Nonvascular VEGF receptor 3 expression by corneal epithelium maintains avascularity and vision

Abstract

Transparency of the cornea, the window of the eye, is a prerequisite for vision. Angiogenesis into the normally avascular cornea is incompatible with good vision and, therefore, the cornea is one of the few tissues in the human body where avascularity is actively maintained. Here, we provide evidence for a critical mechanism contributing to corneal avascularity. VEGF receptor 3, normally present on lymphatic and proliferating blood vascular endothelium, is strongly constitutively expressed by corneal epithelium and is mechanistically responsible for suppressing inflammatory corneal angiogenesis.

The posterior structures of the eye, such as the choroid, are among the most heavily vascularized tissues. Yet ocular vascularity abruptly comes to an end at the cornea, the normally avascular and transparent “window” of the eye, which also serves as its main optical surface (Fig. 6, which is published as supporting information on the PNAS web site). Indeed, corneal clarity is critical for vision and is actively maintained in all animal species that require high visual acuity (1, 2). Conversely, blood vessel growth into the cornea is incompatible with good vision and is associated with the leading causes of corneal blindness both worldwide (trachoma) and in industrialized nations (herpetic keratitis).

Because of its lack of vascularity, the cornea has served as the principal in vivo model system for studying vasculogenic processes, specifically corneal hemangiogenesis (CHA) and more recently lymphangiogenesis (3–8). However, the mechanisms underlying maintenance of corneal avascularity remain poorly understood (2). Several angiogenic growth factors, especially of the VEGF family, have been implicated in mediating corneal angiogenesis (9, 10). As a potential counterbalance, several antiangiogenic factors including thrombospondins 1 and 2, endostatin, pigment epithelium-derived factor, and tissue inhibitor of metalloproteinases have been identified in the cornea (2). In addition, soluble VEGF receptor 1 (VEGFR1) (interacting with VEGF-A) is thought to be involved in corneal avascularity (11–13). However, to date, no single factor has been identified as being critically responsible for maintaining corneal avascularity.

Recently, we observed that intact corneal epithelium can suppress CHA (14) and VEGFR3 is constitutively expressed on normal human corneal epithelial cells (15). Because VEGFR3 binds VEGF-C and VEGF-D, and both of these factors promote lymphangiogenesis and hemangiogenesis and are additionally chemotactic for inflammatory cells that secrete VEGF-A (8, 16), we hypothesized that this ectopic VEGFR3 expression on the corneal epithelium promotes avascularity of the normal cornea by serving as a “sink” for VEGFR3 ligands.

Results

To test the hypothesis that VEGFR3 expression on the corneal epithelium promotes avascularity of the normal cornea by serving as a sink for VEGFR3 ligands, we first analyzed the presence of VEGFR3 protein in murine corneal epithelial cells (MCE) by using immunohistochemistry (Fig. 1A). Staining was intense in the epithelial layer and only weakly positive in the corneal endothelium and stroma. FACS analysis revealed strong expression of VEGFR3 on corneal epithelial cells specifically marked with the keratin-12 (K-12) marker (Fig. 1B). Strong gene transcription of VEGFR3 in normal corneal epithelium was confirmed by RT-PCR (Fig. 1C). Quantitative PCR demonstrated higher expression levels of VEGFR3 in the corneal epithelium compared with stroma/endothelium (1.5 times higher; Fig. 1D). Finally, we determined whether VEGF-C can indeed bind to ectopically expressed VEGFR3 on corneal epithelial cells and whether it causes intracellular phosphorylation events. First, we confirmed that MCE also express VEGFR3 (data not shown). Then, serum-starved MCE were treated with VEGF-C, and VEGFR3 was immunoprecipitated with anti-VEGFR3 antisera followed by Western blotting with antisera against phosphotyrosine. Results revealed that VEGF-C can bind to epithelial VEGFR3 and significantly increase the phosphorylation level of VEGFR3 (Fig. 1E). Surprisingly, there was also some basal epithelial VEGFR3 phosphorylation even in the absence of exogenous VEGF-C. These data demonstrate that epithelial VEGFR3 is able to bind VEGF-C.

Fig. 1.

