Diabetic Retinal Neurodegeneration Is Associated With Mitochondrial Oxidative Stress and Is Improved by an Angiotensin Receptor Blocker in a Model Combining Hypertension and Diabetes

The protocol complies with the guidelines of the Brazilian College of Animal Experimentation (COBEA) and the Statement for the Use of Animals in Ophthalmic and Vision Research (ARVO), and it was approved by the local committee for ethics in animal research (CEEA/IB/Unicamp). The SHRs and normotensive control Wistar Kyoto (WKY) rats used were provided by Taconic (Germantown, NY) and bred in our animal facility. The rats were housed at a constant temperature (22°C) on a 12-h light/dark cycle with ad libitum access to food and tap water.

Experimental diabetes was induced in 12-week-old hypertensive male SHRs and WKY rats with a single intravenous injection of streptozotocin (50 mg/kg in sodium citrate buffer, pH 4.5; Sigma, St. Louis, MO). From the day after diabetes induction, the diabetic SHRs were randomly assigned to receive no antihypertensive treatment or to be treated with the ARB losartan (200 mg/l; Merck Sharp & Dome Farmaceutica, São Paulo, Brazil) in drinking water. During the study, diabetic rats received 2 units of insulin (human insulin HI-0310; Lilly) three times per week subcutaneously to promote survival and prevent ketoacidosis. Control rats received only vehicle. Blood glucose levels were measured by the colorimetric GOD-PAP assay (Merck, Darmstadt, Germany). Values ≥15 mmol/l were indicative of diabetes. Systolic blood pressure (SBP) was obtained by tail-cuff plethysmography (Physiograph MK-III-S; Narco Bio-System, Houston, TX) as previously reported (14). Body weight, blood glucose levels, and SBP were measured at 0, 4, 8, and 12 weeks of duration of diabetes. At 12 weeks after diabetes induction, the rats were killed, and the retinas in one eye were detached from the retinal pigmented epithelium cell layer and used for protein extraction or colorimetric assays; the other eye was used for immunohistochemical or immunofluorescence assays.

To determine whether retinal cell apoptosis was influenced by diabetes or rat strain, the terminal deoxynucleotidyl transferase (TdT)–mediated dUTP nick-end labeling (TUNEL) method for detecting DNA breaks in situ was applied to retinal tissue. After quenching endogenous peroxidase, the sections were rinsed in One-Phor-All buffer (Amersham Pharmacia Biotech, Piscataway, NJ) and incubated with TdT (Amersham Pharmacia Biotech) and biotinylated dUTP (Gibco/BRL, Life Technologies, Grand Island, NY) in TdT buffer. Labeled nuclei were detected with ABC Vectastain (Vector Laboratories, Burlingame, CA) and diaminobenzidine tetrahydrochloride/chloride/hydrogen peroxide and counterstained with hematoxylin. As a positive control, some slides were treated with DNase (Sigma). The quantitative analysis for TUNEL-positive cells was performed by an observer with no knowledge of the studied groups and expressed as the number of positive cells per retinal section.

The eyes were enucleated and fixed with 4% paraformaldehyde, cryoprotected in 30% sucrose in phosphate buffer, frozen in embedding medium (OCT; Sakura Finetek, Torrance, CA), and cut perpendicularly to the vitreal surface. The slides were blocked with BSA and the sections were stained with both primary antibodies to polyclonal cleaved caspase-3 (1:10; Cell Signaling), a marker of apoptosis, and polyclonal anti–glial fibrillary acidic protein (GFAP) antibody (1:10; Santa Cruz Biotechnology, Santa Cruz, CA) for Müller glial cell or anti-rat nestin monoclonal antibody (1:10; BD Pharmingen, Franklin Lakes, NJ) for neural cell identification. The sections were then incubated with the appropriate secondary antibodies. Afterwards, the sections were rinsed and cover-slipped with Vectashield antifading medium containing 4′,6′-diamino-2-phenylindole used for nuclei staining (Vector). The sections were examined with a confocal laser scanning microscope (LSM510; Zeiss) using appropriate emission filters. Digital images were captured using specific software (LSM; Zeiss).

The immunofluorescence labeling of GFAP was performed as described above in double-immunofluorescence staining assay. The sections were examined using an Olympus BX51 fluorescence microscope. Digital images were captured using specific software (Image Pro Express 6.0). The GFAP analyses were determined using the public domain program Image J (National Institutes of Health, available online at http://rsb.info.nih.gov/ij). The semiquantitative analyses were performed as mentioned above in TUNEL assay. The fluorescence of GFAP was expressed as the percentage of fluorescence per millimeter squared of retina.

