Ординатура / Офтальмология / Английские материалы / Retinal and Choroidal Angiogenesis_Penn_2008

Chapter 4

ANIMAL MODELS OF DIABETIC

RETINOPATHY

Timothy S. Kern

Case Western Reserve University, Cleveland, Ohio

1.INTRODUCTION

Diabetic retinopathy is a major complication of Type 1 and Type 2 diabetes mellitus, being observed in most patients after 15 years of diabetes, and increasing the risk of blindness 25-fold above normal.1,2 The natural history of clinically demonstrable retinopathy has been carefully documented, and

81

J.S. Penn (ed.), Retinal and Choroidal Angiogenesis, 81–102.

© Springer Science+Business Media B.V. 2008

important stages have been identified: vascular occlusion, formation of capillary microaneurysms, excessive vascular permeability, proliferation of new vessels and fibrous tissue, and contraction of the fibrovascular proliferations.3 This chapter will focus on the histological lesions that develop in animal models, and their relation to the lesions that develop in diabetic patients. Physiological abnormalities such as retinal blood flow and permeability have been reviewed elsewhere.4,5

The general picture that has emerged of the pathogenesis of vision loss in diabetic retinopathy focuses primarily on increased capillary permeability, which leads to retinal edema, and neovascularization. Retinal edema can result in appreciable visual impairment, presumably due to physical distortion of the retina. Neovascularization can prevent light from reaching the photoreceptors secondary to development of a fibrovascular membrane in front of the retina.

Figure 4-1. Simplified scheme postulated for the pathogenesis of diabetic retinopathy.

The nonproliferative stage of the retinopathy includes capillary cell death and capillary obliteration, microaneurysms, pericyte loss, and increased permeability. Pericyte loss was once believed to be the initial and most important lesion of the retinopathy, but it has since been demonstrated that both retinal endothelial cells and pericytes die at approximately the same rate

in diabetes. These “background” changes precede and are believed to be necessary for progression to the later neovascular changes (Figure 1). The early stages of the retinopathy (before microaneurysms are present) generally are not apparent clinically, even using sensitive techniques such as fluorescein angiography. At even earlier stages, however, these lesions are beginning to appear, and they can be studied histologically using eyes collected at autopsy or at surgery (Figure 2).

Figure 4-2. Microaneurysm (MA), acellular capillaries (long arrows), and pericyte ghosts (thick arrows) in dog diabetic for 5 years.

Nonperfused capillaries in diabetic retinopathy commonly lack endothelial cells or pericytes, and thus appear to be acellular. These “acellular capillaries” are the remnant basement membrane skeleton of a degenerate capillary from which all capillary cells have disappeared. Importantly, they are a histological marker of capillary nonperfusion, since acellular capillaries are not perfused.6

Thus, acellular capillaries are morphological lesions that have physiological significance and can be quantitated by light microscopy in animal studies of the retinopathy. Vaso-obliteration of retinal capillaries in diabetes begins in capillaries and can progress “upstream” to the arterioles and their side-branches. Potential causes of capillary occlusion include leukostasis, excessive platelet aggregation, endothelial swelling, endothelial death, and glial invasion of the capillary lumen.

2.ANIMAL MODELS OF DIABETIC RETINOPATHY

Animal models of diabetic retinopathy have proven valuable in efforts to unravel the pathogenesis of retinopathy, and to identify therapies to inhibit it.

Species used in studies of the effect of diabetes on the retina include spontaneously diabetic animals, including fish,7 mice,8,9 rats,10,11 cats,12

dogs,13-15 and apes,16 but these reports generally have been only descriptive in nature. Many investigations also have relied on experimental induction of

diabetes with alloxan, streptozotocin, growth hormone, or by pancreatectomy. Early studies of animal models have been reviewed elsewhere,12,17-19 and the

present review will focus on studies reported during the past decade.

2.1Dogs and cats

The anatomical features of retinopathy in diabetic dogs have been shown repeatedly to be morphologically indistinguishable from those of background retinopathy seen in diabetic patients. They include capillary microaneurysms, acellular (and nonperfused) capillaries, pericyte ghosts, varicose and dilated capillaries (also called intraretinal microvascular

abnormalities or IRMAs), and dot and blot hemorrhages.18,20-22 Arteriolar smooth muscle cell loss also has been observed in humans and dogs.23,24 The

lesions in diabetic dogs are secondary to insulin deficiency, since they develop irrespective of how diabetes was induced (alloxan, growth hormone, pancreatectomy), and can be inhibited by strict regulation of glycemia with exogenous insulin.25

Microaneurysms, leukocyte and platelet plugging of aneurysms and venules, and degenerating endothelial cells likewise were observed in cats after several years of diabetes.26,27 These histological abnormalities were confined to small regions, and these animals developed hypoxia in at least some areas of retina early in the development of diabetic retinopathy, before capillary dropout was evident clinically. Hypoxia was correlated with endothelial cell death, leukocyte plugging of vessels, and microaneurysms.

