Mechanistic Insights into Glaucoma Provided by Experimental Genetics The Cogan Lecture

The objective of this lecture-based review is to summarize some of the mechanistic insights into glaucoma that have been derived from research in mice in my laboratory over the past 10 years. The mouse system provides powerful genetic tools for studying complex, asynchronous diseases such as glaucoma (see Refs. ¹ , ² ). Because of the high degree of conservation of physiological and homeostatic mechanisms between mammals, mice often have the same diseases as do humans, and these diseases develop in a relatively short span of time. Mouse and human diseases often are caused by the same mutant gene or pathway. Although mouse and human diseases are not always identical, mouse studies are providing an important complement to those in humans and other species. This article illustrates several ways in which we have used mice to study different aspects of glaucoma, but it does not attempt to provide a comprehensive literature review. I first focus the discussion on our studies of risk factors for high intraocular pressure (IOP) and then turn to mechanisms that kill retinal ganglion cells (RGCs). 

Intraocular Pressure and Glaucoma: An Introduction

Glaucoma is a group of diseases that are characterized by the death of RGCs, specific visual field deficits, and optic nerve atrophy. ^{3 4} It is a major cause of blindness worldwide. ^{5 6} An IOP that is high compared with the population mean (elevated IOP) is often associated with glaucoma. Along with increasing age, high IOP is one of the strongest known risk factors for glaucoma. Experimental elevation of IOP can induce glaucoma in animals. Nevertheless, high IOP is not necessary for glaucoma to develop. In some individuals, glaucoma develops despite IOPs in the normal range. ⁷ In addition, high IOP by itself is not sufficient to cause glaucoma. ^{8 9} Most individuals with high IOP do not have glaucomatous visual loss. ^{9 10 11} In some individuals with high IOP, glaucoma does not develop, even over a period of long-term follow-up (see Refs. ⁸ , ¹¹ ). There are also profound differences in the rate of glaucomatous progression and the response to treatment between individual patients. ^{12 13 14 15 16} This indicates the existence of risk factors that determine both the magnitude of IOP that is harmful to each individual and that determine the ultimate severity and speed of visual damage. 

It is clear that many forms of glaucoma have a genetic component. ^{17 18 19 20} Although several glaucoma genes have been identified, ^{19 21} we are far from understanding the genetic or physiologic etiology of susceptibility to glaucoma. ¹⁹ Characterizing genetic factors that contribute to glaucoma will facilitate the early identification of individuals who are at increased risk of disease and who should be observed closely. Understanding the molecular mechanisms killing RGCs and understanding the role elevated IOP and other risk factors play in these processes is important in designing treatments to prevent visual deterioration in patients. For these reasons, our research has focused on understanding risk factors that contribute to IOP elevation and risk factors that determine the susceptibility of RGCs to glaucomatous neurodegeneration. A primary goal is to define genetic aspects of glaucoma that then act as an entry point for a molecular understanding of pathogenesis. 

Mice Provide an Appropriate Model System

It is important to discuss the mouse as a model system. When we started our studies, mice had hardly been used to study glaucoma, because of several factors, including the need for adequate tools for analyzing small eyes and the historic belief that mice were not anatomically appropriate models. We addressed some of these problems and helped to stimulate a growing interest in the use of mice in the study of glaucoma. To help develop the mouse to investigate factors that affect IOP and glaucoma, we developed and refined methods of clinical analysis of the mouse anterior chamber, fundus, and optic nerve, and we developed a reliable method for measuring IOP. ^{22 23 24 25} I will now discuss IOP and its effects. Issues relating to the optic nerve anatomy of mice are discussed later in the review. 

Using our IOP measuring procedure, we demonstrated an almost twofold range of IOP between different mouse strains (≈10–20 mm Hg, similar to that in humans, Fig. 1 ). We also found that age, time of day, and diabetes have similar effects on IOP in mice and humans. ²⁶ Regarding the anatomic relevance of mouse eyes for glaucoma, it was originally thought that the anatomy and development of the aqueous humor outflow pathway was too dissimilar to that in humans to make IOP studies useful. It now has been established that the drainage structures of mice have the same general organization as in primates. The mouse has an endothelial lined canal of Schlemm and a trabecular meshwork that consists of layers of well-organized trabecular beams that are covered with trabecular cells (Fig. 1) . ^{27 28} The similarity extends to drainage structure development and the genes that influence it. ^{21 27 29} Regarding aqueous humor drainage, the biggest anatomic difference between mice and primates is that mice have a poorly developed ciliary muscle. Nevertheless, similar to its effect in humans, the prostaglandin analogue latanaprost lowers mouse IOP. ^{30 31} In addition, aqueous humor dynamics and the effects of adenosine receptors on IOP are similar between mice and primates (Aihara M, et al. IOVS 2002;43:ARVO E-Abstract 3423). ^{32 33} The documented similarities between mice and humans in drainage structure anatomy, in functional responses to drugs that inhibit aqueous production and facilitate outflow, and in values for various outflow parameters indicate that mice represent suitable models for studying IOP. 

