Immunofluorescence staining for desmoglein 3 (green) and propidium iodide (red) in the limbal epithelial crypt (original magnification ×200). A, Central cells. B, Basal cells. C, Limbus. D, Cornea. E, Cornea (negative control).
Immunofluorescence staining for tenascin C (red) and 4′,6-diamidino-2-phenylindole (blue) (original magnification ×100). A, Cornea. B, Limbus. C, Limbal epithelial crypt corresponding to zone 1. D, Cornea. E, Limbus corresponding to zone 2. F, Cornea. G, Limbus corresponding to zone 3.
Diagrammatic representation of limbal regional variations. 1 indicates zone 1, corresponding to the area of limbus with the limbal epithelial crypt; 2, zone 2, corresponding to the area of limbus adjoining the limbal epithelial crypt; and 3, zone 3, corresponding to the area of limbus distant to the limbal epithelial crypt. Curved black line indicates cornea; yellow, limbus associated with the limbal epithelial crypt; dark blue, limbal epithelial crypt; light blue, sclera; and white, cornea.
Diagrammatic representation of stem cells within the limbal epithelial crypt (LEC) in relation to limbus and cornea. Black circles indicate stem cells; red circles, transient amplifying cells; light blue circles, terminally differentiated cells; dark blue, LEC; yellow, limbus; black line, cornea; and arrow, direction of cell migration.
Immunofluorescence staining for connexin 43 (green) and propidium iodide (red) (original magnification ×200). A, Peripheral cornea. Note the abrupt cessation of staining for connexin 43 at the junction of peripheral cornea and limbus (corresponding to zone 3, as described in the “Results” section). B, The limbal epithelial crypt and adjacent limbus. Note the connexin 43–positive cells within the limbal epithelial crypt and in adjacent limbus (corresponding to zone 1). C, Central cornea, showing connexin 43 staining as expected. Reprinted courtesy of the British Journal of Ophthalmology.17
Yeung AM, Schlötzer-Schrehardt U, Kulkarni B, Tint NL, Hopkinson A, Dua HS. Limbal Epithelial CryptA Model for Corneal Epithelial Maintenance and Novel Limbal Regional Variations. Arch Ophthalmol. 2008;126(5):665-669. doi:10.1001/archopht.126.5.665
Copyright 2008 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2008
To determine the distribution of cell membrane proteins and extracellular matrix proteins around the limbal epithelial crypt (LEC) compared with adjacent limbus and corneal epithelium.
Serial histological sections of human corneoscleral limbus rims were stained with antibodies of interest by standard immunohistochemistry.
Superficial cells of the limbus were desmoglein 3 positive, compared with the negative basal cells of the limbus that correspond to cells with more stemlike properties. The LEC had a much lower proportion of desmoglein 3 staining in comparison. Tenascin C staining demonstrated regional variations of the limbus depending on their association with the LEC. Limbus that was associated with or adjacent to the LEC had a greater tenascin C expression compared with normal limbus, whereas the LEC demonstrated the greatest tenascin C expression.
Based on these and similar results previously reported for connexin 43, we propose a novel model on the mechanism of corneal surface epithelium maintenance involving 3 different limbal regions: zone 1, limbus including the LEC; zone 2, limbus associated with the LEC; and zone 3, limbus distant to the LEC.
The noted limbal variations may influence the selection of the donor site for limbal grafts in the future.
Stem cells (SCs) are slow-cycling cells that divide to create daughter cells, which in turn can proliferate rapidly, differentiate, and replenish cells lost through natural turnover or injury. They thus serve to repair and regenerate tissues and have the potential for clinical applications in regenerative medicine and tissue engineering. The corneoscleral limbus has long been regarded as a reservoir of SCs that serve the corneal epithelium.1,2 Since the first report of successful limbal transplantation,3 several variations of limbal autografts and allografts have been described with good reconstruction of the corneal epithelial surface.4- 6
The SC niche (SCN) is a specialized microenvironment where SCs are nurtured and protected and have self-regulatory control in their dividing capacity. The SCNs have been well characterized in bone marrow for hematopoietic SCs, in the intestinal crypts within the gastrointestinal tract,7,8 and in the bulge region of the hair follicle for the skin.9,10 The SCN concept was originally proposed in 1978,11 and further studies on Drosophila have shown that Drosophila E-cadherin–mediated adhesion is required to maintain SCs in an undifferentiated state,12 indicating the importance in cell-cell adhesion interaction in the maintenance of stemness of SCs.
