Caspase Activation in Retinas of Diabetic and Galactosemic Mice and Diabetic Patients

Caspase substrates for cas-1 (YVAD-AFC), cas-2 (VDVAD-AFC), cas-3 (DEVD-AFC), cas-4 (LEVD-AFC), cas-5 (WEHD-AFC), cas-6 (VEID-AFC), cas-8 (IETD-AFC), and cas-9 (LEHD-AFC) were purchased from Calbiochem. Primary antibody against mouse cas-1 was from Zymed, and antibody against subunits of cas-3 was from New England Biochemicals. All secondary antibodies were obtained from Santa Cruz. In Situ Cell Death Detection Kit was from Promega. Annexin V was purchased from R&D Systems. Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), endothelial cell growth factor (ECGF), and penicillin-streptomycin were from Fisher. Endothelial cell basal medium (EBM), plasma-derived horse serum, heparin, and all other chemicals were obtained from Sigma.

Male mice (C57BL/6) that weighed 20 g were randomly assigned to be made diabetic, galactosemic, or control. Diabetes was induced by streptozotocin (STZ) injections (60 mg/kg body wt i.p. on 5 consecutive days), and insulin was given as needed to achieve slow weight gain without preventing hyperglycemia and glycosuria (0.1–0.2 units of NPH insulin subcutaneously, 2–3 times a week). Experimental galactosemia was produced by feeding a diet of Teklad 7004 (Harlan, WI) enriched with 30% d-galactose. Animals were caged in pairs, had free access to food and water, and were maintained under a 14-h on/10-h off light cycle. Body weight was measured weekly. Treatment of animals conformed to the Association for Research in Vision and Ophthalmology Resolution on Treatment of Animals in Research. Shortly before the animals were killed, the severity of blood hexose elevation was estimated by measuring the level of nonenzymatic GHb using affinity chromatography (Glyc-Affin; Pierce, Rockford, IL). Animals in whom GHb was less than twice the normal value were excluded from additional experiments.

Animals were killed after 2, 4, and 8 months of diabetes or galactosemia. These durations precede the appearance of lesions of retinopathy in mice (T.S.K., unpublished observation). Age-matched normal animals served as controls. The retinas of both eyes were isolated and immediately resuspended in 200 μl of lysate buffer (100 mmol/l HEPES buffer [pH 7.5] containing 0.1% 3-[ (3-cholamidopropyl)dimethyl-ammonio]-1-propanesulfonate, 10% sucrose, 1 mmol/l EDTA, 1 mmol/l phenylmethylsulfonyl fluoride, and 1 μg/ml leupeptin) and placed on ice for 30 min. After 10 s of sonication on ice, lysates were centrifuged at 9,000g for 5 min at 4°C. The supernatants were frozen for additional experiments. Protein measurement was done using the BioRad assay.

Human eyes were obtained through the National Disease Research Interchange (Philadelphia, PA). The donors remained anonymous, but age, presence or absence of diabetes, and cause of death were noted. Severity of diabetes and severity of retinopathy were not available, but none of the patients had laser photocoagulation. The characteristics of the patients with diabetes and nondiabetic control subjects are presented in Table 3. The eyes of the donors with and without diabetes were enucleated, packed on wet ice, and obtained by our laboratory 15 ± 3 h after death. For the experiments described below, the retina was isolated and two pieces from the temporal/superior part of the retina were resuspended in 200 μl of lysate buffer, placed on ice for 30 min, sonicated for 10 s, and centrifuged at 9,000g for 5 min at 4°C, followed by protein measurement of the supernatant, as described above.

Retinal Müller cells (rMC-1 cells/4 × 10⁶ cells) were grown in 100-mm Petri dishes in 10 ml of DMEM (5 mmol/l glucose) supplemented with 10% FBS and 1% penicillin/streptomycin (growth medium) at 37°C/ 5% CO₂ overnight. The next day, medium was replaced by 10 ml of DMEM containing 2% FBS, 1% penicillin/streptomycin, and either 5 or 25 mmol/l glucose (treatment medium). Medium was changed every other day. Only cells from passage 20 and lower were used for the experiments.

Bovine retinal endothelial cells were isolated as described (15). Isolated endothelial cells were grown in 100-mm Petri dishes coated with 0.1% gelatin in 10 ml of EBM medium supplemented with 10% plasma-derived horse serum, 5% ECGF, 1% penicillin/streptomycin, and 5% heparin (growth medium) at 37°C/5% CO₂ until 75% confluence. Then, medium was replaced by 10 ml of EBM medium containing 2% FBS, 5% ECGF, 1% penicillin/streptomycin, 5% heparin, and either 5 or 25 mmol/l glucose (treatment medium). Medium was changed every other day. Only cells from passage 8 and lower were used for the experiments.

