Limbal epithelial stem cells of the cornea

7.1. The LESC niche

The stem cell niche, or microenvironment consisting of cellular and extracellular components, is hypothesised to prevent stem cell differentiation and thus regulates their fate (Schofield, 1983; Watt and Hogan, 2000). When a stem cell divides asymmetrically, one daughter may leave the niche to enter a differentiation pathway under the influence of different environmental stimuli. The limbus differs from cornea both anatomically and functionally and hence could differentially determine stem cell fate.

Within the limbal region of the cornea, the LESC niche is thought to be located within the palisades of Vogt (PV) – an undulating region of increased surface area. The palisades are highly pigmented with melanocytes (Davanger and Evenson, 1971; Higa et al., 2005) and are infiltrated with Langerhan's cells (Baum, 1970) and T-lymphocytes (Vantrappen et al., 1985). The melanin pigmentation is thought to shield LESCs from damaging ultraviolet light and the resultant generation of reactive oxygen species (Shimmura and Tsubota, 1997). The deep undulations of the Palisades of Vogt at the limbus provide LESC with an environment that protect them from shearing forces (Gipson, 1989). Furthermore the crypts described by Shortt et al., predominantly occur on the superior and inferior cornea where they are normally covered by the eye lids. (Shortt et al., 2007a) This may reflect the evolution of a protective environment for LESCs in humans. The basement membrane lining the LESC niche contains papillae of stroma that project upwards (Shortt et al., 2007a). The limbal and corneal basement membrane components also differ, with the limbal region containing laminin-1,5 and α2β2 chains not found in the cornea. Furthermore, type IV collagen α1, α2 and α5 chains are found in the limbal region whereas α3 and α5 are located in the cornea (Ljubimov et al., 1995; Tuori et al., 1996). A more recent study by Schlötzer-Schrehardt et al., found patchy immunolocalisation of laminin γ3 chain, BM40/SPARC and tenancin C, that was also found to co-localise with ABCG2/p63/K19-positive cell clusters. These factors may be involved in retaining cell stemness (Schlotzer-Schrehardt et al., 2007).

The basement membrane beneath the LESC may also act to sequester and therefore modulate growth factors and cytokines involved in LESC regulation and function (Klenkler and Sheardown, 2004). Although the surface of the cornea is exposed to atmospheric oxygen, the LESC niche lies beneath a number of cell layers where the oxygen tension is likely to be lower. Interestingly, hypoxic in vitro conditions have been found to produce larger, less differentiated limbal epithelial cell colonies suggesting that low oxygen levels may induce selective proliferation of undifferentiated cells (Miyashita et al., 2007).

The limbal niche is vascularised and highly innervated (Lawrenson and Ruskell, 1991) unlike the avascular cornea and therefore is a potential source of nutrients and growth factors for LESC. Limbal fibroblasts in the underlying stroma are heterogeneous and express secreted protein acidic and rich in cysteine (SPARC) that may contribute to LESC adhesion (Shimmura et al., 2006). Furthermore, Nakamura et al., identified a population of bone marrow-derived cells located in the limbal stroma following transplantation of GFP labelled bone marrow cells into nude mice (Nakamura et al., 2005). It is possible therefore that these cells are able to migrate into the limbal stroma, although any potential functionality remains unclear.

Sonic hedgehog, Wnt/β-catenin, TGF-β and Notch signalling pathways have all being implicated in niche control of stem cells, however little is known of their potential roles in the LESC niche. Mice lacking in expression of Dkk2, a Wnt pathway inhibitor, display epidermal differentiation on the ocular surface. The lack of Dkk2, leads to increased Wnt/β-catenin signalling in the limbal stroma. This demonstrates the importance of limbal niche control over LESC differentiation during development. PAX6 expression is also lost in the corneal epithelial cells of these mice, suggesting it is downstream of Dkk2 (Mukhopadhyay et al., 2006). Deficiencies in PAX6 leads to aniridia resulting in impaired corneal epithelial function and eventual LESC failure, which may be due to altered niche development.

7.2. Putative positive and negative LESC markers

The literature reflects many attempts to prospectively identify LESC using a specific marker. As yet no single, reliable marker has been found. However, the expression of a combination of several features seems to allow for greater specificity.

Putative ‘markers’ can either be positive (present) or negative (absent). Limbal basal cells lack differentiation markers such as the 64 kDa cytokeratin 3 (CK3) that is present in all other layers of the corneal epithelium and the suprabasal layers of the limbal epithelium (Schermer et al., 1986). The corneal specific 55 kD protein, cytokeratin 12 (CK12) is also expressed in a similar pattern (Chaloin-Dufau et al., 1990). Furthermore, connexin 43 (Shortt et al., 2007a; Matic et al., 1997) and involucrin (Chen et al., 2004), both markers of cells destined for differentiation, are also absent.

