The Ins2^Akita Mouse as a Model of Early Retinal Complications in Diabetes

Abstract

purpose. This study tested the Ins2^Akita mouse as an animal model of retinal complications in diabetes. The Ins2^Akita mutation results in a single amino acid substitution in the insulin 2 gene that causes misfolding of the insulin protein. The mutation arose and is maintained on the C57BL/6J background. Male mice heterozygous for this mutation have progressive loss of β-cell function, decreased pancreatic β-cell density, and significant hyperglycemia, as early as 4 weeks of age.

methods. Heterozygous Ins2^Akita mice were bred to C57BL/6J mice, and male offspring were monitored for hyperglycemia, beginning at 4.5 weeks of age. After 4 to 36 weeks of hyperglycemia, the retinas were analyzed for vascular permeability, vascular lesions, leukostasis, morphologic changes of micro- and macroglia, apoptosis, retinal degeneration, and insulin receptor kinase activity.

results. The mean blood glucose of Ins2^Akita mice was significantly elevated, whereas the body weight at death was reduced compared with that of control animals. Compared with sibling control mice, the Ins2^Akita mice had increased retinal vascular permeability after 12 weeks of hyperglycemia (P < 0.005), a modest increase in acellular capillaries after 36 weeks of hyperglycemia (P < 0.0008), and alterations in the morphology of astrocytes and microglia, but no changes in expression of Müller cell glial fibrillary acidic protein. Increased apoptosis was identified by immunoreactivity for active caspase-3 after 4 weeks of hyperglycemia (P < 0.01). After 22 weeks of hyperglycemia, there was a 16.7% central and 27% peripheral reduction in the thickness of the inner plexiform layer, a 15.6% peripheral reduction in the thickness of the inner nuclear layer (P < 0.001), and a 23.4% reduction in the number of cell bodies in the retinal ganglion cell layer (P < 0.005). In vitro insulin receptor kinase activity was reduced (P < 0.05) after 12 weeks of hyperglycemia.

conclusions. The retinas of heterozygous male Ins2^Akita mice exhibit vascular, neural, and glial abnormalities generally consistent with clinical observations and other animal models of diabetes. In light of the relatively early, spontaneous onset of the disease and the popularity of the C57BL/6J inbred strain as a background for the generation and study of other genetic alterations, combining the Ins2^Akita mutation with other engineered mutations will be of great use for studying the molecular basis of retinal complications of diabetes.

Diabetic retinopathy is the leading cause of vision loss in developed countries. ¹ There has been great progress toward revealing the mechanisms of the disease by using animal models of diabetes, but the lack of a good model of spontaneous diabetes has limited the use of genetically altered mice. The most commonly used animal models include rodents, dogs, and primates, with diabetes induced by chemical toxins such as streptozotocin or alloxan. ^{2 3} This approach replicates some of the early symptoms of diabetic retinopathy and has the advantage that the onset of diabetes can be defined as the time of injection of the toxin. However, toxin-induced diabetes in mice has been less successful because of strain-dependent resistance to streptozotocin. ⁴ There is a need for a well-characterized mouse model of diabetic retinopathy to be used in transgenic studies. The power of a mouse model of any disease phenotype lies in the ease with which additional genetic changes can be introduced. The availability of the Ins2^Akita mouse line, containing a dominant mutation that induces spontaneous diabetes with a rapid onset, on a popular inbred genetic background that does not compromise fertility, has great potential for the molecular dissection of diabetic complications, including diabetic retinopathy. 

The Mody4 locus on chromosome 7 has been demonstrated to have a point mutation in the insulin 2 gene, now referred to as the Ins2^Akita allele. Mice heterozygous for the mutation are not obese and show development of hyperglycemia and hypoinsulinemia that are detectable after 4 weeks of age, progressing to early mortality (i.e., before the end of the first year). ⁵ The Ins2^Akita mutation replaces a cysteine with tyrosine at the seventh amino acid of the A chain of the insulin 2 gene product, blocking the formation of an essential disulfide bond between the A and B chains of the mature protein. This mutation causes a conformational change in the protein, leading to its accumulation in the endoplasmic reticulum of pancreatic β cells, which triggers the unfolded protein response and ultimately β-cell death. Loss of β cells in the pancreas results in systemic hypoinsulinemia and hyperglycemia. ^{6 7 8}  

Few studies have examined the potential for the Ins2^Akita mouse as a model of diabetic complications. Expression of αB-crystallin is increased in oligodendrocytes of the corpus callosum, neurofilament orientation in the sciatic nerve is abnormal, and the basement membrane in endoneural microvessels is thicker, suggesting the utility of the Ins2^Akita mouse as a model of diabetic neuropathy. ⁹ However, to date no studies have examined the retinas of Ins2^Akita mice. 

