Expression of pigment epithelium‐derived factor and thrombospondin‐1 regulate proliferation and migration of retinal pigment epithelial cells

Abstract Age‐related macular degeneration (AMD) is the leading cause of vision loss among elderly. Although the pathogenesis of AMD is associated with retinal pigmented epithelium (RPE) dysfunction and abnormal neovascularization the detailed mechanisms remain unresolved. RPE is a specialized monolayer of epithelial cells with important functions in ocular homeostasis. Pathological RPE damage contributes to major ocular conditions including retinal degeneration and irreversible loss of vision in AMD. RPE cells also assist in the maintenance of the ocular angiogenic balance by production of positive and negative regulatory factors including vascular endothelial growth factor (VEGF), thrombospondin‐1 (TSP1), and pigment epithelium‐derived factor (PEDF). The altered production of PEDF and TSP1, as endogenous inhibitors of angiogenesis and inflammation, by RPE cells have been linked to pathogenesis of AMD and choroidal and retinal neovascularization. However, lack of simple methods for isolation and culture of mouse RPE cells has resulted in limited knowledge regarding the cell autonomous role of TSP1 and PEDF in RPE cell function. Here, we describe a method for routine isolation and propagation of RPE cells from wild‐type, TSP1, and PEDF‐deficient mice, and have investigated their impact on RPE cell function. We showed that expression of TSP1 and PEDF significantly impacted RPE cell proliferation, migration, adhesion, oxidative state, and phagocytic activity with minimal effect on their basal rate of apoptosis. Together, our results indicated that the expression of PEDF and TSP1 by RPE cells play crucial roles not only in regulation of ocular vascular homeostasis but also have significant impact on their cellular function.


Abstract
Age-related macular degeneration (AMD) is the leading cause of vision loss among elderly. Although the pathogenesis of AMD is associated with retinal pigmented epithelium (RPE) dysfunction and abnormal neovascularization the detailed mechanisms remain unresolved. RPE is a specialized monolayer of epithelial cells with important functions in ocular homeostasis. Pathological RPE damage contributes to major ocular conditions including retinal degeneration and irreversible loss of vision in AMD. RPE cells also assist in the maintenance of the ocular angiogenic balance by production of positive and negative regulatory factors including vascular endothelial growth factor (VEGF), thrombospondin-1 (TSP1), and pigment epithelium-derived factor (PEDF). The altered production of PEDF and TSP1, as endogenous inhibitors of angiogenesis and inflammation, by RPE cells have been linked to pathogenesis of AMD and choroidal and retinal neovascularization. However, lack of simple methods for isolation and culture of mouse RPE cells has resulted in limited knowledge regarding the cell autonomous role of TSP1 and PEDF in RPE cell function. Here, we describe a method for routine isolation and propagation of RPE cells from wild-type, TSP1, and PEDF-deficient mice, and have investigated their impact on RPE cell function. We showed that expression of TSP1 and PEDF significantly impacted RPE cell proliferation, migration, adhesion, oxidative state, and phagocytic activity with minimal effect on their basal rate of apoptosis. Together, our results indicated that the expression of PEDF and TSP1 by RPE cells play crucial roles not only in regulation of ocular vascular homeostasis but also have significant impact on their cellular function.

Introduction
Age-related macular degeneration (AMD) is one of the major causes of visual impairment in the elderly population worldwide. Despite the high prevalence of AMD, the etiology of this disease remains largely unknown. AMD is presented in two major forms, the dry form which is associated with degeneration of retinal pigment epithelium (RPE), and the exudative or wet form which presents the formation of choroidal neovascularization (CNV) (Hua et al. 2012).The impairment of RPE cell function is an early and crucial event in the molecular pathways leading to clinically relevant AMD changes associated with increased production of vascular endothelial growth factor (VEGF) and CNV. However, the detailed mechanisms impacting RPE cell function and pathogenesis of AMD remain poorly defined.
The retinal pigment epithelium is a single layer of epithelial cells with essential roles in photoreceptor function and defensive immune mechanism of macula (Korte et al. 1984). The RPE cells are involved in phagocytosis of photoreceptor outer segments, oxidative stress response, photoreceptor renewal, and preservation of their photo transduction (Korte et al. 1984;Sparrow et al. 2010;Klettner 2012). The RPE cells have evolved a preventative mechanism against high levels of antioxidants and photooxidation through light filtration and absorption (Strauss 2005;Kevany and Palczewski 2010). In addition, RPE cells selectively regulate the transport of metabolites, ions, nutrients, and water between the retina and choriocapillaris, and maintain outer retinal-blood barrier (Strauss 2005). The failure of RPE cell function results in major ocular clinical changes such as retinal degeneration and irreversible vision loss (Korte et al. 1984;Yang et al. 2006a,b).
The RPE cells, as a major source of ocular angioregulatory proteins, play an important role in maintaining the ocular angiogenic homeostasis through a balanced production of positive and negative regulatory factors including VEGF, thrombospondin-1 (TSP1), and pigment epithelium-derived factor (PEDF). Altered production of these factors in RPE cells may contribute to various eye diseases with a neovascular component including diabetic retinopathy (DR) and exudative AMD (Campochiaro 2000;Duh et al. 2004;Strauss 2005;Sparrow et al. 2010). The increased production of VEGF has been identified as an essential factor in the development and progression of CNV. In addition, many studies have reported impaired production of TSP1 and PEDF in vascular retinopathies including diabetic retinopathy and exudative AMD. However, how these factors impact RPE cell function needs further investigation.
In human eyes, TSP1 is present at high levels in the vitreous and is a major component of the Bruch's membrane (Sheibani et al. 2000;Uno et al. 2006). Decreased TSP1 expression in Bruch's membrane and choroidal vessels during AMD is suggested as a potential regulatory factor, which favors the formation of CNV (Uno et al. 2006). Expression of TSP1 is crucial for normal postnatal development of retinal vasculature and CNV (Wang et al. 2003(Wang et al. , 2009(Wang et al. , 2012. TSP1 is a homo-trimeric matricellular protein produced by various cell types including RPE cells (Adams and Lawler 2004;He et al. 2006). It modulates proliferation, migration, differentiation, and apoptosis in a cell type-specific manner. Thus, TSP1 plays a critical role in the regulation of various biological functions including vascular homeostasis, immunity, and wound healing (Sheibani and Frazier 1999;Mochizuki et al. 2013;Masli et al. 2014). Appropriate production of TSP1 contributes to maintenance of the optical clarity and normal ocular angiogenesis function of the eye (Hiscott et al. 1996a,b;Sheibani et al. 2000;Saika et al. 2004).We previously reported that pathogenesis and progression of DR due to TSP1 deficiency may be influenced by expression of PEDF ). Thus, a relationship between TSP1 and PEDF expression may exist.
A number of studies have demonstrated important roles for PEDF in modulation of vascular leakage and angiogenesis in AMD and DR (Papapetropoulos et al. 1997;Yang et al. 2006a,b;Rizzolo et al. 2011). A significant decrease in the PEDF plasma level has been demonstrated in patients with the dry form of AMD. In contrast in the wet AMD patients, a strong positive correlation between VEGF and PEDF concentrations was observed (Machali nska et al. 2012). PEDF, a glycoprotein (50 kDa), is also an endogenous inhibitor of angiogenesis and is present in the vitreous at high levels (Karakousis et al. 2001). PEDF was originally described by its ability to stimulate differentiation of retinoblastoma cells (Barnstable and Tombran-Tink 2004). However, recent studies have established important roles for PEDF in inflammation, maintenance of normal extracellular matrix, promotion of cell attachment, stability of the endothelium of choriocapillaris, and neurite outgrowths (Yamagishi et al. 2003;Bernard et al. 2009). It also protects retinal neurons from light damage and oxidative stress, as well as inhibiting EC migration and growth of new blood vessels (Bhutto and Lutty 2012). PEDF is also considered as a crucial factor associated with absence of vascularity in the cornea, vitreous, and outer retina (Barnstable and Tombran-Tink 2004). We have previously shown that mouse RPE cells are a major source of PEDF production (Al-Shabrawey et al. 2011). How PEDF impacts RPE cell function and contributes to pathogenesis of AMD and CNV requires further elucidation.
It is now clearly established that RPE cells play a key role in the development and stabilization of retinal structure, and maintenance of the ocular angiogenic balance (Strauss 2005). These activities are accomplished through production of numerous factors but little is known about how these factors impact RPE cell function. These studies are hampered by the lack of availability of methods to readily culture RPE cells from wild-type and transgenic mice retina. Although, RPE cells have been cultured from mouse retina these cells have limited proliferation capacity and/or require special conditions (Kato et al. 1996;Martin et al. 2009), which limit many biochemical studies that require large number of cells. In the present study, we describe a method for routine culturing of RPE cells from wild-type, and TSP1 and PEDF-deficient mice, which can be readily manipulated in culture and maintained for many passages without a significant impact on the expression of their specific markers. Using these cells we show, for the first time, the cell autonomous impact of PEDF and TSP1 on RPE cell function. We found that lack of TSP1 or PEDF in RPE cells impacted their proliferation, adhesion, migration, oxidative state, and phagocytic activity with minimal impact on their basal rate of apoptosis. Collectively, our results indicate important roles for PEDF and TSP1 in modulation of RPE cell functions critical to the role of RPE cells as an important care provider and regulator of various ocular functions. Our results also indicate that these proteins not only play an important role in ocular vascular homeostasis, but also have important role in RPE cell function further emphasizing a key role for these proteins in the pathogenesis of AMD and CNV.

