Expression of Vascular Endothelial Growth Factor and Its Receptors in Inflamed and Vascularized Human Corneas

Abstract

purpose. To help further define the possible role of vascular endothelial growth factor (VEGF) in the pathogenesis of corneal neovascularization, the expression of VEGF and of its receptors Flt-1 and Flk-1 was investigated in various inflammatory corneal diseases.

methods. Polyclonal antibodies to VEGF and its receptors were used for immunohistochemical staining of frozen sections of 38 human corneas with various degrees of neovascularization and inflammation. In addition, a panel of monoclonal antibodies was used to characterize the composition of the inflammatory infiltrates and to confirm the presence of neovascularization. Furthermore, VEGF concentrations were determined in vascularized corneas using a sensitive enzyme-linked immunosorbent assay.

results. VEGF was expressed by epithelial cells, by corneal endothelial cells, by vascular endothelial cells of limbal vessels and of newly formed vessels in the stroma, and weakly by keratocytes. Furthermore, VEGF expression was often markedly increased in inflamed corneas on epithelial cells and on vascular endothelial cells, particularly in the vicinity of macrophage infiltrates, and on fibroblasts in scar tissue. Correspondingly, VEGF concentrations were significantly higher in vascularized corneas compared with normal control corneas (P < 0.001). Expression of both VEGF receptors, Flt-1 and Flk-1, was increased on endothelial cells of newly formed vessels in the stroma of inflamed corneas compared with limbal vessels of normal control corneas. In addition, Flt-1 was also expressed by corneal endothelial cells and by macrophages, whereas Flk-1 expression was lacking.

conclusions. These results demonstrate that VEGF, Flt-1, and Flk-1 are strongly expressed in inflamed and vascularized human corneas and, thus, may play an important role in corneal neovascularization.

Neovascularization is a severe complication particularly of ischemic retinal diseases like diabetic retinopathy, branch and central retinal vein occlusion, and retinopathy of prematurity. However, in various inflammatory corneal diseases, neovascularization may also occur particularly in the chronic course of the disease. Consequences of corneal neovascularization may not only be a severe reduction of visual acuity up to blindness but also a bad prognosis in corneal transplantation due to a loss of the immunologic privilege of the avascular cornea. ^{1 2} However, the pathogenesis of corneal angiogenesis has not yet been clearly defined, and the identity and significance of the actual angiogenic growth factors are unknown. Several studies have shown that vascular endothelial growth factor (VEGF), which was identified about one decade ago, plays a major role in vasculogenesis and in pathologic neovascularization. ^{3 4 5 6 7 8 9 10 11 12 13} This cytokine was originally identified as a 34- to 42-kDa dimeric heparin binding glycoprotein secreted by tumor cells and later by other cells. ^{14 15 16 17 18 19} Four VEGF isoforms (VEGF₁₂₁, VEGF₁₆₅, VEGF₁₈₉, and VEGF₂₀₆) have been identified, which are generated by alternative splicing of messenger RNA (mRNA). ^{20 21} VEGF₁₆₅ is the most abundant molecular species in the majority of tissues. ³ VEGF acts through two high-affinity receptor tyrosine kinases (RTK), Flt-1, VEGF receptor 1 (VEGFR-1) and KDR/Flk-1, or VEGFR-2. ^{22 23 24 25} Unlike other angiogenic factors, such as acidic and basic fibroblast growth factors (aFGF, bFGF) and platelet-derived endothelial cell growth factor, VEGF is a secreted peptide ^{3 21} and has been shown to promote several steps of angiogenesis, including proliferation, migration, proteolytic activity, and capillary tube formation of endothelial cells. ^{3 17 26 27 28 29} Furthermore, this protein stimulates angiogenesis in a noninflammatory model of neovascularization in the mouse cornea ⁴ and was recently identified as a functional endogenous corneal angiogenic factor required for inflammatory neovascularization in a rat model. ⁶ The aim of the present study was to investigate whether and by which cells VEGF and its receptors are expressed in vascularized and inflamed human corneas and, thus, whether they may play a role in corneal neovascularization. 

