Main

It is estimated that glaucoma affects nearly 70 million individuals worldwide, including at least 6.8 million who are bilaterally blind.1 The disease is characterized by the slow progressive degeneration of the retinal ganglion cells (RGCs) and their axons, which are usually associated with elevated intraocular pressure (IOP). Recent studies have shown that glaucoma is affected by multiple genes and environmental factors,2, 3 and there are several inherited and experimentally induced animal models of high IOP glaucoma, including DBA/2J mice and laser-induced chronic ocular hypertension model.4, 5, 6 There is a subtype of glaucoma termed normal tension glaucoma (NTG), however, that presents with statistically normal IOP. The number of NTG patients has been thought to be small relative to the total number of glaucoma patients, but recent studies have revealed an unexpectedly high prevalence of NTG.7 These findings suggest that non-IOP-dependent factors may contribute to disease progression, and elucidating these factors is necessary to better understand the pathogenesis of glaucoma, especially in the context of NTG. For this purpose, an animal model representing disease characteristics of NTG would be extremely useful. To date, some animal models have been introduced, for example, the optic nerve ligation model shows RGC loss with normal IOP,8 but this is more suitable as a model of ischemia or optic nerve injury. In addition, preparation of these artificial models requires a high level of technical skills, but unfortunately, long-term reproducibility seems to be somewhat limited. Thus, there has been a great demand to create suitable animal models of NTG.

In addition to more extensively studied factors such as reduced ocular blood flow and systemic blood pressure changes, excessive stimulation of the glutamatergic system has been proposed to contribute to the death of RGCs in glaucoma. Excessive extracellular concentrations of glutamate induce uncontrolled elevation of intracellular calcium, which enters through chronically activated glutamate receptors. Glutamate uptake by the glial cells is a well-known mechanism to maintain low extracellular levels of glutamate and promote efficient interneuronal signaling in the central nervous system (CNS). Furthermore, the same process is considered to be neuroprotective during neurodegeneration. Clearance of glutamate from the extracellular space is accomplished primarily by the action of glutamate transporters.9 In the CNS, the glutamate/aspartate transporter (GLAST) and glutamate transporter 1 (GLT-1) are Na+-dependent glutamate transporters found in astrocytes. Genetic deletion of GLAST and/or GLT-1 causes abnormal brain development and neurological symptoms such as motor deficits.9, 10, 11 We have previously reported that GLAST, located in Müller glial cells, is the only glial-type glutamate transporter in the retina, whereas GLT-1 is expressed in neurons, including bipolar cells and photoreceptors.12 Not surprisingly, GLAST is more active than GLT-1 in preventing glutamate neurotoxicity after ischemia.12 In addition, we recently found that GLAST-deficient (GLAST−/− and GLAST+/−) mice show spontaneous RGC death and optic nerve degeneration without elevated IOP.13 Interestingly, GLAST is essential not only to keep the extracellular glutamate concentration below a neurotoxic level but also to maintain glutathione levels by transporting glutamate, which is a substrate for glutathione synthesis, into Müller cells. As retinal concentration of glutathione, a major cellular antioxidant in the retina, was decreased in GLAST-deficient mice, both glutamate neurotoxicity and oxidative stress may be involved in NTG-like pathology.13 Together with the evidence that downregulation of GLAST (human EAAT1) in the retina and of glutathione level in the plasma are found in human glaucoma patients,14, 15 it is appropriate to consider GLAST-deficient mice as a valid and adequate model that offer a powerful system to determine the mechanisms of and evaluate new treatments for NTG.

Apoptosis signal-regulating kinase 1 (ASK1) has key roles in human diseases closely related to the dysfunction of cellular responses to oxidative stress and endoplasmic reticulum stressors, including neurodegenerative diseases.16, 17 We have previously reported that ASK1 is primarily expressed in RGCs, and ASK1−/− mice are less susceptible to ischemic injury.18 The role of ASK1 in glaucoma, however, is unknown. In an attempt to identify the apoptotic signals regulating RGC death in GLAST-deficient mice, we generated GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice and determined the effect of ASK1 deficiency on the NTG-like phenotype.

