Retinopathy in a Diet-Induced Type 2 Diabetic Rat Model and Role of Epigenetic Modifications

The worldwide prevalence of diabetes is increasing at a fast pace, and among the population with diabetes, type 2 diabetes accounts for ∼90% of all cases around the world. Physical inactivity, intake of trans fats, and obesity are considered as the major contributor for the rise in diabetes (1,2). Rapid increase in the incidence of diabetes is also escalating the global burden of diabetic complications, including vision loss due to diabetic retinopathy (3), and experimental models are the key to understanding the molecular mechanisms for therapeutical interventions.

The etiology of diabetic retinopathy is complex, with various molecular, biochemical, structural, and functional abnormalities contributing to its development (4). Although hyperglycemia is the main instigator, serum lipids are also associated with the development/progression of diabetic retinopathy (5), and long-term administration of fenofibrate is shown to reduce the progression of diabetic retinopathy in patients with background retinopathy (6). Dysregulation in systemic and local lipid metabolism plays an important role in the development of diabetic retinopathy (7), and experimental models have demonstrated that simultaneous presence of both hyperlipidemia and hyperglycemia augments capillary cell apoptosis and the development of diabetic retinopathy (8). Several animal models of type 2 diabetes have been used for investigating pathophysiology of diabetic complications, including a nonobese spontaneously polygenic Goto-Kakizaki (GK) rat, Zucker diabetic fatty (ZDF) rat, and Nile rat (Arvicanthis niloticus) (9). However, these models fall short of duplicating the general patient population with type 2 diabetes, because unlike patients with diabetes, ZDF rats have a defect in their leptin receptors (10), GK rats are not obese (11), and Nile rats are primarily herbivorous (9). The development of diabetic retinopathy, and its molecular mechanism, in a model that closely mimics the natural history and metabolic characteristics of human type 2 diabetes remains to be elucidated.

Diabetic environment increases oxidative stress, and oxidative stress is considered to play a major role in the development of diabetic retinopathy (12–14). In diabetes, retinal mitochondria are damaged, and DNA damage is more extensive at the displacement loop (D-Loop) region of the mtDNA, the regions important for mtDNA transcription and replication. Furthermore, the expression of many retinal genes and proteins associated with mitochondrial homeostasis is altered (14,15). Experimental models have documented that increase in cytosolic reactive oxygen species (ROS) precedes mitochondrial dysfunction (16), and NADPH oxidase 2 (Nox2) is one of the major enzymes associated with cytosolic ROS production. Rac1, a small-molecular-weight G-protein, is an obligatory component of Nox2 holoenzyme, and in diabetic retinopathy, Rac1-Nox2-ROS signaling is activated before mitochondrial damage (17).

Recent research has documented that the enzymatic machinery responsible for epigenetic modifications, the modifications that can alter gene expression without affecting the DNA sequence, is also altered in diabetes (14,15,18). Enzymes responsible for maintaining DNA methylation status, DNA methyl transferases (Dnmts) and ten-eleven translocation methylcytosine dioxygenases (Tets), are activated, and mtDNA is hypermethylated in the retina and its vasculature in type 1 diabetic rodents and human donors with documented diabetic retinopathy (19). However, how obesity/hyperlipidemia effects DNA methylation and its enzymatic machinery in the hyperglycemic milieu is not clear.

A high-fat–fed/streptozotocin (Stz)–induced hyperglycemic rat model is now being used for neural complications and pharmacologic screening (20,21), but the development of diabetic retinopathy in this animal model is not elucidated. Our goal was to investigate the development of diabetic retinopathy in this type 2 diabetic animal model, which mimics the normal progression and metabolic features of human type 2 diabetes. We have investigated retinal vascular pathology and functional alterations in a high-fat–fed rat model from 2 to 6 months after Stz-induced hyperglycemia and compared it with a type 1 diabetic rat model. To understand the molecular mechanism, mitochondrial damage and epigenetic modifications of mtDNA and Rac1 promoter are evaluated in the retinal microvasculature from these diabetic animal models.

