Aβ-Induced Senescent Retinal Pigment Epithelial Cells Create a Proinflammatory Microenvironment in AMD

Abstract

Purpose.: Chronic inflammation is implicated in the pathogenesis of AMD. The source of chronic inflammation is often attributed to the progressive activation of immune cells over time. However, recent studies have shown that senescent cells can alter tissue microenvironment via secretion of growth factors, proteases, and inflammatory cytokines and might be an additional source of chronic inflammation. We hypothesized that altered secretory pattern in Aβ-induced senescent RPE cells may contribute to compromised RPE barrier integrity and chronic inflammation in AMD.

Methods.: Senescence was assessed by measuring the SA-β-Galactosidase activity, the expressions of p16^INK4a and ATM, and cell cycle analysis. Expressions of IL-8 and MMPs were analyzed by RT-PCR, ELISA, and gelatin zymography. The barrier structures of RPE cells were detected by actin-tracker, ZO-1, claudin-19, occludin immunochemistry, and Western blot; barrier function was analyzed by measuring transepithelial resistance (TER) and transepithelial diffusion rate of FITC-dextran. For inhibitory studies, MMP-9 was inhibited by RNA interference strategy.

Results.: Aβ promotes RPE cells to enter senescence and secrete higher concentrations of IL-8 and MMP-9. Secretion of MMP-9 is associated with compromised barrier integrity and with processing of IL-8 to a more activated form. Silence of MMP-9 preserved the barrier integrity of senescent RPE cells.

Conclusions.: The altered secretory phenotype of senescent RPE cells may contribute to age-related inflammation in AMD.

Chinese Abstract

Introduction

Age-related macular degeneration (AMD) is a leading cause of irreversible visual loss in Western countries. ¹ Drusen represent the earliest sign of AMD and are associated clinically with geographic atrophy of the RPE (dry-AMD) ² and with an increased risk of developing the exudative form of AMD (wet-AMD). ³ Amyloid-beta (Aβ), a known constituent of drusen in AMD patients, ^4,5 is a contributor of developing AMD. Aβ has been implicated in activation of the complement cascade and is a major component of drusen, where it colocalized with activated complement component. ^5,6 Aβ observed in retinas of aging mice was associated with increased recruitment of macrophages. ⁷ Exposure of Aβ to cultured ARPE-19 cell line induced inflammation-associated gene expression changes. ⁸ Although this evidence suggests that Aβ deposition may be associated with activation of chronic inflammation in AMD, the underlying mechanism is not fully explored. 

The greatest risk factor for AMD is aging; several lines of evidence suggest that RPE cells undergo an accelerated ageing process in AMD ^9,10 : (1) RPE cells display senescence when grown continuously in culture ¹¹ ; (2) Senescent RPE cells were observed in aged monkeys ¹² ; (3) Several causes of AMD, such as cigarette smoke, radiation, and complement components, could induce RPE senescence. ^13–15 The loss of RPE cells is the critical event of the atrophic form of AMD leading to irreversible visual loss. ¹⁶ When cells encounter stress, they undergo either apoptosis or senescence, resulting in cell loss. Celluar senescence is defined as an irreversible cell proliferation arrest. Unlike apoptosis, senescent cells remain metabolically active and produce a myriad of factors, including interleukins, such as IL-1α/β, IL-6, and IL-8; growth factors, such as epidermal growth factor (EGF) and VEGF, and several matrix metalloproteinases (MMPs), which create a proinflammatory microenvironment. ^17,18 Chronic inflammation is strongly linked to the pathology of AMD. ¹⁹ Although the immune system plays a major role in modulating the levels of pro- and anti-inflammatory factors, it is not the only source of these factors. It has been shown that the proinflammatory phenotype of senescent cells is another source of inflammation. ²⁰ RPE cells can secrete various molecules, such as growth factors, proteases, and cytokines, that control the stability of the Bruch Membrane (BrM) and the choroid and regulate the immune response. Therefore, we asked whether the altered secretory pattern in senescent RPE cells may contribute to the pathology of AMD. 

MMPs are calcium-requiring, zinc-containing endopeptidases that constitute a major component of the enzyme cascade responsible for degradation of extracellular matrix (ECM) proteins. Verzar ²¹ recognized that aging is associated with ECM alterations. RPE cells produce MMPs, including MMP-2, MMP-9, MMP-3, and MMP-1. ^22–24 BrM is a pentalaminar structure composed of the RPE basement membrane, two collagenous zones divided by a middle elastic layer, and the choriocapillaris basement membrane. With aging, the BrM thickens and type I collagen increases, ²⁵ whereas the type IV collagen in RPE basement membrane declines. ²⁶ MMP-2 and MMP-9 preferentially degrade basement membrane components, such as type IV collagen. It is suggested that MMP-9 might be upregulated with aging. However, the role of MMP-9 in chronic inflammation remains unexplored. Aβ_1-42 peptides have been shown to disrupt RPE barrier integrity. ²⁷ Interestingly, Aβ could induce astrocytes senescence and increased MMP secretion in cultured cells. ^28–30 It has been demonstrated that MMP-9 activities might be associated with altered RPE barrier function. ^22,31 MMP-9–induced pathological conditions cause epithelial barrier disruption not only by basement membrane degradation, but also by cleavage of tight junction proteins (TJs), occludin, ZO-1, and claudins. ^32–34 A breakdown of epithelial barrier is both a stimulus for inflammation in tissue injury and a component of normal inflammatory processes that permit leukocyte influx into areas of tissue damage. ³⁵  

Here we asked whether Aβ could induce RPE senescence and how the altered secretory phenotype of senescent RPE cells contributes to chronic inflammation in AMD. In the present study, we found that exposure of RPE cells to Aβ_1–42 triggered senescence and that senescent RPE cells produced high quantities of MMP-9 and IL-8. We found that MMP-9 contributed to Aβ-induced RPE morphological alterations and barrier dysfunction and that MMP-9 could truncate IL-8 to a shorter form, which has been reported to have 10-fold higher potency in neutrophil activation. ³⁶ Based on this evidence, we proposed that MMP-9 secreted by senescent RPE cells may accelerate age-related inflammation in AMD. 

