ACTA OPHTHALMOLOGICA

Vol. 70 (1992)Suppl. 202

COLLAGENOLYTICIGELATINOLYTIC ENZYMES IN CORNEAL WOUND HEALING M. Elizabeth Fini, Marie T. Girard and Masao Matsubara MGHlHarvard Cutaneous Biology Research Center, Massachusetts General Hospital, and Department of Dermatology, Harvard Medical School, Boston, MA, USA

Abstract. We have documented changes in expression of collagenolytic/gelatinolyticenzymes of the matrix metalloproteinase family (MMP) in healing or ulcerating corneal wounds of rat or rabbit. Correlation of our findings with specific changes in the extracellular matrix of the repair tissue suggests two different roles for the enzymes, MMP-2 and MMP-9. MMP-2 is expressed in undamaged corneal stroma where it may degrade the occasional collagen molecule that becomes damaged. After corneal wounding, expression of this enzyme is increased and much of it appears in the active form. These changes persist for at least seven months, suggesting that MMP-2 is involved in the prolonged process of collagen remodelling in the stromal repair tissue. MMP-9 is expressed in the epithelial layer of repair tissue with a timing suggesting it might participate in controlling resynthesis of the basement membrane. MMP-9 also appears to be involved in degradation of the epithelial basement membrane that precedes corneal ulceration. Key words: collagenase - gelatinase - cornea stroma - basement membrane - collagen - extracelMar matrix - wound healing - corneal ulceration.

Cornea is composed of three distinct layers of cells with their associated extracellular matrices (ECMs). It is the unique properties of these ECMs that help to determine many of the important qualities of the corneal tissue, including, transparency and the capacity of the corneal surface to withstand the abrasive forces of the eyelids. Corneal ECMs are quite static (Davison & Galbavy, 1986); nevertheless the number of cor26

neal pathologies that are problems of ECM maintenance demonstrate the importance of ECM renewal for maintenance of a healthy cornea. In addition, renewal of ECMs becomes a matter of pressing importance after corneal injury. In this paper, we review some new data suggesting mechanisms which could contribute to maintenance and repair of the corneal ECMs.

Cornea contains several collagen types The principle protein component of the cornea is collagen, primarily of the type I variety. Because of its prevalence, the molecular properties of type I collagen determine, to a large degree, the overall corneal structure. However, it is becoming apparent that collagen types that represent a minor component of the corneal collagen play a major part in determining the unique structural and functional properties of cornea. One of these minor collagen types is located in the corneal stroma. Recent studies in chick cornea show that the fibrils of the stromal lamellae, which have previously been thought to be composed entirely of type I collagen, are copolymers of type I and V collagens. Experiments suggests that the heterotypic interaction between type I and V may be at least partially responsible for creating the narrow diameter fibril (Birk et al., 1990) the unique structure of which is thought be be important for corneal transparency. The heterotypic fibril structure is almost certainly true

Collagenolyticlgelatinolytic enzymes in corneal wound healing

in mammalian cornea as well in avian species; type V collagen makes up about 11 070 of the total collagen in rabbit stroma (Cintron et al., 1981).

The typical basement membrane underlying epithelial tissues contains type IV collagen, copolymerized with laminin and heparan sulfate proteoglycan (Timpl, 1989). However, there is some controversy as to whether type IV collagen is found in the basement membrane of corneal epithelium (Kolega et al., 1989). It is certainly clear that the anchoring fibrils, which serve to attach the basement membrane to the underlying stroma, are composed of type VII collagen (Burgeson, 1988). One reason for the development of epithelial defects in the cornea of diabetic patients may be due to decreased penetration of anchoring fibrils into the stroma (Azar et al., 1989), highlighting the importance of this collagen type to a continuous ocular surface. Other corneal ultrastructures also contain minor collagen types. The specialized basement membrane backing the endothelial layer of cornea, Descemet’s membrane, contains collagen types IV and VIII (Sawada et al., 1990). Tendrils of collagen type VI are distributed throughout the corneal stroma (Bruns et al., 1986) where they may serve to stabilize the corneal lamellae. In addition, there have been conflicting reports on the presence of type I11 collagen as a minor component of corneal stroma (Van Wart & Mookhtiar, 1991). We have confined ourselves here to describing the structure of the adult cornea, however, additional collagen types exist transiently in the cornea during development. Each of these collagens must be maintained in the normal cornea by degradative removal of damaged molecules and replacement with new. Collagen replacement in the wounded cornea is a progressive process of synthesis, degradation, and resynthesis, a process which has been termed, remodelling (Gross, 1982).

