In contrast to Western diet, a plant-based, high-fat, low-sugar diet does not exacerbate retinal endothelial injury in streptozotocin-induced diabetes

Abstract

Controversy remains about how diet affects the vascular endothelial dysfunction associated with disordered insulin-glucose homeostasis. It is postulated that the type and level of certain macronutrients contribute to endothelial dysfunction in vascular diabetes complications. However, it is not well understood how specific macronutrients affect the molecular inflammatory response under conditions of hyperglycemia. Here, we examined retinal microvascular endothelial injury in streptozotocin (STZ)-diabetic rats fed a laboratory Western diet (WD). WD, characterized by its high content of saturated fat, cholesterol, and sugar, significantly increased retinal leukocyte accumulation and endothelial injury in the STZ-diabetic rats. Suppression of endothelial NF-κB signaling in the STZ model reduced the WD-induced increase in leukocyte accumulation. To isolate the effect of dietary fat, we generated high-fat diets with varying fatty acid balance and type. These diets contained moderate amounts of carbohydrates but no sugar. We found that neither high levels of saturated or unsaturated fats per se increased retinal leukocyte accumulation and endothelial injury in the STZ-diabetic rat model but that the combination of high levels of dietary cholesterol with specific saturated fatty acids that are abundant in WD exacerbated leukocyte accumulation and endothelial injury in the retinas of STZ-diabetic rats.—Barakat, A., Nakao, S., Zandi, S., Sun, D., Schmidt-Ullrich, R., Hayes, K. C., Hafezi-Moghadam, A. In contrast to Western diet, a plant-based, high-fat, low-sugar diet does not exacerbate retinal endothelial injury in streptozotocin-induced diabetes.

Experimental and clinical evidence supports the hypothesis that chronic low-grade inflammation of the vascular endothelium is an inciting event in the complex pathology of diabetes complications (1–4). At the microscopic level, the inflammatory response elicits leukocyte adhesion to the endothelial lining via classic adhesion molecules, including intercellular adhesion molecule-1 (ICAM-1) on the vasculature and β₂ integrins on the leukocytes. Within 1 wk of streptozotocin (STZ)-induced experimental type 1 diabetes (T1D) in rats, adherent leukocytes are temporally and spatially associated with retinal endothelial cell injury and death, which are early events of endothelial dysfunction (2). When STZ-diabetic rats are treated with ICAM-1– and β₂ integrin–blocking antibodies, leukocyte adhesion is suppressed and retinal endothelial cell injury and death are prevented (2, 5).

The STZ model is an experimental model of insulin deficiency. STZ is toxic to the insulin-producing β cells in the islets of Langerhans in the pancreas. Insulin plays a critical role in the regulation of lipoprotein metabolism in diabetes (6). Hypertriglyceridemia, a marked increase in fasting plasma triglyceride (TG) concentration, is associated with impaired insulin action in conditions of insulin deficiency and T1D, impaired glucose tolerance, and type 2 diabetes (T2D) (7). In insulin deficiency, hypertriglyceridemia is secondary to a defect in the removal of both chylomicrons and very low density lipoprotein (VLDL). Hypertriglyceridemia in STZ-diabetic rats occurs despite a reduction in hepatic VLDL-TG secretion and is accentuated when the rats are fed a high-fat diet (8). Studies in humans with insulin deficiency are consistent with these results, and the phrase “fat-induced” lipemia has long been used to describe hypertriglyceridemia in uncontrolled T1D (9). Although abnormalities in plasma VLDL-TG and chylomicron removal respond to insulin administration in T1D, the reduction in hepatic VLDL-TG synthesis and secretion is more resistant to insulin treatment (7, 10–12). In general, in insulin-treated patients with T1D, plasma cholesterol and plasma concentrations of the lipoproteins VLDL, low density lipoprotein (LDL), and high density lipoprotein (HDL) are comparable to those of nondiabetic persons. However, profound abnormalities in lipoprotein metabolism occur in experimentally induced insulin deficiency. Along with increased concentration of plasma VLDL-TG concentration, there is a fall in plasma LDL- and HDL-cholesterol concentrations (6).

The role of impaired insulin action, even in tissues where insulin does not regulate glucose uptake, and the interplay between abnormalities in glucose and lipoprotein metabolism in the development of vascular diabetes complications is complex and not yet well understood. Even less understood is how macronutrient composition and fatty acid balance and type in the diet may regulate cellular and metabolic abnormalities to preserve endothelial health or exacerbate endothelial dysfunction. To date, no research or clinical trials have evaluated the role of dietary patterns on the development or incidence of microvascular diabetes complications.

