Retina Ferritin But Not Iron Increases in Retina Upon Systemic Iron Overload in Diabetic and Iron-Dextran Injected Mice Aina Bonet,1,2 Judit Pampalona,1 Eduard Jose-Cunilleras,3 Víctor Nacher,2 and Jesús Ruberte1,2 1 Centre for Animal Biotechnology and Gene Therapy (CBATEG), Universitat Autònoma de Barcelona, Bellaterra, Spain 2 Department of Animal Health and Anatomy, Faculty of Veterinary Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain 3 Department of Animal Medicine and Surgery, Faculty of Veterinary Medicine, Universitat Autònoma de Barcelona, Bellaterra, Spain Correspondence: Jesús Ruberte, PURPOSE. Iron overload causes oxidative damage in the retina, and it has been involved CBATEG – Center for Animal in the pathogeny of diabetic retinopathy, which is one of the leading causes of blindness Biotechnology and Gene Therapy, in the adult population worldwide. However, how systemic iron enters the retina during Autonomous University of diabetes and the role of blood retinal barrier (BRB) in this process remains unclear. Barcelona, C/de la Vall Morona, Bellaterra (Cerdanyola del Vallès) METHODS. The db/db mouse, a well-known model of type 2 diabetes, and a model of 08193, Spain; systemic iron overload induced by iron dextran intraperitoneal injection, were used. Perls jesus.ruberte@uab.cat. staining and mass spectrophotometry were used to study iron content. Western blot and AB and JP contributed equally. immunohistochemistry of iron handling proteins were performed to study systemic and retinal iron metabolism. BRB function was assessed by analyzing vascular leakage in Received: March 8, 2022 fundus angiographies, whole retinas, and retinal sections and by studying the status of Accepted: February 13, 2023 Published: March 13, 2023 tight junctions using transmission electron microscopy and Western blot analysis. Citation: Bonet A, Pampalona J, RESULTS. Twenty-week-old db/db mice with systemic iron overload presented ferritin Jose-Cunilleras E, Nacher V, Ruberte overexpression without iron increase in the retina and did not show any sign of BRB J. Ferritin but not iron increases in breakdown. These findings were also observed in iron dextran-injected mice. In those retina upon systemic iron overload animals, after BRB breakdown induced by cryopexy, iron entered massively in the retina. in diabetic and iron-dextran injected CONCLUSIONS. Our results suggested that BRB protects the retina from excessive iron entry mice. Invest Ophthalmol Vis Sci. 2023;64(3):22. in early stages of diabetic retinopathy. Furthermore, ferritin overexpression before iron https://doi.org/10.1167/iovs.64.3.22 increase may prepare the retina for a potential BRB breakdown and iron entry from the systemic circulation. Keywords: iron overload, ferritin, diabetic retinopathy, BRB breakdown, eye cryopexy I ron is one of the most abundant elements in the earth’s crust and is essential for development and survival of organisms. As a metal, iron has the capacity to donate and group has showed that this iron import alternative mecha- nism also works in the retina.9,10 In states of iron overload, the iron regulatory proteins accept electrons in redox reactions that are required in (IRPs) initiate a series of mechanisms to protect the cells, vital enzymatic processes such as oxygen transport, energy increasing the expression of ferritin to store more iron.11–14 production, DNA replication, neurotransmitter signaling, However, when iron levels exceed the limit, free ferrous ion myelin shedding, and steroid hormone synthesis.1 In the triggers the production of oxygen free radicals via the Fenton retina, iron is a co-factor for RPE65 synthesis, involved in the reaction, which are toxic and can cause DNA alterations, phototransduction cascade.2,3 Furthermore, iron-containing protein damage, and lipid peroxidation.17 This can be espe- enzymes are responsible of the shedding and renewal of the cially harmful in the retina, which presents a high oxygen photoreceptor outer segment disc membranes.4 tension and metabolic rate.15 After iron absorption in the duodenum, iron is released Diabetes has been widely associated with systemic iron to the bloodstream and bound to the iron carrier transfer- overload.16,17 Recent studies have showed a positive corre- rin for tissue delivery. Once in the cell membrane, transfer- lation between iron in serum and the risk of develop- rin binds to its transferrin receptor, and the entire complex ing diabetes.18,19 Supporting this association, several mouse enters the cell by endocytosis.5 Then, iron is used for cellu- models of iron overload have been described to develop reti- lar functions, while excess of iron is stored in ferritin. When nal lesions resembling diabetic retinopathy.9,20–22 Further- needed, iron exits the cells through the iron exporter ferro- more, iron overload in diabetic mouse exacerbates the portin.6 In addition, it has been described that ferritin can progression of diabetic retinopathy.23 Nevertheles, wild-type act as an iron importer by binding its L and H subunit to mice in which iron overload was experimentally induced the receptors SCARA5 and TIM2, respectively.7,8 In fact, our by iron dextran injection exhibited that systemic iron had Copyright 2023 The Authors iovs.arvojournals.org | ISSN: 1552-5783 1 This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 2 minimal penetration in the retina.24 Thus the mechanism by iron concentration to the unsaturated iron binding capacity which systemic iron overload induces retinal iron overload values. remains unclear. The aim of this study is to analyze the relation between systemic and retinal iron during diabetes and the implica- Perls Prussian Blue Iron Staining tion of the BRB, as it has been suggested to have a role in Perls stain was performed in 3 μm paraffin sections of the regulation of iron transport into the retina.24 Clarifying eyeballs, that were sectioned along their axis, and spleens, the mechanism of iron entry in the retina could be crucial that were cut transversally. Samples were deparaffinated, for the prevention and treatment of diabetic retinopathy, rehydrated, and immersed in a solution of 5% potassium since it is mainly exacerbated by oxidative damage. For this ferrocyanide (Sigma-Aldrich, St Louis, MO, USA) and 5% purpose, two mouse models have been