Caspase-Independent Component of Retinal Ganglion Cell Death, In Vitro

Abstract

purpose. Although in vitro and in vivo models demonstrate caspase activation in retinal ganglion cells (RGCs) undergoing apoptosis, the caspase-independent component of RGC death is unclear. Identification of the precise mechanisms of cell death in these distinct neurons is essential for the development of effective neuroprotective strategies in glaucoma. Because TNF-α and hypoxia have been implicated in RGC death during glaucomatous optic nerve degeneration, this study was conducted to determine whether RGCs survive exposure to TNF-α or hypoxia in the presence of caspase inhibitor treatment, and whether mitochondrial dysfunction is involved in RGC death induced by these glaucomatous stimuli.

methods. Primary cultures of rat RGCs were exposed to TNF-α or hypoxia for up to 48 hours. The temporal relationship of RGC death with the loss of mitochondrial membrane potential and the release of cell death mediators, including cytochrome c and apoptosis-inducing factor (AIF), was studied in the absence and presence of specific inhibitors of caspases. In addition, treatment with a free-radical scavenger, 4-hydroxytetramethylpiperidine-1-oxyl (tempol; 5 mM), was used in some experiments. Cell viability was assessed using calcein assay, and annexin V binding combined with propidium iodide staining was used for the distinction of apoptotic and necrotic cells. Caspase-3–like protease activity was measured using a fluorometric assay, and for the in situ detection of caspase activity, immunocytochemistry was performed with a cleavage-site–specific antibody. The time course of alterations in the mitochondrial membrane potential and the release of cell death mediators in individual cells undergoing cell death were assessed with a fluorescent tracer and subsequent immunocytochemistry. In addition, a fluorescent dye, dihydroethidium was used to assess the generation of reactive oxygen species (ROS).

results. Findings of this study revealed that the loss of mitochondrial membrane potential and the release of cell death mediators accompanied RGC death induced by TNF-α or hypoxia. Although caspase inhibitor treatment temporarily decreased the rate of apoptosis, caspase inhibition was not adequate to block RGC death if the mitochondrial membrane potential was lost and mitochondrial mediators were released. Despite the inhibited caspase activity, survival rate was less than 70% after a 48-hour incubation with death stimuli, and both apoptotic and necrotic cells were detectable in these cultures. When combined with caspase inhibition, tempol reduced the production of ROS and provided an additional 20% increase in RGC survival.

conclusions. Based on these novel findings, RGC death induced by TNF-α or hypoxia involves a caspase-independent component, and reducing the free-radical generation provides additional protection of RGCs temporarily saved by caspase inhibition. Therefore, neuroprotective strategies in glaucoma should include tools to improve the ability of these neurons to survive the cytotoxic consequences of mitochondrial dysfunction.

Avariety of death signals in mammalian cells converge to activate the executioner proteolysis cascade mediated by caspases, which play a critical role in initiating and sustaining the biochemical events that result in apoptotic cell death. ^{1 2} Whereas different stimuli initiate the caspase cascade by ligand binding (called the receptor-mediated or extrinsic pathway), mitochondrial events, including the loss of mitochondrial membrane potential and the release of cell death mediators, start another chain of events during the early phases of apoptosis, which also play a role in initiation of the apoptotic cascade by activating caspases (called mitochondria-mediated or intrinsic pathway). Death receptor binding can initiate the apoptotic caspase cascade directly, in which caspase-8 is a proximal effector caspase. ³ In contrast, mitochondrial events lead to caspase-2 ⁴ and -9 ⁵ activation, which subsequently activates caspase-3, a common downstream caspase in the extrinsic and intrinsic pathways. ⁶ As a consequence, caspases contribute to cell death through the degradation of DNA repair enzymes and structural elements, such as nuclear lamina and cytoskeleton, and also lead to indirect activation of chromosomal endonucleases. ⁷ In addition, caspases are involved in mitochondrial dysfunction and generation of reactive oxygen species (ROS) through the disruption of the electron transport chain. ⁸  

Previous research indicates a central role of caspases as mediators of neuronal apoptosis induced by different stimuli. ^{9 10} Caspases have also been shown to be involved in retinal ganglion cell (RGC) death induced by optic nerve axotomy ^{11 12} and N-methyl-d-aspartate. ¹³ In addition, RGC death induced by elevated intraocular pressure involves caspase activation (including caspases-3 and -8) as demonstrated using experimental rat models of glaucoma. ¹⁴ In vitro studies provide compelling evidence that the apoptosis of retinal neurons induced by different stimuli share a common caspase cascade, ^{15 16} which can be inhibited using specific inhibitors of caspases. ¹⁷ Gene therapy delivering a caspase inhibitor has also promoted optic nerve axon survival in a rat glaucoma model. ¹⁸  

RGC death induced by retinal ischemia also involves the activation of caspases (including caspases-1, -2, -3, and -8). ^{19 20} Tissue hypoxia, which may result from insufficient blood supply at the retina and optic nerve head caused by elevated intraocular pressure or dysregulation in the microcirculation (mediated by vasospasm or abnormal autoregulation), has been associated with RGC death in glaucoma. There is clinical evidence of vascular abnormalities in the retina and/or optic nerve head of patients with glaucoma. ^{21 22 23 24 25 26} In addition, a recent histopathologic study has provided evidence that hypoxic tissue stress is present in the retina and optic nerve head of glaucomatous eyes, as assessed by the expression of a transcription factor, hypoxia-inducible factor-1α, which is tightly regulated by cellular oxygen concentration. ²⁷  

