Angiopoietin-1 Upregulation by Vascular Endothelial Growth Factor in Human Retinal Pigment Epithelial Cells

Abstract

purpose. To determine whether vascular endothelial growth factor (VEGF) regulates angiopoietin (Ang)-1 and -2 expression in retinal pigment epithelial (RPE) cells.

methods. Expression of VEGF, Ang1, and Ang2 in surgically removed human choroidal neovascular membranes (CNVMs) was analyzed by double-label confocal immunofluorescence microscopy. Total RNA was extracted from cultured human RPE cells treated with VEGF for mRNA analysis. Northern blot analysis was performed to examine the time course and dose response of Ang1 and Ang2 mRNA expression. mRNA stability and nuclear run-on analyses were performed. Secreted Ang1 and Ang2 protein levels in conditioned media from RPE cells were examined by Western blot analysis.

results. Ang1 and Ang2 immunostaining colocalized with VEGF-positive stromal cells in human CNVMs. Ang1 and Ang2 mRNAs were expressed by cultured serum-starved RPE cells. VEGF upregulated Ang1 mRNA in a time- and dose-dependent manner without a significant change in Ang2 mRNA. Ang1 and Ang2 mRNAs in RPE cells were as stable as that of S18. VEGF stimulation further increased the half-life of Ang1 mRNA, but did not alter its transcription rate. VEGF increased the amount of Ang1, but not Ang2, protein secreted into the medium.

conclusions. The colocalization of Ang1 and Ang2 with VEGF in CNVM stromal cells and the upregulation of Ang1 expression by VEGF in cultured RPE cells suggest that VEGF may selectively modulate Ang expression during CNV.

Choroidal neovascularization (CNV) is a major blinding complication of age-related macular degeneration (ARMD). ^{1 2} Increasing evidence suggests that retinal pigment epithelial (RPE) cells play a key role in the pathogenesis of CNV. ^{3 4} In established CNV, multiple angiogenic growth factors, such as vascular endothelial growth factor (VEGF) and angiopoietin (Ang)-1 and -2, have been detected in a variety of cells, including RPE cells. ^{5 6 7 8 9 10} Furthermore, in normal eyes, polarized secretion of VEGF to the choriocapillaris by RPE and polarized localization of VEGF receptors on the inner choriocapillaris have been reported, suggesting that under resting conditions, RPE-derived VEGF plays a role in the maintenance of the choriocapillaris. ¹¹ In addition, several angiogenesis inhibitors, including thrombospondin-1 and pigment epithelium–derived factor, have also been detected in the RPE monolayer. ^{12 13 14} Furthermore, there is evidence that Fas ligand on RPE cells inhibits CNV by Fas-mediated apoptotic killing of choroidal endothelial cells. ¹⁵ Thus, it appears that the choriocapillaris is maintained by a delicate balance of angiogenic activators and inhibitors produced by RPE cells and that CNV may result from a loss of this balance within the subretinal space. 

VEGF, a primary ligand for the endothelial tyrosine kinase receptors Flt-1 and KDR, has been strongly implicated in neovascularization within the eye in clinical and experimental studies. ^{3 16 17 18 19 20 21 22 23 24 25} Ang1 is an agonistic ligand for another endothelial tyrosine kinase receptor, tyrosine kinase with immunoglobulin-like loops and epidermal growth factor homology domains (Tie2). ²⁶ In contrast, Ang2 also binds to Tie2, but inhibits Ang1-mediated Tie2 phosphorylation. ²⁷ The spectrum of roles that Ang1 and Ang2 play in angiogenesis is quite different from that of VEGF, but recent studies have demonstrated the importance of cooperative interaction of Ang1 and Ang2 with VEGF in physiologic and pathologic angiogenesis. ^{27 28 29 30 31} Based on disrupted angiogenesis in mice deficient in Ang1 and Tie2, it has been suggested that Ang1 is an important regulator of angiogenesis at its later stages, including vessel stabilization and maturation. ^{28 32} In contrast, Ang2 probably cooperates with VEGF in actively sprouting vessels by blocking the stabilizing or maturing function of Ang1, thus allowing vessels to revert to, and remain in, a plastic state where they may respond better to a sprouting signal from VEGF. ²⁷ Furthermore, in VEGF-induced neovascularization assayed using the corneal micropocket, Ang1 and Ang2 stimulation results in distinct angiogenic phenotypes. Ang1 with VEGF enlarges the vascular lumen and recruits smooth muscleα -actin–positive cells to the vascular wall, whereas Ang2 markedly enhances VEGF-induced neovascularization. ³⁰ Thus, it appears that sequential regulation of Ang1 and Ang2 in association with VEGF expression within the microenvironment of angiogenic blood vessels is important to ensure precise and stage-appropriate angiogenic signals to endothelial cells. 

