Retinal Ganglion Cell Loss in Superoxide Dismutase 1 Deficiency

Abstract

Purpose.: To investigate the influence of deficiency in superoxide dismutase (SOD) 1, a major antioxidative enzyme, on retinal ganglion cells (RGCs).

Methods.: In the SOD1 total knockout (SOD1-deficient) mice, the level of superoxide anion was measured using dihydroethidium. The number of RGCs was counted in both the retinal sections and the flat-mount retinas after retrograde labeling. Thickness of nerve fiber layer (NFL) was measured in the sections, and the amount of neurofilament protein was measured by immunoblot analysis. Pattern electroretinogram (ERG), which reflects the function of retinal ganglion cells, dark-adapted ERG, and cone ERG were performed. The intraocular pressure (IOP) was measured with an induction-impact tonometer. The levels of SOD-1 and -2 were measured by ELISA, in the serum of 47 newly diagnosed consecutive normal tension glaucoma (NTG) patients and 44 consecutive control subjects.

Results.: The level of superoxide anion in the RGC layer was significantly higher in 24-week-old SOD1-deficient mice than in wild-type mice. The RGC number was significantly reduced in 24-week-old SOD1-deficient mice, although they were not in 8-week-old mice. The NFL thickness and neurofilament protein were reduced in 24-week-old SOD1-deficient mice. The amplitude of pattern ERG was significantly reduced, although dark-adapted and cone ERGs showed no impairment, in 24-week-old SOD1-deficient mice. The IOP level was not changed in the SOD1-deficient mice. The serum level of SOD1, but not SOD2, was significantly lower in the NTG patients than in the healthy controls.

Conclusions.: SOD1 deficiency causes RGC vulnerability, which may be involved in the underlying condition of NTG.

Glaucomatous optic neuropathy involves damage to the retinal ganglion cells (RGCs) and their axons. Epidemiologic reports show that an increase in intraocular pressure (IOP) is the greatest risk factor ¹ ; age and family history also contribute to the risk of developing this neuropathy. An increase in IOP generates oxidative stress in the retina, ² and in vitro data support the association of oxidative stress with RGC death. ^{3 –5} The experiments showed that the apoptosis of cultured RGCs is accelerated by reactive oxygen species (ROS) and that the RGCs can be protected by antioxidant substances, such as coenzyme Q10, thioredoxins, and docosahexaenoic acid. Another risk factor for glaucoma, age, is also related to oxidative stress. Living in an oxygenated environment and consuming intracellular oxygen to produce energy causes ROS to accumulate in cells, and this increase in ROS is thought to be closely connected with cellular lifespan. ⁶ Thus, oxidative stress has been seen as an aggravating factor in glaucomatous optic neuropathy. Vulnerability of the RGCs to the oxidative stress may be involved in the underlying condition of the glaucomatous optic neuropathy. 

Here we focused on superoxide dismutase (SOD) 1, an enzyme that metabolizes superoxide anions, a type of ROS. Three kinds of SODs are reported: SOD1 (Cu,Zn-SOD), SOD2 (Mn-SOD), and SOD3 which is also a Cu,Zn-SOD but is encoded by another gene and has a different protein structure than SOD1. ⁷  

In this study we first confirmed that SOD1-deficient mice showed an increased level of oxidative stress in their RGCs, and we evaluated the number and function of the RGCs in these mice. We previously found by full-field electroretinogram (ERG) that the retinal photoreceptor cells degenerate in SOD1-deficient mice that were >40 weeks old. ⁸ Here we used a more sensitive method, pattern ERG, to measure RGC function, and analyzed the influence of SOD1 deficiency in the RGCs of younger mice. Furthermore, we measured the SOD1 protein level in the serum of human normal tension glaucoma (NTG) patients to discuss the association of SOD1 deficiency in glaucomatous optic neuropathy. 

Materials and Methods

Animals

SOD1 knockout mice (purchased from Jackson Laboratory, Bangor, ME) were backcrossed to C57BL/6 and analyzed at different ages. ^8,9 All the animal experiments were performed in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. 

