Aldose Reductase Inhibitor Fidarestat Prevents Retinal Oxidative Stress and Vascular Endothelial Growth Factor Overexpression in Streptozotocin-Diabetic Rats

The experiments were performed in accordance with regulations specified by The Guiding Principles in the Care and Use of Animals (DHEW Publication, NIH 80-23) and the University of Michigan Protocol for Animal Studies.

Male Wistar rats (Charles River, Wilmington, MA), initial body weight 227–267 g, were fed a standard rat chow diet (ICN Biomedicals, Cleveland, OH) and had ad libitum access to water. Diabetes was induced by a single injection of streptozotocin (55 mg/kg body wt i.p.). Blood samples for measurements of glucose were taken from the tail vein ∼48 h after streptozotocin injection and the day before the rats were killed. Rats with blood glucose of 13.9 mmol/l or more were considered diabetic. The experimental groups included control rats and 6-week diabetic rats treated with or without low (2 mg/kg per day) or high (16 mg/kg per day) doses of fidarestat administered in the diet. The treatments were started after induction of diabetes.

Unless otherwise stated, all chemicals were of reagent-grade quality and were purchased from Sigma Chemical, St. Louis, MO. Methanol (high-performance liquid chromatography grade), perchloric acid, hydrochloric acid, and sodium hydroxide were obtained from Fisher Scientific (Pittsburgh, PA). Ethyl alcohol (200 proof dehydrated alcohol, U.S.P. punctilious) was purchased from McCormick Distilling (Weston, MO). β-d-Glucose, sorbitol, N.F., and d-fructose, U.S.P., were purchased from Pfanstiehl Laboratories (Waukegan, IL). Dulbecco’s modified Eagle’s medium (DMEM) was purchased from GIBCO (Grand Island, NY). 5-(and-6)-chloromethyl-2′,7′-dichlorodihydro-fluorescein diacetate, acetyl ester (CM-H₂DCFDA) was obtained from Molecular Probes (Eugene, OR). Kits for measurements of malondialdehyde (MDA) and 4-hydroxyalkenals (4-HA) and glutathione peroxidase activity were purchased from Oxis International (Portland, OR); Rneasy kits for RNA isolation, Sensiscript Reverse Transcriptase kits, and HotStarTaq DNA Polymerase kits were from Qiagen (Chatsworth, CA); kits for VEGF protein and mRNA measurements and VEGF protein standard were from R&D Systems (Minneapolis, MN); and bicinchoninic acid protein assay kits were from Pierce (Rockford, IL). Deoxycytidine 51-[α-³²P]-triphosphate was purchased from Amersham Pharmacia Biotech (Piscataway, NJ).

Rats were sedated with carbon dioxide and immediately killed by cervical dislocation. Both retinas were rapidly dissected and frozen in liquid nitrogen. Three retinas from three animals were pooled for each VEGF protein measurement, and two retinas from one animal for RNA isolation. Other variables (i.e., glucose, sorbitol, fructose, MDA plus 4-HA, reduced glutathione [GSH], oxidized glutathione [GSSG], total ascorbate [AA], and dehydroascorbate [DHAA] concentrations and superoxide dismutase [SOD], catalase, glutathione peroxidase [GSHPx], glutathione reductase [GSSGRed], and glutathione transferase [GSHTrans] activities) were measured in individual retinas.

The retinas were homogenized, and the homogenates centrifuged as we described previously (17). Fifty microliter aliquots of the supernatant were used for VEGF measurements. VEGF was assayed by a sandwich ELISA with an affinity-purified polyclonal antibody specific for mouse VEGF and mouse VEGF standard according to the manufacturer’s instructions. Good linearities (r_mouse = 0.996 and r_rat = 0.986) were obtained with this antibody using both mouse and rat VEGF standards in the range of 5–500 pg/ml as well as rat VEGF standards in the range of 15–500 pg/ml added to rat retinal homogenates (r = 0.964). VEGF concentrations were normalized to total protein quantified according to the manufacturer’s instructions.

