Corneal neovascularization, the growth of new blood vessels in the cornea, is a leading cause of vision impairment after corneal injury. Neovascularization typically occurs in response to corneal injury such as that caused by infection, physical trauma, chemical burns or in the setting of corneal transplant rejection. The NADPH oxidase enzyme complex is involved in cell signalling for wound-healing angiogenesis, but its role in corneal neovascularization has not been studied. We have now analysed the role of the Nox2 isoform of NADPH oxidase in corneal neovascularization in mice following chemical injury. C57BL/6 mice aged 8–14 weeks were cauterized with an applicator coated with 75% silver nitrate and 25% potassium nitrate for 8 s. Neovascularization extending radially from limbal vessels was observed in corneal whole-mounts from cauterized wild type mice and CD31+ vessels were identified in cauterized corneal sections at day 7. In contrast, in Nox2 knockout (Nox2 KO) mice vascular endothelial growth factor-A (Vegf-A), Flt1 mRNA expression, and the extent of corneal neovascularization were all markedly reduced compared with their wild type controls. The accumulation of Iba-1+ microglia and macrophages in the cornea was significantly less in Nox2 KO than in wild type mice. In conclusion, we have demonstrated that Nox2 is implicated in the inflammatory and neovascular response to corneal chemical injury in mice and clearly VEGF is a mediator of this effect. This work raises the possibility that therapies targeting Nox2 may have potential for suppressing corneal neovascularization and inflammation in humans.
Corneal neovascularization, which frequently occurs following injury and infection, is a major cause of vision loss and blindness. Current therapeutic approaches to the prevention of corneal neovascularization in those at risk are minimally effective and carry risks of significant ocular side effects.
Nox2 is an important source of ROS in the vascular system and it plays a role in new blood vessel formation.
Our study identifies Nox2 as a mediator of neovascularization following chemical cauterization of the mouse cornea. Nox2 acts, at least in part, via the attenuation of inflammatory response and regulation of Vegf-A gene expression. Therapies that target Nox2 may be of value in the management of corneal neovascularization.
Injury or disease often triggers the development of new blood vessels in the cornea, compromising its clarity and impairing vision. Common causes of corneal neovascularization include physical trauma, chemical burns, hypoxia from extended use of contact lenses, bacterial and viral infections, ulceration, corneal graft rejection and local and systemic inflammatory disorders . It has been estimated that blindness associated with corneal neovascularization and scarring affects 4.9 million people, making this the second most common cause of blindness after cataract . Although neovascularization may be an adaptive response to injury in some contexts, the leakage of fluid, proteins and lipids from newly formed vessels in the ordinarily avascular cornea impairs its clarity . Anti-inflammatory steroids are the mainstay of treatment for corneal neovascularization but these have limited therapeutic efficacy and prolonged use of these treatments increases the risk of infection as well as cataract and glaucoma . Thus there is a pressing need for more effective inhibitors of corneal neovascularization.
Equilibrium between endogenous pro-angiogenic and anti-angiogenic factors maintains corneal avascularity . After corneal injury, the level and activity of pro-angiogenic factors exceeds those of anti-angiogenic factors, favouring angiogenesis . Accordingly growth factors, such as vascular endothelial growth factor (VEGF), have been shown to be elevated in experimental models of corneal neovascularization . Although anti-VEGF agents, such as bevacizumab, are now widely used to suppress retinal neovascularization in patients with diabetic retinopathy and neovascular age-related macular degeneration , these agents have only recently been tested in experimental models of corneal neovascularization [6,7] and in numerous small clinical trials in humans with corneal neovascularization [3,8,9]. Anti-VEGF therapies appear to be moderately effective in reducing corneal neovascularization in some contexts. Additional therapeutic approaches are required to prevent vision loss from corneal neovascularization.
