Keratocytes Create Stromal Spaces to Promote Corneal Neovascularization Via MMP13 Expression

Abstract

Purpose.: To investigate the exact mechanism by which keratocytes promote corneal neovascularization.

Methods.: The expression of matrix metalloproteinase 13 (MMP13), cluster of differentiation 146 (CD146), VEGFa, VEGFc, VEGF receptor (r)2, and VEGFr3 by normal and alkali-burned rat corneas was determined via quantitative (q)RT-PCR and/or Western blot analysis or in situ hybridization. Corneal neovascularization was observed under a slit lamp microscope and evaluated via immunohistochemistry. The cells that expressed MMP13 in the corneas were determined via sequential immunohistochemistry and in situ hybridization. The degradation of type I collagen was evaluated via the detection of hydroxyproline content and Western blot analysis. The effects of VEGFa and VEGFc on MMP13 expression were determined via luciferase reporter assay for the MMP13 promoter and primary keratocyte culture.

Results.: Matrix metalloproteinase 13 was predominantly expressed by epithelial cells in normal rat corneas, but it was expressed by cells in corneal stromas after alkali burns. The formation of new blood vessels was consistent with MMP13 expression and attenuated by a selective MMP13 inhibitor in alkali-burned corneas. Keratocytes were the major cells expressing MMP13 in corneal stromas after alkali burns. Through MMP13 expression, keratocytes directly degraded collagen type I to create stromal spaces, which were convenient for newly formed blood vessels to grow into. Expression of MMP13 and collagen type I degradation via keratocytes were induced by VEGFc through VEGFr3 and inhibited by antibodies for VEGFc and VEGFr3.

Conclusions.: Keratocytes could directly degrade type I collagen and create stromal spaces, promoting corneal neovascularization through VEGFc/VEGFr3-induced MMP13 expression.

Introduction

Corneal neovascularization (CNV), a major cause of loss of vision and blindness in a wide range of corneal diseases worldwide, is an incompletely understood process that is also called angiogenesis. It is characterized by an invasion of newly formed blood vessels into the normally avascular stroma from the limbal vascular plexus. For sprouting angiogenesis to occur, the vascular endothelial cells (VECs) must degrade and remodel the extracellular matrix (ECM) by producing ECM-degrading enzymes, especially metalloproteinases (MMPs).^1–3 Matrix metalloproteinases comprise a family of structurally related, zinc-dependent endopeptidases and are able to degrade ECM components such as collagens and proteoglycans.^4,5 It has been reported that some MMPs, including metalloproteinase 2, 3, 9, and 14, are involved in CNV.^6,7 

Matrix metalloproteinase 13, also known as collagenase-3, is a member of the MMP family and was initially discovered in breast carcinomas.⁸ It can cleave intact fibrillar type I, II, and III collagens, which are resistant to most proteases, at a specific site to produce fragments susceptible to hydrolysis by other proteases. In addition to its degrading action on fibrillar collagens, MMP13 is also able to degrade a wide range of other ECM components.^9–13 Therefore, MMP13 is one of the main MMPs responsible for the ECM degradation and remodeling associated with angiogenesis.^4,9 However, as one of collagenolytic MMPs, the contributions of MMP13 to CNV have not been reported. 

Keratocytes, also known as corneal fibroblasts or stromal cells, are normally quiescent and become activated after corneal injury.^14–16 The activated keratocytes, in addition to being involved in extracellular matrix production and the regulation of inflammation, can promote the functional properties of VECs through the paracrine mechanisms of metalloproteinases and other angiogenic factors in corneal neovascularization.⁶ However, the exact function of keratocytes themselves regarding the degradation and remodeling of the extracellular matrix associated with CNV has not yet been revealed, and whether keratocytes can express MMP13 also remains unknown. 

The corneal alkali burn model has been widely used to study the mechanism of and therapies for CNV due to the easy local administration of medicine, its accessible position for observation, and the increased release of MMPs after alkali burn.^17,18 In the present study, we applied this model to investigate the function of keratocytes in the degradation and remodeling of the extracellular matrix associated with CNV. We found that keratocytes were the major MMP13-expressing cells and could directly degrade type I collagen to create stromal spaces into which newly formed blood vessels grew due to MMP13 expression in alkali-burned corneas. Moreover, we revealed that MMP13 expression in keratocytes was induced by VEGFc via VEGFr3. 

Materials and Methods

Antibodies and Other Reagents

Mouse anti-rat CD68 (ED1) antibody was obtained from AbD Serotec (Kidlington, Oxford, UK); Rabbit anti-VEGFc, VEGFa, and VEGFr2 were obtained from Abcam (Cambridge, England); VEGFr3 was obtained from Santa Cruz Biotechnology (Dallas, TX, USA); Rabbit anti-CD146 was obtained from EMD Millipore (Shanghai, China); HRP-conjugated mouse anti-rabbit β-actin, HRP-conjugated goat anti-rabbit immunoglobulins, HRP-conjugated rabbit anti-goat IgG, and fast red substrate tablets were obtained from Sigma-Aldrich Corp. (St. Louis, MO, USA); VEGFa and recombinant human MMP13 were obtained from R&D Systems (Minneapolis, MN, USA); the Liquid DAB+ substrate chromogen system was obtained from Dako (Glostrup, Denmark); recombinant rat VEGFc was obtained from ProSpec (Shenzhen, China); MMP13 selective inhibitor 444283 (pyrimidine-4, 6-dicarboxylic acid, bis-(4-fluoro-3-methyl-benzylamide) was obtained from EMD Millipore. Rabbit anti-rat polymorphonuclear neutrophil (PMN) was obtained from Cantor Fitzgerald (Hong Kong, China). Rabbit anti-Col1α2, ALDH3A1 (aldehyde dehydrogenase 3 family, member A1), and MMP13 were obtained from Proteintech Group, Inc. (Wuhan, China). Transfection reagent (Lipofectamine 2000) was obtained from Invitrogen (Guangzhou, China). A double-luciferase reporter assay kit was obtained from Transgen Biotech (Beijing, China). An MMP13 assay kit was obtained from AnaSpec, Inc. (Fremont, CA, USA). 