In this study, we established the response to diabetes of the vascular, glial, and neural cells of the retina in the Ins2^Akita mouse. Several similarities and few differences were noted in comparison to published changes resulting from streptozotocin (STZ)-induced diabetes in rats. Overall, the data support the use of the Ins2^Akita mouse as an excellent model to explore the molecular mechanisms involved in the initiation and early progression of diabetic retinopathy. 

Methods

Animals

C57BL/6J Ins2^Akita heterozygote mice (Jackson Laboratory, Bar Harbor, ME) were bred in the Penn State College of Medicine Juvenile Diabetes Research Foundation Diabetic Retinopathy Center Animal Core, in accordance with the Penn State College of Medicine Institutional Animal Care and Use Committee guidelines. Only males were used in the studies because disease progression in females is slower and less uniform. ⁵ In all experiments, the control groups were made up of siblings homozygous for the wild-type Ins2 gene. The vascular lesion (trypsin digest) and leukostasis studies were performed at Case Western Reserve University Department of Medicine (Cleveland, OH). All methods involving animals adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Mice were housed in pairs in plastic cages in a pathogen-free environment, with continuous access to food and water on a 12-hour light–dark schedule. Diabetic phenotype and genotype was confirmed 4.5 weeks after birth by blood glucose >250 mg/dL (One-Touch Lifescan meter; Lifescan, Inc., Milpitas, CA) in a drop of blood from a tail puncture. The disease is 100% penetrant in mice with the Ins2 mutation. ⁶ With the exception of a small number of animals in the studies performed at Case Western, mice were not given supplemental insulin. 

BSA-FITC In Vivo Vascular Permeability

Permeability in the retina was measured in vivo with an approach adapted from previous studies on rats. ¹⁰ Bovine serum albumin conjugated with FITC (BSA-FITC, 100 μg/g; Sigma-Aldrich, St. Louis, MO) was injected into the tail vein of mice anesthetized with pentobarbital (100 mg/kg). Thirty minutes later, the mice were rapidly decapitated, and the eyes were enucleated and immersed in ice-cold 4% paraformaldehyde for 1 hour. After fixation, the eyes were flash frozen in OCT (Tissue-Tek; Sakura Finetek, Torrance, CA) by rapid immersion in 2-methylbutane cooled with dry ice. Serial cryostat sections (10 μm) of the eyes were mounted in aqueous medium (Poly/Aquamount; Polysciences, Warrington, PA). 

Trypsin Digest

Trypsin digest of mouse retinas was performed as described previously. ^{11 12} Briefly, the eyes were fixed by immersion in formalin and digested in a crude proteolytic mixture (3% Difco crude trypsin 1:250; Invitrogen-Gibco, Grand Island, NY) containing 0.2 M sodium fluoride at 37°C for 2 hours. The neuroretinal tissue was gently brushed away, and the resultant isolated vascular tree was air dried onto a glass microscope slide. The isolated retinal vessels were stained with periodic acid-Schiff and hematoxylin for histologic evaluation. Pericyte ghosts, which appear where pericytes have been lost, were counted only on capillaries having one or more endothelial cells. Acellular capillaries, defined as basement membrane tubes lacking cell nuclei and maintaining at least one-fourth the normal capillary caliber over their length, were counted in multiple midretinal fields and standardized to retinal area (per square millimeter). 

Leukostasis

Leukostasis was measured after 8 weeks of hyperglycemia, using methods reported by others. ¹³ The left ventricle was cut to allow outflow, and PBS (5 mL) was perfused into the aorta at the normal cardiac output rate (10–14 mL/min) to clear erythrocytes and nonsticking leukocytes. Fluorescein isothiocyanate-coupled concanavalin A lectin (5 mg/kg; 20 μg/mL in PBS [pH 7.4]; Vector Laboratories, Burlingame, CA) was perfused to stain adherent leukocytes and vascular endothelium, followed by another wash with PBS at the same perfusion rate. The retina with attached vitreous was separated from the choroid and sclera, flatmounted on a microscope slide, covered with antifade medium and coverslip, and imaged by fluorescence microscopy. Only whole retinas in which the entire vascular network was stained were used for analysis. Leukocyte location was scored as being either arteriolar, venular, or microvascular. The large vessels emanating from the optic nerve and their first-grade branches were defined as either arterioles or venules. Arterioles were differentiated from venules by virtue of their smaller diameter and microvessels branching off the vessel. All remaining vessels were regarded as microvascular. The total number of adherent leukocytes per retina was counted. 

