Nerve Growth Factor Promotes Corneal Epithelial Migration by Enhancing Expression of Matrix Metalloprotease-9

Abstract

Purpose.: Nerve growth factor (NGF) is a neuropeptide essential for the development, survival, growth, and differentiation of corneal cells. Its effects are mediated by both TrkA and p75 receptors. Clinically relevant use of NGF was introduced to treat neurotrophic ulcerations in patients. Herein, we examine the mechanisms by which NGF enhances epithelial wound healing both in vivo and in vitro.

Methods.: An animal model using adult hens was implemented for the in vivo experiments. Laser ablation keratectomy was performed and animals were observed for up to 7 days. Epithelial healing was measured with fluorescein. In addition, proliferation was measured using BrdU incorporation and both TrkA and matrix metalloprotease-9 (MMP-9) expression were measured by immunohistochemistry (IHC) and Western blot (WB). In vitro experiments were carried out with telomerase-immortalized human corneal epithelial cells (HCLE). The rate of proliferation was measured using a colorimetric assay and BrdU incorporation. Real-time migration was evaluated with an inverted microscope. MMP-9 expression was evaluated by immunocytochemistry (ICC), WB, zymography, and RT-PCR. Finally, beta-4 integrin (β4) expression was assessed by ICC and WB.

Results.: Faster epithelial healing was observed in NGF-treated corneas compared with controls (P < 0.01). These corneas showed increased proliferation, TrkA upregulation, and enhanced MMP-9 presence (P < 0.01). In vitro, faster spreading and migration were observed in response to NGF (P < 0.01). Enhanced proliferation, as well as enhanced TrkA and MMP-9 expression, and decreased β4 levels were observed after adding NGF (P < 0.01).

Conclusions.: NGF plays a major role during the epithelial healing process by promoting migration, a process that is accelerated by cell spreading. This effect is mediated by both the upregulation of MMP-9 and cleavage of β4 integrin.

Introduction

The efficacy of nerve growth factor (NGF) for the treatment of ocular surface pathologies has been clinically tested and introduced for its clinical utility as a therapeutic agent for various diseases. ¹ Clinical experience has shown that topical administration of NGF improves epithelial healing in patients with neurotrophic corneal ulceration ^2–5 or after cataract surgery ⁶ and is also used to treat inflammation. ^7,8 Animal models have also shown NGF benefits in corneal wound healing and nerve regeneration ^9–11 ; however, detailed mechanisms by which NGF elicits its effect on the corneal surface remains unknown. 

NGF is an essential polypeptide for the development, survival, growth, and differentiation of neurons in the nervous system (NS). ¹² NGF also plays a relevant role in skin wound healing. ¹³ NGF-mediated activation of TrkA receptor leads to a variety of direct signaling pathways promoting cell survival, proliferation, and migration. ¹⁴ Conversely, NGF activation of p75NTR may signal apoptosis. ¹⁵  

The cornea is an avascular and transparent external organ of the visual system. Primary functions of the cornea are refraction, transmission of light, and as a protective barrier for the inner structures of the eye. The basic structure of the cornea is composed of three different tissues: epithelium, connective tissue or stroma, and endothelium. These tissues are well-organized layers separated by three additional structures; both the basement membrane (BMZ) and Bowman's layer (BM) separate epithelium from stroma and Descemet's membrane (DM) acts as a barrier between the deep stroma and endothelium. The correct organization of these components is essential for keeping the cornea healthy as well as to maintain transparency. Disruption of any of these components by injury, surgery, or infection may lead to a different pathologic state, such as fibrosis, edema, melting, ulceration, or perforation, resulting in ocular pain, loss of transparency, or blindness. 

When both the epithelial barrier and BMZ are disrupted, adjacent keratocytes undergo necrosis and apoptosis resulting in a significant reduction of keratocyte population around the damaged area. ^16–21 Immediately thereafter, the epithelial barrier is restored by proliferation and migration of surrounding epithelial cells, as well as limbal stem cells. ^22,23 Meanwhile, remaining quiescent keratocytes proliferate and migrate into the wounded area ^24,25 becoming myofibroblasts, which have contractile properties and biosynthetic activity, in order to repair the tissue. ²⁶  

Hens have been demonstrated to reproduce corneal wound healing observed in humans. ^27–36 Hen corneas exhibit a common morphology to human corneas, including the presence of a BM. ²⁸ Herein, we show that the addition of NGF after a compromising injury improves wound healing by inducing epithelial migration and cell proliferation in a well-documented animal model. ³⁰ The same effect was tested in vitro for the first time in Telomerase-immortalized human corneal epithelial cells (HCLE) to conclude that NGF induces epithelial spreading and migration by upregulating of matrix metalloprotease-9 (MMP-9) which cleaves beta-4 integrin (β4) during the epithelial migration process. 

