While the past decade has seen great progress in mapping loci for common diseases, studying how these risk alleles lead to pathology remains a challenge. Age-related macular degeneration (AMD) affects 9 million older Americans, and is characterized by the loss of the retinal pigment epithelium (RPE). Although the closely linked genome-wide association studies ARMS2/HTRA1 genes, located at the chromosome 10q26 locus, are strongly associated with the risk of AMD, their downstream targets are unknown. Low population frequencies of risk alleles in tissue banks make it impractical to study their function in cells derived from autopsied tissue. Moreover, autopsy eyes from end-stage AMD patients, where age-related RPE atrophy and fibrosis are already present, cannot be used to determine how abnormal ARMS2/HTRA1 expression can initiate RPE pathology. Instead, induced pluripotent stem (iPS) cell-derived RPE from patients provides us with earlier stage AMD patient-specific cells and allows us to analyze the underlying mechanisms at this critical time point. An unbiased proteome screen of A2E-aged patient-specific iPS-derived RPE cell lines identified superoxide dismutase 2 (SOD2)-mediated antioxidative defense in the genetic allele's susceptibility of AMD. The AMD-associated risk haplotype (T-in/del-A) impairs the ability of the RPE to defend against aging-related oxidative stress. SOD2 defense is impaired in RPE homozygous for the risk haplotype (T-in/del-A; T-in/del-A), while the effect was less pronounced in RPE homozygous for the protective haplotype (G–Wt–G; G–Wt–G). ARMS2/HTRA1 risk alleles decrease SOD2 defense, making RPE more susceptible to oxidative damage and thereby contributing to AMD pathogenesis.
Age-related macular degeneration (AMD) is one of the most common irreversible causes of severe vision loss in individuals over the age of 55 (1). Despite intensive basic and clinical research, its pathogenesis remains unclear. Studies have shown that both genetic factors and environmental factors, such as persistent oxidative stress and smoking, are involved with the onset of AMD (2). Light exposure in combination with the photosensitizing capability of lipofuscin within the RPE makes the retina especially vulnerable to damage by reactive oxygen species (ROS) and by lipid-derived oxidative protein modifications (3). The resulting increase in oxidative stress to retinal pigment epithelium (RPE) cells elevates the risk of AMD. We hypothesize that antioxidant capacity is influenced by genetic factors.
Despite progress in mapping complex aging-related disease loci, determining how these alleles initiate pathology during aging remains a challenge. Genetic variants at two loci of chromosome 10q26 and 1q31 have been strongly associated with the risk of developing AMD. Genome-wide association studies (GWAS) and linkage studies have identified the Y402H variant in CFH, the rs10490924 single-nucleotide polymorphism (SNP) in ARMS2 (OMIM# 611313) and the rs11200638 SNP in HTRA1 (OMIM# 602194) as potential risk factors for AMD (Fig. 1). The genes ARMS2 and HTRA1 are located on chromosome 10q26 and are in such strong linkage disequilibrium (LD) that their contributions to disease susceptibility are indistinguishable using statistical analysis. The basic biological function of the ARMS2 (age-related maculopathy susceptibility 2) protein still remains unclear (4, 5), and the HTRA1 (high temperature requirement-A) protein is a serine protease; both are expressed in RPE cells. Their underlying molecular mechanisms in AMD pathology remain uncertain (6).
Previous studies of macular diseases utilize post-mortem tissue, which is problematic for several reasons. First, it is difficult to obtain a specific genotype; for example, only 0.5% of Caucasians are double homozygous for the CFH (402H) and ARMS2/HTRA1 (T-in/del-A) risk alleles for AMD and only 25% are double homozygous for the CFH (402Y) and ARMS2/HTRA1 (G–Wt–G) protective alleles. Such low population frequencies make it impractical to study the pathological aging mechanisms associated with these alleles using samples from eye banks. Second, when appropriate post-mortem AMD tissue can be obtained, it is almost always from late-stage donors. Lastly, post-mortem tissue is usually processed under suboptimal experimental conditions, which creates a myriad of research-related problems. Autopsy eyes from end-stage AMD patients, where age-related RPE atrophy and fibrosis are already present, cannot be used to determine how abnormal ARMS2/HTRA1 expression can initiate RPE pathology. To circumvent the human tissue shortage issue, reprogramming technology can be used to convert stem cells from patients homozygous for either the risk haplotype (T-in/del-A; T-in/del-A) or the protective haplotype (G–Wt–G; G–Wt–G) into differentiated retinal cells. Since our research is focused on understanding the molecular pathways that impact early-stage disease, we have differentiated induced pluripotent stem cells (iPSC) obtained from end-stage AMD donors into RPE (7). Using AMD patient-specific stem cells, we were able to model and study the disease.