Materials and Methods

Aβ_1-42 Oligomerization and Dot Blot Assay

Soluble oligomers of Aβ may be better correlated with the severity of the disease than are monomers or insoluble amyloid fibrils. ^37,38 The oligomeric form of Aβ_1-42 was synthesized and determined by dot blot assay as published, ⁸ with some modifications. Briefly, 0.5 mg of lyophilized Aβ_1-42 (Sigma-Aldrich, Shanghai, China) was dissolved in 140 μL hexafluoroisopropanol (HFIP) and incubated for 20 minutes, followed by addition of 900 μL distilled H₂O and 20-minute incubation in the HFIP/water mixture. Then the solvent was evaporated from resulting supernatant under constant stirring at room temperature (RT) for 12 to 72 hours. 

Aliquots of the supernatant were analyzed for confirmation of Aβ oligomers using dot blot assay. A 2-μL sample was applied to a nitrocellulose membrane that had been blocked with 10% nonfat milk in Tris-buffered saline containing 0.05% Tween 20 (TBS-T) for 1 hour, washed three times with TBS-T, and incubated overnight with the oligomer-specific antibody A11. ³⁹ Subsequently, the membranes were washed and incubated with horseradish peroxidase (HRP)-conjugated antirabbit IgG (Abcam, Hong Kong, China) for 1 hour at RT. The blots were washed and developed with chemiluminescent reagent. Membranes were exposed to ImageQuant LAS 4000 (GE Healthcare Life Sciences, Piscataway, NJ). 

Isolation of Human Fetal RPE Cells

Informed consent for tissue donation was obtained from the relatives, and the protocol of the study was approved by the local ethics committee and adhered to the tenets of the Declaration of Helsinki for experiments involving human tissue. Four human fetal eyes were obtained from the eye bank of the Eye & ENT Hospital of Fudan University, Shanghai, China. Human fetal RPE cells were prepared as described. ⁴⁰ In brief, eyes were washed with Hanks' balanced salt solution (HBSS; Invitrogen-Gibco, Carlsbad, CA), excessive tissues were removed. The posterior pole was incubated in accutase solution (Invitrogen, Carlsbad, CA) for 20 minutes and then transferred to fresh HBSS where the retina was removed. The RPE monolayers were carefully peeled off in small sheets and collected in cold minimum essential medium (MEM; Sigma-Aldrich) with 15% fetal bovine serum, taurine (250 mg/L; Sigma-Aldrich), hydrocortisone (20 μg/L; Sigma-Aldrich), triiodothyronine (0.013 μg/L; Sigma-Aldrich), nonessential amino acids (1:100 dilution; Sigma-Aldrich), N1 supplement (1:100 dilution; Sigma-Aldrich), and glutamine-penicillin-streptomycin. After centrifugation, medium was removed, and the RPE sheets were incubated in trypsin solution for 10 minutes followed by gentle pipetting to obtain a cell suspension. Cells were then transferred to a flask with medium containing 15% serum and cultured in a humidified incubator at 37°C, in 5% CO₂ and 95% air. After 2 days of culture, 15% serum was replaced by 5% serum for 2 to 3 weeks until they formed a monolayer and showed pigmentation. RPE cells at passages 2 to 5 were used in the experiments. 

Lactate Dehydrogenase Assay and TUNEL Analysis

To measure cytotoxicity, RPE cells seeded in a 96-well plate were incubated with OAβ at 0.3 μM for different time points (4–7 days). In other experiments, RPE cells transfected with transfection mixture (25 nM, 24 hours) were treated with or without OAβ (0.3 μM, 7 days). Lactate dehydrogenase (LDH) is a cytoplasmic enzyme released by dying cells; therefore, cytotoxicity was determined by monitoring LDH release from RPE cells into the culture medium, with an LDH cytotoxicity assay kit (Beyotime, Shanghai, China). LDH release reagent treatment (1:10 dilution, 1 hour) was used as a positive control to test maximum LDH release according to the manufacturer's protocol. The optical density was measured spectrophotometrically at 490 nm on a microplate reader (ELx800; Bio-Tek, Winooski, VT). 

For TUNEL analysis, the cells were fixed with 4% paraformaldehyde (PFA) in PBS for 1 hour. The cells were washed with PBS and then permeabilized by 0.1% Triton X-100 for FITC end-labeling the fragmented DNA of the apoptotic RPE cells using TUNEL cell apoptosis detection kit (Beyotime). The FITC-labeled TUNEL-positive cells were detected using a fluorescent microscopy. 

Cell Cycle Analysis and SA-β-Galactosidase Staining

To analyze senescence-associated cell cycle arrest, RPE cells were treated with OAβ at 0.3 μM for different time points (4–7 days). Then they were washed with PBS, detached with 0.25% trypsin, and fixed with 75% ethanol overnight. Samples were resuspended with PBS and stained with propidium iodide (PI) in the dark for 30 minutes, and the cell cycle distribution was analyzed by flow cytometry system. The raw data were analyzed using CellQuest software (Becton Dickinson, Mountain View, CA). 

The senescent cells were assayed with an SA-β-Galactosidase (SA-β-Gal) staining kit (K802-250; Biovision, Milpitas, CA) according to the manufacturer's protocol. RPE cells were washed with PBS, fixed with 4% PFA for 15 minutes, then washed with PBS and incubated overnight at 37°C with SA-β-Gal staining solution mix. The cells were then examined for the development of blue color and photographed at ×200 magnification using a light microscope (EVOS xl microscope; AMG, Bothell, WA), and a total of 1000 cells were counted in 20 random fields to determine the percentage of SA-β-Gal–positive cells. 

Protein Extraction and Western Blot Analysis

Total proteins were obtained using radioimmunoprecipitation buffer containing PI (proteinase inhibitor), and nuclear proteins were extracted as protocol from Abcam. To detect different forms of IL-8, supernatants from culture mediums were collected and concentrated 100× by centrifuging at 4000g for 20 minutes at 4°C through Amicon Ultra-15 centrifugal filter devices (Millipore, Bedford, MA). Protein content was determined by bicinchoninic acid (BCA) protein assay (Pierce, Rockford, IL). Equal amounts of total cellular protein were loaded per lane onto 10% SDS-PAGE and transferred onto polyvinylidene difluoride membrane. The membranes were blocked and incubated with first antibodies (Abcam) against p16^INK4a, ataxia telangiectasia mutated (ATM, phospho Ser1981), MMP-9, occludin, ZO-1, claudin-19, F-actin, IL-8, glyceraldehyde 3-phosphate dehydrogenase, or β-actin overnight at 4°C. The membranes were then washed and incubated with HRP-coupled secondary antibodies for 2 hours. Blots were washed and developed with chemiluminescent reagent. Membranes were exposed to ImageQuant LAS 4000 (GE Healthcare Life Sciences); densitometry was performed using Photoshop CS4.0 (Adobe Systems, San Jose, CA). 

