The Complexity of Animal Model Generation for Complex DiseasesANIMAL MODEL GENERATION FOR COMPLEX DISEASES

Commentary

Age-related macular degeneration (AMD) is the most common cause of severe vision loss in older persons. Patients with AMD develop thickening of the Bruch membrane, which is a sheet of connective tissue that separates the retinal pigmented epithelium and retina from the highly vascular choroid. Deposits form along the Bruch membrane called drusen, and gradual degeneration occurs of the retina and the retinal pigmented epithelium. These features are similar to those that occur in some other neurodegenerative diseases and can lead to gradual loss of central vision. Approximately 10% to 20% of patients also develop neovascular AMD, in which abnormal blood vessels from the choroid grow through the Bruch membrane (choroidal neovascularization; CNV), resulting in sudden and severe reversible vision loss from collection of fluid beneath and within the retina and eventual permanent vision loss from scarring. Exactly how and why this complex phenotype develops in some elderly individuals and not others is a mystery. Modeling the disease in animals may help to solve the mystery, but it is a difficult task.

A common strategy for modeling monogenic diseases is to use genetic engineering to replace a wild-type allele with a mutant allele. If this approach results in the disease phenotype, the biochemical steps that occur between production of the mutant protein and the development of the phenotype can be investigated with the hope of identifying molecular targets for intervention. However, the phenotype of a knock-in mouse may vary considerably from a human disease phenotype due to species differences in the disease pathway or simply due to the shorter life span in mice, preventing a slowly progressive disease from developing. An example is Sorsby fundus dystrophy, a monogenic disease that shares some features with AMD. In patients with Sorsby fundus dystrophy, a mutation in the gene encoding tissue inhibitor of metalloproteinase 3 (TIMP3) results in deposits in the Bruch membrane, degeneration of photoreceptors, and CNV.¹ Knock-in mice homozygous for the pathogenic mutation have a different, less severe phenotype.² Such models are still useful, but do not provide all the information needed to completely understand the disease. Expression of mutant proteins in species closer to humans on the phylogenetic tree may provide a model that more closely mimics the human disease, but it is an expensive proposition.

These difficulties in generating models for monogenic diseases are greatly amplified when trying to produce models for complex, multigenic diseases, such as AMD. A forward genetic approach is particularly difficult because disease-associated single-nucleotide polymorphisms may only be markers and not directly involved in pathogenesis. Even when actual mutations are identified, the mechanism by which they generate disease may depend on complex genetic interactions and environmental exposures. Another approach is to use known or suspected risk factors for a disease to try to generate a disease phenotype.

In the February issue of the Archives of Ophthalmology, Albert et al³ used this approach and exposed albino rats to intense light for 12 hours each day for 1, 3, or 6 months. After 1 month, there was extensive degeneration of photoreceptors, staining for markers of oxidative damage, and CNV beneath the retinal pigmented epithelium. After 3 months, almost all photoreceptors had degenerated and there was more advanced CNV extending into the retina. After 6 months, there were no photoreceptors, multiple foci of CNV, and proliferation of retinal pigmented epithelial cells.

Thus, the authors have generated a new model of CNV, but how well does it mimic neovascular AMD? It involves oxidative damage that has been implicated in neovascular AMD,⁴ using excessive light exposure as the source of the damage. Light exposure might be involved in the human disease, but this has not been proven, either because its role is small or because of methodological difficulties in measuring light exposure. Photoreceptor degeneration occurs in AMD, but the extent of degeneration in this model is far greater than that typically seen in AMD patients. However, the biggest shortcoming is that this model is generated solely by environmental exposure, whereas AMD is a multigenic disease in which genetic susceptibility plays a substantial role. Disease risk is increased by many sequence variants, several of which are in genes of the complement pathway, suggesting that complement-mediated damage plays a role in AMD.⁵

Another use for an animal model is to test potential therapies, which does not require that the model mimic all aspects of a human disease, but rather that it recapitulates certain critical features that allow prediction of whether therapeutic agents are likely to provide benefit in the human disease. In addition to neovascular AMD, CNV occurs in other diseases including ocular histoplasmosis, angioid streaks, pathological myopia, and choroidal ruptures. In many of these entities, there are defects in the Bruch membrane through which blood vessels grow from the choroid into the subretinal space. Previous investigators used laser photocoagulation to create defects in the Bruch membrane in monkeys,⁶ and subsequently in mice,⁷ and demonstrated that CNV occurred. In these models, vascular endothelial growth factor (VEGF) was an important stimulus for the development of CNV⁸ ; and another model of subretinal neovascularization was generated by expressing VEGF in photoreceptors.⁹ These findings led to the prediction that VEGF antagonists would provide benefit in patients with CNV, including those with neovascular AMD. Clinical trials proved the hypothesis correct and validated the models as predictive,¹⁰ even though the insult that leads to the development of CNV is not the same as that occurring in patients with AMD. These models are now widely used for preclinical testing of agents being considered for treatment of neovascular AMD. The new model described by Albert et al³ also may be useful in this regard.

Model development for investigation of disease pathogenesis is an iterative process. As new information becomes available, it can be incorporated and provide more complex models that allow testing of additional hypotheses. Albert et al³ have provided a useful contribution demonstrating how a particular environmental exposure can cause disease that mimics some aspects of neovascular AMD. Additional studies to determine how this or similar environmental exposures interact with genetic variants that increase the risk of AMD will help place these findings in perspective and could provide important new insights into the pathogenesis of AMD.