Hepatocyte Growth Factor/Scatter Factor Promotes Retinal Angiogenesis through Increased Urokinase Expression

Abstract

purpose. The purpose of this study was to determine the role of hepatocyte growth factor (HGF) and c-Met in the initiation and development of retinal neovascularization and to determine whether inhibition of this system can suppress the extent of angiogenesis in an animal model.

methods. Retinal tissues from animals with oxygen-induced neovascularization were analyzed for HGF and c-Met expression and localization. The effect of HGF on the migratory and invasive behavior of isolated retinal endothelial cells was quantitated, and the role of the extracellular proteinase urokinase in facilitating this process was determined. Mice were treated with intraocular injections of anti-c-Met antibody, and the extent of neovascularization was quantitated.

results. HGF and c-Met were upregulated in the retinas of mice with hypoxia-induced retinal neovascularization. HGF was active, as evidenced by the increased presence of the phosphorylated form of c-Met in the tissues. c-Met was localized to various cell types in the retina, including vascular cells, and HGF was produced by cells in the ganglion and inner nuclear layers. HGF stimulated the secretion of urokinase and its receptor, uPAR, in isolated retinal endothelial cells. HGF increased the migratory and invasive capacity of these cells, which could be inhibited by the disruption of urokinase/uPAR interactions with the Å6 peptide. Inhibition of c-Met activation in vivo resulted in a 70% decrease in retinal angiogenesis and a 40% decrease in urokinase activity in the retina.

conclusions. These studies suggest that HGF may play an important role in the initial stages of retinal angiogenesis by stimulating a migratory phenotype in endothelial cells mediated by increased urokinase activity.

Angiogenesis is a process by which new blood vessels are formed by the sprouting of endothelial cells from preexisting vessels. This new vessel formation is an important component of many pathologic conditions, and it occurs in the retina in response to prolonged hypoxia, as seen in proliferative diabetic retinopathy, often leading to severe visual impairment. The process of new vessel formation is initiated by growth factor signaling and involves the upregulation of specific proteinases that facilitate matrix remodeling and cell migration. In previous studies, we have demonstrated the increased expression of the matrix metalloproteinases (MMPs), MMP-2 and MMP-9, and the serine proteinase urokinase and its receptor, uPAR, in the retinas of mice with hypoxia-induced neovascularization. ^{1 2 3} These proteins play an important role in the growth of new vessels because their inhibition can result in significant decreases in retinal angiogenesis. ^{1 3 4}  

Many different growth factors have been shown to play a role in the development of retinal neovascularization, including vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), insulinlike growth factor (IGF)-1, and angiopoietin. ^{5 6 7 8 9 10} These factors bind to specific cell surface receptors and initiate intracellular signaling pathways in endothelial cells, leading to altered expression of cell-matrix receptors and enhanced proliferation. 

Another growth factor, hepatocyte growth factor (HGF), has been shown to be involved with the angiogenesis associated with tumor growth and wound healing but has received little attention as a regulator of retinal angiogenesis. ^{11 12} Recent studies, however, have reported increased levels of HGF in the vitreous and serum of patients with proliferative diabetic retinopathy and in the subretinal fluid of patients with stage 5 retinopathy of prematurity, ¹³ raising the possibility of an important role for this growth factor in ocular angiogenesis. ^{14 15 16 17 18} The function of HGF in this context is dependent on the expression and activation of the met receptor, a transmembrane tyrosine kinase encoded by the c-met proto-oncogene. ¹⁹ Activation of c-Met followed by downstream signaling leads to rapid cell scattering and migration and to increased expression of MMPs and urokinase. ^{20 21}  

We have examined the role of HGF and c-Met in retinal angiogenesis using a well-characterized mouse model of retinal neovascularization. Here we report that HGF and c-Met are upregulated in retinas during the angiogenic response. HGF has the ability to regulate retinal endothelial cell migration in a urokinase-dependent manner. In vivo inhibition of c-Met signaling significantly decreases retinal angiogenesis. 

