Retinal Function and Structure in the Hypotransferrinemic Mouse

Abstract

Purpose.: The iron carrier transferrin is expressed at remarkably high levels in normal retinas and is upregulated during retinal degeneration. The authors characterized the consequences of genetically reduced retinal transferrin production on retinal structure and function.

Methods.: Hypotransferrinemic (HPX^−/−) mice treated with weekly intraperitoneal salvage transferrin injections were examined at 1 and 2 months of age. HPX^−/−, HPX^+/−, and wild-type (WT) mice were evaluated by electroretinography, ophthalmoscopy, and histology. Retinal iron content and transferrin levels were measured. RNA levels of genes involved in iron homeostasis and antioxidative response were determined by quantitative PCR. Oxidative injury was assessed by immunostaining for 4-hydroxy-2-nonenal (HNE).

Results.: At 2 months, dark-adapted, mixed rod-cone response b-wave amplitudes were significantly lower in HPX^−/− mice than in WT mice (340 ± 112 μV vs. 624 ± 134 μV [mean ± SEM]; P = 0.002). Oscillatory potentials were significantly suppressed in HPX mice, and ophthalmoscopy demonstrated marked retinal pallor. Quantitative immunostaining revealed a 39% reduction of transferrin content in HPX^−/− compared with WT retinas (P = 0.01). mRNA levels of Tf, Tf receptor, and ceruloplasmin were decreased, whereas mRNA for antioxidant genes were elevated in HPX^−/− retinas. HNE staining was reduced in mice carrying the mutant HPX allele. Histologic examination demonstrated preserved retinal structure, and retinal iron content was similar across the strains.

Conclusions.: Despite the lack of wild-type retinal transferrin production and low levels of retinal transferrin protein, the retinal morphology and retinal iron content in HPX^−/− mice treated by systemic salvage transferrin injections are normal until age 2 months. However, retinal function and gene expression of some of the iron-associated genes are significantly altered.

Iron is an essential element for normal brain and retinal physiology. However, excess labile iron can mediate oxidative injury through the Fenton reaction, ^1,2 and, hence, levels of redox-active iron are normally tightly controlled. Altered iron metabolism leading to oxidative injury has been implicated in the pathogenesis of several neurodegenerative diseases affecting the central nervous system and the retina including Alzheimer's, Parkinson's, and Huntington's diseases, as well as glaucoma and age-related macular degeneration (AMD). ^{2 –5} Evidence of altered iron homeostasis in AMD includes increased levels of chelatable iron, upregulation of transferrin (Tf), and decreased expression of bone morphogenetic protein 6 (Bmp6) in AMD retinas. ^{6 –8}  

Several highly regulated proteins are responsible for maintaining iron homeostasis, among them the iron carrier transferrin and its receptors (TFR1 and TFR2), the storage protein ferritin, and the ferroxidases ceruloplasmin and hephaestin (which facilitate iron loading onto transferrin). Additional genes involved in retinal iron metabolism are the hemochromatosis gene (HFE), which regulates transferrin's interaction with its receptor, hemojuvelin (HFE2), which serves as a coreceptor for BMP6, and hepcidin (HAMP) which is regulated by BMP6 and participates in the regulation of iron export. ^9,10 The importance of intact function of these proteins is underscored by the fact that mutations in the genes encoding HFE, HFE2, TFR2, and HAMP are associated with hemochromatosis, the most common disease of altered iron homeostasis. Furthermore, the importance of tightly controlled retinal iron levels is reflected by the high retinal expression levels of Tf, which are six times those found in the cortex or liver. ¹¹ The functional importance of retinal Tf expression is still unclear, but Tf has a key role in iron transport, and altered iron transport might lead to iron-associated oxidative injury through the Fenton reaction. ¹²  

The association of retinal degeneration with altered iron metabolism and expression of iron metabolism genes has been reported in several animal models. Ceruloplasmin/hephaestin-deficient mice develop retinal iron overload and manifestations reminiscent of macular degeneration. ¹³ Transferrin is degraded in RCS rats, ¹⁴ and major iron metabolism proteins are differentially expressed in rd10 and aged CCR2^−/− mice retinas as retinal degeneration progresses. ¹⁵ In addition, photic injury inflicted on the mouse retina causes ceruloplasmin levels to increase. ^16,17  

