Integrin α5β1 Mediates Attachment, Migration, and Proliferation in Human Retinal Pigment Epithelium: Relevance for Proliferative Retinal Disease

Abstract

Purpose.: The aim of this study was to determine the expression and localization of integrin α5β1 in human retinal pigment epithelium (RPE) and its ability to modulate RPE cell attachment, proliferation, migration, and F-actin cytoskeleton distribution.

Methods.: Expression and localization of α5β1 were analyzed on human RPE by immunoblot/immunofluorescence. Polarized secretion of fibronectin was measured. RPE attachments to different substrates were determined using cell attachment screening kits. BrdU incorporation and wound-healing assays were used to test hfRPE proliferation and migration. F-actin cytoskeleton was visualized with phalloidin.

Results.: Integrin α5β1 was detected in native adult and fetal human RPE. The α5-subunit is predominantly localized at the apical membrane of hfRPE, whereas the β1-subunit is uniformly detected at the apical/basolateral membranes. The authors also found that hfRPE cultures secrete significant amounts of fibronectin to the apical bath. JSM6427, a specific integrin α5β1 antagonist, significantly inhibited hfRPE cell attachment to fibronectin, but not laminin, or collagen I or IV. JSM6427 also showed a strong inhibitory effect on bFGF, PDGF-BB, and serum-induced cell migration and proliferation. Furthermore, JSM6427 induced significant disruption of the F-actin cytoskeleton of dividing RPE cells but had no effect on quiescent cells.

Conclusions.: The apical localization of α5β1 and the secretion of fibronectin to the apical bath suggest the presence of an autocrine loop that can guide the migration of RPE. The strong inhibitory effects of JSM6427 on human RPE cell attachment, proliferation, and migration is probably mediated by F-actin cytoskeletal disruption in proliferating cells and suggests a potential clinical use of this compound in proliferative retinopathies.

Integrins are a family of heterodimeric plasma membrane proteins important for cell–cell and cell–extracellular matrix interaction. ^1,2 They consist of variable α and β subunits whose combinations determine receptor specificity. Thus far, 19 integrin α and eight integrin β subunits have been identified that form at least 25 distinct α/β heterodimers. Integrins provide a link between their extracellular ligand and the cytoskeleton and help modulate various signaling pathways, including cell adhesion, migration, differentiation, angiogenesis, and wound healing. ²  

The retinal pigment epithelium (RPE) is a monolayer of hexagonal cells located between the distal retina and the choroidal blood supply and forms the outer blood-retina barrier. By participation in the visual cycle, photoreceptor outer segment phagocytosis, and transport of nutrients, ions, and fluid between the distal retina and the choriocapillaris, the RPE helps maintain the health and integrity of the distal retina. ^3–5 Studies of RPE cells have shown a wide spectrum of integrin receptor expression ^6,7 ; however, the precise localization and possible polarized distribution of these receptors are controversial or unknown. ^8,9 Proliferative vitreoretinopathy (PVR) is the most common cause of surgical failure in the treatment of rhegmatogenous retinal detachment, characterized by the formation of epiretinal and subretinal membranes on the neuroretinal surface. ^10–12 There is accumulating evidence that the formation of a PVR membrane is a chronic wound-healing process characterized by extracellular matrix (ECM) accumulation, cell adhesion, migration, and proliferation. Proliferative vitreoretinopathy-affected membranes are composed of a mixed population of cells. The RPE cell has been identified as a critical part of PVR membranes in light/electron microscopy and immunohistochemistry studies and is thought to play an important role in the development and onset of PVR. ^13–16 Detachment of RPE cells from their subretinal monolayer after retinal injury, such as retinal detachment or trauma, appears to be the crucial event in early stages of PVR. These detached RPE cells proliferate, migrate, and attach to ECM highly enriched in fibronectin. ^13,17  