Here, we studied the effect of high-fat diets on the early inflammatory events in the vascular retinal endothelium of the STZ-rat, an experimental model of insulin deficiency that exhibits both hyperglycemia and hypertriglyceridemia. In patients with T1D, hypertriglyceridemia is confined to individuals with poor glycemic control. An analysis of the data of the landmark Diabetes Control and Complications Trial (DCCT) and follow-up Epidemiology of Diabetes Interventions and Complications (EDIC) study showed that the severity of diabetic retinopathy (DR) in patients with T1D was positively associated with TGs (13). However, the extent of the relationship between diet and DR in T1D, either directly, or indirectly, via increases in retinopathy-associated risk factors (blood pressure, serum lipids, hemoglobin A1c (HbA1c), body mass index, or insulin utilization) was not determined as part of these studies. A subsequent post hoc analysis of dietary and lifestyle data and retinopathy-associated risk factors in the DCCT revealed that the proportions of dietary carbohydrates (CHOs), protein, and dietary fiber are inversely associated with retinopathy progression, whereas the proportion of cholesterol and dietary fat – particularly saturated and monounsaturated fat – positively correlated with retinopathy progression (14). The correlations reported between dietary fat and CHOs, although statistically significant, were relatively weak (r < 0.25), and correlations of fatty acids according to carbon length with progression of retinopathy were not examined. The relationship between retinopathy-associated risk factors and nutritional intake was similar to that of the direct relationship between diet and retinopathy progression. Based on the retrospective analysis of this study, patients with T1D were advised to reduce their total dietary fat to <30% and the saturated fat to <10%, as was formerly recommended by the American Diabetes Association. To this day, the evidence for the effect of dietary fat intake on DR incidence and progression remains inconclusive, and a potential positive effect of specific saturated fatty acids (SFAs) and unsaturated fatty acids has been raised (15–18).

The diet in the Western industrialized countries entails significant consumption of refined CHOs and fats (19). Based on the macronutrient composition of this diet, the Western diet (WD) was developed for laboratory animals. It contains a relatively moderate proportion of CHOs (44% energy) that is, however, mostly sugar (sucrose), followed by fat (41% energy; most of it saturated), and protein (15% energy). It is also high in dietary cholesterol. To investigate the impact of dietary fat composition on early DR processes in the context of STZ-induced hyperglycemia and its associated hypertriglyceridemia, we performed nutritional studies in STZ-diabetic rats fed chow, WD, or semipurified high-fat diets with and without added cholesterol, and measured retinal leukocyte accumulation and endothelial injury. In vivo, we used our unique molecular imaging technique to quantify the expression of the adhesion molecule ICAM-1 in the retinal microvessels of rats on specific diets.

MATERIALS AND METHODS

Animals

Animal experiments were approved by the Institutional Animal Care Committee (IACUC) of the Brigham and Women’s Hospital and adhered to the standards of Association for Research in Vision and Ophthalmology (ARVO) and Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) regulations.

STZ-induced diabetes model

Long-Evans rats (180–200 g, 6–7 wk old; Charles River Laboratories, Wilmington, MA, USA) had food withheld overnight and intravenously injected with STZ (60 mg/kg; MilliporeSigma, Burlington, MA, USA) diluted in citrate buffer (0.1 M; pH 4.5). C57BL/6J mice (6–7 wk old; 000664; The Jackson Laboratory, Bar Harbor, ME, USA) had food withheld overnight and injected intraperitoneally with STZ (50 mg/kg) for 5 d consecutively. Control animals were injected with citrate buffer. Animals were maintained in an air-conditioned room with a 12-h light/dark cycle with free access to water and food. Diabetic status was evaluated by measuring body weight (BW) and blood glucose (BG) levels. Animals with BG levels higher than 250 mg/dl 24 h after STZ injection were considered diabetic. The mortality rate in the 2 wk after injections was recorded.

Animal diets

We performed nutritional studies with lab chow (Lab Diet 5020; PMI Nutrition, St. Louis, MO, USA), WD (57BD, 45365; TestDiet, St. Louis, MO, USA), and with 4 semipurified diets (Table 1). Chow and WD provided the same number of energy calories per gram but differed in their macronutrient composition [CHO:fat:protein energy percentage (%en)] and fatty acid balance [SFA:monounsaturated fatty acid (MUFA):polyunsaturated fatty acid (PUFA) %en]. The rat chow is a low-fat, high-starch diet, with roughly equal amounts of saturated, mono-, and polyunsaturated fats, and is 0.03% cholesterol. The WD is a moderate-CHO, higher-fat diet, high in both sugar (sucrose) and milk fat (almost all the diet’s fat comes from milk fat and is mostly saturated) and is 0.2% cholesterol. We generated 2 semipurified cholesterol-free, sugar-free, high-fat diets with varying fatty acid balance (Table 1). These diets provided the same number of energy calories per gram as in the WD and have a macronutrient composition similar to WD. One diet, diet A, had a fatty acid balance similar to chow. The other diet, diet B, was a high-saturated-fat diet with just enough PUFAs to prevent essential fatty acid deficiency. We then generated 2 additional diets, C and D, with an added 0.1% cholesterol to diets A and B, respectively. Animals were provided food and water ad libitum.

TABLE 1.