Results

ASK1 deficiency protects visual function in GLAST-deficient mice

To determine whether ASK1 deficiency is capable of preventing the NTG-like phenotype in GLAST-deficient mice,13 GLAST+/−:ASK1+/− mice were interbred and genotyped at weaning. GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice were born in accordance with Mendelian inheritance ratios, survived into adulthood and were fertile. We first examined the visual function of these mice at 3 months of age (3 M) using multifocal electroretinograms (mfERGs), an established noninvasive method.13 Figure 1a and b show the averaged responses of the second-order kernel in each group. The visual function of WT and ASK1−/− mice was indistinguishable (Figure 1c). As we have previously reported, visual function in the GLAST+/− and GLAST−/− mice was impaired in all visual fields, but was clearly improved by ASK1 deficiency (Figure 1a–c). In particular, the amplitude of the secondary kernel in GLAST+/−:ASK1−/− mice (4.7±0.4 nV/deg2; n=8) was not significantly different compared with WT mice (5.0±0.3 nV/deg2; n=9) (P=0.67, Figure 1c). These results suggest that ASK1 deficiency has no harmful effects during development and prevents visual disturbances in GLAST-deficient mice.

Figure 1
figure 1

Effect of ASK1 on visual response in GLAST+/− and GLAST−/− mice. (a) Summed responses of the second-order kernel examined using multifocal electroretinograms. (b) Averaged responses of the second-order kernel are demonstrated using three-dimensional plots. The degree of retinal function is presented in the color bar. A higher score (red) indicates highly sensitive visual function and a lower score (green) indicates retinal dysfunction. (c) Quantitative analysis of the visual response amplitude. Note the improved visual function of GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice compared with GLAST+/− and GLAST−/− mice, respectively. Values are given in nV per square degree (nV/deg2). **P<0.01, *P<0.05

ASK1 deficiency protects retinal neurons in GLAST-deficient mice

We next analyzed the histopathology of the retina. Consistent with the results of the mfERGs, the retina of ASK1−/− mice showed normal organization at 3 weeks (3 W), 3 M and 6 M (Figure 2). Cell number in the ganglion cell layer (GCL) was significantly decreased after 3 M in GLAST+/− mice and after 3 W in GLAST−/− mice (Figures 2 and 3a). In addition, the thickness of the inner retinal layer (IRL) was decreased after 3 M in both strains (Figure 3b). In GLAST+/−:ASK1−/− mice, however, GCL cell number was significantly increased at 3 M and 6 M compared with GLAST+/− mice (Figure 3a). IRL thickness was increased to a normal level (105±11% at 3 M and 95±11% at 6 M; n=6) in GLAST+/−:ASK1−/− mice (Figure 3b). In GLAST−/−:ASK1−/− mice, IRL thickness was significantly increased at 3 M and 6 M compared with GLAST−/− mice (Figure 3b). In addition, GCL cell number was increased at 3 W and 3 M, but not at 6 M (Figure 3a). These results suggest that ASK1 deficiency prevents the loss of RGCs and secondary retinal degeneration in GLAST-deficient mice.

Figure 2
figure 2

Effect of ASK1 on the progress of glaucoma. H&E staining of retinal sections at 3 weeks, 3 months and 8 months. WT, GLAST+/− and GLAST−/− mice were littermates. GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice were littermates. Scale bar: 100 and 400 μm in the upper and lower rows, respectively. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer

Figure 3
figure 3

Quantification of the severity of glaucoma. (a) RGC number in WT, ASK1−/−, GLAST+/−, GLAST+/−:ASK1−/−, GLAST−/− and GLAST−/−:ASK1−/− mice. The number of neurons in the GCL was counted in retinal sections from one ora serrata through the optic nerve to the other ora serrata. (b) Thickness of the inner retinal layer in WT, ASK1−/−, GLAST+/−, GLAST+/−:ASK1−/−, GLAST−/− and GLAST−/−:ASK1−/− mice. **P<0.01, *P<0.05

ASK1 deficiency prevents optic nerve degeneration in GLAST-deficient mice

As nearly half of the cells in the rodent GCL are displaced amacrine cells, we needed to distinguish RGCs from displaced amacrine cells by retrograde labeling.18 As ASK1 deficiency was most effective in 3 M GLAST+/− mice, we examined RGC number in WT, ASK1−/−, GLAST+/− and GLAST+/−:ASK1−/− mice at 3 M (Figure 4a–h). RGC number per square millimeter in ASK1−/− mice (4200±238; n=3) was normal compared with WT mice (4050±170; n=3) (P=0.64, Figure 4m). In GLAST+/− mice, RGC number (3358±180; n=3) was significantly reduced compared with WT mice (P<0.05). However, RGC number in GLAST+/−:ASK1−/− mice (4067±121; n=3) was clearly increased compared with GLAST+/− mice (P<0.05) and in normal range compared with WT mice (P=0.94, Figure 4m). Similarly, RGC number in GLAST−/−:ASK1−/− mice (3392±102; n=3) was increased compared with GLAST−/− mice (2592±269; n=3) (P<0.05,Figure 4m).