The isolation of total RNAs was performed in accordance with the manufacturer’s instructions. The absorbance of isolated RNAs was assessed spectrophotometrically at λ = 260 nm and λ = 280 nm. The absorbance₂₆₀/absorbance₂₈₀ ratio was used as a criterion of the RNA purity. The RNA yield from individual samples was calculated by multiplying the absorbance at 260 nm by 4, i.e., the absorbance of 1 μg/ml RNA solution in the 1-cm path quartz cuvette. The integrity of isolated RNAs was assessed by resolving four randomly selected RNA samples on a formaldehyde-denaturing agarose gel and comparing the quality and distribution of the ribosomal RNA 28S and18S fragments.

The measurements were performed in 3.5-μg RNA samples in duplicate in accordance with the manufacturer’s instructions. Taking into consideration that the assay has been developed for human VEGF mRNA, we have obtained calibration curves with human and rat VEGF mRNA preliminary synthesized from the plasmids for rat and human VEGF probes. Representative calibration curves are presented in Fig. 1. The results of this comparison indicate that the kit can be used for measurements of rat VEGF mRNA.

VEGF-164 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNAs were analyzed by RT-PCR. First, the mRNA template was copied into the complementary single-stranded cDNA (first-strand cDNA synthesis). This was performed using 50 ng retinal RNA, oligo(dT), and Sensiscript Reverse Transcriptase Kit according to the manufacturer’s protocol. Second, VEGF cDNA fragment was amplified by PCR, using forward primer (5′-CCGAAACCATGAACTTTCTGC-3′) and reverse primer (3′-GTGACAAGCCAAGGCGGTGAG-5′). The reverse primer oligonucleotide and the forward primer oligonucleotide AAACCATGAACTTTCTGC correspond to the nucleotide sequences 558–578 and 1–17 of rat VEGF mRNA (GenBank accession no. M32167). The GAPDH cDNA fragment was amplified using the forward primer (5′-CACCACCATGGAGAAGGCTG-3′) and the reverse primer (5′-ATGATGTTCTGGGCTGCCCC-3′) that correspond to 330–349 and 644–625 nucleotide sequences of the rat GAPDH gene, respectively (GenBank accession no. NM017008). The amplification was performed using HotStarTaq DNA Polymerase Kit and 0.25 μCi of deoxycytidine 5′-[α-³²P]-triphosphate according to the manufacturer’s protocol. Amplified PCR products were analyzed by electrophoresis in 2% agarose gel. Gels were photographed using the Quantity One BioRad System (BioRad Laboratories, Hercules, CA). The bands corresponding to VEGF-164 and GAPDH mRNAs were cut and the DNAs extracted using Ultrafree-DA column (Millipore Corporation, Bedford, MA). The VEGF-164 cDNA radioactivities were counted using Beckman LS (6000SC) System (Beckman, Fullerton, CA), and normalized to those of the corresponding GAPDH cDNAs.

Individual retinae were homogenized in 1 ml of 0.1 mol/l sodium-phosphate buffer, pH 6.5. Deproteinization and spectrofluorometric measurements of glucose, sorbitol, and fructose by enzymatic procedures were performed as we described previously (28).

0.2 ml of the retinal homogenates obtained in section five were used for measurements of MDA plus 4-HA perfomed in accordance with the manufacturer’s instructions. Another 0.2 ml were used for GSH and GSSG measurements (29).

0.2 ml of the homogenates described in the section five were mixed with 0.2 ml of ice-cold 10% metaphosphoric acid and centrifuged at 4,000g for 10 min. Total AA and DHAA concentrations were measured as we have described previously (17). AA concentrations were calculated as the difference between total AA and DHAA.

Individual retinae were homogenized in 1 ml of 0.1 mol/l sodium-phosphate buffer, pH 6.5, and centrifuged at 20,000g for 20 min. The supernatants were used for spectrophoto-metric measurements of SOD, catalase, GSSGRed, GSHPx, and GSHTrans activities (29,30).

