Experimental disease models for AMD attempt to mimic what are thought to be important disease-contributing conditions in the retina, but they suffer from one major limitation: AMD is a chronic disease that takes years to develop, while the existing cellular and animal models produce morphological and functional characteristics on a weeks to months timescale. Largely this is out of necessity for timely evaluation of new chemical entities (NCEs), but partly this is due to the difficulty of efficient disease phenotype generation with a weak but long-lasting stimulus, which probably more accurately reflects pathology in man. As with most diseases, in vitro AMD models provide convenient screening tools and frequently use RPE cells, while being limited by the lack of heterogeneous, dynamically interacting components (e.g., RPE, photoreceptors, and Bruch's membrane) that an animal model provides. Nonhuman primates would constitute the preferred in vivo model; however, rodent models are mostly used due to ethical and cost issues and the availability of mouse genetic manipulation. Anatomic differences that may lessen the translatability of results from rodent models include a different surface volume/area ratio of eyes between man and rodents and the fact that rodents do not have a macula and thus technically cannot develop AMD.2 Also, because current disease models focus on a single disease-inducing stimulus, they do not accurately reflect the RPE dysfunction-photoreceptor dropout-choridal capillary neovascular breach of Bruch's membrane-retinal detachment disease etiology, and thus might be expected to overestimate the anti-AMD efficacy of agents that are effective in those models.
In vitro models of dry AMD typically subject an anatomically relevant cell type, such a photoreceptor or an RPE cell, to an AMD risk factor-linked stressor, e.g., high light flux or oxidative insult. The ability to prevent cytotoxicity by genetic manipulation or intervention with a protein or small molecule can then be measured.45 Although the aforementioned lack of a complex, dynamically interacting matrix requires caution to interpretation of cell culture results, these models provide a useful filter for selection of candidates for evaluation in an animal model.
Most current in vivo models of dry AMD can be classified into two categories: genetic models and light-damage models, sometimes in combination.46 Light-damage models typically expose rodents to bright white or blue light for a set period of time.47 Popular genetic models used for drug candidate screening include:
• The Royal College of Surgeons (RCS) rat, which contains a recessive genetic defect that prevents RPE phagocytosis of photoreceptor shed outer segments, leading to photoreceptor and RPE cell death (the specific genetic defect causing retinal degeneration having been determined48).
• A rat with a proline-to-histidine mutation at amino acid position 23 of rhodopsin (P23H rat), which, although is more appropriate as a model for the degenerative disease, retinitis pigmentosa, is used as an AMD model.49
• The abcr ~/ _ mouse in which the ABCR protein, involved in transport of 11-trans-retinaldehyde to the RPE cell for recycling back to the photoactive 11-cis isomer, is knocked out.50
Other genetic animal models of retinal diseases, have been reviewed.51 One mouse model of the highly significant complement dysregulation-AMD link described above, where monocyte chemoattractant protein-1 (MCP-1) or its cognate receptor chemokine receptor-2 (CCR-2) is knocked out.52 Appealingly, unlike other models it reproduces much of the human disease pathology, from drusen accumulation under the RPE to photoreceptor dropout to progression to choroidal neovascularization (i.e., wet AMD).
The two most useful models of wet AMD are the oxygen-induced retinopathy (OIR) model and the laser-induced CNV model.53 In the OIR model, neonatal animals (usually mice or rats) are initially exposed to a high oxygen concentration for 1-2 weeks, after which time room air is introduced. During the early hyperbaric period the disappearance of many capillaries is observed. In contrast, the lower oxygen concentration of room air is perceived as a hypoxic condition by the retina, evoking sprouting of new capillaries from existing ones. The neovascular response is believed to be due to secretion of pro-angiogenic proteins, e.g., VEGF. The OIR model simulates retinopathy of prematurity, a blinding disease sometimes observed in premature infants that are typically on oxygen support therapy at the beginning of life due to insufficient lung development. The development pattern of neovascularization is also somewhat reminiscent of that in proliferative diabetic retinopathy. However, since the OIR model does not involve the CNV response with breach of Bruch's membrane that is a hallmark of wet AMD, there are questions as to whether it is an appropriate stand-in for the disease.
The laser-induced CNV model involves rupture of Bruch's membrane, leading to an inflammatory/wound-healing response and concomitant CNV. Presumably the neovascularization response is due to upregulation of VEGF and other pro-angiogenic factors. This perhaps mimics wet AMD disease pathology better than the OIR model, in that the choroidal capillaries are explicitly involved in the neovascular response, and this model produces a similar angiographic appearance to the disease. This model, however, provides a more powerful angiogenic stimulus than likely occurs in wet AMD due to the more extensive injury to Bruch's membrane.
Was this article helpful?