Production and Accumulation of Thrombospondin-1 in Human Retinal Pigment Epithelial Cells

Abstract

purpose. To investigate the production and release of thrombospondin-1 (TSP-1), a natural inhibitor of angiogenesis, by human retinal pigment epithelial (RPE) cells to clarify the possible role of TSP-1 in maintaining intraocular angiogenesis.

methods. Human RPE cells were isolated from a human cadaveric eye and cultured in medium with 5% newborn calf serum. TSP-1 messages in the purified RNA of RPE cells were analyzed by reverse transcription–polymerase chain reaction (RT-PCR). Intracellular TSP-1 peptides were detected by cytofluorographic analysis. TSP-1 peptides in the culture medium on RPE cells were measured by sandwich enzyme-linked immunosorbent assay (ELISA). TSP-1 specific immunofluorescent staining was tested in RPE cells cultured on glass slides and in a human retinal tissue specimen.

results. mRNA specific for TSP-1 was found in RT-PCR products from RPE cells, and it showed a time-dependent increase from the beginning of the culture. Intracellular staining for TSP-1 was identified by flow cytometry. The sandwich ELISA identified a time-dependent increase of TSP-1 peptides in the culture medium of RPE cells. Immunostaining for TSP-1 was observed in the cytoplasm of RPE cells cultured on glass slides. Positive immunostaining of TSP-1 was observed in the cytoplasm of the RPE layer in the human retinal tissue specimen.

conclusions. RPE cells can produce and release TSP-1 in vitro, and TSP-1 accumulates in the cytoplasm of RPE cells in vivo as well as in vitro. The production of TSP-1 by RPE cells is influenced by the state of proliferation and/or cell density. TSP-1 appears to be an important control factor in retinal and choroidal neovascularization.

Thrombospondin (TSP)-1 is a cell-attachment factor with cell-specific affinity. ^{1 2 3} Of the five known subtypes of thrombospondins, ⁴ TSP-1 is the most common, being present in the α granules of platelets ^{5 6 7} and released in blood clotting. ^{8 9} Several cell types, ^{10 11 12 13} including corneal endothelial cells, ¹⁴ produce TSP-1, which apparently accumulates around the extracellular matrix. TSP-1 enhances cell adhesion of fibroblasts ¹⁵ and various carcinoma cells ¹⁶ but prevents cell invasion of a variety of cell types and prevents tissue angiogenesis. It also inhibits adhesion of vascular endothelial cells ^{17 18} and cell invasion, as well as tube formation of vascular endothelial cells. ^{19 20 21 22} A lower concentration of TSP-1 stimulates tube formation in vivo, ²³ and TSP induces microvascular invasion in vivo. ²⁴ Thus, TSP-1 is considered to be a modulator of angiogenesis. 

In human eyes, TSP-1 was reported to be localized between the RPE layer and Bruch’s membrane ²⁵ and in the epiretinal membrane in several diseases. ^{26 27} Bruch’s membrane separates the microvasculature of choroidal blood vessels from the retina. In their physiological state, choroidal blood vessels do not advance beyond Bruch’s membrane into the retina. However, in several diseases, such as wet-type age-related macular degeneration, new blood vessels develop in the choroidal vasculature and break through Bruch’s membrane. ^{28 29} This progression of new blood vessels is considered a consequence of an anatomic or functional breakdown of Bruch’s membrane. ^{30 31} Functional and morphologic disturbance of RPE cells apparently also precedes this subretinal neovascularization. Thus, a physiological antiangiogenic barrier on Bruch’s membrane and the involvement of RPE cells in maintaining the barrier are suggested. TSP-1, with its antiangiogenic activity, may play some biologic role on Bruch’s membrane. However, the source of the TSP-1 occurring on Bruch’s membrane has not been demonstrated. Therefore, we tried to determine whether TSP-1 is produced by human RPE cells. 

Materials and Methods

Cell Culture

Human RPE cells were isolated from a human cadaveric eye according to the method described by Flood et al. ³² Two cell lines of RPE cells (RPE1 and RPE2) were established from two individuals and cultured in Dulbecco’s minimum essential medium (DMEM) with 5% newborn calf serum (NBCS) at 37°C in 5% CO₂ with 100% humidity. TSP-1–producing HL-60 cells ^{33 34} were cultured in RPMI 1640 culture medium with 10% NBCS and retinoic acid (10 ng/ml). Human T lymphocytes were isolated from human blood preserved at the Japan Red Cross Blood Center, separated as a floating population using a kit (Cellect/Human T-cell; Biotex, Edmonton, Alberta, Canada), and cultured in DMEM with 10% NBCS. 

