High Glucose-Induced Tyrosine Nitration in Endothelial Cells: Role of eNOS Uncoupling and Aldose Reductase Activation

Abstract

purpose. Analyses in diabetic rats have shown that breakdown of the blood–retina barrier is associated with increased formation of peroxynitrite, nitric oxide, and lipid peroxidation. The permeability increase is blocked by treatments that also prevent the increases in reactive oxygen species, suggesting their causal role in vascular dysfunction. The purpose of this study was to determine the specific effects of high glucose and high osmolarity on the formation of nitrotyrosine, nitric oxide, and superoxide anion in retinal vascular endothelial cells and to evaluate the metabolic pathways involved.

methods. Cultured retinal endothelial cells were maintained for 5 days in media with different concentrations of glucose or osmotic control reagents and tested for effects on protein tyrosine nitration and nitric oxide synthase (NOS) expression, using immunoblot techniques. NOS activity was determined by assays for nitrite formation and conversion of arginine to citrulline. Superoxide anion formation was assayed by hydroethidine staining.

results. Increased concentrations of glucose or 3-methyl-o-glucose stimulated formation of nitric oxide (NO) and superoxide induced protein nitration on tyrosine and increased expression and activity of endothelial nitric oxide synthase (eNOS). The effects of glucose were more potent: Inhibiting NOS or aldose reductase (AR), scavenging superoxide or peroxynitrite, or supplementing the NOS substrate l-arginine or cofactor tetrahydrobiopterin blocked the formation of reactive oxygen species and prevented protein tyrosine nitration.

conclusions. Increases in glucose levels and osmotic stress similar to those in diabetic patients increase the formation of nitrotyrosine in retinal endothelial cells because of their actions increasing NOS activity and causing superoxide formation due to eNOS uncoupling and AR activation.

Diabetic retinopathy is characterized by an initial period of vascular injury and increased permeability that is followed by the active proliferation of new vessels. ¹ The relation between diabetes-associated hyperglycemia and vascular disease is well established. However, the molecular mechanisms of hyperglycemia-induced endothelial dysfunction are not fully understood. ² Oxidative stress has been strongly implicated in the macrovascular complications in diabetes. This hypothesis is supported by evidence that many biochemical pathways strictly associated with hyperglycemia (glucose auto-oxidation, polyol pathway, protein glycation) can increase the production of free radicals. ³ Recent studies in diabetic rats have supported the role of reactive oxygen and nitrogen species in retinal vascular disease associated with diabetic retinopathy. ^{4 5 6} Furthermore, studies using cultured aortic endothelial cells have shown that treatment with high glucose increases endothelial nitric oxide synthase (eNOS) activity and promotes superoxide formation, reducing bioactivity of nitric oxide (NO). ^{7 8 9} Studies showing a reduction in NO bioavailability in retinal endothelial cells exposed to high glucose provide further support for this concept. ¹⁰  

Recently, attention has been focused on the role of peroxynitrite in diabetic vascular dysfunction. Peroxynitrite is a highly reactive oxidant formed by the rapid combination of NO with superoxide anion. Peroxynitrite formation contributes to vascular dysfunction indirectly by reducing NO bioavailability. Peroxynitrite initiates a variety of other pathologic processes including inhibition of key metabolic enzymes, lipid peroxidation, nitration of the protein tyrosine residues, reduction of cellular antioxidant defenses by oxidation of thiol pools, and induction of DNA strand breaks, leading to apoptosis. ^{11 12 13} Peroxynitrite can also cause oxidation of tetrahydrobiopterin, an essential cofactor of NOS. ¹⁴ Oxidation of tetrahydrobiopterin can lead to decreases in NO production and to NOS-uncoupled production of superoxide and hydrogen peroxide, resulting in increased oxidative stress and endothelial dysfunction. ¹⁵  

Peroxynitrite is difficult to measure directly because of its short half-life, but nitration of protein tyrosine residues can serve as its biomarker. ^{16 17} Analyses showing nitrotyrosine formation in diabetic rat retinas and in plasma from diabetic patients have provided evidence supporting the role of peroxynitrite in the vascular complications in diabetes. ^{6 18 19} Furthermore, our analyses in streptozotocin-induced diabetic rats have shown that early increases in retinal VEGF expression and vascular permeability are associated with increased formation of proteins nitrated on tyrosine as well as with increases in lipid peroxidation and nitrite-nitrate formation. ¹⁸ These effects were blocked by treatments with N-ω nitro-l-arginine methyl ester (l-NAME), which inhibits NOS activity, or with uric acid, which scavenges peroxynitrite, ²⁰ suggesting a causal role of NOS activity and peroxynitrite formation in the vascular permeability dysfunction. These findings were the basis for the present study to determine the effects of high glucose and high osmolarity on eNOS expression and activity and on the formation of NO, superoxide, and peroxynitrite in retinal endothelial cells. The purpose of this work was to explore the possible role of high-glucose–induced oxidative stress in eNOS uncoupling and the activation of polyol pathway and their impact on increased formation of superoxide anion and peroxynitrite. 