Sprague Dawley male rats (9 to 10 weeks old) were divided into two groups; while rats in group 1 remained on the regular rat chow (catalog number 7001; Envigo, Indianapolis, IN) containing 4.25% kcal as fat, group 2 rats were placed on a high-fat diet (catalog number D12451; Research Diets Inc., New Brunswick, NJ) containing 45% kcal as fat. After 8 weeks on a high-fat diet, half of the rats received a low dose of Stz (30 mg/kg body weight [BW] i.p.) to induce diabetes (T2D) (21,22). Simultaneously, half of the rats on regular rat chow (group 1) were also injected with 60 mg/kg BW Stz (T1D) (16), and the other half remained normal controls (Norm); each group had 12–16 rats. Rats in T2D and HF groups were maintained on the same high-fat diet and T1D and Norm groups on normal rat chow throughout the experiment. They were sacrificed 2, 4, and 6 months after Stz administration. The treatment of the animals was in accordance with the guidelines of The Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and followed our institutional guidelines.

Rats were fasted overnight, weighed, and administered glucose (2 g/kg BW i.p.). Blood glucose was measured using glucose-oxidase reagent strips just before and 15–180 min post–glucose administration (22).

Insulin resistance sensitivity was determined by administering Humulin R (1 IU/kg BW i.p.) (Eli Lilly and Company, Indianapolis, IN) and measuring their blood glucose 0–180 min after insulin administration (23).

Serum cholesterol and triglycerides were quantified using commercial kits from Abcam (catalog numbers ab65390 and ab65336; Cambridge, MA), as described previously (8).

Retina was suspended in 15–20 mL deionized water and incubated in a shaker water bath for 1 h at 37°C. After removing the noncapillary tissue by repeated inspiration and ejection, the preparation was rinsed with sterile PBS (24,25). Each microvessel preparation included retina from Norm, T1D, T2D, and HF groups.

The whole retina was isolated from the formalin-fixed eyes and, after rinsing overnight in running water, incubated for 45–70 min at 37°C in 3% crude trypsin (Invitrogen-Gibco, Grand Island, NY) containing 200 mol/L sodium fluoride. The cleaned vasculature was stained with periodic acid Schiff (PAS) hematoxylin (catalog number 395B; Sigma-Aldrich), and acellular capillaries and pericyte ghosts were counted under a microscope (8,26).

Electroretinogram (ERG) was performed in dark-adapted rats anesthetized with ketamine-xylazine. The pupil was dilated with tropicamide ophthalmic solution, and, after lubricating the cornea with Goniovisc, ERG responses were measured using Ocuscience HMsERG. Ganzfeld flashes with intensities ranging from 100 to 25,000 mcd.s/m² were used, and ERG responses were recorded. The amplitudes and the implicit times of b-wave were measured using ERGView software (8,26).

RNA, isolated from microvessels using TRIzol reagent, was used to synthesize cDNA. Gene transcripts were estimated by SYBR Green–based quantitative PCR (qPCR) using gene-specific primers (Supplementary Table 1). β-Actin was used as a housekeeping gene, and the fold change was calculated using the ΔΔ threshold cycle method.

For mtDNA damage, long (13.4-kb) and short (210-bp) regions of mtDNA were amplified in the genomic DNA obtained from retinal microvessels using semiquantitative PCR, and the products were resolved on a 2% agarose gel. Relative amplification was quantified by normalizing the intensity of the long product to the short product (17).

Mitochondrial copy numbers were determined in the genomic DNA by quantifying the ratio of the mtDNA-encoded gene cytochrome B (CytB) and nuclear DNA–encoded β-actin (27).

Total ROS levels were measured in retinal microvessels (4 to 5 μg protein) fluorimetrically using 2′,7′-dichlorofluorescin diacetate (catalog number D6883; Sigma-Aldrich) (8).