Experimental animals
All the experiments were conducted in accordance with the Association for Research in Vision and Ophthalmology statement for the use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and use Committee of the University of Wisconsin School of Medicine and Public health. Immorto mice expressing a temperature-sensitive simian virus (SV) 40 large T antigen were obtained from Charles Rivers Laboratories (Wilmington, MA). TSP1À/À and PEDFÀ/À mice on a C57BL/6J background were generated and maintained as previously described, and crossed with immorto mice backcrossed to C57BL/6J mice (Lawler et al. 1998;Su et al. 2003). The isolated DNA from tail biopsies was used for screening of various transgenes as previously described (Su et al. 2003;Scheef et al. 2009).
Isolation and culture of RPE cells RPE cells were isolated form one litter (6 or 7 pups) of 4-week-old wild-type, PEDFÀ/À and TSP1À/À Immorto mice (all on C57BL/6J background) by removing retinas, and optic nerve form the posterior portion of the eyes, cutting the remaining cup and collection of RPE layer under a dissecting microscope. RPE sheets were digested in 5 mL of Collagenase type I (Worthington, Lakewood, NJ; 1 mg/mL in serum-free Dulbecco ҆ s modified Eagle ҆ s medium (DMEM)) and incubated at 37°C for 20 min. The digested tissues were washed with DMEM containing 10% fetal bovine serum (FBS), and resuspended in 0.5 mL of RPE cell growth medium [DMEM containing 10% FBS, 2 mmol/L L-glutamine, 100 lg/mL streptomycin, 100 U/mL penicillin and murine recombinant INF-c (R& D Systems, Minneapolis, MN) at 44 U/mL]. Cells were plated in a well of 24-well plate coated with fibronectin (2 lg/mL in serum-free DMEM; BD Bioscience, Bedford, MA) and incubated in a tissue culture incubator at 33°C with 5% CO 2 . Cell were progressively passaged to larger plates and maintained in 60-mm tissue culture plates coated with 1% gelatin. For all experiments, cells were incubated with RPE cell growth medium containing INF-c in a tissue culture incubator at 33°C with 5% CO 2 . However, to confirm the observed results is specifically due to PEDF and/ or TSP1 deficiency, cells were also incubated with RPE cell growth medium without INF-c in a tissue culture incubator at 37°C with 5% CO 2 for 48 h to eliminate large T antigen. Cells allowed to reach 80-90% confluence and then used for experiments. For some experiments cells were allowed to reach confluence (junctional organization) and used for experiments a week later. Three different isolations of RPE cells were used in these studies and all cells were used prior to passage 20.

Cell proliferation studies
Cell proliferation was assessed by counting the number of cells for two weeks. Cells (1 9 10 4 ) were plated in multiple sets of gelatin-coated 60-mm tissue culture plates, fed every other day for the duration of experiment. The number of cells was determined by counting every other day, on days not fed, in triplicates. The rate of DNA synthesis was also assessed using Click-It EDU Alexa Flour 488 as recommended by the supplier (Life technologies, Grand Island, NY). The assay measures DNA synthesis using 5-ethynyl-2 0 -deoxyuridine (EdU) a nucleoside analog of thymidine. The percentage of cells undergoing active DNA synthesis was determined by FACScan caliber flow cytometry (Becton Dickinson).
TdT-dUPT Terminal Nick-End Labeling (TUNEL) was used to assess apoptotic cell death. TUNEL staining was performed using Click-iT-TUNEL Alexa Flour imaging assay as recommended by supplier (Life Technologies). A similar experiment was performed in the presence of 50 lmol/L H 2 O 2 (Fisher Scientific). This concentration was determined based on moderate effect on cell viability after 24-48 h. Positive apoptotic cells counted in 10 highpower fields (9200) and calculated as percentage of total cell number. All samples were prepared in triplicates and repeated twice.

Indirect immunofluorescence studies
Cells (1 9 10 5 ) were plated on fibronectin-coated 4-well chamber slides (5 lg/mL in PBS) and allowed to reach confluence (1-2 days). Cells were rinsed with PBS, fixed with cold acetone for 10 min on ice, permeabilized with PBS containing 0.1% Triton X-100 for 12 min at room temperature, and then blocked with PBS containing 1% BSA at 37°C for 30 min. Following incubation, slides were washed once with PBS and incubated with specific primary antibodies for 2 h at room temperature. The primary antibodies were anti -ZO-1 (Life Technologies), anti-b-catenin, anti-P120-catenin, anti-N-cadherin (BD Bioscience), anti-vinculin and FITC-conjugated phalloidin (Sigma) prepared in PBS (1:200) containing 1% BSA. Following incubation, slides were washed with PBS and incubated with specific Cy3-conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA; (1:800) in PBS containing 1% BSA for 1 h at room temperature. Following incubation, slides were washed four times with PBS and examined using a fluorescence microscope (Carl Zeiss Optical, Germany) and images were taken in digital format.

Scratch wound assays
Cells (1 9 10 6 ) were plated in 60-mm tissue culture dishes and allowed to reach confluent (1-2 days). Plates were wounded using a 1-mL micropipette tip, washed with growth medium twice to remove detached cells, and fed with growth medium containing 1 lmol/L 5-fluorouracil (Sigma) to block cell proliferation. Wound closure was monitored by phase microscopy at different time points (0, 24, 48 h) and images were captured in digital format. The migrated distance as percentage of total distance was determined for quantitative assessment of data as described previouslyDiMaio and Sheibani (2008).

Transwell migration assays
Cell migration was also determined using a transwell migration assay. Costar transwell inserts (8-lm pore size, 6.5-mm membrane, Lowell, MA) were coated with PBS containing fibronectin (2 lg/mL) on the bottom side at 4°C overnight. After washing with PBS, inserts were blocked in PBS containing 1% BSA for 1 h at room temperature and rinsed with PBS. Cells were trypsinized and resuspended in serum-free medium, and 1 9 10 5 cells/ 0.1 mL was added to the top of inserts. The inserts were placed in a 24-well plate containing 0.5 mL of serum-free medium and incubated for 4 h at 33°C. Following incubation, cells were fixed with 2% paraformaldehyde for 10 min at room temperature stained with hematoxylin and eosin (H& E), and the inserts were mounted on a slide with the cell facing down. The number of cells migrated through the membrane was determined by counting 10 high-power fields (9200). Each experiment was done in triplicates and repeated with two different isolation of RPE cells.