Rat Corneal Alkali Burn and Treatment

Adult male Sprague-Dawley rats (180–200 g, quarantined and purchased from Shanghai SLAC Laboratory Animal Company Ltd., Shanghai, China) were used in the study. Animal experiments were performed in accordance with the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the experimental protocol was approved by the experimental animal ethics committee of Xiamen University (approval ID: XMUMC2011-03-3). All rats were confirmed to be free of ocular surface disease before the experiments. Rat corneal alkali burns were created as previously reported.¹⁸ 

To demonstrate the inhibitory effect of 444283 on MMP13 activity and determine the concentration of this inhibitor for topical administration to alkali-burned rat corneas, animals were randomly divided into four equal groups (n = 5) after alkali burns and topically applied with 50, 100, or 200 μM of 444283 (dissolved in dimethyl sulfoxide [DMSO] and further diluted with saline) or saline (containing same volume of DMSO as 200 μM of 444283) using a pipette (two times per day, 10 μL each). After 7 consecutive days of treatment, the eyes were enucleated, and the corneas were dissected for zymogen extraction as previously reported.¹⁹ The MMP13 activity of the rat corneas was detected via MMP13 assay kit according to the manufacturer's instructions. 

To investigate whether MMP13 expression was involved in the formation of new blood vessels and the effect of 444283 on new blood vessels formation in alkali-burned rat corneas, the animals were randomly divided into two equal groups after alkali burns and were topically given saline or 100 μM of 444283 using a pipette (two times per day, 10 μL each) for 7 consecutive days. Normal rats without alkali burn injuries were used as controls. After each treatment time point, 20 animals from each group were euthanized, and their eyes were enucleated. A small number of the enucleated eyes (n = 5) were fixed with cold 4% paraformaldehyde and then embedded in paraffin wax for study via immunohistochemistry (IHC) and in situ hybridization (ISH); other enucleated eyes were given cornea dissection. The dissected corneas were processed for the extraction of protein (n = 5) and RNA (n = 5) or used for the examination of collagen degradation (n = 5). 

Slit Lamp Microscopic Observation

Rats were examined under a slit lamp microscope every day after the alkali burns. The corneal images were taken by an experienced investigator who was blinded to the groups. The images captured via the slit lamp microscope were saved and measured using image analysis software (Image-Pro Plus v6.0; Media Cybernetics, Silver Spring, MD, USA). The area of corneal neovascularization was determined by measuring the radius from the center to the border of vessel growth (I) with a reticule and determining the number of clock hours (C) of limbus involved by using the following formula: S = C/12 × 3.1416 × [r² − (r − I)²], where S is the area of corneal neovascularization and r is the radius of the cornea.^18,20 

IHC and ISH on a Paraffin Section of Rat Corneas

Immunohistochemistry for CD146, which is a marker of vascular endothelial cells (VECs),²¹ ISH for MMP13 mRNA, sequential IHC for specific cell markers (ALDH3A1 for keratocytes,^22,23 CD146 for VECs, ED1 for macrophages, PMN for neutrophils), ISH for MMP13 mRNA, and double IHC for ALDH3A1 and other cell marker (CD146, ED1, or PMN) were performed as previously described,²⁴ except that ISH was performed with a digoxigenin-labeled rat MMP13 cRNA probe that corresponds to nucleotides 616-1154 (GenBank accession number NM_133530) and some sections of double IHC for ALDH3A1 and CD146 were counterstained with hematoxylin. Electronic images of IHC and ISH were visualized under a light/fluorescence microscope (DM2500, Leica, Wetzlar, Germany), captured, and saved to a computer. Red labeling developed using a fast red substrate in IHC was visualized under not only a light microscope but also a fluorescence microscope. The same field as was used in the IHC and ISH images was selected and visualized under the same microscope, and the resulting images were saved for the comparison and analysis of co-localization as previously described.²⁴ 

MMP13 Promoter Assay

A 1.2-kb human MMP13 promoter fragment corresponding to nucleotides 1 through 1202 (GenBank accession number X81640) was generated via PCR using human genomic DNA as a template and subcloned into the KpnI and XhoI sites of the pGL-3 basic luciferase reporter plasmid (Promega, Shanghai, China). The primer pairs used were 5′-cggggtacctctagaatcagtactaagtttctctttatgg-3′(forward) and 5′-ccgctcgagtctagaaagagtggaactcttcatctt-3′(reverse). The sequence of the human MMP13 promoter fragment was verified via DNA sequencing. The constructed MMP13 promoter-luciferase reporter plasmid was transiently cotransfected into 293T cells with pRL-TK containing the Renilla luciferase gene (Promega) by using transfection reagent (Invitrogen) according to the manufacturer's instructions. Twenty-four hours following the transfection, the medium was replaced with serum-free medium with or without the agent (30 ng/mL VEGFc or 50 ng/mL VEGFa). After 24 hours, the transfected cells were washed with PBS. Both luciferase activities were measured with a double-luciferase reporter assay kit as described by the manufacturer, using a 96-microplate luminometer (Glomax; Promega, Madison, WI, USA). 

The Culture and Treatment of Primary Rat Keratocytes

Primary rat corneal keratocytes were isolated and cultured in three-dimensional collagen gel as previously described,²⁵ with some modifications. In brief, the endothelial layer of the cornea was mechanically removed, and the tissue was then incubated with dispase (2 mg/mL in Dulbecco's modified Eagle's medium [DMEM]) at 0°C overnight. After the mechanical removal of the epithelial sheet, the detached corneal stromas were digested with 2 mg/mL collagenase in DMEM containing 10% fetal bovine serum (FBS) at 37°C until a single-cell suspension of keratocytes was obtained. Isolated keratocytes were cultured with DMEM containing 10% FBS in a 37°C humidified incubator with 5% CO₂. The keratocytes were used for experiments between passages 3 and 5. Acid-solubilized type I collagen (3 mg/mL), 5 × DMEM, reconstitution buffer (0.05 M NaOH, 0.2 M HEPES), and the keratocyte suspension (2 × 10⁶ cells/mL in DMEM) were mixed on ice at a ratio of 7:2:1:1, transferred into six-well plates, and incubated at 37°C for 2 hours for collagen polymerization. After incubation with serum-free DMEM overnight, the keratocytes were separately treated with increasing concentrations (7.5, 15, and 30 ng/mL) of VEGFc, neutralized VEGFc (VEGFc 15 ng/mL + VEGFc antibody 200 ng/mL), and VEGFc (15 ng/mL) for 24 hours after pretreatment with VEGFr3 antibody (200 ng/mL) for 30 minutes. The same amount of vehicle was added to the control keratocytes. The culture media were collected and ultrafiltrated to remove native collagen fibrils with molecular weights of more than 100 kDa. The ultrafiltrates were stored at −80°C to determine of the amount of degraded collagen and secreted MMP13. Acid-solubilized type I rat collagen was prepared as previously reported.²⁶ 