Immunohistochemistry

Cryosections (8 μm) of the retina were tangentially cut through the optic nerve head and ora serrata. The sections were incubated with rabbit anti-Cu/Zn SOD polyclonal antibody (1:100; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) overnight after blocking with a TNB blocking buffer in 0.1% triton for 1 hour, followed by Alexa 488-conjugated goat anti-rabbit IgG (1:300, Invitrogen-Molecular Probes, Eugene, OR). Apoptotic cells were analyzed with a TUNEL kit (Chemicom; Millipore, Billerica, MA). The sections were examined using a microscope equipped with a digital camera (Axio Imager; Carl Zeiss, Oberkochen, Germany). 

Immunoblot Analysis

Isolated retina was placed into lysis buffer with protease inhibitors. Each sample was separated by SDS-PAGE after homogenization and electroblotted onto a polyvinylidene fluoride membrane (Millipore, Billerica, MA). After being blocked in 4% skim milk, the membrane was incubated at 4°C overnight with rabbit anti-Cu/Zn SOD polyclonal antibody (1:1000; Santa Cruz Biotechnology, Inc.), mouse anti-neurofilament H (1:1.000; Cell Signaling Technology, Inc., Danvers, MA), or mouse anti-α-tubulin (1:1.000; Sigma-Aldrich, St. Louis, MO) antibodies. The membrane was then incubated with a biotin-conjugated antibody against mouse immunogloblins. The signals were visualized by chemiluminescence (ECL Blotting Analysis System; GE Health Care Japan, Tokyo, Japan) using a dedicated CCD camera system (LAS-4000 mini; Fuji Film, Tokyo, Japan), measured with commercially available software (ImageJ software, version 1.37; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html.), and normalized to α-tubulin. 

Measurement of ROS

Eyes were enucleated and immediately frozen (OCT compound; Sakura Finetek, Torrance, CA). Unfixed cryosections (10 μm) were incubated with 5 μM dihydroethidium (DHE; Invitrogen-Molecular Probes, Tokyo, Japan) for 20 minutes at 37°C, as previously reported. ^10,11 DHE is used to sensitize intracellular superoxide anion, ^{12 –16} by being converted to the red fluorescent compound 2-hydroxyethidium. Sections were examined using a microscope equipped with a digital camera with a filter (Filter set 43 HE; excitation, BP550/25HE; emission, BP605/70HE; Carl Zeiss) in the same exposure conditions, and the intensity of the staining in the ganglion cell layer was measured (ImageJ) in the retina at two points, one on either side of the optic nerve head, that were 0.2 cm apart. All the procedures in each sample, from preparing animals to taking photographs, were performed at the same time in parallel. 

RGC Count and Layer Thickness Measurement

The RGC number was quantified in two ways. First, the number of cells in the ganglion cell layer was counted in cryosections (8 μm) of the retina cut through the optic nerve head and ora serrata, and stained with hematoxylin and eosin. Second, the number was counted in flat-mount retinas, after retrograde labeling from the superior colliculus with Fluoro-Gold (Fluorochrome, Denver, CO). Seven days after application, the eyes were enucleated, and the retinas were detached to prepare flattened whole-mounts in 0.1 M PBS solution. Four standard areas (0.04 mm²) of each retina 0.1 mm from the optic disc were randomly chosen, and the labeled cells were counted by observers blinded to the identity of the mice. The thickness of the inner nuclear layer (INL) and outer nuclear layer (ONL) was measured in the posterior retina at four points, two on either side of the optic nerve, 0.2 and 0.5 cm away from the optic nerve (ImageJ). The thickness of retinal nerve fiber layer (NFL) was measured at the boundary between the retina and optic nerve (ImageJ). In all cases the retina was examined with a microscope equipped with a digital camera (Carl Zeiss). 

ERG

Mice were anesthetized with pentobarbital sodium at a dose of 70 mg/kg body weight and placed on a heating pad that maintained their body temperature at 35–36°C throughout the experiments. The ground electrode was a subcutaneous needle in the tail, and the reference electrode was placed subcutaneously between the eyes. The active contact lens electrodes (Mayo, Inazawa, Japan) were placed on the cornea, and recording was performed (PowerLab system 2/25; AD Instruments, New South Wales, Australia). A small drop of balanced saline maintained the cornea and lens in excellent condition during the recording. 