PEDF and TSP1 reexpression studies
To express PEDF in PEDFÀ/À RPE cells, 2.5 9 10 5 cells were plated in 35-mm tissue culture dishes. The next day, lenti viruses encoding PEDF or GFP (100 pfu/cell, 20 lL) and 20 lL of lenti booster (Sirion Biotech, Martinsried, Germany) including 10 lL each of 1:100 dilution of solution A and B in PBS, were mixed and diluted in 500 lL of opti-MEM (Life Technologies) and incubated for 15 min at room temperature. Following incubation, the tissue culture plates were washed twice with serum-free medium and incubated with 0.5 mL of lenti virus and lenti booster mixture overnight. The next day medium containing lenti virus and lenti booster mixture were removed, fresh growth medium was added to the plates and incubated for 4 days before they were used for further analysis.
To express TSP1 in TSP1À/À RPE cells, 2.5 9 10 5 cells were plated in 35-mm tissue culture dishes. The next day, adenoviruses encoding TSP1 or GFP (625 pfu/cell; 50 lL) and adeno booster (10 lL, Sirion Biotech) were mixed and diluted in 500 lL of opti-MEM (Life Technologies) and incubated for 15 min at room temperature. Following incubation, the tissue culture plates were washed twice with serum-free medium and incubated with 0.5 mL of adenovirus and adeno booster mixture overnight. The next day, medium containing virus and booster mixture were removed, fresh growth medium was added to the plates and incubated for 3 days before they were used for further analysis.

Cell adhesion assays
Cell adhesion assays were conducted using 96-well plates (Maxisorb Nunc Immunoplate, Fisher Scientific) coated with different concentration of collagen I, collagen IV, vitronectin and fibronectin (BD Biosciences), diluted in TBS (50 lL/well) containing 2 mmol/L CaCl 2 , 2 mmol/L MgCl 2 (Ca/Mg) overnight at 4°C. Plates were rinsed four times with TBS containing Ca/Mg (200 lL/well), blocked using TBS with Ca/Mg containing 1% BSA (200 lL/well) at room temperature for 1 h. Cells were collected from tissue culture plates using 2 mL of dissociation buffer (2 mmol/L EDTA, 0.05% BSA in TBS), rinsed with TBS and resuspended in cell-binding buffer (150 mmol/L NaCl, 20 mmol/L HEPES, 4 mg/mL BSA, pH 7.4) at 5 9 10 5 cells/mL. The coated plates were washed with TBS containing Ca/Mg, and incubated with equal amount (50 lL/well) of cell suspension and TBS with Ca/Mg for 2 h at 37°C. Following incubation, plates were washed with TBS with Ca/Mg to remove nonadherent cells. The number of adherent cells was quantified by measuring intracellular acid phosphatase activity as described previouslyPark et al. (2010). These assays were performed in triplicates and repeated with two different isolations of RPE cells.

Phagocytosis assays
Phagocytosis activity was assayed using the unique pHrodo TMbased system (Life technologies) based on the acidification of the particles as they are ingested. pHrodo TM dye is a fluorogenic dye that greatly increases in fluorescence intensity when the pH becomes more acidic. RPE cells in 35-mm culture plates were loaded with medium containing pHrodo TM Green E. coli BioParticles conjugates and incubated for different time points (5 and 24 h). Following incubation, cells were rinsed with PBS containing 0.04% EDTA, and incubated with 1.5 mL of cell dissociation solution (Sigma). Cells were then washed, collected from plates, washed twice with PBS, resuspended in 0.5 mL of PBS, and analyzed by a FAC-Scan caliber flow cytometry (Becton Dickinson).

Proteasome peptidase assays
Proteasome peptidase assays were performed as previously described by us Aghdam et al. (2013). The cells were collected and lysed using the lysis buffer containing 50 mmol/L Tris-HCl pH 7.4, 1 mmol/L ATP, 10% glycerol, 0.1% NP40, 2 mmol/L MgCl 2 , 1.5 mmol/L DTT, and 0.03% SDS. BCA protein assay (Bio-Rad) was used to determine protein concentrations. The protein lysates (50 lg/well) were added into a dark 96-well microplate and probed with the fluorogenic peptide substrates: Z-Leu-Leu-Glu-AMC for caspase-like, Ac-Arg-Leu-Arg-AMC for trypsin-like, and Suc-Leu-Leu-Val-Tyr-AMC for chymotrypsin-like (Enzo Life Sciences, Farmingdale, NY) at the final concentration of 100 lmol/L in a total volume of 0.1 mL. The plates were incubated in the dark for 30 min at 37°C. Following incubation, the reaction was stopped using 100% ethanol (100 lL/well). The fluorescent intensity was measured using a plate reader (PerkinElmer, Wellesley, MA). The excitation and emission wavelengths were 355 and 460 nm, respectively.

Secreted VEGF levels
The level of VEGF produced by WT, TSP1À/À, and PEDFÀ/À RPE cells were determined using the Mouse VEGF Immunoassay Kit (R& D Systems). Cells (1 9 10 6 ) were plated in 60-mm tissue culture dishes and allowed to reach 90% confluence. The cells were washed with serum-free DMEM and incubated with 2 mL of growth medium without serum for 2 days. The CM were collected, centrifuged at 400 9 g for 5 min to remove cell debris, and used for VEGF measurements as recommended by the supplier.

Capillary morphogenesis assays
Tissue culture plates (35 mm) were coated with 0.5 mL Matrigel (10 mg/mL, BD Biosciences) and incubated at 37°C for at least 30 min to allow the gel harden. Mouse choroidal EC (ChEC) were prepared and maintained as previously described by us Lavine et al. (2013). The ChEC were removed by trypsin-EDTA, washed with DMEM containing 10% FBS. Cells (1 9 10 5 cells/mL) were resuspended in collected conditioned medium (CM) form wild type, PEDFÀ/À and TSP1À/À RPE cells. Cells (2 mL) were applied to the Matrigel-coated plates, incubated at 37°C, photographed after 14 h using a Nikon microscope in a digital format. For quantitative assessment of the data, the mean number of branch points was determined by counting the branch points in five high-power fields (9100).
Standard curves were generated from known quantities for each of the target gene of linearized plasmid DNA. Ten times dilution series were used for each known target, which were amplified using SYBR-Green qPCR. The linear regression line for ng of DNA was determined from relative fluorescent units (RFU) at a threshold fluorescence value (Ct) to quantify gene targets from cell extracts by comparing the RFU at the Ct to the standard curve, normalized by the simultaneous amplification of RpL13a, a housekeeping gene. The following primers for RpL13A 5'-TCTCAAGGTTGTTCGGCTGAA-3 0 (forward) and RpL13A 5 0 -CCAGACG CCCCAGGTA-3 0 (reverse) were used.

Ex vivo sprouting of RPE-choroid complex
Choroidal explants were prepared and cultured as described previously Kobayashi et al. (1998) modifications. Briefly, postnatal day 21 (P21) mice were anesthetized using isoflurane and killed by cervical dislocation. Eyes were enucleated, washed three times, and kept in ice-cold serum-free DMEM. Attached tissues to the outer surface of the eyeball (blood vessels, fatty and connective tissues) were shaved in ice-cold DMEM under a dissection microscope. The cornea, lens, and corpus vitreum were removed before the intermediate segment containing the sclera, choroid, RPE, and the retina was dissected along the whole circumference. The neuroretinal was removed before the sclera, and the RPE-choroid complex were sectioned into 0.5-1.0 mm pieces. These pieces were finally transferred into 35-mm culture dishes coated with 0.5 mL of Matrigel (10 mg/mL) (BD Biosciences) and allowed to harden. Tissue preparations were transferred into a 37°C cell culture incubator without medium for 20 min. Endothelial cell growth medium (DMEM containing 10% FBS, 2 mmol/L L-glutamine, 2 mmol/L sodium pyruvate, 20 mmol/L HEPES, 1% nonessential amino acids, 100 lg streptomycin, 100 U/mL penicillin, 55 U/mL heparin, and endothelial growth supplement 100 lg/mL; Sigma) was then added (2 mL/dish) and incubated at 37°C with 5% CO 2 for eight days. Explants were fed every other day. After 8 days, preparations were fixed in 4% PFA for 30 min at room temperature, washed three times in PBS before they were imaged using a Nikon microscope. Ten explants per eye were prepared and cultured in a single dish, and at least three mice per genotype were used.

NO measurements
The intracellular NO levels produced by WT, TSP1À/À, and PEDF À/À RPE cells was determined using 4-amino-5-methylamino-2, 7-difluoroflourescein diacetate (DAF-FM diacetate; Invitrogen). DAF-FM is produced after deacetylation of DAF-FM diacetate by intracellular esterases. The increased fluorescence intensity of DAF-FM due to the reaction with NO is measured with a fluorescein filter. Cells (5 9 10 3 cells/0.1 mL) were plated on gelatincoated 96-well black/clear bottom plates (BD Biosciences) and incubated overnight. The next day, medium was removed and 0.1 mL of RPE medium containing 30 lmol/L DAF-FM was added to each well and incubated for 40 min. The medium was replaced with fresh RPE medium and incubated for an additional 30 min. Following incubation, the cells were washed with TBS (0.1 mL, twice) and fluorescence intensity was measured at an excitation of 458 nm and an emission of 535 nm using a fluorescent microplate reader (Victa21420 Multilabel Counter, PerkinElmer). This assay was conducted in triplicates and repeated with two different isolation of RPE cells.

