Nitric Oxide Prevents Aldose Reductase Activation and Sorbitol Accumulation During Diabetes

Sprague Dawley rats (200–300 g) were housed in accordance with institutional guidelines. Streptozotocin (STZ), S-nitrosoglutathione (GSNO), 3-morpholinosydnonimine (SIN-1), the protease inhibitor cocktail (4-[2-aminoethyl]benzenesulfonyl fluoride hydrochloride [AEBSF], leupeptin, bestatin, E-64, and pepstatin-A) were obtained from Sigma Chemical. The NO donors, (± )-S-nitroso-N-acetylpenicillamine (SNAP), diethylamine NONOate (NONOate), and GSNO mono-ethyl ester (GSNO-ester), N^G-nitro-l-arginine methyl ester (l-NAME), and l-arginine were purchased from Calbiochem. Sorbinil and tolrestat were obtained from Pfizer and American Home Products, respectively. Deriva-Sil was purchased from Regis Technologies, and transdermal nitroglycerine patches were obtained from Hercon Labs.

Diabetes was induced by a single intraperitoneal injection of STZ (65 mg/kg). Blood glucose was monitored daily, and only the rats with blood glucose >400 mg/dl on the 3rd day after the STZ injection were used for further experiments. The rats were killed by injecting pentobarbital 6 days after the STZ injection (i.p.). Blood was collected by cardiac puncture in heparinized tubes and washed three times with PBS, pH 7.0. All animal protocols were approved by the institutional animal care committee. Erythrocytes (0.2 ml) from normal and hyperglycemic rats were incubated with NO donor or AR inhibitors in 1.0 ml of PBS containing 1 mmol/l EDTA and 5.5 mmol/l glucose. The samples were incubated at 25°C in a shaking water bath for 2 h. Wherever indicated, glucose was added to the medium to a final concentration of 40 mmol/l, and incubations were continued for another 4 h at 37°C in a shaking water bath. At the end of the incubation period, the samples were centrifuged at 3,000g for 5 min, and the pellet was washed three times with PBS. The cells were lysed by adding 0.4 ml water, and the proteins were precipitated by adding 0.4 ml each of 0.15 mol/l Ba(OH)₂ and 0.15 mol/l ZnSO₄. The precipitate was removed by centrifugation at 10,000g for 5 min. The supernatant was ultrafiltered through a YM-10 Microcon (Millipore) by centrifugation at 10,000g. An aliquot of the filtrate was lyophilized in a SpeedVac to complete dryness and stored in a vacuum dessicator containing calcium chloride for at least 24 h.

To measure sorbitol content, the dried samples were derivatized by adding 0.1 ml of the Deriva-Sil solvent under anhydrous conditions. The derivatized mixture (1 μl) was injected into a Varian Gas Chromatograph (GC system 3,000) equipped with hydrogen flame ionization detector and Chromopack capillary column packed with CP Sil 24CB. The column temperature was set at 140°C, and the temperature gradient was set from 140°C to 170°C increasing 4°C per min and from 170°C to 250°C and then 50°C per min. The column was maintained at this temperature for an additional 3 min. The injection port was maintained at 250°C, and the detector temperature was set at 300°C. The amount of sorbitol present in the sample was calculated using reagent sorbitol measured by gas chromatography under similar conditions. The sorbitol peak was confirmed by mass spectrometry.

The l-arginine analog l-NAME was used to inhibit NO synthesis (22) in a dose range shown to be effective in rats (50 mg · kg^–1 · day^-1, i.p.) (23). The synthesis of NO was enhanced treating rats with l-arginine (200 mg · kg^-1 · day^-1, i.p.), which is a substrate of NO synthase (22), and has been shown to increase NO generation in vivo (24). Additionally, nitroglycerine patches, which release 200 ng of NO per min (25), were applied on the preshaved dorsal neck region of the rats. The nitroglycerine patches were replaced everyday. After 10 days of treatment, the rats were killed and several tissues (lung, heart, liver, kidney, testis, sciatic nerve, skeletal muscle, and brain) were harvested and their erythrocytes collected. The tissues were homogenized in 1.0 ml ice-cold PBS containing 20 μl protease inhibitor cocktail. The AR activity and sorbitol content of the homogenates were measured. In a separate series of experiments, erythrocytes from normal or diabetic rats were incubated with 40 mmol/l glucose in PBS at 37°C for 6 h before measuring AR activity and sorbitol. Data are presented as the means ± SE. Data were analyzed with unpaired Student’s t test using Excel spreadsheets.

