310
(CaN), which promotes apoptosis, occurs in two rodent models of glaucoma (induced rat model and DBA/2J mice), but not after optic nerve crush (Huang et al., 2005). Finally, heterozygosity for a Bax null allele protects RGC somas from apoptosis after a glaucomatous insult that causes axonal degeneration, but not after crush (Libby et al., 2005b).
Recently, Yang and colleagues performed a microarray analysis between RGC death induced by optic nerve transection and pressure-induced neurodegeneration (translimbal laser photocoagulation) (Yang et al., 2007). By comparing gene expression changes following crush and during experimental glaucoma and then filtering out genes that changed in both models, they report gene expression changes that are suggested to be specific to pressure-induced injury (Yang et al., 2007).
Mouse models to characterize processes involved in glaucomatous neurodegeneration
The death of RGCs and the associated degeneration of the optic nerve are unifying features of glaucoma, but the underlying mechanisms are poorly understood. In recent years, the DBA/2J strain has become the most widely used inherited mouse model to study glaucomatous neurodegeneration. In this section, we describe how the DBA/2J model (and the related strain, DBA/2NNia, collectively referred to as DBA/2) is being used to understand mechanisms involved in neurodegeneration during glaucoma.
Similar patterns of glaucomatous damage occur in humans and mice
A characteristic feature of human glaucoma is the occurrence of focal visual defects due to regionspecific loss or impairment of RGCs. The most consistent regional defects are arcuate scotomas, which are detected by visual field tests (Shields, 1997). The arcuate nerve fibers originate in the temporal region of the retina and arch above or below the fovea to the optic nerve head. In the mouse, the RGC axons do not curve across the retinal surface but radiate straight toward the optic nerve. Considering this, patterns of regional
damage equivalent to that in human patients occur in mice. In DBA/2 mice, RGCs and their axons are lost in ‘‘fan-shaped’’ or ‘‘patchy’’ regions (Danias et al., 2003; Jakobs et al., 2005; Schlamp et al., 2006). These fan-shaped regions of RGC loss are likely to be analogous to the arcuate scotomas seen in human glaucoma. This regional pattern of RGC loss during glaucoma is likely to result from focal damage to discrete’’bundles’’ of axons within the lamina.
The lamina cribrosa is an important site of early glaucomatous damage
Important studies in humans and primates established that early glaucomatous damage affects the RGC axon segments within the lamina cribrosa (Anderson and Hendrickson, 1974, 1977; Quigley and Anderson, 1977; Quigley et al., 1979, 1980, 1981, 1983; Quigley and Addicks, 1980, 1981). In humans and primates, the lamina cribrosa is composed of plates of extracellular matrix (ECM) that provide support for the axons as they pass through the posterior wall of the eye (Fig. 2a). It was hypothesized that in response to elevated IOP, bowing of the ECM plates would damage axon bundles by mechanical stress. Mechanical distortion of the ECM plates has also been suggested to contribute to glaucoma by damaging blood vessels (Quigley and Addicks, 1981; Maumenee, 1983; Fechtner and Weinreb, 1994). The ECM plates are covered by astrocytes that provide neurotrophic and other forms of support to the neurons.
Historically, the mouse was not regarded as a useful model for glaucoma because it was reported to lack a lamina cribrosa (Tansley, 1956; Fujita et al., 2000; Morcos and Chan-Ling, 2000; May and Lutjen-Drecoll, 2002). However, the mouse does have an astrocyte-rich structure in the same position as the human lamina cribrosa (Xie et al., 2005; Petros et al., 2006; Schlamp et al., 2006; Howell et al., 2007). The astrocytes of this region form an enmeshing network of glial cells through which the RGC axons pass, and they are intimately associated with the axons. There is no evidence of collagenous ECM plates in this region of the mouse optic nerve. To reflect the equivalent location compared to the human lamina cribrosa,
311
Fig. 2. (A–C) The lamina cribrosa has robust ECM plates (A, shown in blue, Masson’s Trichrome) through which bundles of axons pass. In contrast, the mouse glial lamina has no collagenous plates (B), although collagens are clearly visible in blood vessel walls (arrowheads). Similar to the human lamina, the mouse glial lamina has an extensive meshwork of astrocytes (C, stained positive for GFAP). (D–F) Early focal damage at the glial lamina is visualized using DBA/2J.Thy1-CFP mice. In these mice, RGC axons appear green when viewed by confocal microscopy. Because axonal contents accumulate in regions of damage, early axon damage is evident as very brightly fluorescent axon segments (arrows). (D) No damage is seen in preglaucomatous young eyes. (E) Obvious axonal swellings were evident specifically in the lamina of eyes that were at early stages of glaucoma. (F) Focal regions of damage are clearly visible as bright, slightly swollen axonal regions (arrowheads). Scale bars: (A) 1 mm; (B, C) 50 mm; (D–F) 20 mm. (A) Adapted with permission from Karim et al. (2004). (B–F) Adapted with permission from Howell et al. (2007). (See Color Plate 22.2 in color plate section.)
the similar arrangement of glial cells, but the absence of ECM plates, this region of the mouse optic nerve has been termed the glial lamina (Howell et al., 2007).