ELISA

The protein concentration of IL-8 was determined using ELISA kits (R&D Systems, Minneapolis, MN) and was calculated based on standard curves. The optical density was measured at 490 nm on a microplate reader (ELx800; Bio-Tek). 

Real-Time Quantitative PCR

Total RNA was isolated using TRIzol (Invitrogen), and reversed transcribed to cDNA with RT Master Mix (Takara Biotechnology, Dalian, China). Expressions of IL-8 and MMP-9 were measured using quantitative PCR mix (Takara Biotechnology). PCR primers were listed as follows: IL-8: sense, 5′-ttgccaaggagtgctaaagaa-3′; antisense, 5′-gccctcttcaaaaacttctcc-3′; MMP-9: sense, 5′-caaagacctgaaaacctccaac-3′; antisense, 5′-gactgcttctctcccatcatct-3′. Real-time PCR was performed in ABI7500 real-time PCR system. β-actin (sense,5′-tgggcatgggtcagaaggattcc-3′; antisense, 5′-ccacacgcagctcattgtagaagg-3′) served to normalize and the relative gene expression was expressed as “fold change,” calculated using the ΔΔCt method. 

Gelatin Zymography

Gelatin zymography was done as described. ²² Briefly, culture media were collected after treatment and subjected to SDS-PAGE in 10% polyacrylamide gels copolymerized with 1 mg/mL gelatin. After electrophoresis, gels were incubated in 2.5% Triton X-100 (1 hour, 37°C) followed by overnight incubation in 50 mM Tris-HCl (pH 7.8), 5 mM CaCl2, 0.02% NaN3, 0.02% Brij. Gels were stained with 2.5% Coomassie blue R-250 (Beyotime) for 45 minutes followed by destaining in deionized water with 10% acetic acid and 20% methanol. Gels were scanned, and density analyses of the bands were performed using Photoshop CS4.0 (Adobe Systems). 

Silence of MMP-9 With RNA Interference

The MMP-9 silencer small interfering RNA (siRNA) was synthesized in Genepharma (Shanghai, China). The sequences were as follows: MMP-9 siRNA: sense, 5′-cuaugguccucgcccugaatt-3′; antisense, 5′-uucagggcgaggaccauagag-3′. Negative siRNA (siRNA-N): sense, 5′-uucuccgaacgugucacgu-3′; antisense, 5′-acgugacacguucggagaa-3′. siRNA transfection was performed according to the protocol of Ref. 41 with modifications. Briefly, siRNA at a final concentration of 25 nM was combined with 10 μL lipofectamine 2000 (Invitrogen) in 500 μL MEM/F12 and allowed to complex by incubation for 20 minutes. Then the RPE cells were incubated with the transfection mixture for 24 hours. 

Cell Morphology and Immunofluorescence Staining

To investigate the morphological alterations in senescent cells, RPE morphology, structures of F-actin, and location of TJs were examined by light microscope (EVOS xl microscope; AMG) and confocal microscope (Leica TCS SP5; Leica Microsystems), respectively. Then cells were fixed in 4% PFA for 30 minutes and blocked with 1% BSA in TBS for 1 hour at RT, then incubated with mouse antioccludin antibody or mouse anti-ZO-1 antibody or rabbit anticlaudin-19 antibody (Abcam) for 1 hour. After washing, they were incubated with AlexaFluor goat-antirabbit 594-conjugated secondary antibody or AlexaFluor (1:500; Invitrogen) or goat-antimouse 488-conjugated secondary antibody (1:500; Invitrogen) and 4′,6-diamidino-2-phenylindole (DAPI) (1:1000) at 25°C for 1 hour. Changes in F-actin structures were detected by FITC-labeled phalloidin (1:250; Beyotime). Then slides were viewed on a confocal microscope (Leica TCS SP5; Leica Microsystems). 

Measurement of Transepithelial Resistance

For RPE monolayer culture, the cells were seeded at 2 × 10⁵ cells/cm² onto Transwell filters (12-mm diameter, 0.4-μm pore size; Costar, Corning, NY), which had been coated with 160 μL of a 1:30 dilution of Matrigel (BD Biosciences, Bedford, MA) in medium and air-dried overnight. Transepithelial resistance (TER) was measured using Millicell-ERS (Millicell; Millipore) and calculated by subtracting the value of a matrigel-coated filter without cells from the experimental value. Final TER (Ω×cm²) was calculated as the formula: Final TER = (the measured resistance) × (the area of the Transwell filter). Fifteen days after TER stabilization, the monolayers were incubated with Aβ (0.3 μM) for 7 days. The medium was changed every day. Measurements were repeated at least four times for each filter, and each experiment was repeated at least four times. 

Permeability Assay

The permeability assays were performed by measuring the passive permeation of FITC-dextran (4 kDa; Sigma-Aldrich) across confluent cells grown on filters. Fifteen days later, the monolayers were treated with Aβ (0.3 μM) for 7 days; 500 mg/mL FITC-dextran was added to the upper chamber on day 21. Samples (100 μL) were taken from the lower and upper chamber 24 hours after addition FITC-dextran. The concentration of FITC-dextran in these samples was quantified by microplate reader (ELx800; Bio-Tek). The diffusion rate was calculated as (amount of dextran lower chamber)/(amount of dextran upper chamber) ⁴² and expressed as fold-change. Each experiment was repeated four times. 

Statistical Analysis

Results were expressed as mean ± SD. Values were processed for statistical analysis by unpaired t-test or by one-way ANOVA, and differences were considered significant at P less than 0.01 (two asterisks) and P less than 0.05 (one asterisk), respectively. 