The hypotransferrinemic (HPX) mouse provides a platform for the assessment of transferrin function and importance in vivo. The naturally occurring HPX mutation, causing a splicing defect in the Tf gene, results in unstable Tf mRNA and reduced circulating transferrin levels in homozygote mice to approximately 1% of normal levels. ¹⁸ Although untreated HPX homozygotes die shortly after birth, systemic administration of transferrin can partially salvage these mice. HPX^−/− mice treated in this manner have liver iron overload, anemia, and abnormal iron uptake and distribution in the brain. ^{18 –21} The retinal consequences of the HPX mutation and its associated iron metabolism alterations are unknown. To gain insight into the effects of altered retinal Tf expression, we evaluated retinal function and structure in HPX mice up to the age of 2 months. 

Methods

Animals

HPX^−/− mice were obtained by mating heterozygotes carrying the mutation. The founders for the colony were generously provided by Andrew McKie (Kings College, London, United Kingdom). To ensure their survival, homozygotes were treated with weekly intraperitoneal injections of 0.75 mg human apotransferrin (Kamada, Beit Kama, Israel), as previously described. ²⁰ HPX^−/− mice were evaluated 1 week after the last transferrin injection. BALB/c littermates served as controls. The HPX mutation was detected by genotyping the region containing the single nucleotide mutation, as described by Trenor et al. ¹⁹  

All mice were maintained in a 12-hour light (white fluorescence, 30 cd/m²)/12-hour dark cycle of illumination and had unlimited access to food and water. Animals were treated in accordance to the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and all experiments were conducted with the approval of the institution's animal ethics committee. Before each procedure, mice were anesthetized by intraperitoneal injection of a mixture of ketamine (Bedford Laboratories, Bedford, OH) and xylazine (VMD, Arendonk, Belgium). Blood samples were collected for CBC analysis, performed at the central laboratory of the Hadassah-Hebrew University Medical Center (Jerusalem, Israel) using an automated system. 

Electroretinography

Full-field ERGs were recorded in anesthetized mice after overnight dark adaptation. Pupils were dilated with 1% tropicamide and 2.5% phenylephrine, and benoxinate HCl 0.4% (all from Fisher Pharmaceuticals, Tel-Aviv, Israel) drops were topically administered for local anesthesia. ERGs were recorded through active gold-wire electrodes placed on the center of each cornea. A reference electrode was placed on the tongue, and the ground electrode was inserted into the thigh muscles. The mouse was positioned facing the center of a Ganzfeld bowl, ensuring equal, simultaneous illumination of both eyes. ERGs were recorded inside a Faraday cage, under dim red lighting, using a computerized system (Espion E²; Diagnosys LLC, Littleton, MA). At 1 and 2 months of age, dark-adapted rod and mixed cone-rod responses as well as light-adapted (30 cd/m² background light) 1-Hz cone responses to increasing white flashes were recorded. All ERG responses were filtered at 0.3 to 500 Hz, and signal averaging was used. Oscillatory potentials (OPs) were recorded under dark-adapted conditions at a flash intensity of 48 cd · s/m² using 100- to 300-Hz filtering. 

Ophthalmoscopy

Ophthalmoscopy was performed after pupil dilatation with 1% tropicamide and 2.5% phenylephrine, and corneas were anesthetized with benoxinate HCl 0.4%. Photographs were taken with a fundus camera (Kowa, Tokyo, Japan) through a 90-diopter lens. 

Retinal Histology

For frozen sections eyes were enucleated and immediately placed in Davidson solution (25 mL glacial acetic acid, 71.25 mL ethanol; both from Bio-Lab, Jerusalem, Israel), 50 mL 10% neutral-buffered formalin (EMS), and 78.75 mL double-distilled water for overnight fixation. The following day, eyes were transferred to 30% sucrose for 24 hours, after which they were frozen in blocks of optimum cutting temperature embedding compound (Tissue-Tek; Sakura, Torrance, CA), and frozen on dry ice. Eyes were sliced into 6-μm thick sections, along the corneal-optic nerve axis, using a cryostat (CM 1100; Leica, Heidelberg, Germany). Sections were stained with hematoxylin and eosin or used for immunostaining as detailed below. Retinal thickness was measured in five retinas from each group at five distance points from both sides of the optic nerve in each eye, and average values were calculated. 

