ally associated with neurodegenerative diseases, which can be broadly categorized into histological, biochemical, and functional pathologies. This chapter will present evidence for retinal neurodegeneration in diabetes, segregated according to these three categories, and will finish by including a brief summary of the theorized mechanisms.

HISTOLOGICAL EVIDENCE

Early Pathology Studies

Early efforts to characterize the histology of diabetic retinopathy were the first to identify potential neuropathy accompanying the vascular changes. An early study of histological sections from postmortem specimens noted atrophy of retinal ganglion cells (RGCs) as one of the pathological changes that accompanied vascular lesions, and suggested that diabetes may induce a gradual loss of neurons as the disease progresses [2]. A similar study on a larger number of specimens also identified degeneration of the inner plexiform and ganglion cell layers as common features in humans with diabetic retinopathy [3]. Later, a paper by Bresnick suggested that neurodegeneration could possibly be viewed as a neurosensory disorder that involved degeneration of the neural retina, possibly preceding the vascular lesions [4]. One common feature of diabetic retinopathy that can be recognized by clinical observation is the appearance of “cotton wool spots” which are thought to be the axoplasmic debris from atrophied neurons in the nerve fiber layer (NFL) [5], and can appear as an early pathological feature in some patients [6].

Histological Evidence of Apoptosis

Studies of tissue from human and animals with diabetes identified apoptotic cells in the retina. In some cases, these included RGCs reviewed recently by Kern and Barber [7]. Many histological studies of apoptosis have used the classic technique of DNA terminal dUTP nick end labeling (TUNEL), which most commonly uses terminal transferase to label nuclei-containing DNA nicks in fixed tissue sections [8–10]. An early study using TUNEL identified apoptotic cells in cross sections of retinas from streptozotocin (STZ)- diabetic rats, although quantification of the numbers of neurons was not possible in this study [11]. Another study, using the trypsin-digest approach to specifically examine the vasculature of rat retinas, indicated a modest increase in TUNEL-labeled nuclei in rats after 6–8 months of STZ diabetes, suggesting that vascular cells also underwent apoptosis [12]; a finding that has been confirmed by others [13–15].

While trypsin digest makes it possible to specifically examine the vasculature of the retina, TUNEL labeling in intact flat-mounted retinas from STZ-diabetic rats made it possible to quantify the numbers of cells undergoing apoptosis in the entire retina. Using this approach, it was found that diabetic rats had significantly more TUNEL positive cells with a similar rate of cell death in groups of rats after 1, 3, 6, and 12 months of hyperglycemia (Fig. 1). The absolute number of positive cells was greater than in the trypsin-digest studies, suggesting that neurons and glial cells were also involved [16]. Others showed that TUNEL labeling was also increased in mouse models of both type I and type II diabetes [17, 18], and quantification of TUNEL labeling in whole retinas from Ins2Akita mice, a spontaneously diabetic genetic model, found a frequency of apoptosis similar to the STZ-diabetic rats [19].

Fig. 1. Diabetes increased apoptosis in whole rat retinas. Apoptotic cells were identified by TUNEL in whole retinas of STZ (streptozotocin)-diabetic rats after 1, 3, 6, and 12 months of hyperglycemia. The total number of positive nuclei in each retina was counted by microscopy. There were significantly more apoptotic cells in the retinas from diabetic rats (black circles) compared to controls (white circles), *p < 0.01, **p < 0.001, 1-way ANOVA with Newman-Keuls test. Taken from Barber et al. [16].

A variety of other histological studies have confirmed the increase in TUNEL labeling in diabetic animals, although the types of cells and the degree of apoptosis vary widely. An early phase of TUNEL labeling in photoreceptors was indicated in one study, accompanied by several indications of degeneration in amacrine, horizontal, and ganglion cells [20], although one study reported no significant increase in apoptosis of nonvascular cells in STZ-mouse retinas [21]. The rate of retinal apoptosis in diabetic rats was further increased by experimentally induced intraocular hypertension, similar to that in glaucoma [22].