Figure 4
figure 4

Effect of ASK1 on RGC and optic nerve degeneration. (ah) Retrogradely labeled RGCs in WT, ASK1−/−, GLAST+/− and GLAST+/−:ASK1−/− mice. (eh) Magnified images of (ad), respectively. (il) Staining of semi-thin sections with toluidine blue revealed the presence of abnormally dark axonal profiles (arrowheads) and reduced axons in GLAST+/− mice (k), which was ameliorated in GLAST+/−:ASK1−/− mice (l). (m) Quantification of labeled RGC number in WT, ASK1−/−, GLAST+/−, GLAST+/−:ASK1−/−, GLAST−/− and GLAST−/−:ASK1−/− mice. Scale bar: 1 mm (ad); 200 μm (eh); 20 μm (il). *P<0.05

Degeneration of the optic nerve is one of the hallmarks of glaucoma. To analyze morphological changes in the optic nerve, semi-thin transverse sections were cut and stained with toluidine blue (Figure 4i–l). Consistent with severe RGC loss, the degenerating axons in 3 M GLAST+/− mice had abnormally dark axonal profiles (arrowheads in Figure 4k). Such degenerating axons, however, were almost absent in GLAST+/−:ASK1−/− mice (Figure 4l). Taken together, these results demonstrate that ASK1 deficiency protects against RGC loss and optic nerve degeneration in GLAST-deficient mice, which leads to improved visual function as detected by mfERG (Figure 1).

IOP measurement in GLAST/ASK1 double-deficient mice

We have previously reported that GLAST-deficient mice show normal IOP compared with WT mice.13 To determine the effect of ASK1 on IOP, we examined the IOP of ASK1−/−, GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice. IOP measurements were carried out at around 2100 hours, when IOP is highest in mouse eyes.19 The IOP values of ASK1−/−, GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice were not significantly decreased compared with WT and GLAST−/− mice (Figure 5). These results suggest that the recovery of NTG-like pathology in GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice is IOP independent.

Figure 5
figure 5

Effect of ASK1 on intraocular pressure in GLAST+/− and GLAST−/− mice. Intraocular pressure in GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice was not significantly decreased compared with GLAST−/− mice

Role of oxidative stress and glutamate neurotoxicity in GLAST/ASK1 double-deficient mice

Oxidative stress has been proposed to contribute to RGC death in glaucoma, and a reduction in glutathione levels was reported in the plasma of human glaucoma patients.15 Consistent with these findings, we have previously reported a decreased glutathione concentration in the retina of GLAST−/− mice.13 To determine the effect of ASK1 on glutathione synthesis, we examined the glutathione concentration in the retina of 6 M ASK1−/− and GLAST−/−:ASK1−/− mice and found that it was not significantly increased compared with WT and GLAST−/− mice, respectively (Figure 6a). In addition, the malondialdehyde concentration in the retina of ASK1−/− and GLAST−/−:ASK1−/− mice was indistinguishable from that of WT and GLAST−/− mice, respectively (Figure 6b).

Figure 6
figure 6

Effect of ASK1 on oxidative stress in GLAST+/− and GLAST−/− mice. (a) Mean glutathione concentration in whole retinas from WT, ASK1−/−, GLAST+/−, GLAST+/−:ASK1−/−, GLAST−/− and GLAST−/−:ASK1−/− mice. (b) Malondialdehyde concentration in whole retinas from WT, ASK1−/−, GLAST+/−, GLAST+/−:ASK1−/−, GLAST−/− and GLAST−/−:ASK1−/− mice. *P<0.05

We have previously reported that intravitreal glutamate concentration is normal, but memantine, N-methyl-D-aspartate receptor antagonist, partially protected RGCs in GLAST−/− mice.13 In addition, we showed that GLAST has a major role in glutamate uptake into Müller glial cells.20 To explore the possibility that ASK1 is involved in glutamate transport, we examined glutamate uptake activity in Müller glial cells prepared from ASK1−/− and GLAST+/−:ASK1−/− mice, and found that it was not significantly increased compared with WT and GLAST+/− mice, respectively (Figure 7). These findings suggest that ASK1 deficiency attenuates NTG-like degeneration without affecting the conditions of oxidative stress and glutamate neurotoxicity in GLAST-deficient mice.