As is true in diabetic humans, there is a long interval before retinopathy becomes manifest in diabetic dogs or cats; capillary aneurysms usually begin to appear in these animals about 2 to 3 years after induction of elevated hexose levels. Likewise, after about 2 years of hyperglycemia in diabetic dogs, increasing numbers of retinal capillaries possess endothelial cells but few or no pericytes. Gradual obliteration of retinal vessels is apparent histologically from the increasing numbers of acellular capillaries that are scattered singly and in small groups about the retinal vasculature, especially

in temporal retina. Within 5 years of insulin-deficient diabetes, all dogs have a marked retinopathy. The reason for the prolonged interval before retinopathy develops is unknown, but any explanation of this latent period might offer valuable insight into the pathogenesis of the retinopathy. Improved glycemic control significantly inhibits the development and progression of retinopathy in diabetic dogs25,28 and in patients.29,30

The retinopathy that develops in cats and dogs more closely resembles lesions in human retinopathy than that of other species studied to date. Neovascularization has been observed to develop in diabetic dogs, albeit only within the retina (see Section 5). However, the cost, slow development of lesions, and lack of availability of antibodies or molecular biology techniques have made dog and cat models less used for the study of retinopathy in recent years.

2.2Rats

During the past decade, streptozotocin-diabetic or alloxan-diabetic rats have been the primary model for research into the pathogenesis of the vascular lesions of diabetic retinopathy.31-50 Spontaneously diabetic BB rats and rats made diabetic with alloxan or streptozotocin exhibit similar retinal lesions: pericyte loss, basement membrane thickening, and an absence of microaneurysms after about 14 months of hyperglycemia.51 However, later stages of retinal microvascular disease do not develop reproducibly (microaneurysms) or at all (IRMA, hemorrhages, and neovascularization).

As a model of diabetic retinopathy, the rat offers practical advantages over the dog and other large animals in terms of costs, housing requirements, and available reagents. Moreover, the early stages of retinopathy develop relatively quickly in the rat; pericyte loss and acellular capillaries are apparent after as little as 6 months of diabetes. A potential concern about this model is that the lens of rats has unusually high levels of aldose reductase compared to other species;52 whether or not this is true in other tissues has not tissues has not been reported.

2.3Mice

In the 1970’s and 80’s, there were a number of attempts to determine

whether or not diabetic mice developed diabetic retinopathy, but the results were controversial.8,9,53-55 Since then, it is surprising that mice have been

little studied with respect to diabetic retinopathy until recently, especially considering the widespread generation and use of genetically modified mice. Recent studies have begun to characterize the development of retinopathy in the streptozotocin-diabetic C57BL/6J mouse. This model develops the early

vascular pathology characteristic of diabetic retinopathy (acellular capillaries, pericyte loss, and capillary cell apoptosis) beginning at about 6 months of diabetes, and the acellular capillaries and pericyte ghosts become more numerous with increasing duration of diabetes (through 18 months of diabetes).56,57 These vascular abnormalities characteristic of retinopathy occurred despite the apparent lack of neuronal loss and Müller glial cell activation,57 although others have reported loss of cells in the ganglion cell layer in mice diabetic only 14 weeks58 (see section 2). Diabetes-induced retinal neovascularization has not been detected in any mouse model to date.

Genetically modified mice are beginning to be used to explore the role of adhesion molecules and leukostasis in the pathogenesis of diabetes-induced retinal vascular disease.59 Mice deficient in the genes encoding adhesion molecules CD18 and ICAM-1 were made diabetic or experimentally galactosemic, and studied after durations of up to 11 months (diabetic) or 22 months (galactosemic). Wild-type diabetic or galactosemic animals developed acellular capillaries and pericyte loss, as well as associated abnormalities including leukostasis, increased capillary permeability, and capillary basement membrane thickening. In contrast, CD18-/- and ICAM-1-/- mice developed significantly fewer of each of these abnormalities, thus providing novel insight that adhesion molecules play an important role in the pathogenesis of the retinopathy.

There are both advantages and disadvantages to the use of mice as models of diabetic retinopathy. The principal advantages are cost, availability of reagents, and ability to generate (or availability of) genetically modified animals for study. The principal disadvantages are the small size of the retina (and consequently the quantitatively small number of lesions that can be detected per retina) and the extreme difficulty in unambiguously identifying pericyte ghosts for quantitation.

