Intraocular iron injection induces oxidative stress followed by elements of geographic atrophy and sympathetic ophthalmia

Abstract Iron has been implicated in the pathogenesis of age‐related retinal diseases, including age‐related macular degeneration (AMD). Previous work showed that intravitreal (IVT) injection of iron induces acute photoreceptor death, lipid peroxidation, and autofluorescence (AF). Herein, we extend this work, finding surprising chronic features of the model: geographic atrophy and sympathetic ophthalmia. We provide new mechanistic insights derived from focal AF in the photoreceptors, quantification of bisretinoids, and localization of carboxyethyl pyrrole, an oxidized adduct of docosahexaenoic acid associated with AMD. In mice given IVT ferric ammonium citrate (FAC), RPE died in patches that slowly expanded at their borders, like human geographic atrophy. There was green AF in the photoreceptor ellipsoid, a mitochondria‐rich region, 4 h after injection, followed later by gold AF in rod outer segments, RPE and subretinal myeloid cells. The green AF signature is consistent with flavin adenine dinucleotide, while measured increases in the bisretinoid all‐trans‐retinal dimer are consistent with the gold AF. FAC induced formation carboxyethyl pyrrole accumulation first in photoreceptors, then in RPE and myeloid cells. Quantitative PCR on neural retina and RPE indicated antioxidant upregulation and inflammation. Unexpectedly, reminiscent of sympathetic ophthalmia, autofluorescent myeloid cells containing abundant iron infiltrated the saline‐injected fellow eyes only if the contralateral eye had received IVT FAC. These findings provide mechanistic insights into the potential toxicity caused by AMD‐associated retinal iron accumulation. The mouse model will be useful for testing antioxidants, iron chelators, ferroptosis inhibitors, anti‐inflammatory medications, and choroidal neovascularization inhibitors.


| INTRODUC TI ON
Iron is widely distributed throughout the human retina and plays an essential role in physiological processes, including respiration and the visual cycle (Picard et al., 2020). Yet, when dysregulated, it can produce the most reactive of the free radicals: hydroxyl radical (Wong et al., 2007).
Iron has been shown to cause or exacerbate multiple retinal diseases. For example, iron accumulation has been reported in the retinas of patients with age-related macular degeneration (AMD) (Biesemeier et al., 2015;Hahn et al., 2003). Ocular siderosis due to an iron-containing intraocular foreign body causes rapid retinal degeneration, as photoreceptors are especially sensitive to iron toxicity, most likely due to their high concentration of oxygen and easily oxidized polyunsaturated fatty acids (SanGiovanni & Chew, 2005).

Retinal iron accumulation also causes retinal degeneration in mice
with hereditary retinal iron overload (Hadziahmetovic et al., 2008Hahn et al., 2004), which is protected by systemic treatment with the iron chelator deferiprone (Song et al., 2014).
Previous studies focusing on acute effects reported that intravitreal (IVT) iron injection leads to lipid peroxidation, photoreceptor degeneration, and increased retinal pigment epithelium (RPE) lipofuscin (Dunaief, 2006;Rogers et al., 2007;Shu et al., 2020). Anderson et al. (1984) reported that frogs with IVT iron injection had decreased polyunsaturated fatty acids and increased lipid hydroperoxides in isolated rod outer segments. Hiramitsu et al. (1976) reported photoreceptor degeneration induced by IVT injections of linoleic acid hydroperoxide in rabbits. Ferroptosis, an irondependent programmed cell death pathway, is accompanied by lipid peroxide accumulation (Yang & Stockwell, 2016) and occurs in the sodium iodate retinal degeneration model (Tang et al., 2021). Taken together, iron may induce retinal degeneration via peroxidation of polyunsaturated fatty acids.
Iron chelation by deferiprone decreased A2E oxidative degradation and protected against cell death (Ueda et al., 2018).
Previously, we studied the acute effects of IVT ferrous sulfate injection, which induced photoreceptor degeneration with increased oxidative stress and lipid peroxidation (Shu et al., 2020). Here, we used ferric ammonium citrate (FAC), which is commonly used experimentally to load cells with iron, is more stable than ferrous sulfate, and yields more reproducible amounts of retinal damage. We extended prior studies with IVT iron by elucidating novel iron-induced changes in bisretinoids and production of carboxyethyl pyrrole, a specific docosahexanoic acid oxidation product implicated in AMD.
We also found, for the first time, progressive geographic atrophy of the RPE, and, unexpectedly, iron-induced myeloid cells infiltrating the saline-injected control eyes of mice that had FAC injections in their fellow eyes.