The transcription factor p63 is required for formation of epidermis and has been proposed as a putative positive LESC marker (Pellegrini et al., 2001). In vitro,p63 was found to be expressed in limbal epithelial cell derived holoclones with little or no expression in meroclones and paraclones. In vivo, p63 was located in the limbal basal epithelium. However, since these initial observations a number of reports have suggested that p63 is not sufficiently specific to act as an LESC marker as it has also been localised to basal cells of the peripheral and central cornea in humans (Chen et al., 2004; Dua et al., 2003) and in rats (Chee et al., 2006). However, limbal epithelial cells expressing high levels of p63 with a high nuclear to cytoplasmic ratio appear to be more stem like (Arpitha et al., 2005). Further work has since indicated that the ΔNp63α isoform may more specifically label LESC (Di Iorio et al., 2005).

Many types of organ-specific stem cells, including LESC have been recently shown to exhibit a side population (SP) phenotype. The SP cells are able to efflux Hoechst 33342 dye through the ATP-binding cassette transporter Bcrp1/ABCG2. ABCG2 has therefore been proposed to be a universal marker for stem cells (Zhou et al., 2001; Watanabe et al., 2004). In putative LESCs, this protein has been immunolocalised to the cell membrane and cytoplasm of a population of limbal basal cells and a few suprabasal cells (Chen et al., 2004). Furthermore, ABCG2 positive cells produce higher colony forming efficiency values in vitro than their negative counterparts (de Paiva et al., 2005). Our laboratory has localised ABCG2 to the outer edge of holoclones where it is thought that the stem cells reside.

Clusters of cells expressing the integrin α9 have been localised to the limbal basal epithelium (Stepp et al., 1995). However, upregulation of α9 in wounded murine corneas have since indicated this integrin to be associated with TAC's (Stepp and Zhu, 1997). Integrin β1 was originally suggested to be a keratinocyte marker (Jones and Watt, 1993). Cells that rapidly adhere to the integrin β1 ligand, collagen IV also display LESC properties (Li and Lu, 2005). Limbal basal epithelial cells are described as β1 integrin bright as are the stem cells of the epidermis suggesting a gradient of expression that decreases with differentiation. The integrins α2, α6 and β4 are negative in the limbal basal epithelial cells (Schlotzer-Schrehardt and Kruse, 2005).

N-cadherin is an important mediator of cell-cell adhesion and may play a key role in the maintenance of haemopoietic stem cells by facilitating adhesion to osteoblasts in the bone marrow niche (Calvi et al., 2003; Zhang et al., 2003). Hayashi et al found expression of N-cadherin in a subpopulation of limbal epithelial basal cells and in adjacent melancytes implying N-cadherin plays an important role in interactions between LESC and their corresponding niche cells (Hayashi et al., 2007).

Even though the limbal epithelium is derived from the surface ectoderm a number of neural stem cell markers have been suggested as LESC markers. Recent in depth immunological studies of neurotrophic factors and their receptors in the human has found NGF, glial cell-derived neurotrophic factor (GDNF) and their corresponding receptors TrkA and GDNF family receptor alpha (GFRα)-1 to be exclusively expressed in the limbus (Qi et al., 2008).

Notch 1 is a ligand-activated transmembrane receptor that has been shown to maintain progenitor cells in a number of tissues. The role of Notch signalling in the cornea is unclear. However, cell clusters in the palisades of Vogt have been found with some co-localisation with ABCG2 (Thomas et al., 2007). Using Notch 1 deficient mice, Vauclair et al, demonstrated Notch 1 signalling is required for cell fate maintenance during corneal epithelial wound healing linking this to regulation of vitamin A metabolism (Vauclair et al., 2007). Notch 1, other Notch family members and their down-stream targets have been identified throughout the cornea suggesting a role in differentiation (Ma et al., 2007). More recently, Nakamura et al., has found Hes1, a major target in Notch1 signalling, to be localised to the basal limbal epithelium in adult mice (Nakamura et al., 2008). It is likely that Notch signalling, perhaps under synergistic regulation with the Wnt signalling pathway, controls the balance between LESC self-renewal and daughter cell commitment to differentiation. The cell cycle arrest transcription factor C/EBPΔ has also been implicated in the regulation of LESC self-renewal. Limbal epithelial basal cells that express C/EBPΔ co-express Bmi1 (which is involved in stem cell self renewal) and ΔNp63α (Barbaro et al., 2007).