Factors Contributing to High IOP

Open-Angle Glaucomas

Primary open-angle glaucoma (POAG) is the most common subset of glaucoma. ³⁴ It is called open-angle, because the angle and drainage routes are clinically observed to be unimpeded. Only a small proportion of POAG is genetically accounted for, ^{19 35} and we have little understanding of the cell biology underlying it. The disease is affected by multiple interacting factors, and this complexity can confound studies (especially human studies, in which many factors cannot be controlled; see Refs. ¹⁹ , ²¹ ). For these reasons, animal experiments provide an important complement to human studies. Because of the ability to study effects of specific factors in defined genetic backgrounds and controlled environments, mouse studies are likely to be particularly valuable. ^{1 36 37} Thus, although mouse studies have barely contributed to our understanding of POAG to date, they are likely to become significantly more important in the future. Mouse models with potential relevance to POAG include C57BL/6J, which has mutations in Optn (a human glaucoma gene), and the Col1a1 mutant mouse (see Refs. ¹⁹ , ^{38 39 40 41} ). 

An excellent example of how in vivo mouse and human studies and cell culture systems complement each other is provided by research involving the human glaucoma gene MYOC. ⁴² MYOC mutations exist in approximately 3% of patients with late-onset POAG and a greater proportion of patients with juvenile open-angle glaucoma (JOAG; an earlier and more severe form of POAG). ^{43 44} Despite the importance of MYOC in glaucoma, its normal functions are controversial and its pathogenic mechanisms are just becoming clearer. ^{19 45} Available mouse and human data agree, in that null alleles in both species do not induce disease. ^{46 47 48} In addition, findings in our mouse studies have indicated that overproduction of MYOC has no consequence for IOP and glaucoma (Fig. 1) . ⁴⁹ Together, these data suggest that MYOC does not have a normal role in IOP homeostasis. Rather, they suggest that abnormal protein molecules are pathologic and necessary to induce disease. This is in agreement with most cell culture experiments, where expression of mutant MYOC proteins (corresponding to human mutations) results in abnormal protein accumulation inside cells and cell death (see Refs. ¹⁹ , ⁴⁹ ). Similarly, protein accumulations were detected in vivo and induced cataracts when a human mutation was expressed in the mouse lens. ⁵⁰ It should be noted that cell death can be prevented by conditions that promote protein folding and secretion. ⁵¹  

Despite this general agreement, caution is necessary when interpreting the mouse data. It remains possible that the studied mice were somehow protected from the glaucomatous effects of MYOC, although there is as yet no good evidence supporting this. If true, then it is conceivable that MYOC overproduction may be able to induce glaucoma in humans or in other mouse strains. The severity of MYOC-induced disease dramatically varies from person to person, and there is evidence to support involvement of modifier genes. ^{19 52} Modifier genes can determine whether mutations in a particular gene cause disease in an individual or can profoundly alter age of onset and rate of progression. The analysis of mice engineered to have human disease-associated mutations will determine whether there are species differences in the ability of MYOC to induce glaucoma. If there are, it will be possible to study the species differences. It then may become possible to use the resultant information for the benefit of patients. Alternatively, if differences exist between individual mouse strains, crosses between these strains may identify important modifier genes, again with the potential for helping patients. 

Developmental Glaucomas

Developmental glaucoma refers to the subset of glaucomas associated with anterior segment dysgenesis (ASD). ⁵³ Although these phenotypes often exhibit autosomal dominant or recessive inheritance, variable expressivity and incomplete penetrance point to a multifactorial etiology (see Ref. ²¹ ). Many of these conditions involve obvious dysgenesis of readily visible anterior chamber structures such as the iris and pupil. In others (e.g., primary congenital glaucoma [PCG]), the defects are subtle, involving abnormal development of Schlemm’s canal and the trabecular meshwork drainage structures of the iridocorneal angle. Although many patients with ASD have elevated IOP, the severity of clinically visible dysgenesis does not correlate with IOP, and the etiology of glaucoma development is not understood. Several ASD genes have been identified, and we recently reviewed the genes and relevant developmental pathways. ^{21 29} Therefore, although we have studied different ASD genes using mice, ^{54 55 56 57} I will only discuss the identification of a modifier gene here. 

PCG is a severe form of early-onset glaucoma. Many PCG cases are caused by recessive mutations in the CYP1B1 gene. ^{19 58} Striking phenotypic differences between individuals with CYP1B1 mutations suggests the effects of a modifier gene(s). ^{59 60} Motivated by these observations, we identified a modifier gene that alters the phenotype in Cyp1b1 mutant mice. ⁵⁶ These mice have focal angle abnormalities similar to those in patients with PCG but do not develop high IOP and glaucoma. Our modifier gene approach identified the tyrosinase gene (Tyr) as a modifier of angle dysgenesis. Cyp1b1-deficient mice completely deficient in Tyr had more severe angle malformations than Cyp1b1-deficient mice with functional Tyr (Fig. 2) . ⁵⁶ Tyr also modifies the phenotype in Foxc1-deficient mice, another gene with an orthologue that causes human glaucoma. Tyrosinase produces l-DOPA, and administration of l-DOPA in drinking water substantially alleviates the developmental abnormalities in Cyp1b1 mutant mice (Fig. 2) . ⁵⁶ This shows that l-DOPA is a critical molecule for normal angle development. 

These experiments raise the possibility that mutations in multiple genes contributing to developmental glaucomas affect DOPA levels. DOPA levels may be altered in the neural crest cells from which the angle structures and iris stroma derive. Tyrosine hydroxylase (TH) is the major enzyme responsible for producing DOPA from tyrosine. Remarkably, several genes that are known to cause anterior segment dysgenesis and glaucoma in humans and/or mice are known to promote either TH expression or the proliferation of TH-expressing neural crest cells during development (Fig. 2 ; see Ref. ²⁹ ). These findings demonstrate the utility of mice for defining multifactorial genetic interactions and for defining new pathways that are relevant to glaucoma. 

