Molecular Vision 2000; 6:15-23 <http://www.molvis.org/molvis/v6/a4/> Received 20 January 2000 | Accepted 24 February 2000 | Published 7 March 2000

Download Reprint

Extracellular matrix components in retrocorneal fibrous membrane in comparison to corneal endothelium and Descemet's membrane

Eamon W. Leung,¹ Lawrence Rife,¹ Ronald E. Smith,^1,2 EunDuck P. Kay^1,2
 
 

¹Doheny Eye Institute and ²Department of Ophthalmology, Keck School of Medicine of the University of Southern California, Los Angeles, CA

Correspondence to: EunDuck P. Kay, D.D.S., Ph. D., Doheny Eye Institute, 1450 San Pablo Street DVRC #203, Los Angeles, CA, 90033; Phone: (213) 342-6625; FAX: (213) 342-6688; email: ekay@hsc.usc.edu

Abstract

Purpose: To investigate the extracellular matrix macromolecules found in Descemet's membrane and in retrocorneal fibrous membrane (RCFM), and to examine whether the corneal endothelium has the capacity to produce both basement and non-basement membrane phenotypes.

Methods: Rabbit corneas with and without RCFM were analyzed by immunofluorescence using antibodies to 8 different collagens (basement membrane collagens: types IV and VIII; fibrillar collagens: types I and III; interfibrillar collagens: type VI and two spliced variant forms of type XII and one anchoring fiber: type VII), proteoglycans (perlecan and decorin), big-h3 and laminin-1.

Results: Normal corneal endothelium stains positively for all of the tested collagen types except type VII collagen. On the other hand, Descemet's membrane reacts positively only to the type IV collagen antibody. When non-collagenous components in normal cornea were examined, corneal endothelium stained positively for perlecan, decorin, big-h3 and laminin, whereas Descemet's membrane staining for these proteins was negative. When collagenous components of RCFM were examined, RCFM stained positively for all of the tested collagen types except type IV collagen. When non-collagneous components of RCFM were examined, RCFM demonstrated a strong positive staining with decorin, big-h3 and laminin, while perlecan staining was weak.

Conclusions: These observations suggest that corneal endothelium is able to produce both basement membrane phenotypes and non-basement membrane, fibrillar phenotypes. This in vivo study confirms our in vitro model of endothelial mesenchymal transformation, in which corneal endothelial cells are transformed to fibroblasts that are responsible for fibrosis.

Introduction

The retrocorneal fibrous membrane (RCFM), first described by Fuchs in 1901 [1], has been observed in various clinical conditions associated with disease and damage to the corneal endothelium [2-4]. The presence of RCFM posterior to the preexisting Descemet's membrane (DM) is thought to represent an end-stage disease process of the corneal endothelium that results in functional alterations of the corneal endothelium, leading to corneal opacity and blindness. Our earlier study demonstrated that fibroblasts isolated from RCFM synthesized and secreted type I collagen, the major constituent of RCFM [5]. To determine the origin of the fibroblasts present within this membrane, we established an in vitro model, which demonstrated that corneal endothelial cells (CEC) undergo endothelial mesenchymal transformation by the action of fibroblast growth factor 2 (FGF-2) and a protein factor released by polymorphonuclear leukocytes [6-9]. As a consequence of this transformation, the transformed CEC produce type I collagen, the major component of the fibrillar extracellular matrix.