We used patient-specific iPSC-derived RPE cell lines containing either or both the protective haplotype (G–Wt–G) and risk haplotype (T-in/del-A). To mimic RPE cell senescence, we allowed the cells to accumulate A2E, one of the lipofuscin fluorophores that accumulate in RPE cells with age. A substantial portion of the lipofuscin that accumulates with age in RPE cells consists of bisretinoids that form as a consequence of reactions of vitamin A aldehyde in photoreceptor outer segments. Photoreceptors shed apical portions of their light-transducing outer segments daily. The RPE cells play a critical function by phagocytosing the shed outer segments, thereby taking up the bisretinoid that has formed in photoreceptor cells. Upon blue light irradiation, A2E generates singlet oxygen, superoxide anion and hydrogen peroxide, agents that are widely believed to contribute to aging processes in human RPE cells (8). By introducing A2E accumulation as a component of the RPE aging process (9, 10), we sought to determine whether oxidative stress defense is associated with AMD gene susceptibility factors.
Reprogramming patient-specific fibroblasts and generation of patient-specific RPE cells
Fibroblasts were cultured from skin biopsies taken from two unaffected controls homozygous for protective haplotype and two high-risk AMD patient donors [one heterozygote (T-in/del-A; G–Wt–G), and one homozygous (T-in/del-A; T–in/del–A) for risk haplotype]. Colonies of iPS cells began to appear between 14 and 21 days after initiation of cellular reprogramming using lentiviral vectors expressing OCT4, SOX2, KLF4 and MYC. Purified RPE cells formed after 8–12 weeks of differentiation. Through light microscopy, hexagonal-shaped cells and dark perinuclear melanin granules typically found in naïve RPE were seen in iPSC-derived RPE cells, indicating that human iPSC-derived RPE cells were indistinguishable from primary cultures of human RPE (Fig. 2A). RPE-specific marker RPE65 and tight junction cell membrane protein ZO-1 immunofluorescence reactions on iPSC-derived RPE cell lines were positive (Fig. 2B). All iPSC-RPE cell lines expressed similar levels of BEST1, CRALBP, MITF and RPE65 (Fig. 2C). Furthermore, reverse transcription–PCR (RT–PCR) analysis confirmed positive expression of BEST1, RPE65 and MFRP in the iPSC-RPE cell lines. Skin biopsy fibroblast cells and iPS cells prior to differentiation were tested concurrently with the negative control. Autopsy-RPE cells' RT–PCRs are shown as the positive controls and GAPDH as the loading control (Fig. 2D). Because phagocytosis is one of the primary roles of RPE cells, we tested for phagocytic ability in the iPSC-RPE cell lines and obtained positive results (Fig. 2E). Additionally, we observed the apical microvilli and intracellular pigment by transmission electron microscopy (Fig. 2F). These tests laid a strong foundation for in vitro AMD RPE cell modeling using iPSC-RPE.
Model of aged RPE and autofluorescent imaging
After delivering A2E (10 μM in media) for 10 days, the accumulation of the fluorophore within the cells was evidenced by autofluorescent granules in light micrographs (Fig. 3A′) and lipofuscin deposits in electron micrographs (Fig. 3B′). Importantly, these A2E-aged iPSC-derived RPE cells resembled intact, aged (24-year-old) monkey RPE (Fig. 3C′) in that they both exhibited a high density of intracellular deposits of melanin and lipofuscin mixtures, which is a typical feature of aged RPE cells (11). On the other hand, cultured iPSC-derived RPE cells that were not aged by A2E exhibited negative autofluorescence (Fig. 3A) and a very low density of the mixture of lipofuscin and melanin deposits (Fig. 3B), similar to RPE in juvenile (1-year-old) monkeys (Fig. 3C).