As an alternative or additional approach to using TUNEL to detect cells undergoing apoptosis, some investigators have used antibodies to the activated form of caspase enzymes in histological sections of retina. Caspases-3 and -7 are often referred to as “executioner enzymes” because they cleave target proteins at specific aspartate recognition sequences. Antibodies raised to identify only the active form of caspase-3 can be used for immunohistochemical detection of cells undergoing apoptosis at the time of tissue fixation [23]. Using this approach, the number of cells positive for active caspase-3 was found to be elevated in the ganglion cell layer of retinas from mice after 2, 6, and 12 weeks of STZ diabetes [17]. A similar approach labeling for active caspase-3 in wholemount retinas from Ins2Akita mice found that, after 4 weeks of hyperglycemia, there were significantly more positive cells compared to nondiabetic age–matched litter mates [24]. There is also evidence that caspase-3 is activated in ganglion cells of postmortem retinas from subjects with diabetes [25]. Similarly, caspase-3 and -9 immunohistochemistry in human postmortem retinas colocalized with Fluoro-Jade B, an indicator of degenerating neurons, and was most abundant in cell bodies in the RGC layer [26]. Caspase-3 immunoreactivity was also found to colocalize with several other neuronal markers in flat-mount retinas of Ins2Akita diabetic mice, suggesting that the cells undergoing apop-

Fig. 2. Immunoreactivity for active caspase-3 did not localize to the vasculature. Whole retinas from STZ-diabetic rats were labeled by immunofluorescence for agrin, a vascular basement membrane glycoprotein (green) and active caspase-3 (red). The majority of caspase-3 positive cells were located away from blood vessels, suggesting that they were neural in origin, scale bar = 50 mm. Taken from Gastinger et al. [38].

tosis included RGCs, amacrine cells, and photoreceptors. Quantification of caspase-3 positive cells in these mice yielded a similar estimate of the total number of apoptotic cells compared to data obtained by TUNEL, and the majority of caspase-3 positive cells did not colocalize with agrin immunoreactivity in vascular basement membrane, indicating that the dying cells were mostly not vascular in origin [19] (Fig. 2).

Gross Morphological Changes in the Retina

Many studies investigating loss of neurons in the retina use measures of the thickness of retinal layers as a measure of cell loss [27]. STZ-diabetic rats had significantly reduced thickness of the inner plexiform and inner nuclear layers, 7.5 months after the onset of hyperglycemia [16] (Fig. 3), suggesting that the increased apoptosis identified in this study leads to an accumulated loss of cells making up the inner retina. A different study suggested that reduction in the thickness of the inner plexiform layer was accompanied by loss of the outer nuclear layers and increased TUNEL labeling primarily among photoreceptors in STZ rats after 6 months of diabetes [20]. Decreased thickness of the outer retina was also noted after 12 and 24 weeks of diabetes in STZ-diabetic rats [28].

Similar reductions in retinal layer thickness were measured in diabetic mice. The inner retina was reduced in Ins2Akita diabetic mice after 5 months of hyperglycemia, although the reduction was limited to the peripheral retina, suggesting that the cell loss may occur more slowly in the central region of the retina in this mouse model [24]. The reduction in inner retina thickness was comparable to previous observations in STZdiabetic mice that were diabetic for 10 weeks [17].

Fig. 3. Diabetes reduced the thickness of the inner retina in rat. The thickness of the inner plexiform layer (IPL), inner nuclear layer (INL), combined outer plexiform and outer nuclear layers (OPL + ONL), and entire retina (RET) were measured as a ratio with the choroid in H&E sections of eyes from rats after 7.5 months of diabetes (shaded bars) and compared to eyes from control rats (white bars). The IPL and INL were significantly thinner in diabetic rats (*p < 0.001) compared to controls. Taken from Vanguilder et al. [56].

Several studies have reported cell loss by measuring cell layer thicknesses in rodent models of diabetes; however, there are disparities in the rate of cell loss and whether the degeneration is predominantly inner or outer retina. The differences between these studies are difficult to explain but may be due to variations in the degree of induced diabetes, genetic background of the animals, and variations in animal husbandry, including the fat content of the diet [29], differences in handling, or exposure to environmental pathogens.

Reductions in Numbers of Surviving Amacrine Cells

Results of several morphological studies indicate that diabetes may deplete the number of amacrine cells in the retina. Tyrosine hydroxylase immunoreactivity, a marker of dopaminergic neurons, was reduced in amacrine cells of the obese sand rat, which becomes moderately hyperglycemic [30]. Necrosis of amacrine cells was also reported in STZ rats, along with photoreceptors, ganglion cells, and other neurons [20]. Tyrosine hydroxylase protein levels were also depleted by approximately 50%, accompanied by significant reduction in the density of dopaminergic amacrine cells [31]. Labeling of neuronal nitric oxide synthase (nNOS)-positive amacrine cells was also reduced in diabetic rats, suggesting a down regulation in the enzyme expression, or a loss of amacrine cells [32]. While the morphology of surviving amacrine cells appeared to be normal in whole-mount retinas from diabetic Ins2Akita mice after 6 months of hyperglycemia, they were reduced in number by 16–20% [19].

Reductions in neurotransmitters and associated enzyme activity also imply a loss of amacrine cells. The concentration of dopamine was significantly reduced in STZdiabetic rats after 3 weeks of hyperglycemia, while there was no change in tyrosine