TNF-α through the binding of TNF receptor-1, a death receptor, has recently been identified to be one of the mediators of RGC death in a number of neurodegenerative injuries, including glaucoma. Evidence supporting the role of TNF-α as a mediator of RGC death shows that intravitreal injections of TNF-α can produce axonal degeneration in the optic nerve. ²⁸ More recently, studies using mice deficient in TNF receptor-1 have identified that TNF-α, through the binding of this receptor, is involved in RGC death after retinal hypoxia ²⁹ or optic nerve injury. ³⁰ Regarding glaucoma, information obtained from primary cocultures of RGCs and glial cells has provided direct evidence that TNF-α secreted by glial cells in response to glaucomatous stressors can induce apoptotic death of RGCs. ¹⁶ Glial production of TNF-α is increased, and TNF receptor-1 is upregulated in RGCs and their axons in glaucomatous donor eyes. ^{31 32 33} Similar to glaucomatous human eyes, TNF-α and TNF receptor-1 are upregulated after experimental elevation of intraocular pressure in rats (Tezel and Yang, unpublished observation, 2003). Activation of retinal caspase-8 (a proximal effector protein in the TNF receptor family cell death pathway) ^{14 33} strongly suggests the role of TNF-α signaling in neuronal loss detected in these hypertensive rat eyes. In addition, a recent study using microarray analysis in hypertensive rat eyes ³⁴ has detected upregulated retinal expression of a transcription factor (Litaf), which regulates TNF-α gene expression. ³⁵ Continuing in vivo observations using experimental models of glaucoma, along with genetic studies demonstrating optineurin gene mutation ³⁶ or TNF-α-308 gene polymorphism ³⁷ in glaucoma patients, also support the role of TNF-α signaling in glaucoma. 

Because apoptosis of RGCs can be induced by a broad array of different stimuli, which share a common caspase cascade, inhibition of caspases has been suggested to provide a means to protect and/or rescue RGCs from apoptotic cell death, regardless of the causative event. However, in many models of apoptosis, cells saved by caspase inhibition may not be able to recover, but eventually go on to die due to mitochondrial dysfunction. ^{38 39 40} Although caspase activation has been identified in RGCs exposed to different apoptotic stimuli, the caspase-independent component of RGC death is not well known. A better understanding of caspase-dependent versus caspase-independent components of cell death in RGCs could facilitate determining the role of caspase inhibitors to provide neuroprotection in several neurodegenerative injuries, including glaucoma. 

In this study, we have taken advantage of our ability to produce highly purified RGCs, ¹⁶ to investigate the caspase-independent component of cell death in these distinct neurons. Evidence has shown that the survival of these axotomized RGCs can be promoted in culture in the presence of a defined medium containing a combination of trophic factors and antioxidants. ^{16 41} In the present study, we specifically sought to determine whether these neurons survive the exposure to TNF-α or hypoxia (two different stimuli known to trigger preferentially the receptor-mediated or mitochondria-mediated cell death pathways, respectively) in the presence of caspase inhibitor treatment, and whether mitochondrial dysfunction accompanies RGC death after exposure to these noxious stimuli. Our findings revealed that the loss of mitochondrial membrane potential and the release of cell death mediators, including cytochrome c and apoptosis-inducing factor (AIF), accompanied RGC death after exposure to TNF-α or hypoxia for up to 48 hours. Although caspase inhibitor treatment temporarily decreased the rate of RGC apoptosis after exposure to these stimuli, inhibition of caspases was not adequate to block RGC death if the mitochondrial membrane potential was lost and the mitochondrial mediators were released. Our findings also revealed that reducing the free-radical generation by additional antioxidants provided further protection against the harmful consequences of mitochondrial dysfunction in RGCs. 

Materials and Methods

RGC Culture

Primary cultures of RGCs were derived from early postnatal rat retinas as previously described. ¹⁶ All the animals were handled according to the regulations of the Institutional Animal Care and Use Committee, and all procedures adhered to the tenets of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Briefly, enucleated eyes of 5- to 7-day-old Sprague-Dawley rats were rinsed with CO₂-independent culture medium (Invitrogen, Carlsbad, CA) and retinas were mechanically dissected under a microscope. To prepare retinal cell suspensions, tissues were dissociated in Eagle’s minimum essential medium containing 20 U/mL papain, 1 mM l-cysteine, 0.5 mM EDTA, and DNase (0.005%; Worthington Biochemicals, Lakewood, NJ) at 37°C for 40 minutes. Then, retinas were rinsed in an inhibitor solution containing Eagle’s minimum essential medium, ovomucoid (0.2%; U.S. Biological, Swampscott, MA), DNAase (0.04%), and bovine serum albumin (0.1%; Sigma-Aldrich, St. Louis, MO). At the end of the treatment period, tissues were triturated through a 1 mL plastic pipette to yield a suspension of single cells. Immunomagnetic selection of Thy-1.1–positive RGCs was performed with antibody-coated magnetic beads (Dynal, Oslo, Norway) in a two-step process, as previously described. ¹⁶ Selected cells were seeded on extracellular-matrix–coated 24-well plates (BD Biosciences, Bedford, MA) at a density of 3 × 10⁴ cells/well, and incubated in a serum-free culture medium, which was prepared using B27-supplemented Neurobasal medium (Invitrogen), containing bovine serum albumin (100 μg/mL), progesterone (60 ng/mL), insulin (5 μg/mL), pyruvate (1 mM), glutamine (1 mM), putrescine (16 μg/mL), sodium selenite (40 ng/mL), transferrin (100 μg/mL), triiodo-thyronine (30 ng/mL), brain-derived neurotrophic factor (BDNF; 50 ng/mL), ciliary neurotrophic factor (CNTF; 20 ng/mL), bFGF (10 ng/mL), forskolin (5 μM), inosine (100 μM), and antibiotics (Sigma-Aldrich). RGCs in these cultures were identified on the basis of retrograde labeling, cell morphology, and the expression of cell markers. ¹⁶  

