Protein Kinase C-δ Mediates Neuronal Apoptosis in the Retinas of Diabetic Rats via the Akt Signaling Pathway

Six-week-old male OLETF and Long-Evans Tokushima Otsuka (LETO) rats were obtained from the Otsuka Pharmaceutical Tokushima Research Institute (Tokushima, Japan). We used 8 LETO and 8 OLETF rats at 24 weeks of age and 28 LETO and 28 OLETF rats at 35 weeks of age. Rats were housed in groups of three animals and supplied with water and food ad libitum under ambient temperature conditions (22 ± 2°C) and a 12-h light/dark cycle, in accordance with the protocol of the institutional review board. We randomly selected five LETO and five OLETF rats at 24 and 35 weeks of age and measured their body weights and blood glucose levels. Blood samples were obtained by tail snipping after a 2-h fasting period. Blood glucose levels were measured using the SureStep (LifeScan, Milpitas, CA).

Rats were anesthetized with an intraperitoneal injection of 50 mg/kg sodium pentobarbital followed by topical application of 0.5% proparacaine to the eye. For intravitreal injection, a 30-gauge needle was inserted into the vitreous 2 mm posterior to the limbus through the pars plana using a microscope, without damaging the lens and the retina. Injections were covered by the institutional animal care and use committee of Gyeongsang National University.

Rottlerin (Sigma, St Louis, MO), a highly specific PKC-δ inhibitor, was dissolved in 0.5% dimethyl sulfoxide (DMSO), and 3 μl rottlerin (5 μmol/l) was used for intravitreal injection into the right eye of 35-week-old LETO and OLETF rats. DMSO (3 μl) was injected into the left vitreous as a control. All rats were killed 1 day after the injection.

For PKC-δ and HSP90 gene silencing, we used commercially available small interfering RNAs (siRNAs) from Dharmacon (ON-TARGET plus Duplex J-080142-05-0050 for Rat PRKCD and J-102259-01-0020 for Rat Hspcb; Dharmacon, Chicago). The sense and antisense strands of the PKC-δ and HSP90 siRNAs were as follows: PKC-δ, 5′-GCAACGCUGCCAUCCAUAAUU-3′ (sense) and 5′-PUUAUGGAUGGCAGCGUUGCUU-3′ (antisense); HSP90, 5′-GCUUUGAGGUGGUAUACAUUU-3′ (sense) and 5′-PAUGUAUACCACCUCAAAGCUU-3′ (antisense). siRNAs were completely dissolved in RNase-free distilled water (Dharmacon) at a final concentration of 500 μmol/l before injection. To assess the effects of PKC-δ and HSP90 siRNAs on retinas, 1 and 3 μl siRNAs, each at a concentration of 500 μmol/l, were intravitreally injected into the right eye of OLETF and LETO rats at 35 weeks. Control rats received 1 and 3 μl distilled water into the left eye. Rats were killed at 1, 2, and 5 days after the injection, and the effects of siRNAs on PKC-δ and HSP90 were determined by immunoblotting. Data are representative of four independent experiments.

Mouse monoclonal antibodies against PKC-δ, PI 3-kinase p85 (regulatory subunit), HSP90, and Akt; goat polyclonal anti-Thy-1 antibody; and rabbit polyclonal anti-PP2A (catalytic subunit, 36 kDa) and anti-Akt antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antibodies against PP2B and phospho-Akt were obtained from BD Biosciences (San Jose, CA) and Cell Signaling (Danvers, MA), and mouse monoclonal anti–α-tubulin antibody was purchased from Sigma. Horseradish peroxidase–conjugated secondary antibodies were purchased from Pierce (Rockford, IL). Cy 3–conjugated donkey anti-rabbit and anti-mouse IgGs and Alexa Fluor 405–conjugated chicken anti-goat IgG were obtained from Amersham Biosciences (Piscataway, NJ) and Invitrogen (Carlsbad, CA), respectively.

Retinal protein extraction and immunoblot analysis were performed as described previously (23). Total protein (30 μg) from LETO and OLETF rat retinas was subjected to SDS-PAGE and then transferred to a nitrocellulose membrane. Antibody incubations and washing were performed on the membranes, and the immunoreactive proteins were visualized using an enhanced chemiluminescent kit (Amersham Biosciences). Each membrane was then stripped and reblotted with anti–α-tubulin antibody as a control. Data are representative of four independent experiments. The fold changes in protein levels are indicated below the blots in each figure.

