Strain-Dependent Increases in Retinal Inflammatory Proteins and Photoreceptor FGF-2 Expression in Streptozotocin-Induced Diabetic Rats

Abstract

Purpose.: Inflammation is thought to play a role in disease progression and vision loss in diabetic retinopathy (DR). However, the level of inflammation and the role of cytokines and growth factors in the early stages of this disease are poorly understood. Streptozotocin (STZ)-induced hyperglycemia in rats is widely used as a model of diabetic retinopathy, and therefore this model was used to better define the inflammatory response and the impact of the genetic background.

Methods.: The expression of a panel of 57 inflammatory proteins and growth factors in the retina of three rat strains was compared by using a highly sensitive flow cytometry–based assay. Hyperglycemia was induced in Brown Norway (BN), Long-Evans (LE), and Sprague-Dawley (SD) rats, and protein expression in the retina was measured 4 weeks and 3 months later.

Results.: The data revealed a subtle, but reproducible, inflammatory response in the retina of SD, but not in those of BN or LE, rats. Upregulation of fibroblast growth factor (FGF)-2 in the photoreceptor nuclear layer coincided with the inflammatory response in SD rats and may constitute a neuroprotective mechanism. Reduced expression of genes involved in the phototransduction pathway indicates altered photoreceptor function.

Conclusions.: Taken together, these data show that inflammatory changes in the diabetic rat retina are highly strain dependent, and SD rats exhibit low-level inflammation similar to that observed in diabetic patients. Therefore, SD rats may be a good model for the study of early inflammatory changes in human diabetic retinopathy.

Diabetic retinopathy (DR) is the number one cause of blindness in working-age people in the United States today, and its prevalence is expected to rise, with a projected increase in the number of patients with type II diabetes. Currently, the only available treatment for DR is laser photocoagulation, and although it is highly effective in preserving sight, it can have serious side effects. ¹ Furthermore, this treatment is efficacious only in proliferative DR and cannot restore vision already lost to the disease. ² Most patients with diabetes develop retinopathy at some point, and the only proven method of slowing its progression is by tightly controlling blood glucose levels. ¹ Despite significant morbidity and effect on quality of life, there is currently no approved pharmacotherapy for this sight-threatening disease. ^1,3  

In the past, DR was considered primarily a vascular disease. It has been described as and is still classified by abnormalities of the retinal vasculature that progress from hyperpermeability of the vessels, to capillary nonperfusion and hemorrhaging, followed by neovascularization of the retina. Macular edema can develop independently of other changes and poses a serious threat to central vision. ^1,2 However, there is mounting evidence that neuropathy develops before and independent of vascular complications and contributes to the disease. ⁴ Inflammatory molecules such as cytokines can be found in the retina of diabetic patients, ⁵ although it is presently unclear whether inflammation contributes to DR or is secondary to it. 

Streptozotocin (STZ)-induced diabetes in the rat is commonly used as an experimental model of DR. The STZ rat model mimics the human disease by inducing hyperglycemia after destruction of the β-cells in the pancreas. ⁶ Although there are vascular changes in this model, the vasculopathy does not progress to neovascularization as is observed in humans. The length of time that passes between the beginning of the hyperglycemic insult and the development of symptoms is also a challenge as are the number of different mechanisms that may cause damage. ⁷ In addition, inflammatory changes are subtle and therefore sometimes difficult to demonstrate reproducibly. Further complicating matters is the fact that different rat strains show different responses to hyperglycemia and other insults to the retina. ^8–10  

In this study, we characterized the inflammatory changes in the retina of diabetic rats, while exploring the effect of genetic background, including pigmentation, on such changes. A highly sensitive flow cytometry-based assay (Luminex Corp., Austin, TX) was used to screen for a large number of cytokines and growth factors in the retina of diabetic Brown Norway (BN), Long-Evans (LE), and Sprague-Dawley (SD) rats. These rat strains were chosen due to their documented use in the STZ-induced DR model and their various levels of pigmentation, with SD being the least pigmented and BN the most pigmented rat strain. To date LE and BN rats have been used mostly to study vascular abnormalities in diabetes such as increased leukostasis ^11–16 and blood–retina barrier breakdown. ^{9,14,15,17–21} However, electrophysiological studies indicate functional deficits in the neural retina ^22–24 and results from studies on γ-aminobutyric acid signaling ^24,25 and microarray data in LE ²⁶ as well as Müller cell abnormalities in BN rats ²⁷ suggest that other hallmarks of experimental DR occur and that these strains merit further investigation. 