Enzymes involved in degrading new collagen types Specialized mechanisms appear to be required for the degradation of collagen ultrastructures because of their highly polymerized nature. The best understood is the mechanism for removal of

structures containing type I collagen. Cells in tissues undergoing resorption or remodelling of collagen structures actively mediate type I collagenolysis (Gross, 1982). This activity has been attributed to secretion of interstitial collagenases, the only enzymes which can catalyze degradation of collagen at the neutral pH of the extracellular space. Type I collagenolytic activity has been demonstrated in corneas healing after injury (Brown & Weller, 1970), indicating operation of this basic collagenolytic mechanism in cornea as in other tissues. Interstitial collagenase can degrade only collagen types I, 11, and 111; however, a mechanism for degradation of the minor collagen types has been suggested by identification of related ECM degrading enzymes (Woessner, 1991). Each enzyme shares with collagenase a common requirement for a metal cofactor and an action against ECM components, prompting the designation of matrix metalloproteinase (MMP). Cloning and sequencing studies have demonstrated that these enzymes are structurally related probably as a result of descent from a common, ancestral gene (Brinckerhoff & Fini, 1989). MMPs are secreted by cells as inactive proenzymes which must be converted to an activated form in the extracellular space by a mechanism thought to involve proteolytic cleavage of the amino-terminus, the “pro” portion of the enzyme. Once activated, all members of the MMP family can be inhibited by binding to members of the Tissue Inhibitor of Metalloproteinase (TIMP) family. To date, seven different MMPs with structural similarity to collagenase have been characterized by cloning and sequencing (Table 1). The two gelatinases, MMP-2 (65 kD gelatinase) and MMP-9 (92 kD gelatinase) have the capacity to further degrade types I, 11, or I11 collagens after collagenase cleavage and subsequent denaturation of the three collagen chains (gelatinization) that compose the collagen triple helix. Importantly, these enzymes also have specificity for native types IV, V, and VII collagen, suggesting their participation in remodelling of ECMs containing these important, collagen types. No differences appear to exist between the two gelatinases in their capacity to cleave ECM components; however, our studies in a corneal cell culture system have demonstrated that expression of each enzyme is independently altered in response to 27

M. Elisabeth Fini et al.

Table 1. The matrix metalloproteinases (MMPs). Enzyme name

Matrix substrates

1. Interstitial collagenase

I, 11, collagens

Vertebrate collagenase Fibroblast collagenase (MMP-I) 2. Neutrophil collagenase (MMP-8)

I, 11, 111 collagens

3. 65 kD (72 kD) gelatinase 65 (72 kD) type IV collagenase (MMP-2)

IV, V, VII collagens fibronectin gelatins

4. 92 kD gelatinase 92 kD type IV collagenase type V collagenase (MMP-9)

IV, V, VII collagens fibronectin gelatins

5 . Stromelysin transin proteoglycanase procollagenase activator (MMP-3)

proteoglycans laminin fibronectin 111, IV, V collagens gelatins

6. Stromelysin-2 transin-2 (MMP-10)

111, IV, V collagen

7. PUMP-1 small MMP of uterus (MMP-7)

fibronectin gelatins gelatins fibronectin

Adapted from Nagase H , Barrett A J & Woessner J F, Jr: Nomenclature and glossary of the matrix metalloproteinases. Matrix (Suppl. l), in press.

various external agents (Fini & Girard, 1990a, b). This suggests that MMP-2 and MMP-9 could be used in a tissue for different purposes at different times.

Wound models for studying collagen turnover mechanisms Corneal wounding stimulates entry into a vigorous remodelling process providing a good model for examining mechanisms of collagen turnover. Much data has been gathered using the penetrating keratectomy wound model in rabbits (Cintron & Kublin, 1977; Cintron et al., 1981). Trephination of a 2 mm disk from the central cornea creates a space that is quickly filled with a fibrin plug formed from precursors in the aqueous humor. Within a few days, epithelium from the area peripheral to the wound migrates cen28