RGCs exhibiting contact of the neuritic processes were incubated in the presence of TNF-α or hypoxia. A concentration of 0.05 ng/mL recombinant rat TNF-α (R&D Systems, Minneapolis, MN) was used based on a reproducible dose–response curve for TNF-α–mediated RGC death. Hypoxia was maintained by placing the plates in a dedicated culture incubator with a controlled flow of N₂ and CO₂ at a setting of 5% CO₂ and 5% O₂. To examine the time course of cellular responses, the noxious stimuli were maintained for 6, 12, 24, or 48 hours. Control cultures prepared using an identical passage of cells were simultaneously incubated in a regular tissue culture incubator with 95% air/5% CO₂ at 37°C in the absence of TNF-α. At the end of the incubation period, cells were immediately subjected to experiments, which were performed in duplicate wells and repeated at least three times in each experimental condition. All counts were conducted in a masked fashion. Data are reported as the mean ± SD. 

Examination of Cell Death

Cell viability was determined with a kit containing calcein AM (Live/Dead Cell Viability Kit; Molecular Probes, Eugene, OR), as previously described. ¹⁶ The kit relies on the intracellular esterase activity within living cells, through which the calcein AM, a cell-permeable fluorogenic esterase substrate, hydrolyzes to a green fluorescent product, calcein. Living cells showing green fluorescence were counted in at least 10 random fields of each well at 200× magnification (∼150 cells per well) using a fluorescence microscope (Carl Zeiss Microimaging, Inc., Thornwood, NY). Survival rate was expressed as the percentage of the total number of cells in control wells (treated with no additives) at each time point. 

In addition, Alexa Fluor 488-conjugated annexin V binding combined with propidium iodide (Molecular Probes) labeling was performed for the distinction of necrotic (annexin V+ and propidium iodide+) and apoptotic (annexin V+/ propidium iodide−) cells. ⁴² Data are presented as the percentage of total number of cells in control wells. 

Examination of Caspase Activity

Caspase-3–like protease activity was measured in a fluorometric assay by measuring the extent of cleavage of the fluorometric peptide substrate as previously described. ^{16 17} Briefly, cell lysates were incubated with Ac-Asp-Glu-Val-Asp-7-amino-4-trifluoro-methyl coumarin (Ac-DEVD-AMC, 50 μM), a fluorometric substrate for caspase-3 and other caspase-3–like proteases, including caspase-7. Positive controls included purified recombinant caspase-3 (0.1 μg; Upstate Biotechnology, Lake Placid, NY). Fluorescence was measured at an excitation wavelength of 360 nm and an emission wavelength of 460 nm in a fluorescence plate reader at different time points up to 180 minutes. Protease activity was expressed as picomoles substrate per milligram protein per minute as calculated using the linear range of the assay. 

Time-dependent changes in caspase activity were also determined by immunolabeling with a cleavage-site–specific caspase-3 antibody (Alexis, San Diego, CA), using a previously described immunocytochemistry protocol. ¹⁶ Briefly, cells grown on extracellular-matrix–coated coverslips were washed in phosphate-buffered saline solution and fixed with 4% paraformaldehyde solution for 30 minutes at room temperature. After washing, they were permeabilized with 0.1% Triton X-100 in 0.1% sodium citrate solution on ice for 4 minutes. Cells were then treated with 3% bovine serum albumin for 30 minutes to block nonspecific binding sites and were incubated with primary antibody (10 μg/mL) at 37°C for 2 hours. At the end of the incubation period, cells were washed and incubated with the secondary antibody, Alexa Fluor 568-conjugated anti-rabbit IgG (2 μg/mL; Molecular Probes), and after washing and mounting, they were examined with a fluorescence microscope (Carl Zeiss Microimaging, Inc.). 

To determine the caspase-independent component of cell death, the experiments were performed in the absence and presence of irreversible, cell-permeable caspase inhibitors. Caspase inhibitors used included Z-IETD-FMK, a caspase-8 inhibitor; Z-LEHD-FMK, a caspase-9 inhibitor; and Boc-D(Ome)-FMK (BAF), a broad-spectrum caspase inhibitor (100 μM, Alexis). In addition, 4-hydroxytetramethylpiperidine-1-oxyl (Tempol; 5 mM; Calbiochem, San Diego, CA), a cell-permeable free-radical scavenger, ^{43 44} was used in some experiments. The concentrations used were chosen based on dose–response experiments in RGCs (data not shown). 

Assessment of Mitochondrial Dysfunction

A fluorescent label (Mitotracker Orange, [CM-H₂TMRos]; Molecular Probes) was used to assess the mitochondrial membrane potential as previously described. ⁴⁰ Briefly, after incubations in the absence and presence of the noxious stimuli, cultures were loaded with 1 μM freshly prepared Mitotracker for 1 hour at 37°C in the dark, just before the end of each experimental period at specific time points. At the end of the incubation period, cells were washed with phosphate-buffered saline solution and fixed in 4% paraformaldehyde for 30 minutes. To determine the localization of cytochrome c and AIF simultaneously with the status of mitochondrial membrane potential in individual cells, cells were also processed for immunofluorescence labeling with cytochrome c or AIF antibodies (1 μg/mL; PharMingen, San Diego, CA) and Alexa Fluor 488–conjugated secondary antibodies (2 μg/mL; Molecular Probes). Nuclei were counterstained with Hoechst 33342 (Molecular Probes). Cells first were imaged on the basis of their nuclear labeling under a fluorescence microscope (Carl Zeiss Microimaging, Inc.), and then scored as positive or negative for Mitotracker labeling and cytochrome c or AIF immunolabeling by a masked observer. Only cells that had completely lost all punctuate labeling were scored as negative. 