Integrin α5β1 is a specific receptor of fibronectin through its arginine-glycine-aspartic acid (RGD) binding site. ² Together they play an important role in the development of the vascular system during embryogenesis. ¹⁸ Knockout of the gene encoding the α5 subunit leads to lethal vasculature and cardiac defects. ¹⁹ Similarly, knockout mice for fibronectin died early during development from a variety of defects, including an improperly formed vasculature. ²⁰ Interestingly, fibronectin and integrin α5β1 were upregulated in growth factor- or tumor-induced neovascularization, ²¹ whereas expression of integrin α5β1 was low in quiescent vascular cells. ^22,23 Robbin et al. ^24,25 showed the presence of the α5 subunit on pigmented cells (probably of RPE origin) from PVR membrane and concluded that its presence was abnormal compared with the lack of α5 staining on normal retina, consistent with recent findings. ²⁶ JSM6427 is a specific inhibitor of integrin α5β1, ²⁷ and recent studies indicate that systemic administration of JSM6427 suppressed laser-induced choroidal neovascularization in which integrin α5β1 was upregulated. ²⁸ A similar inhibitory effect was also observed in hypoxia-induced neovascularization. ²⁹ Most recently, Farber et al. ³⁰ reported the attenuation effect of this compound on glioma cell growth. 

The present experiments show that α5β1 is localized mainly to the apical membrane of primary human fetal RPE culture and that fibronectin is constitutively secreted to the apical bath, suggesting an autocrine signaling pathway that can mediate proliferative disease. We also show that the inhibitory effects of JSM6427 on dividing RPE cells are most likely mediated through F-actin, which supports a therapeutic role for this antagonist. 

Materials and Methods

Cell Culture and Reagents

The research followed the tenets of the Declaration of Helsinki and the National Institutes of Health Institutional Review Board. Fetal eyes (gestation, 16–18 weeks) were obtained from Advanced Bioscience Resources (Alameda, CA), and adult eyes were obtained from Analytical Biological Services Inc. (Wilmington, DE). Human fetal retinal pigment epithelial (hfRPE) cells were isolated and cultured as described previously. ³¹ The culture medium was changed every 3 days, and passages 1 to 2 were used for all studies. For immunofluorescence localization and fibronectin secretion experiments, cells were seeded in polyester membrane inserts (Transwell, Costar 0.4-μm pores; Corning Incorporated, Corning, NY) and were maintained for 6 to 8 weeks before experiment. 

JSM6427 (3-(2-{1-alkyl-5-[(pyridine-2-ylamino)-methyl]-pyrrolidin-3-yloxy}-acetylamino)-2-(alkyl-amino)-propionic acid), a selective integrin α5β1 antagonist, ²⁷ and JSM8009 (3-(4-(3-arylureido)phenoxy)butanamido)-3-arylic-propanoic acid), an inactive control compound, were supplied by Jerini AG Company. 

Immunoblot Analysis

Confluent monolayers of primary cultures of hfRPE (6–8 weeks) and native human adult RPE cells were lysed using lysis buffer (RIPA; Sigma-Aldrich, St. Louis, MO) supplemented with proteinase inhibitor cocktail (Roche, Indianapolis, IN). Cell lysate was centrifuged at 14,000g for 10 minutes, and the supernatant was collected. Protein concentration was determined using protein assay (BCA; Pierce Biotechnology, Rockford, IL). Protein (8–30 μg) was electrophoresed on a 4% to 12% Bis-Tris gradient gel (NuPAGE; Invitrogen, Carlsbad, CA) under nonreduced conditions and was electroblotted onto the nitrocellular membranes (XCell II Blot Module; Invitrogen). Nonspecific binding sites were blocked (StartingBlock T20 [TBS]; Pierce Biotechnology), and membranes were probed with rabbit anti-integrin α5 polyclonal antibody (catalog no. AB1928), mouse anti-human integrin β1 monoclonal antibody (clone LM534), and mouse anti-human integrin α5β1 monoclonal antibody (clone JBS5; all from Chemicon International Inc., Temecula, CA). Membranes were then incubated with horseradish peroxidase (HRP)-conjugated secondary antibody (Pierce Biotechnology), developed with extended-duration substrate (Supersignal; West Dura; Pierce Biotechnology), and imaged (Autochemie; UVP, Upland, CA). 

