Inhibition of conjunctival transdifferentiation by topical retinoids

During the healing of a total corneal epithelial defect extending beyond the limbus, conjunctival transdifferentiation can be inhibited by corneal vascularization as evidenced by the lack of morphological transformation of the conjunctival epithelium into a cornea-like epithelium and the persistence of goblet cells on the corneal surface. We speculated that corneal vascularization might play a causative role in inhibiting conjunctival transdifferentiation, and examined the hypothesis that vitamin A or retinoids might be one of the blood-borne factors in modulating this process. To test this hypothesis, we created total corneal epithelial defects extending 3 mm beyond the limbus in rabbits using n-heptanol, and segregated the resultant corneas into nonvascularized and vascularized groups. After re-epithelialization, both groups received topical 0.1% Etretinate (Roche-Hoffmann, Nutley, NJ) or 13-cis retinoic acid in corn oil three times a day for 8 weeks. Controls received corn oil only. The extent of transdifferentiation was analyzed by assaying goblet cell density and distribution using flat-mount preparations and Alcian blue and periodic acid-Schiff stains (Fischer Scientific Co., Fair Lawn, NJ) and by conventional histology. Topical retinoid application inhibited conjunctival transdifferentiation in nonvascularized corneas to the same extent as that caused by corneal vascularization, suggesting that vitamin A is an important blood-borne factor for goblet cell maintenance. Its relative deficiency in the normal avascular cornea may explain why conjunctival transdifferentiation occurs.     

Goblet cell density and vascularization during conjunctival transdifferentiation

After debridement of the entire corneal epithelium with n-heptanol, two groups of rabbit corneas were segregated according to the extent of corneal neovascularization. Using a new topographic goblet-cell counting method and routine histology, the authors have reexamined the process of conjunctival transdifferentiation and compared the changes of goblet-cell density and morphology between nonvascularized and vascularized groups for a follow-up period of 167 days. Analysis of the total goblet-cell density disclosed that no goblet cells appeared on the corneal surface during the entire period of reepithelialization. After that, two phases were identified with respect to goblet-cell density: phase I (day 0-17) and phase II (after day 17). In phase I, both groups had a similar surge of goblet cells, with the peak occurring between days 7 and 11, suggesting little correlation with vascularization. Morphologic studies indicated the presence of a prominent centripetal cellular migration. In phase II, the nonvascularized group showed a rapid decline in goblet-cell density, and as a result the morphologic transdifferentiation into a cornea-like epithelium was completed on day 43. The changes of goblet cells to a smaller size and the presence of a more acidic mucin in the centrifugal receding zone, suggested that transdifferentiation on nonvascularized corneas is a process involving changes of cellular differentiation. In contrast, the vascularized group maintained a high plateau of goblet-cell density and an epithelium with conjunctival characteristics until day 167. This result disclosed that retardation of conjunctival transdifferentiation by corneal vascularization was in phase II. The possible role of vascularization in the modulation of conjunctival transdifferentiation is discussed.     

Conjunctival transdifferentiation induced by systemic vitamin A deficiency in vascularized rabbit corneas

Conjunctival transdifferentiation, the process in which conjunctival epithelium transforms into a cornea-like epithelium with the loss of goblet cells during the healing of a total corneal epithelial defect, can be retarded or reversed by corneal neovascularization. We have previously shown that this process normally occurring on non-vascularized corneas can be retarded or reversed by topical retinoids, suggesting that vitamin A may be one of the factors from blood circulation which is responsible for modulating transdifferentiation. Herein, we have examined the effect of systemic vitamin A deficiency on vascularized corneas starting 4 months after epithelial denudation, and compared this deficient group with their vascularized and non-vascularized controls. Mean serum retinol level (microgram/dl) (n = 4) measured by HPLC was gradually reduced from 83 of the controls to 20 in a 10 month follow-up. Topographical analysis disclosed a centrifugal loss of goblet cell density with time. Histology showed complete transdifferentiation in vascularized areas at 9 months, initiated by the loss of mucin contents from receding zones first noted at 2 months. Using impression cytology, all corneas were not keratinized and all conjunctivas maintained a normal goblet cell density at 10 months. These results indicate that conjunctival epithelium on corneal surface is more sensitive to the decrease of serum vitamin A levels than that on conjunctiva, and support the hypothesis that the relative vitamin A deficiency on vascularized corneas can also result in the conjunctival transdifferentiation.     