Results

Characterization of Aβ_1-42 Oligomers and Morphology of RPE Cells

Samples were examined after incubation for 12 or 72 hours at 37°C. Dot blot assay with A11 further confirmed the presence of oligomeric structures as it reacted positively with our OAβ sample (Fig. 1A). In contrast, the A11 antibodies did not react against a monomer (Fig. 1B). Spherical oligomers are initially absent from freshly prepared Aβ_1–42 solutions or monomers (aggregated for less than 24 hours), the oligomer-specific antibodies (A11) have been reported to specifically recognize a generic epitope common to prefibrillar oligomers and not fibrils or monomers (freshly prepared Aβ_1–42 solutions). ³⁸  

Figure 1

 

Confirmation of oligomeric form of Aβ_1–42 and images of primary RPE cells. Lyophilized Aβ peptide was dissolved as previously described, and the fresh aliquots were incubated at 37°C for 12 or 72 hours, respectively. Dot blot with the use of the A11 antibody shows positive staining (A) in the aliquot incubated for 72 hours to conformation oligomers, compared with negative staining (B) in monomer of Aβ samples, which were incubated for 12 hours only. Fluorescence microscope image (C) and “a” channel image in LAB color mode (D) of RPE cells cultured in a flask demonstrated that the primary RPE cells were hexagonal in shape and densely pigmented throughout. Magnification: ×200. Scale bar: 100 μm.

To confirm that the isolated RPE cells displayed classic morphology (uniform hexagonal arrays of cells), confluence, and uniform pigmentation as in native tissue, the morphology of the cells was evaluated using fluorescence microscope (IX51; Olympus, Tokyo, Japan) (Fig. 1C). Then the photograph taken by fluorescence microscope was analyzed using image editing software (Photoshop CS4) in the “a” channel of LAB color mode; apparent regular hexagonal morphology of the cells was observed (Fig. 1D). This result confirmed that the isolated RPE cells exhibited heavy pigmentation (Fig. 1C) and hexagonal epithelial morphology (Figs. 1C, 1D) as in native tissue. 

Cytotoxicity of OAβ_1-42 on RPE Cells

To examine the cytotoxicity of OAβ on human fetal RPE cells, released LDH activities were measured; 0.3 μM OAβ did not induce significant LDH release into RPE culture medium after 4 or 7 days of treatment (Fig. 2A). LDH release reagent treatment was used as a positive control of LDH release, which was 4.5-fold higher than that in the cells treated or untreated with OAβ (Fig. 2A). This result indicated that exposure of RPE cells to OAβ for 4 to 7 days did not induce cell death. When cells encounter stress, they undergo either apoptosis or senescence; therefore, TUNEL assay was performed to exclude apoptosis (Fig. 2B). RPE cells were treated with 0.3 μM OAβ for 4 to 7 days or with 10 mM H₂O₂ for 4 hours (the positive control of apoptosis). No TUNEL-positive staining was observed in cells treated with OAβ compared with H₂O₂-treated cells (green: TUNEL; blue: DAPI). It suggested that chronic exposure to Aβ did not induce cell apoptosis. 

Figure 2

 

Effects of OAβ on RPE cell death. RPE cells were treated with OAβ (0.3 μM, 4–7 days). (A) Cell death was quantitatively assessed by measuring LDH activity released into the medium from damaged cells. OAβ treatment did not induce significant LDH release from RPE cells. LDH release reagent treatment was used as a positive control of LDH release, which was almost 4.5-fold higher than that in the cells treated or untreated with OAβ. (B) TUNEL analysis on cells treated with 0.3 μM of OAβ or 10 mM H₂O₂ (positive control of apoptosis). No TUNEL-positive staining was observed in the cells treated with OAβ. (Results are mean ± SD values for four independent experiments and are analyzed by one-way ANOVA followed by t-test. Differences were considered significant at **P < 0.01.)

Aβ_1-42 Peptide Induces Senescence in RPE Cells

Hallmarks of senescent cells include an essentially irreversible growth arrest, expression of SA-β-Gal and p16^INK4a, and/or persistent DDR (DNA damage response) signaling. ⁴³ To determine whether OAβ could induce senescence, RPE cells were treated with OAβ (0.3 μM), and the markers of cellular senescence were evaluated. RPE cells incubated with OAβ (0.3 μM) displayed higher frequency of SA-β-Gal staining (4 days: 37.32% ± 7.12%; 7 days: 62.31% ± 8.82%) than cells in control (5.24% ± 1.98%) (Fig. 3A). Senescent cells are irreversibly cell cycle arrested mainly in G1 phase. ⁴⁴ To analyze senescence-associated cell cycle arrest, the cell cycle profiles of Aβ-treated and untreated RPE cells were analyzed by flow cytometry (Fig. 3B, left), and the percentage of cells at G1, S, and G2/M phases was calculated and presented in the bar chart (Fig. 3B, right). The percentage of cells at G0G1 phase was significantly increased in cells incubated with OAβ (4 days: 75.9%; 7 days: 89.02%) compared with the control (57.25%). This result suggested that cell cycle of OAβ-stimulated cells was arrested at the G0G1 phase. Since flow cytometry analysis suggested that the cell cycle of Aβ-treated cells was delayed in G1 phase, to explore the potential molecular genetic alterations associated with Aβ and growth arrest, we therefore examined the protein levels of p16^INK4a and phospho-ATM, both of which are biomarkers of senescence ^45,46 (Fig. 3C). Incubation with OAβ for 4 and 7 days led to a 1.8- and 2.4-fold increase in expression of p16^INK4a compared with control. No differences in levels of phospho-ATM were detected between the Aβ-treated cells and the control. The result demonstrated that Aβ-induced growth arrest was regulated by p16^INK4a pathway but independent of activation of ATM, which is a signal of DNA damage response. ⁴⁷  

Figure 3

 

Chronic exposure to OAβ induced RPE senescence. (A) RPE cells were treated with OAβ (0.3 μM) for 4 or 7 days, the SA-β-Gal staining was screened by microscopy ([A], above) and the degree of cell senescence was quantified as percentage of SA-β-Gal positive cells ([A], below). After 4 and 7 days of Aβ exposure, approximately 37.32% ± 7.12% and 62.31% ± 8.82% of the cells displayed positive staining of SA-β-Gal. (B) Cell cycle profiles of Aβ-treated and untreated RPE cells were analyzed by flow cytometry ([B], left) and the percentage of cells at G1, S, and G2/M phases was quantified ([B], right). Incubation with OAβ induced an increase in the percentage of cells at G0G1 phase as well as a dramatic decrease in the S phase. (C) Expressions of phospho-ATM and p16^INK4a were analyzed by Western blot ([C], above) and were quantified ([C], below). Incubation with OAβ for 4 and 7 days led to a 1.82- and 2.45-fold increase in expression of p16^INK4a compared with that in the control cells. (Results are mean ± SD values for four independent experiments and are analyzed by one-way ANOVA followed by t-test. Differences were considered significant at **P < 0.01, *P < 0.05.)

