Role of Matrix Metalloproteinase-9 in Ex Vivo Expansion of Human Limbal Epithelial Cells Cultured on Human Amniotic Membrane

Abstract

purpose. To investigate the expression and pivotal role of matrix metalloproteinase (MMP)-9 in the ex vivo expansion of human limbal explants with or without amniotic membrane (AM).

methods. Corneoscleral buttons were cultured on intact, denuded AM or plastic dishes for 3 weeks. To determine the role of MMP-9 in cell migration, either the MMP inhibitor GM6001 or an MMP-9 antibody was used. Expression of MMP-9 was determined by gelatin zymography, reverse transcription–polymerase chain reaction, and immunohistochemical staining.

results. The expression of MMP-9 in all culture conditions increased in a time-dependent manner. However, the active form of MMP-9 emerged only in cultures on both intact and denuded AM from the second week. The averaged corrected ratio of MMP-9 expression in cultures on intact AM versus those on denuded AM or plastic dishes was 2.76 ± 0.69- or 4.25 ± 0.30-fold, respectively, when total RNA was used as an internal control. MMP-9 transcripts were upregulated in cultures on intact AM compared with the other two culture conditions. Immunohistochemical staining demonstrated that the MMP-9 protein was located on the limbal epithelial cells. Upregulation of MMP-9 associated with cell migration was significantly attenuated by both GM6001 and MMP-9 antibody, consistent with the inhibition of MMP-9 activity, as determined by gelatin zymography. In contrast, the sizes of limbal outgrowth were not different between the control and MMP-9 antibody–treated plastic dishes.

conclusions. These results demonstrated that MMP-9 not only was upregulated, it was also involved in the outgrowth of limbal epithelial cells. These results suggest that cell–cell matrix interaction is involved in the expansion of limbal epithelial cells on intact AM, and MMP-9 may be a key element.

Stem cells (SCs) for the corneal epithelium are considered to be located in the basal epithelium of the corneoscleral limbus. ¹ Limbal SC deficiency (LSCD), which may have diverse causes, is associated with conjunctivalization of the cornea and is complicated by persistent epithelial defects, neovascularization, scarring, ulceration, stromal melting, and perforation of the cornea. ² In patients with total LSCD, transplantation of limbal epithelial SCs is necessary to restore normal corneal phenotype and to improve visual acuity. ³ In those who have unilateral LSCD, transplantation of a contralateral conjunctival limbal autograft (CLAU) is possible and the clinical outcome is also promising. ⁴ However, the potential complication of limbal deficiency in the donor’s eye should be a major concern, because a relatively large portion of limbal tissue is removed for transplantation. ^{5 6} Moreover, CLAU is not feasible in patients with bilateral LSCD and under such circumstances, either a living-related conjunctival limbal allograft (lr-CLAL) or a cadaveric keratolimbal allograft (KLAL) is an alternative choice of treatment. ^{7 8 9} Nevertheless, rejection and use of immunosuppressants are other problems in transplantation of allografts. ¹⁰ To circumvent the problems encountered in traditional limbal transplantations, a new technique termed ex vivo expansion of limbal epithelial SCs has been developed recently, ¹¹ which expands limbal epithelial progenitor cells from a small biopsy using amniotic membrane (AM) as a supporting matrix. Such a “composite graft,” as named by Grueterich et al., ¹² has been successfully used in reconstructing the ocular surface in partial or total LSCD in several human studies. ^{11 13 14 15 16}  

Several culturing techniques were developed for ex vivo expansion of limbal epithelial cells, including those that used intact or denuded AM as a supporting matrix ^{11 17} ; adopted an airlifting technique to promote epithelial differentiation ^{14 15} ; and prepared 3T3 fibroblasts as feeder layers. ^{18 19} Previous studies have shown that limbal epithelial cells expanded on denuded AM grow faster than those on intact AM, and distinct morphologic differences are also noted between these two culture conditions. ²⁰ Recent studies indicate that intact AM preserves the limbal epithelial phenotype, whereas denuded AM promotes corneal phenotype. ²¹ Taken together, these facts reflect that the different status of microenvironments of the AM may modulate the outgrowth of limbal epithelial cells cultured on AM. Because human AM contains many components of the extracellular matrix (ECM), such as collagen types IV and V, fibronectin, and laminin, ²² it is reasonable to speculate that migration and expansion of limbal epithelial cells in this coculture model involve an interaction between these cells and the surrounding ECM. Therefore, investigation of the interactions between the limbal explants and the AM may be very important to reveal the intracellular signaling mechanisms involved in this coculture model. 