Conjunctival transdifferentiation is due to the incomplete removal of limbal basal epithelium

Previous studies have shown that using n-heptanol to create a total corneal epithelial defect beyond the limbus results in two different healing patterns with an unpredictable incidence. Between 14-68% of these wounded rabbit corneas (n = 287, combining various reports) showed extensive vascularization and conjunctivalization, whereas the remaining were not vascularized and had conjunctival transdifferentiation with a cornea-like epithelium. To investigate the role of the limbal epithelium in these two healing patterns, the authors treated rabbit eyes for various durations with n-heptanol and additional scraping. Histology showed that treatment for up to 120 seconds removed both the corneal and conjunctival epithelia but left the limbal basal cells intact. To prove viability, they cultured the treated limbal explants on collagen gel. After 14 days of culture, increased stratification of the limbal epithelium and an epithelial outgrowth onto the corneal stroma was observed. The latter was proven to be of corneal origin (positive to AE-5 but negative to AM-3 monoclonal antibody staining). The authors then surgically removed the entire limbal zone including 2 mm of peripheral cornea and 3 mm of adjacent conjunctiva in addition to n-heptanol debridement of the entire corneal epithelium in 54 rabbit eyes and observed a high incidence (96%) of corneal vascularization and conjunctivalization of the resultant epithelial phenotype (positive to AM-3, but negative to AE-5 monoclonal antibody staining). These results support the hypothesis that corneal epithelial stem cells are located in the limbus and indicate that an incomplete removal of the basal limbal epithelium by n-heptanol leads to unvascularized corneas with conjunctival transdifferentiation. Conversely, complete removal of such cells results in corneal vascularization and conjunctivalization.     

Alteration of epithelial paracellular permeability during corneal epithelial wound healing.

We studied the paracellular permeability to mannitol of corneas with epithelium of corneal, limbal, or conjunctival origin. Corneas with epithelial defects reepithelialized by corneal or limbal epithelium were nonvascularized; the corneal permeability was initially increased and returned to normal 3 days later. When epithelial defects extended beyond the limbus, they were healed by conjunctival epithelium. If corneas remained avascular or minimally vascularized, the conjunctiva-derived epithelium underwent a transdifferentiation process into a cornealike morphology in which the corneal permeability was initially increased upon complete reepithelialization, and gradually decreased to a level similar to that of normal cornea, 4 weeks after healing. However, when corneas became vascularized, the conjunctiva-derived epithelium retained its original phenotype, and corneal permeability remained increased throughout the 8-month period of study. The deranged barrier functions noted in the above vascularized cornea were demonstrated further by horseradish peroxidase tracer, which was found in the intercellular spaces of conjunctiva-derived epithelium of vascularized corneas but not in the avascular corneas with epithelia of corneal or limbal origin, or transdifferentiated conjunctival epithelium. To study further the effect of subsequent ocular surface trauma, conjunctival biopsy was performed on transdifferentiated avascular corneas 3 months after initial wounding. The biopsy resulted in extensive vascularization in three of eight previously nonvascularized corneas. Two weeks later, the corneal permeability was increased to a level similar to that of conjunctiva. These results indicate that corneal epithelial paracellular permeability correlates well with the status of the epithelial phenotype.     

Reversal of conjunctival transdifferentiation by topical retinoic acid

After resurfacing a total corneal epithelial defect extending 2-3 mm beyond the limbus, conjunctival epithelium gradually loses goblet cells and transforms into a corneal-like epithelium. We examined the effect of topical retinoic acid on the reversal of transdifferentiation on nonvascularized corneas. Four months after total denudation of corneal epithelium using n-heptanol, rabbit corneas without vascularization received topical drops of 0.1% (wt/vol) all-trans retinoic acid in corn oil 3 times a day. Before treatment, the transdifferentiation was complete, as evidenced by the absence of goblet cells on the corneal surface using a topographical assay and routine histology. After treatment for 15 days, goblet cells reappeared 3 mm into the peripheral cornea, and extended in a centripetal density to 4.5 mm after 32 days. To prove that retinoic acid was not angiogenic, retinoid-bearing Elvax-40 pellets were implanted into normal corneal stroma. Taken together, these data indicate that vitamin A or retinoids may be an important factor in the modulation of conjunctival epithelial transdifferentiation.     