Wholemount staining studies
RPE layers were prepared form 4-week-old wild-type, PEDFÀ/À, and TSP1 À/À mice. Eyes were fixed in 4% PFA in PBS for 2 h. Following fixation, RPE, choroid, and sclera complex was prepared by removing the retina and cutting the optic nerve from the posterior portion of the eyes. The remaining cup was washed with PBS three times for 10 min, and blocked in blocking solution (50% FCS, 20% normal goat serum; NGS, in PBS containing 0.01% Triton 100X) at RT for 1 h. After blocking samples were washed with PBS three times for 10 min and incubated with primary antibodies including anti-iNOS, anti-HNE (Cell Signaling), anti-eNOS (SC-654) (Santa Cruz), and anti-ZO-1 (Life Technology) in PBS containing 20% FCS, 20% NGS, and 0.01% TX-100 overnight at 4°C. Samples were then washed with PBS and incubated with specific Cy3-conjugated secondary antibody (1:800; Jackson ImmunoResearch) in PBS containing 20% FCS, 20% NGS, and 0.01% TX-100 at room temperature for 2-4 h. Following incubation, samples were washed five times with PBS, flattened by four radial cuts, mounted on the slide, and examined using a fluorescence microscope (Carl Zeiss Optical) and the images were taken in digital format.

Assessments of reactive oxygen species
The level of cellular reactive oxygen species was measured with dihydroethidium staining (DHE). DHE is a weak blue fluorescent dye which binds to DNA upon oxidation to red fluorescent ethidium by O À 2 . Cells (3 9 10 4 ) were plated on fibronectin-coated chamber slide (5 lg/mL; BD Bioscience) and incubated for 24 h in growth medium. Following incubation, the cells were exposed to 10 lmol/L DHE for 15-20 min, washed with growth medium and incubated twice with growth medium each for 30 min. A similar experiment was performed in the presence of 50 lmol/L H 2 O 2 (positive control; Fisher Scientific). Digital images were captured at the same exposure time for all samples. For quantitative analysis, images were analyzed with Image J software (National Institute of Health, Bethesda, MD). The mean fluorescent intensity was determined from intensities of multiple cells (100 cells). A similar experiment was performed using cells treated with 50 lmol/L H 2 O 2 for 24 h, and three independent experiments with two different isolation of RPE cells were performed.

Statistical analysis
Statistical differences between control and treated samples were evaluated using Graphpad Prism software (La Jolla,

Results
Isolation and characterization of PEDFÀ/À and TSP1À/À RPE cells RPE cells were prepared from 4-week-old wild-type, TSP1À/À and PEDFÀ/À Immorto mice as described in Material and Methods. The expression of RPE-specific markers was confirmed by FACS analysis (Fig. 1A). All cells expressed significant amount of bestrophin and RPE 65, as expected. The results are representative of sample mean from three different isolations, and there were no statistically significant differences between the cells. The PEDFÀ/À RPE cells exhibited a more spindle-shaped and elongated morphology compared to wild-type cells on gelatin-coated plates, while TSP1À/À RPE cells exhibited a similar morphology as wild-type RPE cells (Fig. 1B).
To determine the properties of PEDFÀ/À, TSP1À/À, and wild-type RPE cells, we next analyzed the expression of other RPE cell markers by FACS ( Fig. 2A) and Western blot (Fig. 2B) analysis. Alteration in some of these markers including (ICAM-1, VCAM-1, and VEGF receptors) has been reported as underlying factors in the development of AMD. The decreased expression level of ICAM-1 was observed in PEDF À/À RPE and TSP1À/À compared with wild-type cells. However, the PEDFÀ/À RPE cells represented the most dramatic change compared to wild-type cells. The expression of other markers including CD36, CD47, VEGFR-1, and VCAM-1 was observed in all RPE cells with modest variations among the cell types. We also determined VEGFR-2, endoglin, and ICAM-2 expression, which were undetectable in all cells ( Fig. 2A). PDGF-Rb signaling plays a key role in the RPE cell migration and proliferation through activation of the downstream signaling pathways (Schlingemann 2004). The TSP1À/À RPE cells exhibited a dramatic increase in PDGF-Rb expression compared with PEDFÀ/À and wild-type cells ( Fig. 2A). Similarly increased levels of PDGF-Ra expression was observed in TSP1À/À and PEDFÀ/À RPE cells compared to wild-type cells.
In the retina, the secreted milk fat globule-EGF factor 8 (MFG-E8) is involved in clearance of shed photoreceptor outer segment (POS) by ligating avb5 integrin. Absence of either MFG-E8 ligand or avb5 receptor results in abolished RPE phagocytosis (Finnemann et al. 1997;Nandrot et al. 2007). MFG-E8 also plays an important role in apoptotic cell clearance, angiogenesis, and adaptive immunity (Deroide et al. 2013). MFG-E8 protein levels were examined by Western blot analysis. We observed a modest decrease in expression level of MFG-E8 in PEDFÀ/À, but not TSP1À/À, RPE cells compared with wild-type cells. We also examined protein disulfide isomerase (PDI) expression in RPE cells. PDI is a crucial factor for cell viability, which plays a critical function during protein folding (Hahm et al. 2013). The downregulation of PDI in human RPE cells may lead to protein aggregation and endoplasmic reticulum stress (Yokoyama et al. 2006). PEDFÀ/À RPE cells expressed increased level of PDI compared with wild-type and TSP1À/À RPE cells. A significant decrease in the level of ICAM-1 was observed in TSP1À/À and PEDFÀ/À cells consistent with FACS results.
Adherens junctions facilitate intercellular adhesions, a process with major roles in maintaining tissue integrity and normal morphology of RPE cells (Bailey et al. 2004).The altered morphology of PEDFÀ/À RPE cells suggested altered cell-cell interactions. We next determined the localization and expression of junctional protein complexes by indirect immunofluorescence staining (Fig. 3A) and Western blot analysis (Fig. 3B). N-cadherin is reported as the dominant cadherin in RPE cells with A B C Figure 3. Cellular localization and expression of junctional proteins. (A) The localization of ZO-1, N-cadherin, b-catenin, and P-120 catenin was determined by immunofluorescence staining. The WT, PEDFÀ/À, and TSP1À/À RPE cells were plated on fibronectin-coated chamber slides and stained with specific antibodies as detailed in Material and Methods. No staining was observed in the absence of primary antibody (not shown). Please note the absence of N-cadherin from the sites of cell-cell contact in PEDFÀ/À RPE cells, as well as reduced junctional localization in TSP1À/À RPE cells. (B) Western blot analysis of junctional proteins. Total cell lysates were prepared from WT, PEDFÀ/À, and TSP1À/À RPE cells and analyzed for expression level of ZO-1, N-cadherin, bcatenin, P-120 catenin, and b-actin. Please note decreased expression of ZO-1 proteins in PEDFÀ/À RPE cells. The TSP1À/À RPE cells expressed increased levels of N-cadherin and P-120 catenin. The b-catenin level was increased in TSP1À/À RPE cells. The b-actin was used for loading control. The quantification of data is shown in (C) (**P < 0.01, ***P < 0.001, n = 3). These experiments were repeated using three different isolations of RPE cells with similar results. important roles in cell migration and photoreceptor survival, and localization at cell-cell junctions (Kaida et al. 2000;Chen and Ma 2007). Here, we observed lack of N-cadherin junctional localization in PEDFÀ/À RPE cells compared with the wild-type cells (Fig. 3A), while no significant difference was observed in the N-cadherin expression level compared with wild-type cells ( Fig. 3B and C). TSP1À/À RPE cells showed increased levels of N-cadherin compared with wild-type or PEDFÀ/À cells.
In addition, a band with higher molecular weight was detected in TSP1À/À RPE cells for N-cadherin. Thus, the absence of PEDF and TSP1 was associated with defects in cell-cell interactions and junction formation in RPE cells.
Tight junctions are complex, dynamic structures that regulate cell proliferation, polarity, and paracellular diffusion and permeability. Formation of tight junctions is important for appropriate barrier function of the RPE cells. ZO-1 plays a major role in the formation and stabilization of tight junctions (Rizzolo et al. 2011). The PEDF À/À and TSP1À/À RPE cells exhibited decreased levels of ZO-1 (Fig. 3B). However, no significant changes in junctional localization of ZO-1 were observed in TSP1À/À and PEDFÀ/À RPE cells compared with the wild-type cells (Fig. 3A). In addition, ZO-1 showed enhanced nuclear localization in TSP1À/À RPE cells compared to wild-type and PEDFÀ/À cells, without impacting the localization of other components of adherens junction complexes. Similar results were observed in RPE-choroid complex prepared from TSP1À/À mice in vivo (not shown). We did not observe any effect on ZO-1 nuclear localization in PEDFÀ/À RPE cells. Junctional localization of p120-catenin, another adherens junctional protein, was similarly observed in all RPE cells. However, an increase in p120-catenin level was observed in PEDFÀ/À and TSP1À/À RPE cells (Fig. 3B). The b-catenin level was significantly increased in PEDFÀ/À RPE cells compared to wild-type cells. However, b-catenin localization was not affected by PEDF and TSP1 deficiency in RPE cells ( Fig. 3B and C). To roll out the possible impact of large T antigen and INF-c present in growth medium, similar experiments were conducted in the absence of INF-c at 37°C to eliminate large T antigen expression with similar results (not shown). Collectively, our results indicated that deficiency in PEDF or TSP1 had a significant impact on junctional properties of RPE cells.
PEDFÀ/À and TSP1À/À RPE cells displayed a greater rate of proliferation RPE cell integrity is essential for proper retina function. The demonstrated impact of PEDF and TSP1 on junctional properties of RPE cells suggested alterations in proliferation of these cells. RPE cells are normally postmitotic in adults and their activation is associated with proliferative vitro retinopathy (PVR) (Hecquet et al. 2002). Choroidal neovascularization (CNV), a major pathologic condition in patient with AMD, at the early stage is associated with hyperplastic RPE cells. However, in the late stages of CNV the RPE cells proliferate in a monolayer to enclose CNV and promote CNV regression (Bhutto and Lutty 2012).The rate of cell proliferation was determined by counting the number of cells for two weeks. An increase in the rate of proliferation was observed in PEDFÀ/À and TSP1À/À RPE cells compared to the wild-type cells (Fig. 4A). We next examined whether the increase in rate of proliferation was the result of enhanced DNA synthesis. The percentage of cells undergoing active DNA synthesis was determined by EdU labeling, a synthetic nucleoside analog. The rate of DNA synthesized in PEDFÀ/À and TSP1À/À RPE cells was significantly greater than that observed in wild-type cells (Fig. 4B). Thus, PEDF or TSP1 deficiency is associated with enhanced RPE cell proliferation.
The apoptosis of RPE cells plays an important role in the development of different retinal degenerative diseases (Mascarelli et al. 2001). Human AMD studies suggest that RPE cells, photoreceptors, and inner nuclear layer cells die by apoptosis (geographic atrophy; GA) (Dunaief et al. 2002). We next determined whether lack of PEDF and TSP1 in RPE cells affects their rate of apoptosis. TUNEL staining was used to determine the rate of apoptosis in RPE cells under basal and challenged conditions. A similar rate of apoptosis was observed in all RPE cells at the basal level (Fig. 4C). However, under stress conditions (50 lmol/L H 2 O 2 ) PEDFÀ/À and TSP1À/À RPE cells exhibited significantly higher rate of apoptosis compared with wild-type cells ( Fig. 4D; P < 0.05; n = 5). Thus, lack of TSP1 or PEDF results in increased sensitivity of RPE cells to oxidative challenge.