An understanding of the corneal endothelium/Descemet's membrane complex, including physiologic functions, transparency, and hydration, as well as disorders such as wound healing and some dystrophies, requires a fundamental knowledge of extracellular matrices. Despite the fact that the presence of RCFM in clinical conditions has long been known, a systematic evaluation regarding the composition of extracellular matrix in RCFM is essentially nonexistent. Furthermore, the evaluation of extracellular matrix proteins in DM has focused predominantly on type IV collagen and laminin, the major constituents of the basement membrane. In recent years, it has been known that the corneal endothelium is able to synthesize fibrillar collagens, such as type I collagen, and that these fibrillar collagens are subsequently destroyed inside the cell [10]. This unexpected observation suggests the complexity of the cellular activity of the corneal endothelium and further suggests that those collagens that are synthesized by the corneal endothelium are not necessarily coincidental to the components of DM. An elucidation of the biochemical composition of the matrices is fundamental to an understanding of corneal function and its pathobiology. In this report, we determined the extracellular matrix proteins present in the corneal endothelium, DM, and RCFM, including a variety of collagen types, proteoglycans, and laminin.

Methods

Production of RCFM

This procedure was performed as described previously [5]. In brief, 8-week-old New Zealand white rabbits were used for all experiments. Animals were sedated with ketamine, and anesthesia was induced with intravenous sodium pentobarbital administered through an ear vein. Proparacaine solution (Alcaine: Alcon Laboratory, Inc., Ft. Worth, TX) was applied topically to the cornea, and a lid speculum was inserted. Transcorneal freezing was performed with an Alcon surgical cryostat unit (DNE-300 U) using nitrous oxide with a probe temperature of -80 °C. The 2.5 mm probe was first applied to the midcornea for 30 sec. The superior, inferior, temporal, and nasal corneal quadrants were then frozen in a similar manner, with slight overlap of the central application spot. After the procedure, atropine and gentamicin sulfate (Garamycin; Schering Corp., Kenilworth, NJ) were instilled in each eye. The freezing procedure was performed at evenly spaced biweekly intervals for four sessions. The animals were followed by external eye examination and slit-lamp microscopy and were sacrificed 1 week after the final cryoapplication. At least one eye per cryoapplication group was examined histologically to determine whether there was contamination by corneal keratocytes. Corneas from normal animals were used in comparison studies. All experiments were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the University of Southern California Animal Care and Use Committee.

Tissue Preparation

Fresh rabbit corneas and corneas containing RCFM were fixed with 3.5% formalin, paraffin-embedded, and then cut in 5 mm-thick sections onto slides. The slides were heated at 60 °C for 1 h and then deparaffinized through three 2-min changes each of xylene, absolute alcohol, and 95% ethanol. Tissues were then rinsed in 0.1 M phosphate-buffered saline (PBS) for 2 min. In some cases, freshly removed rabbit corneas were embedded in OCT compound (Miles, Elkhart, IN) followed by freezing in liquid nitrogen. Eight mm sections were cut on a Cryocut 1800 (Reichert-Jung, West Germany) and collected on glass slides coated with polylysine.

Immunohistochemistry

All washes and incubations were carried out in PBS at room temperature. Tissue sections were treated with buffer A (1% bovine serum albumin [BSA], 0.1% Triton X-100 in PBS) for 15 min at room temperature. Tissues were incubated with primary antibody (1:10 dilution in buffer A) for 1 h at 37 °C in a moist chamber, then washed with PBS. Tissues were next incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody (1:100 dilution in buffer A) for 45 min at room temperature. Following extensive washes, antibody labeling was examined using a Zeiss LSM 210 laser scanning confocal microscope equipped with a barrier filter for fluorescein epi fluorescence. A plan-neofluar x40 (N.A. 1.3) oil immersion objective lens was used for imaging of fluorescently labeled tissues. Image analysis was performed using the standard system operating software provided with the Zeiss LSM microscope (Version 2.08). Color photomicrographs were taken using a Sony printer connected to the video output of the microscope.