To determine the relationship between the status of Mitotracker labeling and rescue with caspase inhibition, at the end of 24-hour incubation in the presence of death stimuli, RGCs were incubated with 250 nM fluorescent tracer for 1 hour at 37°C. After the labeling period, cells were washed and incubated in the presence of identical stimuli for an additional 24 hours in the presence of BAF. Representative fields of cells marked at the beginning of the experiments were photographed at 24 hours by using phase-contrast microscopy to visualize cells and fluorescence microscopy to visualize the Mitotracker labeling. Cells were returned to a tissue culture incubator, and phase-contrast photographs of the same fields of cells were taken again at the end of an additional 24-hour incubation. Care was taken to ensure minimum exposure of cultures to light in all subsequent manipulations (particularly while taking photographs), because exposure to intense light was found to be toxic to the Mitotracker-labeled cells. It was for the same reason that the concentration of Mitotracker was lower than that used in previous experiments described above. Because no immunolabeling steps with washes followed the labeling in these experiments, the lower concentration was sufficient to provide an adequate signal. Approximately 200 cells were scored per experiment. 

To assess ROS generation, cultures were incubated with dihydroethidium (1 mM; Molecular Probes) for 30 minutes. When reacting with intracellular ROS, this cell-permeable dye fluoresces in the rhodamine spectrum within the nuclei. ⁴⁵ After labeling, cells washed and replenished with fresh medium were imaged under a fluorescence microscope (Carl Zeiss Microimaging, Inc.). The percentage of calcein+ viable cells that showed red fluorescent nuclei was assessed in each experimental setting, as previously described. ⁴⁶  

Results

RGC Death

After incubations of RGCs in the presence of TNF-α or hypoxia for up to 48 hours, the survival rate was decreased in a time-dependent manner. Figure 1A demonstrates the morphology of cells incubated in normal conditions or in the presence of cell death stimuli. Whereas the percentage of living cells was approximately 90% in control cultures incubated under normal conditions, less than 60% of RGCs survived (Mann-Whitney test, P = 0.0002) the 48-hour incubation period with death stimuli (Fig. 1B) . 

Annexin V binding combined with propidium iodide labeling was performed for the distinction of necrotic (annexin V+/propidium iodide+) and apoptotic (annexin V+/propidium iodide−) cells. After incubations with death stimuli for 48 hours, 34.2% ± 4.1% of RGCs were annexin V+/propidium iodide−. However, a similar pattern of labeling was detectable in approximately 5% of RGCs incubated in normal conditions. The percentage of annexin V+/propidium iodide+ RGCs were 10.6% ± 2.3% in cultures exposed to noxious stimuli for 48 hours, but less than 5% in RGCs incubated under normal condition. 

Caspase Activity

After incubations with death stimuli for up to 48 hours, immunocytochemistry using a cleavage-site–specific antibody detected immunolabeling of RGCs for active caspase-3 (Fig. 2A) . To measure the cleavage of Ac-DEVD-AMC, which reflects caspase-3–like protease activity, fluorometric analysis was performed in RGCs undergoing cell death. The amount of DEVD-AMC cleaving activity was approximately nine times higher in RGCs incubated in the presence of TNF-α or hypoxia for 48 hours compared with control cultures incubated in normal conditions (Fig. 2B) . 

RGCs were then incubated in identical conditions in the presence of caspase inhibitors. In cultures exposed to TNF-α, treatment with Z-IETD-FMK or Z-LEHD-FMK resulted in an approximately 61% and 22% decrease in caspase activity, respectively. In cultures exposed to hypoxia, treatment with Z-IETD-FMK or Z-LEHD-FMK resulted in an approximately 11% and 69% decrease in caspase activity, respectively. When BAF, a broad-spectrum caspase inhibitor, was used, the caspase activity was similarly decreased by approximately 90% in cultures exposed to TNF-α or hypoxia (Mann-Whitney test, P = 0.0002; Fig. 2B ). 

Based on these data, BAF was used for the following experiments to assure maximum caspase inhibition with a broad spectrum. Thus, the caspase-inhibiting treatment was effective in decreasing caspase activity in RGCs during the incubation period of up to 48 hours. However, despite inhibited caspase activity, BAF treatment provided an approximately 25% increase in RGC survival of 48-hour incubation with death stimuli (Fig. 2C) . An important point to note is that the protection provided by caspase inhibition in RGCs exposed to noxious stimuli for 24 hours was at least 40% greater than that provided in RGCs incubated under identical conditions for 48 hours (Mann-Whitney test, P = 0.005). Caspase-inhibiting treatment using BAF increased RGC survival from 60% to 84% and 64% to 84% in cultures incubated in the presence of TNF-α or hypoxia for 24 hours, respectively. However, the survival rate was improved from 49% to 63% and 56% to 68% after incubation in the presence of TNF-α or hypoxia for 48 hours, respectively. These findings indicate that caspase inhibition provides an early improvement in the survival of RGCs exposed to different death stimuli, but this protective effect decreases after 24 hours (Fig. 2C) . 

Thus, although some early protection was provided by caspase inhibitor treatment, the survival rate was less than 70% after a 48-hour incubation with death stimuli in the presence of caspase inhibitors. In contrast, approximately 90% of RGCs survived in control cultures incubated under normal conditions for 48 hours. Examination of annexin V binding and propidium iodide labeling in RGC cultures incubated with death stimuli in the presence of caspase inhibitor treatment revealed that both annexin V+/propidium iodide− apoptotic cells (16.2% ± 2.5%) and annexin V+/propidium iodide + necrotic cells (17.1% ± 2.1%) were detectable. Apoptotic cells showed green fluorescence (Fig. 2D ; annexin V+), necrotic cells showed red and green fluorescence (annexin V+/propidium iodide+), and living cells showed no fluorescence. Thus, although the morphology of cell death exhibited a change that was characterized by a decrease in the percentage of apoptotic cells in the presence versus absence of caspase inhibitor treatment, caspase inhibition did not completely protect RGCs from cell death. 

