Oxidative stress inactivates VEGF survival signaling in retinal endothelial cells via PI 3-kinase tyrosine nitration

The initial stages of diabetic retinopathy are believed to lead to sight-threatening ischemic and proliferative retinopathy when a critical number of retinal capillaries become nonperfused. It has been reported that retinal endothelial cells in diabetic patients and animals undergo accelerated death by a process consistent with apoptosis (Mizutani et al., 1996). Vascular endothelial growth factor (VEGF), an angiogenic cytokine secreted by a variety of cells, is a potent survival factor for endothelial cells in vivo and in vitro (Gerber et al., 1998a). Paradoxically, even though levels of VEGF and VEGF receptor (VEGFR2) are substantially increased in diabetic retinas (Aiello et al., 1994; Duh and Aiello, 1999; El-Remessy et al., 2003b; Gardner et al., 2002; Joussen et al., 2002; Obrosova et al., 2001; Obrosova et al., 2003) and retinal endothelial cell death is also increased.

Vascular endothelial cells are an important target of hyperglycemic damage (Nishikawa et al., 2000). It has been reported that high glucose is associated with formation of reactive oxygen species and that glucose-induced oxidative stress contributes to the development of hyperglycemia-induced vascular injury (Ceriello et al., 2001; Ceriello et al., 2002; Cosentino et al., 1997; Graier et al., 1999). We have shown increased formation of nitric oxide, superoxide anion and their combination product peroxynitrite in retinal endothelial cells maintained in high glucose (El-Remessy et al., 2003a).

Peroxynitrite is a highly reactive oxidant that mediates a variety of biological processes including inhibition of key metabolic enzymes, lipid peroxidation, nitration of the protein tyrosine residues and reduction of cellular antioxidant defenses by oxidation of thiol pools (Misko et al., 1998; Salgo et al., 1995). Recently, it has been shown that increased levels of nitrotyrosine correlate with cell death in pancreas (Suarez-Pinzon et al., 1997) and in a model for oxygen-induced retinopathy (Brooks et al., 2001). It has been reported that high glucose treatment stimulated endothelial cell death in isolated hearts (Ceriello et al., 2002), in cultured endothelial cells (Morishita et al., 1997; Nakagami et al., 2001; Du et al., 1999; Zou et al., 2002) and in retinas of diabetic mice and patients (Kern et al., 2000; Mohr et al., 2002). Moreover, we have shown that increased formation of peroxynitrite mediates apoptosis in endothelial cells cultured in high oxygen (Gu et al., 2003). However, molecular mechanisms underlying diabetes-induced peroxynitrite formation and acceleration of apoptosis in endothelial cells have not been elucidated. The paradox between increases in VEGF expression and retinal cell death under diabetic condition has prompted us to study the effects of high glucose-induced oxidative stress and exogenous peroxynitrite on cell death in retinal endothelial cells in order to define the molecular mechanism by which peroxynitrite alters the survival signaling pathway of VEGF under high glucose conditions.

Retinal endothelial cells were isolated from bovine retinas by homogenization and a series of filtration steps following an established protocol in our laboratory (Behzadian et al., 1998). The cells (passages 6-8) were plated on gelatin-coated dishes or 4-well chamber slides, grown to confluence and then treated for 5 days in conditions of normal glucose (NG, 5 mM glucose) or high glucose (HG, 25 mM glucose). Various osmotic control agents including L-glucose, a glucose stereoisomer (LG, 25 mM) and 3-methyl-O-glucose, a glucose analogue that is transported into the cells but not metabolized (3mG, 25 mM) were examined. Cells were cultured in NG, HG, LG or 3mG serum-containing media for the initial 3 days of treatment and then cells were washed with PBS twice and switched to corresponding serum free media for 2 days. Serum starvation is known to induce apoptosis in endothelial cells (Gratton et al., 2001). Cell culture reagents were purchased from Gibco-BRL (Grand Island, NY), with the exception of fetal bovine serum (FBS), which was purchased from HyClone Laboratories, Inc (Logan, UT) and CS-C complete medium, which was from Cell Systems Corporations (Kirkland, WA). Inhibitors including L-NAME, uric acid and superoxide dismutase-polyethylene glycol (SOD-PEG) were obtained from Sigma-Aldrich (Saint Louis, MO). Inhibitors for peroxynitrite (FeTPPS) and for p38 MAP kinase (SB203580) were obtained from Calbiochem (La Jolla, CA).