Chemical injury of the rodent cornea with silver nitrate or sodium hydroxide is a commonly used model of wound-healing angiogenesis in the cornea [7,10–12]. Work with this model has demonstrated that topical antioxidants suppress the growth of blood vessels in rats , implicating reactive oxygen species (ROS) in corneal neovascularization. NADPH oxidase is an important source of ROS that mediates redox signalling in the wound-healing response . Several isoforms of the Nox subunit have been identified and the prototypical Nox is composed of membrane-bound Nox2 and p22phox subunits, as well as the cytosolic subunits p40phox, p47phox, p67phox and the small GTPase Rac (Supplementary Figure S1). There is evidence that VEGF can act to increase ROS generation by activating the catalytic domain of Nox2 . Furthermore, it is known that Nox2 can also trigger VEGF signalling to regulate angiogenesis and vascular permeability via up-regulation of VEGF expression [16–18] as well as activation of its receptor, KDR/Flk-1 (VEGFR2) . Thus there is a complex interplay between ROS and VEGF signalling in angiogenesis, and the roles played by different Nox isoforms in corneal angiogenesis are not yet defined. Nox2 is expressed in inflammatory and endothelial cells and both cell types are central to corneal angiogenesis. In this work we have used a mouse model of chemical cauterization to study the role of Nox2 in corneal neovascularization.
MATERIAL AND METHODS
Corneal model of neovascularization
All surgical procedures that were performed on mice were approved by the institutional animal care and use committee (St Vincent's Animal Ethics Committee protocol no. 003/14) to ensure that the animals did not suffer unduly during and after the experimental procedure. Wild type C57BL/6J mice and Nox2 knockout mice (Nox2 KO) on a C57BL/6J background were supplied by Professor Mary Dinauer (Department of Paediatrics, Washington University of School of Medicine in St Louis, MO, USA)  and bred at the Experimental Medical and Surgical Research Unit (EMSU) mouse facility (Fitzroy, Victoria, Australia). Mice aged between 8 and 14 weeks were used. Nox2 genotype was confirmed with conventional genotyping (Supplementary Figure S2). Animals were anaesthetized with intraperitoneal injection of a combination of ketamine (100 mg/kg) and xylazine (10 mg/kg) supplemented with a topical corneal anaesthetic agent (oxybuprocaine hydrochloride 0.4%). Corneal neovascularization was induced by chemical cauterization by a trained operator masked to the genotype of mice. The tip of an applicator containing silver nitrate/potassium nitrate (25%/75%; Grafco) was applied to the centre of the cornea for 8 s. Excess nitrate was rinsed off with 5 ml of saline solution (0.9% sodium chloride) immediately after cauterization. The fellow eye was not treated and used as a control to the cauterized eye. Mice were killed and eyes were harvested on days 1, 2 and 7 post-cauterization.
Assessment of burn stimulus response and neovascularization
Seven days after cauterization, the burn stimulus response and extent of neovascularization were assessed in anaesthetized mice by an ophthalmologist masked to mouse genotype under a dissecting microscope (Zeiss), using a modified grading system . The burn stimulus response was graded according to the size of the corneal blister with a score range from 0 to 3 (0=no blister; 1=blister raised slightly above corneal surface; 2=blister raised moderately above corneal surface; 3=large blister) . Vascularization was scored from 0 to 6 (0=no visible vessels; 1=1/4 distance to edge of burn; 2=1/3 distance to edge of burn; 3=1/2 distance to edge of burn; 4=2/3 distance to edge of burn; 5=3/4 distance to edge of burn; 6=vessels reach edge of burn) in each quadrant of the cornea according to the extent of vessel growth from the limbus to the burn edge . A score of neovascularization was determined by averaging the mode of scores from four quadrants.