used: (i) the db/db hydrochloric acid (Panreac Química SLU, Castellar del Vallés, mouse, a widely accepted model for type 2 diabetes25–27 Spain). Nuclear fast red (Sigma-Aldrich) was applied for and (ii) iron dextran-injected mice, a model of systemic iron nuclear counterstaining. Sections were washed, dehydrated, overload.28,29 and mounted in DPX medium (Sigma-Aldrich) before anal- ysis. To assess iron content in the spleen, Prussian Blue precipitate was quantified with ImageJ software (National MATERIAL AND METHODS Institutes of Health, Bethesda, MD, USA). Mice Male diabetic mice db/db (BKS.Cg-Dock7m +/+ Leprdb /J), Detection of Total Retinal Iron by Inductively nondiabetic littermates db/+ (BKS.Cg-Dock7m + Leprdb /+) Coupled Plasma Mass Spectrometry (ICP-MS) and C57BL/6J wild-type (WT) mice of 20 weeks old were used in this work. The animal facility had a controlled envi- Eyes were enucleated and retinas were dissected in 1x phos- ronment (20°C temperature, 60% humidity, and 12 hours phate buffered saline (PBS) and collected in vials. A blank light/dark cycles) and mice were fed ad libitum with a with 1x PBS was also included to normalize the values. For standard diet (2018S TEKLAD Global; Harlan Teklad, Madi- retinal digestion, samples were treated with 65% nitric acid son, WI, USA). Mice were euthanized with an overdose of and heated at 120°C for 30 minutes. Retinal iron concen- inhaled isoflurane or anesthetized with an injected solu- tration was determined by inductively coupled plasma mass tion of ketamine (100 mg/kg) and xylazine (12 mg/kg) spectrometer Agilent 7500ce (Agilent, Santa Clara, CA, USA). followed by cervical dislocation. Experiment procedures were performed following the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the Injection of Iron Dextran and BRB Breakdown by ARRIVE (Animal Research: Reporting In Vivo Experiments) Cryopexy guidelines 2.0. All the procedures were approved by the To induce acute systemic iron overload, 1g/kg of iron Ethics Committee in Animal and Human Experimentation of dextran (Sigma-Aldrich) was injected intraperitoneally in WT the Universitat Autònoma de Barcelona (UAB) (FUE-2018- mice. Saline solution was injected instead of iron dextran in 00717286). a control group. After 4 hours, animals were anesthetized. BRB breakdown by cryopexy was induced in half of the mice injected with iron dextran, by applying a piece of dry Glycaemia Measurement ice in the scleral conjunctiva at the lateral palpebral commis- The db/db mice start to show hyperglycemia at four to eight sure for 10–20 seconds.31 Eye and spleen samples were weeks of age.30 Glycemia was measured in nonfasted condi- collected for paraffin embedding and eyes for ICP-MS anal- tions with a glucometer (Bayer, Leverkusen, Germany) in ysis. For ICP-MS analysis, mice were subjected to a vascu- blood collected from the tail vein. Mice were considered lar washout to prevent measuring the iron inside the retinal diabetic when glycemia values were greater than 250 mg/dL blood vessels, which could modify the results of iron content in, at least, in two consecutive measurements. According in the retinal parenchyma homogenates. For this purpose, to this, 20-week-old mice presented a significant statistical thorax was opened to expose the heart, 2 ml 1x PBS were difference in body weight (30.21 ± 0.665 g, n = 16, in injected in the left ventricle, and the caudal vena cava was db/+ vs. 52.69 ± 1.117 g, n = 13, in db/db; P < 0.0001) cut to interrupt venous return. and glycemia (121.2 ± 3.204 mg/dL, n = 16, in db/+ vs. 519.1 ± 21.12 mg/dL, n = 13, in db/db; P < 0.0001), confirming that db/db mice used in this study had type 2 Assessment of BRB Breakdown by Evans Blue diabetes. and Bis-Benzimide Evans Blue and bis-benzamide (Hoechst) were used to vali- date BRB breakdown after cryopexy performed on the right Analysis of Iron Parameters in Systemic Blood eye of Ketamine-Xylazine anesthetized mice (100 mg/kg and Mice were anesthetized using inhaled isoflurane, and blood 12 mg/kg, respectively). The left eye was used as a control. samples were collected by intracardiac puncture, placed in Thirty minutes after cryopexy, the thorax of the anesthetized a lithium heparin tube, and spun in a centrifuge to obtain mice was opened to expose the heart and 0.5 ml of a solu- plasma for biochemistry analysis. Serum iron concentration tion containing Evans blue 8% and bis-benzamide (1 mg/ml) and unsaturated iron binding capacity were determined with was injected directly into the left ventricle. Then, the eye was an automated biochemistry analyzer (Chemistry Analyzer enucleated, and the retina dissected in 1xPBS for vessel visu- AU400; Olympus, Tokyo, Japan) and enzymatic colorimet- alization and image analysis under the microscope (Nikon ric assays (Beckman Coulter, Inc, Southfield, MI, USA). Total Eclipse, E800), using the ACT-1 software. Heparin (10 u/gr) iron binding capacity (TIBC) was calculated adding serum was injected prior injection to prevent blood coagulation. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 3 Western Blot camide (Alcon, Barcelona, Spain) were used to induce mydri- asis, and 10 μl of 5% sodium fluorescein (Sigma-Aldrich) Retina and spleen samples were homogenized in RIPA buffer were injected subcutaneously in each mouse to observe with a protease inhibitor (Roche, Basel, Switzerland). For the eye fundus vascularization. Custom-made mouse lenses sample preparation, 50 mg of protein were resuspended were placed on the corneas while visualizing retinal blood in 2x Laemmli loading buffer (pH 6.7) (Sigma-Aldrich) and vessels in the HRA2 scanning laser ophthalmoscope (Heidel- heated at 95°C for 5 minutes. Electrophoresis was performed berg Engineering, Heidelberg, Germany). Angiography was by loading 12 μL of sample in a 12% pre-cast SDS-PAGE gel performed in both eyes and in the three retinal vascular (Bio-Rad Laboratories, Hercules, CA, USA), which was later plexi. transferred to a polyvinylidene difluoride (PVDF) membrane (Merck Millipore, Billerica, MA, USA). After 1 hour blocking with either 5% non-fat dried milk or 2% bovine serum albu- Transmission Electron Microscopy (TEM) min (Sigma-Aldrich) in Tris buffered saline, 0.05% Tween Retinas were dissected in 1x PBS and cut into 1 mm2 frag- 20, membranes were incubated overnight