2.4Lesser used models including primates

The early stages of diabetic retinopathy develop in all species that have diabetes for long durations. Thus, it seems unlikely that the ability to develop microvascular lesions of diabetic retinopathy is in any way unique. However, some laboratory species, such as guinea pigs and rabbits, are inherently of limited usefulness for the study of diabetic retinopathy. The guinea pig retina is avascular, as also is much of the rabbit retina, and the retinal vessels in rabbits are tortuous and limited chiefly to the most superficial inner layers of nasal and temporal retina.

Diabetic hamsters develop the usual spectrum of lesions, including acellular capillaries, pericyte loss, and endothelial proliferation, but lack microaneurysms and neovascularization.60

In spite of many similarities between subhuman primates and humans, diabetic retinopathy has been little studied in primates. In the past decade, these studies have been limited to identification of microaneurysms and other lesions in aged, spontaneously diabetic monkeys.16

3.DIABETES-INDUCED ABNORMALITIES

IN NONVASCULAR CELLS OF THE RETINA

Several decades ago, damage to nonvascular cells of the retina (including ganglion cells) in diabetic humans was detected both ultrastructurally61 and functionally,62,63 and the possible role of neural disease in the pathogenesis of diabetic retinopathy was postulated.64,65 Recently, there has been renewed

appreciation of diabetes-induced damage to nonvascular cells of the retina also in animals. Diabetic rats lose ganglion cells,44,58,66-75 and this

neurodegeneration has been detected at as early as one month of diabetes.69 This nonvascular abnormality precedes the development of the vascular cell changes,69 raising the possibility that neurodegeneration might contribute to the pathogenesis of the vascular disease. This has yet to be conclusively studied.

Retinal glial cells also undergo changes in diabetes in some species. Müller glial cells in diabetic rats became apoptotic in one study.68 In other studies, these cells changed from a quiescent to an injury-associated phenotype with high levels of expressed glial fibrillary acidic protein

(GFAP)—a hallmark of glial cell activation—after a few months of diabetes.44,68,71,76-81 Alterations in GFAP expression patterns in Müller glial

cells have also been observed in the human retina during early diabetes.77

In diabetic C57BL/6J mice, transient damage was noted in retinal ganglion cells (TUNEL-positive with activation of caspase 3) at about 4 weeks of diabetes. These abnormalities quickly returned to normal, however, and ultimately, no detectable loss of retinal ganglion cells or activation of

Müller glial cells was noted in the retinas, even after one year of diabetes.44,57 In contrast, others have reported that 14 weeks of diabetes was

sufficient to cause a 20-25% reduction in the number of cells in the ganglion cell layer compared to age-matched nondiabetic mice,58 and pro-apoptotic caspases were found to increase with increasing duration of diabetes.56 Likewise, C57BL/6J diabetic mice do not show GFAP activation in diabetes,44 other than a transient increase soon after induction of diabetes.57 Müller glial cells from mice show nuclear translocation of GAPDH in diabetes, a change that has been strongly linked to apoptosis.75 Ins2Akita diabetic mice also have increased retinal vascular permeability, greater than

normal numbers of caspase-3 positive cells, and unchanged GFAP immunoreactivity.82

Horizontal cells,71,74 amacrine cells, and photoreceptors74 also have been reported to undergo degeneration in diabetic rats. These changes are not known to be characteristic of retinal changes seen in diabetic patients, however, and the significance of these changes in animals remains to be learned.

4.NONDIABETIC MODELS THAT DEVELOP A DIABETIC-LIKE RETINOPATHY

4.1Galactose feeding

The importance of hyperglycemia per se in the pathogenesis of diabetic

retinopathy was demonstrated a number of years ago by study of normal, nondiabetic dogs fed a galactose-rich diet.83,84 During the 3 to 5 years of

study, normal dogs fed a diet enriched with 30% galactose developed a retinopathy that was indistinguishable from that of diabetic dogs and patients, including microaneurysms, vaso-obliteration, pericyte ghosts, and hemorrhages.22,35,83-93 Likewise, experimental galactosemia has also been shown to cause diabetic-like retinal lesions in rats and mice. Rats fed a 50% or 30% galactose diet for more than 1.5 years develop a significantly greater than normal prevalence of acellular capillaries and pericyte ghosts, excessive thickening of capillary basement membrane and, eventually, IRMAs.35,94-100 Mice fed 30% galactose also develop diabeticlike retinopathy, including rare but unmistakable saccular microaneurysms, as well as acellular capillaries, pericyte ghosts, and capillary basement membrane thickening.59,101

The galactose-retinopathy model has been utilized extensively for studies

of the role of aldose reductase in the pathogenesis of “diabetic-like” retinopathy,22,35,85-98 but more recently the model has been used in studies of

the role of leukostasis in retinopathy,59 and the ability of aminoguanidine,

antioxidants, and antisense mRNA against fibronectin to inhibit retinopathy.99,100,102

