Induced models have been created through five methods: surgical removal of the pancreas, administration of the drug alloxan, administration of the drug streptozotocin (STZ), high-galactose diets, and laser or chemical damage to the eye [
27,
35‐
46]. While all methods of induction are still studied today, the most common is STZ administration, as it results in the fastest rate of disease development [
41]. Alloxan is considered to be less efficient in diabetic induction, and dietary methods require the most time for disease progression [
43]. Surgery- and damage-induced models are the most technically challenging, limiting their use historically. The most frequently used models for inducing DR are mice and rats, but dogs, cats, pigs, rabbits, monkeys, and zebrafish are also used. Presentation of induced DR pathology is generally slower in larger animals, making rodents and, recently, zebrafish more favored models. A comparison of the available induced models can be found in Table
1.
Table 1
Phenotypes of induced and genetic animal models of DR
Induced models |
Pancreatectomy |
Dog | X | X | | | | | | | | X | | | | | | | |
Cat | X | X | | | | | | X | X | | | | | | X | | Capillary nonperfusion [ 47‐ 49] |
Monkey | X | X | | | | | | | | | | X | | | | | |
Alloxan |
Mouse | X | X | X | X | | X | | | X | | | | | X | | | Functional defects [ 51, 52] |
Rat | X | X | X | X | | | | X | | | | X | | X | X | | |
Dog | X | X | X | | | | | | X | X | | | | X | | | |
Pig | X | X | X | X | | | | | | | | | | | | | |
STZ |
Mouse | X | X | X | X | X | X | | | | | | | | X | X | | |
Rat | X | X | X | | | X | X | X | | | X | X | | X | | | |
Zebrafish | X | X | | | X | | | | | | | | | | | | |
Rabbit | X | X | | | | | | | | X | | | | | | | Vascular lesions [ 27, 64] |
Dog* | X | X | X | | | | | X | | | | | | | | | *Alloxan/STZ-induced [ 68, 76] |
Monkey | X | X | | | | | | | | X | | | | X | | | Ischemic retinopathy [ 27, 31] |
Pig | X | X | | | X | X | | X | | | X | | | | | | |
Diet |
Mouse | X | | X | | | | X | X | X | | | | | X | | | |
Rat | X | | X | X | | | | X | X | | | | | X | | | |
Rabbit | X | | | | | | | | X | X | | | | | | | |
Dog | X | | X | | | | | X | X | X | | | | X | X | | |
Zebrafish | X | | | | X | | | | | | | | | | | | |
Hypoxic damage |
Mouse | | | | | | | | | X | | | | | | X | | |
Rat | | | | | X | | | | | | | | | | X | | Abnormal vascular tufts [ 89‐ 93] |
Monkey | | | | | | | | | X | X | X | | | | X | | |
Zebrafish | | | | | | | | | | | | | | | X | | |
Genetic models |
Mouse |
Ins2Akita
| | X | | X | X | X | | | X | X | X | | | X | | | |
NOD | X | X | X | | X | X | | X | | | | | | | | | Disorder focal proliferation vessels [ 105‐ 113] |
db/db | X | X | X | | | X | | X | X | | X | X | | | | | |
Kimba | | | X | | X | | | | | | | | | | | | |
Akimba | | | X | | X | | | | X | | X | | X | | X | X | |
Rat |
BB | | X | X | | | | | X | X | | | X | | | | | |
ZDF | X | | X | | | | | X | | | | | | | | |
Lepr; increased capillary density [ 128, 129] |
OLETF | X | | X | | | | X | X | X | | | | | | | |
GPR10; leukocyte entrapment [ 131‐ 134] |
WBN/Kob | | | | | | | | | | | | | | X | X | | |
SDT | X | | X | | X | X | | | | | | | | | | | |
GK | X | | | | | | | | | | | | | | | | Increased endothelial cells [ 156‐ 161] |
Zebrafish |
Vhl
−/−
| | | | | | | | | | | | | X | | X | X |
Vhl; increased hyaloids [ 162] |
Pancreatectomy
One of the oldest methods used to induce diabetes in animal models is pancreatectomy, the removal of the pancreas or removal of β cells from the pancreas. Pancreatectomy was observed as early as 1922 to increase blood sugar levels in dogs [
35], and by 1968–1971, a technique of complete pancreatectomy in adult dogs had been developed to induce diabetes [
36,
37]. This technically difficult method is usually applied to large animals such as cats and monkeys [
27]. In adult cats, hyperglycemia develops within 3 weeks postsurgery; this time can be reduced to within 1 week by combining pancreatectomy with administration of alloxan [
47] (described in the “
Alloxan” section). Thickening of the basement capillary membrane can occur from 3 months after the onset of hyperglycemia [
48]. Other DR symptoms, including microaneurysm, intraretinal hemorrhages, capillary nonperfusion, and neovascularization, may take 5–9 years to develop [
49]. Maintenance of this model thus requires an extended period of time.