Immunoprecipitations were performed as described previously (24). Pre-cleared immune complexes were collected using protein-G/A-agarose beads and washed with radioimmunoprecipitation assay buffer (50 mmol/l Tris-HCl [pH 8.0], 150 mmol/l NaCl, 0.1% SDS, 0.5% sodium deoxycholate, and 1% Nonidet P-40) containing protease inhibitors. SDS-PAGE sample buffer was added to the beads, and final fractions were subjected to immunoblot analysis. All immunoblots were reprobed with the immunoprecipitating antibody to account for loading differences in protein levels, and each reciprocal analysis was performed. Data are representative of three to four independent experiments.

We performed PKC-δ kinase assay using the SignaTECT PKC Assay System (Promega, Madison, WI) according to the manufacturer's protocol, as described previously (23). Briefly, PKC-δ immune complexes were collected using protein G/A-agarose beads, and then the beads were resuspended in 20 μl kinase reaction buffer (25 mmol/l Tris-HCl [pH 7.5], 5 mmol/l β-glycerol phosphate, 2 mmol/l dithiothreitol, 0.1 mmol/l sodium orthovanadate, 10 mmol/l MgCl₂, and 0.5 μCi [γ-³²P]ATP [3,000 Ci/mmol]). Kinase activity was determined using a scintillation counter. Data are representative of four independent experiments.

PP2A activity was determined using a PP2A-IP phosphatase assay kit (catalog no. D-001810-01-20; Upstate, Temecula, CA) according to the manufacturer's protocol. Total protein (300 μg) from retinas was incubated with 4 μg anti–PP2A-C (catalytic subunit) antibody and 40 μl protein A-agarose beads for 2 h at 4°C with constant rocking. The immune complexes were washed three times in Tris-buffered saline and once with Ser/Thr assay buffer (50 mmol/l Tris-HCl [pH 7.0] and 100 μmol/l CaCl₂). The phosphatase reaction was initiated by the addition of 60 μl phosphopeptide substrate (final reaction concentration of 750 μmol/l) and allowed to proceed for 30 min in a shaking incubator. The reaction mixture was centrifuged briefly, and the supernatant was transferred to a 96-well microplate. Malachite green phosphate detection solution was added to each well and allowed to develop for 15 min at room temperature. Free phosphate was quantified by measuring the absorbance of each well at 650 nm using a microplate reader. Data are representative of four independent experiments.

Akt activity was measured using a nonradioactive Akt kinase assay kit (Cell Signaling) according to the manufacturer's protocol without any modification. Akt immune complexes from total protein (300 μg) from retinas were incubated with recombinant glycogen synthase kinase (GSK)-3α/β fusion protein (30 kDa). Phosphorylation of GSK-3 was measured by immunoblotting using anti–phospho-GSK-3α/β (Ser21/9) antibody. Data represent the results of four independent experiments.

The preparation of frozen retinal sections and immunohistochemical staining were performed as described previously (24). After blocking, retinal sections were incubated with primary antibodies against HSP90, PP2A, and phospho-Akt (Ser473) and biotinylated secondary antibodies. The sections were washed in PBS, incubated with an avidin-biotinylated horseradish peroxidase complex (ABC; Vector Laboratories, Burlingame, CA), and developed using 0.025% 3,3′-diaminobenzidine tetrahydrochloride (DAB; Sigma)/0.003% H₂O₂ in PBS.

To confirm ganglion cell–specific expression of HSP90, PP2A, and phospho-Akt, double-immunofluorescent staining was performed with Thy-1, a ganglion cell marker, as described previously (24). Briefly, retinal sections were incubated in a mixture of primary antibodies, rinsed in PBS, incubated in a mixture of secondary antibodies, and then wet-mounted. Images were obtained using a BH-2 Olympus microscope (Melville, NY) at a point that was ∼0.8–1 mm from the optic nerve head. Data are representative of three independent experiments.

Cell death was determined using a Terminal dUTP transferase nick end labeling (TUNEL) assay (In Situ Cell Death Detection kit; Roche, Mannheim, Germany) according to the manufacturer's protocol, as described previously (25). To assess ganglion cell death, the TUNEL assay was performed after Thy-1 immunofluorescent staining on retinal sections or flat mounts. All sections were stained with the nuclear marker 40,6-diamidino-2-phenylindole dihydrochloride (DAPI; Invitrogen) before being wet-mounted. Whole retinal preparation and flat-mounting were carried out as described previously (26). TUNEL-positive images were observed using a confocal microscope (Axioplan2 Imaging; Zeiss). The number of cells that had co-positive signal of TUNEL and Thy-1 was quantified using the Soft Imaging System (Soft Imaging System; Münster, Germany). Data are representative of four independent experiments from different retinal sections or flat mounts.

