The pathophysiology of some of the more common optic neuropathies associated with cecocentral scotomas might be explained by a unifying hypothesis. This hypothesis is based on clinical features of these optic neuropathies, laboratory studies of the pathophysiology of how retinal ganglion cells (RGCs) die, some of which is unpublished, and biochemistry of reactive oxygen species generation within some of these disorders.
OPTIC NEUROPATHIES AND CECOCENTRAL SCOTOMAS
Diseases of the optic nerve are associated with abnormalities of vision, primarily visual acuity, color vision, contrast sensitivity, and most relevant to this discussion, the visual field. The nature of the visual field defect usually reflects the location at which the disease affects the optic nerve. For example, glaucomatous optic neuropathy primarily affects RGC axons at the optic nerve head. A focal thinning of the optic disc will affect those axons which originated from RGCs defined by the pattern of the retinal nerve fiber layer (1). Compression of the chiasm from a pituitary adenoma that affects primarily crossing fibers will generate unilateral or bilateral temporal visual field defects.
Cecocentral scotomas are distinctive in that there is visual field loss centrally, but there is involvement that includes the blind spot, that is, a temporal predominance. This is different from pure central loss which only involves the blind spot as a result of the defect itself being large enough to subsume it. What is interesting about diseases associated with cecocentral scotomas is that they are relatively few in number and share common clinical features. The main optic neuropathies with cecocentral scotomas are:
- Leber hereditary optic neuropathy (LHON)
- Nutritional optic neuropathy
- Toxic optic neuropathy
- Other hereditary optic neuropathies.
My focus will be on the first 3, creating a hypothetical framework for understanding their pathophysiology.
CLINICAL PRESENTATION OF CECOCENTRAL SCOTOMAS
The optic neuropathies associated with cecocentral scotoma are painless and bilateral, and usually progressive. LHON is an exception in that the visual loss frequently starts unilaterally but eventually becomes bilateral in all cases. When only 1 eye is affected initially, the fellow eye follows within weeks to months although sometimes longer.
Optic disc edema is unusual in the optic neuropathies associated with cecocentral scotomas. In LHON, there may be a swollen optic nerve head with telangiectatic small vessels, but it is not true disc edema because it does not show evidence of vascular leakage on fluorescein angiography. Occasionally, acute toxic or nutritional optic neuropathies show elevation of the disc.
CECOCENTRAL SCOTOMAS AND PATHOPHYSIOLOGY
A cecocentral scotoma is a visual field defect that can be thought of as representing a biomarker of an underlying pathophysiological process. The retinotopic distribution of the cecocentral scotoma classically has been believed to reflect damage to the papillomacular bundle. However, the papillomacular bundle is not a single well-defined set of axons but rather a concentration of axons that are primarily small in diameter and are within the area between the optic disc and the perifoveal macula (2). As pointed out by Plant and Perry (2) and others, the concept of the papillomacular bundle was a reverse induction from the clinical and histological findings associated with toxic optic neuropathy. It is tautologous that the papillomacular bundle is involved in similar diseases, that is, those with cecocentral scotomas. For the sake of convenience and common usage in the rest of this article, I will use the term papillomacular bundle to refer to the set of fibers arising from the foveal and parafoveal regions and approaching the temporal part of the optic disc, both directly and through an arcuate pathway.
The critical assumption underlying the framework being developed is that the common clinical presentation of optic neuropathies associated with cecocentral scotomas implies a common pathophysiological process. Specifically, my hypothesis is that these disorders have in common the generation of superoxide anion, or superoxide (O2–). Superoxide is an oxygen-containing free radical generated within cells in various processes. Relevant to this hypothesis, one of the main sources of superoxide is the reaction of molecular oxygen (O2) with a free electron.
Superoxide is reactive, and among its reactions is that with nitric oxide (NO) to form peroxynitrite (ONOO−). Peroxynitrite can react with other molecules, such as tyrosines, resulting in their nitration. Superoxide also reacts directly with macromolecules such as proteins and nucleic acids. The scavenging of superoxide typically occurs through superoxide dismutases of which there are 3 major types: intracellular (SOD-1 [Cu/Zn-SOD], mitochondrial [SOD-2 Mn-SOD], and extracellular [SOD-3]).
Besides causing oxidative damage, superoxide has an important role as a signaling molecule. Cells use small molecules such as NO to activate or suppress various processes within the cell. Superoxide has been recognized for many years to be such a molecule. For example, it is known to initiate mitosis in certain cells (3).
Over the last decade, our group demonstrated that superoxide plays a special role relevant to RGCs. Specifically, we showed that it signals the death of the cell body, or soma, when the axon is injured (4,5). In other words, injury to the axon of the RGC caused an increase in superoxide within its soma. This finding was accomplished by imaging superoxide using specialized fluorescent probes in rat retinas after optic nerve transection. The increase in superoxide could be seen a few days after axonal injury, followed a day later by death of the RGC.
