Survival of Retinal Pigment Epithelium after Exposure to Prolonged Oxidative Injury: A Detailed Gene Expression and Cellular Analysis

Abstract

purpose. To detail, by DNA microarrays and cellular structure labeling, the in vitro responses of retinal pigment epithelial (RPE) cells to a nonlethal dose of the oxidant agent hydroquinone (HQ).

methods. The viability of growth-quiescent ARPE-19 cells after treatment with HQ was measured by XTT conversion, ³H-leucine incorporation, trypan blue exclusion, and the presence of DNA laddering. The effect of a nonlethal dose of HQ on the localization of apoptosis-induced factor (AIF) and phosphorylation of stress-activated kinase-2/p38 (SAPK2/p38) was detected by immunocytochemistry. Actin structures were visualized by phalloidin staining. Cell membrane blebbing was detected using GFP-membrane–labeled RPE cells (ARPE-GFP-c’-rRas). Changes in gene expression patterns of RPE cells within 48 hours of prolonged treatment with a nonlethal dose of HQ were evaluated by microarray analysis and confirmed by Northern blotting.

results. The viability of RPE after a prolonged sublethal injury dose of HQ was determined by multiple assays and confirmed by the absence of AIF translocation or DNA laddering. Prolonged exposure (16 hours) of RPE cells to a nonlethal dose of HQ resulted in actin rearrangement into globular aggregates and cell membrane blebbing. Kinetic microarray analysis at several time points over a 48-hour recovery period revealed significant upregulation of genes involved in ameliorating the oxidative stress, chaperone proteins, anti-apoptotic factors, and DNA repair factors, and downregulation of pro-apoptotic genes. Genes involved in extracellular matrix functions were also dysregulated. Recovery of RPE cells after the injury was confirmed by the normalization of gene expression dysregulation back to baseline levels within 48 hours.

conclusions. RPE cells avoided cell death and recovered from prolonged oxidative injury by activating a host of defense mechanisms while simultaneously triggering genes and cellular responses that may be involved in RPE disease development.

Age-related macular degeneration (AMD) is the most common cause of vision loss in patients over the age of 60 years. Although the pathophysiology of AMD is not well understood, there is growing evidence that cumulative oxidative injury caused by reactive oxygen intermediates (ROI) could play an important role in the development of AMD. One of the current hypotheses on early AMD development focuses on oxidative changes affecting the retinal pigment epithelial (RPE) cells, a nonrenewable multifunctional cell layer of the retina. RPE cells are subjected to a particularly high level of oxidative stress from several sources, including prolonged exposure to visible light, contact with a high oxygen environment, phagocytosis, and degradation of outer segments of polyunsaturated fatty acid–rich photoreceptors. ^{1 2 3} In addition, accumulation of lipofuscin, a product of lysosomal degradation, increases in aged RPE cells and can contribute to the blue light-induced generation of ROIs and oxidative damage. ^{4 5} Cumulative prolonged oxidative damage at the RPE level may contribute to the development of certain anatomic changes characteristic of early AMD, which include pigmentary abnormalities of the RPE and the formation of subRPE extracellular deposits. ^{6 7}  

Whereas most studies of the relationship between oxidative stress, RPE cells, and AMD have focused on the elucidation of mechanisms of oxidative stress inducing apoptosis, ^{8 9 10 11} widespread RPE cell death is not generally seen in early AMD. Epidemiologic data has demonstrated that RPE cell death, termed geographic atrophy, occurs in only 10–15% of AMD patients and only at the latter stages of the disease. ¹²  

RPE cells are able to tolerate oxidative stress without initiation of cell death. ¹³ These cells exhibit a distinct set of physiological responses, including actin rearrangement and cell membrane blebbing when subjected to a nonlethal oxidative injury. ¹⁴ At the same time prolonged oxidative injury can affect the integrity and function of RPE tight junctions, potentially leading to the disruption of the blood–retinal barrier. ¹⁵ Therefore, it may be possible that the response of the RPE to a prolonged nonlethal oxidative injury could be associated with the pathophysiology of early AMD. 

The exact mechanisms conferring the ability of RPE cells to recover from an oxidative injury without activation of cell death are largely unknown. Antioxidative genes are upregulated both in RPE cells ^{16 17} and other cell types ^{18 19 20 21} when exposed to oxidative stimuli, diminishing the accompanying oxidative cell damage; yet alterations in genes directly involved in cell death pathways have not been shown. 

In the present study, the cellular and molecular responses of RPE cells after exposure to prolonged nonlethal oxidative injury were examined. Viability of the RPE cells after oxidative injury was confirmed by the presence of trypan blue exclusion and XTT conversion, and absence of apoptosis-induced factor (AIF) translocation, DNA laddering, or changes in mitochondrial potential. Cellular responses to the oxidative injury included an increase in protein synthesis, actin rearrangement, and activation of SAPK2/p38 pathways. In addition, oxidative stress induced specific patterns of gene dysregulation, as determined by microarray and confirmed by Northern blot analysis, which resulted in an active avoidance of cell death. This response included upregulation of injury-protective genes and downregulation of pro-apoptotic genes. Interestingly, several cellular and molecular responses, hypothesized to be involved in AMD development, were also detected, including cell membrane blebbing and changes in extracellular matrix proteins (ECM) expression. These results indicated that prolonged treatment with nonlethal oxidative stimulus induces an active survival response in RPE cells and activates genes that may be involved in RPE disease development. 

Materials and Methods

Materials

All cell culture reagents were purchased from Gibco-BRL (Gaithersburg, MD). Trans-retinoic acid, tetramethyl-rhodamine isothiocyanate (TRITC) conjugated phalloidin, and hydroquinone (HQ; 1.4 benzenediol) were purchased from Sigma (St. Louis, MO). Microarray filter membranes were a kind gift of Kevin Becker (DNA Array Unit, National Institute of Aging, Baltimore, MD). ARPE-19 cells were kindly provided by Leonard Hjelmeland (University of California, Davis, CA). 

