Human Corneal Epithelial Cells Require MMP-1 for HGF-Mediated Migration on Collagen I

Abstract

purpose. To investigate the potential regulation of matrix metalloproteinases (MMPs) by hepatocyte growth factor (HGF), and to identify individual MMPs essential for migration of human corneal epithelial cells.

methods. Migration of human corneal epithelial cells (HCECs) was measured with a colony dispersion assay in response to concentrations of HGF (0–50 ng/mL). MMP activity in the conditioned media collected from the dispersion assay was assessed by zymography. The broad-spectrum MMP inhibitor ilomastat (1–100 μM) or an MMP-9–neutralizing antibody (1–10 μg/mL) were included in the dispersion assay to determine their effects on HCEC migration. Immunocytochemistry and in situ hybridization were used to localize MMP-1 in HCECs in the colony dispersion assay and in a human ex vivo corneal wound-healing model, respectively. ELISA for MMP-1 was performed on conditioned medium from migrating HCECs. Neutralizing antibodies to MMP-1 and -9 were added to an in vitro scratch-wound model to assess the effect on HCEC healing.

results. HCEC migration (P < 0.05) and MMP-2 and -9 released into the medium increased in response to HGF in a dose-dependent manner up to 20 ng/mL. Broad-spectrum MMP inhibition significantly reduced HCEC migration (P < 0.05). In contrast, neutralization of MMP-9 increased migration (P < 0.05). MMP-1 was found in association with HCECs at the migratory leading edge in both the dispersion and the ex vivo wound-healing experiments, and was found to be stimulated above basal levels by HGF. Neutralization of MMP-1 significantly decreased (P < 0.05), whereas neutralization of MMP-9 significantly increased (P < 0.05), scratch-wound closure.

conclusions. This study provided novel data regarding HCEC migration in response to HGF and highlighted the importance of MMPs, particularly MMP-1 in migration and possibly reepithelialization in vivo. MMP-9 and/or -2 may be released by HCECs to remodel matrix behind the leading migratory front. Studies such as this are essential to assist in the safe and efficacious design of MMP inhibitors for therapeutic use in the eye.

The matrix metalloproteinases (MMPs) are a family of extracellular-matrix-degrading enzymes. ¹ The participation of MMPs in a variety of diseases has been documented in various processes, including ulcers, ^{2 3} rheumatic disorders, ⁴ and tumor angiogenesis, and metastasis. ⁵ In the cornea, MMPs including MMP-9 (gelatinase B), -2 (gelatinase A), and -1 (collagenase) have been documented in the pathogenesis of ulceration ^{6 7 8} and pterygia. ⁹ Despite playing a destructive role in many diseases, it is becoming apparent that MMPs are essential to normal cellular functions, such as movement. ¹⁰  

After corneal injury, epithelial cells migrate to cover the wound bed before differentiating into new multilayered epithelium. This reepithelialization process is essential to prevent potentially blinding ulceration and scarring. ¹¹ Previous animal model studies have demonstrated the presence of MMP-9 in migrating epithelial cells after injury. ¹² We have also recently demonstrated, using a human ex vivo corneal tissue, that, after injury, migrating epithelial cells express MMP-1, -9, and -10 during reepithelialization over the stroma. ¹³ The necessity of individual MMPs in this process is still not fully understood. 

Previous reports have demonstrated the requirement of MMP-1 for cutaneous keratinocyte migration on type I collagen. ^{14 15} Because collagen I is the most abundant extracellular matrix molecule in the corneal stroma, we wanted to assess whether MMP-1 is necessary for reepithelialization of corneal wounds involving damage to the basement membrane and exposure of the underlying stroma. 