Antibodies

The following antibodies were used: goat anti-type I collagen antibody (Chemicon International Inc., Temecula, CA); sheep anti-type III collagen antibody (a gift from Edward MacCarak, University of Pennsylvania, Philadelphia, PA); monoclonal mouse anti-type IV collagen antibody and monoclonal mouse anti-type XII (long form and short form) antibodies (gifts from Dr. Nirmala SundarRaj, University of Pittsburgh, Pittsburgh, PA); monoclonal mouse anti-type VI collagen antibody (a gift from Eva Engvall, The Burnham Institute, La Jolla, CA); monoclonal mouse anti-type VII collagen antibody (a gift from Robert Burgeson, Harvard Medical School, Boston, MA); monoclonal mouse anti-type VIII collagen antibody (Seikagaku America, Inc, Ijamsville, MD); monoclonal mouse anti-perlecan antibody (Zymed Laboratories Inc., South San Francisco, CA); sheep anti-decorin antibody (United States Biological, Swampscott, MA); chicken anti-big-h3 antibody (a gift from Charles Cintron, Schepens Eye Institute, Boston, MA); goat anti-laminin antibody (a gift from Dr. Yoshihiko Yamada, NIH, Bethesda, MD). FITC-conjugated donkey antibody against goat IgG, FITC-conjugated rabbit antibody against chicken IgG, FITC-conjugated donkey antibody against sheep IgG, and FITC-conjugated goat antibody against mouse IgG were purchased from Vector Laboratories Inc (Burlingame, CA).

Data Presentation

Negative controls for all experiments were stained in the absence of the primary antibody in parallel. In some cases, we further examined negative controls using FITC-conjugated secondary antibodies made in different species from the primary antibody. The routinely performed specificity controls were negative in all experiments; therefore, some of these were not shown. The phase-contrast micrographs were also examined in parallel to the immunohistochemical analysis in all experiments and representative phase-contrast micrographs were shown in Figure 1A,D. The data are representative of 2-4 immunohistochemical analyses per antibody staining.

Results

In situ collagen types in corneal endothelium/Descemet's membrane and RCFM

Light micrographs of corneal endothelium, DM, and RCFM were analyzed using hematoxylin and eosin staining (data not shown): tissue integrity was confirmed in all tissues and the RCFMs were highly cellular, as observed previously [5]. Phase-contrast micrographs showed the integrity of the tissues in corneal endothelium, DM and RCFM (Figure 1A,D). Phase-contrast micrographs were not shown for the subsequent experiments, since the tissue morphologies were very similar to those shown in Figure 1A,D. The staining pattern of type IV collagen, a major basement membrane collagen type, was examined: antibody against type IV collagen, composed of a1(IV) and a2(IV), strongly stained both DM and corneal endothelium (Figure 1C). Control tissues stained in the absence of the primary antibody showed negative staining in corneal endothelium and DM (Figure 1B) and RCFM (Figure 1E). The RCFM-containing cornea showed positive staining with the anti-type IV collagen antibody in DM (Figure 1F). On the other hand, the stromal tissue anterior to DM and the RCFM tissue posterior to DM were not stained with anti-type IV collagen antibody (Figure 1F). The staining pattern for type VIII collagen, another major basement membrane collagen, was also determined: anti-type VIII collagen antibody strongly stained corneal endothelium but not DM (Figure 2B). Interestingly, keratocytes in the corneal stroma showed a positive reaction to the antibody (Figure 2B). Although RCFM tissue showed positive staining with the antibody, the same antibody did not stain DM in between RCFM and the corneal stroma (Figure 2D). Omission of the primary antibody removed the staining potential from corneal endothelium, DM and RCFM (Figure 2A,C).