Cells were switched to serum free medium, treated with 0.5 mM peroxynitrite and cultured for 16 hours. Peroxynitrite was purchased from Upstate Biotechnology (Lake Placid, NY). Stock concentrations of peroxynitrite were freshly prepared in 0.3 N NaOH. Peroxynitrite concentration was determined by spectrophotometry as described by Zou et al. (Zou et al., 2003). An equal amount of 0.3 N NaOH or decomposed peroxynitrite was used for control experiments. Neither of these had an effect on any of the parameters analyzed.

Caspase-3 is an intracellular cysteine protease that exists as a proenzyme, becoming activated during the cascade of apoptosis. The activity of the enzyme was determined using a kit from R&D systems (Minneapolis, MN) according to the manufacturer's recommendation. Briefly cells were lysed on ice for 10 minutes with lysis buffer provided with the kit. To 50 μl of cell lysate, 50 μl of 2× reaction buffer were added followed by 5 μl of caspase-3 fluorogenic substrate (DEVD-AFC). The mixture was incubated for 2 hours at 37°C and fluorescence was measured with a Cytofluor-4000 spectrophotometer (Foster City, CA) with an excitation of 400 nm and an emission of 505 nm. The assay was done with non-induced cells (serum-supplemented cells) for comparative analysis.

Apoptotic cells in cultures treated as described above were detected by nuclear staining with Hoechst 33258 (2.5 μg/ml in PBS). Hoechst 33258 was purchased from Molecular Probes (Eugene, OR). Cells were reacted with the stain for 30 minutes at room temperature, washed twice with PBS, examined and photographed in a Zeiss Axioskop microscope. At least 15 fields were counted for each treatment. Each field contained over 130 cells. Cells with condensed chromatin, shrunken, irregular or fragmented nuclei were considered apoptotic. The number of apoptotic nuclei was calculated relative to the total number of nuclei. Total cell density was also calculated for each treatment condition as an indicator of potential differences in cytotoxicity.

Retinal endothelial cells were harvested after various treatments and lysed in modified RIPA buffer (20 mM Tris, 2.5 mM EDTA, 1% Triton X-100, 1% deoxycholate, 0.1% SDS, 40 mM NaF, 10 mM Na₄P₂O₇, and 1 mM PMSF) for 30 minutes on ice. Insoluble material was removed by centrifugation at 12,000 g at 4°C for 30 minutes. Antibodies for phospho-p38 MAP kinase, p38 MAP kinase, phospho-Akt-1, Akt-1, PARP and Akt-1 kinase kit were purchased from Cell Signaling Technology, Inc (Beverly, MA). For analysis of cleaved poly ADP-ribose polymerase (PARP), phospho-p38 MAP kinase, p38 MAP kinase, phospho-Akt-1 or Akt-1, 50 μg of total protein was boiled in 2× Laemmli sample buffer, separated on a 10-12% SDS-polyacrylamide gel by electrophoresis, transferred to nitrocellulose and reacted with specific antibody. The primary antibody was detected using a horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibody (Amersham Biosciences, Buckinghamshire, UK) and enhanced chemiluminescence. Intensity of immunoreactivity was measured using densitometry.