Eyes were fixed in 4% paraformaldehyde for 1 h and corneal whole-mounts were prepared . Corneas were incubated in acetone for 20 min at 4°C, followed by incubation in proteinase K (Dako) for 5 min at room temperature. Whole-mounts were blocked with total protein block solution (Dako) for 30 min at room temperature and then stained with rat anti-mouse CD31 for endothelial cells (1:150, MEC13.3/#553370, BD Biosciences Pharmingen) overnight at 4°C. Whole-mounts were then incubated with biotinylated rabbit anti-rat IgG (1:200 dilutions, Dako) for 1 h, followed by HRP-streptavidin conjugated with Alexafluor 488 (1:400 dilutions, Dako) for 1 h, or incubated with goat anti-rabbit antibody conjugated with Cy3 (1:300 dilutions, Vector Laboratories) for 1 h. Whole-mounts were washed in phosphate-buffered saline (PBS) and mounted in fluorescent mounting medium (Dako). Images of each whole-mount were captured with a fluorescence microscope (Zeiss) and a confocal microscope (Nikon A1, Victoria, Australia). The degree of neovascularization (CD31+ area) was then quantified with ImageJ  and expressed as a percentage of the corneal area. To identify microglia and macrophages, acetone-fixed and total protein block treated whole-mounts were stained with an antibody with specificity for Iba-1 (1:500 dilutions, ionized calcium binding adaptor molecule 1) expressed by macrophages and microglia (rabbit anti-mouse polyclonal Iba-1 antibody, catalogue no. 019-19741, Wako Pure Chemical Industry) overnight at 4°C. Sections were then incubated with a secondary antibody [goat anti-rabbit Cy3 antibody (1:300 dilutions, Vector Laboratories)] for 1 h, and then counterstained with 4',6-diamidino-2-phenylindole (DAPI) (Dako; 1:1000 dilutions) and mounted in fluorescent mounting medium (Dako). Four images (frame size of 680 μm x 510 μm) per whole-mount were taken randomly at 20× at regions between the burn edge and corneal limbus (Olympus microscope). Iba-1 positive (Iba-1+) cells were counted in the four images and the number of cells per field was averaged.
Histopathology of corneal sections
Paraffin embedded sections (4 μm) were stained with haematoxylin and eosin (H&E) to examine tissue morphology. Central corneal thickness was measured in three corneal sections from each eye using ImageJ.
Immunohistochemistry was used to identify blood vessels in corneal sections as follows. Endogenous peroxidase activity of sections was quenched with 3% H2O2 for 5 min prior to incubation in 0.1% proteinase K (pH 7.8, Dako) for 3–6 min. Corneal sections were blocked with total protein block solution (Dako) for 30 min, and incubated with rat anti-mouse CD31 (1:150 dilutions, MEC13.3/#553370, BD Biosciences Pharmingen) for 1 h. Sections were incubated with biotinylated rabbit anti-rat IgG for 30 min, followed by avidin–biotinylated–peroxidase complex (Vectastain Elite ABC kit, Vector Laboratories). Positive tissues were then detected using a 3,3' diaminobenzidine (DAB) chromogen (Dako), counterstained with haematoxylin and mounted in DPX (VWR International). Rat IgG (Dako) was used as a negative control (Supplementary Figure S3).
The area of neovascularization in corneal cross-sections was measured on CD31 labelled sections (haematoxylin counterstained) using video microscopy with a computer-assisted stereo investigator system (MBF Bioscience) . For each cornea, three sections, 200 μm apart were counted with a 20× magnification objective. Using systematic random sampling, 12-point grids (200 μm x 200 μm) were superimposed on randomly selected fields representing 25% of the total corneal cross-sectional area and each point within the grid was recorded as positive or negative on the basis of CD31 staining. The percentage of points falling on CD31+ area in a mouse corneal section was used to calculate the vascularized area. All counting was completed by a trained operator masked to the identity of the sections.
Gene expression detected by real time-PCR
Total corneal RNA was purified using commercial kits in accordance with the manufacturer's instructions (RNeasy Mini Kit, Qiagen, Victoria, Australia). Briefly, a single cornea was lysed and homogenized in the lysate and RNA was purified using a column system. RNA was reverse-transcribed to cDNA (100 ng) using a high capacity cDNA reverse transcription kit (catalogue no. 4374996, Life Technologies). Real-time PCR reactions were performed (7300 real-time PCR systems, Life Technologies) using a TaqMan Universal PCR master mix and commercially available probe and primer sets (TaqMan Gene Expression Assay, Life Technologies) for mouse Nox2 (Mm00432775_m1), Nox4 (Mm00479246_m1), Vegf-A (Mm00437304_m1), Flt1 (Mm00438980_m1), Kdr/Flk-1 (Mm01222421_m1), transforming growth factor-beta 1 (Tgf-β1) (Mm01227699_m1), matrix metalloproteinase 2 (Mmp2) (Mm00439498_m1), Mmp9 (Mm00442991_m1), interleukin-1 beta (IL-1β) (Mm00434228_m1), thrombospondin 1 (Thbs1) (Mm00449032_g1), thrombospondin 2 (Thbs2) (Mm01279240_m1), Timp3 (Mm00441826_m1) and a disintegrin and metalloproteinase with thrombospondin motifs 1 (Adamts1) (Mm00724059_m1). Mouse GAPDH (4352339E) was used as a reference gene. Gene expression changes in the corneas of the cauterized eyes were normalized using values from untreated fellow eyes.