The G-LISA colorimetric assay kit (catalog number BK-128; Cytoskeleton, Inc., Denver, CO) was used to quantify Rac1 activity in 20–25 μg protein (8,28). The values from Norm group were considered as 1.

5-Methylcytosine (5mC) at D-Loop and 5-hydroxymethylcytosine (5hmC) at the Rac1 promoter were quantified in the immunoprecipitated genomic DNA using EpiQuik methylated and hydroxymethylated DNA Immunoprecipitation Kits (catalog numbers P-1015 and P-1038, respectively, from EpiGentek, Farmingdale, NY), as described previously (19,24).

The binding of Tet2, Dnmt1, or transcription factor nuclear factor-κB (NF-κB) (p65 subunit) at Rac1 promoter was determined by chromatin immunoprecipitation assay (19,24) using IgG as an antibody control. The antibodies used for immunoprecipitation—Tet2 (catalog number ab135087), Dnmt1 (catalog number ab13537), p65 (catalog number ab7970), and IgG (catalog number ab171780)—were obtained from Abcam.

The EpiQuik Epigenase 5mC-Hydroxylase TET Activity/Inhibition Assay Kit (catalog number P-3087; EpiGentek) was used to determine Tets activity in the retinal nuclear fraction (15–20 μg protein), as described previously (19).

Data are presented as means ± SD. Comparison between groups was made using one-way ANOVA followed by Dunn t test; P < 0.05 was considered significant.

The data supporting the findings of this study are available from the corresponding author upon request. No applicable resources were generated during this study.

Initial weight of the rats in each group was 190–210 g, and after 8 weeks, the rats on the high-fat diet were 25% heavier compared with the rats on normal chow (492 vs. 396 g). High-fat diet rats had some impairments in their glucose tolerance, and their triglycerides were significantly higher (Table 1 and Fig. 1A–D). Compared with T1D group, age-matched rats in Norm, T2D, and HF groups had significantly higher BWs throughout the experiment, and, compared with T2D group, rats in HF group were heavier. T1D and T2D groups had comparable nonfasting blood glucose, and the values were greater than five times higher compared with Norm and HF groups. T2D and HF groups had two- to threefold higher serum triglycerides and cholesterol values compared with T1D and Norm groups (Table 1). Glucose utilization and insulin tolerance were impaired in both T1D and T2D groups, but the severity of impairment in insulin tolerance was significantly higher in T2D group compared with T1D group (Fig. 1E).

While T1D, HF, and Norm groups had similar numbers of acellular capillaries and pericyte ghosts at 4 months’ duration, they were significantly higher in T2D group. However, at 6 months of diabetes, histopathology was comparable in T1D and T2D groups, with similar numbers of acellular capillaries and pericyte ghosts in their retinal vasculature (Fig. 2A and B). Figure 2C and D represents PAS-stained trypsin-digested retinal microvasculature from rats at 4 and 6 months of diabetes, respectively.

Retinal microvasculature is considered as the main sight of pathology associated with diabetic retinopathy, but retinal dysfunction, including deficits in ERG, is observed before vascular histopathology (29). Despite no effect of high-fat diet alone (HF group), 2–6 months of diabetes decreased b-wave amplitude and increased implicit time. As shown in Fig. 3A, at 4 months’ duration, these changes were higher in T2D group compared with T1D group. However, at 6 months’ duration, T1D and T2D groups had comparable abnormalities in their ERG function (Fig. 3B).