Ectopic VEGFR3 expression in corneal epithelium. (A–D) Strong ectopic expression of both VEGFR3 protein (A and B) and mRNA (C and D) in normal corneal epithelium. (A) Immunofluorescence at ×200. (Left) Anti-VEGFR3 antibody. (Right) Control. Arrows indicate corneal epithelium. (B) Two-color FACS staining against epithelial marker K-12 (phycoerythrin) and VEGFR3 (FITC). Gray indicates isotype; black indicates VEGFR3 staining. Analysis was gated on K-12-phycoerythrin. Arrow indicates VEGFR3 and K12 costained corneal epithelium. (C) RT-PCR for VEGFR3 in normal corneal epithelium (expected sizes: VEGFR3, 290 bp; GAPDH, 245 bp). (D) Quantitative real-time PCR demonstrates higher levels of VEGFR3 in corneal epithelium (Left) compared with stroma and corneal endothelium (Right). (E) VEGF-C binds to corneal epithelial VEGFR3 and leads to VEGFR3 activation. Serum-starved cultured MCE were treated with VEGF-C and VEGFR3 immunoprecipitated from cell lysates by using a polyclonal anti-VEGFR3 antibody (M20) and protein A-Sepharose. Immunoprecipitated proteins were resolved on an 8% SDS/polyacrylamide gel and transferred to nitrocellulose membrane. Phosphotyrosine residues were detected by immunoblotting using PY20 and AG10 antibodies. After stripping the membrane, total VEGFR3 was detected by using M20 antibody. The phosphorylation level of VEGFR3 was quantified by densitometry and corrected to the amount of VEGFR3. A significant increase in epithelial VEGFR3 phosphorylation after exposition to VEGF-C was shown.

Next, to investigate a potential antiangiogenic role of corneal epithelium in vivo, we used an established inflammatory stimulus that does not cause CHA in the presence of an intact corneal epithelium (Fig. 2 F); mild cautery of the cornea is known to cause inflammatory cell influx into the cornea but fails to invoke a neovascular response (ref. 17 and Fig. 2B). Similarly, removal of the corneal epithelium alone (de-epithelialization) did not cause CHA (Fig. 2C). In contrast, cauterization of de-epithelialized corneas caused an angiogenic response, which by morphometry was significantly greater compared with mice with de-epithelialization alone (P < 0.0001) or cautery alone (P < 0.001), suggesting that intact corneal epithelium inhibits angiogenesis (Fig. 2 E and F and Table 1).

Fig. 2.

Antiangiogenic effect of corneal epithelium I. (A) Normal corneal–conjunctival border. (B and D) A normally nonangiogenic inflammatory stimulus (cautery; B) causes corneal neovascularization in the absence of epithelium (D). (C) De-epithelialized cornea. Representative segments from corneal flat mounts at the border between normally vascularized conjunctiva (Left) and normally avascular cornea (Right) immunostained with CD31. (E) Morphometry. (F) Diagram of experimental design. (Left) Cross section of normal eye. (Right) Enlargement of cornea. Cells in corneal stroma are keratocytes; red lines are invading blood vessels. (Magnification: A–D, ×100.)

Table 1.

Summary of results from the experiment described in Fig. 2F

Cornea	Inflammatory insult	Epithelial VEGFR3	Angiogenic response
Untreated	Absent	Present	No
De-epithelialized	Absent	Absent	Minimal
Cauterized	Present	Present	Minimal
Both	Present	Absent	High

To study whether corneal epithelium could inhibit inflammatory cell influx into the cornea after a stimulus, the number of inflammatory cells was compared at 72 h after cautery between mice receiving cautery in the presence or absence of corneal epithelium. There were significantly more inflammatory cells (mostly GR1⁺ and CD11b⁺; data not shown) in de-epithelialized corneas receiving cautery (352 ± 82 per section) compared with the epithelialized corneas of mice that received cautery alone (18 ± 6 per section; P < 0.01).