Western blot analysis
Protein extracts were isolated from mouse corneas using lysis buffer containing 150 mM NaCl, 50 mM HEPES pH 7, 1% Triton X-100, 10% glycerol, 1.5 mM MgCl2, 1 mM EGTA and protease inhibitors (Roche Applied Science). After separation in SDS/12.5% PAGE, proteins were transferred on to polyvinylidene fluoride membrane (PVDF) using a blotting apparatus. The membrane was blocked with 5% skim milk in tris-buffered saline-tween 20 (TBS-T) for 1 h at room temperature, then incubated with antibodies for Vegf-A (147; rabbit polyclonal, 1:500 dilutions; Santa Cruz Biotechnology) or Flt-1 (C-17; rabbit polyclonal, 1:500 dilutions; Santa Cruz Biotechnology) overnight at 4°C or β-actin (mouse monoclonal, 1:2000 dilutions; Millipore, MA, USA) for 30 min at room temperature. After incubation with secondary antibodies conjugated with horseradish peroxidase (1:5000 dilutions in 5% skim milk) for 30 min, the signals on membrane were detected using ECL-plus luminol solution (Pharmacia) and exposed to X-ray film for autoradiography. ImageJ software was used for scanning and quantitative analysis.
Data are expressed as mean ± S.E.M. Mean data were analysed with unpaired t tests or two-way analysis of variance (ANOVA) followed by post-hoc Tukey analysis (GraphPad Prism 6.0). A value of P<0.05 was regarded as statistically significant.
Chemical cautery burn response and neovascularization are reduced in Nox2 KO mice
To investigate the role of Nox2 in corneal neovascularization, we first compared the extent of neovascularization and the burn response in cauterized corneas between Nox2 KO and wild type mice on day 7. Wild type mice had a burn score of 2 (blister raised moderately above corneal surface; n=16) and a neovascular score of 5 (3/4 distance to edge of burn; n=16). Both scores were significantly reduced in Nox2 KO mice: burn score of 1 (blister raised slightly above corneal surface; n=14) and neovascular score of 3 (1/2 distance to edge of burn; n=14). Digital photographs captured under a stereomicroscope illustrate the radial growth of blood vessels from the corneal limbus towards the central burn area (Figure 1). Corneal whole-mounts were further stained with an antibody against CD31, an endothelial cell marker, to identify blood vessels (Figure 2). At baseline corneal vascularization was comparable in both groups of mice (Nox2 KO: 12.9±2.5%, n=8; wild type: 14.7±2.3%, n=5; Figure 2G). Seven days after cauterization, the extent of corneal neovascularization in Nox2 KO mice was less than in wild type mice (33.7±12% of the corneal surface area, n=8 compared with 58.6±4.6% of the corneal surface area, n=5; P<0.05; Figure 2G). The difference in the extent of neovascularization was even more pronounced 14 days after corneal cautery (Nox2 KO: 29.7±6.5% of the corneal surface area, n=5 compared with wild type: 74.6±2.3% of the corneal surface area, n=3; P<0.01; Figure 2H).
Burn stimulus and neovascular response from Nox2 KO and wild type mice at day 7 following chemical cauterization
Corneal whole-mounts from Nox2 KO and wild type mice following chemical cauterization
Histological analysis of corneal sections demonstrated that the corneal thickness at the cautery site was less in Nox2 KO mice than in wild type mice (Nox2 KO: 115±6 μm, n=5 compared with wild type: 164±12 μm, n=4; P<0.05; Figure 3A) due to more extensive fibrovascular proliferation in wild type than in Nox2 KO mice (Figure 3B).
Histological assessment of corneal sections with H&E staining
To further quantify the extent of neovascularization following chemical cautery, corneal cross-sectional vascular area was measured by CD31 immunohistochemistry. Less neovascularization was observed in the corneas of Nox2 KO mice than in wild type mice (Nox2 KO: 9.6±0.8%, n=4 compared with wild type: 16.9±2.1%, n=4; P<0.05; Figure 4A). As expected, neither group of mice exhibited corneal neovascularization in untreated control eyes (Figures 4B and 4C). The cross-sectional area of corneal neovascularization was similar in both groups for the peripheral corneas (Figures 4D and 4E), but was significantly lower in Nox2 KO mice for the central corneas (Figures 4F and 4G). This reflects the reduced radial extent of new vessel growth in Nox2 KO mice.