| FAC injection induced acute pan-retinal AF and photoreceptor degeneration
To test our hypothesis that FAC, unlike ferric sulfate, would cause retinal toxicity because of FAC's higher solubility, we measured total soluble iron and ferrous iron in phosphate-buffered saline at pH 7.4 with vitreous levels of ascorbate (2 mM). The solution with 50 μM FAC (monoferric) added had 42.5 ± 3.5 μM soluble iron with 31.2 ± 1.4 μM ferrous iron. In contrast, the solution with 25 μM ferric sulfate (diferric) added had only 33.0 ± 0.7 μM soluble iron with 13.0 ± 1.6 μM ferrous iron.
Following IVT FAC injection, in vivo imaging was used to visualize retinal phenotypes. FAC injection caused pan-retinal hypopigmentation detected with color fundus photography and diffuse green emission in the AF photographs at 4 h after injection.
By 2 days after FAC injection, the pan-retinal hypopigmentation and AF became more pisciform and included some red AF emission. At 7 days after FAC injection, diffusely distributed hypopigmented spots were prominent on color fundus photography, and a mixture of gold or green-emitting hexagonal RPE cells, as well as F I G U R E 1 IVT ferric ammonium citrate (FAC) induced pan-retinal autofluorescence (AF) and photoreceptor degeneration. Color fundus photographs and fundus AF photographs were obtained at 4 h, 2 days, and 7 days after FAC or control saline injection (a). Horizontal optical coherence tomography scans through optic nerve acquired at 2 days (b) and 7 days (c) after injection. Outer nuclear layer area in horizontal optical coherence tomography scans (c, yellow line) was quantified using ImageJ (d). 55° wide field infrared AF cSLO images were acquired from the central retina and superior peripheral retina 7 days after FAC injection (e). Optical coherence tomography images acquired from the same mouse and time points as in panel e are shown in panel f. The positions of selected optical coherence tomography scans (g) were indicated in the optical coherence tomography fundus view (f, green line). Long posterior ciliary arteries (red dotted line) were used as a natural landmark for the orientation of images. 102° ultra-wide field blue-AF and infrared AF cSLO images were acquired at 7 days after FAC and saline injection (h). A magnified image of the 102° blue AF cSLO image (h1, yellow box) is shown (i). ERG was conducted at 2 weeks after injection, (j). For each set of images, representative images were chosen from N = 8-10. All yellow arrows indicate hyper-autofluorescent spots, red arrows indicate hypo-autofluorescent spots, and green arrows indicate hyper-autofluorescent binuclear RPEs. OCT, optical coherence tomography; ONL, outer nuclear layer; BAF, blue AF; IRAF, infrared AF; Sup, superior retina. Error bars indicate mean ± SEM. *p < 0.05, **p < 0.01, N = 3-5/ group for ONL thickness measurement gold-emitting myeloid cells were visible on fundus AF (Figure 1a and Figure 4 for histologic verification). A few hypopigmented patches appeared in the retina at both 2 days and 7 days after control saline injections, but these were larger sectoral lesions rather than small spots, and lacked AF, so they were quite distinct from FAC-induced lesions ( Figure 1a). In optical coherence tomography scans, the outer retina became hyper-reflective at 2 days after FAC injection ( Figure 1b), and the outer nuclear layer was markedly thinned by 7 days (Figure 1c). The outer nuclear area (yellow lines) was significantly decreased after FAC injection (Figure 1d). To evaluate fundus AF, scanning laser ophthalmoscopy short-wavelength (blue) AF and near-infrared AF imaging were performed. Wide-field near-infrared AF images ( Figure 1e) and optical coherence tomography volume intensity projections (Figure 1f) from the same eye at 7 days were selected to display the FAC-induced changes. The near-infrared AF images displayed intense pan-retinal hyper-autofluorescent (yellow arrows) and hypo-autofluorescent spots (red arrows), corresponding to hyper-reflective subretinal foci (yellow arrows) and hyporeflective spots (red arrows), respectively, on optical coherence tomography scans (Figure 1g). Atrophic or amelanotic RPE cells have been reported to be hypo-autofluorescent in infrared AF imaging (Duncker et al., 2014;Sunness et al., 2020). RPE atrophy was present in optical coherence tomography scans in the superior retina after FAC injection (right side of scan, yellow box). 102° ultra-wide field blue and infrared AF images showed FAC induced pan-retinal hyper-and hypo-autofluorescent spots, most likely representing myeloid cells (hyper-AF) and RPE atrophy (hypo-AF) (Figure 1h). An enlarged blue AF image from the superior retina ( Figure 1h, yellow box) displays clustered hyper-autofluorescent spots, representing RPE, myeloid cells, and photoreceptor layer undulations (yellow arrows), hyper-autofluorescent binuclear RPEs (green arrows), and hypo-autofluorescent foci of RPE atrophy (red arrows) (Figure 1i).

