Deletion of Aldose Reductase from Mice Inhibits Diabetes-Induced Retinal Capillary Degeneration and Superoxide Generation

Purpose Pharmacologic inhibition of aldose reductase (AR) previously has been studied with respect to diabetic retinopathy with mixed results. Since drugs can have off-target effects, we studied the effects of AR deletion on the development and molecular abnormalities that contribute to diabetic retinopathy. Since recent data suggests an important role for leukocytes in the development of the retinopathy, we determined also if AR in leukocytes contributes to leukocyte-mediated death of retinal endothelial cells in diabetes. Methods Wild-type (WT; C57BL/6J) and AR deficient (AR−/−) mice were made diabetic with streptozotocin. Mice were sacrificed at 2 and 10 months of diabetes to evaluate retinal vascular histopathology, to quantify retinal superoxide production and biochemical and physiological abnormalities in the retina, and to assess the number of retinal endothelial cells killed by blood leukocytes in a co-culture system. Results Diabetes in WT mice developed the expected degeneration of retinal capillaries, and increased generation of superoxide by the retina. Leukocytes from diabetic WT mice also killed more retinal endothelial cells than did leukocytes from nondiabetic animals (p<0.0001). Deletion of AR largely (P<0.05) inhibited the diabetes-induced degeneration of retinal capillaries, as well as the increase in superoxide production by retina. AR-deficiency significantly inhibited the diabetes-induced increase in expression of inducible nitric oxide synthase (iNOS) in retina, but had no significant effect on expression of intercellular adhesion molecule-1 (ICAM-1), phosphorylated p38 MAPK, or killing of retinal endothelial cells by leukocytes. Conclusions AR contributes to the degeneration of retinal capillaries in diabetic mice. Deletion of the enzyme inhibits the diabetes-induced increase in expression of iNOS and of superoxide production, but does not correct a variety of other pro-inflammatory abnormalities associated with the development of diabetic retinopathy.


Introduction
Diabetic retinopathy is a common complication of diabetes, and is the principal cause of blindness in working-aged adults. Hyperglycemia clearly initiates the disease process, but which of the sequelae of hyperglycemia are causal in development of the retinopathy is not clear.
Activation of aldose reductase (AR) by elevated glucose was one of the first biochemical mechanisms postulated to explain the pathogenesis of diabetic complications [1]. Particularly with regard to cataractogenesis, evidence suggested that pathology occurred as a result of osmotic consequences of AR-mediated reduction of glucose to its polyol, sorbitol [2]. Since then, AR has been found also to regulate a variety of additional abnormalities (including oxidative stress and inflammation) that have been implicated in the pathogenesis of various complications of diabetes [3][4][5] and other diseases [6][7][8][9][10][11][12][13][14][15][16].
Efforts to inhibit this pathway to inhibit diabetic retinopathy in patients have relied heavily on pharmacologic inhibitors of the enzyme [17]. Nevertheless, results of these studies with regard to the retinopathy have been inconsistent and controversial. Some studies of diabetic or galactosemic rodents or dogs showed beneficial effects of AR inhibitors on lesions of the retinpathy [18][19][20][21][22][23], whereas no significant effects were detected in clinical studies in patients and other studies of diabetic and galactosemic dogs and rodents [24][25][26][27]. Possibilities that have been discussed to explain the lack of agreement among studies include differences in the degree of inhibition of AR, the presence of isoforms of the enzyme that respond differently to therapies, genetic differences among individuals, and possible off-target effects of the different AR inhibitors [17].
Generation of animals that are totally deficient in AR is one way to overcome the shortcomings associated with the use of pharmacologic inhibitors of AR. Moreover, AR 2/2 animals allow the opportunity to investigate the molecular pathways by which AR acts in hyperglycemia. Thus, the availability of AR 2/2 animals makes it worthwhile to revisit this topic. In the present study, we investigated the effect of AR deficiency on diabetesinduced degeneration of retinal capillaries in early diabetic retinopathy, and on diabetes-induced pro-inflammatory and prooxidant changes in the retina.