RNA Isolation and Quantitative RT-PCR

Total RNA was extracted from rat corneas using TRIzol reagent (Invitrogen, Guangzhou, China) according to the manufacturer's instructions and reverse transcribed to cDNA using a cDNA synthesis kit (RevertAid First Strand cDNA Synthesis Kit; Thermo Scientific, Beijing, China). Quantitative real-time PCR was performed with a PCR detection system (StepOne Real-Time PCR; Applied Biosystems, Carlsbad, CA, USA) using a commercial detection kit (SYBR Premix Ex Taq Kit; Takara Bio Co., Dalian, China) according to the manufacturer's instructions. The amplification program included an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of denaturation at 95°C for 10 seconds and annealing and extension at 60°C for 30 seconds. The fluorescence signals were measured after each extension step, and the specificity of the amplification was evaluated via melting curve analysis. The specific gene products were amplified using the following primer pairs: β-actin, 5-agccatgtacgtagccatcc-3, and 5-ctctcagctgtggtggtgaa-3; VEGFc, 5-ctacagatgtgggggttgct-3, and 5-gctgcctgacactgtggtaa-3; VEGFr3, 5-ctccaacttcttgcgtgtca-3, and 5-acaaggtcctccatggtcag-3; MMP13, 5-aggccttcagaaaagccttc-3, and 5-gagctgcttgtccaggtttc-3; CD146, 5-cagcaaaggagaggaaggtg-3, and 5-ctctggggtcacaaggacat-3; VEGFa, 5-caatgatgaagccctggagt-3, and 5-tttcttgcgctttcgttttt-3; VEGFr2, 5-ccaagctcagcacacaaaaa-3, and 5-ccaaccactctgggaactgt-3. A nontemplate control was included to evaluate the level of DNA contamination. The results of the relative qPCR were analyzed via the comparative Ct method and normalized using β-actin. 

Western Blotting Assay

Western blotting (WB) assays were performed as previously described¹⁸ for proteins extracted from rat corneas. Total proteins were extracted from rat corneas with cold RIPA buffer containing a proteinase inhibitor cocktail (Roche Applied Science, Mannheim, Germany). Equal amounts of rat corneal protein and ultrafiltrates from the culture media of the rat keratocytes were subjected to electrophoresis on 10% sodium dodecyl sulfate-polyacrylamide gels. The standard Western blot assay protocol was applied. Specific primary antibodies (VEGFc, VEGFa, VEGFr2, VEGFr3, CD146, MMP13, Col1α2, and β-actin) and secondary antibodies (HRP-conjugated rabbit anti-goat IgG and HRP-conjugated goat anti-rabbit IgG) were used. Finally, the specific bands were detected by enhanced chemiluminescence reagent. The images were recorded and assessed by a gel imaging system (Molecular Imager ChemiDoc XRS; Bio-Rad, Hercules, CA, USA). Each quantified band was normalized to the corresponding β-actin level. 

The Examination of Collagen Degradation

Hydroxyproline (HYP) content has been used as an indicator to determine the amount of collagen present.²⁷ Collagen degradation was therefore determined by measuring the HYP content in the samples. The content of HYP in rat corneal stromas and the ultrafiltrates of the rat keratocytes culture media was measured spectrophotometrically as previously described,^25,28 using a hydroxyproline assay kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) according to the manufacturer's instructions. 

Statistical Analysis

All experiments described above were repeated three times. Summary data were reported as mean ± SD. An appropriate version of the t-test was used to compare the group mean data, with P < 0.05 considered statistically significant. The statistical analysis was performed with commercial software (Prism 4.0; GraphPad Software, San Diego, CA, USA). 

Results

MMP13 Expression in Normal and Alkali-Burned Rat Corneas

A previous study had shown that MMP13 mRNA was only expressed in epithelial basal cells of wounded rat corneas from 6 hours to 3 days after undergoing laser keratectomy and was not expressed in normal rat corneas.²⁹ To clarify whether MMP13 is expressed in normal rat corneas, we performed qRT-PCR and WB. Expression of MMP13 mRNA was clearly detected via qRT-PCR in normal rat corneas (Figs. 1A, 1A[a]). Sequencing the qRT-PCR product demonstrated that the sequence completely matched rat MMP13 cDNA (data not shown). Western blotting also showed that MMP13 protein was indeed expressed in normal corneas (Figs. 1A, 1A[b], 1A[c]). To determine the location of the MMP13 mRNA expression, we performed ISH for MMP13 mRNA expression on paraffin sections of normal rat corneas. The results showed that little to no MMP13 expression was detectable in keratocytes and the endothelium; expression was predominantly in epithelial cells, but it was not limited to epithelial basal cells (Figs. 1B, 1B[a], 1B[b], purple). We examined MMP13 mRNA via ISH, instead of examining MMP13 protein via IHC, because staining with the available antibodies against MMP13 was suboptimal on paraffin-embedded tissues. 

Figure 1

 

The expression of MMP13 in normal and alkali-burned rat corneas. (A) The expression of MMP13 mRNA and protein in normal and alkali-burned rat corneas was determined via qRT-PCR (a) and WB (b). The expression of MMP13 mRNA and protein was gradually and significantly increased from 1 to 7 days after the alkali burns. *P < 0.05. ***P < 0.001. (c) Graphic representation of densitometry of MMP13 protein expression. The expression levels of MMP13 mRNA and protein were normalized to β-actin and calculated as described in the Materials and Methods section. (B) The expression of MMP13 mRNA in normal and alkali-burned rat corneas was determined via ISH. In normal corneas (a, b), little to no MMP13 mRNA was detectable in keratocytes (some indicated by brown arrows) and endothelia (some indicated by blue arrows). Matrix metalloproteinase 13 mRNA was predominant in epithelial cells but not limited in epithelial basal cells (some indicated by black arrows). In alkali-burned corneas (c–j), MMP13 mRNA was predominantly expressed in corneal stroma cells, including spindle-shaped fibroblast-like activated keratocytes (some indicated by brown arrows in [c–j]) and endothelia (some indicated by blue arrows in [d, e, g, i]), whereas the MMP13 mRNA expression of corneal epithelial cells appeared to experience a slight increase at day 1 (some indicated by black arrows in [c, d]) and a decrease at day 7 (some indicated by black arrows in [i, j]); MMP13 sense control for ISH in normal corneas and alkali-burned corneas at day 7 is in (k) and (l), respectively. These results were representative examples of five independent experiments. Scale bars: 100 μm (B[a], [c], [e], [g], [i], [k], [l]); 50 μm (B[b], [d], [f], [h], [j]). epi: epithelium; endo: endothelium; st: stroma.

