Use of Mitochondrial Antioxidant Defenses for Rescue of Cells With a Leber Hereditary Optic Neuropathy–Causing Mutation FREE

A G-to-A transition at nucleotide 11778 in mitochondrial DNA (mtDNA)in the gene specifying the reduced form of nicotinamide adenine dinucleotide:ubiquinone oxidoreductase subunit 4 (ND4) of complex I was the first pathogenic point mutation linked to Leber hereditary optic neuropathy (LHON).¹ At present, approximately 45 other mutations in mtDNA have been ascribed to LHON.² Most LHON mutations affect the ND1, ND4, and ND6 complex I subunits in the oxidative phosphorylation pathway, where electrons first enter the electron transport chain.^{3 - 5}

Although reductions in oxidative phosphorylation are invariably present in LHON, cell death appears to be mediated by oxidative stress via apoptotic mechanisms.^{6 - 11} Misdirected electrons from the electron transport chain may react with molecular oxygen,thus increasing generation of reactive oxygen species.^{12 - 15} Increases in reactive oxygen species activity and diminished mitochondrial antioxidant defenses in LHON⁸ suggested to us that treatment of LHON may be possible by bolstering antioxidant defenses locally. In this report, we genetically increased mitochondrial defenses against superoxide to rescue LHON cells homoplasmic for the G11778A mutation in mtDNA because this mutation in the ND4 subunit of complex I is responsible for approximately half of all LHON cases, and visual loss in these patients has the worst prognosis for spontaneous recovery.

We constructed an adeno-associated virus (AAV) vector using the AAV vector plasmid pTR-UF12 regulated by the 381–base pair (bp) cytomegalovirus enhancer immediate early gene enhancer and the 1352-bp chicken β-actin promoter-exon1-intron1 driving expression of the human mitochondrial superoxide dismutase (SOD2) complementary DNA (Figure 1A and B). This plasmid was linked to green fluorescent protein (GFP) via a 637-bp poliovirus internal ribosomal entry site. The SOD2-containing plasmid and the parent pTR-UF12 plasmid were amplified and purified by means of cesium chloride gradient centrifugation and then packaged into AAV-2 capsids by transfection into human 293 cells using standard procedures.¹⁶ Genome titers of the recombinant AAV were determined using real-time polymerase chain reaction and assayed for infectious particles.¹⁷ Each virus preparation contained 10¹¹ to 10¹² vector genome particles/mL and 10⁹ to 10¹⁰ infectious center U/mL.

Homoplasmic 143B osteosarcoma cells (cybrids) containing 100% mutated (11778A) mtDNA were grown in Dulbecco modified eagle medium (Fisher Scientific,Hampton, NH) supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin streptomycin (Sigma-Aldrich Corp, St Louis, Mo) at 37°C with 5% carbon dioxide. The cybrids were created by fusion of enucleated cells from patients with mutated mtDNA, in this case the G11778A mutation, with osteosarcoma (143B.TK)–derived human cells containing wild-type mtDNA cells that were depleted of their mtDNA by chronic exposure to ethidium bromide (ρ0 cells).^{8 ,18} The LHON cybrids were seeded in two 6-well or two 96-well dishes. For AAV infections, cybrid cells at approximately 50% confluency were infected at multiplicities of infection of 5000 viral particles per cell, one 6-well dish or one 96-well dish with AAV-SOD2, and one 6-well dish or one 96-well dish with AAV-GFP. Two days after the AAV infections, the high-glucose medium was replaced with glucose-free galactose medium as previously described.¹⁸ This selective medium forces the cells to use oxidative phosphorylation to produce adenosine triphosphate. After 2 days of growth in glucose-deficient galactose medium, the SOD2-infected cells from each of 6 wells and the GFP-infected cells from each of 6 wells were trypsinized and counted using an automated particle counter (Z-100; Coulter Diagnostics, Hialeah, Fla). After 3 days of growth in glucose-deficient galactose medium, the SOD2-infected cells from each of 10 wells and the GFP-infected cells from each of 10 wells were trypsinized and counted.

