Biochimica et Biophysica Acta 1842 (2014) 603–612

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbad is

Regulation of the cholesterol efflux transporters ABCA1 and ABCG1 inretina in hemochromatosis and by the endogenous siderophore2,5-dihydroxybenzoic acid☆

Sudha Ananth a, Jaya P. Gnana-Prakasam a, Yangzom D. Bhutia a, Rajalakshmi Veeranan-Karmegam a,Pamela M. Martin a, Sylvia B. Smith b, Vadivel Ganapathy a,⁎a Department of Biochemistry and Molecular Biology, Georgia Regents University, Augusta, GA 30912, USAb Department of Cellular Biology and Anatomy, Georgia Regents University, Augusta, GA 30912, USA

Abbreviations: RPE, retinal pigment epithelium; ABCporter A1; ABCG1, ATP-binding cassette transporterdehydrogenase-2; 2,5-DHBA, 2,5-dihydroxybenzoic acid;generation; pRPE, primary cultures of mouse retinal pigmry cultures of mouseMuller cells; pGC, primary cultures ofAza-C, 5-azacytidine; TfR, transferrin receptor; HAMP, hep☆ Disclosure: All authors declare that they have no cohave no financial interest in the information contained in⁎ Corresponding author. Tel.: +1 706 721 7652; fax: +

E-mail address: vganapat@gru.edu (V. Ganapathy).

0925-4439/$ – see front matter © 2014 Elsevier B.V. All rhttp://dx.doi.org/10.1016/j.bbadis.2014.01.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 12 August 2013Received in revised form 8 January 2014Accepted 13 January 2014Available online 23 January 2014

Keywords:Cholesterol transportHemochromatosisMammalian siderophoreβ-Hydroxybutyrate dehydrogenase-2Retinal pigment epithelial cells

Hypercholesterolemia and polymorphisms in the cholesterol exporter ABCA1 are linked to age-related maculardegeneration (AMD). Excessive iron in retina also has a link to AMD pathogenesis. Whether these findings meana biological/molecular connection between iron and cholesterol is not known. Here we examined the relationshipbetween retinal iron and cholesterol using a mouse model (Hfe−/−) of hemochromatosis, a genetic disorder ofiron overload.We compared the expression of the cholesterol efflux transporters ABCA1 andABCG1 and cholesterolcontent in wild type and Hfe−/− mouse retinas. We also investigated the expression of Bdh2, the rate-limitingenzyme in the synthesis of the endogenous siderophore 2,5-dihydroxybenzoic acid (2,5-DHBA) in wild type andHfe−/− mouse retinas, and the influence of this siderophore on ABCA1/ABCG1 expression in retinal pigmentepithelium. We found that ABCA1 and ABCG1 were expressed in all retinal cell types, and that their expressionwas decreased in Hfe−/− retina. This was accompanied with an increase in retinal cholesterol content. Bdh2 wasalso expressed in all retinal cell types, and its expression was decreased in hemochromatosis. In ARPE-19 cells,2,5-DHBA increased ABCA1/ABCG1 expression and decreased cholesterol content. This was not due to depletionof free iron because 2,5-DHBA (a siderophore) and deferiprone (an iron chelator) had opposite effects on transferrinreceptor expression and ferritin levels. We conclude that iron is a regulator of cholesterol homeostasis in retina andthat removal of cholesterol from retinal cells is impaired in hemochromatosis. Since excessive cholesterol ispro-inflammatory, hemochromatosis might promote retinal inflammation via cholesterol in AMD.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

There is an overwhelming evidence for the involvement of oxidativestress and inflammation in the pathogenesis of age-related macular de-generation (AMD), a leading cause of blindness in adults in developedcountries [1–4]. There are two forms of AMD, dry and wet, the formerwith accumulation of drusen deposits underneath the basal side of theretinal pigment epithelium and the latter with abnormal developmentof new blood vessels, which invade into the retina through theBruch's membrane. There is no consensus on whether or not a similarmechanism participates in the pathogenesis of both forms, but the

A1, ATP-binding cassette trans-G1; Bdh2, β-hydroxybutyrateAMD, age-related macular de-

ent epithelial cells; pMC, prima-mouse retinal ganglion cells; 5-atic anti-microbial peptidenflict of interest, and that theythe present manuscript.1 706 721 9947.

ights reserved.

involvement of oxidative stress and inflammation has been implicatedin dry as well as wet AMD. Accordingly, most of the animal models ofAMD are based either on disruption of inflammatory pathways suchas the complement system or chemokine signaling or on oxidativedamage [5].

Iron is essential for the survival of all cells, but excessive iron isdetrimental because free iron generates potent reactive oxygen speciessuch as hydroxyl radicals. Numerous studies have suggested a role forexcessive accumulation of iron in the retina as an important contributorto the pathogenesis of AMD [6–8]. Retina expressesmost of the proteinsthat are known to participate in the regulation of iron homeostasis, afeature obligatory for maintenance of iron levels optimal for cellularfunction without the risk of oxidative stress [9]. Disruptions in theexpression and function of these iron-regulatory proteins alter iron levelsin the retina, involving all retinal cell types. Hemochromatosis is a hered-itary disorder caused by loss-of-functionmutations infive different genescoding for iron-regulatory proteins, namely HFE, hemojuvelin (also calledHFE2), hepcidin (also called HAMP), ferroportin, and transferrin receptor 2(TfR2) [10–13]. This disorder represents one of themost prevalent genet-ic diseases in humans, with homozygosity or compound heterozygosityaccounting for 1 in 300 in the general population. Mutations in HFE are

604 S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

seen in N85% of patients with hemochromatosis, the remaining ~15%arising from mutations in the other four genes. Irrespective of thegene involved, hemochromatosis is associatedwith excessive iron accu-mulation in multiple systemic organs, leading to oxidative stress andconsequent organ dysfunction. It was believed for a long time thatthe retina and the brain are spared in this disease because of theblood-retinal barrier or blood–brain barrier, but emerging evidencestrongly indicates otherwise [9,14]. Mousemodels of hemochromatosis(Hfe−/−, Hjv−/−, and hepcidin−/− mice) have provided unequivocal ev-idence for excessive iron accumulation in the retina in this disease[15–17]. These mice exhibit morphological and biological changes inthe retina that are similar to those found in AMD [15–17], indicatingthat iron-mediated oxidative stress is likely to be a critical determinantin the pathogenesis and/or progression of AMD.

With regard to the role of inflammation in AMD, available evidencepoints to the involvement of diverse immune cell types, cytokines, andsignaling pathways [3,4]. Cholesterol is receiving increasing attentionin recent years as a critical determinant of inflammatory pathways[18–20], and two transporters, ABCA1 and ABCG1, both participatingin the efflux of cholesterol from peripheral tissues to load it to on HDL,have been implicated in this process [21,22]. Furthermore, there isstrong evidence from animal models of hypercholesterolemia forexcessive cholesterol build-up in the retina as a causative factor inAMD [23–26].