Methods

Tissues

Thirty-eight vascularized corneal buttons were obtained at the time of penetrating keratoplasty in various inflammatory corneal diseases. Informed consent was obtained from all patients after the nature and possible consequences of the study had been explained. The research followed the tenets of the Declaration of Helsinki and was approved by the institutional human experimentation committee. Only corneas with significant vascularization in at least 2 or more quadrants were included in the study. The number of corneas investigated in the present study and the indications for keratoplasty are listed in Table 1 . Because the pathogenesis of corneal neovascularization may vary according to the etiology of the disease, particularly according to the presence of limbal deficiency, some sections of each cornea were stained with hematoxylin and eosin and periodic acid–Schiff (PAS) to confirm morphologic observations made on immunohistochemical stained slides and to investigate the presence of conjunctival goblet cells in the epithelium for detection of corneas with limbal deficiency. ^{30 31 32} The clinical diagnoses and the number of corneas with limbal deficiency are also detailed in Table 1 . For comparison with the inflamed corneas, we used 10 normal human corneas excised with a scleral rim obtained from donor eyes. The central part of these control corneas was used for keratoplasty (6–7.1 mm transplants), whereas the remaining corneoscleral tissue was used for immunohistochemistry. Furthermore, to detect topographical variations in VEGF expression in normal corneas we performed additional immunohistochemical studies on two whole corneas from donor eyes and on 3 whole normal corneas from eyes that underwent enucleation due to central choroidal melanoma. All these corneas were clear and uninflamed. Immediately after trephination the corneal buttons and the control corneas were bisected with a razor blade, snap-frozen, and stored in liquid nitrogen until processed further. 

Antibodies

Polyclonal Ab to VEGF (anti–VEGF₁₆₅; R&D Systems, Wiesbaden, Germany) and its receptors Flt-1 and Flk-1/KDR (anti–Flt-1 and anti–Flk-1; Santa Cruz Biotechnology, Santa Cruz, CA) were used for immunohistochemical staining of the tissue specimens. To confirm the presence of neovascularization in the inflamed corneas, we used a monoclonal antibody (mAb) to the von Willebrand factor (factor VIII–related antigen [Dako–vWf]; Dakopatts, Copenhagen, Denmark), which selectively stains vascular endothelial cells. ³³ In addition, we used a panel of mAbs (Dakopatts) to characterize the composition of the inflammatory infiltrates in the inflamed corneas. Anti-CD68 mAb (Dako–EBM 11) was used to detect monocytes/macrophages, anti-CD2 mAb (Dako–T11) and anti-CD3 mAb (Dako–T3) to detect mature peripheral T cells, anti-CD15 mAb (Dako–M1) to detect granulocytes, and anti-CD22 mAb (Dako–CD 22) to mark B cells. 

Immunohistochemical Staining Technique

Six-micron-thick frozen sections were cut from the specimens in a Reichert Jung cryostat (Leica–Reichert, Vienna, Austria) at −20°C and mounted on poly-l-lysine–coated slides. The sections were fixed in acetone at 4°C for 10 minutes and stained with the streptavidin–biotin–peroxidase method that has been described previously in detail. ³⁴ In brief, the sections were immersed in a solution of 0.3% H₂O₂ in distilled H₂O for 20 minutes to block endogenous peroxidase activity and then incubated with the selected primary Ab for 60 minutes at room temperature. The optimal dilution of Ab was determined by titration. The sections were then incubated for 30 minutes either with biotinylated rabbit anti-mouse immunoglobulins, biotinylated rabbit anti-goat immunoglobulins, or biotinylated swine anti-rabbit immunoglobulins (all diluted 1:300 in Tris-buffered saline solution[ TBS]; Dakopatts) corresponding to the animal species used for production of the primary Ab, and, finally, incubated with a freshly prepared streptavidin–biotin–peroxidase complex (Dakopatts) for 30 minutes at room temperature. The sections were then immersed in a solution of 3-amino-9-ethylcarbazole, dimethylformamide, and H₂O₂ in acetate buffer (pH 5.2; Dakopatts) and finally counterstained with Mayer’s hematoxylin. The slides were examined under an Olympus microscope (Olympus, Vienna, Austria). Negative controls were prepared by substituting nonimmune mouse serum and equivalent amounts of irrelevant mouse Ab to cytomegalovirus for the primary Abs. Furthermore, to test the specificity of polyclonal anti-VEGF Ab, some sections were stained with this Ab preabsorbed with VEGF. 