The data show that despite the absence of gross inflammatory infiltrates in the retina, markers of inflammation are present in SD rats, but not the other two strains. Furthermore, the data show that the cytokine response coincides with upregulation of fibroblast growth factor (FGF-2) in photoreceptors, and gene expression analysis suggests impairment of photoreceptor function. 

Materials and Methods

Animals and Induction of Diabetes

Diabetes was induced in male BN, LE, and SD rats (Charles River Laboratories, Wilmington, MA) weighing 250 to 300 g (BN: 244 ± 19 g; LE: 298 ± 25 g; SD: 282 ± 26 g) by intraperitoneal injection of 65 mg/kg bodyweight STZ in 0.09 M citrate buffer (pH 4.8; both Sigma-Aldrich, St. Louis, MO). Age-matched control rats were injected with buffer only. At the time of injection, BN rats were moderately, but significantly (P < 0.05) smaller then LE or SD rats. Blood glucose was measured (Ascencia control system; Bayer, Tarrytown, NY), and rats with blood glucose levels above 220 mg/dL after 48 hours were deemed diabetic. Nondiabetic STZ-injected rats were reinjected with a second dose on day 2. ^28–30 There was no significant difference in weight or blood glucose levels 6 days after the initial injection between rats injected once and those receiving a second dose. No insulin was supplied at any time. Occasionally, glucose levels exceeded the upper limit of detection, and this unknown value was replaced by 600 mg/dL. Glucose spikes to these high levels were relatively rare and not limited to any particular rat strain. Intraocular blood glucose was measured by making a 4- to 6-mm diagonal incision across the cornea, removing the lens and then sampling the liquid in the eye cup using the capillary action of the control test stick (Ascencia; Bayer). The sample most likely contains a mixture of aqueous and vitreous humor. Animals were killed 4 weeks and 3 months after induction of diabetes. They were housed in plastic box cages in a specific pathogen-free, AALAC certified facility on a 12-hour light/dark cycle. Room temperature was maintained between 16°C and 23°C, average daily humidity between 30% and 70%, and airflow between 10 and 30 air changes per hour. All animals were fed standard chow and treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Protein Analysis.

Isolated neural retinas were snap frozen in liquid nitrogen and samples were stored at −70°C until processing. One retina each from three individual rats was combined for protein analysis. Protein was extracted by sonication of three pooled retinas in 300 μL Tris HCl (pH 7.5; Mediatech, Inc. Herndon, VA) containing Complete, Mini protease inhibitor cocktail (Roche, Mannheim, Germany). Serum was collected by letting blood clot for 1 hour at room temperature, followed by a 10-minute centrifugation at 10,000g. Protein concentration was then determined (RodentMA Antigen 1.6 panel; Rules Based Medicine Austin, TX). The assay uses mouse antibodies that cross-react with rat cytokines. ^31,32 Briefly, samples are incubated with microsphere multiplexes, which contain covalently linked assay specific capture reagents, for 1 hour at room temperature before biotinylated reporter antibodies are added for an additional hour at room temperature. After incubation for 1 hour with streptavidin-phycoerythrin solution and vacuum filtration, the samples are diluted in matrix buffer and analyzed with a multiplex bio-assay analyzer (100 IS System; Luminex Corp.). 

Histopathology and Immunohistochemistry.