Keratocytes were activated and proliferated after wounding and migrated into the wound area to repopulate the wound.^14–16 To investigate MMP13 expression in wounded rat corneas, the corneas were treated with saline at various time points following alkali burns, then dissected, and finally processed for qRT-PCR, WB, and ISH analyses. Western blotting and qRT-PCR showed that the MMP13 mRNA (Figs. 1A, 1A[a]) and protein (Figs. 1A, 1A[b], 1A[c]) expression is gradually and significantly increased from 1 to 7 days after the alkali burns, compared with normal corneas. In situ hybridization showed that MMP13 mRNA was predominantly expressed in corneal stroma cells, including spindle-shaped fibroblast-like activated keratocytes and some cells in the endothelium, whereas MMP13 mRNA expression in corneal epithelial cells appeared to increase slightly at day 1 and decrease at day 7 (Figs. 1B, 1B[c], 1B[d], 1B[i], 1B[j] versus Figs. 1B, 1B[a], 1B[b]). These results indicated that the increased MMP13 expression was predominantly contributed by cells in corneal stromas after alkali burns. 

MMP13 Expression Promotes Corneal Neovascularization After Alkali Burns

Matrix metalloproteinase 13 is able to directly stimulate new blood vessel formation in the chorioallantoic membrane, as determined by angiogenesis assay.⁹ To investigate whether MMP13 expression was involved in the development of CNV after alkali burns, we first determined the relationship between new blood vessel formation and MMP13 expression in rat corneas treated with saline at various time points after alkali burns (Figs. 2A, 2A[a–e]). The neovascularization area gradually increased from day 1 to 7 after the alkali burns (Fig. 2B), similar to the expression of MMP13 mRNA and protein. We then investigated whether the inhibition of MMP13 activity using 444283 could attenuate new blood vessel formation in alkali-burned rat corneas; 444283 is a derivative of pyrimidine dicarboxamide, and it very selectively inhibits MMP13 activity.^19,30,31 It has been demonstrated to efficiently and specifically inhibit MMP13 activity in cultured C2C12 cells at a concentration of 10 μM.¹⁹ To demonstrate that 444283 can also inhibit MMP13 activity in alkali-burned rat corneas and determine the concentration of this inhibitor to use for topical administration, rats were given 444283 concentrations of 50, 100, and 200 μM or saline topically for 7 days after alkali burns. As shown in Figure 2C, 50 μM of 444283 inhibited only approximately 30% of MMP13 activity, but 100 and 200 μM of 444283 caused approximately a 60% reduction of MMP13 activity in alkali-burned rat corneas. 100 μM of 444283 was therefore determined to be appropriate for topical administration. The results showed that the topical administration of 444283 could significantly inhibit neovessel ingrowth (Figs. 2A, 2A[f–j]). The neovascularization area was much lower from days 3 to 7 in the 444283-treated groups compared with the saline-treated groups (Figs. 2A, 2A[h–j] versus 2A[c–e]; Fig. 2B). The inhibitory effect of 444283 treatment on corneal neovessel ingrowth was also shown by the examination of IHC for CD146, in which there were many CD146+ neovessels (Figs. 2D, 2D[a] and 2D[c], red) in the saline-treated groups, but only a few neovessels in the 444283-treated groups (Figs. 2D, 2D[b] and 2D[d], red) 7 days after the alkali burns. These data show a relationship between MMP13 expression and neovascularization. 

Figure 2

 

Expression of MMP13 promotes corneal neovascularization after alkali burns. (A) Representative slit lamp examination images of normal corneas (a, f) and alkali-burned corneas at days 1, 3, 5, and 7 with saline (b–e) or 444283 (g–j) treatments. In the saline-treated groups (a–e), the onset of peripheral neovascularization occurred on day 1 after the alkali burns (b). There was an ingrowth of newly formed blood vessels toward the central corneas on day 3 (c). newly formed blood vessels reached the central burn on day 7 (e). The neovascularization areas gradually increased from 1 day to 7 days after the alkali burns (b–e). In contrast, in the 444283-treated groups (f–j), new blood vessel formation was inhibited, and the neovascularization areas were much lower from days 3 to 7 (h–j) compared with those of the saline groups (c–e). (B) Graphic representation of the quantitative analysis of CNV areas in the saline- and 444283-treated groups. *P < 0.05. **P < 0.01. ***P < 0.001. (C) The inhibitory effect of 444283 on MMP13 activity in alkali-burned rat corneas at day 7. Fifty μM of 444283 inhibited approximately 30% of MMP13 activity, but 100 and 200 μM of 444283 caused approximately a 60% reduction in MMP13 activity compared with saline. *P < 0.05. ***P < 0.001. (D) New blood vessel formation was detected via IHC for CD146; new blood vessels were well-indicated by CD146+ vessels (red labeling) in alkali-burned corneas at day 7 for both the saline (a, c) and 444283 (b, d) treatments. There were prominent CD146+ vessels (red labeling in [a], [c]) in the saline-treated groups, whereas there were only a few CD146+ vessels (red labeling in [b], [d]) in the 444283-treated groups. These results were representative examples of five independent experiments. Scale bars: 1.5 mm (A); 200 μm (D[a], D[b]); 50 μm (D[c], D[d]).

Keratocytes Are Major Cells Expressing MMP13 in Corneal Stromas After Alkali Burns

New blood vessel formation is a complex process; it is also called sprouting angiogenesis. During this process, the invasion of sprouting VECs into the surrounding stroma depends on the expression of proteinases, particularly MMPs, to degrade and remodel ECM.^32,33 We showed above that MMP13, which can promote CNV and MMP13 mRNA, was predominantly expressed in corneal stroma cells after the alkali burns. To investigate whether the increased expression of MMP13 was caused by invading neovessel VECs, we investigated the relationship between CD146 and MMP13 mRNA and protein expression in rat corneas at various time points after alkali burns. We found that the expression of CD146 mRNA and protein was delayed by 2 days because detectable CD146 expression appeared after 3 days and was gradually increased from day 5 to 7, whereas MMP13 was markedly expressed on day 1 and gradually increased in expression from day 3 to 7 after the alkali burns (Figs. 3A, 3A[a–c]). These results showed that the initial increase in MMP13 expression was caused by other types of cells, rather than invading VECs, in the rat corneas shortly after alkali burns. 