BREC cultures were established from fresh calf eyes as we described previously (31). Passages 2–10 were used for all experiments. Purity of cultures was confirmed by characteristic endothelial cell morphology and by either >90% uptake of acetylated LDL or >90% immunopositivity for factor VIII, both endothelial cell-specific markers (31).

The BREC were cultured (eight plates per condition) for 7 days in DMEM containing either 5 or 30 mmol/l glucose with or without 1 μmol/l fidarestat added for 4 h at the end of the experiment. 10 μl of 10 μmol/l CM-H₂DCFDA, the dichlorofluorescein derivative with the best retention properties among all the studied analogs, was added for 30 min at the end of the experiment. The cells were then washed and trypsinized, and the CM-H₂DCFDA fluorescence, an index of ROS generation, was measured by flow cytometry (λ excitation 480 nm; λ emission 520 nm). After flow cytometry, the BREC were counted, and CM-H₂DCFDA fluorescence was expressed per 10⁶ cells. Parallel cultures (eight plates per condition) were grown without CM-H₂DCFDA and used for measurements of the sorbitol pathway intermediates.

The BREC (eight plates per condition) were cultured for 2 days in DMEM containing 5 mmol/l glucose with or without one of the following pro-oxidants: l-buthionine(S,R)-sulfoximine (BSO, 100 μmol/l), diethyl maleate (DEM, 60 μmol/l), or primaquine (50 μmol/l). The pro-oxidants were added for 24 h at the end of the experiment. Parallel pro-oxidant-treated BREC cultures contained 1 μmol/l fidarestat added for 4 h at the end of the experiment. ROS generation was assessed by the CM-H₂DCFDA fluorescence assay as described above.

The results are expressed as mean ± SE. Data were subjected to equality of variance F test and then to log transformation if necessary, before one-way ANOVA. When overall significance (P < 0.05) was attained, individual between-group comparisons were made using the Student-Newman-Keuls multiple range test. Significance was defined at P ≤ 0.05. When between-group variance differences could not be normalized by log transformation, the data were analyzed by the nonparametric Kruskal-Wallis one-way ANOVA, followed by the Bonferroni-Dunn test for multiple comparisons.

The final body weights were lower in diabetic rats than in the control group (Table 1). The initial body weights were similar in control and diabetic groups. No statistically significant difference was found between final body weights in diabetic rats treated with either low or high doses of fidarestat and the corresponding untreated group.

Blood glucose concentration was increased 4.5-fold in diabetic rats compared with controls. Blood glucose concentrations in diabetic rats were not affected by the low- or high-dose fidarestat treatments.

Retinal glucose, sorbitol, and fructose concentrations were 8.3-fold, 5.1-fold, and 4.3-fold higher in untreated diabetic rats than in the control rats (Table 2). Diabetes-induced increase in retinal glucose concentrations was not affected by treatments with either low or high doses of fidarestat. Retinal sorbitol concentrations were reduced 2.5-fold but not completely normalized by the low-dose fidarestat treatment. Fructose concentrations tended to decrease, but the difference with the untreated diabetic group did not achieve statistical significance. High dose of fidarestat completely normalized retinal sorbitol concentration, whereas retinal fructose concentration remained slightly elevated, but the difference with the control group did not achieve statistical significance.

Retinal VEGF protein concentration was increased twofold in diabetic rats compared with the control group, and this increase was dose-dependently prevented by fidarestat (Fig. 2A). High dose fidarestat completely arrested diabetes-induced retinal VEGF protein overexpression. Total VEGF mRNA (Fig. 2B) and VEGF-164 mRNA abundance (Fig. 3A and B) were similar among the experimental groups.