Densitometric analysis of immunoblots was performed using SigmaGel 1.0 (Jandel Scientific, Erkrath, Germany) and SigmaPlot 4.0 (SPSS, Chicago). All data are presented as means ± SE. Statistical significance was determined using one-way ANOVA followed by a Tukey post hoc test (SAS Institute, Cary, NC) and the Mann-Whitney U test (SPSS). P < 0.05 was considered to be statistically significant.

During the course of this study, OLETF rats gained weight faster than the control LETO rats. The mean body weights of OLETF and LETO rats at 24 weeks were 688 ± 10.5 and 471 ± 8.2 g, respectively, and the difference in weight was increased significantly at 35 weeks (732 ± 20.1 and 520 ± 11.2 g, respectively; P < 0.05, n = 5). Blood glucose levels in OLETF and LETO rats were 14.5 ± 0.5 and 6.2 ± 0.3 mmol/l (P < 0.05; n = 5, respectively) at 24 weeks and 21.6 ± 1.12 and 6.6 ± 0.5 mmol/l (P < 0.05; n = 5, respectively) at 35 weeks. OLETF rats exhibited a steady increase in glucose levels from week 10, whereas LETO rats sustained normoglycemia throughout the period of study (data not shown).

The number of TUNEL-positive ganglion cells in 35-week-old OLETF rats was significantly higher (3.5-fold; P < 0.01; n = 4) than in 24-week-old LETO rats (Fig. 1G). There were no significant differences between 24- or 35-week-old LETO and 24-week-old OLETF rats.

PKC-δ activity was significantly higher (4.9-fold; P < 0.01; n = 4) in 35-week OLETF retinas than 24-week LETO retinas (Fig. 2). There were no significant differences between 24- or 35-week-old LETO and 24-week-old OLETF rats. PKC-δ protein levels were similar in all groups (data not shown).

The protein levels of PI 3-kinase p85 and HSP90 were increased in 24-week OLETF retinas compared with LETO retinas (Fig. 3A–C); however, HSP90 levels were lower in 35-week OLETF retinas than in LETO retinas, and there was no significant difference between PI 3-kinase levels in both age-groups of OLETF rats. The levels of PP2A (catalytic subunit) and cleaved PP2B (48 kDa) were not significantly different in 24-week-old LETO and OLETF rats but were greatly increased in retinas from 35-week-old OLETF rats (Fig. 3D). Phospho-Akt (Thr308) and -Akt (Ser473) levels were increased (1.4- and 2.3-fold; P < 0.05 and 0.01, respectively; n = 4) in 24-week OLETF retinas compared with LETO retinas and decreased significantly (1.7- and 2.5-fold; P < 0.05 and 0.01, respectively; n = 4) in 35-week OLETF retinas (Fig. 3E–H). Akt activity, based on phospho-GSK–3α/β level (30 kDa), was significantly lower (twofold; P < 0.01; n = 4) in 35-week-old OLETF rats than 24-week-old LETO rats (Fig. 3I).

To assess whether PKC-δ affects the association of Akt with its binding partners, we subjected Akt immune complexes to immunoblot analysis using anti-HSP90, -PP2A, and -PP2B antibodies (Fig. 4). Akt binding to HSP90 or PP2A was similar in 24-week LETO and OLETF retinas; however, in 35-week OLETF retinas, this association was significantly decreased or increased more than threefold (P < 0.01; n = 4), respectively, compared with 24-week-old LETO rats. Neither PI 3-kinase binding to HSP90 nor PP2A or HSP90 binding to PP2A was detectable in all groups, there were no differences in PI 3-kinase binding to PKC-δ among groups, and PKC-δ–PP2A binding appeared only in 35-week OLETF rat retinas (data not shown).

HSP90 immunoreactivity was specific only in the ganglion cell layer (GCL), and PP2A- and phospho-Akt (Ser473) signals were positive in the nerve fiber layer (NFL), the inner segment layer, and the GCL in 35-week LETO and OLETF retinas (Fig. 5). HSP90 and phospho-Akt signals in the GCL (Fig. 5, large arrows and arrowheads) were decreased and PP2A signals (Fig. 5, small arrows) were increased in 35-week-old OLETF rats compared with LETO rats. By double-immunostaining with Thy-1 of HSP90, PP2A, and phospho-Akt (Ser473), we confirmed that these positive signals colocalized to ganglion cells (Fig. 5, right panels, insets). PKC-δ immunoreactivity was also observed throughout the retina, including the GCL, and there was no significant difference among 24- and 35-week-old LETO and OLETF rats (data not shown).