Matrix metalloproteinases (MMPs) are a family of zinc-dependent proteases that degrade ECM proteins, such as laminin, fibronectin, and collagen. The expression of MMPs has been correlated with both physiological and pathologic processes including postnatal development, reproduction, tissue remodeling, cancer, and wound healing. ^{23 24 25 26 27} Increasing evidence suggests that regulation of signaling molecules by MMPs contributes to the communication between cells and their microenvironment. ²⁸ To date, there are at least 20 different MMPs expressed in vertebrates. ²⁸ Among them, MMP-2 (72-kDa gelatinase A) and MMP-9 (92-kDa gelatinase B), which are responsible for the degradation of basement membrane collagen types IV and V, are the primary matrix-degrading enzymes produced by the corneal epithelium. ²⁹ Moreover, MMP-2 and -9 have been shown to play a key role in regulating cell migration of the corneal and lens epithelia. ^{27 30 31} However, the question of whether the interaction between human limbal explants and AM results in the induction of MMP-2 and/or -9 which are associated with expansion of corneal epithelial cells has not been documented. Therefore, the purpose of the present study is to identify the expression of MMP-2 and -9 in this ex vivo culture model determined by gelatin zymography, reverse transcription–polymerase chain reaction (RT-PCR), and immunohistochemical staining. Furthermore, the potential role of MMP-9 in cell migration and proliferation was also investigated in this culture system. 

Materials and Methods

Tissue culture dishes (35 mm) were purchased from Orange Scientific (Waterloo, Belgium); DMEM/F-12 medium and fetal bovine serum (FBS) from Invitrogen (Carlsbad, CA); MMP-9 neutralizing antibody (clone GE-213, serum and sodium azide free) from NeoMarkers (Fremont, CA); and the broad-spectrum MMP inhibitor GM6001 from Biomol (Plymouth Meeting, PA). Enzymes and other chemicals were from Sigma-Aldrich (St. Louis, MO). 

Human Limbal Explants Cultured on Human AM

Human tissue was handled according to the tenets of the Declaration of Helsinki. Corneoscleral buttons from human donor eyes, aged 15–65 years, were obtained from the Chang Gung Memorial Hospital Eye Bank. The tissue was rinsed three times with DMEM/F-12 containing 50 μg/mL gentamicin and 1.25 μg/mL amphotericin B. After careful removal of excess sclera, iris, corneal endothelium, conjunctiva, and Tenon’s capsule, the remaining tissue was placed in a culture dish and cut into cubes of approximately 1.5 × 2 × 3 mm with a scalpel. Human AM was obtained by elective cesarean section from Chang Gung Memorial Hospital (Keelung, Taiwan) with properly informed consent and was processed as described. ³² Briefly, the AM was aseptically washed three times in 200 mL of phosphate-buffered saline (PBS) containing 50 μg/mL penicillin, 50 μg/mL streptomycin, 2.5 μg/mL amphotericin B, and 25 ng/mL gentamicin. The AM was preserved sterile in DMEM/F-12 with 50% glycerin at −80°C. Before use, the AM was thawed and placed on a culture dish with the basement membrane side up and incubated at 37°C in a humidified incubator under 5% CO₂ and 95% air overnight. In preparation of denuded AM, the membranes were treated with 0.1% EDTA for 30 minutes and then gently scrubbed with an epithelial scrubber to remove the amniotic epithelium without breaking the underlying basement membrane. ²¹ On the center of the AM, a limbal explant was placed and cultured in a medium made of an equal volume of HEPES-buffered DMEM containing bicarbonate and F-12 and supplemented with 5% FBS, 0.5% dimethyl sulfoxide, 2 ng/mL mouse epidermal growth factor (EGF), 5 μg/mL insulin, 5 μg/mL transferrin, 5 ng/mL selenium, 0.5 μg/mL hydrocortisone, 30 ng/mL cholera toxin, 50 μg/mL gentamicin, and 1.25 μg/mL amphotericin B. Cultures were incubated at 37°C under 5% CO₂ and 95% air, the medium was changed and saved for gelatin zymographic analysis of MMPs every 2 to 3 days, and the extent of each outgrowth was monitored with a phase-contrast microscope. 