| IVT FAC caused photoreceptor and RPE AF and an increase in the bisretinoid all-transretinal dimer
Cryosections obtained from FAC-and saline-injected eyes were used to identify and localize FAC-induced autofluorescent cells.
FAC induced AF at 4 h, 2 days, and 7 days and 1 month after injection ( Figure 2a and b, white arrows). By 4 h after FAC injection, green-emitting AF was present in photoreceptor inner segments, and broad wavelength-emitting (both green and red) AF was apparent in some photoreceptor outer segments. To localize the FACinduced AF in the photoreceptor inner segments, immunolabeling with anti-TOMM20 was used to label photoreceptor mitochondria and imaged by confocal microscopy (Figure 2c,d). At 4 h after FAC injection, the green AF in photoreceptor inner segments co-localized with the TOMM20 labeling, suggesting FAC induced green-emitting AF (possibly flavin adenine dinucleotide, see Discussion) in the mitochondria ( Figure 2d). By 2 days after injection, the photoreceptor inner segment AF diminished but photoreceptor outer segment AF remained (Figure 2e,g). Undulating folds of the photoreceptors were apparent, most likely producing some of the autofluorescent spots or flecks seen with in vivo imaging. By 7 days and 1 month, FAC had induced severe loss of photoreceptors with very thin outer nuclear layer and red/green-emitting AF in the RPE layer. Labeling with peanut agglutinin was used to visualize cone photoreceptors. The AF (white arrows) did not co-localize well with the cone photoreceptors (magenta arrows), suggesting FAC induced AF was emitted from rod photoreceptor outer segments (Figure 2h,f).
TUNEL labeling was conducted on cryosections prepared 2 days after FAC or saline injection to evaluate retinal cell death. Many photoreceptor nuclei were TUNEL positive after FAC injection ( Figure 2i).
Since bisretinoids constitute autofluorescent RPE lipofuscin (Sparrow et al., 2012), we assessed bisretinoid levels in whole globes at 7 days. Total A2E and A2-GPE showed no difference compared to saline, A2DHP-PE was significantly reduced, but all-trans retinal dimer (atRALdi) was significantly increased (Figure 2j