Label-free mass spectrometric protein profiling
Shotgun proteomic mass spectrometry-based measurements were performed in triplicate on one homozygous for protective haplotype (wild-type) and one heterozygous iPSC-RPE cell line, with and without A2E. A liquid chromatography-mass spectrometry (LC-MS) approach was used for the relative quantitation and simultaneous identification of proteins within our samples. This included triplicate LC/MS/MS chromatograms (technical replicates) collected for each of the three biological replicates. Label-free shotgun proteomic profiling provided us with data to identify one or more key proteins from thousands of proteins and discern differences in protein expression phenotypes across cell lines with varying genotypes (12). Mitochondrial superoxide dismutase (SOD2) emerged as the most significant differentially expressed protein in RPE lines as it had a unique differential expression pattern (Supplementary Material, Table S1 and Supplementary Material, Results—Proteomics). SOD2 protein level was 37-fold higher in RPE with the homozygous protective haplotype after A2E treatment compared with untreated cells (P < 0.0004, false discovery rate correction for multiple testing). No other protein in this data set has this striking pattern of expression (i.e. high-fold expression in wild-type + A2E and negligible expression in other treatments/genotypes). The mean individual SOD2 protein abundance values for wild-type cells before and after A2E aging were 83 and 3129, respectively, and values for AMD high-risk cells before and after aging were 154 and 68, respectively. Moreover, the TransOmics software generated a mean peptide ion score of 9.4 (database match quality) for the eight peptides representing SOD2. This was well above the mean peptide ion quality score of 6.8 for this experiment (range of 3.5–14.6), strongly supporting the reproducibility of this identification. Peptide-integrated intensity volumes for extracted ion chromatograms formed the basis of protein abundance measurements. An example of this type of protein abundance measurement is given in a three-dimensional contour plot for one isotopic cluster (Fig. 4A) of a single SOD2 peptide highly expressed in wild-type RPE cells after A2E treatment, but not in A2E-treated high-risk RPE cells. For each protein abundance measurement, multiple peptides are measured in this manner. Hence, SOD2 emerged as a candidate based on the differential expression screening of RPE lines derived from patients' iPSC with different genotypes. Immunoblot analyses confirmed the results from the proteomic profiling: similar patterns were observed in SOD2 expression (Fig. 4B), with increased expression in cells homozygous for the protective haplotype after treatment (+A2E) and negligible changes in heterozygous cells after treatment (Fig. 4C).
SOD2 activity and antioxidant capacity test
We next examined SOD2 activity of all three genotypes (i.e. homozygous for protective haplotype, heterozygous and homozygous for risk haplotype) in iPS-RPE and autopsy-RPE cells, before and after A2E treatment; activity measurements were made using a colorimetric assay (Abcam, Cambridge, MA, USA) (13–16). SOD2 activity data for each type of A2E-aged cell was normalized to its corresponding non-aged counterpart (Fig. 5A) in order for us to examine how genotype affects the response of cultured RPE to A2E aging. SOD2 activity in RPE homozygous for the protective haplotype was dramatically increased in response to the aging treatment. In contrast, SOD2 activity in RPE from high-risk (heterozygous and homozygous for risk haplotype) AMD patients was decreased. This finding held regardless of whether the RPE came from human iPS cells or from post-mortem tissue-derived RPE cells.
We next tested the hypothesis that low SOD2 expression and activity, as observed in cells homozygous for risk haplotype, compromise antioxidant capacity (Fig. 5B). To do this, we cultured iPSC-RPE cells of the three different genotypes that were treated with A2E and exposed to blue light, or untreated as controls. Cell lysates were then tested in an oxygen radical absorbance capacity assay (STA-345, Cell Biolabs, San Diego, CA, USA). The assay works by the addition of peroxyl radicals (via a free-radical generator) followed by the measurement of resulting fluorescein decay over a 60 min period. The decay is slowed by SOD2, which neutralizes peroxyl radicals. After A2E treatment, the decay was found to be faster in AMD high-risk (heterozygous and homozygous for risk haplotype) samples, consistent with their decrease in SOD2 antioxidant capacity. We qualified the decay by calculating the area between the two decay curves (with and without A2E treatment), which represents the change in antioxidant capacity. After integrating the area between the two curves, we found that in the lysate from cells homozygous for the protective haplotype, there was a net increase in antioxidant capacity (+2.49) as a result of A2E treatment. Conversely, antioxidant capacity decreased in the lysates from cells heterozygous and homozygous for the risk haplotypes (−0.47 and −0.85, respectively).