Next, the established mouse model of suture-induced inflammatory CHA was used (as a second in vivo model) to further study the anti-inflammatory, antiangiogenic effects of corneal epithelium in vivo. In this model three 11-0 sutures placed in the paracentral corneal stroma induced robust CHA within the first week postoperatively (ref. 8 and Fig. 3A and B). We compared the degree of CHA after suture placement in mice whose corneas were intact versus those that were de-epithelialized. Absence of corneal epithelium significantly enhanced the neovascular response compared with corneas with intact epithelium (P < 0.0001; Fig. 3E). Furthermore, sutures placed in corneas with intact corneal epithelium displayed significantly reduced CD45 inflammatory cell recruitment compared with de-epithelialized controls (Fig. 3 C and D). We next layered corneal epithelium onto corneas that received central sutures immediately after de-epithelialization and compared the neovascular response with controls that were left de-epithelialized. Morphometry of the area covered by blood vessels at day 7 demonstrated a significant inhibitory effect of corneal epithelium on inflammatory CHA (32 ± 9.8%) compared with mice without epithelial reapplication (51.8 ± 5.2%; P < 0.001).

Fig. 3.

Angiosuppressive effect of corneal epithelium II. (A–D) Absence of corneal epithelium (A) significantly enhances the neovascular response (arrowhead: blood vessel) in the model of suture-induced inflammatory angiogenesis [in parallel with increased influx of CD45⁺ inflammatory cells (C and D; CD45 immunostaining, arrows)] (B). (E) Results of morphometry. (F) Up-regulation of VEGFR3 ligands in inflammatory corneal angiogenesis. The ligands of VEGFR3, VEGF-C, and VEGF–D are significantly up-regulated in conditions associated with inflammatory corneal angiogenesis (suture model from B; P < 0.05 for both ligands; lane 1, control; lane 2, suture-induced inflammatory angiogenesis; expected sizes: VEGF-C, 531 bp; VEGF-D, 307 bp; GAPDH, 245 bp). (Magnification: C and D, ×200.)

After confirming the significant suppressive effect of the epithelium on inflammatory CHA in vivo, we explored whether this effect is mediated by VEGFR3–ligand interactions. First, we determined whether an inflammatory angiogenic stimulus in the cornea is associated with up-regulation of VEGFR3 ligands VEGF-C and VEGF-D. Corneal suturing induced a significant (≈3-fold) up-regulation of both VEGF-C and VEFG-D mRNA (Fig. 3F), whereas the VEGFR3 mRNA level remained unchanged (data not shown).

Because the data described above established that (i) VEGFR3 is strongly expressed by the corneal epithelium, (ii) the epithelium displays strong antiangiogenic effects in two in vivo models, and (iii) VEGFR3 ligands VEGF-C and VEGF–D are up-regulated in conditions leading to inflammatory CHA, we next tested directly whether epithelial VEGFR3 expression accounts for the antiangiogenic and anti-inflammatory properties of corneal epithelium in vivo. We administered recombinant mouse VEGFR3/Fc chimeric protein (ref. 18; comprising the amino acid residues 25–770 of the extracellular domain of mouse VEGFR3 coupled to human Fc IgG₁) subconjunctivally to mouse eyes from which the corneal epithelium had been removed and cautery applied. The Fc protein served as control. Subconjunctival injection of neither Fc protein nor the chimeric protein induced CHA (data not shown). However, whereas cauterized corneas that received control subconjunctival injection displayed a strong angiogenic (Fig. 4) response after de-epithelialization, cauterized deepithelialized corneas that were treated with VEGFR3 chimeric protein displayed significantly reduced CHA (P < 0.0001), demonstrating that the constitutive angiostatic effect of the corneal epithelium could be recreated in the de-epithelialized cornea by a VEGFR3 chimeric molecule.

Fig. 4.

Antiangiogenic effect of a VEGFR3 chimeric protein. A VEGFR3 chimeric protein, ligating VEGF-C and VEGF-D, can substitute for the antiangiogenic effect of VEGFR3-expressing corneal epithelium. The neovascular response after cautery of de-epithelialized corneas (representative segment from CD31-stained corneal flat mount) (A) is significantly diminished (C) when a VEGFR3 chimeric protein is administered locally (B). (Magnification: A and B, ×100.)