Rats were killed after 3 months of diabetes, and the eyes were immediately fixed in Davidson's solution (BBC Biochemical, Mount Vernon, WA) for 24 hours. After fixation, the eyes were transferred to 70% alcohol, processed routinely for paraffin embedding, and sectioned at 4 μm. For routine histopathologic evaluation the sections were stained with hematoxylin and eosin. For immunohistochemistry, paraffin-embedded sections were placed on charged slides and dried overnight at room temperature. After deparaffination and rehydration, slides were placed in citrate plus buffer (Scytek, Logan, UT) and brought to 125°C and 10 to 15 psi for 30 seconds in a Pascal pressure cooker, for antigen retrieval, followed by washing in phosphate-buffered saline Tween (PBST) buffer (Scytek). Immunohistochemistry staining was performed at room temperature with an autostainer (Dako, Carpinteria, CA). Tissue slides were treated for 10 minutes with 3% hydrogen peroxide (VWR, West Chester, PA), washed briefly with distilled water, treated with a blocking agent (SuperBlock; Scytek) for 10 minutes, then reacted with mouse-anti-FGF-2 antibody (Clone bFM-2; Millipore, Billerica, CA) at 1:200 dilution for 60 minutes. After they were washed with PBST, the tissues were reacted with biotinylated donkey anti-mouse antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) for 20 minutes, washed with PBST, and incubated for 20 minutes with horseradish-peroxidase streptavidin (Covance, Dedham, MA) and rinsed in PBST. Signal was detected with AEC chromogen (Romulin; BioCare Medical, Concord, CA), then washed with PBST. Hematoxylin was applied for 5 minutes followed by a rinse with distilled water. 

RNA Extraction and Real-Time PCR

Retinas were removed and placed individually into tubes containing 200 μL RNA stabilizer (RNAlater; Ambion, Austin, TX). After 24 hours at 4°C, excess fluid was removed and the samples were stored at −70°C until analysis. Frozen samples were placed into 1 mL lysis reagent (Qiazol; Qiagen, Valencia, CA) and homogenized (Tissue-Tearor; Biospec, Bartlesville, OK). RNA was extracted (RNeasy Mini Kit; Qiagen) according to the manufacturers' instructions. RNA quality was then assessed (RNA 6000 Nano Kit; Agilent Technologies, Waldbronn, Germany) and quantity determined by spectrophotometry. Reverse transcription was performed on 1 μg of RNA using 25 mM MgCl₂, 10 mM dNTP mixture, 25 U RNasin, and 15 U AMV reverse transcriptase in reverse transcription buffer (all Promega, Madison, WI) in 20 μL for 20 minutes at 42°C, followed by a 5-minute denaturing step at 95°C and 5 minutes at 4°C. After reverse transcription, ddH₂O was added to a final volume of 200 μL. 

Real-time PCR was performed (7900HT Real-Time PCR System; Applied Biosystems [ABI], Foster City, CA), with probe and primer sets (Taqman; ABI) for Actb (ID: Rn00667869_m1), Arr3 (ID: Rn01424383_m1), Gnb3 (Rn00516381_m1), Grk1 (ID: Rn00579470_m1), and Opn1mw (ID: Rn00585560_m1). All primers span exon junctions and therefore do not recognize genomic DNA. Gene expression was normalized and converted to a linearized value by the following formula: unit = 1.8 ⋀ (Ct_actb − Ct_{gene x} ) × 100. ³³  

Statistical Analysis

Experiments analyzing retinal proteins were repeated at least three times, each group containing three rats, making data the average of at least nine rats from at least three independent experiments. Significance was determined by Student's t-test, unless otherwise noted. P < 0.05 was considered significant. 

Results

STZ-Induced Hyperglycemia in BN, LE, and SD Rats

Diabetes was induced with STZ in three rat strains: BN, LE, and SD. Over the 3-month observation period, average blood glucose levels ranged from 330 to 542 mg/dL in the three rat strains (Fig. 1A), similar to previously published results in this model. ^34–36 In contrast, vehicle-treated rats exhibited tightly controlled blood glucose levels in all strains (63–89 mg/dL; Fig. 1A). Intraocular levels were also measured to determine glucose concentration adjacent to the neural retina. Intraocular levels were similar to those in blood in diabetic and control animals (Fig. 1B), suggesting that ocular humor is a major route of retinal glucose exposure in diabetes. Of note, SD rats showed moderately higher levels of intraocular glucose than did the other two strains, a difference that reached statistical significance (P < 0.05). Glucose levels in the blood of diabetic animals, however, did not show consistent significant differences over time between the three strains. Control BN, LE, and SD rats gained weight during the 3 months of observation, however, the BN strain showed significantly less weight gain. In comparison, hyperglycemic rats of all strains failed to gain weight over the same period (Fig. 1C). These data show that STZ induces hyperglycemia in blood and the eye in all three rat strains, and therefore indicate that any strain-specific molecular or functional changes in the retina are not due to large differences in glucose exposure. 