Increased ROS/superoxide level in AMD risk haplotype RPE cells
Red fluorescent products of varying brightness levels, indicative of ROS, were generated across all samples using the Total ROS/Superoxide Detection Kit (ENZ-51010, Enzo, Farmingdale, NY, USA) (Fig. 6). In both autopsy-RPE and iPSC-derived RPE, we observed a trend that was congruous with the previous observation that superoxide levels were greatest in risk haplotype homozygotes, followed by heterozygotes and finally lowest in protective haplotype homozygotes. Quantiﬁcation of fluorescence density was measured using ImageJ (1.48e). The fluorescence values of autopsy-RPE cells were: 44.86 (G/G), 161.66 (T/G) and 217 (T/T). The fluorescence values for iPSC-RPE cells were: 66.1 (G/G), 113.7 (T/G) and 197.6 (T/T). There was a statistically significant difference in fluorescence values between the samples homozygous for the protective haplotype and homozygous for the risk haplotype (iPSC-RPE: P < 0.01, t = 6.4341, unpaired t-test; autopsy-RPE: P < 0.05, t = 3.1209, unpaired t-test). These results showed that in the presence of ROS inducers, the levels of ROS/superoxide exhibited are related to the risk factor genotype, most likely due to differential SOD2 activity among the various cell genotypes.
Oxidative damage of RPE and disruption of β-catenin
Forkhead Box O transcription factors (FOXO) are a family of ROS level sensors (17). Increased ratios of phospho-FOXO3A (FOXO3A phosphorylated on Thr-32) to total FOXO3A after A2E treatment with blue light exposure were found in samples homozygous for the risk haplotypes in both iPSC-derived RPE cell lines and autopsy human cell lines (Fig. 7B and C), but not in homozygous for protective haplotype cells lines after identical A2E treatment (autopsy-RPE: P < 0.01, t = 6.2104, unpaired t-test; iPSC-RPE: P < 0.05, t = 3.6924, unpaired t-test). Additionally, β-catenin expression levels decreased after A2E treatment in cells homozygous for risk haplotype (autopsy-RPE: P < 0.01, t = 6.6568, unpaired t-test; iPSC-RPE: P < 0.05, t = 4.5795, unpaired t-test). Immunostaining for the adherence junction protein β-catenin, which maintains the barrier integrity of the RPE, showed the disruption after A2E treatment with blue light exposure, demonstrating movement of β-catenin from the periphery to the inside of RPE cells (Fig. 7A). Overall, the results show that RPE cells homozygous for risk haplotype, with decreased SOD2 activity and higher ROS levels, have increased FOXO3A but decreased β-catenin levels in response to A2E treatment.
AMD is a leading cause of loss of independence in activities of daily living for older adults. Nine million Americans are diagnosed with AMD, and the incidence is expected to double in the next decade—eventually affecting 20% of Americans between the ages of 65 and 75, and 30% over 75 (18). There is no effective treatment and the greatest barrier to developing treatments is the lack of knowledge about the molecular basis of disease pathogenesis. iPS cells reprogrammed from somatic cells have allowed for the generation of patient-specific disease cells in vitro. Interest in generating human iPS cells for stem cell modeling of disease has overtaken that for patient-specific human embryonic stem cells due to the latter's ethical, technical and political concerns. By providing a platform to study patient-specific targeted disease cells on a research bench, iPS cells have great potential in regenerative medicine and modeling of human disease (19, 20). Studies of the in vitro phenotypes of disease-specific iPSC-derived cells can aid in filling the gap between clinical phenotypes and molecular or cellular mechanisms, with further applications including new strategies for drug screening and the development of novel therapeutic agents. Because an AMD-associated risk haplotype at the ARMS2/HTRA1 locus is known to be a risk factor for AMD, we created specifically genotyped AMD disease models using iPSC technology for use as valuable tools in elucidating its molecular pathways. In conjunction with iPSC-RPE, eye bank autopsy-RPE, genotyped to have AMD high-risk and protective AMD alleles, served as experimental controls in various assays.
RPE cells generated from human iPS cells have been shown to exhibit an RPE fate (7, 21). To recapitulate the phenotype of a late-onset degenerative disease such as AMD, we accelerated the aging process in iPSC-derived and autopsy-RPE cell lines using a combination of 10 μM A2E and blue light, which provides us with an ideal system for modeling AMD (22–25).