Erythrocytes were isolated from normal rats and were incubated with 5.5 or 40 mmol/l glucose. After 4 h, the sorbitol levels in these cells were 0.024 ± 0.004 and 0.18 ± 0.013 nmol/mg protein, respectively. When the erythrocytes isolated from diabetic rats were incubated with 40 mmol/l glucose, their sorbitol content was found to be 0.41 ± 0.04 nmol/mg protein, which was twofold higher than the sorbitol content of the erythrocytes from similarly treated nondiabetic rats. These results indicate that even at the same concentration of extracellular glucose, the diabetic cells accumulate more sorbitol than nondiabetic cells, which may be in part due to the upregulation of AR during diabetes. To examine whether sorbitol accumulation in these cells was due to AR, the erythrocytes were incubated with glucose in the presence of the AR inhibitors sorbinil or tolrestat (1 mmol/l each). As shown in Table 1, both these inhibitors prevented, nearly completely, the accumulation of sorbitol, indicating that erythrocyte sorbitol is derived entirely from the AR-catalyzed reduction of glucose and that the sorbitol content is a valid surrogate measure of cumulative AR catalysis in these cells. Interestingly, the formation of sorbitol in diabetic erythrocytes was less affected by sorbinil and tolrestat (77–84% inhibition) compared with the cells isolated from nondiabetic animals, in which 97–98% inhibition of sorbitol synthesis was observed with these inhibitors (Fig. 1). These data are consistent with our work showing that diabetes diminishes the inhibitor sensitivity of AR (6–8).

To determine whether NO could inhibit AR activity and prevent sorbitol synthesis in erythrocytes, we used NO donors, which provide more uniform steady-state levels of NO than NO gas itself (26). Erythrocytes isolated from both normal and diabetic rats were incubated with several NO donors to rule out donor-specific effects and to compare relative efficacies. All the donors examined (at 1 mmol/l each) prevented sorbitol accumulation to nearly the same extent (70–86%) (Table 1), although SNAP was found to be the most effective compound of this series. Because all the NO donors inhibited sorbitol formation, it appears likely that this inhibition is related to NO and is not unique to the chemical nature of the donor molecule. The NO donors also inhibited sorbitol accumulation in diabetic erythrocytes (Fig. 1); however, as compared with normal cells, the extent of inhibition of sorbitol accumulation in the diabetic erythrocytes was much less (28–58% vs. 70–86% inhibition). The difference between the two groups was statistically significant (P < 0.01). It is noteworthy that the sorbitol content of erythrocytes from diabetic rats incubated with 40 mmol/l glucose and sorbinil or tolrestat was 11- or 33-fold higher than that of the erythrocytes from normal rats incubated with the AR inhibitors under identical conditions, suggesting that even a small fraction of AR remaining uninhibited during diabetes could result in profound sorbitol accumulation. Together, these observations suggest that NO inhibits AR activity and prevents sorbitol accumulation in normal as well as diabetic erythrocytes. As with sorbinil and tolrestat, the higher resistance of AR in diabetic cells to inhibition by NO donors may be reflective of diabetic changes in the AR protein.

To investigate the in vivo effects of NO, nondiabetic and diabetic rats were treated either with the NO synthase (NOS) substrate, l-arginine, or the NOS inhibitor, l-NAME (27,28). In addition, we also examined the consequences of delivering NO via a transdermal nitroglycerine patch, which releases NO slowly into the circulation (25). Eight groups of rats were examined: groups I–IV were euglycemic and groups V–VIII were made diabetic by a single injection of STZ. Group I received no treatment, while group II was treated with l-NAME, group III with the nitroglycerine patch, and group IV with l-arginine. Group V received no treatment. Groups VI, VII, and VIII were treated with l-NAME, nitroglycerine patch, and l-arginine, respectively. Ten days after diabetes and/or the indicated treatments, the rats were killed. Their blood was collected and their tissues harvested. As shown in Table 2, measurable levels of AR activity and sorbitol were present in several tissues and organs. The highest AR activity was observed in the sciatic nerve followed by lung, brain, muscle, kidney, and heart (in descending order). The pattern of relative tissue distribution of AR is similar to that reported before (1). Detectable amounts of sorbitol were recovered from the tissues that expressed AR, and in general the sorbitol content reflected the tissue AR activity. Among the tissues examined, the highest sorbitol content was detected in sciatic nerve, followed by lungs, brain, and skeletal muscle.