Figure 1.

 

Body weight, blood glucose, and vitreous glucose levels in BN, LE, and SD rats. (A) Blood glucose levels in rats of different strains injected with STZ (solid line, filled symbols) or buffer only (broken line, open symbols) at multiple time points after injection. (B) Ocular glucose levels at the time of death measured in the eye cup after lens removal. (C) Body weight of rats with induced diabetes and control rats. Data are the average of results in at least three experiments with three rats per group (n > 9). Error bars, SD.

Protein Measurements in Normal and Diabetic Rat Retinas

A modified bio-assay (Luminex) was used to measure the levels of 57 proteins of interest in retinas of normal and STZ-treated BN, LE, and SD rats 1 month and 3 months after diabetes induction. Of these 57 proteins, 24 were below detection limits in normal and STZ-treated retinas, and the concentration of 22 proteins, though present at detectable levels, was not significantly changed by STZ treatment (Table 1). 

Table 1.

 

Proteins Not Significantly Changed in Diabetic Retina

Detectable (but unchanged)	Undetectable
Apolipoprotein A1	CD40,CD40L
Calbindin	EGF
Endothelin-1	GCP-2
Factor VII	GM-CSF
GSTμ	Growth Hormone
IgA	GSTα
IL−1α,−1β,−2,−5,−10	Interferon−γ
Leukemia inhibitory factor	IL−3,−4,−7,−11,−12,−17
MIP−1α,−3β	KC/Groα
Myeloperoxidase	Leptin
Myoglobin	Lymphotactin
NGAL	MCP-5
Stem Cell Factor	MIP−1β,−1γ,−2
SGOT	MMP-9
Tissue Factor	Oncostatin M
Thrombopoietin	RANTES
VCAM

 Proteins in the retina were not significantly affected by hyperglycemia. Complete panel of proteins tested in all strains at 4 weeks and three months,which did not show changes or were below the detectable level in the retina.

 Abbreviations: GST, glutathione transferase; Ig, immunoglobulin; IL, interleukin; MIP, macrophage inflammatory protein; NGAL, neutrophil gelatinase-associated lipocalin; SGOT, serum glutamic oxaloacetic transaminase; VCAM, vascular cell adhesion molecule; EGF, epidermal growth factor; GCP, granulocyte chemotactic protein; GM-CSF, granulocyte macrophage colony-stimulating factor; MMP, matrix metalloproteinase; RANTES, regulated upon activation, normal T-cell expressed and secreted.

Hyperglycemia-Induced Retinal Inflammation in SD Rats

Several markers of inflammation have been reported in the neural retina in experimental models of diabetes ^37,38 and in diabetic human vitreous ^39–41 and tissues. ^42–44 Although there was no evidence of a measurable inflammatory response in diabetic LE and BN rats, SD rats did show a time-dependent increase in several inflammatory proteins. The chemokine eotaxin (CCL11) was significantly increased in diabetic SD retina after 1 month to 1.6 times the level in control retina. Although the same trend could be observed at 3 months, this change was not statistically significant (P = 0.052; Fig. 2A). Eotaxin was also significantly upregulated in serum after 3 months of diabetes (Table 2), as well as at 4 weeks (data not shown) in SD rats. Macrophage colony-stimulating factor (M-CSF), monocyte chemoattractant protein-1 (MCP-1; CCL-2), and MCP-3 (CCL-7) which act on monocytes and macrophages, were significantly upregulated after 3 months of diabetes in SD rats (Figs. 2B–D). Although protein levels of these cytokines appeared to be low compared with their levels in overtly inflammatory diseases, the increase is statistically significant and highly reproducible. 

Figure 2.

 

Retinal cytokine protein expression at 28 and 84 days of diabetes and in age-matched control rats. The concentration of eotaxin (A) was increased in SD rats after 4 weeks of diabetes. Protein levels of M-CSF (B), MCP-1 (C), and MCP-3 (D) were increased in SD rats after 3 months of diabetes. Protein expression was measured in three independent experiments. Each symbol represents data from three pooled diabetic retinas of three individual diabetic or control rats. Statistical significance was determined with a Student's t-test, and significant probabilities are noted.