Our previous studies, using immunohistochemical and biochemical analysis, showed that type I collagen is synthesized by CEC but is not secreted into the DM [10]. Type XII collagen, a member of the fibril-associated collagen with interrupted triple-helices (FACIT), is also reported to be present in corneal endothelium but not in the DM [11]. Therefore, we examined in situ collagen types that involve collagen fibril formation. The in vivo localization of type I collagen in corneal endothelium, as previously reported [10], was confirmed; the corneal endothelium was stained with anti-type I collagen antibody, whereas the underlying DM demonstrated no staining (Figure 3B). In contrast, RCFM tissue showed a strong positive reaction to the anti-type I collagen antibody, whereas DM anterior to RCFM was not stained with the antibody (Figure 3D). Control tissues stained in the absence of the primary antibodies showed no staining in these tissues (Figure 3A,C). When in vivo localization of type III collagen was examined, the corneal endothelium showed a faint positive reaction with anti-type III collagen antibody, while the underlying DM did not react with the antibody (Figure 4B). On the other hand, RCFM showed a strong positive reaction and DM did not react with the antibody (Figure 4D). The negative controls demonstrated no staining in the absence of the primary antibodies (Figure 4A,C).

Type XII collagen, which is known to be associated with collagen fibrils [12-14], is expressed in at least two alternatively spliced variant forms consisting of 340-kDa (long form) and 220-kDa (short form) chains [15,16]. The staining profiles of the two-splice variant forms of type XII collagen were analyzed using respective antibodies (2E4 antibody for long variant form and 3C7 antibody for short variant form). Although corneal endothelium reacted with the 2E4 antibody, the underlying DM did not react (Figure 5B); and the staining profile of long variant form of type XII collagen is identical to the published data [11]. RCFM tissue showed a faint positive staining with the antibody, while the anterior DM did not react with the antibody (Figure 5D). The corneal stroma demonstrated a faint positive staining with the antibody (Figure 5B,D). The negative controls performed in the absence of the primary antibody showed no staining with the antibody in corneal endothelium, DM, or RCFM (data not shown). We employed another set of negative control experiments in which we used FITC-conjugated secondary antibodies made in different species from the primary antibody: in Figure 5A and Figure 5C, none of the three tissues (endothelium, DM and RCFM) was stained by this procedure. When the short variant form of type XII collagen was examined, the staining profiles were similar to those of the long variant form of type XII collagen. Corneal endothelium reacted strongly with the antibody, but the underlying DM did not react with the antibody (Figure 6B). The corneal stroma was faintly stained with the antibody. RCFM showed a positive reaction with the antibody, while DM in between RCFM and corneal stroma did not react with the antibody (Figure 6D). The corneal stroma demonstrated a faint positive staining with the antibody (Figure 6B,D). The negative controls showed no staining in the absence of the primary antibodies (Figure 6A,C).

Type VI collagen, forming beaded filaments, is abundant in the mature stroma and type VII collagen, forming the anchoring fibrils, is present as an epithelial attachment complex. Although there has been no report of these collagens in DM, their in vivo localization in the corneal endothelium, DM, and RCFM was examined to determine whether corneal endothelium is able to produce these collagens as it does fibrillar collagens. Corneal endothelium reacted strongly with anti-type VI collagen antibody, but the underlying DM did not react with the antibody (Figure 7B). RCFM and the corneal stroma reacted strongly with the antibody (Figure 7D). The staining pattern of the negative controls was negligible (Figure 7A,C). On the other hand, none of these tissues (endothelium, DM and RCFM) reacted with anti-type VII collagen antibody (Figure 7E,F), while the corneal stroma showed a faint positive staining with the antibody. The negative controls demonstrated no staining (data not shown).

In situ localization of non-collagenous extracellular matrix macromolecules in corneal endothelium, Descemet's membrane, and RCFM