In monkeys, pancreatectomy at various ages between 6 and 15 years resulted in insulin dependency and hyperglycemia, which was then deliberately uncontrolled [
50]. This was observed to lead to BRB leakage within 1 year of hyperglycemia onset. However, 10 years postinduction, monkeys still did not develop proliferative retinopathy [
50]. Monkeys thus appear to be surprisingly resilient to induction of DR despite removal of the pancreas, long-term diabetes, and poor control of blood sugar levels. This is a similar phenomenon to humans where only 30% of individuals with diabetes develop DR, suggesting primates may have additional biological mechanisms to re-regulate homeostatic state in the presence of chronic insult. Understanding the regulatory factors that contribute to the physiological differences in species is important for developing appropriate disease models.
Alloxan
The first drug found to induce diabetes, alloxan, was discovered by Dunn and McLetchie in 1942 [
38]. Alloxan is a derivative of uric acid and directly targets β cells found in the pancreas [
39], and was first produced by Wöhler and Liebig through a reaction of uric acid with nitric acid. While conducting rabbit studies focused on kidney disorders, Dunn and McLetchie found that intravenous injection of alloxan resulted in hypoglycemia due to necrosis in the islets of Langerhans in the pancreas. Death of β cells led to the release of insulin stores in these cells, causing the observed hypoglycemia followed by onset of diabetes within 24 h. Dunn and McLetchie also created the diabetic rat model induced by alloxan via intraperitoneal administration. While the diabetic rabbits appeared listless and lost weight, rats that were given alloxan ate voraciously and presented with polydipsia, polyuria, glycosuria, and hyperglycemia, characteristic of diabetes [
38].
Alloxan-directed cell death is mediated by inhibition of glucokinase, an enzyme involved in the glucose-insulin regulatory pathway and expressed in the liver and pancreas. The drug can be toxic to liver and kidney cells, but with proper dosing, toxicity can be avoided. The action of alloxan in the pancreas is specific to β cells, with no toxic effect on a α, δ, or pancreatic exocrine cells. The compound is also unstable in water at room and body temperature, making it difficult to administer [
38,
40]. In recent times, alloxan has fallen in popularity in favor of STZ, described in the “Streptozotocin” section, due to the latter’s greater ease of use and efficacy.
Alloxan has been used to induce DR in a large variety of animals including mice, rats, dogs, and pigs, as well as the rabbits and rats [
28,
51•,
52‐
58]. All models experience damage to pancreatic β cells. Mice aged 8–10 weeks can be given a single dose of alloxan to induce hyperglycemia leading to diabetes [
51•]. It was previously believed that the alloxan-induced diabetic mouse did not develop cellular or vascular lesions, but a recent study found that alloxan does induce pericyte ghosts and loss of retinal ganglion cells (RGCs) within 7 days and microaneurysms with increased acellular capillaries by 21 days in mice from the FOT_FB strain [
51•]. Alloxan also induced microglial changes, with thicker cell bodies and shorter dendrites by 3 months of age in the same animals [
52].
Induction of DR by alloxan in rats is determined by weight (180–200 g weight) [
53,
54]. Within a week of alloxan administration, hyperglycemia and diabetes develop [
54,
55]. Neovascularization occurs between 2 and 9 months postinduction [
54] and cataracts within a year [
59]. Similar to the phenotypes observed for mice, pericyte ghosts, acellular capillaries, and thickened capillary basement membrane are observed by 15 months postinduction [
59,
60]. In addition, the alloxan-induced diabetic rat exhibits BRB breakdown [
55], expansion of Müller glia, and endothelial swelling [
54]. This model is typically studied for up to 22 months [
60].
Alloxan induces diabetes in young dogs by once a week administration for 5 weeks. This results in retinopathy remarkably similar to DR in humans however, dogs can take up to 53 to 69 months after onset of alloxan diabetes to develop DR [
28]. Following disease onset, alloxan-treated dogs present with hemorrhages, acellular capillaries, pericyte loss, and microaneurysms, making this a viable model of PDR. This phenotype persists for 11 months.
The porcine alloxan-induced DR model, in contrast to the dog models, develops hyperglycemia within 48 h [
61]. The molecular phenotype following induction is Müller cell contraction-promoting activity that is detectable as early as 30 days after alloxan administration, and sustains for up to 90 days. Alloxan-induced pigs also develop cataracts by 60 days following alloxan administration [
61] as well as BRB breakdown, capillary collapse, and pericyte ghosts were detected by 20 weeks [
56]. In contrast to other alloxan models that exhibit PDR like DR disease, the porcine alloxan-induced model of DR recapitulates several important markers of NPDR.