Retinal endothelial cells were cultured in high glucose in the presence or absence of peroxynitrite decomposition catalyst, or treated with peroxynitrite for the desired time. Cell lysates were prepared as described above for immunoblotting. Antibodies for nitrotyrosine, p85 and p110 subunits of PI 3-kinase were purchased from Upstate Biotechnology (Lake Placid, NY). For PI 3-kinase tyrosine nitration, 500 μg protein was incubated with p85. The precipitated proteins were analyzed by SDS-PAGE and blotted with nitrotyrosine antibody or p85 for equal loading. To study the effects on the interaction of the regulatory subunit with the catalytic subunit of PI 3-kinase, 500 μg protein was incubated with p110. The precipitated proteins were analyzed by SDS-PAGE and blotted by p85 antibody or p110 antibody for equal loading.

Retinal endothelial cells were cultured in high glucose in the presence or absence of ROS inhibitors, or treated with peroxynitrite for the desired time. For immunoprecipitation, 500 μg protein was incubated with immobilized Akt-1G1 (Cell Signaling Technology, Beverly, MA) monoclonal antibody-containing beads overnight at 4°C with gentle rocking. The beads were washed twice with lysis buffer and once with kinase buffer (25 mM Tris, 5 mM β-glycerolphosphate, 2 mM DTT, 0.1 mM Na₃VO₄, 10 mM MgCl₂). Beads were further incubated in 40 μl kinase buffer supplemented with 200 μM ATP and 1 μg GSK-3 fusion protein for 30 minutes at 30°C. The reaction was terminated with 20 μl 3× SDS sample buffer. After boiling and centrifuging, the supernatant was separated on a 12% SDS-PAGE gel, transferred to nitrocellulose membrane and western immunoblotting was carried out with phopho-GSK-3α/β (Ser21/9) antibody as the primary antibody. Horseradish peroxidase-conjugated goat anti-rabbit antibody and enhanced chemiluminescence were used to detect the primary antibody. Intensity of the immunoreactivity was measured using densitometry.

β-Galactosidase and C-terminal HA-tagged constitutively active Akt (myr-Akt) were generated as described previously (Fulton et al., 1999). Retinal endothelial cells were infected with adenovirus [multiplicity of infection (m.o.i.) of 100] containing the β-galactosidase and myr-Akt. The virus was removed and cells were left to recover for 12 hours in complete medium. These conditions resulted in uniform expression of the transgenes in close to 95% of the cells (determined by infection with β-galactosidase, followed by staining for β-galactosidase activity). Western blot analysis using antibodies against phospho Akt-1 and GSK-3 (Cell Signaling Technology, Inc. Beverly, MA) confirmed the over-expression and activity of Akt-1 in cells transfected with myr-Akt but not in cells transfected with β-galactosidase.

The results are expressed as mean ± s.e.m. Differences among experimental groups were evaluated by ANOVA and the significance of differences between these groups was assessed by Fisher's post-hoc least significant difference test. Significance was defined at P<0.05.

Our previous studies in retinal endothelial cells showed that cultures maintained in high glucose had significant increases in superoxide anion, nitric oxide and peroxynitrite formation (El-Remessy et al., 2003a). In the present study, we examined the effects of high glucose-induced oxidative stress and exogenous peroxynitrite on VEGF pro-survival function. As shown in Fig. 1, high glucose and peroxynitrite treatment accelerated serum starvation-induced endothelial cell death as indicated by significant increases in caspase-3 activity compared to normal glucose. Treatment with vehicle or decomposed peroxynitrite had no effect on capsase-3 activity (data not shown). Caspase-3 is an intracellular cysteine protease that exists as a pro-enzyme, becoming activated during the cascade of intracellular signaling events that culminates in apoptosis. VEGF (40 ng/ml) inhibited serum starvation-induced activation of caspase-3 in control cultures maintained in normal glucose, but had no effect on cultures treated with either high glucose or exogenous peroxynitrite. Moreover, the specific peroxynitrite decomposition catalyst FeTPPS (2.5 μM) restored the pro-survival effects of VEGF on cultures maintained in high glucose. This finding suggests that the effect of high glucose in increasing apoptosis in retinal endothelial cells may be due to the action of peroxynitrite in altering the VEGF cell survival pathway.