At times indicated, cells were lysed in 200 μl of lysate buffer, sonicated for 10 s, and centrifuged at 9,000g for 5 min at 4°C, followed by protein measurement of the supernatant, as described above.

rMC-1 cells (1 × 10⁵) were grown on coverslips and incubated in treatment medium containing 5 or 25 mmol/l glucose as described above. Annexin V staining was done according to the manufacturer’s instructions. Briefly, after incubation of rMC-1 cells for 48 and 96 h, the medium was removed and cells were washed twice with annexin V binding buffer (2 ml). For staining, cells were incubated in 500 μl of annexin V staining solution (1:50 dilution of annexin V-phycoerythrin in annexin V binding buffer) for 15 min in the dark at 37°C/5% CO₂. The annexin V staining solution was removed, and coverslips were immediately mounted on slides using an antifading fluorescent mounting medium (Vector Shield) and analyzed by fluorescence microscopy (Zeiss Axiophot; Karl Zeiss, Thornwood, NY) at excitation of 540 nm and emission of 600 nm.

Caspase activity was measured as described previously (16). Briefly, equal amounts of proteins (usually 15 μg) were incubated in lysate buffer containing the fluorogenic caspase substrate (2.5 μ mol/l) in a total volume of 100 μl at 32°C for 1 h. Cleavage of the substrate emits a fluorescence signal that was quantified by a Tecan Spectra FluorPlus fluorescence plate reader (excitation: 400 nm, emission: 505 nm). Activities of cas-4 and cas-5 were averaged together, because evidence suggests that they cleave similar substrates (17). The average value is represented as cas-4/5.

Equal amounts of proteins (50 μg) were separated in a prepared 4–20% SDS gradient gel (BioRad) by electrophoresis and blotted onto nitrocellulose membrane (100 V, 90 min). The membrane was washed three times (5 min) with PBS containing 0.05% Tween 20, blocked with 5% dry milk in PBS/0.05% Tween 20 (2 h, room temperature), and incubated with a primary antibody (usually 1:1,000 dilution in PBS/ 0.05% Tween 20) overnight at 4°C. After primary incubation, the membranes were washed five times (5 min) and incubated with secondary antibody (1:3,000 dilution) for 1 h at room temperature. After washing (5 × 5 min) of the membranes in PBS/ 0.05% Tween 20, membranes were developed using enhanced chemiluminescence (Amersham).

Cell death in retinal trypsin digests was detected with the TUNEL method using fluorescein isothiocyanate (FITC)–based In Situ Cell Death Detection Kit. The retinal vasculature was isolated by the trypsin digest technique, as reported previously by us, after 5 days of formalin fixation (2). The digestion buffer contained 0.2 mol/l sodium fluoride to inhibit DNA fragmentation as a result of DNAses contaminating the crude trypsin preparation. The digests were rehydrated in PBS and stained according to the manufacturer’s instructions. In each experiment, the negative control received only the label solution without the terminal transferase, and the positive control was exposed to DNase (1 μg/ml in 40 mmol/l Tris-HCl, 6 mmol/l MgCl₂ buffer [pH 7.5]) for 10 min at room temperature before the TUNEL reaction. TUNEL-positive cells were detected by the presence of FITC-positive stain using a Zeiss Axiophot fluorescence microscope.

Experimental groups were compared statistically using the Mann-Whitney U test (two-tailed). Means with P < 0.05 were considered statistically significant.

Blood hexose levels in diabetic and galactose-fed mice were elevated, as evidenced by significantly elevated GHb percentage and abnormal body weight compared with normal age-matched animals. Diabetic animals also had significantly elevated blood glucose levels. Table 1 summarizes the characteristics of the experimental groups at different durations (2, 4, and 8 months).

It was shown previously that apoptosis of retinal cells precedes the development of morphologic lesions of diabetic retinopathy. Activation of caspase is necessary to initiate or execute apoptosis. To identify caspase or a possible caspase cascade involved in the apoptotic process in the retina of diabetic patients, we measured caspase activities of eight caspases (cas-1, -2, -3, -4, -5, -6, -8, and -9) in the retinas of diabetic and normal animals using commercially available substrates. The activities detected using substrates for the human cas-4 and -5 likely represent the activity of mouse cas-11 (for which no substrate is available), because human cas-4 and -5 are close homologues of mouse cas-11 (17). Figure 1 shows the time course of caspase activation in the retina of diabetic mice compared with normal age-matched animals.