Immunohistochemical detection of neovessels in corneal section at day 7
Attenuated inflammatory response in corneas of Nox2 KO mice
In addition to endothelial cells, inflammatory cells such as macrophages are a source of Nox2  and chemical cautery induces the accumulation of inflammatory cells, including macrophages and neutrophils, in corneas [25,26]. We therefore measured the expression of pro-inflammatory cytokine, IL-1β by real-time-PCR and evaluated the accumulation of macrophages and microglia in corneal whole-mounts using an Iba-1 antibody  at days 1, 2 and 7. We found an increase in IL-1β mRNA in cauterized corneas from wild type mice at each time point when compared with untreated control corneas (408-fold at day 1, n=5; 216-fold at day 2, n=5; 32-fold at day 7, n=6; Figure 5A). The up-regulation of IL-1β mRNA in cauterized corneas was significantly reduced in Nox2-deficient mice at days 1 and 2 in comparison with wild type mice (Nox2 KO: 290-fold compared with wild type: 408-fold at day 1, n=4-5; Nox2 KO: 95-fold compared with wild type: 216-fold at day 2, n=4–5; P<0.05 for each a comparison; Figure 5A). Few Iba-1+ cells were observed in the untreated control (not cauterized) corneas of wild type and Nox2 KO mice. A significant increase in the number of Iba-1+ cells in the corneas of wild type mice was observed at each time point following cautery relative to untreated control corneas. In contrast, fewer Iba-1+ cells were seen in the cauterized corneas of Nox2 KO mice at each time point (Figures 5B and 5C) than those of wild types. Thus Nox2 may act as a mediator in the inflammatory response in cauterized cornea.
Effect of Nox2 KO in the inflammatory response following chemical cauterization
Nox2 regulates Vegf-A and its receptor (Flt1) in cornea following chemical cautery
We and others have shown that Nox2 and other isoforms, including Nox1 and Nox4, are involved in the pathological growth of blood vessels in the hypoxic retina in association with increased expression of Vegf-A [16,28,29], a key regulator of angiogenesis . We therefore measured the gene expression of Nox2 and other Nox isoforms in corneas following chemical cautery at days 1, 2 and 7 by real-time-PCR. We found an increase in Nox2 mRNA in cauterized corneas from wild type mice at each time point when compared with untreated control corneas (5.6-fold at day 1, n=5; 13.9-fold at day 2, n=5; 16.9-fold at day 7, n=6–9; P<0.05; Figure 6A). This was accompanied by a significant reduction in Nox4 mRNA expression (Figure 6B). Nox1 and Nox3 expression could not be detected in cauterized corneas from wild type mice.
Gene expression of Nox2 and Nox4 in cornea with chemical cauterization
To explore the association between Nox2 and angiogenic factors and inhibitors that are known to be involved in wound-healing response in cauterized cornea [30,31], we measured the gene expression of (1) angiogenic factors including: Vegf-A, Vegf receptors (Flt1 and Kdr/Flk-1), Tgf-β1, Mmp2 and Mmp9; and (2) inhibitors of angiogenesis including: Thbs1, Thbs2, metalloproteinase inhibitor 3 (Timp3) and a Adamts1 in untreated and cauterized corneas from wild type and Nox2 KO mice at days 1, 2 and 7. Gene expression of Vegf-A, Mmp9 and Thbs1 were increased in the cauterized corneas of wild type mice at day 1 and this was sustained to day 7 (Figures 7A, 7F and 7G). The gene expression of Flt1, Kdr/Flk-1, Mmp2, Thbs1 and Adamts1 but not Tgf-β1 was increased in the cauterized corneas of wild type mice at day 7 (Figures 7B, 7C, 7E, 7H and 7J). Moreover, the gene expression of Timp3 was significantly reduced over the 7 days (Figure 7I). In contrast, the up-regulation of Vegf-A gene expression at days 2 and 7 in cauterized corneas of Nox2-deficient mice was significantly less than that observed in wild type mice (Nox2 KO: 1.2-fold compared with wild type: 2.1-fold at day 2, n=5–7; Nox2 KO: 2.5-fold compared with wild type: 3.3-fold at day 7, n=5–6; P<0.05 for each a comparison; Figure 8A). Moreover, the up-regulation of Flt1 gene expression in cauterized corneas of Nox2-deficient mice was reduced at day 7 in comparison with wild type mice (Nox2 KO: 5.5-fold compared with wild type: 11.6-fold at day 7, n=5–6; P<0.05 for each a comparison; Figure 8B). In contrast, the changes in gene expression for Nox4, Kdr/Flk-1, Mmp2, Mmp9, Thbs1, Thbs2, Timp3 and Adamts1 in the cauterized corneas were similar in wild type in comparison with Nox2-deficient mice (Supplementary Figure S4).