Immunohistochemistry and Confocal Microscopy

Immunohistochemistry was performed on whole retinas, as described previously. ¹⁴ Retinas were rinsed in PBS and permeabilized with 0.3% Triton in 10% donkey serum (Jackson ImmunoResearch, West Grove, PA) block solution. The retinas were incubated in 200 μL primary antibody reagent for 3 days at 4°C and washed in PBS overnight. Primary antibodies were rabbit polyclonal anti-active caspase-3 (1:10,000; CM-1; BD Biosciences, Mountain View, CA); rabbit polyclonal anti-Iba1 (1:500; Wako, Neuss, Germany); anti-GFAP (glial fibrillary acidic protein; 1:50; Roche, Indianapolis, IN); rabbit polyclonal anti-occludin (1:200; Zymed, S. San Francisco, CA). The secondary reagent was a mixture of CY2-conjugated donkey anti-mouse (1:2000) and RRX-conjugated donkey anti-rabbit F(ab′) fragments (1:1000; Jackson ImmunoResearch). Stained retinas were flattened onto 2% 3-aminopropyltriethoxy saline (AES)–coated microscope slides before mounting with aqueous medium (Aqua/Polymount; Polysciences). All images were obtained with a confocal microscope (TCS SP2 AOBS; Leica, Deerfield, IL), at 512 × 512-pixel resolution. Images were maximum projections of z-stacks. The brightness and contrast of some images were adjusted with image-analysis software (Photoshop; Adobe, Mountain View, CA), and the digital resolution was held constant. 

Morphometric Analysis

Mice were anesthetized with pentobarbital (100 mg/kg, intraperitoneally) and decapitated. Both eyes were enucleated and fixed for 1 hour by immersion in 4% paraformaldehyde for 15 minutes and then frozen in 2-methylbutane on dry ice. Morphometric measurements of retinal layers were made as described previously. ¹⁵ Six pairs of cryostat sections (10 μm) were taken from each eye at 100-μm intervals, by using a tape-transfer system (Cryojane; Instrumedics, Hackensack, NJ), which preserves the morphologic structure of cryostat sections more efficiently than does regular cryostat sectioning. One of each pair of sections was stained with hematoxylin and eosin (H&E), and the other was stained with Hoechst (Sigma-Aldrich). Images of H&E sections were captured with software (Optimas, Bothell, WA) running on a computer linked to a video camera (DXC-960MD; Sony, Tokyo, Japan) mounted on a microscope (BH-2; Olympus, Lake Success, NY) Five microscope fields were sampled from each section—two from the peripheral regions and three from the central region—with a 20× objective. Calibrated lines were drawn perpendicular to each layer ofthe retina, and all measurements were expressed relative to the thickness of the choroid and RPE in each field. The number of retinal ganglion cells was estimated in similar sections from the same eyes by counting large cell bodies stained with Hoechst dye. 

IR Kinase Assay

Insulin receptor kinase activity was measured as described previously for rat retinas. ¹⁶ Retinas were rapidly removed and lysed in cold immunoprecipitation buffer. Aliquots containing 75 μg protein were immunoprecipitated with an antibody to the β subunit of the insulin receptor (Santa Cruz Biotechnology, Santa Cruz, CA) overnight at 4°C. A mock immunoprecipitate containing no tissue lysate served to blank the assay. The immune complex was washed twice with kinase buffer, and then the kinase reaction was performed at room temperature in 200 μL total volume with 100 mM adenosine triphosphate (ATP), 3 mg/mL poly Glu:Tyr (Sigma-Aldrich), and 25 μCi/mL [³²P]ATP (Amersham, Piscataway, NJ) for 45 minutes with constant mixing. The reaction was stopped by brief centrifugation, and 25 μL of supernatant was spotted on p81 phosphocellulose paper (Whatman, Clifton, NJ). Filter papers were washed three times for 5 minutes in 0.75% phosphoric acid and once for 5 minutes in acetone before counting on a scintillation counter (model LS 6000SC; Beckman-Coulter, Fullerton, CA). 