A detailed description of the pattern ERG is available elsewhere. ^{17 –19} Briefly, a visual stimulus of contrast-reversing bars (field area, 50°×58°; mean luminance, 50 cd/m²; spatial frequency, 0.05 cycle/deg; contrast 98%; and temporal frequency, 1 Hz) was aligned with the projection of the undilated pupil at a 20-cm distance. Although the eyes were not refracted for the viewing distance, the pinhole pupils of mice permit them to have a large depth of focus and form a proper image on the retina. ¹⁹ The retinal signals were amplified (10,000-fold) and bandpass filtered (1–30 Hz). Three consecutive responses to 600 contrast reversals each were recorded. The responses were superimposed to check them for consistency and then averaged. The amplitude of the pattern ERG was measured from the baseline to the trough of the positive wave. 

Single-flash ERG was performed as previously described. ^20,21 The mice were dark-adapted for 12 hours, and the pupils were dilated with a mixed solution of 0.5% tropicamide and 0.5% phenylephrine (Mydrin-P; Santen, Osaka, Japan) just before the light exposure. Light pulses of 800 cd · s/m² and 4-ms duration were delivered via a commercial Ganzfeld stimulator (Ganzfeld System SG-2002; LKC Technologies, Gaithersburg, MD). 

The cone ERG was measured as previously reported ⁸ with dilated pupils. The uniform stimuli for cone ERG recording consisted of strobe flash stimuli of 20 cd/m² per second superimposed on a steady background light (12 cd/m²) in a Ganzfeld stimulator. ¹⁷  

IOP Measurement

The IOP was measured with an induction-impact tonometer (Tonolab Colonial Medical Supply, Franconia, NH) between 10 AM and noon. ^22,23 Mice were anesthetized with pentobarbital sodium at a dose of 70 mg/kg body weight and placed on a heating pad throughout the experiment. The probe tip was set with the optical axis of the eye at a 1- to 2-mm distance. Five consecutive IOP measurements were averaged. The impact of the probe on the cornea is minimal, so local corneal anesthesia was not necessary. 

Patients

This study followed the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of the Keio University School of Medicine. Written informed consent was obtained from all the subjects before their entry into the study. The inclusion and exclusion criteria for NTG were as in a previous report. ²⁴ Among all the subjects suspected to have primary open-angle glaucoma who came to the glaucoma subspeciality clinic of Keio University Hospital from January 1, 2005, to May 31, 2007, primary open-angle glaucoma was newly diagnosed in 60 patients (28 males, 32 females; mean age ± SD, 59.9 ± 9.8 years). The patients then underwent 24-hour IOP measurements. The IOP measurements were performed 7 times (at 6:00, 9:00 [AM], 12:00, 3:00, 6:00, 9:00 [PM], and 12:00 [AM] as 24-hour IOP measurements), and 13 subjects whose highest IOP was over 21 mm Hg were excluded; therefore, 47 patients were enrolled as NTG (18 males, 29 females; mean age ± SD, 59.5 ± 10.1 years). Forty-four consecutive control subjects were also recruited (16 males, 28 females; mean age ± SD, 62.7 ± 14.8 years). 

Blood Sampling

Blood from the forearm vein of the patients and control subjects was collected in 5-ml tubes before noon after a 12-hr fasting period. The blood samples were then spun at 3000g for 10 minutes and stored at −80°C before analyses. The serum levels of SOD-1 and -2 (Cu, Zn-SOD ELISA system and Mn-SOD ELISA system, NDF Corporation, Tokyo, Japan), fructosamine, total cholesterol, and triglyceride were measured using kits (Roche, Basel, Switzerland). 

Statistical Analysis

In Mice.

For the statistical comparison of two samples, we used the unpaired t-test or Mann-Whitney U test. Data are presented as the mean ± SE. P < 0.05 was regarded as statistically significant. 

In Human Subjects.