Table 2.

 

Serum Protein Levels

	Diabetic	Control
B2M (μg/mL)	69 (±14)	65 (±11)
Clusterin (μg/mL)	183 (±37)*	118 (±30)
Eotaxin (pg/mL)	1401 (±343)*	811 (±126)
FGF-9 (ng/mL)	0.3 (±0.7)	0.4 (±0.5)
FGF-2 (ng/mL)	2.1 (±0.1)	2.3 (±0.2)
MCP-1 (pg/mL)	696 (±290)	846 (±91)
MCP-3 (pg/mL)	386 (±128)	468 (±61)
M-CSF (ng/mL)	0.8 (±0.1)	0.8 (±0.1)
TIMP-1 (ng/mL)	4.4 (±4.7)	3.3 (±3.5)
VEGF (pg/mL)	238 (±44)	211 (±70)
vWF (ng/mL)	532 (±165)*	195 (±44)

 Levels of proteins (±SD) significantly changed in the retina of diabetic rats. Serum was collected from SD rats after 3 months of diabetes and protein levels determined. Results are average of three separate samples each from two independent experiments (n = 6).

 *P < 0.05 compared with control.

Further evidence of inflammation in the retina of SD rats was the upregulation of clusterin after 4 weeks and 3 months of diabetes (Fig. 3A). These results confirm published reports of increased clusterin expression in the diabetic SD rat, ⁴⁵ but show the observation to be highly strain dependent. Tissue inhibitor of metalloproteinase (TIMP)-1 showed a similarly moderate, but significant upregulation after 4 weeks of diabetes and more robust changes after 3 months (Fig. 3B). Although this protein was increased in all SD diabetic samples compared with the control, the concentration in diabetic retinas varied widely between different experiments. Therefore, the Mann-Whitney U test was used to compare the median values between the diabetic and control group, showing a significant increase. In addition, β-2 microglobulin (B2M; Fig. 3C) and von Willebrand factor (vWF; Fig. 3D) were significantly upregulated after 3 months of diabetes, further indicating an ongoing inflammatory response. Clusterin and vWF also showed significantly increased levels in the serum of SD rats at 3 months, whereas none of the other proteins that were significantly changed in the retina were altered in the serum at this time (Table 2). These data indicate subtle, but highly reproducible and strain-dependent, inflammatory protein increases in the retinas of diabetic rats. 

Figure 3.

 

Expression of inflammatory proteins in the diabetic rat retina. Concentrations of proteins upregulated during inflammation were measured in the retina as in Fig. 2. In SD rats, both clusterin (A) and TIMP-1 (B) were upregulated 4 weeks and 3 months after the onset of diabetes. B2M (C) and vWF (D) were increased after 3 months of diabetes in SD rats. Each symbol represents data from three pooled diabetic or control rats. A Student's t-test was used to determine statistical significance. *Mann-Whitney U test.

Changes in Growth Factor Expression in the Retinas of Diabetic Rats

Growth factors, particularly those with angiogenic effects, are of interest because of their role in proliferative diabetic retinopathy. ⁴⁶ In BN rats, there was a small but significant increase in vascular endothelial growth factor (VEGF) protein expression in the retina after 4 weeks of diabetes. However, a significant increase was not observed after 3 months. Although there was a trend toward increased VEGF levels in LE and SD rats at 4 weeks, it did not reach statistical significance (P = 0.115 and P = 0.057 respectively; Fig. 4A). FGF-9 exhibited the same temporal pattern as VEGF, though induction of this growth factor was of a greater magnitude in BN rats. By 3 months of diabetes, FGF-9 levels in the retina of diabetic rats were similar to that in control rats (Fig. 4B). In contrast, and confirming previously published data, ⁴⁷ FGF-2 was significantly increased in the retina of diabetic SD rats after 3 months. This increase in FGF-2 protein was observed only in SD rats and, although measurable, FGF-2 levels were unchanged in BN or LE rats (Fig. 4C). 

Figure 4.