An insult occurs to the axons of RGCs within the lamina in glaucoma
The mechanisms by which RGC somata die are different from those involved in axon degeneration. This was shown by genetically ablating the function of BAX, a proapoptotic molecule, in DBA/2J mice. BAX deficiency prevented the death of essentially all RGC soma, but the axons of these mice were still degenerated (Libby et al., 2005b). Therefore, in these mice at least, RGC soma death is a BAX-dependent apoptotic process, whereas axon degeneration is a BAX-independent process.
Similar to studies in other species (see above), we determined that the first sign of axon damage occurs within the lamina in DBA/2J mice (Fig. 2) (Howell et al., 2007). Demonstrating that the first
signs of axon damage occur within the lamina is not proof that axons are insulted within the lamina. In the general case, it is well established that the first site of neuronal damage may be remote from the site of insult (reviewed in Conforti et al., 2007). For example, in transected motor axons, neuromuscular junctions that are many centimeters from the lesion degenerate first. The axons immediately adjacent to the lesion remain intact for two to three times longer than the distant terminals (Beirowski et al., 2005). We have demonstrated that RGC axons survive up to, but not into, the lamina in BAX-deficient mice that retain all of their RGC soma. This provides strong experimental evidence supporting a direct insult to RGC axons within the lamina during glaucoma (Guo et al., 2007; Howell et al., 2007). Given that DBA/2J mice develop glaucoma with regional cell death and characteristic optic nerve excavation, the lack of ECM plates in the lamina suggests mechanical distortion of ECM plates is not necessary to damage axons within the lamina
312
during glaucoma. Although it remains possible that the ECM plates may modulate damage in human glaucoma (as the nerve is much larger), these findings strongly focus attention on other components of the lamina.
What is the nature of the insult at the lamina?
One current challenge is to identify mechanism(s) initiating axon damage at the lamina and not just the processes involved in the propagation of axon degeneration. This is of particular importance when considering improved human therapies. It is possible that the initiating events are either intrinsic to the RGC itself or extrinsic involving other cell types. Other important cell types that are present in the optic nerve are glial cells (such as astrocytes, microglia, and oligodendrocytes). These cell types ordinarily play important roles in maintaining healthy, fully functioning RGCs, but in response to stressful conditions can become harmful. Examples of pathways/molecules known to be active in these different cell types during glaucomatous injury are given below to highlight the possible importance of all of these cell types in glaucomatous neurodegeneration.
Intrinsic changes in RGCs leading to axon degeneration in glaucoma are reported. An interesting line of research relates to amyloid precursor protein (APP) and amyloid-b (Ab). Mutations in APP cause Alzheimer’s disease (AD), and recent studies imply that there is a significantly higher incidence of glaucoma among patients with AD (Bayer et al., 2002). The same abnormal Ab peptide found in AD was found in AqH of 40% of the assessed glaucoma patients. In addition, in a rat model of glaucoma, caspase-3, a major activator of the apoptotic cascade, is activated in RGCs and cleaves APP to produce neurotoxic fragments that include Ab (McKinnon et al., 2002; McKinnon, 2003). Supporting this, Goldblum and colleagues have shown that APP and Ab increase in the RGC layer, lamina, and pia/dura layer in aged DBA/2 mice (15 months old) compared to young DBA/2J and C57BL/6J controls (Goldblum et al., 2007). They suggest a disruption of the homeostatic properties of secreted APP with consecutive Ab cytotoxicity as a contributing
factor to RGC loss in glaucoma. This would suggest that glaucoma and AD share common features, and suggests mechanisms identified as important in one disease should be investigated in the other. A promising study suggests that targeting different components of Ab formation and aggregation pathway can effectively reduce glaucomatous RGC apoptosis as shown in an experimental glaucoma in rats (Guo et al., 2007).
Astrocytes form a cellular network in the glial lamina in the mouse. It has been suggested that IOP elevation can alter astrocytes so that they damage RGC axons (Hernandez, 2000). Astrocytes become reactive in glaucoma. An increase in glial fibrillary acidic protein (GFAP) is considered a hallmark of reactive astrocytes (Pekny and Nilsson, 2005). GFAP is upregulated in experimental models of glaucoma in primates and rats (Tanihara et al., 1997; Wang et al., 2000), as well as in human glaucomatous eyes (Tezel et al., 2003). Two independent microarray experiments, the first in an experimentally induced rat model, the second using DBA/2J mice, showed increases in the expression of astro- cyte-related genes in response to elevated IOP (Ahmed et al., 2004; Steele et al., 2006), and we have made similar observations (unpublished data).
Microglia are another cell type present in the optic nerve and may contribute to glaucoma (Tezel and Wax, 2004; May and Mittag, 2006; Nakazawa et al., 2006). Microglia increase in numbers as glaucoma progresses, and this is true in DBA/2J mice (Inman and Horner, 2007). It is possible that microglia are necessary to initiate or propagate damage in the retina or optic nerve. Individual microglia may contribute to highly local insults within specific regions of the lamina and could conceivably underlie the fan-shaped patterns of cell death.
Oligodendrocytes were recently suggested to participate in glaucoma (Nakazawa et al., 2006). Oligodendrocytes are numerous in the optic nerve from the start of the myelinated portion (approximately 100 mm behind the glial lamina in mice). Axons within the glial lamina are unmyelinated and no myelin-producing oligodendrocytes have been shown to be present within the glial lamina. Based on findings using an induced model of high IOP, an intriguing model involving TNFa and oligodendrocyte death was suggested to damage RGCs in