at 4°C with the ments. Retinal fragments were fixed in 2,5% glutaraldehyde following primary antibodies: rabbit anti-mouse L-ferritin and 2% paraformaldehyde for 2 hours at 4°C and then (1:1000; ab69090, Abcam, Cambridge, UK); rabbit anti- washed in PBI. Post fixation was performed with 1% osmium mouse H-ferritin (1:1000; ab65080, Abcam); rabbit anti- tetroxide for 2 hours. Samples were dehydrated in a graded mouse SCARA5 (1:1000; ab76720, Abcam); rabbit anti- acetone solution and infiltrated with Spurr resin (Sigma- mouse transferrin (1:2000; PA3-913, Thermo Fisher Scien- Aldrich). Ultra-sections of 60–80 nm were obtained with tific, Wilmington, DE, USA); rat anti-mouse transferrin recep- ultramicrotome Leica EM UC6 and contrasted with Reynolds’ tor (1:1000; ab63333, Abcam); rabbit anti-mouse claudin- stain and 2% aqueous uranyl acetate. Finally, samples were 5 (1:1000; 34-1600, Thermo Fisher Scientific); and rabbit analyzed with a transmission electron microscope Jeol JEM- anti-mouse VEGF (1:100; ab46154, Abcam). For protein 1400 ( Jeol Ltd., Tokyo, Japan). normalization, rabbit anti-mouse α-tubulin (1:80000; ab4074, Abcam) or rabbit anti-human α smooth muscle actin (1:10000; ab5694, Abcam) were used as a loading control. Retinal Thickness Measurement Secondary antibodies goat anti-rabbit (Bionova Scientific, To evaluate retinal thickening, an indicator of retinal Fremont, CA, USA) and sheep anti-rat (LifeSpan, Provi- oedema,32 paraffined eyes were sectioned through their axis dence, RI, USA) horseradish peroxidase at 1:5000, were used. and were stained with hematoxylin and eosin. Retinal thick- Immobilon Crescendo Western HRP Substrate (Merck Milli- ness was measured between the internal limiting membrane pore) and the ImageJ software were used for band detection and the outer limiting membrane. A total of six areas, at and quantification, respectively. 100 μm, 500 μm, and 1000 μm on each side of the optic nerve head, were analyzed. ImageJ was used for morpho- Immunohistochemistry metric analysis by an operator blinded to the experimental groups. Paraffin embedded mice eyes and spleens were sectioned, deparaffinized, and rehydrated. Antigen retrieval was performed in a solution with 3.38% citric acid and 24.4% Statistical Analysis sodium citrate. Sections were washed in PBI (1x PBS, For statistical analyses R Software (version 4.1.0; R Foun- 0.05% Igepal). The following primary antibodies were dation for Statistical Computing, Vienna, Austria) was used. used: rabbit anti-mouse L-ferritin (1:100; ab69090 Abcam); Shapiro-Wilk test was performed for evaluation of the rabbit anti-mouse H-ferritin (1:100; ab65080 and ab183781 data distribution and F test for the variance homogene- Abcam); goat anti-mouse collagen IV (1:20; AB769, Merck ity. Data with normal distribution was analyzed using the Millipore); and rabbit anti-mouse albumin (1:2000; A001, parametric method unpaired t-test and data not normally DAKO, Glostrup, Denmark). All primary antibodies were distributed was analyzed with Wilcoxon test. Data are diluted in 1x PBS, 0.5% bovine serum albumin, 0.05% expressed as mean ± standard error of the mean (SEM). Igepal with 10% of donkey serum (Sigma Aldrich), except Differences between groups were considered significant at anti-albumin antibody that was diluted in PBI. Negative P-value<0.05. controls were included by omitting the primary antibody in sequential tissue sections. Secondary antibodies donkey anti-goat Alexa 488 (1:1000; Life Technology, Carlsbad, RESULTS Ca, USA) and goat anti-rabbit 568 (1:1000; Invitrogen, Systemic Iron Was Increased in 20-Week-Old Carlsbad, CA, USA) were incubated for 2 hours. Samples were counterstained with Hoechst (Sigma-Aldrich), mounted Diabetic Mice in Fluoromount (Sigma-Aldrich) and analyzed in a SP5 To assess systemic iron status in diabetic mice we first laser scanning confocal microscope (Leica Microsystems analyzed the iron content in the spleen using Perls stain. GmbH). The results revealed that db/db mice presented a significant increase of iron content compared to db/+ (0.452 ± 0.136 In Vivo Observation of Eye Fundus arbitrary units (au) in db/+ mice vs. 1.489 ± 0.387 au in Vascularization db/db mice; p = 0.0142; n = 9) (Fig. 1A). As expected, iron was mostly accumulated in the red pulp (Fig. 1A), where Mice were anaesthetized with an intraperitoneal injection erythrophagocytosis takes place. Another widely used indi- of a ketamine and xylazine cocktail. During anesthesia, cator to assess iron content in tissues is ferritin concen- Viscofresh (Allergan, Dublin, Ireland) was applied in the tration.33–35 To assess if the increment of iron in db/db eyes to prevent them from drying. Eye drops of tropi- spleen correlated with a higher L-ferritin expression, western Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 4 FIGURE 1. Spleen and serum iron, TIBC and spleen ferritin were increased in 20-week-old db/db mice. (A) Perls stain showed a significant increment in iron content in the spleen of db/db mice, mostly in the red pulp. (B) Western blot analysis also revealed higher levels of L-ferritin expression in spleen of db/db mice. α-smooth muscle actin was used as a loading control. (C) Immunofluorescence representative images showed increased L-ferritin expression in the red pulp of db/db mouse spleen. Nuclei were counterstained with Hoechst. (D) Serum iron and TIBC, which is an indicator of serum transferrin concentration, were significatively increased in db/db mice. RP: red pulp; WP: white pulp. *P < 0.01; ***P < 0.001; ****P < 0.0001, respectively. Scale bar: 100 μm (inset A), 30 μm (C). Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 5 blot was performed and as expected, db/db mice exhibited (Fig. 3A). Altogether, these results showed that, despite higher amounts of L- ferritin (0.372 ± 0.133 au in db/+ mice the systemic iron overload and the increase of ferritin, the vs. 2.123 ± 0.364 au in db/db mice; p = 0.0107; n = 3) iron levels and the iron import machinery were not incre- (Fig. 1B). Immunohistochemistry showed that the increase mented in diabetic retinas, thus suggesting the existence of L-ferritin expression occurred also in the red pulp, where of a mechanism that protects the retina from massive iron iron is accumulated (Fig. 1C). entry. The increment in iron and ferritin content found in the db/db mouse spleen showed concordance with the high levels of serum iron (20.17 ± 1.072 μmol/L in db/+, BRB Breakdown Signs Were Not Observed in n = 15 vs. 28.92 ± 2.323 μmol/L in db/db, n = 9; p = 0.0008) 20-Week-Old Diabetic Mice (Fig. 1D). This finding was accompanied by an increase The db/db mice develop complete BRB breakdown at 15 of serum TIBC in db/db mice (48.90 ± 0.932 μmol/L in months of age.37 To assess BRB integrity in 20-week-old db/+, n = 15 vs. 65.28 ± 3.006 μmol/L in db/db, n = 9; diabetic mice, we first analyzed eye fundus with Scan- P < 0.0001; Fig. 1D), which is an indicator of serum trans- ning Laser Ophthalmoscopy (SLO), which revealed that ferrin concentration and iron delivery into the tissues.36 injected fluorescein remained inside the vessels and none of db/+ nor db/db mice presented vascular leakage (n = 16) Retinas Showed Increased Ferritin, But Not Iron (Fig. 3B). In addition, serum albumin detected by immuno- Overload in 20-Week-Old Diabetic Mice histochemistry was confined to the vessel lumen, indicat- ing the absence of retinal edema in 20-week-old diabetic To assess if, as happened in the spleen, retinas from diabetic mice (Fig. 3C). VEGF, which increases vascular permeability mice presented an elevated expression of ferritin, levels of in diabetic retinopathy38 , was also analyzed by Western blot both L- and H-ferritin were analyzed by Western blot analysis analysis, and no statistical differences were found between in db/+ and db/db retinas. An increase of twofold change db/+ and db/db retinas (1.003 ± 0.142 au in db/+ vs. 0.8553 in L-ferritin (0.484 ± 0.488 au in db/+ vs. 0.9947 ± 0.146 ± 0.183 au in db/db; n = 3; P = 0.559) (Fig. 3D). Further- au in db/db; n = 3; P = 0.029) and a 3.3-fold change in H- more, the comparison of retinal thickness between db/+ and ferritin (0.3168 ± 0.051 au in db/+ vs. 0.986 ± 0.052 au in db/db mice confirmed that diabetic retinas did not present db/db; n = 3; P = 0.0008) were observed in diabetic retinas edema (183.38 ± 2.102 μm in db/+, n = 12 vs. 187.12 ± (Fig. 2A). Immunohistochemistry confirmed an increase in 4.696 μm in db/db; n = 6; P = 0.41) (Fig. 4A). the expression of both ferritin subunits in diabetic retinas Next, to evaluate tight junctions, the key player in the (Fig. 2B). L-ferritin immunohistochemistry was quantified in maintenance of BRB integrity39 , ultrathin sections of the terms of pixel intensity showing a significant increase in the retina were analyzed by TEM. Endothelial tight junctions db/db mouse retinas (0.08152 ± 0.0082 db/+, n = 8, vs. in db/db retinas appeared morphologically normal, with- 0.1145 ± 0.0094 in db/db, n = 6; P = 0.02; Fig. 2C). These out any open space between membranes of neighboring results agreed with those obtained by a recent study.23 endothelial cells and with the characteristic electron density Because Perl’s stain was not sensitive enough to detect produced by the tight junction proteins (Fig. 4B). Addition- iron in db/+ and db/db retinas (Supplementary Fig), ICP- ally, one of the most important tight junction proteins in MS was used to ensure that retinal ferritin levels corre- the BRB, claudin-540 was analyzed by Western blot analysis. lated with iron levels. Unexpectedly, ICP-MS analysis did No differences in claudin-5 expression were observed when not show any difference in the iron content between db/+ comparing db/+ with db/db retinas (0.989 ± 0.134 AU in and db/db retinas (2.264 ± 0.212 μg iron/g retina in db/+ vs. 1.058 ± 0.243 AU in db/db; n = 3; P = 0.818) db/+ vs. 2.162 ± 0.071 μg iron/g retina in db/db; (Fig. 4C). Altogether, these results showed that BRB was n = 5; P = 0.744) (Fig. 2D), revealing that increased retinal intact in 20-week-old diabetic mice and that could be ferritin was not accompanied with increased iron content in preventing the entrance of high amounts of circulating iron the diabetic retina. into the retina. Iron Import Proteins Were Not Increased in BRB Protected Retina From Iron Entrance During 20-Week-Old Diabetic Mouse Retinas Systemic Iron Overload Because ferritin expression, but not iron content, was To assess the BRB role in the protection of the increased in the retina, we next analyzed the proteins retina during iron overload, we experimentally induced involved in iron import within the retina. In two independent systemic iron overload by injection iron dextran intraperi- experiments, Western blot analysis showed that transferrin toneally to C57BL/6J WT mice. At four hours after (1.013 ± 0.071 au in db/+ vs. 1.079 ± 0.12 au in db/db; n injection, iron accumulation in the spleen was quanti- = 3; P = 0.665, in Figure 3A and 1.393 ± 0.156 au in db/+, fied with Perls staining, which demonstrated that iron n = 3, vs. 1.399 ± 0.202 au in db/db; n = 4; P = 0.985, in dextran-injected mice presented systemic iron overload Supplementary Fig. SB) was not increased in diabetic retinas. (0.8051 ± 0.46 au in the group injected with saline solution Similarly, transferrin receptor was also not increased (1.079 vs. 2.281 ± 0.13 au in the group injected with iron ± 0.237 au in db/+ vs. 0.773 ± 0.058 au in db/db; n = 3; dextran; n = 3; P = 0.037) (Fig. 5A). In contrast, P = 0.278, in Figure 3A and 0.217 ± 0.004 au in db/+ vs. Perls staining performed in the retina of dextran- 0.232 ± 0.013 in db/db; n = 4; P = 0.313, in Supplementary injected mice only detected iron inside the blood Fig. SB). In addition, the receptor of circulating L-ferritin, vessels and not in the parenchyma (Fig. 5B). ICP-MS SCARA5, known to mediate iron import into the retina10 confirmed that no statistical differences were found in was also not increased in diabetic retinas (0.9799 ± 0.076 iron content in the retinal parenchyma between groups au in db/+ vs. 1.099 ± 0.087 au in db/db; n = 3; P = 0.363) (1.768 ± 0.521 μg iron/g retina in saline solution–injected Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 6 FIGURE 2. L- and H-ferritin, but not iron, were increased in 20-week-old db/db mouse retinas. (A) Western blot analysis revealed that L-ferritin expression was twofold increased and H-ferritin was 3.3-fold increased in db/db mouse retinas. As a loading control, α-tubulin was used. (B) Immunofluorescence representative images showed an increase in both L- and H-ferritin expression in db/db mouse retinas. (C) L-ferritin immunohistochemistry was quantified in terms of pixel intensity showing higher levels of ferritin in the db/db mouse retinas. Nuclei were counterstained with Hoechst. (D) There were no differences between db/+ and db/db retinas in total iron content analyzed with ICP-MS. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; IS, inner segments of photoreceptors. *P < 0.05; ***P < 0.001. Scale bar: 32.05 μm. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 7 FIGURE 3. Iron import proteins were not increased, and no signs of BRB breakdown were found in 20-week-old db/db mouse retinas. (A) Western blot analysis revealed that neither transferrin nor its receptor was increased in db/db mouse retinas. SCARA5, the receptor of L-ferritin, was also not increased in db/db mouse retinas. As a loading control, α-tubulin was used. (B) Analysis of the eye fundus by Scanning Laser Ophthalmoscopy (SLO) showed that injected fluorescein remained inside the retinal blood vessels in db/db mice showing no vascular leakage (n = 16). (C) To assess retinal vessels leakage microscopically, immunohistochemistry against serum albumin and collagen IV, a marker for blood basement membrane, was performed. The results showed that serum albumin was restricted inside the vessel lumen in db/db mice. (D) Western blot analysis of VEGF expression, which induces vascular permeability, showed no statistical differences between db/+ and db/db retinas. As a loading control, α-tubulin was used. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer. Scale bar: 29.41 μm (inset: 11.49 μm). Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 8 FIGURE 4. No alterations in thickness and ultrastructure were found in 20-week-old db/db mouse retinas. (A) To assess retinal edema, paraffined sections stained with hematoxylin and eosin were measured from the internal limiting membrane to the outer limiting membrane (arrowheads) in six different areas (at 100 μm, 500 μm, and 1000 μm on each side of the optic nerve head), and no differences between db/+ and db/db were found. (B) TEM analysis revealed a normal ultrastructure and electron density of tight junctions (arrows) in db/db retinal vessels. (C) Western blot analysis did not show changes in the expression of the tight junction protein claudin-5 in db/db mice retinas. As a loading control, α-tubulin was used. OD, optic disc; ON, optic nerve; GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; ILM, internal limiting membrane; OLM, outer nuclear membrane. Scale bar: 41.15 μm. EC, endothelial cell; L, vessel lumen; TJ, tight junction. Scale bar: 41.15 μm in A, 200 nm in B (inset: 217 nm). Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 9 FIGURE 5. Intraperitoneal injection of iron dextran induced acute systemic iron overload but did not cause retinal iron overload in WT mice. (A) At four hours after injection, Perls staining was performed in the mouse spleen showing an increase in iron accumulation in the red pulp of mice injected with iron dextran. (B) However, in the retina, Perls stain revealed iron exclusively inside the blood vessels but not in the retinal parenchyma. (C) After a vascular washout to remove intravascular iron dextran, ICP-MS was used to quantify iron. (D) Only after BRB breakdown was induced by cryopexy, immunohistochemistry against serum albumin and collagen IV showed vascular leakage (arrowhead), demonstrating BRB breakdown in retinas from cryopexed eyes. (E) injection of Evans blue together with bis-benzamide revealed leakage outside the blood vessel in cryopexed eyes, in blue (arrowheads in the bottom left image) and red (arrowheads in bottom right image), respectively. Bis-benzamide stained cell nuclei. RP, red pulp; WP, white pulp; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; ONL, outer nuclear layer. *P < 0.05. Scale bar: 82 μm in A, 36.49 μm in B (inset: 10 μm), 13.7 μm in D, and 35.09 in E. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 10 FIGURE 6. BRB breakdown induced by cryopexy caused retinal iron overload in mice injected with iron dextran. (A) Perls staining in paraffin sections showed iron leakage (arrowheads) into the retinal parenchyma in cryopexed eyes. (B) ICP-MS was used to quantify iron content in the retinal parenchyma, showing, as expected, an increase in retinal iron in cryopexed eyes. GCL, ganglion cell layer. **P < 0.01. Scale bar: 22.72 μm. mice, n = 5, vs. 2.755 ± 1.4 μg iron/g retina in the iron thy, retinal ferritin and transferrin receptor expression four dextran-injected mice; n = 4; P = 0.905) (Fig. 5C). Altogether, hours after iron dextran injection in WT mice was assessed these results supported the hypothesis that the BRB has a by Western blot analysis. The results showed a 1.8-fold protective effect in preventing retinal iron overload. increased L-ferritin expression in the retinas of iron dextran- Finally, to confirm our hypothesis, BRB was experimen- injected mice (0.595 ± 0.136 au in the saline solution– tally broken by cryopexy, a well-established method to injected mice vs. 1.040 ± 0.071 au in the iron dextran- produce BRB breakdown.31 Cryopexy was performed in the injected mice; n = 4; P = 0.0272) and no change for the right eye of iron dextran-injected mice, whereas the left transferrin receptor levels (1.151 ± 0.074 au in the saline eye served as a control. BRB breakdown was confirmed solution–injected mice vs. 1.384 ± 0.097 au in the iron by immunohistochemistry against serum albumin (Fig. 5D) dextran-injected mice; n = 4; P = 0.105) (Fig. 7A). Further- and Evans Blue/bis-benzamide vascular leakage (Fig. 5E). more, to confirm that ferritin was increased in the whole Furthermore, Prussian blue staining showed iron leaking retinal parenchyma and not only in the endothelial cells from the retinal vessels (Fig. 6A). One hour after cryopexy, in contact with the intravascular iron, an immunohisto- ICP-MS was used to quantify total iron content in the retinal chemistry against L-ferritin and collagen IV was performed parenchyma. The results showed that iron was significantly (Fig. 7B). The results obtained showed that L-ferritin expres- increased in cryopexed retinas (2.755 ± 1.4 μg iron/g retina sion was not accumulated in the endothelial cells, therefore in control eyes, n = 4, vs. 12.052 ± 1.764 μg iron/g retina confirming