Pigmentary Glaucomas

Pigment Dispersion Syndrome.

Pigment dispersion syndrome (PDS) is a common condition that often progresses to pigmentary glaucoma (PG). ^{61 62 63} PDS involves focal iris pigment epithelial atrophy and dispersal of liberated pigment onto anterior chamber structures and into the ocular drainage structures. Hallmarks of PDS are accumulation of pigment on anterior chamber structures and a radial slitlike pattern of iris transillumination. Various mechanisms have been suggested to account for the pigment dispersion, including the iris’s rubbing against the zonules or lens, developmental defects, inherited disease of the pigment epithelium, and hypovascularity of the iris (see Refs. ^{61 62 63} ). Nevertheless, the exact mechanism(s) of pigment dispersion are not conclusively established (see Ref. ⁶⁴ ). At a minimum, the etiology of PDS is more complex than simply iris rubbing and likely involves additional risk factors. ^{61 63 64}  

Genetics of Pigment Dispersion.

Although autosomal dominant inheritance has been reported, most PDS cases do not exhibit clear Mendelian inheritance and they most likely have a multifactorial etiology (see Ref. ¹⁹ ). A few mouse strains are reported to have phenotypes relevant to pigmentary glaucoma and may help define mechanisms of disease. ^{65 66 67 68 69}  

In DBA/2J mice, a pigmentary form of glaucoma develops that is characterized by a pigment-liberating iris disease, increased IOP, and optic nerve degeneration. ^{67 68} The degree of pigment dispersion and iris destruction in DBA/2J mice is much greater than occurs in humans with PDS. This difference is explained by our discovery that DBA/2J mice harbor two mutant genes that can independently cause disease but, when inherited together, interact to cause the severe DBA/2J phenotype. ⁶⁷ The DBA/2J disease is induced by the b allele of tyrosinase-related protein 1 (Tyrp1 ^b ) and a stop codon mutation in the glycoprotein (transmembrane) nmb gene (Gpnmb ^R150X ). ^{66 67} Whereas mice homozygous for both Tyrp1 ^b and Gpnmb ^R150X exhibit development of the very severe DBA/2J disease, single homozygotes are more mildly affected and have distinct phenotypes. Mice homozygous for Tyrp1 ^b exhibit an iris stromal atrophy phenotype. In mice homozygous for Gpnmb ^R150X , an iris pigment dispersion phenotype develops that involves deterioration of the iris pigment epithelium (Fig. 3) . ^{67 68} The clinical phenotype in Gpnmb ^R150X/R150X mice includes radial slitlike transillumination defects that resemble those in humans with PDS/PG. ⁶⁷  

Melanosomal Toxicity.

The Tyrp1 gene encodes a melanosomal protein previously ascribed to have both enzymatic and structural functions. ⁷⁰ The Gpnmb gene encodes a heavily glycosylated protein that seems to be present in multiple cell types with lysosomal-related organelles, including melanosomes. ^{71 72} The finding that both of these gene encode melanosomal proteins suggests that melanosomal cytotoxicity ⁷³ may kill iris cells, resulting in pigment dispersion. A key function of melanosomes is to sequester the cytotoxic intermediates produced during melanin production. ^{74 75} It seems possible that mutant Tyrp1 ^b and Gpnmb ^R150X may somehow permit toxic intermediates of pigment production to leak from melanosomes (Fig. 3) . The leakage may contribute to cell death and iris stromal atrophy. This hypothesis surmises that pigment production plays an important role in disease etiology. That this is the case has been demonstrated by the fact that tyrosinase mutations that abrogate flow through the pigment production pathway prevent the diseases caused by both Tyrp1 ^b and Gpnmb ^R150X (Anderson M, John SWM, unpublished data, 2004). ⁶⁷ In case similar mechanisms are active in humans, it was necessary to assess the benefit of reducing, as opposed to abrogating, flow through this pathway. (Drug treatments would be likely to reduce but not abrogate flow.) To test this benefit, we used hypopigmentation mutations including the pearl (pe) mutation. These hypopigmentation mutations completely prevented the iris disease and glaucoma in DBA/2J mice (Fig. 3) . 

Melanosomes and Humans.

As discussed earlier, GPNMB and TYRP1 are transmembrane melanosomal proteins, and their mutation appears to promote iris disease by allowing leakage of toxic intermediates from melanosomes. The melanosomes of DBA/2J mice are structurally abnormal and many appear immature. ^{19 76 77} Suggesting human relevance, Rodrigues et al. ⁷⁸ found an increased number of immature-appearing melanosomes in the iris pigment epithelium of patients with pigmentary glaucoma. Abnormal melanosomes have also been observed by others. ⁷⁹ A recent report that PDS associates with hair color in a specific population ⁸⁰ also may suggest that the melanosomal basis of pigment dispersion in DBA/2J mice has relevance to human forms of the disease. However, this study should be extended. Together, these observations suggest that genes affecting melanosomal functions and/or the survival of melanosome-containing cells are reasonable candidates to assess in human patients. To help prioritize appropriate candidates and learn more about pathologic roles, we are aging mice with a variety of mutations that affect melanin biology. 

The Immune Genotype.