Materials and Methods

Cell Culture

Bovine retinal endothelial (BRE) cells were isolated by using a protocol established in our laboratory. ²¹ Cells (passages 4–8) were treated for 5 days in media containing different concentrations of glucose and osmotic control agents. For all experiments, unless specified, cells were grown in M199 medium (Invitrogen Corp., Carlsbad, CA) containing penicillin-streptomycin, 10% fetal bovine serum and 15% CS-C (Cell Systems Corporation, Kirkland, WA). Medium was changed on the third day of treatment. Four osmotic control agents were examined: 3-methyl-o-glucose (MG), l-glucose (LG), mannitol, and dextran (DN). MG is a glucose analogue that is transported into the cells, but is not metabolized. l-glucose, a glucose stereoisomer, and mannitol, a sugar alcohol, are not transported into the cells and are restricted to the extracellular space. Mannitol is commonly used as an osmotic control for high-glucose treatment. However, it can act as a free radical scavenger, thereby reducing levels of reactive oxygen species. ²² DN is a high molecular weight polysaccharide that is not transported into cells and therefore increases extracellular osmotic pressure. The media conditions were: normal glucose 5 mM (d-glucose [NG]), high glucose (HG; including 16.5 HG [16.5 mM d-glucose] and 25 HG [25 mM d-glucose]), 20 mM mannitol (5 mM d-glucose+20 mM mannitol), 20 mM DN (5 mM d-glucose+20 mM DN), high LG (5 mM d-glucose+20 mM lg), or high MG (including 11.5 MG [5 mM d-glucose + 11.5 mM MG] and 20 MG [5 mM d-glucose+20 mM MG]). 

Nitrite Assay

To investigate treatment effects on formation of NO and peroxynitrite, levels of nitrite, the oxidized product of NO, were determined in culture supernatants by a fluorometric assay using the 2,3-diaminonaphthalene (DAN) reagent as described. ²³ Briefly, cell homogenate (250 μL) was incubated with 25 μL DAN (633 μM in 0.67 N HCl) at room temperature in the dark for 10 minutes. The mixture was adjusted to pH 11.5 to 12 with 1 N NaOH. Fluorescence was measured with a spectrophotometer (CytoFluor 4000; Perspective Biosystems, Foster City, CA) with excitation of 365 nm and emission of 405 nm. Nitrite concentrations in the different samples were calculated by comparison with a standard curve (0.02–3.2 μM sodium nitrite). 

NOS Protein Expression

Protein from treated cells was extracted in lysis buffer (20 mM Tris, [pH 7.4]; 2.5 mM EDTA, [pH 8]; 1% Triton X-100, 1% deoxycholate, 1% sodium dodecyl sulfate, 50 mM sodium fluoride, and 10 mM sodium pyrophosphate) containing 1 mM phenylmethylsulfonyl fluoride. Protein samples equated for total protein were separated on 10% SDS gel (Bio-Rad, Hercules, CA) and transferred to nitrocellulose membranes. Levels of eNOS and neuronal (n)NOS proteins were analyzed by immunoblot analysis with eNOS, nNOS, and inducible (i)NOS monoclonal antibodies (Transduction Laboratories, Lexington, KY) followed by reaction with peroxidase-labeled goat anti-mouse IgG. Antibodies were detected using enhanced chemiluminescence (ECL; Amersham, Buckinghamshire, UK). The samples were normalized to an internal standard (β-actin) and analyzed by densitometry. 

NOS Activity

NOS activity was quantified by an assay for the conversion of [³H]-l-arginine to [³H]-l-citrulline in living cells, as described previously. ²⁴ Briefly, confluent cultures were incubated for 12 hours in l-arginine–free medium (containing the appropriate glucose and osmotic control reagent concentrations) and then switched to HEPES buffer with the following composition (mM): NaCl, 125; KCl, 5; NaHCO₃, 25; MgSO₄, 1.2; KH₂PO₄.H₂O, 1.19; CaCl₂.2H₂O, 2.54; glucose, 11; and HEPES, 10 (pH 7.4). Immediately, l-[2,3-³H]-arginine (2 μCi) and 10 μM cold l-arginine were added to each well. After 2 minutes, some cultures were treated with 10⁻⁶ M of Ca⁺² ionophore A23187. After 20 minutes, the reaction was stopped by a wash in cold buffer containing 20 mM HEPES, 5 × 10⁻⁶ M l-arginine and 4 × 10⁻³ M EDTA. In control experiments, cultures were pretreated with the NOS inhibitor N ^γ-monomethyl-l-arginine(l-NMMA, 10⁻³ M) for 15 minutes. At the end of the treatments, cell lysate was harvested, applied to 2-mL (Na form) columns (Dowex 50W-8; Dow Chemical Co. Midland, MI), and eluted with washing buffer. The amount of eluted [³H]-l-citrulline activity was determined by liquid scintillation counting (LS75; Beckman Instruments, Fullerton, CA). Basal formation of [³H]-l-citrulline was reduced by 71.7% ± 5.4% in the presence of 10⁻³ M l-NMMA. Cellular uptake of l-arginine was determined by counting the total cellular lysates mixed with scintillation fluid using scintillation-counting spectroscopy. ²⁵  

Hydroethidine Staining Assay for Superoxide Formation

Superoxide anion was measured by digital imaging microfluorometry of the oxidation of hydroethidine (HE) into ethidium. ²⁶ The culture media were replaced with HEPES buffer containing 2 μL HE and incubated for 10 minutes at room temperature. Cultures were washed with 37°C HEPES, and images of the fluorescent ethidium were collected immediately by confocal microscope. An imaging system (MetaMorph; Universal Imaging Corp., West Chester, PA) was used for quantitative analysis of fluorescence intensity. Ethidium optical density was normalized to the number of BRE cells. 