Cell-cell communication is facilitated by gap junctions. Connexins 43 and connexin 50 are present in the corneal epithelium (Dong et al., 1994). Cx 43 is expressed by corneal basal cells except that of the limbus, implying it is utilised by more early TACs. The lack of intracellular communication has been suggested to help maintain stem cells and their niche (Matic et al., 1997) by protecting the cells from damage affecting adjacent neighbours (Chee et al., 2006). Like the stratified squamous epithelia, (Watt and Green, 1981) involucrin is also expressed in the corneal epithelium (Chen et al., 2004) and in larger cells in vitro suggesting it is a marker of differentiation.

The RNA binding protein, Musashi-1 is produced in the developing and adult eye (Raji et al., 2007) and has recently been found in putative LESCs co-cultured with amniotic epithelial cells as feeders (Chen et al., 2007).

7.3. Clinical consequences of LESC failure and cultured stem cell therapy

LESC deficiency can occur as a result of primary or acquired insults. Partial or full LESC deficiency leads to deleterious effects on corneal wound healing and surface integrity (Chen and Tseng, 1991; Dua et al., 2003). Deficiency can arise following injuries including chemical or thermal burns and through diseases such as aniridia and Stevens Johnson syndrome (see Figure 5). As a result of LESC deficiency conjunctivalisation, neovascularisation, chronic inflammation, recurrent erosions, ulceration and stromal scarring can occur causing painful vision loss (Holland and Schwartz, 1996; Kenyon and Tseng, 1989; Puangsricharern and Tseng, 1995). Long term restoration of visual function requires renewal of the corneal epithelium, through replacement of the stem cell population has traditionally been achieved by grafting limbal auto- or allografts (Kenyon and Tseng, 1989 {Ramaesh, 2003 #1138). Each procedure carries a risk of complication such as damage to healthy eye by removal of autologous tissue for transplantation or side effects from long-term immunosuppression with allogenic tissue. As an alternative, cultured LESC therapy has been developed where LESCs are expanded in vitro for therapeutic application in patients in a variety of protocols utilising amniotic membrane or fibrin, in the presence or absence of growth arrested 3T3 fibroblast feeder layers (Lindberg et al., 1993; Pellegrini et al., 1997; Grueterich et al., 2002; Koizumi et al., 2001; Shortt et al., 2007b; Tsai et al., 2000). Cultured autologous mucosal epithelial cell grafts have also been used to reconstruct the ocular surface of LESC deficient patients with some success (Nakamura et al., 2003). Recently it has been demonstrated that other stem cell populations including human embryonic stem cells (Ahmad et al., 2007) and hair follicle stem cells (Blazejekska et al., 2008) can be driven towards a corneal epithelial-like phenotype. These exciting data may lead to alternative therapeutic strategies in the future for patients blinded by ocular surface disease cause by failure of LESC function.

Figure 5.

Chemical burn injury.

An alkali burn to the human cornea can cause ocular surface failure with neovascularisation, opacification and blindness resulting from LESC deficiency.

The biological mechanisms of efficacy experienced by recipients of the cultured LESCs are unclear yet the clinical results are promising. It has been suggested that bone marrow derived stem cells may be recruited to the cornea to repair the damage caused by LESC failure (Daya et al., 2005) since no long-term survival of allogeneic cultured LESCs has been demonstrated. Our hypothesis is that the transplanted cultured limbal epithelium may act, at least in some patients, by ‘kick-starting’ the recipient's own ailing LESC.

One of the causes of blindness in children with aniridia is due to progressive ocular surface failure. The majority of cases are caused by PAX6 haploinsufficiency being a result of heterozygous null mutations (Van Heyningen and Williamson, 2002). The disease is a pan-ocular, bilateral condition most prominently characterised by iris hypoplasia and varies from a relatively normal iris to the complete lack of an iris. Aniridia is often associated with cataracts, corneal vascularisation and glaucoma, with a significant number of cases of visual morbidity being due to corneal abnormalities. The underlying process of these abnormalities is poorly understood and is thought to be due to stem cell failure (Mackman et al., 1979; Nishida et al., 1995; Tseng and Li, 1996). However, it has also been proposed that it may be due to a deficiency in the stem cell niche and adjacent corneal stroma (Ramaesh et al., 2005). More recently downregulation of Pax6 has been linked to abnormal epidermal differentiation of cornea epithelial cells (Li et al., 2008). Treatment usually involves replacement of LESC using limbal allografts and/or corneal grafts or more recently ex vivo cultured LESC grafts (Holland et al., 2003).