RPE cells express both VEGF and VEGF receptors and respond to VEGF stimulation in vitro. ^{33 34 35} In experimental choroidal neovascular membranes (CNVMs), RPE cells express mRNAs for VEGF and its receptors. ³⁶ Also, RPE cells increase their VEGF expression in response to pathogenic stimuli. ^{37 38} We considered the possibility that VEGF may control production of Ang1 and Ang2 in RPE cells in an autocrine manner to achieve efficient temporal and spatial cooperation between the VEGF and Ang systems during CNV. In the present study, Ang1 and Ang2 expression colocalized with that of VEGF in human choroidal neovascular membranes. VEGF upregulated Ang1 mRNA and protein levels in human RPE cells by increasing its mRNA stability. 

Materials and Methods

Collection of Choroidal Neovascular Membranes

Surgical excision of ARMD-related, subfoveal CNVMs was performed in 12 eyes of 12 patients. All specimens were obtained by one of the authors (JIL, five specimens) during the course of the patients’ treatments, or the specimens were obtained from our frozen archives (seven specimens). The tenets of the Declaration of Helsinki were followed, informed consent was obtained after explanation of the nature and possible consequences of this study, and institutional review board (University of Southern California) approval was granted for this study. Each of the fresh, surgically excised CNVMs was placed in isotonic saline at 4°C, then snap frozen in optimum cutting temperature compound (Ames/Miles Inc., Elkhart, IN) within 1 hour. Each specimen was serially sectioned on a cryostat into 6-μm frozen sections on glass slides. The sections were fixed in reagent-grade acetone for 5 minutes at room temperature and stored at −80°C. Sections from 7 of the 12 specimens had been evaluated previously for expression of other growth factors. ⁷  

Double-Label Confocal Immunofluorescence Studies

Thawed sections were air dried, fixed with reagent-grade acetone for 5 minutes, and washed with Tris buffer (pH 7.4). Sections were blocked for 15 minutes with 1% bovine serum albumin (Sigma, St. Louis, MO) in Tris buffer after endogenous peroxide was blocked by 0.3% hydrogen peroxide. For double labeling, sections were incubated first with a polyclonal antibody for 30 minutes and rhodamine-labeled anti-rabbit or anti-goat secondary antibody. Sections were then washed profusely with Tris buffer, followed by monoclonal antibody immunohistochemistry using the FITC-labeled anti-mouse secondary antibody. Negative controls included omission of primary antibody or an irrelevant polyclonal or isotype-matched monoclonal primary antibody. In all cases negative control specimens showed only faint, insignificant staining. Rabbit polyclonal antibodies against human Ang1 and Ang2 (1:500 dilution) were provided by George D. Yancopoulos of Regeneron Pharmaceuticals, Inc. (Tarrytown, NY) and purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Specificity of the Regeneron Ang1 and Ang2 antibodies was confirmed by an absorption test involving an excess of the peptides used to raise the antibodies (Ang1: N-terminal peptide, NQRRSPENSGRRYNRIQHGQ; Ang2: N-terminal peptide, NFRKSMDSIGKKQYQVQHGS). Mouse monoclonal antibody against human VEGF (1:100 dilution) was obtained from Santa Cruz Biotechnology and mouse monoclonal antibody against pancytokeratin from Dako Corp. (Carpinteria, CA). Stained sections were observed with a high-resolution laser-scanning microscope (model LSM210; Carl Zeiss, Jena, Germany). 