Our previous studies have shown that osmotic control agents including the glucose stereoisomer (LG) and the glucose analog, 3-O-methyl glucose (3mG) also cause increases of superoxide anion and peroxynitrite formation in retinal endothelial cells, though to a lesser extent than high glucose. This effect was found to be due to activation of aldose reductase, a rate-limiting step in the polyol pathway (El-Remessy et al., 2003a). As shown in Table 1, caspase-3 activity was increased in cells cultured in high osmolarity (5 mM glucose + 20 mM LG or 5 mM glucose + 20 mM 3mG) compared to cells cultured in normal glucose. The effects of high glucose/osmotic stress were blocked by the specific aldose reductase inhibitor Zopolrestat (10 μM) and by the peroxynitrite scavengers uric acid (1 mM) and FeTPPS (2.5 μM). The caspase-3 activity was also blocked by inhibiting NOS using L-NAME (0.5 mM) and significantly reduced by superoxide dismutase (SOD, 100 U/ml) confirming the involvement of peroxynitrite in high glucose and high osmolarity-induced cell death.

The oxidative stress-induced cell death was additionally assessed by nuclear staining with Hoechst 33258 which showed that the cultures maintained in medium with high glucose or high osmolarity (3mG or LG) had significantly increased numbers of cells with shrunken or irregular nuclei characteristic of apoptosis (32%, 28% and 26% respectively, P<0.01, Table 2). Fig. 2A shows images of Hoechst nuclear stain of cells undergoing apoptosis. The increase in apoptotic cell profiles was significantly inhibited by treatment with uric acid, superoxide dismutase, or L-NAME providing further support for the role of peroxynitrite in the apoptotic process. The oxidative stress-induced cell apoptosis was further confirmed using western blot analysis of cleaved poly ADP-ribose polymerase (PARP). Detection of cleaved PARP (89 kDa), a cleavage target of caspase-3, is accepted as a marker for cell apoptosis. A representative western blot and normalized PARP/actin expression are shown in Fig. 2B. Cultures maintained in high glucose, 3-O-methyl glucose or exogenous peroxynitrite showed significant increases in PARP expression compared to cultures maintained in normal glucose.

The above data indicate that peroxynitrite-mediated oxidative stress accelerates retinal endothelial cell death in cultures maintained in high glucose and inactivates VEGF pro-survival function. Activation of the PI 3-kinase/Akt-1 pathway plays a critical role in the VEGF-survival pathway, whereas activation of p38 MAP kinase can induce cell death. We, therefore, examined whether the high glucose-induced impairment of VEGF-mediated cell survival function could be the result of alterations of the pro-survival Akt-1 pathway or the proapoptotic p38 MAP kinase pathway. Analysis of the VEGF-induced phosphorylation of Akt-1 showed that the proapoptotic effect of either high glucose-induced oxidative stress or exogenous peroxynitrite was associated with significant decreases in phosphorylation of Akt-1 in basal and VEGF-stimulated conditions (Fig. 3A). Analysis of the VEGF-induced phosphorylation of P38 MAP kinase showed that the pro-apoptotic effect of either high glucose or peroxynitrite was associated with significant increases in phosphorylation of p38 MAP kinase in the presence or absence of exogenous VEGF (Fig. 3B). Treatment of cultures maintained in high glucose with the specific peroxynitrite decomposition catalyst FeTTPS blocked the increases in p38 MAP kinase phosphorylation in both basal and VEGF-stimulated conditions. Control experiments with decomposed peroxynitrite showed no significant difference compared to normal glucose (data not shown).