From what has been described above, it is likely that excessive ironin retina may contribute to the pathogenesis/progression of AMDthrough oxidative damage and that excessive cholesterol in retinamay also contribute to the pathogenesis/progression of AMD throughinflammation. However, whether there is any link between iron andcholesterol in health and disease is not known. The purpose of thepresent study was to investigate the expression of the cholesterol effluxtransporters ABCA1 and ABCG1 in the retina in normal mice and in amouse model of hemochromatosis (Hfe−/− mouse) and also to deter-mine how the iron status in retina and RPE influences the expressionof these two transporters. Furthermore, there has been a recent devel-opment in the area of iron homeostasis in mammalian cells, which in-volves identification of an endogenous siderophore (i.e., iron carrier)and the enzyme critical for its synthesis [27]. The siderophore is2,5-dihydroxybenzoic acid (2,5-DHBA) and the enzyme is the cytosolicβ-hydroxybutyrate dehydrogenase, known as Bdh2. This siderophorebinds iron and facilitates its entry through plasma membrane andmitochondrial inner membrane [27]. There is no information availableat present on the expression of Bdh2 and on the role, if any, of thenewly discovered endogenous siderophore in the retina; therefore,here we studied the expression of this important iron-regulatoryenzyme in the retina in wild type and Hfe−/− mice and also examinedthe influence of the siderophore 2,5-DHBA on the expression ofABCA1 and ABCG1 in RPE cells.

2. Materials and methods

2.1. Materials

Antibodies were obtained from the following sources: rabbit poly-clonal anti-ABCA1 and rabbit polyclonal anti-ABCG1 (Novus Biologicals,Littleton, CO, USA), goat polyclonal anti-Bdh2 (Abcam, Cambridge, MA,USA), mouse monoclonal anti-vimentin (Millipore, Billerica, MA, USA)and chicken anti-MCT1 (Alpha Diagnostic International, San Antonio,TX, USA), goat anti-rabbit IgG coupled to Alexa Fluor 568, donkeyanti-goat IgG coupled to Alexa Fluor 568, goat anti-chicken IgG coupledto Alexa Fluor 568, and goat anti-rabbit and anti-mouse IgG coupled toAlexa Fluor 488 (Molecular Probes, Carlsbad, CA, USA). The dilutions ofthe antibodies used for immunofluorescence experimentswere: 1:1000for ABCA1, 1:25 for ABCG1, 1:50 for vimentin, 1:50 for Bdh2, and 1:1000for MCT1. Rabbit polyclonal antibodies specific for L-ferritin and

H-ferritin were provided by Professor P. Arosio (Dipartimento MaternoInfantile e Tecnologie Biomediche, Universita di Brescia, Brescia, Italy).

2.2. Animals

Breeding pairs of Hfe+/− mice were obtained from the JacksonLaboratory (Bar Harbor, ME, USA). Age- and gender-matched wildtype and Hfe−/− mice were obtained from the same litter originatingfrom the mating of heterozygous mice. Albino Balb/c mice (6-week-old) were used for immunofluorescence analyses in some studies.Mice were purchased from Harlan–Sprague Dawley, Inc. (Indianapolis,IN, USA). All experimental procedures involving these animals adheredto the “Principles of Laboratory Animal Care” (National Institutes ofHealth publication #85-23, revised in 1985) and were approved bythe Institutional Committee for Animal Use in Research and Education.

2.3. Immunofluorescence analysis

Eyes were embedded in OCT compound and frozen at –20 °C. Sec-tions (8-μm thick) were used for immunostaining by fixing in 4% para-formaldehyde for 10 min, washed with 0.01 M phosphate-bufferedsaline (pH 7.4), and blocked with 1× Power Block for 60 min. Sectionswere then incubated overnight at 4 °C with appropriate primary anti-bodies. Negative controls involved omission of the primary antibodies.Sections were rinsed and incubated for 1 h with appropriate secondaryantibodies. Coverslips weremounted after stainingwith Hoechst nucle-ar stain and sections were examined by epifluorescence microscopy(Axioplan-2 microscope, equipped with an HRM camera and the Axio-Vision imaging program; Carl Zeiss, Jena, Germany).

2.4. Primary RPE, Müller, and ganglion cell cultures from mouse eyes

Primary cultures of RPE were prepared as described previously[15,17]. Three-week-oldwild type andHfe−/−micewere used to estab-lish primary cultures of RPE. The purity of the culture was verified byimmunodetection of RPE-65 (retinal pigment epithelial protein 65)and CRALBP (cellular retinaldehyde binding protein). Müller cellswere prepared from 7- to 10-day-old C57BL/6 mice by a methodadapted fromHicks and Courtois [28] and described in one of our previ-ous publications [29]. Staining for the Müller cell markers glutaminesynthetase, glutamate transporter EAAT1, and CRALBP confirmed thepurity of the primary cultures. Retinal ganglion cells were isolated byimmunopanning from 2-day-old C57BL/6 mice by themethod of Barreset al. [30] as described previously in our publications [31,32]. The purityof the cultures was confirmed by immunostaining for Thy-1, a ganglioncell marker.

2.5. Real time PCR

The following real time primers were used: 5′-AGTTTCGGTATGGCGGGTTT-3′ (forward) and 5′-AGCATGCCAGCCCTTGTTAT-3′ (reverse)for mouse ABCA1; 5′-ACCTACCACAACCCAGCAGACTTT-3′ (forward)and 5′-GGTGCCAAAGAAACGGGTTCACAT-3′ (reverse) for mouseABCG1; 5′-GATGCAACTGTGTGTGTCCAGGAA-3′ (forward) and 5′-ACAGGGTTGCCAGTTACATAGGCT-3′ (reverse) for mouse Bdh2; 5′-GAAGTACATCAGAACATGGGC-3′ (forward) and 5′-GATCAAAGCCATGGCTGTAG-3′ (reverse) for human ABCA1; 5′-CAGGAAGATTAGACACTGTGG-3′ (forward) and 5′-GAAAGGGGAATGGAGAGAAG-3′ (reverse)for human ABCG1; 5″-GAGGACGCGCTAGTGTTCTT-3′ (forward) and5′-TGTGACATTGGCCTTTGTGTT-3′ (reverse) for human transferrin re-ceptor 1 (TfR1). 18S and hypoxanthine phosphoribosyl transferase 1(HPRT1) were used as internal controls. Real-time amplificationsusing SYBR Green detection chemistry were run in triplicate on 96-well reaction plates. Reactions were prepared in a total volume of25 μl containing: 5 μl cDNA, 0.5 μl of each 20 μM primer, 12.5 μl ofSYBR® Green Supermix and 6.5 μl RNase/DNase-free sterile water.

605S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

Blank controls were run in triplicate for eachmaster mix. The cycle con-ditions were set as follows: initial template denaturation at 95 °C for1 min, followed by 40 cycles of denaturation at 95 °C for 10 s, and com-bined primer annealing/elongation at 60 °C for 30 s and 70 °C for 45 s.This cycle was followed by a melting curve analysis, ranging from56 °C to 95 °C, with temperature increasing by steps of 0.5 °C every10 s. Baseline and threshold values were automatically determined forall plates using the StepOne software. Raw Ct values were transformedto quantities using an Excel spreadsheet. The resulting data were con-verted into correct input files, according to the requirements of the soft-ware, and analyzed for fold change using the Excel spreadsheet.