Diet compositions

Ingredient	Diet (g/kg)
	Plant-fat mimicking	Plant-fat mimicking + cholesterol	Animal-fat mimicking	Animal-fat mimicking + cholesterol	WD
PUFA (%en)	13.5	13.5	2	2	0
CHO:Fat:Protein (%en)	43:40:17	43:40:17	43:40:17	43:40:17	44:41:15
Energy (kcal/g)	4.5	4.5	4.5	4.5	4.41
Casein	100	100	100	100	—
Lactalbumin	100	100	100	100	—
Dextrose	200	200	200	200	—
Cornstarch	225 (+60/gel)	224 (+60/gel)	225 (+60/gel)	224 (+60/gel)	—
Cellulose	50	50	50	50	—
Fat SFA:MUFA:PUFA (%en)	(13.5:14:13.5)	(13.5:14:13.5)	(35:4:2)	(35:4:2)	(13:5:1)
Soybean oil	85	85	—	—	—
Palm olein	68	68	—	—	—
Palm oil	47	47	—	—	—
Palm kernel oil	—	—	191	191	—
Safflower oil (high linoleic 18:2)	—	—	9	9	—
Mineral mix	50	50	50	50	—
Vitamin mix	12	12	12	12	—
Choline chloride	3	3	3	3	—
Cholesterol	0	1	0	1	2

Plasma measurements

Blood glucose (BG), hemoglobin A1c (HbA1c), total cholesterol (TC), TG, free fatty acid (FFA), adiponectin, and vascular endothelial growth factor A (VEGF-A) were measured in normal and diabetic animals. BG levels were measured in tail-blood samples taken randomly or after 16 h of overnight food withdrawal with an Elite Glucometer (Bayer, Mishawaka, IN, USA). Animals with repeated BG of >250 mg/dl were considered diabetic. At the end of the experiments, blood was collected from anesthetized animals by cardiac puncture in EDTA-treated syringes. Glycated hemoglobin was measured by a Glyco-Tek affinity column kit (Helena Laboratories, Beaumont, TX, USA). HbA1c was calculated according to the equation provided by the manufacturer. Subsequently, blood was centrifuged at 12,000 g for 10 min at 4°C and plasma was collected. Plasma levels of TCs and TGs were measured by the enzymatic assay provided in the Thermo Infinity Kit (Thermo Fisher Scientific, Waltham, MA, USA). Adiponectin, FFA, and VEGF-A were measured using a rat total adiponectin-Acrp30 Quantikine ELISA Kit (RRP300; R&D Systems, Minneapolis, MN, USA), FFA Fluorometric Assay Kit (700310; Cayman Chemicals, Ann Arbor, MI, USA), and rat VEGF Quantikine ELISA Kit (RRV00; R&D Systems), respectively.

Leukocyte adhesion in the retinal microvessels

Retinal vessels and adhering leukocytes in control and diabetic animals were labeled with FITC-conjugated concanavalin A (ConA) lectin (Vector Laboratories, Burlingame, CA, USA). Briefly, rats were perfused with 50 ml PBS for 5–10 min to remove intravascular content including erythrocytes and nonadherent leukocytes. To allow drainage, a 16-gauge needle was placed into the right atrium. The perfusion was continued with FITC-coupled ConA (20 µg/ml in PBS). Retinal flat mounts were prepared for evaluation of leukocyte accumulation. The total number of accumulated leukocytes per retina was counted using epifluorescence microscopy.

Endothelial damage in rat retinas

Endothelial cell injury was visualized by in vivo staining with propidium iodide (PI; Molecular Probes, Eugene, OR, USA). PI (1 mg/ml) and DAPI (10 mg/ml) were injected intravenously via the femoral vein. After 12 h, the vasculature and adherent leukocytes in rat retinas were labeled with ConA. The retinas were then studied under a fluorescence microscope. The total number of PI-positive cells per retina was counted.

Western blot analysis

Animals were perfused with PBS (500 ml/kg BW) and eyes were enucleated. Retina and choroid were microsurgically isolated and placed into 150 µl of lysis buffer (Mammalian Cell Lysis Kit; MCL1; MilliporeSigma) supplemented with protease and phosphatase inhibitors (MilliporeSigma) and sonicated. The lysate was centrifuged (12,000 rpm; 15 min; 4°C) and the supernatant was collected. Each sample containing an equal amount of total protein quantified by protein assay (Bio-Rad, Hercules, CA, USA) was separated by SDS-PAGE and electroblotted to PVDF membranes (Thermo Fisher Scientific). Membranes were washed with 5% skim milk and subsequently incubated with a rabbit pAb against ICAM-1 (5 µg/ml; M-19; sc-1511; Santa Cruz Biotechnology, Dallas, TX, USA), phosphorylated (phospho–)NF-κB (3033; Cell Signaling Technology, Danvers, MA, USA), NF-κB (3034; Cell Signaling Technology), phospho-PKC (pan; βII Ser660; 9371; Cell Signaling Technology), or an mAb against β-tubulin (1.5 µg/ml; Abcam, Cambridge, MA, USA) at 4°C overnight, followed by incubation with a horseradish peroxidase–conjugated donkey or sheep antibody against rabbit or mouse IgG (1:2000; GE Healthcare, Waukesha, WI, USA). The signals were visualized with chemiluminescence (ECL kit; GE Healthcare) according to the manufacturer’s protocol.