Cell Culture

ARPE-19 cells were maintained in Dulbecco’s modified Eagle’s medium-DMEM/F12 supplemented with 10% fetal bovine serum (FBS) and containing 100 μ/mL each of penicillin and streptomycin. Cells were used for the experiments between passages 10 and 20. For all experiments, ARPE-19 cells were allowed to grow to confluence and to differentiate in DMEM/F12 media supplemented with 1% FBS and 25 nM trans-retinoic acid without antibiotics for a minimum of 21 days. For the microarray experiments, ARPE-19 cells were plated on growth factor (GF)-reduced matrigel (1:16 dilution in DMEM; BD Biosciences, Palo Alto, CA) at confluence and growth quiescence for a minimum of 21 days. For treatment with HQ, growth-quiescent cells were exposed to various concentrations of HQ in serum-free media for 16 hours, followed by washing and replacement with 1% FBS media. Untreated control cells were handled in a similar fashion save for the addition of serum-free media without HQ. 

Apoptosis-Induced Factor/Phosphorylated p38 Immunohistochemistry

Cells were fixed with methanol:acetone 50:50 (vol/vol) for 10 minutes at −20°C and blocked with 10% normal donkey serum containing 0.6% Triton X-100 in phosphate buffer (PBS; pH 7.4) for 1 hour at room temperature. Slides were incubated with either anti-AIF (Santa Cruz Biotechnologies, Santa Cruz, CA) or anti-phosphorylated p38 (Cell Signaling, Beverly, MA) antibody and detected with a FITC-conjugated secondary antibody (Jackson Laboratories, Burlingame, CA). The cells were mounted with 4′,6-diamidino-2-phenylindole (DAPI) containing anti-fade mounting media (Vectashield; Vector Laboratories, Inc., Burlingame, CA) and viewed with an epi-fluorescent microscope (Olympus Optical Co., Ltd, Melville, NY). Images were digitally captured using OpenLab software (Improvision, Lexington, MA) and recompiled in Adobe Photoshop (Adobe System, San Jose, CA). Negative controls were incubated with secondary antibody without primary. 

RPE Cell Membrane Blebbing

ARPE-19 cells stably expressing green fluorescent protein (GFP) targeted to the inner leaflet of the plasma membrane (ARPE-GFP-c’-rRas) were generated as described, ¹⁴ and 2 × 10⁴ cells were seeded onto collagen-coated 8-well chamber slides (LAB-TEK, Naperville, IL). The cells were treated with various concentrations of HQ, fixed in 4% paraformaldehyde, and examined under a dual-channel laser scanning confocal microscope (Leica, Exton, PA). The percentage of cells with membrane blebbing and AIF translocation was determined by the average of three fields per slide in which approximately 150 GFP positive cells were counted by epi-fluorescent microscopy (Olympus Optical Co., Ltd). 

Visualization of F-Actin

Treated or untreated growth-quiescent ARPE-19 cells grown on collagen-coated slides were fixed with 4% paraformaldehyde and permeabilized with 0.1% saponin. Changes in F-actin structures were detected by incubating the cells for 30 minutes at RT with 0.33 μM TRITC-labeled phalloidin, mounting with DAPI containing anti-fade mounting media (Vectashield, Vector Laboratories, Inc.) and examining the cells under a dual-channel laser scanning confocal microscope (Leica). 

XTT Assay

ARPE-19 cells were seeded in 96-well collagen-coated microculture plates and treated with varying concentrations of HQ for 16 hours, followed by a 24-hour recovery period in 1% media. At the 40-hour point, the number of viable cells was then determined by the addition of XTT- (sodium 3′-[phenylaminocarbonil)-3; 4-tetrasodium]-bis (4-methoxy-6-nitro) benzene sulfonic acid hydrate) for 24 hours using the manufacturer’s instructions (Cell Proliferation Kit II; Roche Molecular Biochemicals, Indianapolis, IN). 

Incorporation of ³H-Amino Acids

After treatment with varying concentrations of HQ for 16 hours, followed by a 24-hour recovery period in 1% media, protein synthesis in ARPE-19 cells was measured by incorporation of labeled amino acids as previously described. ²² In brief, cells, grown on 6-well collagen-coated dishes, were incubated continuously without refeeding for either 24 or 48 hours in serum-free DMEM/F12 medium containing 2 mM l-methionine and 2 μCi/mL l-methionine [methyl ³H], 2 mM l-tyrosine and 2 μCi/mL l-tyrosine [ring-3; 5-³H], and 2 mM of L-leucine and 2 μCi/mL of L-leucine [3; 4; 5-³H(N)] (Amersham Bioscience, Piscataway, NJ). Cells were washed twice with ice-cold PBS and three times with 5% ice-cold trichloroacetic acid. Proteins were then solubilized in 0.5 N NaOH for 15 minutes and neutralized with equal amounts of 0.5 N HCl. The uptake of ³H was determined by scintillation counting (Packard, Albertville, MN). In parallel experiments, equal cell counts in each well of the plate after HQ treatment was confirmed (data not shown). 

RPE Cell Proliferation Assay

ARPE-19 cells were seeded in 96-well collagen-coated flat bottom microculture plates in a total volume of 200 μL of 1% serum DMEM-F12 media. Cells were treated in serum-free conditions with varying concentrations of HQ for 16 hours, followed by incubation with [³H]thymidine (2.5 μCi/ 50 μL per well; Amersham) for an additional 48 hours. Cells were then washed three times with ice-cold PBS, trypsinized, and incorporated radioactivity was measured by a scintillation counter. Data were expressed as percent of relative [³H]thymidine incorporation in treated cells compared with the control. 