RNA isolated fromwhole retinawas used for qPCR. Mice were killedby cervical dislocation under isoflurane anesthesia and the eyes wereremoved. Extraocular tissues connected to the eyeball were removedwith scissors, and then a small cut was made on the anterior side ofthe eyeball to remove the lens and the vitreous. The remainingpart consisting of the neural retina, RPE, and choroid was used forpreparation of RNA.

2.6. Treatment of primary RPE cells with 5-azacytidine

Wild type andHfe−/− primary RPE cells were seeded in 6well platesand treated with the DNA-demethylating agent, 5-azacytidine (5-AzaC;2 μg/ml) for 24 h. RNA was isolated from the cells for real-time PCR.

2.7. Analysis of ABCG1 and ABCG1 expression in control and DNMT1−/−,DNMT3b−/−, and DNMT1−/−/DNMT3b−/− cancer cells

Control HCT116 cells, a human colon cancer cell line, and isogenicHCT116 cells with deletion of DNMT1 (DNMT1−/−), DNMT3b(DNMT3b−/−), or both (double-knockout or DKO) were originally pro-vided by Dr. Bert Vogelstein (Johns Hopkins University, Baltimore,MD, USA). RNA prepared from these cells was used for RT-PCR to deter-mine the role of DNMT1 and DNMT3b in the control of expression ofABCA1 and ABCG1.

2.8. Quantification of cholesterol levels

The levels of total cholesterol in retinal tissue and ARPE-19 cellswere measured using a commercially available assay kit (Cell Biolabs,Inc., San Diego, CA, USA). This is a fluorometric assay, which involvesthe extraction of total cholesterol with chloroform/isopropanol/NP-40.In the case of retinal tissue, lens and vitreous were removed from theeye and the rest of the tissue was used for the assay. The data arepresented μM in the tissue/cell extracts; based on the volumes of thetissue extracts that were used in the technique, 1 μM equals 0.8 nmolof cholesterol/retina and 0.67 nmol/mg of protein for ARPE-19 cells.

2.9. Influence of 2,5-DHBA (a siderophore) and deferiprone (an ironchelator) on TfR1 mRNA levels and ferritin protein levels

ARPE-19 cells were cultured in 6-well culture dishes for two daysand then treated with 2,5-DHBA (250 μM) or deferiprone (250 μM)for 16 h and then the cells were used for preparation of RNA and proteinextracts. Untreated cells were used as the control. Steady-state levels ofTfR1mRNAweremonitored by RT-PCR and qPCR. To determine the fer-ritin levels, protein was extracted from the cells using the lysis buffer(10 mM Tris/HCl buffer, pH 7.6, 50 mM NaCl, 1% Triton X-100, 5 mMethylenediamine tetraacetate) containing a cocktail of protease andphosphatase inhibitors (Thermo Scientific, Waltham, MA, USA). Thelevels of ferritin (heavy chain H aswell as light chain L)weremonitoredby Western blot using specific antibodies. β-Actin was used as theinternal control for protein loading in Western blot analysis.

2.10. Statistics

Statistical differenceswere calculated by paired Student's t test or byANOVA as appropriate. A p b 0.05 was considered as statisticallysignificant.

3. Results

3.1. Expression of ABCA1 and ABCG1 in mouse retina and their polarizeddistribution in RPE

We examined the expression of ABCA1 and ABCG1 in mouse retinaby immunofluorescence analysis. We used the albino Balb/c mice forthis purpose because the lack of pigment in RPEmakes it easy to visual-ize the polarized distribution of proteins in the basolateral membraneversus apical membrane. Both transporters were robustly expressed inthe retina with almost a similar expression pattern (Fig. 1A). The pro-teins were detectable in all cell layers of the retina, including the RPE.In photoreceptor cells, ABCA1 was detected in the inner segment andABCG1 in the outer segment. Since RPE is a polarized cell, we examinedby confocal microscopy the expression of ABCA1 and ABCG1 in the twopoles of the RPE plasma membrane, the apical membrane, which facesthe subretinal space and the basolateral membrane, which facesthe choroid. We used double-labeling for this purpose with MCT1(monocarboxylate transporter 1) as the marker for the RPE cell apicalmembrane [33]. We have used this marker successfully in many of ourpreviously published studies to determine the polarized expression ofseveral transporters and receptors [34–37]. We found ABCA1 as wellas ABCG1 to colocalize with MCT1, indicating the presence of bothtransporters in the apical membrane (Fig. 1B). The basal membranealso was immunopositive for both transporters, but the protein levelsin the apical membraneweremuch higher than in the basalmembrane.We confirmed the expression of ABCA1 and ABCG1 in primary mouseretinal ganglion cells, Muller cells, and RPE cells by real-time RT-PCR(Fig. 1C). ABCA1 expression was much higher in pRPE and pMC thanin pGC; in contrast, ABCG1 expression was much higher in pRPE thanin pMC and pGC.

3.2. Downregulation of ABCA1 and ABCG1 in hemochromatosis retina

We then compared the expression of both transporters betweenwild type mouse retina and Hfe−/− mouse retina. We used 18-month-old mice for these experiments because of our earlier findings that ret-inal degeneration becomes evident in Hfe−/− mice only at ages olderthan 1 year [15]. We found the expression of ABCA1 and ABCG1 to bedecreased in Hfe−/− mouse retinas (Fig. 2). The decrease was evidentin all cell layers of the retina. This phenomenon was confirmed by cor-responding changes in steady-state mRNA levels for both transportersin whole retina (Fig. 3A).

3.3. Involvement of DNA methylation in the suppression of ABCA1 andABCG1 expression in Hfe−/− mouse retina

There is evidence for the regulation of ABCA1 expression by promot-er methylation [38,39]; hypermethylation of ABCA1 in the promoterregion silences the gene expression. We have shown recently that thelevels of DNA methyltransferases are elevated in Hfe−/− mouse RPEcells [40]. These findings led us to hypothesize that the suppression ofABCA1 and ABCG1 observed in Hfe−/− retinas may occur epigeneticallythrough DNA methylation. We tested this hypothesis by studying theinfluence of the DNA methylation inhibitor 5′-azacytidine on thesteady-state levels of mRNA for ABCA1 and ABCG1 in pRPE cellsprepared fromwild type and Hfe−/− mouse retinas. Without any treat-ment, the levels of ABCA1 and ABCG1 expression were significantlylower in RPE cells from Hfe−/− mice than in RPE cells from wild typemice (Fig. 3B). Treatment of wild type RPE cells with the DNA

Fig. 1.Expression of ABCA1 andABCG1 inmouse retina. (A) The expression of ABCA1 andABCG1 inmouse (Balb/c) frozen retinal sectionswas studied by immunofluorescence. Red signalsindicate the protein expression and blue signals indicate nuclei. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. (B) Thepolarized expression of ABCA1 and ABCG1 in RPE cell apical membrane versus basolateral membrane was studied by double-labeling immunofluorescence using MCT1 as a marker forRPE cell apical membrane. Green, ABCA1 or ABCG1; red, MCT1; blue, nuclei. (C) Relative levels of ABCA1 and ABCG1 mRNAs in primary cultures of mouse retinal pigment epithelialcells (pRPE), Muller cells (pMC), and ganglion cells (pGC) as assessed by real-time RT-PCR. The mRNA level in pRPE was taken as 1.