Cell Culture

Human RPE cells were isolated from human fetal eyes (Anatomic Gift Foundation) or from adult (>60 years) donor eyes without evidence of macular disease, as previously described, ³⁹ and cultured in Dulbecco’s modified Eagle’s medium (Irvine Scientific, Santa Ana, CA) supplemented with 10% fetal bovine serum (Gemini Bioproducts, Calabasas, CA), 2 mM l-glutamine (Omega Scientific, Tarzana, CA), 100 U/ml penicillin and 100 μg/ml streptomycin. Cells were grown at 37°C in a humidified atmosphere of 5% CO₂. The purity of primary cultured RPE cells was determined by counting the number of the cells that were immunoreactive for cytokeratin (Dako). Only cultures that had more than 98% cytokeratin-positive cells were used for analysis (data not shown). 

Northern Blot

Total RNA was extracted from RPE cells using the Trizol reagent (Gibco BRL, Gaithersburg, MD) according to the manufacturer’s instructions. Equal amounts of total RNA (10–20 μg per lane) were fractionated on 1% (wt/vol) agarose/6.3%formaldehyde gels and blotted on a nylon membrane (Nylon Duralon-UV; Stratagene, La Jolla, CA), by using the traditional capillary system, in 10× SSPE (1.5 M NaCl, 100 mM sodium phosphate, and 10 mM Na₂ EDTA). Filters were cross-linked (UV Stratalinker 1800; Stratagene) and hybridized at 68°C for 2 hours (ExpressHyb Hybridization Solution; Clontech Laboratories, Palo Alto, CA) containing 0.1 mg/ml denatured salmon sperm DNA with 2 to 3 × 10⁶ counts per minute (cpm)/ml ³²P-labeled Ang1 (a 0.57-kb SpeI-EcoRI fragment of human Ang1 cDNA), Ang2 (a 0.64-kb EcoRI-HindIII fragment of human Ang2 cDNA), and VEGF (a 0.6-kb BamHI fragment of human VEGF cDNA). Ang1, Ang2, and VEGF cDNAs were kindly provided by Regeneron Pharmaceuticals, Inc. The cDNAs were labeled (specific activity of 1 × 10⁹ cpm/μg) using a random primer labeling kit (Rediprime; Amersham Pharmacia Biotech, Piscataway, NJ) according to the manufacturer’s instructions. After hybridization, filters were washed three times in 2× SSPE/0.1% SDS for 15 minutes at room temperature and then in 0.1× SSPE/0.1% SDS for 30 minutes at 60°C. To correct for differences in RNA loading, the filters were stripped and rehybridized with a human S18 rRNA probe (Ambion, Austin, TX). The filters were scanned, and radioactivity was quantified by computer imaging (PhosphoImager with ImageQuant software; Molecular Dynamics, Sunnyvale, CA). 

mRNA Stability Analysis

RPE cells were exposed to vehicle or VEGF (10 ng/ml) for 8 hours and then incubated with actinomycin D (10 μg/ml) to stop RNA synthesis. Total RNA was isolated at the indicated time points and used for Northern hybridization, as described. 