Although the potential involvement of MMPs in the epithelial cell migration process is very important, the cells also need chemotactic stimuli to move. In skin, hepatocyte growth factor (HGF) stimulates expression of MMP-1 and -3 in migrating keratinocytes in a dose- and matrix-dependent manner, whereas a splice variant of HGF (HGF/NK2) inhibits MMP-1 synthesis. ¹⁶ HGF, also known as scatter factor, is a fibroblast-derived protein causing separation of contiguous epithelial cell sheets. ^{17 18} HGF provides a motogenic, rather than mitogenic, signal to many types of epithelial cells, including those in the lens and respiratory tract. ^{19 20} The role of HGF in the eye has been extensively reviewed. ^{21 22} During corneal wound healing, HGF was found to be upregulated in the tissue of rabbits and tears of humans after anterior segment surgery. ²³ In a mouse model, HGF mRNA levels were markedly upregulated by keratocytes (stromal fibroblasts) and remained elevated for at least 7 days after epithelial wounding. ²⁴ HGF is produced by fibroblasts in the peripheral cornea, whereas the receptors for HGF (c-Met) are predominantly expressed by epithelial cells. ¹⁸ In addition, HGF has been reported to increase the rate of corneal reepithelialization in organ cultured rabbit corneas. ²⁵ These data make HGF a likely candidate for stimulating human corneal epithelial cell movement after injury, therefore HGF was chosen as the chemotactic stimuli for this study. 

Previous studies have shown that MMP-9-mediated cutaneous keratinocyte migration on type I collagen is stimulated by HGF, ²⁶ but so far no evidence linking MMP-mediated epithelial cell migration and HGF in the cornea is available. On this basis, it is essential to understand the role and importance of individual MMPs in vital processes such as epithelial cell migration before potential therapeutic wound-healing modulating agents, such as MMP inhibitors, can be safely used. In this study, we investigated the potential regulation of MMPs by HGF and sought to identify the individual MMPs that are essential for human corneal epithelial cell migration. 

Methods

Cells and Culture

Human corneas supplied by Moorfields Eye Hospital National Health Service (NHS) Trust Eye Bank were used in the study, in accordance with the Declaration of Helsinki for the use of human tissue. Normal corneas became available for research purposes from cold storage culture within 48 hours after death, because a low endothelial cell count made them unsuitable for transplantation. Human corneal epithelial cells (HCECs) were isolated and cultured in keratinocyte serum-free medium (K-SFM) supplemented with bovine pituitary extract (50 μg/mL), epidermal growth factor (5 ng/mL; all Gibco Life Technologies, Paisley, Scotland, UK), and calcium chloride (0.03 mM), as previously described ^{27 28} (without a 3T3 fibroblast feeder layer). Experiments were performed with cells between passages 2 and 3. 

Colony Dispersion

The dispersion assay described by Pilcher et al., ¹⁴ was used, with modifications, to measure cell migration. HCECs were seeded and cultured to confluence with supplemented K-SFM in cloning rings (Flexiperm; Heraeus Instruments, Brentwood, UK) on acid-extracted rat tail collagen I-coated plates (6 μg/cm²; Sigma-Aldrich Chemical Co., Ltd., Poole, UK). The cells were then cultured for a further 24 hours in the presence of 100 μM hydroxyurea to induce growth arrest ¹⁴ and then starved for 18 hours with K-SFM supplemented with only 1% bovine serum albumin (BSA, wt/vol; Sigma-Aldrich Co., Ltd.). After removal of the rings, the cells were thoroughly washed with phosphate-buffered saline (PBS; Sigma-Aldrich Co., Ltd.) and fed with concentrations of hepatocyte growth factor (0–50 ng/mL; R&D Systems Europe Ltd., Oxon, UK) or epidermal growth factor (0–20 ng/mL EGF; R&D Systems Europe, Ltd.) in the presence of 1% BSA (wt/vol) in triplicate wells. Migration was permitted for 3 days. Conditioned medium was collected, pooled from triplicate wells, and stored in siliconized tubes at −20°C. The cells were washed with PBS three times, fixed with 90% (vol/vol) methanol and stained with Harris hematoxylin (Shandon Life Sciences International, Europe, Ltd., UK). Dispersion areas were photographed (Casio Computer Co., Ltd., Tokyo, Japan) and measured in pixels with image-analysis software (UTHSCSA, San Antonio, TX). 