As a means for producing a model of diabetic retinopathy in animals, experimental galactosemia can be advantageous because it is easily established and requires less nursing care than experimental diabetes. Not to be overlooked, however, is the expense of the galactose diet, which can be costly if animals are large or numerous. Moreover, the galactose-induced retinopathy has at least two important differences from that in diabetes. First,

it develops despite the absence of many of the systemic metabolic abnormalities that are characteristic of diabetes (such as those involving concentrations of glucose, insulin, fatty acids, etc.).84 This is valuable, in that it demonstrates that excessive blood hexose (either glucose or galactose) is important in the initiation of retinopathy. The second difference between the retinopathies induced by diabetes and galactose feeding is a different response to at least one therapy. Aminoguanidine has been shown several

times to inhibit retinal microvascular disease in diabetic dogs and rats,99,103-105 but has not been found to do so in galactose-fed rats.99,106

Moreover, caspases activated in diabetic mice differ from those induced in galactose-fed mice.56 Thus, although the final histopathology induced by galactosemia seems morphologically identical to that in diabetes, the biochemical steps leading to that pathology apparently differ between the two models. The galactose model of retinopathy is a valuable source of comparison to diabetes, but it should not be assumed to respond to therapy like diabetic animals or patients would without comparing the two models first. Neurodegeneration has not yet been assessed in galactosemic models.

4.2Sucrose or fructose feeding

Nondiabetic rats fed very high concentrations of sucrose or fructose (approximately 70% in the diet) also have been reported to develop retinal

lesions, including loss of pericytes and endothelial cells, and formation of capillary strands,107,108 but these models have been used little in the past

decade.

4.3VEGF overexpression

Vascular endothelial growth factor (VEGF) was injected into the eyes of normal cynomolgus monkeys, and as a result, capillary nonperfusion and vessel dilation and tortuosity developed.109 Preretinal neovascularization was observed throughout peripheral retina, but not in the posterior pole. Arterioles demonstrated endothelial cell hyperplasia and microaneurysmal dilations. Thus, pharmacological doses of VEGF alone were able produce many features of nonproliferative and proliferative diabetic retinopathy.

4.4IGF overexpression

Normoglycemic/normoinsulinemic transgenic mice overexpressing insulin-like growth factor-1 (IGF-1) in the retina developed several vascular alterations characteristic of diabetic retinopathy, including nonproliferative lesions (pericyte loss, thickened capillary basement

membrane, intraretinal microvascular abnormalities), proliferative retinopathy, and retinal detachment.110

4.5Sympathetic denervation

Several retinal lesions consistent with diabetic retinopathy also have been detected after sympathectomy.111 Experimental elimination of sympathetic innervation to the eye by removal of the right superior cervical ganglion resulted in an increase in glial fibrillary acidic protein (GFAP) staining in Müller cells, reduced number of capillary pericytes, and alteration in expression of proteins found in basement membrane.

5.RETINAL NEOVASCULARIZATION

To date, diabetic animal models (without other genetic modifications or experimental manipulations) have not been demonstrated to reproducibly progress to preretinal neovascularization, and some have criticized the available diabetic models for this failure. In fairness, however, most patients do not develop preretinal neovascularization even after many years of diabetes. Moreover, diabetic or galactosemic animals (at least dogs) have been demonstrated to develop new intraretinal vessels (the precursor to preretinal neovascularization). The new vessels, identified by their lack of basement membrane112 and their characteristic ‘chicken-wire’ pattern, developed within the retina in diabetic dogs and experimentally galactosemic dogs, but did not extend into the vitreous during the initial 5 years of study. New vessels extending into the vitreous have been reported in 2 of 9 dogs fed galactose for 6 to 7 years.89 The diabetic Ren-2 rat develops proliferation of retinal endothelial cells, but overt neovascularization has not yet been demonstrated. This endothelial proliferation can be attenuated by RAS blockade via VEGF-dependent pathways.113 In general, the overall evidence indicates that diabetic or galactosemic animals are not a good model for studies of preretinal angiogenesis.

Although preretinal neovascularization has not been detected in diabetic

rodents, diabetes is known to increase retinal concentrations of VEGF106,114-118 and other growth factors119,120 in these animals.

Why diabetes or hyperglycemia in animal models fails to elicit preretinal neovascularization such as that which occurs in diabetic patients is an important question. Vaso-obliteration and subsequent retinal ischemia are believed to be major causes of neovascularization in the retina. Thus, one likely reason that the diabetic animal models do not develop preretinal neovascularization is that much less vaso-obliteration occurs in the retina of