Retinal lipid peroxidation product concentration was increased 1.6-fold in the diabetic rats compared with the control group, and this increase was totally arrested by both low and high doses of fidarestat (Table 3). Retinal GSH, GSSG, total AA, and AA and DHAA concentrations were similar among control rats and diabetic rats treated with or without fidarestat. Neither glutathione redox state nor ascorbate redox state (expressed as GSSG/GSH and DHAA/AA ratios, respectively) were affected by diabetes or fidarestat treatment. SOD, GSSR Red, and GSHTrans activities were decreased in the retina of diabetic rats compared with controls. Catalase activity remained unchanged. GSHPx tended to decrease, but the difference between diabetic and control groups did not achieve statistical significance. Low dose fidarestat did not affect SOD activity, whereas high dose supranormalized it. Catalase activities were indistinguishable among diabetic rats treated with or without low or high doses of the ARI. GSSGRed and GSHPx activities tended to increase with the low-dose fidarestat treatment but the difference with the untreated diabetic group did not achieve statistical significance. GSSGRed activity was essentially normalized by the high dose of fidarestat, whereas GSHPx responded similarly to the low-dose and high-dose fidarestat treatments. In contrast, GSHTrans was normalized by low dose, but not high dose, of fidarestat. It tended to increase with the high-dose fidarestat treatment, but the difference with the untreated diabetic group did not achieve statistical significance.

Glucose, sorbitol, and fructose concentrations were 24.5-fold, 11.7-fold, and 6.7-fold higher in BREC cultured in 30 mmol/l glucose than in those cultured in 5 mmol/l glucose (Table 4). High glucose-induced increase in BREC glucose concentrations was not affected by fidarestat treatment, whereas sorbitol and fructose concentrations were reduced 2.3-fold and 2-fold, respectively.

Intracellular ROS abundance was 2.4-fold higher in BRECs cultured in 30 mmol/l glucose compared with those cultured in 5 mmol/l glucose (Fig. 4A). Hyperglycemia-induced increase in ROS generation was essentially corrected by 1 μmol/l fidarestat.

Intracellular ROS abundance was 3.2-fold, 5.4-fold, and 16.7-fold higher in BREC cultured in 5 mmol/l glucose supplemented with BSO, DEM, or primaquine, respectively, than in 5 mmol/l glucose without pro-oxidants (Fig. 4B). 1 μmol/l fidarestat did not affect oxidative stress induced by any of the three pro-oxidants.

The present study provides the first experimental evidence of the major contribution of increased AR activity to enhanced oxidative stress and VEGF protein expression in the retina in short-term diabetes. The relation between increased AR activity and oxidative stress deserves the most thorough consideration because, in the last several years, the continuing debate about a “primary mechanism” of diabetic complications has centered on oxidative stress and its relation with other hyperglycemia-initiated factors. Several groups (32,33) have suggested that the key physiological role of AR is the detoxification of lipid peroxidation products, and thus, AR inhibition in the diabetic conditions could be detrimental rather than beneficial. However, this view is not supported by the studies in ARI-treated nondiabetic animals that did not reveal any appearance of oxidative stress or diabetes-like complications (21,22), as observed with modulation of antioxidant deficiency (34) or pro-oxidants (3,4). Furthermore, it is well known that the effects of ARIs and antioxidants in diabetic animal models are unidirectional, i.e., both classes of agents delay, prevent, or correct the development of diabetic complications including DR (1–4,7,9–11,13,16,19,21–23). One should keep in mind that although AR has been found to metabolize 4-hydroxynonenal, in in situ perfused organs (33,35), the aforementioned studies employed additions of relatively high concentrations of 4-hydroxynonenal, 100–200 μmol/l. Until now, no evidence of elevated concentrations of the products of 4-hydroxynonenal metabolism by AR, i.e., glutathione-4-hydroxynonenal conjugate, 1,4-dihydroxy-2-nonene, and 4-hydroxy-2-nonenoic acid, in tissue-sites for diabetic complications, and no evidence of the presence of those products in vivo, has been generated. Recently, one group hypothesized that three pathways leading to diabetic complications, i.e., increased sorbitol pathway activity, nonenzymatic glycation/glycoxidation, and protein kinase C (PKC) activation, originate from oxidative stress and, in particular, production of superoxide anion radicals in mitochondria (20). However, this concept, at least the part related to the sorbitol pathway, is not supported by experimental studies demonstrating the absence of any suppression of tissue sorbitol pathway activity by antioxidants, including those neutralizing superoxide anion radicals (DL-α-lipoic acid and taurine) in diabetic animal models (28,36). Based on evidence of contribution of increased AR activity (4,23,37,38), the Maillard reaction (39), the interaction of advanced glycation end products (AGE) with their receptors (40), and recently, PKC activation (41) to hyperglycemia-induced oxidative injury, it would be more logical to assume that the common component for the pathways leading to diabetic complications, i.e., oxidative stress, is localized downstream from primary hyperglycemia-initiated mechanism(s). The importance of extramitochondrial versus intramitochondrial mechanisms of high glucose-induced ROS generation remains to be clarified. A recent study indicates the predominantly extramitochondrial origin (xanthine oxidase) of oxidative stress in patients with type 1 diabetes (42). At least, two extramitochondrial mechanisms of free radical production, i.e., xanthine oxidase and semicarbazide-sensitive amine oxidase, are of key importance in vasa nervorum (4). Our findings suggest that the extramitochondrial enzyme AR has an important role in diabetes-associated retinal oxidative stress.