Immunofluorescence Localization

For localization experiments, primary antibodies against integrin α5, β1 subunits, α5β1, or ZO-1 (Invitrogen) were fluorescently labeled with fluorochrome technology (Zenon; Invitrogen) according to the manufacturer's instructions. Primary cultures (6–8 weeks) of hfRPE monolayers on polyester membrane inserts (Transwell; Corning) were fixed with 4% formaldehyde (Ted Pella Inc., Redding, CA), permeabilized for 10 minutes with 0.2% Triton X-100 (Sigma-Aldrich), and blocked with a signal enhancer (Image-iT FX; Invitrogen). RPE monolayers were incubated with antibodies prelabeled with fluorophores, and normal mouse serum or rabbit serum was used as the negative control. Samples were mounted on glass slides with antifade reagent containing DAPI (Prolong Gold; Invitrogen) and were imaged with a microscope (Axioplan 2; Carl Zeiss, Oberkochen, Germany) with slider module (ApoTome; Carl Zeiss) and software (Axiovision 3.4; Carl Zeiss). 

Attachment Assay

Cell attachment assays were performed with a screening kit (CytoMatrix; Chemicon International Inc.) according to the manufacturer's instructions. Briefly, 96-well plates were coated with fibronectin, vitronectin, laminin, collagen I, collagen IV, or BSA (negative assay control for each plate). Subconfluent hfRPE cells were pretreated with integrin α5β1 antagonist JSM6427 (0.1, 0.4, 1.6, 6.3, 25, 100 μM) or inactive control compound JSM8009 for 16 hours. Cell were collected using trypsin and resuspended in serum-free medium (SFM; MEM-α modified medium; Sigma-Aldrich) containing nonessential amino acids (Sigma-Aldrich) and glutamine-penicillin-streptomycin (2 × 10⁵ cells/mL; Invitrogen) containing JSM6427 or JSM8009. Cells were seeded to precoated 96-well plates (100 μL/well) and put back to CO₂ incubator (37°C) for 60 minutes. Plates were washed with PBS containing Ca²⁺/Mg²⁺, and 0.2% crystal violet in 10% ethanol (100 μL/well) was added and incubated for 5 minutes at room temperature. After washing with PBS, solubilization buffer (50/50 mixture of 0.1 M NaH₂PO₄, pH 4.5 and 50% ethanol) was added (100 μL/well). Plates were placed on a shaker for 10 minutes until the cell-bound stain was completely solubilized. Absorbance of the cells was measured at 540 to 570 nm with a spectrophotometric microplate reader (Safire 3; Tecan Trading AG, Männendorf, Switzerland). 

Bromodeoxyuridine Incorporation Assay

Before the experiments, 96-well tissue culture plates (Primaria; Sigma-Aldrich) were coated with 10 μg/mL fibronectin. RPE cells were seeded in tissue culture plates (2.5 × 10³/well) for 24 hours and serum starved in SFM for another 16 hours. Cells were then treated with different concentrations of JSM6427 or inactive control compound JSM8009. SFM was used as the negative control, and SFM supplemented with 5% serum was used as positive control. After 48 hours of treatment, RPE cells were incubated with BrdU for another 24 hours. The proliferation rate was evaluated using a cell proliferation ELISA BrdU Kit (Roche). Similar experiments were performed using noncoated tissue culture plates as described. Quadruplicates were used for each condition, and the experiments were repeated using cell cultures from two donors. 

