Protective Effect of Perindopril on Diabetic Retinopathy Is Associated With Decreased Vascular Endothelial Growth Factor–to–Pigment Epithelium–Derived Factor Ratio: Involvement of a Mitochondria–Reactive Oxygen Species Pathway

All experiments in this study comply with the requirements of the Association for Research in Vision and Ophthalmology statement with regard to the “Use of Animals in Ophthalmic and Vision Research.” All chemicals were reagent grade quality and were purchased from Sigma Chemicals (St. Louis, MO) unless stated otherwise.

Eight-week-old male Sprague-Dawley rats weighing ∼200 g (Shanghai Laboratory Animal Center, Chinese Academy of Sciences) were randomly assigned to receive either 60 mg/kg STZ intraperitoneally or citrate buffer alone. Rats were categorized as diabetic when the blood glucose exceeded 16.7 mmol/l at 48 h after STZ administration. One week after the injection of STZ, diabetic rats were randomly assigned to groups receiving either 2 mg·kg⁻¹·day⁻¹ perindopril (Servier, Tianjin, China) by drinking water for 24 weeks or no treatment at all. Age-matched rats receiving no STZ served as controls. All rats had free access to standard rat food and drinking water. Diabetic rats received subcutaneous insulin (Humulin-N; Eli Lilly & Co., Indianapolis, IN) twice a week to maintain body weight and maximize survival rate (0–4 units).

The primary culture of bovine retinal capillary endothelial cells (BRECs) was done as described in our previous study (19). The endothelial cells of three to four passages were used in the following experiments. In each case, the confluent-cultured BRECs were maintained in free-serum DMEM. The cells were exposed to normal glucose (5 mmol/l), normal glucose plus mannitol (25 mmol/l), normal glucose plus perindopril (10 μmol/l), normal glucose plus H₂O₂ (500 μmol/l), high glucose (30 mmol/l), high glucose plus perindopril (10 μmol/l), high glucose plus ROS scavenger N-acetylcysteine (NAC; 10 mmol/l), high glucose plus perindopril plus NAC, or high glucose plus perindopril plus NAC plus GW9662 (an inhibitor of PPARγ; 20 μmol/l).

UCP-2 antisense oligonucleotide synthesis and treatment were conducted as described previously (20). UCP-2 antisense oligonucleotide sequence was 5′-TGAGATCTGCAATACA-3′, and the corresponding sense oligonucleotide sequence was 5′-TGTATTGCAGATCTCA-3′. After 24 h, the medium was removed, free-serum DMEM in high glucose was added, and the cells were allowed to recover for 30 min. Then BRECs were washed once with free-serum DMEM and exposed to either normal glucose (5 mmol/l) or high glucose (30 mmol/l) for 24 h.

Harvested eyes were immediately placed in 4% buffered paraformaldehyde for 24 h. Retinal trypsin digestion was performed according to the method described by Cogan and associates (21). Preparations of retinal vascular networks were placed onto polylysine-coated glass slides in distilled water and then dried. The preparations were stored at −20°C until periodic acid Schiff (PAS) and hematoxylin staining. The capillary network was investigated to determine the numbers of pericytes and acellular capillaries as previously described (22).

Tissue processing, electron microscopy, morphometric measurements (retinal capillary basement membrane thickness [BMT]), and statistics were performed as detailed in our previous study (6). Enucleated eyes were fixed in 2.5% glutaraldehyde in 0.1 mol/l cacodylate buffer (pH 7.4) containing 0.2% tannic acid, washed in the same buffer, and postfixed in 0.5% osmium tetroxide. Tissue sections were block stained with uranyl acetate, lead stained, dehydrated through a graded series of ethanol, and embedded in epon. One-micrometer-thick sections were examined with a JEM-1200EX transmission electron microscope (JEOL, Akishima, Japan). Computer-assisted morphometric measurements (The Image Center of Beijing University of Aeronautics & Astronautics, Beijing, China) were done on electron micrographs taken from 12 randomly selected capillaries of the outer plexiform layer from four different tissue blocks of the same retina. Only cross-sectioned capillaries were considered. A total of 96 capillaries were evaluated in each experiment group.