Reverse Transcription–Polymerase Chain Reaction (RT-PCR) Analysis for TSP-1

Total cellular RNA was prepared as follows. The cells were homogenized with a direct application of 1.5 ml RNAzol B (Tel-Test, Friendswood, TX) in a tissue-culture flask with a 75-cm² bottom. Cellular RNA was purified according to the instruction manuals for RNAzol B. The RNA concentration was measured by optical density at 260 nm. 

Single-stranded cDNA was synthesized from 5 μg purified RNA by adding murine reverse transcriptase and dNTPs from a commercially available kit (Ready to Go T-Primed First Strand Kit, Pharmacia Biotech, Foster City, CA). 

PCR was performed in a solution containing final concentrations as follows: 1.25 units Taq DNA polymerase (Perkin Elmer, Foster City, CA), 10 mM Tris-HCl (pH 9.0), 50 mM KCl, 1.5 mM MgCl₂, 200 μM dNTP (Perkin Elmer), and 50 pM each of forward and reverse primers. TSP-1 primers used were 5′-AAC CGC ATT CCA GAG TCT GG-3′ and 5′-TTC ACC ACG TTG TTG TCA AGG GT-3′. ³⁵ PCR (Perkin Elmer) was conducted for 10 seconds at 95°C, for 15 seconds at 56°C, and for 40 seconds at 72°C for 25 cycles. 

The amplified products were electrophoresed on 0.8% (wt/vol) agarose gels before staining with ethidium bromide, and the stained gels were exposed to ultraviolet light for observation and photographed. The experiment was repeated twice. 

Antibodies for TSP

A mouse monoclonal anti-human TSP-1 antibody and a rabbit polyclonal anti-human TSP antibody were obtained from Sigma (St. Louis, MO), ^{36 37} and Athens Research and Technology (Athens, GA), respectively. The specificity of the antibodies was examined by Western blot analysis according to the method of Qabar et al. ³⁸ As shown in Figure 1 , both polyclonal and monoclonal antibodies were bound to TSP-1 of approximately 180 kDa purified from platelets (lanes 2 and 4). From the spent medium of the RPE cells, only one band with the same apparent molecular weight was detected with both antibodies (lanes 3 and 5). 

Cytofluorographic Analysis of Intracellular TSP-1

RPE cells separated from the 25-cm² culture flasks by 10 mM EDTA and retinoic acid–treated HL-60 cells ^{33 34} suspended in calcium- and magnesium-free phosphate-buffered saline (PBS). After centrifugation at 400g for 5 minutes, approximately 1.5 × 10⁶ cells were resuspended in 0.1 ml PBS, fixed by 5.5% formaldehyde, and permeabilized by saponin-based permeabilizing medium (IntraPrep, Immunotech, Marseilles, France). ³⁹ The cells were then incubated with 40 μl rabbit polyclonal anti-human TSP-1 IgG (5 μg/ml) or with control unimmunized rabbit IgG at 4°C for 30 minutes, washed with PBS, and incubated with 40 μl fluorescein-conjugated goat anti-rabbit IgG antibody (10 μg/ml; Cappel, Aurora, OH) at 4°C for 30 minutes. The cells were washed and resuspended in 1 ml PBS, and single-color fluorescence flow cytometry was performed (FACS Calibur; Becton Dickinson, San Jose, CA). The experiment was repeated twice. 

Collection of the Culture Medium of RPE Cells

RPE cells (1.2–.3 × 10⁵/ml) in DMEM with 5% NBCS were seeded onto 24-well plates and incubated at 37°C in 5% CO₂. At 4, 8, 12, 16, and 24 hours after incubation, culture medium was collected and frozen in −20°C until it was measured. The cells were washed twice with PBS and tripsinized. The number of cells in each well was counted with an automated cell counter (Tokyo Kohden, Tokyo, Japan). Samples were produced in triplicate, and the experiment was repeated three times. 

