(A–C) A double-labeling assay that identifies fragmented DNA using fluorescently labeled dUTP (A) and detects chromatin condensation by binding of the dye YOYO-1 (B) was used to assess the presence of these hallmarks of apoptosis in glaucomatous DBA/2J eyes at 10–11 mo of age (a time when many RGCs die). A cell in the retinal ganglion cell layer (GCL, arrowhead) has both of these features of apoptosis as indicated by double labeling (C). INL, inner nuclear layer.
(D–F) Electron microscopy provided further evidence for apoptosis. (D) An example of a healthy RGC. (E) Chromatin condensation (a hallmark of apoptosis) along the inner surface of the nuclear envelope in a ganglion cell (arrows). The internal limiting membrane of the retina is indicated by arrowheads. (F) An apoptotic body in the ganglion cell layer (arrows) containing a nuclear fragment with prominent condensed chromatin (asterisk) and other cell remnants.
(G) A TUNEL assay (see Materials and Methods) was used to assess the prevalence of cell death at different ages. TUNEL labeling was not detected at 7 mo (an age prior to glaucomatous cell death) and peaked at 10–13 mo, when most RGCs die. No TUNEL-positive cells were detected in nonglaucomatous, age-matched control mice. These results support an important role of apoptosis in RGC death in spontaneous glaucoma. Scale bar, 1 μm.

To assess the effects of Bax deficiency on optic nerve degeneration, we analyzed PPD-stained optic nerve cross sections from Bax^+/+ and Bax^−/− mice (n > 49 for each genotype; see Materials and Methods).
(A and B) Before the DBA/2J glaucoma damages RGCs, the optic nerves of both Bax^+/+ (A) and Bax^−/− mice (B) had a normal organization. The axons appeared healthy with a clear axoplasm and darkly stained myelin sheath.
(C and D) BAX deficiency did not prevent glaucomatous optic nerve damage. Severe degeneration involving extensive to complete axon loss and scarring occurred in both Bax^+/+ (C) and Bax^−/− mice (D). The majority of mice of both genotypes had this severe degree of damage by 12 mo of age. These experiments show that BAX is not required for glaucomatous axon degeneration. Scale bar, 50 μm.

To determine the effects of BAX deficiency on RGC death in glaucoma, we analyzed RGC layer cells at stages with and without glaucomatous optic nerve damage (see Materials and Methods). All shown images are from a similar region of the superior, peripheral retina.
(A and B) In both Bax^+/+ (A) and Bax^−/− (B) mice without glaucomatous optic nerve damage, the retinas appear healthy. The retinas of both genotypes are similar except that Bax^−/− mice have extra RGCs (since BAX is important in normal developmental RGC death [39]).
(C and D) In contrast, an obvious difference was evident between the retinas of Bax^+/+ and Bax^−/− mice that had all suffered severe glaucomatous damage with 95% or more axon degeneration. As expected, for Bax^+/+ retinas (C) from eyes with 95% or more optic nerve axon loss, there was a noticeable decrease in RGC layer cells (compare [C] to [A]). In contrast, retinas from Bax^−/− mice with correspondingly damaged optic nerves (D) had suffered no obvious loss of RGC layer cells (compare [D] to [B]). This suggests that BAX is required for RGC death in DBA/2J glaucoma. As is well established for both RGCs and other neurons, Bax^−/− RGCs that survive without axons have a shrunken morphology [38,82]. This is clearly evident in the Bax^−/− glaucomatous mice (D).
(E) RGC layer cell counts for eyes with 95% or more axon degeneration confirmed that BAX is necessary for RGC death in this glaucoma. To allow comparison between genotypes, the percent of surviving cells is shown (% soma in mice with 95% or more axon loss compared to mice of the same genotype without glaucomatous damage). At 12 mo of age, Bax^+/+ mice had 61.4% ± 3.8% of their RGC layer cells remaining, Bax^−/− mice had no appreciable cell loss (101% ± 5.3%). The RGC layer cells of Bax^+/− mice were also protected (89.2% ± 5.7%). The p values comparing differences in cell counts between nonglaucomatous and very severely glaucomatous (≥ 95% axon loss) eyes of the same genotype were: Bax^+/+, p < 0.001; Bax^+/−, p = 0.207; Bax^−/−, p = 0.426. No cell loss was seen in Bax^−/− mice even out to 18 mo (94.8% ± 4.4% cells surviving, p = 0.524 compared to nonglaucomatous Bax^−/− mice). These findings show that BAX gene dosage has an important effect on the susceptibility of RGCs to glaucomatous death. Scale bar, 50 μm.

To assess the possibility that Bax deficiency may delay axon degeneration by lessening the glaucomatous insult to which RGCs are exposed, we analyzed IOP at key ages of IOP elevation. (A) The average IOP for each genotype (± SEM) and (B) the actual IOP values recorded. Bax deficiency did not prevent IOP elevation. At both 9 mo and 10.5 mo, the average IOP of both Bax^+/− and Bax^−/− mice was significantly elevated compared to preglaucomatous DBA/2J mice (p < 0.001). The width of the horizontal line (black in [A], gray in [B]) represents the mean IOP ± SEM of a group of wild-type preglaucomatous DBA/2J mice that were 3 mo old. The degree of IOP elevation, however, was altered in Bax^−/− mice. In both 9- and 10.5-mo-old Bax^−/− mice, the average IOP was less than that of Bax^+/+ and Bax^+/− mice. This reduction in IOP elevation was significant at 10.5 mo (p < 0.01). The IOP of Bax^+/− mice did not differ from wild type at either 9 or 10.5 mo (p > 0.82). By 12 mo, there was no difference in IOP between mice of any genotype (p > 0.25).

BAX deficiency did not prevent glaucomatous axon degeneration. Nevertheless, the beneficial effect on RGC survival raised the possibility that it may have a protective effect on the axon and delay optic nerve damage. To assess this, three investigators masked to genotype determined the severity of optic nerve damage for mice of each Bax genotype at each age (n = 49–71 per genotype at each age; see Materials and Methods). At 10.5 mo, both Bax^+/− and Bax^−/− mice had significantly less optic nerve damage than Bax^+/+ mice. Mild damage was evident in 53% of Bax^−/− and 44% of Bax^+/− optic nerves, compared to only 20% of Bax^+/+ (p < 0.001 for both Bax^+/− and Bax^−/− compared to Bax^+/+; Chi² test). With disease progression to 12 mo of age, the distribution of optic nerve damage became indistinguishable among mice of different Bax genotypes (p > 0.10). Quantitative assessment of a random subset of nerves assigned each damage level (more than eight of each) demonstrates that the number of axons that remain in optic nerves having each damage level are clearly different (see Materials and Methods).

To help distinguish between the likely roles of mechanical axon insult and excitotoxicity in cell death induction in spontaneous glaucoma, we subjected preglaucomatous DBA/2J mice of each Bax genotype to either controlled optic nerve crush or NMDA-mediated excitotoxicity. For controlled crush and NMDA, the percent RGC survival in the manipulated eye compared to the contralateral control eye is shown. For ease of comparison, the data for glaucomatous damage are the same as shown in Figure 3. In contrast to the spontaneous glaucoma, NMDA-mediated RGC death is not dependent on BAX, as evident by the complete lack of protection from death in Bax^−/− mice. As for the spontaneous glaucoma, RGC death induced by controlled optic nerve crush was completely dependent on BAX and prevented in both Bax^+/− and Bax^−/− mice. Overall, the effects of BAX in the face of spontaneous glaucoma and controlled crush were remarkably similar.