ROS production in the cells was assessed using the fluorescent probe 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate acetyl ester (CM-H2DCFDA; Molecular Probes, Eugene, OR). CM-H2DCFDA (λ_ex, 488 nm and λ_em, 520 nm) is a cell-permeable indicator of ROS that remains nonfluorescent until acetate groups are removed (H2DCF) by intracellular esterases and oxidation occurs within the cells. Two hundred microliters of the cell suspension were loaded into the wells of a FluoroNunc 96-well polystyrene plate together with CM-H2DCFDA (10 μmol/l) for 45 min at 37°C. Intracellular ROS production was calculated using an H₂O₂ standard curve (10–200 nmol/ml). Retinal mitochondria ROS production was detected as described by Benani et al. (23). Retinal tissues were harvested in cold-buffered medium (5 mmol/l HEPES in PBS) and immediately frozen in liquid nitrogen to improve the following probe diffusion. After rapid thawing, medium was discarded. Samples were exposed to 8 μmol/l CM-H2DCFDA dissolved in 400 μl fresh medium and were incubated at 37°C for 30 min under agitation. Medium was then removed, and samples were further incubated in a lysis buffer (0.1% SDS, Tris-HCl, pH 7.4) for 15 min at 4°C. After homogenization, samples were centrifuged at 6,000g for 20 min at 4°C. Supernatants were collected and subjected to fluorescence analysis as stated previously.

5,5′,6,6′-tetrachloro1,1′,3,3′-tetraethyl-benzimidazolylcarbocyanine iodide (JC-1; Molecular Probes) is a potentiometric dye that exhibits membrane potential–dependent loss as J-aggregates (polarized mitochondria) that are converted to JC-1 monomers (depolarized mitochondria) as indicated by the fluorescence emission shift from red to green. Mitochondrial depolarization is indicated by an increase in the green/red fluorescence intensity ratio. Mitochondrial membrane potential (Δψm) measurement in BRECs or retinal tissues was performed using flow cytometry (Coulter Epics XL; Beckman-Coulter) as described in our previous study and by Hassouna and associates (19,24).

Total RNA was extracted from rat retinal tissue and BRECs using TRIZOL reagent (Invitrogen Life Technologies, Gaithersburg, MD) and stored at −80°C. The DyNAmo Flash SYBR Green qPCR kit (Finnzymes Oy, Espoo, Finland) was used according to the manufacturer's instructions. The primer sequences (sense/antisense) used were as follows: PEDF, 5′-CAGAAGAACCTCAAGAGTGCC-3′/5′-CTTCATCCAAGTAGAAATCC-3′; VEGF, 5′-GCGGGCTGCTGCAATG-3′/5′-TGCAACGCGAGTCTGTGTTT-3′; UCP-2, 5′-TCTGACCATGGTGCGTACTGA-3′/5′-GACAATGGCATTACGAGCAAC-3′; PPARγ1, 5′-TTCTGACAGGACTGTGTGACAG-3′/5′-ATAAGGTGGAGATGCAGGTTC-3′; PPARγ2, 5′-GCTGTTATGGGTGAAACTCTG-3′/5′-ATAAGGTGGAGATGCAGGTTC-3′; and β-actin, 5′-GCACCGCAAATGCTTCTA-3′/5′-GGTCTTTACGGATGTCAACG-3′. The specificity of the amplification product was determined by a melting curve analysis. Standard curves were generated for the expression of each gene by preparing serial dilutions with known quantities of each cDNA template. Relative quantification of the signals was performed by normalizing the signals of different genes with the β-actin signal. Signal intensities in the control lanes were arbitrarily assigned a value of 1.0.

Retinas were removed rapidly and frozen in liquid nitrogen. The frozen tissues were homogenized with lysis buffer (50 mmol/l Tris-HCl [pH 7.4], 10% glycerol, 2 mmol/l EDTA, 150 mmol/l NaCl, 1 mmol/l MgCl₂, 50 mmol/l glycerophosphate, 2 mmol/l Na₃VO₄, 20 mmol/l NaF, 1 mmol/l phenylmethylsulfonylfluoride, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 1% Nonidet P-40) and were centrifuged at 12,000 rpm for 15 min at 4°C. BREC extracts were prepared with the same lysis buffer. The protein concentrations in the supernatants were measured using the Bio-Rad DC protein assay. Fifty micrograms of protein from each sample (retinas or BRECs) were subjected to SDS-PAGE using a Bio-Rad miniature slab gel apparatus and electrophoretically transferred onto a nitrocellulose sheet. The sheet was blocked in 5% nonfat dried milk solution and incubated overnight with partially purified goat anti-PEDF (Chemicon International) monoclonal antibody, mouse anti-VEGF (Chemicon, Temecula, CA) monoclonal antibody, rabbit anti–UCP-2 polyclonal antibody (Merck & Co.), and rabbit anti-PPARγ polyclonal antibody (MILLIPORE). β-Actin (monoclonal anti-β-actin; Sigma Chemical) expression was used as an internal control.