Homogeneity of the distributions of the variables between the NTG patient and control groups was analyzed using the t-test, Mann-Whitney U test, or Fisher's exact test, depending on the variables. Data are presented as the mean ± SD. The association between NTG and the serum level of SOD1 was examined using a logistic regression model that contained some confounders as factors. The adjusted odds ratios and 95% CIs were estimated using the logistic regression model. All data were analyzed with commercially available software (SPSS version 18.0; SPSS, Chicago, IL). 

Results

Increased Superoxide Anion Level in the RGCs of SOD1-Deficient Mice

Immunohistochemical labeling of SOD1 showed high levels of expression in the RGC and inner nuclear layer of wild-type (WT) mice (Fig. 1A). SOD1 was localized to the cytoplasm, consistent with a previous report. ²⁵ No immunostaining was observed in the retina of SOD1-deficient (SOD1^−/−) mice (Fig. 1 B). Immunoblot analysis revealed the absence of SOD1 protein in the retina of SOD1-deficient mice (Fig. 1C). The level of ROS in the retinal ganglion cell layer was measured by the fluorescence level of DHE and was significantly higher in 24-week-old SOD1-deficient mice than in WT mice (Fig. 1D–F; P < 0.01). 

Figure 1.

 

SOD1 and superoxide anion in RGCs. Immunostaining of SOD1 in the retina of WT and SOD1-deficient (SOD1^−/−) mice (A, B). SOD1 was localized to the cytoplasm. Hoechst staining representing nuclei shown for a guide to the retinal cell layers (A, B). Immunoblot analysis of SOD1 using whole retinal lysate confirmed absence of SOD1 protein in the retina of SOD1-deficient mice (C). The level of superoxide anion in the RGC layer was analyzed using DHE. Very slight DHE fluorescence was observed in the retina of 24-week-old WT mice (D); however, it was significantly increased in the 24-week-old SOD1-deficient mice (E). The relative fluorescence intensity was determined (F). Scale bar, 50 μm. WT mice, n = 11; SOD1-deficient mice, n = 10. RGC, retinal ganglion cell; DHE, dihydroethidium; WT, wild-type. *P < 0.01.

Reduced Number of RGCs and Thinning of NFL in SOD1-Deficient Mice

The number of RGCs was counted in cross sections of the retina (Fig. 2A–C). Strikingly, the number of cells in the RGC layer was significantly reduced in 24-week-old SOD1-deficient mice compared with WT mice (P < 0.01). We further counted the RGCs in flat-mount retinas using retrograde labeling, which distinguishes RGCs from displaced amacrine cells, ²⁶ since nearly half the cells in the rodent ganglion cell layer are known to be displaced amacrine cells (Fig. 2D–H). ²⁷ By this method also, the number of RGCs in the 24-week-old SOD1-deficient mice was significantly lower than in 24-week-old WT mice (Fig. 2F–H; P < 0.001). The RGC number in the retina of 8-week-old SOD1-deficient mice was the same as in WT mice (Fig. 2D, 2E, 2H), indicating that the reduction progressed with age. 

Figure 2.

 

Reduction of RGCs in SOD1-deficient mice. Histology of the retina of 24-week-old mice showed that the number of cells in the RGC layer was reduced in SOD1-deficient mice (B) compared with WT mice (A). The cells were counted, and the number is shown graphically (C). WT mice, n = 14; SOD1-deficient mice, n = 16. **P < 0.01. Retrogradely labeled RGCs in a flat-mounted retina showed no significant difference in the number of RGCs between WT (D) and SOD1-deficient mice (E) at the age of 8 weeks. However, the RGCs were decreased in the SOD1-deficient mice (G) compared with WT mice (F) at the age of 24 weeks. Retrogradely labeled RGCs were counted (H). Scale bar, 50 μm. Eight-week-old WT mice, n = 18; SOD1-deficient mice, n =16. Twenty-four-week-old WT mice, n = 18; SOD1-deficient mice, n = 20. *P < 0.05; ***P < 0.001. The NFL thickness at the disc margin was significantly decreased in the SOD1-deficient mice (J) compared with WT mice (I) at the age of 24 weeks. The thickness was measured (K). WT mice, n = 13; SOD1-deficient mice, n = 18. **P < 0.01. Immunoblot analysis showed that the level of neurofilament protein, a marker of nerve fibers, in the retina was significantly reduced in the SOD1-deficient mice compared with WT mice at the age of 24 weeks (L, M). WT mice, n = 5; SOD1-deficient mice, n = 6. *P < 0.05. No significant difference was observed in the thickness of the ONL (A, B, N) or INL (A, B, O) between WT and SOD1-deficient mice at the age of 24 weeks. WT mice, n = 14; SOD1-deficient mice, n = 17. RGC, retinal ganglion cell; INL, inner nuclear layer; NFL, nerve fiber layer; ONL, outer nuclear layer.