that systemic iron overload causes an increase of with cryopexy; n = 6; P = 0.005) (Fig. 6B), confirming that L-ferritin expression in the retinal parenchyma without BRB massive iron entry only occurs when BRB is compromised. breakdown and retinal parenchyma iron increase. DISCUSSION Ferritin Expression in the Retina Was Increased During Systemic Iron Overload Without Retinal Previous reports have described the association between Iron Overload iron overload and diabetic retinopathy.16,41 Our results showed that 20-week-old db/db mice had systemic iron over- To understand why ferritin expression was increased while load and increased ferritin expression in the retina, which iron content and other iron handling proteins remained at fully agree with the results observed by Chaudhary et al. in normal levels in diabetic retinas, and to determine whether 2018.23 However, unexpectedly, analysis of retinal iron by this was a specific finding that occurred in diabetic retinopa- ICP-MS did not show higher iron content in the retina of Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 11 FIGURE 7. L-ferritin expression increased in retinas but not in endothelial cells of iron dextran-injected mice. (A) Western blot analysis revealed a significant 1.8-fold increase in L-ferritin expression in the retinas of mouse injected with iron dextran compared to the saline solution–injected controls, whereas transferrin receptor levels remained unchanged in these mice. α-tubulin was used as a loading control. *P < 0.05. (B) Immunohistochemistry against L-ferritin and collagen IV showed that L-ferritin expression was similar in saline solution injected mice and iron dextran-injected mice, indicating that L-ferritin was not being accumulated in vascular endothelial cells of iron dextran-injected mice. Scale bar: 7.35 μm. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 12 diabetic mice. In addition, no changes were found in retinal unaltered. Ferritin levels are post-transcriptionally regulated iron import machinery of db/db mice, which supports the by the so-called IRPs, which bind to ferritin mRNA and fact that retinal iron content was not increased. prevent its expression. When intracellular iron is high, IRPs Classical methods for iron detection, such as Prussian lose the capacity to bind to ferritin mRNA, and thus ferritin blue staining, are not sensitive enough to detect iron in the levels increase.54 As for the retina, this mechanism has been retina. There are high precision techniques to locate and described using a model of IRP deficiency in which ferritin quantify iron in the retina, such as particle-induced X-ray expression in the retina was increased.55 In contrast, on iron emission and synchrotron X-ray fluorescence (SXRF),42,43 overload, transferrin receptor mRNA is no longer protected although these have a very limited access. ICP-MS is also by IRPs and becomes unstable to avoid more iron entrance. a high sensitivity method for iron detection, and it is more Surprisingly, retinas from db/db and iron dextran-injected available, but it does not provide information about the iron mice, which had increased ferritin expression and no change location within the retinal layers. Thus in situ detection of in the transferrin receptor levels, did not present a higher retinal iron can be challenging. For this reason, and because iron content that could trigger IRP depletion and ferritin ferritin expression changes proportionally to iron levels in increase. This suggests the existence of a signaling mech- systemic tissues,33 most studies use ferritin expression as an anism between the retina and the rest of the body. The indirect indicator for retinal iron content,23,24,44,45 justifying retina could be detecting systemic iron overload and activat- that an increase of ferritin expression in the retina is equiv- ing ferritin synthesis as a compensatory response in front alent to retinal iron overload. Nevertheless, our results point a potential BRB breakdown, given that a single ferritin out that ferritin expression might not always be a reliable complex can harbor up to 4500 molecules of iron and act marker to estimate iron content in the retina as it has been as a natural iron chelator.56 used so far. Apart from sequestering deleterious free iron, ferritin is We have shown that in diabetic mice with systemic iron also overexpressed in chronic inflammatory diseases such as overload, the retina remains protected from iron entry, at type 2 diabetes mellitus.57 Ferritin induces anti-inflammatory least until the age of 20 weeks, but how is the retina cytokines and works as an immunomodulator, although this protected? In a recent study of long-term iron overload applies specially to H-ferritin.58 Nevertheless, the fact that induced by iron dextran injections in mice,24 BRB was increased ferritin was present in both diabetic and iron proposed to prevent the retina from iron entry, so we next dextran-injected mice retinas and that both models had analyzed BRB status in db/db mice. BRB breakdown is a systemic iron overload suggests that ferritin expression in hallmark of diabetic retinopathy, and its breakage occurs the retina might be regulated by systemic iron excess. as a consequence of different stresses, like hypoxia, oxida- All together our results demonstrated for the first time tive damage and/or inflammation. BRB disruption has been that ferritin expression cannot always be extrapolated to linked to VEGF overexpression, endothelial tight junction iron content in the retina. Diabetic mice with an intact BRB alterations, and vascular leakage leading to retinal edema.46 and systemic iron overload showed increased retinal ferritin, In or study, diabetic mice did no present any sign of BRB but no changes were found in iron content nor iron import breakdown or any related alterations, suggesting an early machinery in the retina. In addition, systemic iron overload stage of diabetic retinopathy in 20-week-old db/db mice. induced by an injection of iron dextran confirmed that reti- However, we can assume that, eventually, db/db mice will nal iron only increased when BRB was compromised. Over- develop BRB breakdown and iron will enter in the retina, all, we hypothesized that the retina might be protected by, producing iron overload, as happened when we experimen- at