 

Changes in growth factor expression in the hyperglycemic rat retina. (A) A small, but highly reproducible increase in VEGF expression was found in BN rats after 4 weeks of diabetes. (B) FGF-9 was also increased in BN after 1 month, but not significantly changed in any other strain or time point. (C) Protein levels of FGF-2 were elevated in SD rats after 3 months of diabetes. Each symbol represents data from three pooled diabetic or control retinas. Statistical significance was determined using Student's t-test.

Histopathologic Changes after 3 Months of Diabetes

No gross disease was found in any of the rat strains by histologic examination (Figs. 5A–F and Ref. 48), and there was no change in retinal thickness after 3 months of diabetes (data not shown). Despite an increase in inflammatory proteins in SD rats, no inflammatory infiltrate was detected in the retinas of any strain (Figs. 5A–F). Staining with specific antibodies also revealed no evidence of macrophages, CD4 T cells, CD8 T cells, or B cells in either control or diabetic retinas (data not shown). Immunohistochemistry for FGF-2 localized the protein to the inner nuclear layer (INL) in the retina of control and diabetic SD rats, and showed a dramatic increase in FGF-2 expression in the photoreceptor nuclear layer (PNL) after 3 months of diabetes. These data suggest subtle inflammatory changes in the absence of peripheral immune cells in the retina of diabetic SD rats that coincide with FGF-2 upregulation in photoreceptors. 

Figure 5.

 

H&E stain of the retina and localization of FGF-2. H&E staining of retinas from diabetic (A, C, E) and control (B, D, F) rats after 3 months of diabetes in BN (A, B), LE (C, D), and SD (E, F) showed no inflammatory infiltrates or changes in retinal thickness in any of the sections. FGF-2 immunostaining reveals an increase in this protein localized to photoreceptor nuclear layer (PNL) in diabetic SD rats (H) compared with control animals (G). Results are representative of two independent experiments with three rats per group.

Effect of Hyperglycemia on Gene Expression in Photoreceptors

Gene expression in the retina was normalized to β-actin and compared in diabetic and control BN, LE, and SD rats, to evaluate the effect of hyperglycemia on photoreceptor function. After 3 months of diabetes, probe arrays (TaqMan; ABI) revealed a decrease in photoreceptor-specific opsin (Opn1mw) mRNA in all three strains, suggesting reduced photoreceptor function. ⁴⁹ Arrestin (Arr3) and rhodopsin kinase (Grk1) expression are reduced in BN and SD, but not LE rats, further indicating strain-dependent differences in the diabetic retina beyond the inflammatory response. Transducin (Gnb3) mRNA was increased in the retina of diabetic SD rats only (Fig. 6). The decrease of inhibitory molecules arrestin and rhodopsin kinase mRNA contemporaneous with an upregulation of transducin gene interaction may indicate a compensatory mechanism in SD rat phototransduction. 

Figure 6.

 

Gene expression patterns of proteins associated with the phototransduction pathway. Expression of photoreceptor-specific genes as well as those essential to the phototransduction pathway were measured by real-time PCR. Levels of Opsin (Opn1mw), arrestin (Arr3), rhodopsin kinase (Grk1), and transducin (Gnb3) were compared in diabetic versus control animals. Expression is denoted in relative units normalized to Actb mRNA. Data are average of results in three independent experiments with three rats per group (n = 9). Error bars, SE. *P < 0.05, compared with the control, by Student's t-test.

Discussion

In this study, we report inflammatory responses in the retina of diabetic SD rats, but not LE or BN strains. Over 50 inflammatory proteins and growth factors in the retina of diabetic BN, LE, and SD rats were analyzed, and the data showed a subclinical inflammatory response in the retina of diabetic SD rats that coincided temporally with the upregulation of FGF-2 in photoreceptors as well as evidence of photoreceptor dysfunction. It is the first direct comparison of three commonly used rat strains in the study of diabetic retinopathy and demonstrates the highly strain-dependent nature of this animal model. 

STZ-induced diabetes in rats and mice has been used extensively to study DR. In the rat model, SD rats are a commonly used strain. However, the lack of pigmentation in these albino rats clearly interferes with the function of the visual system, ⁵⁰ and is therefore of limited use in some studies, particularly those examining visual function. 