Retinal mitochondrial dysfunction is intimately associated with the development of diabetic retinopathy (14,15); to understand how high fat exacerbates diabetes-induced retinal pathology and dysfunction, mitochondrial damage was examined. Although at 2 months of diabetes, compared with Norm, no changes in mitochondrial damage and copy numbers were observed in T1D, T2D, and HF groups, at 4 months, unlike HF and T1D groups, T2D group had increased mtDNA damage and decreased CytB transcripts. At 6 months, both T1D and T2D groups had mtDNA damage, but T2D had higher mtDNA damage and lower CytB transcripts compared with T1D group (Fig. 4A and B). Consistent with our previous results showing increased mtDNA biogenesis in the early stages of diabetes (30), mitochondrial copy numbers were about twofold higher at 2 months’ duration, and at 4 months, Norm, T1D, and HF groups had similar mitochondrial copy numbers, but they were reduced by 20% in T2D group (Fig. 4C). At 6 months’ duration, compared with Norm group, mitochondrial copy numbers were decreased in the other three groups, and the decrease was significantly higher in T2D compared with T1D and HF groups (Fig. 4C).

Mitochondrial damage in diabetic retinopathy is preceded by an increased accumulation of cytosolic ROS that are produced mainly by Rac1-Nox2 signaling (17). To determine the role of cytosolic ROS in accelerated retinal histopathology, ROS were quantified. ROS levels were significantly higher in T1D group at 2 months of diabetes compared with HF group, but, compared with T1D, T2D group had approximately twofold higher ROS levels. Similar patterns were observed at 4 and 6 months of diabetes (Fig. 5A). Consistent with ROS, compared with T1D group, T2D group had higher Rac1 activity and gene transcripts at all of the three durations of diabetes (Fig. 5B and C).

Epigenetic modifications are implicated in mtDNA damage and its transcriptional suppression and in activation of Rac1 (15,24). To investigate the role of epigenetic modifications in exacerbated/accelerated mtDNA damage and Rac1 activation in T2D group, methylation of D-Loop and Rac1 promoter DNA was quantified. While at 2 months’ duration, D-Loop methylation was similar in all of the four groups, at 4 and 6 months’ durations, it was significantly higher in T2D group compared with T1D or HF groups (Fig. 6A). However, in the same samples, Dnmt1 was increased significantly as early as 2 months of diabetes, and, compared with T1D and HF groups, T2D group continued to present significantly higher values (Fig. 6B).

Dnmts methylate cytosine and Tets hydroxymethylate 5mC, forming 5hmC, which opens up the chromatin for transcription factor binding (31), and in diabetes, Rac1 promoter is hydroxymethylated (24). Figure 7A shows that Rac1 promoter had a 55–85% increase in 5hmC at 2 months’ duration in HF and T1D groups compared with Norm, but the values were increased by more than twofold higher in T2D group. Similarly, T2D group had significantly higher 5hmC at Rac1 promoter at 4 and 6 months’ durations. Consistent with 5hmC, binding of Tet2 and the transcription factor (p65 subunit of NF-κB) at Rac1 promoter was also about twofold higher in T2D group compared with HF or T1D groups at 2–6 months of diabetes (Fig. 7B and C). Values obtained from IgG antibody control were <1% of those obtained from antibodies against Tet2 or p65. Tet2 gene transcripts were also significantly higher in T2D compared with T1D and HF groups (P < 0.05) (Fig. 7D). Similarly, activity of Tets was 2- to 3.5-fold higher in T1D and T2D groups, but T2D group had significantly higher Tets activity compared with T1D group. Figure 7E represents Tets activity in retinal samples at 4 months’ duration.