A2E creates oxidative stress by light-mediated generation of singlet oxygen and superoxide (22, 26). An excess of reactive forms of oxygen results from either their overproduction or insufficient activity of antioxidant defense systems. A2E accumulates in RPE cells in every individual and yet only certain patients—those with risk haplotypes—develop AMD. Therefore, we hypothesized that RPE with the AMD-associated risk haplotype (T-in/del-A) should have reduced antioxidant capacity, whereas wild-type RPE with the protective haplotype have a greater antioxidant protective capacity. In our RPE cells modeling AMD, antioxidant defense ability was tested by the activity of mitochondrial SOD, which is localized in mitochondria and protects the cells against oxidative stress. Because impaired SOD2 stress responses in RPE occur in cells containing the risk haplotype (Fig. 5), we conclude that oxidative stress defense is associated with GWAS risk factors.
Low SOD2 activity in high-risk RPE cells may lead to increased susceptibility to the development of AMD. In our unbiased proteome screen of high-risk AMD cells, SOD2 was identified as one of the downstream targets of ARMS2/HTRA1 pathology (Supplementary Material, Fig. S1). Plasma SOD activity has been shown to be lower in neovascular AMD patients than in controls (13). Previous studies have also been performed to test reactions of SOD2-deficient animals to oxidative injury; RPE and retinal cells in SOD2-deficient mice showed greater instances of morphologic abnormalities, including features of AMD, compared with wild-type controls (15, 16, 27). We showed that SOD2 activity in RPE from patients homozygous for the protective haplotype (G-Wt-G; G-Wt-G) increased in response to the A2E aging treatment. In contrast, SOD2 activity in RPE from high-risk AMD patients (both homozygous for risk haplotype and heterozygous) was significantly decreased. We also showed that decreased SOD2 activity corresponded to diminished antioxidant capacity in both high-risk RPE lysates, as verified by the oxygen radical absorbance capacity test. These results were not limited to A2E oxidant stress, as superoxide level responses to the ROS inducer pyocyanin showed the same trend in both autopsy-RPE and iPSC-RPE, with cells homozygous for the risk haplotype having the highest superoxide levels.
Diminished SOD2 protective responses in high-risk RPE cells may also be linked to the role of the FOXO family of transcription factors in regulating ROS stress. The FOXO family is a key component in sensing and responding to ROS levels, as evidenced by an increase in phosphorylated FOXO3A under conditions of oxidative stress (28, 29). One function of FOXO is to regulate pro-apoptotic genes; studies have shown that a FOXO3a mutant suppresses ROS-induced cell death (30). In our study, RPE homozygous for protective haplotype all showed negligible changes in ratio levels of phospho-FOXO3A to FOXO3A expression after A2E treatment, while RPE homozygous for risk haplotype exhibited increased ratios after treatment. Our data show that decreased SOD2 activity is accompanied by increased phospho-FOXO3A to FOXO3A ratios, confirming that the ROS stress levels are indeed closely linked to the genotype mediated by SOD2 antioxidant activity.
The FOXO family of transcription factors inhibits β-catenin-mediated antioxidative defense (31), which is consistent with our findings. The phospho-FOXO3A protein levels increased in RPE cells homozygous for the risk haplotype and coincided with greater decreases in β-catenin levels. Since β-catenin is an adherence junction protein that plays an important role in the integrity of the RPE cell layer, decreased levels of β-catenin resulting from oxidative stress disrupts RPE junctions (32, 33), which has a stronger negative effect on cell barrier function and may be a mechanism leading to RPE detachment and subsequent choroidal neovascularization (33) (Fig. 8).
Dissection of the pathways affected by ARMS2/HTRA1 alleles in iPSC-derived RPE should be considered when assessing molecular targets for AMD intervention. The AMD-associated risk haplotype impaired the protective SOD2 activity under conditions of oxidative stress generated by A2E. Since there were no statistical differences between the HTRA1 expression in AMD cells homozygous for risk haplotype and RPE homozygous for protective haplotype (Supplementary Material, Fig. S2), the biological relationship between genotype and SOD2 activity was not the main focus of this report.