Using a variety of techniques, we established that the superoxide was not simply a result of the cell in the final throes of death but, rather, a signal that both preceded and was necessary for death. We did this by showing that other drugs that decrease the levels of superoxide (e.g., pegylated superoxide dismutase); both decreased the levels of superoxide and prevented the death of the RGC. Based on these studies, we concluded that superoxide was a signal for RGC death after axonal injury (5,6).
SUPEROXIDE AND CECOCENTRAL SCOTOMAS
How is superoxide relevant to optic neuropathies in which cecocentral scotomas occur? The next sections will discuss 3 specific optic neuropathies in turn, namely, LHON, the optic neuropathy associated with vitamin B12 deficiency and the toxic optic neuropathy caused by ethambutol. Each of these is associated with cecocentral scotomas, and for each of these, we have evidence that superoxide is generated at increased levels.
Leber Hereditary Optic Neuropathy
In at least 95% of patients with LHON, there is a mutation at 1 of 3 sites within the mitochondrial DNA (mtDNA). The most common is the 11778 mutation, followed by the 3460 and 14484 mutations. There is a slow-to-rapid development of loss of vision in 1 or both eyes. If only 1 eye is involved initially, the second eye follows within weeks to months. The loss of vision typically occurs in the late teens or early 20s, but it can occur at any age. There is a strong male predominance, which is as yet unexplained, especially as mtDNA is present in all cells (except mature erythrocytes) of both genders.
All of the 3 primary mutations that produce LHON are contained in mtDNA coding for components of the complex I of the mitochondrial electron transport chain. Complex I serves as an NADH:ubiquinone oxidoreductase. In other words, NADH, a product of the Krebs cycle, is oxidized at the same time that ubiquinone, a molecule that carries electrons between complexes, is reduced. Functionally, 2 electrons are transported from NADH to ubiquinone at the same time that 4 protons (H+) are transported across the inner membrane of the mitochondria into the mitochondrial intermembrane space.
This set of reactions has 2 results. First, electrons are transported to various loci in the electron transport chain. Second, protons are pumped across the inner membrane to form a voltage gradient. The voltage gradient, or electromotive potential, is the basis for another complex, complex V, to generate adenosine triphosphate (ATP) in the process of transferring protons back across the inner membrane into the mitochondrial matrix.
For many years, it was assumed that the mechanism by which LHON mtDNA mutations in complex I caused visual loss was a deficiency of ATP, for example, by decreased pumping of protons or some other mechanism. This made sense, given that ATP is essential for cellular function, and RGCs, with their very long axons relative to the size of their somas, would presumably have high ATP demands. However, there are some flaws in this assumption, related to the fact that RGCs are specifically affected in LHON, whereas other cells are less commonly involved. We pointed out several years ago that diseases characterized by decreased ATP production from mutations in mtDNA usually are not associated with death of RGCs (7). Instead, they are associated with death of photoreceptors, as occurs in mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes (MELAS; mtDNA position 3243) or neuropathy, ataxia, and retinitis pigmentosa (NARP; mtDNA position 8993). Frequently, there is abnormal extraocular muscle function, as well as other systemic involvement such as abnormal cardiac conduction deficits and pancreatic dysfunction. Tissues such as photoreceptors and extraocular muscles are highly energy-consuming, and any disruption of ATP production in MELAS or NARP usually will affect patients early in the course of their disease. Therefore, if there were a deficiency in ATP production in LHON, there should be effects on photoreceptors and not, or at least not only, on RGCs.
The second reason to suspect that ATP production is less relevant to LHON than previously believed is that studies of LHON cybrids demonstrate only a relatively small effect on ATP production. Cybrids are cells in which the mitochondria are from one cell source, usually having mutant mtDNA, whereas the rest of the cell is from another source, usually a wild-type dividing cell. This allows study of the specific effects of mtDNA mutations without affecting other aspects of cellular function. Surprisingly, the respiration deficits that would decrease ATP production are not particularly marked (8–12), although this is not always the case (13). Note that it is currently not possible to make cybrids from isolated RGCs because a dividing cell is necessary, and RGCs, like other neurons, do not divide.
Superoxide is produced in LHON cybrids that are made in neuronal cell lines (14). For example, Wong and colleagues showed an approximate doubling of superoxide production in LHON 11778 or 3460 cybrids compared with control, nonmutant cybrids (14). They suggested that this effect was specific to differentiated neurons. However, they used a neuronal cell line that was different from RGCs, and thus, their findings would not explain the specific damage to RGCs in LHON.