Treatment of nondiabetic rats with l-NAME led to an increase in AR activity and sorbitol accumulation (Table 2). The highest increase in sorbitol activity was observed in sciatic nerve (threefold), followed by kidney (1.7-fold), brain (1.6-fold), skeletal muscle (1.6-fold), and heart (1.4-fold), suggesting that endogenous generation of NO in these tissues maintains AR in the repressed state, which is relieved upon inhibiting NO synthesis. The repressing effect of NO was also observed on AR activity, and treatment with l-NAME led to a sixfold increase in the enzyme activity in the lungs, a 5.2-fold increase in the heart, and a 1.3- to 2.0-fold increase in other tissues. However, when the rats were treated with either the nitroglycerin patch (group III) or l-arginine (group IV), no significant inhibition of AR activity or sorbitol accumulation was observed, indicating that the NO regulation of the polyol pathway in nondiabetic rats is maximal and cannot be enhanced further by exogenous delivery of NO.

As expected, tissues removed from diabetic rats (group V) displayed higher sorbitol content and AR activity. As with untreated nondiabetic controls, the highest level of sorbitol in the diabetic rats was found in sciatic nerve (2,499 ± 253 nmol/mg protein) followed by lung, heart, brain, muscle, and kidney in the range of 17–59 nmol/mg protein. Trace levels of sorbitol were observed in liver and testis (Table 3). Treatment of diabetic rats with l-NAME (group VI) resulted in a 2- to 3.8-fold increase in the sorbitol content of all tissues examined compared with untreated diabetic rats (group V): 382, 297, 286, and 275% increase in brain, heart, lung, and testes, respectively. The increase in sorbitol accumulation was accompanied by a corresponding increase in AR activity.

Treatment of diabetic rats with agents that enhance NO levels (nitroglycerine patch or l-arginine) uniformly led to an inhibition of AR activity and sorbitol accumulation. As shown in Table 3, in the tissues isolated from l-arginine-treated rats (group VIII), the decrease in sorbitol content was maximum in heart (94%), followed by brain (90%), testis (89%), erythrocytes (89%), kidney (84%), and sciatic nerve (80%) compared with untreated diabetic rats (group V). Treatment with nitroglycerine patch (group VII) also had an inhibitory effect on sorbitol content and AR activity. The decrease in sorbitol content in tissues from group VII rats compared with group V rats was most prominent in heart (68%) followed by sciatic nerve (67%) and erythrocytes (62%).

To determine whether the inhibition of AR by NO is reversible, erythrocytes isolated from group I–VIII rats were incubated ex vivo with 40 mmol/l glucose, and after 6 h the AR activity and the sorbitol content were determined. In case of erythrocytes from nondiabetic rats (groups I–IV), the basal levels of AR activity and sorbitol were below the detection limit of our assays. However, after 6 h of incubation with 40 mmol/l glucose, both the AR activity and the sorbitol content increased dramatically but were comparable among groups I–IV (data not shown). In case of diabetic rats (groups V–VIII), the AR activity and sorbitol content in erythrocytes isolated from l-NAME-treated animals were more than twofold higher and those from l-arginine and nitroglycerine patch-treated animals were inhibited >80 and 60%, respectively as compared with untreated diabetic controls (group V). The differences between the groups were significantly diminished after 6 h of incubation (Fig. 2), indicating that the stimulatory or inhibitory effects of NO on AR are reversible and that once NO-generation is normalized, AR activity and sorbitol synthesis revert back to their basal levels.

Increased activity of AR and the polyol pathway has been suggested to be the underlying biochemical cause of secondary diabetes complications (1–3). The role of AR in the pathogenesis of diabetes complications is supported by extensive evidence showing that inhibition of the enzyme prevents and/or delays the development of diabetic cataracts, neuropathy, or nephropathy (1–3). However, in cell culture studies, hyperglycemia induces progressive resistance of the enzyme to inhibition (29), and AR isolated from diabetic humans or animals is more resistant to pharmacological inhibition than that from normal, euglycemic tissues (6–8). Moreover, in clinical trials, AR inhibitors display high nonspecific toxicity and low efficacy (4,5), suggesting that our understanding of the enzyme and its regulation during diabetes is incomplete and that additional investigations are required to rationally redesign anti-AR interventions.