Although the precise roles of HTRA1 and ARMS2 in AMD pathogenesis are unknown, our data suggest that the wild-type ‘protective haplotype’ can decrease oxidative stress by means of increased SOD2 activity, as evidenced by five observations of ‘protected’ cells after A2E treatment: (i) higher SOD2 expression levels; (ii) greater SOD2 activity; (iii) lower ROS levels; (iv) no increase in the ratio of phospho-FOXO3A to total FOXO3A; and (v) a smaller decrease in protective β-catenin levels. The differential antioxidant capacities among haplotypes of each of the three chromosome 10q26 locus variants can account for genetic susceptibility to AMD. In this manner, therapeutic strategies aimed at potentiating SOD2 activity may be a reasonable approach to reducing the influence of the AMD-associated variant in ARMS2/HTRA1 on the development of AMD.
MATERIALS AND METHODS
AMD-afflicted and unaffected individuals were enrolled in the study under protocol #IRB-AAAF1849 after we obtained full consent. The Institutional Review Board at Columbia University approved the protocol and this study conformed to the tenets of the Declaration of Helsinki. Each patient received a complete clinical evaluation by the corresponding author.
Skin biopsy samples were obtained from patients and healthy subject using lidocaine anesthesia (APP Fresenius Kabi, Germany) and biopsy-punch (McKesson, VA, USA). Samples were processed and cultured according to protocol (7).
iPSC-derived RPE cell lines were obtained from wild-type (unaffected), heterozygous high-risk (affected) AMD patients and homozygous high-risk (affected) AMD patients fibroblasts. The homozygous high-risk fibroblasts were donated by the New York Stem Cell Foundation Laboratory. We used lentiviral vectors carrying transcription factors OCT4, SOX2, KLF4 and MYC to transduce fibroblasts into iPS cell lines in accordance with established protocols using human embryonic stem cell medium (HUESM) with 10 μM basic fibroblast growth factor (bFGF). iPS cells were co-cultured with mitomycin-C-treated stromal cells from the PA6 line. Next, we further incubated the cells under 5% CO2 at 37°C in differentiation medium (HUESM–bFGF with 10 nm Nicotinamide) from day 0 to day 20 and 20 ng/ml Activin A from day 20 to day 40. The media were replaced twice a week. Upon pigmentation of the cells, we chose pigmented colonies to be replated on Matrigel-coated plate in RPE culture medium as per protocol.
Autopsy-RPE cell cultures were generated using human eye samples with the anterior portions removed obtained from the Eye-Bank of New York. Following the RPE cell culture procedure developed by Maminishkis and Miller (7), we first incubated the eye cup at 37°C with 5% CO2 for 40–60 min. Next, the retina was removed from the optic nerve with retinal scissors, and forceps were used to separate RPE-Bruch's membrane from the choroidal tissue layer. RPE sheets were then placed into cold trypsin–EDTA (Gibco) solution in a 15 ml tube. Upon collection of RPE cells, the tissue-containing tubes were transferred into a water bath for 10–15 min at 37°C in order to separate the RPE into small clusters. Using a clinical centrifuge, we spun down the RPE cells at 1.4 rpm (Rotor radius 15 cm) for 4 min, removed the supernatant and re-suspended the cells in 15% RPE media (9 ml total). Three milliliters of cell suspension were placed into Primeria flasks, to which 2 ml of fresh 15% RPE media was then added. The flask was incubated at 37°C, 5% CO2 until the following day.
To confirm the RPE protein marker expression on our iPSC-RPE, immunoblot analyses were conducted before A2E treatment on one protective and all high-risk iPSC-RPE cell lines. Extracted total cellular protein was analyzed using sodium dodecyl sulfate–polyacrylamide gel electrophoresis. After transferring the samples to nitrocellulose membranes, we incubated them with rabbit anti-RPE65 monoclonal antibody (1:1000; a gift from T. Michael Redmond, National Institutes of Health), mouse anti-MiTF antibody, rabbit anti-Bestrophin 1 antibody (1:1000; Abcam) and rabbit anti-CRALBP (1:1000; Santa Cruz Biotechnology, Santa Cruz, CA, USA). A chemiluminescence assay (Immobilon Western; EMD Millipore, Billerica, MA, USA) was used to detect blots, whereupon multiple exposures were taken using Kodak BioMax film (Kodak, New York, NY, USA) and developed with a Konica SRX-101A medical film processor (Konica Minolta Medical Imaging USA Inc., Wayne, NJ, USA). After A2E treatment and blue light exposure, immunoblot analysis with anti-SOD2 antibody (1:1000; Abcam) was performed to confirm the results of the label-free mass spectrometric protein profiling test. Anti-FOXO3a (phosphorylated on T32) and anti-β-catenin were also tested on samples with and without treatment (1:1000; Abcam).