Previous work in vitro has shown that the p85 regulatory subunit of PI 3-kinase is one target for peroxynitrite^-induced protein nitration on tyrosine (Hellberg et al., 1998). Our immunoprecipitation studies with an antibody that recognizes the p85 regulatory subunit of PI 3-kinase showed that cultures maintained in high glucose or treated with peroxynitrite had a significant increase in tyrosine nitration of p85 compared to cultures in normal glucose (Fig. 4A). These increases in tyrosine nitration were blocked by the specific peroxynitrite decomposition catalyst FeTTPs (2.5 μM) and by the specific nitration inhibitor epicatechin (100 μM). Epicatechin, a dietary flavanol, has been used to selectively block the nitrating effect of peroxynitrite on tyrosine residues (Schroeder et al., 2001; Schroeder et al., 2003). In order to investigate the effect of tyrosine nitration in causing PI 3-kinase dysfunction, we determined the effects of high glucose and exogenous peroxynitrite on the association between the regulatory subunit of p85 and the catalytic subunit p110 in the presence or absence of peroxynitrite decomposition catalyst (FeTPPs, 2.5 μM). We stimulated cells with VEGF (40 ng/ml) and immunoprecipitated the p110 catalytic subunit and detected the presence of p85 in normal glucose controls (Fig. 4B). However, in treatments with high glucose or peroxynitrite, p85 could no longer be detected in the immunoprecipitates. These results indicate the dissociation of p85 from p110, because reprobing the blots with p110 antibodies revealed the presence of p110. Treatments with FeTPPs restored the association between p85 and p110 in cultures of high glucose or peroxynitrite-treated cells and did not affect normal glucose controls. These results suggest that ONOO^– could modulate the cell survival responses mediated by PI 3-kinase activation.

In order to further investigate the physiological role of nitration of the PI 3-kinase subunits, we determined the effects of high glucose and exogenous peroxynitrite on the activity of Akt-1 kinase, a downstream target of PI 3-kinase in the survival pathway. Analysis of Akt-1 kinase activity in phosphorylating GSK-3 showed that both high glucose and exogenous peroxynitrite inhibited Akt-1 kinase activity. The effect of high glucose was reversed by the NOS inhibitor (L-NAME, 0.5 mM), superoxide dismutase (SOD, 100 U/ml) or specific peroxynitrite decomposition catalyst FeTPPS (2.5 μM, Fig. 4C).

To evaluate the causal relationship between the nitration of PI 3-kinase and the pro-apoptotic effects of high glucose and peroxynitrite, we tested the effects of the nitration inhibitor epicatechin (100 μM) in blocking activation of caspase-3. These experiments showed that inhibiting tyrosine nitration completely blocks the pro-apoptotic effects of high glucose or exogenous peroxynitrite as indicated by reduced caspase-3 activity (Fig. 4D). We verified the effects of blocking PI 3-kinase on blocking cell survival and inducing cell apoptosis by measuring caspase-3 activity. Cells were treated with the specific PI 3-kinase inhibitors wortmannin (100 nM) or LY294002 (10 μM) in the presence or absence of VEGF (40 ng/ml) and the activity of caspase-3 was determined. Under basal condition, treatments with wortmanin or LY294002 resulted in significant increases in caspase-3 activity (12.7±1.3 and 14.1±1.1 fold compared to 3.6±0.8 fold in control). Stimulation with VEGF rescued control cells and significantly reduced caspase-2 activity (1.7±0.2). However, in cells treated with wortmanin or LY294002, VEGF did not rescue cells and the activity of caspase-3 was significantly increased (13.1±1.1 and 16.9±1.7). Together, these findings confirm the correlation between the inhibitory effects of PI 3-kinase nitration, the decreases in Akt-1 activity and the proapoptotic effects of high glucose and peroxynitrite.