| IVT FAC induced iron accumulation in Müller glia and myeloid cells, and the formation of lipid peroxidation products
Immunolabeling for ferritin light chain (L-Ft) and Perls' staining were conducted on cryosections to localize FAC-induced iron accumulation in the retina. L-Ft protein levels can be used as an indicator of iron, since L-Ft levels are increased in response to elevated intracellular iron (Song et al., 2014). While Figure 2 showed FAC-induced green and red emitting AF visible with long fluorescence microscopy exposure times, we used a shorter exposure time for all immunofluorescence imaging to avoid detecting the AF. No primary controls were used to verify that the secondary antibody was not binding non-specifically and to show whether any AF was visible with the short exposure times. Co-labeling of glial fibrillary acidic protein (GFAP) and L-Ft was conducted on cryosections prepared at 2 days after injection. Müller cells overexpress GFAP in a gliotic response to many retinal injuries, pathological conditions, and aging.
By 2 days after saline injection, L-Ft weakly labeled the ganglion cell F I G U R E 2 Ferric ammonium citrate (FAC) induced autofluorescence (AF) in mitochondria, rod outer segments, and atRALdi accumulation. Epifluorescence photomicrographs (unless indicated as confocal) of AF on cryosections obtained after FAC injection (a). Enlarged images for AF (b). Confocal imaging of green-emitting AF and immunolabeling for TOMM20 (mitochondrial marker) were performed on cryosections prepared at 4 h after FAC and saline injection (c). Enlarged images of AF and TOMM20 (d). Green-emitting AF and TOMM20 labeling was performed on cryosections at 2 days after injection (e). Green-emitting AF and peanut agglutinin labeling for cones performed on cryosections at 2 days after injection (f). Enlarged images for green-emitting AF and labeling with TOMM20 (g). Enlarged images for green-emitting AF and labeling with peanut agglutinin (h). TUNEL labeling was conducted on cryosections prepared on 2 days after injection (i). DNase I was used as the positive control for TUNEL labeling. Quantification of bisretinoids by high pressure liquid chromatography from whole eyes at 7 days after FAC and saline injection (j-l). All white arrows indicate AF, magenta triangles indicate immunolabeling for TOMM20, and magenta arrows indicate labeling with peanut agglutinin. White triangles indicate AF in a. White triangles indicate positive TUNEL labeling in i. ONL, outer nuclear layer; IS, inner segment; OS, outer segment; RPE, retinal pigmented epithelium; PNA, peanut agglutinin. Error bars indicate mean ± SEM. *p < 0.05, **p < 0.01, N = 3-5/group for bisretinoid measurements. Representative immunolabeling images are shown from N = 3 mice per group. Scale bar: 50 µm layer, outer plexiform layer, and inner segment layers ( Figure 3a). In contrast, by 2 days after FAC injection, increased L-Ft staining was observed in the ganglion cell layer, outer plexiform layer, and Müller cells co-labeled with GFAP ( Figure  The E06 antibody reacts with oxidized phospholipids. At 5 days, increased immunolabeling for E06 was present mainly in the RPE and Iba1-labeled myeloid cells among the photoreceptor outer segments ( Figure 3g and j, orange arrows). Immunolabeling for malondialdehyde showed no difference between FAC saline-injected eyes at 2 days after injection (data not shown), but was present in CD68+ myeloid cells at 5 days after injection (Figure 3h and k, green arrows).
Taken together, these results suggest IVT iron induced an accumulation of lipid peroxidation products, appearing first in the photoreceptors, then in subretinal myeloid cells and RPE cells, most likely due to phagocytosis of lipid peroxidation products in photoreceptor outer segments.