Molecular imaging of retinal and choroidal vessels

We introduced molecular imaging of retinal and choroidal microvessels for quantitative evaluation of endothelial molecules in the living animals (20–25). Imaging probes were prepared and characterized as previously described (21, 25). Rats were anesthetized with xylazine hydrochloride (10 mg/kg) and ketamine hydrochloride (50 mg/kg) and their pupils were dilated with 0.5% tropicamide and 2.5% phenylephrine hydrochloride. A contact lens was used to retain corneal clarity throughout the experiment. Functionalized probes (3 × 10⁸/ml in saline) were continuously injected into the tail vein within 1 min through a 30½-gauge needle. To visualize probe interactions in the retinal vessels and the choriocapillaris in control and diabetic rats, a scanning laser ophthalmoscope (SLO; HRA2; Heidelberg Engineering, Heidelberg, Germany) was used to make continuous high-resolution fundus images. An argon blue laser was used as the excitation light with a regular emission filter for fluorescein angiography because the excitation (441 nm) and emission (488 nm) maxima of the probes were comparable with those of sodium fluorescein. After imaging probes were injected, SLO images were obtained at a 30° angle at 15 frames per second and digitally recorded for further analysis. Images were recorded 30 min after probe injections. The bright spots in SLO micrographs were counted as a measure for the number of accumulated probes. ImageJ software (v.1.41; National Institutes of Health, Bethesda, MD, USA) was used for analysis. For automated quantification of the number of bound imaging probes in the choriocapillaris microvasculature, the confocal images were merged and, subsequently, the area of the bright spots were measured in ImageJ software.

Mice lacking endothelial NF-κB signaling

NF-κB is a transcription factor that regulates genes necessary for cell survival, differentiation, immunity, and inflammation (26). The RelA (p65) subunit is the major transcriptional activator of NF-κB. Lack of RelA is not compatible with life (27). To study the contribution of endothelial NF-κB activity to leukocyte accumulation in diabetic vascular injury, we generated a mouse that lacks endothelial NF-κB signaling (28, 29). This was achieved by selective expression of IκB, the endogenous inhibitor of NF-κB, in the vascular endothelium. Briefly, Henke et al. (29) crossed mice carrying loxP-flanked alleles encoding for the human IκBαΔN with Cre knock-in mice expressing the Cre recombinase under the control of the endothelial-specific Tie-1 promoter to generate the Tie-1–Cre/IκB-αΔN (Tie1∆N) mice (30). As a result of the IκB-αΔN expression in the endothelial cells, these mice lack NF-κB signaling in their vascular endothelium (29, 31). Endothelial NF-κB can act pro- or antiangiogenic depending on context (32). In the laser-induced choroidal neovascularization, the suppression of endothelial NF-κB signaling in the Tie1∆N mice did not affect angiogenic outcome (31). In this study, the Tie1∆N mice were used to study retinal leukocyte accumulation in the context of diabetes.

Statistical analysis

All values are expressed as means ± sem. Data were analyzed by Student’s t test. Differences between the experimental groups were considered statistically significant when the probability value was P < 0.05 or P < 0.01.

RESULTS

Characterization of the STZ-diabetic rats on normal diet and WD

BW, BG, TG, TC, and HbA1c were measured in the control and STZ-diabetic rats. Compared with normal controls, diabetic rats showed significantly lower BW and higher BG levels 2 wk after diabetes induction (Fig. 1A, B). WD significantly increased BW in normal controls but not in diabetic rats (Fig. 1A). BG and HbA1c were significantly higher in diabetic rats on chow and WD when compared with normal controls (Fig. 1B, C). Plasma TGs were significantly higher in diabetic rats on chow compared with nondiabetic rats on chow. WD increased TGs in both nondiabetic and diabetic rats, albeit drastically more in the diabetic rats (Fig. 1D). TC was highest in diabetic animals on WD but was also significantly higher in normal animals on WD (Fig. 1E). In control rats, WD did not affect plasma FFAs, but FFAs were significantly higher in diabetic animals on WD compared with controls and diabetic rats on chow (Fig. 1F). Plasma adiponectin was significantly reduced in diabetic rats on chow and WD (Fig. 1G). Plasma VEGF-A was slightly increased in diabetic animals on WD (Fig. 1H).

Figure 1.

Characterization of systemic parameters in normal and diabetic animals on chow or WD. Normal and STZ-induced diabetic rats were fed normal chow or WD for 2 wk. Physiologic parameters and blood values were measured in each animal after this period. BW (A), BG (B), HbA1c (C), plasma TGs (D), TC (E), FFA (F), adiponectin (G), and VEGF-A (H). *P < 0.05, **P < 0.01 (n = 9–12 in each group).

Leukocyte accumulation and endothelial damage in retinal microvessels

To investigate the impact of diet on early inflammatory processes in the retina in the context of STZ-induced diabetes, we quantified firm leukocyte adhesion in the retinal vessels using the ConA staining technique. In line with previous reports (5, 25), significantly more leukocytes accumulated in diabetic rats compared with normal controls (Fig. 2A, B). WD significantly increased leukocyte accumulation in diabetic rats but not in normal controls (Fig. 2A, B). To examine the impact of diet on endothelial injury in the retinas of diabetic rats, we performed in vivo injections of PI and quantified the number of injured or dead endothelial cells in histologic flat mounts. Diabetic rats showed significantly more PI⁺ cells per retina compared with the normal controls (Fig. 2C, D). WD significantly increased the number of PI⁺ endothelial cells in both normal and diabetic animals (Fig. 2C, D). These data demonstrate that WD increases leukocyte accumulation and endothelial damage in the retinas of STZ-diabetic rats.

Figure 2.