The above data imply that the p38 MAP kinase-driven proapoptotic pathway becomes important when the PI 3-kinase/Akt-1 pro-survival pathway is inhibited. To confirm the role of PI 3-kinase/Akt-1 inhibition in pro-apoptotic effects of high glucose or peroxynitrite, we used adenoviral-mediated gene transfer of constitutively active Akt-1 (myr-Akt) into retinal endothelial cells. Fig. 5A shows that infection with a myr-Akt blocked the high glucose- or peroxynitrite-induced apoptosis as compared to control cells transfected with β-galactosidase constructs. Moreover, the anti-apoptotic effect of myr-Akt was associated with significant decreases in phosphorylation of p38 MAP kinase in high glucose or peroxynitrite-treated cultures compared to cultures infected with β-galactosidase constructs. Fig. 5B shows a representative image of phospho p38 MAP kinase western blot and statistical analysis of three independent experiments.

To confirm the role of p38 MAP kinase activation in the pro-apoptotic effects of high glucose or exogenous peroxynitrite, we treated the cells with the specific p38 MAP kinase inhibitor SB203580 (25 μM) and assessed apoptosis by measurement of caspase-3 activity. The p38 MAP kinase inhibitor did not affect cell death in the normal glucose control but reduced significantly the apoptosis induced in high glucose-maintained or exogenous peroxynitrite-treated cultures (Fig. 6A). Moreover, inhibition of p38 MAP kinase restored basal levels of Akt-1 phosphorylation in high glucose or peroxynitrite-treated cultures, but had no significant effect on normal cultures (Fig. 6B). This result suggests that cross-talk between PI 3-kinase/Akt-1 and p38 MAP kinase signaling pathways contributes to the balance of pro- vs anti-apoptotic signaling pathway and hence the survival of endothelial cells.

The present study provides a novel mechanism of oxidative stress-induced apoptosis via tyrosine nitration-mediated inhibition of PI-3kinase/Akt-1 and activation of p38 MAP kinase pathway. In this study, we demonstrate that tyrosine nitration of the PI 3-kinase p85 subunit inactivates the survival signaling pathway in retinal endothelial cells cultured in high glucose or exogenous peroxynitrite. These effects are blocked by the specific peroxynitrite decomposition catalyst FeTPPs and by the specific nitration inhibitor epicatechin. We show also for the first time that the pro-apoptotic effect of high glucose or peroxynitrite is associated with imbalance between Akt-1 and p38 MAP kinase phosphorylation and that blocking p38 MAP kinase or over-expression of constitutively active Akt-1 masks the pro-apoptotic effect of high glucose or peroxynitrite and restores survival function in retinal endothelial cells.

High glucose concentrations mimicking the diabetic situation have been shown to trigger endothelial cell death (Baumgartner-Parzer et al., 1995; Du et al., 1999; Ido et al., 2002; Mizutani et al., 1996; Mohr et al., 2002; Quagliaro et al., 2003) and to increase formation of reactive oxygen and nitrogen species (Cosentino et al., 1997; El-Remessy et al., 2003a; Quagliaro et al., 2003; Zou et al., 2002). Paradoxically, high glucose also stimulates formation of the endothelial cell survival factor VEGF (Behzadian et al., 2003; Kim et al., 2000; Williams et al., 1997). Thus, we tested the hypothesis that high glucose increases apoptosis in endothelial cells through the action of peroxynitrite altering the VEGF cell survival pathway. We examined the pro-survival effects of exogenous VEGF (40 ng/ml) on rescuing endothelial cells from serum withdrawal-induced apoptosis under various conditions by determining caspase-3 activity. VEGF protected endothelial cells cultured in normal glucose but not those cultured in high glucose or peroxynitrite. The peroxynitrite decomposition catalyst (FeTPPs) restored the pro-survival effect of VEGF in cells cultured in high glucose, suggesting that high glucose alters pro-survival signaling of VEGF via peroxynitrite formation. Cell death by apoptosis was also confirmed by morphological study of nuclear changes, using the nuclear stain Hoechst 33258 and western blot analysis of cleaved poly (ADP-ribose) polymerase (PARP). The pro-apoptotic effects of high glucose and peroxynitrite were associated with activation of p38 MAP kinase in the presence or absence of exogenous VEGF. These results are in agreement with previous reports (Igarashi et al., 1999; Nakagami et al., 2001). Moreover, the peroxynitrite decomposition catalyst FeTPPS blocked the increases in p38 MAP kinase phosphorylation in high glucose cultures under basal or VEGF-stimulated conditions, further supporting a causal role of peroxynitrite inactivating VEGF pro-survival signal and causing cell death.