Methods

Animals and Surgery

Forty-eight Lohmann Classic hens (18 weeks old) were purchased from a local vendor. Hens were quarantined and housed in a specific pathogen-free environment at the Valladolid Medical School animal facility. All procedures were approved by the Institutional Animal Care and Use Committee. Animals were treated according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Forty-eight hens were anesthetized with an intramuscular injection of ketamine hydrochloride (37.5 mg/kg, Ketolar; Parke-Davis S.A., Barcelona, Spain) and xylazine hydrochloride (5 mg/kg, Rompun; Bayer AG, Leverkusen, Germany). Prior to surgery, one drop of 0.5% tetracaine chlorhydrate and 1 mg of oxybuprocaine (Colircusí Anestésico Doble; Alcon Cusí, S.A., Barcelona, Spain) were applied to the ocular surface. Standardized ablations (68 μm depth, 6 mm diameter) were performed in both eyes using an Apex Plus excimer laser (Summit, Waltham, MA). Prior to ablation, corneal epithelium was removed in a 7-mm premarked area. Triple antibiotic ophthalmic ointment (Vetropolycin, Melville, NY) was applied to both eyes immediately after surgery. Three animals were used as normal controls. Under general anesthesia (37.5 mg/kg, 5 mg/kg), animals were euthanized with an intramuscular injection of 200 mg/kg of pentobarbital sodium at different time points (1, 2, 3, 5, and 7 days after ablation). 

NGF Topical Treatment and Follow-up

Animals were randomized into three groups. Group 1 received one drop (40 μL) of 0.2% murine NGF in balanced salt solution (BSS) (murine mouse β-NGF; prepared/purified according to the standard procedure reported by Bocchini and Angeletti ³⁷ ). Animals in group 2 were treated with BSS as a vehicle control in the same manner, and animals in group 3 served as a second, untreated control. Eye drops were administered immediately after surgery and every 4 hours for 5 days. Nightly rest (8 hours) was permitted for all animals. 

Animals were observed under a surgical microscope (Model OM-5; Takagi, Nakano, Japan) before and after laser ablation and once daily thereafter. The following parameters were evaluated: 

1.  
   Epithelial closure with sodium fluorescein (Fluotest; Alcon Cusí, Barcelona, Spain).
2.  
   Epithelial erosion or ulcers (present or not present).
3.  
   Corneal edema (present or not present).

Determination of Wound Healing Rates

The healing process was monitored every 8 hours with a surgical microscope (Model OM-5; Takagi). Corneas were photographed before and after 2% sodium fluorescein stain with a charge-coupled device camera (ADV-5400CH; Sony, Tokyo, Japan). The rate of epithelial healing was evaluated by measuring the wound size with image-processing software (ImageJ v.1.5; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD; provided in the public domain at http://rsb.info.nih.gov/ij). The total de-epithelialized area was estimated as 38.48 mm² according to the formula A = π ^. r ² (where A = surface, r = radius; 3.5). Data were averaged and analyzed for significant variations. 

Tissue Processing and Light Microscopy

Corneas were removed on days 1, 2, 3, 5, and 7 (n = 6) following ablation and were radially sectioned into equal halves. The epithelium from one half was scraped from the stroma and flash frozen in nitrogen. The other half was fixed in 10% PBS buffered formalin and then embedded in paraffin. Sections (7 μm) were stained with hematoxylin-eosin (H-E) and Masson's trichrome (MT). In addition, immunohistochemistry (IHC) was performed on 7-μm-thick tissue slides. Briefly, sections were deparaffinized and, after washing, cells were blocked with PBS/5% goat serum (Millipore, Billerica, MA). Slides were then incubated overnight with the following primary antibodies: rabbit anti-TrkA (1:1000 in PBS/5% goat serum; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and rabbit anti-MMP-9 (1:1000 in PBS/5% goat serum; Abcam, Cambridge, UK). Biotinylated goat antirabbit IgG (1:200 PBS/5% goat serum; Dako, Carpinteria, CA) was applied for 1 hour and, after washing, slides were incubated for 30 minutes with horseradish peroxidase (HRP)-conjugated streptavidin (Dako). Finally, samples were incubated for 3 minutes with 0.01% diamine benzidine tetrahydrochloride (DAB Substrate Kit for Peroxidase; Vector, Burlingame, CA). To avoid false-positive results, a series of tissue sections were stained, omitting the primary antibody. Also, irrelevant antibodies of the same isotype were compared to ensure specificity. 

Cell Proliferation

One hour before euthanasia, hens received an intramuscular injection of 5 mL/kg of 5-bromo-2′-deoxyuridine (BrdU) (10 mg/mL) (Sigma-Aldrich, St. Louis, MO). Sections were deparaffinized and treated with HCl (2 N, 37°C, 1 hour) and rinsed with tris-buffered saline (TBS). Prior to blocking with 5% goat serum (Sigma-Aldrich) in TBS, sections were incubated with mouse monoclonal IgG anti-BrdU (1:20 dilution in TBS; Dako) for 1 hour at room temperature. A secondary goat antimouse IgG Texas Red–conjugated antibody (Molecular Probes, Leiden, The Netherlands) was used. Sections were examined under an Axiophot fluorescence-incorporated microscope (Zeiss, Oberkochen, Germany) and photographed with a SPOT Digital Camera (Diagnostic Instruments, Inc., Sterling Heights, MI). Five photographs of different areas were taken in each cross-section/slide: (1) limbus-peripheral, (2) paracentral, (3) central, (4) paracentral, (5) peripheral-limbus. The number of epithelial BrdU-positive cells in the photographs was blind-hand-counted. 