MMP-2 and -9 Gelatin Zymography

The culture medium was collected and centrifuged at 14,000 rpm for 5 minutes at 4°C to remove cell debris. The supernatant was mixed with 5× nonreducing sample buffer (4:1, vol/vol) and electrophoresed on a 10% SDS-polyacrylamide gel containing 0.1% gelatin as a substrate for MMP-2 and -9. After electrophoresis, gels were washed in 3% Triton X-100 for 1 hour to remove SDS and then incubated for 16 hours at 25°C in developing buffer (50 mM Tris, 40 mM HCl, 200 mM NaCl, 5 mM CaCl₂, and 0.2% Briji) on a rotary shaker. After incubation, gels were stained in 30% methanol, 10% acetic acid, and 0.5% (wt/vol) Coomassie brilliant blue for 1 hour followed by destaining. Mixed human MMP-2 and -9 were used as positive controls. Gelatinolytic activity manifested as horizontal white bands on a blue background. 

Immunohistochemical Staining of MMP-9

Six-micrometer-thick, formalin-fixed, paraffin-embedded sections taken from the AM containing the limbal epithelial outgrowth were mounted on silane-coated slides. Sections were dewaxed in xylene and rehydrated through graded alcohols. Endogenous peroxidase activity was blocked by placing sections in 2% hydrogen peroxide for 30 minutes. Sections were rinsed in deionized water and then in Tris-buffered saline containing 0.1% BSA. To block nonspecific staining, slides were incubated in 20% normal goat serum for 10 minutes. Sections were incubated overnight at 4°C with MMP-9 antibody at a dilution of 1:100. Sections were washed in Tris-buffered saline and then incubated sequentially with biotinylated rabbit anti-mouse IgG (Dako, Carpinteria, CA) at a dilution of 1:400, followed by streptavidin combined in vitro with biotinylated horseradish peroxidase at a dilution of 1:1000 (Dako). The reaction product was developed with diaminobenzidine tetrahydrochloride. Sections were counterstained with hematoxylin, dehydrated through graded alcohols, and mounted in resinous mountant. Known positive controls were included with each run, and negative controls had the primary antibody omitted. 

RNA Extraction and RT-PCR Analysis

Total cellular RNA was isolated by lysis in a guanidinium isothiocyanate buffer followed by single-step phenol-chloroform-isoamyl alcohol extraction. ³³ Briefly, cells are harvested and lysed in solution D containing 4 M guanidium isothiocyanate, 25 mM sodium citrate (pH 7.0), 0.5% sodium sarcosine, and 0.1 M β-mercaptoethanol. Sequentially, a 1:10 volume of 2 M sodium acetate (pH 4.0), one volume of phenol, and 1:5 volume of chloroform-isoamyl alcohol (49:1, vol/vol) were added to the homogenate. After it was vigorously vortexes for 30 seconds, the solution was centrifuged at 10,000g for 15 minutes at 4°C. RNA in the aqueous phase was precipitated by the addition of 0.5 mL isopropanol. One microgram total RNA was reverse transcribed into cDNA by incubating with 200 U of reverse transcriptase in 20 μL reaction buffer containing 0.25 μg random primers and 0.8 mM dNTPs at 42°C for 1 hour. Two microliters of the cDNA was used for the PCR reaction as templates. The PCR was performed in buffer containing 10 mM Tris (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM dNTPs, 1 μM of each primer, and 5 U Taq DNA polymerase for 30 cycles of denaturation at 94°C for 1 minute, annealing at 55°C for 1 minute, and extension at 72°C for 2 minutes. The resultant PCR product was analyzed by 1.5% agarose gel electrophoresis. Sequences for the specific primers used in the PCR are summarized in Table 1 . 

Statistical Analysis

Data were analyzed on computer (Prism; GraphPad, San Diego, CA), expressed as the mean ± SEM, and analyzed with a two-tailed Student’s t-test with P < 0.05 set as the level of significance. 