Enzyme-Linked Immunosorbent Assay of VEGF Protein

VEGF levels were analyzed in the remaining halves of 4 normal and 9 vascularized human corneas, the other halves of which were used for immunohistochemistry. The diagnoses of the latter corneas, which had significant neovascularization in at least 3 quadrants, are listed in Table 2 . 

Half corneal buttons (7 mm in diameter) were sonicated for 10 seconds (Branson Sonifier 250; Danbury, CT) in 500 μl calibrator diluent containing 5 μl proteinase inhibitor cocktail (Sigma, Deisenhofen, Germany). Samples were centrifuged for 10 minutes at 13,000 rpm, the supernatant collected and the pellet used for protein determination with the Bradford protein assay (Bio–Rad, Vienna, Austria). VEGF was determined with a highly sensitive commercially available enzyme-linked immunosorbent assay (ELISA) kit (R&D Systems), which recognizes the165-amino acid splice variant of human VEGF. The assay was performed according to the manufacturer’s instructions. Briefly, standards or tissue samples (200 μl) were pipetted into the (Ab-coated) 96 wells containing 50-μl assay diluent and incubated for 2 hours at room temperature on a shaker. The wells were then washed 3 times with wash buffer, then 200 μl VEGF conjugate was added and the samples again incubated for 2 hours at room temperature on a shaker. Samples were then again washed (3 times), 200 μl substrate buffer was added, the samples incubated for 20 minutes at room temperature, the reaction stopped, and the absorption measured in an ELISA reader (MWG, Ebersberg, Germany) at 450 nm. All measurements were performed in duplicate. The tissue sample concentration was calculated from the standard curve and corrected for the protein concentration. The lower detection limit of the ELISA was 5.0 pg/ml. Mean VEGF levels of normal and vascularized corneas were statistically compared using Student’s unpaired t-test. 

Results

Immunohistochemical Findings

Negative controls, prepared either with nonimmune mouse serum or irrelevant Abs of the same subclass as the primary Abs, revealed no specific staining. 

Furthermore, no specific staining could be observed using anti-VEGF Abs preabsorbed with VEGF. 

Normal Corneas

In normal corneas excised with a scleral rim, VEGF was expressed by epithelial and corneal endothelial cells and by vascular endothelial cells of limbal vessels, and weakly by some keratocytes (Table 3 , Fig. 1 ). Topographical analysis revealed that VEGF was equally expressed throughout the epithelium of normal corneas, including the center and the periphery. 

Both VEGF receptors, Flt-1 and Flk-1/KDR, were moderately expressed on vascular endothelial cells of limbal vessels (Table 4) . However, weak expression of Flt-1 only was found on corneal endothelial cells, whereas Flk-1/KDR expression was lacking (Table 4) . 

Inflamed and Vascularized Corneas

There was evidence of marked neovascularization in virtually all of these inflamed corneas as was confirmed by staining with mAb, anti–factor VIII–related antigen (Dako–vWf). 

According to the presence of goblet cells on the corneal surface, ³¹ 4 cases with limbal deficiency could be identified, 3 of which were due to severe chemical burns and 1 due to repeated graft failures after multiple keratoplasties in a patient with atopic keratoconjunctivitis (Table 1) . All of these cases showed conjunctival epithelial ingrowth (conjunctivalization) with corresponding neovascularization, particularly in the anterior stroma of all 4 quadrants. In addition, 3 of these corneas disclosed severe destruction of Bowman’s membrane with dense scar tissue in the anterior stroma and central epithelial defects. 