Aldose Reductase Activity

Aldose reductase (AR) activity was quantified by fluorometric assay of nicotinamide adenine dinucleotide phosphate (NADP), which results from reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidation and an aldose reduction as described. ²⁷ Cells were incubated in hypotonic buffer containing 0.1% Triton X on ice for 30 minutes and then homogenized (Dounce homogenizer; Kontes Glass Co., Vineland, NJ). AR activity was assayed in the clear supernatant. The total volume of the reaction mixture for the assay was 200 μL, in which 50 μL of cell lysate was preincubated with 50 mM potassium phosphate (pH 6.0), 0.4 M lithium sulfate, and 5 mM 2-mercaptoethanol for 3 minutes at 37°C. Subsequently, 0.1 mM NADPH was added, and the reaction mixture was incubated for 1 minute. The reaction was started by the addition of 10 mM dl-glyceraldehyde. The enzymatic reaction was stopped after 15 minutes by the addition of 0.1 mL 0.5 N HCl. After 10 minutes, 0.2 mL 6 N NaOH containing 10 mM imidazole was added, and the mixture was cooled to room temperature. The fluorescence was determined using a spectrophotometer (Cytofluor 4000; Applied Biosystems) with excitation of 360 nm and emission of 460 nm. Sample concentrations of AR activity were calculated using a standard curve of NADP (100–1000 picomoles) along with reagent blanks and expressed as milliunits per milligram protein. 

Nitrotyrosine Formation

The relative amounts of proteins nitrated on tyrosine were measured by slot-blot analysis, as described previously. ¹⁷ The results were confirmed by immunocytochemistry. For solid-phase immunoradiochemical assay, duplicate samples of cellular proteins were immobilized onto polyvinylidene difluoride (PVDF) membranes using a slot-blot micro filtration unit (Bio-Rad). A dilution series of peroxynitrite-modified BSA (Cayman Chemical Co., Ann Arbor, MI) was loaded in duplicate to generate a standard curve. After blocking, the nitrocellulose membrane was reacted with a monoclonal anti-nitrotyrosine antibody (Cayman Chemical Co.) followed by peroxidase-labeled goat anti-mouse IgG and ECL. Relative levels of nitrotyrosine immunoreactivity were determined by densitometry and comparison with the standard curve generated from peroxynitrite-modified BSA. This analysis showed that nitrotyrosine formation in normal glucose-treated cells was equivalent to 0.1 mg/mL versus 4.8 mg/mL nitrosylated BSA in positive control cells treated with 1000 μM peroxynitrite. For immunocytochemistry, cultures were fixed with 4% paraformaldehyde and then reacted with a polyclonal rabbit anti-nitrotyrosine antibody (Upstate Biotechnology, Lake Placid, NY). Oregon green–conjugated goat anti-rabbit antibody (Molecular Probes, Eugene, OR) was used to visualize the primary antibody. Data were analyzed with a morphometric computer program (MetaMorph) and fluorescence microscopy to quantify intensity of immunostaining. 

Statistical Analysis

The results are expressed as the mean ± SE. Differences among experimental groups were evaluated by ANOVA, and the significance of differences between groups was assessed by Fisher’s protected least-significant difference (PLSD) when indicated. Significance was defined as P < 0.05. 

Results

Endothelial Cell Formation of Nitrotyrosine

Peroxynitrite is a short-lived molecule at physiologic pH, but it produces stable nitration of protein tyrosine residues. Thus, nitrotyrosine immunoreactivity can be used to assay for peroxynitrite formation. ¹⁶ We determined the relative amounts of proteins nitrated on tyrosine by using an immunoblot procedure. The data showed that 25 mM glucose increased nitrotyrosine immunoreactivity approximately 1.8-fold above the level in the cultures maintained in 5 mM glucose (Table 1) . Nitrotyrosine reactivity was also significantly increased, though to a lesser extent, by 20 mM MG and 20 mM LG (∼1.6 and 1.7, respectively). In contrast, the cultures maintained in the 20-mM mannitol medium had nitrotyrosine levels even lower than the 5-mM glucose medium. The 20-mM DN medium-treated cultures were unchanged from the 5 mM glucose control. These results were confirmed by parallel studies, using quantitative immunocytochemical techniques (data not shown). 