Figure 3

 

Keratocytes are the major cells expressing MMP13 in rat corneas after alkali burns. (A) The mRNA and protein expression of MMP13 and CD146 in normal and alkali-burned rat corneas were determined via qRT-PCR (a) and WB (b). After alkali burns, MMP13 mRNA and protein were markedly expressed at day 1 and gradually increased from day 1 to 7, whereas the expression of CD146 mRNA and protein appeared at day 3 and gradually increased from day 3 to day 7; thus, the expression of CD146 mRNA and protein had a delay of 2 days (a, b). (c) Graphic representation of densitometry of MMP13 and CD146 protein expression. (B) The types of cell expressing MMP13 in corneal stromas were studied via sequential IHC and ISH on the same sections: IHC for spindle-shaped fibroblast-like ALDH3A1+ keratocytes in corneal stromas at day 5 after alkali burns (red labeling, some indicated by purple arrows in [a]), CD146+VECs in limbus at day 1 after alkali burns (red labeling, one indicated by red arrow in [b]), ED1+ macrophages in corneal stroma at day 5 after alkali burns (red labeling, some indicated by green arrows in [c]), PMN+ neutrophils in corneal stroma at day 5 after alkali burns (red labeling, some indicated by white arrows in [d]), isotype antibody control (e); ISH showed MMP13 mRNA expression in ALDH3A1+ keratocytes (purple labeling, some indicated by brown arrows in [f]), CD146+ VECs (purple labeling, one indicated by red arrow in [g]), and ED1+ macrophages (purple labeling, some indicated by green arrows in [h]), but not in neutrophils (some indicated by white arrows in [i]); MMP13 sense control (j). Counterstaining with DAPI in the right column showed cellular nuclei. The panels of each row represent the results for the same section of an alkali-burned rat cornea. These results are representative examples of five independent experiments. (C) The types of cells expressing ALDH3A1 were studied via double IHC in alkali-burned rat corneas at day 5. The results showed that ALDH3A1 was expressed only by spindle-shaped fibroblast-like keratocytes (brown labeling, some indicated by purple arrows in [a–c]), not by CD146+ VECs (red labeling, some indicated by red arrows in [a]), ED1+ macrophages (red labeling, some indicated by green arrows in [b]), or PMN+ neutrophils (red labeling, some indicated by white arrows in [c]). Isotype antibody control for the double IHC is shown in (d). The green irregular dotted circles (in B[a], [f], C[a–c]) indicated some ALDH3A1+ keratocytes lining the lumens. Scale bars: 50 μm.

To identify what types of cells expressed MMP13 in corneal stromas and thus determine their contributions to CNV, we performed sequential IHC for ALDH3A1, CD146—as well as ED1 or PMN and ISH for MMP13 mRNA—on the same section of alkali-burned rat corneas. Regarded as a keratocyte marker,^22,23 ALDH3A1 is abundantly expressed in epithelial cells and stromal keratocytes of mammalian corneas.^34,35 As shown in Figure 3B, ALDH3A1+cells were spindle-shaped fibroblast-like activated keratocytes (Fig. 3B, red in 3B[a]), and they all expressed MMP13 (Fig. 3B, purple in 3B[f]). Although MMP13 was also expressed by CD146+ VECs (Fig. 3B, red in 3B[b] and purple in 3B[g]) and ED1+ macrophages (Fig. 3B, red in 3B[c] and purple in 3B[h]), not by PMN+ neutrophils (Fig. 3B, red in 3B[d] and 3B[i]), Spindle-shaped fibroblast-like activated keratocytes were the predominant cells expressing MMP13 in alkali-burned rat corneal stromas (Fig. 3B, purple in 3B[f–i]). To exclude the possibility that other cells in the corneal stromas also express ALDH3A1, we performed double IHC for ALDH3A1 and CD146, ED1, and PMN in alkali-burned rat corneas. The results confirmed that ALDH3A1 was expressed only by spindle-shaped fibroblast-like activated keratocytes in rat corneal stromas (Figs. 3C, 3C[a–c]). These data indicated that keratocytes were the major contributors that expressed MMP13 to promote CNV after alkali burns. 

Keratocytes Directly Create Stromal Spaces to Promote Corneal Neovascularization Through MMP13 Expression

To investigate how keratocytes promoted CNV via MMP13 expression after alkali burns, we checked the sections of the alkali-burned corneas and found that there were many ALDH3A1 + keratocytes lining the lumens in the stroma of alkali-burned corneas (Figs. 4A, 4A[a–d], also seen in Figs. 3B, 3B[a], 3B[f], and Figs. 3C, 3C[a–c], green irregular dotted circles). These ALDH3A1+ keratocytes lining the lumens were in front of or around CD146+ neovessels (Figs. 4A, 4A[a–d], also seen in Figs. 3C, 3C[a], red), whereas there were no ALDH3A1+ keratocytes lining the lumens in normal rat corneas (Figs. 4A, 4A[e], 4A[f]). The results indicated that ALDH3A1+ keratocytes lining the lumens in the stroma of alkali-burned corneas were spaces created by ECM degradation rather than an artificial phenomenon created by tissue processing. Moreover, some of the ALDH3A1+ keratocytes lining the lumens obviously contained red blood cells (Figs. 4A, 4A[c], 4A[d], yellowish). The existence of red blood cells in ALDH3A1+ keratocytes lining the lumens further supported the notion that these lumens were spaces created by ECM degradation rather than an artificial phenomenon caused by tissue processing because newly formed vessels are very fragile and susceptible to hemorrhage. They have no basement membrane or pericytes,¹ and red blood cells can flow from leaky blood vessels into the surrounding spaces following the leakage of plasma. Therefore, these data indicated that, as the major MMP13-expressing cells in corneal stromas after alkali burns, keratocytes can at least partially degrade and remodel the ECM to create spaces into which VECs can migrate through MMP13 expression. 

Figure 4

 