trally to surface the plug. Keratocytes in the stroma adjacent to the damaged area undergo transformation into fibroblasts and migrate into the wound directly under the new epithelium. A number of fibroblasts have accumulated within 7 days; these cells multiply over the next few weeks to fill the space originally occupied by fibrin. Collagen is synthesized and deposited by the wound fibroblasts, but not in the characteristic parallel arrays. Instead, collagen deposition occurs in a more haphazard, lamellar pattern resembling an onion skin (Cintron et al., 1978). However, this early repair tissue is progressively remodelled, over a period of years, until the parallel lamellar layers are recreated across the region where the incision had interrupted them (Davison & Galbavy, 1986). Remodelling in cornea, as in other tissues involves synthesis, degradation, and resynthesis of collagens (Cionni et al., 1986). During this process, collagen fibril size becomes more regular and the stromal fibrils attain a more orderly arrangement (Cintron et al., 1978). These changes are thought to contribute to the return of normal corneal transparency in the damaged area, which can occur in some species such as the rabbit (Cintron et al., 1977). The epithelial basement membrane that is deposited in the early repair tissue is also progressively remodelled. This has been most clearly described in the superficial keratectomy wound of rabbits (Gipson et al., 1989). As visualized by microscopy, basement membrane is replaced in segments synchronous with deposition of immunoreactive type VII collagen. Newly deposited type VII collagen is not immediately assembled into defined anchoring fibrils, however. In fact, these structures cannot be visualized by electron microscopy until 2-4 weeks after wound healing has begun, despite the fact that deposition of type VII collagen occurs almost immediately after wounding. With time, new stromal matrix also begins to accumulate in the wound bed displacing the epithelium upward. Basement membrane is abandoned by the epithelium in its upward shift; this material accumulates in strata within the anterior stroma, and a new basement membrane is synthesized under the epithelium. The strata can persist for up to 6 months, but are gradually removed in the stromal remodelling process.

Collegenolyticlgelatinolytic

enzymes in corcorneal wound healing

Fig. I. (Top) Schematic diagram on the left depicts cross-section through cornea which has just undergone a 2 mm penetrating keratectomy. The right diagram depicts the same cornea after healing for two weeks. To collect repair tissue for analysis, a trephine was aligned on the boundary between the damaged and undamaged cornea and inserted to obtain a full thickness disk of tissue. A similar sized corneal disk was collected from an area just adjacent to the repairing area, and from the center of the contralateral cornea. (Bottom) Gelatinases were extracted from each of the three corneal disks obtained from six different rabbits (1-6) and electrophoresed on a gelatin zymogram. The left, middle, and right lane of each set represents gelatinases (clear bands) from the undamaged cornea, the repair adjacent tissue of the wounded cornea, and the repair tissue of the wounded cornea, respectively. Molecular sizes of the bands were calculated from molecular size standards and are indicated to the left of the panel in kD (kiloDaltons).

The role of gelatinases in corneal ECM remodelling after wounding In a recent study, we used the superficial and penetrating keratectomy wound models in rabbits to examine our hypothesis that gelatinases are involved in deposition and remodelling of repair tissue collagen after injury (Matsubara et al., 1991a). Trephined, full thickness disks of an identical, 2 mm diameter were collected from the central region of unwounded corneas and from the repairing and repair-adjacent regions of wounded corneas. The disks were directly extracted with a 2 070 solution of SDS, and an equal portion of each extract was analyzed by substrate gel electrophoresis, also called zymography (Fig. 1). This method reveals the presence of specific enzymes in a complex mixture of proteins by their capacity to digest a substrate which is copolymer-

ized in a polyacrylamide gel. In our experiment, corneal extracts were electrophoresed on gels containing 0.1 070 denatured, type I collagen (gelatin) and the gels were developed under conditions that would allow gelatinase activity to take place (37°C in 10 mM calcium buffer at neutral pH). Once the gels were stained, the position of enzymes could be determined as cleared areas in the stained gelatin background, and enzyme sizes were calculated with respect to molecular weight standards that had been co-electrophoresed in a parallel gel lane. Progelatinases as well as activated enzyme forms can be visualized with this method. The SDS in the gel unfolds the proenzyme; during gel development, it renatures in an active form. Gelatinases were preliminarily identified as MMP-9 or MMP-2 by molecular size, but this identification was confirmed by testing for binding with specific antisera. The large 29