How do LHON mutations result in an increase in superoxide? As mentioned previously, the function of complex I is as an NADH:ubiquinone oxidoreductase. If the transfer of electrons to ubiquinone is disrupted, then complex I will retain the extra electron, which then can react with free oxygen to form superoxide. In other words, the problem with the LHON mutations and their effects on complex I is less likely due to decreased ATP production, but instead due to an accumulation of superoxide from failure to transfer electrons down the electron transport chain to other complexes. This, in turn, results in those electrons reacting with molecular oxygen to form superoxide.
Why should an increase in superoxide in LHON lead to RGC death but not death of other cells? Mitochondria are present in all cells in the body except mature erythrocytes, and therefore, superoxide should be increasing in those cells as well. The specificity to RGCs is, therefore, unexpected.
The answer is that superoxide is not only a free radical that can react with other molecules to cause oxidative damage, but it also is a signaling molecule. As discussed previously, superoxide is a signal for RGC bodies to die after the axon is injured. Our hypothesis (7) is that the increased superoxide “tricks” the soma of the RGC into thinking that axon injury has occurred, although it has not. The RGC soma then initiates the process of apoptosis, misled by the aberrant superoxide signal, and eventually dies. Other cells in the body may not use superoxide as a signal for cell death and, therefore, remain unaffected.
Partial support for this theory can be found in a study of mice in which superoxide dismutase-1 is knocked out, resulting in the levels of superoxide within the cell bodies being higher than normal (15). These animals have decreased numbers of RGCs, consistent with the premise (relevant to LHON) that increasing superoxide levels in cells other than RGCs nevertheless causes the RGCs to die, even when there is no axon injury and no direct injury to the RGCs.
In summary, LHON likely represents a disease in which superoxide is abnormally produced because of mutations in complex I of the mitochondrial electron transport chain. RGCs may be specifically sensitive to increased superoxide because it is a signaling molecule for cell death after axon injury. The increased superoxide ultimately would result in specific loss of RGCs, with relative preservation of other cells within the nervous system and elsewhere in the body.
Vitamin B12 Deficiency Optic Neuropathy
“Nutritional amblyopia” is a term used for the optic neuropathy that arises from malnutrition. Examples include starved prisoners of war in Asia during World War II, severe malnutrition in alcoholics, those on inadequately supplemented vegan diets, patients with various gastrointestinal disorders or who have had intestinal surgery resulting in inadequate absorption of needed nutrients, and pernicious anemia in which, antibodies to intrinsic factor decrease absorption of vitamin B12.
In virtually all cases, the causative factor is an abnormality of vitamin B12, either from its absence in the diet or failure of absorption in the gastrointestinal tract. Occasionally, abnormalities in metabolism or cell uptake of vitamin B12 are causative. There are occasional cases of nutritional amblyopia from a deficiency of other nutrients, most commonly folate and, rarely, vitamin B6.
The clinical picture of vitamin B12 deficiency–related optic neuropathy is slowly progressive bilateral visual loss, decreased color vision, cecocentral scotomas, and the eventual appearance of temporal disc pallor, sometimes with excavation of the temporal disc. Interestingly, these disc features are similar to those seen in patients with LHON.
Vitamin B12 has several roles within cells. Its absence causes megaloblastic anemia, subacute degeneration of the spinal cord, neuropsychiatric disorders, and optic neuropathy. It is presumed that most of these abnormalities are due to its role as a cofactor for 2 enzymes, methionine synthase and methylmalonyl-CoA mutase, so that its deficiency will result in increased levels of methionine and methylmalonic acid, respectively. In some patients, vitamin B12 levels in the blood may be in the “normal” range, but there are elevated serum levels of homocysteine and/or methylmalonic acid, implying that there is a functional vitamin B12 deficiency despite the normal serum levels.
The treatment of severe vitamin B12 deficiency–related optic neuropathy is usually large amounts of vitamin B12 delivered through a parenteral route, followed by continued supplementation at a lower dose. The commercially available form of vitamin B12 in tablets is cyanocobalamin. Many believe that hydroxocobalamin is preferable because of its longer half-life, the cyanide group on cyanocobalamin could be toxic to RGCs, or hydroxycobalamin can detoxify free cyanide. Interestingly, the World Health Organization lists hydroxocobalamin and not cyanocobalamin in its list of essential drugs (16).
Vitamin B12 as a Superoxide Scavenger
As part of a collaboration with Professor Zeev Gross of the Technion in Haifa, Israel, we assessed novel chemical entities to serve as superoxide scavengers for decreasing RGC death after axon injury. We tested metallocorroles, molecules that Gross and his laboratory had invented for redox chemistry. We tested their role as potential neuroprotective compounds and showed that some metallocorroles served as excellent superoxide scavengers in RGCs (17). They also were effective in decreasing RGC death after optic nerve transection in adult rats (18).