In the present study, measurement of a large number of cytokines and growth factors as well as other inflammatory markers showed that most proteins were either below detectable limits or unchanged in the eyes of diabetic rats. These results match prior published data that failed to show a large scale inflammatory response in this model. In fact, in BN and LE rats, there is no evidence of cytokine upregulation through three months of diabetes. However, SD rats show signs of a subtle, but reproducible inflammatory response. Four weeks after the induction of diabetes, eotaxin, a basophile and eosinophil chemoattractant, was upregulated in the diabetic retina of SD rats compared with their nondiabetic counterparts. Although increased eotaxin expression in the retina was transient, significantly elevated levels were present in the serum of diabetic rats after 3 months of diabetes. This excludes serum as the source of eotaxin in the retina, since the levels in this tissue are decreased at a time when serum levels hit their peak. Increased protein levels of M-CSF, which act on monocytes and stimulate their growth, further indicate a response tailored to early immune mediators. This upregulation in SD rats occurred at the same time as increased MCP-1 and -3 protein expression, cytokines that attract early inflammatory cells such as monocytes. ^51,52 Although MCP-3 is undetectable in vitreous or serum of diabetic patients, ⁵³ MCP-1 has been found to be increased in the vitreous of patients with diabetic retinopathy ^41,53–55 and its level correlates with the presence of retinopathy. ⁵⁶ These inflammatory changes are subtle, and concentrations of all four cytokines are generally low, but the results are highly reproducible and clearly restricted to only one rat strain. Despite the lack of overt inflammation, there are other indicators that hyperglycemia activates an immune response in SD rats as early as 4 weeks after the induction of diabetes. Similar to an earlier report in diabetic SD rats, ⁴⁵ clusterin was overexpressed through 3 months of hyperglycemia and indicated early retinal damage. Of interest, though control levels of clusterin were similar between rat strains, there was no increase in either BN or LE rats. Like clusterin, TIMP-1 also showed a small but significant upregulation after 4 weeks of diabetes followed by a more robust change after 3 months in SD rats. Protein levels in the other two rat strains were low in both control and diabetic rats. Increased mRNA of the TIMP-1 ligand MMP-9 have been reported in SD rats, ⁵⁷ and although the proteins can be found in humans in normal interphotoreceptor matrix ⁵⁸ an increase correlates with vitreous hemorrhage in diabetic patients. ⁵⁹ The upregulation of B2M was another indicator of an ongoing immune response in SD rats, as was the increase in vWF after 3 months. Although some of the proteins upregulated in the retina were also increased in serum, these phenomena were most likely independent of each other. Only 3 of the 11 proteins that changed in the retina were also upregulated in serum and more important, eotaxin levels did not reach their peak in serum until 3 months of diabetes, a time at which retina levels had already decreased. 

The flow cytometry-based technique (IS 100 system; Luminex) used in this study did not measure a significant increase in total VEGF in the diabetic retina of SD or LE rats at the times tested, but did show an increase in BN retina. This difference may be due to a less robust VEGF upregulation in SD than in BN rats, which has been demonstrated in retinal ischemia. ^8,9 Furthermore, vascular permeability in STZ-induced diabetes has been reported to be an early and transient phenomenon in SD rats, whereas the effect is prolonged in BN rats. ¹⁰ Similar to VEGF expression, FGF-9 protein was upregulated in BN rat retinas, and this effect was transient and only observed at 4 weeks. 

FGF-2, on the other hand, was upregulated after 3 months in SD rats, confirming published reports. ⁴⁷ Immunohistochemistry showed no inflammatory infiltrates in the retina, suggesting low-grade inflammation most likely originating in the retina. The amount of cytokines produced is evidently not high enough to induce transmigration of leukocytes into the retina. This result is in agreement with published data in both animals and patients, where increased leukostasis ^12,60 by neutrophils and macrophages ^42,60 in retinal vessels is reported without convincing evidence of extravasation of leukocytes into the retina. Although Miyamoto et al. ⁶¹ report transmigration, the data presented do not lend strong support to such a claim. 