Zymography

MMP activity in conditioned medium was demonstrated by gelatin zymography (10% zymogram gelatin gels), using the manufacturer’s buffers and instructions (Mini Cell; Invitrogen, Groningen, The Netherlands). Briefly, samples were diluted in sample buffer (1:1) and electrophoresed through gelatin-impregnated zymogram gels at 150 V for 90 minutes. Kaleidoscope molecular weight markers (Bio-Rad, Hemel Hempstead, UK) were also included. The gels were incubated at room temperature in renaturing buffer for 30 minutes and washed in developing buffer for a further 30 minutes. Fresh developing buffer was added, and the gels were incubated for 16 hours at 37°C. Zymograms were stained with 0.5% Coomassie blue (Bio-Rad) for 90 minutes before destaining, until clear bands of MMP activity appeared against a blue background. 

Inhibition of MMP

To evaluate the effects on HCEC migration in the presence of HGF (20 ng/mL) concentrations of the broad-spectrum MMP inhibitor ilomastat ^{29 30} (a generous gift from Glycomed, San Diego, CA; 0–100 μM) diluted in dimethyl sulfoxide (DMSO), neutralizing antibodies to MMP-9 (sheep anti-pig MMP-9 provided by author GM; 0–100 μg/mL), and a commercial preparation (0–10 μg/mL; monoclonal mouse anti-human; Oncogene Research Products, Cambridge, MA) were included in the dispersion assay in triplicate wells. The control for ilomastat contained 1:100 DMSO and the antibody control was mouse or pig γ-immunoglobulin. 

Immunocytochemistry

HCECs migrating in the presence of HGF (20 ng/mL) in the dispersion assay were washed three times with PBS, fixed for 15 minutes at room temperature with 4% paraformaldehyde (wt/vol), and treated with 20% sucrose (wt/vol) before storage at −20°C. Immunocytochemistry for MMP-1, -2, and -9 using mouse anti-human monoclonal antibodies (Cambridge Bioscience, Cambridge, UK) and mouse immunoglobulins as negative controls were performed in triplicate with the streptavidin-avidin-alkaline phosphatase and vector red detection technique (Vector Laboratories, Peterborough, UK), as previously described. ³¹  

Ex Vivo Wound-Healing Model

Seven human corneas were cultured for 4 days with keratinocyte culture medium (KCM) formulated by Rheinwald and Green. ³² The corneas were cut into segments. One segment from each was reserved for histologic confirmation of original tissue integrity (hematoxylin and eosin staining), whereas the rest were wounded by removal of the epithelium with a corneal brush (Algerbrush II; Algerbrush Co., Inc., Lago Vista, TX). The tissues were recultured in KCM and segments removed at intervals to obtain tissue demonstrating reepithelialization. Tissues were fixed with 4% paraformaldehyde at room temperature overnight and dehydrated through a series of various concentrations of alcohol, and embedded in paraffin. Tissue sections (5 μm) were cut onto microscope slides (Superfrost Plus; BDH Laboratory Supplies, Poole, UK). 

In Situ Hybridization

The production and specificity of the antisense human MMP-1 probe has been demonstrated, ^{33 34} and in situ hybridization was performed as previously described. ³⁵ Briefly, samples were pretreated with proteinase K (1 μg/mL; Sigma Aldrich Co., Ltd.) then washed in 0.1 M triethanolamine buffer containing 0.25% acetic anhydride. The sections were covered with 35 μL of hybridization buffer containing 2.5 × 10⁴/μL ³⁵S-labeled antisense or sense RNA probe and incubated at 55°C to 60°C for 18 hours in a humidified chamber. After hybridization, the slides were washed under stringent conditions, including treatment with RNase A to remove unhybridized probe. After 15 to 45 days of autoradiography, the photographic emulsion was developed, and the slides were stained with hematoxylin and eosin. Cutaneous wounds known to express MMP-1 were used as the positive control, and each sample with several sections was hybridized in two different experiments. Sections were viewed with both dark-field and bright-field microscopy and assessed by two independent investigators. 