Further analysis of cultures treated with increasing concentrations of glucose or MG revealed differences in dose-dependent effects of increasing glucose and osmotic pressure (Fig. 1) . Cultures maintained in 16.5 mM glucose showed significant increases in nitrotyrosine formation above the levels observed in the 5 mM glucose controls, whereas those maintained in 5 mM glucose+11.5 mM MG were unchanged from the 5 mM glucose control. However, as shown in the initial experiments, 25 mM glucose, 20 mM LG, and 20 mM MG media caused significant increases in nitrotyrosine formation. Therefore, increases in glucose concentration and osmotic pressure both increase nitrotyrosine formation, but the magnitude of the effect with MG or LG is less than that for glucose. 

Expression and Activity of eNOS

To see whether the increase in peroxynitrite formation was associated with increases in eNOS protein expression or activity we determined the effects of the various treatments on eNOS protein levels and on conversion of [³H]-l-arginine to [³H]-l-citrulline. Western blot analysis showed a twofold increase in eNOS protein levels in the cultures maintained in 25 mM glucose versus 5 mM glucose (Fig. 2) . Levels of eNOS protein in the cultures maintained in 20 mM MG or 20 mM DN were comparable with the 5 mM glucose control. Neither nNOS nor iNOS was detected in any of the treatment conditions. The analysis of the conversion of [³H]-l-arginine to [³H]-l-citrulline in living cells showed that basal NOS activity was significantly increased by culturing cells in 25 mM glucose (1.8-fold) and to a lesser extent by culture in 20 mM MG (1.4-fold; Table 1 ). DN-treated cultures did not show any change in NOS activity in comparison to the 5 mM glucose control cultures. Ca²⁺ ionophore-stimulated NOS activity was also increased significantly in the cultures treated with either 25 mM glucose (1.5-fold) or 20 mM MG (1.3-fold) in comparison to the other groups. Similarly, uptake of l-arginine was increased in the cultures maintained in either 25 mM glucose (1.9-fold) or 20 mM MG (1.9-fold) compared with those cultured in 5 mM glucose or 20 mM DN (Table 1) . This increase in uptake of l-arginine is consistent with the increase in NOS activity in both 25 mM glucose and 20 mM MG conditions. 

Analysis of the concentration-dependency of the glucose and osmolarity effects on basal NOS activity showed patterns of glucose-induced and osmolarity-augmented increases similar to those in the analysis of nitrotyrosine formation (Fig. 3) . In contrast, Ca²⁺-stimulated NOS activity was increased above the 5 mM glucose control cultures by either 25 mM glucose or 20 mM MG, but was not significantly changed by either 16.5 mM glucose or 11.5 mM MG (data not shown). 

These increases in NOS activity were confirmed by using the DAN reaction assay to measure nitrite accumulation in media conditioned by BRE cells maintained in different concentrations of glucose and osmolarity controls. Initial analyses showed that treatment with either 25 mM glucose, 20 mM LG, 20 mM MG, or 20 mM mannitol caused significant increases in nitrite formation compared with 5 mM glucose or 20 mM DN (Table 1) . The 25 mM glucose treatment also increased nitrite formation significantly above that seen with either 20 mM MG or 20 mM mannitol. 

Analysis of the concentration dependency of the glucose and osmolarity effects on nitrite formation showed the same trend observed in the analysis of nitrotyrosine formation. The 16.5 mM glucose treatment caused a significant increase in nitrite formation, whereas the 11.5 mM MG concentration did not alter nitrite formation. The 25 mM glucose and 20 mM MG treatments significantly increased nitrite formation (Fig. 4) . 

Superoxide Anion Formation

HE imaging was used to determine the effects of high glucose treatment on the formation of superoxide anion. HE is oxidized to ethidium by reaction with superoxide, resulting in an increase in fluorescence intensity. The specificity of the HE method for detection of superoxide was verified by control experiments in which ethidium formation was completely blocked by acute application of superoxide dismutase (100 U/mL) and unaffected by the acute application of catalase (100 U/mL). Control experiments testing the effect of l-NAME (0.5 mM) applied acutely showed an increase in ethidium formation confirming the well-established action of NO as superoxide scavenger. Quantitative analysis of the fluorescence intensity in the chronically treated cultures showed that both 25 mM glucose and 20 mM MG induced significant increases in endothelial superoxide production in comparison with 5 mM glucose (Fig. 5A) . Significant increases in superoxide anion formation were also observed in BRE cells treated with 20 mM LG. The levels of superoxide anion were significantly less in cells treated with 20 mM mannitol, whereas they were not altered in cells treated with 20 mM DN (data not shown). 

We next evaluated the metabolic pathway underlying the increase in superoxide formation. Possible sources of superoxide formation in endothelial cells include both eNOS and AR. AR is the rate-limiting step in the polyol pathway. AR is a NADPH-dependent oxidoreductase that reduces the aldehyde form of glucose to sorbitol. AR activity involves reduction of an aldose to its corresponding alcohol and consumption of NADPH. Depletion of NADPH inhibits the function of the antioxidant enzyme glutathione reductase, leading to decreased levels of glutathione and increased oxidative stress. To test whether AR contributes to superoxide formation in BRE cells, we treated the cultures with the AR inhibitor zopolrestat and measured superoxide formation. The results showed that zopolrestat significantly reduced the superoxide formation in cultures treated with either 25 mM glucose or 20 mM MG (Fig. 5A) . The AR inhibitor did not affect superoxide formation in the control cultures maintained in 5 mM glucose. AR activity was measured by NADP production and was significantly increased in cultures treated with 25 mM glucose or 20 mM MG compared with the 5 mM glucose control cultures (Fig. 5B) . NADP production indicating AR activity was totally blocked by zopolrestat. 