Nuclear Run-On Analysis

RPE cells were treated with vehicle or VEGF (10 ng/ml) for 8 hours. The cells were washed twice with ice-cold phosphate-buffered saline, scraped off the dish in ice-cold SSC (150 mM sodium chloride and 15 mM sodium citrate) and collected in a 15-ml tube by centrifugation at 500g for 5 minutes at 4°C. Subsequent steps were performed at 4°C. The cells were resuspended in 4 ml lysis buffer (10 mM Tris-HCl [pH 7.4], 10 mM NaCl, 3 mM MgCl₂, and 0.5% Nonidet P-40) and then were disrupted with homogenizers (10 strokes; Dounce; Kontes Glass, Vineland, NJ). Nuclei were pelleted by centrifugation at 500g for 5 minutes, resuspended in 100 μl of glycerol storage buffer (10 mM Tris-HCl, [pH 8.3], 40% [vol/vol] glycerol, 5 mM MgCl₂, and 0.1 mM EDTA), and frozen in liquid N₂. The nuclear suspension was mixed with 0.1 ml 2× reaction buffer (100 mM HEPES [pH 8.0]; 10 mM MgCl₂; 300 mM KCl; 200 U/ml RNasin [Roche Molecular Biochemicals, Indianapolis, IN]; 1 mM each ATP, GTP, and CTP; and 150 μCi (1 μCi = 37 kBq)[ ³²P]UTP [3000 Ci/mmol; Amersham Pharmacia Biotech]) and incubated for 30 minutes at 30°C. Transcription was stopped by adding 20 μg DNase I, followed by 80 μg proteinase K. The ³²P-labeled RNA was purified by extraction with phenol-chloroform and two sequential precipitations with ammonium acetate. Equal amounts of ³²P-labeled RNA were hybridized in 50% formamide, 5× SSC, 5× Denhardt’s solution, 1% SDS (1× SSC, pH 7.0) at 42°C for 72 hours. Filters contained 0.1 to 5 μg each of linearized plasmids immobilized on membranes (Zeta-Probe GT; Bio-Rad Laboratories, Hercules, CA) after blotting in 12× SSPE with a microfiltration apparatus (Bio-Dot SF; Bio-Rad). Filters were washed three times with 2× SSC and 0.1% SDS at 42^oC for 5 minutes, followed by two washes with 0.2× SSC and 0.1% SDS at 65°C for 15 minutes, and then analyzed as for Northern blot analysis (PhosphoImager; Molecular Dynamics). 

Western Blot

Confluent RPE cells were incubated with serum-free medium for 24 hours, whereupon the cells were exposed to vehicle or VEGF (10 ng/ml). The conditioned medium was collected, and cell debris was removed by centrifugation at 14,000g. Total protein concentrations of the cell-free supernatant were measured by a Bradford-based assay kit (Bio-Rad), and equal amounts of protein were fractionated by 10% SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Immobilon; Millipore Corp., Bedford, MA) and then probed with polyclonal rabbit anti-human Ang1 (Regeneron, 0.45 μg/ml) or anti-human Ang2 antibodies (Regeneron, 0.22 μg/ml). Bound antibodies were developed using a chemiluminescence kit, as described by the manufacturer (ECL; Amersham Pharmacia Biotech). The membranes were scanned, and blue fluorescence intensity was quantified (Storm PhosphoImager with ImageQuant software; Molecular Dynamics). 

Statistical Analysis

Experiments were performed in triplicate at least three times. Values are expressed as mean ± SEM. Factorial ANOVA was performed followed by Fisher’s least-significant difference test. Statistical significance was defined as α < 0.05. 

Results

Immunofluorescence Analysis of Human CNVMs

To determine whether Ang1 and Ang2 expression colocalizes with that of VEGF in human CNVMs, we performed confocal double-label immunofluorescent analysis of human CNVMs. These membranes have strong autofluorescence derived from lipofuscin, causing some false-positive staining when immunofluorescent methods are used. Therefore, in this study, we observed unstained adjacent sections to confirm that the positive staining was not derived from the autofluorescence. As previously reported, there were numerous cells positive for VEGF, Ang1, and Ang2, to a varying extent, in all surgically excised CNVMs. ¹⁰ In the stromal regions, a majority of Ang1- or Ang2-positive cells were immunoreactive for pancytokeratin, indicating that migrating RPE cells within the CNVMs produce Ang1 and Ang2 (Fig. 1A) . Only a minor population of cytokeratin-negative cells were immunoreactive for Ang1 and Ang2. In each CMVM, VEGF immunostaining highly colocalized with that of Ang1 and Ang2 (Fig. 1B) . There was no apparent difference in the degree of colocalization between VEGF-Ang1 and VEGF-Ang2. Similar results were obtained, whether using the Ang1 and Ang2 antibodies obtained from Regeneron or from Santa Cruz Biotechnology. 