MMP-1 ELISA

Conditioned medium was collected from the dispersion assay involving HCECs migrating over type I collagen in response to concentrations of HGF (0–50 ng/mL) on day 3. Total MMP-1 protein was measured in the samples with an MMP-1 human ELISA system (Biotrak; Amersham Pharmacia Biotech UK, Ltd., Bucks, UK). Paired t-tests were used to compare the amount of MMP-1 present in the HGF-treated cells with basal production of MMP-1 in K-SFM supplemented with 1% BSA (wt/vol) only. P < 0.05 was considered to be statistically significant. 

Scratch-Wound Model

HCECs cultured to confluence on collagen I (6 μg/cm²) in 24-well plates were starved for 24 hours and then scratch wounded with a 100 μL pipette tip. Concentrations of neutralizing antibodies to MMP-9 (0–100 μg/mL sheep anti-pig MMP-9) and MMP-1 (mouse anti-human monoclonal Chemicon International Ltd., Harrow, UK) were added together with HGF (20 ng/mL). The wound areas in a marked field of view were measured with a graticule at the start and subsequently until closure. An average of five measurements per field of view were recorded. Mouse and sheep immunoglobulins were included in the control wells. 

Statistical Analysis

Experiments were repeated at least three times. One-way analysis of variance (ANOVA) was performed on computer (SPSS for Windows; SPSS Inc., Chicago, IL) unless otherwise stated. The observed significance levels were adjusted with the Bonferroni test for multiple comparisons. P < 0.05 was considered significant. 

Results

Corneal Epithelial Cell Migration and MMP Production in Response to HGF

Increasing HCEC migration on type I collagen correlated with increasing HGF concentrations up to 20 ng/mL (P < 0.05; Fig. 1A ). At 50 ng/mL HGF became inhibitory to migration with dispersion areas not significantly different to the baseline BSA control, and trypan blue dye exclusion confirmed that this inhibition was not due to cytotoxicity (data not shown). The optimal HGF concentration of 20 ng/mL was therefore used in subsequent assays. After the trend for migration, MMP-9 and -2 production also increased in response to concentrations of HGF up to 20 ng/mL and subsequently decreased in the presence of 50 ng/mL (Fig. 1B) . Despite some variation in the actual dispersion areas and MMP production levels between HCEC cultures from different donors, the response trends were reproducible throughout all the assays conducted. 

Corneal Epithelial Cell Migration and MMP Production in Response to EGF

To determine whether MMP production during HGF-mediated HCEC migration was growth-factor-specific the dispersion assay was repeated with EGF. Migration in response to EGF on type I collagen significantly increased with each concentration tested (P < 0.05; Fig. 2A ). The dispersion areas resulting from incubation of HCECs with 20 ng/mL EGF was greater than that with 20 ng/mL HGF. Again, an increase in production of MMP-9 correlated with increased EGF, but unlike that seen with HGF, MMP-2 production levels remained constant (Fig. 2B) . 

Effect of MMP Neutralization on Corneal Epithelial Cell Migration

Increasing concentrations of the broad-spectrum MMP inhibitor ilomastat significantly decreased HCEC migration on type I collagen in response to HGF (20 ng/mL; Fig. 3A ). Trypan blue dye exclusion confirmed that this inhibition was not due to cytotoxicity (data not shown). However, after addition of MMP-9-neutralizing antibodies in the dispersion assay, HCEC migration increased (P < 0.05; Fig. 3B ). A concentration of 10 μg/mL was the most effective with the pig anti-sheep antibody. Similar results were obtained with a commercially prepared MMP-9-neutralizing antibody at a concentration of 10 μg/mL (data not shown). 