F I G U R E 3
Ferric ammonium citrate (FAC) induced retinal iron accumulation and lipid peroxidation products. Epifluorescence photomicrographs of co-labeling for GFAP and L-Ft at 2 days after injection (a). Co-labeling for anti-Iba1 and L-Ft at 5 days after injection (b). Perls' Prussian Blue staining at 2 days and 7 days after injection (c). Autofluorescence (AF) and immunolabeling for 8-OHdG at 2 days after injection (d). AF and immunolabeling for carboxyethyl pyrrole (CEP) at 2 days (e) after injection. Co-labeling for anti-Iba1 and CEP at 5 days after injection (f). Co-labeling for anti-Iba1 and E06 at 5 days after injection (g). Co-labeling for anti-CD68 and MDA at 5 days after injection (h). Enlarged image of co-labeling for Iba1 and CEP (white box, f1-f3) (i). Enlarged image of co-labeling for Iba1 and E06 (white box, g1-g3) (j). Enlarged image of co-labeling for CD68 and MDA (white box, h1-h3) (k). White arrows indicate co-labeling for GFAP and L-Ft in a. Magenta arrows indicate immunolabeling for Iba1 and white arrows indicate immunolabeling for L-Ft in b. Red arrows indicate Perls' staining, and black arrows indicate migrated RPEs in c. Yellow arrows indicate immunolabeling for 8-OHdG in d; blue arrows indicate immunolabeling for Iba1 and CEP in e. Blue arrows indicate co-labeling for anti-Iba1 and CEP in f and i. Orange arrows indicate co-labeling for anti-Iba1 and E06 in g and j. White arrows indicate co-labeling for anti-CD68 and MDA in h and k. INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigmented epithelium; CEP, carboxyethyl pyrrole; MDA, malondialdehyde. Representative images are shown from N = 3 mice per group. Scale bar: 50 µm F I G U R E 4 Ferric ammonium citrate (FAC) induced myeloid cell infiltration in FAC-injected eyes and in saline-injected fellow eyes of FACinjected eyes. Epifluorescence photomicrographs showing co-labeling for Iba1 and CD68 on cryosections prepared on 2 days (a) and 7 days (b) after FAC or saline injection. Confocal imaging following immunolabeling for Iba1 was performed on retina flat mounts prepared at 5 days after FAC and saline injection (c). 102° ultra-wide field blue-AF cSLO images acquired at 3 months and 6 months after saline injection (d). Co-labeling for Iba1 and L-Ft on cryosections prepared at 6 months after injection (e). Autofluorescence (AF) on cryosections prepared at 6 months after saline injection (f). Perls' staining at 6 months after injection (g). TUNEL labeling at 6 months after injection (h). DNase I was used as a positive control for TUNEL labeling. White arrows indicate immunolabeling for CD68 in a and b. Magenta arrows indicate immunolabeling for Iba1, blue arrows indicate co-labeling for Iba1 and CD68, and yellow arrows indicate co-labeling for Iba1 and L-Ft. Green arrows indicate AF in f.

| IVT FAC induced expression changes in antioxidant, iron-regulating, and cell-type-specific genes in RPE and neural retina
To assess intracellular iron levels after FAC injection, mRNA levels of transferrin receptor (Tfrc) were measured by qPCR. Tfrc levels are inversely related to intracellular iron levels, so Tfrc measurement by qPCR is a reliable indicator of intracellular iron .
In the neural retina, at both 2 days and 7 days after FAC injection, Tfrc mRNA levels were significantly decreased. Levels were also decreased in the isolated RPE at 7 days but not yet at 2 days. To assess rod and cone stress and differentiation, mRNA levels of the rod-specific gene rhodopsin (Rho), cone opsin1 medium wave sensitive and short wave sensitive (Opn1mw and Opn1sw) were measured. mRNA levels of Rho, Opn1mw, and Opn1sw were significantly decreased after FAC injection compared to saline at both time points ( Figure S1a, d). Retinal pigment epithelium 65 (Rpe65) in RPE cells was quantified to detect RPE stress and differentiation. By 7 days, Rpe65 was significantly decreased after FAC treatment compared to saline control (*p < 0.05, Figure S1f).
To investigate FAC-induced oxidative stress, mRNA levels of antioxidants heme oxygenase 1 (Hmox1), catalase (Cat), and superoxide dismutase 1 (Sod1) were measured. Hmox1, Cat, and Sod1 were significantly increased in FAC-treated neural retina at both 2 days and 7 days, compared to saline ( Figure S1b and e). In RPEs, the mRNA level of Hmox1 was significantly upregulated at both 2 days and 7 days; however, both Sod1 and Cat decreased at 2 days then recovered at 7 days, in FAC compared to saline. (**p < 0.01, Figure S1c and f). To investigate FAC-induced inflammation, relative mRNA levels of interleukin 1 beta (IL-1β), interleukin 6 (IL-6), cluster of differentiation 68 (Cd68), and complement C3 were detected by qPCR. IL-1β, IL-6, and C3 were significantly increased in the neural retina at 2 days and 7 days (**p < 0.01, Figure S1b and e). In RPE cells, consistent with the immunofluorescence results, Cd68 mRNA was significantly upregulated at 7 days, in IVT FAC compared with saline-treated controls.
To investigate whether FAC-induced cell death was associated with changes in ferroptosis-related genes, mRNA levels of Gpx4 and Slc7a11 were detected by qPCR in both neural retina and RPE cells. The mRNA level of Gpx4 in FAC-injected neural retina was significantly upregulated at 2 days compared to saline controls and significantly decreased at 7 days. Slc7a11 was significantly increased in FAC-treated neural retina at both 2 days and 7 days. In the RPE, there was no significant change to either of these genes ( Figure S1a, c, d, and f).