Lucigenin (bis-N-methylacridinium nitrate; Invitrogen, Eugene, OR) was used to measure superoxide anion production (23). Briefly, the retinas were isolated and placed into tubes containing RPMI-1640 medium (Gibco/BRL, Life Technologies, Gaithersburg, MD) at 37°C in a humidified atmosphere of 95% air/5% CO₂. Then, 25 mmol/l of lucigenin was added, and photon emission was measured over 10 s; repeated measurements were made over a 3-min period using a luminometer (TD 20-E; Turner). Superoxide production was expressed as the relative luminescence units (RLU) per minute per milligram protein. Protein concentration was measured using the Bradford method (24) using BSA standard.

Retinal glutathione (GSH) levels were measured using a method described previously (25), with a few modifications described previously (14). The absorbance was read at 412 nm and the GSH concentration expressed as micromoles per liter GSH per microgram of retinal protein. GSH was used as an external standard for preparation of a standard curve.

Briefly (14), after quenching endogenous peroxidase, the sections were incubated with nonfat milk. Tissue sections were incubated with polyclonal rabbit anti-nitrotyrosine antibody (Upstate Cell Signaling Solutions, Lake Placid, NY) and a mouse monoclonal anti–8-OHdG antibody (1:50, N45.1; Japan Institute for the Control of Aging). Afterwards, secondary appropriated antibodies were applied to the sections. Labeled nuclei were detected as described above in the section on immunohistochemistry for TUNEL. Staining was performed, omitting the primary antibody for negative controls. For nitrotyrosine, quantitative analyses were performed as a positivity percentage in all retinal layers, grading from 0 for no positivity to 4 for ≥80% of positivity (14). The 8-OHdG analyses were determined using the public domain program Image J (National Institutes of Health) and expressed by percentage of positivity per retinal field. Semiquantitative analyses were performed as described above in the section on GFAP immunofluorescence assay.

The retinas were lysed in a buffer containing 2% SDS and 60 mmol Tris-HCl (pH 6.8) supplemented with a protease inhibitor cocktail (Complete; Boehringer-Mannheim, Indianapolis, IN). After centrifugation, the protein concentrations were measured by the Bradford method. For analysis, 100 μg of total retinal protein was loaded into SDS polyacrylamide gels. Molecular weight markers (Rainbow; Amersham Pharmacia) were used as standards. Proteins were transferred to nitrocellulose membranes (Bio-Rad, Hercules, CA). Membranes were blocked in nonfat milk, incubated with primary antibody (rabbit polyclonal anti–Bcl-2; Santa Cruz), subsequently incubated with goat anti-rabbit IgG horseradish peroxidase (HRP) secondary antibody, and developed by chemiluminescence method (Super Signal CL-HRP substrate system; Pierce, Rockford, IL). Exposed films were scanned with a densitometer (Bio-Rad) and analyzed quantitatively with Multi-Analyst Macintosh Software for Image Analysis Systems. Equal loading and transfer were ascertained by reprobing the membranes for β-actin.

The retinas were lysed directly with buffer A containing 100 mmol/l Tris, 10 mmol/l sodium pyrophosphate, 100 mmol/l sodium fluoride, 10 mmol/l EDTA, 2 mmol/l phenylmethylsulfonyl fluoride, 10 mmol/l sodium ortovanadate, and 1% Triton X-100. The protein concentrations were measured by the Bradford method (24). Samples containing 1 mg of total protein were incubated with antibody goat polyclonal IgG uncoupling protein-2 (UCP-2; Santa Cruz) overnight, followed by the addition of protein A Sepharose for 1 h. After centrifugation, the pellets were repeatedly washed in buffer C (100 mmol/l Tris, 10 mmol/l sodium vanadate, 10 mmol/l EDTA, and 1% Triton X-100). For immunoblotting, 400 μg of protein was loaded into 5% glycerol/0.03% bromophenol blue/10 mmol dithiothreitol and then loaded onto 15% SDS polyacrylamide gels. Molecular weight markers were used as standards. Proteins were transferred to nitrocellulose membranes (Bio-Rad), and the membranes were blocked in nonfat milk and then incubated with primary antibody goat polyclonal IgG UCP-2 (1:500). The blots were subsequently incubated with secondary antibody donkey anti-goat IgG HRP and developed using a chemiluminescence method. Equal loading and transfer were ascertained by Ponceau S staining.