Trypan Blue Exclusion

After HQ treatment, cells were removed from the plate with 0.025% trypsin-EDTA solution and suspended in PBS. Cells were stained with 0.4% trypan blue for 30 minutes and counted using a Bright line cell counting chamber (Hausser Scientific, Horsham, PA). 

Northern Blot Analysis

Total RNA was isolated from cells using the Atlas Pure Total RNA Labeling System (Clontech, Palo Alto, CA). Total RNA (20 μg) was subjected to electrophoresis on formaldehyde denaturing gels and transferred to a 0.45-μm Nytran membrane (Schleicher and Schuell, Keene, NH). After cross-linking, membranes were probed with ³²P-labeled full length cDNA probes of TRX1, PRDX1, TXNRD1, or GADPH, and washed in 0.2X SSC, 0.1% SDS, 65°C for 30 minutes. Filters were then exposed to Kodak X-OMAT AR film (Eastman Kodak, Rochester, NY) at −80°C and developed. 

Pulsed-field DNA Gel-Electrophoresis

Genomic DNA was extracted according to procedures recommended by the manufacturer (Genomic DNA Preparation Kit, Promega, Madison, WI). For detecting oligonucleosomal DNA, DNA was fractionated on a 1.2% agarose/EtBr gel using a Bio-Rad powerPAC/1000 (constant voltage 100 V, 1.5 hours). For the detection of large-scale DNA fragments, the DNA sample was analyzed in a Bio-Rad CHEF-DR III gel (1% agarose; 0.5 XTBE; 200 V; 15 h; pulse wave 60 seconds; 120° angle; Bio-Rad Laboratories, Richmond, CA) using low range pulsed-field gel electrophoresis (PFG) marker (New England Biolab, Beverly, MA) as the molecular weight standard. 

Microarray RNA Labeling

For probe preparation, 5 μg of total RNA for each sample was radiolabeled with [³³P] dCTP in a reverse-transcription (RT) reaction. RNA was annealed in 16 μL H₂O, with 1 μg 24-mer poly (dT) primer (Research Genetics, Huntsville, AL), by heating at 65°C for 10 minutes, and cooling on ice for 2 minutes. The RT reaction was performed by adding 8 μL of 5X first strand RT buffer (Life Technologies, Rockville, MD), 4 μL of 20 mmol/L dNTPs (minus dCTP; Pharmacia, Piscataway, NJ), 4 μL 0.1 M dithiothreitol (DTT), 40 U RNAseOUT (Life Technologies), 6 μL 3000 Ci/mmol alpha [³³P] dCTP (ICN Biomedicals, Costa Mesa, CA) to the RNA/primer mixture to a final volume of 40 μL. Two microliters (400 U) of Superscript II RT (Life Technologies) was then added, and the sample was incubated for 30 minutes at 42°C, followed by additional 2 μL of Superscript II RT and stopped with 5 μL 0.5 M EDTA after 30 minutes of incubation. The samples were incubated at 65°C for 30 minutes after addition of 10 μL 0.1 M NaOH to hydrolyze and remove RNA. The samples were pH-neutralized by the addition of 45 μL 0.5 M Tris (pH 8.0) and purified using Bio-Rad 6 columns (Bio-Rad, Hercules, CA). 

Array Construction and Hybridization

Microarray construction and hybridization were performed as previously described. ²³ Briefly, NIA-Human Focused Arrays including 2742 nonredundant genes enriched for involvement in stress response, were printed on Nytran + Supercharge nylon membranes (Schleicher and Schuell) in duplicates. To demonstrate reproducibility and to reduce technical variations of procedure, each pooled RNA sample was hybridized twice to different arrays producing two replica points for each condition. Membranes were hybridized with α -³³P-dCTP-labeled cDNA probes overnight at 50°C in 4 mL of hybridization solution as previously described ²⁴ (protocols available at http://www.grc.nia.nih.gov/ branch/rrb/rrb.htm; accessed August 12, 2004). Hybridized arrays were rinsed in 50 mL 2X SSC and 1% SDS twice at 55°C, followed by one to two times of washing in 2X SSC and 0.1% SDS at 55°C for 15 minutes each. The microarrays were exposed to phosphorimager screens for 1 to 3 days. The screens were then scanned in a Molecular Dynamics STORM PhosphorImager (Sunnyvale, CA) at 50 μm resolution. ArrayPro software (MediaCybernetics, Silver Spring, MD) was used to convert the hybridization signals from the image into raw intensity values, and the data thus generated were transferred into Microsoft Excel spreadsheets (Microsoft, Seattle, WA), predesigned to associate the ArrayPro data format to the correct gene identities. 

Array Data Analysis

ImageQuant software (Amersham Bioscience) was used to convert the hybridization signals into raw values. The z score normalization method was applied to the raw expression values. To achieve that first, intensity data were log₁₀ transformed to reduce the variance due to the extreme values. Mean and SD of the log₁₀ score were calculated and used to calculate the z score by subtracting the overall average from each data point in the experiment and dividing the result by the overall SD. The distribution of z scores was calculated for all the genes in each array. Gene expression differences between any two experiments were calculated by subtracting the corresponding z scores for each gene. To facilitate comparison between several experiments, z ratios were calculated by dividing the observed z differences between untreated cells and treated RPE cells at each time point of recovery by SD of the differences. A z ratio of ± 1.96 was considered to be significant (P < 0.05). Since there were two independent technical repeats for each experiment, a z test was also performed to give a two-tailed P value for each difference with a significance threshold of P < 0.001. ²⁴  

Results

Induction of Cellular Changes in RPE Cells by Prolonged Nonlethal Oxidative Stress