Senescent RPE Cells Produce High Levels of MMP-9 and IL-8

Unlike apoptotic cells, senescent cells remain metabolically active and could produce a proinflammatory microenvironment for as long as they persist. ²⁰ Therefore, it is important to characterize the secretory pattern that is produced during cellular senescence. RPE monolayers cultured on Transwell filters were stimulated with OAβ (0.3 μM, 7 days) in the upper and lower compartments to trigger senescence. Expression of IL-8 was analyzed using RT-PCR and ELISA. ELISA (Fig. 4A) revealed that a slight secretion of IL-8 at both apical and basal sides was found in control cells (apical: 92 ± 41 pg/mL; basal: 83 ± 36 pg/mL), stimulation with OAβ significantly increased the secretion level of IL-8 (apical: 1875 ± 243 pg/mL; basal: 485 ± 116 pg/mL), statistical analysis revealed that senescent RPE cells secreted IL-8 in a polarized fashion toward the apical side. PCR (Fig. 4B) revealed that IL-8 expressions in senescent cells were significantly higher (10.3 ± 1.8-fold) than that in control (1.0 ± 0.5-fold). 

Figure 4

 

Expressions of IL-8 and MMP-9 in senescent RPE cells. Senescence was induced by continuous exposure of RPE cells to OAβ (0.3 μM, 7 days) in both compartments of the RPE monolayer. (A) The amount of IL-8 in the upper or lower compartment was measured. ELISA analysis showed that RPE cells secreted IL-8 in a polarized fashion toward the apical side. (B) RT-PCR exposed that mRNA expression of IL-8 was elevated to 10.3 ± 1.8-fold in senescent RPE cells compared with control. (C) Supernatants from both compartments were collected and total MMP-2/-9 activities were assessed by gelatin zymography and the band intensity values were calculated. (D) Secretion level of MMP-2 was quantified by calculating the band intensity. No significant difference in MMP-2 secretion was detected between the control cells and senescent cells. (E) Secretion level of MMP-9 was quantified by calculating the band intensity. The secretion levels of pro-MMP-9 and active-MMP-9 in senescent cells were 2.36-fold and 6.36-fold higher than that in controls. (Results are mean ± SD values for four independent experiments and are analyzed by one-way ANOVA followed by t-test. Differences were considered significant at **P < 0.01, *P < 0.05.)

Gelatin zymography exposed that MMP secretion from control cells was characterized by low levels of pro-MMP-9, absent levels of active-MMP-9, considerable output of pro-MMP-2, and absence of active-MMP-2 (Fig. 4C). The secreted levels of pro-MMP-9 (2.36-fold) and active-MMP-9 (6.36-fold) in senescent cells were significantly higher than those of control cells (Fig. 4E), but no significant changes were observed in pro-MMP-2 or activated MMP-2 production between senescent cells and control cells (Fig. 4D). 

These results suggested that OAβ-induced senescent RPE cells produced higher concentrations of IL-8 in a polarized fashion toward the apical side. In addition, the result demonstrated that senescent cells could release higher concentrations of pro-MMP-9 and active-MMP-9. 

MMP-9 Alters TJ Distribution in RPE Cells

RPE cells transfected with MMP-9 siRNA or siRNA-N were treated with or without OAβ; neither siRNA transfection nor OAβ treatment induced significant LDH release from RPE cells (Fig. 5A), this finding showed that cell viability was not affected under these conditions. RT-PCR (Fig. 5B) and Western blot (Fig. 5C) exposed that MMP-9 expression in MMP-9–silenced cells was significantly lower than that in controls. Compared with controls, transfection of RPE cells with siRNA-N did not affect the expression of MMP-9 in mRNA or protein level. In MMP-9–silenced RPE cells, OAβ treatment did not increase MMP-9 expression. In contrast, OAβ significantly increased MMP-9 expression in siRNA-N–transfected RPE cells. 

Figure 5

 

Effects of MMP-9 knockdown on cell viability and MMP-9 expression. RPE cells were transfected with MMP-9 siRNA or negative siRNA (25 nM, 24 hours), then the siRNA-transfected RPE cells were exposed to OAβ (0.3 μM, 7 days). (A) The cytotoxicity of siRNA transfection was evaluated using the LDH release colorimetric method. Neither siRNA transfection nor OAβ treatment induced significant LDH release from RPE cells. (B) RT-PCR analysis on mRNA expression level of MMP-9. (C) Western-blot analysis on protein level of MMP-9. The mRNA and protein levels of MMP-9 in MMP-9-silenced cells were significantly lower than that in controls. Compared with control, siRNA-N did not affect the expression of MMP-9 in mRNA or protein level. (Results are mean ± SD values for four independent experiments and are analyzed by one-way ANOVA followed by t-test. Differences were considered significant at **P < 0.01, *P < 0.05.)

In the control cells, actin filaments were regular with no breaks in the staining pattern, and the distribution of occludin, ZO-1, and claudin-19 was continuous around the cells (Fig. 6A). Cells that were exposed to OAβ showed marked irregularity and discontinuity in the distribution of occludin, ZO-1, claudin-19, and F-actin (Fig. 6B). It showed that OAβ greatly modified the cell morphology and location of TJs. Silence of MMP-9 significantly attenuated Aβ-induced dislocation of occludin and ZO-1, but slightly reversed Aβ-induced dislocation of claudin-19 and rearrangement of F-actin (Fig. 6C). Transfection with negative siRNA did not reverse Aβ-induced dislocation of TJs and F-actin. These results suggested that MMP-9 secreted by senescent cells may be related to the blood-retinal barrier (BRB) breakdown by disrupting barrier integrity of RPE cells. 

