a) catalase; (b) peroxyredoxin-6 (Prx6); and (c) hemeoxygenase-1 (HO-1) in baseline wild-type and GPx1-deficient mice, and 1 day after light insult in wild-type and GPx1-deficient mice. n = 4 retinas/group. Analysis by Kruskal-Wallis ANOVA; significant post hoc comparisons: ***P < 0.001 versus baseline wild-type retinas, **P < 0.01 versus baseline GPx1-deficient retinas, ##P < 0.01 light-damaged GPx1-deficient retinas versus light-damaged wild-type retinas. No significant differences were detected for any other groups. Bars, mean ± SEM.

Figure 7.

 

Retinal gene expression for (a) glutaredoxin-1 (Grx1); (b) glutaredoxin-2 (Grx2); (c) thioredoxin-1 (Trx1), and (d) thioredoxin-2 (Trx2) in baseline wild-type and GPx1-deficient mice, and 1 day after light insult in wild-type and GPx1-deficient mice. n = 4 retinas/group. Analysis by Kruskal-Wallis ANOVA; significant post hoc comparisons: ##P < 0.01 versus baseline wild-type retinas. *P < 0.05 versus baseline GPx1-deficient in (b) and baseline wild-type retinas in (c). No significant differences were detected for any other groups. Bar, mean ± SEM.

Figure 8.

 

Retinal lipid peroxides in wild-type (n = 10) and GPx1-deficient retinas (n = 14). Concentration was determined by the TBARS assay, which measures the amount of TBA reactivity, with MDA formed during acid hydrolysis of lipid peroxides. Data are mean ± SEM. Analysis reveals significantly increased levels of lipid peroxides in GPx1-deficient retinas. ***P < 0.0001.

The authors thank Algis Vingrys (Department of Optometry and Vision Sciences, University of Melbourne) for his kind assistance in calibrating the photoflash stimulator used for murine electroretinography.