Wound Healing Assay

The wound healing assay was used to study the effects of integrin α5β1 antagonist on hfRPE cell migration, as described previously. ³² hfRPE cells (100 × 10³ cells/well) were seeded into fibronectin-coated or -uncoated 24-well tissue culture plates (Primaria; Sigma-Aldrich) and were grown for 4 weeks to confluence. Cell proliferation was suppressed by incubation with 10 μg/mL mitomycin C (Sigma-Aldrich) for 2 hours before all the experiments. A circular denuded area (7-mm diameter) was made in each well using a custom-designed cell scraper. Cells were treated with JSM6427 or inactive control compound JSM8009 (100, 400 μM) for 48 hours; cells were fixed in cold methanol and stained with ethidium homodimer-1 (EthD-1; Invitrogen). To test the effects of JSM6427 on growth factor-induced cell migration, cells were treated with EGF, PDGF-BB, bFGF, or a combination of growth factors and JSM6427 (50 μM) in SFM condition. Cell migration was quantitated by counting the average number of cells that migrated into the denuded area in 16 microscope fields surrounding the circumference of the denuded area. Each condition was tested in triplicate and was repeated using cells from different donors. Cell viability was evaluated using a cell viability assay (Live/Dead Viability/Cytotoxicity Kit; Invitrogen). 

F-Actin Cytoskeleton Staining

Confluent monolayers of cells grown in polyester membrane inserts (Transwell; Corning) for 6 to 8 weeks were scratched with 1-mL pipette tips, washed, and grown for 24 hours. Cells were then treated with integrin α5β1 antagonist JSM6427 or inactive compound JSM8009 (100, 400 μM) for another 24 hours. Cells were fixed with 4% formaldehyde, permeabilized for 10 minutes with 0.2% Triton X-100, and blocked with a signal enhancer (Image-iT FX; Invitrogen). Cells were then stained with Alexa Fluor 488-conjugated phalloidin (Invitrogen) and Alexa Fluor 555-conjugated mouse anti-human ZO-1 antibody. Samples were mounted using antifade reagent containing DAPI (Prolong Gold; Invitrogen) and were imaged with a fluorescence microscope (Axioplan 2; Carl Zeiss) with a slider module (ApoTome; Carl Zeiss) and software (Axiovision 3.4; Carl Zeiss). 

Fibronectin Secretion

Confluent monolayers of hfRPE cultured in polyester membrane inserts (Transwell; Corning) for 6 to 8 weeks were washed with SFM, incubated overnight, and replaced with fresh SFM. Cells were cultured for another 24 hours, and supernatants from both apical and basal compartments were collected. Fibronectin secretion level was assayed using a commercial technology (SearchLight; Pierce Biotechnology), as described previously. ³³ Quadruplicates were used, and the final concentrations were adjusted to normalize for the volume difference in apical and basal compartments of the polyester membrane inserts (Transwell; Corning). 

Statistical Analysis

Data are expressed as mean ± SEM; statistical significance (Student's t-test, unpaired, two-tailed) was accepted as P < 0.05. 

Results

Localization of Integrin α5β1 and Fibronectin Secretion in hfRPE

Integrin α5β1 subunits were detected in native human adult RPE and native and cultured hfRPE by microarray analysis (Wang F, et al. IOVS 2006;47:ARVO E-Abstract 2855). In Figure 1, the protein expression of α5, β1 subunits was further confirmed by immunoblots, which show antibody-specific bands of 140 kDa (α5) and 110 kDa (β1). Figure 2 shows the immunofluorescence staining of α5, β1 subunits on hfRPE. Nuclei are stained with DAPI (blue), the tight junction marker ZO-1 is stained red (Fig. 2A) or green (Figs. 2B, 2C), the α5 subunits are stained green (Fig. 2A), the β1 subunits are stained red (Fig. 2B), and α5β1 subunits are stained red (Fig. 2C). The middle part of each panel is an en face view of the monolayer, shown as a maximum intensity projection through the Z-axis. It also shows a uniform hexagonal pattern of ZO-1, typical of epithelial cells. Integrin α5β1 appear as punctuate staining visible throughout the cells. The top and right sides of each panel show a cross-section through the Z-plane. In these cross-sections, ZO-1 serves as a tight junction marker separating the apical and basolateral sides of the epithelial cells. Nuclei (blue) are located close to the basal side and serve as a marker to help define basal localization. High-gain images of the cross-section through the Z-plane are shown at the top of each panel. Integrin α5 subunit (Fig. 2A) was detected mainly on the apical side, though some expression can be detected at the basolateral membrane. In contrast, integrin β1 subunit was uniformly detected at both the apical and the basolateral membranes. In separate experiments using another antibody that targets α5β1, localization was detected mainly at the apical membrane, consistent with the localization of α5 (Fig. 2C). Fibronectin is a specific ligand for this receptor, and, as shown in Figure 3, intact monolayers of hfRPE constitutively secrete significant amounts of fibronectin to the apical bath (1.1 μg/mL). This combination suggests a possible autocrine signaling pathway at the apical membrane. 