606 S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

methylation inhibitor did not show any significant effect on the expres-sion of both transporters whereas under similar conditions the DNAmethylation inhibitor increased the expression of both transporters inHfe−/− RPE cells (Fig. 3B). In fact, the decrease inmRNA levels observedin Hfe−/− RPE cells compared to wild type RPE cells disappeared aftertreatmentwith 5′-azacytidine, suggesting thatDNAmethylationwas al-most entirely responsible for the hemochromatosis-associated decreasein the retinal expression of ABCA1 andABCG1. To determine the identityof DNMT isoform that was responsible for this phenomenon, we used ahuman colon cancer cell line (HCT116), which is available with andwithout DNMT1, DNMT3b, or both. These are isogenic cell lines, theonly difference being the presence or absence of specific DNMT iso-forms.We found the expression of ABCA1 andABCG1 to bemarkedly in-creased in DNMT1−/−/DNMT3b−/− double-knockout cells (data notshown), suggesting that the hemochromatosis-associated silencing ofthe cholesterol efflux transporters may similarly require DNMT1 orDNMT3b.

3.4. Retinal and RPE cholesterol levels in wild type and Hfe−/− mice

The function of ABCA1 and ABCG1 is to remove cholesterol from thecells by an active mechanism coupled to ATP hydrolysis. Therefore, thedecreased expression of these two transporters in hemochromatosismouse retina should be associated with increased cholesterol levels.This was indeed the case. The levels of cholesterol in whole retinawere significantly higher in Hfe−/− mice than in wild type mice(Fig. 4A). This phenomenon was also observed in the primary culturesof RPE cells prepared from wild type mice and Hfe−/− mice (Fig. 4B).

3.5. Effect of the mammalian siderophore 2,5-DHBA on the expression ofABCA1 and ABCG1 in RPE

It has recently been shown that mammalian cells synthesize anendogenous siderophore (i.e., an iron carrier), known as 2,5-dihydroxybenzoic acid (2,5-DHBA), which facilitates the transfer ofiron across the plasmamembrane aswell as across the innermitochon-drial membrane [27]. Since the expression of ABCA1 and ABCG1 in theretina is altered in the iron-overload disease hemochromatosis, wewondered whether treatment of RPE cells with the endogenoussiderophore would have any effect on the expression of these trans-porters. To address this issue, we treated the human RPE cell lineARPE-19, a widely used model for RPE [41], with 2,5-DHBA for 16 hand then monitored the levels of ABCA1 and ABCG1 mRNAs. These ex-periments showed that the expression of both transporters was upreg-ulated significantly in the RPE cell line when treated with 2,5-DHBA(Fig. 5A). This was associated with a decrease in the cellular cholesterolcontent as expected from the known function of these transporters inthe efflux of cholesterol from the cells (Fig. 5B). Since 2,5-DHBA is notsimply an iron chelator but rather an iron carrier responsible for ironentry into the cell and also into the mitochondria, we asked whetherthe observed effects of this siderophore on the expression of the choles-terol efflux transporters is due to depletion of free iron. To address thisquestion, we compared the influence of 2,5-DHBA with that ofdeferiprone, which functions simply as an iron chelator rather than asan iron carrier. ARPE-19 cells were treated with 250 μM of either 2,5-DHBA or deferiprone for 16 h and then the levels of TfR1 mRNA weremeasured by RT-PCR and qPCR and ferritin protein levels were mea-sured by Western blot. Steady-state levels of TfR1 mRNA as well as H

Fig. 2. Comparison of ABCA1 and ABCG1 protein expression in the retina betweenwild type mice (HFE-WT) and Hfe−/−mice (HFE-KO). (A) Expression of ABCA1 inmouse frozen retinalsectionswas studied by immunofluorescence. Red signals indicate ABCA1 protein and blue signals indicate nuclei. (B) Expression of ABCG1 inmouse frozen retinal sectionswas studied byimmunofluorescence. Red signals indicate ABCG1 protein and blue signals indicate nuclei. In both cases, age-matched (18-month-old) wild type and Hfe−/− mice were used for prepara-tion of retinal sections.

Fig. 3. ABCA1 and ABCG1mRNA levels inwild type andHfe−/−mouse retinas and the role of DNAmethylation on the expression of ABCA1 and ABCG1 in RPE. (A)mRNA levels for ABCA1andABCG1were compared betweenwild type (HFE-WT) andHfe−/− (HFE-KO)mouse retinas by real-time RT-PCR. *,p b 0.05 compared toHFE-WTbypaired Student's t test. (B) Influenceof DNA methylation on ABCA1 and ABCG1 expression in primary cultures of RPE cells. Cells prepared from wild type and Hfe−/− mice were subjected to treatment with and without5′-azacytidine (2 μg/ml) in the culture medium for 24 h. Following the treatment, RNA was prepared and used for real-time RT-PCR to quantify the levels of ABAC1 and ABCG1mRNAs. *, p b 0.05 as assessed by ANOVA.

607S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

Fig. 4. Retinal and RPE cholesterol levels in wild type (HFE-WT) and Hfe−/− (HFE-KO)mice. (A) Age-matched (18-month-old) wild type and Hfe−/− mouse retinas were usedfor measurement of cholesterol. **, p b 0.01 compared to wild type control by pairedStudent's t test. (B) Cholesterol content in primary cultures of RPE cells prepared from wildtype and Hfe−/− mice. **, p b 0.01 compared to wild type control by paired Student's t test.

Fig. 6. Influence of 2,5-DHBA (a siderophore) and deferiprone (an iron chelator) on thelevels of TfR1 mRNA and ferritin protein (heavy chain H and light chain L) in RPE. Thehuman RPE cell line ARPE-19 was exposed to 250 μM of either 2,5-DHBA or deferipronefor 16 h and then total RNA was isolated and protein extracts prepared. Untreated cells

608 S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

subunit of ferritin were decreased with 2,5-DHBA (Fig. 6). These datawere unexpected; while the changes in H-ferritin suggest depletion offree iron, the changes in TfR1 mRNA suggest the opposite. This was incontrast with the findings that treatment of the cells with deferiproneincreased the levels of TfR1 mRNA and decreased the levels ofH-ferritin,which indicate that deferiprone does indeed deplete intracel-lular levels of free iron under similar conditions as expected of an ironchelator.

were used as the control. The changes in the steady-state levels of TfR1 mRNAwere mon-itored by RT-PCR and qPCR (A & B) and those in ferritin (heavy chain H and light chain L)were monitored by Western blot (C).