Time- and Dose-Dependent Regulation of mRNA Expression for Ang1 and Ang2 by VEGF in Fetal RPE Cells

Northern blot analysis revealed that Ang1 and Ang2 mRNAs were expressed in serum-starved RPE cells. Stimulation of these cells with 10 ng/ml VEGF increased Ang1 mRNA levels gradually to a maximum of 2.8-fold at 12 hours (P < 0.01; Fig. 2 ). In contrast, Ang2 mRNA levels were unchanged over 24 hours. Ang1 mRNA was expressed at low levels in retinal and choroidal endothelial cells, but the same VEGF stimulation did not change its level (data not shown). Thus, the upregulation of Ang1 in response to VEGF was specific to RPE cells in the cell types examined. To confirm that the effects of VEGF on the induction of Ang1 mRNA were dose dependent, cells were stimulated with various concentrations of VEGF for 12 hours. The upregulation of Ang1 mRNA was dose dependent with a maximal 2.5-fold increase (P < 0.01; Fig. 3 ). 

Increase in Stability of Ang1 mRNA by VEGF

To elucidate the mechanism by which VEGF upregulates Ang1 mRNA, we measured mRNA stability. Unexpectedly, Ang1 and Ang2 mRNA levels normalized by those of S18 remained unchanged for 8 hours after actinomycin-D treatment, suggesting that Ang1 and Ang2 mRNAs in RPE cells are as stable as that of S18 (Figs. 4A 4B 4C) . The approximate half-life of mRNA decay usually can be calculated based on the decay curves for the mRNA levels normalized by a stable housekeeping gene such as S18. Because ribosomal RNA primarily represents total RNA and is more stable than other housekeeping genes, such as the β-actin and tubulin genes, the use of S18 as an internal control seems to be appropriate to examine stability of stable genes. ^{40 41}  

As a reference, under the same experimental conditions, VEGF mRNA levels decreased rapidly, and the half-life of VEGF mRNA (4.8 hours) was obtained by using the decay curve of VEGF mRNA levels normalized to S18 mRNA levels on a logarithmic scale (Figs. 4A 4D) . However, Ang1 and Ang2 mRNAs were so stable that reliable measurements of their half-lives could not be obtained by using the decay curves of the mRNA levels normalized to S18. Nonetheless, it is clear that the stability of Ang1 and Ang2 mRNAs in RPE cells was extremely high compared with that in bovine choroidal endothelial cells (Ang1 half-life, 3.7 hours; Ang2 half-life, 1.7 hours; data not shown). Addition of VEGF (10 ng/ml) further increased the stability only of Ang1 mRNA (Fig. 4) . In contrast, the same VEGF stimulation had little if any effects on Ang2 mRNA stability and decreased VEGF mRNA stability. 

Nuclear run-on analysis was also performed. Although in parallel assays prostaglandin E₂ increased the transcription rate of the VEGF gene in RPE cells and tumor necrosis factor-α increased transcription rate of Ang1 and Ang2 genes in choroidal endothelial cells (data not shown), VEGF treatment (10 ng/ml) of RPE cells did not change the transcription rate of the tested angiogenic growth factors and their receptors, including Ang1 (Fig. 5) , except for a decrease in the transcription rate of the VEGF gene. Taken together, these experiments demonstrate that Ang1 mRNA induction by VEGF in RPE cells was due to an increase in Ang1 mRNA stability. 

Effects of VEGF on Ang1 and Ang2 Protein Levels in Fetal RPE Cells

In preliminary experiments, Ang1 and Ang2 proteins were detected primarily in conditioned media collected from serum-starved RPE cells, but were found at much lower levels in cell lysates. Therefore, the effects of VEGF on Ang1 and Ang2 secretion into conditioned media were examined by Western blot analysis. Stimulation of RPE cells with 10 ng/ml of VEGF for 24 to 48 hours did not change the cell number but increased Ang1 protein levels by 1.8-fold (P < 0.01) at 24 hours (Fig. 6 and by 1.5-fold (P < 0.05) at 48 hours. The same stimulation did not change Ang2 protein levels over the same time course. These results indicate that, after VEGF stimulation, an increase in Ang1 mRNA was associated with an increase in Ang1 protein in RPE cells. 