Figure 7
figure 7

Glutamate uptake activity in Müller cells derived from WT, ASK1−/−, GLAST+/− and GLAST+/−:ASK1−/− mice. The data are presented as means±S.E.M. of eight samples for each group. *P<0.05

Effect of ASK1-p38 mitogen-activated protein kinase (MAPK) signaling in Müller glial cells and RGCs

ASK1 is activated in response to cytotoxic stresses, including reactive oxygen species (ROS) and tumor necrosis factor (TNF), and relays these signals to p38 MAPK.16, 17 To determine whether this pathway is active in Müller glial cells, we first examined the effects of TNF on cultured Müller cells from WT and ASK1−/− mice. Western blot analysis demonstrated that stimulation of WT Müller cells with TNF leads to strong phosphorylation of p38 in a dose-dependent manner (Figure 8a). The activation of p38, however, was significantly suppressed in ASK1-deficient Müller cells (Figure 8a). Nitric oxide (NO) generated by inducible nitric oxide synthase (iNOS) is involved in retinal neuronal cell death,21, 22 and a previous study has reported that TNF-induced iNOS expression and NO release are suppressed by a specific inhibitor of p38 in mouse astrocytes.23 These results suggest that the ASK1-p38 pathway regulates TNF-induced iNOS expression in Müller cells. To evaluate this possibility, we next examined iNOS protein levels in cultured Müller cells. In untreated Müller cells, iNOS protein was almost absent, but TNF clearly increased iNOS expression levels (Figure 8b). Similar iNOS induction was detected in GLAST-deficient Müller cells (Figure 8b). However, TNF-induced iNOS expression was completely suppressed in ASK1-deficient Müller cells (Figure 8b). These results suggest that the ASK1-p38 pathway is required in Müller cells for the TNF-induced iNOS production, which may lead to the death of retinal neurons including RGCs. We further examined the direct effect of TNF on cultured RGCs.18 TNF-induced cell death in cultured RGCs from ASK1-deficient mice was significantly decreased (41±9%; n=6) compared with that from WT mice (P<0.05, Figure 8c). Taken together, loss of ASK1 prevents TNF-induced RGC death through both the direct pathway and the indirect pathway through Müller cells that is independent of GLAST.

Figure 8
figure 8

ASK1 is required for TNF-induced p38 activation and iNOS production in Müller glial cells. (a) Effect of ASK1 on TNF-induced p38 activation. Müller cells derived from WT or ASK1−/− mice were stimulated with the indicated concentration of TNF for 20 min, followed by immunoblot analysis of total and phosphorylated p38 in cell lysates. (b) Effect of GLAST and ASK1 on TNF-induced iNOS production. Müller cells derived from WT, GLAST−/− or ASK1−/− mice were stimulated for 16 h with TNF (50 ng/ml) or left unstimulated. (c) Effect of TNF on RGC death. RGCs derived from WT or ASK1−/− mice were treated for 2 days with TNF (400 ng/ml), and the number of PI-positive cells was counted. The data are presented as mean ± S.E.M. of six samples for each group

Discussion

In this study, we show that ASK1 is associated with progressive RGC loss, glaucomatous optic nerve degeneration and visual disturbances in GLAST-deficient mice. We previously suggested the possibility that dysfunction of GLAST (human EAAT1) has a role in RGC death in human NTG.13 It has been reported that EAAT1 is downregulated in the retinas of human patients with glaucoma14 and in fibroblasts from patients with Alzheimer's disease.24 Considering the high frequency of glaucoma in Alzheimer's disease patients,25 common mechanisms such as GLAST dysfunction might contribute to both diseases. Interestingly, ROS-mediated ASK1 activation is a key mechanism for amyloid beta (Aβ)-induced neurotoxicity, which has a central role in Alzheimer’s disease.26 The accumulation of Aβ is also observed in apoptotic RGCs in a rat model of glaucoma due to high IOP.27 In this model, inhibiting amyloidogenic pathways by agents affecting multiple stages of the Aβ pathway reduces RGC apoptosis in vivo.27 This suggests a new hypothesis for RGC death in glaucoma involving ASK1-dependent Aβ neurotoxicity, mimicking Alzheimer's disease.28 In addition, multiple single-nucleotide polymorphisms in the Toll-like receptor 4 (TLR4) gene have been associated with the risk of NTG.29 As ASK1 is required for innate immune responses dependent on TLR4,30 TLR4-ASK1 signaling may be involved in the development of NTG. Taken together, these findings suggest that ASK1 may have roles in various neurodegenerative disorders, including glaucoma.16, 17, 18