Immunohistochemistry

Immunostaining for transferrin was performed using polyclonal rabbit anti-transferrin antibody (DakoCytomation, Carpinteria, CA), and immunostaining with rabbit anti 4-hydroxy-2-nonenal (HNE) (Alpha Diagnostics, San Antonio, TX) antibody was performed to assess the extent of oxidative damage as we have previously described. ¹⁵ Briefly, slides were placed in a humidity chamber and submerged in PBS for 15 minutes. Blocking was performed by incubation in 1% bovine serum albumin (BSA; Amresco, Solon, OH) in PBS with 0.1% Triton X (Sigma-Aldrich, St. Louis, MO) and 3% normal goat serum for 1 hour in a humidity chamber. Slides were then incubated with primary antibody (diluted 1:100 in 1% BSA) for 1 hour at room temperature. Control sections were processed without primary antibody. After three consecutive 5-minute PBS rinses, the secondary antibody Cy3-conjugated goat anti-rabbit antibody (Jackson ImmunoResearch, West Grove, PA), diluted 1:200, was added, and sections were again incubated for 1 hour in the dark. Slides were rinsed again, followed by counterstaining with DAPI (Santa Cruz Biotechnology, Santa Cruz, CA). Sections were imaged on a fluorescence microscope (BX41; Olympus, Tokyo, Japan), using the appropriate filters. Background was controlled by setting the exposure parameters such that no detectable signal was obtained from the control sections, and these same parameters were maintained while capturing images from the test sections. Images were taken with a digital camera (DP70; Olympus). To ensure the staining of a similar retinal region from each eye, the same serial section number was chosen from each eye for the staining procedures. Quantification of immunochemistry was performed using ImageJ software (developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; available at http://rsb.info.nih.gov/ij/index.html) ²² by calculating the mean of the average staining intensity per pixel measured in five replicas. In each section, staining intensity was measured in an area of equal length, with the width set to include the entire thickness of the retina from the inner limiting membrane to the RPE. 

Real-Time Quantitative-PCR

For PCR and biochemical analysis, retinas were gently separated from freshly enucleated eyes under a dissecting microscope and immediately frozen in liquid nitrogen. Tissue samples were stored at −80°C until further use. Total RNA was extracted from pairs of frozen retinas using reagent (TRIzol; Sigma) according to the manufacturer's instructions and then were treated with DNAse (Turbo DNA-free; Ambion, Austin, TX). Reverse transcription–polymerase chain reaction was performed using a first-strand synthesis kit (Reverse-iT; ABgene, Epsom, UK). Anchored oligo dT primers were used to prepare cDNA from 1 μg RNA. Quantitative PCR was performed to measure mRNA levels of genes involved in iron metabolism and of antioxidant enzymes. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the endogenous control. Two sets of reactions were performed. In the first set, all reactions were carried out in duplicate at a total volume of 15 μL. Wells contained 40 ng cDNA template, 0.75 μL TaqMan Gene Expression assay (ceruloplasmin, Mm00432654_m1; transferrin, Mm01230431_m1; TFR1, Mm00441941_m1), 7.5 μL TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA]) and were completed with double-distilled water. An additional set of reactions was performed in triplicate at a total volume of 10 μL. Wells contained 30 ng cDNA template, 0.5 μL gene expression assay (TaqMan [hemojuvelin (HFE2), Mm01265683_m1; HFE, Mm00439314_m1; HAMP, Mm00519025_m1; hephaestin, Mm00515970_m1; TFR2, Mm00443703_m1; GAPDH, Mm99999915_g1]), and 5 μL universal master mix (TaqMan; Applied Biosystems) and were completed with double-distilled water. When using SYBR Green, each well contained amounts of cDNA and concentrations of primers determined as optimal by calibration for each gene (heme oxygenase 1 (Hmox1), 40 ng cDNA, 500 nM: forward CACGCATATACCCGCTACCT, reverse AAGGCGGTCTTAGCCTCTTC; GAPDH 0.5 ng, 500 nM: forward GTATGACTCCACTCACGGCAAA, reverse GGTCTCGCTCCTGGAAGATG; glutathione peroxidase 1 (GPX1) 15 ng, 700 nM: forward ACAGTCCACCGTGTATGCCTTC, reverse CTCTTCATTCTTGCCATTCTCCTG; superoxide dismutase 1 (SOD1) 5 ng, 700 nM: forward ATGGGGACAATACACAAGGCTG, reverse CAATGATGGAATGCTCTCCTGAG ²³ ), 7.5 μL SYBR Green quantitative PCR (qPCR) mix (DyNAmo HS; Finnzymes, Espoo, Finland), and double-distilled water. Amplification was measured throughout 40 cycles of 60°C for 15 seconds, followed by 95°C for 15 seconds. Fluorescent signals were measured with a PCR system (7900HT Fast Real-Time; Applied Biosystems) and analyzed with appropriate software (SDS 2.3 and RQ Manager 1.2; Applied Biosystems). Expression levels were compared according to relative quantification (2^−ΔΔCt) values. 