The cellular and genetic responses to a nonlethal oxidative injury in vitro were assessed in growth-quiescent human retina epithelial cells (ARPE-19) cultured using previously reported protocols. ^{25 26 27} Markers of RPE cells were verified and included the presence of keratin, RPE-65, and cellular retinaldehyde-binding protein (CRALBP; data not shown). Oxidative injury in RPE cells was produced using HQ. To determine the appropriate duration of exposure to hydrophilic HQ, a time course of HQ exposure and determination of lethality was performed. As can be seen in Figure 1A , exposure to high doses of HQ resulted in RPE cell death only when present for >4 hours on the cells, confirming previous reports of the relatively slow time course of HQ entrance and intracellular generation of oxidative reactants. ^{28 29} To induce conditions of prolonged injury, ¹⁵ 16 hours of HQ exposure was chosen. To verify the viability range of increasing doses of HQ in RPE cells, various determinants of cell viability were used. Assays for mitochondrial function (XTT) and cell membrane integrity (trypan blue exclusions) demonstrated full cellular viability at a dose of 100 μM or lower of HQ with notable cell death at higher doses (Figs. 1B and 1C ). To further verify the absence of cell death at a nonlethal dose of HQ, more direct cell death assays were used. Specifically, DNA laddering and AIF nuclear translocation, both markers of RPE cell death ³⁰ were not detected after exposure of the cells to 100 μM HQ (Figs. 2B 2D ). AIF translocation and the appearance of 50 kb DNA laddering were seen only after treatment with a lethal concentration of HQ (500 μM; Figs. 2C and 2D ). 

Viable cells should be able to support a number of internal functions including protein synthesis. To assay protein synthesis function, RPE cells were incubated for 24 and 48 hours with ³H-amino acids and the level of protein incorporation was determined. Figure 1D demonstrates an increase in the level of amino acid incorporation into total cell protein in RPE cells under nonlethal oxidative stress conditions (100 μM HQ). A substantial decrease in protein synthesis was detected at higher concentrations of HQ (500 μM). For both concentrations of HQ, rates of amino acid incorporation did not demonstrate any statistical difference after 24 or 48 hours incubation with labeled amino acids. To control for possible HQ-induced proliferation of RPE cells, rates of DNA synthesis was determined by [³H]-thymidine incorporation. There was no significant increase in RPE cell proliferation after treatment with nonlethal doses of HQ (50, 75, or 100 μM) compared to the untreated control level, whereas higher doses of HQ caused a sharp decline in thymidine incorporation, consistent with cell death (data not shown). 

To assess the effects of prolonged nonlethal oxidative stress on RPE morphology, growth-quiescent RPE cells were subjected to nonlethal (100 μM) and lethal (500 μM) doses of HQ. To track cell membrane blebbing, growth quiescent ARPE-GFP-c’-rRas expressing GFP in the inner leaflet of the plasma membrane ¹⁴ were used. Exposure to a nonlethal dose of HQ initiated a number of cellular changes. Whereas control RPE cells showed absence of any significant cell membrane blebbing, stress-activated kinase-2/p38 (SAPK2/p38) phosphorylation or nuclear DNA condensation, and preservation of actin stress fibers (Figs. 3A 3D and 3G ), injured (100 μM HQ) RPE cells demonstrated the presence of small surface membrane blebs, disassociation of actin into globular aggregates, and p38 phosphorylation (Figs. 3B 3E and 3H) . However, no nuclear DNA condensation was seen indicating an absence of cell death (Figs. 3B and 3H ). Blebbing and actin disassociation were fully reversible after exposure to HQ, returning to the untreated appearance within 48 hours (data not shown). By contrast, a lethal dose of HQ (500 μM) induced extensive membrane blebbing with large blebs, loss of F-actin cytoskeletal structures, extensive p38 phosphorylation, and nuclear DNA condensation (Figs. 3C 3F and 3I ). 

To confirm the apparent disassociation of membrane blebbing response from cell death mechanisms, the numbers of RPE cells demonstrating cell membrane blebbing and AIF translocation were counted. As can be seen in Figure 2E , under nonlethal conditions, approximately 45% of all cells demonstrated cell membrane blebs but without AIF translocation. However, under lethal conditions, all cells demonstrated cell membrane blebbing and AIF translocation. These results indicate that the extensive cellular changes occurring in RPE cells when exposed to a nonlethal oxidative stress do not lead to the initiation of cell death. 

Specific Changes in Gene Expression Pattern in Growth-Quiescent RPE Cells Initiated by Prolonged Nonlethal Oxidative Injury

To define the genetic responses of RPE cells to a prolonged nonlethal injury systematically, genetic profiles of untreated and injured cell cultures were performed using microarray analysis. To examine the kinetics of gene expression after the oxidative injury, gene expression profiles were collected at various times during the recovery period (0, 6, 16, or 48 hours). Since biological variation between the cells cultured under similar conditions and undergoing the same number of passages was assumed to be limited, variations in the HQ treatment conditions were expected to be the main source of variability between the samples. To address this issue, equal amounts of total RNA from RPE cells injured with a nonlethal dose of HQ (100 μM) from three independent experiments were mixed. The resulting total RNA pool was used for the labeling and hybridization. Pooling of cRNA samples from independent tissue samples has been shown to be a valid approach in minimizing biological variations in expression profiles. ³¹  