Increase of Ang1 mRNA in Adult RPE Cells by VEGF

Experiments were performed using fetal RPE because of the ease of obtaining the large numbers of cells that are required for such experiments. To demonstrate that these data are applicable to adult cells, we also examined human adult RPE cells obtained from eyes of donors older than 60 years, for the induction of Ang1 mRNA by VEGF. Stimulation of the cultured adult cells with 10 ng/ml VEGF increased Ang1 mRNA levels by 3.1-fold at 8 hours (P < 0.05; Fig. 7 ), whereas Ang2 mRNA levels were steady. These data suggest that VEGF upregulates Ang1 not only in fetal RPE cells but also in adult RPE cells. 

Discussion

The Ang-Tie2 system is unique, in that a natural antagonist for Ang1 exists and that both Ang1 and Ang2 contribute to angiogenesis at distinct stages. ^{27 29} Ang1 is constitutively expressed primarily by nonendothelial (mesenchymal, smooth muscle, and tumor) cells associated with blood vessels during physiologic and pathologic angiogenesis. ^{28 31 42} Tie2 is expressed in quiescent blood vessels throughout the body and in angiogenic blood vessels. ⁴³ Increasing evidence suggests that Ang1-mediated Tie2 activation is required for angiogenesis, especially at its later stages, to produce structurally and functionally mature blood vessels. ^{28 29 30 44 45} In contrast, Ang2 is induced in endothelial cells within actively sprouting vessels or zones of vessel regression, suggesting that Ang2-mediated autocrine Tie2 inactivation or other unknown effects of Ang2 are required for active sprouting or vessel regression. ^{27 29 31} Thus, the relative levels of Ang1 and Ang2 are regulated, depending on the stage of angiogenesis. However, it has been shown that both Ang1 and Ang2 are expressed to a similar extent in the same human CNVMs. ¹⁰ CNVMs consist of a mixture of variously vascularized and fibrotic regions. It is possible that relative levels of Ang1 and Ang2 are delicately regulated depending on the regional microenvironment of CNV. Our results indicate that RPE cells are capable of producing Ang1 and Ang2 and that VEGF can upregulate RPE cell expression of Ang1 without a significant change in Ang2 expression. This regulation may help induce an Ang1-dominant local microenvironment in CNVMs. 

Fetal RPE cells were used in this study, because they divide rapidly and thus are suitable to provide the large number of cells required for the extensive analyses reported. Fetal human RPE cells are frequently used as a source of cultured RPE cells for in vitro experiments, and they appropriately express a wide range of cytokine and growth factor receptors including receptors for VEGF. ^{46 47} Because morphologic and functional aging changes in the RPE are well described, it was necessary to determine whether the effects of VEGF on Ang expression are similar in fetal and adult cells. ⁴⁸ In the present study, VEGF upregulated Ang1 but not Ang2 in both fetal and adult RPE, suggesting that aging changes do not interfere with the ability of RPE to respond appropriately to VEGF. Whether RPE cells derived from patients with ARMD respond in a similar way was not determined in this study, because of the extreme difficulty in culturing large numbers of cells from these eyes. 

Our previous observation of experimental CNV in monkeys revealed the importance of RPE cells in the development and maturation of CNV. ⁴⁹ Newly formed vessels are leaky with abundant fenestrated vascular walls and open (permeable) interendothelial junctions. ⁵⁰ This early active stage is followed by maturating stages during which newly formed vessels become less leaky with decreased fenestrations and acquire a dense basement membrane and tightly closed interendothelial junctions. During the maturation processes, RPE cells migrate and proliferate around the new vessels and finally envelop them. ⁴⁹ Thus, RPE cells are believed to play an essential role in the maturation and/or suppression of newly formed subretinal vessels. However, little is known about the molecular mechanisms involved in interactions of RPE and endothelial cells during the maturation processes. It has been shown that a combination of Ang1 and VEGF can increase luminal diameter of angiogenic blood vessels and recruit smooth muscle α-actin–positive cells to vascular walls, suggesting cooperative roles between Ang1 and VEGF in vessel maturation. ³⁰ Furthermore, it has recently been shown that Ang1 plays a key role in establishing leakage-resistant blood vessels and that Ang1 can protect blood vessels against VEGF-induced plasma leakage. ^{45 51} RPE-derived Ang1 and VEGF in CNVMs may collaborate to induce structural and functional maturation in such enveloped new vessels. 