Streptozotocin
In 1963, Rakieten et al. reported that STZ administration causes diabetes in rats and dogs [
41]. STZ is an antibiotic produced by
Streptomyces achromogenes and was studied for use in cancer chemotherapy [
62]. Rakieten et al. studied intraperitoneal administration of STZ in rats and intravenous injection of STZ in dogs, both of which led to sustained hyperglycemia in each species, along with polyuria and polydipsia characteristic of diabetes [
41]. The mechanism of diabetes mellitus induction was found to be the disruption of pancreatic islets of Langerhans and loss of β cells due to STZ [
41]. β Cells take up STZ specifically because they express the low affinity glucose transporter 2 (GLUT2), and STZ is structurally similar to glucose and
N-acetyl glucosamine [
42]. Other cells that also express GLUT2, including hepatocytes and renal tubular cells, experience similar damage with STZ administration [
42]. STZ mechanism of action is cell death by DNA fragmentation.
Induction of DR by STZ has been observed in multiple models including mice, rabbits, pigs, rats, dogs, zebrafish, and monkeys [
31,
63-
70]. STZ is now generally preferred over alloxan, as it is more effective in recapitulating the diabetic disease state, though both drugs are still commonly used [
41]. Several protocols for STZ induction of diabetes in mice have been developed, ranging from 1 to 5 doses delivering a total of 150 to 400 mg/kg of STZ [
27]. Hyperglycemia onset typically occurs within 2 weeks, regardless of dosage [
27] and can be maintained for up to 22 months [
71]. DR phenotypes observed in STZ mice include increased number of astrocytes and gliosis 4–5 weeks after onset hyperglycemia [
63,
71], RGC loss at 6 weeks [
56], retinal inner nuclear layer (INL) and outer nuclear layer (ONL) thinning at 10 weeks [
72], neovascularization at 16 weeks [
73], and acellular capillaries and pericyte ghosts at 6 months [
71].
In contrast to mice, rats require lower doses of STZ to develop diabetes [
27]. The onset of retinal lesions differs between rat strains, but several observed phenotypes include BRB breakdown 2 weeks after diabetes onset [
74,
75], ONL thinning beginning the in the fourth week following induction [
74], increased acellular capillaries, decreased numbers of both pericytes and endothelial cells after 8 weeks [
67], and basement membrane thickening after 1 year [
68]. STZ-induced DR rat models are typically studied for up to 20 weeks [
75].
While rodents are commonly used for STZ-induced diabetes, several other models have been studied with various outcomes and onset of disease. Adult zebrafish, 4–6 months of age, injected with multiple doses of STZ intraperitoneally or through direct caudal fin injection over one or several weeks, develop hyperglycemia and within 3 weeks, and display inner plexiform layer (IPL) thinning, photoreceptor segment layer (PSL) thinning, cone receptor dysfunction, and neuronal damage by 4 weeks [
69,
70]. This model is maintained approximately 80 days after induction of diabetes [
70].
Larger animal models such as rabbits, dogs, and nonhuman primates use a single dose protocol for STZ induction. A single dose of STZ can be given to rabbits weighing 1.5 kg to induce hyperglycemia, which after 135 days results in retinal and preretinal hemorrhages, vascular lesions, venous thrombosis, and proliferative retinopathy [
27,
64]. Beagles ranging in age from 4.5 to 17 months and weighing between 11 and 24 kg given a single dose alloxan/STZ cocktail develop hyperglycemia within 2 days [
68]. Alloxan/STZ-induced diabetic dogs present with basement membrane thickening after 1 year and pericyte ghosts and smooth muscle cell loss after 4–5 years [
76]. This model is studied for 7 years [
76]. Interestingly, monkeys treated with a single dose of STZ at age 12 develop diabetes, then ischemic retinopathy with cotton-wool spots and hyperfluorescent spots after 10 years [
27,
31]. Interestingly, the induction of hypertension is required for retinopathogenesis in this model, as monkeys without hypertension fail to develop retinopathy [
31]. The porcine model of STZ is induced, at 20 kg, with STZ administration for three consecutive days [
65]. Induced pigs develop hyperglycemia within 1 week and are studied for up to 32 weeks. Diabetes lasting 4–8 months after STZ induction results in increased BRB permeability, thinning of the INL and ganglion cell layer (GCL), and thickening of the capillary basement membrane [
65,
77]. When STZ-induced pigs are subject to hyperlipidemic diets, they acquire dyslipidemia similar to that experienced by patients with type 2 diabetes. Diabetic pigs also experience increased BRB permeability, as well as compromised tight junctions in the retina. The pig’s large size and hierarchal vascular structures make its metabolic and circulatory functions highly similar to humans [
66], thus making it a common model for DR.