Results

Outgrowth of Human Limbal Epithelial Cells from Limbal Explants Cultured on AM

The outgrowth of limbal epithelial cells from limbal explants cultured on intact AM was not observed until 1 week after culture (Fig. 1A) . In fact, we noted that limbal epithelial cells did not migrate onto intact AM until 4 to 5 days after culture. It was also apparent that the initial migration of epithelial cells took place from the limbal region, implying that the limbus was the reservoir of corneal epithelial SCs. After 2 weeks in culture, the leading edge of the outgrowth of epithelial cells on intact AM formed a well-defined demarcation circle, which increased in size through 3 weeks in culture (Figs. 1B 1C) . On the basis of microscopic examination, the demarcation line was evident, because the outermost epithelial cells piled up at the leading edge (see Fig. 3 ). In contrast, limbal explants cultured on denuded AM or plastic dishes began growing as early as 2 days after culture, a finding consistent with those of others. ²⁰  

Gelatin Zymography of MMP-2 and -9

To investigate whether the expression of MMP-2 and -9 were induced by the interaction between human limbal explants and AM, we determined the activities of MMP-2 and -9 by gelatin zymography in the condition media collected from three different culture conditions: human limbal explants cultured on intact or denuded AM (H/A^intact or H/A^denuded group), and limbal explants on plastic dishes (H group). The expression of MMP-9 in all three culture conditions increased in a time-dependent manner, whereas MMP-2 expression remained unchanged throughout the culture period. However, the expression of MMP-9 in active form varied among three different groups, with a significant amount induced in the H/A^intact group, weakly detectable in the H/A^denuded group, and little in the H group (Figs. 2A 2B 2C 2F) . Zymography performed either in the AM alone (A group) or the medium alone demonstrated barely detectable MMP-9 activity (Figs. 2D 2E) , indicating that the MMP-9 activity was not expressed or induced by the AM itself. 

To compare the respective differences in the expression of MMP-9 at the indicated times, the conditioned media collected from three groups of culture conditions were analyzed on the same gel in a time sequence. As shown in Figure 2F , MMP-9 activity in the H group seemed more intensely expressed than that in the H/A^intact group. However, as we know that due to the different cell growth rate among the different culture conditions, the outgrowth areas in the H/A^denuded or H group were usually larger than those in the H/A^intact group at the end of culture. Because the expression of MMP-9 also paralleled with cellular mass, a corrected ratio for “real” MMP-9 expression should be calculated to reveal actual differences among the various culturing conditions. In our preliminary data, there was hardly any RNA isolated from AM alone, we therefore used the amount of extracted total RNA instead of total protein from each culture as an internal control of cellular mass at the end of 3 weeks to avoid the potential problem of contamination with AM proteins during sample extraction. In this way, we found that the corrected ratio of MMP-9 activity of the H/A^intact group versus the H/A^denuded and H groups was 2.76 ± 0.69- and 4.25 ± 0.30-fold, respectively. 

Immunohistochemical Staining of MMP-9 in Human Limbal Epithelial Cells Cultured on AM

To localize the expression of MMP-9 in the coculture model, we stained the paraffin-embedded tissue sections prepared from the H/A^intact group with an MMP-9 antibody. As shown in Figure 3B , the stain of MMP-9 was denser on the limbal epithelial cells than on the AM. The light-brownish staining on the AM indicated the presence of MMP-9 proteins in a low amount on the AM. However, as shown in Figure 2D , these MMP-9 proteins on AM were not released into the medium and exerted no activity at all. We also noted that the devitalized amniotic epithelial cells under expanded limbal epithelial cells were barely discernible compared with those on naive intact AM (Fig. 3C) . 

MMP-9 mRNA Expression Analyzed by RT-PCR

Having shown that the activity of MMP-9 was upregulated more in the H/A^intact group than in H/A^denuded group or H group, we wanted to verify whether the induced MMP-9 expression was also upregulated at the transcriptional level. Thus, we performed RT-PCR to compare the expression of MMP-9 mRNA among the three groups. As shown in Figure 4 , when normalized to the level of β-actin, the MMP-9 transcripts were increased in the H/A^intact group compared with those in the H/A^denuded or H groups (P < 0.05, n = 4). There was no significant difference between the H/A^denuded and H groups (P = 0.70, n = 4). Taken together, theses data suggest that upregulation of MMP-9 occurs at the transcriptional level when limbal epithelial cells are expanded on intact AM. 