| DISCUSS ION
This study revealed mechanisms of iron-induced retinal toxicity as well as chronic effects of IVT FAC. We found temporal changes in the retinal AF spectrum, accumulation of carboxyethyl pyrrole and atRALdi and slowly progressive degeneration of the RPE, producing a new model of geographic atrophy. Surprisingly, we observed infiltration of iron-laden myeloid cells into fellow eyes, which is in some ways similar to sympathetic ophthalmia, a disease that can affect fellow eyes of patients with ruptured globes.
In a prior study, we found that ferric sulfate was not acutely retina-toxic. Herein, we found that equimolar ferric ammonium citrate (FAC) was toxic. The most likely explanation is that FAC is known to be more soluble than ferric sulfate (National Center for Biotechnology Information, 2021a, 2021b) and is reduced to toxic ferrous iron by the abundant (2 mM) ascorbate in the vitreous, which we confirmed in solution.
IVT FAC caused acute photoreceptor degeneration by 7 days.
By 1 month after injection, RPE degeneration developed reproducibly (N = 10) into a large "kidney bean" shaped area of geographic atrophy specifically in the superior retina and several independent small geographic atrophy lesions throughout the retina. As in human geographic atrophy (Moschos et al., 2014), these atrophic areas expanded and fused over time. Mechanisms of RPE degeneration likely F I G U R E 6 Ferric ammonium citrate (FAC) induces retinal and choroidal neovascularization. Optical coherence tomography images acquired at 1 month (a) and 3 months (b) after FAC injection. Simultaneous fluorescein angiography and indocyanine green angiography with cSLO imaging system at 4 months after FAC and saline injection (c-e). Enlarged images of fluorescein angiography (c, yellow boxes) (d). Enlarged images of fluorescein angiography and indocyanine green angiography (c, red boxes) (e). Fluorescence imaging and immunolabeling for ZO-1 conducted on RPE flat mount prepared at 4 months after injection (f). Enlarged images of autofluorescence (AF) and immunolabeling for ZO-1 (f, white box) (g). Brightfield image of cryosection prepared at 6 months after FAC injection (h). Toluidine blue staining conducted on plastic sections prepared at 6 months after FAC injection (i). Red arrows indicate neovascularization. White arrows indicate infiltrated cells and yellow arrows indicate neovascularization in f. OCT, coherence tomography; FA, fluorescein angiography; ICGA, indocyanine green angiography. Representative in vivo images are shown from N = 10 mice per group. Representative histology images are shown from N = 3 mice per group involve damage from toxic lipid peroxidation products and carbonylcarrying degradation products of bisretinoids. These products may impair RPE phago-lysosomal processing. This may trigger lysosomogenesis, as we found an increase in the lysosomal marker Lamp1 in RPE lysates following IVT FAC (not shown).
IVT FAC also induced green-emitting AF in mitochondria of the photoreceptor inner segments, and broad wavelength-emitting AF in rod photoreceptor outer segments by 4 h after injection.
