Search

Keratocytes produce the extensive stromal matrix of the cornea during the late embryonic and neonatal time periods. We propose to test the hypothesis that their biosynthetic activity declines during this process. Keratocytes were isolated from corneas of 6–8-week-old rabbits and corneas of 1–2-year-old cows and their ability to proliferate and synthesize collagen in serum-free media was determined. Rabbit keratocyte cultures increased 38% in DNA content after one week and deposited collagen type I and IGF-II in the media. Bovine keratocyte cultures, in contrast, did not increase in DNA or produce detectable collagen and IGF-II. Bovine keratocytes cultured in media previously conditioned by rabbit keratocytes, however, increased 56% in DNA content, and deposited collagen type I into the media. Microarray analysis of mRNA from neonatal and adult mouse keratocytes was used to confirm these differences. Compared to adult mouse keratocytes, neonatal keratocytes showed high expression levels of IGF-I, IGF-II and collagen types III and V. Since previous studies showed that IGFs stimulate bovine keratocytes to proliferate and to synthesize procollagen type I, we therefore propose that the results of this study suggests that the IGFs may play an important role in regulating early corneal growth in vivo.

Previously, pharmacological levels of insulin have been shown to stimulate the synthesis of normal corneal stromal collagen and proteoglycans by bovine keratocytes in culture. Here we compared insulin to physiological levels of IGF-I and found that IGF-I also stimulated the synthesis of these extracellular matrix components, but less than that of insulin. Keratocytes in monolayer culture secreted most of the collagen synthesized into the media in the form of procollagen, a precursor of collagen. We found that an overlay of 3% agarose on the keratocytes in culture enhanced the conversion of procollagen to collagen and increased the deposition of collagen and proteoglycans into the cell layer. The extracellular matrix associated with the keratocytes cultured under agarose exhibited a corneal stromal-like architecture. These results suggest that enhancing the conversion of procollagen to collagen is a key step in the formation of extracellular matrix by keratocytes in vitro. Agarose overlay of insulin activated keratocytes in culture is a useful model for studying corneal stromal extracellular matrix assembly in vitro.

Fetal bovine serum has commonly been used to expand the population of keratocytes in culture. Tissue extracts, however, have also been used to grow other cell types. We prepared a DMEM/F12 extract of corneal stroma and compared the growth and morphology of collagenase-isolated keratocytes cultured in DMEM/F12, or DMEM/F12 containing either stromal extract or fetal bovine serum. Cell proliferation was measured by 3H-thymidine and BrdU incorporation as well as by DNA quantitation. The extract was fractionated by gel filtration. Cell morphology was assessed by phase-contrast microscopy. Culture in both extract and serum stimulated keratocytes to proliferate, but keratocytes cultured in the extract grew more slowly due to a longer cell cycle and to a lower final density because of greater sensitivity to contact inhibition. Keratocytes cultured in serum became fibroblastic while those cultured in extract retained the dendritic morphology of quiescent keratocytes. The stimulating factors in the corneal extract were more sensitive to heat inactivation and of higher molecular weight than the stimulating factors in serum. These results indicate that the mitogenic activity in extract and serum are different and that the phenotypes resulting from growth in serum and extract are also different. Keratocytes cultured at low cell densities in the corneal extract may mimic keratocyte activation, an initial and crucial event for keratocytes during the corneal wound healing process.

The corneal proteoglycans belong to the Leu-rich proteoglycan (LRP) gene family and contain chondroitin/dermatan (CS/DS) or keratan sulfate (KS) chains. These proteoglycans play a critical role in generating and maintaining a transparent matrix within the corneal stroma. Decorin which has CS/DS chains and lumican which has KS chains, were first to be identified in the cornea. Two other corneal KS proteoglycans (KSPGs), keratocan and osteoglycin/mimecan were recently identified in bovine corneas. We cloned and sequenced chick osteoglycin/mimecan and found it to contain a stretch of 60 amino acids that showed no identity to the presumed mammalian homolog. The 177 base pair DNA coding for this unique sequence shows 47% identity to an 189 base pair sequence between exons 4 and 5 of the bovine osteoglycin/mimecan gene. This indicates that this cDNA represents an alternatively spliced form of osteoglycin/mimecan containing a unique N-terminal sequence. The expression of each of the three corneal KSPGs in the developing and mature chick cornea was investigated by competitive PCR and immuno-biochemical analysis of corneal extracts. Competitive PCR was used to determine the message levels for chick lumican, keratocan and osteoglycin in embryonic day 9, 12, 15, 18 and adult corneas. Results showed that lumican mRNA fluctuated during development but remained at a relatively high level while keratocan and osteoglycin message levels declined steadily from day 9 to adult. Additionally, lumican mRNA was present at higher levels, during all stages of corneal development, than keratocan and at much higher levels than osteoglycin. Antibodies shown to be specific for each KSPG were used to characterize proteoglycans isolated from embryonic and adult chick corneas. KSPGs from embryonic corneas eluted 1–2 fractions earlier on Q-Sepharose than KSPG from adult corneas. Additionally, Western blot analysis showed that embryonic KSPGs were more keratanase-resistant, endo-β-galactosidase sensitive than adult KSPGs. The results of this study indicate an alteration in sulfation or the fine structure of the glycosaminoglycan chains occurs during corneal maturation for the 3 KSPGs.