Methods

Mouse Model of Retinal Neovascularization

Specific pathogen-free C57Bl/6J mice were bred at the University of New Mexico Animal Research Facility. All experiments were consistent with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were performed in accordance with institutional animal care and use guidelines. Briefly, litters of 7-day-old C57Bl/6J mice were placed with their nursing mothers in an oxygen chamber maintained at 75% oxygen until postnatal day (P)12. ²² Mice were removed from the chamber at day 12 and were maintained in room air until day 15 or day 17. By day 17, retinal neovascularization was present in 100% of the experimental animals. Newborn mice exposed only to room air served as controls. 

Real-Time Polymerase Chain Reaction

Whole retinas were collected from three experimental and three control mice on days 12, 15, and 17. Total mRNA was extracted (Trizol reagent; Invitrogen, Carlsbad, CA), and cDNA was synthesized (TaqMan Reverse Transcriptase reagents; Applied Biosystems, Foster City, CA). Real time RT-PCR (TaqMan) assays for c-Met: Ma00434924, HGF: Mm0135192, and 18S: Hs99999901 were obtained (Applied Biosystems), and amplification and detection were performed (7500 Fast system; Applied Biosystems). Data were derived using the comparative Ct method of triplicate reactions. ²³  

Western Blotting

Western blot analysis was performed on proteins extracted from three experimental and three control whole mouse retinas on days 12, 15, and 17 with sample buffer (12.5 mM Tris, pH 6.8, 2% glycerol, 4% SDS) containing 0.1 M Na orthovanadate and 1× proteinase inhibitor cocktail (Complete Mini; Roche, Mannheim, Germany). Protein concentrations were determined with an assay (micro-BCA; Pierce, Rockford, IL), and equal amounts of total protein were loaded onto 10% polyacrylamide gels and confirmed with a β-tubulin loading control. Separated proteins were transferred onto polyvinylidene difluoride (PVDF) membranes and blocked for 45 minutes at room temperature with 5% milk in TBS/0.1% Tween 20. Membranes were incubated with either rabbit anti-human phospho c-Met or rabbit anti-mouse c-Met (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). As a loading control, the membranes were stripped and reprobed with an anti-human β-tubulin antibody (Novus Biologicals, Littleton, CO). Membranes were washed with TBS/0.1% Tween, incubated for 1 hour at room temperature with a horseradish peroxidase (HRP)-labeled secondary antibody, and developed with the enhanced chemiluminescence (ECL) reagent (Pierce). Bands were visualized with a chemiluminescence system and quantitated (GeneGnome; Syngene, Frederick, MD). 

Immunohistochemistry

Experimental and control mice were humanely killed on day 17, and the eyes were enucleated, fixed in 10% formalin overnight, and embedded in paraffin. Sections were deparaffinized, and antigen retrieval was performed by boiling in 10 mM sodium citrate, pH 6.0, for 5 minutes. The sections were blocked with 10% normal goat serum and incubated with rabbit anti-mouse c-Met (1:50) for 30 minutes (Santa Cruz Biotechnology, Inc.). Sections were washed with TBS/0.1% Tween 20 and incubated with an Alexa 488-labeled goat anti-rabbit secondary antibody (1:100) for 30 minutes (A11008; Molecular Probes, Eugene, OR). After a final wash, the sections were coverslipped with mounting medium containing DAPI (Vectashield; Vector Laboratories, Burlingame, CA) and were examined using a confocal microscope (LSM 510; Carl Zeiss, Inc., Thornwood, NY). 