Figure 6

 

MMP-9 secreted by senescent RPE cells alters TJ distribution. RPE cells were transfected with MMP-9 siRNA to evaluate the effect of MMP-9 on junctional integrity of RPE cells. RPE senescence was induced by continues exposure to 0.3 μM Aβ for 7 days. Live cell images and locations of ZO-1, occludin, claudin-19, and F-actin were detected in control cells (A), senescent cells (B), MMP-9–silenced senescent cells (C), and negative siRNA-transfected senescent cells (D). Light microcopy showed that senescent cells (B) and negative siRNA-transfected senescent cells (D) displayed irregular morphology compared with control (A), but inhibition of MMP-9 by RNAi avoided the morphological alteration in senescent cells (C). In control cells, immunostaining of occludin, ZO-1, and claudin-19 showed that the localization of these TJs perfectly matched the typical cobblestonelike morphology of native RPE cells. F-actin staining was continuous without interruption (A), whereas it showed not only disorganization of the TJs, but also disappearance of typical cobblestonelike morphology in senescent cells (B) and negative siRNA-transfected senescent cells (D). Silence of MMP-9 significantly attenuated Aβ-induced dislocation of ZO-1 and occludin, but slightly reversed Aβ-induced dislocation of claudin-19 and rearrangement of F-actin. (Light microcopy: Magnification: ×400. Scale bar: 50 μm. Confocal microcopy of F-actin, occluding, and ZO-1: Magnification: ×630. Scale bar: 25 μm.)

MMP-9 Degrades Occludin and ZO-1 in RPE Cells

MMP-9 has been reported to degrade occludin, ZO-1, and claudins, ^32–34 therefore the protein levels of the TJs and F-actin were measured using Western blot (Fig. 7A). We found that occludin and ZO-1, not claudin-19 and F-actin, were significantly decreased in OAβ-stimulated cells compared with the control and that inhibition of MMP-9 by siRNA reversed the loss of occludin and ZO-1. Transfection with siRNA-N did not attenuate OAβ-induced loss of occludin and ZO-1. 

Figure 7

 

MMP-9 is associated with loss of TJs and RPE barrier dysfunction. (A) The expressions of ZO-1, occludin, claudin-19, and F-actin were evaluated by Western blot. OAβ induced a significant reduction of occludin and ZO-1. In OAβ-induced senescent cells, inhibition of MMP-9 by siRNA abrogated OAβ-induced loss of occludin and ZO-1, whereas negative siRNA transfection could not attenuate OAβ-induced reduction of occludin and ZO-1. (B) Evaluation of TER in control (opened square), senescent cells (filled square), MMP-9–silenced senescent cells (filled triangle), and negative siRNA-transfected cells (opened triangle). Cells were incubated with OAβ to induce senescence. The OAβ-induced senescent cells displayed a significant decrease of TER at day 18 and thereafter, but inhibition of MMP-9 by RNAi partially prevented Aβ-induced decrease of TER. In contrast, siRNA-N transfection did not abrogate OAβ-induced decrease of TER. (C) Evaluation of epithelial permeability by measuring the passive permeation of FITC-dextran. In accordance with the result of TER, senescence induced a significant increase of FITC-dextran flux compared with control. Silence of MMP-9 inhibited OAβ-induced high permeability, whereas transfection with siRNA-N could not ameliorate OAβ-induced high permeability. (Results are mean ± SD values for four independent experiments and are analyzed by one-way ANOVA followed by t-test. Differences were considered significant at **P < 0.01, *P < 0.05.)

MMP-9 Affects Epithelial Barrier Function

The further study was performed to examine whether the MMP-9–mediated alterations of TJ proteins was associated with a compromised barrier function following induction of RPE senescence. The TER (Fig. 7B) was recorded to determine the stability of TJs, and the transepithelial diffusion rate of FITC-dextran (Fig. 7C) was measured to evaluate the permeability of the monolayers. The result revealed that the TER increased rapidly during the initial 12 days of standard culture and reached a plateau within the following 3 days (Fig. 7B). Then the monolayer was incubated with Aβ (0.3 μM, 7 days) to induce senescence, and the TER was measured every day. It was found that Aβ gradually decreased the TER, and significantly decreased the TER on day 18 and thereafter, but silence of MMP-9 partially reversed this effect. The permeability assays were performed with cells cultured on day 22 under the same conditions as those used for the TER measurements (Fig. 7C). A significantly increased diffusion rate of FITC-dextran was observed in Aβ-induced senescent cells, but silence of MMP-9 significantly inhibited Aβ-induced high permeability. In contrast, transfection of RPE cells with negative siRNA did not attenuated OAβ-induced low TER and high permeability (Figs. 7B, 7C). This result suggested that inhibition of MMP-9 activation partially protected RPE cells from senescence-associated barrier dysfunction. 

Detection of IL-8 Cleavage Products

MMP-9 could regulate chemoattractant activities of chemokines. ³⁵ It has been reported that shorter forms of IL-8 are more active in neutrophil activation than the full-length form. ³⁶ Therefore, secreted IL-8 from medium supernatants was collected, concentrated, and evaluated by Western blot (Fig. 8). In control cells, the two protein variants IL-8(1-77) and IL-8(6-77) were separated by SDS-PAGE. OAβ treatment induced a conversion of IL-8(1-77) to a shorter form, identified as IL-8(7-77). Silence of MMP-9 inhibited Aβ-induced IL-8 truncation, whereas transfection with negative siRNA did not inhibit Aβ-induced IL-8 truncation. This result exposed that MMP-9 produced by senescent cells could truncate or process intact IL-8 to a shorter form. 

Figure 8

 

MMP-9 processes intact IL-8 to a shorter form. Different lengths of secreted IL-8 were detected by Western blot. In control cells, the two protein variants IL-8(1-77) and IL-8(6-77) were separated by SDS-PAGE. OAβ treatment induced a conversion of IL-8(1-77) to a shorter form, identified as IL-8(7-77). Silence of MMP-9 before Aβ treatment significantly blocked the appearance of the truncated IL-8 form. However, transfection with negative siRNA did not inhibit Aβ-induced IL-8 truncation. (S indicates relative molecular mass standard. IL-8(1-77): full length of IL-8; IL-8(6-77): the most prominent form of IL-8; IL-8(7-77): the truncated form of IL-8.)