Next, we evaluated the damage to the RGC axons. The thickness of NFL measured at the disc margin in the cryosections was significantly reduced in 24-week SOD1-deficient mice compared with WT mice (Fig. 2I–K; P < 0.01). Moreover, immunoblot analysis showed that the level of neurofilament protein which is a marker of nerve fibers was reduced in 24-week SOD1-deficient mice compared with WT mice (Fig. 2L, 2M; P < 0.05). 

Since we previously found that SOD1-deficient mice show progressive photoreceptor cell death after 40 weeks of age, ⁸ we examined the thickness of both the ONL, which is composed of photoreceptor cells, and the INL of the 24-week-old SOD1-deficient mice and found no difference at this age (Fig. 2A, 2B, 2N, 2O). Moreover, TUNEL-positive cells were hardly observed in ONL and INL of both the 24-week-old SOD1-deficient and WT mice (data not shown). Therefore, in the retina of the 24-week-old SOD1-deficient mice, RGC reduction was primarily caused and was not a secondary effect of photoreceptor cell death. 

Impairment of RGC Function in SOD1-Deficient Mice

To observe the functional loss of RGCs in SOD1-deficient mice, we performed pattern ERG, which can detect early RGC dysfunction in both humans ¹⁸ and experimental animals, including primates, ²⁸ cats, ²⁹ rats, ^30,31 and mice. ^{17 –19} Importantly, the amplitude of the pattern ERG of 24-week-old SOD1-deficient mice was significantly reduced compared with that of WT mice (Fig. 3A–C; P < 0.0001). We also measured the dark-adapted ERG to examine the functions of rod photoreceptor cells and inner retinal cells (Fig. 3D–F), and the cone ERG to examine cone photoreceptor cells (Fig. 3G, 3H), to find that none of these cell functions were changed in the SOD1-deficient mice at 24 weeks of age. Therefore, the RGCs in SOD1-deficient mice became dysfunctional before the other retinal cells. 

Figure 3.

 

Impairment in RGC function in SOD1-deficient mice. (A–C) Pattern ERG. Representative wave responses from an individual mouse (A) and the superimposed wave responses of either WT or SOD1-deficient mice (B). The amplitude of the pattern ERG of 24-week-old SOD1-deficient mice was significantly lower than that of WT mice (C). WT mice, n = 12; SOD1-deficient mice, n = 15. P < 0.0001. (D–F) Full-field ERG. Representative wave responses from an individual mouse to one flash (D). No significant difference was observed in the a-wave amplitude (E) or b-wave amplitude (F) between 24-week-old WT and SOD1-deficient mice. WT mice, n = 9; SOD1-deficient mice, n = 11. (G, H) Full-field cone ERG. Representative wave responses from an individual mouse in each group. No significant difference was observed between 24-week-old WT and SOD1-deficient mice. WT mice, n = 12; SOD1-deficient mice, n = 15. ERG, electroretinogram; SOD1, superoxide dismutase 1; WT, wild-type.

IOP Measurement in the SOD1-Deficient Mice

To determine the influence of the SOD1 deficiency on the IOP, which might induce RGC death when elevated, we next examined the IOP in 8-, 16-, 20-, and 24-week-old SOD1-deficient mice (Fig. 4). Interestingly, there was no significant difference in the IOP level between the SOD1-deficient and WT mice at any time point. These data indicated that the decrease in RGCs and optic neuropathy in the SOD1-deficient mice were independent of the IOP level. 