The PEDF concentrations in cell media were measured using a 2-antibody sandwich ELISA. Assays were performed in 96-well immunoplates (Chemicon International). VEGF ELISA was performed using a Quantikine VEGF assay kit (R & D Systems) to quantify the levels of VEGF in cell media. Serial dilutions of recombinant human VEGF and PEDF were included in all assays to serve as standards.

The experimental data were expressed as means ± SD. Group means were compared by a one-way ANOVA using the GraphPad Prism 4.0 software system (GraphPad, San Diego, CA) and the statistical software program SPSS13.0 for Windows (Chicago, IL). Pearson correlation tests were also performed. P values <0.05 were considered significant in all cases.

The retinal expression of PEDF and VEGF was significantly increased in the diabetic rats compared with that in the nondiabetic rats at both mRNA and protein levels, whereas the VEGF-to-PEDF ratio was still significantly increased in diabetic rats compared with that in the nondiabetic rats (P < 0.01) (Fig. 1,A–C). Pathological damage of the retina occurred in diabetic rats at an early stage (Fig. 2,A–C). Compared with the nondiabetic rats, the number of pericytes in the retinas of diabetic rats was significantly reduced (P < 0.01), and the number of acellular capillary segments and BMT were significantly increased (P < 0.01) (Fig. 2,D–F). Statistical analysis showed that the VEGF-to-PEDF ratio was negatively correlated with the number of pericytes and positively correlated with acellular capillary segments and BMT before and after perindopril treatment (Table 1). Perindopril significantly inhibited the increase of VEGF-to-PEDF ratio in diabetic rats (Fig. 1,A–C) and attenuated the damages to the retinas (Fig. 2,A–F), and the decrease of the VEGF-to-PEDF ratio was significantly correlated with the attenuation of retinal damage (Table 1). In addition, we found that the nonfasting blood glucose was markedly higher in the diabetic rats than in the nondiabetic rats (28.5 ± 4.3 mmol/l versus 4.3 ± 0.5 mmol/l, P < 0.01), and perindopril showed no effect on hyperglycemia in diabetic rats (27.3 ± 3.8 mmol/l versus 28.5 ± 4.3 mmol/l, P > 0.05). The body weights of diabetic rats and perindopril-treated diabetic rats were significantly lower than that of nondiabetic rats (377 ± 28 g or 392 ± 36 g versus 594 ± 45 g, P < 0.05).

Compared with nondiabetic rats, the production of mitochondrial ROS, Δψm, and the expression of UCP-2 and PPARγ protein were significantly increased in the retinas of diabetic rats (P < 0.01). Perindopril treatment reduced ROS production and Δψm but further upregulated the expression of UCP-2 and PPARγ protein (Fig. 3,A–D). Pearson correlation analysis indicated that the decrease of the VEGF-to-PEDF ratio was correlated with the reduction of mitochondrial ROS in the perindopril-treated group (Fig. 3 B) (r = 0.749, P = 0.032). In addition, statistical analysis showed that there was a correlation between mitochondrial ROS and Δψm (r = 0.902, P = 0.003), between Δψm and UCP-2 (r = 0.823, P = 0.012), and between UCP-2 and PPARγ levels (r = 0.887, P = 0.005) in the perindopril-treated group.

To understand the mechanism underling the inhibitory effect of perindopril on the VEGF-to-PEDF ratio, an in vitro study was performed with BRECs. Exposure of BRECs to high glucose increased ROS production, upregulated VEGF mRNA and protein in cell extracts or media, downregulated PEDF mRNA and protein in cell extracts or media, and elevated the VEGF-to-PEDF ratio (protein in cell extracts or media), whereas these changes were significantly inhibited by 10 μmol/l perindopril (Fig. 4,A–F). Our results also showed that NAC, an ROS scavenger, arrested the elevation of the VEGF-to-PEDF ratio (upregulating VEGF and downregulating PEDF). We also found that incubation with H₂O₂ upregulated VEGF, downregulated PEDF, and elevated VEGF/PEDF ratio in BREC extracts or media (Fig. 5 A–F). The effect of H₂O₂ on VEGF-to-PEDF ratio was time- and dose-dependent (data not shown). To rule out the influence of osmolarity on the ratio, mannitol was used to treat the cells, and the results showed that mannitol had no effect on the VEGF-to-PEDF ratio (data not shown).