Nearly all patients with type 1 and >60% of patients with type 2 diabetes have retinopathy within 20 years of diabetes, but ufortunately, >20% of patients with type 2 diabetes at the time of their diagnosis of diabetes have some form of retinopathy (32). Obesity is considered a major independent risk factor for developing diabetes, and type 2 diabetes is three to seven times higher in obese patients than in normal-weight individuals. Hyperlipidemia is also closely associated with the development of diabetic retinopathy (5,33), and lipid-lowering therapy has been shown to decrease the number of laser treatments in patients with diabetes with proliferative retinopathy (34). The type 2 diabetic animal model, developed by feeding a high-fat diet, followed by induction of hyperglycemia closely duplicates the features of the common population with type 2 diabetes. This model has elucidated a molecular mechanism in the diabetic neuropathy field, including changes in corneal nerve innervation and sensitivity (20,21). In this study, we document the development of retinopathy in this type 2 diabetic animal model presenting obesity, hyperlipidemia, and impaired glucose and insulin tolerance. Compared with 6 months of hyperglycemia in type 1 diabetic model, retinal vascular pathology is observed within 4 months in type 2 diabetic model, and ERG impairments are also intensified. The mechanistic insight into the development of diabetic retinopathy illustrates higher mitochondrial damage and lower mtDNA copy numbers and increased Rac1-ROS activation in their retinal vasculature. Our study documents the role of epigenetics in obesity/hyperlipidemia-induced increased ROS production with hypermethylation of mtDNA and hydroxymethylation of Rac1 promoter in T2D group. The results confirm the validity of an animal model of diabetic retinopathy closely mimicking the population with type 2 diabetes and document an important role of obesity/hyperlipidemia in diabetes-induced epigenetic modifications in diabetic retinopathy.

Many type 2 diabetic animal models have been used to evaluate diabetic retinopathy, including ZDF, GK, Otsuka Long-Evans Tokushima fatty, spontaneously diabetic Torii, and Nile grass rats and db/db mouse (8,9,35–37). These models have provided a wealth of information about the pathogenesis of the disease, but they fall short of closely replicating the entire complexity of the disease of human type 2 diabetes (11). Insulin resistance is a common feature of type 2 diabetes, which is increased by obesity, and glucose tolerance is impaired (38). In this study, the rats in T2D group have the general characteristics of the patient population with type 2 diabetes with significantly high BW and hyperlipidemia, both before and after induction of diabetes, and overt hyperglycemia and hyperlipidemia after induction of diabetes. Although no retinal microvascular histopathology is detectable at 4 months of diabetes in rats in T1D group, age-matched rats in T2D group with similar duration of hyperglycemia show significant increase in acellular capillaries and pericyte ghosts. In accordance, the number of acellular capillaries in the retinal vasculature is significantly increased at 20 weeks of age in ZDF rats, a duration when they are not seen in the age- and hyperglycemia duration-matched type 1 diabetic rat model (8). While histopathology of diabetic retinopathy is observed in the retinal vasculature, nonvascular cells of the retina also experience considerable structural and functional damage, and ERG shows impairments with delayed implicit time and reduced amplitudes (29). The type 2 diabetic animal model used in this study shows that, compared with a type 1 diabetic animal model, during early stages of diabetes (2 months), abnormalities in ERG are higher. However, extending the duration of diabetes to 6 months results in comparable retinal histopathology and retinal functional abnormalities between these two animal models, suggesting that initial obesity could have an important role in potentiating the development of diabetic retinopathy. In support, changes in BMI in patients with type 2 diabetes have an impact on ERG functional changes (39). Furthermore, a docosahexaenoic acid–rich diet prevents diabetic retinopathy in rats via an ameliorating increase in the acid sphingomyelinase in endothelial progenitor cells (40), and very-long-chain ceramides are considered important for stabilizing tight junctions and preventing retinal vascular permeability (41).

Mitochondrial damage plays a significant role in the pathogenesis of diabetic retinopathy (4,15). Data presented in this study show that in the early stages of diabetes, addition of obesity/hyperlipidemia does not produce any effect on mtDNA and its transcription, but as the obesity/hyperlipidemia continues, mtDNA damage begins, and its transcription decreases. However, at a duration when histopathology is detectable in T1D group, addition of obesity/hyperlipidemia further damages mtDNA and reduces its transcription. In support, high fat/obesity is associated with mitochondrial damage (42,43). Activation of cytosolic Nox2 generates excessive ROS, and ROS damage the mitochondria, accelerating capillary cell apoptosis (15,17,44). The results show that obesity/hyperlipidemia, without hyperglycemia (HF group), continues to activate Rac1-ROS signaling, and hyperglycemia (T2D group) further increases the signaling. Consistent with our results in the whole retina from the ZDF model (8), the retinal vasculature in type 2 diabetic animal model has consistently higher Rac1-ROS compared with type 1 diabetic model, and this continues until at least 6 months of diabetes. Similar to the retinal microvascular histopathology in HF group, despite some increase in Rac1-ROS signaling in retinal vasculature, mitochondria are not damaged, further confirming the role of mitochondrial damage in the development of diabetic retinopathy. These results also raise a possibility of a threshold level of ROS required for damaging the mitochondria, which the HF group fails to achieve.