We noticed that the chemical structures of metallocorroles and vitamin B12 were very similar, with both having a corrin ring. Based on this similarity, we hypothesized that vitamin B12 could serve as a superoxide scavenger. A literature search found an article from the laboratory of Nicola E. Brasch at Kent State University showing that cobalamin is a superoxide scavenger with catalytic activity approaching that of superoxide dismutase (19). We therefore embarked on a series of experiments to examine the ability of vitamin B12, in the form of cyanocobalamin, to scavenge superoxide in RGCs and rescue them from cell death. While our results have been submitted for publication, the main findings are:
- In the test tube, vitamin B12 is an excellent scavenger of superoxide produced by reacting xanthine with xanthine oxidase.
- Superoxide scavenges RGCs in neuronal-like cells (we used the RGC-5 photoreceptor-derived cell line that, despite its name, is not an RGC).
- Intravitreal vitamin B12 scavenges superoxide produced in RGCs after axon transection.
- Intravitreal superoxide decreases RGC death after optic nerve transection.
Implications for Understanding Vitamin B12 Optic Neuropathy
We hypothesize that vitamin B12 has, in addition to its enzymatic cofactor roles with respect to methionine synthase and methylmalonyl-CoA mutase, a role as an endogenous superoxide scavenger in RGCs. Its deficiency would be specifically deleterious to RGCs because, as noted in the previous section on LHON, RGCs use superoxide as an intracellular signaling molecule for inducing soma apoptosis after axon injury. Given that superoxide is continuously being generated within cells as part of the mitochondrial electron transport chain, if superoxide is a signaling molecule, an RGC would require strict control of intracellular concentrations. Otherwise, the deregulated superoxide might induce aberrant cell death. The usual superoxide scavengers within cells, such as the various superoxide dismutases, theoretically should be sufficient, as they are for other cells. Given that RGCs do not divide and therefore cannot be replenished if they die, it is not surprising that they would develop a system for regulating superoxide levels that is more stringent than that of other cells.
We have suggested a role for vitamin B12 as an additional superoxide scavenger. This is hypothetical, and further work will be needed to prove that this is indeed a mechanism relevant to RGCs. It is very difficult to induce optic neuropathy from vitamin B12 deficiency in rats, and even more than 12 months of such deficiency does not cause an obvious loss of RGCs. Thus, it is possible that vitamin B12 does not play the same role in rats as it does in humans. This is akin to other differences between rats and humans with respect to optic neuropathies, for example, rats given methanol develop primarily a photoreceptor toxicity not an RGC toxicity. Another possibility is that rats are able to upregulate other superoxide scavenging molecules within cells in response to vitamin B12 deficiency. Overall, this is an area of active research.
In summary, these data are consistent with the hypothesis that vitamin B12 deficiency, the known cause of nutritional optic neuropathy, is due to an abnormality in superoxide scavenging and that elevated levels of superoxide within RGCs would be responsible for the progressive bilateral visual loss associated with cecocentral scotomas that occurs from RGC death in this disease.
Ethambutol Optic Neuropathy
Ethambutol is commonly used to treat tuberculosis and other mycobacterial infections, especially multiple drug-resistant tuberculosis. Ethambutol is the most common cause of toxic optic neuropathy (20,21) and is characterized by progressive bilateral loss of visual acuity, acquired dyschromatopsia, cecocentral scotomas, and optic disc pallor. In some cases, there is a significant temporal preponderance of the visual field defect that sometimes respects the vertical meridian, suggesting that the damage is occurring primarily at or near the optic chiasm. This is different from the classic cecocentral scotoma associated with presumed involvement of the papillomacular bundle.
The potential for ethambutol to produce an optic neuropathy limits its use in patients who have visual symptoms, underlining the need to understand this optic neuropathy and find mitigating strategies. Major problems are that it is difficult to predict who will develop the optic neuropathy, how it can be detected before RGCs are irreversibly damaged, and, once present, how to treat the visual loss other than by stopping the medication. In many cases of ethambutol optic neuropathy, there is reversibility when the visual loss is detected early. However, the literature is replete with cases demonstrating failure to completely or even partially reverse the loss of vision. Two studies demonstrated that 40%–50% of patients with ethambutol-related optic neuropathy had no visual improvement after ethambutol was discontinued, an unacceptably high failure rate for the only known therapeutic intervention (22,23).
A priori, there is no specific reason to suspect that ethambutol toxicity is associated with superoxide induction. However, in preliminary experiments that we have performed in rats, we found increased levels of superoxide after intravitreal injection of ethambutol, using the oxidation of hydroethidine as a marker for superoxide induction.
In Vitro Detection of Superoxide Induced by Ethambutol by Fluorescent Microscopy
We purified rat RGCs by sequential immunopanning and plating on poly-D-lysine/laminin-coated plates to provide a substrate for adherence and neurite outgrowth. The cells were then exposed to 3 mM ethambutol for 1 hour, with hydroethidine (1 mM) added in the last 30 minutes to detect superoxide generation. Hydroethidine reacts with superoxide to form 2-oxy-ethidium, which has specific excitation and emission characteristics. The cells then were imaged by fluorescent microscopy or with a fluorescent plate reader.