Mitochondrial Dysfunction

Fluorescence labeling with Mitotracker Orange (Molecular Probes) and subsequent immunolabeling for cytochrome c or AIF allowed us to assess the status of mitochondrial membrane potential and localization of cytochrome c and AIF simultaneously in individual RGCs undergoing cell death. RGCs incubated under normal condition showed an identical punctuate Mitotracker labeling pattern. Similarly, immunolabeling for cytochrome c and AIF appeared punctuate, as expected for their mitochondrial localization. However, incubation of RGCs in the presence of death stimuli for up to 48 hours resulted in the loss of both the Mitotracker labeling and the cytochrome c immunolabeling in most of the cells. After incubations in the presence of TNF-α or hypoxia for up to 48 hours, AIF was similarly released from the mitochondria in association with the loss of mitochondrial membrane potential and translocated to the nuclei. We did not observe any cells that maintained mitochondrial cytochrome c but lost the Mitotracker labeling at any time point during the experimental period. The release of cytochrome c also preceded the release of AIF. Within 24 hours of incubation with death stimuli, more than 35% of cells had already lost cytochrome c immunolabeling, but there was only a modest decrease in the fraction of cells exhibiting the mitochondrial AIF immunolabeling (10%) or the Mitotracker labeling (25%). This indicates that individual RGCs first released cytochrome c and subsequently lost mitochondrial membrane potential and AIF. After 48 hours incubation in the presence of death stimuli, approximately 50% of cells lost both cytochrome c and Mitotracker labelings, and 30% lost AIF. Many of the cells that lost the mitochondrial AIF immunolabeling, but not all, exhibited AIF accumulation in the nucleus. Figure 3 demonstrates the loss of punctuate Mitotracker labeling, the release of cytochrome c and AIF from the mitochondria, and the diffuse nuclear AIF immunolabeling in individual RGCs after exposure to TNF-α for 48 hours. It was also notable that the blue fluorescent Hoechst 33342 more brightly stained the condensed chromatin in the nuclei of apoptotic cells relative to dim labeling of the normal chromatin in living control cells. Thus, these findings reveal that the loss of mitochondrial membrane potential and the release of mitochondrial cytochrome c and AIF accompanied the death of RGCs after exposure to death stimuli. These mitochondrial alterations were detectable consistently in the absence and presence of caspase inhibiting treatment. 

Figure 4A shows time-dependent quantitative evaluation of the loss of Mitotracker labeling in RGCs incubated under normal condition or exposed to TNF-α in the absence or presence of BAF. The time course of the loss of mitochondrial membrane potential was constant with the time-dependent decrease in the protective effect of caspase inhibitor treatment shown in Figure 2C . The consistency in the loss of mitochondrial membrane potential in the presence of caspase inhibitor treatment may explain why cell death could only be partially prevented by caspase inhibition. 

To assess the relationship between the status of mitochondrial membrane potential and the protective ability of caspase inhibitor treatment, RGCs incubated under normal condition or exposed to death stimuli for 24 hours were loaded with Mitotracker and photographed to document which cells had lost or maintained the fluorescent labeling. Approximately 25% of RGCs had lost the labeling at that time. BAF was then added to cultures, and they were incubated in identical death stimuli for an additional 24 hours in the presence of this caspase inhibitor. At the end of the incubation period, the identical fields were photographed again. This experiment demonstrated that the addition of BAF blocked cell death in these RGCs. However, BAF rescued only the RGCs that were positive for Mitotracker at the beginning of the treatment. Figure 4B exemplifies the correlation seen between fluorescence labeling and rescue with caspase inhibition. In the field shown, all fluorescence-positive RGCs were rescued with caspase inhibition, whereas fluorescence-negative cells were not. After 48-hour incubation, the rescued RGCs were clearly identifiable with large and phase-bright cell bodies and branched neurites of varying length, whereas the nonrescued cells became atrophic and appeared degenerated. As shown in Figure 4C , quantitative evaluation demonstrated that more than 70% of RGCs that were Mitotracker-positive were rescued by BAF. In contrast, approximately 30% of the Mitotracker-negative RGCs were rescued by this treatment (Mann-Whitney test, P = 0.0002). Thus, based on time course experiments, Mitotracker labeling predicts the rescue of RGCs with caspase inhibitor treatment, and the inhibition of caspase activity is not adequate to block RGC death if the cell death mediators are released and the mitochondrial membrane potential is lost. 

Because mitochondrial dysfunction leads to ROS production, which could contribute to cell death, in another set of experiments, we examined whether ROS are generated during the caspase-independent death of RGCs. To assess ROS generation, we used a fluorescent dye, dihydroethidium. We detected augmented ROS accumulation in RGCs after their incubation with death stimuli in the absence or presence of caspase inhibitor treatment. Whereas less than 5% of control cells incubated under normal conditions exhibited dihydroethidium labeling, approximately 55% of the cell population in cultures exposed to death stimuli for 48 hours were positive for dihydroethidium labeling. As shown in Figure 5A , a significant percentage of RGCs incubated with TNF-α in the presence of BAF accumulated ROS over time. The time course of ROS generation showed that this was a relatively late event. RGCs were then incubated with death stimuli in the presence of both BAF and an antioxidant, tempol. Addition of tempol significantly reduced ROS generation in RGCs compared with treatment with BAF alone (Mann-Whitney test, P = 0.001; Fig. 5A ). However, no prominent effect of tempol treatment on the loss of the Mitotracker labeling or the cytochrome c or AIF immunolabeling was detectable (data not shown). 