M. Elisabeth Fini et al.

amounts of the wound gelatinases required for immunological analysis were obtained by organ culturing corneal disks; enzyme was secreted into the culture medium and could be collected after a few days. Easily detectable levels of the 65 kD, MMP-2 proenzyme could be extracted from the unwounded cornea. This enzyme was secreted into organ culture, demonstrating that it must be localized extracellularly in tissue and is not an intracellular storage form of the enzyme. Separation of epithelium by scraping from stroma/ endothelium before extraction of gelatinases revealed that enzyme is localized to the stroma/ endothelium; no enzyme could be extracted from the scrape-isolated epithelium. After keratectomy, no immediate change in MMP-2 levels or forms could be detected; the acellular fibrin plug contained similar levels of MMP-2 proenzymes as did the normal corneal stroma, but the enzyme is likely to be derived, in this case, from the aqueous humor. An increase in MMP-2 proenzyme levels, however, could be detected as fibroblasts entered the repair tissue and proliferated over the first week after wounding. In addition, a lower molecular weight form of MMP-2 having a size appropriate to be an activated form of the enzyme appeared. The increase in the MMP-2 forms was precisely localized to the repair tissue; no change could be observed in tissue even immediately adjacent to the repairing area. New enzyme levels peaked at one week after wounding and then began a steady decline; however, even 7 months after wounding, the level of MMP-2 pro- and activated forms was still above normal. A second change in the corneal gelatinases that could be detected after wounding was the appearance of MMP-9. Enzyme was precisely localized to the repair tissue, much as was altered expression of MMP-2. However, appearance of enzyme was not gradual and could be extracted from corneal tissue by the earliest time point examined after wounding (2 days). The MMP-9 level did not change substantially over the first week after wounding. Thereafter, the enzyme was lost from the repair tissue at a fairly rapid rate and was no longer detectable within 2 weeks after superficial keratectomy or within 4 weeks after penetrating keratectomy. MMP-9 was localized both in the epithelial' and the stromal/endothelial layers of repairing corneal tissue, unlike MMP-2. In ad-

30

dition, essentially all of the extractable enzyme was an appropriate size (92 kD) to be the proenzyme form. This was also the predominant form secreted into organ culture, indicating that proenzyme is not simply an intracellular storage form of MMP-9, but must be the predominant form in the extracellular space of repair tissue. Correlation of the individual expression patterns of the two gelatinases with changes in ECM after wounding suggest two different roles for these enzymes in the wound healing process. The fact that MMP-2 proenzyme is a normal constituent of corneal stroma, a connective tissue which is unusually static in structure, demonstrates that simple presence of MMP-2 proenzyme does not initiate the remodeling process. We previously suggested that this enzyme might play a surveillance role in normal cornea, becoming locally activated to degrade the occasional collagen molecule which might become damaged (Fini & Girard, 1990a). After wounding, MMP-2 expression is not immediately altered. However, when new matrix deposition begins, following entry of matrix-synthesizing fibroblasts into the repair tissue, more MMP-2, much of it in the active form, appears. These changes correlate with the timing of increased collagen degradation in the repair tissue (Cionni et al., 1986), signalling the initiation of the collagen remodelling process. Importantly, the changes in MMP-2, including activation, localize exclusively to the repair tissue itself, where collagen turnover is occurring. The persistence of increased MMP-2 forms for as long as seven months after wounding is consistent with the gradual loss of basement membrane strata in the repairing corneal stroma, which continues for up to 6 months, and the process of collagen fibril remodelling, which continues for at least 1.5 years (Cintron & Kublin, 1977). MMP-2 makes its appearance early in the repair process in association with re-epithelialization of the ocular surface and disappears from the repair tissue within 2-4 weeks after wounding during a time of high collagen turnover in the stroma. This is not suggestive of a role in stromal remodelling. Maybe of importance however, is the coincidence of enzyme loss with appearance of visibly recognizable anchoring fibrils of type VII collagen (a substrate for MMP-9) in the epithelial basement membrane zone. A role for