In Situ Hybridization

For the preparation of riboprobes, HGF and c-Met cDNA fragments were obtained by PCR of total RNA from mouse liver. mRNA was reverse transcribed with the use of a preamplification system according to the manufacturer’s instructions (SuperScript; Invitrogen). AmpliTaq-Gold (Perkin Elmer, Boston, MA) was used for PCR amplification with the following primers: HGF forward, 5′-AGT ATT TAC GGC TGG GGC TAC AC-3′; HGF reverse, 5′-AGG ACG ATT TGG GAT GGC AC-3′; c-Met forward, 5′-GAA AGA CTT CAG CCA TCC CAA TG-3′; and c-Met reverse, 5′-GCA CAC CAA AGG ACC ACA CAT C-3′. PCR products were ligated into the PGEM easy vector (Promega, Madison, WI), and plasmids were purified. Resultant plasmids were sequenced and confirmed to contain the HGF and c-Met gene sequences. Sense and antisense riboprobes were generated and labeled with digoxigenin using RNA polymerase (SP6 and T7; Roche, Mannheim, Germany). Probes were stored in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 7.4) and were used within 2 months of synthesis. All steps before and during the hybridization procedure were carried out under RNase-free conditions. Eyes used for in situ hybridization were fixed in 4% paraformaldehyde in PBS overnight at 4°C and embedded in paraffin. Sections were treated with serine protease (Proteinase K; Sigma-Aldrich, St. Louis, MO) in 50 mM Tris, pH 7.4, for 30 minutes at 37°C, rinsed in water, and briefly fixed in 4% paraformaldehyde in PBS. Slides were rinsed three times, for 10 minutes each, in PBS and were treated with freshly made triethanolamine/acetic anhydride solution (100 mL water, 0.92 g NaCl, 1.48 mL triethanolamine, 0.5 mL acetic anhydride) for 7 minutes at room temperature. Slides were rinsed again in water and equilibrated in 2× SSC (3 M NaCl, 0.3 M Na₃-citrate, 2 H₂O autoclaved) for 10 minutes at room temperature. Sections were prehybridized for 30 minutes at room temperature in a 1:1 prehybridization buffer (P1415, 50 mL; Sigma-Aldrich, St. Louis, MO) and formamide. Sections were covered with probes diluted to 2 ng/mL in hybridization buffer (H7782; Sigma-Aldrich) and were incubated overnight in a humidified chamber at 60°C. After hybridization, sections were washed in 2× SSC followed by treatment with 10 μg/mL RNase A at 37°C for 30 minutes. Sections were washed with 2× SSC and a final wash of 0.5× SSC at 56°C. Endogenous peroxidase activity was quenched with 3% H₂O₂ in 0.5× SSC for 15 minutes. Sections were blocked with TNE blocking buffer (Perkin Elmer, Boston, MA) and incubated overnight at 4°C with an HRP-labeled polyclonal rabbit antidigoxigenin antibody (P5104; Dako, Copenhagen, Denmark). Tyramide amplification was performed according to the manufacturer’s instructions, and sections were developed with diaminobenzidine (DAB; Vector). 

Cell Migration and Invasion Assay

Experiments were performed using bovine retinal microvessel endothelial cells (BRECs; VEC Technologies, Inc., Rensselaer, NY) between passages 3 and 7 and were grown in MCDB-131 complete media with 10% fetal bovine serum and antibiotics. For migration and invasion assays, 2 × 10⁴ cells were plated onto solubilized basement membrane preparation (Matrigel; BD Biosciences, Bedford, MA)-coated cell culture inserts containing 8.0-μm pores in serum-free media containing insulin, transferrin, and sodium selenite (ITS; Sigma-Aldrich) and antibiotics. The bottom chamber was filled with serum-free media, serum-free media plus 10 ng/mL HGF (GF116; Chemicon, Temecula, CA), or complete media (MDCB-131; Life Technologies, Gaithersburg, MD). In some cases, 100 mM peptide (Å6; Angstrom Pharmaceuticals, San Diego, CA) was added to the cells in the upper chamber. Cultures were collected after 16 hours of incubation, and cells in the upper chamber were removed with a cotton swab. Remaining cells on the lower surfaces of the membranes were fixed with 100% methanol for 5 minutes, and the membranes were excised from the insert, mounted on glass microscope slides, and stained with DAPI (Vector Laboratories). Images were collected from the membranes at 12, 3, 6, and 9 o’clock at 20× magnification, and the number of cells nuclei per field was determined. 