In cultures exposed to death stimuli, cotreatment of RGCs with BAF and tempol resulted in at least a 20% additional increase in cell survival compared with cultures treated with BAF alone. However, in control cultures incubated in the absence of death stimuli, tempol treatment resulted in a less than 5% increase in cell survival (Mann-Whitney test, P > 0.05). In the presence of cotreatment, the survival rate was approximately 76% and 86% in cultures incubated with TNF-α or hypoxia for 48 hours, respectively (Fig. 5B) . Thus, tempol was effective in preventing ROS generation and RGC death induced by TNF-α or hypoxia. These findings indicate that reducing free-radical generation provides additional protection against the caspase-independent component of RGC death. 

Discussion

Findings of this study revealed that although caspase inhibitor treatment provides some early protection and decreases the rate of apoptosis in RGCs exposed to two different stimuli, TNF-α and hypoxia, inhibition of caspases is not adequate to block RGC death if the mitochondrial membrane potential is lost and cell death mediators, including cytochrome c and AIF, are released. Caspase-8 is at the apex of the hierarchy of caspases, cleaving and activating all other caspases involved in the TNF receptor family cell death pathway. ³ Therefore, inhibition of this proximal caspase may be expected to block receptor-mediated caspase activity, before mitochondrial dysfunction. In accordance with this, we observed that the inhibition of caspase-8 in RGCs exposed to TNF-α resulted in a relatively greater decrease in protease activity than in RGCs exposed to hypoxia. However, there is in vitro evidence suggesting that caspase-8 activation may also occur downstream of mitochondrial dysfunction. ^{47 48} Although most of these observations are associated with anticancer-drug–or ionizing-radiation–induced apoptosis in tumor cells, such an activation of caspase-8 through caspase-9 ⁴⁸ or -3, ⁴⁷ independent from the death receptor signaling, may have been associated with the amplification of caspase-mediated cell death. In any case, the caspase cascade in both the extrinsic and intrinsic pathways of apoptosis eventually activated caspase-3, which is one of the key caspases, being responsible for the proteolytic cleavage of many proteins. This may explain a relatively similar effect when a broad-spectrum caspase inhibitor was used in RGC cultures exposed to two different stimuli, TNF-α or hypoxia. However, we found that even simultaneous inhibition of a broad spectrum of caspases with BAF could not completely prevent RGC death. These observations provide evidence that RGC death induced by TNF-α or hypoxia involves a caspase-independent component. 

Our observations revealed that caspase inhibitor treatment of RGCs results in a delay in the cell death process rather than providing a permanent and complete rescue. Consistent with our observations in RGCs, earlier studies have demonstrated that caspase-independent neuronal cell death often occurs with a delayed time course. ^{40 49 50} In addition, although caspase inhibition has an early beneficial effect on the survival of adult rat RGCs after optic nerve axotomy, ¹¹ the positive effect decreases after 4 weeks. ⁵¹ In accordance with previous studies in different systems, ^{38 39} our findings provide evidence that in RGCs undergoing apoptosis, caspase inhibitors block the characteristic features of apoptosis, but not the commitment to death. We found that although the rate of annexin V+/propidium iodide− apoptotic cells was decreased with caspase inhibition, an increase in the rate of annexin V+/propidium iodide+ necrotic cells was detectable in RGC cultures exposed to TNF-α or hypoxia in the presence of caspase inhibitor treatment. Thus, despite a central role of caspases in the regulation of RGC apoptosis, caspase inhibition may turn the morphology of cell death, even the receptor-mediated cell death, from apoptotic into necrotic without preventing death itself. Obviously, the long-term fate of an injured RGC, in which the execution of apoptosis was blocked by caspase inhibition, depends on several other factors. 

Mitochondria have been found to be essential in controlling cell death. ⁵² Cytochrome c release from the mitochondria has been involved in neuronal apoptosis, which leads to caspase activation. ^{53 54} In addition to the release of cytochrome c, several other mediators may play a role in cell death by interacting with mitochondria. These mediators include AIF, ^{55 56} bid, Smac/Diablo, ⁵⁷ endonuclease G, ⁵⁸ and Omi/HtrA2. ⁵⁹ In response to an apoptotic stimulus, AIF translocates from the mitochondrial intermembrane space to the nucleus and causes chromatin condensation, DNA fragmentation, oxidative damage, and cell death by autophagic degeneration. ^{60 61} AIF has been identified to be a component of caspase-independent cell death in the brain, including that induced by hypoxic injury. ^{62 63} Our findings revealed that after exposure to TNF-α or hypoxia, mitochondrial alterations, including the release of cytochrome c and AIF, are turned on and lead to RGC demise through the caspase-independent events. 