Identification of Mutant Tf mRNA

cDNA was prepared as described. PCR was performed using primers (forward CGTTAAACTTCCAGAGGGTACCAC, reverse CTGTCTCCACCACAGTGGCAACC ¹⁹ ) flanking the splice junction of exons 16 and 17 of the Tf transcript. Reaction tubes contained 10 μL master mix (Reddy Mix; ABgene), 1 μL each primer, 1 μL cDNA, and 7 μL double-distilled water. Amplification was carried out for 40 cycles at an annealing temperature of 55°C. PCR products were analyzed by electrophoresis on agarose gel and visualized by ethidium bromide staining under UV light. 

Protein Isolation

Lysis buffer containing 1% deionized Triton X-100 and 0.1% sodium azide in 50 mM Tris-HCl (Sigma), pH 7.5, was incubated (with Chelex-100; Bio-Rad, Hercules, CA) for 24 hours. Immediately before use, 0.25 mM phenylmethylsulfonyl fluoride was added (1:1000; Sigma). Pools of retinal tissue were homogenized in the buffer, sonicated at 10 W for 1 minute, and stored on ice for half an hour during vortexing every 5 minutes. Samples were then centrifuged at 2750g for 15 minutes at 4°C, after which supernatant was separated and refrozen at −80°C. Protein content was estimated using the BCA Protein Assay Kit (Pierce, Rockford, IL). 

Total Iron Content

Five samples (each containing two retinas homogenized together in 100 μL lysis buffer, as described) were spun in a vacuum centrifuge until dry and then dissolved in concentrated nitric acid (J.T. Baker, Phillipsburg, NJ). Samples were incubated at 37°C for 48 hours and diluted fivefold with double-distilled water. A calibration curve was calculated from serial dilutions of Fe³⁺ (50–1600 ng/mL). Each standard/sample was split and measured in duplicate. Samples were mixed 1:1 with a mixture containing 1.2 N HCl (Frutarom, Haifa, Israel), 10% trichloroacetic acid (TCA), and 3% thioglycolic acid (both from Sigma) and were left for 15 minutes at room temperature, followed by centrifugation at 2500g for 30 minutes. In a curette, equal volumes of a solution containing 0.025% bathophenanthroline sulfate (BPS) and sodium acetate 2 M in double-distilled water were added to the supernatant. The iron content was determined by measuring the optical absorption (λ = 535 nm) of the colored complex formed by Fe²⁺ ions and BPS. ²⁴  

Statistical Analysis

All data are presented as mean ± SEM. Statistical significance was calculated with scientific statistics software (Instat; GraphPad, San Diego, CA). The two-tailed t-test was applied when results were of normal distribution; otherwise, the nonparametric Mann-Whitney U test was performed. 

Results

Systemic Findings

HPX^−/− mice were frail, pale, and smaller than their heterozygote HPX^+/− and WT littermates. Despite transferrin supplementation, only a minority of HPX^−/− mice survived to the 2-month time point of the study (Supplementary Fig. S1A). Autopsy of HPX^−/− mice revealed markedly enlarged spleens and hearts and smaller and discolored livers compared to those of WT littermates. Complete blood count analysis of blood drawn 1 week after the last transferrin injection demonstrated 2.2-fold lower hemoglobin levels in 2-month-old HPX^−/− mice compared with WT mice. 