Corneal epithelial wound healing in the absence of limbal epithelium.

Corneal epithelial stem cells are thought to be at the limbus. The limbal epithelium was surgically removed in 12 New Zealand white rabbits. After 6 months, four showed mild vascularization. To challenge the remaining proliferative reserve, two consecutive 7.5-mm epithelial woundings were created 3 weeks apart in 11 limbal-deficient corneas and 11 controls. After the first wounding, five of the limbal-deficient corneas showed delayed healing, and seven became moderately vascularized; the controls healed normally. After the second wounding, eight experimental corneas showed intense vascularization; the controls did not. Recurrent erosions with delays in healing were noted in nine experimental animals but not in controls. Flat-mount preparation and impression cytology revealed centripetal migration of conjunctival epithelium with goblet cells onto the experimental corneas. These results indicate that only limited proliferative capacity of corneal epithelium remains in the absence of limbus. The constellation of delayed healing with recurrent erosion, corneal vascularization, and conjunctival epithelial ingrowth can be considered possible signs of limbal stem cell dysfunction.     

Comparison of limbal and conjunctival autograft transplantation in corneal surface reconstruction in rabbits.

Destruction of corneal surface was created in one eye of 24 rabbits by n-heptanol corneal epithelial debridement and surgical removal of limbal zone. One month later, the animals were equally subdivided into three groups of eight for limbal transplantation, conjunctival transplantation, and control without transplantation. During a 6-month postoperative follow-up, all corneas in the control group showed progressive vascularization and conjunctivalization. All corneas with limbal transplantation showed progressive decrease of vascularity, verified by fluorescein angiography. In contrast, all but one of the eight corneas of conjunctival transplantation showed progressive vascularization (P = 0.01). More important, the resultant epithelia showed corneal phenotype in limbal transplantation, but remained conjunctival in conjunctival transplantation, as verified by monoclonal antibodies AM-3, APSM-1, and AE-5. These results support the concept of the limbal location of corneal epithelial stem cells, and indicate that complete destruction of the limbal zone resulted in corneal vascularization and conjunctivalization, and that limbal transplantation has a better efficiency than conjunctival transplantation in restoring such destroyed corneal surface.     

Inhibition of vascularization in rabbit corneas by heparin: cortisone pellets

The purpose of this work was to study the effectiveness of heparin plus cortisone, and of cortisone alone in control of corneal vascularization in rabbit eyes. Corneal vascularization was induced by de-epithelialization of the cornea and limbus and part of the bulbar conjunctiva with concurrent trephination and excision of a central 2-mm diameter corneal button. Inhibition of vascularization by polymer pellets impregnated with heparin (Panheprin, Abbott Laboratories; Chicago, IL) and cortisone, or neither drug was studied by implanting the pellets into the eyes at the time of injury and following the eye clinically and histologically. Wounded corneas with empty pellets developed vascularization extending from the limbus to the central cornea within 3 wk (n = 10). In other wounded eyes, heparin:cortisone pellets prevented vascularization (n = 10) while cortisone pellets slowed, but did not totally inhibit vascularization (n = 6). In other eyes, clear autografts were transplanted into vascularized eyes; and the ability of the drug-impregnated pellets to inhibit grafts vascularization was evaluated. In eyes with heparin:cortisone pellets inserted into the donor button at the time of keratoplasty, the autografts remained clear for at least 6 wk (n = 10) but subsequently vascularized if the sutures were not removed, while cortisone pellets slowed but did not block vascularization (n = 6). If heparin:cortisone pellets were inserted into the vascularized host tissue, rather than into the donor button, vascularization of the graft occurred (n = 6). Thus, heparin (Panheprin, Abbott Laboratories; Chicago IL) plus cortisone inhibited vascularization in rabbit cornea in the models studied: The effect of other commercially available heparins remains to be studied.     