Experimental Animals
AR 2/2 mice were prepared [28] and backcrossed with C57BL/ 6 mice for seven generations. Male C57Bl/6J mice and AR 2/2 mice were randomly assigned to become diabetic or remain as nondiabetic group. Diabetes was induced by 5 sequential daily intraperitoneal injections of a freshly prepared solution of streptozotocin in citrate buffer (pH 4.5) at 45 mg/kg of body weight. Insulin was given as needed to prevent weight loss without preventing hyperglycemia and glucosuria (0-0.2 units of NPH insulin subcutaneously, 0-3 times per week). Glycohemoglobin (GHb) was measured by Bio-Rad Total Glycated Hemoglobin Assay (Bio-Rad Laboratories, Inc, Hercules, CA, USA) every 2-3 mos and just before animals were sacrificed. Food consumption and body weight were measured weekly. Animals were studied for durations of 10 months or 2 months of diabetes in order to investigate effects of the therapy on retinal histopathology, or molecular and physiologic changes, respectively. These durations were chosen because 2 mos diabetes has been found to result in numerous metabolic alterations which precede (and likely contribute to) the later appearance of vascular histopathology, and 10 mos duration of diabetes has been shown to have developed robust vascular histopathology characteristic of the early stages of the retinopathy.

Ethics Statement
Treatment of animals conformed to the ARVO Resolution on Treatment of Animals in Research, as well as to institutional guidelines (Case Western Reserve University IACUC # 2010-0156).
Retinal histopathology. After 10 months of diabetes, one retina was fixed in 10% buffered formalin, washed in running water overnight, and digested in elastase (Calbiochem, Cat#324682) buffer (40 unit/ml elastase in 100 mM sodium phosphate buffer (containing 150 mM sodium chloride and 5 mM ethylenediamene tetraacetic acid (EDTA; pH 6.5) in an agitating water bath at 37u for 1K h [29]. Then, retinas were washed overnight in 100 mM Tris-HCI (pH 8.5) at room temperature, and cleaning was finished the next day. When totally cleaned of neural cells, the isolated vasculature was laid out on a glass microscope slide, dried overnight, stained with hematoxylin and periodic acid-Schiff, dehydrated and coverslipped. Degenerate (acellular) capillaries were quantitated in 6-7 field areas corresponding to the mid-retina in a masked manner. Acellular capillaries were identified as capillary-sized vessel tubes having no nuclei anywhere along their length, and were reported per square millimeter of retinal area.

Superoxide Measurement
At 2 months of diabetes, fresh retinas from animals were analyzed for superoxide production as previously described [30,31]. Briefly, whole retinas were placed in 0.2 ml Krebs/Hepes buffer and allowed to equilibrate in the dark for 30 mins at 37uC under 5% CO 2 . To each tube, 0.5 mM lucigenin (Sigma Chemical Company, St. Louis, MO) was added, and photon emission was detected over the following 10 min using a luminometer. Retinal protein was quantified per samples (Bio-Rad Laboratories, Inc, Hercules, CA, USA) and luminescence was expressed per milligram protein.

Reduced glutathione (GSH)
At 2 months of diabetes, tissues of interest were homogenized in 0.1 M phosphate buffer (pH 8.0) containing 5 mM EDTA. After centrifugation, a quantity of the resulting supernatant was removed for storage at 280uC for subsequent protein determination by the bicinchoninic acid protein assay method (Pierce Biotechnology, Inc.). GSH determinations were carried out on the freshly prepared tissue extract following the fluorometric method described previously [32]. Fluorometric measurements were carried out using a Biotek Synergy4 microplate reader and data fitted by linear regression to a standard curve generated using purified GSH as a quantitation standard. In all cases, analytes were measured in triplicate.

Co-culture
A mouse retinal endothelial cell line (mREC; generated from Immortomice [33]) was grown in DMEM containing 10% FBS and 5.5 or 25 mM glucose. Cells were cultured at 37uC in 5% CO 2 and 95% air, and the media was changed every other day for 5 days. When mRECs were 80% confluent (app 500,000 cells), leukocytes (100,000; purified from blood with RBC lysis buffer) from male C57Bl/6J mice and AR 2/2 mice groups were added to the mREC and incubated for 24 hrs. After incubation, the leukocytes were carefully removed by gentle washing, and viability of remaining retinal endothelial cells was measured by trypan blue extrusion. Briefly, an aliquot of the endothelial cell suspension was diluted 1:1 (vol/vol) with 0.1% trypan blue, and the cells were counted with a hemocytometer. Cell death was reported as the percentage of blue-stained cells (dead cells) of the total number of cells. Approximately 200-400 cells were counted in each sample.

Statistical analyses
Data are expressed as mean 6 SD. Statistical analysis was performed using ANOVA, followed by Fischer's test. A value of p,0.05 was considered statistically significant.