Uncoupling of eNOS

Endothelial NOS produces superoxide under uncoupling conditions where its substrate l-arginine and/or cofactor tetrahydrobiopterin (BH4) is limited. ^{28 29} To test whether eNOS contributes to superoxide anion formation in high glucose-cultured BRE cells, we treated the cultures with the NOS inhibitor, l-NAME; the NOS substrate, l-arginine; or the BH4 precursor, sepiapterin (Sep) and measured superoxide formation. The results showed that l-NAME but not d-NAME blocked the increase of superoxide formation induced by 20 mM glucose (Figs. 5C 5D) . Moreover, supplementation of sepiapterin or l-arginine but not d-arginine also significantly reduced the increase in superoxide formation in cultures treated with 25 mM glucose (Figs. 5C 5D) . Similar results were obtained with cultures treated with 20 mM MG (Fig. 5D) . 

Effect of Inhibition of AR or Scavenging Superoxide on Nitrotyrosine Formation

The results described thus far suggest that increases in glucose concentrations and high osmolarity cause increases in peroxynitrite formation due to their actions increasing AR activity and stimulating superoxide anion formation. We tested this concept by determining the effects of inhibiting AR or scavenging superoxide on nitrotyrosine formation. Cultures were maintained in 25 mM glucose, 20 mM MG, or 5 mM glucose in the presence or absence of superoxide dismutase and zopolrestat. Additional cultures were treated with the peroxynitrite scavenger uric acid as a positive control. The results of this experiment showed that, as expected, uric acid totally blocked the high glucose– and high osmolarity–induced increases in tyrosine nitration (Fig. 6A) . Moreover, treatment with superoxide dismutase was equally as effective as uric acid. Zopolrestat also significantly reduced tyrosine nitration, but the reduction was smaller than that induced by SOD and uric acid. 

Effect of Inhibition of NOS or Prevention of Uncoupling on Nitrotyrosine Formation

To test the role of eNOS uncoupling under high-glucose–induced oxidative stress and its impact on peroxynitrite formation, cultures were maintained in 25 mM glucose or 5 mM glucose in the presence or absence of l-NAME, the BH4 precursor sepiapterin (0.1 mM) or supplemental l-arginine (1 mM), and effects on tyrosine nitration were determined. The results of this experiment showed that tyrosine nitration was blocked by l-NAME and significantly reduced by either the addition of sepiapterin or supplemental l-arginine (Fig 6B) . This finding is consistent with the data in Figure 5C , showing that these treatments inhibited formation of the superoxide anion. It is noteworthy that neither d-NAME nor d-arginine altered nitrotyrosine formation (Fig 6C) . 

Effect of Scavenging Superoxide or Peroxynitrite on Nitrite Formation

As shown in Table 1 , 25mM glucose and 20 mM MG caused significant increases in nitrite formation, compared with 5 mM glucose, indicating increases in NOS activity in those treated cells. To further evaluate the role of NOS activity in formation of peroxynitrite, we also analyzed the effects of the ROS inhibitors on nitrite release by the treated cultures. This analysis showed that l-NAME and uric acid but not d-NAME inhibited the increase in nitrite formation in both 25 mM glucose and 20 mM MG treated cells (Fig 6D) . Cells maintained in 25 mM glucose or 20 mM MG and treated with SOD also showed significant increases in nitrite formation compared with cells treated with 5 mM glucose. The latter result suggests that most of the nitrite increase detected in our analysis reflects the formation of NO rather than peroxynitrite. We did not use nitrate reductase to convert nitrate to nitrite in our experiments, because the endothelial cell culture system lacks oxyhemoglobin, which normally converts nitrite to nitrate in vivo. ³⁰ However, it has been suggested that peroxynitrite can break down to form nitrate in some culture conditions. ³¹ Thus, the nitrite assay may have underestimated peroxynitrite formation in our experiments. 

Discussion

In the present study, (1) high glucose stimulates a dose-dependent increase in formation of peroxynitrite in retinal endothelial cells by increasing formation of superoxide anion and NO; (2) increases in osmolarity contribute to this effect, but the osmolarity effect is less prominent than glucose’s metabolic actions; and (3) the sites of increases in superoxide anion formation induced by high glucose include both eNOS and AR. 