Desmoglein 3, a cell membrane protein, is a member of the group desmogleins, which are part of the family of cadherins. It functions to assist in the formation of desmosomes that allow cell-cell adhesion. Clusters of basal cells at the tips of rete ridges in human palm skin that demonstrate very low levels of desmoglein 3 expression on the cell surface have high colony-forming efficiency and clonogenecity,13 which are indicative of cells with a stemlike nature. Altered or modified cell-cell adhesion appears to be a characteristic of SCs; cell-cell adhesion may thus be a characteristic that SCs portray. Tenascin C is an extracellular matrix protein that is situated in the basement membrane and is thought to maintain limbal SCs in a particular state of slow-cycling ability and to divide when necessary. Human cornea tenascin C has been shown to be widely expressed in the preterm infant and less so in neonates, and it is restricted to the limbus in the child and adult.14,15 Therefore, it may play a role in growth and differentiation of SCs and in corneal development. Both desmoglein 3 and tenascin C have been shown to have a role in SC attachment and may be related to SCN functions.
Our group made the novel discovery that in certain areas of the human limbus, solid cords of epithelial cells emanate from the posterior end of the limbal palisades and extend into the underlying stroma. These structures were termed the limbal epithelial crypt16 (LEC) and were further characterized in 2007,17 demonstrating that basal cells of the LEC were smaller than basal cells in adjacent rete pegs and also smaller than suprabasal limbal and central corneal epithelial cells. Furthermore, basal cells of the LEC demonstrated a higher nuclear to cytoplasmic ratio, thus supporting the hypothesis of the LEC being a putative SCN.
The aim of this study was to examine the distribution of desmoglein 3 and tenascin C in the LEC compared with the adjacent limbus and corneal epithelium. Regional variations in the distribution of the proteins examined together with existing knowledge of differences between the LEC, limbus, and peripheral cornea have enabled us to propose a model by which the LEC may maintain the corneal surface epithelium.
Tissue samples were collected and prepared as previously described.17 In brief, 3 pairs of consented human cadaver eyes removed within 48 hours were used with the approval of the local ethics committee. There was no evidence of any disease, desiccation, or damage. A 360° circumferential frill of conjunctiva was retained and a corneoscleral disc was then punched out using a 17-mm trephine. The disc was divided into 8 equal parts, snap frozen in optimum temperature compound (Emitech Ltd, East Sussex, England) with liquid nitrogen, and stored at −80°C until further use. Each of the blocks was sectioned serially in 5- to 7-μm sections with a cryostat (Leica Microsystems Ltd, Milton Keynes, England) and monitored for the presence of the LEC under the light microscope. On identification of an LEC, preceding and subsequent unstained sections were used for immunofluorescence staining. All of the experiments on human tissue were approved by our institutional research ethics committee.
Samples were air dried and fixed in acetone, followed by blocking in normal goat serum (dilution, 1:10). Samples were incubated with primary antibodies against desmoglein 3 (clone 5G11; dilution, 1:25; mouse; Zymed Laboratories, Inc, South San Francisco, California) and tenascin C (clone H-300; dilution, 1:250; rabbit; Santa Cruz Biotechnology, Inc, Santa Cruz, California) overnight at 4°C. The samples were washed and incubated with secondary antibodies Alexa Fluor 555 (antirabbit) and 488 (antimouse) (Invitrogen Ltd, Paisley, Scotland) for 1 hour. Samples were counterstained with propidium iodide and 4′,6-diamidino-2-phenylindole, mounted with glycerol mounting medium (Dako, Glostrup, Denmark), and examined under an Alphaphot-2 fluorescent microscope (Nikon UK, Surrey, England) or a DMRB microscope (Leica Microsystems Ltd). Images were either captured on a D70S digital camera (Nikon UK) using Camera Control software (Nikon UK) or a C4742-95 digital camera (Hamamatsu, Middlesex, England) using Openlab software (Improvision, Coventry, England). All of the images were edited on Photoshop CS2 software (Adobe Systems Inc, San Jose, California).
The more superficial and central body of the LEC was mostly desmoglein 3 negative (Figure 1A), with some regions of positive staining; however, these areas were not as bright as the staining of superficial limbal cells. When examining the deepest part of the LEC, the basal regions were found to be desmoglein 3 negative (Figure 1B), especially within the deepest recess of the LEC. At the limbus, the superficial cells were desmoglein 3 positive, in contrast to the basal cells of the limbus (Figure 1C). Desmoglein 3 staining of the cornea was consistent throughout (Figure 1D).
Tenascin C staining demonstrated regional variations depending on the zone from which sections of the limbus were taken (Figure 2). Zone 1 (Figure 3) consists of a cross section through clear cornea, limbus adjacent to the LEC, and the LEC. Zone 2 consists of a section through clear cornea and limbus that is adjacent to the LEC but not involving the LEC. Zone 3 consists of clear cornea and normal limbus that has no relationship to the LEC. Cornea in all of the zones was negative for tenascin C (Figure 2A, D, and F). Zone 1 LEC demonstrated the strongest staining of tenascin C (Figure 2C) compared with its adjacent limbus (Figure 2B). Zone 2 limbus (Figure 2E) showed tenascin C expression similar to that in zone 1 limbus. Tenascin C expression in zone 3 limbus (Figure 2G) was weakly positive.