Enzyme-Linked Immunosorbent Assay for TSP-1

An enzyme-linked immunosorbent assay (ELISA) for TSP-1 was performed using a sandwich method. Each well of a 96-well ELISA plate (Corning, Cambridge, MA) was incubated with 100 μl of 1 μg/ml rabbit polyclonal anti-human TSP-1 IgG overnight at room temperature and blocked with blocking solution (Block Ace, Dainihon, Osaka, Japan). The sample solution (100 μl) was added to each well and incubated for 1 hour at 37°C and washed twice with a washing buffer (0.05% NP-40 in barbiturate-buffered saline). The wells were incubated with 1μ g/ml mouse monoclonal antibody for 1 hour at 37°C, washed, and then incubated successively with 1 μg/ml biotinylated horse polyclonal anti-mouse IgG (Vector, Burlingame, CA) and 1 μg/ml horseradish peroxidase-avidin complex (Vector). The bound antibody was detected by the addition of o-phenylenediamine diluted in methanol with 0.03% hydrogen peroxide. The absorbance at 490 nm of each well was measured by a microplate reader (Bio-Rad, Hercules, CA). A standard concentration curve was plotted from the measurement of purified human TSP-1 in graded concentrations from 1 to 1024 ng/ml. The concentration of TSP-1 in each sample solution was determined from the standard curve, and the production rate of TSP-1 at each time point was calculated. 

In the standard culture medium with 5% fetal calf serum, a maximum of 2 ng/ml TSP-1 was detected with ELISA in a cell-free condition. Therefore, the value for secreted TSP-1 in culture medium was calculated by subtracting the value in cell-free medium from the value of spent culture medium. 

Culture of RPE Cells on Glass Slides

The cells were trypsinized, separated from the flask, collected into the culture medium, and seeded onto glass slides of four-chamber separated wells (Nunc, Napierville, IL) and cultured in DMEM with 5% NBCS for 12 hours at 37°C in 5% CO₂. 

Human Retinal Tissue Specimen

A human eyeball obtained at autopsy from a 61-year-old Japanese male donor without ocular disease was dissected, and the posterior segments were embedded and fresh frozen in optimal cutting temperature (OCT) compound. Eight-micrometer-thick sections were cut by a cryomicrotome for histopathologic examination. 

Immunofluorescent Staining of TSP-1 in Cultured RPE Cells and Human Retina

The histologic sections were air dried, fixed in 4% paraformaldehyde, and washed in 50 mM tris-buffered saline. The sections were then incubated with blocking solution (Block Ace; Dainihon) for 30 minutes at room temperature. RPE cells and retinal tissues were incubated with rabbit polyclonal anti-human TSP-1 IgG (5μ g/ml) for 1 hour at room temperature. RPE cells were then incubated with biotinylated swine polyclonal anti-rabbit IgG (Vector) and alkaline phosphatase (Vector). For monoclonal antibody staining, cells and tissue sections were incubated with mouse monoclonal anti-human TSP-1 IgG κ chain (1 μg/ml) and biotinylated goat polyclonal anti-mouse IgG antibody (Vector). Alkaline phosphatase was reacted with fuchsin (Merck, Darmstadt, Germany) for 7 minutes at room temperature and washed. The nucleus was stained with Meyer’s hematoxylin. The sections were covered by a coverslip and observed by light microscopy. Sections of human retina were incubated with rhodamine-conjugated goat polyclonal anti-rabbit IgG (Vector) for 1 hour at room temperature and washed three times. Specimens were sealed by a water-soluble, nonfluorescent sealant and covered with a coverslip. Finally, the dried specimens were observed under a confocal scanning laser microscope (Zeiss, Oberkochen, Germany) and photographed. 

Results

Expression of mRNA for TSP-1 on RPE Cells

On agarose gel electrophoresis, mRNA specific for TSP-1 was found in RT-PCR products obtained from both of the two cell lines of cultured human RPE cells as a single band corresponding to a length of 688 bp (Fig. 2) . The intensity of the band on RPE2 cells was slightly higher than that of RPE1 cells. The band on RPE1 cell line before incubation was faint but became stronger with time, reaching peak intensity 6 hours after incubation (Fig. 3) . A single band was also found on the RNA from HL-60 cells used as a positive control. No band was found on the gel of RNA from T cells (which are known to be incapable of producing TSP-1). A single band representing β-actin, which was probed as an internal control, was found on all RT-PCR products from different cell types. 

Intracellular Staining for TSP-1

Intracellular staining on flow cytometry identified a positive pattern of TSP-1–specific fluorescence in the two cell lines of RPE cells incubated with fluorescein-conjugated antibody to TSP-1 or the vehicle buffer (Fig. 4) . Intracellular staining was also positive on HL-60 cells (Fig. 4) . 

Release of TSP-1 into the Culture Medium

A sandwich ELISA assay identified a time-dependent increase of the TSP-1 peptides in the culture medium of RPE cells (Fig. 5) . Four hours after the start of the culture, 6.7 ng/ml TSP-1 was detected in the culture medium. At 8 hours, 14.3 ng/ml TSP-1 was detected in the same well. The concentrations of TSP-1 in each well gradually increased with time, reaching 24.2 ng/ml at 12 hours and peaking at 41.7 ng/ml at 16 hours. The rate of TSP-1 release began decreasing at 24 hours, although each cell number increased with time. 