Retinal leukocyte accumulation and endothelial injury. Normal and STZ-induced diabetic rats fed regular chow or WD for 2 wk were perfused with FITC-conjugated ConA. Retinas were harvested and flat mounted for histologic analysis. To illustrate a larger region of the retinal flat mount at higher resolution, composite micrographs were generated by merging the digital images from adjacent regions of the retina in a mosaic fashion. A) Representative composite micrographs of accumulated leukocytes (arrowheads) in normal and diabetic animals. Scale bar, 200 µm. B) Quantitation of the number of accumulated leukocytes per retina in normal and diabetic animals on standard chow or WD; n = 7–9 in each group. C) To illustrate endothelial injury, in vivo PI staining (red) was performed. Representative micrographs illustrate injured endothelial cells (arrows) in normal and diabetic animals on standard chow or WD 2 wk after diabetes induction. Scale bar, 200 µm. D) Quantitation of the number of PI-stained endothelial cells per entire retina in normal and diabetic animals on standard chow or WD 2 wk after diabetes induction. *P < 0.05, **P < 0.01 (n = 10–16 in each group).

Adhesion molecule expression in the retinas of STZ-diabetic rats on WD

To investigate the expression of the endothelial adhesion molecule ICAM-1 in the retinal vessels in response to diet, we performed Western blotting of retinal microvascular tissues. ICAM-1 was up-regulated in the STZ-diabetic rats compared with the normal controls (Fig. 3A). Although Western blotting revealed the total amount of protein in the retinal tissues, we used our molecular imaging technique to quantify the expression of these molecules in the inner lumina of the retinal microvessels (20–25 and Fig. 3B). We generated molecular imaging probes against ICAM-1 and visualized their interaction in vivo in normal and diabetic animals fed chow or WD (Fig. 3C). Molecular imaging revealed significantly more anti–ICAM-1 imaging probe accumulation in control rats fed WD, comparable to the level in diabetic rats fed regular chow, and diabetic rats fed WD expressed the highest level of ICAM-1 (Fig. 3D).

Figure 3.

Adhesion molecule expression and in vivo molecular imaging of diabetic retinal endothelial injury on chow or WD. A) Representative Western blots of retinas of STZ-induced diabetic and nondiabetic rats for ICAM-1 (n = 2 animals in each group) 2 wk after being fed chow or WD. The housekeeping protein β-tubulin was blotted as internal control. B) Schematic of the in vivo molecular imaging experiments. ICAM-1–targeted nanoprobes were generated by conjugating an anti–ICAM-1 mAb and recombinant Fc-conjugated P-selectin glycoprotein ligand-1 (PSGL-1) as a means to initiate rolling on the endothelium in a 1:1 ratio to the nanoprobes (21, 25). Nanoprobes were then injected into the tail vein of normal and diabetic animals under anesthesia. The retinal microcirculation was then imaged using SLO. C) Representative micrographs from SLO imaging of the funds of normal and diabetic rats on chow or WD for 2 wk. Scale bar, 250 µm. D) Quantitation of the number of αICAM-1 per retina in diabetic animals and nondiabetic animals. *P < 0.05, **P < 0.01 (n = 6).

Differential role of dietary fats in retinal microvascular injury in STZ-induced diabetes

To isolate the effects of the high amounts of saturated fat and cholesterol in WD on the development of inflammatory retinal endothelial injury in the STZ-diabetic rats, control and diabetic rats were kept on either chow or each of our semipurified diets, with or without added cholesterol, for 2 wk (Table 1). Retinal leukocyte accumulations in these rats were quantified in histologic flat mounts after ConA perfusion (Fig. 4A, B). Endothelial injury was quantified using PI staining (Fig. 4C, D). Microvascular injury, as determined by leukocyte accumulation and the number of PI⁺ endothelial cells, was significantly higher in diabetic rats compared with normal controls. Cholesterol-free diet A, with its fatty acid balance similar to chow and macronutrient balance similar to WD, did not affect the number of firmly adhering leukocytes or injured endothelial cells in the STZ-diabetic rats, which remained true after adding a high level of cholesterol (0.1%). This was also true for cholesterol-free, high-saturated-fat diet B. Only the STZ-diabetic rats fed diet B with high levels of added cholesterol (0.1%) showed an increase in leukocyte accumulation and the number of PI⁺ endothelial cells (Fig. 4C, D).

Figure 4.

Retinal leukostasis and endothelial damage in animals on semipurified diets. Normal and STZ-induced diabetic animals were fed the indicated semipurified diets for 2 wk. The diets were designed to be high in fats, mimicking the composition of plant-based and animal-based sources with and without added cholesterol (0.1% of the total diet). At the end of the period, animals were perfused with FITC-conjugated ConA lectin; subsequently, retinal flat mounts were generated for histologic examinations. A) Quantitation of the number of firmly adhering leukocytes. B) Composite micrographs of representative sections of retinas from indicated groups. C) PI-stained injured or dead endothelial cells per retina in STZ-induced diabetic rats and nondiabetic controls fed indicated diets. D) Representative micrographs showing PI-positive cells in retinal flat mounts. Scale bar, 200 µm. *P < 0.05 (n = 7–14).