Western Blot

Corneal epithelium was homogenized in lysis buffer (0.5 M Tris-HCL, pH 6.8, 20% glycine, 10% SDS) for TrkA, and lysis buffer plus 5% β-mercaptoethanol for MMP-9 in the presence of a protease inhibitor (Complete Mini protease inhibitor cocktail tablets; Roche, Indianapolis, IN). Samples were equally resolved on a denaturing 10% SDS-polyacrylamide gel and later transferred to a nitrocellulose membrane (Amersham, Piscataway, NJ). Membranes were blocked with 5% milk in TBS and then incubated at 4°C overnight, with the following primary antibodies: rabbit anti-TrkA (1:1000; Santa Cruz Biotechnology, Inc.), anti-β-actin (1:1000; Santa Cruz Biotechnology, Inc.), and rabbit anti-MMP-9 (1:1000; Abcam), and an HRP-conjugated secondary donkey antirabbit IgG antibody (1:5000; Jackson, West Grove, PA). Immunoreactive bands were visualized on the Lumino Analyzer LAS1000 (Fuji Film, Tokyo, Japan) using enhanced chemiluminescence Western blotting detection reagents (Millipore). ImageJ was used to measure intensity of the bands (ImageJ v.1.5; developed by Wayne Rasband, National Institutes of Health; provided in the public domain at http://rsb.info.nih.gov/ij). 

Tissue Culture

Telomerase-immortalized HCLEs were kindly provided by Ilene Gipson, PhD. ³⁸ Cells were grown until confluent in keratinocyte serum-free medium (Life Technologies, Carlsbad, CA) supplemented with 0.4 mM CaCl2, 25 μg/mL bovine pituitary extract, 0.2 ng/mL epidermal growth factor, 1% penicillin/streptomycin (hereafter referred to as KSFM), and maintained at 37°C in 5% CO2. Cells were passed and plated in either 96-well plates (3 × 10³ cells/100 μL/well) or 12-well plates (3 × 10⁴ cells/1 mL/well) using the same medium described above. 

Proliferation and Viability Assay

Cells were plated in 96-well plates (3 × 10³ cells/100 μL/well) using the medium described above (KSFM) alone, or challenged with either concentrations of NGF (50, 100, 250, and 500 ng; No. 354009; BD Bioscience, San Jose, CA), 250 ng NGF plus TrkA inhibitor (sc-311,553, 1 μg/mL; Santa Cruz Biotechnology, Inc.), or 250 ng NGF plus mitomycin C (MMC) (0.02%). A cell proliferation assay kit (Millipore) was used for in vitro quantitative detection of newly synthesized DNA of actively proliferative cells. As per the manufacturer's instructions, BrdU was added to the cell culture for up to 12 hours. One day after plating, the medium was carefully aspirated and the culture cells fixed following the protocol provided. After three washes, 100 μL/well mouse anti-BrdU–diluted antibody was added for 1 hour at room temperature. Next, a goat antimouse IgG peroxidase conjugate (1:2000) was added for 30 minutes and cells were incubated with 100 μL tetramethyl benzidine (TMB) peroxidase substrate for 30 minutes. The reaction was stopped with phosphoric acid (1 M), and absorbance (450 nm) was read with a Tecan GENios microplate reader (Tecan, Männedorf, Switzerland). 

For the viability assay, cells were plated using the same method as described above. The experiment was performed as per the manufacturer's instructions using a cell counting kit-8 (CCK-8) (Dojindo, Kumamoto, Japan). After 24 hours in the incubator, 10 μL CCK-8 solution was added to each well of the plate. The plates were placed back in the incubator and, after 2 hours, absorbance (450 nm) was read in a Tecan GENios microplate reader (Tecan). Each group consisted of six wells and both experiments were repeated in triplicate. 

Migration Assay

The effects of NGF on corneal epithelial migration were measured over HCLE confluent monolayers in 12-well plates. The confluent monolayer was wounded manually by scraping using a 1-mL blue tip. Briefly, the tip was positioned near to the border and perpendicularly to the plate. Next, the tip was manually drawn from one side through the center to the opposite side of the plate. Only plates with a wound width of approximately 1 μm were used. Medium was aspirated and replaced with fresh KSFM alone, or challenged with concentrations of NGF (50, 100, 250, and 500 ng), 250 ng NGF plus TrkA inhibitor, or 250 ng NGF plus a concentration of MMP-9 inhibitor I (5 nM, 1 nM, or 0.01 nM; No. 444278; Millipore). Plates were placed in a Leica DMI6000CS (Leica Microsystems, Wetzlar, Germany) inverted microscope with an incorporated live cell chamber (Ludin cube) and maintained at 37°C in 5% CO2 for up 2 days. The microscope was programmed to acquire pictures every 10 minutes from the center of the plate (×20 magnification). Movie files were automatically generated using ALAS AF software (Leica Microsystems). Using the same software, measurements were taken in different areas within the most advanced points independently whether single cells or leading edge. Each group consisted of three plates and experiments were repeated in triplicate. 