least, two different autonomous mechanisms in front of tally broke the BRB by cryopexy in iron dextran-injected systemic iron overload in diabetic retinopathy. The first limit- mice. Altogether, the results obtained showed that in the ing factor for iron entry would be the regulation at the BRB, retina a massive entrance of iron only occurs when BRB is and the second would be the increase of ferritin expression compromised. This mechanism has been recently showed in before BRB breakdown and retinal iron increase. the brain of aging mice with systemic iron overload, where, Monitoring iron in patients with type 2 diabetes is a after blood brain barrier breakdown, the brain becomes common practice in the follow-up of the disease.59 Our find- unprotected and suffers iron overload.47 ings show that systemic iron overload appears before retinal Signs of retinal alterations in db/db mice are reported iron overload, meaning that early detection and treatment to appear at variable stages of diabetic progression.37,48–50 with iron chelators could stop the passage of iron into the For instance, decrease of retinal thickness in db/db mice retina and prevent iron-induced retinal damage. has been described to appear in 20-month-old individuals in some reports.27,51 In contrast, our results did not show retinal thinning or thickening in 20-week-old mice, similar Acknowledgments to the results published by Yang et al.,52 where retinal thin- The authors truly thank Lorena Noya, Verónica Melgarejo, and ning appeared later on. The upregulation of VEGF has also Ángel Vázquez for their technical assistance. been reported in db/db retinas from 13-week-old mice.48 However, our results did not reveal any increase in VEGF Supported by the Ministerio de Ciencia e Innovación, Spain expression in 20-week-old mice, which is consistent with (grant number PID2021-122545OB-I00) and Fondo Europeo de an unaltered BRB. The variability in the onset of lesions Desarrollo Regional (FEDER). during diabetic retinopathy in mice could be explained by the different diets available among different animal facilities, Disclosure: A. Bonet, None; J. Pampalona, None; E. Jose- because the administration of neuroprotective or antioxidant Cunilleras, None; V. Nacher, None; J. Ruberte, None compounds present in the diet can delay the BRB disruption and related alterations.53 Another unexpected finding in our study was the overex- References pression of ferritin in the db/db mouse retinas without iron 1. Abbaspour N, Hurrell R, Kelishadi R. Review on iron and its increase, whereas other iron-handling proteins remained importance for human health. J Res Med Sci. 2014;19:164. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 13 2. Yau KW, Baylor DA. Cyclic GMP-activated conduc- 23. Chaudhary K, Promsote W, Ananth S, et al. Iron overload tance of retinal photoreceptor cells. Annu Rev Neurosci. accelerates the progression of diabetic retinopathy in asso- 1989;12:289–327. ciation with increased retinal renin expression. Sci Rep. 3. Moiseyev G, Takahashi Y, Chen Y, et al. RPE65 is an iron(II)- 2018;8(1):3025. dependent isomerohydrolase in the retinoid visual cycle. J 24. Shu W, Baumann BH, Song Y, Liu Y, Wu X, Dunaief JL. Biol Chem. 2006;281:2835–2840. Iron accumulates in retinal vascular endothelial cells but has 4. Adijanto J, Du J, Moffat C, Seifert EL, Hurley JB, Philp NJ. minimal retinal penetration after IP iron dextran injection The retinal pigment epithelium utilizes fatty acids for keto- in mice. Invest Ophthalmol Vis Sci. 2019;60:4378–4387. genesis. J Biol Chem. 2014;289:20570–20582. 25. Belke DD, Severson DL. Diabetes in mice with monogenic 5. Mayle KM, Le AM, Kamei DT. The intracellular traf- obesity: The db/db mouse and its use in the study of cardiac ficking pathway of transferrin. Biochim Biophys Acta. consequences. Methods Mol Biol. 2012;933:47–57. 2012;1820:264. 26. Burke SJ, Batdorf HM, Burk DH, et al. db/db Mice exhibit 6. Dunaief JL, Hahn P, Rouault T, Harris ZL. Retinal iron features of human type 2 diabetes that are not present metabolism and its possible role in AMD pathogenesis. in weight-matched C57BL/6J mice fed a western diet. J Invest Ophthalmol Vis Sci. 2004;45:2291–2291. Diabetes Res. 2017;2017:8503754. 7. Chen TT, Li L, Chung DH, et al. TIM-2 is expressed on B 27. Bogdanov P, Corraliza L, Villena JA, et al. The db/db mouse: cells and in liver and kidney and is a receptor for H-ferritin A useful model for the study of diabetic retinal neurodegen- endocytosis. J Exp Med. 2005;202:955–965. eration. PLoS One. 2014;9(5):e97302. 8. Li JY, Paragas N, Ned RM, et al. Scara5 is a ferritin 28. Musumeci M, Maccari S, Massimi A, et al. Iron excretion in receptor mediating non-transferrin iron delivery. Dev Cell. iron dextran-overloaded mice. Blood Transfus. 2014;12:485. 2009;16:35–46. 29. Wichaiyo S, Yatmark P, Morales Vargas RE, et al. Effect 9. Valença A, Mendes-Jorge L, Bonet A, et al. TIM2 modulates of iron overload on furin expression in wild-type and β- retinal iron levels and is involved in blood-retinal barrier thalassemic mice. Toxicol Reports. 2015;2:415–422. breakdown. Exp Eye Res. 2021;202(October 2020):108292. 30. Robinson R, Barathi VA, Chaurasia SS, Wong TY, Kern TS. 10. Mendes-Jorge L, Ramos D, Valença A, et al. L-ferritin binding Update on animal models of diabetic retinopathy: From to Scara5: A new iron traffic pathway potentially implicated molecular approaches to mice and higher mammals. DMM in retinopathy. PLoS One. 2014;9(9):e106974. Dis Model Mech. 2012;5:444–456. 11. Zhou ZD, Tan EK. Iron regulatory protein (IRP)-iron respon- 31. Derevjanik NL, Vinores SA, Xiao WH, et al. Quantitative sive element (IRE) signaling pathway in human neurode- assessment of the integrity of the blood-retinal barrier in generative diseases. Mol Neurodegener. 2017;12(1):75. mice. Invest Ophthalmol Vis Sci. 2002;43:2462–2467. 12. Gray NK, Hentze MW. Iron regulatory protein prevents bind- 32. Goebel W, Kretzchmar-Gross T. Retinal thickness in diabetic ing of the 43S translation pre-initiation complex to ferritin retinopathy: A study using optical coherence tomography and eALAS mRNAs. EMBO J. 1994;13:3882–3891. (OCT). Retina. 2002;22:759–767. 