Diabetes- and high-glucose–induced vascular dysfunction has long been thought to involve the inactivation of endothelium-derived NO by its combination with superoxide anion to form peroxynitrite. This concept has been supported by research showing nitrotyrosine formation in the plasma of diabetic patients ¹⁹ and in blood vessels and retinas of diabetic rats, ^{3 4 5 6} as well as by our research in diabetic rat retinas, which showed changes similar to those in cultured BRE cells in the current study, including increases in retinal nitrotyrosine levels, lipid peroxidation, eNOS expression, and NOS activity. ¹⁸ Other pathways of tyrosine nitration have been described in addition to peroxynitrite formation, but nitrotyrosine is considered to be a likely indicator for peroxynitrite under conditions of simultaneous production of NO and superoxide. ^{32 33 34}  

To our knowledge our results are the first to show that treatment of cultured retinal endothelial cells with high glucose or high osmolarity conditions induces increases in formation of superoxide and NO as well as in protein nitration on tyrosine. The effect of high glucose on tyrosine nitration is in agreement with previous work in other endothelial cell types. ^{3 31} Our observation that the 20 mM LG or 20 mM MG treatment also stimulated the formation of superoxide and NO and tyrosine nitration suggests that osmotic pressure also may have a role in the high glucose effects. Increases in superoxide anion formation have been observed in porcine aortic endothelial cells treated with high glucose or MG media. ^{7 8 9} However, the observed increases in superoxide formation were diminished by the metal chelator desferal, but were not affected by coincubation of cultures with inhibitors of cyclooxygenase, lipoxygenase, cytochrome P450, or constitutive NOS. ⁹ It was concluded that autoxidation and/or transition-metal-mediated d-glucose oxidation might account for superoxide anion formation. ^{7 9} Further work is needed to elucidate the mechanism by which 3-methylglucose and l-glucose mimic the glucose effect in increasing superoxide anion formation. 

The results of our experiments comparing effects of increasing concentrations of glucose and methyl glucose revealed that high glucose was more effective than high methyl glucose in increasing NOS activity and peroxynitrite formation (Figs. 1 3 4) . Furthermore, treatment with the intermediate glucose concentration (16.5 mM) caused smaller, but still significant increases in NOS activity and formation of nitrite and peroxynitrite, whereas the intermediate methyl glucose concentration had no significant effect on these parameters. Thus, the metabolic action of glucose to stimulate formation of reactive oxygen species is more prominent than its action increasing osmotic stress. In contrast with the high MG and LG, mannitol significantly reduced nitrotyrosine formation, and DN had no effect on any parameter measured. The mannitol effect can be explained by the fact that mannitol can act as a free radical scavenger to reduce levels of reactive oxygen species. ²² The lack of an effect of the DN treatment indicates that increasing extracellular osmotic pressure does not contribute to oxidative stress in BRE cells. 

Our findings of increased superoxide formation in retinal endothelial cells treated with high glucose are consistent with previous work in other endothelial cell types. ^{3 7 8 9 31 35} The mitochondrial electron transport chain has been reported to be an initial site of hyperglycemia-induced superoxide anion production. ³⁵ The effect of oxidative stress on the production and normal functions of NO has not been fully elucidated. However, high glucose-induced increases in superoxide anion formation cause increases in intracellular calcium levels, leading to activation of eNOS. ⁸ Our data are consistent with this observation. We found that BRE cells treated with high glucose or high MG had significant increases in NOS activity, as shown directly by an assay for the conversion of l-arginine to l-citrulline and indirectly by an assay for nitrite formation. 

In addition to increasing NOS activity, high glucose treatment also promoted increases in eNOS protein levels, which probably also contributed to the increase in eNOS activity. Protein levels of eNOS were not altered by any of the osmotic control treatments. Neither nNOS nor iNOS protein was detectable in the cultured BRE cells under any of the treatment conditions. The high glucose-induced increase in eNOS protein is consistent with our previous studies showing substantial increases in eNOS protein expression in diabetic rat retinas. ¹⁸ Significant increases in eNOS protein expression have also been reported in aortic endothelial cells cultured in high glucose ³⁶ or treated with superoxide in the presence of superoxide dismutase or with hydrogen peroxide. ^{37 38}  

Our finding that formation of excess superoxide anion and peroxynitrite can be prevented by inhibiting eNOS with l-NAME indicates that eNOS uncoupling is a primary source of these reactive oxygen species. eNOS uncoupling refers to a process in which the enzyme generates superoxide rather than l-citrulline when either its substrate l-arginine or its cofactor BH4 is limited. ^{14 28 29} Both of these phenomena have been implicated in diabetic vascular dysfunction, ^{39 40} yet their direct effect on eNOS uncoupling and superoxide anion formation in vascular endothelial cells has not been demonstrated previously. This was approached by either blocking NOS activity or by supplementation of the NOS substrate or cofactor precursor and measuring the formation of superoxide anion. To the best of our knowledge, we are the first to show that the glucose-induced increases in superoxide and peroxynitrite formation in endothelial cells can be blocked by inhibiting NOS or providing supplemental BH4 precursor or l-arginine. These findings clearly implicate eNOS uncoupling in the glucose-induced oxidative injury of the retinal vasculature. Our results are in good agreement with the beneficial effect of l-arginine or BH4 precursor supplementation in reversing endothelial dysfunction induced by diabetes or hyperglycemia and hyperlipidemia in vivo. ^{39 40 41}  

Activation of the polyol pathway may also contribute to the glucose-induced increases in formation of superoxide and peroxynitrite. During hyperglycemia, glucose flux through the polyol pathway is increased from 10% to 30% due to saturation of hexokinase with ambient glucose. ⁴² AR, a member of NADPH-dependent aldo-keto reductase family, is the first and rate-limiting step in the polyol pathway. ⁴³ It has also been shown that AR exhibits broad substrate specificity and that its expression is regulated by osmolarity, ⁴⁴ suggesting that exposure to osmotic stress could activate this pathway. This assumption was confirmed by our data showing significant increases in AR activity in retinal endothelial cells treated with high methyl glucose. Moreover, the AR inhibitor zopolrestat inhibited the high glucose/osmolarity-induced increases in superoxide anion and nitrotyrosine formation. 