To minimize any technical variations, hybridization was performed for each time point in duplicate using two independent arrays. Z ratios were calculated for each gene of interest as the ratio of fold changes between untreated and treated conditions of RPE cells at each time point. Since there were two independent technical repeats for each experiment, a z test was also performed to give a two-tailed P value for each difference with a significance threshold of P < 0.001. Analysis of these differences revealed that from the 2742 genes spotted on the stress gene array between 1% to 1.6% of genes (specifically, 29, 31, 39, or 46 genes at each time point) were found to be significantly upregulated (z ratio >2; P < 0.001), whereas between 1.8% and 5% of genes (specifically, 58, 72, 110, or 128 genes at each time point) were significantly downregulated (z ratio <−2; P < 0.001) compared to controls over the time course after the injury. Only those genes which expression values were changed at least twofold and the P value was <0.001 at the same time for one or more time points during the course of recovery were considered for future analysis. Low abundance genes downregulated more than sixfold after treatment in which verification assays (Northern blot) were not possible were omitted from the analysis. Analysis of 368 downregulated and 145 upregulated genes meeting these criteria demonstrated that several functional gene groups, possibly associated with AMD pathogenesis, were affected by prolonged oxidative injury. Interestingly, changes in expression of apoptosis related protein, extracellular matrix protein, and stress-related genes followed a specific pattern. Most of the genes, which exhibited a pattern of significant elevation immediately at the end of the injury, were involved, either directly or indirectly, in the protection of the cells from the oxidative stress (Fig. 4 , middle panel). An apoptosis-related group of genes was significantly downregulated after the injury (Fig. 4 , top panel). Other functional gene expression groups did not demonstrate any noticeable pattern except for an overall increase in the degree of gene dysregulation, as can be seen in the case of extracellular matrix related proteins (Fig. 4 , bottom panel). A detailed analysis revealed a pattern of dysregulation reflecting a possible mechanism of successful recovery from the oxidative injury in RPE cells (Table 1) . A group of genes responsible for directly ameliorating the oxidative stress (MGST1, TRX, TXNRD1, PRDX1, FTH1) as well as antiapoptotic genes (BIRC3, ALDH2, TNFRSF1A) was significantly upregulated. The microarray expression pattern of three of these genes (TRX, TXNRD1, PRDX1) was confirmed by Northern blot analysis (Fig. 5) and demonstrated close correlation with the z ratios over the various time points. In addition, genes responsible for controlling oxidative protein damage (HSPA70, DNAJB1, DNAJA1) or DNA damage (RECQL4) were upregulated as well as a transcriptional factor (ATF4) known to be responsible for the regulation of a number of stress-related pathways. Other genes (CASP3, CASP6, CASP9, TNFRSF10, TNFRSF12, TNSF12, BAD, TRAF4, cFOS, FADD, DAP, and NGFR) linked to cell death pathways were significantly downregulated in the RPE cells after the injury. 

Of note, the changes in the z ratio of 22 from the above 26 genes had returned to a baseline control levels 48 hours after the injury (Table 1) , further demonstrating complete cell recovery from the injury. 

Change in Expression of Extracellular-Related Genes

Because of the importance of the extracellular matrix in RPE-related diseases, these genes were identified and their expression profile examined after prolonged HQ injury in RPE cells. A number of these genes were shown to be dysregulated (Table 2) . 

Specifically, genes involved in matrix turnover and composition (FN1, MMP15, MMP3, FBN1, and CCN1) were altered. Interestingly, several of these genes appeared to be upregulated only 48 hours after the injury, a time when the majority of other gene expression dysregulation had returned to baseline levels. This may be due to a prolonged feedback loop, as in the case with growth factor–induced expression of CCN1 ^{32 33} or possibly due to a steady state alteration in the gene expression profile of the cell. 

Discussion

RPE cells are subjected to a high degree of oxidative injury in vivo because of their anatomic position and function. Prolonged oxidative stress has been suggested as one of the causes of a number of retinal pathologic conditions, including AMD. ^{1 2} In the present report, the genetic and cellular responses of growth-quiescent RPE cells to a nonlethal injury were examined in vitro. The results indicated that a nonlethal oxidative injury triggered a set of genetic responses that appear to aid the RPE cell in ameliorating the oxidative stress and avoiding cell death. On the cellular level, RPE cells demonstrated cellular membrane blebbing and cytoskeletal rearrangement. In addition, changes in cell expression of extracellular matrix genes, responses that might have direct involvement in RPE disease pathogenesis, were detected. Most of the above changes were fully reversible without evidence of programmed cell death activation. 

Hydroquinone (HQ), the nonlethal oxidant trigger used in the present study, causes a dose-dependent production of oxidant byproducts, including superoxide and hydroxyl radical anions and hydrogen peroxide. ³⁴ A longer period of treatment with a nonlethal dose of oxidative stimuli was chosen to evaluate mechanisms of resistance of RPE cells to a prolonged injury. The concentration of 100 μM HQ used in this study was shown not to be lethal by multiple assays indicating maintenance of cell membrane integrity (trypan blue exclusion) and mitochondrial dehyrogenase activity (XTT assay). In addition, both AIF translocation and DNA laddering, known markers of RPE apoptosis, ³⁰ were not detected. Compared with untreated cells, total protein synthesis was increased, without accompanying DNA synthesis, a phenomenon seen in other cell types exposed to nonlethal stress. ^{35 36} This response can in part be explained by transcriptional upregulation of genes in response to the oxidative injury in RPE cells. For example, ATF4, a molecule regulating transcription of multiple genes involved in the defense of the cell from oxidative stress, amino acid import, and translation, ³⁷ was upregulated in RPE cells after the nonlethal stress. 