Immunocytochemical Detection

After aspirating the culture medium, cells were washed with cold PBS and prefixed with 1% paraformaldehyde (PFA) for 2 hours. Then cells were fixed with chilled 100% methanol and kept overnight at −20°C. After washing, cells were blocked with tween tris buffered saline (TTBS)/5% goat serum (Millipore). The following primary antibodies were used (incubated at 4°C overnight): MMP-9 (whole molecule-ab38898, 2 μg/mL in TTBS/5% goat serum; Abcam), TrkA (sc-118, 5 μg/mL in TTBS/5% goat serum; Santa Cruz Biotechnology, Inc.), and β4 (clone 346-11A, 5 μg/mL in TTBS/5% goat serum; BD Pharmingen, San Diego, CA). Secondary antibodies, goat antirabbit IgG antibody, and HRP conjugate (AP307P, 5 μg/mL; Millipore), were incubated for 30 minutes. After washing, cells were incubated with TMB/E single reagent, blue color, HRP (Millipore) for 30 minutes. The reaction was stopped with phosphoric acid (1 M) and absorbance (450 nm) was read using a Tecan GENios microplate reader (Tecan). Each group consisted of three plates and experiments were repeated in triplicate. 

Western Blot and Zymography

Culture medium was aspirated and fast frozen for zymography gelatinase assay. Cells were immediately homogenized in lysis buffer plus 5% β-mercaptoethanol in the presence of a protease inhibitor (complete mini protease inhibitor cocktail tablets; Roche). Samples were equally resolved by SDS-PAGE, transferred to the membrane, and visualized as described above. Membranes were blocked with 5% milk in TBS and then incubated at 4°C overnight, with the following primary antibodies: goat anti-β4 integrin (1:1000, C-terminus; Santa Cruz Biotechnology, Inc.), anti-β-actin (1:1000; Santa Cruz Biotechnology, Inc.), rabbit anti-MMP-9 (1:1000; Abcam), and an HRP-conjugated secondary donkey antigoat or antirabbit IgG antibody (1:5000; Jackson). 

For zymography assay, equal amounts of culture medium were resolved using ReadyGel Precast Zymography Gels (Bio-Rad, Hercules, CA) following manufacturer instructions. After 16 hours of incubation at 37°C, bands were developed, and gels stained with Coomassie blue were washed and photographed. The experiment was repeated three times using different medium from three experiments. 

Real-Time PCR

Differential MMP-9 expression was confirmed using quantitative real-time PCR (qPCR). Total RNA was isolated with Trizol (Invitrogen, Grand Island, NY) followed by purification using the RNeasy Microkit (Qiagen, Valencia, CA). The complementary strand of DNA (cDNA) was generated by employing SuperScript III Reverse Transcriptase (Invitrogen) and random hexamer primers (Invitrogen). Quantitative real-time reaction was performed in duplicate with the same concentration of cDNA using Taqman Universal PCR Mastermix and FAM-MGB dye-labeled predesigned primers for MMP-9 (Hs00957559_g1; Applied Biosystems, Inc., Foster City, CA). Glyceraldehyde 3-phosphate dehydrogenase (Hs99999905_m1) was used as an internal control. Differential gene expression was calculated according to the Comparative Ct method, as outlined in Applied Biosystems User Bulletin 2 (updated, 2001). The relative expression levels of each sample were expressed as fold change relative to the normal control. Each group consisted of three plates and both experiments were repeated in triplicate. 

Statistical Analysis

Statistical analyses included 1-way ANOVA and Bonferroni's multiple comparison test, in addition to the two-tailed Student's t-test. SEM and SD of the mean were also calculated. A P value less than 0.01 was considered statistically significant and less than 0.001 highly statistically significant. Results are reported as mean ± SD (Prism 5.0; GraphPad Software, La Jolla, CA). 