Although other inflamed corneas also showed severe neovascularization, epithelial irregularities, and, often, defects of Bowman’s membrane with stromal scarring, limbal deficiency based on detection of PAS-positive conjunctival goblet cells in the epithelium could not be verified in these corneas. However, because the diameter of corneal explants varied between 6 and 7.1 mm, some cases with partial limbal deficiency may have been overlooked because goblet cells indicating limbal deficiency may have only been present in the perilimbal area of the graft bed and not on the surface of the excised corneal buttons. ³¹  

Inflammatory Infiltrates

By analyzing inflammatory infiltrates in the diseased corneas with mAbs, we were able to show that inflammatory infiltrates of corneas with vascularized scars resulting from bacterial corneal ulcers consisted of high densities of macrophages and low densities of polymorphonuclear leukocytes. Corneas with chemical burns disclosed high densities of macrophages and low densities of T cells, particularly in their anterior stroma. Similarly, all corneas with limbal deficiency disclosed high densities of macrophages and low densities of T cells, particularly in the anterior stroma, and lower densities of these inflammatory cells also in the epithelium. The cellular infiltrates in the rejected corneal allografts and in corneas with herpetic stromal keratitis and zoster keratitis also were composed predominantly of macrophages and T cells. As was to be expected, there was considerable individual variation in the density and composition of the inflammatory infiltrates in the inflamed corneas, possibly according to the cause, the time course, and the severity of the disease. 

Expression of VEGF, Flt-1, and Flk-1 in Inflamed and Vascularized Corneas

VEGF was expressed moderately to strongly by epithelial cells and by corneal endothelial cells, weakly to moderately by keratocytes (fibroblasts), and moderately to strongly by vascular endothelial cells of newly formed vessels (Figs. 2 3 4A ; Table 3 ), and by macrophages (Fig. 3) in corneas with vascularized posttraumatic scars; rejected corneal allografts; corneas with herpetic stromal keratitis, zoster keratitis, chemical burns (in both, with and without limbal deficiency), and atopic keratoconjunctivitis; in corneas with vascularized scars resulting from bacterial corneal ulcers, and in corneas with fungal keratitis. Compared with normal corneas, expression of VEGF was often markedly increased in inflamed corneas on epithelial cells and on vascular endothelial cells of newly formed vessels, particularly in the vicinity of macrophage infiltrates (Fig. 3) , on fibroblasts in scar tissue, and on epithelial cells of all 4 corneas with limbal deficiency. As regards expression of VEGF receptors Flt-1 and Flk-1/KDR (Figs. 4B 5 ; Table 4 ), immunostaining for both VEGF receptors was often markedly increased on endothelial cells of newly formed vessels in the stroma of inflamed and vascularized corneas compared with limbal vessels of normal control corneas. 

In addition, Flt-1 but not Flk-1/KDR was moderately expressed by macrophages and moderately to strongly by corneal endothelial cells (Figs. 4B 5) . 

VEGF Levels in Normal and Vascularized Human Corneas

VEGF protein was quantified from protein extracts of normal and vascularized corneas using a sensitive ELISA (Table 2) . The mean ± SD total amount of VEGF protein in vascularized corneas (109.4 ± 86.8 pg) was significantly higher than that (10.0 ± 4.3 pg) of normal control corneas of the same size (half corneal buttons, 7 mm in diameter; Table 2 , P < 0.01). Correspondingly, the mean VEGF concentration (±SD) in vascularized corneas (166 ± 98.9 pg/mg protein) was significantly higher than that in normal corneas (12.1 ± 2.7 pg/mg protein; Table 2 , P < 0.001). 

Discussion

In the present study we were able to show that VEGF and its receptors are expressed in increased intensities in inflamed and vascularized human corneas compared with normal corneas, suggesting that this angiogenic cytokine may be involved in the pathogenesis of corneal neovascularization. Furthermore, we found significantly higher levels of VEGF in vascularized corneas compared with normal control corneas. If the mean volume of half corneal buttons (7 mm in diameter), which were used for VEGF ELISA in the present study, is estimated as 12.9 μl (according to an average thickness of 0.67 mm as determined in histologic sections), and the mean amount of total VEGF protein was 109.4 pg/half corneal button, a mean VEGF level of 8.5 ng/ml (187 pM) can be calculated for the vascularized corneas investigated in the present study. This estimated VEGF concentration is well above the level required to promote the growth of vascular endothelial cells, because half-maximal stimulation of endothelial cell growth is obtained at 100 to 150 pg/ml (2–3 pM) VEGF and a maximal effect at concentrations of 1 to 4 ng/ml (22–88 pM). ²⁸  