Type I collagen degradation and MMP13 expression in rat corneas after alkali burns. (A) The detection of double IHC for ALDH3A1 and CD146 followed by hematoxylin counterstaining for nuclei on the same sections, showing ALDH3A1+ keratocytes lining the lumens in alkali-burned corneas (a–d), but not normal corneas (e, f). In the stroma of alkali-burned rat corneas at day 3 (a–d), some CD146+ neovessels and VECs (red labeling, some indicated by red arrows) were visible in (a, b) (under the light microscope) and also visible in (c, d) (under the fluorescence microscope); ALDH3A1+ keratocytes lining the lumens (brown labeling, some indicated by purple arrows) were visible in (a, b). The green irregular dotted circles in (b, d) indicate some ALDH3A1+ keratocytes lining the lumens that were in front of or around CD146+ neovessels. The cells having no nuclei in ALDH3A1+ keratocyte lining the lumens or CD146+ vessels, some indicated by yellow arrows and visible in (b), and also being biconcave and having a yellowish fluorescence, visible in (c, d), were red blood cells. In the stroma of normal rat corneas (e, f), there were ALDH3A1+ keratocytes (brown labeling, some indicated by purple arrows, visible in [f]), but not CD146+ neovessels, VECs, or ALDH3A1+ keratocytes lining the lumens. (B) Type I collagen degradation and MMP13 expression in normal and alkali-burned rat corneas. (a) Graphic representation of detected HYP contents in normal and alkali-burned rat corneas. In alkali-burned rat corneas treated with saline, HYP content exhibited a gradual decrease from day 1 to 7 as compared with normal corneas. However, in alkali-burned rat corneas treated with 444283, the gradual decrease in HYP content was significantly inhibited from day 3 to 7. *P < 0.05. **P < 0.01. ***P < 0.001. (b) The levels of Col1α2 and MMP13 detected by WB; Col1α2 was gradually decreased in the saline treatment group from day 1 to 7, whereas the gradual decrease of Col1α2 was also significantly inhibited from day 3 to 7 in rat corneas treated with 444283. *P < 0.05. **P < 0.01. ***P < 0.001. Matrix metalloproteinase 13 was gradually and markedly increased after alkali burns. (c) Graphic representation of the densitometry of Col1α2 and MMP13 protein. These results were representative examples of five independent experiments. Scale bars: 100 μm (A[a], [c], [e]); 50 μm (A[b], [d], [f]). H, hematoxylin; RBC, red blood cells.

The ECM of the corneal stromas is primarily composed of type I collagen, a heterotrimer consisting of two α1 chains and one α2 chain, with lesser amounts of other collagens and small leucine-rich proteoglycans.^36,37 Matrix metalloproteinase 13 can degrade these ECM components.^9–13 To prove that ALDH3A1+ keratocyte lining the lumens were formed by collagen degradation, we detected the HYP content in rat corneas treated with saline or 444283 at various time points after alkali burns. We detected HYP content because it has been used as an indicator of collagen amount.^27,38 The results showed that the HYP content was gradually decreased in rat corneas treated with saline from day 1 to 7 after alkali burns compared with normal corneas. However, the gradual decrease in HYP content was significantly inhibited in rat corneas treated with 444283 (Figs. 4B, 4B[a]). To confirm that type I collagen was degraded in the corneal stromas, we also detected the level of type I collagen α2 polypeptide (Col1α2) in rat corneas treated with saline or 444283 at various time points after alkali burns via WB and found that Col1α2 was gradually decreased from day 1 to 7 after the alkali burns, in contrast to MMP13 expression. The gradual decrease of Col1α2 was also significantly inhibited in rat corneas treated with 444283 (Figs. 4B, 4B[b], 4B[c]). These data evidently supported the notion that keratocytes can at least partially degrade type I collagen to create stromal spaces into which VECs migrate, (i.e., encouraging the growth of newly formed blood vessels through MMP13 expression after alkali burns). 

MMP13 Expression Is Consistent With VEGFc and VEGFr3 Expression in Alkali-Burned Rat Corneas

The expression of MMP13 is regulated by certain cytokines, such as interleukin (IL)-1, IL-6, tumor necrosis factor alpha, transforming growth factor beta, basic fibroblast growth factor, and VEGFa.^12,39–42 The most important family of cytokines implicated in the neovascularization processes is the VEGF family.⁴³ Vascular endothelial growth factor A and VEGFc are members of the VEGF family and have been demonstrated to promote CNV. Vascular endothelial growth factor A, by binding VEGFr2, and VEGFc, by binding VEGFr2 and VEGFr3, stimulate new blood vessel formation.^1,2,43 Previous studies have indicated that VEGFa is correlated with MMP13 expression.^4,41,44,45 However, the correlation between VEGFc and MMP13 expression has not yet been reported. Because VEGFa, VEGFc, VEGFr2, and VEGFr3 have all been showed to be expressed in the cornea, in addition to VECs,^1,2,46,47 we investigated whether their expression was related to MMP13 expression in alkali-burned rat corneas via qRT-PCR and WB. Interestingly, we found that the mRNA and protein expression of VEGFa and VEGFr2 was not significantly different between normal corneas and alkali-burned corneas, although their expression was clear (Figs. 5A, 5A[a–c]). However, the mRNA and protein expression of VEGFc and VEGFr3 showed a gradual increase from day 1 to 7 after the alkali burns, similar to the expression of MMP13 mRNA and protein (Figs. 5A, 5A[d–f] versus Figs. 1A, 1A[a–c]). These data indicated that MMP13 expression was related to the expression of VEGFc and VEGFr3, but not VEGFa and VEGFr2, in alkali-burned corneas. Because keratocytes were the major cells expressing MMP13 in alkali-burned corneas, to confirm that VEGFc andVEGFr3 could be expressed by keratocytes, in addition to VECs, we performed the sequential detection of IHC for CD146 and ISH for VEGFc and VEGFr3 in normal corneas and alkali-burned corneas at day 3. The results showed that VEGFc (Figs. 5B[d], 5B[j], purple) and VEGFr3 (Figs. 5B[e], 5B[k], purple) mRNAs were indeed predominantly expressed in spindle-shaped fibroblast-like keratocytes (some indicated by brown arrows). There was some expression in CD146+ VECs (Figs. 5B[g], 5B[j], 5B[h], 5B[k]; some indicated by red arrows) of neovessels and other cells in alkali-burned corneas. They were also expressed in the epithelium (Figs. 5B[d], 5B[e], 5B[j], 5B[k]; some indicated by black arrows) and endothelium (Figs. 5B[d], 5B[e], 5B[j], 5B[k]; some indicated by blue arrows) in normal and alkali-burned corneas, similar to MMP13 mRNA expression (Figs. 5B[f], 5B[l], purple). The correlation between the expression of VEGFc/VEGFr3 and MMP13 in alkali-burned corneas strongly indicated that the MMP13 expression of keratocytes could be regulated by VEGFc via VEGFr3 but not VEGFa via VEGFr2. Our results also indicated that the MMP13 expression of keratocytes could be regulated by VEGFc from them keratocytes themselves (autocrine) or other cells in the corneas (paracrine). 