The present study demonstrated that after a prolonged oxidative treatment of RPE cells, rearrangement of F-stress fibrils into globular aggregates was accompanied by intensive blebbing. Disruption of the F-actin network can be associated with membrane bleb formation, as has been detected in other cell types such as hepatocytes ³⁸ and MDCK cells. ³⁹ Both of these processes may be reversible. ^{40 41} Oxidant injury can affect actin structure by various pathways including activating mitogen-activated protein (MAP) kinase stress-activated kinase-2/p38 (SAPK2/p38) with subsequent phosphorylation of the actin modulator heat shock protein 25/27 or redox alterations of actin or actin regulatory proteins. ⁴² We have previously shown that RPE cells express high levels of Hsp27 ¹⁴ and the present study further demonstrated that activation of p38 occurs under injury conditions. Cellular blebbing observed in the present study is a well-defined injury response that can occur in both physiological and pathologic situations and can be part of the apoptosis pathway, related to injury, or induced by physiological stimuli. ⁴³ In the present study, blebbing occurred at a nonlethal HQ concentration in RPE cells without AIF translocation, or DNA degradation, known markers of program cell death. ³⁰ These results confirm previous observations that blebbing can be distinct from programmed cell death ^{41 44} and may be an early response to nonlethal injury activated independent of apoptosis. ⁴⁵  

A number of genes involved in antioxidation, protection, detoxification, and reparation (MGST1, TRX, TXNRD1, PRDX1, FTH1, DNAJB1, DNAJA1, and HSP70) were transcriptionally upregulated in the present study in response to the nonlethal oxidative injury. Upregulation of proteins in the glutathione and thioredoxin related pathways in response to an oxidative injury was described in several studies. ^{46 47 48} Glutathione is the most abundant intracellular thiol-based antioxidant ⁴⁹ and appears to be protective in RPE cells against oxidative damage. ⁵⁰ Similarly, after a nonlethal injury in RPE cells, the present study demonstrated upregulation of glutathione-related genes (MGST1) and thioredoxin-related genes (TRX, TXNRD1, PRDX1) with subsequent normalization of the RNA levels to baseline levels within 48 hours. TRX in conjunction with TXNRD1 are part of the oxidoreductase system with antioxidant properties found in many mammalian cells and can be induced by oxidative stress in many cell types including RPE cells. ⁴⁹ Upregulation of ferritin mRNA in RPE cells may also aid in antioxidative defenses by controlling intracellular iron levels and minimizing hydroxyl radical formation, important contributors to oxygen free-radical toxicity. ^{51 52}  

The protein chaperones Hsp70 and different forms of Hsp40 (DNAJB1 and DNAJA1) have an indispensable role in preventing aggregation of partially denatured proteins. ^{53 54 55} Both were upregulated after the injury in RPE cells. Hsp70 may play a direct role in prevention of apoptosis in RPE cells by binding to AIF, ⁵⁶ a major determiner of apoptosis in RPE cells. ³⁰  

Oxidative stress affects not only cellular proteins, but can also induce cell death through multiple mechanisms including direct DNA damage and activation of programmed cell death pathways. Oxidative stress also affects the ability of the cells to undergo homologous recombination properly and increases the possibility of DNA mutagenesis. ^{57 58} Transcriptional upregulation of RECQL4, a gene involved in suppressing illegitimate recombination, ⁵⁹ was detected after the oxidative stress and might play a role in maintaining DNA genomic stability in RPE cells. 

In addition to the upregulation of the genes involved in protection and detoxification, the present study also demonstrated that RPE cells upregulate genes directly involved in inhibition of apoptosis (BIRC3, TNFRSF1A, ALDH2) and downregulate genes involved in active apoptotic pathways (caspase3, caspase 6, caspase 9, TRAF4, TNFRSF10, TNFRSF12, TNFSF12, NGFR, BNIP3, cFOS, FADD, BAD, and DAP). Significant upregulation of the transcription of the inhibitor of apoptosis (IAP) molecule containing third BIR domain (BIRC3) was detected immediately after the oxidative injury in RPE cells. The BIR3 domain of the XIAP potently inhibits the activity of the processed caspase-9. ⁶⁰ TNFRSF1A (TRADD), a multifunctional intracellular signaling protein is recruited to the TNF receptor 1 and is involved in NF-κB nuclear translocation, a well-known mechanism for cell survival. Family members of the TNF superfamily (TRAF4, TNFRSF-10 and TNFRSF-12, TNFSF12, NGFR) were downregulated after the oxidative injury. These proteins are secreted ligands with known pro-apoptotic functions. ^{61 62} BNIP3, which has intrinsic pro-apoptotic activity and inhibits the activity of the anti-apoptotic protein BCL-X_L ⁶³ and Fas (TNFRSF6)-associated via death domain (FADD) which is involved in autocleavage and activation of caspase-8 and initiation of apoptosis ⁶⁴ were both downregulated after the injury. cFOS plays an important role in neuronal cell death, ⁶⁵ and light-induced apoptosis of photoreceptors ^{66 67} and was also downregulated after the oxidative injury in RPE cells. Dysregulation of genes directly linked to apoptosis has not previously been described as a component of the survival mechanism in other cell types and appears to aid the RPE in avoiding cell death as recovery from the oxidative injury is taking place. 

Extracellular matrix-related genes (FBN1, CYR61, FN1, MMP3, MMP15) were also dysregulated after the oxidative stress, with some of these genes remaining upregulated at the 48-hour time point. Upregulation of MMP3 and MMP15, zinc dependent proteases that are responsible for the degradation and remodeling of the extracellular matrix, was detected in injured RPE cells. MMP15 is expressed in RPE cells ⁶⁸ and capable of degrading laminin, fibronectin, tenascin, nidogen, aggrecan, and perlecan. ⁶⁹ MMP3, known as stromelysin 1, also has a broad range of substrate activity. ⁷⁰ Fibronectin is upregulated by oxidative stress. ³⁶ Fibrillin 1 in association with other proteins forms extracellular complexes in elastic tissues. ⁷¹ Alterations in extracellular elastin under the RPE have been found in early AMD. ⁷² Another protein that associates with the extracellular matrix and is dysregulated by oxidative stress in RPE cells was Cyr61. Cyr61 is capable of mediating diverse functions such as monocyte adhesion, ³³ matrix turnover, and cell survival. ³²  

What novel genetic responses were detected in RPE cells using the present experimental design? A number of studies have demonstrated the usefulness of microarray analysis to examine global changes in genetic expression patterns in response to oxidative stress in a variety of cell types thought to be involved in tissue specific diseases. ^{73 74 75 76 77} However, in general, these studies have not evaluated comprehensively the temporal kinetics of the gene dysregulation. ^{73 74 75 76 77} In the present report, the time-dependent kinetics of changes in gene expression pattern after a prolonged nonlethal oxidative injury demonstrated complex and integrated recovery mechanisms in RPE cells. In addition, the specifics of both the cellular and genetic responses of RPE cells to a prolonged oxidative treatment have not been previously assessed. Whereas a number of studies ^{17 18 19 20 78} have reported similar findings of upregulation of detoxification and antioxidant genes after exposure to an oxidative stress, the present study is the first to detail, as part of the survival equilibrium in RPE cells, the downregulation of apoptosis-related genes. 