Our in vitro studies show that due to the presence of an active site cysteine (cys-298), AR is very sensitive to oxidants (12,13, 15). Several thiol-modifying reagents alter AR catalysis and substrate and inhibitor binding, suggesting that in vivo the enzyme may be under redox regulation. Although thiol-disulfide regulation may represent one such mechanism (30), recent evidence suggests that the multipotent regulator, NO, which controls several diverse functions, could initiate and participate in a number of redox reactions, including nitrosation, nitration, and S-thiolation (31–33). Indeed, our studies show that exposure to NO donors results in rapid and selective modification of AR at cys-298 (17–19). These observations suggest that NO may be a physiological regulator of AR activity in vivo and that changes in NO availability during diabetes could alter AR catalysis and consequently the polyol pathway activity. In agreement with this, the present studies show that NO donors inhibit AR activity and sorbitol accumulation in erythrocytes. These observations are consistent with the view that NO is a physiological regulator of AR. In the in vitro studies, NO donors induce a variety of changes in AR catalysis, and substrate and inhibitor binding (17–19). The NO-releasing agents, such as SNAP or NONOate, cause nitrosation of AR protein, and the nitrosated enzyme is more active and more resistant to inhibitors than the native form (19). In contrast, nitrosothiols such as GSNO and SNAP inhibit AR by S-thiolating cys-298 (18). Hence, depending on the nature of the NO donor, AR could be either nitrosated or thiolated and activated or inhibited. However, in glutathione-proficient cells, the paracrine effects of NO are likely to be mediated by nitrosothiols (32,33). Hence, the predominant form of AR modification induced in these cells is likely to be glutathiolation, which will lead to enzyme inactivation. Indeed, all of the NO donors examined in the present study inhibited AR (Table 1), suggesting that under euglycemic conditions, NO inhibits rather than activates AR catalysis in vivo.

The inhibitory effect of NO on AR catalysis and the polyol pathway activity was also evident in diabetes. The erythrocytes of diabetic animals when incubated with glucose ex vivo or tissues isolated from diabetic rats accumulated significantly less sorbitol when treated with NO donors. However, the extent of inhibition of sorbitol formation in the diabetic erythrocytes was much less than that observed in normal, euglycemic cells, indicating that diabetes increases the inhibitor resistance of AR. Although the specific modifications of the AR protein remain unidentified, we speculate that under glutathione-replete conditions NO S-thiolates AR. However, during hyperglycemia NO availability is diminished, and because of a concurrent loss of glutathione, the residual NO (instead of S-thiolating) causes nitrosation of AR. Although speculative, this view accounts for the inhibitor resistance of AR and for the increased sorbitol accumulation in diabetic tissues.

In addition to post-translational changes in AR protein, NO could also affect the transcription of the AR gene. It has been shown that stimulation of vascular smooth muscle cells in culture with NO donors results in an increase in AR mRNA and activity and that inhibiting NO synthesis prevents the increase in AR in lipopolysaccharide or interferon-γ- stimulated macrophages (34). Although the in vivo significance of these results remains to be established, they suggest that the expression of AR should increase with an increase in NO generation or availability. However, our observation that NO donors do not increase AR activity (Tables 2 and 3) suggests that the transcriptional activation of the AR gene in vivo by NO is marginal or that this effect is overcome by the post-translational inhibition of AR activity. Nevertheless, the regulation of the AR gene by other stimuli could not be ruled out. The AR gene is stimulated by the tonicity of high glucose (35,36), as well as growth factors, mitogens (37), cytokines (38), and oxidants (37,39). Many of these stimuli are affected by diabetes and could change AR expression during diabetes. However, our observation that NO donors acutely inhibit AR activity in erythrocytes provides strong evidence supporting a post-translational mechanism. And collectively, our results are consistent with the view that NO represses, rather than activates, AR and that AR is stimulated by a lack of NO, resulting in an increase in the polyol pathway activity. During diabetes NO availability is decreased (even though the NO synthesis per se may be enhanced), and therefore the net effect will be further de-repression of AR and stimulation of sorbitol synthesis.