Immunohistological analysis of proteoglycans (perlecan as a basement membrane component and decorin as a non-member of basement membrane) in corneal endothelium, DM and RCFM showed that the perlecan antibody reacted primarily with the corneal endothelium, while the underlying DM did not react with the antibody (Figure 8B), when compared to the staining intensity of the negative controls (Figure 8A). RCFM showed positive reaction with the antibody (Figure 8D) and the negative controls showed no staining (Figure 8C). On the other hand, the decorin antibody strongly stained corneal endothelium but did not stain DM (Figure 9B). RCFM strongly reacted with the decorin antibody, whereas the anterior and posterior borders of DM showed a faint reaction with the antibody (Figure 9D). The corneal stroma demonstrated a strong positive staining with the antibody (Figure 9D). Control tissues stained in the absence of the primary antibodies showed no staining in corneal endothelium, DM or RCFM (Figure 9A,C). The protein big-h3 has been found in the ECM of the cornea, where it appears to be associated with several different types of collagens [17-19]. We examined the in situ localization of big-h3 in corneal endothelium, the underlying DM and RCFM. Corneal endothelium strongly reacted with the anti-big-h3 antibody but DM did not react with the antibody (Figure 10B). RCFM, on the other hand, was stained with the antibody, whereas the anterior DM was not stained (Figure 10D). Keratocytes in the corneal stroma showed a positive staining with big-h3 antibody (Figure 10B). Negative controls demonstrated no staining (Figure 10A,C). When the in situ localization of laminin-1 was examined, corneal endothelium was stained with anti-laminin-1 antibody, but the underlying DM was not stained with the antibody (Figure 11B). It is of interest that RCFM, the stromal keratocytes, the stromal side of DM, and the posterior borders of DM adjacent to RCFM were all strongly stained with the antibody (Figure 11D). Negative controls showed no positive staining (Figure 11A,C).

Discussion

Descemet's membrane is a specialized matrix that separates the corneal endothelium and the stromal matrix. This membrane is called the basement membrane of the corneal endothelium because of its location and ultrastructral appearance. Unlike other basement membranes, which are predominantly composed of laminin and type IV collagen, DM contains mostly type VIII collagen in addition to type IV collagen [20-23]. When a cornea is exposed to a variety of insults, the corneal endothelium can heal by forming a scar, clinically called RCFM [1-3,24,25]. This pathological membrane consists of fibrous ECM and elongated cells that are responsible for the fibrillar phenotypes. While it is known that type I collagen is the major constituent of RCFM, the biochemical composition of RCFM has not yet been determined. Therefore, we attempted to identify the ECM components of RCFM in comparison with those in DM using immunofluorescence microscopic analysis. We further examined whether corneal endothelium is responsible for the expression of both basement membrane and non-basement membrane phenotypes. For this purpose, we used paraformaldehyde-fixed tissues, in which corneal endothelial layers were well preserved.

When the distribution of the classic type IV collagen, composed of a1(IV) and a2(IV) chains, was studied, the molecule was seen in the entire DM and in the corneal endothelium, whereas RCFM did not react with the type IV collagen antibody. The staining pattern of DM with the type IV collagen antibody was in agreement with the previous report [26] in which the cryostat sections were pretreated with 0.1 M acetic acid prior to antibody staining. These staining patterns in the entire DM differ from those reported by Ljubimov et al [22], in which the antibody made against the a2(IV) collagen chain predominantly stained the stromal face of the DM. The authors subsequently proposed that keratocytes produce the classic type IV collagen [22]. The difference in the staining pattern of the classic type IV collagen among these studies may not be due to the spatial segregation of a1(IV) and a2(IV) collagen chains produced by the stromal keratocytes. The positive staining of type IV collagen in corneal endothelium (Figure 1B) does not confirm the proposition. Furthermore, we have reported that corneal endothelium contains a2(IV) collagen mRNA and that a1(IV) and a2(IV) collagen chains are synthesized and secreted by CEC [27,28]. The present finding, shown in Figure 1C, confirms that the classic type IV collagen, a product of corneal endothelium, may be present in the entire DM [26].