Figure 5

 

Expression of MMP13 is consistent with VEGFc and VEGFr3 expression in alkali-burned rat corneas. (A) The expression of VEGFa/VEGFr2 and VEGFc/VEGFr3 mRNA and protein in normal and alkali-burned rat corneas. The analysis of the qRT-PCR (a) and WB (b) showed that the expression of mRNA and protein for VEGFa/VEGFr2 was not significantly different for normal corneas and alkali-burned corneas, although they were clearly expressed, whereas the expression of VEGFc/VEGFr3 mRNA and protein determined by qRT-PCR (d) and WB (e) showed a gradual increase from day 1 to day 7 in alkali-burned corneas as compared with normal corneas. *P < 0.05. ***P < 0.001. (c, f) Graphic representations of the densitometry of the VEGFa/VEGFr2 and VEGFc/VEGFr3 protein expression, respectively. (B) Sequential detection of IHC for CD146 and ISH for VEGFc, VEGFr3, and MMP13 mRNA in normal rat corneas (a–f) and alkali-burned rat corneas at day 3 (g–l). Examination of IHC showed that CD146 was not expressed in normal rat corneas (a–c), but was expressed in the VECs of neovessels in alkali-burned rat corneas (red labeling, some indicated by red arrows in [g–i]). Detection of ISH showed that VEGFc (purple labeling in [d, j]) and VEGFr3 (purple labeling in [e, k]) mRNAs were expressed in epithelia (some indicated by black arrows), spindle-shaped fibroblast-like keratocytes (some indicated by brown arrows), endothelia (some indicated by blue arrows), and CD146+VECs (Figs. 5B[g], [h], [j], [k]; some indicated by red arrows) in normal rat corneas and alkali-burned rat corneas, which was similar to MMP13 mRNA expression (purple labeling in [f, l]). These results were representative examples of five independent experiments. Scale bars: 100 μm.

VEGFc Induces Primary Keratocytes to Express MMP13 and Degrade Type I Collagen Via VEGFr3

Vascular endothelial growth factor C has been demonstrated to upregulate the expression of other MMPs, such as MMP14.⁴³ To determine that the MMP13 expression of keratocytes is induced by VEGFc via VEGFr3 but not VEGFa via VEGFr2, we first investigated the effects of VEGFc and VEGFa on the transcriptional activation of the MMP13 promoter; 293T cells were transiently transfected with the MMP13 promoter-luciferase reporter plasmid, followed by treatment with or without rat VEGFc or human VEGFa. The transcriptional activation of the MMP13 promoter was examinated via luciferase activity; 293T cells were chosen because they were highly transfectable, whereas primary cultured rat keratocytes have been found to be notoriously difficult to transfect. In addition, the human MMP13 promoter has a high degree of functional and sequence homology (86%) with rat MMP13,⁴⁸ and there is only a one-amino-acid difference between rat and mature human VEGFc (dNdC VEGFc).⁴⁹ The promoters of MMP13 and mature VEGFcs of humans and rats can therefore replace one another functionally. The results showed that VEGFc treatment caused a significant increase in the transcriptional activation of the MMP13 promoter as compared with vehicle treatment, whereas VEGFa treatment had little effect on the transcriptional activation of the MMP13 promoter (Fig. 6A). These results demonstrated that VEGFc, but not VEGFa, could activate the MMP13 promoter. 

Figure 6

 

VEGFc induces primary keratocytes to express MMP13 and degrade type I collagen via VEGFr3. (a) Effects of VEGFc and VEGFa on MMP13 promoter activity. The MMP13 promoter was significantly activated by VEGFc, but not VEGFa, compared with no treatment. ***P < 0.001. The transfection of 293T cells with a 1.2-kb human MMP13 promoter-luciferase reporter plasmid, which was then either followed by VEGFc and VEGFa treatment or not, and the detection of luciferase activity were described in Materials and Methods section. (b) The level of MMP13 secretion in the culture supernatants from VEGFc-treated keratocytes. Upper: MMP13 secreted in culture supernatants was determined via WB. Lower: Graphic representation of the densitometry of the secreted MMP13. Vascular endothelial growth factor C promoted MMP13 secretion from keratocytes in a dose-dependent manner. ***P < 0.001. (c) The detected contents of HYP in the culture supernatants collected from VEGFc-treated keratocytes. The contents of HYP showed dose-dependent increases in culture supernatants from VEGFc-treated keratocytes. ***P < 0.001. (d) The level of MMP13 secretion in the culture supernatants from keratocytes given various treatments for 24 hours. Upper: Keratocytes were treated with a vehicle, 15 ng/mL VEGFc + 200 ng/mL VEGFc antibody, 15 ng/mL VEGFc, and 15 ng/mL VEGFc + 200 ng/mL VEGFr3 antibody, respectively. Lower: MMP13 secreted in culture supernatants was determined via WB. MMP13 secretion was significantly decreased in culture supernatants from keratocytes treated with VEGFc antibody or VEGFr3 antibody. (e) Graphic representation of the densitometry of the secreted MMP13. ***P < 0.001. (f) The detected contents of HYP in culture supernatants from (d). Content of HYP was significantly decrease in culture supernatants from keratocytes treated with VEGFc antibody or VEGFr3 antibody. ***P < 0.001. These results were representative examples of five independent experiments.

We then investigated whether VEGFc can induce keratocytes to express MMP13 and degrade type I collagen through VEGFr3. After primary keratocytes were cultured in three-dimensional type I collagen gel, followed by various treatments, the culture media were collected and the levels of MMP13 secretion and HYP content were detected. The results showed that the levels of MMP13 secretion and HYP content experienced dose-dependent increases in the culture supernatants from VEGFc alone–treated keratocytes compared with those from vehicle-treated keratocytes (Figs. 6B, 6C), whereas the levels of MMP13 secretion and HYP content were significantly reduced in culture supernatants from either keratocytes treated with VEGFc antibodies or keratocytes treated with VEGFr3 antibodies (Figs. 6D–F). These results strongly demonstrated that VEGFc indeed could induce keratocytes to express MMP13 and degrade type I collagen via VEGFr3 and further supported the notion that activated keratocytes could at least partially express MMP13 to directly degrade type I collagen and create stromal spaces into which newly formed blood vessels can grow when induced by VEGFc in alkali-burned rat corneas. 

Discussion

Corneal neovascularization is the ingrowth of newly formed blood vessels into the normally avascular stroma from the limbal vascular plexus. The formation of new blood vessels is traditionally explained by the incompletely understood mechanism of sprouting angiogenesis, which underlies the invasion, migration, proliferation, and tube formation of VECs stimulated by angiogenic factors. The well-characterized stages of sprouting angiogenesis begin with the release of proteases by VECs and the degradation of the vessel basement membrane and the surrounding ECM.^32,33,50,51 In previous studies, keratocytes were commonly found to promote the functional properties of VECs via a paracrine mechanism, including enhanced secretion of ECM-degrading enzymes (MMP2, MMP3, MMP14, tissue- and urokinase-type plasminogen activators); angiogenic factors (VEGF); chemokines; and biomechanical forces or decreased expression of antiangiogenic factors (pigment epithelium-derived factor and thrombospondins).⁶ However, the exact function of keratocytes themselves has not yet been revealed in terms of the degradation and remodeling of corneal stromas associated with the formation of new blood vessels. In the present study, we elucidated the fact that keratocytes may not be limited to promoting the functional properties of VECs through the paracrine mechanism of VEGFc and MMP13, but could also directly degrade type I collagen to create spaces convenient for neovessel growth via the mechanism of MMP13 expression. To our knowledge, this is the first report delineating the fact that keratocytes can directly degrade ECM to create spaces convenient for neovessel growth during CNV; this is also the first report revealing that keratocytes can express MMP13 through the induction of VEGFc and VEGFr3. 