Figure 1.

 

Western blot analysis identifying constitutive expression of integrin α5β1 in human RPE. 10 μg (native adult RPE) or 30 μg (hfRPE) protein was loaded and electrophoresed. In each sample, prominent antibody-specific bands (arrow) for α5, β1 subunits are shown at approximately 140 and 110 kDa, respectively.

Figure 2.

 

Immunofluorescence localization of integrin α5β1 in hfRPE. Central part of each panel is an en face view of a cell culture monolayer shown as a maximum intensity projection through the Z-axis. (A–C, top and right) Cross-section through the Z-plane of each panel. Nuclei were stained with DAPI (blue) and ZO-1 tight junction marker (A, red; B, C, green). High-gain images of the cross-section through the Z-plane are shown at the top of each panel. Integrin α5 (A, green) was mainly detected on the apical side, and integrin β1 (B, red) was detected at both the apical and the basolateral membranes. Anti-α5β1 antibody staining (C, red) shows that integrin α5β1 was detected mainly on the apical membrane.

Figure 3.

 

Fibronectin secretion in hfRPE. Array technology was used to detect the secretion of fibronectin to apical and basal baths. Final concentrations were adjusted to normalize for the volume difference in apical and basal baths of the polyester membrane insert assembly. Secretion to apical bath ≈ 450-fold greater than to basal bath (n = 4).

Effect of JSM6427, a Specific α5β1 Antagonist, on RPE Attachment to Different Extracellular Matrixes

The data summarized in Figure 4B show that this antagonist significantly inhibits RPE cell attachment to fibronectin in the range from ≈2 to 100 μM (P < 0.05); this inhibitory effect is monotonic with concentration. In contrast, an inactive form of this inhibitor (JSM8009) showed no effect over the entire range tested from 0.1 to 100 μM (Figs. 4C, 4D). 

Figure 4.

 

Effects of integrin α5β1 antagonist on hfRPE attachment to fibronectin, vitronectin, laminin, collagen I, and collagen IV. (A, B) Cells treated with various concentrations of JSM6427 (0.1–100 μM) or (C, D) inactive control compound JSM8009. BSA-coated wells served as a negative assay control for each plate (gray shaded bars). For comparison we provide both the original raw data before normalization (A, C) and the data normalized to untreated control (B, D). JSM6427 significantly inhibited RPE cell attachment to fibronectin (P < 0.05; n = 4).

Although the attachment to vitronectin was also inhibited by JSM6427 in the range from 0.4 to 100 μM (P < 0.02), the level of RPE attachment to fibronectin, in absolute terms, as measured colorimetrically (see Materials and Methods), was ≈16-fold greater than the attachment to vitronectin (Fig. 4A). The observations that JSM6427 has no significant effects on RPE attachment to collagen I, collagen IV, laminin, or vitronectin confirm the specificity of JSM6427 for α5β1. 

Effect of JSM6427 on RPE Cell Proliferation and Migration

The data summarized in Figure 5A show that JSM6427 (solid bars), but not JSM8009 (hatched bars), significantly inhibits the serum-induced proliferation of hfRPE cells cultured on fibronectin precoated 96-well plates (72 hours). This inhibition extends in a dose-dependent manner from ≈ 6 to 400 μM, with some variability at the lower concentrations. At all concentrations of the inactive analogue, there was no significant decrease of FBS-induced cell proliferation. In contrast, the active compound (JSM6427) caused monotonic and significant inhibition starting at 50 μM. The effect of the inactive compound (JSM8009), which serves as a control for the effect of the active compound and validates its specificity, was constant over all concentrations. In a separate set of experiments (Fig. 5B), in the absence of coating, we found that the effect of the inactive analogue was also constant over all tested concentrations. In this case, the active compound was much less effective and exhibited significant inhibition of proliferation only at the highest concentration. This inhibitory effect can be assumed to be caused by constitutive fibronectin secretion by RPE (Fig. 3). These observations reflect a specific blockade of RPE proliferation by integrin α5β1 antagonist and highlight the importance of the fibronectin substrate. 