Statistical Analysis

Quantifications were performed in a masked and randomized fashion. Statistical comparisons were by unpaired t-test for data sets with only two groups (Excel 2000; Microsoft, Redmond, WA) and by one-way analysis of variance with post hoc Student-Newman-Keuls multiple-comparisons test for data sets containing more than two groups (Instat 3.0 for the Macintosh; GraphPad Software, San Diego, CA). Quantification of acellular capillaries and pericyte ghosts was analyzed with the Mann-Whitney test. 

Results

Animals

After 4 weeks of age, the heterozygote Ins2^Akita mice had significantly elevated concentrations of blood glucose compared with those of control mice (Fig. 1) . At death, the Ins2^Akita mice also weighed significantly less than the control animals (Table 1) . These data agree with results in previous reports. ⁵  

Vascular Permeability to BSA-FITC

Increases in retinal vascular permeability in animal models of diabetes have been reported by multiple investigators who used a variety of different techniques. ^{17 18 19 20 21 22 23} Retinal vascular permeability was measured by extravasation of BSA-FITC, as described previously for STZ-diabetic rats. ¹⁰ After 12 weeks of hyperglycemia, mice were injected systemically with BSA-FITC, and after 30 minutes the eyes were removed, fixed, and sectioned. The relative fluorescence was measured by image analysis in multiple regions of retinal cross sections from control and Ins2^Akita diabetic mice. There was significantly more fluorescence in the retinas of Ins2^Akita mice than in control retinas (Fig. 2 ; n = 4 controls, n = 5 Ins2^Akita, P < 0.01). This experiment was performed three times with similar results, equivalent to measures of BSA-FITC permeability in STZ-diabetic rats. ¹⁰  

Occludin Immunoreactivity

Increased vascular permeability is accompanied by modifications in the immunoreactivity and distribution of the tight junction protein occludin. ^{10 17 24 25 26} Immunohistochemistry for occludin was performed on whole retinas from Ins2^Akita diabetic and control mice after 8 weeks of hyperglycemia. The overall intensity of occludin immunoreactivity in the Ins2^Akita retinas was generally similar to that in the control retinas, but several arterioles contained punctate, discontinuous immunoreactivity similar to that reported previously in STZ-diabetic rat retinas (Fig. 3) . ¹⁰  

Vascular Lesions and Leukostasis

Acellular capillaries and pericyte ghosts were counted in retinal vasculature isolated from the neural tissue by trypsin digest in mice at between 31 and 36 weeks of hyperglycemia. There was a modest elevation in the number of acellular capillaries in the Ins2^Akita mice compared with control retinas (Fig. 4A ; P < 0.0008, Mann-Whitney test), but no significant increase in the number of pericyte ghosts in the Ins2^Akita retinas (n = 17) compared with control retinas (n = 13). There was a significant increase in the number of leukocytes per retina in the Ins2^Akita mice compared with the control (Fig. 4B ; P < 0.05). 

GFAP Immunoreactivity in Astrocytes

Immunohistochemistry for GFAP was performed in whole retinas from Ins2^Akita diabetic mice after 22 weeks of hyperglycemia (n = 4) and in age-matched control mice (n = 6). In control mice, GFAP immunoreactivity was restricted to astrocytes (Fig. 5A) . GFAP immunoreactivity was reduced in some regions of retinas from Ins2^Akita diabetic retinas (Fig. 5B) . Inspection at higher magnification revealed that projections from astrocytes in control retinas were wrapped around blood vessels (Fig. 5C) . In Ins2^Akita retinas, astrocytes close to large-caliber superficial vessels had short projections that did not conjoin with the blood vessels (Fig. 5D) . Similar results were also obtained after 8 weeks of hyperglycemia (data not shown). 

Microglial Reactivity

Microglia were visualized in whole retinas from control and Ins2^Akita mice after 8 weeks of hyperglycemia, by immunohistochemistry for the ionized calcium-binding adaptor molecule (Iba1), which is unique to microglia and macrophages. ²⁷ Microglia in control retinas had a typical nonreactive morphology with long, thin processes (Figs. 6A 6C) . In contrast, retinas from Ins2^Akita diabetic mice had discrete regions containing microglia with retracted, swollen processes, indicating a reactive morphology (Figs. 6B 6D) . Similar observations have been made in STZ-diabetic rat retinas. ²⁸  

Active Caspase-3 Immunoreactivity

Immunohistochemistry for the active form of caspase-3 was performed in whole retinas of control and Ins2^Akita mice after 4 weeks of hyperglycemia. Cells with positive immunoreactivity were quantified by visually scanning the entire retina in a raster pattern, with a fluorescence microscope and 40× objective, and the results were expressed as the number of cells per 0.5 cm², as described previously. ¹⁵ There were significantly more cells with active caspase-3 immunoreactivity in the retinas from Ins2^Akita mice than in control retinas (Fig. 7 ; n = 6 per group; P < 0.005). 