It has been reported that AR activity is regulated by NO and that as NO bioavailability decreases, AR activity increases. ⁴⁵ Therefore, the high glucose/MG-induced increase in superoxide anion may also stimulate AR activity indirectly by reducing NO bioavailability. AR activation leads to increased consumption of NADPH, which can contribute to diabetes-induced vascular dysfunction by reducing activity of glutathione reductase leading to diminished glutathione levels and further increases in formation of superoxide anion and peroxynitrite. 

Taken together, our results indicate that the formation of NO, superoxide, and peroxynitrite can be stimulated by increases in glucose levels commonly encountered in diabetic patients and that the glucose effect is augmented by glucose-induced osmotic stress. A suggested working model showing potential cellular mechanisms that could be involved in the glucose/osmotic stress–induced formation of superoxide and peroxynitrite is presented in Figure 7 . Additional study is needed to evaluate these mechanisms further. 

 Supported by National Eye Institute Grants EY04618 and EY11766, an unrestricted grant from Research to Prevent Blindness, American Heart Association Grant 0130154N, and postdoctoral fellowships from the American Heart Association and ARVO/Ciba Vision Inc. Peter Oates, Pfizer Pharmaceutical Corp., provided zopolrestat to ABE.

 Submitted for publication October 4, 2002; revised December 20, 2002, and February 6, 2003; accepted February 20, 2003.

 Disclosure: A.B. El-Remessy, Pfizer Pharmaceutical Corporation (F); G. Abou-Mohamed, None; R.W. Caldwell, None; R.B. Caldwell, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Effects of Glucose and Osmolarity Treatments on Formation of Peroxynitrite and Nitrite, NOS Activity, and Uptake of l-Arginine

	5 mM Glucose	25 mM Glucose	20 mM Methyl Glucose	20 mM Dextran	20 mM Mannitol	20 mM l-Glucose
Peroxynitrite (optical density, nitrotyrosine)	100.0 ± 8.5	179.8 ± 19.5*	159.3 ± 13.8*	105.0 ± 3.8	85.0 ± 9.5*	169.2 ± 11.8*
Basal NOS activity (pM/min/mg protein)	2.17 ± 0.12	3.84 ± 0.26*	2.98 ± 0.09*	2.22 ± 0.31	ND	ND
Ca2+-stimulated NOS activity(pM/min/mg protein)	4.05 ± 0.33	5.97 ± 0.34*	4.98 ± 0.10*	4.13 ± 0.52	ND	ND
l-Arginine uptake (total counts/mg protein)	2.74 ± 0.34	5.27 ± 0.62*	5.2 ± 0.53*	2.49 ± 0.23	ND	ND
Nitrite (mM/million BRE cells)	0.91 ± 0.11	2.28 ± 0.14*	1.81 ± 0.15*	1.14 ± 0.08	1.59 ± 0.1*	1.92 ± 0.14*

Figure 1.

 

Effects of glucose and MG on formation of peroxynitrite in BRE cells. Inset: A representative slot-blot analysis of nitrotyrosine. Quantification of nitrotyrosine immunoblots showed significant increases in nitrotyrosine formation in cells treated with 16.5 or 25 mM glucose and in cells treated with 5 mM glucose plus 20 mM MG in comparison with control cultures treated with 5 mM glucose. Results shown represent three independent experiments. Optical density of NG is equivalent to 0.1 mg nitrosylated BSA. Relative optical density mean ± SE for 5 mM glucose = 100.0 ± 8.5; *P < 0.01 compared with 5 mM glucose control; n = 9.

Figure 2.

 

Effects of glucose and osmolarity control media on eNOS protein levels in cultured BRE cells. Window shows representative Western blot images of eNOS and β-actin protein levels. Results shown represent three independent experiments. Statistical analysis of relative optical density showed that cells treated with 25 mM glucose (HG) had significant increases in eNOS expression in comparison to control cultures treated with 5 mM glucose (NG). Cells treated with 5 mM glucose and either 20 mM MG, 20 mM mannitol (HM), or 20 mM DN were not significantly different from the controls. *P < 0.01 compared with 5 mM glucose control; n = 6.

Figure 3.

 

Effects of glucose and MG on basal NOS activity using an assay for the conversion of [³H]-l-arginine to [³H]-l-citrulline in BRE cells. Results shown represent two independent experiments. Basal NOS activity is increased significantly in cells treated with 16.5 or 25 mM glucose or with 5 mM glucose+20 mM MG in comparison to 5 mM glucose control. Mean ± SE for 5 mM glucose = 3.05 ± 0.9 pM/min per milligram protein, *P < 0.01 compared with 5 mM glucose control, n = 8.

Figure 4.