Mitochondrial oxidative phosphorylation is compromised after the release of cell death mediators, which leads to production of ROS, cytoplasmic vacuolation, plasma membrane permeability, and subsequent cell death. ^{64 65} Whether a caspase-inhibited neuron survives the mitochondrial dysfunction depends on several factors, which include its ability to generate adenosine triphosphate (ATP) and its energetic requirements for recovering and maintaining normal physiological functions. ^{46 66} Many direct or indirect consequences of mitochondrial dysfunction may similarly contribute to the caspase-independent death of RGCs. It should be noted that the medium we used (Neurobasal; Invitrogen) contains high amounts of glucose and is supplemented by B-27, which contains antioxidants. Because this medium is essential for the in vitro maintenance of RGCs during the experimental period, we could not test whether providing an alternative source of ATP (namely glucolysis) by supplementation with glucose or reducing free-radical generation by addition of antioxidants would improve the survival of RGCs incubated in a glucose-free or antioxidant-free medium. However, despite the presence of high glucose and antioxidant supplements in culture medium, many BAF-treated RGCs did not survive the continued presence of death stimuli. Therefore, although we could not supplement our cultures with more glucose, we decided to further supplement our cultures with an antioxidant, because RGCs are known to be differentially susceptible to certain ROS, ⁶⁷ and RGC survival is critically sensitive to the oxidative redox state. ⁶⁸ For this purpose, we used tempol, a cell-permeable free-radical scavenger that interacts with peroxynitrite, a reactive oxidant formed by the combination of superoxide and nitric oxide. ^{43 44} Tempol has been shown to reduce cerebral injury caused by transient cerebral ischemia in vivo ⁶⁹ and to prevent ROS generation and caspase-independent cell death in neurons in vitro. ⁴⁶ It has also been found to protect retinal neurons against N-methyl-d-aspartate–induced neurotoxicity. ⁷⁰ Our observations revealed that tempol provided an additional increase in the survival of RGCs treated with BAF. This is consistent with previous studies demonstrating the role of free-radical generation in the caspase-independent death of embryonic cortical neurons exposed to camptothecin, ⁴⁶ as well as studies demonstrating the role of ROS generation in cell death induced by TNF-α or hypoxia. ^{71 72 73} Thus, reducing free-radical generation protects, in part, against the harmful consequences of mitochondrial dysfunction and offers a beneficial tool to protect RGCs temporarily saved by caspase inhibition during the continued presence of death stimulus. 

As an alternative measure to prevent harmful consequences of mitochondrial dysfunction, the opening of the permeability transition pore (PTP) can be inhibited using pharmacologic agents, such as cyclosporin A. However, as also supported by our observations in the present study, the loss of mitochondrial membrane potential in neurons is downstream of cytochrome c release. ^{74 75} More importantly, previous studies using primary mixed retinal cultures demonstrated the differential regulation of the PTP opening in RGCs in which cyclosporin A did not prevent mitochondrial depolarization or apoptosis, but induced cell death. ⁷⁶  

Growing evidence suggests the involvement of the mitochondria in RGC death induced by different stimuli. For example, p53, a transcription factor activating Bax required for cytochrome c release from the mitochondria, has been involved in neuronal apoptosis in glaucoma. ⁷⁷ There is also evidence suggesting that mitochondrial dysfunction contributes to neuronal apoptosis in an experimental rat glaucoma model, ^{78 79} and that free-radical scavenging when combined with trophic factors can provide neuroprotection in hypertensive rat eyes. ⁸⁰ In addition, amplification of ROS, which is associated with mitochondrial dysfunction, has been suggested to be involved in signaling of RGC apoptosis after axonal injury. ⁸¹ We studied RGC death induced by two different stimuli: TNF-α and hypoxia. Hypoxia-induced neuronal cell death has been associated with mitochondria through energy depletion, altered ionic homeostasis, and/or oxygen-sensing molecules. ^{82 83} Our findings revealed that similar to hypoxia-induced RGC death, receptor-mediated death of RGCs through TNF-α involves caspase-dependent and -independent components of the mitochondrial cell death pathway. This is consistent with previous studies demonstrating that caspase-8 cleaves a proapoptotic member of the Bcl-2 family of proteins, Bid, and the activated Bid participates in the activation of the mitochondrial cell death pathway. ^{84 85} In addition, ceramide generated after TNF death receptor binding may initiate mitochondrial events. ⁷²  

Although TNF-α is a mediator of RGC death, growing evidence demonstrates that the bioactivities of TNF-α, even through TNF receptor-1 (a death receptor) can promote both cell death and survival in neuronal tissues. ^{29 86} Cross talk between the death-promoting and survival-promoting signals is critical for maintaining the balance that determines whether a cell should live or die in response to TNF-α. For example, inhibition of caspase activity has been suggested to amplify the TNF-α–induced survival-promoting signaling through the facilitated activation of NF-κB. ⁸⁷ However, our recent studies have revealed that JNK activity after optic nerve injury is associated with the amplification of TNF-α–mediated death signaling, thereby switching the life balance toward cell death. ³⁰ Because of the diverse bioactivities of TNF-α, the inhibition of cell death signaling as in the present study (and/or amplification of survival signaling), rather than the inhibition of death receptor binding, appears to be more rational for accomplishing neuroprotection against TNF-α–mediated RGC death, because such strategies would not impede the survival-promoting signaling triggered by TNF-α. 

Identification of precise cellular mechanisms requires the isolation and primary culture of specific cell types. In the case of glaucoma, this clearly warrants the use of RGCs. However, in vitro experiments using primary cultures of RGCs are not easy to perform, mainly because of the limited yield and the typically postmitotic feature of these neurons. Therefore, early postnatal tissues are used as an attempt to optimize cell number and survival in culture. In this study, we used primary cultures of RGCs similarly derived from early postnatal rat retinas. It has been well-recognized that despite differences between early postnatal and adult tissues, these cultures have unique informative value for glaucoma research. In addition, we did not coculture RGCs with glia as we originally described, ¹⁶ to identify specific mechanisms in RGCs triggered by selected stimuli without any interference of the results with cellular events triggered by glia-associated factors. Thus, many features of our culture system make it a unique tool for in vitro studies to identify specific mechanisms in RGCs relevant to glaucomatous neurodegeneration. 