Immunolocalization of TSP-1 on Cultured RPE Cells and Human Retina Tissue

On immunostaining of a histologic section of RPE cells cultured on glass slides, positive staining for TSP-1 was observed in the cytoplasm of the cells (Fig. 6) . On histologic section of the human retinal tissue, strong fluorescence was observed in the cytoplasm of the RPE layer, and patchy fluorescence was observed on some parts of Bruch’s membrane (Fig. 7) . No remarkable fluorescence was observed on the neuroretina, choroid, or sclera (Fig. 7) . Although immunoreaction was more marked on the cells and tissue sections stained with the polyclonal antibody than that stained with the monoclonal antibody, localization of immunoreactions was identical. 

Discussion

Our results show that RPE cells produced and released TSP-1 in vitro and that TSP-1 accumulated in the cytoplasm of RPE cells in vitro and in vivo. 

Several different cell types are known to produce TSP-1. Among ocular cells, corneal endothelial cells probably produce TSP-1, because positive immunostaining for TSP-1 has been observed in them. ^{14 40} Although corneal tissue is avascular, vascular growth can be induced by several pathologic processes in which the tissue’s physical structure or cell function are disturbed. It has been suggested that corneal cells may play some role in maintaining the transparency of the cornea by inhibiting vascular growth. ⁴¹  

Also, pathologic choroidal neovascularization with progression toward the neuroretina is frequently seen in eyes in which RPE cells are undergoing senescent functional and morphologic changes. ²⁸ These findings suggest that RPE cells play a role of inhibiting choroidal angiogenesis. TSP-1 is known to inhibit angiogenesis profoundly, both in vivo and in vitro. ^{19 20 21 22} TSP-1 molecules and RPE cells coexist on the avascular epiretinal membrane in several diseases. ^{26 27} Possible natural mechanisms for halting angiogenesis have been found in vivo and in vitro. Cytokines mediate cell-to-cell signals for regulating cell proliferation. ^{42 43} Pericytes inhibit angiogenic activity by contacting vascular endothelial cells through the activation of a latent form of transforming growth factor-β. ^{44 45 46} Alteration of the extracellular matrix can induce morphologic changes of angiogenic vascular endothelial cells, also halting angiogenesis. ⁴⁷ These findings suggest that RPE cells may modulate choroidal vascular growth by supplying TSP-1. Bornstein ⁴⁸ has defined extracellular protein molecules dedicated to modulation of cell behavior by interacting with many extracellular molecules and with cell surface receptor as matricellular proteins. TSP-1 was expressed by migrating RPE in proliferative vitreoretinopathy membranes and RPE-derived TSP-1, together with other matricellular proteins, has been considered to play a role in development of proliferative vitreoretinopathy. ⁴⁹ Our results suggest an additional or alternative role for RPE-derived TSP-1, because it is also expressed on nonmigratory RPE cells. 

The biological influence of TSP on angiogenesis is still controversial. ^{4 23} It has been reported that lower concentrations of TSP-1 induces the tube formation of vascular endothelial cells in vitro, whereas inhibition occurs in higher concentrations. Furthermore, hypoxic stimulation and increased concentration of extracellular matrices including TSP-1 enhance the release of angiogenic growth factors as vascular endothelial growth factor and fibroblast growth factor-2 by RPE cells. ⁵⁰ It is hard to determine whether the TSP-1 produced by RPE cells stimulates or inhibits the angiogenesis in vivo. However, it has been clear that neither choroidal blood vessels nor retinal blood vessels proliferate and migrate toward the RPE layer though TSP-1 accumulated surrounding the RPE layer. Therefore, TSP-1 may play some role in maintaining differentiated blood vessels. 

Although data are specific to TSP-1, it has been reported that the biologic activity of all types of TSPs is similar, ^{51 52} and TSP-1 is the most prevalent TSP molecule found in vivo. Therefore, it is reasonable to speculate that RPE cells may produce other varieties of TSP. Also, TSP-1 may be produced by retinal cells other than RPE cells. That there is no evident accumulation of TSP-1 on retinal blood vessels and retinal neuronal and glial cells ⁵³ suggests that RPE cells are the major retinal components producing TSP-1 physiologically. 