Results

NGF Induces Epithelial Migration In Vivo and In Vitro

The capability of NGF to promote corneal re-epithelialization has been well documented.^2–6 To confirm this effect in our model, we analyzed the influence of NGF on epithelial wound healing by measuring epithelium closure with sodium fluorescein: resurfacing was significantly faster in corneas treated with NGF compared with both controls (BSS and untreated; P < 0.01; Fig. 1A). Differences were highly significant beginning from hour 16 until the epithelium closed. NGF-treated corneas reached epithelium confluence 8 to 16 hours faster than controls (Fig. 1B). The migration process was monitored in vitro using a wound-healing experiment designed to reproduce in vivo conditions (Fig. 2A). Significantly faster migration was observed in plates supplemented with NGF compared with the KSFM controls (P < 0.01; Fig. 2B). The effects of NGF were observed in a dose-dependent fashion. Based on these experiments, we selected a concentration of 250 μg/mL for further experiments. The addition of a TrkA inhibitor (1 μg/mL) suppressed the effects of 250 ng NGF (Fig. 2B). However, the most provocative result was the real-time observation of single cell spreading and migration patterns in cultures supplemented with NGF. Immediately (2 hours) after NGF contact, the cell monolayer became disaggregated (Fig. 2A; Supplementary Videos S1, S2). The cells moved 100 μm/h individually to the opposite side of the edge, crossing approximately 1 mm in 10 hours. In KSFM plates, cells moved significantly slower as a unit, and single-cell migration was observed only when the edges were nearly confluent. 

Figure 1

 

(A) Effect of NGF on epithelial wound closure in vivo. The time for reepithelialization was significantly reduced in the NGF group relative to both BSS and untreated controls (P < 0.01) (B). A de-organized nonconcentric but faster pattern of migration was observed in NGF corneas compared with the concentric pattern observed in both controls. (C) Corneal epithelial proliferation is shown 48 hours after surgery. More BrdU-positive cells (red) can be seen in NGF cross-sections (a) compared with both controls (b, c). Differences were statistically significant at 48 hours and 3 days (P < 0.01) (D). (E) MT stained cross-sections show a thicker epithelium with hypertrophic basal epithelial cells as well as stromal repopulation in the NGF-treated corneas (d) compared with controls that exhibit a thinner epithelium and lack of keratocytes (e, f). **P < 0.01, ***P < 0.001.

Figure 2

 

In vitro epithelial migration and wound healing. (A) The time to confluence was significantly reduced in plates supplied with 250 ng NGF relative to KSFM controls (P < 0.01). (B) The speed of migration was substantially higher (∼100 μm/h) compared with controls (40–50 μm/h) (P < 0.01). The addition of a TrkA inhibitor considerably decreased the effect of exogenous NGF. (C) Epithelial proliferation is shown by optical density 24 hours after scratching. More BrdU incorporation was observed, in a dose dependent manner, in plates supplied with NGF; however, differences were not significant compared with KSFM (P > 0.05). ***P < 0.001.

NGF Induces Epithelial Proliferation

Corneal epithelium wound repair involves the flattening and elongation of cells during migration and also proliferation to repopulate the wound area. ^39,40 Migration and proliferation are two compartmentalized responses. Cells migrating to cover the wound do not proliferate; however, distal epithelial cells increase their proliferative rate. ^41,42 Intramuscular injection of BrdU, 1 hour before euthanasia, provides the ratio of corneal epithelial cell proliferation in vivo. More proliferation was observed in the periphery and limbus than surrounding the wound at 24 hours (data not shown), but differences were not significant between groups (P > 0.05; Fig. 1D). However, at 48 (Fig. 1C) and 72 hours, differences were significantly higher between the NGF group and both controls (P < 0.01; Fig. 1D). Consequently, a thicker epithelium was observed at days 5 and 7 in NGF-treated corneas compared with both controls (Fig. 1E). 

Similar results were obtained in vitro: after plating 3 × 10³ cells/100 μL/well in a 96-well plate, more cell density was observed on microscopy in wells supplemented with higher concentrations of NGF compared with wells containing only KSFM (Fig. 3A). This observation was confirmed using a CCK8 kit assay (P < 0.01; Fig. 3B). The addition of a TrkA inhibitor repressed NGF effects, and MMC decreased viability (Fig. 3B). A parallel proliferation assay using BrdU incorporation was carried out and similar results were obtained. Significantly more proliferation was detected in wells supplemented with higher concentration of NGF compared with the KSFM plates (P < 0.01; Fig. 3C). As noted above, adding a TrkA inhibitor suppressed NGF effects and MMC inhibited proliferation. Also, we investigated if NGF would induce more proliferation during wound healing in vitro after scratching. More proliferation was observed in plates complemented with NGF, but differences compared with KSFM plates were not significant at 24 hours (P > 0.05; Fig. 2C). 

Figure 3

 

In vitro epithelial cell proliferation 24 hours after plating (3 × 10³ cells/100 μL/well). (A) Higher cell density was observed by microscope in plates supplemented with higher concentrations of NGF (500 ng) relative to KSFM controls. (B) CCK8 kit assay of cell viability 24 hours after plating is shown. The number of viable cells was significantly higher in plates supplemented with higher doses of NGF relative to KSFM (P < 0.01). A TrkA inhibitor reverted the effect of exogenous NGF and MMC compromised viability. (C) BrdU incorporation measured by ELISA. Significant differences can be observed in NGF-supplemented plates compared with KSFM controls (P < 0.01). The addition of TrkA inhibitor reverted the effects of exogenous NGF and MMC suppressed proliferation. **P < 0.01, ***P < 0.001.