Retinal NF-κB signaling in STZ-diabetic mice

NF-κB signaling regulates expression of endothelial ICAM-1 (33). We performed Western blotting and found that phospho–NF-κB (p65) was higher in the retinas of diabetic mice fed chow and further increased in diabetic mice fed WD (Fig. 5A). To examine the contribution of NF-κB to leukocyte adhesion, we analyzed leukocyte accumulation in retinas of the Tie1ΔN mice, which lacked endothelial NF-κB signaling. As established above, WD exacerbated retinal leukocyte accumulation in STZ-induced diabetes. In the STZ-diabetic Tie1ΔN mice, however, WD did not cause an increase in leukocyte accumulation (Fig. 5B, C).

Figure 5.

Examination of NF-κB–mediated retinal inflammation in WD-fed diabetic animals. A) Representative Western blots of retinas in chow- or WD-fed diabetic mice and nondiabetic mice for phospho–NF-κB, NF-κB, phospho-PKC, and β-tubulin. B) Representative pictures of adhered leukocytes with ConA staining (green) in diabetic wild-type or Tie1ΔN mice 2 wk after induction of standard chow or WD. Arrows indicate adhered leukocytes. Scale bar, 100 µm. C) Quantitation of the number of adhered leukocytes per retina in diabetic wild-type or Tie1ΔN mice 2 wk after induction of standard chow or WD. **P < 0.01 (n = 8).

DISCUSSION

Hyperglycemia is a common symptom of T1D and T2D, however, the diseases differ in their etiology and pathogenesis. T1D is caused by autoimmune destruction of the pancreatic β cells, leading to insulin deficiency, whereas T2D is characterized by a defect in, or resistance to, insulin action primarily on muscle and adipose tissue. The greater the level of insulin resistance, the more insulin needs to be secreted to maintain glucose homeostasis. It has been argued that T2D only evolves in a relatively small portion of insulin-resistant individuals whose β cells are unable to adequately compensate with increased insulin secretion to prevent gross decompensation of glucose tolerance (34–36). T1D and T2D are quite different syndromes in many other respects, including their impact on lipoprotein metabolism, which insulin plays a central role in regulating in a not-yet-fully understood manner.

STZ administered at toxically high levels in animals ultimately leads to almost complete ablation of β cells, causing a severe deficiency in insulin production, leading to stable hyperglycemia and weight loss and thus experimentally reproducing the main symptoms of T1D (37). T1D and T2D are associated with microvascular complications, including retinopathy, nephropathy, and neuropathy. Because these complications are often clinically observed in patients with diabetes after long exposure to elevated BG concentrations, the study of the pathogenesis of these complications in animal models has traditionally put less emphasis on the manner in which the animals become hyperglycemic. However, there is mounting evidence that ambient insulin levels and plasma lipoprotein kinetics and concentrations play a role in the pathogenesis of vascular diabetes complications. Models of DR include STZ-induced models of hyperglycemia as well as spontaneous T1D and T2D models. Because of the robustness of the STZ-induced hyperglycemia, it is used in most mechanistic studies that are focused on glucose-mediated microvascular damage in DR. However, in this model, the contribution to DR of other components of the complex T1D pathology, including insulin deficiency, hypertriglyceridemia, and other lipoprotein abnormalities, is far less understood.

We introduced the Nile grass rat (NGR) as a spontaneous nutritional model of T2D (38, 39). We established the development of DR in the NGR and found several key differences between DR pathologies in this model and the STZ-diabetic rat model (40). Notably, although in the STZ-rat model retinal leukocyte accumulation increases within days of STZ administration, remains unchanged in degree with the duration of hyperglycemia, and correlates with the increased expression of retinal ICAM-1 (5), we found in the NGR retina an initial increase and subsequent decrease in both ICAM-1 and leukocyte accumulation that are associated with declining insulin levels and rising hyperglycemia (40). Furthermore, as opposed to the STZ model, most leukocytes accumulated in the retinal arteries of diabetic NGRs vs. in veins (41). Interestingly, there is also an inverse correlation between fasting plasma TG levels and arterial leukocyte accumulation in the NGR retinas. Leukocyte adhesion and elevated ICAM-1 expression have been documented in retinal vessels of deceased patients with T1D and T2D (1), but their levels during, and the contribution to, the evolution of human DR have not been fully examined and remain unclear. In the STZ-diabetic rat model, adherent leukocytes were shown to cause capillary occlusion, endothelial cell injury and death, and blood-retina barrier breakdown; VEGF was shown to increase retinal vessel permeability and capillary nonperfusion, in part through its up-regulation of ICAM-1 and subsequent leukocyte adhesion (4, 5, 42, 43).

As previously discussed, hypertriglyceridemia in the STZ-diabetic rat is secondary to a defect in the removal of both chylomicrons and plasma VLDL-TGs, despite a reduction in hepatic VLDL-TG secretion, and is accentuated by a high-fat diet (8). In the current work, we set out to study the effect of diet macronutrient composition and fatty acid balance and type on early DR pathology in the insulin-deficient, hyperglycemic, and hypertriglyceridemic STZ-rat model.