Collegenolyticlgelatinolytic enzymes in corcorneal wound healing

MMP-9 in controlling the timing of basement membrane assembly is also consistent with localization of enzyme to the epithelial layer of the repair tissue. It may be important that MMP-9 appears to be predominantly in the proenzyme form. This is not inconsistent with degradation of substrate; previous studies with cells grown on collagen films have demonstrated that collagenolysis can occur even when essentially all the detectable enzyme is in an inactive form (Thomson et al., 1989). However, the predominance of the proenzyme form of MMP-9 is in striking contrast to the large amounts of activated MMP-2 that accumulate in corneal wounds. Perhaps this indicates that MMP-9 is very locally activated, possibly in a protected environment beneath the epithelium. Products of epithelial cells control early events in corneal wound healing, including the activation of the quiescent corneal keratocytes at the wound edge to an active fibroblastic morphology, and the chemotactic attraction of these cells into the repairing area (Weimar, 1957). Basement membrane can act as a selective barrier to macromolecules and timing of its assembly is very likely to control these processes.

with increasing frequency by investigators since it was revealed in the clinic that thermal treatment to resculpt the cornea in such disorders as keratoconus, often leads to stromal ulceration (Kenyon, 1985). To produce the wound, a 4.5 mm thermal probe, heated to a temperature of 130"C, was touched to the central corneal surface for one second; two applications were performed. Half of the corneas treated in this way developed stromal ulcers after three days. As visualized by electron microscopy and immunolight microscopy, basement membrane was not removed by the initial application of the thermal probe. One day post-burn, after the epithelium had resurfaced the burned area, basement membrane was still intact. However, three days after thermal burn, the basement membrane had disappeared from under the repair epithelium in all corneas examined. Basement membrane dissolution also occurred if the corneas were removed to organ culture immediately after wounding; however, dissolution was blocked by addition of cycloheximide (an inhibitor of protein synthesis) to the culture medium. These results revealed that basement membrane loss following thermal burn is an active degradation process mediated by living cells and their products. Furthermore, it A possible role for MMP-9 in corneal could be concluded that basement membrane degradation can be initiated by the resident corulceration neal cells themselves and does not require parUlceration of the corneal stroma is a devastat- ticipation by PMNs which could not have invading disorder which responds poorly to current ed the organ-cultured corneas. It was our hypothesis that one of the gelatreatment modalities. A clinical observation is that recurrent or persistent defects in the corneal tinases might play a role in basement membrane epithelium precede stromal degradation. Ex- degradation following thermal burn. To test this, perimental studies in animals using the chemical we examined changes in gelatinase expression, usor thermal burn wound models have further ing the zymography method described above. documented this correlation (Kenyon, 1985). An- The results were compared to those obtained folother important finding from the animal studies lowing creation of a simple epithelial scrape is that loss of the epithelial basement membrane wound, which does not involve any damage to occurs subsequent to epithelial defect formation, ECM and heals without any matrix turnover but prior to stromal ulceration. This observation (Gipson & Kiorpes, 1982). Direct extraction of has lead to the hypothesis that basement mem- corneal tissue demonstrated that MMP-9, in the brane dissolution, rather than epithelial defect proenzyme form, appeared in thermally-burned formation, might be the important step initiat- corneas shortly after wounding and continued to ing stromal degradation (Berman et al., 1988). persist in the cornea over the period in which We recently undertook a study aimed at basement membrane was lost. Importantly, the elucidating the mechanism of basement mem- enzyme did not appear in corneal tissue after brane loss preceding corneal ulceration (Mat- epithelial scrape, consistent with retention of the subara et al., 1991b). Our wound model was the original basement membrane. Corneas removed thermal burn in rat. This model has been used to organ-culture immediately after wounding also 31

M. Elisabeth Fini et al.

produced MMP-9 demonstrating that the enzyme is not simply a product of the inflammatory cells that accumulate in the cornea after thermal burn, but is synthesized by corneal cells themselves. This is also in agreement with our finding that resident corneal cells, and not PMNs, are required for basement membrane degradation. Normal and abnormal processes of cellular invasion often involve the local degradation of eithelial basement membranes and underlying stroma. For example, the cytotrophoblast of the mammalian uterus must first penetrate the basement membrane of the uterine epithelium in order to invade the uterine stroma and establish a connection between fetal and maternal tissue (Fisher et al., 1989). Likewise, the progression from carcinoma in situ to invasive carcinoma occurs when the tumor cells have acquired the capacity to invade the basement membrane (Liotta et al., 1986). A specific role for MMP-9 in invasion of a basement membrane-like gel by cultured cytotrophoblasts has been recently demonstrated (Librach et al., 1991). The induction of MMP-9 synthesis by the resident cells of thermally-burned corneas suggests that the basement membrane degradation preceding corneal stromal ulceration may occur via mechanisms similar to those operating in other normal and abnormal invasion processes.