Retinal Function

Full-field ERGs were recorded from BALB/c (WT), HPX^+/− and HPX^−/− mice at 1 and 2 months of age to assess retinal function. WT and heterozygotes displayed similar responses under all stimulus conditions, whereas HPX^−/− mice showed significantly reduced scotopic and photopic ERG amplitudes at both ages (Figs. 1A, 1B). There was significant suppression of the OPs at both 1 and 2 months of age in the HPX retinas compared with WT controls (Figs. 1C, 1D). 

Figure 1.

 

Dark- and light-adapted ERG responses recorded from BALB/c, HPX^+/−, and HPX^−/− mice at 1 (A) and 2 (B) months of age. a- and b-wave amplitudes represent rod-derived (at lower intensities) and mixed rod-cone (at higher intensities) responses recorded under dark-adapted (scotopic) conditions. The cone 1-Hz responses were derived under photopic conditions (30 cd/m² background light). P < 0.05 (A, B) for all comparisons of HPX^−/− with BALB/c WT except for the maximal intensity b-wave and the 1-HZ stimuli at 1 month and the lowest stimulus intensity a-wave at 2 months. Representative OP wave for HPX and BALB/c mouse at 2 months of age show reduced OPs in the HPX mouse (C). Summed OP amplitudes from normal BALB/c and HPX^−/− mice at 1 and 2 months of age show reduced OPs in HPX^−/− mice (D). Results are presented as mean ± SEM. *P < 0.05 (n = 6–12 in each group).

Funduscopy and Retinal Morphology

Ophthalmoscopy revealed marked pallor of the HPX^−/− fundus compared with BALB/c mice, potentially attributable to severe anemia (Fig. 2). Optic discs and blood vessels appeared normal, and other alterations such as drusen-like deposits and pigmentary alterations were not detected up to the age of 2 months. 

Figure 2.

 

Color fundus photographs from BALB/c (A) and HPX^−/− (B) mice. Marked retinal pallor is evident in the HPX fundus.

Retinal morphology was evaluated in hematoxylin and eosin–stained cryosections. Despite the anemia, the retinal structure appeared unaltered (Fig. 3). Retinal thickness was measured at five distance points from both sides of the optic nerve in each eye, and average values were calculated for five eyes in each group. Thickness of the entire neuroretina (WT, 200 ± 6 μm; HPX^+/−, 201 ± 2.6; HPX^−/−, 195.5 ± 4.9; P = 0.64) and of the outer nuclear layer (WT, 51 ± 1.5 μm; HPX^+/−, 51 ± 1; HPX^−/−, 51.3 ± 2.3; P = 0.96) were similar across the strains. 

Figure 3.

 

Hematoxylin and eosin staining of frozen retinal sections from 2-month-old BALB/c, HPX^+/− and HPX^−/− mice. Similar retinal structure and thickness were observed across the mice strains. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium.

Expression of Iron Metabolism-Associated Genes

Retinal mRNA levels of genes involved in iron metabolism were measured in retinas of BALB/c, HPX^+/−, and HPX^−/− mice. Transferrin levels decreased in correlation to the number of mutant alleles: heterozygotes maintained 54% (P = 0.029) and homozygotes maintained 38% (P = 0.037) of Tf mRNA levels compared with WT (Fig. 4). Expression levels of ceruloplasmin and transferrin receptor did not significantly differ between HPX^+/− and WT mice. In HPX^−/− retinas, the level of ceruloplasmin mRNA was 63% (P = 0.047) and the level of transferrin receptor mRNA was 45% (P = 0.003) of that present in WT mice (Fig. 4). Expression levels of hephaestin, hemojuvelin, hemochromatosis gene, hepcidin, and transferrin receptor 2 were not significantly different between WT and HPX^+/− mice (hemochromatosis gene, P = 0.87; hemojuvelin, P = 0.30; hepcidin, P = 0.43; hephaestin, P = 0.24; transferrin receptor 2, P = 0.19; t-test) or between WT and HPX^−/− mice (hemochromatosis gene, P = 0.53; hemojuvelin, P = 0.33; hepcidin, P = 0.44; hephaestin, P = 0.58; transferrin receptor 2, P = 0.59; t-test; Supplementary Table S1). 

Figure 4.