The addition of ethambutol resulted in a robust fluorescence signal, compared with vehicle control. To prove that ethambutol led to superoxide generation and not generation of some other reactive oxygen species that might also oxidize hydroethidine, some wells were incubated simultaneously with pegylated superoxide dismutase (500 U/mL), which enters cells and scavenges only superoxide and not other reactive oxygen species. This eliminated the ethambutol-induced fluorescence, implying that the fluorescence reflected an increase in superoxide alone.
The interpretation of these experiments is that ethambutol induces superoxide in cultured rat RGCs.
In Vivo Detection of Superoxide Induced by Ethambutol Using In Vivo Confocal Imaging
One eye of adult Long-Evans rats under anesthesia received an intravitreal injection of ethambutol (final concentrations 0, 0.3, 1, or 3 mM), along with hydroethidine (final concentration 100 μM) to detect superoxide. The other eye was not injected and acted as control. Rats were imaged 24 hours later with a Heidelberg HRA-2 confocal scanning laser ophthalmoscope using the 488-nm laser for excitation and a broad-spectrum 500–600 nm filter for emission detection. Eyes injected with 3 mM ethambutol demonstrated small but definite production of superoxide in cells within the RGC layer that were not fluorescent before ethambutol injection, whereas equivocal induction was seen in eyes injected with the 0.3 or 1 mM concentrations of ethambutol. No fluorescence evidence of superoxide induction was seen in control eyes.
These data suggest that ethambutol can induce superoxide in RGCs and presumably lead to their death. Note that these experiments used an acute exposure to high concentrations of the drug. Chronic exposure experiments need to be performed to allow translational validation in patients. Furthermore, chronic exposure experiments to very high doses of ethambutol in rats demonstrated chiasmal lesions (24), and whether or not this chiasmal predilection is related to superoxide production is unclear. Finally, the mechanism for ethambutol causing an increase in superoxide is currently unknown but could be related to zinc chelation effects of ethambutol and the interaction of zinc with complex I (25).
ISSUES RELATED TO THE SUPEROXIDE HYPOTHESIS FOR CECOCENTRAL SCOTOMAS
Why the Papillomacular Bundle?
Even if superoxide is a common pathophysiological factor for LHON, vitamin B12–deficiency optic neuropathy, and ethambutol optic neuropathy, this does not explain why a cecocentral scotoma develops. In other words, why should the papillomacular bundle be preferentially involved, and why should axons or their RGCs elsewhere in the retina be relatively spared?
The answer probably relates to the fact that the axons in the papillomacular bundle are smaller than average. Although there is controversy regarding this issue (2), there is strong evidence to suggest that 1) RGCs forming the papillomacular bundle are predominantly midget cells with small axons, 2) the axons are small for most of their course toward the disc, 3) these features are different from fibers approaching the disc from other directions (26–28).
Sadun and colleagues have made convincing arguments that the size of RGC axons is relevant to the effects of a decrease in ATP production in the setting of LHON, with a greater mismatch between the ATP produced and the ATP needed in small vs large fibers of RGCs (29,30). However, as indicated above in the section on LHON, a deficit in ATP production is unlikely to be the main causative factor in this disorder, and presumably, also in vitamin B12–deficiency optic neuropathy or ethambutol optic neuropathy.
Instead, a parallel analytical process used by Pan et al (30) for comparing small and large axons can be applied to the dynamics of superoxide production, as follows:
- Most of the ATP needed for axon conduction is for renormalizing the sodium and potassium concentrations at the nodes of Ranvier, primarily after an action potential has occurred. Sodium enters the axon and potassium leaves the axon as the 2 major ionic fluxes associated with the formation of an axon potential. The Na+–K+-ATPase renormalizes these concentrations by shuttling sodium and potassium ions to the extracellular and intracellular spaces, respectively.
- The renormalization of sodium and potassium concentrations is an ATP-dependent step, and therefore, the amount of ATP being consumed will be proportional to the amount of ionic flux, which takes place at the surface of the axon, at the nodes of Ranvier.
- Superoxide is a byproduct of ATP production in the mitochondrial electron transport chain, and therefore, the amount of superoxide being produced will largely be proportional to the amount of ATP needed to support the ionic flux that takes place at the axon surface.
- The detoxification of superoxide should be proportional to the axon volume, given that most superoxide is detoxified by intracellular superoxide dismutases.
- The area/volume ratio is greater in small fibers than in large fibers because the surface area is proportional to the radius and the volume is proportional to the square of the radius.
- Given that the amount of superoxide production is proportional to sodium and potassium flux at the axon surface area and that the superoxide detoxification is proportional to axon volume, there will be relatively more superoxide production than detoxification in small fibers compared with large fibers.