In a previous study, we found that PKC-δ activity was greatly increased (4.9-fold; P < 0.01; n = 4) in 35-week OLETF retinas compared with LETO retinas, and 5 μmol/l rottlerin abrogated this effect (22). In this study, 5 μmol/l rottlerin treatment also significantly decreased ganglion cell death in 35-week OLETF retinas (2.4-fold; P < 0.05; n = 4) compared with DMSO-treated OLETF retinas (Fig. 6). However, rottlerin had no effect on PKC-δ activity or cell death in LETO rats (data not shown).

Rottlerin did not significantly affect PI 3-kinase p85 and phospho-Akt (Thr308) protein levels, although it modestly increased HSP90 levels in 35-week OLETF retinas (Fig. 7A–C). Furthermore, rottlerin greatly decreased PP2A protein levels (2.5-fold; P < 0.01; n = 4; Fig. 7D) and increased phospho-Akt (Ser473) levels (2.8-fold; P < 0.01; n = 4; Fig. 7E–G). Rottlerin also blocked the decrease in Akt activity in 35-week OLETF retinas (Fig. 7H). Akt binding to HSP90 or PP2A was significantly increased or decreased, respectively, (2.7- and 2-fold; P < 0.01, respectively; n = 4; Fig. 7I–M) in rottlerin-treated 35-week OLETF retinas compared with the untreated group.

We next examined the effect of PKC-δ and HSP90 siRNA treatment on 35-week OLETF and LETO rat retinas. Three microliters siRNA significantly reduced PKC-δ and HSP90 protein expression in OLETF and LETO rats 1 day (30.2 ± 1.1 and 11.1 ± 0.8%; P < 0.05, respectively; n = 4) and 2 days (70 ± 2.1 and 50.1 ± 1.8%; P < 0.01 and 0.05, respectively; n = 4) after the injection, whereas 1 μl siRNA did not show these effects. No effects were shown at 5 days after siRNA injection (data not shown). Distilled water treatment did not show any effect on PKC-δ and HSP90 protein expression in the retinas. Figures 8 and 9 show data at 2 days after 3-μl siRNA (500 μmol/l) injection. PKC-δ siRNA significantly decreased PP2A protein levels and phosphatase activity and Akt-PP2A binding in 35-week OLETF retinas (Fig. 8A–C), whereas it modestly increased Akt-HSP90 binding (Fig. 8D–F). PKC-δ silencing also significantly increased phospho-Akt (Ser473) levels and Akt activity but did not affect phospho-Akt (Thr308) levels (Fig. 8G and H). We also observed that ganglion cell death in 35-week OLETF retinas was blocked by PKC-δ siRNA treatment (Fig. 8I–K). Additionally, we found that HSP90-silenced LETO retinas show significant decreases in Akt-HSP90 binding and Akt activity and increases in Akt-PP2A binding and ganglion cell death (Fig. 9A–G).

We have demonstrated here that PKC-δ activation is responsible for neuro-retinal apoptosis in diabetic OLETF rats via the inactivation of Akt.

Previously, we found that PKC-δ acts as a selective mediator of neuronal apoptosis in the retinas of 35-week-old OLETF rats (22). In the current study, we demonstrated that apoptosis occurs only in ganglion cells of the 35-week OLETF retinas (Fig. 1) and that PKC-δ activity is greatly increased in 35-week OLETF retinas compared with 24- or 35-week LETO and 24-week OLETF retinas (Fig. 2). PKC-δ immunoreactivity was observed throughout the retina and was highest in the GCL group (data not shown). These results indicate that PKC-δ activation is involved in ganglion cell death in OLETF rat retinas.