The DM contains stacks of hexagonal lattices that are arranged parallel to the surface of the membrane [29]. Type VIII collagen most likely forms this hexagonal nodal/internodal network [20,30]. When corneal tissue was stained with anti-type VIII collagen antibody, the antibody stained corneal endothelium but did not stain DM. The absence of staining in DM in the present study differs from previous studies [20,26] in which the entire DM was stained with the type VIII collagen antibody. What causes the difference in the staining patterns is not known; this discrepancy may be due to the masking of epitopes of type VIII collagen in rabbit DM. However, the masking of epitopes is not caused by the use of fixatives, because one of the previous studies [20] was performed with 4% paraformaldehyde-fixed adult bovine corneas, as was the present study. When corneas containing RCFM were stained with the antibody, positive staining of RCFM was observed, suggesting that the transformed endothelial cells in RCFM have the ability to produce type VIII collagen, which is subsequently deposited into the ECM. Hogan reported the characteristic hexagonal figures within the DM (most likely RCFM tissue) in Fuch's endothelial dystrophy before type VIII collagen and its supramolecular organization had been determined [31]. The positive staining of type VIII collagen in RCFM confirms Hogan's finding in Fuch's endothelial dystrophy.

In our previous study [10], we reported that corneal endothelium produces type I procollagen, which is subsequently subjected to intracellular degradation. Type XII collagen is also found in corneal endothelium but not in DM [11]. Ultrastructural studies on the development of rabbit corneal endothelium demonstrate that fibrils deposited on the basal side of the corneal endothelium and subsequently incorporated into DM are synthesized by corneal endothelium [32]. Taken together, these findings suggest that corneal endothelium is able to synthesize not only basement membrane collagens but non-basement membrane collagen types. However, in normal adult corneal endothelium, unlike the fetal corneal endothelium, these non-basement membrane phenotypes should be intracellularly degraded before they are secreted into the ECM for subsequent deposit in DM. The in vivo localization of the fibrillar collagens (I and III), interfibrillar collagens (two splice variant forms of XII and VI), and type VII collagen (anchoring fiber), therefore, was determined in corneal endothelium, DM and RCFM. Types I, III, and VI and both splice-variant forms of type XII collagen are present in corneal endothelium but not in DM, while all of these collagens are present as components of RCFM. The absence of these collagens in DM indicates that corneal endothelium produces non-basement membrane collagens and destroys them before their secretion into the DM. This phenomenon has been studied in detail with type I collagen [10,33], and the present study further indicates that a similar mechanism of intracellular degradation may be applicable to the other non-basement membrane collagens in corneal endothelium. On the other hand, type VII collagen is absent in corneal endothelium, DM and RCFM. It is likely that type VII collagen is present only in the epithelial attachment complex, in which the molecule stabilizes the epithelial basement membrane to the underlying stroma. Linsenmayer et al. suggest that type IV collagen in DM may be a functional homologue of type VII collagen in corneal epithelium [30].

Matrix assembly is dependent on molecular interactions between fibrillar components (mostly collagen fibrils) and interfibrillar macromolecules that bind to fibril surfaces. Among the interfibrillar non-collagenous components that are abundant in corneal tissues are proteoglycans and big-h3. big-h3 is disulfide linked to the globular domain of native type VI collagen; thus the protein is copurified with native type VI collagen [17]. Because of the characteristic behavior of big-h3, we examined its distribution. As shown by its staining pattern, big-h3, like type VI collagen, is present in corneal endothelium but not in DM. RCFM shows a strong positive staining. When the distribution of the two proteoglycan molecules, perlecan (basement membrane phenotype) and decorin (non-basement membrane phenotype), was examined, perlecan was observed in corneal endothelium but not in DM. This staining pattern is different from the one discussed in the previous report [22], in which the endothelial surface of DM is positively stained, as opposed to the lack of staining of DM in the present study. Since corneal endothelium was not present in the previous work [22], a direct comparison of the two studies may hamper the information obtained from these studies. Decorin is seen in corneal endothelium, in the stromal side of DM and in the RCFM. The findings for decorin confirm that corneal endothelium is able to produce a non-basement membrane type of proteoglycan that exists predominantly in the corneal stroma. When the distribution of laminin was examined, corneal endothelium and RCFM were positive for staining with the classic laminin (laminin-1), whereas DM was negative for laminin staining. This observation is different from that reported in the previous work [22], in which the endothelial face of DM demonstrates positive staining. The differences in the staining pattern of basement membrane phenotypes (type IV collagen, perlecan and type VIII collagen) between the present study and the previous report [22] requires close examination.