Our study is apparently different from previous studies regarding MMP13 expression in rat corneas. Previous studies showed that MMP13 mRNA was detected in wounded rat corneas 3 days after they underwent laser keratectomy, but not in normal rat corneas via RT-PCR, and that MMP13 mRNA was localized only in basal cells of the epithelia of wounded corneas from 6 hours to 3 days after laser keratectomy, but not in normal corneas or at 7 days after wounding via ISH.²⁹ In this study, we showed that MMP13 mRNA and protein were both found to be expressed in normal rat corneas and gradually increased from day 1 to 7 after alkali burns via qRT-PCR and WB. We also showed that MMP13 mRNA expression was predominantly contributed by epithelial cells in normal rat corneas and also by keratocytes in alkali-burned rat corneas, although MMP13 mRNA expression in corneal epithelial cells appears to experience a slight increase at day 1 when investigated via ISH. The reason that our finding was inconsistent with those of previous studies is difficult to analyze because ISH is a complex technique that is distinct from immunohistochemistry. However, comprehensive gene expression analysis of MMPs in a corneal model of epithelial resurfacing has shown that MMP13 expression is detectable in normal mouse corneal epithelia and significantly upregulated in regenerating epithelia at 18 and 24 hours after mouse corneal abrasion.⁵² Moreover, as an interstitial collagenase,^53–56 MMP13 has been shown to be expressed not only in stromal fibroblasts surrounding tumor cells (it was initially discovered in breast carcinomas and is overexpressed in a variety of malignant tumors),^4,8,10,44 but also in various other types of cells, including vascular endothelial cells,^3,57 uterine smooth muscle cells,⁵⁸ chondrocytes, and osteoblasts,⁵⁹ macrophages⁵³ and Kupffer cells,⁵⁴ neurons,⁶⁰ skin fibroblasts,⁶¹ stromal cells of the gastric mucosa,⁶² and hepatic stellate cells.⁵⁵ These studies evidently support our findings. The expression profile and function of MMP13 in normal and injured corneas, especially in the corneal epithelium, remains ill-defined because there have been few research projects in this area. 

Our study is also apparently different from previous studies regarding the regulation of MMP13 expression. The VEGF family is the most important family of cytokines involved in the neovascularization process.⁴³ In previous studies, VEGFa was indicated to have a causal role in CNV. It has been shown that VEGF expression is significantly higher in vascularized corneas than in normal corneas, and corneal epithelial cells, keratocytes, corneal endothelial cells, VECs, macrophages, and infiltrating leukocytes are the main sources of VEGFa production during CNV.^1,2,63 Vascular endothelial growth factor A may promote MMP13 expression by ovarian cancer cells.^40,41 However, there is a discrepancy regarding the correlation between VEGFa and MMP13 because some reports have showed that MMP13 could promote VEGFa expression or release from chondrocytes, fibroblasts, endothelial cells, and the ECM.^4,44,45 In the present study, we showed that the expression of VEGFa and VEGFr2 mRNAs and proteins was not significantly different in rat corneas before and after alkali burns, although they were indeed clearly detected via qRT-PCR and WB. Furthermore, we found that VEGFa had little effect on MMP13 gene transcription via MMP13 promoter assay. Therefore, we regard it as unlikely that the upregulation of MMP13 expression is a response to VEGFa expression in rat corneas after alkali burns. For the same reason, it is also unlikely that the upregulation of MMP13 expression can promote the expression or release of VEGFa. Interestingly, in the present study, we found that the mRNA and protein expression of VEGFc and VEGFr3 was consistent with that of MMP13 in terms of pattern and time course and that MMP13, VEGFc, and VEGFr3 mRNAs were all expressed by not only epithelial cells in normal rat corneas, but also keratocytes and VECs in alkali-burned rat corneas. We demonstrated that VEGFc could significantly promote MMP13 gene transcription via MMP13 promoter assay and that the blocking of VEGFc and VEGFr3 by their corresponding antibodies could markedly attenuate the MMP13 expression of cultured primary keratocytes. Our studies indicated that MMP13 expression could be regulated by VEGFc via autocrine or paracrine mechanisms. Vascular endothelial growth factor C is another member of the VEGF family and a ligand for VEGFr3 and VEGFr2.² Previous studies had shown that VEGFc could promote angiogenesis in mouse corneas, in the chorioallantoic membrane of chicks, and in the ischemic hind limbs of rabbits by inducing the proliferation, migration, and survival of endothelial cells and that it was a potent angiogenic factor regulating physiological and pathological angiogenesis, although it had been described as a relatively selective growth factor for lymphatic vessels. Vascular endothelial growth factor C was only expressed in wounded corneas, not in normal corneas, while VEGFr3 was expressed in both normal and wounded corneas.^2,64 In the present study, we revealed that VEGFc was expressed not only in wounded corneas, but also in normal corneas and can promote CNV through upregulating MMP13 in alkali-burned rat corneas. 

In conclusion, we present in vitro and in vivo data revealing that keratocytes are the major MMP13-expressing cells and can directly degrade the ECM to create stromal spaces into which new blood vessels grow through MMP13 expression in alkali-burned corneas. Moreover, we found that MMP13 expression in keratocytes was induced by VEGFc through VEGFr3. As a result, we discovered a new mechanism by which keratocytes promote CNV. 

Acknowledgments

We thank Hui He and Rongrong Zong for their technical support, Junqi Wang for photo editing, and Genomeditech Co. Ltd. (Shanghai) for the construction of an MMP13 promoter-luciferase reporter plasmid. 

Supported by the National Key Basic Research Program of China (2013CB967502) and the National Natural Science Foundation of China (81370991, 81070075, 81270978, 81330022, U1205025). 

Disclosure: J. Ma, None; D. Zhou, None; M. Fan, None; H. Wang, None; C. Huang, None; Z. Zhang, None; Y. Wu, None; W. Li, None; Y. Chen, None; Z. Liu, None 

Footnotes

 JM and DZ contributed equally to the work presented here and should therefore be regarded as equivalent authors.