Next, we directly tested whether the antiangiogenic effect of epithelial VEGFR3 could be suppressed in corneas receiving an inflammatory stimulus. Syngeneic corneal epithelium was treated ex vivo with a blocking anti-VEGFR3 antibody (ref. 19; mF4–31C1; 30 min; 2.1 mg/ml) or control IgG, and then layered onto corneas of BALB/c mice immediately after de-epithelialization. When evaluated after 7 days (Fig. 5), ex vivo blockade of corneal epithelium with anti-VEGFR3 significantly inhibited its antiangiogenic capacity. Specifically, the vascularized area after corneal suturing was significantly greater in the group receiving the anti-VEGFR3-treated epithelium than in controls (P < 0.001), providing definitive and direct support for VEGFR3-mediated (rather than another epithelium-specific mechanism) suppression of CHA by corneal epithelium.

Fig. 5.

Antiangiogenic effect of corneal epithelium critically depends on epithelial VEGFR3. Direct inhibition of epithelial VEGFR3 using ex vivo treatment with neutralizing antibodies (19) diminishes the epithelium’s ability to dampen angiogenesis. (A) Experimental design. (B and C) The neovascular response to an inflammatory stimulus is significantly increased after ex vivo treatment with a neutralizing anti-VEGFR3 antibody. Representative segments from CD31-stained corneal flat mounts (C) are compared with corneas receiving epithelial transplants that received ex vivo treatment with control IgG (B). (D) Morphometry. (Magnification: B and C, ×100.)

Discussion

The results provided here allow three important conclusions to be drawn: First, corneal epithelial VEGFR3 expression and the capacity to bind angiogenic growth factors VEGF-C and VEGF-D constitutes a potent mechanism inhibiting inflammation-induced angiogenesis. Second, the data provide evidence for an antiangiogenic role for VEGFR3. VEGFR3 may not only provide proangiogenic signaling mediating hemangiogenesis and lymphangiogenesis via ligation of VEGF-C and VEGF-D while expressed on endothelium (as has been shown extensively; refs. 16 and 20–23), but VEGFR3 may also display antiangiogenic properties when expressed at an avascular site by nonendothelial (i.e., here epithelial) cells, where it can act as a decoy receptor. Third, this regulation of corneal angiogenesis is critical for maintenance of corneal clarity.

To date, no one factor has been identified as singularly critical to maintaining corneal avascularity, a unique feature that the cornea shares only with cartilage. The challenge of avascularity is particularly critical for the cornea, which because of its anatomically exposed position and unkeratinized surface, is constantly confronted with potential inflammatory stimuli such as particles landing onto the eye surface and mechanical stresses from the microabrasive effects of blinking and rubbing. The cornea therefore needs to balance its usually high threshold for CHA with its ability to react, when required, to sight-threatening injuries (e.g., aggressive microbial invasion) with a robust and swift angiogenic response to enhance the immune defense against these threats (1). A system that buffers low concentrations of angiogenic factors, but on the other hand allows for angiogenesis to occur if a certain threshold is reached, fits with the specificities of the VEGFR3 sink described herein.

Why then is there not a “trap” for VEGF-A instead? VEGF-A, the principal hemangiogenic growth factor binding to VEGFR1/2, has been implicated as the key player in mediating CHA (3, 5, 10). But there is no significant expression of VEGFR1/2 in the normal cornea (11) and the physiologic VEGFR3 sink proposed herein is unable to bind and neutralize VEGF-A. One likely explanation for this seeming paradox is the fact that most of the minor insults to which the cornea is constantly exposed, which would normally cause unnecessary but potentially vision-threatening CHA, are inflammatory (rather than hypoxic) in nature; hence the angiogenic response that needs to be regulated is inflammatory-driven VEGF-C in contrast to (primarily) hypoxia-driven VEGF-A (which can also be up-regulated by inflammatory cytokines; ref. 24). Other explanations for this paradox exist; for example, it may well be that the initial (e.g., IL-1 triggered) release of inflammatory VEGF-C and VEGF-D is upstream of a subsequent VEGF-A release by recruited neutrophils and macrophages (8). Therefore, an initial (decoy receptor-mediated) neutralization of VEGF-C and VEGF-D could interfere with this mechanism, that is, an initial VEGF-C/-D sink might be enough to prevent the angiogenic cascade if the causative stimulus is low intensity. According to this concept, corneal epithelial VEGFR3 may be a physiologic trap to prevent low-grade inflammation-induced, but potentially vision-threatening and physiologically unnecessary, CHA. Additionally, VEGFR3 signaling is important for maintenance of newly outgrown blood vessels (25). Hence, in addition to VEGFR3 overexpression by epithelium acting as a sink to deplete VEGFR2-binding ligands VEGF-C and VEGF–D, corneal epithelial VEGFR3 offers a mechanism to promote the regression of new blood vessels if the angiogenic stimulus is not overwhelmingly strong or sustained. Indeed, multiple and longstanding clinical and animal experimental evidence relating persistent epithelial defects in the setting of inflammation with potent CHA support this concept of an epithelial “defense” against blood vessel in-growth.