RNA was isolated from one protective and all high-risk iPSC-RPE cell lines using the RNeasy kit (QIAGEN, Hilden, Germany), and genomic DNA contamination was removed using a DNA-free kit (Invitrogen). One microgram of total RNA was used for cDNA synthesis using Superscript III Reverse Transcriptase and Oligo (dT) primers (Invitrogen). The quality of cDNA synthesis was assessed by PCR amplification of GAPDH, as a housekeeping gene, using a combination of forward (5′-ATCACCATCTTCCAGGAGCG-3′) and reverse (5′-TGATGACCCTTTTGGCTCCC-3′) primers. The RPE65 and bestrophin-1 PCR were then performed using primer sets Hs01071462_m1 for RPE65 (4331182) and Hs00188249_m1 for bestrophin-1 (4331182) (Applied Biosystems; Life Technologies). The MFRP test used a combination of forward (5′-CAAGATCGAAGCCCTCAGCA-3′) and reverse (5′-GGTACCAGGCATGGAAACCA-3′) primers.
iPSC-derived RPE cells were placed in 4% paraformaldehyde (Sigma) in PBS at 21°C for 2 h. We performed anti-RPE65, anti-ZO-1 immunofluorescence staining (1:100) in order to characterize the iPSC-derived RPE generated from the fibroblasts taken from AMD patients. The nuclei were stained with DAPI (4′, 6-diamidino-2-phenylindole). Alexa Fluor 488 goat anti-rabbit or Alexa Fluor 555 goat anti-mouse IgG were conjugated by secondary antibodies (1:1000; Invitrogen; Life Technologies). We obtained images for all antibody labels using a fluorescence microscope (Leica DM 5000 B) under the same settings. A2E-treated RPE were maintained in RPE medium for 20 days before β-catenin immunofluorescence studies were performed. Confocal images (Zeiss LSM510) of β-catenin-stained autopsy human RPE cells were taken before and after A2E treatment.
A2E (C42H58NO, molecular weight 592) was synthesized from all-trans-retinal (ATR) and ethanolamine combined in a 2:1 ratio (34–36). iPSC-derived RPE lines and autopsy-RPE were seeded in 3.5 cm dishes and 4-well chamber slides. Then, they were cultured for 10 days in the presence of 10 μM A2E. Where indicated, the cells were exposed for 10 min to 430 nm (blue) light. The light source was calibrated with a power meter (Scientech, Boulder, CO, USA). We determined the cellular autofluorescent levels using stereomicroscopy (Zeiss LSM 510 META) (23, 24, 26), and A2E levels in cultured RPE cells by HPLC (37). The same seeding density was used for all cell lines and cells not treated with A2E served as negative controls.
iPSC-derived RPE cells were seeded at 24-well transwells pre-coated with poly-L-Lysine. After 10 days of A2E treatment, we performed electron microscopy according to established protocols. Prior to observation by electron microscopy, areas of interest were trimmed for ultrathin sectioning and stained with uranyl acetate. A2E-treated and untreated cells' EM morphologies were compared with each other and with 1- and 24-year-old monkey RPE cells (gifts from Professor Peter Gouras).
Label-free mass spectrometric protein profiling
We extracted total proteins from one G/G genotype at rs10490924 (homozygous for protective haplotype) iPSC-derived RPE cell line and one T/G genotype at rs10490924 (heterozygous) iPSC-derived RPE cell line, with and without A2E treatments. For each of these cell lines, three biological replicates were prepared representing three separate cultures derived from each cell line, and replicates were also done separately for A2E-treated samples. Proteins were extracted with TRIS-buffered saline with sodium dodecyl sulfate as previously described (38). The proteins were precipitated with chloroform–methanol and dissolved in 0.1% RapiGest detergent in 50 mM ammonium bicarbonate. All protein extracts were reduced and alkylated prior to tryptic digestion, and RapiGest was cleaved with acid. The resulting peptides were analyzed with Synapt G2 quadrupole-time-of-flight mass spectrometer (Waters Corp.) using MSE data-independent scanning (38). Initial data processing was with ProteinLynx Global Server (Version 2.5 RC9, Waters Corp.) for quality assurance testing of samples and verifying instrument performance as described (38). Further analysis was with TransOmics software (Waters Corp.). The data processing method is described in the Supplementary Material, Methods—Proteomics section.