PEDFÀ/À and TSP1À/À RPE cells were less migratory
Migration and proliferation of RPE cells contribute to a number of pathologies including PVR (Pastor et al. 2002). Scratch wound assays were conducted to investigate the impacts of PEDF or TSP1 deficiency on RPE cell migration. A significant delay in wound closure was observed in PEDFÀ/À and TSP1À/À RPE cells compared to wild-type cells (Fig. 5A). The quantitative assessment of the data is shown in Fig. 5B. Similar results were observed using a transwell migration assay (Fig. 5C). Reexpression of PEDF or TSP1 restored normal migration of PEDFÀ/À and TSP1À/À RPE cells PEDFÀ/À and TSP1À/À cells displayed decreased migration compared with wild-type cells. To examine if reexpression of PEDF or TSP1 is sufficient to restore normal migration in PEDFÀ/À or TSP1À/À cells, we reexpressed PEDF and TSP1 in PEDFÀ/À and TSP1À/À cells. Reexpression of PEDF and TSP1 was confirmed by Western blot analysis of cell lysates and conditioned medium ( Fig. 5D and E). A significant improvement in migration of RPE cells was observed in PEDFÀ/À and TSP1À/À cells upon reexpression of PEDF and TSP1 (Fig. 5F).
To confirm the change in migratory phenotype of PEDFÀ/À and TSP1À/À RPE cells, we examined the formation of actin stress fibers and focal adhesions using indirect immunofluorescence staining. Increased number of focal adhesions was observed in PEDFÀ/À and TSP1À/À RPE cells compared to wild-type cells (Fig. 6A). The TSP1À/À and PEDFÀ/À RPE cells were also spread more with stronger peripheral actin staining. These results were further confirmed by increased expression of FAK in PEDFÀ/À and TSP1À/À RPE cells compared with the wild-type cells (Fig. 6B). A significant increase in vinculin level was also observed in TSP1À/À RPE cells. These results are consistent with increased presence of focal adhesions, spreading, and reduced migration of TSP1À/À and PEDFÀ/À RPE cells.

PEDFÀ/À and TSP1À/À RPE cells were more adherent
Altered migration in PEDFÀ/À and TSP1À/À RPE cells suggested changes in their adhesion properties. We next examined the adhesion of RPE cells to various ECM proteins. PEDFÀ/À RPE cells exhibited stronger adhesion on collagen IV, collagen I, fibronectin, and vitronectin compared with wild-type cells (Fig. 7A). TSP1À/À RPE cells were also more adherent on fibronectin, vitronectin, collagen IV, and collagen I compared with the wild-type cells (Fig. 7A).
To determine if changes in adhesion were mediated by alterations in integrin expression, we next determined the expression level of different integrins by FACS analysis. Similar expression levels for a2, a3, a5, av, b1, b3, b5,

(D and E)
The re-expression of PEDF and TSP1 in PEDFÀ/À and TSP1À/À RPE cells. PEDFÀ/À and TSP1À/À cells were infected with viruses encoding PEDF or TSP1 cDNAs as detailed in Material and Methods. The expression of PEDF and TSP1 was confirmed by Western blot analysis. These lanes are all from the same membrane and extraneous lanes were removed for uniform labeling of the Figure. (F) Reexpression of PEDF and TSP1 in PEDFÀ/À and TSP1À/À RPE cells restored their migration to near WT RPE cell as determined by transwell migration assay (*P < 0.05, ****P < 0.0001, n = 3). a5b1, and avb3 were observed in these cells (Fig. 7B). There were no dramatic changes in the integrin expression. However, the possibility of alteration in affinity and avidity of these integrins cannot be excluded. RPE cells had undetectable levels of a1 integrin. The PEDFÀ/À RPE cells showed no detectable expression of b8 integrin. However, TSP1À/À and wild-type RPE cells expressed low levels of b8 integrin. Diverse biological events including wound closure, inflammatory responses, cell migration, and various developmental processes and angiogenesis are affected by changes in production of various ECM proteins. We next examined the level of various ECM proteins including fibronectin, collagen IV, tenascin C, SPARC, TSP1, TSP2, periostin, opticin, and PEDF in RPE cells (Fig. 8). TSP1À/À and PEDFÀ/À RPE cells expressed increased levels of fibronectin compared with wild-type cells. The level of cell associated collagen IV was decreased in PEDFÀ/À and TSP1À/À RPE cells compared to wildtype cells. PEDFÀ/À RPE cells showed a reduction in TSP1 production compared with wild-type cells. TSP2 expression was only detected in TSP1À/À RPE cells. The PEDFÀ/À RPE cells showed lower TSP1 production compared with wild-type cells. In contrast, increased expression of PEDF was observed in TSP1À/À RPE cells compared with wild-type cells. The TSP1À/À and PEDFÀ/À RPE cells expressed increased levels of tenascin C in both lysates and CM compared with wild-type cells.