At 2 months’ duration of diabetes, activities of cas-1, -2, -4/5, -6, -8, and -9 were significantly elevated in comparison with normal age-matched animals (Fig. 1, top panel). Activities of initiator caspases such as cas-2, -8, and -9 increased by 7.5–15% above normal at this duration of diabetes. The activity of the executioner cas-6 increased by 15%, whereas cas-3, another executioner caspase and the most common marker for apoptosis, was not yet activated. The activities of cas-1 and -4/5, members of the subfamily that is most closely associated with proinflammatory cytokine production, were strongly elevated (6 and 16%, respectively) at 2 months of diabetes.

At 4 months of diabetes, the pattern of caspase activities changed. Cas-1 activity slightly increased up to 8%, whereas all other caspase activities, especially those of other initiators, declined (but were still significantly elevated above normal). The activity of cas-2 could not be detected anymore. The activity of cas-3 (3.5%) was significantly detectable for the first time (Fig. 1, middle panel).

At 8 months’ duration of diabetes, members of the cas-1 subfamily, such as cas-1 and -4/5, were still active. Activities of all initiator caspases were undetectable at this duration. The activity of the executioner cas-3 increased up to 8%, whereas that of the executioner cas-6 remained unchanged (Fig. 1, bottom panel).

Figure 2 shows a linear correlation of cas-1 activity in the retinas of diabetic animals (n = 9) with the severity of diabetes at 2 months’ duration. The correlation was statistically significant (P < 0.025) with a correlation factor of r = 0.742.

Because we identified caspase activation in the retina of diabetic animals, evaluation of caspase activity in another model for diabetic retinopathy, experimental galactosemia, would provide insight on which caspases were closely associated with the development of lesions like those characteristic of diabetic retinopathy.

At 2 months of galactose feeding, only cas-1 and -4/5 were strongly activated (14 and 6.4% above normal, respectively). Thereafter, cas-6 was activated at 4 months and cas-9 at 8 months of galactosemia. Thus, the pattern of caspase activation observed in retinas of galactosemic animals differed from those observed in diabetic animals. Cas-1, -4/5, and -6 activities disappeared in the retinas of galactosemic animals with longer duration. At 8 months of galactosemia, cas-1 activity even dropped below normal (− 7.3%). Cas-3 activity was not detected in the retinas of galactose-fed mice at these different durations (Table 2).

Evaluation of which caspases are activated on the basis of measurement of enzyme activities depends heavily on the specificity of the substrates for individual caspases. To confirm activation of several caspases, we assessed the cleavage of the caspase proenzyme into its subunits by Western blot.

Figure 3 shows the proenzyme and the active p20 subunit of cas-1 in the retinal lysates of normal, diabetic, and galactosemic mice at 2 and 8 months, respectively (Figs. 3A and B). In the retinas of normal animals, the p20 cas-1 subunit could not be detected at any time point.

Cas-3 activation was confirmed using a cas-3 antibody specifically detecting cleavage products of cas-3, such as p20 (inactive) and p17 (active) subunits. The smaller p10 subunit of cas-3 cannot be detected by this antibody. Activation of cas-3 into its active subunits occurs via an intermediate inactive p20 subunit that needs to be cleaved further to the active p17 subunit.

As Fig. 4 shows, the active p17 subunit of cas-3 could be detected in the lysates of retinas of animals that were diabetic for 8 months. In contrast, neither normal nor galactosemic animals showed any cas-3 cleavage products at 8 months’ duration. The lack of an active cas-3 subunit in the retinas of galactose-fed animals is consistent with the absence of cas-3 activity in the retina of these animals. Cas-6 activation also was confirmed by Western blot (data not shown).

The frequency of TUNEL-positive cells in retinal capillaries of diabetic mice was slightly greater than normal at 24 weeks: 2.3 ± 0.9 TUNEL-positive capillary cells in diabetic animals (n = 4) compared with 1.0 ± 0.8 TUNEL-positive capillary cells in normal animals (n = 4; P < 0.05). The frequency of TUNEL-positive cells in the retinal vasculature tended to be greater at 48 weeks of diabetes (7.9 ± 1.6 in diabetic animals and 1.3 ± 0.6 in normal; both n = 2).

Caspase activities were measured in lysates of human retinas obtained postmortem from patients with (n = 5) and without (n = 4) diabetes. Table 3 summarizes the characteristics of the patients.

Consistent with the findings in diabetic mice, the activities of some caspases in retinas of patients with diabetes, such as cas-1 (5.2%), cas-3 (5%), cas-4 (4.5%), and cas-6 (13%), were significantly greater than normal (P < 0.05, two-tailed). Activities of other caspases, however, remained equivalent to that observed in normal patients (Fig. 5).