Endothelin-1 and ETA/ETB receptor protein and mRNA: expression in normal and vascularized human corneas

PURPOSE: Neovascularization of the cornea causes blindness and increases the risk of immune rejections after keratoplasty. The purpose of this study was to investigate involvement of the potent angiogenic growth factor endothelin (ET)-1 and its receptors, ETA and ETB, in corneal neovascularization. METHODS: ET-1, ETA, and ETB receptor protein expression was evaluated in nonvascularized and vascularized human corneas by immunohistochemistry. Epithelial ET-1 protein expression of both groups was compared using a semiquantitative scoring system. Double immunofluorescence was used to colocalize ETA and ETB receptor with CD31. In situ hybridization and immunoelectron microscopy analyzed ET-1 and its receptors in normal and vascularized corneas. RESULTS: Nonvascularized corneas displayed ET-1 and ETA/ETB receptor protein and mRNA in epithelial and some corneal endothelial cells. ETA more than ETB receptors were expressed on some keratocytes. In vascularized corneas, ET-1 and ETA/ETB receptor expression was found in the endothelial lining of new blood vessels (as shown by CD31-colocalization). ET-1 protein expression was significantly increased in the epithelium of vascularized corneas (P < 0.001). Immunogold localized ET-1 and its receptors to the nuclear/perinuclear space and to the luminal side of endothelial cells of new blood vessels. CONCLUSIONS: In corneal neovascularization, ET-1 protein and mRNA expression is upregulated in epithelial cells. Together with ET-1, ETA, and ETB receptor expression on endothelial cells of ingrown new blood vessels, this points to an involvement of ET-1 and its receptors in corneal angiogenesis. As potent ETA and ETB receptors are available, the endothelin system may represent an additional target for corneal antiangiogenic therapy.     

Corneal epithelial wound healing in partial limbal deficiency.

Previous studies have shown that the corneal epithelial stem cells are located at the limbal basal layer. The limbal stem cells are regarded as the ultimate source for corneal epithelial cell proliferation and differentiation. This paper examines epithelial wound healing in rabbit corneas with partial limbal deficiency (PLD), which was created by the surgical removal of two-thirds of the limbal zone (superior and inferior). Four to eight months after PLD creation, all corneas appeared normal, without vascularization. The residual stem cell capacity then was challenged by two sizes of corneal epithelial debridement created with combined n-heptanol and mechanical scraping. In the first group, two consecutive 6-mm defects were created 1 month apart. After the first wounding, three of eight PLD corneas had delayed wound healing and two of the three had vascularization, as compared to controls (n = 7). After the second wounding, both controls (n = 7) and the remaining PLD (n = 5) corneas showed similar rapid healing. In the second group, a large defect of up to 1 mm within the limbus was created. Healing was completed in 25-40 days in PLD (n = 6) corneas, a more marked delay compared to the 10-12 days for controls (n = 6) (P = 0.001). In addition, all PLD corneas showed increased vascularization and had epithelium of the conjunctival phenotype, verified by the immunofluorescent staining positive to AM-3 monoclonal antibody but negative to AE-5 monoclonal antibody. Thus, a deficiency of limbal stem cells contributes to the triad of conjunctival epithelial ingrowth, corneal vascularization, and delayed healing with recurrent erosion. In PLD, corneal epithelium is still compromised, particularly when a large epithelial cell mass is removed.     

Outgrowth of cells from human conjunctival explants onto cornea in vitro.

The epithelium was removed from human corneas, and samples of conjunctival tissue were cultured as explants on the denuded corneal surface for 1 and 2 weeks. Cells migrating from the conjunctival explants onto the corneal surface produced a multilayer where cells on the surface generally showed a flattened appearance. The apical membrane of these cells demonstrated villi as well as microplicae. Surface projections were also detected on cells in the deeper layers of the epithelium. Neighbouring cells were connected by junctional complexes. After 2 weeks, however, a lack of intercellular junctions in some areas resulted in the formation of intraepithelial cystoid spaces. Basal cells were connected to the underlying basement membrane by hemidesmosomes. Although transdifferentiation of the cells into a corneal epithelium was not observed within the 2 weeks, the present system provides a tool for studies on factors affecting reepithelialization of corneal epithelial defects by conjunctival cells.     

Vascularization of corneas of hairless mutant mice.