 

Expression levels of transferrin, ceruloplasmin, and transferrin receptor measured by qPCR in 2-month-old BALB/c, HPX^+/−, and HPX^−/− mice. Results are presented as mean ± SEM. RQ, relative quantification (2^−ΔΔCt). *P < 0.05 compared with WT (n = 5 in each group).

Identification of Mutant Tf mRNA

The HPX mutation consists of a single nucleotide G→A switch in the splicing junction between exons 16 and 17 of the transferrin gene. As a result, the splicing takes place at an alternative site, 27 base pairs upstream, causing the formation of faulty mRNA. ¹⁹ Although the levels of Tf measured in the HPX^−/− retina were much lower than in the WT, they were considerably higher than transferrin protein levels reported in HPX^−/− circulation. ¹⁸ To determine whether the Tf mRNA formed by the HPX^−/− mouse retinas is mutant, we evaluated retinal Tf mRNA. PCR of cDNA from WT and HPX^+/− mice yielded the expected WT product size, whereas PCR from HPX^−/− mice yielded a smaller fragment compatible with the mutant Tf mRNA (Fig. 5). 

Figure 5.

 

Photograph of a gel showing results of RT-PCR for transferrin mRNA. The PCR product spans across the splice junction of exons 16 and 17. WT spliced products are 163 bp long, whereas the HPX mutation results in a 27-bp truncation, yielding a 136-bp product.

Transferrin Protein in the HPX Retina

The relative amount and spatial distribution of transferrin protein throughout the retina were assessed by semiquantitative immunohistochemistry (IHC). The staining quantified the supplemented human transferrin and the mutated transferrin produced in the mouse retina. Fluorescence intensity was measured on photographs captured using the same exposure parameters for all sections. Transferrin was abundant in the RPE, the inner segments of the photoreceptors, and the outer plexiform layer in the WT retina. Signal intensity was 85% (P = 0.3) in the HPX^+/− retina and 61% (P = 0.01) in the HPX^−/− retina compared with the WT retina. The majority of the signal in HPX^−/− retinas stems from the RPE and the inner segments of the photoreceptors (Fig. 6). 

Figure 6.

 

Fluorescent IHC for transferrin in BALB/c, HPX^+/−, and HPX^−/− retinas at 2 months of age. Red: positive staining. Negative control was photographed using the same imaging parameters and merged with (blue) DAPI nuclear stain for tissue orientation. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. Staining intensity was quantified for the total retina and separately for the RPE and inner segments. *P < 0.01 compared with WT (n = 5 in each group).

Total Iron Content

The effect of low transferrin levels on retinal iron was evaluated by measuring the total iron content of BALB/c, HPX^+/−, and HPX^−/− retinas. At 1 month of age, the mean retinal iron content was similar in all three groups of mice (WT, 51.14 ± 7.26 Fe_(ng)/retina; HPX^+/−, 52.56 ± 3.12; HPX^−/−, 45.95 ± 3.70; P = 0.54; n = 5 for each group). 

Oxidative Damage and Expression of Antioxidant Enzymes

Oxidative injury was measured by semiquantitative IHC for the lipid peroxidation product HNE. In HPX^+/− and HPX^−/− mice, the intensity of fluorescent staining was 22% lower than in WT mice (P = 0.06; Fig. 7). mRNA levels of the antioxidant enzymes glutathione peroxidase 1, heme oxygenase 1, and superoxide dismutase 1 were measured by qPCR to determine the degree of activation of the endogenous defense mechanisms against oxidative stress. Expression levels of GPX1 and SOD1 did not differ between heterozygote HPX and WT mice; Hmox1 was 1.3-fold higher in HPX^+/− retinas than in WT retinas (P = 0.025). In HPX^−/− mice, all three enzymes were significantly upregulated (GPX1, 1.9-fold, P = 0.007; Hmox1, 3.6-fold, P = 0.016; SOD1, 1.2-fold, P = 0.026; Fig. 8). 

Figure 7.

 

Fluorescent IHC for HNE (4-hydroxy-2-nonenal) in BALB/c, HPX^+/−, and HPX^−/− retinas at 2 months of age. Red: positive staining. Negative control was photographed using the same imaging parameters and merged with DAPI nuclear stain (blue) for tissue orientation. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium.

Figure 8.

 

mRNA levels of genes encoding antioxidant enzymes measured by qPCR. Results are presented as mean ± SEM. RQ, relative quantification (2^−ΔΔCt). *P < 0.03 compared with WT (n = 5 in each group).