For the experiments described above, caspase activities were measured from lysates of whole retinas. To identify possible caspase cascades in different cell types, we studied the effect of hyperglycemia on caspase activation in vitro using two different cell types. The rat rMC-1 and primary bovine retinal endothelial cells (BRECs) were incubated in high glucose (25 mmol/l) and low glucose (5 mmol/l) for up to 7 days. Because of the faster growth of rMC-1 cells, cells were incubated for only 5 days to avoid apoptosis induced by overgrowth or starvation. Figure 6A shows that high glucose induced the activation of cas-1 family members in rMC-1 cells, as well as cas-8 (6.4%/day 3) and subsequent cas-9 (6.1%/day 5). Hyperglycemia led to a strong activation of the executioner cas-6 (17.3%/day 5) and a modest activation of cas-3. However, the activity of cas-3 induced by hyperglycemia varied and became completely suppressed in higher passages, possibly as a result of changes in the SV 40 transfected cell line.

In contrast, hyperglycemia did not activate any member of the cas-1 subfamily in BRECs. Cas-3 activity was strongly induced (22%) at 7 days of high glucose treatment (Fig. 6B). Activation of cas-8 (11.3%/day 5) was stronger in BRECs than in rMC-1 cells, but activity of cas-9 (3.1%/day 5) was not significantly induced in endothelial cells.

Inhibition of caspase activities using the broad-spectrum caspase inhibitor zVAD-fmk (50 μmol/l) inhibited hyperglycemia-induced apoptosis in rMC-1 cells, as shown by annexin V staining (Fig. 7). DMSO (solvent) had no effect on glucose-induced apoptosis.

Apoptosis of retinal capillary cells and subsequent nonperfusion of these capillaries have been postulated to play an important role in the development of diabetic retinopathy (2). Caspase activation is crucial to initiate and execute apoptotic processes within cells (6,8). Our study reveals that, indeed, caspases become activated in retinas of diabetic, galactose-fed mice and in retinas of patients with diabetes, and inhibition of caspases prevented cell death caused by elevated glucose concentrations. To determine caspase activation in our study, we actually measured caspases activities, rather than relying on techniques that are able to detect only structural evidence of caspase activation (e.g., Western blot). Measuring caspase activities has the advantage of determining whether activation of a specific caspase actually led to an active, functional enzyme. Detection of a cleaved caspase active subunit (activation) does not always guarantee increased enzymatic activity, because caspases are proteolytic enzymes that contain a highly reactive cysteine in their active site and are known to be very sensitive toward oxidative as well as nitrosative stress (18). However, Western blots were done to test the specificity of the substrates used.

Initiator caspases, such as cas-8, -9, and -2, are known to be able to activate proteolytically executioner caspases, such as cas-3, -6, or -7 (19,20). Consistent with this, a time course at different stages of diabetes has shown a distinct pattern of caspase activation and deactivation in the retina. The increase in caspase activities was modest but statistically significant and consistent with the slow rate of apoptosis and the slow development of diabetic retinopathy. The activation of several initiator caspases occurred early in the course of diabetes. Apoptosis initiator caspases induce distinct signaling pathways. Diabetes activates cas-8 and -9, suggesting the activation of both an extrinsic and an intrinsic pathway. Cas-8 is known to become activated via death receptors leading to two possible subsequent pathways that induce apoptosis (21–23). In the first, type I (extrinsic), cas-8 is strongly activated via a tight death-induced signaling complex formation at the receptor. The strong activation of cas-8 allows the direct activation of cas-3. In the second, type II (extrinsic/intrinsic), cas-8 is less strongly activated, and the signal is amplified via the mitochondria, in which cas-9 is activated by Apaf-1 and cytochrome c that has been released from the mitochondria subsequently activating downstream executioners (24–26). In addition, DNA damage, oxidative stress, and drugs can activate both cas-9 isoforms (mitochondrial and nuclear) directly (intrinsic) (27,28). Very little is known about cas-2, but it has been found to be an inducer of apoptosis in neuronal cells and retinal ganglion cell (29, 30). Because of its long prodomain, it has been considered an upstream caspase binding to anchor molecules of death receptors, but other reports have also demonstrated a very downstream activation of cas-2 (31).

The activation of cas-3, a downstream executioner, occurred later than the activation of other caspases. Cas-3 is the executioner caspase known to activate the apoptotic endonuclease DFF40, which is responsible for the DNA laddering, a characteristic of apoptosis (32,33). Compared with the late activation of cas-3, our results have demonstrated an earlier activation of cas-6, another executioner. Reports have shown rapid and slow activation of cas-6, suggesting that the place of cas-6 in a cascade might depend on the cell type (34,35). Cas-6 activation has been closely associated with neuronal apoptosis and the development of neuronal diseases such as Alzheimer’s disease (34,36). Likewise, perhaps the early activation of cas-6 is related to the early detection of apoptotic retinal ganglion cells in diabetic rats (3).