During the course of experiments examining the immunobiology of corneal transplants, the corneas of athymic, nude mice (nu/nu) were found to contain blood vessels that extended through the entire superficial stroma into the centermost portion of the cornea. The presence of corneal vessels was not related to the immunodeficient condition of the nude mouse since corneas from the severe combined immunodeficiency (SCID) mutant mouse strain were avascular and indistinguishable from corneas obtained from immunocompetent BALB/c mice. Furthermore, Langerhans cells were not found to accompany the blood vessels in the corneas of any of the nude mice examined. Corneal vascularization that was similar to that seen in the nude mouse was found in the cuthymic, hairless mutant mouse strain (SKH1; hr/hr). Although vascularization of the corneal stroma was associated with the heritable loss of hair, the genes responsible for hair loss in these two mutant mouse strains reside on different chromosomes. Understanding the processes involved in either promoting or preventing corneal vascularization may have significant impact in preventing corneal allograft rejection and in controlling inflammatory diseases of the corneal surface. The two mutant mouse strains described here may serve as valuable tools for such investigations.     

Corneal argon laser photocoagulation for neovascularization in penetrating keratoplasty

Corneal argon laser photocoagulation (CALP) was used in 13 patients to treat deep stromal vascular ingrowth. Eight patients had undergone successful penetrating keratoplasty but had developed deep stromal vessels into the graft associated with signs of graft rejection, which did not improve with steroid treatment alone (group 1). After CALP, there was marked regression of the neovascularization with reversal of graft rejection in all eyes. Three additional patients with vascularized corneas, referred for penetrating keratoplasty, underwent CALP preoperatively with obliteration of the vessels (group 2). Two of these patients have since undergone keratoplasty and, in both, the grafts have remained avascular and clear over a 21-month follow-up. Two other patients with corneal injury and progressive corneal opacification and vascularization have also been treated with CALP (group 3). CALP may be a useful adjunct in the treatment of corneal neovascularization. Further clinical studies are needed to define its exact role.     

Fine needle diathermy occlusion of corneal vessels

PURPOSE: To develop a novel technique, fine needle diathermy (FND), for the occlusion of corneal vessels and to evaluate its safety and efficacy in a series of patients. METHODS: Fourteen patients were treated with FND to occlude corneal vessels. Patients were categorized into four groups: group 1 (n = 4), high risk patients with stromal vascularization before keratoplasty; group 2 (n = 2), patients with progressive lipid keratopathy; group 3 (n = 4), post keratoplasty patients with active rejection episodes associated with vessels; and group 4 (n = 4), patients with disciform vascularized scars with recurrent inflammation. The success of the treatment in terms of vessel occlusion and the clinical outcome were monitored. RESULTS: All patients in group 1 had successful corneal transplantation, and the grafts remained clear without graft rejection. Patients in group 2 with lipid keratopathy had 100% obliteration of vessels with stabilization of corneal scar. All four patients in group 3 had complete regression of vessels with reversal of graft rejection. Patients with vascularized disciform scar had resolution of the inflammation without recurrence. Average follow-up was 10.3 months (minimum, 6 months; maximum, 24 months). No serious complications were observed with FND. CONCLUSIONS: FND is a useful and inexpensive technique that can serve as an adjunct or alternative to laser occlusion for the treatment of established corneal vessels. It is fairly safe and effective, although complications such as intrastromal bleeding and crystalline deposits can occur and at times it may have to be repeated once or twice to achieve the desired result.     

Expression of vascular endothelial growth factor and its receptors in inflamed and vascularized human corneas

PURPOSE: To help further define the possible role of vascular endothelial growth factor (VEGF) in the pathogenesis of corneal neovascularization, the expression of VEGF and of its receptors Flt-1 and Flk-1 was investigated in various inflammatory corneal diseases. METHODS: Polyclonal antibodies to VEGF and its receptors were used for immunohistochemical staining of frozen sections of 38 human corneas with various degrees of neovascularization and inflammation. In addition, a panel of monoclonal antibodies was used to characterize the composition of the inflammatory infiltrates and to confirm the presence of neovascularization. Furthermore, VEGF concentrations were determined in vascularized corneas using a sensitive enzyme-linked immunosorbent assay. RESULTS: VEGF was expressed by epithelial cells, by corneal endothelial cells, by vascular endothelial cells of limbal vessels and of newly formed vessels in the stroma, and weakly by keratocytes. Furthermore, VEGF expression was often markedly increased in inflamed corneas on epithelial cells and on vascular endothelial cells, particularly in the vicinity of macrophage infiltrates, and on fibroblasts in scar tissue. Correspondingly, VEGF concentrations were significantly higher in vascularized corneas compared with normal control corneas (P < 0.001). Expression of both VEGF receptors, Flt-1 and Flk-1, was increased on endothelial cells of newly formed vessels in the stroma of inflamed corneas compared with limbal vessels of normal control corneas. In addition, Flt-1 was also expressed by corneal endothelial cells and by macrophages, whereas Flk-1 expression was lacking. CONCLUSIONS: These results demonstrate that VEGF, Flt-1, and Flk-1 are strongly expressed in inflamed and vascularized human corneas and, thus, may play an important role in corneal neovascularization.     