The activity of Akt is regulated by its association with a variety of binding partners, and Akt binding to PP2A results in the dephosphorylation and inactivation of Akt, consistent with our results. Moreover, HSP90 physically associates with Akt and disrupts the PP2A-Akt complex, stabilizing Akt activity (13,15,16). As expected, protein levels of PI 3-kinase, HSP90, and phospho-Akt and Akt activity were moderately increased in 24-week OLETF retinas compared with LETO retinas, but these were significantly decreased, with the exception of PI 3-kinase levels, in 35-week-old OLETF rats (Figs. 3 and 4). PI 3-kinase–Akt survival signals were differently regulated in OLETF rat retinas at 24 and 35 weeks. The significant reduction in these signaling components in 35-week-old OLETF rats may reflect the retinal damage associated with the pathological progression of diabetes, whereas these increases in 24-week-old OLETF rats may relate to functional compensation for diabetes-induced cellular stress. Either PP2A or PP2B cleavage was specifically increased in 35-week OLETF retinas. Furthermore, Akt binding to HSP90 was greatly decreased and its association with PP2A was greatly increased in 35-week OLETF retinas compared with other groups. The levels of Akt-PP2B complexes were similar between LETO and 24-week OLETF retinas, and PI 3-kinase binding to HSP90 or PP2A was not detected in any of the groups (data not shown). Interestingly, the immunoreactivity of HSP90, PP2A, and phospho-Akt (Ser473) was altered specifically in ganglion cells of 35-week OLETF retinas compared with LETO retinas (Fig. 5), in agreement with the results of the TUNEL assay (Fig. 1). Taken together, our results suggest that Akt inactivation is due to the upregulation of PP2A rather than a PI 3-kinase–dependent pathway and that PP2A plays an important role in ganglion cell death in OLETF rat retinas.

PKC-δ inhibition or knockdown by rottlerin or siRNA treatment significantly abrogated not only PKC-δ activation and ganglion cell death but also PP2A activation and its association with Akt. PKC-δ inhibition also decreased HSP90, phospho-Akt (Ser473), and phospho-GSK levels and the association of Akt with HSP90 in 35-week OELTF retinas (Figs. 6–8). However, these PKC-δ downregulations had no significant effects on the levels of PI 3-kinase and phospho-Akt (Thr308) and PKC-δ–PI 3-kinase binding. Also, we found that HSP90 knockdown decreases Akt-HSP90 binding and Akt activity and increases Akt-PP2A binding and ganglion cell death in 35-week LETO retinas (Fig. 9A–G). Therefore, our findings indicate that PKC-δ regulates ganglion cell death in OLETF rats through upregulating PP2A that dephosphorylates phospho-Akt (Ser473), competing with Akt stabilization by HSP90. A recent study reported that PKC-δ enhances PP2A by its direct binding (27). Consistently, downregulation of PKC-δ by rottlerin or siRNA decreased both PP2A protein levels and phosphatase activity in 35-week OLETF rat retinas. Furthermore, PKC-δ also bound to PP2A in 35-week OLETF retinas (data not shown). Therefore, we suggest that PKC-δ directly acts on PP2A in the OLETF rat retina, resulting in retinal apoptosis.

PI 3-kinase–dependent activation of Akt is mediated by PKC-δ, which is required for cell survival in various cancer and immune cells (1,2,28). However, Zhong et al. suggested that PKC-δ downregulation suppresses apoptotic signals through a novel PI 3-kinase–independent survival pathway (29). As such, we found that PKC-δ inhibition did not affect PI 3-kinase levels, but it abrogated neuronal apoptosis, increasing Akt signaling including Akt disassociation from PP2A and phosphorylation of Akt on Ser473. Thus, it seems likely that PI 3-kinase–independent Akt pathway mediates PKC-δ–induced apoptosis during diabetes.

Rottlerin has virtually no effect on PKC-δ activity in cells in culture (30–33), whereas it specifically inhibits PKC-δ activity and PKC-δ–mediated apoptosis at concentrations of 3–6 μmol/l (5,34,35), consistent with our results. In addition, rottlerin can inhibit many other kinases, including mitogen-activated protein kinase–activated protein kinase 2 (MAPKAP-K2) and p38MAPK (30,36), in a dose-dependent manner, whereas it has no significant effect on these kinases in OLETF rat retinas (data not shown). Therefore, this discrepancy may be due to differences in the doses of rottlerin used or in the physiological and pathological state of the animals.

In conclusion, our data suggest that PKC-δ mediates neuronal death in retinas of diabetic rats via PP2A activation and Akt signaling inhibition. Ganglion cell death occurs early as an initial event in diabetic retinopathy, and the mechanism of this cell death is unknown. Therefore, our data provide new insights into the mechanism of diabetes-associated neuro-retinal damage, showing that specific PKC-δ inhibitors may have potential for therapeutic agents for the prevention of human diabetic retinopathy.

This work has been supported by a grant from the Korean Research Foundation (KRF-2006-005-J04201) and partially by the Medical Research Center program of the Ministry of Science and Technology/Korea Science and Engineering Foundation (R13-2005-012-01001-1).