Localization of MMP-1 in Migrating Corneal Epithelial Cells In Vitro

Because MMP-1 was not detectable by zymography but is known to be involved in keratinocyte migration on type I collagen, ¹⁴ immunocytochemistry was performed to attempt to detect MMP-1 in the dispersion assay. The leading HCECs migrating away from the edge the of the dispersion colony stained positively for MMP-1 (Fig. 4A) , whereas few cells within the main colony were positive. A few cells located behind the leading edge stained positively for MMP-2 (Fig. 4C) and MMP-9 (Fig. 4E) . No significant staining for MMP-2 was found within the dispersion colony for MMP-2 (Fig. 4D) , whereas some rounded cells that appeared to be detaching from the surface of the culture intensely stained positively for MMP-9 (Fig. 4F) . 

Production of MMP-1 in Response to HGF

MMP-1 was detected by ELISA in conditioned media collected from HCECs migrating on type I collagen. HGF (10 and 20 ng/mL) significantly increased (P < 0.05) the amount of MMP-1 produced by the cells compared with the basal level present with BSA (Fig. 5) . 

MMP-1 Production by Migrating Corneal Epithelial Cells Ex Vivo

The ex vivo wounded corneal tissue demonstrated reepithelialization over the corneal stroma (Fig. 6A) and the localization of MMP-1 in the leading migratory epithelial cells is shown in Figure 6B . No signal for MMP-1 was detected with the sense control probe (Fig. 6C) . 

Effect of Neutralization of MMP-9 and -1 on Wound Closure by Corneal Epithelial Cells

The addition of the pig anti-sheep MMP-9-neutralizing antibody (10–100 μg/mL) significantly increased the rate of HCEC closure of scratch wounds on type I collagen in response to HGF (20 ng/mL) by 72 hours (P < 0.05; Fig. 7A ). Conversely, the same concentrations of a neutralizing antibody to MMP-1 significantly delayed HCEC wound closure (P < 0.05; Fig. 7B ). Differences occurred in the rate of wound closure between the negative control groups treated with mouse immunoglobulin (for MMP-1) or sheep immunoglobulin (for MMP-9); however, rates of wound closure were dependent on concentration within each treatment group as described. 

Discussion

The purpose of this study was to investigate the potential regulation of MMPs by HGF and the necessity for the presence of individual MMPs during human corneal epithelial cell migration. The in vitro experiments were performed with HCECs seeded on type I collagen. The purpose was to provide the cells with the extracellular matrix stimulus most likely to be encountered by HCECs after injury to the cornea involving the basement membrane. HGF was used as the chemotactic stimulus in the in vitro assays, because it is very likely that this growth factor facilitates reepithelialization in vivo. ^{23 25}  

This study provided the first demonstration that MMP-9 and -2 are increased in a dose-dependent manner during HCEC migration in response to HGF. The farther the cells migrated, the more MMP-9 and -2 were released from the cells into the medium. However, beyond the optimal growth factor concentration for migration, movement of HCECs slowed and production of MMP was also reduced. Our data provide contrasting information to the data in the study by Li et al., ³⁶ which showed that no increase in gelatinase activity occurred in response to HGF. Our dispersion experiments were conducted with HCECs cultured on type I collagen, indicating that the HCECs may also require extracellular matrix signals to alter their gelatinolytic profile in response to concentrations of HGF. Our MMP-9 data correlate with previous findings in cutaneous keratinocytes ²⁶ and suggest a functional link between HGF-stimulated migration and utilization of MMP during cell movement. MMP-2 is constitutively expressed in the cornea during wound-healing; however, its potential regulation in epithelial cell migration by HGF has not been reported before. After repetition of the experiments with EGF, a more potent chemotactic stimulus to HCECs, it was revealed that upregulation of MMP-2 was specific to these cells in response to HGF. 

The MMP inhibitor ilomastat, significantly reduced migration of the HCECs. Even though migration was not completely halted, these data suggest that the involvement of MMPs in HCEC movement was critical. Ilomastat is thought to be a broad-spectrum inhibitor of MMPs. ^{29 30} When this inhibitor was discovered, fewer MMPs had been identified, and it is possible that there are members of the MMP family involved in epithelial cell migration that are not inhibited by ilomastat. 