Expression of tissue inhibitor of metalloproteinase-4 in normal human corneal cells and experimental corneal neovascularization

PURPOSE: To study the expression of TIMP-4 in cultured corneal cells and in corneal neovascularization. METHODS: Human limbo-corneal epithelial cells, fibroblasts, and endothelial cells were cultured in serum-free, PMA- or basic fibroblast growth factor (bFGF)-treated condition. Neovascularization in rat cornea was induced by suturing. The expression of TIMP-4 was examined by immunohistochemistry, Western blot and RT-PCR. RESULTS: TIMP-4 was constitutively expressed in cultured human corneal cells. The expression was only mildly enhanced after mitogen treatment. TIMP-4 immunoreactivity was predominantly expressed in normal rat corneal epithelium, and also in ingrowing blood vessels following suturing, which persisted up to day 28. Increased staining in corneal epithelium and blood vessels were also noted in vascularized human corneas. CONCLUSIONS: TIMP-4 is expressed in the cornea, which may play a role in modulating extracellular matrix remodeling associated with corneal wound healing and angiogenesis.     

Histology of human conjunctival transplantation

The technique of conjunctival transplantation has proven very successful in reestablishing an intact ocular surface in patients with severe ocular surface disease in whom conventional treatment has failed. We present follow-up on the histology of the transplanted conjunctival tissue in four alkali-burned patients who underwent penetrating keratoplasty 3-28 months following conjunctival transplantation. The corneal button was re-epithelialized in all patients. The fate of the transplanted tissue agreed with experimental observations made in the rabbit model, i.e., under areas of vascularization and inflammation, the transplanted tissue resembled conjunctiva with numerous goblet cells and 3-4 cell layers of nonkeratinized, stratified epithelium. In one patient, a definite "transition zone" from conjunctival to corneal appearing epithelium was seen in an area with minimal inflammation and vascularization, again confirming the animal results showing that transdifferentiation of conjunctival to corneal epithelium can occur.     

The conjunctiva in corneal epithelial wound healing

BACKGROUND/AIMS—During the healing of corneal epithelial wounds with limbal involvement, conjunctival epithelium often migrates across the denuded limbus to cover the corneal surface. It is believed that, over a period of time, conjunctival epithelium covering the cornea assumes characteristics of corneal epithelium by a process referred to as conjunctival transdifferentiation. The purpose of this study was to examine, clinically, the fate of conjunctival epithelial cells covering the cornea and to assess the healing of corneal epithelial wounds when the conjunctival epithelium was removed or actively prevented from crossing the limbus and extending onto the cornea.
METHODS—10 patients with conjunctivalisation of the cornea were followed for an average of 7.5 months. Five patients in this group had their conjunctival epithelium removed from the corneal surface and allowed to heal from the remaining intact corneal epithelium. In another four patients with corneal epithelial defects, the conjunctival epithelium was actively prevented from crossing the limbus by mechanically scraping it off.
RESULTS—The area of cornea covered by conjunctival epithelium appeared thin, irregular, attracted new vessels and was prone to recurrent erosions. Conjunctivalisation of the visual axis affected vision. Removal of conjunctival epithelium from the cornea allowed cells of corneal epithelial phenotype to cover the denuded area with alleviation of symptoms and improvement of vision. It was also established that migration of conjunctival epithelium onto corneal surface could be anticipated by close monitoring of the healing of corneal epithelial wounds, and prevented by scraping off conjunctival epithelium before it reached the limbus.
CONCLUSION—This study shows that there is little clinical evidence to support the concept that conjunctival transdifferentiation per se, occurs in humans. "Replacement" of conjunctival epithelium by corneal epithelial cells may be an important mechanism by which conjunctival "transdifferentiation" may occur. In patients with partial stem cell deficiency this approach can be a useful and effective alternative to partial limbal transplantation, as is currently practised.

Keywords: corneal epithelium; conjunctiva; stem cells; transdifferentiation