I have already discussed how two genes conspire to induce the DBA/2J iris disease via a mechanism of melanosmomal toxicity. Adding another layer of complexity, immune cell genotype also appears to be an important risk factor. In addition to melanosomes, Gpnmb is present in dendritic cells that control immune responses. ^{81 82} Dendritic cells are normally present in the iris and ocular outflow pathway, ^{83 84} which suggests the possibility that the Gpnmb mutation alters dendritic cell function(s) and promotes iris disease through ocular immune abnormalities. DBA/2J eyes have deficiencies in some aspects of immune privilege before dispersed pigment is evident. ⁸⁵ For example, they are deficient in anterior chamber associated immune deviation (ACAID; Fig. 4A ). ACAID is an active physiologic process that acts to suppress proinflammatory responses to antigens that are first detected in the eye. ⁸⁶ Although DBA/2J eyes lack clinical signs of overt inflammation, a chronic and mild form of inflammation attacks the iris (Figs. 4B 4C 4D) . ⁸⁵ Furthermore, reconstituting the immune system of DBA/2J mice with cells that express wild-type Gpnmb (through bone marrow transfer) substantially alleviates the pigment dispersion ⁸⁵ (Anderson M, John SWM, unpublished observations, 2004; Fig. 4E ). The mutant immune system is not sufficient to induce disease, however, in otherwise wild-type mice, (Fig. 4E) . This supports the need for both abnormal melanosomes and a susceptible immune genotype for iris disease propagation. Overall, our experiments suggest that DBA/2J iris damage is initiated by leakage of toxic molecules from melanosomes, and that abnormal immune suppression resulting from a susceptible immune genotype then allows an inflammatory response to propagate the melanosome-initiated disease (Fig. 4F) . 

The Immune Genotype and Humans.

Based on our findings in the mouse, it seems reasonable that mild subclinical inflammation and abnormal ocular immunity may contribute to human PDS. It is possible that at least some PDS represents the mild end of a spectrum of inflammatory diseases that attack the iris and cause glaucoma (including anterior uveitis, ⁸⁷ Behçet’s disease ⁸⁸ and Vogt-Koyanagi-Harada [VKH] disease ⁸⁹ ). There is no current evidence to support a role of inflammation in PDS, but the possibility has not really been considered. Pigment-containing macrophages are reported in histologic studies of eyes from patients with PDS ^{79 90 91} and may not be simply “cleaning up.” If immune genotype and inflammatory components are important in PDS susceptibility, factors affecting iris configuration and rubbing could cause the initial damage that primes the disease (either alone or by exacerbating the effects of melanosomal toxicity). Alternatively, but not exclusively, viral or other trauma could have the same effect. 

In many cases, PDS progresses to high IOP and causes pigmentary glaucoma. Perfusion of monkey eyes with uveal pigment only transiently elevates IOP, ⁹² however, and many patients with PDS do not progress to high IOP. Together, these findings suggest that factors in addition to direct obstruction by pigment are necessary to cause sustained elevation of IOP. It is likely that a genetic susceptibility of the drainage tissues to a pigment/cell debris-induced adverse reaction is needed for progression to pathologic IOP elevation. The molecular events are not well understood, and more knowledge is critical. As discussed earlier, mutations in Tyrp1 and Gpnmb induce a pigment dispersion phenotype in DBA/2J mice that progresses to high IOP and glaucoma. We have now introduced these mutations into a genetically different strain background. Even though the mice were housed under similar conditions, on this new strain background these genes induce the iris disease, but there is rarely progression to high IOP (Anderson M, Libby R, John SWM, unpublished data, 2004). This observation strongly suggests that genetic susceptibility factors determine the likelihood that pigment dispersion will progress to elevated IOP (Fig. 4F) . We are conducting experiments to identify relevant factors that differ between these strains. 

Apoptosis and IOP Elevation

Because of the loss of drainage structure cells in old individuals and at late stages of glaucoma, cell death has been speculated to contribute to common forms of human glaucoma. ^{93 94} There has been no clear evidence to indicate whether cell loss contributes to elevated IOP, however, or whether it is merely a secondary consequence of high IOP or treatment protocols. Thus, a primary role for cell death in elevation of IOP in human glaucoma has not been established. During recent experiments to assess the role of the proapoptotic molecule BAX in DBA/2J glaucoma, we found that complete BAX deficiency has a protective effect against elevation of IOP. Although elevated IOP and glaucoma developed in homozygous mutants, the elevation of IOP was limited, in that it tended to be less than in wild-type mice (heterozygotes were indistinguishable from wild type). ⁹⁵ These results suggest that the apoptotic death of cells affecting aqueous humor drainage can contribute to the elevation of IOP, at least in a secondary glaucoma in which the drainage structures are insulted by pigment and cell debris. Cell death was recently implicated in glaucomatous elevation of IOP through in vitro culture experiments (as discussed for MYOC earlier in the article ⁵¹ ). Together with our finding that complete BAX deficiency limits elevation of IOP in a glaucoma setting, these results support further investigation of apoptotic pathways and effects of antiapoptotic drugs on elevated IOP in human glaucoma. 