Our study did not allow us to determine whether RPE cells in the physiological state continuously produce TSP-1, although the positive immunostaining of the intracellular TSP-1 of RPE layer suggests that they do. However, a decrease of TSP-1 after an initial increase was shown in both the cultured medium and mRNA of RPE cells. This finding suggests that the production of TSP-1 may increase in pathologic conditions involving RPE proliferation and migration. Decreased production of TSP-1 in conditions of cellular quiescence as the cell density increases has been reported for other TSP-producing cells. ⁵⁴ The decrease of TSP-1 on the culture medium may be due to degradation after secretion. Furthermore, it is also known that the antiangiogenic activity of TSP-1 is mediated by CD36, a TSP-1–binding receptor. ^{55 56 57} TSP-1 expression has been shown to be upregulated by stimulation to CD36. Several TSP-1–producing cells express CD36, ⁵⁸ suggesting that TSP-1 is a factor in the autocrine–paracrine mechanism. ^{42 43} Finally, it has been reported that CD36 is expressed on RPE cells and mediates the modulatory action of phagocytosis of RPE cells. ⁵⁹ Further accumulation of knowledge on modulation of TSP-1 production by RPE cells and its biologic effects may elucidate key physiological maintenance mechanisms and pathologic processes. 

 Submitted for publication February 23, 1999; revised June 8 and August 10, 1999; accepted September 13, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Specificity of rabbit polyclonal antihuman TSP antibody and mouse monoclonal anti-human TSP-1 antibody. TSP-1 purified from human platelet (lanes 2 and 4) and concentrated culture medium of RPE cells (lanes 3 and 5) were electrophoresed on 7.5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel, and transferred to nitrocellulose membrane. TSP-1 was detected with rabbit anti-TSP (lane 2 and 3) and monoclonal anti-TSP-1 (lanes 4 and 5) as described in the Materials and Methods section. Lane 1: molecular weight markers.

Figure 2.

 

Expression of TSP-1 and β-actin mRNA on two lines of human RPE cells (RPE1, RPE2), HL-60 cells, and T lymphocytes. A TSP-1–specific PCR product is seen as a 688-bp band on agarose gel, indicated on cDNA from RPE1 cells (lane 2), RPE2 cells (lane 3), and HL-60 cells (lane 4), but not on cDNA from T lymphocytes (lane 5). β-Actin–specific products as an internal standard were seen as a 525-bp band on cDNA from RPE1 cells (lane 6), RPE2 cells (lane 7), HL-60 cells (lane 8), and T lymphocytes (lane 9). Lane 1: molecular weight marker.

Figure 3.

 

Expression of TSP-1 and β-actin mRNA in RPE1 cells at various intervals after the start of incubation. TSP-1 mRNA in RPE cells was seen at 0 hours (lane 2), 3 hours (lane 3), 6 hours (lane 4), 12 hours (lane 5), and 24 hours (lane 6) after incubation. The expressions of TSP-1 mRNA gradually increased from time 0, reached peak intensity at 6 hours, and decreased at 12 hours. No difference was seen in the expression of β-actin mRNA at 0 hours (lane 7), 3 hours (lane 8), 6 hours (lane 9), 12 hours (lane 10), and 24 hours (lane 11) after incubation. TSP-1 mRNA reached peak intensity at 6 hours after incubation. Lane 1: molecular weight marker.

Figure 4.

 

Intracellular expression of TSP-1 on RPE1 cells, RPE2 cells, and HL-60 cells. The shift of flow cytometry fluorescence histograms of RPE cells that were incubated with unimmunized IgG (thin line) and rabbit polyclonal anti-human TSP-1 (thick line) represents intracellular expression of TSP-1.

Figure 5.

 

Release of TSP-1 from RPE cells to the culture medium and the number of RPE cells at the beginning of the culture. RPE cells in DMEM containing 5% NBCS were seeded in a 24-well plate. The TSP-1 concentrations at each time point in the culture medium were measured by sandwich ELISA.○ , average number of cells (vertical bars, ± SD) in a culture well at each time interval.

Figure 6.

 

A light micrograph of cultured RPE cells on the glass slides immunostained for TSP-1. Positive immunostaining (purple) for TSP-1 was seen in the cytoplasm of cultured RPE1 cells (A) and RPE2 cells (B). (C) Negative control using nonspecific isotypic antibody.

Figure 7.

 

An accumulation of TSP-1 in the cytoplasm of RPE cells in a section of the human chorioretinal layer. A micrograph of phase-difference coherent images overlaid on laser-scanned fluorescent images from a histopathologic section of the human chorioretinal layer. Rhodamine fluorescence was observed on the cytoplasm of the RPE layer. No remarkable fluorescence was observed on the neuroretina, choroid, or sclera.