3.6. Expression pattern of Bdh2, the enzyme critical for the biosynthesis of

2,5-DHBA, in the retina

The enzyme that is critical for the synthesis of 2,5-DHBA inmamma-lian cells is β-hydroxybutyrate dehydrogenase-2 (Bdh2) [27]. Weexamined the expression pattern of this enzyme in mouse retina. Wefound robust expression of the enzyme throughout the retina in all

Fig. 5. Influence of the endogenous iron chelator 2,5-dihydroxybenzoic acid (2,5-DHBA) on thwere treatedwith 250 μM2,5-DHBA for 24 h and the levels of ABCA1 and ABCG1mRNAswere tStudent's t test. (B) The cells were treatedwith 250 μM2,5-DHBA for 24 h and the cellular levelstest.

cell layers, including ganglion cells, Muller cells, and RPE cells (Fig. 7).The radial distribution pattern of the immune-positive signals in the ret-inal sections indicated the expression of the enzyme inMuller cells. Thiswas confirmed by double-labeling with vimentin, a marker for retinalMuller cells.WeusedRNA isolated fromprimarymouse retinal ganglion

e expression of ABCA1 and ABCG1 and cholesterol content in ARPE-19 cells. (A) The cellshen determined by real-time RT-PCR. *, p b 0.05; **, p b 0.01 compared to control by pairedof cholesterolwere then determined. **, p b 0.01 compared to control by paired Student's t

Fig. 7. Expression pattern of Bdh2 in mouse retina. Mouse (Balb/c) frozen retinal sections were used for immunofluorescence to determine the expression pattern of Bdh2 (red fluores-cence signals). Vimentin (green fluorescence signals) was used as a marker for Muller cells and Hoechst stain (blue fluorescence signals) for nuclei. GCL, ganglion cell layer; INL, innernuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium.

609S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

cells, Muller cells, and RPE cells to determine the relative expressionlevels of this enzyme in the three retinal cell types (Figs. 8A, B). Theexpression in RPE was maximal, followed by Muller cells and thenganglion cells.

3.7. Regulation of Bdh2 expression in the retina in hemochromatosis

Since 2,5-DHBA as a siderophore is a critical component of ironhomeostasis in mammalian cells, we wondered whether the synthesisof this compound is altered in the iron-overload disease hemochroma-tosis. To address this issue, we examined the expression of Bdh2 inwild type mouse retina and Hfe−/− mouse retina. We found markedreduction in Bdh2 mRNA levels in hemochromatosis retinas comparedto wild type retinas (Fig. 8C). This was also evident at the protein level(Fig. 9). The decrease in Bdh2 expression observed in Hfe−/− mouseretina was evident in all cell layers.

4. Discussion

Themost salient findings of the present study can be summarized asfollows: (a) the cholesterol efflux transporters ABCA1 and ABCG1 areexpressed in three important cell types in the retina, namely theganglion cells, Muller cells, and RPE cells; (b) the expression of bothtransporters is downregulated to a significant extent in the iron-overload disease hemochromatosis, and the decrease in the expressionis evident in all three cell types (RPE, Muller cells, and ganglion cells);(c) the disease-associated silencing of ABCA1 and ABCG1 in the retina

Fig. 8. Relative expression of Bdh2 in primary cultures of mouse retinal ganglion cells, Muller ceKO)mice. (A) RT-PCR analysis of Bdh2mRNA levels in primary cultures ofmouse RPE cells (pRP(B) Real-time RT-PCR analysis of relative levels of Bdh2 mRNA in primary cultures of mouse RPtaken as 1. (C) Comparison of Bdh2mRNA in retinas from age-matched (18-month-old)wild typcontrol by paired Student's t test.

occurs most likely via epigenetic mechanism involving DNA methyla-tion; (d) the decrease in the expression of the two transporters inhemochromatosismouse retina is associatedwith an increase in choles-terol content in the tissue; (e) the treatment of the RPE cell lineARPE-19with the recently discovered endogenous siderophore 2,5-DHBA en-hances the expression of ABCA1 and ABCG1 with resultant decrease incellular content of cholesterol; (f) the mouse retina and several impor-tant retinal cell types (ganglion cells, Muller cells, and RPE cells) expressBdh2, which is the critical rate-limiting enzyme in the synthesis of2,5-DHBA; and (g) the expression of this enzyme is downregulated inhemochromatosis mouse retina.

The findings of this study are significant for several reasons. Thesehave relevance to retinal biology both in health and disease states. Cho-lesterol is an important component of cell membranes, not only as astructural constituent but also as a modulator of membrane permeabil-ity, and the state and function of microdomains such as the lipid rafts,which are critical for the transmission of intracellular signals from vari-ous cell-surface receptors. Therefore, it is essential to understand howthe cellular and tissue content of cholesterol are regulated in the retina.ABCA1 and ABCG1 represent two important transporters that mediatethe ATP-coupled efflux of cholesterol from the cells. There is some evi-dence in the literature for the expression of ABCA1 in the retina. Duncanet al. [42] and Simon et al. [43] have shown that ABCA1 is expressed inthe retina and RPE cells. The study by Duncan et al. [42] has demonstrat-ed the expression of ABCA1 protein in the inner segment of photorecep-tor cells by immunofluorescence analysis and in the basolateralmembrane of RPE cells by functional analysis. Our studies not only

lls, and RPE cells and retinal expression of Bdh2 in wild type (HFE-WT) and Hfe−/− (HFE-E),Muller cells (pMC), and ganglion cells (pGC). 18SmRNAwas used as an internal control.E cells (pRPE), Muller cells (pMC), and ganglion cells (pGC). The mRNA level in pRPE wase andHfe−/−mice asmonitored by real-time RT-PCR. ***, p b 0.001 compared towild type

Fig. 9. Comparison of Bdh2 protein expression in the retina betweenwild type (HFE-WT) andHfe−/− (HFE-KO)mice. Bdh2 protein expressionwasmonitored in frozen retinal sections byimmunofluorescence (red fluorescence signals) with Hoechst stain (blue fluorescence signals) to visualize nuclei. Age-matched (18-month-old) mice were used.

610 S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

confirm these earlier observations but also provide new evidence forthe presence of ABCA1 in the apical membrane of RPE cells. The expres-sion of ABCG1 has never been studied in the retina, and the presentstudy is the first to demonstrate the expression of this cholesterol effluxtransporter in the retina and retinal cell types. Even though the retinalexpression pattern of the two transporters is similar for most part,which includes expression in several important cell types (ganglioncells, Muller cells, and RPE cells) and presence in the apical as well asbasolateral membrane of RPE cells, the two transporters differ in oneimportant aspect. While ABCA1 is expressed in the inner segment ofthe photoreceptor cells, ABCG1 is restricted to the outer segment.

The functional significance of ABCA1 and ABCG1 in RPE is readilyapparent, particularly the findings that the transporters are expressedmore robustly in the apical membrane than in the basolateral mem-brane. RPE cells handle a large load of cholesterol arising from phagocy-tosis of the outer segments of photoreceptor cells; as such, RPE cellsmust possess transport mechanisms to remove cholesterol as a meansto prevent excessive accumulation of this lipid inside the cells. ABCA1and ABCG1 are likely to serve this important role. The quantitativelymore robust expression of both transporters in the apical membraneof RPE cells underlines the essential function of this cell in supplyingcholesterol to the neural retina. We speculate that the photoreceptor-derived cholesterol in RPE cells following phagocytosis is not onlyremoved from the retina via ABCA1 and ABCG1 in the basolateral mem-brane but is also effluxed into neural retina via ABCA1 and ABCG1 in theapical membrane for re-use by the photoreceptor cells. The supply ofcholesterol to the neural retina may not be the only function of ABCA1and ABCG1 in the apical membrane of RPE cells. There is evidence sug-gesting an essential role for ABCA1 in delivering the dietary lipids luteinand zeaxanthin to the retina [44,45]. Defects in the function of thistransporter are associated with a significant decrease in retinal levelsof these lipids. Therefore, in addition to the role in the re-circulation ofcholesterol between photoreceptor cells and RPE cells, the apically lo-cated ABCA1 and ABCG1 in RPE cells may also function in the supplyof lutein and zeaxanthin to neural retina.