Two days after AAV infections, we harvested AAV-SOD2–transfected cybrids, control cells infected with AAV-GFP,¹⁹ or LHON cells that were not exposed to either AAV.Briefly, this involved washing the trypsinized cells in cold phosphate-buffered saline solution. Cells were then manually homogenized and stored at −80°C for later analysis. For immunodetection, 15 μg of total protein was separated on a 10% sodium dodecyl sulfate–polyacrylamide gel and electrotransferred to a polyvinylidene fluoride membrane (BioRad Laboratories, Hercules, Calif).The protein content of the samples was measured using a DC protein assay (BioRad Laboratories). We immunostained the membrane with polyclonal anti-SOD2 antibodies (Stressgen Bioreagents, Victoria, British Columbia) and then goat antirabbit IgG horseradish peroxidase–conjugated secondary antibodies (Sigma-Aldrich Corp). We detected complexes using the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Piscataway, NJ). Antimouse β-actin antibody was used as an internal control for protein loading.

We used the fluorescent probe dihydroethidium (DHE) to detect intracellular superoxide (Molecular Probes, Eugene, Ore). Superoxide oxidizes the weakly blue fluorescent DHE to a bright red fluorescent signal. Cybrids were seeded into 48 wells of the 96-well plates. Cells in 24 wells were transfected with SOD2, and cells in the other 24 wells were transfected with GFP. Two days later, the medium was replaced with glucose-free galactose medium. After 24 or 48 hours, cells were incubated with 1μM DHE for 20minutes at 37°C. They were washed and then observed under a fluorescence microscope (Leitz, Wetzlar, Germany). The intensity of fluorescence was quantitated using a fluorophotometer (Eclipse; Varian Medical Systems, Palo Alto, Calif)with excitation at 480 nm and emission at 560 nm (red). Wells were counted in duplicate or greater. Protein content of the samples was measured using the DC protein assay (BioRad Laboratories), and the intensity of fluorescence was adjusted to the sample protein content.

We selected DHE not only because of its specificity for detection of intracellular superoxide²⁰ but also because other commercially available fluorophores such as dichlorodihydrofluorescein have a green emission similar to that of GFP and may interfere with detection of the oxidized green fluorescence of dichlorodihydrofluorescein. In contrast,the peak of red fluorescent DHE oxidized by superoxide and used herein was easily distinguished from the other emission at 520 nm from the green fluorescence of GFP.

Cybrids were seeded into 48 wells of the 96-well plates. Cells in 24wells were transfected with AAV-SOD2, and cells in the remaining 24 wells were transfected with AAV-GFP. Two days later, the high-glucose medium was exchanged for glucose-free galactose medium. After 1 day (24 wells) and 2 days (24 wells) in this restrictive medium, apoptotic cell death was assessed with a TUNEL (terminal deoxynucleotidyl transferase–mediated biotin–deoxyuridine triphosphate nick-end labeling) reaction kit, according to the manufacturer's specifications (Roche Diagnostics Corp, Indianapolis,Ind). The red TUNEL-positive cells (emission, 560 nm) were visualized and quantitated as described for superoxide.

We compared the AAV-SOD2–transfected cells with controls inoculated with AAV-GFP. Statistical analysis was performed by analysis of variance. P<.05 was considered significant.

Immunoblots of AAV-SOD2–infected LHON cells showed increased manganese SOD expression relative to the control uninfected cybrids and those infected with AAV-GFP (Figure 1C). Fluorescence micrographs confirmed a decrease in superoxide-induced fluorescence following AAV-SOD2 infection. Treatment with AAV-SOD2 decreased superoxide-induced DHE fluorescence in LHON cells after 1 day (Figure 2A)or 2 days (Figure 2C) in the restrictive medium, relative to infection with AAV-GFP (Figure 2B and D). After 1 day of growth in the glucose-free galactose medium,quantitative analysis of the emission at 560 nm that was distinct from the green emission of GFP at 520 nm revealed that superoxide-induced DHE fluorescence decreased 15% relative to AAV infection with AAV-GFP (Figure 2E). This difference was not statistically significant. However,after 2 days of growth in this restrictive medium, superoxide-induced DHE fluorescence decreased 26% relative to the LHON cells infected with the control AAV. This difference was significant (P = .003).Clearly, SOD2 suppressed cellular production of superoxide.

Because mitochondrial oxidative stress is closely linked to apoptotic cell death, we assayed for TUNEL-positive cells as early as 1 day after growth in the galactose medium. Treatment with AAV-SOD2 decreased TUNEL-positive LHON cells after 1 day (Figure 3A) or 2 days (Figure 3C)in the restrictive medium, relative to infection with AAV-GFP (Figure 3B and D). Quantitative analysis revealed that, relative to the control AAV infection, the intensity of TUNEL fluorescence was diminished by 34% (not significant) after 1 day and 21% (P = .048)with SOD2 infection after 2 days in the galactose medium (Figure 3E). Clearly, SOD2 infection protected LHON cells against apoptotic cell death.