Factors Contributing to RGC Degeneration

Individual susceptibility factors are important in determining whether a person will have glaucoma and in determining the rate of progression and ultimate severity. In addition to factors directly affecting IOP, other factors that affect RGC injury and survival are likely to be important. Different insults are proposed to injure RGCs in glaucoma, including ischemia, excitotoxicity, axonal injury, glial activation, and autoimmunity. ^{96 97 98 99 100 101 102} Some combination of these and/or other insults may be active in any patient. The relative significance of each specific insult is likely to vary between patients, depending on the kinetics and magnitude of IOP elevation, the individual constellation of susceptibility factors in the optic nerve and retina, and the effects of environmental factors and lifestyle. We are largely using a genetic approach to understand important susceptibility factors for glaucomatous neurodegeneration. Genetic variations that alter intrinsic RGC susceptibility or that modify the nature and magnitude of damaging insults within the optic nerve and retina are important determinants of glaucoma susceptibility. ¹⁹  

Mice as Models of Glaucomatous RGC Degeneration

In addition to models in other species, there are now several published mouse models of pressure-induced RGC death. These include inherited ^{40 65 68 103 104 105} and experimentally induced ^{39 106 107} models. The specific degeneration pathway that is triggered depends on both the nature and magnitude of neuronal insult and on genetic context. Thus, the roles of specific molecules in RGC death may differ among individuals, different types of glaucoma, and different glaucoma models. It is essential, therefore, to study a variety of models. These models have different advantages and disadvantages, and useful knowledge will be derived from each. ¹⁰⁸  

Although age-related inherited models can be slower and more complex to work with compared with induced models, my group has mainly worked on developing such models. This focus has come about because these features describe most human glaucoma. The best characterized of the models we work on is the DBA/2J model of glaucomatous neurodegeneration. ^{68 109} The DBA/2J disease is characterized by hallmarks of glaucoma, including RGC death and atrophic excavation of the optic nerve (Fig. 5) . This inherited model also has progressive age-related elevation of IOP and subsequent variable and asynchronous degeneration of RGCs, similar to human glaucoma. In our colony, the RGCs are by far the most susceptible retinal cell type. In most DBA/2J eyes affected by severe glaucoma, there is no obvious loss of inner or outer nuclear layer neurons. In addition, despite the mechanisms contributing to their IOP elevation (discussed earlier), we have found no evidence of involvement of inflammatory cells in this RGC degeneration. 

Despite these similarities, it would be shortsighted to claim that DBA/2J glaucoma is a perfect model of human glaucoma or that the relative importance of specific neurodegenerative processes is the same. In fact, differences in factors that determine glaucoma susceptibility (and thus the relative importance of specific neurodegenerative pathways) almost certainly exist between humans and different model systems, as they do between individuals. Although studying models with different genetic constitutions will help, there may always be some important species differences between humans and mice (or rodents in general). 

Importance of the Lamina Cribrosa.

Although not demonstrated to affect the neurodegenerative mechanisms, species differences in optic nerve and retinal vasculature or in lamina cribrosa structure may alter the disease. The lamina cribrosa (LC) consists of a network of collagen, elastic tissue, and glial cells that is partly continuous with the sclera and through which the RGC axons run as they exit the eye. This network is located at the weak point in the ocular wall and has been implicated in glaucomatous RGC death. ¹¹⁰ There are known differences in LC architecture between mouse and rat strains and humans and possibly between different rat strains. ^{111 112 113 114} The mouse is sometime said to lack an LC because the appropriate region contains little extracellular matrix. At various times, we have indicated the presence of the mouse LC at the level of the sclera where the optic nerve exits the eye. Although this seems contradictory, the difference is simply due to differing definitions of what comprises the LC. Although essentially lacking extracellular matrix components, the mouse LC has a well-formed meshlike network of glial processes (densely packed with cytoskeletal filaments) through which the RGC axons pass (Smith R, John SWM, unpublished observations, 2004). ¹¹⁵ Thus, I suggest the use of the term cellular lamina to describe the mouse LC. 

The extracellular matrix (ECM) components of the LC are suggested to be important in RGC axon damage. Differences in the lamina cribrosa ECM may explain the increased susceptibility of axons located superiorly in the human optic nerve ¹¹⁶ and possibly contribute to individual differences in glaucoma susceptibility. ¹¹⁷ Therefore, species differences in the LC may alter aspects of the neurodegeneration and the relative importance of specific damaging mechanisms. Nevertheless, glaucoma develops in different mouse strains, and mouse axons located in the superior optic nerve have recently been reported to be the most susceptible to glaucoma (in a laser-induced mouse model). ¹¹⁸ Thus, it is also possible that the ECM of the LC does not have a central role in glaucoma. Alternatively, the ECM and/or glial elements of the LC may have differing levels of importance in different patients. 

Distinct Pathways Involved in the Degeneration of Somata and Axons in DBA/2J Glaucoma

Somata.

Although the distinctions are not always clear cut, neurons can degenerate through different processes, including apoptosis, necrosis, and autophagy. ^{119 120 121 122} Furthermore, it has become clear that, within a single neuron, distinct, spatially confined degeneration programs can be activated in different compartments (e.g., soma, axon). ^{123 124} These findings may be relevant to glaucoma. ¹²⁵ RGCs die by apoptosis in experimentally induced glaucoma and are expected to do so in inherited animal models and human glaucoma. ^{126 127 128 129 130 131} Supporting this, we have found that the proapoptotic molecule BAX is necessary for RGC death in inherited DBA/2J glaucoma. ⁹⁵ BAX deficiency completely stopped RGC somal degeneration in this model (Fig. 6) . Heterozygous Bax deficiency was also highly protective. Heterozygous deficiency models quantitative variation in BAX levels that may occur in the human population. Thus, our data suggest that BAX is a reasonable candidate gene to assess as a modulator of susceptibility to RGC death in human glaucoma. The heterozygous effect also encourages the testing of BAX inhibitors as neuroprotective agents, because these agents would be unlikely to abrogate BAX activity completely. The BAX deficient RGC somata survive for months in vivo without their axons in the DBA/2J model. ⁹⁵ Thus, assuming conservation of mechanisms, it is interesting to speculate that BAX inhibition may allow a patient’s somata to be maintained until future axon regenerative therapies are possible. 