We propose a novel model for corneal epithelial cell regeneration and maintenance (Figure 3) with 3 distinct limbal zones determined by the relationship of the limbus with the LEC. The LEC has specific characteristics that support its role as a potential limbal SCN. The role it plays with corneal surface epithelial maintenance is summarized in Figure 4. Limbal SCs reside in the LEC. When regulatory factors are released to stimulate SCs to divide, the SCs divide and their daughter cells migrate from out of the LEC and toward the adjacent limbus to become transient amplifying cells. Here, they may divide and again will be recruited to move toward the cornea when required, where they differentiate into terminally differentiated cells.
The LEC demonstrates a different distribution of desmoglein 3 staining compared with that of the adjacent limbus. The negative desmoglein 3 staining of basal cells of the limbus (Figure 1C) could represent the potential SC population that enables current successful transplantation of limbal grafts for SC deficiency. As the LEC has a higher number of desmoglein 3–negative cells, it harbors a higher proportion of stemlike cells. Most cells of the LEC that are desmoglein 3 negative are localized to the deepest recess and appear tightly packed together (Figure 1B). This may function to protect the SCs from external factors similar to the SCN as demonstrated in the SC crypts of the gastrointestinal system and the primary keratinocytes seen in the skin.7,13
Tenascin C expression has been detected in normal and constant regenerating tissue such as at the epithelial-mesenchymal interface in the gastrointestinal tract18 and transiently in the placenta,19 a tissue that has a high turnover of cellular activity. Tenascin C staining was strongest around the LEC, followed by that in adjacent limbus associated with the LEC, and was weakest in normal limbus (Table). Consistent with previous studies, tenascin C staining in our cornea samples was negative. These results suggest that there are some variations in the limbus depending on whether it is associated with adjacent LEC. Zones of the limbus that are associated with the LEC (marked in yellow in Figure 3) demonstrate a higher level of tenascin C staining and therefore may indicate a higher ability of stemness compared with normal limbus not associated with the LEC. Therefore, depending on where sections are cut in the limbus according to specific zones, different characteristics may be acquired as seen by the varying results shown by different groups. Reviewing published literature, no authors to our knowledge describe variations of limbal immunostaining. When these regional variations are taken into consideration, all of the published data on limbal epithelial markers are compatible with and support our proposed model.
It is widely accepted dogma that the central cornea expresses connexin 43, but with minimal or no expression at the limbus.20 However, we have recently demonstrated that basal cells at the limbus adjacent to the LEC and cells within the LEC are connexin 43 positive,17 whereas limbus distant to the LEC does not express connexin 43. Connexin 43 expression in the peripheral cornea stops abruptly at the limbus distant from the LEC (Figure 5). The connexin 43 gap junction network exists within the bone marrow of certain stromal cells, which is upregulated before hematopoietic SCs divide.21,22 Therefore, connexin 43 expression within the LEC would further substantiate its role as an SCN. Both connexin 43 and tenascin C expression support our proposed model. Limbal basal cells directly adjacent to the LEC may possess similar properties to the LEC and thus are likely to have a greater proliferative potential as compared with limbus tissue that is not associated with any LEC.
Currently, it is common to select limbal SC donor grafts from the superior and inferior limbus.23 Our novel finding may have implications in deciding the selection of donor sites for limbal SC grafting. Donor limbal explants that include a section of limbus that is associated with the LEC but not necessarily including the LEC (zone 2 limbus) in addition to limbus that includes the LEC (zone 1) may be better suited for corneal epithelial regeneration in patients with limbal SC deficiency. Current in vivo confocal microscopy techniques allow for identification of limbal palisades but are not refined enough to identify the crypts. Once this is achieved, a more precise localization of limbus segments with greater proliferative potential should be possible prior to transplantation. We propose that during normal corneal epithelial maintenance, most terminally differentiated cells originate from zone 2 limbus while cells in the LEC remain quiescent. In circumstances where there is extensive epithelial loss, SCs in the LEC may be recruited to meet the extra demands for corneal epithelial cells.
Cell membrane protein and extracellular matrix protein play important roles in cell structure and adhesion, and our early results and those of others24 suggest that they may modulate SCs in their niche. Differences between the LEC and the limbus continue to emerge as further studies are carried out to investigate this unique anatomical structure.
Correspondence: Harminder S. Dua, PhD, FRCS, FRCOphth, Division of Ophthalmology and Visual Sciences, Eye, Ear, Nose, and Throat Bldg, Queen's Medical Centre, Nottingham NG7 2UH, England (firstname.lastname@example.org).
Submitted for Publication: November 20, 2007; final revision received January 4, 2008; accepted January 8, 2008.
Financial Disclosure: None reported.
Author Contributions: Dr Yeung had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Additional Contributions: Friedrich E. Kruse, MD, Department of Ophthalmology, University of Erlangen-Nürnberg, Erlangen, Germany, facilitated the collaboration and Elke Meyer, MD, Department of Ophthalmology, University of Erlangen-Nürnberg, provided technicalassistance.