Whereas several known inhibitors of angiogenesis have been located within the cornea, the precise role of each of these factors in maintaining corneal avascularity is unknown. The fact that several inhibitors are present, and that genetic deletion of one or several of them does not induce spontaneous corneal neovascularization, suggests that a redundant system is maintaining corneal avascularity. The fact that VEGF-C is the main angiogenic factor up-regulated in an inflammatory milieu, and that it binds to its ectopically expressed VEGFR3 on epithelial cells, suggests in the aggregate that the VEGFR3 sink is important for maintaining corneal avascularity in the setting of inflammation. Soluble VEGFR1 binding VEGF-A in contrast may be more suitable for blocking hypoxia-driven neovascularization.

VEGFR3 during development is expressed on venous and lymphatic vascular endothelium but later becomes largely restricted to lymphatics (21), acting as the prime mediator of lymphangiogenesis in the adult (16). For this reason it has commanded significant interest because of its potential use in therapies for lymphatic disorders and in oncology where tumor cell access to lymphatics is critical for their metastasis (16, 26, 27). Interestingly, VEGFR3 expression by an epithelium has not been reported previously to our knowledge in an adult tissue (28). This concept of an “ectopic” expression of a molecule that can in turn regulate a key tissue response fits into a small, but growing, body of evidence that tissue parenchymal cells may use cytokine and growth factor receptors as sinks, or decoys, to provide an additional level of regulation for the potentially pathological role of these factors (29). Because we demonstrate that VEGF-C can ligate epithelial VEGFR3 and induce receptor phosphorylation we cannot exclude an “antiangiogenic” signaling role for corneal epithelial VEGFR3 (apart from its sequestering bioactive ligand from endothelial VEGFR2/3). However, because a nonsignaling VEGFR3 chimeric protein in our study could substitute for the antiangiogenic effect of corneal epithelial VEGFR3 suggests that membrane-bound receptor signaling is not essential for the antiangiogenic effect of this receptor in the cornea, and that similar soluble receptors could be used therapeutically to suppress pathological tissue responses.

The broad implications of our study are several-fold: First, these data suggest a “check and balance” system involved in regulating the angiogenic response to inflammation in a tissue whose avascularity is critical for vision via constitutively high ectopic overexpression of VEGFR3 by the epithelium. Second, the data suggest a dual and differential role of VEGFR3 in regulating angiogenesis so that it can serve not only as a mediator but also as an inhibitor of angiogenesis. Finally, these findings provide potential venues for inhibition of sight-threatening corneal angiogenesis. Potential therapeutic applications for the eye include overexpression of VEGFR3 on transplanted corneas to inhibit posttransplantation angiogenesis and secondary immune rejection or application of (cultured) corneal epithelium onto common nonhealing corneal ulcers to promote wound healing without angiogenesis. Nonocular applications could include induction of VEGFR3 overexpression by tumor epithelia to suppress angiogenesis, just to name a few.

Materials and Methods

Mice and Anesthesia.

BALB/c mice aged 6–8 weeks (Taconic Farms) were used in all experiments and treated in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals. Mice were anesthetized with a mixture of ketamine and xylazine (120 mg/kg body weight and 20 mg/kg body weight, respectively).

Surgical Manipulations of the Corneal Surface and Autologous Corneal Epithelial Transplantation.

Corneal de-epithelialization, cauterization, and combined treatment.