Future work Our evidence for the role of MMPs in normal and abnormal wound healing processes is purely correlative at this point. Proof of enzyme involvement in the disease process might be obtained by blocking enzyme synthesis or activity by corneal cells in wounded corneas. We recently showed that the transforming growth factor-p (TGF-B) represses MMP expression by corneal fibroblasts (Girard et al., 1991), suggesting that this cytokine might be useful in such experiments. Another potentially useful agent, HONHCOCH,(i-Bu)COTyr(0Me)-NHMe, has been previously characterized as an inhibitor of tumor cell collagenase activity, and is effective in blocking invasion of a basement membrane-like gel by cultured cytotrophoblasts (Fisher et al., 1991). We recently showed that this inhibitor can efficiently block activity of MMP-9 purified from cultures of rab32

bit corneal epithelial cells (Fini et al., 1991). Considering the close structural relationship between MMP-9 and type I collagenase, it might be expected that many other substances that block collagenase activity also block activity of MMP-9. We suggest that recent studies, which attribute the efficacy of anti-ulcer agents to their capacity to inhibit the activity of type I collagenase (Burns et al., 1988; Schultz et al., 1990), should be reevaluated in light of our findings. In fact, we might further suggest that the combination in a therapeutic cocktail of MMP activity inhibitors along with inhibitors of MMP synthesis, such as TGF-B, might prove to have the most effective action against corneal ulceration.

Acknowledgments Supported by a grant from the National Institutes of Health (EYO8408) to MEF, and by an agreement with the Shiseido Corporation of Japan. MM was a fellow of Bausch and Lomb.

References Azar D T, Spurr-Michaud S J, Tisdale A S & Gipson I K (1989): Decreased penetration of anchoring fibrils into the diabetic stroma. Arch Ophthalmol 107: 1520-1523. Berman M B, Kenyon K, Hayashi K & L’Hernault N (1988): The pathogenesis of epithelial defects and stromal ulceration. In: Cavanagh D (ed.). The Cornea: Transactions of the World Congress on the Cornea 111, p. 35-43. Raven Press, New York. Birk D E, Fitch J M, Babiarz J P , Doane K J & Linsenmayer T F (1990): Collagen fibrillogenesis in vitro: interaction of type I and V collagen regulates fibril diameter. J Cell Sci 95, 649-655. Brinckerhoff C E & Fini M E (1989): Molecular cloning of collagenase and activator/stromelysin. In: 01sen B & Nimmi M E (eds.). Collagen Vol IV, Molecular Biology, p. 65-84. CRC Press, Boca Raton, FL. Brown S I & Weller C A (1970): Cell origin of collagenase in normal and wounded corneas. Arch Ophthalmol 83: 74-77. Bruns R R, Press W, Engvall E, Timpl R & Gross J (1986): Type VI collagen in extracellular, 100 nm periodic filament and fibrils: Identification by immunoelectron microscopy. J Cell Biol 103: 393404.