Previous experiments involving antibody neutralization of MMP-9 have suggested that this MMP is essential for migration of cutaneous keratinocytes on type I collagen in response to HGF ^{26 37} ; however, MMP-9 was not found at the leading edge of migratory HCECs in our experiments. Pulmonary epithelial cells also need MMP-9 for migration. ³⁸ Previous corneal experiments in animals have suggested a role for MMP-9 in basement membrane remodeling. ^{39 40} Our data demonstrated a moderate increase in HCEC migration with both a commercial and in-house MMP-9-neutralizing antibody. In addition, epithelial scratch wounds healed more quickly in the presence of the neutralizing antibody to MMP-9. Our data correlate with those in a recent in vivo study showing that reepithelialization of corneal stroma and skin occurs more quickly in MMP-9-null mice than in wild-type mice. ⁴¹ Alterations in the rate of epithelial cell proliferation have been shown to contribute to the accelerated reepithelialization of cornea and skin in the MMP-9 knock-out mouse. ³⁷ In addition, the MMP-9-null mouse displays defects in the ability to remodel extracellular matrix at the epithelial basement membrane zone. ⁴¹ It is possible that neutralization of MMP-9 in HCECs in vitro prevents efficient matrix remodeling behind the leading migratory cell front, therefore facilitating rapid cell migration as the cells neglect their remodeling function. The location of the epithelial cells stained positively for MMP-2 and -9—that is, behind the leading edge cells—suggests that these MMPs may play a role in remodeling the degraded collagen matrix (gelatin) that perhaps has been cleaved by the leading cells. Apparent differences in the literature regarding the necessity for the presence of individual MMPs for epithelial cell migration to occur may reflect the availability to the cell of several factors, such as cytokines, to perform certain functions. 

Because the reduction of HCEC migration in the presence of the broad-spectrum MMP inhibitor ilomastat cannot be explained by neutralization of MMP-9, further studies are needed. Although it was not possible to observe MMP-1 clearly in culture supernatants on the zymograms that showed MMP-2 and -9, previous work with cutaneous keratinocytes ¹⁴ had suggested its present in HCECs. In fact, this was found to be true. Only the HCECs at the leading edge of the dispersion colonies, those not surrounded by other HCECs, expressed MMP-1. It is possible that similar to cutaneous keratinocytes, ¹⁴ HCECs use MMP-1 to cleave the type I collagen substrate, allowing migration and further remodeling of the matrix by the gelatinases (MMP-2 and -9). Indeed, if MMP-1 is used for this purpose, it may explain why relatively small amounts were detectable in the conditioned medium, therefore preventing large-scale type I collagen degradation by gelatinases. 

It has been suggested that cleavage of the collagen triple helix by MMP-1 gives the migrating epithelial cells directionality. ¹⁴ Our data from studies of ex vivo wounded human corneal tissue confirmed the presence of MMP-1 in the leading HCECs during reepithelialization over stroma. This evidence supports the hypothesis that MMP-1 may be essential for reepithelialization to occur. Chimeric enzyme studies have indicated that both the C- and N-terminal ends of the MMP-1 molecule are required for it to cleave native fibrillar collagens. ^{42 43} The hemopexin domain is thought to participate in the initial binding and orientation of the collagen fibril and local unwinding of the triple helix, allowing subsequent cleavage. ⁴⁴ In fact, recent evidence suggests that the hemopexin domain of MMP-1 interacts with the α₂ domain of the type I collagen receptor α_2β1 integrin, confining the proteinase activity to points of cell contact with collagen. ^{45 46} The scratch-wound analyses suggested that MMP-1 rather than MMP-9 may be an essential component for initiation of HCEC cell movement in response to HGF, in that the highest levels of MMP-1 in conditioned media were detected in association with the farthest-migrating cells. In all likelihood, a precise combination of MMP-1, -2, and -9 activities is needed for optimal maintenance of HCEC migration on type I collagen. It is likely that MMP-1 was induced in HCECs through intracellular signaling, perhaps to initiate migration, in that it is known, for example, that HGF signaling occurs through the Ras-mitogen-activated protein kinase pathway (Ras-MAPK) in HCECs. ⁴⁷  