To induce an inflammatory but nonvascularizing response in the cornea (17), a fine diathermy tip (Fine Ophthalmic Tip, Aaron, St. Petersburg, FL) was placed briefly on eight separate points for 1 s each within the central 2 mm of the cornea. To examine the effect of the corneal epithelium on modulating CHA, the central 2 mm of the corneal surface was marked with a 2-mm trephine (Acuderm, Fort Lauderdale, FL), after which the epithelium was removed with a microsurgical knife (5 mm; Surgistar; Windsor Medical, Portsmouth, NH) by gently scraping over the corneal surface in the demarcated area without injuring the underlying corneal stroma. The completeness of de-epithelialization was verified by using hematoxylin/eosin serial sections and fluorescein sodium staining in vivo (data not shown). Application of cautery to these de-epithelialized corneas was performed to determine whether the neovascular response had been altered.

Transplantation of corneal epithelial cell sheets.

To further study the antiangiogenic effects of the corneal epithelium, corneal epithelium was reapplied to the denuded area of the de-epithelialized eyes to determine whether the putative antiangiogenic effect of the corneal epithelium could be re-established. To accomplish this, the central 2 mm of syngeneic BALB/c donor corneas was excised and incubated in 2% EDTA at 37°C for 1 h to enable separation of the corneal epithelium from stroma as described (14). After three washes with PBS for 5 min each, the epithelial sheet (referred to subsequently as epithelium) was applied, basal layer down, to the de-epithelialized surface of the recipient cornea. A surgical lid closure was performed to secure the epithelium in place.

Mouse model of suture-induced, inflammatory corneal angiogenesis.

The mouse model of suture-induced inflammatory CHA (which unlike cautery leads to profound CHA) was adopted as a contrasting model to the cauterization model, which induces inflammation without CHA (8). Briefly, a 2-mm corneal trephine was gently placed on the cornea to mark the central corneal area. Three 11-0 sutures (Sharpoint Nylon 11-0; Surgical Specialty, Reading, PA) were then placed intrastromally with two stromal incursions extending over 120° of corneal circumference each. The outer point of suture placement was chosen as halfway between the limbus and the line outlined by the 2-mm trephine; the inner suture point was at the same distance from the 2-mm trephine line to obtain standardized angiogenic responses. Sutures were left in place for 7 days. Mice were killed, and the cornea was then excised and a flat-mount double-immunohistochemistry was performed as described below (8).

Immunohistochemistry and Morphometry of Corneal Angiogenesis.

Briefly, corneal flat mounts were rinsed several times in PBS, fixed in acetone, rinsed in PBS, blocked with 2% BSA, stained with an FITC-conjugated CD31/PECAM-1 antibody overnight (1:100; Santa Cruz Biotechnology), washed (three times for 5 min with PBS), blocked, and stained with anti-LYVE-1 antibody overnight (1:500; a lymphatic endothelium-specific hyaluronic acid receptor; gift from D. Jackson, University of Oxford, Oxford, U.K.), which was visualized with a Cy3-conjugated secondary antibody (1:100; Jackson ImmunoResearch). Double-stained sections were analyzed with a Zeiss Axiophot microscope. Digital pictures of the flat mounts were taken with the Spot Image Analysis system, and the area covered by CD31⁺/LYVE-1⁻ blood vessels (8) was measured with National Institutes of Health image software. LYVE-1 staining was performed to make sure that only CD31⁺/LYVE-1⁻ blood vessels were measured and not CD31⁺/LYVE-1⁺ lymphatic vessels. The total corneal area was outlined by using the innermost vessel of the limbal arcade as the border, and the area of CHA was then calculated and normalized to the total corneal area (expressed as a percentage of the cornea covered by blood vessels). The mean area of blood vessels extending beyond the limbus caused by the normal irregularity of the limbal border architecture was subtracted from these values (i.e., 11%). Paraffin embedding of corneas and immunostaining was done as described (15).

Immunohistochemistry for VEGFR3.

To verify the expression of VEGFR3 in MCE, immunohistochemistry on frozen and paraffin-embedded sections of normal mouse eyes was performed with a polyclonal rabbit–anti-mouse antibody (FLT-4, M20; 1:200; Santa Cruz Biotechnology) and a polyclonal goat–anti-mouse antibody (FLT-4, 102806 and 102804; 1:50; R & D Systems) as described (8, 15). EDTA treatment for 1 h leading to preparation of corneal epithelial sheets did not reduce staining intensity for VEGFR3 on the epithelium.