Proteoglycan distribution during corneal stromal healing in growing corneas of young chicks were histologically and immunohistochemically analysed. Single linear incisions to produce partial-thickness wounds were made in the corneas of 5 day old chicks. The corneas were harvested at different times after wounding and processed for either histochemical analyses using periodic acid-Schiff's reaction (PAS) or for indirect immunofluorescence analyses of lumican, keratocan, keratan sulfate, perlecan and laminin. Linear corneal stromal incisions were completely covered by migrated stratified epithelium by day 2 post wounding and resulted in a gaping wound with a thinner stroma. New stromal scar tissue formed between the epithelium and the original stroma that resulted in partial restoration of stromal thickness. During the first two to three weeks of healing, the stromal tissue filling the depression formed from the gaping wound, was hypercellular and PAS positive, indicating significantly higher levels of glycoprotein content but no new Bowman's membrane was formed. By four weeks, the scar tissue occupied a 2-3 mm wide region. Immunofluorescence analyses indicated that other major differences in the healing and normally growing stroma were the increased synthesis and deposition of perlecan and laminin. No differences were evident in the immunofluorescence for keratocan or keratan sulfate in the scar tissue, but the scar tissue did contain markedly decreased levels of lumican. Thus, the regulation of proteoglycan and glycoprotein synthesis is altered in the keratocytes that are recruited to the wounded regions in the growing corneal stroma of post-hatched young chicks. While synthesis and deposition of adhesive molecules including laminin and perlecan are elevated, the synthesis of one of the keratan sulfate proteoglycans, lumican, is reduced in the scar tissue as compared to the normally growing stroma.

Extracts of bovine corneal stroma have been shown to activate keratocytes in culture to proliferate. We fractionated stromal extract on a column of Sephacryl S-300 and tested the fractions for mitogenic activity using cell culture and for the presence of IGF-II and its binding protein IGFBP-2 by Western blot. We found that the mitogenic activity in the extract separated into major and minor peaks and that immunologically detectable IGF-II and IGFBP-2 co-eluted with the minor peak. We also compared the effects of 10 ng IGF-II/ml on keratocytes in culture to that of 2 ng TGF-beta/ml over a 7-day culture period. We found that IGF-II and TGF-beta, alone or combined, increased both (3)H-thymidine incorporation and DNA content of the cultures. The phenotype of the cells was determined by using antibodies to alpha-SM (smooth muscle) actin, fibronectin, SPARC, lumican and keratocan in Western blots of cell layers of media. Keratocytes cultured in IGF-II expressed no alpha-SM actin or fibronectin, low levels of SPARC and high levels of lumican and keratocan, indicating a native phenotype. Keratocytes in TGF-beta expressed alpha-SM actin, fibronectin, SPARC and lumican, and expressed no or low levels of keratocan, indicating a myofibroblast phenotype. Keratocytes cultured in IGF-II plus TGF-beta, however, expressed alpha-SM actin, fibronectin, SPARC, lumican, and keratocan by day 7 of culture. The results of this study show that IGF-II to be present in the corneal stroma, to stimulate keratocyte proliferation while maintaining native phenotype and to override the TGF-beta mediated down regulation of keratocan production. The IGF-II in the stroma may serve as a mechanism to immediately activate keratocytes upon wounding and to ameliorate the scarring effects of TGF-beta.

Keratocytes can become fibroblasts and myofibroblasts during corneal injury and wound healing. We used the in vitro bovine keratocyte repair model system, which involves culturing collagenase-isolated keratocytes in serum-free media and then adding serum or serum plus TGF-beta to the culture media to induce the fibroblast and myofibroblast phenotypes, respectively, to evaluate the synthesis of secreted products by the cells. Serum and serum plus TGF-beta rapidly induced the fibroblast morphology and alpha smooth muscle actin, a marker of myofibroblasts. Keratocytes cultured in serum and serum plus TGF-beta also increased the synthesis of several high molecular weight products (approximately 100kD and larger) and the accumulation of a 43kD protein shown to be osteonectin/SPARC by both sequencing tryptic peptides from the protein and by reaction with antisera to osteonectin/SPARC. Immunohistochemical staining of mouse corneas with antisera to SPARC seven days post-wounding also demonstrated an increased accumulation of SPARC in the regions undergoing repair. These results indicate SPARC accumulation is a marker for stromal repair.