The role of the PI 3-kinase/Akt-1 pathway in VEGF's pro-survival function was confirmed by studies showing that the specific inhibitors of PI 3-kinase, including wortmanin and LY294002, caused retinal endothelial cells to undergo apoptosis as indicated by significant increases in caspase-3 activity in the presence or absence of exogenous VEGF. These results are in agreement with several reports showing that VEGF does not rescue endothelial cells from serum starvation-induced apoptosis in the presence of a specific inhibitor of PI 3-kinase (Fujio and Walsh, 1999; Gerber et al., 1998b; Gratton et al., 2001; Suhara et al., 2001). Previous work in vitro has shown that the p85 regulatory subunit of PI 3-kinase is a target for peroxynitrite-induced tyrosine nitration (Hellberg et al., 1998). Our data showed that cultures treated with high glucose or exogenous peroxynitrite had significant increases in tyrosine nitration of the p85 subunit. This effect was accompanied by dramatic decreases in the association of the regulatory p85 subunit from the catalytic p110 subunit, as well as by significant decreases in Akt-1 phosphorylation and Akt-1 kinase activity, confirming an inhibitory effect of the nitration of PI 3-kinase. Moreover, the specific peroxynitrite decomposition catalyst FeTPPs and the nitration inhibitor epicatechin blocked tyrosine nitration, restored p85/p110 interaction of PI 3-kinase and Akt-1 kinase and survival promoting activity. These results confirm the relationship between nitration of p85, the decreases in Akt-1 activity and the pro-apoptotic effects of high glucose and peroxynitrite. Akt-1 kinase activity was restored also by inhibiting NOS and by SOD. Our findings lend further support to previous reports of significant increases in tyrosine nitration in diabetic patients (Ceriello et al., 2001), in high glucose models (Ceriello et al., 2002; Zou et al., 2002) and in experimental diabetes (El-Remessy et al., 2003b; Turko et al., 2001). However, our study is the first we know of to elucidate the molecular mechanism of inhibitory tyrosine nitration of p85 leading to inactivation of the pro-survival effects of VEGF/Akt-1in a physiological model of high glucose-induced oxidative stress.

Further work is needed to determine the effects of high glucose and peroxynitrite-induced tyrosine nitration on other steps in the VEGF signal transduction pathway. Studies now in progress in our lab indicate that high glucose and peroxynitrite also induce tyrosine nitrosylation of VEGFR2. Interestingly, the high glucose treatment also causes increases in tyrosine phosphorylation of VEGFR2. This suggests that tyrosine nitration induces activation of VEGFR2, as has been shown previously for PDGF and EGF (Klotz et al., 2000; van der Vliet et al., 1998). More work is needed to define the specific mechanism of VEGFR2 activation by high glucose and to determine the molecular and physiological consequences. However, this observation suggests that the inhibitory effects of high glucose and peroxynitrite on cell survival are specific to a direct action of the tyrosine nitration inhibiting the PI 3K/Akt signaling pathway.

Parallel activation of survival and death pathways has recently been documented in response to VEGF, FGF and cytokines such as TNF-α (Cardier and Erickson-Miller, 2002; Gratton et al., 2001; Harfouche et al., 2003; Matsumoto et al., 2002). Thus, it is likely that high glucose and peroxynitrite-induced alterations in the Akt-1 and p38 MAP kinase interaction could alter cell responses to other cell survival stimuli in addition to VEGF. While further study is needed to directly test this hypothesis, our previous study has shown that peroxynitrite blocks the effects of either basic FGF or serum in activating the PI 3-kinase/Akt cell survival pathway (Gu et al., 2003).