Flow Cytometry.

Epithelial cells of normal BALB/c mice were obtained and used for flow analysis as described (30). Two-color staining was used with the corneal epithelial-specific K-12 staining red [phycoerythrin (PE); gift of Tung-Tien Sun, New York University Medical School, New York, NY] and the anti-VEGFR3 antibody staining green (FITC). The analysis was done by gating on K-12⁺ cells using appropriate isotype and cell culture controls to adjust color compensation and gating parameters. Cells were washed and analyzed with an Epics XL flow cytometer (Beckmann Coulter). The proportion of K-12⁺ cells that were also VEGFR3⁺ was quantified. As controls, we used rabbit IgG (for PE) and goat IgG (for FITC; Santa Cruz Biotechnology).

Histological Quantification of Inflammatory Cells.

The presence of inflammatory cells in normal corneas and their recruitment into corneas after microsurgical manipulations was quantified in hematoxylin/eosin-stained serial sections of plastic-embedded corneas, fixed in 10% paraformaldehyde after enucleation. In addition, corneal whole mounts and frozen sections were stained for the macrophage markers CD11b (Pharmingen) and CD68 (Santa Cruz Biotechnology), the panleukocyte marker CD45 (Pharmingen), and the neutrophil marker GR1 (Pharmingen).

VEGFR3/Fc Chimeric Protein.

To investigate whether exogenous VEGFR3 inhibits inflammatory CHA and substitutes for the putative function of corneal epithelial VEGFR3, we used a recombinant mouse VEGFR3 (FLT-4)/Fc chimeric protein (R & D Systems) in mouse corneas subjected to de-epithelialization- and cautery-induced CHA. For this, a DNA sequence encoding the signal peptide from human CD33 joined with amino acid residues 25–770 of the extracellular domain of mouse VEGFR3 (18) was fused to the 6× histidine-tagged Fc of human IgG₁ via a polypeptide linker. This chimeric protein was expressed in Sf21 cells. After corneal de-epithelialization and cautery, one group of mice was treated subconjunctivally with 10 μl of VEGFR3 chimeric protein immediately after surgery and 2, 4, and 6 days later. Control mice received only the human IgG₁-Fc instead (R & D Systems). The degree of CHA was quantified 1 week after surgery and 24 h after the last injection, using the above-described immunohistochemical flat-mount morphometry. Each group comprised four mice, and the experiment was performed twice with similar results.

Neutralizing Anti-VEGFR3 Antibody.

A neutralizing anti-VEGFR3 antibody (mF4–31C1; ImClone Systems; ref. 19) was used to directly test the role of corneal epithelial VEGFR3 expression on angiogenesis. Using this anti-VEGFR3 antibody, we compared the effect of epithelium incubated ex vivo in blocking anti-VEGFR3 antibody (2.1 mg/ml for 30 min) with the effect of similar epithelium incubated in control IgG solution (Jackson ImmunoResearch) at the same concentration. After extensive PBS washing, the epithelium was layered onto inflamed and de-epithelialized corneas, and the degree of CHA was measured.

RT-PCR for VEGF-C, VEGF-D, and VEGFR3 and Real-Time PCR for VEGFR3.