The results are expressed as the means ± SD. The groups were compared by one-way ANOVA, followed by the Fisher protected least-significant difference test. StatView statistics software for Macintosh was used for all comparisons with a significance value of P < 0.05.

The final body weight was lower after streptozotocin injection in both WKY and SHRs (P < 0.0001), and it was not affected by treatment with losartan in diabetic SHRs. As expected, SBP was significantly higher in SHRs than in WKY rats, and it was reduced in the treated SHRs (P < 0.0001). Blood glucose levels were higher in diabetic rats compared with nondiabetic groups, and these were not modified by ARB treatment (P < 0.0001) (Table 1).

TUNEL staining was a rare event in the retina from control rats. After 12 weeks of diabetes in WKY rats, an increment of retinal cells staining positive for TUNEL was observed (P = 0.03). The diabetic SHRs exhibited an increased number of TUNEL-positive cells in the retina compared with diabetic WKY rats (P = 0.01) and control SHRs (P = 0.0003). Losartan significantly reduced the number of TUNEL-positive cells in all retinal layers compared with diabetic SHRs (P = 0.01) (Fig. 1).

To further characterize the apoptotic cells in the retina of adult rats, we labeled retinal sections for GFAP and nestin. A caspase-3–positive cell colabeled with the GFAP marker was observed (Fig. 2,A), and another caspase-3–positive cell expressed nestin (Fig. 2 B); both cells were in the outer nuclear layer. These findings indicate that the apoptotic retinal cells are of glial and neural origin.

Retinal glial reaction, demonstrated by a local increase in GFAP expression, is an early marker in the pathogenesis of diabetic retinopathy (26). In the retina of control WKY rats, GFAP positivity is minimally apparent. In contrast, after diabetes induction there was an accentuated increase in glial reactivity (P < 0.0001). Similarly observed in diabetic WKY rats, there was a moderate glial reaction throughout the retina in control SHRs (P = 0.4), and the concomitance of both diabetes and hypertension extensively exacerbated the GFAP staining in retinal tissue (P < 0.0001). The treatment prevented the retinal glial reaction seen in diabetic SHRs, which remained similar to control levels (P < 0.0001) (Fig. 3).

After 12 weeks of diabetes, a significant increase of superoxide anion production was observed in diabetic WKY rats compared with the control WKY rats (0.9 ± 0.3 vs. 2.0 ± 0.8 RLU · min⁻¹ · mg protein⁻¹, P = 0.03), and the concomitance of diabetes and hypertension exacerbated the superoxide production compared with the other groups (P = 0.0002). To identify the source of superoxide production, we used diphenyliodonium, an inhibitor of flavin-containing oxidases, and rotenone, an inhibitor of complex I of the mitochondrial respiratory chain, in vials containing retina from diabetic SHRs. Preincubation of the retinal tissue with diphenyliodonium (20 μmol/l) did not affect superoxide production, whereas preincubation with rotenone (100 μmol/l) resulted in a marked reduction in superoxide production. This indicates that mitochondria are an important source of the superoxide in retinal tissue (Fig. 4,A). Therefore, we evaluated the effect of the treatment with losartan on retinal superoxide production. It was observed that the ARB restored the superoxide production in retina in diabetic SHRs to control WKY levels (P = 0.0003) (Fig. 4 B).

Antioxidant defense was examined using the quantitative measurement of GSH levels in the retinal tissue; it was diminished in diabetic WKY rats compared with control WKY rats, but not significantly (P = 0.1). However, the concomitance of diabetes and hypertension led to a marked diminution in GSH concentration (∼2.5-fold decrease) compared with control SHRs (P = 0.0005). The treatment with losartan reestablished this parameter to control SHR levels (P = 0.006) (Fig. 5).

The nitration of tyrosine, an effect of peroxynitrite on proteins, was assessed by nitrotyrosine expression. The immunohistochemistry for nitrotyrosine showed stronger staining throughout retinal layers in diabetic WKY rats and control SHRs compared with control WKY rats (P < 0.0001 and P = 0.0003, respectively). The presence of diabetes alone showed an increment in nitrotyrosine values compared with hypertension alone (P = 0.004). The concomitance of diabetes and hypertension further increased nitrotyrosine expression in all retinal layers compared with other groups (P = 0.001). Losartan completely reestablished retinal nitrosative status in the retinas of diabetic SHRs (P < 0.0001). These findings suggest that losartan protects the retinal tissue against nitrosative stress in diabetic SHRs (Fig. 6 A).