Gene expression is also regulated by epigenetic modifications; formation of methylated cytosine is as a mark of gene repression and hydroxylation of methylated cytosine a mark of gene induction (15,18,45). DNA methylation is a highly dynamic process, and both Dnmts and Tets are activated in the retinal vasculature in diabetes (19,31). Rac1 promoter DNA undergoes dynamic methylation–hydroxymethylation, resulting in increased 5hmC levels and transcription factor binding (24,46). Lipids can influence DNA methylation machinery (e.g., blood lipids influence DNA methylation in the circulating cells) (47,48), and maternal dyslipidemia increases Dnmt activity and global DNA methylation in the liver of the offspring (49). In this study, we show that obesity/hyperlipidemia, in addition to aggravating Dnmts, exacerbates Tet2 and alters methylation of mtDNA and Rac1. Transcription of mtDNA is further compromised, and mtDNA-encoded CytB continues to be subnormal in T2D group, further aggravating ROS accumulation. In support, a strong relationship is reported between high BMI and global DNA methylation of genes associated with oxidative phosphorylation and electron transport chain (50). However, despite similar abnormalities in glucose tolerance and retinal histopathology at 6 months of diabetes in T1D and T2D groups, redox-sensitive Dnmt1 and Rac1 promoter DNA methylation remain significantly higher in T2D group. This raises the possibility that the early damage initiated by the impaired insulin/glucose tolerance could be further worsened by the continuous presence of hyperlipidemia, contributing to the sustained difference between T1D and T2D groups.

One limitation of this study is the use of only male rats, and the possibility that under similar experimental conditions female rats may not develop retinopathy, or the severity of hyperglycemia/hyperlipidemia and retinopathy may be higher than that seen in male rats, cannot be excluded. Furthermore, the mechanism of the development of diabetic retinopathy in female rats may be different than the one proposed in this study for male rats; future studies will focus on addressing sex-based variations in this novel type 2 diabetic animal model of retinopathy.

Obesity is considered as a strong risk factor of type 2 diabetes, and in this study, we have clearly demonstrated the development of retinopathy in an animal model in which obesity is followed by hyperglycemia, closely mimicking the general patient population with type 2 diabetes. Epigenetic modifications are greatly influenced by external factors, including exercise and diet, and we show that obesity/hyperlipidemia further augments epigenetic modifications of mtDNA and Rac1 promoter, accelerating mitochondrial damage and the development of diabetic retinopathy. Thus, strategies focusing on maintaining good lifestyle and BMI, in addition to ameliorating hyperlipidemia, could also be beneficial in regulating epigenetic modifications and preventing/retarding retinopathy in patients with diabetes.

Acknowledgments. The author thanks Dr. Duraisamy and Dr. Sunita (Wayne State University) for help with initial experiments and Dr. Mohammad and Gina Polsinelli (Wayne State University) for analyzing the histology and technical help, respectively.

Funding. This study was supported in parts by grants from the National Eye Institute (EY014370, EY017313, and EY022230) and The Thomas Foundation to R.A.K. and an unrestricted grant from Research to Prevent Blindness to the Ophthalmology Department, Wayne State University.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.A.K. was responsible for the experimental plan, literature search, and manuscript writing and editing. R.A.K. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.