Effect of GM6001 on the Outgrowth of Limbal Epithelial Cells on AM

To clarify the functional importance of MMP-9 in the model of ex vivo expansion of limbal epithelial cells on intact AM, we used a broad-spectrum MMP inhibitor, GM6001, to differentiate the role of MMP-9 on limbal epithelial outgrowth. As shown in Figures 5A and 5B , cultures at the end of the second week with similar sizes of limbal epithelial outgrowth were treated with or without GM6001 (10 μM). One week after the addition of GM6001, the limbal epithelial outgrowth in both the control group and the GM6001-treated group increased in size. However, the expansion of limbal epithelial outgrowth was significantly retarded by the treatment with GM6001 (Figs. 5C 5D) . Statistical analysis from at least six independent experiments showed that treatment with GM6001 caused a significant inhibition on the outgrowth of limbal epithelial cells (Fig. 5E ; P < 0.01, n = 6). Zymographic analysis of conditioned media also confirmed the elimination of MMP-9 in both latent and active forms in the culture system treated with GM6001 at days 18 and 21 (Fig. 5F) . 

Neutralization of MMP-9 on the Outgrowth of Limbal Epithelial Cells on AM or Plastic Dishes

Because GM6001 is a broad-spectrum MMP inhibitor, we wanted to confirm further whether MMP-9 expression is essential for limbal epithelial outgrowth on intact AM using an MMP-9-neutralizing antibody specifically to block MMP-9 activity in this coculture model. The sizes of limbal epithelial outgrowth in both conditions before treatment with (Figs. 6B 6F)or without (Figs. 6A 6E)MMP-9 antibody were almost the same. One week later, the inhibition of limbal epithelial outgrowth was more prominent in dishes treated with MMP-9 antibody (Figs. 6D 6H)compared with the control (Figs. 6C 6G) . Moreover, the area of limbal epithelial outgrowth was consistently decreased with increasing concentrations of MMP-9 antibody, which showed a concentration-related response in at least six independent experiments (Fig. 6I) . To demonstrate that MMP-9 is specifically and critically involved in the outgrowth of limbal epithelial cells on AM, we also treated expanded epithelial cells grown for 1 week on plastic dishes with MMP-9 antibody. For better visualization of the extent of limbal outgrowth, the dishes were fixed with 4% paraformaldehyde followed by contrast blue staining at the end of treatment. As shown in Figures 7A and 7B , the sizes of limbal outgrowth were not different between control and MMP-9 antibody-treated dishes 1 week later. No significant difference was noted from at least six independent experiments (Fig. 7C ; P = 0.67, n = 6). 

Discussion

In this study, both latent and active forms of MMP-9 were upregulated during the outgrowth of human limbal explants cultured on intact AM compared with those cultured on denuded AM or plastic dishes. GM6001, a broad-spectrum MMP inhibitor, or a specific MMP-9 antibody significantly reduced the outgrowth of the limbal epithelial cells when MMP-9 activity was inhibited, as determined by gelatin zymography. Collectively, our data indicate a strong correlation between MMP-9 expression and the outgrowth of limbal epithelial cells in this coculture model of human limbal explants on intact AM. 

The expression of MMP-9 is regulated at several levels. In addition to the gene regulation at the transcriptional level, MMP-9 is secreted by cells as inactive proenzymes and must bind to the cell surface, where they are processed and activated. ³⁴ In general, the basal level of MMP-9 is usually low, and its expression can be induced by various cytokines or growth factors. ³⁵ The present study showed that in addition to gelatin zymography, upregulation of MMP-9 gene transcripts in the H/A^intact group was further confirmed by RT-PCR (Fig. 4) . Because hardly any RNA could be extracted from frozen human AM alone (data not shown), we believe that the upregulation of MMP-9 gene transcript was mainly produced by expanded limbal epithelial cells but not from AM. Moreover, immunohistochemical study revealed that most of the MMP-9 stain was located on the limbal epithelial cells, rather than on the AM (Fig. 3B) . In contrast to MMP-9, MMP-2 is usually constitutively expressed. ²⁹ In this study, the expression of MMP-9 in the H/A^intact group paralleled with the outgrowth area of the limbal explants (Figs. 1 2) , whereas the expression of MMP-2 remained unchanged during the culture period. We also used a specific MMP-2 antibody in this coculture model and found that it was not as effective as MMP-9 antibody for inhibiting the outgrowth of limbal explants (data not shown). Taken together, these data demonstrated that MMP-9 may play an important role in the process of ex vivo expansion of limbal SCs on AM. 