The de-repression of AR due to loss of NO is supported by our data showing that treatment with l-NAME, which inhibits NO synthesis, caused a considerable increase in AR activity and sorbitol formation in nondiabetic as well as the diabetic tissues (Tables 2 and 3). These results suggest that a significant fraction of AR remains inhibited by NO, and if this inhibition is removed AR is activated. Fundamentally, these data demonstrate that AR activity, or more importantly the flux of glucose through the polyol pathway leading to NADPH depletion and sorbitol accumulation, is not a simple function of extracellular glucose concentration. Under euglycemic conditions NO represses AR by maintaining the enzyme in an inactive, S-thiolated state; however, during diabetes NO availability is reduced and the enzyme is liberated from this repression (40).

Previous studies have shown that under euglycemic conditions the polyol pathway activity represents <3% of the total glucose flux. However, during hyperglycemia, metabolism via AR accounts for >30% of the total glucose utilized (41). This increase could be due to a simple mass effect; at low glucose concentrations low levels of sorbitol are synthesized, whereas high glucose leads to higher polyol pathway activity. Any nonlinearity in the response could be accounted for by the low affinity of AR for glucose. Because AR has a high K_m glucose, the flux through the polyol pathway will be significant only at high glucose concentration (1). Although our results do not invalidate these considerations, they indicate that regulation by NO imparts another level of complexity to the relationship between glucose concentration and AR catalysis, such that even at the same concentration of the extracellular glucose, the level of AR catalysis as well as the flux of glucose through the polyol pathway could change depending on the availability of NO.

Our results with l-arginine and the nitroglycerine patch show that under euglycemic conditions exogenous delivery of NO does not inhibit basal AR activity or sorbitol accumulation. We suggest that in normal rats, the inhibition of AR by NO is maintained at a set level, which could not be exceeded by exogenous application of NO-donors. However, the observation that AR inhibition upon treatment of diabetic animals with l-arginine or nitroglycerine patches suggests that during diabetes there is partial inhibition of AR by NO, which could be significantly enhanced by delivering NO or increasing the endogenous synthesis of NO. Nonetheless, the effects of NO on other processes, such as the regulation of AR expression (34), sorbitol dehydrogenase, sorbitol efflux, and glucose availability and metabolism, cannot be completely ruled out and are currently under investigation in our laboratory.

Regardless of the mechanism, our results suggest a new approach for specifically inhibiting AR during diabetes. As evident from Table 3, treatment with either the nitroglycerine patch or l-arginine inhibited AR and prevented sorbitol accumulation in diabetic animals, however, no significant effect of these interventions was observed in nondiabetic animals (Table 2), demonstrating that the inhibitory effects of NO appear to be specific to hyperglycemia and NO-donors do not perturb AR under euglycemia. Moreover, in addition to inhibiting the polyol pathway, enhancing NO during diabetes may provide additional benefits, such as restoration of endothelial function and blood pressure (42–44). Therefore, NO therapy may be useful in simultaneously preventing and treating several aspects of vascular pathology associated with long-term diabetes.

In summary, the results of the present study demonstrate for the first time that endogenous NO maintains AR in a partially inhibited state in vivo and prevents AR-dependent accumulation of sorbitol in cells exposed to high glucose. These observations provide a critical link between our previous work showing on one hand that the kinetic and the ligand binding properties of AR are highly sensitive to thiol oxidation (9,10, 12,13,15) and NO (17–19) and on the other hand that these properties of the enzyme are dramatically affected by diabetes (6–8). On the basis of the current investigations, we propose that the changes in AR associated with diabetes are due to perturbations in the regulation of this enzyme by NO. During euglycemia, AR is partially inhibited by nitrosothiols, whereas in diabetes, AR is activated, presumably by nitrosation and/or the loss of NO-mediated repression. This increase in AR activity could be prevented by exogenous delivery of NO or by enhancing NO synthesis. Thus, treatment with nitroglycerine patch or l-arginine may be useful therapeutic approaches for controlling or preventing diabetes complications.

This work was supported in part by National Institutes of Health Grants DK36118, EY01677, and HL55477.