Collegenolyticlgelatinolytic enzymes in corcorneal wound healing

Burgeson R E (1988): New collagens, new concepts. Ann Rev Cell Biol 4: 551-577. Burns F R, Gray R D & Paterson C A (1990). Inhibition of alkali-induced corneal ulceration and perforation by a thiol peptide. Invest Ophthalmol Vis Sci 31: 107-114. Cintron C & Kublin C L (1977): Regeneration of corneal tissue. Dev Biol 61: 346-357. Cintron C, Hassinger L C, Kublin C L & Cannon D J (1978): Biochemical and ultrastructural changes in collagen during corneal wound healing. J Ultrstruct Res 65: 13-22. Cintron C, Hong B-S & Kublin C L (1981): Quantitative analysis of collagen from normal developing corneas and corneal scars. Current Eye Res 1: 1-7. Cionni R J, Katakami C, Lavrich J B & Kao W-Y (1986): Collagen metabolism following corneal laceration in rabbits. Current Eye Res 5: 549-558. Davison P F & Galbavy E J (1986): Connective tissue remodelling in corneal and scleral wounds. Invest Ophthalmol Vis Sci 27: 1478-1484. Fini M E & Girard M T (1990): Expression of collagenolytic/gelatinolytic metalloproteinases by normal cornea. Invest Ophthalmol Vis Sci 31: 17791788. Fini M E & Girard M T (1990): The pattern of metalloproteinase expression by corneal fibroblasts is altered with passage in cell culture. J Cell Sci 97: 373-383. Fini M E, Cui T-y, Mouldovan A, Grobelny D, Galardy R E & Fisher S J (1991): An inhibitor of corneal epithelial cell gelatinase. Invest Ophthalmol Vis Sci 32: 151-155. Fisher S J, Cui T-y, Zhang L, Hartman L, Grahl K, Guo-Yang Z, Tarpey J & Damsky C H (1989): Adhesive and degradative properties of human placental cytotrophoblast cells in vitro. J Cell Biol 109: 891-902. Freeman I L (1982): The eye. In: Weiss J B & Jayson M V (eds.). Collagen in Health and Disease, p. 388-403. Churchill Livingston, New York. Gipson I K & Kiorpes T C (1982): Epithelial sheet movement: protein and glycoprotein synthesis. Dev Biol 92: 259-262. Gipson I K, Spurr-Michaud S, Tisdale A & Keough M (1989): Reassembly of the anchoring structures of the corneal epithelium during wound repair in the rabbit. Invest Ophthal Vis Sci 30: 425-434. Girard M E, Matsubara M & Fini M E (1991): Transforming growth factor-(3 and 11-1 modulate expression of metalloproteinases by corneal stromal cells. Invest Ophthalmol Vis Sci 32: 2441-2454. Gross J (1982): An essay on biological degradation fo collagen. In: Hay E D (ed.). Cell Biology of the Extracellular Matrix, p. 217-258. Plenum Press, New York.

3

Kenyon K R (1985): Inflammatory mechanisms in corneal ulceration. Trans Am Ophthalmol SOC83: 6 10-663. Kolega J, Manabe M & Sun T-T (1989): Basement membrane heterogeneity and variation in corneal epithelial differentiation. Differentiation 42: 54-63. Librach C L, Werb Z, Fitzgerald M L, Chiu K, Corwin N M, Esteves R A, Grobelny D, Galardy R, Damsky C H & Fisher S J: 92 kD type IV collagenase mediates invasion of human cytotrophoblasts. J Cell Biol 113: 437-449. Liotta L A, Rao C N & Wewer U M (1986): Biochemical interactions of tumor cells with the basement membrane. Annual Rev Biochem 55: 1037-1055. Matsubara M, Girard M T, Kublin C L, Cintron C, and Fini M E (1991a): Differential roles for two gelatinolytic enzymes of the matrix metalloproteinase family in the remodelling cornea. Devel Biol 147: 425-439. Matsubara M, Zieske J & Fini M E (1991b): Mechanism of basement membrane dissolution preceding corneal ulceration. Invest Ophthalmol Vis Sci 32: 322 1-3237. Sawada H, Konomi H & Hirosawa K (1990): Characterization of the collagen in the hexagonal lattice of Descemet’s membrane: Its relation to type VIII collagen. J Cell Biol 110: 219-227. Schultz G,Strelow S, Stern G,Galardy R & Grobelny D (1990): Inhibition of corneal ulceration in alkali burned raobit corneas by a synthetic collagenase inhibitor. Invest Ophthalmol Vis Sci (suppl) 31: 559. Thomsom B M, Atkinson S J, McGarrity A M, Hembry R M, Reynolds J J & Meikle M C (1989): Type I collagen degradation by mouse calvarial osteoblasts stimulated with 1,25-dihydroxyvitamin D-3: evidence for a plasminogen-plasmin-metalloproteinase cascade. Biochem Biophys Acta 1014: 125-132. Timpl R (1989): Structure and biological activity of basement membrane proteins. Eur J Biochem 180: 487-501. Weimar V (1957): The transformation of corneal stromal cells to fibroblasts in corneal wound healing. Am J Ophthalmol 44: 173-182. Woessner J F (1991): Matrix metalloproteinases and their inhibitors in connective tissue remodelling. FASEB J 5: 2145-2154.

Author’s address: Elizabeth Fini, Ph.D. CBRC Massachusetts General Hospital Building 149-third floor 13th Street Charlestown, MA 02129 USA

33

gelatinolytic enzymes in corneal wound healing.

We have documented changes in expression of collagenolytic/gelatinolytic enzymes of the matrix metalloproteinase family (MMP) in healing or ulcerating...
740KB Sizes 0 Downloads 0 Views