Despite the broad clinical applications of transplantations with ex vivo expanded limbal SCs, the mechanisms of how ex vivo expanded human limbal epithelial cells interact with underlying AM (i.e., ECM) remain unknown. In the present study, when limbal explants were cultured on intact AM, the activity of MMP-9 was 2.76 ± 0.69- and 4.25 ± 0.30-fold higher, respectively, than in those cultured on denuded AM or plastic dishes at the third week in culture. The question raised is how and why the expression of MMP-9 is upregulated in the H/A^intact group. There are several possibilities. The most common reason for the differences among the three culture conditions may be that more matrix-degrading enzymes were needed to dissolve amniotic epithelial cells and some components of the amniotic basement membrane in the H/A^intact group to facilitate subsequent attachment and migration of the limbal epithelial cells on underlying amniotic basement membrane. In fact, the functions of MMPs mimicked those of scavengers, which carved and swept off the debris from cells and the ECM, to build up the connections between cells and surrounding ECM. Based on the connections settled by MMPs, adhesion molecules or other receptors on the cell surface could bind to ECM proteins and transduce the signaling pathways into the cells to induce migration and proliferation of cells. This hypothesis was in part supported by the notion in histologic staining that the devitalized amniotic epithelial cells under expanded limbal epithelial cells were barely discernible compared with those on naive intact AM (Fig. 3C) . However, this hypothesis seems not very convincing in this coculture model, because previous studies have demonstrated that expanded limbal epithelial cells appeared to migrate over the top of amniotic epithelial cells ²⁰ and that ex vivo expanded limbal epithelial cells with AM retain an intact amniotic basement membrane on transplantation in human patients. ^{16 36} Another possible explanation for the upregulation of MMP-9 in cells grown on intact AM is probably related to the interaction of limbal epithelial cells and devitalized amniotic epithelial cells instead of the interaction between limbal epithelial cells and AM. However, our results indicate that interaction between limbal epithelial cells and denuded AM also could induce the expression of MMP-9, both in latent and active forms (Fig. 2B) . It suggests that MMP-9 activity is also necessary for limbal epithelial cells to expand on AM matrix. Collectively, the evidence implies that other mechanistic actions may underlie the upregulated MMP-9 expression. 

Outgrowth of limbal explants on AM involves serial and complex processes. After attachment of the budding limbal epithelial cells to the underlying AM, migration of the expanded cells is necessary for subsequent proliferation of the limbal epithelial cells. During the culture period, we found that, apart from pro-MMP-9 overexpression, upregulation of active-form MMP-9 was also observed at the second week in the H/A^intact group and to a lesser extent was noted in the H/A^denuded group (Fig. 2) . However, only the latent, not the active, form of MMP-9 was expressed in the H group, implying that the presence of AM (either intact or denuded) was responsible for the induction of the active form of MMP-9. The functional significance of active-form MMP-9 in this coculture model is not clear at present. A recent report has demonstrated that active-form MMP-9, but not active- or proform MMP-2, together with activation of integrins, is involved in migration of metastatic breast cancer cells. ³⁷ Integrins are αβ heterodimers that function as the major receptors for cell adhesion to the ECM. ³⁸ Integrins α3β1 and α6β4 are major ligands to the basement membrane component laminin-5, ³⁹ which is also a component of the amniotic basement membrane. ⁴⁰ Several lines of evidence have indicated that the interaction of laminin-5 with integrin α3β1, which is involved in epithelial cell adhesion and migration, ⁴¹ is associated with the secretion of MMP-9. ^{42 43} Therefore, it seems plausible to speculate that migration of expanded limbal epithelial cells on AM might involve a causal relationship between the interaction of specific ECM components (such as laminin-5) and integrins, which results in the enzymatic activation of MMP-9. This hypothesis may explain why the active form of MMP-9 was expressed only in the H/A^intact and H/A^denuded groups, but not in the H group (Fig. 2) . However, further studies are mandatory to verify the relationship between the overexpression of active-form MMP-9 and the migration of limbal epithelial cells in this coculture model and to identify how limbal epithelial cells migrate on intact AM through a devitalized layer of amniotic epithelial cells without direct contacts with underlying amniotic basement membrane. 

The relatively higher MMP-9 activity noted in the H/A^denuded group (Fig. 2B)compared with that in the H group (Fig. 2C)indicates that MMP-9 activity is also likely to be involved the outgrowth of limbal epithelial cells on denuded AM. However, the results of RT-PCR did not reveal a significance difference between the H/A^denuded and H groups (Fig. 4) . The analysis suggested that the upregulation of MMP-9 activity in limbal epithelial cells grown on denuded AM is at the posttranscriptional level, which is different from cells on intact AM. This interesting finding may directly or indirectly provide a potential explanation for the differential phenotype expression of cells grown in different AM microenvironments and deserves further study in detail. 