Polymorphisms in ABCA1 have been implicated in the pathogenesisof AMD [46–49]. However, a major part of the effort to explain the linkbetween defective ABCA1 and AMD at the functional level has focused

on the relevance of this transporter to reverse-cholesterol transport inmediating cholesterol efflux from macrophages to load it on high-density lipoprotein (HDL). A recent study has shown that senescentmacrophages exhibit decreased expression and activity of ABCA1 andthat the resultant accumulation of cholesterol within the aged macro-phages polarizes these cells to an abnormal phenotype that promotespathologic vascular proliferation [50]. Deletion of Abca1 selectively inmacrophages mimics the phenotype of senescent macrophages. In con-trast, disruption of cholesterol homeostasis in retinal cells such as RPE asa potential contributor to the pathogenesis of AMD has received less at-tention. Our present studies reporting on the robustwidespread expres-sion of the cholesterol efflux transporters ABCA1 and ABCG1 in theretina are likely to form the basis for future research focusing on thepotential role of altered cholesterol homeostasis within retinal cells inthe pathogenesis of AMD. There is strong evidence supporting a patho-genic role of excessive cholesterol in AMD [23–26]; therefore, defects inthe function of the two cholesterol efflux transporters examined in thepresent study are expected to disrupt cholesterol homeostasis in theretina and hence may contribute to the progression of AMD.

The present studies also report for the first time on the associationbetween the genetic disease hemochromatosis and cholesterol accumu-lation within the retina and RPE. To the best of our knowledge, therehave been no studies in the literature indicating any role for iron inthe regulation of cholesterol homeostasis. Our studies show that ironis an important regulator of retinal cholesterol levels in health and dis-ease. Hemochromatosis is a disease of iron overload in systemic organsaswell as in retina. The present findings that excessive iron in the retinain amousemodel of hemochromatosis disrupts cholesterol homeostasisthrough downregulation of the cholesterol efflux transporters ABCA1and ABCG1 are interesting because of the increasing evidence of the in-volvement of excessive iron in the progression of AMD. Themechanismunderlying the decrease in ABCA1 and ABCG1 expression in hemochro-matosis mouse retina involves DNA methylation. Even though ourprevious studies showing an increase in the expression of DNAmethyl-transferases in hemochromatosis mouse RPE cells [40] offer amolecularmechanism for the silencing of the two cholesterol efflux transporters,how excessive iron increases the expression of DNMTs is not known.Nonetheless, the importance of the findings to retinal cholesterol

611S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

homeostasis as well as AMD is clear. The prevalence of hemochromato-sis is very high in general population. Interestingly, even though this isa genetic disease, the appearance of clinical complications is age-dependent. In patients with this disease, it takes decades for iron toaccumulate in tissues to pathogenic levels, precipitating clinical symp-toms. This is true not only for the pathological symptoms associatedwith systemic organs but also for the biochemical and morphologicalchanges associated with the retina. Given the similarities between theappearance of pathological complications in hemochromatosis andAMD in terms of age-dependence, we speculate that hemochromatosismay contribute to the progression of AMD through excessive iron. Accu-mulation of iron in tissues to abnormally high levels is known to causeoxidative damage, which could be a contributing factor in the progres-sion of AMD. The findings of the present study linking excessive ironto disruption of cholesterol homeostasis in the retina offer an additional,but hitherto unknown, potential molecular link between hemochroma-tosis and AMD.

Deletion ofHfe inmice leads to significant biochemical andmorpho-logical alterations in the retina [15,36,37]. Some of these changes aresimilar to those found in AMD. To what extent these similarities inretinal features seen in hemochromatosis and AMD can be explainedin terms of the changes in retinal cholesterol content is not known.The retinal pathology begins to appear in hemochromatosis mice onlyat N18 months of age [15,36]. It would be interesting to determine if theprogression of the retinal pathology is accelerated in Hfe−/−/Abca1−/−

mice or Hfe−/−/Abcg1−/− mice.Another important observation of the present study is the evidence

that the retina expresses the enzyme Bdh2 that is critical for the synthe-sis of the endogenous siderophore 2,5-DHBA, that the expression of thisiron-regulatory enzyme is altered in the retina in the iron-overload dis-ease hemochromatosis, and that the newly discovered siderophore 2,5-DHBA regulates cholesterol homeostasis in RPE. The expression of theenzyme is highest in RPE, much higher than inMuller cells and ganglioncells. These findings suggest that RPE and other cell types in the retinaare capable of synthesizing 2,5-DHBA. Being a siderophore, this com-pound is likely to play a critical role in the biology of iron in the retina,particularly with regard to subcellular distribution of iron in the cyto-plasm versus mitochondria due to the function of this compound as acarrier of iron across the plasmamembrane and the innermitochondrialmembrane [27]. It is quite intriguing to note that so many genes havealready been identified as key players in iron homeostasis and thatloss-of-function mutations in five different genes can lead to the iron-overload disease hemochromatosis. BDH2 represents the newest addi-tion to the already known plethora of genes that control iron homeosta-sis. The function of iron as a regulator of cholesterol homeostasis in RPEis evident from the present findings that exposure of RPE cells to the en-dogenous siderophore 2,5-DHBA influences the expression of ABCA1and ABCG1. Interestingly, the effect of 2,5-DHBA on the steady-statelevels of TfR1 mRNA is different from that of the iron chelatordeferiprone, suggesting that the consequences of 2,5-DHBA-inducedperturbation in subcellular distribution of iron inside the cell are notthe same as those of a global decrease in intracellular levels of freeiron caused by an iron chelator. However, 2,5-DHBA and deferipronehad similar effects on the cellular levels of H-ferritin. There is no infor-mation in the literature as to how changes in subcellular distributionof iron inside the cell influence TfR1 mRNA and ferritin protein levels.The results presented here are the first to describe the influence of theendogenous siderophore 2,5-DHBA on TfR1 mRNA and ferritin levels.Equally interesting is the observation that the retinal expression ofBdh2 is downregulated in hemochromatosis. This has significant impli-cations in thepathology of the disease because impaired synthesis of theendogenous siderophore is expected to decrease the delivery of iron tothe mitochondria. Since excessive iron in mitochondria is toxic to thecell, it is likely that the downregulation of the synthesis of thissiderophore in hemochromatosis represents a compensatory mecha-nism to blunt the toxic effects of excessive iron accumulation associated

with the disease. A recent study has demonstrated the downregulationof BDH2 in the liver of patients with hemochromatosis [51]. Ours is thefirst report on the expression of this enzyme in normal retina and on thesuppression of its expression in hemochromatosis, thus providing an-other important link between hemochromatosis and retinal pathology,particularly AMD.