Retinal capillaries are composed of endothelial cells resting on a basement membrane, in which are embedded pericytes. In diabetes mellitus, the basement membrane becomes thickened, and there is a loss of pericytes. The relative contributions of endothelial cells and pericytes to the synthesis of extracellular matrix proteins which are components of the basement membrane are not well-characterized. To determine how a selective loss of pericytes might affect the composition of retinal capillary basement membranes, we used primary cultures of bovine retinal capillary endothelial cells and pericytes to determine the forms and quantify the amounts of laminin and fibronectin synthesized and secreted by these cell types as well as to determine how high glucose concentrations alter these parameters. Results of ELISAs showed that pericyte cell/matrix layers contained nearly ten times more fibronectin than endothelial cells (288 +/- 24 vs. 34 +/- 5 ng micrograms-1 DNA, P0.001), but the amounts of laminin were similar. D-glucose (40 mM) tripled the amount of fibronectin incorporated into the endothelial cell/matrix layer (102 +/- 4 vs. 34 +/- 5 ng micrograms-1 DNA, P0.05), but had a lesser effect on pericytes. The non-metabolizable analogue L-glucose, also increased the amount of fibronectin incorporated in both pericyte and endothelial cell/matrix layers. The effects of D- and L-glucose on fibronectin secreted into the medium by both cell types were similar to the effects on incorporation of fibronectin into cell/matrix layers. Glucose had no effect on laminin synthesis. [35S]methionine radiolabeling and immunoprecipitation showed that pericytes and endothelial cells synthesize different forms of fibronectin. Both pericytes and endothelial cells synthesized an A and two B chains of laminin which were of similar apparent size, but the two cell types post-translationally modified the subunits differently. We conclude that pericytes and endothelial cells may contribute different forms and amounts of fibronectin and laminin to the retinal capillary basement membrane, so the preferential loss of pericytes in diabetes could result in basement membrane abnormalities which might lead to endothelial cell dysfunction.

Corneal transparency is dependent on the size and arrangement of collagen fibrils within the corneal stroma. The corneal stroma is composed primarily of collagen type I fibrils and two proteoglycans; one with chondroitin/dermatan sulfate side-chains (decorin) and one with keratan sulfate side-chains (lumican). We investigated the effects of the corneal proteoglycans on corneal collagen fibrillogenesis, utilizing an in vitro assay for fibril formation. Collagen was extracted from bovine corneal stromas with 0·1 M acetic acid and monomers purified by NaCl precipitation, Decorin and lumican were extracted from bovine cornea/stroma with either 0·7 M NaCl or 4 M guanidine HCl and purified by DEAE and Sepharose CL-4B chromatography. Decorin and lumican from both extracts inhibited the rate of collagen fibrillogenesis and the development of turbidity in fibrillogenesis samples. Furthermore, the core proteins of decorin and lumican were shown to be as effective as the intact proteoglycans in inhibiting fibrillogenesis. The decorin core protein isolated from the 0·7 M NaCl extract was determined to be a 20 kDa fragment which lacks the C-terminal half of the core protein. This fragment was ≈ 1/36 as effective in inhibiting fibrillogenesis as intact decorin isolated from guanidine extracts. This suggests that the C-terminal half of the decorin core plays an important role in the interaction of this proteoglycan with collagen. Lumican extracted with 0·7 m NaCl was slightly smaller and was only one-sixth as effective in inhibiting collagen fibril formation as 4 M guanidine extracted lumican. Furthermore reduction and alkylation of lumican core protein abolished the inhibitory activity of the core protein on collagen fibrillogenesis. Electron microscopic examination indicated that fibrils formed in the presence of lumican and lumican core protein were significantly thinner than fibrils formed in the absence of proteoglycans. The results of these studies indicate that in addition to decorin, lumican retards corneal collagen fibrillogenesis and results in the formation of collagen fibrils which are significantly thinner than those formed in the absence of any proteoglycan. The inhibitory activity of lumican or decorin on collagen fibrillogenesis resides in the core proteins of these proteoglycans, not the glycosaminoglycan side chains, and that interaction of the lumican core protein with collagen appears to be dependent on the presence of disulfide bridges within the protein core.

The retinal pigment epithelium of various species, including man, can synthesize extracellular matrix components in vitro. We studied extracellular matrix molecule production by human, rhesus and cynomolgus monkey pigment epithelium in vitro, using a panel of specific antibodies to collagen types I–V, fibronectin, laminin, the basement membrane heparan sulfate poteoglycan, human alpha elastin, glial fibrillary acidic protein (GFAP), and Factor VIII in a modified indirect immunofluorescence reaction. Results were similar between human and both monkey species in early passage cultures with respect to the positive detection of collagen thpes I, III, IV, V, laminin, basement membrane proteoglycan, and fibronectin. The fluorescence was strongest in the center of confluent monolayers and was absent in border or isolated cells. Collagen type II, elastin, GFAP, or Factor VIII were not detected. Based on these findings, human and monkey pigment epithelium in vitro should be useful for the study of the synthesis, deposition and supramolecular interactions of a variety of extracellular matrix molecules.