Our results of increased VEGF levels in vascularized corneas favorably compare with an animal model of cautery-induced corneal angiogenesis, suggesting that increased VEGF concentrations may be required for the induction and maintenance of new vessels in the cornea. ³⁵  

Furthermore, it was recently shown that VEGF is a functional endogenous corneal angiogenic factor and that it may be required for neovascularization in a rat model of wound- and inflammation-related corneal angiogenesis. ⁶ However, in contrast to other tissues, little is known about the regulation of the synthesis and expression of VEGF and its receptors in the human cornea. 

In investigation of the development of retinal vasculature and ischemic retinal diseases it was shown that hypoxia alone may induce increased synthesis and expression of VEGF in various cells (e.g., pigment epithelial cells ³⁶ and several cells of all retinal layers ^{11 12 37 38} ). Furthermore, it has been shown that hypoxia may increase the expression of VEGF and its receptors. ^{39 40 41 42 43 44} In this context it is important to note that tissue hypoxia is a known side effect of contact lens overwear often leading to corneal neovascularization. ^{45 46} Thus, there is emerging evidence that VEGF may play an important role in contact lens–induced corneal neovascularization. However, in chronic inflammatory corneal diseases characterized by infiltration of various densities, particularly of mononuclear inflammatory cells, ⁴⁷ and by expression of numerous cytokines, ^{48 49 50 51 52 53} some other factors may be responsible for the increased synthesis and expression of VEGF and its receptors. In this context it is important to note that it has been shown in cell cultures ^{54 55 56 57} and in skin diseases (e.g., in psoriasis ⁵⁸ ) that transforming growth factor (TGF)-α and -β, bFGF, and platelet-derived growth factor-BB (PDGF-BB) all can induce or augment VEGF synthesis in various cells. In the cornea a variety of cytokines has been detected, including the above-mentioned ones, ^{48 51 52 53} which may be produced in increased amounts by resident tissue-based corneal cells like epithelial cells and keratocytes during inflammation ^{49 50} and may be involved in the regulation of VEGF production in the cornea. However, the most important source of VEGF in chronic inflammatory corneal diseases may be activated macrophages, which are found in high densities particularly in the stroma of inflamed and vascularized corneas. ⁴⁷ Thus, macrophages may play a major role in the pathogenesis of corneal neovascularization by secreting VEGF directly and by producing other angiogenic factors (e.g., bFGF, TGF-β, and PDGF-BB), which may also trigger VEGF production by other cells (e.g., epithelial cells and keratocytes). ⁵⁹ On the other hand, VEGF is chemotactic for macrophages ^{60 61} and has been shown to induce their activation based on expression of procoagulant activity on their surface caused by de novo synthesis of the potent initiator of coagulation, tissue factor. ⁶⁰ These effects are mediated by a specific interaction of VEGF with a single class of binding site, the VEGFR-1/Flt-1 receptor, ^{61 62 63} which was shown to be the only VEGF receptor on cells of the monocyte/macrophage lineage. However, little is known about the regulation of the Flt-1 receptor on these cells. 

The coexpression of VEGF and its receptor Flt-1 on monocytes/macrophages raises the question of an autocrine stimulation or of autoregulation of VEGF via its receptor Flt-1. To the best of our knowledge such a mechanism has not yet been shown for cells of the monocyte/macrophage lineage. However, it has been clearly demonstrated that VEGF itself may upregulate its receptor Flt-1 on human vascular endothelial cells. ⁶⁴ If a similar mechanism existed for monocytes/macrophages the induced enhanced Flt-1 expression might result in increased tissue factor production and procoagulant activity, and in an increased chemotactic response and migration of these cells after stimulation with VEGF. 