 

Effects of glucose and MG on formation of nitrite as determined using the DAN assay. Results shown represent three independent experiments. Cells treated with 16.5 or 25 mM glucose or with 5 mM glucose+20 mM MG had significant increases in nitrite formation compared to control cultures treated with 5 mM glucose (Mean ± SE for 5 mM glucose = 0.98 ± 0.05; *P < 0.01 compared with 5 mM glucose control, n = 8).

Figure 5.

 

Effects of glucose and MG on AR activity and formation of superoxide in BRE cells, as detected by HE. Results shown represent three independent experiments. (A) Representative image and densitometric analysis of HE staining normalized to the number of BRE cells treated with 25 mM glucose (HG) or 5 mM glucose+20 mM MG showed significant increases in superoxide formation that were significantly reduced by the AR inhibitor zopolrestat (ARI, 1 mM). *P < 0.01 compared with 5 mM glucose, n = 9. (B) Effects of glucose and MG on AR activity in BRE cells as determined using a fluorometric assay of NADP. Cells treated with 25 mM glucose (HG) or with 5 mM glucose+20 mM MG had significant increases in AR activity compared with control cultures treated with 5 mM glucose (NG). *P < 0.01 compared with 5 mM glucose control, n = 9. (C) Representative image and densitometric analysis of HE staining densitometry normalized to the number of BRE cells treated with 25 mM glucose (HG) showed significant increases in superoxide formation that was blocked by either the NOS inhibitor, l-NAME, or the NOS cofactor precursor, sepiapterin (Sep, 0.1 mM) and significantly reduced by and l-arginine (1 mM). *P < 0.01 compared with 5 mM glucose; #P < 0.05 compared with 25 mM glucose, n = 9. (D) Representative image and densitometric analysis of HE staining of BRE cells treated with 25 mM HG or 5 mM HG+20 mM MG showed significant increases in superoxide formation that was not altered by either d-NAME (0.5 mM) or d-arginine (1 mM). *P < 0.01 compared with 5 mM glucose control, n = 6.

Figure 6.

 

Effects of blocking AR activity or scavenging superoxide or peroxynitrite on formation of nitrotyrosine and nitrite. Results shown represent three independent experiments. (A) Quantification of nitrotyrosine immunoblots showed significant increases in nitrotyrosine formation in cells treated with 25 mM HG or 5 mM HG+20 mM MG in comparison with control cultures treated with 5 mM glucose (NG). These increases were prevented by treatment of the cells with superoxide dismutase (100 U/mL, SOD) or uric acid (1 mM). The AR inhibitor (ARI), zopolrestat reduced significantly the formation of nitrotyrosine. *P < 0.01 compared with 5 mM glucose and #P < 0.05 compared with 25 mM glucose or MG, n = 9. (B) Quantification of nitrotyrosine immunoblots showed significant increases in nitrotyrosine formation in cells treated with 25 mM HG in comparison to control cultures treated with 5 mM glucose (NG). These increases were prevented by treatment of the cells with the NOS inhibitor, l-NAME (0.5 mM) or NOS cofactor precursor, sepiapterin (Sep, 0.1 mM) or NOS substrate, l-arginine. *P < 0.01 compared with 5 mM glucose control, #P < 0.01 compared with HG, n = 9. (C) Quantification of nitrotyrosine immunoblots showed significant increases in nitrotyrosine formation in cells treated with 25 mM glucose (HG) in comparison to control cultures treated with 5 mM glucose (NG). These increases were not prevented by treatment of the cells with d-NAME (0.5 mM) or d-arginine (d-Arg). *P < 0.01 compared with 5 mM glucose control, n = 9. (D) Effects of scavenging superoxide or peroxynitrite on formation of nitrite as determined using the DAN assay. Cells treated with 25 mM glucose or with 5 mM glucose+20 mM MG had significant increases in nitrite formation compared with control cultures treated with 5 mM glucose. These increases were inhibited by the NOS inhibitor l-NAME (0.5 mM) or the peroxynitrite scavenger uric acid (1 mM), but not with SOD (100 U/mL). *P < 0.01 compared with 5 mM glucose control; #P < 0.01 compared with HG or MG, n = 9.

Figure 7.

 

Suggested working model showing proposed scheme for the effects of high glucose increasing formation of superoxide anion and peroxynitrite (ONOO⁻) in BRE cells. High glucose generates increased formation of superoxide anion (O \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\cdot}}{2}\) \end{document} ) due to glucose mitochondrial electron chain transport and/or depletion of NADPH by activation of the polyol pathway. O \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\cdot}}{2}\) \end{document} causes an increase in intracellular Ca²⁺, leading to eNOS activation and NO formation. NO and O \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\cdot}}{2}\) \end{document} combine to form ONOO⁻. ONOO⁻ oxidizes cellular components that may include the NOS substrate BH4 and the l-arginine transporter (T). Reduction in availability of BH4 and/or l-arginine uncouples NOS, leading to further increases in formation of O \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\cdot}}{2}\) \end{document} and ONOO⁻. This cycle can be interrupted by treatments with scavengers of O \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\frac{{\cdot}}{2}\) \end{document} or ONOO⁻ or inhibitors of NOS activity.