Figure 5.

 

Dose-response of integrin α5β1 antagonist on hfRPE proliferation (6.3–400 μM). (A) Experiments performed on fibronectin-coated plate. (B) Experiments performed on uncoated plate. JSM6427 dose dependently inhibited the proliferation of hfRPE cultured on plates coated with fibronectin. (A, B, open bars) Controls for 0% and 5% FBS. *P < 0.05, **P < 0.01, ***P < 0.001 compared with control (5% FBS).

Figure 6A shows that JSM6427 completely abolished the stimulatory effect of 5% serum on hfRPE migration. In contrast, the inactive control compound, JSM8009, showed no effect on cell migration. In addition, 50 μM JSM6427 significantly inhibited bFGF, and PDGF-BB induced hfRPE cell migration (Fig. 6B). In all these experiments, cell viability was evaluated and confirmed using an appropriate assay. Figure 7 shows that the percentage of viable cells was greater than 96% in every group examined, and this percentage was not significantly decreased after incubation with JSM6427 or the control compound (JSM8009). 

Figure 6.

 

Effects of integrin α5β1 antagonist on hfRPE migration. (A) Integrin α5β1 antagonist JSM6427 (100 μM, 400 μM) completely abolished the stimulatory effect of 5% serum-induced hfRPE migration (uncoated plate). (B) JSM6427 (50 μM) significantly inhibited bFGF, and PDGF-BB induced hfRPE migration on fibronectin-coated plates. *P < 0.05 compared with SFM control. #P < 0.05.

Figure 7.

 

Cell cytotoxicity assay. JSM6427 had no appreciable effect on cell viability up to 400 μM.

Effect of JSM6427 on hfRPE F-Actin Cytoskeleton

Figures 8A–C show immunofluorescence staining of postconfluent monolayers of hfRPE. In all images, F-actin filaments were stained with phalloidin (green), junctional complexes were stained with ZO-1 (red), and nuclei were stained with DAPI (blue). Neither the integrin antagonist JSM6427 (Fig. 8B) nor the control compound JSM8009 (Fig. 8C) had any appreciable effect on F-actin cytoskeleton of quiescent postconfluent RPE cells compared with the untreated control. The middle set of images (Figs. 8D–F) are from cells in the process of reaching confluence and, like the postconfluent cells, do not show any effect of treatment with either JSM6427 or its control, JSM8009. In contrast, Figures 8G–L illustrate cells that are nonconfluent along a wound edge. For these cells, the addition of JSM6427 caused the aggregation of F-actin filaments around the circumference of the cell (Figs. 8H, 8K, white arrow). Panel II shows an enlarged view of Figures 8J–L illustrating the effect of JSM6427 on cytoskeleton organization and ZO-1 distribution. To better visualize the ZO-1 staining, images Figures 8M–O only show the red and blue channels. In addition to the circumferential accumulation of F-actin, JSM6427 also caused the loss of the “interdigitated” structure of ZO-1 observed in the regions of cell–cell contact (Fig. 8, panel II; compare insets in Figs. 8M, 8N). Taken together, panels I and II indicate that this antagonist can affect only actively dividing cells. 

Figure 8.

 

Phalloidin staining of F-actin cytoskeleton. Postconfluent monolayers of hfRPE cells were scratched and either untreated or treated with 100 μM JSM6427 or JSM8009 for 48 hours. Panel I: (A–C) Postconfluent area. (D–F) Subconfluent area. (G–L) Nonconfluent (edge area). Nuclei were stained with DAPI (blue), ZO-1 tight junction marker (red), and phalloidin staining of F-actin (green). White arrows: areas of JSM6427-induced cytoskeletal changes. Panel II: Enlarged view of images (J–L) showing the effect of JSM6427 on cytoskeleton organization and ZO-1 distribution. Images including the red and blue channels are also shown in the bottom of panel II for better visualization of ZO-1. Arrowheads: detailed views (insets) of the ZO-1 interdigitation.