To determine whether MMP-9 was involved in the ex vivo expansion of limbal epithelial cells on AM, we added either GM6001 or MMP-9 antibody to the growth medium at the end of the second week in culture and continued culturing for 1 week. Cultures at the end of the second week were used in this study because the activity of MMP-9 and the growth of limbal epithelial cells were more prominent during the third week in culture. Our results demonstrated that the outgrowth of limbal epithelial cells cultured on intact AM was significantly inhibited accompanied by the attenuation of both the latent and active forms of enzymatic activity of MMP-9 (Figs. 5 6) . On the contrary, the outgrowth of limbal epithelial cells on plastic dishes was not suppressed by MMP-9 neutralizing antibody (Fig. 7) . These results suggest that MMP-9 plays a pivotal role in ex vivo expansion of limbal explants on intact AM. Several lines of evidence have indicated that MMPs may be implicated in the processing of precursor forms of growth factors and cytokines as well as their receptors and/or may regulate the release of matrix-associated growth factors, ²⁸ which may have an important influence on cell signaling. For example, MMP-9 has been shown to activate transforming growth factor-β proteolytically. ⁴⁴ Recent studies have also demonstrated that MMP-mediated release of EGF receptor ligand is essential for corneal epithelial wound closure. ^{45 46} These functions provide evidence that MMP-9 can be an important regulator of cellular activity and thereby a mechanism for cells to interact reciprocally with their ECM. Therefore, we speculate that expression of MMP-9 is essential to maintain the limbal epithelial outgrowth in this ex vivo expansion model. 

In summary, using the human limbal tissue as primary culture explants, we have demonstrated that the activity of MMP-9 was upregulated in the cultures with intact AM as an underlying matrix. Based on the inhibitory effects of a synthetic MMP inhibitor, GM6001, as well as a specific MMP-9 antibody, we conclude that MMP-9 plays an important role in the outgrowth of limbal epithelial cells. Taken together, these results indicate that MMP-9, whose functional importance had been thought to be limited to the physical destruction of ECM barriers, may provide a signaling mechanism for limbal epithelial cells to begin migration and proliferation on AM. 

 Supported by Grants CMRPG32079 (C-CS) from the Chang Gung Medical Research Foundation, NSC91-2321-B182-00002 from the National Science Council, and EMRPD2E11 from the Ministry of Education (C-MY), Taiwan.

 Submitted for publication April 2, 2004; revised August 17 and November 2, 2004; accepted November 23, 2004.

 Disclosure: C.-C. Sun, None; C.-Y. Cheng, None; C.-S. Chien, None; J.-H. Su Pang, None; W.-C. Ku, None; P.Y.-F. Chen, None; C.-M. Yang, None

 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.

 

Table 1.

 

RT-PCR Primer Sequences

	Primer Sequence	PCR Product (bp)
MMP-9
Sense	5′-GACCTCAAGTGGCACCACCA-3′	440 bp
Antisense	5′-GTGGTACTGCACCAGGGCAA-3′
β-actin
Sense	5′-GACGGGGTCACCCACACTGTGCCCATCTA-3′	636 bp
Antisense	5′-CTAGAAGCATTTGCGGTGGACGATGGAGGG-3′

Figure 1.

 

Outgrowth of human limbal epithelial cells from limbal explants cocultured with intact AM. (A) The epithelial cells initially migrated from the limbal area (arrows) after 1 week in culture, indicating the location of corneal epithelial stem cells. The leading edge of expanded limbal epithelial cells formed a demarcated circle over the AM at (B) 2 and (C) 3 weeks in culture.

Figure 3.

 

Expression of MMP-9 in human limbal epithelial cells expanded on intact AM determined by immunohistochemical staining. After coculture of human limbal explants with intact AM for 3 weeks, paraffin-embedded sections were taken, incubated overnight at 4°C with MMP-9 antibody at a dilution of 1:100, and developed. Compared with the control without primary antibody (A), positive staining was noted on the expanded limbal epithelial cells (B, arrows), rather than on the AM. The expanded limbal epithelial cells piled up at the leading edge (A, arrowhead). Also note the barely detectable devitalized amniotic epithelial cells under expanded limbal epithelial cells (C, arrowheads) compared with those on nearby intact AM (arrow). Scale bars, 20 μm. Magnification: (A, B) ×200; (C) × 400.