STZ-diabetic and control rats were fed normal laboratory rat chow or WD. Both diets provided the same number of energy calories per gram but differed in their macronutrient composition and fatty acid balance and type. The rat chow is a low-fat, high-starch diet, with a CHO:fat:protein %en of 58:13:29, an SFA:MUFA:PUFA %en of roughly 1:1:1, and is 0.03% cholesterol. The WD is high in both sugar (32%) and milk fat (21%), with a CHO:fat:protein %en of 44:41:15, an SFA:MUFA:PUFA %en of roughly 13:5:1, and is 0.2% cholesterol. The WD significantly increased plasma TGs and TCs in control and diabetic rats. Plasma FFAs were significantly increased in diabetic rats fed WD but not in control rats fed WD, and plasma adiponectin was significantly reduced in diabetic rats fed chow and WD. The numbers of accumulated leukocytes and injured endothelial cells were significantly higher in the retinal vessels of diabetic rats fed WD compared with diabetic rats fed chow and control rats fed WD.

For in vivo visualization of leukocyte recruitment and accumulation in retinal microvessels, we developed fluorescent leukocyte-mimicking nanoprobes that targeted the endothelial adhesion molecule ICAM-1 for adhesion and endothelial P-selectin for initiation of rolling (21). Imaging nanoprobes injected into the bloodstream of live animals circulated throughout the animals’ vasculatures, including the retinas. When excited in light-based fundus imaging, the nanoprobes in the retinal microvessels emit a fluorescent signal that can be detected in vivo. The probes’ interactions with the inner vessel wall, mediated by single molecular interactions, provide an unprecedented temporal and spatial resolution that gives precise knowledge about the presence of ICAM-1 in the retinal microvessels. Our in vivo molecular imaging technique showed higher expressions of ICAM-1 in the retinal microvessels of diabetic rats fed chow. WD further increased ICAM-1 in diabetic rats, which could explain the higher numbers of leukocyte accumulation and endothelial injury in diabetic animals fed WD. Future work will address the mechanisms by which nutrients in the diet modulate surface expression of ICAM-1 and other adhesion molecules on vascular endothelial cells, where they likely exert most of their physiologic role.

As noted above, the fat content in laboratory chow and WD has very different fatty acid composition, namely 1:1:1 vs. 13:5:1 (SFA:MUFA:PUFA %en). In chow, the CHOs consisted mostly of starch, as opposed to mostly sugar in the WD. Studies in humans have demonstrated that fasting TG levels were lowered by unsaturated fats and raised by saturated fat, irrespective of the source of CHO in the diet (44). Furthermore, in STZ-diabetic rats as well as in patients with T1D, cholesterol absorption is elevated and synthesis down-regulated (45, 46). These observations led us to hypothesize that, irrespective of the sugar content of the rat diet (i.e., whether the CHOs consisted of mostly starch or sugar), a high-fat diet with a high ratio of saturated to unsaturated fat, as is found in milk and other animal fat, together with the addition of a large amount of cholesterol to the diet, significantly increases the levels of plasma cholesterol and fasting TGs in rats. Furthermore, we hypothesized that such a diet high in saturated fat and cholesterol in STZ-diabetic rats, with their defect in chylomicron and VLDL-TG removal, would have an adverse effect on DR development in those rats.

To test the above hypothesis on the role of fatty acid balance and dietary cholesterol in the development of inflammatory retinal endothelial injury in the STZ-diabetic rats, we generated 2 semipurified cholesterol-free, high-fat diets with varying fatty acid balances. These diets provide the same number of energy calories per gram as the WD. Although they have a macronutrient composition of 43:40:17 (CHO:Fat:Protein %en) similar to WD, unlike the WD, they contain no sugar. One diet, diet A, has a fatty acid balance of 1:1:1 (SFA:MUFA:PUFA %en), similar to chow. The other diet, diet B, is a high saturated-fat diet with fatty acid balance of 35:4:2 (SFA:MUFA:PUFA %en), containing mostly C-12:0 lauric, C-14:0 myristic, C-16:0 palmitic acids, and just enough PUFA (mostly C-18:2 linoleic acid) to prevent essential fatty acid deficiency. These 3 SFAs account for the majority of fat intake in human WDs and come primarily from high dairy and meat consumption in that diet. As noted above, most of the fat content of laboratory WD comes from milk fat, which is high in myristic and palmitic acids. We formed 10 groups of control and diabetic rats that we fed either rat chow, diet A, or diet B, or each of the latter with added 0.1% cholesterol. Neither diet A, diet B, or diet A with added cholesterol increased leukocyte accumulation or endothelial injury in retinal vessels. Only in the diabetic group fed diet B with added cholesterol was there a significant increase in retinal leukocyte accumulation and endothelial injury.

Dietary CHOs and dietary fats have related and overlapping metabolic effects. It is well-known that dietary sugar (sucrose) raises the concentrations of plasma TGs in both humans and rats (47). In normal rats, the tendency of sugar to elevate plasma TG levels is caused by increased hepatic lipogenesis, without a parallel increase in plasma TG removal (48, 49). Furthermore, it has been found that the presence of saturated fat in a high-sugar diet of normal rats further elevates their plasma TG levels, despite dietary fat having a depressant effect on hepatic lipogenic enzymes, irrespective of its degree of saturation (48, 49). Thus, the higher plasma TG levels found in normal rats fed saturated fat seem to be caused by a difference in chylomicron and VLDL-TG clearance rather than in hepatic fat synthesis. It has also been shown that lipoprotein lipase activity in adipose tissue is higher in rats fed unsaturated fat compared with those fed saturated fat (50). This would account for the additive influence of saturated fat and sugar on plasma TG levels. The insulin-deficient, hypertriglyceridemic STZ-diabetic rat already exhibits a defect in chylomicron and VLDL-TG removal as well as depressed hepatic VLDL-TG synthesis and secretion. It is then plausible to speculate that an additive action of saturated fat and sugar in the diet would further exacerbate early endothelial injury in the retinal vessels of this STZ-induced T1D model. This hypothesis, along with mechanisms of the additive action of saturated fat and sugar on the retinal vessel walls and the incitement of an inflammatory response under conditions of hypertriglyceridemia and insulin-glucose dysregulation, will be addressed in our future work.