In a previous report using bovine retinal microvascular endothelial cells, Ang2 expression was upregulated by VEGF and hypoxia, whereas Ang1 mRNA expression was sustained. ⁵² It is interesting that regulation of Ang1 and Ang2 expression by VEGF differed between endothelial cells and vessel-associated nonendothelial cells in the eye. This distinction may be due to the different contributions of endothelial and RPE cells in the neovascularization processes. Of importance, in our results, Ang1 and Ang2 mRNAs in RPE cells were much more stable than Ang1 and Ang2 mRNAs in endothelial cells, including retinal (Ang2 half-life, 3.8 hours) ⁵² and choroidal (Ang1 half-life, 3.7 hours; Ang2 half-life, 1.7 hours) endothelial cells. In our study, VEGF further increased the stability of Ang1 mRNA. In contrast, upregulation of Ang2 in retinal endothelial cells was rapid and short lived and was due to an increase in transcription rate. Evidence has been provided that active sprouting requires prompt and short Ang2-mediated autocrine blockade of Tie2 activation to allow the endothelial cells to better respond to a sprouting signal by VEGF. For this purpose, the rapid and short-lived induction of endothelial Ang2 and its autocrine action to endothelial cells is advantageous. In contrast, at the later stage of neovascularization, endothelial cells may require sustained Tie2 activation for vessel stabilization, maturation, and survival. For this purpose, stable expression of Ang1 by RPE cells is advantageous. At this later stage of CNV, VEGF may shift the balance of Ang1 and Ang2 expression in RPE cells to an Ang1-dominant state by stabilizing Ang1 mRNA and may achieve more effective cooperation with Ang1 to establish structurally and functionally mature blood vessels. 

Human CNVMs show strong expression of both Fas and Fas ligand and contain numerous apoptotic cells. ⁵³ A functional study revealed that Fas ligand on RPE cells controls experimental CNV by Fas-mediated killing of choroidal endothelial cells. ¹⁵ Thus, apoptotic endothelial cell death may be one important mechanism by which the retina controls CNV. There is increasing evidence that Ang1 and VEGF can protect endothelial cells against apoptotic death. ^{31 54 55} The sustained expression of Ang1 by RPE cells may promote stabilization of CNV by protecting choroidal endothelial cells against apoptosis in cooperation with VEGF. Upregulation of Ang1 by VEGF in RPE cells could enhance these antiapoptotic activities to further stabilize CNV. 

The sustained presence of CNVMs beneath the retina causes damage to the retina. On the one hand, actively leaking CNV results in macular edema and subretinal hemorrhage, which are serious causes of vision loss. On the other hand, less leaky, fibrotic CNVMs block various supporting activities of the RPE-choriocapillaris for photoreceptors. Accordingly, it is important to identify the factors that maintain newly formed blood vessels as well as those that stimulate CNV to initiate rational therapeutic interventions. Although RPE-derived Ang1 and VEGF may be the important factors that maintain CNV, it has been shown that Ang1 also plays a key role in establishing leakage-resistant blood vessels and preventing VEGF-induced plasma leakage. ^{45 51} Thus, RPE-derived Ang1 could be both beneficial and detrimental to the retina. Whether blocking Ang1 in CNVMs would promote or inhibit the development of retinal damage remains an important question to be resolved. 

 Supported by National Institutes of Health Grants EY01545 and EY03040 and grants from the JG Foundation and Research to Prevent Blindness. MH was supported by Kyoto University Foundation, Japan National Society for the Prevention of Blindness, and Nippon Eye Bank Association.

 Submitted for publication August 15, 2000; revised January 12, 2001; accepted February 7, 2001.

 Commercial relationships policy: F.

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

 

Figure 1.