Acknowledgements

This work was supported by the National Institutes of Health grantEY 019672.

References

[1] X. Cai, J.F. McGinnes, Oxidative stress: the achilles' heel of neurodegenerativediseases of the retina, Front. Biosci. 17 (2012) 1976–1995.

[2] S.G. Jarrett, M.E. Boulton, Consequences of oxidative stress in age-related maculardegeneration, Mol. Aspects Med. 33 (2012) 399–417.

[3] J. Ambati, J.P. Atkinson, B.D. Gelfand, Immunology of age-relatedmacular degeneration,Nat. Rev. Immunol. 13 (2013) 438–451.

[4] J. Ambati, B.J. Fowler, Mechanisms of age-related macular degeneration, Neuron 75(2012) 26–39.

[5] M.E. Pennesi, M. Neuringer, R.J. Courtney, Animal models of age related maculardegeneration, Mol. Aspects Med. 33 (2012) 487–509.

[6] X. He, P. Hahn, J. Iacovelli, R. Wong, C. King, R. Bhisitkul, M. Massaro-Giordano, J.L.Dunaief, Iron homeostasis and toxicity in retinal degeneration, Prog. Retin. EyeRes. 26 (2007) 649–673.

[7] J. Blasiak, J. Szaflik, J.P. Szaflik, Implications of altered iron homeostasis forage-related macular degeneration, Front. Biosci. 16 (2011) 1551–1559.

[8] D. Song, J.L. Dunaief, Retinal iron homeostasis in health and disease, Front. AgingNeurosci. 5 (2013) 24, http://dx.doi.org/10.3389/fnagi.2013.00024.

[9] J.P. Gnana-Prakasam, P.M. Martin, S.B. Smith, V. Ganapathy, Expression and functionof iron-regulatory proteins in retina, IUBMB Life 62 (2010) 363–370.

[10] J.L. Babitt, H.Y. Lin, The molecular pathogenesis of hereditary hemochromatosis,Semin. Liver Dis. 31 (2011) 280–292.

[11] A. Pietrangelo, A. Caleffi, E. Corradini, Non-HFE hepatic iron overload, Semin. LiverDis. 31 (2011) 302–318.

[12] J. Kaplan, D.M.Ward, I. De Domenico, Themolecular basis of iron overload disordersand iron-linked anemias, Int. J. Hematol. 93 (2011) 14–20.

[13] T. Ganz, E. Nemeth, Hepcidin and iron homeostasis, Biochim. Biophys. Acta 1823(2012) 1434–1443.

[14] W. Nander, J.R. Connor, HFE gene variants affect iron in the brain, J. Nutr. 141 (2011)729S–739S.

[15] J.P. Gnana-Prakasam, M. Thangaraju, K. Liu, Y. Ha, P.M. Martin, S.B. Smith, V.Ganapathy, Absence of iron-regulatory protein Hfe results in hyperproliferation ofretinal pigment epithelium: role of cystine/glutamate exchanger, Biochem. J. 424(2009) 243–252.

[16] M. Hadziahmetovic, Y. Song, P. Ponnuru, J. Iacovelli, A. Hunter, N. Haddad, J. Beard,J.R. Connor, S. Vaulont, J.L. Dunaief, Age-dependent retinal iron accumulation anddegeneration in hepcidin knockout mice, Invest. Ophthalmol. Vis. Sci. 52 (2011)109–118.

[17] J.P. Gnana-Prakasam, A. Tawfik, M. Romej, S. Ananth, P.M. Martin, S.B. Smith, V.Ganapathy, Iron-mediated retinal degeneration in haemojuvelin-knockout mice,Biochem. J. 441 (2012) 599–608.

[18] M.T. Montero-Vega, The inflammatory process underlying atherosclerosis, Crit. Rev.Immunol. 32 (2012) 373–462.

[19] M.G. Sorci-Thomas, M.J. Thomas, High density lipoprotein biogenesis, cholesterolefflux, and immune cell function, Arterioscler. Thromb. Vasc. Biol. 32 (2012)2561–2565.

[20] G.D. Norata, A. Pirillo, A.L. Catapano, HDLs, immunity, and atherosclerosis, Curr.Opin. Lipidol. 22 (2011) 410–416.

[21] L. Yvan-Charvet, N. Wang, A.R. Tall, Role of HDL, ABCA1, and ABCG1 transporters incholesterol efflux and immune responses, Arterioscler. Thromb. Vasc. Biol. 30(2010) 139–143.

[22] D. Ye, B. Lammers, Y. Zhao, I. Meurs, T.J. Van Berkel, M. Van Eck, ATP-binding cassettetransporters A1 and G1, HDL metabolism, cholesterol efflux, and inflammation:important targets for treatment of atherosclerosis, Curr. Drug Targets 12 (2011)647–660.

[23] S. Dithmar, C.A. Curcio, N.A. Le, S. Brown, H.E. Grossniklaus, Ultrastructural changesin Bruch's membrane of apolipoprotein E-deficient mice, Invest. Ophthalmol. Vis.Sci. 41 (2000) 2035–2042.

[24] M. Rudolf, B.Winkler, Z. Aherrahou, L.C. Doehring, P. Kaczmarek, U. Schmidt-Erfurth, In-creased expression of vascular endothelial growth factor associatedwith accumulationof lipids in Bruch's membrane of LDL receptor knockout mice, Br. J. Ophthalmol. 89(2005) 1627–1630.

[25] W. Hu, A. Jiang, J. Liang, H. Meng, B. Chang, H. Gao, X. Qiao, Expression of VLDLR inthe retina and evolution of subretinal neovascularization in the knockout mousemodel's retinal angiomatous proliferation, Invest. Ophthalmol. Vis. Sci. 49 (2008)407–415.

[26] M. Fujihara, E. Bartels, L.B. Nielsen, J.T. Handa, A human apoB100 transgenic mouseexpresses human apoB100 in the RPE and develops features of early AMD, Exp. EyeRes. 88 (2009) 1115–1123.

612 S. Ananth et al. / Biochimica et Biophysica Acta 1842 (2014) 603–612

[27] L.R. Davireddy, D.O. Hart, D.H. Goetz, M.R. Green, Amammalian siderophore synthe-sized by an enzyme with a bacterial homolog involved in enterobactin production,Cell 141 (2010) 1006–1017.

[28] D. Hicks, Y. Courtois, The growth and behaviour of rat retinal Müller cells in vitro. 1.An improved method for isolation and culture, Exp. Eye Res. 51 (1990) 119–129.

[29] N.S. Umapathy, W. Li, B.A. Mysona, S.B. Smith, V. Ganapathy, Expression and func-tion of glutamine transporters SN1 (SNAT3) and SN2 (SNAT5) in retinal Müllercells, Invest. Ophthalmol. Vis. Sci. 46 (2005) 3980–3987.

[30] B.A. Barres, B.E. Silverstein, D.P. Corey, L.L. Chun, Immunological, morphological, andelectrophysiological variation among retinal ganglion cells purified by panning,Neuron 1 (1988) 791–803.