Reducing apoptotic cell death by protection against mitochondrial oxidative stress with AAV-SOD2 increased the survival of LHON cybrids. After 2 days of growth in the galactose medium, we found that LHON cell survival increased by 25% with AAV-SOD2 infection relative to the control infection with AAV-expressing GFP (P = .05) (Figure 4A-C).Although the population of cells dwindled relative to 2 days of growth in the galactose medium, after 3 days of growth in this restrictive medium, we found that AAV-SOD2 increased LHON cell survival by 89% relative to the controls (P = .006)(Figure 4C). Clearly, increasing mitochondrial antioxidant defenses rescued LHON cells.

Our findings show that the superoxide anion is involved in LHON cell death and suggest that increasing mitochondrial antioxidant defenses may be a potential treatment for LHON. Reactive oxygen species that include superoxide anion, hydrogen peroxide, nitric oxide, and peroxynitrite are major initiators of the apoptotic pathway leading to cell death in LHON cells.^{7 - 8} Although tissue levels of SOD2 expression and activity in the optic nerves of patients with LHON have yet to be determined, a decrease in mitochondrial SOD activity has been detected in the LHON cybrid cell line.⁸ Mitochondria mitigate oxygen toxicity predominantly via enzymatic antioxidants that include SOD and glutathione peroxidase. Lowered levels of mitochondrial SOD activity likely increase cellular injury and induce optic neuropathy in mitochondrial disorders, particularly those like LHON that are related to a loss of complex I activity.^{9 ,14 ,21 - 22}

Bolstering anti–reactive oxygen species defenses may suppress the death of retinal ganglion cells in LHON.⁸ Rescue of our animal model of complex I deficiency with SOD2 suggests that antioxidant gene therapy may be useful for patients with complex I deficiencies such as LHON.²³ In that model system, suppression of reactive oxygen species inhibited apoptotic death of retinal ganglion cells,a phenomenon that is also involved in the pathogenesis of disease caused by the mutated human ND4 complex I subunit gene. Apoptotic cell death associated with complex I impairment induced by rotenone can also be blocked by overexpression of SOD2, further supporting our work described in this report.¹¹

Treatment options for patients with LHON and those with other mitochondrial disorders are limited at present.²⁴ The most direct approach to treatment would be to correct the mutated mitochondrial DNA. Although genes have been inserted into the nucleus and cytoplasm through the use of vectors, the technology to introduce a gene into the mitochondria is not yet possible.²⁵ Because it is expression of the mutant complex I subunit at the protein level that causes the biochemical defect of LHON, an alternative and feasible approach is to import a normal protein allotopically into the mitochondria to complement the defective protein encoded by the mutated mtDNA.^{18 ,26 - 28} Our previous study showing allotopic rescue of this same LHON cell line with mutated G11778A mtDNA supports this form of intervention.¹⁸ However,a different allotopic construct would be needed for the 3 mitochondrial genes containing mutations in ND1, ND4, or ND6 responsible for 85% of LHON cases.

Recent studies showing subtle retinal and optic nerve injury in families harboring the G11778A mtDNA mutation^{29 - 30} suggest that treatment may be necessary before symptoms actually develop. Nevertheless,many patients with LHON are found at the initial examination to have optic disc edema and predominantly unilateral visual loss. Thus, there is a window of opportunity of several months for prophylactic intervention in the fellow eye³¹ with SOD2 gene therapy before it too loses vision. Still, the early retinal changes detected in LHON carriers before apoplectic visual loss²⁹ suggest that this approach may have the best chance for success if it is initiated at the earliest stages of disease. The aim would be to reduce the accumulation of optic nerve damage so that injury does not progress to a point beyond which loss of function becomes irreversible.

Correspondence: John Guy, MD, Box 100284,Department of Ophthalmology, College of Medicine, University of Florida, Gainesville,FL 32610-0284 (johnguy@eye.ufl.edu).

Submitted for Publication: June 9, 2006; final revision received August 14, 2006; accepted August 30, 2006.

Financial Disclosure: Dr Hauswirth and the University of Florida have a financial interest in the use of AAV vectors for treating retinal diseases associated with their involvement with Applied Genetic Technologies Corporation.

Funding/Support: This study was supported by grant EY 12355 from the National Eye Institute (Dr Guy).

Acknowledgment: We thank Valerio Carelli, MD,PhD, for the generous gift of the cybrids and Mabel Wilson for editing the manuscript.