It is important to remember, however, that major limitations of the use of microarray exist. Primary is the focus of this technology on quantitative differences in RNA levels. Critical post-transcriptional events involved in the regulation of protein levels and activity were not analyzed. For example, while IAP RNA levels may be increased, translation of IAP proteins can be inhibited by apoptosis-inducing proteins such as Reaper. ⁷⁹ Translational modifications and activation of the MMPs are thought to play a critical role in the subsequent protein activity levels of the MMPs. ⁸⁰  

In summary, the present work points to the importance of studying prolonged nonlethal oxidative stress in RPE cells, a condition which elicits a wide range of cellular and genetic responses. In the study of Alzheimer’s, a chronic degenerative disease of aging, investigators have pointed out that apoptosis is not a prominent feature of the disease, but it is the cytopathology occurring in neuronal cells before late stage apoptosis which can result in prolonged cellular dysfunction and lead to disease symptoms. ^{81 82 83} Similarly, it may be that the study of nonlethal injury to RPE diseases will provide important clues to our understanding of RPE–related diseases. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2003.

 Submitted for publication March 18, 2004; revised June 24, 2004; accepted July 4, 2004.

 Disclosure: N. Strunnikova, None; C. Zhang, None; D. Teichberg, None; S.W. Cousins, None; J. Baffi, None; K.G. Becker, None; K.G. Csaky, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Viability of growth quiescent ARPE-19 cells after 16 hours of treatment with HQ. XTT colorimetric assay expressed as a percentage of live cells relative to the control after (A) varying duration of exposures or (B) varying doses of 16 hour exposure of HQ. Each data point represents the mean obtained from three independent experiments performed in triplicate ± STD. (C) Cell membrane integrity was measured by trypan blue exclusion. A minimum of 250 cells were counted for each data point. The data are expressed as the percentage of viable cells relative to the total number of cells in control sample ± STD. Three independent experiments were performed for each condition. (D) Protein synthesis after HQ exposure was measured by incorporation of l-methionine [methyl ³H], l-tyrosine [ring-3; 5-³H], and L-leucine [3; 4; 5-³H(N)] amino acids for 24 or 48 hours after the treatment as in (B).

Figure 2.

 

Localization of AIF and DNA laddering pattern in RPE cells after nonlethal oxidative treatment. RPE cells were treated with nonlethal (100 μM) and lethal (500 μM) concentrations of HQ for 16 hours. Cellular membrane blebbing was visualized in ARPE-GFP-c’-Ras cells. In control cells (A, arrows), AIF demonstrated punctual cytoplasmic pattern corresponding to the mitochondrial localization with absence of AIF in the DAPI labeled nuclei. Injured RPE cells demonstrated a similar localization of AIF (B, arrows). Lethal treatment caused redistribution of AIF to the nuclei (C, arrows). Scale bar, 50 μm. In parallel experiments, RPE cells were treated with the same doses of HQ and used to analyze the presence of large-scale DNA fragmentation and count the number of blebs. Pulse field gel electrophoresis was performed on 10 μg of genomic DNA extracted from the cells subjected to the same treatment with HQ (D). Note absence of large scale DNA fragmentation after nonlethal treatment. Data represent results from two independent experiments. Blebbing cells and cells with redistributed AIF were counted in the control and treated groups. The values are the percentage of cells displaying either cell membrane blebbing or nuclear AIF localization obtained by an average of three fields per slide and 150 cells per slide were counted (E). The results shown are mean ± STD, representative of three independent experiments (P < 0.001, ANOVA/t-test, unpaired).

Figure 3.

 

Cellular and morphologic changes caused by nonlethal oxidative injury in growth-quiescent RPE cells. RPE cells were treated with nonlethal (100 μM) and lethal (500 μM) concentrations of HQ for 16 hours. Cellular membrane blebbing was visualized in ARPE-GFP-c’-Ras cells. No significant blebbing was observed in control cells (A), while injured RPE cells demonstrated small surface membrane blebs (B, arrows and inset) morphologically different from large membrane lethal blebbing (C, stars and inset). Corresponding changes in the F-actin structure were visualized after the treatment. Note that F-actin fibrils in control cells (D, arrows) were partially rearranged into the aggregate structures after the injury (E, arrows) and were completely disassembled after the lethal treatment (F, arrows). Phosphorylation of p38 was detected immediately after the treatment at both nonlethal (H) and lethal (I) concentrations. Note nuclear DNA condensation (I, arrows). Images represent data from three independent experiments. Scale bar, 50 μm.

Figure 4.

 

Patterns of gene dysregulation in RPE cells after prolonged nonlethal oxidative injury, differing between functional groups of genes. After a nonlethal oxidative injury (100 μM HQ) a number of significantly down- (z ratio < −2, P < 0.001) or upregulated (z ratio >2, P < 0.001) genes were detected in apoptosis-related (top panel), stress-related (middle panel), and extracellular matrix-related groups (bottom panel) at each recovery time period (0, 6, 16, and 48 hours).