RT-PCR was carried out as described (2). Briefly, total RNA was extracted from central corneal epithelium immediately after the deaths of the mice with RNAStat-60 (Tel-Test, Friendswood, TX). The central epithelium of corneas of 10 mice (20 corneas) was gently removed by scraping and then pooled (similar results were obtained by pooling epithelium, which were retrieved by 1 h of EDTA incubation; data not shown). In addition, full-thickness corneas from 20 normal eyes and 20 eyes bearing intrastromal sutures for 12 h were obtained to compare VEGF-C and VEGF-D expression. From 1 μg of mRNA, cDNA was synthesized with Moloney murine leukemia virus reverse transcriptase (Promega) according to the manufacturer’s instructions. The following primers were used for PCR from 5′ to 3′: GAPDH sense, GGTGAAGGTCGGTGTGAACGGA; GAPDH antisense, TGTTAGTGGGGTCTCGCTCCTG; VEGFR3 sense, GCGACAGGGTTCTCATAA; VEGFR3 antisense, CGTTGCCTCATTGTGATTAG; VEGF-C sense, GTCTGTGTCCAGCGTAGATG; VEGF-C antisense, GCTGGCAGAGAACGTCTAAT; VEGF-D sense, GCGGCAACTTTCTATGACA; and VEGF-D antisense, AGCACTTACAACCCGTATGG. All primers were designed by Genosys (The Woodlands, TX). PCR was carried out under the following conditions: denaturation at 94°C, annealing at 55°C, and extension at 72°C. After 40 cycles of amplification (AmpliTaq DNA polymerase; Applied Biosystems), PCR products were electrophoresed in 2% agarose gel and visualized by ethidium bromide staining (0.5 μg/ml ethidium bromide) for 40 min. Photographs of the gel were taken with a high-resolution camera, and the density of the bands was analyzed on the gel by using UV illumination and image one image analysis software (Bio-Rad). The expression level of mRNA was standardized by the expression of GAPDH as an internal control. The predicted sizes of PCR products are 245 bp for GAPDH, 531 bp for VEGF-C, 307 bp for VEGF-D, and 290 bp for VEGFR3. For real-time PCR for comparison of VEGFR3 levels in central corneal epithelium versus central stroma and endothelium, first-strand cDNA was synthesized from 1 μg of total RNA with random hexamers by using Super Script III (Invitrogen) according to the manufacturer’s protocol. Real-time PCR was performed with FAM-MGB dye-labeled predesigned primers (Applied Biosystems) for VEGFR3 (Assay ID Mm00433354_m1) according to the manufacturer’s recommendations: 1 μl of cDNA was loaded in each well, and assays were performed in duplicate. A nontemplate control was included in all of the experiments to evaluate DNA contamination of the isolated RNA and the reagents used. The comparative C_T (threshold cycle) method was used to determine the difference (ΔC_T) between the C_T of normal corneal stroma and the C_T of normal corneal epithelium. Before subtraction, the C_T was normalized by the C_T of the endogenous reference gene, GAPDH.

Immunoprecipitation and Western Blot Analysis.

Cultured MCE (gift of J. Niederkorn, University of Texas Southwestern Medical Center, Dallas) were maintained in DMEM supplemented with 10% FBS, 2 mM glutamine, and antibiotics at 37°C and CO₂. To assay VEGFR3 phosphorylation, subconfluent cells were incubated overnight in serum-free media, incubated for 1 h in serum-free media containing 100 μM sodium orthovanadate, then stimulated for 8 min with 100 ng/ml of recombinant VEGF-C (R&D Systems). Reactions were terminated by washing in cold PBS. Cells were collected in lysis buffer (10 mM Tris·HCl, pH 7.4/5 mM EDTA/50 mM NaCl/1% Triton X-100/50 mM NaF/1 mM PMSF/2 mM Na₃VO₄/20 mg/ml aprotinin). Lysates were precleared with protein A-Sepharose (Amersham Pharmacia Biosciences) for 1 h at 4°C. VEGFR3 was immunoprecipitated by using a polyclonal rabbit anti-mouse VEGFR3 antibody (M20; Santa Cruz Biotechnology) overnight at 4°C. Immunoprecipitated complexes were analyzed by SDS/PAGE using a mixture (1:1) of antiphosphotyrosine antibodies: PY20 (Transduction Laboratories, Lexington, KY) and 4G10 (Upstate USA, Chicago). The membranes were stripped by incubation for 30 min in 6.25 mM Tris·HCl (pH 6.8), 2% SDS, and 100 mM β-mercaptoethanol at 50°C and reprobed with polyclonal VEGFR3 antibody (M20) to detect total VEGFR3.

Statistical Analyses.

Statistical significance was analyzed by Mann–Whitney test. Differences were considered significant at P < 0.05. Each experiment was performed at least twice with similar results. Graphs were drawn with GraphPad (San Diego, CA) prism, version 3.02.

Abbreviations

CHA

corneal hemangiogenesis

VEGFR

VEGF receptor

K-12

keratin-12

MCE

murine corneal epithelial cells

C_T

threshold cycle.