In conclusion, this study provided novel data regarding HCEC migration in response to HGF. It is possible that MMP-1 is produced to initiate migration by “nicking” the collagen triple helix and that MMP-2 and -9 may serve to remodel the matrix behind the leading migratory front. Although previous in vitro work in keratinocytes has suggested a role for MMP-2 in epithelial cell migration, ⁴⁸ this has not yet been found in vivo. ^{49 50} The apparent novel regulation of MMP-2 by HGF during HCEC migration is unclear and warrants further investigation. 

Finally, this study has highlighted the importance of MMPs, particularly MMP-1 in HCEC migration and possibly reepithelialization in vivo. Although MMP-1 has been shown to play a pathologic role in corneal disease associated with rheumatoid arthritis, ⁶ its apparently essential role in HCEC migration should not be overlooked during the design of potentially therapeutic MMP inhibitors for corneal or other diseases. 

 Supported by The Royal National Institute for the Blind, the Eranda Foundation, the Hayman Trust (JTD), The Wellcome Trust (GAL), the Academy of Finland, the Sigrid Juselius Foundation, and Helsinki University Central Hospital Research Funds (US-K). JTD and PTK received a proportion of their funding from the National Health Service Executive. The views expressed in this article are those of the authors and do not necessarily reflect those of the grant-awarding bodies.

 Submitted for publication May 3, 2002; revised July 30, 2002; accepted September 20, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) HCEC migration at day 3 in response to various concentrations of HGF in a colony dispersion assay. Error bars, SEM. *Significant differences (P < 0.05) between HGF concentrations and basal rates of migration. (B) Zymogram showing the production of MMP-2 and -9 during migration in response to HGF.

Figure 2.

 

HCEC migration at day 3 in response to various concentrations of EGF shown in a colony dispersion assay (A). Error bars, SEM. *Significant differences (P < 0.05) between EGF concentrations and basal rates of migration. (B) Zymogram showing the production of MMP-2 and -9 during migration in response to EGF.

Figure 3.

 

HCEC migration at day 3 in response to HGF (20 ng/mL) as shown in a colony dispersion assay in the presence of concentrations of the broad-spectrum MMP inhibitor ilomastat (A) or MMP-9 neutralizing antibody (B). Error bars, some too small to be seen, represent SEM. (B). *Significant differences (P < 0.05) between migration in response to HGF only and each inhibitor.

Figure 4.

 

MMP-1 was strongly immunolocalized to the leading migratory HCECs in the dispersion assay in response to HGF (A; arrows). Much less MMP-1 was detected in the cells within the confluent colony (B). MMP-2 and -9 were detected in HCECs behind the leading migratory edge (arrows, C, E, respectively). MMP-2 was not detected in the confluent colony (D); however, some rounded cells contained MMP-9 (F). Scale bars, 10 μm.

Figure 5.

 

ELISA showing total production of MMP-1 by HCECs migrating on type I collagen in response to concentrations of HGF. Error bars, SEM. *Significant differences (P < 0.05) between basal (0 ng/mL) and HGF-stimulated MMP-1 levels.

Figure 6.

 

(A) HCECs migrating over the stroma of wounded corneal tissue (arrows). Hematoxylin and eosin. (B) MMP-1 mRNA expression, detected by silver grains after in situ hybridization, by migrating HCECs (arrow). (C) No signal was detected with the sense control probe. Scale bars, 10 μm.

Figure 7.

 

The rate of scratch-wound closure by HCECs in response to HGF (20 ng/mL) in the presence of concentrations of neutralizing antibodies to MMP-9 (A) and -1 (B). Concentrations: (▾) 0, (•) 1, (▪) 10, and (♦) 100 μg/mL. Error bars, some too small to be seen, represent SEM. *Significant differences (P < 0.05) between nonoverlapping error bars are indicated at 72 hours.