SOD2 activity and antioxidant capacity test
SOD2 activity measurements in high-risk homozygous (T/T), heterozygous (T/G) and protective (G/G) human autopsy-RPE and iPSC-RPE cells, with and without A2E treatment, were made using a colorimetric assay (Abcam). The measurements were based on inhibition of xanthine oxidase (XO) activity, which is linearly related to superoxide anion concentration. SOD2 activity was measured in triplicate cultures for each of the six cell types. The triplicate measures were averaged and then normalized (A2E-aged cells were normalized to non-aged cells). Antioxidant capacity was determined by the oxygen radical absorbance capacity assay (STA-345, Cell Biolabs). The assay works by adding peroxyl radicals (via a free-radical generator) and then measuring the resulting fluorescein decay over a 60 min period.
ROS/superoxide level test
To observe the effect of risk factor genotype on ROS/superoxide levels in response to oxidative stressors other than A2E, we prepared A2E-untreated RPE samples for fluorescence microscopy using the Total ROS/Superoxide Detection Kit (ENZ-51010, Enzo, Farmingdale, NY, USA). The samples were classified by AMD allele risk level (i.e. homozygote G/G protective, heterozygote T/G high-risk and homozygote T/T high-risk), and by origin (i.e. iPSC-derived and autopsy). First, RPE cells cultured on glass slides were loaded with 30 nm Superoxide Detection Reagent, a cell-permeable probe that reacts with superoxide to produce a fluorescent product. Next, we treated the cells with 200 μM ROS inducer (Pyocyanin) for 20 min, and then washed the cells in accordance with protocol. Using fluorescence microscopy, we visualized the ROS/superoxide levels in the cells using 80 ms exposure. Quantiﬁcation of the density of fluorescence was measured by software ImageJ (1.48e).
J.Y., J.R.S. and S.H.T. designed research; J.Y., Y.L. and D.E. contributed to iPSC-RPE cell differentiation; J.Y. and W.-H.W. collected autopsy-RPE; J.Y., Y.-T.T., L.M.B., C.-W.H and W.-H.W. performed research; J.R.S. provided A2E treatment; J.Y., X.L. and S.H.T. analyzed data; J.Y., L.C., H.V.N., J.R.S. and S.H.T. wrote the paper.
This work was supported by the National Institute of Health Core (5P30EY019007), National Cancer Institute Core (5P30CA013696) and unrestricted funds from Research to Prevent Blindness, New York, NY, USA. S.H.T. is a member of the RD-CURE Consortium and is supported by the Tistou and Charlotte Kerstan Foundation, the National Institute of Health (R01EY018213), the Research to Prevent Blindness Physician-Scientist Award, Association for Research in Vision and Ophthalmology (ARVO) Foundation, Macular Society, Retina Research Foundation Cox Macular Research Project, the Bernard and Shirlee Brown Family Fund, the Schneeweiss Stem Cell Fund, New York State (N09G-302), the Foundation Fighting Blindness New York Regional Research Center Grant (C-NY05-0705-0312), the Joel Hoffman Fund, the Professor Gertrude Rothschild Stem Cell Foundation and the Gebroe Family Foundation. H.V.N. is supported by the Research to Prevent Blindness Medical Student Fellowship.
We thank all the individuals and their families who donated their skin biopsies and eyes for the study. We thank Peter Gouras and members of their laboratories for sharing ideas, histological specimens from monkeys, antisera, equipment and for critically reading the manuscript. We thank Katherine J. Wert, Richard Davis, Deniz Erol and LiJuan Zhang for guidance and advice; and members of the Barbara and Donald Jonas Laboratory for support, especially our financial supporters.
Conflict of Interest statement. The authors declare no competing financial interests.
- oxidative stress
- stem cells
- gene frequency
- tissue banks
- age-related macular degeneration
- superoxide dismutase
- genome-wide association study
- htra1 gene
- retinal pigment epithelium
- retinal pigment cell