Animals
Diabetic mice were hyperglycemic over the entire duration of the 10 month experiment, and failed to gain weight at normal rate. Serum glucose levels of the diabetic groups (356658 mg/dl and 344662 in in wildtype and AR 2/2 groups, respectively) were all significantly greater than nondiabetic values (132-137 mg/dl). The degree of hyperglycemia, as denoted by glycated hemoglobin, did not vary among diabetic mice (wild-type, 10.1%61.0 ; AR 2/2 , 10.2%61.1 respectively), and was significantly greater in diabetic groups as compared with the corresponding nondiabetic groups (wildtype, 3.060.2; AR 2/2 2.960.3). Body weight at 10 months of diabetes averaged 28 g61in the wildtype group and 27 g62 in AR 2/2 group, and these values were about half of those in nondiabetic mice of each group (51-60 g). Data from the 2 month experiment was similar (not shown). Deletion of AR had no apparent effect on glycemia or health of any animals. Crosssections of retina in nondiabetic animals (12 mos of age) indicated that deletion of AR had no significant effects on retinal thickness or retinal architecture, and no effect on vascular density in nondiabetic animals.

Inhibition of diabetes-induced retinal histopathology by AR deletion
Wild-type diabetics developed the expected increase in numbers of degenerate retinal capillaries by 10 mos of study (p,0.0001). In contrast, retinas from AR 2/2 diabetic animals exhibited significantly fewer diabetes-induced acellular capillaries (Figure 1). Nondiabetic AR 2/2 mice showed a normal retinal vasculature pattern. Diabetes of 10 mos duration did not cause a significant decrease in the number of cells in the retinal ganglion cell layer in these C57Bl/6 mice (not shown), so we could not assess any effect of the AR deletion on neurodegeneration.
In an effort to investigate the mechanism by which the AR deletion mediated the beneficial effect on degeneration of retinal capillaries in diabetes, we measured several parameters related especially to inflammation and nitrative (oxidative) stress, these being physiologic and molecular abnormalities that have been found in other studies to be associated with (and possibly causally related to) the development of the early stages of diabetic retinopathy.

Oxidative stress
Oxidative stress was evaluated in the present study by measuring superoxide generation by the freshly isolated retina. Diabetes significantly increased retinal superoxide production by almost two-fold (Figure 2a), whereas retinas from AR 2/2 mice were totally protected from diabetes-induced increase in superoxide production (p,0.0001). In contrast, retinal GSH tended to be subnormal in wild-type diabetics, but deletion of AR had little effect on this diabetes-induced reduction (Figure 2b). The accumulation of nitrotyrosine, a marker of nitrative stress, tended to be greater than normal in wild-type diabetic animals and tended to be inhibited in the absence of AR, but the results were not statistically significant (Figure 2c).

Pro-inflammatory proteins
Expression of iNOS and ICAM-1 in whole retina were significantly increased by diabetes compared to that of wildtype nondiabetic animals (both p,0.05), but deletion of AR inhibited the diabetes-induced enzyme induction only for iNOS ( Figure 3). The expression of phosphorylated p38 MAPK, and activity of poly-ADP ribose synthase (PARS) tended to be greater than normal in wild-type diabetic animals, but the increases were not statistical significant and deletion of AR had no apparent effect (not shown).

Endothelial cells killed by white blood cells
Previous studies have demonstrated an important role of leukocytes in development of early stages of diabetic retinopathy [34,35]. To investigate if white blood cells might contribute to the AR-mediated degeneration of retinal capillaries in diabetes, mouse retinal endothelial cells (mREC) were incubated with peripheral blood leukocytes from non-diabetic or diabetic mice. After 24 hrs of co-culture, the number of endothelial cells killed by leukocytes from wild-type diabetic mice was significantly greater than those killed by incubation with leukocytes from nondiabetic wild-type mice (1.7 fold increase; p,0.05; Figure 4). In contrast, leukocytes from diabetic animals lacking AR killed fewer endothelial cells than did leukocytes from wild-type diabetics, but this decrease did not achieve statistical significance.