[31] Y. Dun, B. Mysona, T. Van Ells, L. Amarnath, M.S. Ola, V. Ganapathy, S.B. Smith, Ex-pression of the cystine–glutamate exchanger (xc−) in retinal ganglion cells andregulation by nitric oxide and oxidative stress, Cell Tissue Res. 324 (2006) 189–202.

[32] P.S. Ganapathy, R.E. White, Y. Ha, B.R. Bozard, P.L. McNeil, R.W. Caldwell, S. Kumar,S.M. Black, S.B. Smith, The role of N-methyl-D-aspartate receptor activation inhomocysteine-induced death of retinal ganglion cells, Invest. Ophthalmol. Vis. Sci.52 (2011) 5515–5524.

[33] N.J. Philp, D. Wang, H. Yoon, L.M. Hjelmeland, Polarized expression of monocarbox-ylate transporters in human retinal pigment epithelium and ARPE-19 cells, Invest.Ophthalmol. Vis. Sci. 44 (2003) 1716–1721.

[34] P.M. Martin, J.P. Gnana-Prakasam, P. Roon, R.G. Smith, S.B. Smith, V. Ganapathy, Ex-pression and polarized localization of the hemochromatosis gene product HFE inretinal pigment epithelium, Invest. Ophthalmol. Vis. Sci. 47 (2006) 4238–4244.

[35] J.P. Gnana-Prakasam, M. Zhang, P.M. Martin, S.S. Atherton, S.B. Smith, V. Ganapathy,Expression of the iron-regulatory protein haemojuvelin in retina and its regulationduring cytomegalovirus infection, Biochem. J. 419 (2009) 533–543.

[36] J.P. Gnana-Prakasam, S. Ananth, P.D. Prasad, M. Zhang, S.S. Atherton, P.M. Martin,S.B. Smith, V. Ganapathy, Expression and iron-dependent regulation of succinate re-ceptor GPR91 in retinal pigment epithelium, Invest. Ophthalmol. Vis. Sci. 52 (2011)3751–3758.

[37] J.P. Gnana-Prakasam, S.K. Reddy, R. Veeranan-Karmegam, S.B. Smith, P.M. Martin, V.Ganapathy, Polarized distribution of heme transporters in retinal pigment epitheli-um and their regulation in the iron-overload disease hemochromatosis, Invest.Ophthalmol. Vis. Sci. 52 (2011) 9279–9286.

[38] S.P. Guay, D. Brisson, J. Munger, B. Lamarche, D. Gaudet, L. Bouchard, ABCA1 genepromoter DNA methylation is associated with HDL particle profile and coronary ar-tery disease in familial hypercholesterolemia, Epigenetics 7 (2012) 464–472.

[39] B.H. Lee, M.G. Taylor, P. Robinet, J.D. Smith, J. Schweitzer, E. Sehayek, S.M. Falzarano,C. Magi-Galluzzi, E.A. Klein, A.H. Ting, Dysregulation of cholesterol homeostasis inhuman prostate cancer through loss of ABCA1, Cancer Res. 73 (2013) 1211–1218.

[40] J.P. Gnana-Prakasam, R. Veeranan-Karmegam, V. Coothankandaswamy, S.K. Reddy,P.M. Martin, M. Thangaraju, S.B. Smith, V. Ganapathy, Loss of Hfe leads to

progression of tumor phenotype in primary retinal pigment epithelial cells, Invest.Ophthalmol. Vis. Sci. 54 (2013) 63–71.

[41] K.C. Dunn, A.E. Aotaki-Keen, F.R. Pulkey, L.M. Hjelmeland, ARPE-19, a human retinalpigment epithelial cell line with differentiated properties, Exp. Eye Res. 62 (1996)155–169.

[42] K.G. Duncan, K. Hosseini, K.R. Bailey, H. Yang, R.J. Lowe, M.T. Matthes, J.P. Kane, M.M.LaVail, D.M. Schwartz, J.L. Duncan, Expression of reverse cholesterol transport pro-teins ATP-binding cassette A1 (ABCA1) and scavenger receptor BI (SR-BI) in the ret-ina and retinal pigment epithelium, Br. J. Ophthalmol. 93 (2009) 1116–1120.

[43] E. Simon, B. Bardet, S. Gregoire, N. Acar, A.M. Bron, C.P. Creuzot-Garcher, L.Bretillon, Decreasing dietary linoleic acid promotes long chain omega-3 fattyacid incorporation into rat retina and modifies gene expression, Exp. Eye Res.93 (2011) 628–635.

[44] W.E. Conner, P.B. Duell, R. Kean, Y. Wang, The prime role of HDL to transport luteininto the retina: evidence from HDL-deficient WHAM chicks having a mutant ABCA1transporter, Invest. Ophthalmol. Vis. Sci. 48 (2007) 4226–4231.

[45] E. Yonova-Doing, P.G. Hysi, C. Venturini, K.M. Williams, A. Nag, S. Beatty, S.H. Liew,C.E. Gilbert, C.J. Hammond, Candidate gene study of macular response to supple-mental lutein and zeaxanthin, Exp. Eye Res. 115 (2013) 172–177.

[46] S. Zareparsi, M. Buraczynska, K.E. Branham, S. Shah, D. Eng, M. Li, H. Pawar, B.M.Yashar, S.E. Moroi, P.R. Lichter, H.R. Petty, J.E. Richards, G.R. Abecasis, V.M.Elner, A. Swaroop, Toll-like receptor 4 variant D299G is associated with sus-ceptibility to age-related macular degeneration, Hum. Mol. Genet. 14 (2005)1449–1455.

[47] W. Chen, D. Stambolian, A.O. Edwards, K.E. Branham, M. Othman, et al., Genetic var-iants near TIMP3 and high-density lipoprotein-associated loci influence susceptibil-ity to age-related macular degeneration, Proc. Natl. Acad. Sci. U. S. A. 107 (2010)7401–7406.

[48] I. Peter, G.S. Huggins, J.M. Ordovas, M. Haan, J.M. Seddon, Evaluation of new andestablished age-related macular degeneration susceptibility genes in the Women'sHealth Initiative Sight Exam (WHI-SE) study, Am J. Ophthalmol. 152 (2011)1005–1013.

[49] Y. Yu, R. Reynolds, B. Rosner, M.J. Daly, J.M. Seddon, Prospective assessment ofgenetic effects on progression to different stages of age-related macular degen-eration using multistate Markov models, Invest. Ophthalmol. Vis. Sci. 53 (2012)1548–1556.

[50] A. Sene, A.A. Khan, D. Cox, R.E.I. Nakamura, A. Santeford, B.M. Kim, R. Sidhu, M.D.Onken, J.W. Harbour, S. Hagbi-Levi, I. Chowers, P.A. Edwards, A. Baldan, J.S. Parks,D.S. Ory, R.S. Apte, Impaired cholesterol efflux in senescent macrophages promotesage-related macular degeneration, Cell Metab. 17 (2013) 549–561.

[51] Z. Liu, R. Lanford, S. Mueller, G.S. Gerhard, S. Luscieti, M. Sanchez, L. Devireddy,Siderophore-mediated iron trafficking in humans is regulated by iron, J. Mol. Med.90 (2012) 1209–1221.