The demonstration of the importance of AR in diabetes-induced retinal oxidative stress complement previous findings in the lens and peripheral nerve obtained with a number of ARIs (4,19,23,37,43). These reports complemented by studies in AR-overexpressing (38) mice implicate AR in increased lipid peroxidation (19,23,37,38), depletion of major nonenzymatic antioxidants, i.e., GSH (4,19,23,37,38,43), ascorbate (19,23), and taurine (19), and downregulation of superoxide dismutase activity (19) in the two tissue sites for diabetic complications. Fidarestat appeared remarkably effective in arresting diabetes-induced retinal lipid peroxidation. Preservation of normal MDA plus 4-HA concentration was achieved with 2 mg/kg per day, i.e., the dose that was at least 50-fold lower than the effective doses of direct antioxidants previously tested in a rat model of diabetes (17,28,29,35). Note, that unlike in lens or peripheral nerve, the arrest of retinal lipid peroxidation with AR inhibition was not due to a “sparing/regeneration” of the key nonenzymatic antioxidants, GSH, and AA. In contrast to other tissues of rats with short-term diabetes (4,19,23,28,37,43), retinal GSH and AA concentrations are not depleted, and the glutathione and ascorbate redox states are not compromised. It is quite plausible that the remarkable efficacy of the low dose of fidarestat on retinal lipid peroxidation is due to an upregulation of GSHTrans, which neuralizes 4-HA by their conjugation with GSH, combined with a faster turnover of the glutathione redox cycle due to increase in both GSSGRed and GSHPx activities. Note that retinal SOD activity was supranormalized by the complete inhibition of the sorbitol pathway hyperactivity. These results are consistent with 1) the attenuation of diabetes-induced downregulation of SOD activity in the peripheral nerve of diabetic rats by the ARI sorbinil and 2) the upregulation of NADH oxidase, the superoxide-generating enzyme, by an excessive sorbitol accumulation induced by a sorbitol dehydrogenase inhibitor in the lens of diabetic rats, as described in our previous studies (19,30).

Although the presence of AR in endothelial cells remained a subject of debate for a long time, a rigorous study (44) has generated a positive answer. Our findings indicate a major contribution of AR to high glucose-induced oxidative stress in retinal endothelial cells. The arrest of hyperglycemia-induced ROS generation by fidarestat is not due to its direct antioxidant properties, which are not related to an ability to inhibit the sorbitol pathway of glucose metabolism. Fidarestat did not affect oxidative stress induced by three unrelated pro-oxidants with different mechanisms of action that included hydrogen peroxide generation and initiation of hydrogen peroxide-stimulated lipid peroxidation (primaquine) and GSH depletion due to inhibition of glutathione biosynthesis (BSO) or monovalent thiol oxidation (DEM). Therefore, fidarestat does not interfere with increased production or compromised neutralization of ROS caused by mechanisms other than hyperglycemia. The present study, together with the findings obtained with another structurally diverse ARI zopolrestat (24), support an important role for AR activity in diabetes-induced ROS generation in vascular endothelium. Furthermore, the major contribution of AR to hyperglycemia-induced endothelial superoxide production (24) strongly implies that AR is also involved in generation of the highly reactive hydroxyl radicals produced from superoxide and hydrogen peroxide in the Fenton and Haber-Weiss reactions.