13. Meneghini R. Iron homeostasis, oxidative stress, and DNA 33. Daru J, Colman K, Stanworth SJ, De La Salle B, Wood EM, damage. Free Radic Biol Med. 1997;23:783–792. Pasricha SR. Serum ferritin as an indicator of iron status: 14. Zhao T, Guo X, Sun Y. Iron Accumulation and lipid peroxi- What do we need to know? Am J Clin Nutr. 2017;106(Suppl dation in the aging retina: Implication of ferroptosis in age- 6):1634S–1639S. related macular degeneration. Aging Dis. 2021;12:529–551. 34. Munro HN. Iron regulation of ferritin gene expression. J Cell 15. Hughes JM, Groot AJ, Van Groep P Der et al. Active Biochem. 1990;44:107–115. HIF-1 in the normal human retina. J Histochem Cytochem. 35. Wang J, Pantopoulos K. Regulation of cellular iron 2010;58:247. metabolism. Biochem J. 2011;434(Pt 3):365. 16. Ciudin A, Hernández C, Simó R. Iron overload in diabetic 36. Faruqi A, Mukkamalla SKR. Iron binding capacity. In: Stat- retinopathy: A cause or a consequence of impaired mecha- Pearls [Internet]. Treasure Island, FL: StatPearls Publishing; nisms? Exp Diabetes Res. 2010;2010:714108. 2022. Available from: https://www.ncbi.nlm.nih.gov/books/ 17. Wu W, Yuan J, Shen Y, et al. Iron overload is related to NBK559119/. elevated blood glucose levels in obese children and aggra- 37. Cheung AKH, Fung MKL, Lo ACY, et al. Aldose reductase vates high glucose-induced endothelial cell dysfunction in deficiency prevents diabetes-induced blood-retinal barrier vitro. BMJ Open Diabetes Res Care. 2020;8(1):e001426. breakdown, apoptosis, and glial reactivation in the retina of 18. Misra G, Bhatter SK, Kumar A, Gupta V, Khan MY. Iron db/db mice. Diabetes. 2005;54:3119–3125. profile and glycaemic control in patients with type 2 38. Boyer DS, Hopkins JJ, Sorof J, Ehrlich JS. Anti-vascular diabetes mellitus. Med Sci. 2016;4(4):22. endothelial growth factor therapy for diabetic macular 19. Liu J, Li Q, Yang Y, Ma L. Iron metabolism and type 2 edema. Ther Adv Endocrinol Metab. 2013;4(6):151–169. diabetes mellitus: A meta-analysis and systematic review. 39. Campbell M, Humphries P. The blood-retina barrier tight J Diabetes Invest. 2020;11:946–955. junctions and barrier modulation. Adv Exp Med Biol. 20. Hahn P, Chen L, Beard JL, Harris ZL, Dunaief JL. Disrup- 2013;763:70–84. tion of Ceruloplasmin and Hephaestin in mice causes reti- 40. Tian R, Luo Y, Liu Q, et al. The effect of claudin-5 overex- nal iron overload and retinal degeneration with features pression on the interactions of claudin-1 and -2 and barrier of age–related macular degeneration. Invest Ophthalmol Vis function in retinal cells. Curr Mol Med. 2014;14:1226–1237. Sci. 2004;45:2292–2292. 41. Gnana-Prakasam JP, Chaudhary K, Promsote W, et al. Reti- 21. Gnana-Prakasam JP, Ananth S, Prasad PD, et al. Expression nal iron overload during diabetic retinopathy accelerates and iron-dependent regulation of succinate receptor GPR91 ganglion cell death. Invest Ophthalmol Vis Sci. 2016;57:751– in retinal pigment epithelium. Invest Ophthalmol Vis Sci. 751. 2011;52:3751–3758. 42. Ugarte M, Grime GW, Osborne NN. Distribution of trace 22. Arjunan P, Gnanaprakasam JP, Ananth S, et al. Increased elements in the mammalian retina and cornea by use of retinal expression of the pro-angiogenic receptor GPR91 particle-induced X-ray emission (PIXE): Localisation of zinc via BMP6 in a mouse model of juvenile hemochromatosis. does not correlate with that of metallothioneins. Metal- Invest Ophthalmol Vis Sci. 2016;57:1612–1619. lomics. 2014;6:274–278. Retinal Protection in Front Iron Overload IOVS | March 2023 | Vol. 64 | No. 3 | Article 22 | 14 43. Ugarte M, Grime GW, Lord G, et al. Concentration of 51. Sheskey SR, Antonetti DA, Rentería RC, Lin CM. Correla- various trace elements in the rat retina and their distri- tion of retinal structure and visual function assessments bution in different structures. Metallomics. 2012;4:1245– in mouse diabetes models. Invest Ophthalmol Vis Sci. 1254. 2021;62(10):20. 44. Hahn P, Qian Y, Dentchev T, et al. Disruption of cerulo- 52. Yang Q, Xu Y, Xie P, et al. Retinal neurodegeneration in plasmin and hephaestin in mice causes retinal iron over- db/db mice at the early period of diabetes. J Ophthalmol. load and retinal degeneration with features of age-related 2015;2015:757412. macular degeneration. Proc Natl Acad Sci. 2004;101:13850– 53. Rossino MG, Dal Monte M, Casini G. Relationships 13855. between neurodegeneration and vascular damage in 45. Zhao L, Li Y, Song D, et al. A high serum iron level causes diabetic retinopathy. Front Neurosci. 2019;13:1172. mouse retinal iron accumulation despite an intact blood- 54. Hentze MW, Kühn LC. Molecular control of vertebrate iron retinal barrier. Am J Pathol. 2014;184:2862–2867. metabolism: mRNA-based regulatory circuits operated by 46. Zhang X, Zeng H, Bao S, Wang N, Gillies MC. Diabetic macu- iron, nitric oxide, and oxidative stress. Proc Natl Acad Sci lar edema: New concepts in patho-physiology and treat- USA. 1996;93:8175. ment. Cell Biosci 2014;4:1–14. 55. Hahn P, Dentchev T, Qian Y, Rouault T, Harris ZL, Dunaief 47. Mezzanotte M, Ammirata G, Boido M, Stanga S, Roetto JL. Immunolocalization and regulation of iron handling A. BBB damage in aging causes brain iron deposits proteins ferritin and ferroportin in the retina. Mol Vis. via astrocyte-neuron crosstalk and Hepc/Fpn1 pathway. 2004;10:598–607. bioRxiv. 2021:2021–07. 56. Aisen P, Enns C, Wessling-Resnick M. Chemistry and biol- 48. Li J, Wang JJ, Yu Q, Chen K, Mahadev K, Zhang SX. Inhi- ogy of eukaryotic iron metabolism. Int J Biochem Cell Biol. bition of reactive oxygen species by lovastatin downreg- 2001;33:940–959. ulates vascular endothelial growth factor expression and 57. Ferńandez-Real JM, Mcclain D, Review MM. Mechanisms ameliorates blood-retinal barrier breakdown in db/db mice. linking glucose homeostasis and iron metabolism toward Diabetes. 2010;59:1528–1538. the onset and progression of type 2 diabetes. Diabetes Care. 49. Xiong SQ, Jiang HB, Xu HZ, Xia XB. Effect of pyridone agent 2015;38:2169–2176. on blood-retinal barrier in diabetic mice. Int J Ophthalmol. 58. Kernan KF, Carcillo JA. Hyperferritinemia and inflammation. 2017;10:890–895. Int Immunol. 2017;29:401. 50. Kim J, Kim C-S, Lee IS, et al. Extract of Litsea japonica 59. Saha S, Murgod R. Evaluation of iron profile in type ameliorates blood–retinal barrier breakdown in db/db mice. II diabetes mellitus cases. Int J Biotechnol Biochem. Endocrine. 2014;46:462–469. 2019;15:27–37.