There was also broad-spectrum AF in myeloid and RPE cells at 7 days, presumably because these cells phagocytosed the AF material within degenerating photoreceptor outer segments. Previous studies of IVT iron reported an increase in broad-spectrum AF in photoreceptors and RPEs (Dunaief, 2006;Shu et al., 2020). Katz et al. (1994) reported that IVT ferrous sulfate induced vitamin A dependent AF in photoreceptor outer segments and RPEs, consistent with the possibility that the AF results from bisretinoid formation. This is aligned with our finding of increased atRALdi at 7 days after IVT FAC.
The green-emitting AF in mitochondria within photoreceptor inner segments may come from flavin adenine dinucleotide (FAD).
Oxidized FAD is fluorescent, while reduced FADH 2 is not (Heikal, 2010). FAD has fluorescence excitation at 450 nm and green emission maxima at 535 nm, which is consistent with the spectra observed by fluorescence microscopy in retinal cryosections of FACtreated eyes. Impaired reduction of FAD to FADH 2 in the citric acid cycle could account for increased FAD fluorescence.
Bisretinoids are photosensitizers that can damage or kill the RPE. Here, we found IVT iron was associated with elevated at-RALdi, a well-characterized RPE lipofuscin bisretinoid. AtRAL di has been reported to be more abundant than A2E in the retinas from Abca4−/−Rdh8−/− mice, a model with features of recessive Stargardt disease (Zhao et al., 2017). Unconjugated atRALdi serves as a photosensitizer for the generation of singlet oxygen leading to oxidative damage; atRALdi is more susceptible to photooxidation than A2E (Kim et al., 2007). Elevated levels of atRALdi can induce apoptosis of cultured RPEs (Zhao et al., 2017).
Neuroinflammation and complement activation have been reported in AMD. Here, we report that IVT iron increased mRNA levels of IL-1β, IL-6, and complement factor C3 in the neural retina.
Further, oxidative stress and Fenton reaction-associated formation of carbonyl-compounds from bisretinoid degradation can induce RPE cells to produce cytokines, including IL-1, IL-6, TNFα, and others, which recruit inflammatory cells including microglia and macrophages (Krizhanovsky et al., 2008). IVT iron induced Iba1+/CD68+ and Iba1−/CD68+ myeloid cell infiltration into the outer retina, consistent with increased mRNA levels of CD68 in both neural retina and RPEs. These myeloid cells most likely consist of both bloodborne macrophages and activated resident microglia. Consistent with this, Moos et al. (2011) reported macrophage infiltration into eyes injected with iron. Herein, there was also increased labeling for CD68 within the RPEs, and RPEs migration into the neural retina after iron injection. This mirrors observations from AMD patients where migrated RPEs were immunoreactive for CD68, suggesting transdifferentiation (Cao et al., 2020). Surprisingly, we found progressive accumulation of autofluorescent cells in saline-injected eyes when the fellow eye was injected with FAC. This pan-retinal autofluorescent cell infiltration was observed by in vivo imaging beginning 2 months after the injections, with the number of infiltrating cells increasing over time.
Immunohistochemistry revealed that these subretinal autofluores-