Discussion

The main results of the present study are as follows: (1) Aβ_1–42 induced RPE cell senescence, (2) senescent RPE cells produced higher concentrations of MMP-9 and IL-8, (3) activation of MMP-9 was associated with barrier dysfunction in senescent cells, and (4) MMP-9 could truncate intact IL-8 to a shorter form. 

When cells encounter stress, they undergo either apoptosis or senescence. Using two methods, including measurement of LDH release (Fig. 2A) and TUNEL staining (Fig. 2B), we found that exposure of RPE cells to 0.3 μM OAβ for 4 or 7 days did not induce cell death or apoptosis. This is consistent with the recent study, showing that OAβ (10 μM, 24 hours) is not cytotoxic to ARPE-19 cells. ²⁷ In contrast, some studies suggested that OAβ could induce cell apoptosis. ⁴⁸ However, the mechanism by which OAβ modulates the fate of the cells, such as apoptosis or senescence, is largely unknown. 

Persistent, sublethal oxidative stress has been shown to induce RPE cell senescence. ¹⁵ Therefore, we asked whether chronic exposure of RPE cells to subtoxic OAβ could induce cell senescence. SA-β-Gal staining was used to identify replicative senescence in cultured RPE cells. ¹¹ Exposure of astrocytes to OAβ_1–42 increased the percentage of β-Gal–positive cells in vitro. ³⁰ Similarly, Aβ peptides induced increased β-Gal activities in endothelial cells in vitro and in vivo in the zebrafish model. ⁴⁹ The present study demonstrated that chronic exposure to OAβ_1–42 induced increased SA-β-Gal staining in RPE cells (Fig. 3A). It has been shown that oxidative stress leads to an increase in SA-β-Gal staining in cultured RPE cells. ¹⁵ Recent studies investigated the in vitro effects of Aβ_1-42 in RPE cells and found that it increased the production of reactive oxygen species (ROS) and mitochondrial dysfunction. ²⁷ During senescence, free radical levels increase and these reactive species damage mitochondria. ⁵⁰ We thus supposed that Aβ may induce oxidative stress to trigger the senescent program in RPE cells. Sustained oxidative stress leads to activation of p16^INK4a, which has been characterized for its ability to decelerate cell progression from G1 to S phase. ⁵¹ In the present study, OAβ induced an increase in the percentage of cells at G0G1 phase as well as a dramatic decrease in the S phase and accompanying increased expression of p16^INK4a (Figs. 3B, 3C). The results suggested that Aβ-induced cell cycle arrest or senescence was regulated by p16^INK4a. Consistent with our results, a similar increase in p16^INK4a expression after exposure to Aβ has been reported in human astrocytes. ³⁰ p16^INK4a expression was also increased in replicative senescent RPE cells. ⁵²  

The present study observed that expressions of IL-8 and MMP-9 were increased in Aβ-induced senescent cells (Fig. 4). Altered secretory pattern in senescent cells is a possible source of age-related inflammation. In the absence of stimuli, RPE cells secreted lower levels of IL-8 into the basal and apical bath. In Aβ-induced senescent cells, secretion of significant quantities of IL-8 preferentially into the apical bath may provide signaling in vivo to attract immune cells across the RPE to induce chronic inflammation (Fig. 4A). The preferential secretion of IL-8 into the apical side creates a chemotactic gradient and is responsible for the migration of neutrophils across the RPE monolayer to the compartment with the highest IL-8 levels. ⁵³ Polarized secretion of IL-8 has been observed in other epithelia. Preferential apical secretion was shown in mesothelial and female reproductive tract epithelia. ^53,54 In contrast, renal and colonic epithelia show a preferential basolateral secretion, ^55,56 whereas controversial results were obtained in RPE cells. ^57,58 The present study demonstrated that senescent RPE cells showed a preferential apical secretion of IL-8, which is consistent with the recent study, showing that stimulation of RPE cells with inflammatory cytokine mixture induced a marked secretion of IL-8 mainly across the RPE apical membrane. ⁵⁸ Higher concentrations of MMP-9 have been detected in aqueous humour and plasma of AMD patients. ^59,60 MMP-9 has been demonstrated as a biomarker of subretinal fluid in AMD. ⁶¹ There is evidence suggesting MMP-9 is associated with components of the BrM-RPE complex and is secreted by RPE cells, and secretion of MMP-9 may result in defects in BrM. ⁶² The present study suggested that Aβ-induced senescent RPE cells may be a source of MMP-9 production (Figs. 4C–E). In other studies, it has been demonstrated that Aβ (5 μM, 24 hours)-induced MMP-9 secretion was mediated by RAGE (the receptor for advanced glycation end products) through intracellular Ca²⁺-calcineurin signaling. ⁶³ However, we confirmed that a low concentration of 0.3 μM Aβ_1–42 (7 days) also induced MMP-9 secretion (Fig. 4) and accompanying structural alteration in TJs (Fig. 6). Actually, pro-MMPs are kept in a catalytically inactive state by the interaction between the thiol group of a pro-domain cysteine residue and the zinc ion of the catalytic site. They are converted to active proteinases by oxidants through oxidation of the pro-domain thiol group. ⁶⁴ In senescent cells, defective mitochondria are known to produce increased amounts of ROS. ⁶⁵ In addition, Aβ_1–42 could induce oxidative stress in RPE cells. ²⁷ We supposed that Aβ-induced MMP-9 activation might be partially mediated by chronic oxidative stress. Our study found that Aβ_1–42 could not induce MMP-2 activation, which is consistent with the study, suggesting that Aβ_1–42 only induced MMP-9 activation, but not MMP-2. ⁶⁶ Another study suggested that oligomeric Aβ_1–42 downregulates MMP-2 expression. ⁶⁷ However, in murine endothelial cell line, Aβ_1–42 monomers have been reported to increase MMP-2 and MMP-9 expression. ⁶³ However, the mechanism by which Aβ_1–42 regulates MMP-2 or MMP-9 activation is largely unknown. 