In the present study VEGF was also expressed by corneal epithelial and endothelial cells and by vascular endothelial cells of limbal vessels in normal corneas, albeit in lower intensities than in inflamed corneas. A similar finding was also made by van Setten ⁶⁵ who detected VEGF expression in the epithelium of normal corneas and by Bednarz et al. who detected the corresponding gene. ⁶⁶ Nevertheless, the normal cornea remains avascular even though significant levels of VEGF are expressed, particularly in epithelial cells. An explanation for this interesting finding may be that the angiogenic response depends on the balance of production of positive angiogenic regulators and inhibitors of angiogenesis ⁶⁷ and that in normal corneas potent antiangiogenic factor(s) possibly may block the angiogenic effects of VEGF. Such presumed antiangiogenic factor(s) are probably produced by corneal and particularly by limbal epithelial cells, because it has recently been shown in vitro that in contrast to conjunctival epithelial cells, limbal epithelial cells exert antiangiogenic activity based on inhibition of endothelial cell proliferation and morphogenesis. ^{68 69 70} However, the function of VEGF derived from normal corneal epithelium is not yet clear. Because it has previously been shown that VEGF (in subangiogenic concentrations) plays an important role in the physiology of normal vessels ³⁸ and because it has been suggested to be a survival factor responsible for the maintenance of vascular networks, ⁷¹ we primarily speculated that the normal production of VEGF in the cornea might also be involved in the physiology and maintenance of limbal vessels. However, if this were true one would expect a stronger expression of VEGF in the limbal and peripheral corneal epithelia than in the central region. To test this hypothesis we performed additional immunohistochemical studies on normal corneas including the corneoscleral rim areas. However, because VEGF was equally expressed throughout the epithelium of the normal cornea including the center and the periphery, the above-mentioned hypothesis could not be confirmed by these studies. 

Yet there exists another more plausible hypothesis regarding the function of VEGF derived from the corneal epithelium. Although in a healthy normal cornea, potent antiangiogenic factor(s) presumably may neutralize the angiogenic effect of VEGF derived from corneal epithelial cells and may keep the cornea avascular, ^{68 69 70} the observation of increased expression of VEGF on epithelial cells near corneal ulcers in the present study, and the recently published finding of enhanced VEGF staining on epithelial cells adjacent to corneal erosions ⁶⁵ strongly suggests that the ability of corneal epithelial cells to produce VEGF could be of great importance during certain corneal surface diseases and wound healing. Rapid secretion of preformed VEGF and increased new production of this cytokine could induce prompt repair mechanisms by increasing permeability of limbal vessels, attracting monocytes, and inducing angiogenesis. Furthermore, epithelial cells probably play a key role in the pathogenesis of vascularization in corneas with limbal deficiency. Although it has previously been shown that in the normal cornea the stem cell–containing limbal epithelium maintains a barrier between corneal and conjunctival epithelia, ^{32 72} it is not fully understood why conjunctival epithelial ingrowth is closely associated with the development of corneal vascularization in corneas with limbal deficiency. ^{31 73} The results of the present study showing increased VEGF expression by epithelial cells in corneas with conjunctivalization, together with the recently published finding of an antiangiogenic activity of limbocorneal epithelial cells versus conjunctival epithelial cells, ^{68 69 70} suggest that both angiogenic factors (at least in part VEGF) and the loss of antiangiogenic factor(s) due to conjunctival epithelial ingrowth may be involved in vascularization of these corneas. 

However, it must be mentioned that similar to other inflamed and vascularized corneas, VEGF (and probably also other cytokines like bFGF) also derived from keratocytes (fibroblasts) and macrophages, which often strongly express this cytokine particularly in the anterior stroma of such corneas with limbal deficiency, may be involved in the pathogenesis of neovascularization in this entity. 