Proteinase Assays

Bovine retinal endothelial cells were grown to confluence, transferred to serum-free MCDB-131, and treated for 24 hours with 10 ng/mL human recombinant HGF. Urokinase levels in the conditioned media and cell extracts of treated and untreated cell cultures were analyzed by zymography. Total protein levels were determined (Micro BCA Kit; Pierce), and equal amounts of protein were loaded onto gels. Samples of media or cell extract were electrophoresed in 10% polyacrylamide minigels, into which casein and plasminogen were cross-linked. After electrophoresis, the gels were incubated in a solution of 2.5% Triton X-100 followed by multiple rinses with distilled water. Gels were incubated in 100 mM Tris, pH 8.0, at 37°C for 3 hours with gentle rocking. The zones of proteolysis corresponding to the presence of urokinase were visualized by staining the gel with 0.125% Coomassie blue, and the area of the lysis zones was quantified with the use of image analysis software (Alpha Imager; Alpha Innotech Corp., San Leandro, CA) and was expressed as the integrated density value (IDV). 

In Vivo Anti-c-Met Injection and Quantitation of Retinal Angiogenesis

Retinal angiogenesis was induced in mice, as described. On day 13, 1 day after removal from high oxygen, mice were anesthetized and injected intravitreally with 400 ng c-Met neutralizing antibody (AF527; R&D Systems, Minneapolis, MN) or normal goat IgG (AB-108-C; R&D Systems). Eyes were collected on day 17, fixed with 4% paraformaldehyde, and embedded in paraffin. Sections were stained with anti-type IV collagen antibody (Rockland Immunochemicals, Gilbertsville, PA) to identify the retinal vasculature. Sections were examined using a fluorescence microscope (Axiovert 200M; Carl Zeiss), and the number of nuclei associated with type IV collagen-positive vessel profiles were quantitated. The average number of nuclei per section was determined, and statistical analysis was performed (Prism 4; GraphPad, San Diego CA). 

Results

Expression of HGF and c-Met in the Retina during Hypoxia-Mediated Angiogenesis

The well-established oxygen-induced retinopathy (OIR) mouse model was used for these studies. ²² Retinas of mice exposed to a high-oxygen environment underwent a stereotypical pattern of vaso-obliteration, leading to tissue hypoxia when the mice were removed to the room air environment on day 12. Almost 100% of these animals developed retinal neovascularization by day 17. The level of HGF mRNA in the retinas of experimental animals was found to be increased threefold on day 15 and maintained increased expression through day 17 (Fig. 1A) . 

Mice with retinal neovascularization showed a similar pattern of total c-Met expression compared with that of HGF (Fig. 1B) . On day 12, c-Met mRNA was more than twofold higher in experimental retinas than in room air controls. This further increased on day 15 (threefold higher than controls) and increased yet further on day 17 (fourfold higher than controls). 

We next examined the levels of activated c-Met protein in the retinas of experimental animals as a measure of HGF activity. Active c-Met was determined by blotting with an antibody that recognized the phosphotyrosine 1234 of activated c-Met. The degree of c-Met activation was expressed as a ratio of total c-Met within the tissue to account for the loss of vessels in the experimental animals after high-oxygen treatment. Quantitation of band densities indicated more than a twofold increase in the ratio of active to total c-Met in the experimental animals on day 12 compared with controls (Fig. 2) . Although active c-Met was still present in the retinas of experimental animals after day 12, the levels as a ratio of total c-Met were no different from those of controls. 

To determine which cells were producing the HGF protein in the retina, in situ hybridization was performed on day 15, when HGF mRNA expression was highest. HGF mRNA was localized predominantly to cells of the ganglion and inner nuclear layers (Fig. 3A) . Control retinas showed a low level of HGF expression in the ganglion cell layer, and staining in the inner nuclear layer was decreased compared with experimental retinas (Fig. 3B) . 

The site of HGF action in the retina was determined by the localization of c-Met by immunostaining. At days 12 and 15, particularly intense staining was seen associated with retinal blood vessels and cells within the ganglion cell layer (Figs. 4A 4B) . Some degree of staining also occurred within the outer plexiform layer and the inner nuclear layer of the retina. The capillary staining that occurred supports the hypothesis that c-Met and HGF play a role in retinal angiogenesis and may facilitate an increase in endothelial cell proteinase expression, migration, and proliferation. 