Discussion

Transferrin is highly expressed in the normal retina, and its levels further rise during retinal and macular degeneration. ^8,11 Transferrin's role as a major component of the iron homeostasis system, combined with iron's potential for generation of oxidative injury, suggests that transferrin expression is important for the prevention of iron-associated retinal injury and the maintenance of normal retinal function. In accordance with this, Picard et al. ¹² recently showed that transferrin overexpression or systemic administration of transferrin can enhance photoreceptor survival in a mouse model of inherited retinal degeneration. This possibility is underscored by the existence of altered iron homeostasis during retinal and macular degeneration and by the involvement of oxidative injury in such diseases. ^{2,6 –8,10,25,26}  

The present study demonstrated altered retinal function in HPX^−/− mice that manifested as subnormal ERG responses despite weekly transferrin supplementation. Although such transferrin replacement treatment allowed short-term survival of the animals up to the age of 2 months and development of a retina with normal morphology and thickness, treated HPX^−/− mice remained anemic and demonstrated retinal pallor on ophthalmoscopy. 

The subnormal ERG responses in HPX^−/− mice may stem from transferrin deficiency or from hypoxia secondary to the severe anemia in these animals. The presence of normal retinal morphology suggests that retinal degeneration caused by transferrin deficiency did not underlie the altered ERG levels at these ages. On the other hand, the markedly suppressed OPs in HPX^−/− mice support the possibility that retinal ischemia is at play. OPs are thought to be generated by the amacrine cells and are sensitive indicators of retinal ischemia in several retinal pathologies, including diabetic retinopathy and vascular occlusions. ²⁷ In accordance with that, Burian described abnormal ERG recordings in anemic patients ²⁸ and speculated that the altered function may be secondary to relative anoxia caused by low hemoglobin levels. Thus, it is feasible that hypoxia rather than reduced transferrin levels underlie the subnormal ERG responses in HPX mice up to the age of 2 months. Whether reduced retinal transferrin levels or chronic hypoxia will result in retinal degeneration at older ages is unknown. 

Although we found that retinal Tf mRNA levels in HPX^−/− retinas are reduced to 38% of the levels present in WT retinas, Trenor et al. ¹⁹ reported that Tf mRNA levels in the liver of HPX^−/− mice are only 5% of normal values. The relatively higher retinal Tf mRNA levels that we report may be attributed to technical variations among the studies or to increased expression of Tf mRNA secondary to exogenous transferrin supplementation ¹² ; however, such differences can also reflect tissue-specific differences in Tf mRNA regulation that are not blunted by the HPX mutation. 

Semiquantitative IHC revealed reduced levels of transferrin protein in HPX^−/− retinas (61% of WT levels). Such transferrin immunostaining may label exogenous human transferrin and mutated mouse transferrin. Previous reports from HPX mice showed that transferrin levels in the systemic circulation, at the same time interval after transferrin supplementation that we evaluated, are reduced by >50% of WT levels in HPX^+/− mice, whereas homozygotes have <1% of WT circulating transferrin levels. ¹⁹ Raja et al. ²⁹ discovered that heterozygous HPX^+/− mice manifest a normal transferrin synthesis rate from the normal allele in association with slower transferrin degradation. On the other hand, transferrin in the HPX^−/− mouse is so sparse that it is immediately absorbed by the tissue, leading to an accelerated serum clearance rate. ²⁹ Furthermore, Dickinson et al. ³⁰ showed that human transferrin crosses the blood-brain barrier in HPX^−/− mice. ³⁰ Potentially, despite the fact that we tested the mice 1 week after the last transferrin supplementation, residual exogenous transferrin protein may still exist in the retina. 

We show that the retinal Tf mRNA formed by the HPX^−/− mouse is mutant. Transferrin protein encoded by such mutated mRNA is expected to be shorter than WT transferrin by nine amino acids located in proximity to the C terminus. This area of the protein is not involved in iron binding; consequently, its absence may not disrupt the function of transferrin as an iron carrier. However, to the best of our knowledge, the function of the mutated transferrin in this animal model has not been characterized in the literature thus far. 