UCP-2 expression has been reported to be associated with ROS generation (25), and PPARγ has been shown to modulate the transcription activities of genes involved in energy metabolism, including the mitochondrial uncoupling proteins, UCP-1, UCP-2, and UCP-3 (26). To investigate the mechanism by which perindopril reduces ROS production, we examined the mRNA and protein expression of UCP-2 and PPARγ in BRECs exposed to normal and high glucose. We found that the mitochondrial membrane was hyperpolarized under high glucose conditions and the production of ROS was increased; the production of ROS was positively correlated with Δψm (r = 0.779, P = 0.023); we also found that perindopril significantly decreased hyperglycemia-induced mitochondrial membrane hyperpolarization and the subsequent increased ROS production (Figs. 4,A and 6 A and B). We found that hyperglycemia induced upregulation of UCP-2 mRNA and protein; perindopril also upregulated UCP-2 expression in BRECs exposed to normal or high glucose with or without NAC (Fig. 6 C and D). Incubation with UCP-2 antisense oligonucleotide greatly enhanced the hyperglycemia-induced Δψm and ROS production, whereas it blocked the inhibitory effect of perindopril on the Δψm and ROS production (Figs. 4,A and 6,A and B). In addition, as shown in Fig. 7,A–C, hyperglycemia increased the expression of PPARγ1 and PPARγ2 mRNA and PPARγ protein; perindopril could upregulate their expression under normal conditions and further upregulate their expression under hyperglycemia conditions. We also found that high glucose combined with GW9662, an inhibitor of PPARγ, inhibited the upregulation of UCP-2 in BRECs; besides, GW9662 blocked the upregulating effect of perindopril on UCP-2 (Fig. 7 A–C).

Recently, the protective effect of ACEI on DR has increasingly become a focus of study (2,,,–6). We observed the changes of the VEGF-to-PEDF ratio in the rat retinal tissues and BRECs in the presence of normal or high glucose; we also observed the effect of perindopril on changes of the VEGF/PEDF ratio in diabetic rats and in BRECs exposed to high glucose. We confirmed that ACEI exerted a protective effect on DR (2,,–5,27,28), and for the first time, we found that this protective effect was associated with a decreased VEGF-to-PEDF ratio (downregulating VEGF and upregulating PEDF). We found the decreased VEGF-to-PEDF ratio was a result of reduced mitochondrial ROS production, and the reduced ROS production was attributable to decreased Δψm, which was a result of ACEI-induced upregulation of PPARγ and UCP-2 expression.

We found that VEGF was upregulated and PEDF was downregulated in the vitreous of patients with PDR, which subsequently elevated the VEGF-to-PEDF ratio and was accompanied by retinal neovascularization (data not shown), whereas in diabetic Sprague-Dawley rats, both VEGF and PEDF were upregulated. (A similar result was also noticed in a study with spontaneously diabetic Torii rats, an animal model of type 2 diabetes [(29)].) Nevertheless, as a result of even more upregulation of VEGF, the VEGF/PEDF ratio was still significantly elevated compared with that of the controls; the elevation was also accompanied by early vascular damages; and a significant correlation was found between the severity of damage and the VEGF/PEDF ratio. In this study, we found that ACEI perindopril downregulated VEGF and upregulated PEDF, which subsequently resulted in a lowered VEGF-to-PEDF ratio and relieved vascular damage in the retina of diabetic rats; and the lowering of VEGF-to-PEDF ratio was significantly correlated with the relief of the vascular damage. It is therefore indicated that the protective effect of ACEI on DR was associated with the decreased VEGF-to-PEDF ratio, which may serve as an effective marker for the development, progression, and treatment outcome of DR.

To further explore the mechanism by which ACEI exerts its protective effect on DR, we investigated the possible role of mitochondrial ROS, which is taken as a key initiator for all of the four pathogenic pathways of diabetic microangiopathy (15,–17). We found that high glucose could induce ROS production, which was consistent with our previous study (19); we also noticed that ACEI inhibited the retinal and cellular production of ROS and blocked the elevation of the VEGF-to-PEDF ratio. We discovered that H₂O₂ upregulated VEGF, downregulated PEDF, and elevated the VEGF-to-PEDF ratio and that NAC, an ROS scavenger, could inhibit hyperglycemia-induced changes of VEGF and PEDF expression and elevation of the VEGF-to-PEDF ratio. These findings suggest that ROS is an upstream molecule of VEGF and PEDF and that ACEI lowers the VEGF-to-PEDF ratio through inhibiting hyperglycemia-induced ROS production.