Figure 4.

 

IOP measurement. No significant difference was observed in the IOP between WT and SOD1-deficient mice at the age of 8, 16, 20, or 24 weeks. IOP (mm Hg) in WT vs. SOD1-deficient mice: 15.7 ± 2.2 (n = 5) vs. 16.7 ± 1.6 (n = 5), P = 0.56 (8 weeks); 16.4 ± 0.8 (n = 10) vs. 15.5 ± 1.2 (n = 6), P = 0.72 (16 weeks); 16.5 ± 2.7 (n = 9) vs. 16.8 ± 0.93 (n = 8) P = 0.81 (20 weeks); and 15.7 ± 0.7 (n = 14) vs. 14.4 ± 0.6 (n = 17), P = 0.18 (24 weeks). IOP, intraocular pressure; WT, wild-type.

Reduced Serum Level of SOD1 Protein in Human NTG Patients

We next investigated the serum level of SOD1 in human patients with NTG, in which glaucoma occurs with no IOP elevation, and compared it with that of normal healthy controls. Examination of the average mean deviation of the visual field showed a value of −6.6 ± 7.9 dB in the right eye and −6.1 ± 6.9 dB in the left for the NTG patients. The clinical characteristics of the NTG patients and control subjects are shown in Table 1. The age, gender, height, weight, and body mass index, prevalence of hypertension and diabetes mellitus, systolic and diastolic blood pressure, and serum levels of fructosamine, total cholesterol, and triglyceride were not significantly different between the NTG patients and controls. The IOP was higher in the NTG patients than in the controls (right eyes, P = 0.001; left eyes, P = 0.005); however, the level in the NTG patients was still in the normal range. The central corneal thickness in the NTG patients (right eyes, 526.6 ± 29.5 μm; left eyes, 523.9 ± 29.1 μm) was almost the same as in the normal Japanese population. ³²  

Table 1.

 

Demographics of the Normal Tension Glaucoma and Control Groups

Demographics	NTG (n = 47)	Control (n = 44)	P
Age, y	59.5 ± 10.1	62.7 ± 14.9	0.22
Male/female ratio	18/29	16/28	1.00
Height, cm	160.2 ± 8.6	156.5 ± 9.4	0.13
Weight, kg	57.6 ± 9.1	56.0 ± 11.5	0.56
Body mass Index	22.4 ± 2.5	22.8 ± 3.5	0.61
Hypertension, n (%)	9 (19.1)	3 (6.8)	0.12
Diabetes mellitus, n (%)	4 (8.5)	4 (9.1)	1.00
Systolic blood pressure, mm Hg	125.7 ± 16.3	121.5 ± 15.0	0.51
Diastolic blood pressure, mm Hg	75.1 ± 14.1	70.9 ± 13.4	0.45
IOP, mm Hg
Right eye	15.9 ± 2.3	13.7 ± 2.5	0.001
Left eye	15.6 ± 2.3	14.0 ± 2.8	0.005
Fructosamine, μmol/L	253.7 ± 22.2	260.1 ± 32.3	0.27
Cholesterol, mg/dL	223.0 ± 46.9	217.1 ± 38.4	0.51
Triglyceride, mg/dL	149.3 ± 91.6	125.4 ± 68.6	0.17

 Data are expressed as mean ± SD unless marked otherwise. IOP, intraocular pressure; NTG, normal tension glaucoma.

Interestingly, the serum level of SOD1 was significantly lower in the NTG patients than in the healthy controls (Fig. 5, control, 34.2 ± 27.4 ng/mL; NTG, 18.9 ± 15.0 ng/mL; P = 0.002). In contrast, there was no significant difference in SOD2 (control, 146.6 ± 47.8 ng/mL; NTG, 140.2 ± 43.1 ng/mL). To identify correlations between clinical factors by multivariate analysis using logistic regression, we chose independent factors from categorized or continuous variables shown in Table 1. However, using the selected independent variables, no significant correlations were observed, except for between normal tension glaucoma and the serum SOD1 level (P = 0.007; odds ratio 0.95; 95% confidence interval; 0.92–0.99). Therefore, the reduction in serum SOD1 protein level was linked to the development of NTG. 