The role of MMP-9 on choroidal neovascularization of AMD has been elucidated. ^22,68,69 However, recent studies suggested that MMP-9 contributes to inflammatory processes by regulating physical barriers, modulating inflammatory mediators, such as cytokines and chemokines, and establishing chemokine gradients in tissues that regulate the movement of leukocytes at sites of injury. ⁷⁰ In the present study, we explored the two roles of MMP-9 in inflammation: (1) MMP-9 regulates epithelial integrity via proteolysis of occludin and ZO-1. The fact that silence of MMP-9 attenuated Aβ-induced disruption of epithelial structure (Fig. 6) and barrier dysfunction (Figs. 7B, 7C) suggests that the breakdown of epithelial barrier integrity involved an MMP-9–dependent cleavage of occludin and ZO-1 (Fig. 7A). The barrier functional integrity of RPE monolayer is dependent on the structures of TJs, including the transmembrane protein occludin, claudins, and junctional adhesion molecules. Zonula occludens (ZO-1, ZO-2, and ZO-3) interact with the transmembrane proteins at the cytoplasmic face of the cell membrane and serve to anchor them to the actin cytoskeleton. Occludin consists of four transmembrane domains, intracellular N- and C-termini, and two extracellular domains that might interact with cell membranes of vicinal cells, thus sealing the intercellular clefts. ⁷¹ The first extracellular loop displayed type IV collagenase–sensitive motives, implying that occludin cleavage might be caused by MMP-9. ⁷² Although ZO-1 and claudins have been reported to be substrates of MMP-9, ^73,74 the exact cleavage site remains to be determined. Our results are consistent with recent studies, MMP-9 disrupted barrier integrity of airway epithelial, RPE, and corneal epithelial via proteolysis of occludin and ZO-1. ^32,34,75 In our study, claudin-19 dislocation under the impact of Aβ treatment was slightly attenuated by silence of MMP-9 (Fig. 6) and MMP-9–dependent cleavage of claudin-19 was not detected (Fig. 7A). The protein level of F-actin was also not altered in the experiments (Fig. 7A). Therefore, the alterations in claudin-19 location and F-actin rearrangement may be independent of MMP-9. However, MMP–dependent claudin-5 cleavage has been reported in other cell types. ^74,76 The results may be due to the different cell types and stimuli. In addition, the activation of cAMP signaling pathway was significantly weaker in senescent cells. ⁷⁷ Based on recent studies, ^78,79 we supposed that a possible reduced cAMP level might contribute to dislocation of claudin-19 and rearrangement of F-actin. (2) MMP-9 truncated the secreted IL-8 into a shorter form (Fig. 8). The best-known function for MMPs is the degradation of extracellular matrix components; however, other substrates are also degraded by MMPs. CXC chemokines, such as IL-8, are substrates for activated MMP-9: MMP-9 modifies the biological activity of IL-8 by proteolytic processing. MMP-9 truncates the full-length form of IL-8 into shorter forms of IL-8; the truncated shorter forms of IL-8 are more active than full-length IL-8 in recruitment and activation of neutrophils. ³⁶ Therefore, we concluded that MMP-9 secreted by senescent cells may accelerate chronic inflammation via disruption of epithelial barrier integrity and promoting IL-8 to the more activated form. 

Aβ deposition, detected in drusen of AMD patients ^4,5 and in photoreceptor outer segments of aged individuals, ⁷ may trigger inflammatory response in the RPE/choroidal layers of the eye. In addition, it has been reported that oligomeric Aβ staining is observed in cytoplasma of RPE cells. ³⁷ Interestingly, elevated Aβ production has been found in senescent RPE cells. ⁸⁰ Our study found that OAβ_1–42 could induce RPE senescence and altered secretory behavior. Expressions of IL-8 and MMP-9 by Aβ-induced senescent RPE cells may accelerate continuous chronic inflammation in AMD for the following reasons (Fig. 9): (1) The retina is an immune-privileged site where inflammatory responses are suppressed, but opening of epithelial barriers by MMP-9 activity may be a mechanism that allows passage of plasma proteins and inflammatory cells into this privileged compartment. (2) The aqueous humour of AMD patients contains higher concentrations of IL-8, ⁵⁹ which can induce accumulation of immune cells into local tissue via recruitment of neutrophils and natural killer (NK) cells. ⁸¹ It has been reported that chemokines produced by the senescent cells could recruit NK cells to clear senescent cells. ⁸² (3) However, immune clearance is not 100% efficient or the rate at which senescent cells are produced outpaces the rate of clearance. Therefore, continued presence (sometimes over many years) of proinflammatory factors and accumulation of immune cells in the retina may cause chronic inflammation. (4) Immune cells and chronic inflammation may accelerate tissue damage. Neutrophils are phagocytic cells characterized by a segmented lobular nucleus and cytoplasmic granules filled with degradative enzymes, such as metalloproteinase-9, ⁸³ which mediate barrier disruption via degradation of type IV collagen–contained basement membrane and tight junctional proteins. In addition, they can convert molecular oxygen into ROS, which are highly reactive oxidizing agents that destroy tissue cells. ⁸⁴ NK cells have granules that contain a protein called perforin, which creates pores in targeted cell membranes, and enzymes called granzymes, which enter through target cell membranes and induce apoptosis of cells. ⁸⁵  

Figure 9

 

MMP-9 secreted by Aβ-induced senescent cells disrupted RPE barrier structure and contributed to chronic inflammation. Aβ could induce RPE senescence. Senescent RPE cells secrete MMP-9 and chemokines IL-8. MMP-9 contributed to outer Blood-Retinal Barrier (oBRB) breakdown via proteolysis of epithelial tight junction proteins. IL-8 was required for recruitment of immune cells, such as NK cells and neutrophils. All these may contribute to continuous chronic inflammation in the retina.

Together with these observations, our present study suggested that release of MMP-9 by Aβ-induced senescent cells may accelerate chronic inflammation via disruption of epithelial barrier integrity and promoting IL-8 activities. Maintenance of the structural integrity in the RPE monolayers by blocking the action of Aβ or Aβ-induced senescence may thus represent a new approach to the treatment of AMD. 

Supplementary Materials

pdf S1 

Acknowledgments

The authors thank the Biochemistry and Molecular Biology Institute of Shanghai Tenth People's Hospital for technological support, Xiaoqing Liu's Laboratory in Tongji University for experiment facilities and equipment, and Xiujuan Shi for excellent technical assistance. 

Supported by Science and Technology Commission of Shanghai (11JC1409900) and National High Technology Research and Development Program of China (863 Program, S2010GR0002). The authors alone are responsible for the content and writing of the paper. 

Disclosure: L. Cao, None; H. Wang, None; F. Wang, None; D. Xu, None; F. Liu, None; C. Liu, None