In other words, there should be a relatively greater mismatch between superoxide production (relative to ATP synthesis used to maintain axon conduction) and detoxification in small RGC axons than in large axons. This would support a preferential involvement of small axons in the papillomacular bundle. However, as pointed out by others (2), it is not fiber size alone that would explain the concentrated damage in this area. It is possible that the packing density of the axons in the retinal nerve fiber layer near the temporal disc or where the axons enter the disc that is the critical factor.
There are occasional cases of ethambutol optic neuropathy in which the visual field defects are suggestive of chiasmal involvement. There is also a pathological study in rats showing chiasmal axon damage (24). Interestingly, there are cases of LHON in which there is predominantly chiasmal involvement based on clinical, magnetic resonance imaging, or pathological evidence (31,32).
These findings are not directly explained by aberrant superoxide levels causing death of RGCs. One possible reason for preferential involvement of the chiasm is that the crossing fibers within the chiasm theoretically undergo a small deformation as they cross under and over each other, in the same way that a weave deforms the fibers as they cross each other. Under those circumstances, the surface/volume relationship described above would be exacerbated in those locations where the chiasm is flattened. The surface/volume ratio of a flattened fiber is greater than one with a more circular cross-section. The factor of packing density and relation with adjacent fibers also may be relevant.
Other Optic Neuropathies With Cecocentral Scotomas
This hypothesis discusses 3 specific optic neuropathies, but there are other optic neuropathies, such as dominant optic atrophy and methanol-related optic neuropathy, in which the same clinical pattern occurs. The following is highly speculative and is included only in the interest of completeness.
Dominantly inherited optic atrophy is associated most often with mutations in the OPA1 and OPA3 genes, both of which are important for mitochondrial function. Studies in Drosophila have shown evidence that OPA1 deficiency increases reactive oxygen species levels, causes a variety of ocular abnormalities, and is rescued by superoxide dismutase (33). It would not be surprising if OPA1 deficiency in patients with dominant optic atrophy leads to increased superoxide levels in most cells, with preferentially more toxicity in RGCs for the same reason that it occurs in LHON.
Methanol intoxication is another example in which there may be a role for superoxide. Methanol is metabolized to formate, which inhibits mitochondrial cytochrome c oxidase (complex IV). This inhibition leads to electrically reduced upstream complexes in the mitochondrial electron transport chain. The electron-rich reduced complexes can react with molecular oxygen to form superoxide. In experiments conducted in rodents, methanol intoxication does not cause the predominant RGC death that it does in humans, but instead results in primarily photoreceptor death (34). Therefore, it is unclear if studies in rodents can help to clarify the precise mechanism by which methanol causes an optic neuropathy.
Implications for Treatment of Optic Neuropathies Associated With Cecocentral Scotomas
From the foregoing discussion, one might conclude that drugs that reduce superoxide levels within RGCs might be effective treatments for any or all of these diseases. This is a reasonable assumption, but there are several caveats. The most important is that it is the level of intracellular superoxide that is critical, and unless a superoxide scavenger is able to actually enter the RGC, it may not be effective. It also has to enter the cell at high levels. For example, loading patients with high levels of vitamin B12 makes theoretical sense, but unless the intracellular concentration is maintained at a sufficiently high level for the long term, it may not be a sufficiently effective superoxide scavenger to be clinically beneficial. Drugs that our laboratory has studied, such as pegylated superoxide dismutase and metallocorroles, (5,17,18) are theoretically potent, but their use in humans would require a lengthy drug development process.
There has been much interest in drugs such as in idebenone for LHON. A formal clinical trial was performed but did not meet its primary end point for preventing vision loss in LHON (35). However, much of the data from the trial were encouraging because it was a separate nonrandomized study (36). Idebenone was designed as a drug that would serve to shuttle electrons, serving as a synthetic ubiquinone electron carrier. However, biochemical studies show that idebenone also is a superoxide scavenger (37,38). It may actually be efficacious because it reduces intracellular superoxide and not because it restores mitochondrial electric transport in LHON. On this basis, one could speculate whether or not drugs such as idebenone could be used in other optic neuropathies associated with cecocentral scotomas.
I have presented above a hypothesis that ties together several disparate optic neuropathies, all characterized by a similar clinical presentation. The hypothesis is predicated on the formation of intracellular superoxide within RGCs as a common pathological pathway for the type of cell death that occurs. The anatomical predisposition of the papillomacular bundle to have elevated superoxide levels is tied to the size of the fibers involved, a hypothesis that also implicates the crossing fibers of the chiasm. Much of this work is speculative and is an interpretation of several experimental studies that have been performed to date. Hopefully, this hypothesis will be developed further, and its validity tested in both experimental models and, ultimately, in humans.