Retinal Morphometry

The relative thickness of retinal layers was measured in Ins2^Akita diabetic mice after 22 weeks of hyperglycemia (n = 4) and in age-matched control mice (n = 6). The measurements were taken at two peripheral and three central regions of each section. Image analysis of serial sections was used to quantify the thickness of the inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), and outer nuclear layer (ONL). All measurements were expressed relative to the thickness of the choroid and RPE in each section, which was not significantly different in Ins2^Akita diabetic mice compared with control animals (77.7 ± 4.8 and 78.7 ± 2.0 μm respectively; P > 0.8). In the peripheral regions of the Ins2^Akita retinas IPL thickness was 27% less than in the control retinas (P < 0.001; Fig. 8A ) and INL thickness was 15.6% less than in the control (P < 0.01; Fig. 8A ). In the central regions of the Ins2^Akita retinas the thickness of the IPL was 16.7% less than in the control (P < 0.001; Fig. 8B ). The thickness of the OPL and nuclear layers was not altered in either the peripheral or central regions. The thickness of the outer segments of the photoreceptors was not significantly altered in any region (data not shown). The number of Hoechst-labeled nuclei in the retinal ganglion cell layer was significantly reduced (23.4% versus the control; P < 0.005). These data are similar to results in previous reports of loss of cells from the inner retina in STZ-diabetic rats ¹⁵ and humans. ^{29 30}  

Insulin Receptor Kinase Activity

Loss of insulin or insulin receptor activity in the liver, fat, and muscle underlie the metabolic features of diabetes. We determined whether a similar change occurs in the retina by immunoprecipitating the insulin receptor β subunit from retinal lysates of control and Ins2^Akita mouse retinas after 12 weeks of hyperglycemia. Kinase activity in response to an artificial substrate was measured with [³²P]ATP and expressed as a percentage of control activity. The kinase activity was significantly reduced (27.4% less than the control level) in the lysates from Ins2^Akita mice (Fig. 9 ; n = 9 control, n = 8 Ins2^Akita; P < 0.05). 

Discussion

In this study, we tested the spontaneously diabetic Ins2^Akita mouse as a model of early diabetic retinopathy. Several outcomes of diabetes previously established in STZ-diabetic rats were measured in the retinas of the Ins2^Akita mice. There were strong similarities to the STZ-diabetic rat model, but also some differences. At a whole-body level, the Ins2^Akita mice are less catabolic than STZ-diabetic rats because they gain weight (although still at a slower rate than normal). ⁵ Nevertheless, the features of retinal complications are very similar. 

The increase in vascular permeability to exogenous bovine serum albumin was similar in magnitude to that found previously in STZ-diabetic rats using the same technique. ¹⁰ The loss of blood–retinal barrier function was accompanied by disorganized occludin immunoreactivity in superficial arterioles, again as described in STZ-diabetic rat retinas. ^{14 17} Occludin immunoreactivity was similar to that associated with occludin phosphorylation and migration in cell culture, which are proposed as central to the mechanism of increased paracellular endothelial cell permeability. ^{24 26} This is the first study to establish that diabetes increases retinal vascular permeability in mice. 

The moderate increase in acellular capillaries was statistically significant but less than previously reported in STZ-diabetic rats. ^{31 32 33} Pericyte ghosts are more difficult to count in mice and do not have the obvious appearance noted in diabetic rats and larger species. Our inability to detect an effect of prolonged diabetes on pericyte loss raises the possibility that pericyte loss occurs by a different process than that leading to formation of acellular capillaries, or that the pericyte ghosts disappear more rapidly in Ins2^Akita mice than in other species. In a study in which trypsin digests of STZ-diabetic mouse retinas were used, the investigators reported thickening of the vascular basement membrane after 18 months of hyperglycemia, but did not report acellular capillaries and pericyte loss, confirming that these lesions may be less prominent in diabetic mice. ³⁴  

Diabetes has been found to increase the number of leukocytes adhering to the wall of retinal vessels, ³⁵ a finding that is consistent with inflammation. ¹³ The present studies demonstrate that the number of leukocytes adherent to the vascular wall was significantly elevated in the Ins2^Akita mouse, confirming that the vascular inflammatory component of diabetic retinopathy is present in this model. 