Hyperglycemia-induced ROS production is primarily thought to be associated with mitochondria and NADPH oxidase (30). A previous study found that both hyperglycemia and angiotensin II, through activating membrane-bound NADH-/NADPH oxidase, could induce superoxide anion generation in human vascular endothelial cells (31). ACEI was also found to inhibit TGF-β and prevent activation of NADPH oxidase so as to attenuate renal damage (32). Mitochondrial ROS was reported to play a very important role in the pathogenesis of DR (16,17,19). Consistent with these studies, our results also revealed that hyperglycemia induced NADPH oxidase activation (data not shown) and mitochondrial membrane hyperpolarization and promoted ROS production. We noticed that perindopril suppressed hyperglycemia-induced activation of NADPH oxidase in a manner similar to that of diphenyleneiodonium in cultured BRECs under hyperglycemic conditions (data not shown).

We discovered, for the first time, that perindopril inhibited the mitochondrial ROS production by increasing the expression of UCP-2, homologous to UCP-1 (33), found in various human tissues, and one of its major functions is as a sensor and negative regulator of ROS production (25). We found that hyperglycemia could compensatorily upregulate UCP-2 so as to negatively modulate Δψm and ROS because treatment with specific UCP-2 antisense oligonucleotide could increase hyperglycemia-induced mitochondrial membrane hyperpolarization and subsequent ROS generation. We also found that perindopril upregulated UCP-2 and reduced ROS in BRECs exposed to normal glucose, high glucose, or high glucose plus NAC, indicating ACEI can directly upregulate UCP-2; meanwhile, this effect of perindopril could be blocked by the specific UCP-2 antisense oligonucleotide, suggesting that perindopril exerts it effect through UCP-2. It can be concluded that ACEI can attenuate oxidative stress through both the NADPH oxidase pathway and the UCP-2/mitochondrial pathway. In view of the important role of ROS in the pathogenesis DR, the inhibitory effect of ACEI on mitochondrial ROS production might be an important mechanism for treatment of diabetes complications.

We also investigated the specific mechanism by which ACEI induces UCP-2 upregulation and found that PPARγ, a ligand-activated transcription factor belonging to the nuclear receptor superfamily, was involved in the process. PPARγ has two isoforms, γ1 and γ2, and they differ at their NH₂ terminus, which has the same sequence except for the additional 28 amino acids at the NH₂-terminus of PPARγ2. PPARγ2 is rich in the different adipose tissues, whereas PPARγ1 has a broader expression pattern, including the gut, brain, vascular cells, and specific kinds of immune and inflammatory cells (34,35). The measurement of PPARγ protein expression by anti-PPARγ polyclonal antibody could be used to determine the total protein of PPARγ1 and γ2. Owing to its critical role in fat metabolism and the role of PPARγ activators in diabetes prevention, PPARγ has been extensively studied in patients with diabetes (36). It has been reported that PPARγ could benefit vascular function and inhibit neovascularization, retinal leukostasis, and retinal leakage (37,,–40). Until now, no one has studied the effect of ACEI on PPARγ. Our study is the first to find that perindopril could upregulate PPARγ1 and γ2 in BRECs exposed to normal-glucose or high-glucose conditions; besides, PPARγ inhibitor GW9662 could inhibit the upregulation of UCP-2 expression induced by high glucose or perindopril, suggesting that the upregulation of UCP-2 expression is mediated, at least in part, by PPARγ.

In conclusion, our study suggests that ACEI exerts a protective effect on DR and this protective effect can be reflected by a decreased VEGF-to-PEDF ratio, which is a result of reduced mitochondrial ROS production—itself caused by ACEI-induced increase of PPARγ and subsequent upregulation of UCP-2 expression. It is indicated that ACEI possesses a great potential for treatment of DR; a long-term prospective study based on large samples is needed to verify the clinical effect of ACEI for DR.

This work was supported by grants from the Research Fund for the Doctoral Program of Higher Education of China (20060248077) and the National Nature Science Funding of China (30772370 and 30872828).

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

We thank Yu Danghui of Second Military Medical University Press for his polishing of the English language.