HGF Stimulates Endothelial Cell Urokinase Production In Vitro

We have previously shown that upregulation of the serine proteinase urokinase and its receptor, uPAR, is an important component of the angiogenic response in the retina. ³ Bovine retinal endothelial cells stimulated with HGF for 24 hours were found to significantly increase the secretion of urokinase into the conditioned media and demonstrated increased urokinase associated with the cell layer (43% and 67% increases, respectively) compared with cells without HGF stimulation (Figs. 5A 5B) . 

Urokinase is present in both a high (54 kDa) and a low (34 kDa) molecular-weight form. The low molecular-weight form, lacking the amino terminal fragment, cannot associate with the cell surface and is only present in the conditioned media. In contrast, the high molecular-weight form associates with the urokinase receptor at the cell surface and is present in both the media and the cell layer. The increase in urokinase activity seen in response to HGF stimulation was restricted to only the high molecular-weight form (not shown). 

HGF Stimulates Urokinase-Dependent Endothelial Cell Invasion In Vitro

Bovine retinal endothelial cells were plated onto solubilized basement membrane preparation (Matrigel; BD Biosciences)-coated membranes and stimulated with either HGF (10 ng/mL) or 10% serum. Cells exposed to HGF demonstrated a fourfold increase in invasive capacity compared with cells without stimulation (Fig. 6) . In addition, the level of cell invasion was as great as that seen in cells stimulated with serum as a positive control. Cells treated with HGF and the Å6 peptide, which blocks the association of uPA and uPAR, ²⁴ showed decreased invasion of the solubilized basement membrane preparation (Matrigel; BD Biosciences) matrix to levels not significantly different from those seen in the serum-free control. This result indicates that the invasive potential of the cells in response to HGF was dependent on the localization of uPA at the cell surface. 

Inhibiting c-Met In Vivo Results in Decreased Urokinase Activity and Retinal Angiogenesis

To determine whether the inhibition of c-Met signaling results in a decrease in retinal angiogenesis, experimental mice were injected intravitreally with either 400 ng neutralizing c-Met antibody or 400 ng normal goat IgG. Mice were injected on day 13, and angiogenesis was quantified on day 17. 

Interrupting the c-Met signaling pathway with a neutralizing antibody resulted in a significant decrease in retinal angiogenesis (Figs. 7A 7B) . Eyes injected with c-Met antibody had a significantly decreased number of neovascular nuclei (on the vitreal side of the inner limiting membrane) compared with eyes injected with IgG (average of 14 nuclei vs. 47 nuclei, respectively; Fig. 7C ). This result demonstrated a 70% reduction in retinal angiogenesis when the ability of HGF to bind and activate c-Met was inhibited after a single dose of antibody. Retinas that received the anti-c-Met treatments had smaller, more regularly spaced blood vessels that rarely crossed the inner limiting membrane. 

A zymography assay was used to measure urokinase activity in the retinas of IgG and anti-c-Met-treated eyes on day 15 (Figs. 8A 8B) . Eyes that received the anti-c-Met injections had a 42% reduction in urokinase activity compared with eyes that received normal IgG. These data suggested that the activity of c-Met and HGF in the promotion of retinal angiogenesis was caused in part by the upregulation of uPA expression and activity. 

Discussion

Excessive angiogenesis in response to tissue nonperfusion and hypoxia occurs in patients with advanced stages of diabetic retinopathy. We have previously shown that the expression of specific extracellular proteinases produced by activated capillary endothelial cells helps to facilitate the angiogenic process in an experimental animal model of retinal neovascularization. ^{1 2 3} Previous studies have reported that the activation of retinal capillaries is mediated by a variety of growth factors, including VEGF, bFGF, insulinlike growth factor 1 (IGF1), and angiopoietin-2 (Ang-2), leading to basement membrane degradation, endothelial cell migration, and proliferation. ^{6 25 26} In the present study, we have begun to investigate the role of HGF in the regulation of retinal neovascularization. HGF is upregulated early in response to hypoxia, and its receptor c-Met is active during the early stages of the angiogenic process. HGF was also found to influence the migratory and invasive behavior of cultured retinal endothelial cells, partially through the regulation of urokinase expression. These results suggest that the angiogenic process in the retina may be regulated by multiple growth factors. Understanding the interplay between HGF and these other factors might lead to the eventual development of new therapeutic approaches to the treatment of retinal vascular disease. 