High-Sugar Diets
Kern and Engerman first reported an animal model of DR induced by galactose-heavy diet [
43]. Several high-sugar diet models have been developed including mice, rats, rabbits, dogs, and zebrafish [
33,
43,
44,
60,
78‐
83] that were persistently exposed to galactose developing retinopathy similar to that observed in human diabetes. Maintenance of galactose exposure results in continued disease progression. However, galactose-fed animals lack some metabolic abnormalities experienced in diabetes [
20]. Mice developed hyperglycemia by 6 weeks of age following high-galactose diet [
44]. After 15 months of hyperglycemia, endothelial cell loss and increased acellular capillaries were observed [
44,
78]. After 21 months, lesions including pericyte ghosts, microaneurysms, and retinal thickening are observed [
27,
44,
79]. While retinopathy takes longer to develop in these mice, they do live longer than other models, allowing them to be observed over a longer period of time, up to 26 months [
78]. Similarly, rats have been kept on high-galactose diets for over 2 years. Phenotypes observed in rodents on a continuous high-sugar diet include pericyte ghosts, acellular capillaries, and capillary basement membrane thickening by 18 months of hyperglycemia [
60,
80], as well as gliosis and microaneurysm by 28 months [
20]. While rodents can develop diet-induced DR, drug-induced and genetic models are more commonly studied in small animals due to their faster onset of disease.
In contrast, larger animals generally take longer to develop DR whether by drug induction or diet. Rabbits fed a high-sucrose diet for 24 weeks develop hyperfluorescent dots and microaneurysms appeared by the 12th week of the diet [
27,
81]. Dogs fed a diet with 30% increased galactose develop a more complex disease phenotype including DR and cataracts within 1 year; pericyte ghosts, microaneurysms, dot and blot hemorrhages, and acellular capillaries by 32 months; and basement membrane thickening by 60 months [
43]. As observed with all dog models of DR, disease can take many years to develop, but phenotypes in the dog are most similar to those in humans [
27,
29,
30,
80,
84].
Most recently, hyperglycemic zebrafish have been developed as a model for DR. Zebrafish are housed in freshwater with alternating concentration 0 and 2% glucose every other day and develop hyperglycemia after 28 days and IPL thinning [
82]. As this model has only been maintained for 28 days to date, several attributes including similar retinal topography, ease of vascular structures visualization with fluorescent expression [
33,
83], short life span, and large breeding size reduce experimental time and make zebrafish a strong model to study DR [
27].
Hypoxic Damage-Induced Retinopathy
Models of retinal neovascularization and vasculature leakage lacking hyperglycemia have been used in recent years to study DR. These models simulate advanced-stage PDR observed in human patients. In a 1969 study, Dollery, Bulpitt, and Kohner exposed newborn kittens to hyperoxic conditions and found that returning the kittens to normal air made them experience hypoxia, leading to neovascularization [
85]. It was later discovered that retinal damage induced the release of angiogenesis factors [
45]. This discovery led to a number of different damage models for retinal neovascularization using mouse, rats, primates, and zebrafish [
27,
46]. Hyperoxic mouse models are generated by exposing juvenile mice, typically postnatal days 7–12, to hyperoxic conditions, which results in hypoxic conditions of the retina once they return to normal air and the growth of blood vessels in the retina [
86,
87]. These models of oxygen-induced retinopathy (OIR) exhibit neovascularization and nonperfusion, accompanied by the appearance of microaneurysms, which typically occur within 5 days postexposure [
88].
Similar to mice, OIR in nondiabetic rats results in neovascularization. Rat pups were exposed to the hyperoxic conditions between 11 and 14 days [
89,
90]. Neovascularization is apparent immediately once rats are returned to normoxic conditions followed by astrocyte degeneration [
91] and subsequent reduction in INL and IPL thickness, with disorganized outer segments [
92,
93]. A distinct feature of this model is the incomplete development of the vascular plexus and the presence of abnormal endothelial tufts [
91]. Two nonrodent OIR models, monkey and zebrafish, also develop neovascular disease. OIR-induced neovascularization in zebrafish requires for the animal to be placed in normoxic water followed by the gradual reduction of O
2 tension over a period of 48–72 h until reaching 10% of air saturation (820 ppb) [
94]. Zebrafish can be maintained in this environment for up to 15 days [
95]. After exposure, neovascularization is evident as well as reduction in intercapillary distance [
94]. The primate model of OIR-induced neovascularization is distinct in that induction is localized by laser vein occlusion. Thus, focal regions of hypoxia are created rather than a whole organism exposure to hypoxic conditions. Retinal neovascularization in OIR primates typically occurs 4–7 days postexposure. Hypoxic monkeys show vascular leakage, venous occlusion, capillary nonperfusion, venous dilation, and dot and blot hemorrhages, which result from microaneurysm ruptures [
96]. Interestingly, a primate model was used to develop anti-VEGF treatment [
96].