SPARC expression was not affected in RPE cells (not shown). PEDFÀ/À and TSP1À/À RPE cells exhibited increased periostin expression compared with wild type cells. Opticin was not detectable in PEDFÀ/À RPE cells while TSP1À/À cells exhibited a dramatic increase in op-
ticin expression compared to wild-type cells. Thus, our results suggest PEDF and TSP1 expression have major effects on ECM expression impacting adhesive and migratory properties of RPE cells.

RPE cell phagocytosis
The phagocytic function of RPE cells is crucial for removal of toxic metabolites and RPE cell survival. Defects in photoreceptor phagocytosis can results in severe retinal pathologies (Sparrow et al. 2010;Mustafi et al. 2013).To determine the impact of PEDF and TSP1 deficiency on RPE cell phagocytic activity, phagocytosis was assayed using the unique pHrodo TM-based system as detailed in Material and Methods. Enhanced accumulation of phagocytosis activity was observed in PEDFÀ/À RPE cells compared with wild-type and TSP1À/À cells (Fig. 9A). A similar trend was observed at later incubation time. Enhanced accumulation of the phagocytized particles may be associated with altered proteasome and/ or lysosomal activity (Suraweera et al. 2012).
We next assessed the proteasome activity of RPE cells using an in vitro peptidase assay as detailed in Material  a1, a2, a3, a5, av, b1, b2, b3, b5, b8, a5b1, and aVb3 integrin expression was determined by FACS analysis as described in Material and Methods. Please note lack of a1 expression in RPE cells; while b8 integrin was absent in PEDFÀ/À RPE cells. These experiments were repeated using three different isolations of RPE cells with similar results. and Methods. A decrease in caspase-like and trypsin-like activity was observed in PEDFÀ/À RPE cells compared with wild-type cells, while TSP1À/À cells exhibited an increase in caspase-like and trypsin-like activity compared to the wild-type cells. All cells similarly showed lower chymotrypsin-like activity (Fig. 9B). Thus, the lack of PEDF, but not TSP1, in RPE cells was associated with impaired proteasome activity. To determine potential changes in lysosomal system, we also examined the level of cathepsin-B, the major lysosomal cysteine protease, in lysates prepared from RPE cells by Western blot analysis. PEDFÀ/À cells expressed similar level of cathepsin-B as observed in wild-type cells. In contrast, a dramatic decrease in the level of cathepsin-B was observed in TSP1À/À RPE cells suggesting reduced lysosomal activity in TSP1À/À RPE cells.

Alterations in the expression of VEGF and NOS isoforms in RPE cells
Alterations in the secretory activity of RPE cells are associated with proliferative and degenerative diseases in the retina (Strauss 2005). The increased production of VEGF has been identified as essential factor in the development and progression of AMD and CNV (Kliffen et al. 1997;Yamagishi et al. 2003). Many studies have reported that RPE cells are the major source of VEGF (Adamis et al. 1993;Kliffen et al. 1997;Nagineni et al. 2003). VEGF is considered a survival factor for many cell types including RPE cells (Byeon et al. 2010). To determine the impact of PEDF and TSP1 deficiency on VEGF production, we measured VEGF levels in conditioned medium prepared form PEDFÀ/À and TSP1À/À RPE cells. A significant decrease in VEGF level was observed in PEDFÀ/À RPE cells compared to wild-type cells. In contrast, TSP1À/À RPE cells showed increased VEGF levels compared with wild-type cells (Fig. 10A). Both PEDF and TSP1 levels are dramatically downregulated in human's exudative AMD samples. Thus, PEDF and TSP1 deficiency have a significant impact on VEGF production in RPE cells and pathogenesis of exudative AMD. Altered expression of NOS isoforms including reduction in nNOS in RPE nuclei has been reported in AMD (Bhutto et al. 2010). We next examined the expression of various NOS isoforms in PEDF À/À and TSP1À/À RPE cells by Western blot analysis. NO modulates the angiogenic response of various key factors including VEGF (Papapetropoulos et al. 1997). Additionally, NO production is essential for VEGF-mediated angiogenesis (McLaren et al. 2006). We observed no significant difference in NO levels produced by RPE cells (Fig. 10B).
We detected expression of the inducible NOS (iNOS) in PEDFÀ/À RPE cells. Wild-type RPE cells exhibited low level of endothelial NOS (eNOS) expression. While, eNOS expression was undetectable in PEDFÀ/À and TSP1À/À RPE cells (Fig. 10C). Wholemount staining of RPE/choroid complex from wild-type, PEDFÀ/À and TSP1À/À mice showed similar results (Fig. 10D). The expression of nNOS was not detectable in RPE cells by Western blot analysis. A B C Figure 9. Altered Phagocytic activity in PEDFÀ/À RPE cells. (A) RPE cells from 35-mm culture plates were loaded with DMEM medium containing pHrodo TM Green E. coli BioParticles conjugates and incubated for different time points (5 and 24 h). Please note the increased phagocytosis activity in PEDFÀ/À RPE cells compared to WT and TSP1À/À cells. (B) The in vitro proteasome peptidase activity of cultured RPE cells. Chymotrypsin-like activity was not affected by PEDF and TSP1 deficiency. However, trypsin-like activity was significantly decreased in PEDFÀ/À and TSP1À/À RPE cells (*P < 0.05, **P < 0.01, n = 3). Please also note a significant decrease in caspase-like activity in PEDFÀ/À RPE cells, while TSP1À/À RPE cells exhibited increased caspase-like activity (*P < 0.05, **P < 0.01, n = 3). (C) Attenuation of cathepsin-B expression in TSP1À/À RPE cells. Please note the dramatic decrease in the level of cathepsin-B in TSP1À/À RPE cells compared with WT and PEDFÀ/À RPE cells. These experiments were repeated using three different isolations of RPE cells with similar results. Capillary morphogenesis of choroidal endothelial cells The progression of visual loss in the exudative AMD is associated with abnormal CNV membranes that develop under the retina. This results in leaking serous fluid and blood, and ultimately a blinding disciform scar in and under the retina (Ciulla et al. 1999). We showed that PEDF and TSP1 deficiency differentially impact VEGF secretion in RPE cells. We next determined the effects of conditioned medium collected from PEDFÀ/À and TSP1À/À RPE cells on capillary morphogenesis of choroidal endothelial cells (ChEC) and ex vivo sprouting of choroid-RPE complex. Endothelial cells undergo capillary morphogenesis forming a capillary-like network when plated on Matrigel. This mimics the late stages of angiogenesis (Solowiej et al. 2003). ChEC were isolated and cultured as previously described Lavine et al. (2013).
Here, we plated the ChEC, resuspended in CM prepared from wild-type and, PEDF or TSP1 null RPE cells, on Matrigel. Capillary morphogenesis of ChEC was analyzed and quantified as described in Material and Methods. Conditioned medium from PEDFÀ/À RPE cells decreased capillary morphogenesis of ChEC compared with wildtype cells, while CM from TSP1À/À RPE cells increased capillary morphogenesis of ChEC ( Fig. 11A and C). These results are consistent with increased production of VEGF by TSP1À/À RPE cells (Fig. 10A). Representative images of capillary morphogenesis are shown in Fig. 11A. We also assessed the ex vivo sprouting of the choroid-RPE complex prepared from wild-type, TSP1À/À, and PEDFÀ/À mice eyes. In contrast to enhanced capillary morphogenesis of ChEC treated with TSP1À/À CM, we observed no significant differences in degree of sprouting angiogenesis in TSP1À/À and PEDFÀ/À mice compared to wild-type mice ( Fig. 11B and D).