In the present study we used immunohistochemistry and not in situ hybridization and thus could detect only the protein and not the mRNA of VEGF. Because VEGF was densely expressed in the cytoplasm and not just on the surface of epithelial, corneal, and vascular endothelial cells, it was very evident that VEGF may also have been synthesized by these cells. Furthermore, results from other investigators, who have detected either VEGF or the gene coding this cytokine, indicate that VEGF may be produced by corneal epithelial and endothelial cells, vascular endothelial cells, and keratocytes. ^{65 66} As regards expression of VEGF receptors, we were able to clearly show in the present immunohistochemical study that receptors Flt-1 and Flk-1/KDR are expressed often in increased densities on vascular endothelial cells of newly formed vessels in the stroma of vascularized and inflamed corneas and that only Flt-1 was expressed on macrophages and corneal endothelial cells, whereas Flk-1 expression was lacking. The latter results are in agreement with those of other authors who detected expression of genes coding VEGF and Flt-1 but not Flk-1 in corneal endothelial cells and macrophages. ^{61 66} Although VEGF is not mitogenic for corneal endothelial cells, it is assumed that this cytokine may stimulate via its receptor Flt-1 migration of these cells especially during wound healing. ⁶⁶ In general, it is suggested that angiogenesis is induced and regulated by various cytokines (e.g., by bFGF, aFGF, TGF-β, PDGF-BB, and, particularly, VEGF). ^{8 10} All these cytokines may be involved in corneal neovascularization. However, VEGF is unique because it is the only vasoproliferative factor that can be induced by hypoxia alone and appears to play an irreplaceable role in angiogenesis. ²⁸  

The investigation of corneal angiogenesis is not just of theoretical interest. New therapeutic concepts may be developed (e.g., substances that specifically may interfere with VEGF or its receptors). 

 Supported in part by Allergan Austria and Croma Pharma Austria (printing of color figures).

 Submitted for publication March 18, 1999; revised September 9, 1999 and February 22, 2000; accepted March 15, 2000.

 Commercial relationships policy: N.

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1997.

 

Table 1.

 

Indications for Keratoplasty and Number of Corneas with Limbal Deficiency

Total No. of Corneas (n = 38)	Diagnoses (Indications for Keratoplasty)	No. of Corneas with Limbal Deficiency*
4	Vascularized posttraumatic scars
7	Rejected corneal allografts	1, †
8	Herpetic stromal keratitis
3	Zoster keratitis
7	Chemical burns	3
3	Atopic keratoconjunctivitis
4	Vascularized scars after bacterial ulcers
2	Fungal keratitis

*  Diagnoses histologically verified by the presence of conjunctival goblet cells in the corneal epithelium.

†  Limbal deficiency due to multiple graft failures following keratoplasties in a patient with atopic keratoconjunctivitis.

Table 2.

 

VEGF Levels in Normal and Vascularized Human Corneas

Cornea	Diagnosis	Total VEGF Protein, pg/Corneal Sample*	VEGF Level, pg/mg Extracted Protein
1	Normal cornea	10.8	8.17
2	Normal cornea	15.6	13.1
3	Normal cornea	8.1	13.6
4	Normal cornea	5.5	13.7
1–4	Mean± SD	10.0 ± 4.3, †	12.1 ± 2.7, ‡
5	Atopic keratoconjunctivitis	77.5	97.1
6	Chemical burn	284.5	246.7
7	Chemical burn	100.2	194.9
8	Herpetic stromal keratitis	49.2	85.1
9	Bacterial corneal ulcer	157.5	302.3
10	Herpetic stromal keratitis	30.2	100.0
11	Rejected corneal allograft	12	19.9
12	Rejected corneal allograft	85	158.8
13	Rejected corneal allograft	188.2	289.5
5–13, §	Mean± SD	109.4 ± 86.8, †	166.0 ± 98.9, ‡

*  Half corneal buttons with a diameter of 7 mm.

§  Diseased corneas (5–13) had significant vascularization in at least 3 quadrants.

†   P < 0.01.

‡   P < 0.001.

Table 3.