Previous studies suggest that VEGF is the growth factor of primary importance in driving the angiogenic process forward in most tissues. ^{5 8 10 27} In the mouse model of hypoxia-induced retinal angiogenesis, VEGF mRNA had been shown to increase maximally 12 hours after removal from high oxygen. ¹⁰ The VEGF protein was found to localize to the inner nuclear layer, ganglion cell layer, and vascular endothelial cells. ²⁸ In addition, because of the spatial and temporal pattern of HGF and c-Met expression and activation described here, we can now postulate that VEGF and HGF might have complementary or cooperative roles in the induction of angiogenesis in the retina, which is supported by previous studies in other systems. ^{29 30 31} Some studies report that both growth factors are required for in vitro tubulogenesis, ³² whereas in others, HGF appears to be functioning upstream of VEGF. ^{33 34} A similar situation might have existed in the mouse retina undergoing hypoxia-induced neovascularization reported here, whereby HGF expression was found to be elevated during the first 12 hours after the removal of mice from high oxygen. In addition, HGF appears to act early in the angiogenic process because the ratio of active to total c-Met is highest on day 12. Further studies are needed to determine the possible interactions of VEGF and HGF in this system. 

The stimulus for the increased production of HGF in the hypoxic retina and the cells responsible for its production have not yet been determined. Immediately after removal from high oxygen, HGF protein was undetectable in the retina by Western blotting. However, 3 to 9 hours after removal from the high-oxygen environment, HGF expression was induced (Fig. 2) . Regions of the retina that had undergone vaso-obliteration subsequently became hypoxic and induced HGF protein production. Another possibility is that HGF is regulated by changes in the supply of nutrients to the tissue. Previous studies have demonstrated that the angiogenic factor, VEGF, is rapidly upregulated in response to amino acid deprivation. ^{35 36} Nutrient deprivation is likely to occur in the model used for these studies as a portion of the retina undergoes vaso-obliteration and nonperfusion in response to high-oxygen treatment. ²² Further studies will address this possibility and the postulated interactions between HGF and VEGF in this system. 

The c-Met receptor and HGF distribution were determined by immunostaining and in situ hybridization within the retina. HGF mRNA was localized predominantly in the ganglion cell and inner nuclear layers in the experimental animals. Because numerous cell types are present in these layers of the retina, further studies are needed to determine the specific source of this protein. c-Met was specifically localized to capillaries near the inner limiting membrane in addition to other cells within the ganglion cell layer, the outer plexiform layer, and the outer nuclear layer. This staining was decreased in the control retinas, indicating there was a response in the experimental retinas by specific cells to an insult, perhaps hypoxia, that led to the upregulation of c-Met expression. This is in agreement with previously reported studies showing hypoxia-induced c-Met expression by other cell types. ^{19 37} In addition to the promotion of angiogenesis by capillary endothelial cells, HGF and c-Met may serve a protective function in the retina, which may explain their widespread distribution in the retina. Indeed, recent studies have reported a neuroprotective function for HGF, ^{38 39} and this may explain the relatively low level of apoptosis seen in the retina of this model. ^{40 41 42} High levels of HGF have been reported in the vitreous of patients with increasing severity of proliferative diabetic retinopathy. ⁴³ In addition, HGF is thought to function as a potent chemoattractant for c-Met-expressing cells. It is plausible that HGF produced by the ganglion cells and deposited in the vitreous acts as a chemoattractant to the underlying endothelial cells. In the OIR model, the new vessels grow toward the ganglion cell layer and enter the posterior chamber of the eye. 