I thank members of my laboratory who worked in the areas covered by the preceding discussion, including Mohammadali Almasieh, Rachel Beaubien, Maria-Magdalena Catrinescu, Wesley Chan, Megan Crowe, Laura Frassetto, Alireza Ghaffariyeh, Akiyasu Kanamori, Chris Lieven, Katrina Mears, Nicholas Niemuth, Jonathan Ribich, Colin Scott, and Emily Seidler. I am also grateful for helpful discussions with many of my colleagues, including Simmons Lesell, Alfredo Sadun, and Helen Danesh-Meyer.
1. Levin LA. Relevance of the site of injury of glaucoma to neuroprotective strategies. Surv Ophthalmol. 2001;45:S243–S249.
2. Plant GT, Perry VH. The anatomical basis of the caecocentral scotoma. New observations and a review. Brain. 1990;113:1441–1457.
3. Joneson T, Bar-Sagi D. A Rac1 effector site controlling mitogenesis through superoxide production. J Biol Chem. 1998;273:17991–17994.
4. Lieven CJ, Schlieve CR, Hoegger MJ, Levin LA. Retinal ganglion cell axotomy induces an increase in intracellular superoxide anion. Invest Ophthalmol Vis Sci. 2006;47:1477–1485.
5. Kanamori A, Catrinescu MM, Kanamori N, Mears KA, Beaubien R, Levin LA. Superoxide is an associated signal for apoptosis in axonal injury. Brain. 2010;133:2612–2625.
6. Lieven CJ, Thurber KA, Levin EJ, Levin LA. Ordering of neuronal apoptosis signaling: a superoxide burst precedes mitochondrial cytochrome c release in a growth factor deprivation model. Apoptosis. 2012;17:591–599.
7. Levin LA. Mechanisms of retinal ganglion specific-cell death in Leber hereditary optic neuropathy. Trans Am Ophthalmol Soc. 2007;105:379–391.
8. Carelli V, Ghelli A, Ratta M, Bacchilega E, Sangiorgis S, Mancini R, Levzzi V, Cortelli P, Montagna P, Lugaresi E, Degli Esposti M. Leber's hereditary optic neuropathy biochemical effect of 11778/ND4 and 3460/ND1 mutations and correlation with the mitochondrial genotype. Neurology. 1997;48:1623–1632.
9. Carelli V, Ghelli A, Bucchi L, Montagna P, DeNegri A, Levzzi V, Carducci C, Lenaz G, Lugaresi E, Degli Esposti M. Biochemical features of mtDNA 14484 (ND6/M64V) point mutation associated with Leber's hereditary optic neuropathy. Ann Neurol. 1999;45:320–328.
10. Cock HR, Cooper JM, Schapira AH. Functional consequences of the 3460-bp mitochondrial DNA mutation associated with Leber's hereditary optic neuropathy. J Neurol Sci. 1999;165:10–17.
11. Brown MD, Trounce IA, Jun AS, Allen JC, Wallace DC. Functional analysis of lymphoblast and cybrid mitochondria containing the 3460, 11778, or 14484 Leber's hereditary optic neuropathy mitochondrial DNA mutation. J Biol Chem. 2000;275:39831–39836.
12. Brown MD, Allen JC, Van Stavern GP, Newman NJ, Wallace DC. Clinical, genetic, and biochemical characterization of a Leber hereditary optic neuropathy family containing both the 11778 and 14484 primary mutations. Am J Med Genet. 2001;104:331–338.
13. Baracca A, Solaini G, Sgarbi G, Lenaz G, Baruzzi A, Schapira AH, Martinuzzi A, Carelli V. Severe impairment of complex I-driven adenosine triphosphate synthesis in Leber hereditary optic neuropathy cybrids. Arch Neurol. 2005;62:730–736.
14. Wong A, Cavelier L, Collins-Schramm HE, Seldin MF, McGrogan M, Savontaus ML, Cortopassi GA. Differentiation-specific effects of LHON mutations introduced into neuronal NT2 cells. Hum Mol Genet. 2002;11:431–438.
15. Yuki K, Ozawa Y, Yoshida T, Kurihara T, Hirasawa M, Ozeki N, Shiba D, Noda K, Ishida S, Tsubota K. Retinal ganglion cell loss in superoxide dismutase 1 deficiency. Invest Ophthalmol Vis Sci. 2011;52:4143–4150.
17. Kanamori A, Catrinescu MM, Mahammed A, Gross Z, Levin LA. Neuroprotection against superoxide anion radical by metallocorroles in cellular and murine models of optic neuropathy. J Neurochem. 2010;114:488–498.
18. Catrinescu MM, Chan W, Mahammed A, Gross Z, Levin LA. Superoxide signaling and cell death in retinal ganglion cell axotomy: effects of metallocorroles. Exp Eye Res. 2012;97:31–35.