Discussion

α5β1 Localization and RPE Attachment

Human fetal RPE cultures constitutively secrete fibronectin to the apical bath. In Figure 3, the much lower protein secretion to the basal, compared to the apical, bath probably occurred because the ECM-coated porous plastic support could limit the secretion of large proteins, such as fibronectin (≈440 kDa). Fibronectin is a specific ligand for integrin α5β1 that is present on both native and cultured human RPE cells. The α5 subunit is located mainly at the RPE apical membrane, whereas the β1 subunit is localized to both the apical and the basolateral membranes. The integrin α5β1 antagonist JSM6427 significantly and specifically inhibited hfRPE cell attachment to fibronectin; it also inhibited bFGF and PDGF-BB-induced cell migration and serum-induced cell proliferation. In dividing RPE cells, this antagonist significantly altered the F-actin cytoskeleton but had no effect on F-actin filaments of quiescent cells. 

Reports on the presence and localization of α5 in RPE cells have been inconsistent, even within species. For example, using immunohistochemistry techniques, Brem et al. ²⁵ found no α5 in adult native human RPE cells, but another study used immunofluorescence to detect integrin α5 at the basal surface of the RPE. ³⁴ Anderson et al. ⁸ used nonpermeabilized adult human RPE cell cultures and found the fibronectin receptor to be localized to the plasma membrane of apical microvilli in vitro; a similar result was obtained using native monkey RPE. ⁸ A later report using permeabilized adult monkey RPE cells with access to the basolateral surface showed fibronectin receptor (α5β1) localization at the basolateral surface. ⁷ The present experiments used confluent, high-resistance primary cultures of well-characterized human fetal RPE ^{31,33,35–37} to reexamine the question of localization. A panel of antibodies (α5 subunit, β1 subunit, and α5β1) and immunoblots showed that the integrin α5 and β1 subunits are expressed on both native human RPE and primary cultures of hfRPE; immunofluorescence experiments show that α5 is localized mainly on the apical membrane, whereas β1 is localized on both the apical and basolateral membranes of hfRPE. In addition, microarray experiments show that mRNA expression of integrin α5 and β1 subunits are much higher in primary culture of hfRPE cells than in native human RPE cells (Wang F, et al. IOVS 2006;47:ARVO E-Abstract 2855). In addition, the mRNA expression of β1 is much higher than that of α5, indicating other β1 integrins in human RPE. 

The difference in the behavior of primary culture of hfRPE and native human adult RPE may be related to the matrix coating, which contains various ECM components, including fibronectin. The high level of hydrocortisone in the culture media could increase integrin expression level. ³⁸ Early reports suggest that the expression level of integrin α5β1 can be significantly affected by RPE cell state; actively dividing cells express higher levels of integrin α5β1 than do quiescent cells. ³⁴ The protein levels of integrin α5 and β1 subunits in native RPE may depend on donor age and nonreported systemic diseases; this conclusion is based on our limited sample of adult human eyes (n = 2). In another study, Zarbin et al. ³⁹ assessed the differences in expression of integrin subunits in fetal and human adult RPE. Integrin subunits α1, α2, α3, α4, and α5 (mRNA) were significantly lower in uncultured aged native RPE than in primary cultured cells. In fetal RPE, α2, α3, α5, β4, and β5 subunit mRNAs were significantly lower in uncultured than in passaged cultured cells. Taken together, these data indicate that α5β1 integrin expression correlates closely with proliferative capacity of RPE cells. 