The present study confirms that corneal endothelium has the capacity to produce non-basement membrane phenotypes, which are subsequently degraded before secretion. RCFM, on the other hand, mostly expresses fibrillar phenotypes although residual production of basement membrane phenotypes is also observed. This further suggests that the cells responsible for RCFM production are indeed the corneal endothelial cells. This in vivo finding supports our in vitro model of endothelial mesenchymal transformation of corneal endothelial cells.

Acknowledgements

Supported by grants EY 06431 and EY 03040 from the National Institutes of Health, Bethesda, Maryland, and by an unrestricted grant from Research to Prevent Blindness, New York, New York.

References

1. Fuchs E. On keratoplasty. Zeitschrift für Augenheilkunde 1901; 5:1-5.

2. Rodrigues MM, Waring GO, Laibson PR, Weinreb S. Endothelial alterations in congenital corneal dystrophies. Am J Ophthalmol 1975; 80:678-89.

3. Waring GO, Laibson PR, Rodrigues M. Clinical and pathologic alterations of Descemet's membrane: with emphasis on endothelial metaplasia. Surv Ophthalmol 1974; 18:325-68.

4. Brown SI, Kitano S. Pathogenesis of the retrocorneal membrane. Arch Ophthalmol 1966; 75:518-25.

5. Kay ED, Cheung CC, Jester JV, Nimni ME, Smith RE. Type I collagen and fibronectin synthesis by retrocorneal fibrous membrane. Invest Ophthalmol Vis Sci 1982; 22:200-12.

6. Kay EP, Nimni ME, Smith RE. Modulation of endothelial cell morphology and collagen synthesis by polymorphonuclear leukocytes. Invest Ophthalmol Vis Sci 1984; 25:502-12.

7. Kay EP, Gu X, Ninomiya Y, Smith RE. Corneal endothelial modulation: a factor released by leukocytes induces basic fibroblast growth factor that modulates cell shape and collagen. Invest Ophthalmol Vis Sci 1993; 34:663-72.

8. Kay EP, Gu X, Smith RE. Corneal endothelial modulation: bFGF as direct mediator and corneal endothelium modulation factor as inducer. Invest Ophthalmol Vis Sci 1994; 35:2427-35.

9. Gu,X, Seong GJ, Lee YG, Kay EP. Fibroblast growth factor 2 uses distinct signaling pathways for cell proliferation and cell shape changes in corneal endothelial cells. Invest Ophthalmol Vis Sci 1996; 37:2326-34.

10. Kay EP, Gu X, Choi SH, Ninomiya Y. Posttranslational regulation of type I collagen in corneal endothelial cells. Invest Ophthalmol Vis Sci 1996; 37:11-9.

11. Wessel H, Anderson S, Fite D, Halvas E, Hempel J, SundarRaj N. Type XII collagen contributes to diversities in human corneal and limbal extracellular matrices. Invest Ophthalmol Vis Sci 1997; 38:2408-22.

12. Keene DR, Lunstrum GP, Morris NP, Stoddard DW, Burgeson RE. Two type XII-like collagens localize to the surface of banded collagen fibrils. J Cell Biol 1991; 113:971-8.

13. Koch M, Bohrmann B, Matthison M, Hagios C, Trueb B, Chiquet M. Large and small splice variants of collagen XII: differential expression and ligand binding. J Cell Biol 1995; 130:1005-14.

14. Nishiyama T, McDonough AM, Bruns RR, Burgeson RE. Type XII and XIV collagens mediate interactions between banded collagen fibers in vitro and may modulate extracellular matrix deformability. J Biol Chem 1994; 269:28193-9.