 

Expression of VEGF in Normal and Vascularized Human Corneas

No. of Corneas	Diagnosis	EPC	KC	EC	VEC	M
15	Normal cornea	++	−/+	+/++	++*	−
4	Vascularized posttraumatic scars	++	+/++	+/++	++	+/++
7	Rejected corneal allografts	++/+++	+/++	++/+++	++/+++	++/+++
8	Herpetic stromal keratitis	++/+++	+/++	++/+++	++/+++	++/+++
3	Zoster keratitis	++/+++	+/++	++/+++	++/+++	++/+++
7	Chemical burns	+++	+/++	++/+++	++/+++	++/+++
3	Atopic keratoconjunctivitis	++/+++	+/++	++	++/+++	++/+++
4	Vascularized scars after bacterial ulcers	++/+++	+/++	++	++/+++	++/+++
2	Fungal keratitis	++/+++	+/++	++	++/+++	++/+++

 EPC, epithelial cells; KC, keratocytes; EC, corneal endothelial cells; VEC, vascular endothelial cells; M, macrophages. Staining intensity: −, negative; +, weak; ++, moderate; +++, intense.

*  Limbal vessels.

Figure 1.

 

Cryostat sections of cornea excised with a scleral rim (C) of normal donor eye after staining with polyclonal Ab, anti-VEGF. VEGF was expressed by epithelial cells (A, C), by corneal endothelial cells (B), weakly by some keratocytes (B, arrows), and by vascular endothelial cells of limbal vessels (C, arrowheads). Scale bars, (A, B) 25 μm; (C) 50 μm.

Table 4.

 

Expression of Flt-1 (VEGFR-1) and Flk-1/KDR (VEGFR-2) in Normal and Vascularized Human Corneas

No. of Corneas	Diagnosis	Flt-1					Flk-1
				EC	VEC	M	VEC
15	Normal cornea	+	++*	−	++*
4	Vascularized posttraumatic scars	++	++	−/++	++
7	Rejected corneal allografts	++/+++	++/+++	++	++/+++
8	Herpetic stromal keratitis	++/+++	++/+++	++	++/+++
3	Zoster keratitis	++/+++	++/+++	++	++/+++
7	Chemical burns	++/+++	++/+++	++	++/+++
3	Atopic keratoconjunctivitis	++	++/+++	++	++/+++
4	Vascularized scars after bacterial ulcers	++/+++	++/+++	++	++/+++
2	Fungal keratitis	++	++/+++	++	++/+++

 KDR, kinase insert domain-containing receptor; EC, corneal endothelial cells; VEC, vascular endothelial cells; M, macrophages. Staining intensity: −, negative; +, weak; ++, moderate; +++, intense.

*  Limbal vessels.

Figure 2.

 

Cryostat sections of cornea of 45-year-old patient with chronic herpetic stromal keratitis after staining with anti-VEGF Ab. VEGF was strongly expressed by epithelial cells (A), by vascular endothelial cells of newly formed vessels in the stroma (A, B), and moderately by keratocytes (A, B). Scale bars, (A) 50 μm; (B) 25μ m.

Figure 3.

 

Cryostat section of cornea of 64-year-old patient 2 months after reepithelialization of herpetic corneal ulcer after staining with anti-VEGF Ab. Note increased expression of VEGF by epithelial cells particularly in the vicinity of mononuclear inflammatory cells, which also express this cytokine. Mononuclear cell infiltrates predominantly consist of macrophages as was determined by staining with EBM 11 mAb (not shown). Scale bar, 25 μm.

Figure 4.

 

Cryostat sections (taken less than 15 μm apart) of cornea of 36-year-old patient 11 months after the onset of allograft rejection after staining with polyclonal Abs anti-VEGF (A) and anti–Flt-1 (B). VEGF is strongly expressed by epithelial cells, vascular endothelial cells, and corneal endothelial cells (A). Flt-1 was strongly expressed by endothelial cells of the same vessels in the stroma (arrows) and by corneal endothelial cells (B). Scale bar, 100 μm.

Figure 5.

 

Cryostat section of cornea of 53-year-old patient 12 months after alkali burn after staining with anti–Flk-1 Ab. Flk-1 was strongly expressed by vascular endothelial cells in the stroma but not by corneal endothelial cells and inflammatory cells. Note dense inflammatory infiltrates consisting predominantly of macrophages as was determined by staining with EBM 11 mAb (not shown). Scale bar, 50 μm.

The authors thank Erika Bachmann for her technical assistance.