Increasing evidence indicates that such diabetes-associated phenomena as increased retinal AR activity, oxidative stress, VEGF overexpression, and increased vascular permeability are interrelated. The present study, together with the previous observations by our group and others, suggest that AR triggers the whole cascade by causing oxidative stress, which, in turn, leads to overexpression of VEGF (17,45) responsible for increased vascular permeability (13,18). Indeed, all three components of this cascade, i.e., enhanced oxidative stress, VEGF protein overexpression, and increased vascular permeability, have been found preventable by an ARI treatment (2,11,13, and in the present study). Several studies (12,17,45), including one from our group (17), also relate VEGF upregulation and increased vascular permeability to oxidative stress. Our findings revealed that retinal VEGF protein, but not VEGF mRNA expression, is increased in early diabetes. The absence of diabetes-induced retinal VEGF mRNA overexpression in the present study using the commercial kit with a sensitivity of a Northern blotting as well as RT-PCR was also found in another study from our group using a ribonuclease protection assay (17). The present findings, together with our previous observations with two other classes of agents, i.e., the antioxidants DL-α-lipoic acid and taurine (17) and the poly-(ADP-ribose) synthetase inhibitors, 3-aminobenzamide and 1,5-isoquinolinediol (46), suggest that early diabetes-induced upregulation of retinal VEGF protein occurs at a post-transcriptional level. Based on current knowledge, one can speculate that three factors, i.e., increased AR activity, oxidative stress, and poly(ADP-ribosyl)ation, interfere with PKC-dependent phosphorylation of the initiation factor (eIF) 4E, known to play an important role in the translational regulation of VEGF (47). The role for AR (48), oxidative stress (6), and recently, poly(ADP-ribosyl)ation (49) in PKC activation has already been demonstrated. Both low and high doses of fidarestat decreased diabetes-induced retinal VEGF overexpression, which supports the findings of the favorable effects of long-term treatment with 2 mg/kg of fidarestat on retinal pathological changes and increased vascular permeability in streptozotocin-diabetic rats (13). It remains to be established whether this dose of fidarestat will also be effective in preventing VEGF-unrelated phenomena in the pathogenesis of DR, i.e., pericyte loss and formation of acellular capillaries, which ultimately leads to retinal capillary closure. The initial event, pericyte dropout, is likely to occur due to high glucose-induced apoptosis, which is known to be preventable by fidarestat (50).

In conclusion, increased AR activity has a key role in diabetes-induced oxidative stress in the retina and high glucose-induced oxidative stress in retinal endothelial cells. AR is also responsible for retinal VEGF protein expression in early diabetes. Both oxidative stress and VEGF overexpression are prevented by the potent and well-tolerated ARI fidarestat. The findings support the “AR concept” of DR and provide the rationale for further evaluation of fidarestat to prevent the onset and progression of diabetes-induced retinal vascular changes. A recent 52-week multicenter placebo-controlled double-blind parallel group trial (27) demonstrated clinical efficacy of the very low dose of fidarestat (1 mg/day) in improving nerve conduction deficits and symptomatic indexes of diabetic neuropathy.

This study was supported by a grant from Sanwa Kagaku Kenkyusho (I.G.O.) and the Juvenile Diabetes Research Foundation Center for the Study of Complications of Diabetes Grant 4-200-421 (I.G.O., A.K.K., and M.J.S.).