Our present results demonstrate reduced RGC death, decreased axonal loss and mild visual disturbance in GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice, highlighting ASK1 as a potential therapeutic target for NTG. Loss of ASK1 had no effect on IOP (Figure 5), the production of malondialdehyde (Figure 6) or glutamate uptake activity by Müller cells (Figure 7). Whereas TNF-induced iNOS production was suppressed in ASK1-deficient Müller cells, and ASK1-deficient RGCs were more resistant to TNF-induced death compared with WT RGCs (Figure 8). Thus, in combination with conventional treatments to lower IOP, inhibition of ASK1 signaling may be useful in the treatment of glaucoma. In addition, we recently reported that interleukin-1 (IL-1) increases glutamate uptake by Müller cells, primarily through the activation of GLAST, and protects RGCs from glutamate neurotoxicity.20, 31 IL-1 is a mediator of brain injury induced by ischemia or trauma and has been implicated in chronic brain diseases, such as Alzheimer's disease, Parkinson's disease and multiple sclerosis.32 Therefore, we need to examine the beneficial and detrimental roles of IL-1 during glaucoma in vivo. Furthermore, we are undertaking experiments to determine the neuroprotective effect of GLAST against neurotoxicity, axotomy and neuroinflammation in mice overexpressing GLAST.

A recent study has shown that upregulation of GLAST in Müller cells by glial cell line-derived neurotrophic factor (GDNF) and neurturin (NTN) is required to rescue RGCs following optic nerve transection.33 As the receptors for GDNF and NTN are increased in Müller cells after RGC axotomy, the neuroprotective effects of GDNF and NTN may be indirect, at least partly, through the enhancement of glutamate uptake in Müller cells. Similar upregulation of the receptors for GDNF and NTN has been observed in a rat model of photoreceptor degeneration.34, 35 In this animal model, trophic factors such as nerve growth factor, brain-derived neurotrophic factor and basic fibroblast growth factor increase the production of multiple trophic factors in Müller cells, which indirectly leads to photoreceptor survival.35, 36, 37 In addition, nerve growth factor eye drops may prevent the progress of human glaucoma.38 In this study, we found that the loss of ASK1 prevented the activity of p38 and TNF-induced iNOS production in Müller cells (Figure 8). Recent studies have shown that TNF and NO can induce RGC death and participate in the pathophysiology of glaucoma.39, 40, 41 Our results suggest that the ASK1-p38 pathway is involved in the process of TNF-induced RGC degeneration in neighboring glial cells, as well as in the RGC itself (Figure 8).18 Taken together, these findings suggest that such a glial–neuronal network may be functional in various forms of neurodegenerative diseases and that ASK1, NO, GLAST and trophic factors in Müller cells have important roles in this network during glaucoma.42, 43 Thus, further efforts to discover new compounds that can enhance glutamate uptake and inhibit ASK1 signaling for a prolonged period may lead to the development of novel strategies for the management of glaucoma, including NTG.

Materials and Methods

Mice

Experiments were carried out using ASK1−/−,18 GLAST+/− and GLAST−/− mice12, 13 in accordance with the Tokyo Metropolitan Institute for Neuroscience Guidelines for the Care and Use of Animals. After mating ASK1−/− and GLAST−/− mice, GLAST+/−:ASK1+/− mice were interbred to obtain GLAST+/−:ASK1−/− and GLAST−/−:ASK1−/− mice.