NGF Increases Expression of TrkA Receptor in Epithelial Cells

NGF promotes stem cell self-renewal through TrkA. ^43,44 We herein investigated if the effects of NGF are accompanied by an increase in TrkA expression. Both IHC cross-sections and Western blot (WB) analysis revealed an increase in TrkA expression in corneas treated with NGF compared to controls (P < 0.01; Figs. 4A, 4B, 4C). Similar results were observed in vitro in both proliferation and migration assays (Figs. 4D, 4E). A remarkable increase in TrkA expression was detected 24 hours after cells were plated in a 96-well plate with NGF, compared with those without added NGF (P < 0.01; Fig. 4D). A similar trend was observed in wound-healing conditions: TrkA expression increased significantly during the wound-healing process in plates supplemented with NGF relative to KSFM alone (P < 0.01; Fig. 4E). The addition of a TrkA inhibitor to the culture medium reduced the effects of NGF (P < 0.01; Figs. 4D, 4E). 

Figure 4

 

Epithelial expression of the TrkA receptor. (A) Representative WB image showing a significant expression increase in TrkA in the epithelium of corneas treated with NGF, compared with both normal healthy corneas and controls. Results were normalized to β-actin and compared to baseline expression of normal healthy corneas (B). TrkA was significantly upregulated in response to exogenous NGF (P < 0.01). Increased TrkA staining can be observed in the epithelium 5 days after surgery (C). NGF-treated corneas (2) show a stronger stain compared with both controls (3, 4) and normal healthy corneas (1). Notice that NGF epithelium staining (2) is observed in all epithelial layers, whereas in controls (3, 4), staining is limited to the most basal epithelial layers. In vitro TrkA expression was measured by ICC during proliferation (D) and migration (E). Under both conditions, TrkA expression was upregulated by exogenous NGF (P < 0.01). ***P < 0.001.

MMP-9 Is Briefly Upregulated in Response to Exogenous NGF

MMP-9 expression is increased after injury in the edge of the migrating epithelium. ⁴⁵ Assuming that NGF induces MMP-9 expression ⁴⁶ by activating the TrkA receptor, ^47,48 we wondered if the cell spreading observed in our experiment in response to NGF was MMP-9 mediated. Cross-sections revealed predominant expression in the epithelium within the wounded area (Fig. 5A). To assess the amount of MMP-9 involved in wound healing, corneal epithelium was removed by scraping and MMP-9 expression was analyzed by WB. A significant increase in the active form (75 kDa) was found in hen corneas treated with NGF relative to controls (P < 0.01; Fig. 5B). Pro-MMP-9 (92-kDa band) was not detectable in hen corneal epithelium. The increase of MMP-9 activity in vivo in response to NGF was confirmed in vitro. Upregulation of pro-MMP-9 was dramatically increased in cultures supplemented with NGF (P < 0.01; Fig. 5C). Using RT-PCR, we confirmed that MMP-9 gene expression was also increased. Higher levels of MMP-9 mRNA were observed after addition of NGF during wound healing in vitro (P < 0.01; Fig. 5F). Addition of a TrkA inhibitor significantly reduced NGF effects on MMP-9 gene expression. In addition, active MMP-9 (88 kDa) was observed higher in both cell lysates (WB) and culture medium (zymography) from plates supplemented with NGF (P < 0.01; Figs. 5C, 5D). 

Figure 5

 

MMP-9 epithelial expression. (A) MMP-9 staining increased in the epithelium 24 hours after surgery (white arrows). NGF-treated corneas (2) show stronger staining compared with controls (3, 4) and normal healthy corneas (black arrowheads; [1]). (B) Representative WB image showing the expression of active MMP-9 in the epithelium of hen corneas treated with NGF compared with both normal healthy corneas and controls. Results were normalized to β-actin and compared to the baseline expression of normal healthy corneas. MMP-9 was highly upregulated in response to exogenous NGF (P < 0.01). Only the 75-kDa band was observed. (C) Representative WB image showing the significant increase in the expression of both pro and active MMP-9 in response to NGF in culture cells after scratching. Both Pro-MMP-9 and active MMP-9 were highly increased in cells fed with NGF (P < 0.01). (D) Gelatin zymography from culture medium showing MMP-9 and MMP-2 gelatinase activity. MMP-9 gelatinase activity was increased significantly after wounding (3 and 6 hours) in aspirated medium from NGF plates with respect to controls (P < 0.01). (F) MMP-9 gene expression was measured by RT-PCR at 3, 6, and 24 hours after scratching. MMP-9 gene expression was observed more quickly and increased following exogenous NGF administration (P < 0.01). This NGF effect was diminished by a TrkA inhibitor. **P < 0.01, ***P < 0.001.