Figure 5.

 

Reduction in SOD1 protein level in human NTG patients. The SOD1 protein level in the serum of human NTG patients was reduced compared with that of age- and sex-matched controls. Controls, n = 47; NTG patients, n = 44. *P = 0.002. NTG, normal tension glaucoma; SOD1, superoxide dismutase.

Discussion

We demonstrated that the RGC number and NLF thickness were significantly reduced in 24-week-old SOD1-deficient mice compared with WT mice with no difference in IOP. Concomitantly, the amplitude of the pattern ERG, which reflects RGC function, of the SOD1-deficient mice was significantly reduced. Interestingly, the serum SOD1 in human NTG patients was significantly reduced compared with normal healthy controls. 

Previous reports showed that oxidative stress can induce the apoptosis of RGCs. ^{3,4,26,33 –40} Since RGC axonal damage is involved in the glaucomatous optic neuropathy, ⁴⁰ an optic nerve crush model has been used to analyze glaucoma experimentally. In this model, the number of axotomy-induced apoptotic RGCs is reduced in transgenic mice that overexpress SOD1. ⁴¹ Moreover, in the optic nerve transection model, intravitreal injection of pegylated superoxide dismutase, which has a direct effect on RGCs as shown in vitro, ¹⁵ blocks superoxide generation after axotomy and delayed RGC death. ^13,14 These results suggest that RGC axonal damage can be suppressed by artificially overexpressed SOD1. 

Although SOD1 was knocked out in all the retinal cells including RGCs, INL cells, and ONL cells in the SOD1-deficient mice, we demonstrated that obvious changes were detected only in the RGCs at 24 weeks of age, by both histologic (Fig. 2) and functional (Fig. 3) analyses. These changes were not observed at 8 weeks of age, and therefore the loss of RGCs progressed with age. Our results indicated that SOD1 is indispensable for reducing the superoxide anion level generated during physiological cellular activity in RGCs to protect themselves. Given that DHE may also detect other kinds of ROS such as H₂O₂ and ONOO⁻ in addition to superoxide anion, which is reported in a recent paper, ⁴² loss of SOD1 might have induced other kinds of endogenous ROS secondarily to the cellular dysfunction caused by excessive superoxide anion. As shown in our previous report, ROS is also observed throughout the retina of a mouse diabetes model, but apoptosis dominantly occur in RGCs, ¹⁰ supporting the idea that RGCs may have increased susceptibility to ROS. Therefore, as shown in the SOD1-deficient mice of this study, a gradual and progressive loss of RGCs can be induced by endogenous ROS accumulated with age, and the SOD1 deficiency itself causes vulnerability of RGCs. 

Reduction of pattern ERG amplitude were robust; however, that of RGC number and NLF thickness was significant but not very large in the SOD1-deficient mice. It is consistent with the previous report that progressive loss of RGC axons lags behind progressive loss of pattern ERG amplitude by approximately 3 months in DBA2/j mice, which is a genetic model of glaucoma. ^43,44 By 8 months of age, the pattern ERG amplitude has lost approximately 50% of its initial value, although the number of optic nerve axons is still unchanged in DBA2/J mice. The loss of the pattern ERG amplitude is also relatively larger than the estimated loss of retinal NFL thickness measured by optical coherence tomography in human. ⁴⁵ Pattern ERG of SOD1-deficient mice may also have reflected dysfunction of each RGC that is already damaged but still alive. 