The distribution of positive 8-OHdG, a marker of oxidative damage on nucleic acids, was heterogeneous in retinal tissue, presenting higher positivity in the outer nuclear layer of diabetic WKY rats and in all cellular layers of diabetic SHRs. The presence of diabetes or hypertension alone induced a significant increment in 8-OhdG–positive cells in retinal tissue compared with the control WKY group (P = 0.003). The concomitance of both resulted in a marked increase in oxidative DNA damage compared with control SHRs (P < 0.0001). Similarly, as observed with nitrotyrosine, losartan significantly protected the retinal cells against DNA damage (P < 0.0001) (Fig. 6 B).

The overexpression of Bcl-2 mitochondrial protein inhibits the release of cytochrome c into the cytosol, protecting the cells against early death (27). Therefore, the estimation of Bcl-2 may be indicative of mitochondrial involvement in the apoptotic cascade. The expression of Bcl-2 in total retinal lysates was decreased in both diabetic groups when compared with the controls (P = 0.01) and reestablished in diabetic SHRs treated with losartan (Fig. 7 A).

UCP-2 expression reflects the mitochondrial energy metabolism and might play a role in retinal neurodegeneration (28). Similarly, as observed in Bcl-2 protein, the expression of UCP-2 was diminished in both diabetic groups compared with the respective control groups (P = 0.03), and it was restored to the levels of the control groups in losartan-treated diabetic SHRs (P = 0.04) (Fig. 7 B).

The ARB used in the treatment of hypertension exerts a variety of pleiotropic effects, including antioxidant, antiapoptotic, and anti-inflammatory effects (29). However, the possible antioxidant/antiapoptotic effects of ARB in the diabetic retina have never been addressed. In the current study, we investigated the potential effect of the losartan on retinal neurodegeneration in a model that combines diabetes and hypertension. We observed that the apoptotic rate was higher in the retina of diabetic SHRs compared with control WKY rats, and the cells exhibited neural and glial characteristics, as demonstrated by specific antigens. The oxidative imbalance, characterized by an increase in superoxide production and a decrease in reduced GSH levels in retinal tissue, was higher in diabetic rats and accentuated in diabetic SHRs in the presence of mitochondrial involvement, as demonstrated by decreased expression of Bcl-2 and UCP-2 mitochondrial proteins. Losartan treatment led to amelioration of the apoptotic rate in neural and glial retinal cells, reestablishment of redox status by decreasing superoxide production and improving the antioxidative enzymatic system GSH, and restoration of mitochondrial protein expression levels. Therefore, the ARB seemed to offer neural protection, including antiapoptotic and antioxidant benefits, in the retina of diabetic hypertensive rats.

The increased number of TUNEL-positive cells detected in the retinas of diabetic SHRs occurred mainly in the outer nuclear layer, and this may contribute to widespread retinal dysfunction. In line with previous studies of electroretinography, Phipps et al. (30) demonstrated a significant reduction in the rod photoreceptor response in diabetic Ren-2 rats compared with nondiabetic Ren-2 rats, which in turn translated losses to b-wave and oscillatory potentials. Other studies had confirmed that rod photoreceptors are the primary retinal neuron affected by diabetes (31,32). The photoreceptors are most vulnerable to oxidative damage because of the high content of polyunsaturated fatty acids in their membranes; oxidative stress may cause lipid peroxidation reactions (33) and therefore damage their structure and function. In this study, we observed glial reactivity, as evaluated by GFAP immunoreactions, but not apoptosis of photoreceptor, as evaluated by TUNEL, in hypertensive rats. This is explained by the fact that glial cells and photoreceptors are damaged by oxidative stress through different mechanisms. The glial cells possess mechanisms providing high intracellular GSH concentration (34). Depending on the insult, the depletion of GSH in glial cells and the subsequent heme oxygenase-1 induction (35) is associated with the production of bilirubin, a potent free radical scavenger (36), and with the reduction of heme, a powerful pro-oxidant (37). One reason for this is that Müller cells provide metabolic support to adjacent neural cells, providing protective mechanisms for photoreceptors.