In this study, we have demonstrated that HGF upregulates the expression of the serine proteinase urokinase in cultured retinal endothelial cells. In addition to its role in facilitating basement membrane remodeling and cell migration, urokinase may also play an important role in this system as an activator of HGF. HGF is secreted in an inactive form that can be cleaved to its activated state by urokinase ⁴⁴ and by other factors. ⁴⁵ Future studies examining the effect of specific urokinase inhibitors on HGF activation and c-Met activity may elucidate this role for urokinase in this system. 

We did not investigate the downstream signaling pathways that mediate HGF-induced urokinase activity. Numerous studies have described such pathways in a variety of different cell types. ^{19 46 47} In the hypoxic retina, HGF and hypoxia may function together to effect the expression of components of the urokinase proteolytic system. Hypoxia itself has been shown to upregulate urokinase expression through hypoxia inducible factor (HIF)-1 regulation of gene expression. ^{48 49} A recent study also describes a novel mechanism for the hypoxia-independent HGF regulation of urokinase through the activity of HIF-1. ²⁰ This suggests that in the correct context, HGF and hypoxia might function together through HIF-1 transcription and stabilization to increase the expression of urokinase. 

The inhibition of HGF-induced retinal endothelial cell invasion by Å6 provides evidence that, in this system, HGF-induced cell invasion is dependent on the binding of urokinase to uPAR at the cell surface. Previous studies suggest that the localization of proteolytic activity at the cell surface is important for cell motility. ⁵⁰ Additionally, the binding of urokinase to uPAR may induce changes in uPAR function that mediate cell migration and adhesion. ⁵⁰  

Finally, we have demonstrated that inhibition of the HGF/c-Met signaling pathway results in both decreased urokinase expression and decreased retinal angiogenesis. The dramatic reduction of angiogenesis after c-Met inhibition is likely mediated by more than just inhibition of urokinase expression. HGF and c-Met mediate multiple cellular events that deserve further study. For example, previous reports indicate that HGF stimulates the proliferation of endothelial cells and may be antiapoptotic, ⁵¹ processes that are important to successful angiogenesis. 

In summary, the results of these studies suggest an important role for HGF and c-Met in the initiation and development of retinal angiogenesis by stimulating endothelial cell migratory activity through the upregulation of proteinase activity. Anti-VEGF drugs such as pegaptanib and ranibizumab, which are used to treat choroidal neovascularization, are to be used at frequent intervals to stop the regrowth of new vessels. A combination of therapies that target VEGF and other key growth factors may be more important than the monotherapy to achieve this goal of complete inhibition of angiogenesis. Results from these and future studies on the roles of these other angiogenic factors, such as HGF, during retinal angiogenesis may ultimately lead to the development of new therapeutic interventions for the treatment of proliferative retinopathies. 

 Supported by National Institutes of Health Grant RO1 EY12604 and a Physician-Scientist Training Award from the American Diabetes Association.

 Submitted for publication August 4, 2006; revised September 24, 2006; accepted February 20, 2007.

 Disclosure: E.S. Colombo, None; G. Menicucci, None; P.G. McGuire, None; A. Das, Angstrom Pharmaceuticals (F, C)

 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.

 

Expression of HGF and c-Met mRNA in control and experimental OIR mice using the real-time comparative Ct method of quantitation. HGF mRNA levels in retinas from experimental animals was significantly elevated on days 15 (P = 0.0093) and 17 (P = 0.0044) compared with controls when new blood vessels were forming (A). The pattern of c-Met mRNA expression in experimental animals was similar (B). The level of c-Met mRNA increased significantly on day 12 (P = 0.0043) and persisted through day 15 (P = 0.0126) and day 17 (P = 0.0121). Samples are from day 12, 15, and 17 control (C) and experimental (E) animals; n = 3. *Significantly greater than control.

Figure 2.

 

The level of activated c-Met was increased in experimental animals on day 12. Phosphorylated and total c-Met protein levels were quantitated in the retinas of experimental (E) and control (C) animals on days 12, 15, and 17 by Western blotting. Ratios of phosphorylated to total c-Met, as a measure of c-Met activation, were determined and are shown. A significant change in the level of the ratio of phosphorylated c-Met relative to total c-Met was seen on day 12 in experimental animals. By days 15 and 17 in experimental animals, c-Met activation was no different from that in controls. Values are the mean ± SEM. *Significantly greater than control (P = 0.030; n = 3).