19. Suarez-Moreira E, Yun J, Birch CS, Williams JH, McCaddon A, Brasch NE. Vitamin B(12) and redox homeostasis: cob(II)alamin reacts with superoxide at rates approaching superoxide dismutase (SOD). J Am Chem Soc. 2009;131:15078.
20. Sadun AA, Wang MY. Ethambutol optic neuropathy: how we can prevent 100,000 new cases of blindness each year. J Neuroophthalmol. 2008;28:265–268.
21. Wang MY, Sadun AA. Drug-related mitochondrial optic neuropathies. J Neuroophthalmol. 2013;33:172–178.
22. Kho RC, Al-Obailan M, Arnold AC. Bitemporal visual field defects in ethambutol-induced optic neuropathy. J Neuroophthalmol. 2011;31:121–126.
23. Lee EJ, Kim SJ, Choung HK, Kim JH, Yu YS. Incidence and clinical features of ethambutol-induced optic neuropathy in Korea. J Neuroophthalmol. 2008;28:269–277.
24. Lessell S. Histopathology of experimental ethambutol intoxication. Invest Ophthalmol Vis Sci. 1976;15:765–769.
25. Sharpley MS, Hirst J. The inhibition of mitochondrial complex I (NADH:ubiquinone oxidoreductase) by Zn2+. J Biol Chem. 2006;281:34803–34809.
26. Ogden TE. Nerve fiber layer of the primate retina: morphometric analysis. Invest Ophthalmol Vis Sci. 1984;25:19–29.
27. Pavlidis M, Stupp T, Hummeke M, Thanos S. Morphometric examination of human and monkey retinal ganglion cells within the papillomacular area. Retina. 2006;26:445–453.
28. Hiraoka M, Inoue K, Kawano H, Takada M. Localization of papillofoveal bundles in primates. Anat Rec (Hoboken). 2012;295:347–354.
29. Sadun AA, Win PH, Ross-Cisneros FN, Walker SO, Carelli V. Leber's hereditary optic neuropathy differentially affects smaller axons in the optic nerve. Trans Am Ophthalmol Soc. 2000;98:223–232; discussion 232–225.
30. Pan BX, Ross-Cisneros FN, Carelli V, Rue KS, Salomao SR, Moraes-Fiho MN, Moraes MN, Berezovsky A, Belfort R, Sadun AA. Mathematically modeling the involvement of axons in Leber's hereditary optic neuropathy. Invest Ophthalmol Vis Sci. 2012;53:7608–7617.
31. Weiner NC, Newman NJ, Lessell S, Johns DR, Lott MT, Wallace DC. Atypical Leber's hereditary optic neuropathy with molecular confirmation. Arch Neurol. 1993;50:470–473.
32. Phillips PH, Vaphiades M, Glasier CM, Gray LG, Lee AG. Chiasmal enlargement and optic nerve enhancement on magnetic resonance imaging in leber hereditary optic neuropathy. Arch Ophthalmol. 2003;121:577–579.
33. Yarosh W, Monserrate J, Tong JJ, Tse S, Le PK, Nguyen K, Brachmann DC, Huang T. The molecular mechanisms of OPA1-mediated optic atrophy in Drosophila model and prospects for antioxidant treatment. PLoS Genet. 2008;4:e6.
34. Seme MT, Summerfelt P, Henry MM, Neitz J, Eells JT. Formate-induced inhibition of photoreceptor function in methanol intoxication. J Pharmacol Exp Ther. 1999;289:361–370.
35. Klopstock T, Yu-Wai-Man P, Dimitriadis K, Rizzo G, Carbonelli M, DeNegri AM, Sadun F, Carta A, Guerriero S, Simonelli F, Sadun AA, Aggarwal P, Liguori R, Avoni P, Baruzzi A, Zeviani M, Montagna P, Barboni P. A randomized placebo-controlled trial of idebenone in Leber's hereditary optic neuropathy. Brain. 2011;134:2677–2686.
36. Carelli V, La Morgia C, Valentino ML, Rizzo G, Carbonelli M, DeNegri AM, Sadun F, Carta A, Guerriero S, Simonelli F, Sadun AA, Aggarwal P, Liguori R, Avoni P, Baruzzi A, Zeviani M, Montagna P, Barboni P. Idebenone treatment in Leber's hereditary optic neuropathy. Brain. 2011;134:e188.
37. Murakami M, Zs -Nagy I. Superoxide radical scavenging activity of idebenone in vitro studied by ESR spin trapping method and direct ESR measurement at liquid nitrogen temperature. Arch Gerontol Geriatr. 1990;11:199–214.
38. Maroz A, Anderson RF, Smith RA, Murphy MP. Reactivity of ubiquinone and ubiquinol with superoxide and the hydroperoxyl radical: implications for in vivo antioxidant activity. Free Radic Biol Med. 2009;46:105–109.