PEDF and TSP1 deficiency increased ROS generation in RPE cells
Most vascular diseases and their subsequent pathologies are associated with alteration in the cellular oxidative state. Oxidative stress acts as key initiator of inflammation, and is considered a major contributor to the pathogenesis of AMD (Zhou T et al. 2010). Decreased ability of RPE cells to cope with oxidative stress and absence of oxidative defense can promote ocular neovascularization and retinal degeneration (Dong et al. 2009).
Using dihydroethidium staining, we examined ROS production at basal level and following incubation with 50 lmol/L H 2 O 2 . PEDFÀ/À and TSP1À/À RPE cells demonstrated more fluorescent intensity compared with wild-type cells indicating increased ROS production under basal level or stress condition ( Fig. 12A and B). The quantitative assessments of the data are shown in Fig. 12C and D. To further confirm our results, we investigated whether PEDFÀ/À and TSP1À/À RPE cells exhibit an increase in intracellular accumulation of ROS in vivo. Using wholemount RPE/choroid complexes prepared from wild-type, TSP1À/À, and PEDFÀ/À mice the level of oxidative state was determined by 4-hydroxynonenal (4-HNE) staining as described in Material and A B E C D Figure 11. Increased oxidative stress in PEDFÀ/À and TSP1À/À RPE cells. The level of oxidative stress in RPE cells was assessed by dihydroethidium (DHE) staining under basal (A) or challenged conditions (50 lmol/L H 2 O 2 ) (B). A significant increase in the level of ROS was detected under basal (C) and challenged (D) conditions in PEDFÀ/À and TSP1À/À RPE cells compared to WT cells (**P < 0.01, n = 3). (E) Wholemount staining of RPE-choroid tissue prepared from PEDFÀ/À, TSP1À/À and WT mice with antibody to 4-hydroxy-2-nanonal (HNE). Please note increased HNE staining in RPE-choroid tissues from PEDFÀ/À and TSP1À/À mice compared to Wild type mice. These experiments were repeated using three different isolations of RPE cells, and eyes from five different mice with similar results. Methods. The 4-HNE is a key reactive product of lipid peroxidation, and indicator of oxidative stress (Yang et al. 2003). The staining of RPE/choroid complex from PEDFÀ/À and TSP1À/À mice exhibited increased 4-HNE staining compared with wild type mice, indicating increased oxidative stress in the absence of PEDF and TSP1 (Fig. 12E). Our results demonstrated that PEDF and TSP1 deficiency is associated with increased oxidative stress in RPE cells. These results are consistent with decreased levels of these proteins in eye samples from AMD patients.

PEDF and TSP1 deficiency resulted in production of inflammatory mediators in RPE cells
Increased oxidative stress in PEDFÀ/À and TSP1À/À RPE cells suggested potential alteration in the inflammatory response of these cells. We next examined the expression of inflammatory mediators in PEDFÀ/À and TSP1À/À RPE cells. Figure 13 shows a dramatic increase in expression of monocyte chemotactic protein-1 (MCP-1) in both TSP1À/À and PEDFÀ/À RPE cells compared to the wild-type cells. PEDFÀ/À RPE cells exhibited a dramatic increase in production of IL-18 compared with TSP1À/À RPE and wild-type cells. Similar trend was observed for TNF-a in PEDFÀ/À RPE cells, however it was not significant. We also observed increased RANTES production in TSP1À/À RPE cells compared to the PEDFÀ/À RPE and wild-type cells. These results indicated that the absence of PEDF and TSP1 is associated with a distinct inflammatory phenotype of RPE cells. Our results are consistent with previous studies establishing that the inflammatory responses in the retina and RPE play a crucial role in the development and progression of AMD and is further supported by detection of decreased levels of TSP1 and PEDF in patients with exudative AMD. The role of MAPK pathways in oxidative stress signaling is controversial, with MAP kinases JNK and p38 shown to be protective, proapoptotic or not involved (Klettner 2012). The ERK1/2 is considered to be involved in cell proliferation and differentiation. However, it may also contribute to an apoptotic response (Gilley et al. 2009). Increased ERK1/2 activation of RPE cell has been demonstrated in the AMD patients with GA, and its inhibition shown therapeutic benefit (Dridi et al. 2012). TSP1À/À RPE cells demonstrated increased levels of p-ERK1/2, and lower levels of total ERK1/2 compared to PEDFÀ/À and wild-type cells. Thus, there is a potential mechanism whereby decreased TSP1 expression may result in increased ERK1/2 activation in RPE cell of AMD patients. PEDFÀ/À and TSP1À/À RPE cells exhibited an increase in p-JNK level compared with wild-type cells, without affecting the total level of JNK (Fig. 14). The level of p-p38 was increased in TSP1À/À RPE cells compared with PEDFÀ/À and wild-type cells. We did not observe any significant change in the total level of p38 in all cells. Altered proliferation rate in PEDFÀ/À and TSP1À/À suggested a potential change in AKT activation and/or expression. AKT plays a key role in cell survival and proliferation (Klettner 2012). Here, we observed decreased AKT activation in TSP1À/À RPE cells compared with PEDFÀ/À and wild-type cells. However, increased expression of total AKT was detected in PEDFÀ/À and TSP1À/À RPE cells (Fig. 14). Thus, absence of PEDF and TSP1 were associated with altered MAPK and AKT signaling pathways, which may contribute to the altered cellular functions observed in these cells.