Axons.

In glaucoma, it is not clear whether the same or different degeneration pathway(s) are activated in the soma and axon. In our experiments, somal and axonal degeneration were uncoupled. Although BAX deficiency protected the RGC soma, BAX is not necessary for axon degeneration in this model ⁹⁵ (Fig. 6) . Thus, different RGC compartments degenerate by different molecular processes (soma, BAX dependent; axon, BAX independent) in this inherited glaucoma. Because severe axon degeneration occurs in Bax ^−/− mice with no somal loss, these findings clearly demonstrate that axon degeneration is not a consequence of somal apoptosis in this glaucoma. It is not yet clear whether the somal and axonal pathways share common features or are completely distinct. 

Although BAX deficiency did not stop axon degeneration in the DBA/2J model, it did delay optic nerve damage. ⁹⁵ Optic nerve axon degeneration was delayed in both heterozygous and homozygous mutant mice. Finally, because axon degeneration was delayed in heterozygous mice that had indistinguishable IOP compared to wild-type mice, these data suggest that protecting the soma has a beneficial effect on the axons. This notion is consistent with a model in which somal stress contributes at least in part to the demise of the axon. Overall, these data suggest that BAX inhibitors may be able to delay optic nerve axon degeneration. Regardless, when designing therapeutic strategies for human glaucoma, our studies suggest that both somal and axonal degeneration pathways may have to be considered. 

Support for a Role of Direct Axon Injury in DBA/2J Glaucoma Determined by an Integrated Approach

The BAX deficient DBA/2J mice provide a resource for investigating the insults that lead to induction of cell death in an inherited glaucoma. Our integrated approach compares the effects of genetically altering specific pathways on glaucomatous RGC death in DBA/2J mice and on experimental RGC death induced by known insults. An important feature of this approach is that all comparisons are made on the same uniform DBA/2J genetic background. This method allows direct comparison of RGC death induced by distinct insults to the naturally progressing glaucoma without any confounding effects of differing genetic risk factors. Ultimately, this approach must be applied to a panel of genes that affect different processes, but BAX is a good start. The approach should provide valuable insights about the nature and location of insults that kill RGCs in this inherited glaucoma. 

Different glaucoma hypotheses invoke different insults to the RGCs, including direct axon injury and somal excitotoxic injury. ^{98 132} Although the weight of data support a role for direct axon injury, it is not completely clear whether insults to the axon, soma, or both contribute to the degeneration. In the hypothesis of direct optic nerve and axon damage, high pressure places stress on the optic nerve and RGC axons as the axons exit the eye at the weak point in the ocular wall through the LC. Reports that the first damage to RGCs is evident in the axon segment near the LC in the optic nerve head ^{110 133} strongly support the axon as the first site of IOP-induced insult (see Ref. ⁹⁸ ). However, they do not definitively prove that this is so. Although these findings show local axonal abnormalities, the occurrence of initial damage in this region does not conclusively indicate that this is necessarily the first or only site of neuronal insult. Because of the optic nerve head architecture and the stress at the LC, it is conceivable that the axon segment at the LC may take substantial resources to be maintained, especially when IOP is elevated. Thus, somal stress may decrease available resources for axon maintenance/repair and this may be first manifest as defects in the axon segment near the LC. 

We used the BAX-deficient DBA/2J mice in further assessment of these excitotoxic and direct axon injury models. Acute experimental procedures are used to mimic these insults. Intraocular N-methyl-d-aspartate (NMDA) injection is used to model excitotoxic RGC insult, and controlled optic nerve crush is used to mimic direct axon insult. ^{134 135} To assess the roles of these distinct insults in spontaneous glaucoma, we subjected preglaucomatous DBA/2J mice of different Bax genotypes to these procedures. ⁹⁵ Unlike the glaucoma, the Bax genotype had no effect on RGC death initiated by the excitotoxin NMDA. Although these experiments do not rule out a role for an endogenous form of excitotoxicity in glaucoma, they do not support a primary role for NMDA-receptor–mediated events. In contrast, both complete and partial deficiencies of BAX were profoundly protective against optic nerve crush (Fig. 6) . The remarkably similar protective effect of decreasing functional Bax gene dosage on glaucomatous RGC death and on the axonal injury-induced RGC death further supports models of glaucoma involving direct axonal injury. 

Radiation with Syngeneic Bone Marrow Transfer Prevents RGC Loss in DBA/2J Glaucoma

To end, I will very briefly discuss a treatment that has a profound neuroprotective effect against DBA/2J glaucoma. We serendipitously uncovered this effect during the course of experiments to understand the role of bone marrow genotype in the pigment-dispersion phenotype (discussed earlier). When DBA/2J mice receive high-dose irradiation accompanied by syngeneic bone marrow administration at 5 to 8 weeks of age, they are protected from glaucoma out to the oldest age assessed (14 months). The treatment has no detected effect on the iris disease or IOP. ¹³⁶ Glaucoma is prevented in most animals (Fig. 7) . Therefore, a single treatment at 5 to 8 weeks of age has long-lasting benefit and places the animals in a state that is resistant to neurodegeneration until they are at least 14 months of age. Because of the timing of IOP elevation in DBA/2J mice, the neurodegeneration is delayed at least for 3 to 5 months after the neurodegenerative signals arise. 