| Intravitreal injections
Intravitreal injections were performed as previously described . Eyes were injected with 1 μl of 0.5 mM FAC (MP Biomedicals LLC) diluted in 0.9% NaCl (saline) or 1 µl of saline as control.

| In vivo imaging system
Mice were given general anesthesia and placed on a platform. Color fundus photograph and fundus AF were acquired using a Micron III fundus camera (Phoenix Research Laboratories, Inc). The Micron III fundus camera has a filter for color fundus photography between

| Electroretinography
Mice were dark adapted overnight and then anesthetized. Pupils were dilated with 1% tropicamide saline solution (Akorn, Inc.). Two contact lens electrodes made of UV transparent plastic with embedded platinum wires were placed in electrical contact with the corneas. A platinum wire loop placed in the mouth served as the reference and the ground electrode. The electroretinograms were recorded with an Espion E3 system (Diagnosys LLC) with a ganzfeld Color Dome stimulator. The stage was positioned in such a way that the mouse's head was located inside the stimulator, thus ensuring uniform full-field illumination. The flash intensities for recordings of rod a-and b-waves were 500 and 0.01 scot cd m-2 s delivered by the white xenon flash, and green (510-nm maximum) LED, respectively. The cone b-wave was elicited by 500 scot cd second m-2 white xenon flash delivered on a rod-suppressing steady green background of 30 scot cd m-2. All electroretinography was performed at the same time of day.

| Tissue preparation and immunofluorescence
Immunofluorescence was performed on 10 μm cryosections as described previously .

| Bisretinoid analysis by HPLC
Eyes were enucleated and snap frozen on dry ice at 7 days after FAC or saline injection. Frozen eyes were processed as previously described (Sparrow et al., 2010). The extract was redissolved in chloroform/ methanol (2:1), and bisretinoids were measured by HPLC (Alliance system; Waters Corp.) (Kim et al., 2004). Absorbance peaks were identified by comparison to external standards. Molar quantities per eye were calculated from peak areas using standard concentrations determined spectrophotometrically together with published extinction coefficients. Values from each sample were calibrated to the number of eyes in a sample and were expressed as picomoles/eye.

| TUNEL labeling
TUNEL labeling was performed on cryosections prepared as above as described previously using an In Situ Cell Death Detection Kit, Fluorescein (Roche) per the manufacturer's instructions (Shu et al., 2020).

| Perls' Prussian blue and Toluidine blue staining
Four micrometer (4 μm) thick plastic sections were cut in the sagittal plane. Perls' staining was conducted on plastic sections to evaluate retinal iron levels as previously described (Theurl et al., 2016). Plastic sections were stained with toluidine blue to evaluate retinal morphology as previously described (Bhoiwala et al., 2015).

| Neural retina flat mounts and RPE flat mounts
Eyes were enucleated and fixed in 4% PFA for 15 min. Cornea, lens, and retina were removed to make eyecups. Eyecups were incubated with primary antibody at 4°C overnight, rinsed with PBS three times, then incubated with secondary antibody for 1 h at room tempera-

| RNA extraction and Quantitative RT-PCR
Neural retina and purified RPE cells were isolated as previously described . Gene expression changes in the neural retina and purified RPE cells were evaluated (Wolkow et al., 2012). Gapdh was used as an endogenous control. Taqman Probes (ABI) were used as follows: Cat (Mm00437992), Cd68

| Total iron and ferrous iron measurements
Stock solutions of 25 mM (di) ferric sulfate (Fe₂(SO 4 ) 3 ) and 50 mM ferric ammonium citrate (FAC) (C 6 H 8 FeNO 7 ) were prepared in normal saline. Both stock solutions were vortexed for 30 s before use. 20 mM sodium L-ascorbate normal saline solution was prepared. The 20 mM sodium L-ascorbate solution was diluted 1:10 in PBS in two vials into which 10 μl of FAC or ferric sulfate stock solutions were added. The final volume of either sample solution was 10 ml and used in the iron assay after a 5 min incubation at 37°C. The level of total (ferrous and ferric) and ferrous iron was measured using an iron assay kit (MAK025; Sigma) according to the manufacturer's protocol.

| Statistical analysis
Mean ± SEM was calculated for each group. Student's, two-tailed t test was used for the statistical analysis. All statistical analyses were performed using GraphPad Prism 6.0.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

AUTH O R CO NTR I B UTI O N S
YL, BAB, YS, HJK, MG, KZ, and AR performed the experiments. YL and JLD analyzed data and drafted the manuscript. MP contributed essential reagents. All authors provided critical review of the manuscript. GS, JRS, and JLD provided funding for the study.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.