The pattern of caspase activation in the retina differed between diabetic and galactose-fed animals. Cas-9 activation seemed to be delayed in the retinas of galactose-fed mice compared with diabetic mice, and activation of cas-2 and -8 were not detectable in galactosemia at the durations used for the study. Perhaps this difference between diabetes and galactosemia contributes to the observed different response of the two retinopathies to aminoguanidine therapy (37): aminoguanidine was found to inhibit both capillary cell apoptosis and capillary degeneration in diabetes but did not have an effect in galactose-fed animals. It is interesting that common to both diabetes and galactosemia was the early activation of cas-1 and its family members, raising the possibility that activation of this caspase subfamily might play an important role in the formation of histological lesions observed in the two models for diabetic retinopathy. The finding of cas-1 and -4 activation in retinas of patients with diabetes strengthens this idea and is consistent with a growing awareness that diabetic retinopathy has several characteristics of a chronic inflammatory disease (e.g., increased nitric oxide production, vaso-occlusion, and vascular permeability) (38–40).

How cas-1 and its family members cas-4, -5, and -11 (cas-11 is the mouse homologue to human cas-4 and -5) become activated in diabetes and their role in apoptosis are still uncertain. Some reports have shown that activation of cas-1 is involved in the induction of fibroblast apoptosis, and cas-1 activation also seemed to be related to apoptosis in several neurodegenerative diseases and subsequent ischemic events (41,42). Cas-1 has also been shown to activate cas-3 as well as cas-6 (43). However, the major function of cas-1 has been closely related to proinflammatory cytokine production, such as interleukin-1β and -18 (44,45). Some reports have shown that cas-11 is necessary to activate cas-1 in mice, which is consistent with our finding of an early induction of a cas-11–like activity (17,46).

Our in vitro studies have demonstrated that high glucose itself is able to induce the activation of the proinflammatory cas-1 family in rMC cells. Cas-8 and -9 were significantly activated in these cells, indicating an involvement of the mitochondria and type II (extrinsic/intrinsic) signaling pathway. In contrast, elevated levels of glucose did not activate members of the cas-1 family in retinal endothelial cells but led to a strong cas-8 response, suggesting a possible type I (extrinsic) signaling pathway (especially because the cas-9 activation was not significant). The in vitro studies of retinal endothelial and Müller cells strongly indicate that high glucose treatment induces different caspase cascades in these cell types, suggesting that the findings in retinal homogenates represent different caspase cascades occurring in different cell types. Full understanding of caspase cascades and cell type specificity will require more selective substrates and antibodies. As expected, inhibition of caspases prevented high glucose–induced cell death, strengthening that elevated glucose clearly causes caspase-dependent apoptosis but not necrosis.

In total, our data suggest that diabetes or elevated glucose activates several slowly progressing cascades of caspases, in which initiator caspases activate a sequence of other caspases that results in multiplication of the death signal within a cell and ultimately leads to activation of executioner caspases and cell death. The degree of caspase activation by diabetes may seem small but is statistically significant. So far, we are not able to compare the increase in caspase activities with other chronic diseases (where apoptosis is known to be important, e.g., arthritis and Alzheimer’s disease), because these studies do not assess enzymatic activity, only immunohistochemical demonstration of increased caspase cleavage. Our in vitro demonstration that inhibition of caspase activity can prevent cell death caused by elevated glucose offers a new potential target, at which the role of apoptosis in the pathogenesis of diabetic retinopathy might be evaluated. Identifying the role of both the proinflammatory and proapoptotic caspase network might offer new therapeutic strategies for the inhibition of this retinal disease. However, the long-term use of caspase inhibitors has not been evaluated in any disease and will require more extensive studies.

This work was supported by a Career Development Award from the Juvenile Diabetes Research Foundation (2-2000-390 to S.M.) and by grants from the Diabetes Association of Greater Cleveland (415-99 to S.M.) and the National Institutes of Health (EY00300 to T.S.K. and DK57733 to T.S.K. and S.M.). Core Facilities of the CWRU Visual Science Research Center (P30EY11373) facilitated the work. We recognize the technical assistance of Casey Miller, PhD, and thank the National Disease Research Interchange (Philadelphia, PA) for human tissue. We also thank Dr. V. Sarthy (Chicago) for the rMC-1 cell line.