Figure 3.

 

HGF mRNA is localized to the ganglion and inner nuclear layers of the experimental retina by in situ hybridization. Representative section of day 15 experimental retina stained with the antisense HGF riboprobe (A). The mRNA for HGF localizes to cells of the ganglion cell layer (arrow). Additional staining is seen in the inner nuclear layer (arrowhead). Section of day 15 control retina demonstrates less intense ganglion cell (arrow) and inner nuclear layer staining (B). No staining is seen in a section of day 15 experimental retina incubated with the sense riboprobe (C). Bars, 20 μm.

Figure 4.

 

Localization of the c-Met receptor during hypoxia-mediated angiogenesis. Representative sections from the retinas of experimental mice at day 12 (A) and day 15 (B) and control mice at day 12 (C) stained for total c-met. Staining for c-Met is detected in the experimental retina around blood vessels, close to the inner limiting membrane (A, B, arrows). In addition, cells in the ganglion cell layer (A, arrowheads), the outer plexiform layer (A, B, asterisks), and the inner nuclear layer are also positive for c-Met. (D) Section of day 12 experimental retina incubated without primary antibody. Bars, 10 μm.

Figure 5.

 

HGF induces urokinase expression by retinal endothelial cells in vitro. Urokinase activity in bovine retinal endothelial cells (A) and conditioned media (B) was quantitated by casein/plasminogen zymography after stimulation with HGF. The level of high-molecular-weight urokinase was significantly elevated in the HGF-treated cultures in both the cell layer (P = 0.001; n = 3) and the conditioned media (P = 0.001; n = 3). ITS, serum-free media; IDV, integrated density value (mean ± SEM).

Figure 6.

 

HGF stimulates retinal microvascular endothelial cell invasion in a urokinase-dependent manner. Cells were grown in serum-free media and stimulated with either serum-free media (ITS), media containing 10% fetal calf serum (Serum), serum-free media containing 10 ng/mL HGF (HGF), or serum-free media containing 10 ng/mL HGF and 100 mM Å6 peptide (HGF + A6). HGF caused a significant increase in cell invasion compared with serum-free media alone. *Significantly less than HGF (P = 0.0032; n = 3). This effect was blocked by the addition of the Å6 peptide, which inhibited the association of urokinase with its receptor. **Significantly less than HGF (P = 0.0041; n = 3). Values are mean ± SEM.

Figure 7.

 

Inhibition of retinal angiogenesis by reduction of c-met activation. Experimental mice were treated with a single intraocular injection of c-met neutralizing antibody on day 13 and were analyzed for new vessel formation. Representative images of retinas from normal IgG (A) and anti-c-met-treated (B) animals. Numerous large new vessels that penetrated the inner limiting membrane are seen in the IgG-treated experimental animals (A, arrows). When treated with a c-met neutralizing antibody, the vessels infrequently cross the inner limiting membrane (arrowheads), are more evenly spaced, and appear more organized. Quantitation of new vessels reveals a 70% reduction in the degree of angiogenesis in the eyes treated with the c-met antibody (C). *Significantly less than IgG-treated animals (P < 0.0001). Bar, 10 μm.

Figure 8.

 

Retinal urokinase activity is decreased after in vivo c-met inhibition. Experimental mice treated with intraocular injection of neutralizing c-Met antibody demonstrated a 45% reduction in urokinase activity compared with injection of normal goat IgG on day 15. *Significantly less than IgG; P = 0.0441. Representative casein/plasminogen zymogram of retinal extracts from normal IgG-treated (lanes 1–3) and anti-c-Met-treated (lanes 4–6) animals.

The authors thank Terry Jones and Virgil Thompson of Angstrom Pharmaceuticals for the kind gift of Å6 peptide used in these studies, and the American Diabetes Association and the National Institutes of Health for their support.