MMP-9 Cleavages α6β4 Integrin Promoting Epithelial Spreading and Migration

MMP-9 cleaves both α6β4 integrin in hemi-desmosomes (HDs) ⁴⁹ and desmoglein in desmosomes, ⁵⁰ breaking down adherens junctions, desmosomes, and tight junctions, ^51,52 which may promote cellular spreading. By immunocytochemical detection (ICC) we observed a big decrease in the amount of β4 integrin in plates fed with NGF compared with controls in a dose-dependent manner (P < 0.01; Fig. 6A). By blocking TrkA receptor or MMP-9, the amount of β4 integrin remained insignificantly unchanged compared with normal plates. However, when concentrations of the MMP-9 inhibitor were reduced to either 1 nM or 0.1 nM, the presence of β4 integrin significantly decreased. To probe that reduction of β4 integrin was a result of proteolysis, we assessed WB using a C-terminus antibody. Native and four additional proteolytic products of β4 integrin were detected (Fig. 6B). A significant reduction in the 205-kDa band appeared at 3 and 6 hours after wounding in NGF-fed cells. The band is reestablished again after 24 hours. By contrast, no relevant decrease in the 205-kDa band is observed in the KSFM, NGF plus anti-TrkA, or NGF plus proteinase inhibitor groups. As expected, a tremendous increase in the 100-kDa band appeared at 3 and 6 hours in the plates supplemented with 250 ng NGF. Interestingly, a new 180-kDa band was visible at 3 and 6 hours only in the NGF cells. A homogeneous band (160 kDa) with no significant differences between groups and control is observed, suggesting a transitory proteolytic product. Finally, a light increase in the 50-kDa band is observed also in the NGF cells. Proteolytic activities were significantly reverted by addiction of anti-TrkA or MMP-9 inhibitor. 

Figure 6

 

Reduction of β4 integrin in response to NGF. (A) ICC shows a drastic reduction in the β4 integrin staining in response to NGF relative to both normal and KSFM controls (P < 0.01). A TrkA inhibitor reverted the effects of exogenous NGF on β4 presence. The addition of different concentrations of a MMP-9 inhibitor reverted the effects of exogenous NGF in a dose-dependent manner. (B) A representative immunoblot showing different products as result of proteolytic cleavage of β4 integrin in response to NGF. Five different bands were detectable using a C-terminus antibody. A strong reduction in the native 205-kDa band is observed at 3 and 6 hours after scratching in the plates treated with 250 ng NGF (asterisk). A new 180-kDa band appeared at 3 and 6 hours only in cells supplemented with NGF. A third band (160 kDa) was also detectable with no significant differences between groups. However, a fourth band (100 kDa) shows up to 8-fold increase in this product 6 hours after scratching in NGF-treated cells. Finally, a fifth cleavage product (50 kDa) was increased at 6 hours in the same group. Both proteolytic activities were reverted by addiction of anti-TrkA or MMP-9 inhibitor. ***P < 0.001.

Discussion

NGF has been observed to improve epithelium reestablishment in patients with neurotrophic ulcers ^2–5 or after cataract surgery. ⁶ Also, animal studies have shown that NGF increases corneal cell proliferation in rabbits and dogs. ^10,11 In the present study, we used hens as a better model to reproduce in vivo human corneal wound healing. ^27–36 In addition, we tested the in vitro effects of NGF using Telomerase-immortalized HCLEs in contrast to previous works concentrated on the effect of NGF on corneal fibroblasts and conjunctiva epithelial cells. ^46,47,53  

Epithelial resurfacing occurred significantly faster in corneas treated with NGF compared with controls (P < 0.01). Furthermore, epithelial proliferation increased significantly in NGF-treated corneas (P < 0.01). These results clearly suggest that epithelial cells respond quickly to NGF, which promotes wound healing. 

NGF activation of TrkA leads to migration and promotes both cell survival and proliferation. ¹⁴ NGF acts as an autocrine and paracrine factor supporting stem cell self-renewal in corneal epithelial progenitor cells. ^43,44 In the present work, TrkA expression was remarkably enhanced in injured corneas compared with normal healthy tissue. Upregulation of TrkA increased significantly in NGF-treated corneas relative to controls (P < 0.01). These results agree with previous works showing upregulation of TrkA expression in response to NGF. ^46,47,53  

During corneal epithelial resurfacing, cells migrating to cover the wound area drastically reduce proliferative activity. However, cells distal to the original wound exhibit a greatly enhanced level of proliferative activity. ²³ Accordingly, our in vivo model showed a healing process mediated by epithelial migration rather than proliferation. In a previous work, we described that hen corneal epithelial cells start to proliferate 12 hours after photorefractive keratectomy surgery. ³⁰ In the current work, faster epithelial resurfacing was observed in corneas treated with NGF but increase in the proliferation rate was observed at 48 hours when the cornea was already covered by a new epithelium. In consequence, thicker epithelium was observed in those treated corneas. 