The fact that iron overload occurs in some tissues in HPX mice implies that there must be other means for iron uptake that are not dependent on transferrin. ³¹ Accordingly, Takeda et al. ³² evaluated brains of neonatal HPX^−/− mice that were not treated with transferrin supplementation and found three times higher iron content than in nonmutant mice and abnormal iron distribution. Beard et al. ²⁰ demonstrated that the pattern of uptake and the location of iron in the adult HPX^−/− brain were similar to those of WT mice. In the latter study, mice were treated with transferrin injections. These data suggest that the production of normal levels of tissue transferrin is not essential for iron uptake, although it may play an important role in regulating the amount and distribution of iron in the brain. In our study, a similar phenomenon was observed in the retina. The iron content of HPX^−/− retinas from mice treated with weekly injections of transferrin was normal, suggesting that subnormal retinal transferrin levels as observed in the HPX^−/− mouse is sufficient to prevent retinal iron deficiency or overload. 

HPX^−/− mouse retinas exhibit subnormal transferrin receptor and ceruloplasmin mRNA levels. Expression of TfR is suppressed when the intracellular labile iron pool expands. ¹ Ceruloplasmin enables iron clearance from tissues as its oxidative activity transforms the ferrous ion (Fe²⁺) into the ferric form (Fe³⁺), enabling its upload onto transferrin. In the absence of this enzyme and its homolog hephaestin, iron overloading and retinal degeneration occur. ¹³ CNS ceruloplasmin (this protein cannot cross the blood-brain barrier) may also facilitate the release of iron from intracellular reservoirs, located within the supporting cells, for neuronal use. This may be an additional protective mechanism, allowing the nervous system to retain iron homeostasis independently of the systemic condition. ³³ In this context, the local decrease in Cp expression in the HPX^−/− mouse may limit the release of stored iron. Conceivably, the combination of exogenous transferrin supplementation, retinal production of mutated protein, and alteration in expression of transferrin receptor and ceruloplasmin is sufficient to maintain normal retinal iron levels. In accordance with that, mRNA levels of other genes associated with iron metabolism, including transferrin receptor 2, hephaestin, hemojuvelin, hemochromatosis gene, and hepcidin, were not significantly different between WT and HPX mice. 

Interestingly, staining for HNE, a product of lipid peroxidation, was less intense in HPX^+/− and HPX^−/− retinas than in WT retinas. Relatively low metabolic activity due to hypoxia caused by anemia may result in reduced production of HNE. Accordingly, it was shown that HPX^−/− mice are less susceptible than normal mice to metal particle exposure to the lungs. ³⁴ On the other hand, mRNA levels of antioxidant enzymes (gluthatione peroxidase 1, heme oxygenase 1, and superoxide dismutase 1) were elevated in HPX^−/− retinas, indicating the existence of oxidative stress to some extent. Increased expression of such enzymes in combination with the anemia in HPX^−/− mice may be sufficient to prevent oxidative injury, as evident in the HNE data. Of the three antioxidant genes tested, Hmox1 showed the most marked increase, which was also observed in HPX heterozygotes. This enzyme is responsible for the release of iron from molecules and heme proteins for recycling. In the absence of the Hmox1 enzyme, there is an attempt to improve iron transport to red blood cells by raising the amount of transferrin ³⁵ ; it is possible that transferrin deficiency in HPX mice results in the elevation of Hmox1 levels in an attempt to increase iron availability to the retina. It is also possible that increased expression of antioxidant genes is the result of retinal hypoxia in this mouse strain. 

Transferrin is one of the most highly expressed genes in the retina, a fact that underscores its important role in retinal iron metabolism. We show that HPX^−/− mouse retinas up to the age of 2 months are characterized by abnormal function and expression of some of the iron metabolism-associated genes in the presence of normal retinal morphology and iron levels. These facts reflect the success of the mechanisms involved in the regulation of retinal iron levels in maintaining its homeostasis; on the other hand, they demonstrate the scope of the alterations, which are associated with transferrin deficiency. 

Supplementary Materials

Figure sf01, PDF - Figure sf01, PDF 

Table st1, PDF - Table st1, PDF 

Footnotes

 Supported by grants from the Israel Science Foundation and by the Yedidut Research Grant.

Footnotes

 Disclosure: M. Lederman, None; A. Obolensky, None; M. Grunin, None; E. Banin, None; I. Chowers, None