 

Colocalization of Ang1 and Ang2 with VEGF in human CNVMs. Double-labeled confocal immunofluorescence experiments were performed using the anti-VEGF monoclonal antibody (red), the anti-pancytokeratin antibody (red), and the anti-Ang1 or anti-Ang2 polyclonal antibodies (green; Regeneron, Tarrytown, NY). Results of a representative CNV membrane and the respective combined images are shown. Colocalization produces a yellow signal. Magnification, ×50.

Figure 2.

 

Time-course of mRNA induction of Ang1 and Ang2 by VEGF in human fetal RPE cells. VEGF upregulated Ang1 mRNA levels. Total RNA was extracted from RPE cells at the indicated times after stimulation with VEGF (10 ng/ml) and subjected to Northern blot analysis (15 μg total RNA/lane). Results were quantified by computer imaging. Differences in loading were normalized by using the signal intensity of S18. The corrected density was plotted in comparison with the 0-hour value. Results are mean ± SEM from three separate experiments for each time point. A representative autoradiogram is shown (*P < 0.05, **P < 0.01).

Figure 3.

 

Dose–response of mRNA induction of Ang1 by VEGF in human fetal RPE cells. Induction of Ang1 mRNA by VEGF was dose dependent. RPE cells were stimulated with the indicated concentration of VEGF. Total RNA was extracted at 12 hours after the stimulation and analyzed as described in Figure 2 , with results expressed as in Figure 2 (*P < 0.05, **P < 0.01).

Figure 4.

 

Effect of VEGF on Ang1, Ang2, and VEGF mRNA half-lives in human fetal RPE cells. Ang1 (B) and Ang2 (C) mRNAs were very stable compared with that of VEGF (D). VEGF increased the stability of Ang1 mRNA, but not of Ang2 and VEGF mRNAs at the time points examined. RPE cells were exposed to vehicle or VEGF (10 ng/ml) for 8 hours before the addition of actinomycin D (10 μg/ml). Total RNA was extracted from the cells at the indicated times after actinomycin D treatment, and Northern blot analysis was performed as described in Figure 2 . The mean values of the corrected signal intensity were plotted as a percentage of the 0-hour value in logarithmic scale. Results are mean ± SEM from three separate experiments for each time point (*P < 0.05).

Figure 5.

 

Effect of VEGF on the Ang1 transcription rate in human fetal RPE cells. VEGF did not alter the transcription rate for Ang1. RPE cells were exposed to vehicle or VEGF (10 ng/ml) for 8 hours (Ang1). Nuclei were isolated, and in vitro transcription was allowed to resume in the presence of [³²P] UTP. Equal amounts of ³²P-labeled RNA probes were hybridized to nylon membrane on which each cDNA (5 μg) was immobilized. Reduced glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 5 μg) and β-actin (0.1 μg) were blotted as internal controls. The band intensities were measured by computer imaging, and the values of each factor were normalized using the signal intensity of GAPDH. Results are expressed as in Figure 2 .

Figure 6.

 

Western blot analysis of Ang1 and Ang2 proteins in conditioned media from VEGF-treated human fetal RPE cells. VEGF increased Ang1 (A), but not Ang2 (B), protein secretion. RPE cells were exposed to vehicle or VEGF (10 ng/ml), and the conditioned media were collected 24 and 48 hours after the stimulation. Approximately 5 μg protein per lane was resolved by 10% polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane, and immunoblotted using polyclonal antibodies to Ang1 or Ang2. Bound antibodies were detected using a horseradish peroxidase–based chemoluminescence method and quantified by computer imaging. The intensity was plotted in comparison to the vehicle value (C). Results are mean ± SEM from three separate experiments (*P < 0.01; **P < 0.05); representative bands are shown in (A) and (B).

Figure 7.

 

Induction of Ang1 mRNA by VEGF in human adult RPE cells. VEGF upregulated Ang1 mRNA levels in RPE cells from adult donor eyes. Total RNA was extracted from RPE cells at the indicated times after stimulation with VEGF (10 ng/ml) and analyzed, with results expressed as in Figure 2 (*P < 0.05).

The authors thank Susan Clarke for editorial assistance.