Finally, we found elevated levels of the proinflammatory signaling molecules phospho–NF-κB and phospho-PKC in the STZ-diabetic rats fed chow and WD. This led us to examine the role of endothelial NF-κB in the WD-induced diabetic leukocyte accumulation. In transgenic mice that lack NF-κB signaling in the vascular endothelium, we found no WD-induced endothelial injury in STZ-induced diabetes. When compared to the diabetic wild-type mice fed WD, the endothelial-specific NF-κB signaling–deficient, diabetic mice fed WD showed significantly lower numbers of leukocyte accumulation.

CONCLUSIONS

Taken together, our results suggest that in the STZ-diabetic rat model, a high-fat diet per se, irrespective of the degree of fat saturation, does not necessarily exacerbate retinal endothelial injury, and that the components of the WD eliciting an endothelial inflammatory response in this model do so through NF-κB signaling. In particular, a low-sugar, high-fat, plant-based diet, which is naturally cholesterol-free, would have a neutral effect on DR development in the insulin-deficient, hyperglycemic, and hypertriglyceridemic STZ-rat model.

Our findings cannot be prematurely extrapolated to patients with T1D, most of whom are receiving insulin replacement therapy and presenting with a different lipoprotein profile than the insulin-deficient STZ-diabetic rat. In general, the higher the plasma FFA concentration, the greater the hepatic VLDL-TG synthesis and secretion. However, insulin-treated patients with T1D tend to have plasma TG concentration that is within the normal range despite elevated FFA concentrations. As noted above, this can be explained through both the observation that hepatic VLDL-TG secretion is not increased in those patients and the fact that the defect in chylomicron and VLDL removal is sensitive to insulin replacement. Further, plasma FFA levels can be low when intensive glycemic control is achieved in these individuals, and under these conditions, plasma TG concentrations are also low.

It is noteworthy that the effects of dietary fats and CHOs on blood lipids vary considerably in humans, depending at least on ambient plasma insulin levels and the extent of a present defect in insulin action, and in turn the effect of those on lipoprotein metabolism (6). This is also true for proteins because they are converted into amino acids in the intestinal track and can stimulate insulin secretion. In particular, results on the effect of the amounts of dietary macronutrients (as a percent of energy) on blood lipids cannot be extrapolated from healthy individuals to patients with diabetes, or from insulin-resistant individuals or patients with T2D to patients with T1D, or vice versa. In addition, the balance and type of dietary fatty acids exert differential effects on an individual’s lipoprotein profile and metabolic markers.

At present, there is only limited understanding of the relationship between insulin level and action, and lipoprotein kinetics and concentrations. There is even lesser understanding of how this relationship may be modulated by the profile of macronutrients in the diet. Furthermore, the metabolic environment induces a complex dynamic between the level and actions of insulin and other hormones. For example, the secretin of the adipocyte hormone adiponectin is significantly altered in relation to insulin level and action, and this can, in turn, alter insulin sensitivity through modulating glucose uptake in skeletal muscle as well as fatty acid oxidation in liver and skeletal muscle. The relationship between the dysregulation of insulin and other interrelated hormones in vascular diabetes complications is largely unknown. Given the potential importance of hormonal activities and lipoprotein metabolism in the development of vascular endothelial dysfunction in individuals with disordered insulin-glucose homeostasis, and how these may be affected by diet, these areas of research warrant further systematic development.

Glossary

%en

energy percentage

BG

blood glucose

BW

body weight

CHO

carbohydrate

ConA

concanavalin A

DR

diabetic retinopathy

FFA

free fatty acid

HbA1c

hemoglobin A1c

ICAM-1

intercellular adhesion molecule-1

LDL

low density lipoprotein

MUFA

monounsaturated fatty acid

NGR

Nile grass rat

phospho

phosphorylated

PI

propidium iodide

PUFA

polyunsaturated fatty acid

SFA

saturated fatty acid

SLO

scanning laser ophthalmoscope

STZ

streptozotocin

T1D

type 1 diabetes

T2D

type 2 diabetes

TC

total cholesterol

TG

triglyceride

Tie1∆N

Tie-1–Cre/IκB-αΔN

VLDL

very low density lipoprotein

WD

Western diet

AUTHOR CONTRIBUTIONS

A. Barakat, K. C. Hayes, and A. Hafezi-Moghadam designed research; S. Nakao, S. Zandi, and D. Sun performed research and analyzed data; R. Schmidt-Ullrich provided unique reagents; K. C. Hayes generated semipurified diets; and A. Barakat and A. Hafezi-Moghadam wrote the manuscript.