We are not aware of any other neuroprotective effects this substantial. Because bone marrow genotype is not changed, it is not the causative agent. A recent associative study, suggests a protective effect of high-dose radiation without bone marrow transfer in a human population. In an epidemiologic study of atom bomb survivors, radiation appeared to protect against glaucoma. ¹³⁷ The glaucoma was self-reported and so more detailed examinations are needed. Together with our experiments, however, these data suggest that high-dose radiation is neuroprotective and that our finding may not be unique to DBA/2J mice. Considerable effort is needed to understand this effect and a variety of possible mechanisms must be assessed. ¹³⁶ It is important to determine whether this treatment protects against other forms of glaucoma. With increased understanding, it may be possible to eliminate the need for radiation but maintain the beneficial effect. In conclusion, this discovery provides an exciting new tool for research into mechanisms of neuroprotection in glaucoma and possibly other diseases. 

 Supported in part by National Eye Institute Grant EY11721 and National Cancer Institute Grant CA34196, American Health Assistance Foundation Grant AHAF 97437 and Grant G1999023 from the National Institute of General Medical Sciences. SWMJ is an Investigator of the Howard Hughes Medical Institute.

 Disclosure: S.W.M. John, None

 

Figure 1.

 

IOP and angle morphology. (A) There is an almost twofold difference in IOP between genetically different mouse strains housed in the same environment. (B) Normal mouse iridocorneal angle. SC, Schlemm’s canal; TM, trabecular meshwork; AC, anterior chamber; I, iris. (C) Western blot for tissue enriched in ocular drainage structures. Transgenic Myoc Tg/Tg express approximately 15 times the normal amount of MYOC protein. (D) Despite the increased levels of MYOC, Tg/Tg mice do not have elevated IOP (mean ± SEM). (A) Reproduced from Savinova OV, Sugiyama F, Martin JE, et al. IOP in genetically distinct mice: an update and strain survey. BMC Genet. 2001;2:12. (B) Reproduced from Libby RT, Smith RS, Savinova OV, et al. Modification of ocular defects in mouse developmental glaucoma models by tyrosinase. Science. 2003;299:1578–1581. (C, D) Reproduced, with permission, from Gould DB, Miceli-Libby L, Savinova OV, et al. Genetically increasing Myoc expression supports a necessary pathologic role of abnormal proteins in glaucoma. Mol Cell Biol. 2004;24:9019–9025. © American Society of Microbiology.

Figure 2.

 

Tyrosinase and DOPA modify angle development. Eyes of Cyp1b1 ^−/− mice have obvious angle abnormalities (compare with Fig. 1B ). (A) In a pigmented eye (Tyr ^+/+), the trabecular meshwork (TM) is hypoplastic, and material resembling Descemet’s membrane covers a portion of the TM (arrowheads). (B) In an albino eye (Tyr ^−/−), a very small Schlemm’s canal may be present (arrowhead). There is no TM and the iris is attached to the cornea ( ). (C) l-DOPA administration throughout development profoundly alleviates the developmental abnormalities in Cyp1b1 ^−/− mice. The higher the severity grade, the more severe and the more extensive the abnormalities were. (D) Several genes that cause anterior segment dysgenesis and/or glaucoma possibly modulate DOPA levels during ocular development. (A–C) Reproduced from Libby RT, Smith RS, Savinova OV, et al. Modification of ocular defects in mouse developmental glaucoma models by tyrosinase. Science. 2003;299:1578–1581. © American Association for the Advancement of Science. (D) Reproduced, with permission, from Gould DB, Smith RS, John SWM. Anterior segment development relevant to glaucoma. Int J Dev Biol. 2004;48:1015–1029. © UBC Press.

Figure 3.

 

Melanosomes contribute to DBA/2J pigment dispersion. (A–D) Mutations in genes encoding the melanosomal proteins TYRP1 and GPNMB cause iris disease. DBA/2J mice are homozygous mutant for both genes and have the most severe disease (as shown in D). All images are from 24-month-old mice. (E, F) Potentially toxic intermediates of melanin production (shaded) are largely sequestered inside melanosomes and are converted into pigment (E). The Tyrp1 and Gpnmb mutations appear to allow toxic molecules to escape from the melanosomes and damage iris cells (F). (G–J) Heterozygotes for the pe mutation are not hypopigmented and have typical DBA/2J disease (G, I). In contrast, homozygosity for this recessive hypopigmentation mutation prevents iris disease (H) and glaucomatous optic nerve excavation (J). (A–D, G–J) Reproduced from Anderson MG, Smith RS, Hawes NL, et al. Mutations in genes encoding melanosomal proteins cause pigmentary glaucoma in DBA/2J mice. Nat Genet. The Nature Publishing Group 2002;30:81–85 and Chang B, Smith RS, Hawes NL, et al. Interacting loci cause severe iris atrophy and glaucoma in DBA/2J mice. Nat Genet. 1999;21:405–409; http://www.nature.com/naturegenetics/. © The Nature Publishing Group.

Figure 4.