The response of the macroglial cells did not precisely reflect changes in STZ-diabetic rats. Astrocytes in some regions of the Ins2^Akita retinas contained reduced GFAP immunoreactivity with shorter processes, suggesting atrophy or loss of contact with blood vessels, but increased Müller cell expression of GFAP was not observed. In STZ-diabetic rats, there is a dramatic reduction in GFAP expression in the astrocytes, followed by increased expression in Müller cells. ^{14 36 37 38} Furthermore, a study of postmortem human retinas found increased Müller cell GFAP expression. ³⁸ Our data confirm a previous report that GFAP expression was not dramatically altered in STZ-diabetic mice. ³⁹ The reason for this discrepancy between diabetic rats and mice is unclear. 

Regions of the Ins2^Akita diabetic retinas contained microglia with swollen and contracted processes, suggesting a reactive state, as described in STZ-diabetic rat retinas. ^{28 40} The appearance of Iba1 immunoreactivity was comparable to that in brain tissue after facial nerve axotomy or ischemia. ^{41 42} It is not clear what factors initiate microglial reactivity, but it is further indication of an inflammatory response within the neural retina, which may contribute to vascular permeability or neuronal apoptosis. The sporadic distribution of reactive microglia suggests that some pathologic changes may occur in discrete regions rather than homogeneously. In a study involving concanavalin A perfusion, the researchers concluded that vascular permeability also occurs in discrete locations in STZ-diabetic rats. ¹⁷  

Thinning of the inner layers of the retina suggests that chronic degeneration occurs in Ins2^Akita mice after 22 weeks of hyperglycemia. These data parallel reductions in the thickness of the inner layers in STZ-diabetic rats after 7.5 months of diabetes. ¹⁵ The significantly reduced thickness of the peripheral INL suggests loss of horizontal, bipolar and amacrine cell bodies. The thickness of the IPL was significantly reduced in all regions measured, indicating atrophy of the processes between neurons in all parts of the retina. These data imply that atrophy of dendrites occurs before loss of cell bodies in the inner retina. There was no significant loss of the ONL and the outer segments of the photoreceptors, suggesting that the degeneration predominantly occurs in the inner retina. These data agree with a recent study showing reduced retinal thickness in STZ-diabetic mice after 10 weeks of diabetes, ⁴³ but conflict with a study reporting extensive loss of photoreceptors and the outer nuclear layer in STZ-diabetic rats. ⁴⁴ The reason for this inconsistency is not clear. 

An increase in apoptosis, measured by counting cells with active caspase-3 immunoreactivity, was found in Ins2^Akita mice after 4 weeks of hyperglycemia. The magnitude of apoptosis was similar to that in STZ-diabetic rat retinas measured by TUNEL. ^{15 39} These data also confirm the increase in caspase activity measured by substrate cleavage in STZ-diabetic rats and mice. ⁴⁵  

Retinal insulin receptors have higher constitutive kinase activity than do those in the liver. ¹⁶ The basal kinase activity of immunoprecipitated insulin receptors was significantly reduced in the Ins2^Akita mouse retina, similar to STZ-diabetic rat retina (Reiter and Gardner, unpublished data, 2004) and muscle. ⁴⁶ Constitutive phosphorylation of the retinal insulin receptor was elevated in STZ-diabetic mice and increased further in response to a large dose of exogenous systemic insulin. ⁴⁷ Insulin has been shown to reduce apoptosis in R28 cells, which model retinal neurons, and thus disruption of this signaling pathway by diabetes may increase the potential for apoptosis in the retina. ⁴⁸  

Overall, these data suggest that coordinated degeneration of the inner retina and vasculature occurs during the initial months of diabetes in the Ins2^Akita mouse. The specific causes and consequences of each of these changes remains unknown; but, in a broad sense, they imply impaired neural cell-to-cell interactions that are essential for normal vision. Establishing this spontaneously diabetic mouse model of early retinal complications provides, for the first time, the opportunity for molecular–genetic investigation. 

The Ins2^Akita mouse has several important advantages over other animal models. First, the autosomal dominant mutation provides the opportunity to study heterozygotic animals. Second, the mice are fertile and breed well. Third, they have stable insulin-deficient diabetes and can be maintained in a noncatabolic state without exogenous insulin. Fourth, the mechanism of diabetes onset does not involve systemic immunologic alterations, and it is therefore possible to evaluate the metabolic impact on the retina without interference from T- and B-cell abnormalities, as in the NOD mouse. Fifth, the pure C57BL/6J genetic background lacks the rd (pde6) mutations and photoreceptor degeneration found in FVB and other strains. ^{49 50} Together, these features provide an important new opportunity to investigate the pathophysiology of the early stages of diabetic retinopathy. 