Aniridia represents a spectrum of disease, with iris anatomy defects ranging from the total absence of the iris to mild stomal hypoplasia with a pupil of normal appearance. Other associated defects include foveal hypoplasia, optic nerve hypoplasia, nystagmus, glaucoma and cataracts. These conditions may develop with age causing progressive visual loss. Another important factor leading to progressive loss of vision is aniridic-related keratopathy (ARK; Mackman et al., 1979; Margo, 1983) which occurs in 90% of patients. Initially the cornea of patients appears normal during childhood (Nelson et al., 1984; Nishida et al., 1995). Changes occur in patients in their early teenage years, with the disease manifesting as a thickened irregular peripheral epithelium. This is followed by superficial neovascularisation and if left untreated it may result in subepithelial fibrosis and stromal scarring. Furthermore patients develop recurrent erosions, ulcerations, chronic pain and eventual blindness (Holland et al., 2003). Histologically, stromal neovascularisation and infiltration of inflammatory cells is seen with the destruction of Bowman's layer. Additionally, the presence of goblet and conjunctival cells is seen on the corneal surface (Margo, 1983). Traditionally, these clinical and histological manifestations have lead to the consensus that LESC deficiency is largely responsible for corneal abnormalities in aniridia (Dua and Azuara-Blanco, 2000; Margo, 1983).

Based on the clinical and histological manifestation of aniridia, LESC deficiency has been presumed to be the pathogenesis behind ARK (Dua et al., 2000; Margo, 1983; Nishida et al., 1995). As a LESC marker has yet to be definitively identified, a true demonstration of LESC deficiency can not be assumed. Furthermore treatment for these patients involving replacement of LESC, either by keratolimbal allografts or more recently ex vivo expanded LESC grafts, provides a better outcome than corneal transplants (Holland et al., 2003; Shortt et al., 2007b; Tiller et al., 2003). This is consistent with LESC deficiency. However, patients who receive both limbal and corneal tissue seem to have the better outcome, suggesting an abnormality with corneal tissue and not just the limbus. This may be a downstream effect of LESC deficiency. Alternatively low levels of PAX6 may have a generalised effect on the entire cornea. ARK could also be the consequence of abnormal corneal epithelial/stromal healing responses as there is insufficient evidence to indicate that the proliferative potential of LESC is impaired (Ramaesh et al., 2005; Sivak et al., 2000). Recently, studies looking at the regulation of genes downstream of Pax6 in the Pax6 heterozygous mouse, suggests the pathogenesis of ARK is due a number of mechanisms and not solely due to LESC deficiency (Ramaesh et al., 2005). Further studies are needed to elucidate the exact mechanism of ARK progression to allow the use of appropriate treatments.

Mutations in Pax6 result in a distinct small eye syndrome in the small eye (SEY) mouse and rat (Hill et al., 1991; Matsuo et al., 1993). These animals are excellent models for aniridia and the progressive nature of associated corneal abnormalities (Ramaesh et al., 2003; Davis et al., 2003). As the name suggests, mice with semidominant mutations develop small eyes and other ocular deformities. The murine strains Pax6^Sey, Pax6^SeyNeu and Pax6^Coop represent three SEY mice with differing point mutations in the Pax6 gene. Pax6^SeyDey and Pax6^SeyH mice have Pax6 gene deletions (Hill et al., 1991; Hogan et al., 1986; Lyon et al., 2000; Schmahl et al., 1993; Theiler et al., 1980). The SEY mice with semidominant heterozygous phenotypes demonstrate comparable developmental ocular abnormalities. This includes microophthalmia and defects in the iris, lens and retina with phenotypic severity being variable (Callearts et al., 1997; Hill et al., 1991). Cataracts, glaucoma and more importantly corneal abnormalities can develop in mutant SEY during post-natal development and adult life (Lyon et al., 2000). Interestingly, the phenotypic variability seen between mice is also observed within a single SEY strain (Hogan et al., 1986). This can even be detected between two eyes of the same mouse, suggesting a stringent requirement for Pax6 activity to be at specific levels at precise times during development (Hill et al., 1991; Schedl et al., 1996; van Ramsdonk and Tilghman, 2000). Homozygotes generate an ultimately lethal phenotype with no eyes and nasal primordial (Hill et al., 1991). A number of sey mice arose independently all of which are semidominant and by examining comparative mapping studies and phenotypic similarities to aniridia, it was suggested to be the mouse homologue of the human disease (Glaser T et al., 1990). This research led to the discovery that the Pax6 gene was responsible for the Sey phenotype and suggested that it was also responsible for the human disease, aniridia (Hill et al., 1991). These models are helping us to address fundamental questions about LESCs and their niche environment.