We investigated the interaction between Akt-1 and p38 MAP kinase pathways under high glucose or exogenous peroxynitrite treatment. It is interesting that expression of constitutively active Akt-1 (myr-Akt-1) masked the proapoptotic effects of high glucose and exogenous peroxynitrite and restored cell survival function in treated cells. In addition, blocking of p38 MAP kinase also rescued retinal endothelial cells from accelerated cell death. The protective effects of SB203580 were more prominent in high glucose cultures than in peroxynitrite-treated cells, probably because of the magnitude of peroxynitrite-induced cell death. Our results are in agreement with previous reports in other vascular endothelial cells (Fujio and Walsh, 1999; Gratton et al., 2001; Harfouche et al., 2003; Matsumoto et al., 2002; Yue et al., 1999). Furthermore, the anti-apoptotic effect of myr-Akt-1 was associated with significant inhibition of p38 MAP kinase phosphorylation and the protective effect of SB203580 was associated with significant increases in Akt-1 phosphorylation. (see Figs 5 and 6). These findings demonstrate for the first time a cross talk between Akt-1 and the p38 MAP kinase signaling pathway in retinal endothelial cells. It has been reported that inhibition of p38 activation is mediated through phosphorylation and inhibition of MEKK3 by Akt-1 in aortic endothelial cells (Gratton et al., 2001). Similar results were reported (Harfouche et al., 2003) in HUVEC in response to angiopoietin-1 suggesting that the negative influence of Akt-1 on p38 MAP kinase is a general phenomenon in endothelial cells.

Of note, we and others have reported the effects of osmotic stress in increasing oxidative stress (Du et al., 1999; El-Remessy et al., 2003a; Obrosova et al., 2003; Shaw et al., 2003). Mannitol is commonly used as an osmotic control for high glucose treatment. However, it has been shown that mannitol acts as a free radical scavenger, thereby reducing intracellular levels of reactive oxygen species (El-Remessy et al., 2003a; Karasu, 2000). For osmotic control studies, we used the glucose stereoisomer (LG) and the glucose analog, 3-O-methyl glucose (3mG). Both LG and 3mG accelerated cell death though to a lesser extent than that seen with high glucose. These effects were blocked by the NOS inhibitor L-NAME, superoxide dismutase (SOD) and peroxynitrite scavengers (uric acid and FeTPPS) confirming the role of oxidative stress- and peroxynitrite-induced cell death. Moreover, blocking aldose reductase, a rate-limiting step in the polyol pathway, prevented the pro-apoptotic effects of osmotic stress in high glucose or osmotic control cultures. Similar results have been reported in vivo and in vitro (Obrosova et al., 2003; Miwa et al., 2003).

Taken together, our data suggest that the increased formation of peroxynitrite stimulated by high glucose inactivates VEGF pro-survival function in retinal endothelial cell and triggers apoptosis. In endothelial cells, VEGF promotes the parallel alteration of Akt-1 and p38 MAP kinase, anti- and proapoptotic pathways, respectively. It is suggested that inhibition of PI 3-kinase by peroxynitrite-induced tyrosine nitration triggers a molecular switch, reducing Akt-1 activation and facilitating the pro-apoptotic p38 MAP kinase pathway and thus initiating a molecular cascade in which VEGF loses its ability to sustain cell survival in the presence of high glucose or oxidative stress. A schematic representation of the proposed mechanism is shown in Fig. 7. The fact that the peroxynitrite decomposition catalyst FETPPS and the nitration inhibitor epicatechin restored pro-survival function in retinal endothelial cells opens the door for a possible new target in controlling early diabetic retinopathy.

This work was supported by NIH Grants: NIH-EY04618; NIH-EY11766, American Heart Association (AHA) Beginning Grant in Aid and by Postdoctoral Fellowships from Association for Research in Vision and Ophthalmology/Ciba Vision Inc. and AHA. Peter Oates, Pfizer Pharmaceutical Corp. provided Zopolrestat.