 Supported by the Juvenile Diabetes Research Foundation International Diabetic Retinopathy Center Grant and Career Development Award, the Pennsylvania Lions Sight Conservation and Eye Research Foundation, the American Diabetes Association, National Eye Institute Grants EY122021 and EY00300, and Jack and Nancy Turner.

 Submitted for publication November 17, 2004; revised February 7, 2005; accepted February 19, 2005.

 Disclosure: A.J. Barber, None; D.A. Antonetti, None; T.S. Kern, None; C.E.N. Reiter, None; R.S. Soans, None; J.K. Krady, None; S.W. Levison, None; T.W. Gardner, None; S.K. Bronson, None

 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.

 

Figure 1.

 

Average blood glucose levels were significantly elevated after 4 weeks. Blood glucose was measured weekly in control and Ins2^Akita mice beginning at week 3 and ending at week 24. It was significantly elevated by week 4 and thereafter remained relatively constant in the Ins2^Akita mice, compared with the control (n = 6 control, n = 5 Ins2^Akita, *P < 0.01). Data are expressed as the mean ± SEM.

Table 1.

 

Average Weight and Blood Glucose Level at Time of Death

Study	Group	n	Duration (wk), †	Weight (g)	Blood Glucose (mg/dL)
BSA-FITC permeability	Control	4		30 ± 0.9	NA
	Diabetic	5	12	26 ± 0.8, **	NA
Occludin IHC	Control	5		32 ± 1.3	152 ± 9.1
	Diabetic	4	8	26 ± 0.4, **	391 ± 14.0, **
Vascular lesions and leukostasis	Control	13		37 ± 8.0	138 ± 32.0
	Diabetic	17	31–36	25 ± 2.0	469 ± 28.0, **
GFAP IHC and morphometry	Control	6		36 ± 1.7	146 ± 16.8
	Diabetic	4	22	25 ± 0.3, **	447 ± 32.2, **
Microglial IHC	Control	7		27 ± 0.8	144 ± 5.3
	Diabetic	9	8	24 ± 0.4, **	410 ± 14.8, **
Caspase-3 IHC	Control	6		32 ± 1.3	187 ± 12.0
	Diabetic	6	4	28 ± 1.2*	456 ± 23.0, **
IR kinase activity	Control	9		28 ± 0.8	140 ± 8.1
	Diabetic	8	12	24 ± 0.4, **	446 ± 12.2, **

 IHC, immunohistochemistry; NA, not available.

*   P < 0.05,

**   P < 0.01, unpaired t-test relative to corresponding control group.

†  Length of time with significantly elevated hyperglycemia based on onset between 3.5 and 4.5 weeks of age.

Figure 2.

 

Vascular permeability to albumin was increased in the retinas of Ins2^Akita mice. Control and Ins2^Akita mice after 12 weeks of hyperglycemia were injected intravenously with BSA-FITC, and eyes were fixed and sectioned. Images from sections were captured by confocal microscopy, and the average fluorescence intensity of regions between blood vessels was measured by digital image analysis. (A) In sections from control mice the retinal parenchymal tissue was less fluorescent than blood vessels; (B) sections from Ins2^Akita mice contained homogeneous fluorescence throughout the retinal tissue, particularly in the inner and outer plexiform layers and the rod outer segments; (C) the average fluorescence intensity was significantly greater in the sections from Ins2^Akita mice compared to controls (n = 4 control, n = 5 Ins2^Akita; *P < 0.01, t-test). Data expressed as the mean ± SEM.

Figure 3.

 

Redistribution of occludin immunoreactivity in major retinal blood vessels. Whole retinas from Ins2^Akita mice and control subjects were labeled by immunohistochemistry for the tight junction protein occludin and examined by confocal microscopy. (A) In control retinas, large vessels had continuous occludin immunoreactivity at cell borders; (B) in retinas from Ins2^Akita mice, arterioles contained regions of punctate occludin immunofluorescence (arrow).

Figure 4.