The interface between photoreceptors and the RPE is the site of a weak, but functionally important, adhesive interaction. Disruption of this interface can lead to a number of proliferative and degenerative changes, concluding in photoreceptor cell death. ^39,40 The apical distribution of the integrin α5 subunit suggests a possible role for integrin α5β1 in the attachment of RPE to the neuroretina. The apical microvilli of native RPE cells ensheathe the photoreceptor outer segments and phagocytose large numbers of shed ROS tips each day. ⁴¹ Mechanisms involving αvβ5 integrins and other receptors have been implicated in phagocytosis by RPE. ^42–44 Zhao et al. ⁴⁵ have presented data indicating that α5β1 integrin plays a role in the phagocytosis of fibronectin by subconfluent RPE. Whether this conclusion holds for confluent/native RPE remains to be determined. 

Cell attachment of RPE cells to fibronectin, but not collagen I or IV, laminin, or vitronectin, was significantly inhibited by a specific α5β1 antagonist, JSM6427 (1700-fold greater for α5β1 than for other integrins). ²⁶ In diseases such as PVR, ^47,48 PDR, ⁴⁹ and AMD, ⁵⁰ normally quiescent RPE cells can reenter the cell cycle and initiate migration and proliferation. In vitro, integrin α5β1-mediated cell adhesion to fibronectin is particularly efficient in supporting mitogen-dependent proliferation of fibroblastic, epithelial, and endothelial cells in vitro. ⁵¹ In our studies, JSM6427 significantly inhibited bFGF and PDGF-BB-induced RPE cell migration and cell proliferation, suggesting this small molecule as a possible therapeutic agent. 

Clinical Relevance

We demonstrated the inhibitory effect of an integrin α5β1 antagonist on RPE cell attachment, migration, and proliferation. This inhibition is accompanied by a concomitant reorganization of RPE cytoskeleton with distinctive features, including the aggregation of F-actin filaments around the circumference of the cell, resembling the quiescent state. Activated, migrating cells at the wound edges often adopt a more fibroblast-like phenotype that forms actin stress fibers. ⁵² These stress fibers terminate in focal adhesions containing integrins that anchor the cells to the extracellular matrix. Intracellularly, focal adhesions contain numerous structural and signaling molecules. ² These are needed to transmit integrin-mediated signals to the cell during proliferation, apoptosis, or migration. Disrupting these structures leads to the inactivation of these integrin-mediated signaling pathways. Actin is a major component of the cytoskeleton-forming microfilaments and, together with microtubules and intermediate filaments, mediate cell movement. ^53,54 Depending on their structure, integrin receptors can interact with ECM molecules, cytoskeleton, and related proteins such as talin. ⁵⁵ For example, β4 has a unique and very large cytoplasmic domain that can link to keratin filaments, whereas β1 has a short cytoplasmic domain that connects with actin-based filaments. ⁵⁶ This unique difference creates a transmembrane link between the extracellular matrix and the cell cytoskeleton, and adds a regulatory function in a variety of adhesion-related cellular events. JSM6427 showed significant disruption of the F-actin cytoskeleton of dividing RPE cells but had no effect on F-actin filaments of quiescent cells. We also found that JSM6427 had no effect on mitochondrial membrane potential or cell apoptosis of quiescent cells (not shown). Therefore, JSM6427 may provide therapeutic benefit under certain pathophysiological conditions, such as PVR, PDR, and AMD. 

We show that primary cultures of human fetal RPE constitutively secrete fibronectin to the apical bath to create an autocrine signaling loop, with low levels of α5β1 localized to the apical membrane. The present experiments led us to speculate that in vivo, fibronectin secretion may populate the subretinal space. Following a disease stimulus, this secretion would be increased along with a α5β1 receptor expression increase, and this combination of events could provide the basis for RPE cell proliferation and migration in proliferative retinopathies. 

Footnotes

 Supported by the Intramural Research Program of the National Institutes of Health, National Eye Institute.

Footnotes

 Disclosure: R. Li, None; A. Maminishkis, None; G. Zahn, Jerini AG (E, P); D. Vossmeyer, Jerini AG (E, P); S.S. Miller, None

Footnotes

 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.

The authors thank Roland Stragies (Jerini AG) for the synthesis of test compounds.