Immunohistochemistry further showed that FGF-2 is constitutively expressed in the INL, and the increase in FGF-2 was due to upregulation restricted to the PNL. Studies in other animal models have shown that FGF-2 is present in Müller cells and astrocytes, ⁶² but is upregulated in response to injury in the retina only in photoreceptors, ^62,63 and this upregulation has a protective effect on these cells. ⁶⁴ Although there is a significant increase in apoptotic photoreceptors in diabetic SD rats, most of the photoreceptors appear unaffected. ⁶⁵ To compare photoreceptor viability between diabetic and control rats, genes expressed exclusively in photoreceptors were assessed. This gene expression has been shown to correlate to photoreceptor condition. ⁶⁶ Opn1mw mRNA was decreased in photoreceptors of all strains, suggesting degeneration of these cells. Furthermore, rhodopsin kinase and arrestin3 mRNA, which are also expressed exclusively in the photoreceptors in the retina, ^67,68 were downregulated in BN and SD rats, but not in LE rats. Finally, Gnb3 expression increased in diabetic retina of SD, but not in BN or LE rats. These data suggest a negative effect of high glucose on activation of the phototransduction pathway as demonstrated by reduced opsin mRNA while reinforcing the notion of strain-dependent differences in the retina as seen in rhodopsin kinase, arrestin3, and Gnb3 expression. The reduced opsin expression may explain the reported reduction in photoreceptor activation with high flash intensity ERG in diabetic SD rats. ⁶⁹ However, the same study did not find an effect on photoreceptor deactivation ⁶⁹ suggesting that the reduction in rhodopsin kinase and arrestin3, although statistically significant, does not affect function after 3 months of diabetes. Of interest, mRNA of Gnb3 kinase which is part of the transducin complex is increased in SD rats. This increase may indicate a compensatory mechanism in damaged photoreceptors of this strain. 

It is tempting to speculate that hyperglycemia activates retinal glial cells either directly or through the binding of advanced glycation endproduct (AGE)–modified proteins to AGE receptors on these cells. This activation may lead to production of low levels of inflammatory cytokines. Although their concentration is not high enough to induce infiltration of peripheral immune cells into the retina they may play a role in retinal vessel leukostasis. ⁶¹ The cytokines may also constitute a danger signal to photoreceptors, inducing upregulation of FGF-2 and causing decreased function of photoreceptors. Alternatively, the ongoing inflammatory response and high glucose levels may influence photoreceptor gene expression directly. 

The results of the present study clearly show that genetic background plays a significant role in the response to hyperglycemia in the retina and is consistent with findings in other studies of hyperglycemic and hypoxic stress in the retina. ^8–10 This strain-dependent response is not due to differences in glucose exposure in either blood or the area adjacent to the retina, as all three strains show similar levels of hyperglycemia. A possible explanation of the lack of inflammation in LE and BN rats is the presence of melanin in the two pigmented strains as opposed to the lack of it in the albino SD rats. Melanin has been shown to have antioxidant properties by acting as a scavenger for free radicals and reactive oxygen species. ^70,71 Nonpigmented RPE cells show significantly more oxidative changes than pigmented ones ⁷² and high-dose exposure to light damages photoreceptor outer segments to a much larger extent in albino Wistar than LE rats. ⁷³ It is possible that hyperglycemia by itself is not enough to induce inflammatory changes in the rat eye, but that in nonpigmented rats the increase in oxidative stress provides a secondary stimulus that results in upregulation of inflammatory proteins in the retina and increase of FGF-2 expression in photoreceptors in SD rats. 

The findings of this study are the first direct comparison of the retinal expression of a large number of inflammatory proteins and growth factors in three rat strains. It is evident that responses to hyperglycemia are highly strain dependent. We demonstrate a subtle but reproducible inflammatory response in SD rats that shows some overlap with human diabetic retinopathy. This model may be useful in identifying the molecular mechanisms of subclinical inflammation and photoreceptor dysfunction in the early stages of diabetic retinopathy. 

Footnotes

 Supported by Allergan, Inc.

Footnotes

 Disclosure: S.J. Kirwin, Allergan, Inc. (E, F); S.T. Kanaly, Allergan, Inc. (E, F); N.A. Linke, Allergan, Inc. (E, F); J.L. Edelman, Allergan, Inc. (E, F)

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.