 

Vascular lesions and leukostasis. Vascular features associated with diabetic retinopathy were quantified. (A) After 31 to 36 weeks of hyperglycemia, whole retinas were digested with trypsin to isolate the vascular network, stained with periodic acid-Schiff and hematoxylin, and examined by light microscopy. Acellular capillaries were counted per square millimeter, and pericyte ghosts were counted per 1000 capillary cells. There was a modest increase in the number of acellular capillaries (*P < 0.0008; Mann-Whitney test) but no significant change in the number of pericyte ghosts (n = 13 controls; n = 17 Ins2^Akita mice). (B) Leukostasis was measured in whole retinas after 8 weeks of hyperglycemia. There were significantly more static leukocytes in retinas of Ins2^Akita mice than in the control retinas (*P < 0.05). Data are expressed as the mean ± SD.

Figure 5.

 

Astrocyte morphology and expression of GFAP was altered in retinas of Ins2^Akita mice. Whole retinas were labeled by immunofluorescence for GFAP, a glial intermediate filament expressed in astrocytes. The retinas were wholemounted and examined by confocal microscopy. (A) In control retinas, astrocytes had GFAP immunoreactivity and were closely associated with large superficial blood vessels. (B) In retinas from Ins2^Akita mice, GFAP-immunoreactive astrocytes appeared less dense. (C) Higher-magnification images showed that the processes from astrocytes in control retinas made contact with large blood vessels. (D) In Ins2^Akita mouse retinas the astrocytes close to superficial arterioles did not maintain intimate contact with the vessel (arrow).

Figure 6.

 

Microglia in retinas of Ins2^Akita mice became reactive. Whole retinas from Ins2^Akita and mice were labeled by immunohistochemistry for Iba1, a calcium-binding protein unique to microglia, and examined by confocal microscopy. (A) In control retinas, the microglia had a normal distribution and morphology; (B) some regions of Ins2^Akita mouse retinas contained microglia with swollen and retracted processes; (C) higher-magnification images showed fine, extended processes on microglia in control retinas; (D) in Ins2^Akita mouse retinas microglia had shorter, less-branched processes than did the control retinas.

Figure 7.

 

The number of cells immunoreactive for active caspase-3 increased in Ins2^Akita mice. Whole retinas from Ins2^Akita and control mice were labeled by immunohistochemistry for active caspase-3 after 4 weeks of hyperglycemia and counted by fluorescence microscopy. The number of positive cells was standardized to the area of each retina, measured by image analysis, and expressed as cells per 0.5 square centimeter. The number of cells immunoreactive for active caspase-3 was significantly increased in retinas from Ins2^Akita mice compared with the control (n = 6 per group; *P < 0.005, t-test). Data are expressed as the mean ± SEM.

Figure 8.

 

The thickness of the inner retina is reduced in Ins2^Akita mice. Eyes from Ins2^Akita (n = 4) and age-matched controls (n = 6) were fixed and sectioned after 22 weeks of hyperglycemia. The thicknesses of the IPL, INL, OPL, and ONL were measured by image analysis in multiple H&E-stained sections and expressed as a ratio with the thickness of the choroid and RPE. (A) There was a significant reduction in the thickness of the IPL and INL in the peripheral retina compared with that in the control (P < 0.01 and P < 0.05, respectively, ANOVA). (B) The thickness of the IPL was significantly reduced in the central retina compared with that in the control (P < 0.05, ANOVA). (C) The number of cell bodies in the retinal ganglion cell layer was significantly reduced (n = 6 control, n = 4 Ins2^Akita retinas; P < 0.005, t-test). Data are expressed as the mean ± SEM.

Figure 9.

 

Insulin receptor kinase activity was decreased in the retinas of Ins2^Akita mice. The β subunit of the insulin receptor was immunoprecipitated in retinal lysates of control and Ins2^Akita mouse retinas after 12 weeks of hyperglycemia. Receptor kinase activity was measured by in vitro phosphorylation of a poly (Glu:Tyr) substrate in the presence of [³²P]ATP. Radioactive counts were expressed as a percentage of control. Samples from Ins2^Akita mice showed significantly less kinase activity than did the control samples (n = 9 control, n = 8 Ins2^Akita; *P < 0.05, t-test). Data are expressed as the mean ± SEM.

The authors thank Edward H. Leiter, PhD (Jackson Laboratory), for helpful consultations; Casey Miller, PhD (Department of Medicine, Case Western Reserve University), Wendy Holtry, and Neelam Desai (Penn State College of Medicine) for breeding the Ins2^Akita mice; and Rhona Ellis, Ellen Wolpert, and Catherine Stiller for excellent technical assistance.