Histological and morphometric studies

Paraffin retinal sections of 7 μm thickness were cut through the optic nerve and stained with hematoxylin and eosin. RGC number and the extent of retinal degeneration were quantified in three ways.18 First, the thickness of the IRL (between the internal limiting membrane and the interface of the outer plexiform layer and the outer nuclear layer) was analyzed. Second, in the same sections, the number of neurons in GCL was counted from one ora serrata through the optic nerve to the other ora serrata. Third, RGCs were retrogradely labeled from the superior colliculus with Fluoro-Gold (FG; Fluorochrome, Englewood, CO, USA). At 7 days after FG application, the eyes were enucleated, and the retinas were detached and prepared as flattened whole mounts in 4% PFA in 0.1 M PBS solution. The GCL was examined in whole-mounted retinas with fluorescence microscopy to determine RGC density. Four standard areas (0.04 mm2) of each retina at a point 0.1 mm from the optic disc were randomly chosen. Labeled cells were counted by observers blinded to the identity of the mice, and the average number of RGCs/mm2 was calculated. The changes in RGC number were expressed as a percentage of the WT control eyes.

For detailed morphological analysis, optic nerves were fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer overnight at 4°C. After dissection, the pieces of tissue were placed in 1% osmium tetroxide, and after dehydration, the pieces were embedded in Epon (Nisshin EM, Tokyo, Japan). Transverse semi-thin (1 μm) sections were stained with 0.2% toluidine blue in 1.0% sodium borate.13, 44

IOP measurement

IOP was measured by a previously validated commercial rebound tonometer (TonoLab; Colonial Medical Supply, Franconia, NH, USA) in anesthetized mice as reported previously.45 To minimize variation, the data were collected during a time window 4–6 min after injection of the anesthetic during which IOP plateaus. The animals were 6 months of age, and their body weights ranged from 22–36 g at the time of IOP measurement. As the 24-h IOP pattern in mouse eyes is biphasic, with IOP being highest around 2100 hours,19 we examined IOP between 2000 and 2300 hours.

mfERG

Mice (3 months old) were anesthetized by intraperitoneal injection of a mixture of xylazine (10 mg/kg) and ketamine (25 mg/kg). The pupils were dilated with 0.5% phenylephrine hydrochloride and 0.5% tropicamide. mfERGs were recorded using a VERIS 6.0 system (Electro-Diagnostic Imaging, Redwood City, CA, USA). The visual stimulus consisted of seven hexagonal areas scaled with eccentricity. The stimulus array was displayed on a high-resolution black and white monitor driven at a frame rate of 100 Hz. The second-order kernel, which is impaired in patients with glaucoma, was analyzed.13, 44

Malondialdehyde and glutathione assay

The concentrations of malondialdehyde and glutathione were measured using the Bioxytech LPO-586 (Oxis Research, Beverly Hills, CA, USA) and Glutathione Assay Kit (Cayman Chemical, Ann Arbor, MI, USA), according to the manufacturers’ protocols.

Glutamate uptake assay

Glutamate uptake assay in primary cultured Müller cells was carried out as previously reported.20, 31 Müller cells were cultured in 5.5 mM glucose-containing DMEM supplemented with 10% fetal bovine serum. The culture media were replaced with a modified Hanks’ balanced salt solution (HBSS) for a 20-min preincubation, before the addition of 0.025 mCi/ml L-[3H]-glutamate (Amersham, Uppsala, Sweden) and 100 μM unlabeled glutamate to the media. Uptake was terminated after 7 min by three washes in ice-cold HBSS, immediately followed by cell lysis in 0.1 M NaOH. Aliquots were taken for scintillation counting, and protein concentration was determined using BSA standards.

Immunoblot analysis

Primary Müller cells were obtained as previously reported35, 36, 37 and treated with TNF at various concentrations for 20 min or 16 h. Immunoblotting was carried out as previously reported.20, 31 Membranes were incubated with antibodies against p38 (1 : 1000; BD Biosciences Pharmingen, San Diego, CA, USA), phospho-p38 (1 : 1000; BD Biosciences Pharmingen) or iNOS (1 : 1000; BD Biosciences Pharmingen).

Assessment of TNF-induced cell death in cultured RGCs

RGCs derived form WT and ASK1−/− mice were seeded at a density of 5 × 104 cells per well and cultured with 0.1 ml of medium on a 96-well culture plate.18 After 2 days, they were stimulated with 400 ng/ml of TNF for 2 days, and dying RGC number was counted after staining with propidium iodide.

Statistics

For statistical comparison of two samples, we used a two-tailed Student's t-test. Data are presented as the mean±S.E.M. P<0.05 was regarded as statistically significant.