Table 1.

 

Anti-apoptotic, Pro-apoptotic, and Stress-Related Genes Quantified by Microarray Analysis after Nonlethal Oxidative Injury in RPE Cells

Treatment/Recovery Time	16/0	16/6	16/16	16/48	Function	Gene Bank Accession Number	Gene
Gene Name	Z-Ratio
Baculoviral IAP repeat-containing 3	5.94	4.17	2.75	1.42	Caspase inhibitor	NM_001165	BIRC3
Aldehyde dehydrogenase 2	3.57	2.52	2.23	1.46	Detoxification of carbonyls	NM_000690	ALDH2
Peroxiredoxin 1	2.89	0.98	0.79	−0.73	Protection from the oxidative stress	NM_002574	PRDX1
Thioredoxin	5.33	4.02	2.59	0.05	Protection from the oxidative stress	NM_003329	TRX1
Thioredoxin reductase 1	3.47	1.38	0.32	−0.91	Protection from oxidative stress	NM_003330	TXNRD1
Glutathione S-transferase	5.78	2.77	2.58	1.45	Protection from the oxidative stress	AY3688173	MGST1
Ferritin	2.64	2.6	1.04	0.98	Protection from the oxidative stress	NM_002032	FTH1
Activating transcription factor 4	3.19	1.61	1.33	0.58	Transcriptional regulation of the stress response pathways	NM_001675	ATF4
DNAJ (Hsp40) homolog, subfamily B member 1	3.56	2.69	2.01	0.7	Detoxification and protection from the stress	NM_006145	DNAJB1
DNAJ (Hsp40) homolog, subfamily A member 1	2.62	1.32	1.27	−0.42	Detoxification and protection from the stress	NM_001539	DNAJA1
RecQ-like protein 4	2.33	2.64	2.01	0.65	DNA repair	NM_004260	RECQL4
Heat shock protein 70	2.52	1.15	1.62	0.75	Detoxification and protection from the stress	NM_021979	HSPA2
TNF receptor associated death domain protein (TRADD)	−1.28	2.91	2.03	−1.16	Apoptosis, proliferation, differentiation	NM_003789	TNFRSFIA-associated via death domain
Caspase 3	−2.06	−1.37	−2.01	−0.96	Involve in cell death execution	NM_032991	CASP3
Caspase 6	−0.36	−2.5	−1.29	−1.63	Involve in cell death execution	NM_001226	CASP6
Caspase 9	−3.34	−1.93	−3.98	−0.86	Involve in cell death execution	NM_001229	CASP9
BCL2-antagonist of cell death	−1.60	−2.10	−1.40	−0.81	Involved in cell death execution	NM_004322	BAD
Death associated protein	−1	−0.22	−2.06	1.12	Involved in cell death execution	NM_004394	DAP
Fas associated death domain (FADD)	−2.84	−3.74	−3.27	−2.85	Involve in cell death execution	NM_003824	Fas (TNFRSF6)-associated via death domain
TNF superfamily member 16	−2.38	−1.97	−2.05	−2.81	Apoptosis, proliferation, differentiation	NM_002507	NGFR
TNF superfamily member 12	−1.87	−2.21	−2.11	0.12	Apoptosis, proliferation, differentiation	NM_003809	TNFSF12
TNF Receptor 10 Super Family	−2.52	−2.73	−3.44	−2.65	Apoptosis, proliferation, differentiation	NM_147187	TNFRSF10
TNF Receptor 12 Super Family	−2.55	−2.32	−2.2	−1.69	Apoptosis, proliferation, differentiation	NM_003790	TNFRSF12
TNF Receptor Associated Factor-4	−3.26	−1.66	−3.84	0.58	Apoptosis, proliferation, differentiation	NM_004295	TRAF4
c-FOS	−3.38	−3.35	−3.30	−3.08	Apoptosis, proliferation, cell cycle regulation	NM_005252	C-FOS
BCL2/adenovirus E1B 19 kDa interacting protein	−4.84	−3.30	−2.99	−1.90	Pro-apoptotic properties	NM_004052	BNIP3

 Changes in gene expression between the untreated and injured cells were measured following different recovery time (0, 6, 16, and 48 hours) and presented in form of Z ratio.

Figure 5.

 

Northern blot validation of stress-triggered changes in mRNA levels. A fraction of total RNA samples used for the microarray analysis was used for validation by Northern blotting. Hybridizations were performed using 15 μg of total RNA obtained from the untreated RPE cells and at 0, 6, 16, and 48 hours after the treatment. Full-length cDNA probes to PRDX1, TRX, and TXNRD were used for the hybridization. Control hybridizations to detect level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were performed to monitor the evenness of loading and transfer of the Northern blot samples. Numbers shown are actual differences in gene expression between the expression level at the control and each point of recovery based on the microarray data presented in form of z ratios.

Table 2.

 

Dysregulation of Extracellular Matrix-Related Genes in RPE Following Nonlethal Oxidative Injury

Treatment/Recovery Time	15/0	15/6	15/16	15/48	Function	Gene Bank	Gene
Gene Name	Z-Ratio
Matrix components
Cysteine-rich angiogenic inducer	−2.80	−0.65	0.77	2.21	Extracellular matrix and cell adhesion	NM_001554	CYR61 (CNN1)
Fibronectin 1	−2.43	−0.08	−0.5	2.48	Cell adhesion	NM_002026	FN1
Matrix metalloproteinase 15	2.56	3.46	2.23	−0.25	Cell adhesion and neovascularization	NM_002428	MMP15
Matrix metalloproteinase 3	0.9	2.07	2.68	2.08	Cell adhesion and neovascularization	NM_002422	MMP3

 Changes in gene expression between the untreated and injured cells were measured following different recovery times (0, 6, 16, and 48 hrs) and presented in the form of Z ratios.