 

Evidence of an immune contribution to DBA/2J pigment dispersion. (A) DBA/2J mice were initially injected with an antigen (OVA), as shown, and subsequently challenged by antigen administration to the ear. The negative controls were naive mice that did not receive an initial antigen injection. The positive controls were initially injected subcutaneously. Mice that had an initial anterior chamber injection (OVA-AC) exhibited robust ear swelling on subsequent challenge. This indicates that ACAID was not functioning in these mice. (B–D) Inflammatory cells in the iris. At 3 months, the iris is normal (arrow) with no margination of leukocytes in the vessels. At 6 months, neutrophils (arrow) and macrophages (arrowhead) are present. In a severe case at 6.5 months, the peripupillary border of the iris was thickened and contained pigment-bearing macrophages (D, arrowhead) and neutrophils (D, arrow). (E) Bone marrow genotype has an important effect on the phenotype. (F). Summary of experimental findings implicating both melanosomal and immune components in the pigment dispersion. (A–D and photographs in E) Reproduced, with permission, from Mo SJ, Anderson MG, Gregory M, et al. By altering ocular immune privilege, bone marrow-derived cells pathogenically contribute to DBA/2J pigmentary glaucoma. J Exp Med. 2003;10:1335–1344. © Rockefeller University Press.

Figure 5.

 

Glaucoma in DBA/2J mice. (A–D) In many DBA/2J mice, severe glaucoma develops, characterized by profound axon loss (B) and excavation of the optic nerve (D, ). Filled arrowheads demonstrate the thickness of the nerve fiber layer. The cellular lamina is partially evident in (C) (open arrowheads). It appears to extend over only part of the nerve, because of the 1.5 μm section thickness. (E, F) Retinas from the eyes shown in (C) and (D), respectively. There is an obvious loss of RGCs in the eye with severe glaucoma (D, F), but other retinal layers appear unaffected. This is true of most mice in our colony. S, sclera; V, blood vessel; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer.

Figure 6.

 

Effects of BAX deficiency. (A–E) Optic nerve cross sections and retinal flatmounts from mice of the indicated Bax genotypes. (A–E, insets) Sections from young preglaucomatous mice of the same genotype. Except for the insets, all retinal flatmounts are from eyes that had very severe glaucoma (as determined by massive axon loss and optic nerve damage similar to that shown in C). It is very clear that both homozygous and heterozygous BAX deficiencies save RGC somata (D and E compared with B) but do not save the RGC axons (compare C with A). (F) A summary of ganglion cell counts in mice of each Bax genotype that were subjected to the indicated insults. Glauc, inherited glaucoma; Crush, optic nerve crush; NMDA, intraocular NMDA injection. Modified from Libby RT, Li Y, Savinova OV, et al. Susceptibility to neurodegeneration in glaucoma is modified by Bax gene dosage. PLoS Genet. In press. © Public Library of Science.

Figure 7.

 

Protective effect of radiation and bone marrow transfer. (A–D) Typical optic nerves and retinas from aged mice of the indicated treatment groups. (E) There was no detectable difference in IOP between the groups (mean ± SEM; P > 0.3). The thickness of the horizontal bar represents the mean IOP ± SEM for young preglaucomatous DBA/2J mice. (F) The treatment prevented glaucoma in most of the mice. Reproduced from Anderson MG, Libby RT, Gould DB, Smith RS, John SWM. High-dose radiation with bone marrow transfer prevents neurodegeneration in an inherited glaucoma. Proc Natl Acad Sci USA. 2005;102:4566–4571. © The National Academy of Sciences.

It is a great honor to receive the Cogan Award, and I am indebted to my family, mentors, colleagues, friends, and everyone who has helped and guided me throughout the years. I owe a special thank you to everyone who has worked in my laboratory. Your efforts resulted in this award and allowed us to make new discoveries. I especially thank Richard Smith who is a wonderful colleague and friend, my postdoctoral fellows Michael Anderson, Douglas Gould, and Richard Libby, who were responsible for much of the work I discussed, and Robert Ritch for his overflowing enthusiasm and open mind. I thank my former teachers and mentors for nurturing and expanding my interests: Jerome Den Hollander, Dennis Phillips, John Rees, Rima Rozen, Lesley Southerns, Charles Scriver, and Oliver Smithies; Abbot Clark and Robert Nickells for frequent discussions and collaborations; the organizers and faculty of the National Eye Institute–funded “Fundamental Issues in Vision Research” Course at Woods Hole; and individuals who provided encouragement and support as I began to consider working on glaucoma and who have encouraged and supported me during this time (alphabetically): Ruben Adler, Lee Alward, Ben Barres, David Beebe, Joseph and Barbara Cohen, Muriel Davisson, John Dowling, Thomas Freddo, Sharon Freedman, Achim Gossler, Brigid Hogan, Paul Kaufman, Barbara Knowles, Lesley Kozak, Bruce Ksander, Eric Lander, Edward Leiter, Brian Link, Richard Masland, Larry Mobraaten, James Morgan, John Morrison, Robert Nussbaum, Timothy O’Brien, Ken Paigen, David Papermaster, Elio Raviola, Thomas Roderick, Derry Roopenian, Carla Shatz, John Schimenti, David Serreze, Val Scott, Leonard Schultz, Val Sheffield, William Sly, the late J. Wayne Streilein, John Sundberg, Martin Wax, Alan Whitmore, Janey Wiggs, Cookie Willems, and Don Zack. I acknowledge my collaborators, colleagues at The Jackson Laboratory and The Howard Hughes Medical Institute, and the help and support of Norman Hawes and the “eye group” headed by John Heckenlively and Bo Chang. In addition, I thank my assistant Felicia Farley and The Jackson Laboratory Scientific/Administrative Services, including Jennifer Torrance who prepared the figures. It is not possible to list everyone, but I thank you all.