We next examined whether the IOP was elevated in the SOD1-deficient mice, since IOP is the greatest risk factor for glaucomatous optic neuropathy, and its reduction suppresses the progression of optic neuropathy in most cases of conventional human glaucoma. ⁴⁶ IOP is, at least in part, determined by the function of the trabecular meshwork cells, which play an important role in draining the aqueous humor from the anterior chamber to Schlemm's canal. ⁴⁷ The trabecular meshwork cells may also be damaged by ROS, ^48,49 as shown in primary open-angle glaucoma patients; in these patients, the trabecular meshwork cells show increased levels of 8-hydroxy-2-deoxyguanosine, a DNA oxidative damage marker. There are statistically significant correlations among DNA oxidative damage of the trabecular meshwork, IOP elevation, and visual field defects due to optic neuropathy. ⁵⁰ Thus, the SOD activity in human trabecular meshwork cells ⁵¹ could be important for IOP regulation. However, in the present study, the SOD1-deficient mice showed no IOP elevation up to 24 weeks of age. Thus, the RGC degeneration progressed independent of the IOP. 

Interestingly, in human NTG patients, the serum SOD1 level was significantly reduced compared with that of normal healthy controls. Decrease in SOD1 protein possibly involved genetic abnormalities in molecules regulating SOD1 transcription, translation, or degradation, as well as in SOD1 itself. Thus, the change in SOD1 protein level could be induced by secondarily to the other genetic background than SOD1 gene mutation. Although further detailed analyses are required, this finding supports our proposal that the SOD1 insufficiency, and resulting RGC vulnerability is involved in the underlying condition of glaucomatous optic neuropathy. In this condition, slight increase in IOP that does not exceed normal range might be sufficient to progress detectable RGC dysfunction and loss and develop NTG. 

Biomarkers associated with glaucoma have been studied for decades. Increased oxidative stress markers and reduced serum antioxidants are associated with glaucomatous optic neuropathy. ^50,52,53 In regard to SOD1 activity, it was reduced in the serum of psuedoexfoliation glaucoma compared with the healthy controls, ⁵⁴ but was increased in the aqueous humor of primary open-angle glaucomas. ⁵³ Thus, the association of SOD1 and glaucoma has been speculated; however, the underlying mechanism has not been proposed. Our present study showing that the SOD1 deficiency induced the RGC damage, but not an increase in IOP at least up to 24 weeks of age, suggesting that the reduction of SOD1 activity in the psuedoexfoliation glaucoma might be associated with the glaucomatous optic neuropathy, and the change in SOD1 in aqueous humor might not associate with neurodegeneration, directly. A reduced vitamin C level is also observed in NTG patients. ²⁴ If vitamin C was consumed to reduce the endogenous ROS that accumulate under conditions of SOD1 insufficiency, this biomarker might also associate with the neuropathy. 

Some systemic diseases such as cardiovascular disease or peripheral vascular disorder are also reported as related factors for NTG. ^{55 –57} These could be confounding factors that are also caused by the increased ROS due to SOD1 deficiency. 

The RGC number falls by about of 0.5% per year in normal humans. ⁵⁸ In the present study, this reduction was accelerated by SOD1 insufficiency in mice. Other phenotypes seen in SOD1-deficient mice include a short lifespan, ⁵⁹ reduced fertility, ⁶⁰ anemia with autoantibodies, ⁶¹ fatty liver, ⁶² hepatocarcinogenesis, ⁵⁹ hearing loss, ⁶³ skin atrophy, ⁶⁴ and retinal dystrophy, ⁸ which are also aging phenotypes. Thus, the SOD1-deficient mice may be considered aging-accelerated mice, one of the aspects of which is NTG. 

In summary, SOD1 is indispensable in RGCs to suppress ROS and protect from RGC death. The vulnerability of RGC by SOD1 deficiency may be involved in the basis of NTG development. SOD1 and ROS may be potential targets of new therapeutic approaches for neuroprotection in NTG. 

Footnotes

 Supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan; from the Tokyo Biochemical Research Foundation (KY and NO); and a Keio University Grant-in-Aid for Encouragement of Young Medical Scientists.

Footnotes

 Disclosure: K. Yuki, None; Y. Ozawa, None; T. Yoshida, None; T. Kurihara, None; M. Hirasawa, None; N. Ozeki, None; D. Shiba, None; K. Noda, None; S. Ishida, None; K. Tsubota, None

The authors thank Haruna Koizumi-Mabuchi, Shunsuke Kubota, and Seiji Miyake for technical assistance.