Previous studies demonstrated that local angiotensin II expression was extremely elevated during retinal inflammation, thereby influencing the condition of the retinal neural and vascular cells through AT₁ receptor signaling. The ARB was effective in keeping the retinal neural cells from losing their physiological activities and normal electroretinography responses; therefore, the ARB plays a key role in neuroprotection, and it preserves good visual function by reducing inflammatory reactions in the retinal neural and vascular cells, preventing the development of diabetic retinopathy (7,38). This treatment also reduced the accumulation of one of the advanced glycation end products as well as vascular endothelial growth factor expression in a model of type 2 diabetes (39). Other studies also showed that treatment with ARBs in hypertensive diabetic Ren-2 rats prevented acellular capillary and endothelial cell proliferation as well as development of neuronal deficits in diabetes, namely loss of function in photoreceptors and neurons, independent of controlling hypertension (30,40). These findings support the concept that ARBs may be useful as a therapeutic target for diabetic retinopathy. Although in this study we did not treat the diabetic SHRs with another antihypertensive drug that does not act on the renin-angiotensin system, to avoid the hypotensive effect of the losartan treatment, the normotensive diabetic WKY rats still had markers of oxidative stress and mitochondrial dysfunctions that were not present in diabetic SHRs treated with losartan.

The marked effect in reducing the oxidative damage in diabetic retinal tissue of hypertensive rats treated with losartan may be attributable to its antioxidant effect through ARB and an increase in nitric oxide (NO) biodisponibility via the angiotensin II type 2 receptor (41). A recent study demonstrated that intravenous application of angiotensin II reduces plasma NO levels and increases the peroxynitrite concentration, which significantly increases nitrosative stress. Treatment with valsartan, an ARB similar to losartan, suppressed this effect (42). The question of how NO can attenuate oxidative damage is still intriguing. A previous study provided direct evidence that NO inactivates xanthine oxidase by reacting first with the superoxide anion to form peroxynitrite, which in turn reduces both xanthine oxidase activity and superoxide generation (43), thus reestablishing the oxidative status.

It was previously demonstrated that hyperglycemia-induced production of ROS is associated with the development of diabetic microvascular complications (11) and that the normalization of mitochondrial superoxide production blocks the pathways of diabetic damage (44). Normalization of mitochondrial ROS production prevents pathways involved in the development of diabetic microvascular complications (44). In the current study, the blockage of mitochondria complex I using rotenone revealed a significant reduction in its superoxide production. This fact may be explained by the observation of diversity in electron chain transport components present in different neuron subpopulations, as demonstrated by a study where mitochondria from retinal ganglion cells decreased the superoxide production in response to rotenone (45).

Mitochondria are a major endogenous source and target of superoxide and hydroxyl radicals (46). Reactive oxidant intermediates can trigger mitochondria to release cytochrome c, resulting in activation of caspase-3 (27,46). Increasing evidence indicates that mitochondria are intimately associated with the initiation of apoptosis (47,48). In this study, mitochondrial integrity was evaluated by determining Bcl-2 and UCP-2 protein expressions. The mechanism that involves angiotensin II in mitochondrial ROS production is supported by the fact that in vivo preconditioning effects of angiotensin II for cardiac ischemia/reperfusion injury may be mediated by cardiac mitochondria–derived ROS enhanced by NAD(P)H oxidase (49). The superoxide production via NAD(P)H oxidase stimulates the opening of reconstituted mitochondrial ATP-sensitive K⁺channels via a direct action on the sulfhydryl groups of this channel (50). Thus, the overstimulation of angiotensin II and renin-angiotensin system may account for mitochondrial dysfunction, in line with the observed effect that AT₁blockage with losartan reestablished the UCP-2 and Bcl-2 contents in retinal tissue.

In summary, the findings of this study provides evidence for the first time of the benefits of the AT₁blocker losartan in ameliorating diabetic retinal neurodegeneration, mitochondrial function, and oxidative balance.

This work was supported by the State of São Paulo Research Foundation (FAPESP) grants 04/00455-9 and 05/58189-5. K.C.S. was a recipient of a scholarship from the National Council for Scientific and Technological Development (CNPq).

No potential conflicts of interest relevant to this article were reported.

The authors thank Dania E.H. Britto and Aline M. Faria for assistance in double-labeling immunofluorescence assays.