To establish a possible mechanism by which cells migrate faster in vivo, we reproduced the process in vitro using HCLEs. As in the in vivo studies, faster migration was observed in plates supplemented with NGF compared with the KSFM controls (P < 0.01) in a dose-dependent manner. Interestingly, NGF had a very similar effect on migration rate in stromal fibroblasts, ⁴⁷ as we observed in corneal epithelial cells. In both cell types, 250 ng/mL NGF was the most efficient concentration (Fig. 2B, Supplementary Material S1). Real-time observations show that immediately after scratching, cells fed with NGF separate from each other and frenetically move to the opposite side. NGF enhances cell proliferation of seeded single cells (P < 0.05); however, NGF did not promote proliferation of injured confluent epithelial cells (P > 0.05). 

Corneal epithelial migration is promoted by upregulation of MMP-9 at the edge of migrating epithelium. ⁴⁵ The role of MMP-9 on wound healing is controversial, because both too little and too much activity can lead to pathology. ⁵⁴ NGF induces MMP-9 expression by activating the TrkA receptor. ^46–48 In the present experiment, MMP-9 was upregulated in corneal epithelium during the wound-healing process in vivo, with higher expression in the NGF-treated group compared with controls (P < 0.01). It is know that excess of MMP-9 activity might cause erosions or ulceration; however, we followed up with the animals for a period of six months and all the corneas remained healthy (data not shown). The tearing might wash out leftover NGF, which explains that controlled MMP-9 activity promotes faster migration but not erosion. This was confirmed in vitro, where higher amounts of MMP-9 were observed in the plates treated with NGF in a dose-dependent manner compared with those treated with KSFM (P < 0.01; Supplementary Material S2). These results are in accordance with those obtained by Micera et al. ⁴⁷ in keratocytes-fibroblasts. In addition, we observed an MMP-9 time-dependent response: NGF significantly upregulates MMP-9 gene expression immediately (3 hours) compared with KSFM controls (P < 0.01). However, MMP-9 gene expression returns to normal after 6 hours. These findings strongly suggest that NGF released by nerves and epithelial cells following injury induces focal expression of MMP-9 to help epithelial migration but also why its topical application improves healing in patients with neurotrophic ulceration. When the epithelial barrier is reestablished, NGF is relapsed, avoiding excessive MMP-9 activity and consequently preventing ulceration. This was confirmed with both WB and zymography assays. The MMP-9 active form was much more elevated in the NGF-fed cells at 3 and 6 hours with respect to controls (P < 0.01). As expected, active MMP-9 was observed as normal 24 hours later when the epithelium was 100% confluent. The fact that pro-MMP-9 was upregulated by NGF does not explain its activation and further experiments are needed to elucidate MMP-9 activation in response to exogenous NGF. 

The α6β4 integrin is upregulated in the epithelium within the injured area, promoting cell attachment during late stages of the wound-healing process. ⁵⁵ However, it has been described that adherens junctions, desmosomes, and tight junctions are less abundant toward the leading edge at early stages of this process. ^51,52 Next, MMP-9 cleaves α6β4 integrins in HDs ⁴⁹ and desmoglein in desmosomes, ⁵⁰ promoting cell spreading. The amount of β4 integrin (detected by IHC using an ectodomain antibody) decreased significantly after scratching in plates supplemented with NGF (P < 0.01), suggesting an extracellular cleavage of the molecule. This was confirmed using an antibody against the C-terminus cytoplasmic tail of β4 integrin. We observed an increase in different products of β4 integrin as a result of proteolytic activity. Fragments higher than 100 kDa, are products of extracellular proteases, but fragments smaller than 100 kDa are a result of intracellular proteases. ⁴⁹ The 100k-kDa band was increased, indicating extracellular cleavage of β4 integrin in response to NGF. Pal-Ghosh et al. ⁴⁹ described that MMP9 cleavages the β4 integrin ectodomain leading to recurrent epithelial erosions in mice. The current work shows that MMP-9 is upregulated, released, and activated in response to NGF, but thereafter is downregulated to baseline state. This clearly demonstrates that MMP-9 might promote healing or lead pathology depending on growth factors, homeostasis, and understanding the mechanisms for controlling the pattern of MMP-9 expression and activation will be a future challenge. ⁵⁴  

The data herein obtained using both in vivo and in vitro approaches show that adding NGF induces proliferation and migration of epithelial cells in vivo as well as in vitro, NGF effects on proliferation appear directly regulated through an increase in its receptor, TrkA; however, the effects on migration are also mediated and enhanced by MMP-9 upregulation and consequently the cleavage of α6β4 integrin. 

Supplementary Materials

pdf S3 

mov S1 

mov S2 

Acknowledgments

The authors thank Ilene Gipson for the gift of the HCLE cell line, Marta Gonzalez for technical assistance, and HyunSoo Lee and Kishore Reddy for helping with RT-PCR and cell culture. 

Supported by Grant FIS-PI: PIO52841. 

Disclosure: T. Blanco-Mezquita, None; C. Martinez-Garcia, None; R. Proença, None; J.D. Zieske, None; S. Bonini, None; A. Lambiase, None; J. Merayo-Lloves, None