Discussion
The clinical and pathological signs of AMD, especially the dry form or the geographic atrophy, are associated with degeneration and loss of RPE cells. Although an important role for RPE cell dysfunction has also been demonstrated in exudative form of AMD, these roles may be somewhat distinct and need further elucidation. RPE is a monolayer of specialized epithelial cells which perform a wide variety of functions to preserve retina. However, the detailed mechanisms regulating RPE cell function remain poorly understood. One major function of RPE cells is the maintenance of ocular vascular homeostasis. This is accomplished through a balanced production of pro-and antiangiogenic factors including VEGF, PEDF, and TSP1. However, how these factors impact RPE cell function are poorly understood. Here we determined the impact of PEDF and TSP1 deficiency on RPE cell function. Using RPE cells prepared from PEDFÀ/À and TSP1À/À mice, we showed that the PEDF or TSP1 deficiency resulted in increased proliferation of RPE cells. These cells were also more adherent and less migratory, exhibited increased oxidative stress with minimal impact on basal levels of apoptosis. The PEDF and TSP1 deficiency were also associated with significant changes in production of ECM proteins, cell-cell junctional organization, increased oxidative sensitivity, and altered phagocytic activity. A summary of changes observed in PEDFÀ/À and TSP1À/À RPE cells compared with wild-type cells is shown in Table 1. The reexpression of PEDF or TSP1 was sufficient to restore normal migration of PEDFÀ/À and TSP1À/À RPE cells. Thus, the cell autonomous expression and/or activity of PEDF and TSP1 are essential for appropriate RPE cell function. Furthermore, our results provide a better understanding of the underlying mechanisms, such as altered production of PEDF and TSP1, which contribute to RPE cells dysfunction and pathogenesis of AMD.
The ability to gain insight into RPE cell autonomous regulatory mechanisms and function of various genes has been limited by the unavailability of methods for routine culturing of these cells from wild-type and transgenic mice retina. Here, we established a method to isolate and culture RPE cells from PEDFÀ/À, TSP1À/À and wildtype mice crossed with the immorto mouse. These cells allowed us, for the first time, to gain novel insight into the autonomous role PEDF and TSP1 play in regulation of RPE cell function. These cells expressed the RPE cell markers including RPE65 and bestrophin. However, the morphology of PEDFÀ/À RPE cells was different from wild-type and TSP1À/À RPE cells, suggesting altered cell-cell and cell-matrix interactions. The morphology of Figure 14. Alterations in the downstream AKT and MAPK signaling pathways. The activation of MAPK and AKT pathways were evaluated by Western blot analysis using specific antibodies as detailed in Material and Methods. A significant change in these pathways were detected in PEDFÀ/À and TSP1À/À RPE cells compared with WT cells (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n = 3). Please note decreased levels of p-Akt and p-p38 in TSP1À/À RPE cells compared to WT and PEDFÀ/À cells. However, p-JNK was increased in PEDFÀ/À and TSP1À/À RPE cells compared with WT cells. TSP1À/À RPE cells also expressed significantly higher levels of p-ERKs compared to PEDFÀ/À or WT cells. These experiments were repeated with two different isolations of RPE cells with similar results.  PEDFÀ/À RPE cells observed here were consistent with those reported in RPE cells cultured in the absence of retinoic acid. The incubation of RPE cells with retinoic acid resulted in a more hexagonal and epitheloid morphology, instead of an elongated and spindle-like fibroblast appearance. This was mainly attributed to increased levels of PEDF in RPE cells incubated with retinoic acid (Uchida et al. 2005). In addition, the increased level of PEDF in TSP1À/À RPE cells may be responsible, at least in part, for their similar morphology to wild-type cells.
Although the junctional localization of cell adhesion proteins was not dramatically affected in PEDFÀ/À and TSP1À/À RPE cells, a significant decrease in the expression of ZO-1 was observed. The tight junctional protein ZO-1 is critical for normal cell proliferation, differentiation, and homeostasis of RPE cells. Consistent with our results a marked disruption of the tight-junction-associated protein ZO-1 and abnormal RPE morphology has been demonstrated in human AMD patients. Furthermore, the total level of b-catenin was also increased in PEDFÀ/À RPE cells but not TSP1À/À RPE cells, which have significant amounts of PEDF. This is consistent with the proposed role of PEDF in inhibition of Wnt/b-catenin signaling pathway (Park et al. 2011). The pathological role of Wnt signaling pathway in AMD has been previously reported (Zhou T et al. 2010). Thus, lack of PEDF may result in increased levels of b-catenin and sustained activation of Wnt signaling pathway and development of AMD. This may be compensated to some extent by downregulation of TSP1 expression and increased PEDF expression, as observed in TSP1À/À RPE cells. However, the detailed regulatory mechanisms involved in coordinated regulation of PEDF and TSP1 need further investigation.
The N-cadherin level was increased in TSP1À/À RPE cells. However, a significant increase in p120 catenin was observed in both PEDFÀ/À and TSP1À/À cells. Interestingly, the localization of N-cadherin to sites of cell-cell contact was decreased in PEDFÀ/À RPE cells consistent with their elongated spindly morphology and sustained activation of Wnt signaling. The P120 catenin plays an important role in formation of adherence junctions, and its increased production may be a compensatory feedback for defects in cellular adherens and gap junctions. Thus, PEDF and TSP1 may have significant impact on ZO-1 expression with potential impact on its junctional localization, stability, and barrier function.
The proliferation of RPE cells plays a major role in the development of PVR (Pastor et al. 2002;Tamiya et al. 2010). Previous studies demonstrated that cell junctions and junctional proteins have a significant role in the proliferation of RPE cells. Loss of cell-cell contact contributes to RPE cell proliferation. In addition, downregulation of ZO-1 is involved in induction of RPE cell proliferation (Georgiadis et al. 2010;Tamiya et al. 2010). Here, we observed a significant increase in proliferation of PEDFÀ/ À and TSP1À/À RPE cells. These results are consistent with lack of appropriate N-cadherin junctional localization and decreased ZO-1 expression in PEDFÀ/À and TSP1À/À RPE cells. Furthermore, our results are supported by studies demonstrating that increased PEDF and TSP1 expression in RPE cells incubated with retinoic acid attenuates their proliferation (Ma et al. 2012;Du et al. 2013).
PEDF was initially identified as a contact inhibitor factor, since its expression was induced when cells reached confluence and stopped growing. PEDF as a trophic factor may also contribute to attenuation of proliferation by decreasing the number of cells entering S phase of the cell cycle and increasing the number of cells entering G0 (Barnstable and Tombran-Tink 2004). Thus, PEDF may play a significant role in RPE cells proliferation. The impact of TSP1 on cell proliferation is cell specific which is derived from conformational flexibility of TSP1 or different TSP1 receptors (Majack et al. 1986;Adams and Lawler 2011). TSP1 acts as a negative regulator of proliferation in endothelial cells and induces their apoptosis (Tolsma et al. 1993). However, TSP1 expression is essential for PDGF-mediated proliferation of perivascular supporting cells (Scheef et al. 2009). Here, we observed increased proliferation in TSP1À/À RPE cells, which was concomitant with increased levels of PDGF-Rb. Thus, TSP1 expression may also negatively contribute to RPE cell proliferation as occurs in endothelial cell, involving CD36 and CD47 receptors, both of which were expressed in RPE cells. Collectively, our results suggest PEDF and TSP1 expression are involved in modulating RPE cell-cell adhesion and proliferation, and their absence resulted in altered adhesion and increased proliferation. Alteration in RPE cell migration contributes to some pathological conditions including AMD, proliferative diabetic retinopathy (PDR), and PVR (Dong et al. 2009;Chan et al. 2010). Here, we demonstrated that PEDF and TSP1 deficiency resulted in more adhesive and less migratory phenotype of RPE cells. Furthermore, reexpression of PEDF and TSP1 in PEDFÀ/À and TSP1À/À RPE cells was sufficient to restore RPE cell migration. These results are contradictory to those reported by Ma et al. (Ma et al. 2012) who showed increased PEDF expression as a result of over expression of microphthalmia-associated transcription factor or exogenous PEDF inhibited RPE cell migration. These contradictory results may be attributed to the different levels of PEDF in these studies and require further evaluation.
Cell migration is influenced by cell-ECM protein interactions. Thus, the abrogated migratory features of PEDFÀ/À and TSP1À/À RPE cells are consistent with their enhanced adhesion. Although, lack of PEDF and TSP1 was associated with minimal changes in expression of integrins, the possibility of alteration in avidity and affinity of integrins, however, cannot be excluded and requires further investigation. We detected altered expression of ECM proteins in PEDFÀ/À and TSP1À/ À RPE cells, which may contribute to the altered migratory and adhesive features of RPE cells with PEDF and TSP1 deficiency. Taken together, these observations suggest that PEDF and TSP1 play key roles in cell-cell and cell-ECM interactions, which modulate proproliferative and promigratory signaling pathways of RPE cells. Additionally, altered ECM composition of RPE cells due to PEDF and TSP1 deficiency may result in defective ocular angiogenesis such as abnormal CNV in AMD.
VEGF is a proangiogenic factor with prosurvival activity and maintenance of RPE integrity (Ford et al. 2011). The increased production of VEGF has been identified as essential in the development and progression of AMD and CNV. Here, we demonstrated that TSP1À/À cells expressed significantly higher levels of VEGF while, PEDFÀ/À cells produced significantly lower amounts of VEGF compared to wild-type cells. The increased level of VEGF in TSP1À/À RPE cells is consistent with the proangiogenic state of the retinal vasculature in global TSP1 null retina (Hiscott et al. 1996a,b) and enhanced neovascularization observed in mouse laser-induced CNV model (Wang et al. 2012).This was mainly attributed to increased recruitment of microglial cells to the site of lesions, perhaps through enhanced inflammatory signaling through the CCR5/RANTES (CCL5) axis in the absence of TSP1 (Zhou T et al. 2010;Lavalette et al. 2011). This notion is supported by observed increased production of RANTES/CCL5 in TSP1À/À RPE cells compared to PEDFÀ/À or wild-type mice. Furthermore, we have observed no significant differences in degree of CNV in PEDFÀ/À mice subjected to laser-induced CNV (our unpublished results). The reason for significant decrease in VEGF level in PEDF À/À RPE cells is not clear, and is in contrast to reports that increased VEGF level is associated with increased oxidative stress and decreased level of PEDF in diabetic ocular samples (Boehm et al. 2003). These results are, however, consistent with our previous findings where no association was observed between diabetic retinopathy where VEGF level is shown to be increased without a major effect on PEDF levels (Wang et al. 2009). Altogether our results demonstrate that PEDF and TSP1 may share some common pathways in regulation of RPE cell function, they may behave differently development and progression of AMD, perhaps as a result of their different impact on VEGF production in RPE cell.
In summary, consistent with previous studies (Adams and Lawler 2004;Barnstable and Tombran-Tink 2004), we demonstrated that PEDF and TSP1 play key roles in RPE cell function and subsequently in pathogenesis of AMD. We showed TSP1 and PEDF deficiency in RPE cells resulted in their enhanced proliferation and reduced migration. These changes were associated with alterations in adhesion and production of various ECM proteins. Significant changes in the expression and localization of junctional proteins were also observed. The alterations in MAPK and AKT signaling pathways were consistent with abnormalities observed in adhesion, oxidative stress, and sensitivity to oxidative challenge. Collectively, our results imply that PEDF and TSP1 play a critical role in regulation of the RPE cell proliferation and migration through alterations of their inflammatory and oxidative state.