Gene expression of angiogenesis markers and inhibitors in cornea with chemical cauterization
Gene and protein expression of Vegf-A and Flt1 from Nox2 KO and wild type mouse cornea at days 2 and 7 following chemical cauterization
We also quantified the expression of Vegf-A and Flt1 protein by Western blot analysis in untreated and cauterized corneas from wild type and Nox2 KO mice at day 7 (Figures 8C and 8D). These proteins were up-regulated in the cauterized corneas of wild type mice but not in Nox2 KO mice, in keeping with the gene expression findings.
Corneal neovascularization is a significant cause of vision loss and blindness following injury. We have now identified a role for Nox2 in the pathological growth of vessels in the cornea following chemical cauterization. In the present study, we have demonstrated that neovascularization in the cornea is associated with an increase in gene expression of Nox2, Vegf-A and its receptor Flt1 as well as an accumulation of Iba-1+ macrophages and microglia in wild type mice. In Nox2-deficient mice, the response to the burn, extent of neovascularization, accumulation of Iba-1+ cells and expression of Vegf-A and Flt1 were all significantly attenuated relative to wild type mice. Inhibitors of the Nox2 signalling pathway could therefore represent a means of suppressing neovascularization in corneal wound healing.
The NADPH oxidase family of ROS-generating enzymes is the key source of ROS in the vascular system and these enzyme complexes are crucial players in both physiological and pathological angiogenesis [32,33]. We have previously demonstrated that NADPH oxidase facilitates angiogenic responses in wound healing in a variety of tissues in vivo [34,35]. Studies of the roles of different isoforms of NADPH oxidase with genetically modified animals and in vivo siRNA approaches have also highlighted the involvement of Nox in VEGF-dependent neovascularization of hypoxic retinas in animals with oxygen-induced retinopathy [16,28,36]. We now highlight a role for the Nox2 isoform of NADPH oxidase in corneal neovascularization. Although recent studies have demonstrate a role of Nox1 and Nox4 in neovascularization in other parts of the eye, and that Nox5 is expressed by fibroblasts in the human cornea [28,36], here we show that cauterization caused an increase in Nox2 expression but a decrease in Nox4 gene expression and we were unable to detect Nox1 in normal or cauterized corneas.
Corneal cauterization with silver nitrate and potassium nitrate in rats and mice activates tissue repair responses such as inflammation, tissue remodelling and angiogenesis [5,10]. Chemical cauterization instigates an inflammatory insult that leads to a burst of inflammatory cytokines as reflected by a remarkably high mRNA expression of IL-1β and Vegf-A in the corneas of wild type mice at days 1 and 2. This response is also associated with the accumulation of Iba-1+ macrophages and microglia which are important cellular sources of Vegf-A. Our finding is comparable to other experimental models of chemical cauterization that also reveal an infiltration of macrophages within 96 h of chemical injury . The inflammatory response is then followed by angiogenesis by which blood vessels grow radially from the limbal vascular plexus from day 2, which tends to peak at around day 7 [37,38]. Chemical cautery-induced fibrovascular proliferation is reduced in Nox2-deficient mice as is indicated by lower burn severity scores at day 7, as well as reduced corneal neovascularization at days 7 and 14. Although it might be possible that the reductions in corneal neovascularization and Vegf-A observed in Nox2 KO mice were consequences of the less severe corneal cautery in these than in wild type mice, this is unlikely for several reasons: (1) corneal cautery was performed in accordance with a standardized protocol by a single experienced operator masked to the genotype of each mouse; (2) increased Nox2 expression and the generation of ROS are associated with the inflammatory response after tissue injury and regeneration [39–41].