Discussion
AR and the polyol pathway were among the first molecular mechanisms considered to explain how hyperglycemia initiates diabetic retinopathy and other complications of diabetes. At that time, AR was regarded as the enzyme responsible for the reduction of glucose into sorbitol, a molecule that penetrates poorly through cell membranes, thus leading to cellular damage via the ''osmotic hypothesis'' [2]. In an effort to further investigate the role of AR in these complications, a number of pharmacologic inhibitors of the enzyme were developed. Since then, research on this pathway has expanded beyond the initial hypothesis, but results using pharmacologic inhibitors of AR in clinical (and some pre-clinical) studies of diabetic retinopathy have been disappointing. Whether this was due to inadequacies of the pharmacologic inhibitors, limited penetration of target tissues, or unfavorable pharmacokinetic characteristics has remained unclear.
Use of AR inhibitors revealed a variety of diabetes-induced alterations in retinal metabolism that could be manipulated. AR inhibitors were reported to inhibit diabetes-induced abnormalities in oxidative stress, inflammation, and apoptosis in the retina [5,36,37,38,39]. Recent studies using pharmacologic inhibitors of AR found anti-inflammatory actions even in the absence of Figure 2. Effects of 2 mos diabetes and AR2/2 on retinal oxidative stress. (a) Retinas from diabetic WT mice generated significantly more superoxide than nondiabetic WT mice, and the diabetes-induced superoxide generation was inhibited by deletion of AR. Reduced glutathione (GSH) (b) tended to become subnormal, and nitrotyrosine (c) tended to become increased in diabetes, but these changes were not statistically significant, and were not corrected in AR2/2 mice. n = 4-6 per experimental group. doi:10.1371/journal.pone.0062081.g002 hyperglycemia [3,6,[11][12][13][14][15][16][40][41][42][43]. Thus, it is unclear whether the observed effects of AR inhibitors are really mediated via hyperglycemia-induced activation of polyol pathway activity or alternatively via other mechanisms possibly unrelated to glucose metabolism (such as metabolism of lipid-derived aldehydes and retinoids by AR and closely-related aldo-keto reductases [44,45]).
To avoid potential uncertainties associated with pharmacologic inhibitors, mice genetically lacking AR were generated [28,46]. Initial studies of db/db mice deficient in AR suggested that AR contributed to the diabetes-induced abnormalities in the retina, including breakdown of the blood-retinal barrier, loss of pericytes, and even an effect on neovascularization was claimed. The present results using C57Bl/6 mice that are deficient in AR demonstrated that this genetic modification substantially (but not totally) inhibited the diabetes-induced degeneration of retinal capillaries, and totally inhibited superoxide generation by the retina in diabetes. Consistent with most other studies of diabetic mice, however, no retinal neovascularization was observed. Unlike studies using pharmacologic inhibitor, genetic deletion of AR had no significant effect on retinal GSH levels in diabetes in the present study. Consistent with this, reported inhibition of diabetesinduced defects in electroretinogram using pharmacologic inhibitors [47][48][49] was not confirmed using AR 2/2 mice studied by us [50], raising a possibility that some of the observed effects of AR inhibitors were due to off-target effects. In addition, the differences could be secondary to metabolic pathways induced in response to the deletion of AR, thus altering the retinal environment. Deletion of AR also inhibited the diabetes-induced increase in expression of iNOS but not ICAM-1, which is interesting because both are regulated by NF-kB, and both have been shown to be strongly related to the capillary degeneration occurring in retinas of diabetic animals [51,52]. Inasmuch as superoxide and iNOSgenerated NO can lead to the generation of nitrotyrosine on proteins [53,54], it is surprising that AR deletion inhibited both superoxide and iNOS expression in diabetic mice, yet did not correct nitrotyrosine. This finding needs to be confirmed with additional animals.
It has been assumed by many that the molecular abnormalities contributing to vascular pathology in diabetic retinopathy are present in vascular cells themselves. In vitro studies have indicated effects of AR within endothelial cells [35,54,55. Recent evidence indicating that leukocytes play a critical role in the capillary degeneration [Li, 2012 #6144], however, led us to test the possibility that AR present in leukocytes might contribute to the capillary cell death. Our in vitro studies compared leukocytemediated killing of retinal endothelial cells by leukocytes having, and not having, AR. These results suggest that AR contributes only little to the hyperglycemia-induced killing of endothelial cells by leukocytes, suggesting that the beneficial effect of AR deletion might in fact be within cells of the retina itself.
Our results utilizing AR mutant mice, together with results from other investigators using a similar approach [46], demonstrate that AR contributes to the pathogenesis of vascular pathology of early diabetic retinal disease in mice. Surprisingly, the AR knockout did not inhibit several diabetes-induced molecular abnormalities that previously have been found to contribute to the retinopathy [51,56], suggesting that multiple molecular pathways contribute to the vascular lesions of the retinopathy in mice. Perhaps the relative contributions of these various pathways differs among species, and these differences among species contributed to the lack of significant effect of AR inhibitors in diabetic dogs [25] or patients [24], in contrast to more positive effects in rodents. In the absence of information as to which pathways predominate in the development of retinopathy in diabetic humans, future strategies probably will benefit from targeting of multiple pathways that each partially contribute to the pathogenesis of the retinopathy.