The current study elucidated the crutial role of GPx4 in the RPE in mice, revealing that RPE-originated lipid peroxidation leads to RPE degeneration resembling key features of the late stage of dry AMD. Animal models are indispensable in medical research, aiding in the understanding and treatment of human diseases. In the case of wet AMD, the development of laser-induced choroidal neovascularization model was pivotal for the establishment of anti-VEGF therapy. Similarly, various gene-modified mice related to lipid metabolism have been reported to recapitulate age-related early and intermediate changes of AMD.13,15,55,56 Additionally, deficiency of anti-oxidative enzymes, including SOD1,57 SOD2,58 and Nrf259,60 has been linked to early and intermediate AMD phenotypes.
However, the roles of these models in the development of GA, the late-stage manifestation of dry AMD, have remained unclear. Previous models take 6–12 months, often up to 2 years, to sufficiently develop early/intermediate AMD phenotypes, such as drusen and subretinal drusenoid deposits. While these models underscore the importance of age in AMD pathogenesis, their utility as preclinical animal models may be limited. In contrast, the laser-induced choroidal neovascularization model for wet AMD can be applicable regardless of the mouse age, and it takes only one week from laser irradiation to the assessment of neovascularization. In the late stage of dry AMD, the scarcity of clinically relevant animal models has posed a significant challenge, leading researchers to rely on injections of artificial toxicants including NaIO3. Previously reported Dicer1-deficient mice61,62 could have served as a potential genetic mouse model for GA, but wider adoption in research laboratories has yet to be realized. GPx4 cKO mice demonstrate that the toxicity of lipid peroxidation products in the RPE can induce RPE degeneration similar to GA. Although lipid peroxidation in the RPE has long been considered a major pathogenic factor in AMD development,13–15 it has lacked direct in vivo evidence, especially for the late stage of dry AMD or GA.
GPx4 cKO mice exhibited key features resembling the late stage of dry AMD. Initially, RPE cells underwent a loss of polarized structures, including apical microvilli and basal infoldings. As degeneration progressed, we observed the loss of RPE cells, loss of photoreceptors, accumulation of lipofuscin, presence of subretinal/intraretinal melanophages, and activation of complement. Additionally, subretinal voids were observed in GPx4 cKO mice using both light microscopy (Fig. S2) and TEM (Fig. 4D and 4F). Although the significance of these voids remains unclear, similar images have being reported in previous studies on AMD mouse models.40,63 Interestingly, extracellular lipid deposition beneath the RPE (resembling drusen) or in the subretinal space (resembling subretinal drusenoid deposits), characteristics of early/intermediate AMD, were not evident in GPx4 cKO mice. This absence may be attributed to the rapid degeneration of RPE cells, potentially limiting the time required for the gradual lipid deposition in the extracellular space. However, in the RPE-choroid of GPx4 cKO mice, we observed a significant accumulation of toxic lipid peroxidation products, including acrolein and MDA. Acrolein can be efficiently formed from PUFA,34 and is also a component of tobacco smoke, which is recognized as the most significant modifiable risk factor for AMD.35 Additionally, acrolein is utilized in cellular models that mimics smoking-induced RPE cell death during AMD progression.36 MDA, another observed product, is a constituent of lipofuscin in human RPE.37 Recently, increased MDA levels in the RPE has been reported in a slowly progressive early/intermediate AMD model.38 Furthermore, resent research has demonstrated the transport of intracellular lipids from RPE cells to the extracellular space, contributing to the formation of drusen, in a slowly progressive AMD model using a focused-ion beam scanning electron microscopy.42
RPE degeneration in GPx4 cKO mice was alleviated by α-tocopherol (vit E) and Fer-1, accompanied by the observation of shrunk mitochondria, indicating the involvement of the ferroptotic mechanism. Interestingly, RPE degeneration was partially rescued by Nec-1s, suggesting the potential role of necroptosis in this process. Activation of necroptosis pathway was confirmed by presence of activated RIP3 and MLKL, along with inactivated caspase-8. We consider that GPx4 to protects RPE cells from cell death by suppressing both ferroptosis and necroptosis pathways. While most studies have depicted GPx4 as a suppressor of ferroptosis, there are reports suggesting its involvement in inhibiting necroptosis in certain cell types. For example, a study demonstrated the inhibition of necroptosis by GPx4 in erythroid precursor cells.54 Moreover, PUFA treatment on cancer cells induced lipid peroxidation-dependent necroptosis evidenced by elevated phosphorylation of RIP3 and MLKL, which was blocked by Fer-1.64 Furthermore, it has been shown that necroptosis and ferroptosis are alternative pathways in cultured fibroblasts. Deficiency of acyl-CoA synthetase long-chain family member 4 (ACSL4) results in resistance to ferroptosis but sensitivity to necroptosis, whereas deficiency of MLKL results in resistance to necroptosis as well as sensitivity to ferroptosis.65 These findings suggest that the mechanisms for GPx4-mediated cell death prevention may vary depending on the cell types, and it is possible for necroptosis and ferroptosis to coexist. Further research will elucidate the molecular mechanisms underlying lipid peroxidation-induced RPE cell death.
In conclusion, our current study unveiled the critical role of GPx4 in maintaining the homeostasis of RPE in mice. The synthesis of lipid peroxidation products in the RPE led to the manifestation of features resembling the late stage of dry AMD within a relatively short time period. GPx4 cKO mice offer a promising avenue to further explore the intricate relationship between lipid peroxidation and RPE degeneration. Additionally, they provide a valuable new option as a mouse model for preclinical research on GA.