We have now demonstrated that the accumulation of Iba-1+ macrophages and microglia in the cornea with chemical cautery of Nox2 KO mice is significantly lower than that seen in wild type mice at each time point. Nox2 is known to mediate the chemotaxis of macrophages in response to a tissue chemoattractant such as colony stimulating factor (CSF) and Nox2 has been shown to stimulate the CSF-dependent phosphorylation of second messenger protein Erk in these macrophages . Moreover, cerebral microglia cultured from Nox2 KO mice demonstrate a reduction in migration responses to Vegf-A when compared with wild type cells . Expression of Vegf-A and tissue chemoattractants such as CSF  are known to be up-regulated in the injured cornea. CSF-dependent generation of ROS has also been shown to stimulate the proliferation of macrophages . Therefore Nox2 might be involved in the proliferation and migratory process of macrophages in the cauterized corneas.
The increase in the extent of neovascularization observed in the cauterized corneas of wild type mice was associated with elevations in Vegf-A and its receptors Flt1 and Kdr/Flk-1 at day 7. Our findings align with previous studies that show that blocking the activity of Vegf-A with either Vegf-A receptor inhibitors or Vegf neutralizing antibodies suppresses neovascularization following corneal cautery [37,46,47]. In the present study we further demonstrated that Nox2 may play a role in Vegf mediated angiogenic responses following corneal cautery as we found reductions in neovascularization as well as gene and protein expression of Vegf-A and Flt1. NADPH oxidase has been shown to mediate processes that are important to pathological angiogenesis including proliferation and migration of endothelial cells . Inhibiting the activity or expression of Nox2 has indeed been demonstrated to attenuate the proliferation and migration of endothelial cell under the stimulation of Vegf [15,49]. Nox2 has also been found to be involved in the switching of endothelial cell phenotype from quiescent to pro-inflammatory in association with Vegf-A and angiopoietin-2 . Angiopoietin-2 is known to be involved in inflammation-driven neovascularization in corneas and promotes vessel sprouting in the presence of Vegf . Thus activated endothelial cells in the cauterized corneas may also contribute to the inflammatory response. Although we observed a decrease in Iba-1+ macrophages and microglia in the Nox2 KO cauterized corneas, a pro-inflammatory role of endothelial Nox2 remains to be confirmed. Nox2 appears to play multiple roles in Vegf-mediated angiogenesis. Nox2-derived ROS has been shown to cause an inactivation of the negative modulators of Vegf receptors Kdr/Flk-1 . Moreover, Vegf-A has been shown to stimulate the production of ROS from Nox2 through Flt1 and Kdr/Flk-1 in endothelial cells . Therefore Nox2 promotes a feed-forward mechanism to amplify the angiogenic responses to Vegf-A .
In conclusion, this work demonstrates that Nox2 plays an important role in corneal neovascularization following chemical cautery and its effect is mediated, at least in part, by a pro-inflammatory function of Nox2 and via up-regulation of Vegf-A (Figure 9). Therapies that target Nox2 may be of value in the prevention and treatment of corneal neovascularization.
Schematic diagram of Nox2 involvement in neovascularization following chemical cauterization of the cornea
Elsa Chan and Guei-Sheung Liu designed/performed experiments and wrote the manuscript; Peter van Wijngaarden provided technical advice, assisted with experimental design, data interpretation and wrote the manuscript; Elsie Chan, Darleen Ngo and Jiang-Hui Wang assisted with experiments; Hitesh M. Peshavariya and Gregory J. Dusting appraised and revised the manuscript; Gregory J. Dusting provided financial support. All authors approved the submission of the paper.
We thank Ms Cheryl Rangiah for her technical support. The Centre for Eye Research Australia receives Operational Infrastructure Support from the Victorian State Government.
This work was supported by the National Health and Medical Research Council of Australia [grant numbers #1061912, PF 11M 6093 (to P.v.W.) and #1003113 (G.J.D.)]; the Ophthalmic Research Institute of Australia; and the Angior Family Foundation.
1Elsa C. Chan and Peter van Wijngaarden contributed to this work equally as first authors.
2Guei-Sheung Liu and Gregory J. Dusting contributed to this work equally and should therefore be regarded as equal senior authors.