15. Trueb J, Trueb B. The two splice variants of collagen XII share a common 5' end. Biochim Biophys Acta 1992; 1171:97-8.

16. Oh SP, Griffith CM, Hay ED, Olsen BR. Tissue-specific expression of type XII collagen during mouse embryonic development. Dev Dyn 1993; 196:37-46.

17. Gibson MA, Kumaratilake JS, Cleary EG. Immunohistochemical and ultrastructural localization of MP78/70 (betaig-h3) in extracellular matrix of developing and mature bovine tissues. J Histochem Cytochem 1997; 45:1683-96.

18. El-Shabrawi Y, Kublin CL, Cintron C. mRNA levels of alpha1(VI) collagen, alpha1(XII) collagen, and beta ig in rabbit cornea during normal development and healing. Invest Ophthalmol Vis Sci 1998; 39:36-44.

19. Rawe IM, Zhan Q, Burrows R, Bennett K, Cintron C. Beta-ig. Molecular cloning and in situ hybridization in corneal tissues. Invest Ophthalmol Vis Sci 1997; 38:893-900.

20. Sawada H, Konomi H, Hirosawa K. Characterization of the collagen in the hexagonal lattice of Descemet's membrane: its relation to type VIII collagen. J Cell Biol 1990; 110:219-27.

21. Labermeier U, Kenney MC. The presence of EC collagen and type IV collagen in bovine Descemet's membranes. Biochem Biophys Res Commun 1983; 116:619-25.

22. Ljubimov AV, Burgeson RE, Butkowski RJ, Michael AF, Sun TT, Kenney MC. Human corneal basement membrane heterogeneity: topographical differences in the expression of type IV collagen and laminin isoforms. Lab Invest 1995; 72:461-73.

23. Kay EP, Smith RE, Nimni ME. Basement membrane collagen synthesis by rabbit corneal endothelial cells in culture. Evidence for an alpha chain derived from a larger biosynthetic precursor. J Biol Chem 1982; 257:7116-21.

24. Waring GO 3d. Posterior collagenous layer of the cornea. Ultrastructural classification of abnormal collagenous tissue posterior to Descemet's membrane in 30 cases. Arch Ophthalmol 1982; 100:122-34.

25. Bloomfield SE, Jakobiec FA, Iwamoto T. Fibrous ingrowth with retrocorneal membrane. Ophthalmology 1981; 88:459-65.

26. Kapoor R, Bornstein P, Sage EH. Type VIII collagen from bovine Descemet's membrane: structural characterization of a triple-helical domain. Biochemistry 1986; 25:3930-7.

27. Kay EP. Expression of types I and IV collagen genes in normal and in modulated corneal endothelial cells. Invest Ophthalmol Vis Sci 1989; 30:260-8.

28. Kay EP, He YG. Post-transcriptional and transcriptional control of collagen gene expression in normal and modulated rabbit corneal endothelial cells. Invest Ophthalmol Vis Sci 1991; 32:1821-7.

29. Jakus MA. Studies on the cornea. II. The fine structure of Descemet's membrane. The Journal of Biophysical and Biochemical Cytology 1956; 2:S243-52.

30. Linsenmayer TF, Fitch JM, Gordon MK, Cai CX, Igoe F, Marchant JK, Birk DE. Development and roles of collagenous matrices in the embryonic avian cornea. Prog Retin Eye Res 1998; 17:231-65.

31. Hogan MJ, Wood I, Fine M. Fuch's endothelial dystrophy of the cornea. Am J Ophthalmol 1974; 78:363-83.

32. Cintron C, Covington HI, Kublin CL. Morphogenesis of rabbit corneal endothelium. Curr Eye Res 1988; 7:913-29.

33. Ko MK, Kay EP. Hsp47-dependent and -independent intracellular trafficking of type I collagen in corneal endothelial cells. Mol Vis 1999; 5:17 <http://www.molvis.org/molvis/v5/a17/>.