It should also be noted that the caspase activity was inhibited in these cultures by using a common irreversible inhibitor of a broad spectrum of caspases, which resulted in an ∼90% decrease in caspase activity. Thus, our conclusions rely on that the caspase inhibitor treatment was sufficiently potent to block caspase activity with no effects on other cellular components. However, the opposite possibility cannot be excluded, or some caspases may not yet be identified. Nevertheless, it is obvious that providing some early protection, current caspase inhibitor treatment may increase the length of time before RGC death, thereby allowing alternative neuroprotective treatments. Determination of ultimate cell fate in caspase-inhibited RGCs depends on several factors, which include the ability of these neurons to withstand cytotoxic events initiated by mitochondrial dysfunction. Therefore, neuroprotective strategies in several neurodegenerative injuries of RGCs, including glaucoma, should include additional tools to improve the intrinsic ability of these neurons to survive cytotoxic consequences of mitochondrial dysfunction. Current findings provide evidence that antioxidant treatment is a promising complementary strategy to improve the survival of RGCs that are temporarily saved by caspase inhibition. 

 Supported by National Eye Institute Grant 1R01EY013813; the American Health Assistance Foundation, Clarksburg, MD; and an unrestricted grant to the University of Louisville Department of Ophthalmology and Visual Sciences from Research to Prevent Blindness. GT is a Research to Prevent Blindness Sybil B. Harrington Scholar.

 Submitted for publication May 3, 2004; revised July 7, 2004; accepted July 15, 2004.

 Disclosure: G. Tezel, None; X. Yang, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Time course of RGC death. (A) Phase-contrast images of RGCs incubated in normal conditions or in the presence of death stimuli for 48 hours show that after incubation of RGCs in the presence of TNF-α or hypoxia, cell death was induced characterized by cell body shrinkage, compaction of the nucleus, and cell debris. (B) Survival rate, assessed by counting calcein AM–positive cells, was decreased in a time-dependent manner. Experiments were performed in duplicate wells and repeated at least three times for each experimental condition. Data are presented as the mean ± SD.

Figure 2.

 

Caspase activity. For in situ detection of caspase activity, immunocytochemistry was performed with a cleavage site-specific caspase-3 antibody and Alexa Fluor 568-conjugated secondary antibody. (A) Phase-contrast images and corresponding red fluorescent images of RGCs incubated under normal condition or in the presence of TNF-α, respectively. After incubations of RGCs in the presence of TNF-α, active caspase-3 immunolabeling (red fluorescence) was detectable. Fluorometric analysis was then performed to measure the cleavage of Ac-DEVD-AMC. (B) Time-dependent increase in DEVD-AMC cleaving activity in RGCs incubated in the absence or presence of death stimuli. Incubation of RGCs in identical conditions in the presence of a caspase inhibitor (BAF, 100 μM) resulted in a significant decrease in caspase activity (Mann-Whitney test, P = 0.0002). (C) BAF treatment provided a partial increase in RGC survival during the 48-hour incubation with death stimuli, which was most prominent at 24 hours (Mann-Whitney test, P = 0.005). (D) A phase-contrast image from RGC cultures exposed to TNF-α in the presence of BAF and annexin V (green) and propidium iodide (red) labeling in the same field of cells. Both annexin V+/propidium iodide− apoptotic cells (white arrows) and annexin V+/propidium iodide+ necrotic cells (white arrowheads) were detectable in BAF-treated cultures. Note that living cells (black arrows) showed no fluorescence. Data obtained from three independent experiments are presented as the mean ± SD.

Figure 3.

 

Loss of mitochondrial membrane potential. Mitotracker Orange (Molecular Probes) labeling followed by immunolabeling for cytochrome c or AIF was performed to assess the status of mitochondrial membrane potential and localization of cytochrome c and AIF simultaneously in individual RGCs undergoing cell death. Shown are representative images of Mitotracker labeling and cytochrome c or AIF immunolabeling in individual RGCs incubated in the absence or presence of death stimuli. After incubations in the presence of TNF-α or hypoxia, the loss of punctuate Mitotracker labeling (red) was accompanied by the loss of punctuate immunolabeling for cytochrome c or AIF (green) and the diffuse nuclear AIF immunolabeling. Blue fluorescence corresponds to nuclei of these cells labeled with Hoechst 33342.

Figure 4.

 

Association of mitochondrial alterations with cell death. (A) Time course of the loss of mitochondrial membrane potential during RGC death induced by TNF-α. (B) After 24-hour incubation in the presence of TNF-α, RGCs were loaded with Mitotracker Orange (Molecular Probes) and photographed to document the status of fluorescence labeling in the cells. A representative field shows two cells of six that were Mitotracker negative (arrows). Cells were then incubated with TNF-α for an additional 24 hours in the presence of BAF (100 μM), and the same field of cells was photographed after 24 hours. Photographs show that the Mitotracker-positive cells were rescued with the addition of BAF, whereas the negative cells were not. (C) Data showing the correlation between fluorescence positivity and protection with BAF treatment (Mann-Whitney test, P = 0.0002). Cell survival was assessed by counting calcein AM–positive cells in duplicate wells in three independent experiments. Results are presented as the mean ± SD.

Figure 5.

 

(A) ROS generation was assessed with dihydroethidium labeling of RGCs. Despite BAF treatment, a significant percentage of RGCs incubated with death stimuli accumulated ROS over time. RGCs were then incubated with death stimuli in the presence of both BAF (100 μM) and an antioxidant, tempol (5 mM). Addition of tempol significantly reduced ROS generation in RGCs compared with treatment with BAF alone (Mann-Whitney test, P = 0.001). (B) Cotreatment of RGCs with BAF and tempol resulted in an additional increase in the rate of calcein AM–positive surviving cells. Control bars show the survival rate in RGCs incubated under normal condition. Results are presented as the mean ± SD of three independent experiments.