Figure 2.

 

Zymographic analysis of MMP-2 and -9 expression in various culture conditions. The media were collected from (A) limbal explants cultured on intact AM (H/A^intact group), (B) denuded AM (H/A^denuded group), (C) plastic dishes (H group), (D) intact AM (A group), or (E) medium alone, at the times indicated. Equal amounts of proteins were loaded and MMP-2 and -9 expression was assessed by zymography after SDS-PAGE on a 10% gel and quantified by densitometry of the corresponding bands in the linear response range of the zymography. During the culture period, the expression of MMP-9 paralleled the outgrowth of limbal epithelial cells in the H/A and H groups, whereas the expression of MMP-2 remained unchanged (A–C). There was a similar small amount of expression of proform MMP-9 in the conditions with AM or growth medium alone (D, E), which excluded the possibility of the contribution of MMP-9 expression from the AM. Active-form MMP-9 was noted only in the H/A^intact group and to a lesser extent, in the H/A^denuded group at the second week in culture (A, B, F). The zymographs shown represented one of at least three individual experiments.

Figure 4.

 

Differential expression of MMP-9 mRNA of limbal epithelial cells expanded on intact AM (H/A^intact), denuded AM (H/A^denuded), and on plastic dishes (H). After incubation for 3 weeks, the total RNA was extracted and analyzed by RT-PCR. Data are representative results in one of four individual experiments. (A) The intensities of PCR product bands shown in the top panel were quantified by scanning densitometry and normalized to equivalent β-actin mRNA levels (bottom). (B) Data are summarized and expressed as the mean ± SEM of four independent experiments. *P < 0.05, compared with that of human limbal explant cultured on intact AM (H/A^intact). No significant difference was detected between the H/A^denuded and H groups (P = 0.70). M, molecular mass marker.

Figure 5.

 

Effect of GM6001 on the outgrowth of limbal explants cultured on intact AM. The sizes of limbal epithelial outgrowth in both conditions before treatment (2 weeks in culture) without (A) or with (B) 10 μM GM6001 were almost the same. After a 1-week treatment with GM6001 (D), the area of limbal epithelial outgrowth was decreased compared with the control (C). (E) Data are summarized and expressed as the mean ± SEM of results in six independent experiments (*P < 0.05, compared with control). (F) Gelatin zymography performed with supernatants treated at indicated times with 10 μM GM6001 demonstrated elimination of both latent and active forms of MMP-9 activity from the media compared with the culture without adding GM6001. Equal amounts of proteins in the media were loaded on the gel, and MMP-9 expression was assessed by zymography after SDS-PAGE on a 10% gel and quantified by densitometry of the corresponding bands in the linear response range of the zymography. Zymograms shown represent one of six individual experiments.

Figure 6.

 

Neutralization of MMP-9 antibody led to inhibition of outgrowth of limbal explants cultured on intact AM. The sizes of limbal epithelial outgrowth in both conditions before treatment (2 weeks in culture) with (B, F) or without (A, E) an MMP-9 antibody were almost the same. One week later, the inhibition of limbal epithelial outgrowth was more prominent in dishes treated with 8 μg/mL MMP-9 neutralizing antibody (G, H) compared with those treated with 4 μg/mL MMP-9 antibody (C, D). Data are summarized and expressed as the mean ± SEM of six independent experiments shown in (I). *P < 0.05, compared with control.

Figure 7.

 

MMP-9-neutralizing antibody did not inhibit the outgrowth of limbal epithelial cells grown on plastic dishes. The extent of limbal epithelial outgrowth in both conditions before treatment (1 week in culture) with or without an MMP-9 antibody was almost the same. At the end of treatment, the dishes were fixed with 4% paraformaldehyde followed by contrast blue staining for better visualization of the extent of limbal outgrowth. One week later, the limbal epithelial outgrowth was not different between control (A) and MMP-9 antibody-treated (B) dishes. Data are summarized and expressed as the mean ± SEM of six independent experiments shown in (C). P = 0.67, compared with the control.

The authors thank Ray R. F. Tsai for technical support and Wei-Hsuan Yu (National Taiwan University) for a critical reading of the manuscript and for providing invaluable suggestions.